YAP Regulates Hematopoietic Stem Cell Formation in Response to the Biomechanical Forces of Blood Flow

SUMMARY

Hematopoietic stem and progenitor cells (HSPCs), first specified from hemogenic endothelium (HE) in the ventral dorsal aorta (VDA), support lifelong hematopoiesis. Their de novo production promises significant therapeutic value; however, current in vitro approaches cannot efficiently generate multi-potent long-lived HSPCs. Presuming this reflects a lack of extrinsic cues normally impacting the VDA, we devised a human dorsal aorta-on-a-chip platform that identified YAP as a cyclic stretch-induced regulator of HSPC formation. In the zebrafish VDA, inducible Yap-overexpression significantly increased runx1 expression in vivo, and the number of CD41⁺ HSPCs downstream of HE specification. Endogenous Yap activation by lats1/2 knockdown or Rho-GTPase stimulation mimicked Yap-overexpression, and induced HSPCs in embryos lacking blood flow. Notably, in static human iPSC-derived HE culture, compound-mediated YAP activation enhanced RUNX1 levels and hematopoietic colony forming potential. Together, our findings reveal a potent impact of hemodynamic Rho-YAP mechanotransduction on HE fate, relevant to de novo human HSPC production.

eTOC

Hemodynamic forces are required for embryonic hematopoietic stem and progenitor cell (HSPC) formation, however transcriptional mediators of this process are unknown. Using in vitro microfluidics-mediated stimulation of human iPS cells and in vivo functional analyses in zebrafish, Lundin et al. show cyclic stretch-mediated biophysical activation of YAP facilitates HSPC production.

Graphical Abstract

INTRODUCTION

Hematopoietic stem cells (HSCs) form the foundation of the blood system, as they can both self-renew and differentiate into all mature lineages. The production of patient-specific hematopoietic stem and progenitor cells (HSPCs) from pluripotent cells for clinical use has been a long-standing pursuit in the field. However, despite their therapeutic value, methods to derive or expand human HSPCs in vitro remain inefficient (Doulatov et al., 2013; Ditadi, et al., 2015; Sugimura et al., 2017), resulting in limited multipotency and long-term function. HSCs are first produced in the embryo from specialized hemogenic endothelium (HE) along the ventral wall of the dorsal aorta (VDA) (Dzierzak and Speck, 2008) and exhibit the unique and transient ability to expand without loss of “stemness” (Zape et al., 2017). Therefore, a complete understanding of endogenous mechanisms that promote and maintain developmental HSC commitment in vivo is essential for improving efforts toward the de novo production of fully functional human HSCs.

Recent work has revealed the importance of the local embryonic environment in regulating HE specification and HSPC production (Clements et al., 2011; Kwan et al., 2016). In particular, we previously demonstrated that blood flow promotes HSC formation in zebrafish and mouse embryos (Adamo et al., 2009; North et al., 2009), initiating their emergence from HE after the onset of heartbeat. Loss of blood flow in silent heart zebrafish and Ncx1^−/− mice significantly decreased expression of the essential transcriptional regulator of endothelial-to-hematopoietic transition (EHT), RUNX1 (Chen et al., 2009b; Kissa and Herbomel, 2010; North et al., 2002) and HSC number (Adamo et al., 2009; North et al., 2009). In contrast, application of wall shear stress (WSS) to dissociated murine para-aortic splanchnopleura, the precursor of the aorta-gonad-mesonephros (AGM) region, stimulated HSPC production, enabling long-term engraftment and lymphoid potential (Adamo et al., 2009; Diaz et al., 2015a). Nitric oxide (NO), a second messenger induced by WSS, was necessary and sufficient to drive HSPC formation in vivo and in vitro, and rescued HSPC production in embryos lacking a heartbeat and circulation (Adamo et al., 2009; North et al., 2009). Subsequent analyses identified a role for klf2a upstream of NO, as well as the involvement of flow-induced cyclic adenosine monophosphate (cAMP) and protein kinase A (PKA) signaling in HSC emergence (Diaz et al., 2015b; Kim et al., 2015; Wang et al., 2011). However, it remains unclear whether WSS is the only relevant biomechanical force involved in HSC production, how forces are sensed and transduced to impact cell fate, and, most importantly, whether biophysical regulation is relevant to unlocking human HSC function in vivo and in vitro.

Mechanical forces are increasingly recognized as cues that govern cell fate by directing maintenance or differentiation during development or tissue repair (reviewed in Vining & Mooney, 2017). Beyond its role as the transcriptional effector of the Hippo signaling pathway that regulates organ size, Yes-activated protein (YAP) has emerged as a transducer of physical stimuli in the microenvironment (Dupont et al., 2011). YAP translocates to the nucleus to control transcriptional regulation of cell fate in response to biophysical cues, such as cell density and extracellular matrix stiffness (Panciera et al., 2017). Guided by hemodynamics, YAP signaling is involved in endothelial maintenance during angiogenic remodeling (Azad et al., 2018; Nakajima et al., 2017; Wang et al., 2017). Moreover, YAP is required for pluripotency in mouse embryonic stem cells (Lian et al., 2010). While an essential role for YAP in adult HSC homeostasis in the bone marrow has been excluded (Jansson and Larsson, 2012), the potential impact of YAP signaling on HSPC development has not been defined in vivo due to early embryonic lethality in murine models (Morin-Kensicki et al., 2006). Interestingly, a genome-wide study of in vitro hematopoietic differentiation uncovered a YAP/Transcriptional enhancer factor domain (TEAD) signature (Goode et al., 2016) during mouse hematopoiesis. However, a role for YAP in HE biology, including its potential function in mechanotransduction and transcriptional commitment to HSPC fate in response to embryonic blood flow remains unexplored.

Microfluidic organ-on-a-chip technology has emerged as a powerful tool to enable physiologic modeling of functional human organ units that are otherwise prohibitive to study in vivo. Using human induced pluripotent stem cells to generate HE (iPSC-HECs), we engineered an in vitro model of the human dorsal aorta, allowing direct study of the effects of flow-related forces on human HSPC formation. Employing this dorsal aorta-on-a-chip platform, we determined that YAP signaling is activated in HE in response to blood flow-associated circumferential strain (CS). These findings were corroborated in vivo, whereby loss of Yap activity in zebrafish embryos blocked HSPC production in the VDA, and overexpression of constitutively active Yap increased HSPC number. Mechanistically, YAP is essential for maintenance, but not initiation, of hematopoietic commitment in HE, contributing to HSPC maturation and proliferation. Furthermore, epistasis analysis revealed an essential intermediary role for the Rho family of GTPases upstream of YAP, and downstream of CS, which can be directly targeted to stimulate YAP-mediated induction of HSPC production in vivo and in vitro. Collectively, our investigation defines a previously uncharacterized requirement for YAP as a conserved, pharmacologically amenable, mechanotransducer of flow-associated transcriptional commitment to HSPC fate.

RESULTS

Biophysical forces regulate RUNX1 expression and YAP signaling

As blood flow promotes definitive hematopoiesis in mice and zebrafish, we sought to determine the mechanistic impact of biomechanical forces on human HSPC production. Human iPSCs were converted into definitive HE using established protocols (Sturgeon et al., 2014) (Fig S1A) and flow cytometric analysis of embryoid bodies (EBs) on day 7–9 of differentiation (D7–9) identified a robust GlyA⁻/CD45⁻/CD34⁺/KDR⁺ population, indicative of definitive HE, on D7 (Fig S1B–C), which was used for subsequent studies. Hematopoietic potential was assessed by colony forming unit (CFU) assays (Fig S1D) and phenotypic endothelial function was confirmed via a standard tube forming assay, comparing endothelial cord formation from D7 HE to human umbilical vein endothelial cells (Fig S1E). After seeding on thin-layer Matrigel in hematopoietic media, iPSC-HECs permitted to undergo in vitro EHT over the next 7 days (D7+7) (Fig S1F) generated non-adherent CD34⁺/CD45⁺ HSPCs (Fig S1G) with erythro-myeloid potential when plated into CFU media (Fig S1H–I), similar to CD34⁺ umbilical cord blood (CB) or peripheral blood (PB) cells (Fig S1J).

Upon functional validation of our starting population, iPSC-HECs were introduced to a 2-channel microfluidic culture device to create an in vitro dorsal aorta-on-a-chip (Fig 1A). To mimic biophysical forces in the embryonic vasculature, the endothelial monolayer was exposed to shear stress mediated by continuous medium flow through the channel, emulating physiologic WSS, or to cyclic stretching exerted by applying pump-controlled cyclic suction to the side chambers, causing the flexible membrane with the adherent cell layer to stretch and relax, modeling CS (Fig 1B). Based on prior work indicating size similarities of human and mouse embryos at the onset of circulation (Mu et al., 2008; Papaioannou et al., 2010), we presumed human HE experiences a comparable magnitude of force acting along the dorsal aorta to that of the mouse. Thus, we subjected the human HE to WSS of 5dyn/cm², approximating embryonic day 10 in the mouse (Adamo et al., 2009), or to CS at 2Hz (10%), which corresponds to embryonic heart rate when blood flow is established around 7 weeks of human gestation (van Heeswijk et al., 1990). HE exposed to WSS or CS for 72 hours showed characteristic endothelial cell elongation and alignment, in parallel or perpendicular directions, respectively (Fig 1B). Gene expression analysis revealed significantly increased RUNX1 expression by qPCR in response to WSS compared to static control, similar to prior observations in zebrafish and mouse, supporting evolutionary conservation. Interestingly, application of CS also significantly increased RUNX1 expression (Fig 1C). Notably, while elevated WSS of 10dyn/cm² had no effect (data not shown), consistent with prior reports, application of either WSS or CS at lower magnitudes (WSS=1dyn/cm²; CS=1Hz) also did not impact RUNX1 expression, indicating that HE is responsive to biomechanical forces within discrete ranges (Fig S2A–D).

Figure 1. YAP signaling is active in force-mediated RUNX1 expression in human HE.

(A) Schematic approach of engineering dorsal aorta-on-a-chip. Differentiation of human iPSCs into CD34⁺ HE cells (iPSC-HEC) seeded in microchannel devices. Cells subjected to wall shear stress [WSS (5dyn/cm²)] or cyclic strain [CS (2Hz, 10% strain)] to mimic blood flow or circumferential strain, respectively. iPSCs, induced pluripotent stem cells; EBs, embryoid bodies; HE, hemogenic endothelium.

(B) Brightfield images of iPSC-HECs after exposure to WSS or CS for 72h in dorsal aorta-on-chip compared to static control. Arrows indicate direction of force.

(C) RUNX1 expression in HE after 72h of WSS (5dyn/cm²) or CS (2Hz, 10% strain), quantified by qPCR (n=6; unpaired Student’s t-test, **p<0.01. Error bars indicate S.E.M.).

(D) qPCR of HE upon WSS or CS for 72h (n=3 for WSS; 6 for CS; unpaired Student’s t-test, *p<0.05. Error bars indicate S.E.M.).

(E) Immunostaining for YAP, DAPI and F-actin in iPSC-HECs in static control chip or 72h of CS. Enlarged boxed region highlights YAP localization. Scale bar is 400μm.

(F) Quantification of 1E showing the ratio of nucleus over cytoplasm fluorescent intensity. Higher number indicates more co-localization of YAP with DAPI (n=4–23 images/condition from three independent experiments; unpaired Student’s t-test, *p<0.05. Error bars indicate S.E.M.).

(G) CFU assay scored for G/M/GM/E/GEMM colonies from iPSC-HECs after CS and static control (n=2, One-way ANOVA, *p<0.05. Error bars indicate S.E.M.).

To investigate potential downstream mechanisms for the effects of WSS and CS on RUNX1, we analyzed common targets of the transcriptional regulators YAP/TAZ, which are established mediators of mechanical stimuli (Dupont et al., 2011). CS significantly upregulated expression of the YAP targets ANKRD1 and CTGF by qPCR (Fig 1D) and promoted nuclear localization of YAP (Fig 1E–F), suggesting that CS activates YAP signaling and drives RUNX1 expression in human HE cells in vitro. Interestingly, WSS exposure downregulated ANKRD1 and CTGF expression compared to static control, consistent with prior work showing sustained fluid shear stress excludes YAP from the nucleus, suppressing downstream targets (Nakajima et al., 2017). Finally, to assess the functional impact of CS on HE, we performed CFU analysis and found that CS significantly increased multi-potent hematopoietic colony formation (Fig 1G). Together, these findings indicate that WSS and CS impact HSPC production in vitro through distinct mechanisms.

Yap signaling instructs HSPC specification, production, and maturation in vivo

To corroborate the relevance of YAP regulation identified in the human dorsal aorta-on-a-chip assays to HSPC development in vivo, we utilized the zebrafish model. In zebrafish, runx1⁺ HSPCs begin to appear in the VDA shortly after the onset of the heartbeat at 24 hours post fertilization (hpf). Following EHT, HSPCs migrate to the caudal hematopoietic tissue (CHT) to expand and differentiate, before seeding the thymus and kidney marrow (reviewed in Frame et al., 2017). Loss of Yap activity in yap1 ^mw48 [yap1^−/−] zebrafish mutants (Miesfeld et al., 2015) reduced expression of the conserved HSPC markers runx1 and cmyb in the VDA by whole mount in situ hybridization (WISH) at 36hpf compared to wildtype (WT) controls (Fig 2A–B). In addition to decreased expression of Yap target ctgfa by whole-embryo qPCR analysis at 36hpf, yap1^−/− mutants had significantly reduced expression of runx1 and cmyb, as well as the vascular marker kdrl, which is retained on nascent HSPCs (Bertrand et al., 2010) (Fig 2C). Importantly, embryonic heartbeat (data not shown), kdrl⁺ vascular structure and acquisition of arterial-restricted efnb2a expression appeared normal in yap1^−/− mutants (Fig S3A–D). Live-imaging of Tg(flk1:GFP) reporter embryos at 36hpf showed normal patterning and lumenization of the axial and segmental vessels in yap1^−/− fish compared to their WT siblings, with a reduction in GFP⁺ clusters of presumptive HSPCs in the VDA (Fig S3E). These results suggest that the effects of Yap loss are not due to gross vascular abnormalities, improper arterial specification or developmental delay, but rather indicate a defect in HSPC formation in the absence of Yap function.

Figure 2. Yap signaling impacts HSPC production in the zebrafish embryo.

(A) Whole mount in situ hybridization (WISH) for runx1/cmyb in control and yap1^mw48 mutants at 36hpf. Scale bar is 100μm.

(B) Qualitative phenotypic distribution plot of embryos in 2A scored with low, medium or high runx1/cmyb expression in the VDA at 36hpf (n=32 control; 33 yap1^mw48, Chi-squared test, ***p<0.001).

(C) Whole-embryo qPCR on control and yap1^mw48 mutants at 36hpf (unpaired Student’s t-test, *p<0.05, ***p<0.001. Error bars indicate S.E.M.).

(D) WISH for runx1/cmyb at 36hpf following heat-shock mediated overexpression of constitutively active Yap (Yap^S87A, hs:Yap) at 12hpf. Scale bar is 100μm.

(E) Distribution plot of embryos in 2D (n=53 control; 48 hs:Yap, Chi-squared test, ***p<0.001).

(F) qPCR in 36hpf embryos following Yap-OE at 12hpf (unpaired Student’s t-test, *p<0.05, ***p<0.001. Error bars indicate S.E.M.).

(G) In vivo imaging of Cd41:GFP in the CHT at 72hpf after YAP-OE. Scale bar is 200μm.

(H) Quantification of fluorescent images in 2G (n=9 control; 7 hs:Yap, unpaired Student’s t-test, *p<0.05. Error bars indicate S.D.).

(I) rag1 expression in the thymus at 120hpf with Yap-OE (n=21). Scale bar is 100μm.

(J) Quantification of thymic area in 2I (n=8, unpaired Student’s t-test, *p<0.05. Error bars indicate S.D.).

(K) WISH for runx1/cmyb at 36hpf following Yap-OE and sih morpholino (MO). Black bracket denotes VDA; arrowhead indicates uncleared primitive erythrocytes in the vein due to lack of blood flow. Scale bar is 100μm.

(L) Distribution plot of embryos in 2K (n≥30/condition, Chi-squared test, ***p<0.001).

(M) Quantification of runx1/cmyb staining intensity in 2K (n≥30/condition, two-way ANOVA and Tukey’s multiple comparison test, *p<0.05, ***p<0.001. Error bars indicate S.D.)

In contrast, heat-shock mediated overexpression of activated Yap (Tg(Hsp70:Yap^S87A); hereafter referred to as Yap-OE) (Cox et al., 2018) at 12hpf increased runx1 and cmyb expression in the VDA at 36hpf (Fig 2D–E), without impacting Kdrl⁺ vascular structure in Tg(flk1:GFP) reporter embryos (Fig S3F) (Choi et al., 2007) or kdrl expression by qPCR (Fig 2F). Levels of ctgfa increased with Yap-OE, and whereas only a trend towards enhanced runx1 expression was seen, the mature HSPC marker cmyb was already significantly increased at 36hpf (Fig 2F), suggesting Yap activation can accelerate HSPC specification, maturation or expansion. In support of this hypothesis, enhanced runx1/cmyb expression was already present in the VDA at 24 and 30hpf under this heat-shock regimen (Fig S3G–I). Furthermore, Yap-OE led to a modest increase in runx1, cmyb and ctgfa expression by qPCR by 30hpf (Fig S3J). These data are consistent with a role for Yap in promoting the production of HSPCs.

To determine whether Yap activation had an impact on HSPC number in addition to gene expression, we utilized a Tg(−6.0itga2b:GFP) [Cd41:GFP] reporter fish (Bertrand et al., 2008) in combination with the inducible Yap model. Yap-OE at 12hpf significantly increased Cd41⁺ HSPCs present in the CHT at 72hpf (Fig 2G–H), as well as rag1⁺ thymic area at 120hpf (Fig 2I, J), indicative of increased lymphoid progenitor number. Finally, to investigate the physiological relationship between mechanical forces and Yap activation in vivo, we examined silent heart (sih) morphants, which lack a heartbeat and blood flow (Sehnert et al., 2002). WISH revealed very low levels of runx1/cmyb HSPC expression in morphant embryos (Fig 2K, L), consistent with prior studies (North et al., 2009). Strikingly, Yap-OE in sih morphants restored runx1/cmyb expression in the VDA at 36hpf (Fig 2K–M), indicating Yap activation is sufficient to bypass the requirement of blood flow in embryonic HSPC formation.

Endogenous levels of active Yap are sufficient to promote hematopoiesis

As our loss- and gain-of-function experiments suggested Yap regulates HSPC production, we next sought to confirm that Yap signaling was active in the VDA. To this end, we utilized a Tg(ctgfa:EGFP) reporter line whereby GFP positivity acts as a proxy for Yap activity in vivo (Mokalled et al., 2016); reporter efficacy was validated by induction with Yap-OE (Fig S4A, B). At 36hpf, discrete GFP labeling was present throughout the embryo, with strong localization in the notochord above the VDA (Fig 3A, inset). Examination of single confocal slices through the VDA of bigenic Yap- and vascular-reporter embryos at high magnification demonstrated the presence of Kdrl⁺/Ctgfa⁺ cells in the endothelial floor of the dorsal aorta (Fig 3B). The presence of endogenous ctgfa in the VDA was confirmed by fluorescent in situ hybridization, where cells co-expressing both the vascular marker fli1a and ctgfa were identified (Fig 3C, D).

Figure 3. Endogenous Yap signaling is sufficient to induce HSPC production.

(A) Maximum intensity projection (max proj.) of confocal z-stack of double transgenic embryo at 36hpf. Kdrl⁺ endothelium in magenta, Ctgfa⁺ cells in cyan. Dashed box denotes VDA, dotted line highlights region magnified in 3B (scale bar: 500μm in overview; 50μm in inset). DA: dorsal aorta; PCV: posterior cardinal vein

(B) Merge and single-color slices from 3B showing Kdrl⁺ membrane surrounding Ctgfa⁺ cells (yellow arrowheads) in the VDA floor. Scale bar is 20μm.

(C) Max proj. of confocal z-stack of 36hpf WT embryo. Double fluorescent in situ hybridization of endogenous mRNA expression for fli1a (magenta) and ctgfa (cyan). Bright stipe of ctgfa in notochord (yellow arrowhead) mirrors GFP expression in the reporter line. Dashed box denotes VDA region shown in Fig 3D. Scale bar is 50μm.

(D) Merge and single-color slices from confocal z-stack showing a pair of neighboring fli1a⁺ ctgfa⁺ cells in the VDA floor. Scale bar is 10μm.

(E) qPCR in lats1/2 morphants at 36hpf (**p<0.01. Error bars indicate S.E.M.).

(F) In vivo imaging for Cd41:GFP in the CHT at 72hpf in control and lats1/2 morphants. Scale bar is 200μm.

(G) Quantification of fluorescent images in 3F (n=9, unpaired Student’s t-test, **p<0.01. Error bars indicate S.D.).

Active Hippo signaling sequesters Yap in the cytoplasm via action of the Lats1/2 kinases, which phosphorylate Yap to prevent translocation to the nucleus, repressing its transcriptional activity (Zhao et al., 2008). We hypothesized that knockdown of lats1/2 in zebrafish embryos could de-repress endogenous Yap signaling, and asked whether the physiologic levels of Yap present during development were sufficient to drive HSPC production. In vivo microscopy of the Ctgfa:GFP reporter embryos injected with previously validated morpholinos targeting lats1/2 (Chen et al., 2009) showed elevated Yap signaling at 36hpf, including the VDA region (Fig S4C, D), similar to that seen for Yap-OE (Fig S4A, B), implying loss of Lats1/2 regulation phenocopies enforced Yap activation. Whereas WISH for kdrl confirmed that lats1/2 morphants had normal gross vascular structure (Fig S4E, F), expression of runx1, cmyb, and ctgfa were each increased by qPCR at 36hpf (Fig 3E). The effects of Yap de-repression were maintained throughout development as lats1/2 morphants also had significantly increased numbers of Cd41:GFP⁺ HSPCs in the CHT at 72hpf (Fig 3F, G). Therefore, in absence of negative regulation, endogenous embryonic levels of Yap are sufficient to drive HSPC production.

Yap is required for maintenance, not initiation, of the hematopoietic program in HE

We next aimed to define where Yap activity is required in the developmental continuum of EHT from HE specification to HSPC production. Analysis of the pan-endothelial markers kdrl, fli1a and cdh5 at 24hpf showed normal endothelial identity in yap1^−/− mutants compared to WT (Fig S5A); similarly, yap1^−/− mutants showed normal partitioning of tbx20 and flt1 to the dorsal aorta, and restriction of dab2 to the posterior cardinal vein (Fig S5B), indicating Yap is not required for proper arteriovenous patterning. We next analyzed expression of hematopoietic transcription factors showing progressive commitment to HSPC fate: gata2b, runx1, and cmyb at two different timepoints of EHT, 24 and 30hpf, as circulation is initiated, and an intermediate stage where all embryos have hemodynamic flow through the aorta, respectively. At 24hpf, both WT and yap1^−/− mutants have similarly low levels of runx1 and cmyb expression (Fig 4A–B). However, whereas WT embryos show increased levels of expression by 30hpf, this progression is completely blunted in yap1^−/− mutants (Fig 4C–D). In contrast, gata2b expression is detected equally in both WT and yap1^−/− mutants at 24hpf, and expands in the same manner by 30hpf. Importantly, loss of HSPC markers was not due to cell death, as Acridine Orange staining of yap1^−/− mutant embryos at 30hpf showed no difference in apoptosis in the VDA compared to controls (Fig S5C–D). Together, these results suggest that Yap is not required to initiate HE specification, but is essential to maintain or progress toward HSPC production following the onset of blood flow.

Figure 4. YAP activity is required for maintenance, but not acquisition, of hematopoietic commitment in HE.

(A) WISH for runx1, cmyb and gata2b at 24hpf in WT and yap1^mw48 mutants. Black arrowheads indicate putative HE cells expressing cmyb in VDA versus faintly stained primitive erythrocytes not yet mobilized by circulation. Scale bar is 100μm.

(B) Distribution plot of embryos in 4A (n≥30/condition).

(C) WISH for runx1, cmyb and gata2b at 30hpf in WT and yap1^mw48 mutants.

(D) Distribution plot of embryos in 4C (n≥30/condition, Chi-squared test, ***p<0.001).

(E) WISH for runx1/cmyb at 36hpf upon Yap-OE at 12hpf and treatment with DMSO or DAPT from 12–36hpf. Scale bar is 100μm.

(F) Distribution plot of embryos in 4E (n≥30/condition, Chi-squared test, ***p<0.001).

(G) Quantification of runx1/cmyb staining intensity in 4F (n≥30/condition, two-way ANOVA and Tukey’s multiple comparison test, **p<0.01, ***p<0.001. Error bars indicate S.D.)

(H) Max proj. of confocal z-stack of Flk1:GFP embryos (cyan) at 36hpf following Yap-OE. EDU⁺ nuclei shown in magenta. Green arrowheads denote dorsal EDU⁺ endothelial nuclei in the aorta. Yellow arrowheads indicate Flk1⁺ EDU⁺ cells in the VDA and perivascular space between the DA and PCV. Scale bar is 50μm.

(I) Quantification of Flk1⁺ EDU⁺ cells in 4H (n=20, unpaired Student’s t-test, ***p<0.001. Error bars indicate S.D.).

To directly test this hypothesis, epistasis analysis with Notch signaling, required for gata2b⁺ HE specification (Butko et al., 2015; Robert-Moreno et al., 2005), was performed. Yap-OE was not sufficient to overcome loss of Notch activity in DAPT-treated embryos and drive runx1/cmyb expression (Fig 4E–G), consistent with a role downstream of HE specification. To determine if Yap-OE influenced HSPC expansion, we performed proliferation studies. At 36hpf, a ~2-fold increase in the number of EDU⁺/Flk1⁺ cells in the VDA was detected, suggesting that Yap activity impacts HSPC maturation, including subsequent proliferative expansion (Fig 4H–I). Finally, we evaluated whether Yap-OE during EHT would impact the ability of HSPCs to populate the blood system at later stages. Analysis of lineage markers at 120hpf showed that early (14hpf) Yap activation did not hinder the ability of HSPCs to home to secondary hematopoietic organs or preclude differentiation toward the lymphoid, myeloid and erythroid lineages (Fig S5E–G).

Rho-GTPases control in vivo HSPC production by modulating Yap activity

As Yap appears to be a physiologically relevant regulator of HSPC production, we next sought to identify mechanotransducers that relay signals from blood flow to Yap. While Rho-GTPases are traditionally linked to cell migration and proliferation, they are also mechanosensors that interface with YAP signaling (Dupont et al., 2011). Therefore, we postulated that Rho-GTPases are involved in Yap-mediated HSPC development. Treatment with the small G-protein activators CN02, CN03 and CN04 (0.1μg/ml, 1μg/ml and 1μg/ml, respectively) from 12–36hpf increased runx1/cmyb expression by WISH (Fig 5A–B), and elevated Ctgfa:GFP expression in the VDA at 36hpf (Fig S6A–B). CN02 and CN04 also significantly increased runx1 and ctgfa expression by qPCR at 36hpf (Fig 5C), indicating that Rho-GTPase activation induces Yap signaling and promotes HSPC formation in vivo. In contrast, antagonism mediated via exposure to the Rho inhibitor CT04 (1μg/ml; 12–36hpf), decreased runx1/cmyb expression by WISH and significantly lowered runx1 by qPCR (Fig S6C–E); however, at this dose, CT04 did not significantly decrease ctgfa expression, suggesting additional factors may be capable of acting upstream of Yap during hematopoiesis. Importantly, the positive effects of Rho-GTPase activation were maintained throughout development, with CN02 or CN04 treatment (12–72hpf) significantly increasing CD41⁺ HSPCs in the CHT (Fig 5D).

Figure 5. Rho-GTPases drive Yap mechanotransduction and HSPC production in vivo.

(A) WISH for runx1/cmyb at 36hpf in WT embryos after treatment (12–36hpf) with Rho-GTPase activators CN02, CN03 and CN04. Scale bar is 100μm.

(B) Distribution plot of embryos in 5A (n≥50/condition, Chi-squared test, *p<0.05, ***p<0.001).

(C) qPCR after treatment with CN02 or CN04 (unpaired Student’s t-test versus control, *p<0.05, **p<0.01, ***p<0.001. Error bars indicate S.E.M.).

(D) Quantification of Cd41:GFP⁺ cells in images of the CHT at 72hpf (n=30 control; 23 CN02; 22 CN04, unpaired Student’s t-test, ***p<0.001, Error bars indicate S.D.).

(E) WISH for runx1/cmyb at 36hpf in sih morphants after CN02 treatment (12–36hpf). Scale bar is 100μm.

(F) Distribution plot of embryos in 5F (n≥50/condition, Chi-squared test, ***p<0.001).

(G) qPCR in WT embryos and sih morphants at 36hpf after CN02 treatment (unpaired Student’s t-test, **p<0.01, ***p<0.001. Error bars indicate S.E.M.).

We next probed functional intersections between Rho family GTPase activity and blood flow. As anticipated, CN02 treatment (12–36hpf) and sih morpholino injection each had statistically significant, opposing impact on runx1/cmyb expression by WISH and runx1 expression by qPCR (Fig 5E–G). Furthermore, CN02 activation in sih morphants partially rescued the phenotypic levels of runx1/cmyb in the VDA at 36hpf, and significantly increased runx1 expression by qPCR compared to sih morphants alone (Fig 5E–G). Notably, Rho-GTPase activation was unable to induce HSPC production in the absence of Yap function: unlike their WT counterparts, runx1/cmyb expression did not respond positively to CN02 stimulation in yap1^−/− mutants (Fig S6F–G), indicating activation of Rho-GTPase function modulates definitive hematopoiesis in a Yap-dependent manner. Together with our earlier observation for Yap-OE (Fig 2K–M), these findings suggest a pathway in which mechanical forces induced by blood flow stimulate Rho-GTPase function to activate Yap-mediated HSPC production in vivo.

Activation of physiologic YAP signaling increases HSPC potential in vitro

Toward our long-term goal of improving protocols for the generation of clinical-grade HSPCs in vitro, we optimized an alternative strategy to efficiently derive GlyA⁻/CD45⁻/CD34⁺/KDR⁺ HE from iPSCs under serum- and feeder-free conditions (Fig S7A–C). Using this two-dimensional (2D) differentiation system, phenotypic and functional resemblance of these cells to EB-derived HE used in our dorsal aorta-on-a-chip was verified based on colony forming potential and ability to undergo in vitro EHT to produce HSPCs (Fig S7D–F). Employing these iPSC-HECs in conventional, static cultures, we aimed to characterize the impact of enhancing endogenous YAP signaling on definitive hematopoiesis. siRNA-mediated knockdown of the LATS1/2 kinases (siLL) in human HE following overnight culture significantly increased YAP nuclear localization (Fig 6A–B), enhanced expression of the YAP target genes ANKRD1 and CTGF (Fig 6C), and increased total YAP protein levels by Western blot relative to controls (siNC) (Fig 6D), suggestive of de-repressed YAP signaling. Furthermore, knockdown of LATS1/2 significantly upregulated expression of RUNX1 and the early hematopoietic commitment marker CD43 (Fig 6C), and significantly increased total numbers of hematopoietic colonies in CFU assays (Fig 6E), including the proportion of multi-potent GEMM colonies. Together these findings demonstrate that targeted elevation of YAP signaling in iPSC-based cultures promotes the formation of functional human HSPCs from HE.

Figure 6. Endogenous YAP activation promotes human HSPC formation in static culture.

(A) Immunostaining for YAP, DAPI and F-actin in iPSC-HECs in static well plates, 24h after knockdown of LATS1/2 (siLL) or negative control (siNC). Scale bar is 200μm.

(B) Quantification of 6A (n=10 images/condition from two independent experiments; unpaired Student’s t-test, **p<0.01. Error bars indicate S.E.M.).

(C) qPCR of HE 24h after siLL knockdown (n=7; unpaired Student’s t-test, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Error bars indicate S.E.M.).

(D) Western blot analysis for YAP (predicted molecular weight: 65 kDa) in iPSC-HECs 72h after siLL or siNC. GAPDH (predicted molecular weight: 36kDa) used as loading control. Two samples shown are from the same membrane.

(E) CFU assay of iPSC-HECs after siLL knockdown compared to siNC (n=6, One-way ANOVA, ***p<0.001. Error bars indicate S.E.M.).

Rho-GTPases regulate YAP activity and HSPC production in vitro

Having shown endogenous Rho-GTPase stimulation can mediate the connection between blood flow and Yap activation during zebrafish HSPC production, we next asked whether the same was true in the human dorsal aorta-on-a-chip. As expected, exposing iPSC-HECs to CS increased expression of ANKRD1 and CTGF; however, simultaneous CT04-mediated (200 ng/ml) Rho inhibition abolished CS-responsive induction of YAP targets (Fig 7A). Further, immunostaining of HE in static cultures showed that CT04 treatment inhibited nuclear translocation of YAP (Fig 7B–C), together indicating Rho-GTPases are conserved essential intermediaries between biomechanical stimuli and YAP-dependent transcriptional regulation.

Figure 7. YAP drives hematopoietic specification in a Rho-dependent manner in vitro.

(A) qPCR of human HE in the dorsal aorta-on-a-chip 72h after CS exposure with or without the Rho inhibitor CT04 (n=3, One-way ANOVA, *p<0.05, **p<0.01. Error bars indicate S.E.M.).

(B) Immunostaining for YAP, DAPI and F-actin in HE in static well plates, upon 3h stimulation with Rho-GTPase inhibitor CT04 or activators CN03 and CN04. Scale bar is 200μm.

(C) Quantification of 7B (n=10 images/condition from three independent experiments, One-way ANOVA, **p<0.01, ****p<0.0001. Error bars indicate S.E.M.).

(D) qPCR of HE treated with CN03 or CN04 for 24h (n=2–5; unpaired Student’s t-test versus control, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Error bars indicate S.E.M.).

(E) qPCR of HE treated with CN03 or CN04 for 3–5days (n=4; unpaired Student’s t-test versus control, *p<0.05, **p<0.01. Error bars indicate S.E.M.).

(F) Representative FACS analysis for CD43⁺ on CD34⁺ HE treated with CN03 or CN04 for 24–48h.

(G) Quantification of 7F (n=5; unpaired Student’s t-test, *p<0.05. Error bars indicate S.E.M.).

(H) CFU assay of iPSC-HECs following treatment with CN03 or CN04 compared to control (n=4, One-way ANOVA, *p<0.05. Error bars indicate S.E.M.).

We next asked if pharmacologic YAP activation via Rho-GTPase stimulation would be sufficient to drive HSPC fate from human HE in the absence of mechanical stimuli or genetic manipulation. Two-hour exposure of iPSC-HECs to CN03 (250 ng/ml) or CN04 (125 ng/ml) promoted YAP nuclear localization (Fig 7B–C). CN03 or CN04 stimulation for 24 hours during EHT significantly increased gene expression of ANKRD1 and CTGF, similarly to CS; levels of the marker of hematopoietic commitment, CD45, were also significantly elevated (Fig 7D). Moreover, treatment with CN03 or CN04 led to a significant increase in RUNX1, reflective of commitment to HSPC fate (Fig 7E). Consistent with this transcriptional impact of YAP activation, stimulation of endogenous Rho-GTPases significantly expanded the percentage of CD34⁺/CD43⁺ HSPCs produced from iPSC-HECs by flow cytometry (Fig 7F–G). Finally, while long-term exposure to CN04 at this concentration caused cell toxicity (data not shown), functional assessment of hematopoietic potential of HE treated with CN03 showed a significant increase in the production of total hematopoietic colonies (Fig 7H). Together, our data demonstrate that biomimetic YAP activation promotes derivation of transgene-, serum- and feeder-free HSPCs from human pluripotent cells in vitro.

DISCUSSION

The embryonic vasculature is continuously subjected to mechanical forces after the onset of heartbeat and circulation. While we previously showed that blood flow impacts HSPC formation in the embryonic dorsal aorta (Adamo et al., 2009; North et al., 2009), conservation in the human embryo and the underlying molecular mechanisms, including relevant transcriptional mediators, remained undefined. Here, using a specifically designed biomimetic platform, and in vivo zebrafish validation, we identified a physiologically relevant mechanotransductive mediator of blood flow that can significantly enhance functional HSPC production ex vivo. Our data reveal a previously unknown role for YAP-mediated mechanotransduction in definitive hematopoiesis, whereby blood flow stimulates YAP activity via a Rho-GTPase-mediated mechanism in the VDA. Unexpectedly, CS, rather than WSS, acted as the relevant physiological stimulus of YAP activation in vitro. This mechanosensory network is conserved between fish and mammals, and presents a unique pharmacological target to improve human HSPC production from pluripotent cells.

Our study reveals a connection between biomechanical signaling and YAP transcriptional regulation of hematopoietic fate in the vertebrate embryo. Prior investigations indicated Yap was dispensable for maintaining the hematopoietic compartment, or leukemogenesis, in mouse models (Donato et al., 2018; Jansson and Larsson, 2012). In contrast, bioinformatics analysis of murine hematopoietic development identified a YAP signature and nuclear localization in endothelium prior to EHT (Goode et al., 2016). However, neither the mechanism by which YAP was activated in HE nor its functional impact on EHT were determined. Our observation that YAP activation is necessary and sufficient to accelerate commitment of HE to Runx1⁺ HSPC fate in vivo validates and extends those findings. Indeed, the developmentally-cued onset of YAP signaling in Notch-specified gata2⁺ HE directly regulates the selective commitment and/or maturation of HSPCs in the VDA. Given their contemporaneous nature, additional tools and transcriptional profiling will be needed to further evaluate a facilitative or supplementary role in stimulating proliferative expansion of newly specified HSPCs. Finally, in addition to vascular defects known to arise in mice with endothelial-specific YAP/TAZ deletion (Wang et al., 2017) and flow-stimulated Yap stabilization of vessels during angiogenic remodeling in zebrafish (Nakajima et al., 2017), our data reveals an uncharacterized dimension for YAP signaling in the regulation of hematovascular fate in response to hemodynamics, which is conserved across species, including human.

A major finding of our analysis is the discovery that YAP signaling is induced by pulse-mediated CS. Using our biomimetic platform, we can now manipulate the distinct hemodynamic forces of blood flow without affecting other confounding variables, such as substrate stiffness or material. While WSS or CS individually induced RUNX1 expression, YAP was active under CS, but unaffected by WSS. This is in line with recent studies showing transient YAP nuclear localization in response to laminar shear stress, followed by cytoplasmic retention after prolonged WSS (Nakajima et al., 2017). Conversely, YAP was noted to be active in endothelial cells under disturbed flow (Wang et al., 2016), as well as in various adherent cells, such as fibroblasts, endothelial and epithelial cells, under stretch (Benham-Pyle et al., 2015; Elosegui-Artola et al., 2017; Neto et al., 2018). Our results suggest WSS and CS regulate RUNX1 and HSPC induction through distinct mechanisms. While we have previously implicated NO signaling in WSS-induced hematopoiesis (Adamo et al., 2009; North et al., 2009), potential crosstalk between YAP and other force-induced signaling pathways still requires extensive investigation. For example, emerging evidence points to crosstalk between COX2/PGE2 signaling and YAP regulation in a number of cancer lines (Cheng et al., 2016; Guerrant et al., 2016; Kim et al., 2017a), and conditional deletion of YAP/TAZ leads to NOS2 downregulation in sorted brain endothelial cells of mice (Kim et al., 2017b). Beyond these molecular effectors, known to impact HSCs, how YAP interprets and integrates the relative contribution, timing and magnitudes of WSS and CS on HE development and hematopoietic output will be an exciting avenue for further study.

Given the number of clinically-approved small molecules that modulate vascular tone and kinase activity, our finding that Rho-GTPases are essential for YAP-mediated mechanotransduction during developmental hematopoiesis in vivo and in vitro is of significant therapeutic interest. While Rho-GTPases are linked to a large number of processes in the vasculature, precise roles in the blood system have only been modestly investigated: Rac2^−/− mice have elevated numbers of circulating HSCs (Jansen et al., 2005), which show enhanced migration, suggestive of a role in bone marrow retention or migration (Cancelas et al., 2005), and fetal liver hematopoiesis is impaired in Rac1 deficient mice (Ghiaur et al., 2008). Recently, a requirement for actomyosin constriction was shown in cells undergoing EHT (Lancino et al., 2018). This is consistent with our observation that Rho inhibition reduced runx1 expression in vivo, as well as the effect of CS-induced YAP nuclear localization and RUNX1 expression in vitro. Notably, while Rho-GTPase-mediated Yap activation was sufficient to rescue HSPCs in embryos lacking a heartbeat and circulation, further work will be needed to determine which Rho family members are functionally relevant. However, our finding that compound mediated Rho-GTPase activation drove YAP nuclear localization and increased multilineage hematopoietic colony formation serves as a proof-of-concept that the regulatory impact of hemodynamic forces from blood flow can be mimicked in standard static cell culture to drive commitment to functional HSPCs.

In summary, our work delineates a previously unknown role for YAP-mediated mechanotransduction in embryonic hematopoiesis, whereby select mechanical forces induced by blood flow are sensed and transduced in the HE to promote HSPC commitment and expansion. Further characterization of the spatio-temporal dynamics and molecular players in this pathway, and the development of efficient strategies to recapitulate and modify these signaling cascades ex vivo, may greatly improve current protocols for in vitro production of HSPCs from human pluripotent sources, which has the potential to revolutionize our standard of care for leukemias, autoimmune diseases, and genetic blood disorders.

STAR Methods

Contact for Reagents and Resource Sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Trista E. North, PhD (Trista.North@childrens.harvard.edu).

Experimental Model and Subject Details

iPSC culture

The following human iPSC lines were used: CD45-iPSC (Vo et al., 2018), MSC-iPSC1 (Park et al., 2008) and 1157-iPSC (Sugimura et al., 2017) generated by the hESC Core Facility at Boston Children’s Hospital and Dub7-iPSC (Ferrell et al., 2015) generated and kindly provided by Dan Kaufman Laboratory, UCSD. Human iPSCs were maintained on hESC-qualified Matrigel (BD) in mTeSR1 media (Stem Cell Technologies). Cells were fed daily and clump passaged using Dispase or ReLeSR (Stem Cell Technologies).

Embryoid Body Differentiation

iPSC-HECs for in vitro organ-on-chip experiments were derived using embryoid body (EB) differentiation, as previously described (Sturgeon et al., 2014), with minor modifications. Prior to EB formation, human iPSCs were cultured for at least one passage on irradiated mouse embryonic fibroblasts (GlobalStem) in DMEM/F12 supplemented with 20% KnockOut Serum Replacement (Gibco), 1% penicillin/streptomycin, 1 mM L-Glutamine, 1 mM MEM-NEAA, 0.1 mM beta-mercaptoethanol and 10 ng/ml FGF2. Human iPSCs were treated with collagenase IV (Gibco) and gently scraped into small aggregates, which were resuspended in aggregation media consisting of Ham’s F-12:IMDM in a 1:3 ratio supplemented with 1% penicillin/streptomycin, 1% N-2, 2% B-27, 0.05% bovine serum albumin, 50 μg/ml ascorbic acid, 150 μg/ml holo-transferrin and 400 μM monothioglycerol. 10 ng/ml BMP4 was added on day 0 and 5 ng/ml FGF2 on day 1. Aggregation media was replaced on day 2 to include 6 μM SB431542, 3 μM CHIR99021 (Stem Cell Technologies), 5 ng/ml FGF2 and 10 ng/ml BMP4. On day 3 of differentiation, the developing EBs were collected and resuspended in EB media (StemPro-34, 1% penicillin/streptomycin, 2 mM L-Glutamine, 50 μg/ml ascorbic acid, 150 μg/ml holo-transferrin and 400 μM monothioglycerol) containing 5 ng/ml FGF2 and 15 ng/ml VEGF. On day 6, EB media was replaced now containing the following cytokines: 15 ng/ml VEGF (R&D Systems), 2 U/ml EPO (Amgen), 5 ng/ml FGF2, 10 ng/ml IL6, 5 ng/ml IL11, 25 ng/ml IGF1 and 50 ng/ml SCF (PeproTech). EBs were cultured in Ultra-Low Attachment plates in 5% CO₂/5% O₂/90% N₂ until dissociation on day 7.

2D Differentiation

iPSC-HECs for in vitro static experiments were derived using a 2D differentiation approach. Human iPSCs were differentiated in 12-well plates according to the vendor’s protocol using STEMDiff Hematopoietic Kit (Stem Cell Technologies). iPS colonies were seeded in mTeSR, and replaced with media A the next day, following confirmation of appropriate colony number and size. On day 2, half of the media was replaced with fresh media A. On day 3, cells were washed once with PBS and replaced with media B. Half media replacement was done again on day 5. Cells were grown in 5% O₂ and dissociated on day 7 of differentiation.

Hemogenic Endothelium Isolation

EBs were dissociated using 0.05% trypsin for 15 min followed by incubation for up to 1 h in collagenase IV at 37°C with repeated pipetting. Cells grown in 2D were incubated with Accutase for 5 min and dissociated mechanically using a P1000 pipette. Freshly dissociated cells were filtered through a 40 μm cell strainer and incubated with CD34⁺ magnetic beads (Miltenyi) for 30 min on ice. Cells were washed and transferred to LS columns for CD34⁺ cell isolation. For in vitro EHT, cells were counted and seeded onto Matrigel (BD) in EB media supplemented with 10 ng/ml BMP4, 5 ng/ml FGF2, 15 ng/ml VEGF, 10 ng/ml IL6, 5 ng/ml IL11, 25 ng/ml IGF1, 50 ng/ml SCF and 2 units/ml EPO.

Flow Cytometry

Analysis of iPSC-HECs was based on the following markers: CD34-PE Cy7, CD43-PE, CD45-PE Cy5, CD235a-FITC, and VEGFR2-APC. DAPI was included to omit dead cells. For characterization of HSPCs, CD34-PE Cy7, CD43-PE, CD45-PE Cy5 and DAPI were used. Cells were stained in PBS/2% FBS for 30 min at room temperature and analyzed on a BD LSRFortessa using BD FACSDiva software. Graphs were prepared in FlowJo.

Dorsal aorta-on-a-Chip culture

The dorsal aorta-on-a-chip devices were made from PDMS (Dow Corning; Sylgard 184 Silicone Elastomer Kit) similarly to previously described (Huh et al., 2010, 2013). The chips contained a central hollow chamber separated into upper and lower microchannels by a flexible, non-porous membrane; two full-height, hollow side chambers were oriented in parallel on either side of the central channel, through which cyclic suction could be applied using a vacuum pump to exert physiologically relevant cyclic strain on cells attached to the central membrane. The dimensions of the microfluidic channel were 1000 μm wide (W), 200 μm high (h) and 16 mm long. Prior to cell culture, chips were treated with oxygen plasma (Diener Atto; 2 min, 0.5 mbar, 30 W) and coated with Matrigel (1:20; BD) plus 50 μg/ml fibronectin (Sigma-Aldrich) diluted in PBS with calcium and magnesium for 1 hour at 37°C. 90,000 cells were seeded in 13 μl media per microchannel and incubated for two hours until cells had adhered, before more media was added. Chips were cultured overnight or for two days to ensure a complete monolayer before forces were applied. For experiments, the cytokine concentration was reduced to half and 2 μg/ml fibronectin was added to the media to promote cell adhesion. Cyclic membrane stretching was applied at 2 Hz and 10% strain (starting at 0.5 Hz and increasing over at least 12 h), to mimic the heart rate at this stage of development, which is around 120 beats per minute (Doubilet and Benson, 1995; von Steinburg et al., 2013). WSS was applied at 5 dyn/cm² as used previously for mouse cells, described in detail by Adamo et al., 2009. Notably, as the size of the human and mouse embryo is approximately the same at an equivalent time of hematovascular development (Mu et al., 2008; Papaioannou et al., 2010), we assumed comparable aortic diameters between human and mouse at this developmental stage. Media was perfused using a multichannel peristaltic pump (Ismatec) at 294 μl/min (gradually increasing from 29 μl/min over at least 12 h), corresponding to 5 dyn/cm², based on the following equation:

$$\tau =\frac{6\eta \mathcal{Q}}{h^{2}W}$$

where $\mathcal{Q}$ equals flow rate and fluid dynamic viscosity, η, for cell culture media at 37°C was set to 6.8×10⁻³ dyn⋅s/cm².

Colony Forming Assay

Serum-free methylcellulose-based medium with recombinant cytokines, MethoCult H4636, or H4434 (Stem Cell Technologies) supplemented with 50 ng/ml IL6, 50 ng/ml TPO, and 50 ng/ml Flt3L (all PeproTech) were used for colony forming assays. Cells from one microchannel per condition were used to assess CFU potential in dorsal aorta-on-a-chip studies, seeded into three 35 mm dishes, from two biological experiments. Alternatively, 10,000 iPSC-derived cells or 1000 CD34⁺ cells were seeded per 35 mm dish in two replicates. Colonies were scored 14 days later.

Endothelial Cell Tube Formation Assay

Human umbilical vein endothelial cells (HUVECs) were cultured in LSGS-supplemented Medium 200 (Gibco) and the Endothelial Cell Tube Formation Assay (In vitro Angiogenesis) was performed according to the vendor’s protocol. Specifically, 24 well plates were coated with 100 μl Matrigel per well (BD) for 30 min at 37°C, and 70,000 HUVECs or 60,000 iPSC-HECs were seeded in 400 μl media per well. Cells were washed with PBS and fixed in 10% formalin for 15 min at room temperature 20 hours after HUVEC seeding, and 6 hours after HE seeding. Cells were stained for 1 h at room temperature with 1:100 Alexa Fluor 488 phalloidin (Invitrogen) in PBS/0.1% Triton X-100/1% BSA prior to imaging at 4x magnification.

siRNA/Rho Stimulation of Cultured Cells

iPSC-HECs were seeded at 30,000 cells/cm² and treated 1 or 2 days thereafter, when cells had reached around 50% confluency. The following Silencer Select siRNAs were used for siRNA-mediated knockdown: hLATS1 s17393, hLATS2 s25503, and Silencer® Select Negative Control No 1 (Ambion), according to the vendor’s protocol. For each 6-well plate, 20 pmol of siRNA was mixed with 4 μl of Lipofectamine RNAiMax Reagent (Invitrogen) in OPTI-MEM Reduced Serum Medium (Gibco) and incubated at room temperature for 15 min before being added drop-wise to the well. Cells were collected and analyzed 24, 48 and 72 h post transfection. For Rho-GTPase modulation, the following compounds were used: 250 ng/ml CN03, 125 ng/ml CN04, or 200 ng/ml CT04 (Cytoskeleton, Inc).

Zebrafish Husbandry

Zebrafish (detailed line information in key resources table) were maintained in accordance with the Beth Israel Deaconess Medical Center and Institutional Animal Care and Use Committee protocols. Tg(hsp70:mCherry-2A-Flag-YapS87A) (Chi et al., 2008) expression was induced by incubation at 38°C for 30 min. yap1^mw48 mutants were identified by PCR amplification of a 305 bp band and Tfi restriction digest (primer sequences in key resources table). Mutant amplicons digest to 161 bp and 140 bp bands. WT amplicons do not cut.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
CD34-PE Cy7	BD Biosciences	Cat. 348791
CD43-PE	BD Biosciences	Cat. 560199
CD45-PE Cy5	Beckman Coulter	Cat. IM2652U
CD235a-FITC	BioLegend	Cat. 349104
VEGFR2-APC	BD Biosciences	Cat. 560495
DAPI solution	BD Biosciences	Cat. 564907
Monoclonal YAP1	Santa Cruz Biotechnology	Cat. sc-101199
Monoclonal GAPDH	Santa Cruz Biotechnology	Cat. sc-47724
Alexa Fluor 488 Phalloidin	Invitrogen	Cat. A12379
Alexa Fluor 647-AffiniPure donkey anti-mouse IgG	Jackson ImmunoResearch Laboratories	Cat. 715-605-150
Biological Samples
Cord Blood CD34+ Cells	AllCells	Cat. CB008F
mPB CD34+ Cells	AllCells	Cat. mPB015F
Mouse Embryonic Fibroblasts	GlobalStem	Cat. 6001G
Primary Human Umbilical Vein Endothelial Cells	ThermoFisher	Cat. C0155C
Chemicals, Peptides, and Recombinant Proteins
CN02: Rac/Cdc42 Activator II	Cytoskeleton, Inc	Cat. CN02
CN03: Rho Activator II	Cytoskeleton, Inc	Cat. CN03
CN04: Rho/Rac/Cdc42 Activator I	Cytoskeleton, Inc	Cat. CN04
CT04: Rho Inhibitor I	Cytoskeleton, Inc	Cat. CT04
CHIR99021	StemCell Technologies	Cat. 72052
SB431542	StemCell Technologies	Cat. 72232
Recombinant Human BMP4	PeproTech	Cat. 10779-138
Recombinant Human bFGF	Gibco	Cat. PHG0266
Recombinant Human VEGF	R&D Systems	Cat. 293-VE
Recombinant Human IL6	PeproTech	Cat. 200-06
Recombinant Human IL11	PeproTech	Cat. 200-11
Recombinant Human IGF1	PeproTech	Cat. 100-11
Recombinant Human SCF	PeproTech	Cat. 300-07
Recombinant Human EPO	Epogen	Cat. 55513-144-01
Recombinant Human TPO	PeproTech	Cat. 300-18
Recombinant Human Flt3L	PeproTech	Cat. 300-19
Critical Commercial Assays
Click-iT™ EdU Cell Proliferation Kit for Imaging, Alexa Fluor™ 647 dye	Invitrogen	Cat. C10340
TSA Plus Cyanine 3/Fluorescein Kit	Perkin Elmer	Cat. NEL753001KT
Experimental Model: Cell Lines
CD45-iPSC	BCH iPS Core	Vo et al., 2018
MSC-iPSC1	BCH iPS Core	Park et al., 2008
1157-iPSC	BCH iPS Core	Sugimura et al., 2017
Dub7-iPSC	Kaufman Laboratory, UCSD	Ferrell et al., 2015
Experimental Model: Zebrafish
Ctgfa:GFP: Tg2(ctgfa:EGFP)pd96		Mokalled et al., 2016
Hs:YapS87A: Tg(hsp70:mCherry-2A-Flag-YapS87A)		Cox et al., 2018
Yap−/−: yap1mw48/mw48		Miesfeld et al., 2015
Flk1:GFP: Tg(kdrl:gfp)la116		Choi et al., 2007
Kdrl:mCherry: Tg(kdrl:Hsa.HRAS-mCherry)s896		Chi et al., 2008
Kdrl:mCherry: Tg(kdrl:Hsa.HRAS-mCherry)s916		Hogan et al., 2009
Cd41:GFP: Tg(−6.0itga2b:egfp)la2		Bertrand et al., 2008
Oligonucleotides
Standard control morpholino	CCTCTTACCTCAGTTACAATTTATA	GeneTools
lats1 morpholino	CCTCGGGTTTCTCGGCCCTCCTCAT	Chen et al., 2009
lats2 morpholino	CATGAGTGAACTTGGCCTGTTTTCT	Chen et al., 2009
sih morpholino	CATGTTTGCTCTGATCTGACACGCA	Sehnert et al., 2002
hLATS1 siRNA	ThermoFisher	Cat. s17393
hLATS2 siRNA	ThermoFisher	Cat. s25503
Negative Control siRNA	ThermoFisher	No 1
runx1_fw	CGTCTTCACAAACCCTCCTCAA	Carroll et al., 2014
runx1_rv	GCTTTACTGCTTCATCCGGCT	Carroll et al., 2014
cmyb_fw	TGATGCTTCCCAACACAGAG	Carroll et al., 2014
cmyb_rv	TTCAGAGGGAATCGTCTGCT	Carroll et al., 2014
kdr1_fw	CGAACGTGAAGTGACATACGG	Carroll et al., 2014
kdr1_rv	CCCTCTACCAAACCATGTGAAA	Carroll et al., 2014
ctgfa_qPCR_fw	CTACGGCTCCCCAAGTAACC	Cox et al., 2016
ctgfa_qPCR_rv	TCCACTGCGGTACACCATTC	Cox et al., 2016
ctgfa_ISH_fw	TGTGATTGCTCTGCTGTTCC	Mokalled et al., 2016
ctgfa_ISH_rv	GGTGAGGCGATTAGCTTCTG	Mokalled et al., 2016
18s_fw	TCGCTAGTTGGCATCGTTTAT	Esain et al., 2015
18s_rv	CGGAGGTTCGAAGACGATCA	Esain et al., 2015
yap gen FWD	AGTCATGGATCCGAACCAGCACAA	Cox et al., 2018
yap gen REV	GCAGGCTGAAAGTGTGCATTGCC	Cox et al., 2018
ANKRD1	ThermoFisher	Hs00923599_m1
CTGF	ThermoFisher	Hs00170014_m1
LATS1	ThermoFisher	Hs01125524_m1
LATS2	ThermoFisher	Hs01059009_m1
CD43	ThermoFisher	Hs01058311_m1
CD45	ThermoFisher	Hs04189704_m1
RUNX1	ThermoFisher	Hs00231079_m1
18s	ThermoFisher	Hs99999901_s1
Recombinant DNA
Software and Algorithms

Zebrafish chemical exposures and whole-mount in situ hybridization (WISH)

Zebrafish embryos were exposed to 100 ng/ml CN02, 1 μg/ml CN03, 1 μg/ml CN04, 1 μg/ml CT04 (Cytoskeleton, Inc) or 100 μM DAPT in 6 well plates in E3 fish medium. For live-imaging experiments, 0.003% phenylthiourea was added to embryo medium to prevent pigment formation. To visualize cell apoptosis, embryos were dechorionated and incubated in Acridine Orange (10 μg/mL in embryo medium) for 30 minutes, followed by 3 × 10 minute washes to remove staining and immediately imaged. Embryos were analyzed by WISH as previously described (Theodore et al., 2017) using published probes to runx1, cmyb, rag1, flk1, ephrinB2, flt1, cdh5, fli1a, dab2, mpo, ikaros, l-plastin and gata2b. The ctgfa in situ probe was generated by amplifying a 1.4 kb fragment of ctgfa coding sequence from cDNA (using primers published in Mokalled et al., 2016) and cloning into the pCRII TOPO vector for linearization and antisense probe generation by in vitro transcription. Fluorescence in situ hybridization was performed as described (Brend and Holley, JOVE. 2009). Cell proliferation was assayed using the Molecular Probes Click-iT EdU Cell Proliferation Kit for Imaging (Catalog #: C10340). In brief, embryos were incubated on ice with 500 μM EDU/10% DMSO in embryo medium for 1 h at 34 hpf, then allowed to recover at 28.5°C for 1 h before fixation at 36 hpf in 2% PFA. EDU detection and imaging was performed according to kit instructions. Images were taken using Axiovision LE software on a Zeiss Axio Imager A1 or Zeiss Discovery V8 microscope. Images used in a given lettered panel were taken during the same imaging session using identical microscope settings and magnification, with all embryos coming from the same clutch and/or from the same WISH experiment. Adjustments to brightness or contrast were applied equally to all images in a panel. Confocal imaging was done on a Zeiss LSM880 microscope. WISH staining was analyzed qualitatively (≥20 embryos/condition unless otherwise specified, from at least 2–3 independent experiments) for high/medium/low expression compared to the median of matched sibling controls, and displayed graphically as the percentage of embryos within each phenotype bin for each condition. These results were statistically analyzed by Chi-square contingency tests or two-tailed Fisher’s exact test for small sample sizes.

Zebrafish Morpholino Injections

Morpholinos (Gene Tools; sequences indicated in the key resources table) were injected at the 1-cell stage, as previously described (Cortes et al., 2016). Effects were scored similarly to WISH samples, and compared to matched sibling controls. Injection amounts were 4 ng for silent heart MO, 2.5 ng each for lats1/2 MO (5 ng total) and a corresponding amount of standard control MO.

Method Details

Zebrafish qPCR Analysis

RNA (30 pooled embryos/condition) was isolated using the RNAqueous Total RNA isolation kit, purified using the Turbo DNA-free Kit, and 1μg of RNA was used to generate cDNA using the Reverse Transcriptase Supermix (all kits Life Technologies). qRT-PCR was performed on a Bio-Rad CFX384. Samples were run in duplicate or triplicate with ≥3 biological replicate pools/condition using the primers listed in the key resources table. Expression was normalized to the zebrafish 18s housekeeping gene.

Zebrafish Cell Proliferation and Apoptosis Analysis

3D confocal micrographs were rendered and quantitatively analyzed using Imaris software. To measure cell proliferation, EDU+ cells that were labeled with flk1:GFP were counted across 8 somites anterior to the anal pore if they were in the dorsal aorta or in budding/recently budded cells between the dorsal aorta and posterior cardinal vein. Similarly, Acridine Orange-positive cells were counted if they were located in the dorsal aorta or VDA region across 8 somites anterior to the anal pore over the yolk extension.

Intensity analysis of in situ hybridization staining

Quantifications of the intensity of staining for runx1/cmyb expression was performed as described in (Dobrzycki et al., 2018). In brief, regions around the VDA were manually drawn in ImageJ to obtain an average pixel intensity value. This same area was used to obtain a background intensity from trunk area immediately dorsal to the VDA as means of normalizing the signal in each image. The difference in signal from VDA-background was reported for each embryo and used to calculate the mean per condition.

Human cell qPCR Analysis

Total RNA was extracted using RNeasy PLUS Micro or Mini Kits (Qiagen). cDNA was prepared using Maxima First Strand cDNA Synthesis Kit (Thermo Fisher). Gene expression was analyzed using TaqMan Assays (see key resources table) with TaqMan™ Gene Expression Master Mix (Thermo Fisher) and run on Applied Biosystems QuantStudio Flex Real-Time PCR System. Samples were run in triplicates and normalized to the human 18s housekeeping gene.

Western Blot

iPSC-HECs were harvested and lysed in RIPA buffer. 30 μg of protein from the lysate samples was subjected to 4–12% Bis-Tris gel electrophoresis. The gel was transferred to a polyvinylidene difluoride membrane and was cut into two parts based on the ladder indicated size: Upper and lower part membranes were incubated overnight at 4°C with primary antibodies diluted in odyssey blocking buffer for YAP1 (1:500, sc-101199) or GAPDH (1:1000, sc-47724), respectively. The YAP protein bands were detected by secondary immunoblotting with anti-ECL HRP (1:50,000; NAV931) and SuperSignal West Femto Maximum Sensitivity Substrate. The GAPDH protein bands were blotted with anti-IRDye 800CW (1:2000) and LI-COR Odyssey scanner.

Immunocytochemistry

Cells were washed with PBS and fixed in 10% formalin for 15 min at room temperature. Cells were stained overnight at 4°C with YAP1 antibody (1:100; sc-101199) in PBS/0.1% Triton X-100/1% BSA. The next day, cells were washed with PBS/0.1% Triton X-100 and incubated with 1:100 Alexa Fluor 488 phalloidin (Invitrogen), DAPI and Alexa Fluor 647-AffiniPure donkey anti-mouse IgG (1:450, Jackson ImmunoResearch Laboratories; 715–605-150) for 1 h at room temperature. Cells were imaged using EVOS fluorescent imager and intensity ratios were calculated using ImageJ.

Quantification and Statistical Analysis

If not stated otherwise, we reported mean ± SEM values, and performed Student’s t-test (for two groups) or ANOVA (more than two groups) using GraphPad Prism 6 for statistical analysis. Pairing, repeated-measurements, or other test corrections were applied as needed.

Data and Code Availability

This study did not generate data or code requiring public database deposition.

HIGHLIGHTS.

• Biomechanical strain impacts human hematopoietic stem cell specification in vitro

• YAP drives Endothelial-to-Hematopoietic-Transition downstream of blood flow in vivo

• Rho-GTPase stimulation promotes HSC formation in vitro and in vivo via YAP activity