Integrin-αvβ3 regulates thrombopoietin-mediated maintenance of hematopoietic stem cells

Hematopoietic stem cells (HSCs) are clonogenic cells capable of both self-renewal and multilineage differentiation, enabling life-long maintenance of blood cell generation. To maintain HSCs, at least one HSC daughter cell must always be capable of self-renewal and multilineage differentiation, but the mechanism by which HSCs retain that capability is not yet defined. It has been proposed that there is a specialized microenvironment, called “niche,” within the BM, where the balance among self-renewal, differentiation, and quiescence is regulated by both adhesion-dependent (eg, osteopontin [OPN]/integrin-α9β1 or α4β1, Tie-2/angiopoetin-1 interaction) and -independent machinery (eg, thrombopoietin [TPO], TGF-β).^1-5 

Recently, we found that the mouse CD34⁻KSL HSC fraction, but not the CD34⁺KSL hematopoietic progenitor fraction, dominantly expresses the integrin-β3 subunit (CD61) bound to the αv (CD51) or αIIb (CD41) subunit.⁶  In particular, the αvβ3 complex, but not the αIIbβ3 complex, appears to be involved in the function of HSCs, as αv positivity, but not αIIb positivity, enhances long-term repopulating (LTR) activity in CD34⁻KSL HSCs after transplantation.⁷  This led us to conclude that αvβ3 is crucial for HSC function.

Integrins are heterodimeric receptors consisting of an α- and β-subunit, and their active and inactive conformations (forms with higher and lower affinities for ligands) are tightly regulated by “inside-out” signaling, mediated through external stimulation of several receptors on the cell surface.⁸  After integrin activation, specific ligand binding to the protein initiates “outside-in” signaling, which coordinates with signaling cascades initiated through growth factor-, cytokine-, and G protein-coupled receptors to regulate actin reorganization, cell survival, and proliferation.^9-11  With regard to β3-integrin, inside-out signaling results in the binding of talin to the specific binding amino acids within the intracellular tail of the β3 subunit, which is essential for integrin activation.¹²  Of these sequences, tyrosine phosphorylation of the β3 subunit, at least on Tyr747 (pY747), appears to be required for the outside-in signaling cascade, such as that seen during stable thrombus formation in platelets.¹³  Although the roles of bidirectional integrin signaling by β3-integrin in other hematopoietic cells remain unclear, evidence suggests that an interaction between integrins α4β1 and α9β1 and OPN is required for regulation of HSC proliferation, which is indicative of the crucial contribution made by integrins to HSC maintenance.²  Integrin-αvβ3 reportedly interacts with OPN, as well as with vitronectin (VN), fibronectin, and CD31. Interestingly, CD31-null mice exhibit greater numbers of KSL cells but with less functionality than wild-type (WT) mice, which could reflect the absence of interaction with αvβ3-integrin.¹⁴  From these results, it seems apparent that, although integrin-αvβ3 appears to be involved in the regulation of HSC function, its precise role in that process remains unclear.

It is well known that TPO is essential for megakaryopoiesis¹⁵  and also contributes to the maintenance and expansion of HSCs.^16-18  Mice deficient in TPO or its receptor (c-mpl) show not only impaired megakaryopoiesis but also reduced HSC number and function.^16,19,20  Moreover, recent reports indicate that TPO is required for the maintenance of HSCs in a quiescent state within the BM.^3,4 

Here, we demonstrate that integrin-αvβ3 on HSCs plays essential roles in maintaining their stem cell activity. Furthermore, we show that specific ligation of β3-integrin contributes to the maintenance of LTR activity in HSCs through collaboration with TPO/c-mpl-mediated signaling, which inhibits the loss of LTR activity during ex vivo culture. Our approach to clarifying integrin function in HSCs entailed the use of knock-in mutant mice that display defective β3-integrin inside-out or outside-in signaling because of blockade of talin-binding to specific amino acids in the β3-integrin tail.²¹  This enabled us to show that outside-in signaling via pY747 of integrin-β3 (β3^PY747) after activation of αvβ3-integrin by TPO-mediated inside-out signaling is indispensable for TPO-mediated maintenance of HSC activity in vitro. In addition, outside-in signaling via β3^PY747 is also essential for maintenance of LTR activity in vivo within the BM niche.

C57BL/6-Ly5.2 and C57BL/6-Ly5.1 mice were from Sankyo Labo Service, and β3-integrin-deficient mice were from The Jackson Laboratory unless otherwise noted. β3-integrin Y747A and L746A knock-in mutant mice were described previously.²¹  Each strain was back-crossed for more than 5 generations and used at 8 to 10 weeks of age. All animal experiments were approved by the Institutional Review Board for Animal Care and Use at the University of Tokyo.

The following monoclonal antibodies were used for cell sorting and flow cytometric analysis: anti–c-Kit (2B8, BioLegend), anti-CD34 (RAM34, eBiosciences, San Diego, CA), anti-CD150 (TC15-12F12.2, BioLegend), and anti-CD48 (HM48-1, BioLegend). Anti–integrin-β3 (2C9.G2), anti-Sca-1 (E13-161.7), anti-CD45.2,(104) anti-CD45.1 (A20), anti-B220/CD45R (RA3-6B2), anti–Mac-1 (M1/70), anti–Gr-1 (RB6-8C5), anti–CD4 (RM4-5), and anti-CD8 (53-6.72) antibodies were from BD Biosciences (BD Biosciences Pharmingen) unless otherwise noted. For in vitro assays, low endotoxin and azide-free anti–integrin-β3 (2C9.G2, BioLegend) and its isotype control IgG (HTK888, BioLegend) were used.

Suspensions of BM cells were prepared from mice as described previously.^6,7 

We used an EPICS ALTRA (Beckman Coulter) or MoFlo XDP (Beckman Coulter) for cell sorting and flow cytometric analysis, as described previously.^6,7 

Long-term competitive repopulation (LTCR) assays were performed by transplantation of the indicated cells into lethally irradiated (9.5 Gy) C57BL/6-Ly5.2 or C57BL/6-Ly5.1 congenic mice, as described previously.⁶  Twelve or 20 weeks after transplantation, recipient mice with donor cell chimerism (> 1.0% for myeloid and B- and T-lymphoid lineages) were considered to be multilineage-reconstituted mice (positive mice). For serial transplantation, 10⁶ whole BM cells were obtained from primary transplanted mice and transplanted into other irradiated recipient mice.

After sorting 1000 to 1500 CD34⁻KSL cells, SOLiD sequencing was performed as described in the supplemental Methods (available on the Blood Web site; see the Supplemental Materials link at the top of the online article). The whole transcriptomes obtained through SOLiD sequencing were analyzed using GeneSpring (Agilent Technologies). After filtration based on a significant (P < .05 ANOVA) and more than 2-fold changes in expression, the selected genes were subjected to hierarchical cluster analysis. In addition, whole transcriptomes were subjected to gene set enrichment analysis (GSEA) using GSEA Version 2.06 software available from the Broad Institute (http://www.broad.mit.edu/gsea). Changed gene sets were selected based on a threshold set at a P value < .05 and false discovery rate (FDR; q-value < 0.25).

CD34⁻KSL cells were sorted and cultured for 5 days in S-Clone SF-03 medium (Sanko-Junyaku) supplemented with 0.5% BSA (Sigma-Aldrich) and 50 ng/mL mouse SCF and/or 50 ng/mL mouse TPO (all from R&D Systems). To induce integrin signaling via β3^PY747, 2C9.G2 (50 μg/mL; BioLegend) was added to the medium; hamster IgG (BioLegend) was used as a control. Activation of β3-integrin was also exogenously induced by adding to the medium 0.015mM MnCl₂, an activator of integrin receptors. Total cell numbers were then counted under a light-phase microscope after the culture.

CD34⁻KSL cells were cultured for 18 hours with AlexaFluor 647–conjugated human fibrinogen (Invitrogen) in S-Clone SF-03 medium supplemented with 50 ng/mL SCF or 50 ng/mL TPO. Samples without cytokine stimulation served as the control. The cells were washed twice in PBS and analyzed for fluorescence intensity using flow cytometry to assess the degree of fibrinogen binding to cells.

Ninety-six-well plates were coated with 5 μg/mL VN (Molecular Innovations) or 5 μg/mL OPN (R&D Systems) overnight at 4°C. They were then blocked with 1% BSA for 1 hour at 37°C, after which they were used as ligand-coated plates. BSA-coated plates served as the control.

After culture, exactly 10, 30, 50, 100, or 500 whole cultured cells were counted and sorted using a cell sorter and transplanted along with 2 × 10⁵ BMCs from Ly5.2 mice as competitor cells. Twelve weeks after transplantation, HSC frequency was estimated from the input cell number and the percentage of multilineage reconstructed mice (positive mice), based on the Poisson distribution.²²  As a control, single fresh (uncultured) CD34⁻KSL cells were also assessed using the same procedure.

HSC numbers were estimated based on the total cell number counted using a phase-contrast microscope and the HSC frequency determined from limiting dilution assays. Repopulating unit (RU) values were calculated from the results of LTCR assays using 40 fresh CD34⁻KSL cells or their progeny with 5 × 10⁵ Ly5.2 whole bone marrow competitor cells with the following formula:

RU = (% donor chimerism/% competitor chimerism) × 5

In addition, mean activity of stem cell (MAS) values, which reflect the repopulation ability of single HSCs, were estimated by dividing the RU values by the calculated HSC number.

Using CD48⁻KSL cells that were sorted after culture of CD34⁻KSL cells for 5 days with 2C9.G2 antibody or control IgG under the indicated conditions, RNA extraction, amplification, and microarray analyses were performed as described in the supplemental Methods. All microarray data are available for viewing at the Gene Expression Omnibus under accession number GSE33696.

Using 5000 sorted cells from each sample, mRNA expression was assessed using real-time quantitative RT-PCR as described previously.⁶ 

We previously reported that CD34⁻KSL cells expressed higher levels of β3-integrin than CD34⁺KSL cells, whereas other reports suggested that β1 integrin is involved in the maintenance of HSCs.^1,2  Our aim was to investigate why HSCs, and not progenitor cells, dominantly express β3, and whether this integrin has a distinct function in HSCs that differs from its function in other KSL cell populations.

We initially focused on intracellular signaling by β3-integrin in HSCs. It has been demonstrated that β3^−/− mice exhibit osteopetrosis induced by a lack of interaction between the tail of the β3-cytoplasmic domain and c-Src, leading to impaired osteoclast differentiation.²³  Although it has been proposed that dysregulation of osteoclasts leads to impairment of HSCs,^24,25  evaluation of the LTR activity of β3^−/− HSCs after transplantation into irradiated mice in vivo showed that β3-integrin might be involved in HSC function (Figure 1A-C). To clearly rule out the effects of osteoclast dysregulation on specific integrin signaling in HSCs, we simultaneously used L746A (leucine-to-alanine substitution) and Y747A (tyrosine-to-alanine) knock-in mice, which have normal osteoclasts (data not shown).^21,23  β3-Integrin “inside-out” signaling is mediated by talin binding to the Leu746 and Tyr747 residues in the β3 tail.¹²  Consequently, Y747A mutation disrupted both signals, whereas L746A mutation disrupted inside-out signaling without affecting outside-in signaling, for which pY747 is indispensable.²¹  Neither mutation altered the total cell number or the frequency of the HSC population (Figure 1D). In addition, there was also no difference between the expression profiles of integrin-β3 and CD150, a reliable maker of mouse HSCs,²⁶  within CD34⁻KSL cells derived from both mutant mice (Figure 1E).

However, when we performed a serial competitive repopulation assay, chimerism in primary recipients 12 weeks after transplantation showed defective reconstitution with Y747A HSCs, but not with L746A HSCs (Figure 1B). This finding was further confirmed by secondary transplantation (Figure 1C). Given the input HSC number used in the primary transplantation assays, these results indicate that stem cell activity was diminished in Y747A HSCs compared with WT HSCs, which suggests outside-in signaling via at least pY747 of β3^PY747, but not inside-out signaling or formation of an αvβ3/c-Src complex,²⁷  contributes to both the LTR and self-renewal activities of HSCs in vivo.

Given the specificity of the effects of β3^PY747, we wondered why β3^−/− HSCs did not exhibit impaired LTR activity after primary transplantation (Figure 1A-C). Cluster analysis of the whole transcriptome revealed that the gene expression pattern of Y747A HSCs differed from that of WT or β3^−/− HSCs (Figure 2A). GSEA identified 102 gene sets up-regulated (and 0 down-regulated gene sets) in WT HSCs compared with Y747A HSCs, based on thresholds set at a P value < .05 and FDR (q-value) < 0.25 (supplemental Table 1). On the other hand, there was no differential set between WT and β3^−/− HSCs (data not shown). These results indicate that the phenotype of β3^−/− HSCs is significantly closer to that of WT HSCs than Y747A HSCs, suggesting that an as yet undetected molecule (perhaps other integrin receptors) might exert a compensatory effect mitigating the β3 deficiency.

In addition, our GSEA study also showed that Y747A mutation significantly reduced enrichment of a gene set “DORSAM_HOXA9_UP,” which are up-regulated by Hoxa9 (supplemental Table 1), whereas the level of Hox-a9 expression did not significantly differ between Y747A and WT HSCs (data not shown). Hoxa9 is well known to be an essential factor involved in the maintenance of HSCs,²⁸  and TPO signaling reportedly promotes its transfer into the nucleus.²⁹  Moreover, Y747A HSCs also exhibited less enrichment of gene sets involved in “signal transduction” and “cytokine responses,” than WT HSCs (Figure 2B; supplemental Table 1). Thus, our results suggest that impaired cytokine-mediated maintenance of HSC activity in the TPO/c-mpl axis leads to reduced LTR activity in Y747A HSCs, probably resulting in a diminished capacity for engraftment in transplantation assays (Figure 1A-C).

Because outside-in signaling, at least via β3^PY747, appeared to be involved in the TPO-mediated effects on HSCs, we next examined whether this signaling would also mediate the effects of TPO on ex vivo cultures of HSCs. For exogenous manipulation of β3^PY747, we used 2C9.G2, a β3-integrin antibody that acts as a ligand stimulating transduction of αvβ3-integrin–mediated intracellular signaling in smooth muscle cells and neutrophils.^30,31  We also confirmed that 2C9.G2 ligation to β3-integrin induced phosphorylation of β3^PY747 as well as c-Src^PY418 and Syk, hallmarks of outside-in signaling from β3-integrin in mouse platelets (supplemental Figure 1).^9,32  After culture of HSCs in the presence of TPO alone, we assessed the effect of 2C9.G2 on the LTR activity of HSCs 12 and 20 weeks after transplantation. Interestingly, 2C9.G2 treatment in the presence of TPO positively influenced HSC LTR activity (Figure 3A), and this effect was blocked by Y747A mutation (Figure 3B). By contrast, this positive action of 2C9.G2 disappeared during culture in the presence of SCF, a crucial cytokine involved in the maintenance and proliferation of HSCs (Figure 3A). These results indicate that the β3^PY747-mediated effects on LTR activity in HSCs and the positive action by 2C9.G2 on the LTR activity of HSCs are both TPO-dependent. Interestingly, 2C9.G2 treatment had little effect on the frequency of KSL or CD48⁻KSL cells, a population enriched in HSCs after culture³³  (Figure 3C) or on total cell number (Figure 3D) after culture in the presence of TPO. Outside-in signaling via β3^PY747 per se appears to be independent of HSC expansion in the presence of TPO. This suggests that, given the effects of 2C9.G2 on LTR activity and HSC expansion, outside-in signaling via β3^PY747 may contribute to the enhanced LTR activity of individual HSCs only by collaborating with TPO, rather than through HSC amplification.

We have so far proposed a strong link between TPO and β3^PY747 signaling. As for why β3^PY747-mediated function in HSCs is dependent on TPO, we initially wondered whether TPO enhanced expression of integrin-β3 on HSCs. We found that, indeed, TPO-treated HSCs showed slightly less expression of integrin-β3 than fresh HSCs or SCF-treated HSCs (supplemental Figure 2). We next hypothesized that TPO might regulate integrin affinity via a conformational change for activation. Accordingly, purified CD34⁻KSL cells (from WT or β3^−/− mice as negative control) were treated with TPO or SCF for 18 hours, after which the binding of AlexaFlour 647–labeled fibrinogen, a ligand for αvβ3, αIIbβ3, and/or αMβ2, was assessed.³⁴  As shown in Figure 4A, TPO, but not SCF, increased fibrinogen binding to WT HSCs detected with flow cytometry. However, fibrinogen also bound to β3^−/− HSCs, indicating an interaction with another receptor, probably αMβ2 integrin.^34,35  After subtracting the binding to β3^−/− HSCs from that to WT HSCs, we were likely left with the binding to αvβ3 rather than αIIbβ3-integrin (Figure 4A). This is because we used Mg²⁺-containing Ca²⁺-free medium, which was previously shown to exclude binding to αIIbβ3-integrin (data not shown).^36,37  Thus, TPO apparently contributes to the activation of at least αvβ3-integrin via the inside-out intracellular machinery in HSCs.

Although it has already been shown that disrupting inside-out signaling without affecting outside-in signaling through L746A mutation has little effect on the capacity of HSCs for reconstitution in vivo (Figure 1A-C), we again tested whether the L746A mutation is independent of 2C9.G2-mediated β3^PY747 effects on the LTR activity of HSCs. Interestingly, L746A mutation did not appear to increase chimerism by 2C9.G2 treatment in a manner similar to Y747A mutation, even in the presence of TPO (Figure 4B left). Whereas the addition of Mn²⁺, a strong external inducer of integrin activation that acts independently of intracellular signaling, ameliorated the effect of 2C9.G2 on LTR activity in L746A HSCs (Figure 4B right), it had no such effect on Y747A HSCs (Figure 4C). There was also no effect of 2C9.G2 administration on the frequency of KSL or CD48⁻KSL cells (Figure 4D) or on total cell number (Figure 4E) after culture of L746A HSCs in the presence of TPO plus Mn²⁺, again indicating that the collaboration between TPO and outside-in signaling via β3^PY747 is independent of HSC expansion in vitro. Taken together, these results indicate that activation of integrin-αvβ3 and subsequent outside-in signaling via β3^PY747 are required to enhance LTR capability per individual HSC via an integrin-mediated function and suggest that these results are consistent with the in vivo study summarized in Figure 1A-C. On the other hand, forcing integrin activation with Mn²⁺ did not restore 2C9.G2 activity in the presence of SCF in WT HSCs (Figure 4F). We therefore concluded that not only outside-in signaling via β3^PY747, but also activation of integrin-αvβ3 per se, is required for TPO-dependent, but not SCF-dependent, full LTR maintenance in HSCs. We also sought to identify endogenous αvβ3-integrin ligands involved in the HSC function that maintains LTR. VN and OPN are well-known αvβ3 ligands. Interestingly, LTR activity was enhanced when HSCs were cultured with TPO for 5 days on plates coated with VN alone (Figure 4G). On the other hand, OPN exerted a positive effect on LTR activity only when Mn²⁺ was simultaneously administered (Figure 4G). Moreover, these natural ligands had no effect on HSC expansion, as evidenced by the similar frequency of HSC subsets (supplemental Figure 3A) and total cell numbers (supplemental Figure 3B) under all culture conditions. These findings again allowed us to consider the possibility that factors in the microenvironment other than TPO might influence integrin activation status and that full activation of integrin bound to extracellular matrix or other ligands might be required for β3^PY747-mediated enhancement of LTR activity per individual HSC.

We found that bidirectional β3-integrin signaling contributed to the maintenance of HSC activity at the single-cell level through collaboration with TPO. Ema et al previously reported that, under various conditions, SCF and TPO act together to effectively induce ex vivo expansion of HSCs through rapid cell division.³⁸  We therefore examined the effect of β3^PY747 signaling on the LTR activity of HSCs during ex vivo expansion induced by the combination of SCF plus TPO. Application of 2C9.G2 or control IgG to 40 HSCs during 5 days of ex vivo culture in the presence of SCF plus TPO did not alter total cell numbers (Figure 5A), and levels of β3-integrin bound with 2C9.G2 were not affected (Figure 5B). By contrast, the number of CD48⁻KSL cells was reduced by 2C9.G2 treatment (Figure 5C), suggesting that the frequency of HSCs after ex vivo culture was reduced by cellular manipulation via β3^PY747. Indeed, in cultures treated with 2C9.G2 for 5 days, the HSC frequency was lower and the number of HSC in transplantation assays was reduced (Figure 5D; supplemental Table 2), suggesting that outside-in β3-integrin signaling suppresses the amplification of HSCs. But interestingly, 2C9.G2-treated cells eventually showed a tendency toward increased HSC activity compared with cells cultured with IgG in LTRC assays (Figure 5E). This tendency was confirmed through in vivo evaluation of RU values, which imply LTR activity³⁹  and are based on the percentage of donor-derived cells in the peripheral blood of recipients 20 weeks after transplantation (supplemental Table 2), and the MAS values (RU divided by the number of detectable HSCs in recipient mice)⁴⁰  (Figure 5F). MAS values clearly showed that ex vivo culture with IgG reduced LTR activity per single HSC compared with uncultured fresh HSCs (Figure 5F). By contrast, 2C9.G2 treatment enabled purified HSCs to expand ex vivo without diminishing LTR activity (Figure 5F; supplemental Table 2). Thus, β3-integrin signaling may also be able to maintain LTR capability at the single-HSC level during ex vivo expansion.

Finally, to determine the molecules involved in β3-integrin signaling and acting in concert with TPO, we carried out DNA microarray analyses using purified HSCs. In sorted CD48⁻KSL cells after ex vivo culture, we selected genes showing elevated or reduced expression on administration of 2C9.G2 (vs IgG) in the presence of only TPO (Figure 6A-B; supplemental Table 3; supplemental Figure 4). Eleven of the up-regulated gene sets and 5 of the down-regulated sets were picked up by gene ontology (GO) enrichment analysis (Table 1). Interestingly, chromatin modification (GO 0016568), one of up-regulated gene sets, contained Vps72, which is known to be a key gene involved in LTR activity, and Mll1 (Table 1), which is involved in the maintenance of HSCs.^41,42  Similarly, Runx1 and Etv6, which are required for definitive HSC generation and maintenance, were also included among the up-regulated genes (supplemental Table 3).^43,44  Subsequent real-time quantitative RT-PCR confirmed the 2C9.G2-induced enhancement of Vps72, Mll1, and Runx1 expression in the presence of TPO alone, although Etv6 was not apparent (Figure 6C), and that effect was reversed by Y747A mutation in fresh HSCs (Figure 6D). These results suggest that the contribution of β3-integrin signaling to the maintenance of LTR activity by HSCs is related at least to the enhanced expression of these 3 genes.

In the present study, we showed that β3-integrin signaling contributed to HSC regulation by collaborating with TPO signaling. In Figure 7, we propose a model in which HSC maintenance is regulated by αvβ3-integrin-mediated bidirectional (inside-out and outside-in) signaling. Downstream signaling via β3^PY747 in the presence of TPO may also involve Vps72, Mll1, and Runx1, which contribute to the regulation of HSC LTR activity (Figure 6C-D). Perhaps through a still undetected mechanism, treatment with 2C9.G2 or a natural ligand not only contributes to enhanced LTR activity after culture in the presence of TPO alone (Figures 3 and 4), but also to the maintenance of HSC activity at the single-cell level, even after ex vivo expansion induced with SCF plus TPO, by compensating for the reduction in activity that occurs during the culture (Figure 5F).

HSCs treated with 2C9.G2 in the presence of TPO exhibited enhanced expression of genes related to chromatin modification (GO 0016568; Table 1), and the positive effect of 2C9.G2 on the LTR activity of HSCs was dependent on the presence of TPO, but not SCF (Figure 3A). These results suggest that outside-in signaling via β3^PY747 contributes to the maintenance of HSC LTR activity, possibly via chromatin modification in the presence of TPO signal. Consistent with that idea, integrin signaling reportedly acts in concert with cytokine stimulation to influence chromatin modification for the maintenance of tissue-specific function in mammary gland cells.⁴⁵  We therefore propose that a change in the TPO concentration (possibly a large increase) eventually elicits outside-in signaling via β3^PY747 that mediates chromatin modification related to cell division and/or maintenance of the stemness of HSCs. GO enrichment analysis also revealed a number of down-regulated gene sets, 2 of which were “cell division ” (GO 0051301) and “mitosis” (GO 0007067; Table 1). Moreover, 4 genes, Apc, Rb, Rbl2, and Nf2, within the up-regulated gene set “cell cycle ” (GO 0007507) also belonged to the GO term “negative regulation of cell cycle” (GO 0045786; Table 1). These data suggest that TPO and outside-in signaling via β3^PY747 act collaboratively to regulate expression of these genes, which may imply that HSC amplification is repressed during ex vivo expansion through a combination of SCF and TPO signaling (Figure 5D; supplemental Table 2). Indeed, single-cell culture of an HSC showed suppressed cell division in the presence of 2C9.G2 (supplemental Figure 4), again confirming the repressive effects of integrin-β3 signaling on HSC division during the ex vivo expansion trial with SCF and TPO. However, neither BrdU uptake assays nor Hoechst staining revealed a direct link between β3^PY747 and the regulation of the cell cycle in HSCs in vivo (supplemental Figure 5) or during ex vivo expansion (data not shown).

This suppressive action of integrin-β3 signaling was apparent in HSCs undergoing rapid cell division (ie, ex vivo expansion induced by SCF plus TPO; Figure 5D; supplemental Table 2), but not in slowly dividing HSCs (ie, in cultures exposed to TPO alone; Figure 3C-D; Figure 4D-E) or quiescent HSCs (ie, in vivo; Figure 1D). Because the effect of integrin-β3 signaling appears to be small, as discussed in the previous paragraph, we suggest that the suppressive action may be masked by the state of the slowly dividing or quiescent HSCs. By contrast, rapidly dividing HSCs (ie, during ex vivo expansion requiring by both SCF and TPO) may emphasize the suppressive effect of integrin-β3 on HSC expansion (Figure 5D; supplemental Table 2). We therefore suggest that suppression of cell division via a β3^PY747-dependent mechanism could contribute to the maintenance of HSCs through the prevention of excess cell division. But the true mechanism by which β3-integrin ligation associates with TPO signaling and influences HSC function remains to be elucidated.

Fresh L746A HSCs without ex vivo culture exhibited LTR activity similar to that of WT HSCs in primary and secondary transplantation assays (Figure 1A). This confirms the dominant importance of outside-in signaling via β3^PY747, as L746A is only critical for inside-out signaling, which implies a “change in the integrin conformational change.”²¹  But L746A mutants affected impaired outside-in based function (Figure 4B left), whereas ligation with 2C9.G2 in the presence of TPO plus Mn²⁺ augmented the LTR activity of HSCs (Figure 4B right). To explain this apparent discrepancy, we consider that at least outside-in signaling via β3^PY747 may require the fully activated form of αvβ3-integrin during ex vivo culture. Alternatively, stimulants other than TPO may be required for this action. An investigation of extrinsic factors that induce the fully active conformation of αvβ3-integrin in HSCs via a pathway not involving the talin-binding motif, particularly Leu746 of β3-integrin, or an unknown TPO-dependent ligand from surrounding cells, may enable us to suggest or clarify the “true” niche circumstances of BM.

In our limited studies, VN-coated plates favored TPO-dependent enhancement of LTR activity in HSCs without Mn²⁺ (Figure 4G), whereas OPN required Mn²⁺ for its effect (Figure 4G). This suggests that VN, a major component of the BM sinusoid,⁴⁶  rather than OPN, a component of the osteoblastic niche,²  is preferentially involved in HSC division via αvβ3-integrin. As has been claimed previously, the osteoblastic niche may preferentially use the interaction of β1 integrin (α4β1, α9β1) with OPN.^1,2  Alternatively, there may be a TPO gradient within the BM microenvironment (ie, the TPO concentration in the osteoblastic niche is lower than in the sinusoid). Because a higher TPO concentration would accelerate HSC division into progenitors and further megakaryopoiesis, TPO-dependent activation of αvβ3-integrin via inside-out signaling may induce subsequent ligation with an appropriate ligand, leading to Tyr747 phosphorylation and, in turn, inhibition of cell division and maintenance of LTR activity on HSCs.

Our results show that the phenotype of β3^−/− HSCs in transplantation assay and hierarchal cluster analysis is much closer to that of WT HSCs than Y747A HSCs (Figures 1A-C and 2A). This suggests that, in β3^−/− HSCs, an as yet undetected system exerts a compensatory effect in response to the deficiency in the β3 molecular component. By contrast, impairment of outside-in signaling through Y747A point mutation might effectively and negatively affect LTR activity without inducing compensation because the β3 molecular component is retained in Y747A HSCs. Identification of the mechanisms that redundantly compensate for the absence of integrin-β3 signaling in β3^−/− HSCs will require further investigation.

In conclusion, our findings demonstrate, for the first time, the critical role played by bidirectional β3-integrin signaling, especially outside-in signaling via β3^PY747, in the maintenance of LTR activity in HSCs in vitro and in vivo. Moreover, activation of integrin-αvβ3, but not inside-out signaling, is required for β3^PY747-mediated maintenance of HSC activity per each cell. This implies that the induction of activated integrin-αvβ3 can be regarded as a novel function of the HSC niche. We also show that outside-in signaling via β3^PY747 plays an essential role in HSC maintenance mediated by TPO but not SCF, which suggests a novel mechanism for HSC maintenance by TPO. These results thus open a new area of inquiry into the link not only between integrin-αvβ3 activation and the surrounding microenvironment within the BM niche, but also between integrin and cytokine signaling in the maintenance of hematopoietic homeostasis.

The authors thank Dr M. H. Ginsberg (University of California–San Diego, La Jolla, CA) and Bristol-Myers Squibb (New York, NY) for providing L746A and Y747A KI mice, all of the members of our research group for discussion, and Drs S. Mitani and T. Furukawa for their great help in sequencing the whole transcriptome using the SOLiD system.

This work was supported in part by the Formation of Innovation Center for Fusion of Advanced Technologies in the Special Coordination Funds for Promoting Science and Technology Cell Sheet Tissue Engineering Center (M.Y. and T.O.), KIBAN (B, K.E.), and the Ministry of Education, Culture, Sports, Science and Technology, Japan (Special Promoted Grant, K.E.), the Mitsubishi Pharma Foundation (K.E.), and the National Institutes of Health (grant HL078784 RO1, B.G.P.).

National Institutes of Health

Contribution: T.U. and K.E. designed the study; T.U. performed most of the cellular and molecular experiments and wrote the paper; K.E. edited the manuscript; J.I. and M.U. helped with the cellular experiments; H.T. bred and kept the mice; Y.S., M.T., and T.S. helped with transplantation assays; B.G.P. generated integrin knock-in mice; and Y.M., M.Y., K.N., Y.K., H.N., and T.O. supervised aspects of the project and helped with manuscript preparation.

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

The current affiliation for K.N. is Department of Ophthalmology, Osaka University, Graduate School of Medicine, Osaka, Japan.

Correspondence: Masayuki Yamato, Institute of Advanced Biomedical Engineering and Science, Tokyo Women's Medical University, Tokyo 162-8666, Japan; e-mail: myamato@abmes.twmu.ac.jp; and Koji Eto, Clinical Application Section, Center for iPS Cell Research and Application, Kyoto University, 53 Shogoin-Kawaharamachi, Sakyo-ku, Kyoto 606-8507, Japan; e-mail: kojieto@cira.kyoto-u.ac.jp.