Glia maturation factor-γ is required for initiation and maintenance of hematopoietic stem and progenitor cells

gmfg is a potential regulator of HSPC development

Endothelial protein C receptor (EPCR), also known as CD201, has been identified as a novel HSC marker in mouse embryos and human adults [26,27,28,29]. Very recently we established a 3D induction system cocultured with stromal cells, capable of yielding a higher percentage of CD201 + HSC-like cells (Lin-Sca-1 + c-kit + CD201 + , LSKCD201 +) with robust hematopoietic reconstitution potential from mouse PSCs [30], which we defined as PSC-derived HSPCs hereafter. By taking advantage of this model, we previously performed RNA sequencing on mouse undifferentiated PSCs, PSC-derived HSPCs, FL-derived HSPCs (FL-LSKCD201 +), and BM-derived HSPCs (BM-LSKCD201 +) [30]. To dig into the biological generalities and screen potential regulators implicated in HSPC development, we re-analyzed the RNA-seq data and supposed that genes upregulated in HSPCs must play a critical role. According to a rigorous cut-off criterion (Foldchange > 2 and p.adjust < 0.05), 697, 722 and 1116 upregulated DEGs in PSC-, FL- and BM-derived HSPCs as compared to undifferentiated PSCs were identified, respectively. Common upregulated DEGs were defined as genes that were significantly upregulated in all the three populations and thereby 172 genes were discovered (Fig. 1a). Further application of biological process enrichment analysis revealed that these common upregulated DEGs were strongly linked to actin cytoskeleton organization, with 20 related genes involved (Fig. 1b). Afterwards, by searching the public database of BIOGPS (http://biogps.org/#goto=welcome), we found that gmfg was most prominently expressed in the human blood system, and was much more highly expressed in CD34 + HSCs than in their mature counterparts (NK cells, B cells and T cells) (Fig. 1c), indicating a more important role of gmfg in HSCs than in the precursor/mature stages. To confirm the data from RNA-Seq and public databases and determine the exact time window when gmfg began to work during hematopoietic development, we performed real-time PCR analysis as hESCs progressed through hematopoietic differentiation. Results showed that gmfg expression increased sharply at day 6 (D6) of differentiation, and maintained high-level expression as maturation towards hematopoiesis. The expression of CD34, CD43 and CD45, three well-known HSPC markers representing distinct phases of differentiation, was detected to validate that the genetic program for hematopoiesis in our system is activated at D6 (Fig. 1d). Furthermore, stage-specific cell populations at various time points were purified to document the cell-type specificity of gmfg. Consistently, gmfg exhibited a greater than 600-fold upregulation in D6-CD31 + CD34 + hemogenic endothelium progenitors (HEPs) as compared to D3-CD309 + mesoderm cells, and then it was slightly downregulated in D9-CD43 + and D12-CD45 + hematopoietic progenitor cells (HPCs) (Fig. 1e). Therefore, the stage of HEP specification from mesoderm cells was proposed as the appropriate time window that gmfg began to play a role. For illustration of the idea, we evaluated the impact of gmfg deletion on the generation of HEPs and HPCs from hESCs by utilizing a short hairpin RNA oligonucleotides (shRNA)-mediated gmfg knockdown approach. The sequence of control and gmfg shRNAs was listed in Additional file 1: Table S6. gmfg-sh2 and gmfg-sh3 were chosen as the main tools for the next experiments since the mRNA level of gmfg was remarkably decreased in hESCs treated with these two shRNAs (Fig. 1f). As expected, flow cytometry analysis showed that gmfg deletion significantly reduced the fraction of D6-CD31 + CD34 + HEPs and subsequent D9-CD43 + HPCs derived from hESCs (Fig. 1g), confirming the impairment of hESC hematopoietic differentiation.

Fig. 1figure 1

gmfg is a potential regulator of HSPC development. a Venn diagram showing the upregulated DEGs in the PSC-, FL- and BM-derived HSPCs as compared to undifferentiated PSCs (172 common upregulated DEGs). b Top 20 statistically enriched pathways of the common upregulated DEGs in the PSC-, FL- and BM-derived HSPCs. c gmfg expression in human tissues. The abscissa represents different human tissues, the ordinate represents gene expression level, and the histogram represents the expression level of gmfg in specific human tissue. Black arrow denotes gmfg expression in CD34 + cells. d Dynamic analysis of gmfg, CD34, CD43 and CD45 expression with real-time PCR during hESC hematopoietic differentiation. Relative expression is normalized by undifferentiated hESCs (except for CD45 by differentiated hESCs at D9). Data are shown as mean ± SD (n = 3). ns, not significant; *p < 0.05, ***p < 0.001, ****p < 0.0001. e Quantitative RT-PCR analysis of gmfg expression in undifferentiated hESCs, mesoderm (CD309 +), HEPs (CD31 + CD34 +), and HPCs (CD43 + and CD45 +) derived from hESCs. Relative expression was normalized by undifferentiated hESCs. Data are shown as mean ± SD (n = 3). ***p < 0.001, ****p < 0.0001. f Quantitative RT-PCR analysis of gmfg expression in hESCs treated with ctl-sh, gmfg-sh2, and gmfg-sh3. Data are shown as mean ± SD (n = 3). ***p < 0.001, ****p < 0.0001. g Flow cytometry analysis of CD31 + CD34 + HEPs and CD43 + HPCs generated from ctl-sh, gmfg-sh2, and gmfg-sh3 hESCs at day6 and day9 of differentiation, respectively. Data are shown as mean ± SD (n = 3). **p < 0.01, ****p < 0.0001. h Expression pattern of gmfg during zebrafish embryogenesis: The stage examined by whole-mount in situ hybridization (WISH) is shown in each panel: 1‐cell stage, 2‐cell stage, 4 h postfertilization (hpf), 13 hpf (Scale bars: 200 μm), 24 hpf, 36 hpf, 48 hpf and 72 hpf (Scale bars: 100 μm). i (left) Graphic representation of the sorting strategy of kdrl + cmyb- ECs and kdrl + cmyb + HE from kdrl:mCherry;cmyb:GFP transgenic zebrafish by FACS at 48 hpf. (right) Quantitative RT-PCR analysis of gmfg and gmfb expression in FACS-sorted kdrl + cmyb- ECs and kdrl + cmyb + HE. Bars represent mean ± SD (n = 3). ns, not significant; **p < 0.01

Since hematopoietic differentiation of PSCs in vitro well recapitulates embryonic hematopoiesis in vivo [31], above evidence prompted us to further probe the role of gmfg in embryonic hematopoiesis. We therefore analyzed the dynamic expression of gmfg in vivo by WISH during zebrafish embryogenesis. Though ubiquitously expressed from one-cell stage to 13 h postfertilization (13 hpf), gmfg mRNA exhibited a marked enrichment in the AGM from 24 to 48 hpf, and then it was found in the CHT at 72 hpf (Fig. 1h). This specific spatio-temporal expression pattern was reminiscent of the developmental trajectory of embryonic HSPCs. Consistently, kdrl + cmyb- endothelial cells (ECs) and kdrl + cmyb + HE were sorted from 48 hpf kdrl:mCherry;cmyb:GFP embryos and qPCR was performed for gmfg and gmfb, another gmf isoform. Compared to ECs, gmfg expression was much higher in the developing HE, whereas gmfb expression was comparable in both groups of cells (Fig. 1i). Altogether, these data strongly suggested that gmfg is a potential regulator of HSPC development.

gmfg, instead of gmfb, is required for HSPC development

To investigate whether gmfg has a function in HSPC development, loss-of-function experiments were performed by utilizing two different types of antisense morpholino oligonucleotides (MOs) against gmfg, namely, translation blocking MO (gmfg-atgMO) and splice modifying MO (gmfg-spMO). The efficiency of these two MOs was validated by western blot analysis, in which both MOs could significantly reduce the protein level of gmfg at 36 hpf and 4 dpf, respectively (Additional file 1: Fig. S1a), and gmfg-atgMO was chosen as the main tool for subsequent experiments. In particular, of the reference genes described in various publications, β-actin and GAPDH were the most widely used ones. However, previous studies have reported that GAPDH expression varies during zebrafish embryonic development and can't be detected at early stages (no detectable until the Prim 5 stage), rendering it unsuitable as an internal control for zebrafish developmental time course studies [32]. In our study, Actin (β-actin) and GAPDH were simultaneously subjected to western blot analysis and results showed that no significant changes in the protein level of Actin compared with Gapdh was observed between control and gmfg morphants at 48 hpf (Additional file 1: Fig. S3), demonstrating that gmfg deficiency did not grossly affect the expression stability of Actin. Therefore, though gmfg is a regulator in actin dynamics, β-actin is still considered to be superior to GAPDH for normalization in this study.

In zebrafish embryos, HSPCs can be visualized along the axial vessels by the expression of runx1 and cmyb, two conserved HSPC markers. runx1 + and cmyb + HSPCs in the DA were dramatically decreased or even absent in the embryos injected with gmfg-atgMO at 36 hpf (Fig. 2a). This result was supported by western blot analysis of the protein levels of Runx1 and Cmyb, which were also significantly reduced in embryos injected with gmfg-atgMO, and gmfg-spMO phenocopied the gmfg-atgMO results (Fig. 2b). These reductions could be due to a defect in the earliest steps of HSPC initiation, therefore, we analyzed the nascent HSPC marker runx1 expression at earlier time points. A significant reduction in the number of runx1 + HSPCs was observed in gmfg-deficient embryos in or near the floor of DA at 24 and 28 hpf (Fig. 2c), indicating that gmfg plays an important role in HSPC initiation. To determine whether the HSPC loss observed in gmfg morphants resulted from growth retardation, we carried out time course analysis of cmyb expression at later phases. WISH and fluorescence imaging results showed that cmyb expression was always decreased in the DA and CHT in gmfg morphants at 48, 52 and 75–77 hpf, respectively (Additional file 1: Fig. S4a, b). To further confirm that HSPC loss was indeed specific to gmfg, we overexpressed gmfg by injecting the full-length zebrafish gmfg mRNA (escaping from gmfg-atgMO targeting) into embryos at one-cell stage to perform rescue experiments. The specificity of gmfg mRNA was validated by gel electrophoresis and western blot analysis (Additional file 1: Fig. S2a, b). WISH results showed that coinjection of gmfg mRNA with gmfg-atgMO was sufficient to normalize the decreased runx1/cmyb transcript levels in embryos injected with gmfg-atgMO alone at 28 hpf (Fig. 2d).

Fig. 2figure 2

Loss of gmfg, but not gmfb, induces HSPC defects. a WISH results of runx1 and cmyb in the AGM in control and gmfg morphants at 36 hpf. The red arrowheads indicate the expression of runx1 and cmyb. b Western blotting showing the protein level of Runx1 and Cmyb in control and gmfg morphants at 36 hpf. Representative blot is shown in the figure (Full-length blots are presented in Additional file 2: Fig. S1 and S2). Data represent mean ± SEM intensity of indicated blots (n ≥ 3). *p < 0.05, ***p < 0.001, ****p < 0.0001. c WISH analysis showing the expression of runx1 (red arrowheads) in the DA in control and gmfg morphants at 24 and 28 hpf. d WISH analysis showing runx1/cmyb expression (red arrowheads) in the DA in control, gmfg-atgMO, and coinjection of gmfg-atgMO and gmfg mRNA embryos at 28 hpf. e WISH analysis showing the expression of cmyb (red arrowheads) in the DA in embryos injected with control MO and gmfb-MO at two different doses (gmfb-MO1: 9.6 ng and gmfb-MO2: 12.8 ng per embryo) at 30 hpf. f Representative images showing runx1 expression (red arrowheads) in the DA in control, gmfg-atgMO and gmfb-MO2 (12.8 ng per embryo) at 24 hpf. g Maximum projections of 48 hpf kdrl:mCherry; cmyb:GFP double-transgenic embryos injected with control MO, gmfg-atgMO, and gmfg-spMO. Arrowheads denote kdrl + cmyb + HE along the DA. All views: anterior to left. h Enumeration of kdrl + cmyb + HE shown in (g). Bars represent mean ± SD of control MO (n = 10), gmfg-atgMO (n = 21), and gmfg-spMO (n = 8). ***p < 0.001. Numbers at the lower right corner of the picture represent embryos with displayed phenotype/whole embryos. All scale bars, 100 µm

To determine whether gmfb also plays a role in HSPC development, we injected gmfb-MO into embryos at different doses and performed WISH analysis of runx1 and cmyb. gmfb-MO at either dose, including a moderate amount of 9.6 ng (gmfb-MO1) and a maximum amount of 12.8 ng (gmfb-MO2) which caused slight malformation of embryos, showed no effect on the expression of cmyb at 30 hpf (Fig. 2e). Moreover, we injected gmfb-MO2 into zebrafish embryos again, and 1 ng gmfg-atgMO was used as a positive control, to detect runx1 expression in the DA at 24 hpf. runx1 expression remained unchanged in embryos injected with gmfb-MO2 but significantly decreased in embryos injected with 1 ng gmfg-atgMO (Fig. 2f). These results indicated that gmfg, but not gmfb, is required for HSPC development. We therefore focused on gmfg and gmfb was not pursued further in this study. In addition, we directly visualized the emerging HE undergoing EHT in the aortic floor using living kdrl:mCherry;cmyb:GFP double transgenic embryos by confocal microscopy. The number of kdrl + cmyb + HE in the floor of the DA was reduced by approximately 50%-70% in gmfg morphants in comparison to their control siblings (Fig. 2g, h), suggesting that the EHT process was severely impeded.

HSPCs produced by EHT in zebrafish subsequently migrate to and expand in the CHT which mirrors the functions of fetal liver in mammals [33]. We then utilized cd41:eGFP transgenic animals to track the development of HSPCs in the CHT. Confocal microscopy results revealed that the number of cd41 + cells (cd41low HSPCs and cd41high thrombocytes) was markedly reduced in gmfg-deficient larvae at 3 dpf (Fig. 3a, b), which was further verified by flow cytometry quantitation of cd41 + cells from the dissected trunk and tail (Fig. 3c). Furthermore, since T cells exclusively originate from definitive HSPCs, we examined later larval stages by detecting the expression of rag1, a T lymphocyte-specific marker. The transcript level of rag1 was nearly or even completely absent in the gmfg-deficient larvae at 4 dpf (Fig. 3d), whereas the thymic anlage developed normally as assessed by the expression of a thymic epithelial cell marker foxn1 (Fig. 3e). Moreover, lcr:eGFP and lyz:dsRed transgenic animals were utilized to track the developing erythrocytes and neutrophils. Fluorescence imaging and flow cytometry results showed that gmfg deficiency led to a significantly decrease in the number and percentage of erythrocytes/lcr:eGFP cells (Fig. 3f, g) and neutrophils/lyz:dsRed cells (Fig. 3h, i) at 3dpf, respectively, suggesting that to a certain extent, the differentiation potential of HSPCs into erythroid and myeloid lineages was compromised in gmfg morphants, although primitive hematopoiesis- and erythromyeloid progenitor (EMPs)-derived erythrocytes and neutrophils cannot be entirely excluded.

Fig. 3figure 3

Loss of gmfg impairs the differentiation potential of HSPCs. a Maximum projections of 3 dpf cd41:eGFP transgenic embryos injected with control MO, gmfg-atgMO, and gmfg-spMO. White arrowheads denote cd41 + cells in the CHT. All views: anterior to left. Scale bar, 100 μm. b Enumeration of cd41 + cells shown in (a). Bars represent mean ± SD of control (n = 18), gmfg-atgMO (n = 17), and gmfg-spMO (n = 18). ***p < 0.001, ****p < 0.0001. c FACS analysis showing the percentage of cd41 + cells in the dissected trunk and tail of control, gmfg-atgMO, and gmfg-spMO at 3 dpf (n = 3). ****p < 0.0001. d and e WISH for the T lymphocyte marker rag1 (left, red dotted line circle) and thymic epithelial marker foxn1 (right, white dotted line circle), respectively, in embryos injected with control MO and gmfg-atgMO at 4 dpf. All views are lateral, with anteriors to left. Scale bars, 50 µm. f and h lcr:eGFP and lyz:dsRed transgenic embryos injected with control MO and gmfg-atgMO were visualized at 3 dpf. Scale bars, 500 μm. g and i FACS analysis showing the percentage of lcr + erythrocytes and lyz + neutrophils in whole embryos of control and gmfg-atgMO groups at 3 dpf (n = 3). **p < 0.01, ***p < 0.001. Numbers at the lower right corner of the picture represent embryos with displayed phenotype/whole embryos

A widely accepted concept in embryonic hematopoiesis is that at least two waves of hematopoiesis occurred, namely primitive and definitive wave, which take place in anatomically distinct locations at different developmental time and can be further distinguished on the basis of cell types produced [34]. We next investigated whether gmfg is required for the initial wave of hematopoiesis commonly termed ‘primitive hematopoiesis’ due to the absence of upstream multipotent progenitors. In zebrafish, primitive hematopoiesis produces myeloid cells and erythrocytes that maintain early immunity and oxygenation [35]. The number of primitive erythrocytes in the intermediate cell mass (ICM) at 24 hpf was significantly decreased in the gmfg-deficient animals, as assayed by utilizing transgenic lcr:eGFP embryos (Additional file 1: Fig. S5a), and the expression of panleukocyte marker l-plastin and neutrophil marker mpx in the anterior lateral mesoderm (ALM) and the posterior lateral mesoderm (PLM) at 28 hpf was also markedly reduced (Additional file 1: Fig. S5b, c), suggesting that gmfg also is indispensable for primitive hematopoiesis. However, as primitive hematopoiesis is not the focus of our study, a more detailed role and mechanism of gmfg in primitive hematopoiesis is not explored further.

gmfg does not appear to modulate PLM formation, DA specification, and EC proliferation or apoptosis

Posterior lateral mesoderm (PLM) produces both endothelial and hematopoietic lineages [36]. To rule out the possibility that the observed HSPC defects upon gmfg deficiency were a consequence of impaired earlier PLM formation, we analyzed the expression of PLM marker fli1a at 12 hpf. Results showed that fli1a expression in gmfg morphants remained normal when compared to that of control siblings (Fig. 4a), demonstrating that PLM formation was unaffected. Because HSPCs derive from DA and many mutants with impaired arterial specification also display defective hematopoiesis [37], we assessed whether gmfg was required for DA specification. Knockdown of gmfg did not yield any obvious arterial abnormalities at 24 hpf, as evident from the unchanged expression of two DA-specific markers ephrinB2a and dlc [38] in embryos injected with gmfg-atgMO at the doses used in this study (Fig. 4b). However, as embryos developed at the late stage of 28 hpf, the expression of ephrinB2a and dlc within the DA was significantly reduced (Fig. 4c), indicating gmfg is dispensable for DA specification but is required for later DA development.

Fig. 4figure 4

The effect of gmfg deficiency on PLM formation, DA specification, and EC proliferation or apoptosis. a Expression of the PLM marker fli1a in embryos injected with control MO, gmfg-atgMO, and gmfg-spMO analyzed by WISH at 12 hpf. b and c Expression of the DA-specific markers ephrinB2a and dlc in control and gmfg morphants analyzed by WISH at 24 (b) and 28 hpf (c). d TUNEL staining in fli1a:eGFP transgenic embryos injected with control MO, gmfg-atgMO, and gmfg-spMO at 26 and 32 hpf. White dashed lines indicate blood vessels and white arrowheads indicate the apoptosis of fli1a + cells. e Quantification of apoptotic fli1a + TUNEL + cells in (d). Bars represent mean ± SD of control (n = 18), gmfg-atgMO (n = 17), and gmfg-spMO (n = 18). ns, not significant. f PH3 staining in fli1a:eGFP transgenic embryos injected with control MO, gmfg-atgMO, and gmfg-spMO at 26 and 36 hpf. White dashed lines indicate blood vessels and white arrowheads indicate proliferative fli1a + cells. g Quantification of proliferative fli1a + pH3 + cells in (f), Bars represent mean ± SD of control (n = 18), gmfg-atgMO (n = 17), and gmfg-spMO (n = 18). ns, not significant. h Expression of runx1 and cmyb in the AGM in p53 mutant injected with control MO, gmfg-atgMO, and gmfg-spMO at 26 (left) and 36 hpf (right). The red arrowheads mark HSPCs in the AGM. Numbers at the lower right corner of the picture represent embryos with displayed phenotype/whole embryos. All scale bars, 100 µm

To probe whether HSPC loss observed in gmfg morphants could be attributed to abnormal apoptosis or proliferation of ECs, we performed TUNEL assay and immunostaining for proliferative marker PH3 using fli1a:eGFP transgenic animals[39]. Analysis of ECs by confocal microscopy showed that no significant difference in cell proliferation or apoptosis was present in the AGM between gmfg morphants and their control counterparts (Fig. 4d–g). Injection of antisense MOs often induces a nonspecific p53-dependent apoptosis [40], to address this issue, gmfg-atgMO and gmfg-spMO were injected into the p53 mutant at one-cell stage. The expression of runx1 and cmyb was still reduced in gmfg morphants (Fig. 4h), suggesting that the HSPC loss was not caused by p53 activation. Altogether, these observations support the notion that gmfg regulates HSPC development independently of its role in PLM formation, DA specification, and EC proliferation or apoptosis.

Blood flow-regulation of HSPC development is mediated, in part, by gmfg

Since we did not find evidence for the involvement of gmfg in the proliferation or survival of HSPCs in the AGM, we then turned our attention to the cellular mechanisms underlying HSPC reduction upon loss of gmfg. Considering the main function of GMFG in actin dynamics through its actin-severing/depolymerizing activities [13], we determined to examine the subcellular localization of GMFG and its colocalization with F-actin in HUVEC. Results showed that GMFG localized preferentially in the F-actin-rich structures at the forward periphery of cells as well as in cytoplasm and the colocalization of GMFG with F-actin was observed in these two regions (Additional file 1: Fig. S6), implying its putative role in actin cytoskeleton turnover by severing/depolymerizing filaments. On the other hand, elegant work from Lancino et al. demonstrated that blood flow regulated the morphodynamics of EHT-undergoing cells during HSC emergence [6], we therefore hypothesized that gmfg might mediate blood flow-dependent HSPC development through regulation of actin cytoskeleton reorganization. To validate this notion, we monitored gmfg expression variations in silent heart (sih/tnnt2a) embryos lacking a heartbeat and blood circulation [41]. Firstly, we constructed a silent heart model by using a previously verified tnnt2a-MO [42]. In comparison with their control siblings (Additional file 3), heart beating and blood flow were arrested and even invisible in tnnt2a morphants at 48 hpf as assessed by utilizing a lcr:eGFP transgenic line that can track the erythrocytes circulating in blood vessels (Additional file 4). HE was then isolated from control and tnnt2a morphants to perform qPCR for gmfg. As shown in Fig. 5a, the mRNA level of gmfg was significantly reduced in tnnt2a morphants and that this reduction was further confirmed by western blot analysis of Gmfg from dissected trunk and tail (Fig. 5b). Both results suggested that gmfg could well respond to hemodynamic alterations and justified gmfg as a downstream target of blood flow.

Fig. 5figure 5

gmfg mediates blood flow-dependent HSPC development. a Quantitative RT-PCR analysis of gmfg expression in FACS-sorted kdrl + cmyb + HE of control and tnnt2a-MO embryos at 48 hpf. Bars represent mean ± SD (n = 3). *p < 0.05. b Western blotting showing the protein level of Gmfg in the dissected trunk and tail of control and tnnt2a-MO embryos at 48 hpf. Representative blot is shown in the figure (Full-length blots are presented in Additional file 2: Fig. S3). Data represent mean ± SEM intensity of indicated blots (n = 3). *p < 0.05. c Representative images showing runx1/cmyb expression (red arrowheads) in the DA in control, tnnt2a-MO, and coinjection of tnnt2a-MO and gmfg mRNA embryos at 36 hpf. d Enumeration of runx1 + /cmyb + HSPCs shown in (c). Bars represent mean ± SD of control (n = 32), tnnt2a-MO (n = 43), and tnnt2a-MO + gmfg-mRNA (n = 23) embryos. ****p < 0.0001. e Representative images showing cmyb expression (red arrowheads) in the DA of control, tnnt2a-MO, and coinjection of tnnt2a-MO and gmfg mRNA embryos at 48 hpf. f Enumeration of cmyb + HSPCs shown in (e). Bars represent mean ± SD of control (n = 5), tnnt2a-MO (n = 10), and tnnt2a-MO + gmfg-mRNA (n = 10) embryos. ***p < 0.001, ****p < 0.0001. g Maximum projections of 80 hpf cd41:eGFP transgenic embryos injected with control MO, tnnt2a-MO, and tnnt2a-MO + gmfg-mRNA. Arrowheads denote cd41 + cells in the CHT. All views: anterior to left. h Enumeration of cd41 + cells shown in (g). Bars represent mean ± SD of control (n = 11), tnnt2a-MO (n = 12), and tnnt2a-MO + gmfg-mRNA (n = 11) embryos. **p < 0.01, ***p < 0.001. i FACS analysis showing the percentage of cd41 + cells in the dissected trunk and tail of control, tnnt2a-MO, and tnnt2a-MO + gmfg-mRNA embryos at 80 hpf (n = 3). ***p < 0.001, ****p < 0.0001. All scale bars, 100 µm

Next we sought to determine whether gmfg induction could restore HSPC loss in the absence of blood flow, we therefore injected tnnt2a-MO alone or in combination with gmfg mRNA into zebrafish embryos and performed WISH analysis. Compared to their control siblings, embryos injected with tnnt2a-MO had a marked reduction in the number of runx1/cmyb + HSPCs along the DA at 36 hpf and 48 hpf, as demonstrated previously [43]. However, this effect was mitigated by the enforced expression of gmfg (Fig. 5c–f). Correspondingly, parallel results were obtained by utilizing cd41:eGFP transgenic embryos in which gmfg overexpression significantly increased the number and percentage of cd41 + cells in tnnt2a-MO embryos at 80 hpf, as assessed by confocal microscopy (Fig. 5g, h) and flow cytometry (Fig. 5i). The role of gmfg in HSPC development and the requirement of blood flow for gmfg expression in HSPCs and hematopoietic tissues suggest that gmfg signaling is a part of the blood flow-mediated regulatory mechanism underlying HSPC development.

gmfg regulates Yap activity

When pulsatile blood flow goes through a vessel, it generates both shear stress and circumferential strain [44, 45]. Recent studies have suggested that these two different components of hemodynamic forces regulate HSPC development via distinct and separable mechanisms [10, 43]. Thereinto, kruppel-like transcription factor 2 (Klf2) was revealed to be an important mechanical mediator sensitive to fluid shear stress [43, 46] while YAP has been recently identified as a circumferential strain-induced regulator of HSPC formation in zebrafish [10]. To explore how gmfg relays signals from blood flow to HSPCs, we naturally tended to analyze the effects of gmfg deficiency on these two well-known flow-responsive factors. qPCR analysis of klf2a in FACS-sorted HE at 48 hpf demonstrated that the mRNA level of klf2a in gmfg morphants was comparable to that in controls (Fig. 6a). klf2a expression assayed by WISH, specifically in the axial vessels also did not differ significantly between control and gmfg morphants at 36 hpf and 48 hpf (Fig. 6b). Moreover, western blot analysis of the protein level of Klf2a from the dissected trunk and tail at 54 hpf further supported the validity of WISH and qPCR results (Fig. 6c). These data suggest, in 3 independent settings, that gmfg deficiency has little effect on klf2a expression, excluding the possibility of klf2a as a gmfg-target gene. Then, we aimed to characterize the impact of gmfg deficiency on Yap. As shown in Fig. 6d, loss of gmfg had little effect on the mRNA level of yap1 (zebrafish YAP gene) but significantly reduced the expression of two well-known Yap target genes, ctgfa and cyr61, in FACS-sorted HE at 48 hpf, suggestive of repressed Yap signaling. This observation was further supported by western blot analysis of dissected trunk and tail at 48 hpf, in which Yap expression was comparable between control and gmfg morphants but Ctgf expression was remarkably downregulated in gmfg morphants (Fig. 6e, upper and medium panel). Since phosphorylation of YAP on serine 127 (S127) generally underlies YAP inactivation and cytoplasmic sequestration [47], we further analyzed the phosphorylation of Yap (S127) in zebrafish. As anticipated, an increased level of p-Yap (S127) was observed in embryos injected with gmfg-atgMO (Fig. 6e, lower panel), suggesting that gmfg deficiency facilitated cytoplasmic sequestration of YAP.

Fig. 6figure 6

gmfg regulates Yap activity. a Quantitative RT-PCR analysis of klf2a expression in FACS-sorted HE of control and gmfg-atgMO embryos at 48 hpf. Bars represent mean ± SD (n = 3). ns, not significant. b WISH analysis of klf2a expression in the trunk and the cloaca (red arrowheads) of control and gmfg-atgMO embryos at 36 and 48 hpf, respectively. Numbers at the lower right corner of the picture represent embryos with displayed phenotype/whole embryos. Scale bars, 100 µm. c Western blotting showing the protein level of Klf2a in the dissected trunk and tail of control and gmfg-atgMO embryos at 54 hpf. Representative blot is shown in the figure (Full-length blots are presented in Additional file 2: Fig. S4). Data represent mean ± SEM intensity of indicated blots (n = 4). ns, not significant. d Quantitative RT-PCR analysis of yap1, ctgfa and cyr61 expression in FACS-sorted HE of control and gmfg-atgMO embryos at 48 hpf. Bars represent mean ± SD (n = 3). ns, not significant; **p < 0.05, ***p < 0.001. e Western blotting showing the protein level of Yap, Ctgf and p-Yap (S127) in the dissected trunk and tail of control and gmfg-atgMO embryos at 48 hpf. Representative blot is shown in the figure (Full-length blots are presented in Additional file 2: Figs. S5–S7). Data represent mean ± SEM intensity of indicated blots (n = 3). ns, not significant; *p < 0.05. f Western blotting showing the protein level of GMFG, total YAP, CTGF, nuclear and cytoplasmic YAP in HUVEC treated with ctl-sh, gmfg-sh2 and gmfg-sh3. Representative blot is shown in the figure (Full-length blots are presented in Additional file 2: Figs. S8–S12). Data represent mean ± SEM intensity of indicated blots (n ≥ 3). ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001, ****p < 0.0001

In order to further verify the relationship between gmfg and Yap, both of which are downstream effectors of blood flow, we extended our analysis in vitro by utilizing HUVEC cell line. The reason why we selected HUVEC lies in the generation of embryonic HSCs from HE, and to some extent, HE is considered to be a special subtype of endothelium. Besides, as endothelium lining the surface of blood vessels are naturally exposed to blood flow, their mechanotransduction process elicited by flow is relatively well elucidated [

留言 (0)

沒有登入
gif