Many Mechanisms Mediating Mobilization: An Alliterative Review

Abstract

Purpose of review

Blood cell production is maintained by hematopoietic stem cells (HSC) that reside in specialized niches within bone marrow. Treatment with granulocyte-colony stimulating factor (G-CSF) causes HSC egress from bone marrow niches and trafficking to the peripheral blood, a process termed “mobilization”. Although the mobilization phenomenon has been known for some time and is utilized clinically to acquire HSC for transplant, the mechanisms mediating HSC release are not completely understood. We discuss recent advances and controversies in defining mechanisms responsible for G-CSF induced mobilization.

Recent findings

New reports define a role for resident monocytes/macrophages in maintaining niche cells, which is diminished after G-CSF treatment, suggesting a new mechanism for mobilization. While osteoblasts have been reported to be a primary component of the HSC niche, new results suggest a unique niche composed of innervated mesenchymal stem cells. Modulating bioactive lipid signaling also facilitates mobilization, and may define a future therapeutic strategy.

Summary

Hematopoietic mobilization by G-CSF is primarily mediated by alterations to the bone marrow niche by both direct and indirect mechanisms, rather than directly altering HSC function. Further understanding of the processes mediating mobilization will advance our understanding on the cellular and molecular components of the HSC niche.

Introduction

Hematopoietic stem cells (HSC) reside within specialized bone marrow niches, “tethered” through adhesion molecule interactions to a complex of stromal cells, endosteal lining osteoblasts, and mesenchymal stem cells within a matrix of collagens, fibronectin, and proteoglycans, where they produce mature cells that ultimately exit the marrow and enter the peripheral blood. HSC and hematopoietic progenitor cells (HPC) also traffic to the peripheral blood [1–6], leaving open niches that can be repopulated by transplanted HSC [7*]. Based on observations that increased circulating HPC were found in patients after chemotherapy [8,9], we now know that this natural trafficking of HSC and HPC can be modulated, allowing for directed “mobilization” [3,4]. Mobilization is widely used to acquire hematopoietic cells for autologous and allogeneic transplantation, achieved through administration of chemotherapy [8–10], hematopoietic growth factors, chemokines, or small molecule inhibitors or antibodies against niche chemokine receptors and integrins [11–13].

Granulocyte-colony stimulating factor (G-CSF) (Neupogen^R) is widely used clinically to mobilize HSC and HPC for transplantation. G-CSF-mobilized HSC and HPC are associated with more rapid engraftment, shorter hospital stay [14–17], and in some circumstances, superior overall survival compared to bone marrow [18]. The mobilization process is not well understood and many mechanisms mediating the effect have been proposed. This alliterative review focuses on recent findings that add to our understanding of the G-CSF mobilization response, but also introduces new controversies and questions.

CXCR4 controls (hematopoietic) cell circulation

The CXC4 receptor (CXCR4) and its ligand stromal cell-derived factor-1α (SDF-1α) are the most widely explored hematopoietic niche interaction regulating HSC and HPC retention and trafficking. Hematopoietic cells express CXCR4 and are chemo-attracted to and retained within the bone marrow by SDF-1α [19–21]. Genetic knockout of either CXCR4 [22] or SDF-1α [23] is embryonic lethal in mice, with a failure to populate the bone marrow niche during development. Conditional deletion of either CXCR4 [24] or SDF-1α [25*] results in a substantial hematopoietic cell egress from the bone marrow and impaired marrow retention of gene deleted HSC and HPC after transplantation [25*,26].

While numerous reports support a key role for the CXCR4/SDF-1α axis in hematopoietic cell retention/trafficking/mobilization within the bone marrow niche, the predominate source of SDF-1α is still not clear. Osteoblasts [27], reticular cells found in endosteal and vascular niches [28], endothelial cells, and bone itself [29,30] all produce/express SDF-1α. Recently, Mendez-Ferrer and colleagues described a novel population of nestin⁺ mesenchymal stem cells (MSC) that express high levels of SDF-1α mRNA and suggest they form a unique hematopoietic niche [31**]. Early reports demonstrated that osteoblast SDF-1α is reduced after G-CSF treatment [30,32,33], suggesting that osteoblast-derived SDF-1α was a key regulator of hematopoietic cell retention/mobilization. However, other studies question the importance of osteoblast-derived SDF-1α [28,31**,34]. Christopher et al. recently showed reduced SDF-1α production in Col2.3-expressing osteoblasts with no reduction in Col2.3 negative stromal cells, suggesting that reduced osteoblast SDF-1α is a common mechanism of cytokine-induced mobilization [35*]. However, Mendez-Ferrer et al. [31**], using a similar approach, showed substantially reduced SDF-1α in nestin⁺ MSC relative to a similar population of stromal cells described by Christopher et al. [35*], although a direct comparison to defined osteoblasts was not made. This is similar to findings that a mesenchymal precursor cell may form the hematopoietic niche [36*]. Overall, these results suggest that heterogeneous cell populations expressing SDF-1α and other supportive factors create unique niches, each of which can be manipulated to facilitate mobilization. Defining the specific niche(s) responsible for SDF-1α production and HSC retention will aid in development of future therapeutic strategies.

Osteoblasts, osteoclasts, osteomacs and other operators

Osteoblasts express signaling molecules in addition to SDF-1α that may regulate HSC function and niche retention [37–40]. Targeting the interaction between osteoblast vascular cell adhesion molecule-1 (VCAM-1) and HSC very late antigen-4 (VLA-4) with antibodies against VLA-4 [41,42] or VCAM-1 [43,44], or a small molecule inhibitor of VLA-4 (BIO5192) [45*], results in mobilization. Osteoblasts also express significant amounts of osteopontin (OPN), and HSC adhere to OPN via β₁ integrins [46]. Intriguingly, OPN also negatively regulates HSC pool size within the marrow niche [46,47], and OPN knockout mice show enhanced endogenous and G-CSF-induced mobilization [48**].

Hematopoietic mobilization with G-CSF results in suppression of niche osteoblasts [30,33,49**], with increased osteoblast apoptosis [33] and a characteristic osteoblast "flattening" [30], and decreased endosteal niche expression of retention molecules. This suppression has been reported to result from altered sympathetic nervous system (SNS) signaling to osteoblasts [30] (discussed below), however recent reports also suggest an alternate, perhaps parallel, pathway involving bone marrow macrophages. Winkler et al. reported that the marrow niche contains supportive endosteal lining “osteomacs”, and that G-CSF treatment results in a trafficking and reduction of these osteomacs, causing osteoblast suppression [49**]. Macrophage depletion using Mafia transgenic mice or by treatment with clodronate-loaded liposomes (Clo-lip), resulted in hematopoietic mobilization. The osteomac population of cells was characterized as F4/80⁺ Ly-6G⁺ CD11b⁺, largely based on findings that this was the predominate population depleted after Clo-lip treatment.

Two recent reports also suggest a role for resident macrophages/monocytes in mobilization. Chow et al. also demonstrated that macrophage depletion with Clo-lip results in mobilization [50**]; however, in slight contrast to osteomacs, these cells were defined as Gr-1^negative F4/80⁺ CD115^mid CD169⁺. In an animal model where CD169⁺ cells are specifically depleted, mobilization was enhanced. The authors suggest that CD169⁺ macrophages express a soluble, yet to be identified, factor that supports niche cells, specifically nestin⁺ MSC. Similarly, Christopher et al. utilized G-CSF receptor (G-CSFR) knockout mice crossbred to mice expressing G-CSFR under the control of the CD68 (macrosialin) promoter [51**], which restricts G-CSFR expression to the monocyte/macrophage lineages. Treatment of these mice with G-CSF showed a reduction in marrow macrophages, coincident with HPC mobilization. Intriguingly, we previously reported that treatment of mice with an anti-Gr-1 antibody to deplete neutrophils significantly reduced the G-CSF mobilization response [52], yet the animal model used by Christopher et al. showed mobilization to G-CSF despite neutropenia. As noted by Christopher et al., the antibody we used to deplete neutrophils (clone RB6-8C5) would also target monocyte and osteomac populations, possibly accounting for the differing effects seen. However, if this were the case, one would expect increased mobilization due to depletion of these populations, not decreased mobilization. In addition, the anti-Gr-1 antibody would not target the CD169⁺ macrophages [50**]. More likely, mobilization is a complex process coordinated by a balance between cell populations in the bone marrow rather than specifically a function of a single cell lineage. Thus, model systems that target specific populations would be sufficient to induce mobilization, but not necessarily exclusive. Clearly, the cellular players and factors warrant further exploration.

As a prime example of cellular balance, osteoblasts and osteoclasts regulate bone formation/bone resorption, within the bone marrow niche. Kollet et al. reported that RANK ligand treatment, which increases osteoclast activity, produces a moderate mobilization of HPC [53]. Similar results are also seen in an independent report [54*]. Correspondingly, stress models such as bleeding or LPS treatment increased the number of osteoclasts that was coincident with HPC mobilization. G-CSF mobilization was decreased after inhibition of osteoclasts, by treatment with calcitonin or using a genetic knockout model of PTP_ε, further suggesting osteoclast involvement in G-CSF-mediated mobilization. The authors proposed that osteoclast-derived proteolytic enzymes, such as Cathepsin K, degrade important niche interaction components including SDF-1α and OPN [53]. In a more recent study they demonstrated reduced osteoclast maturation and activity in CD45 knockout mice that correlated with reduced mobilization to RANK ligand and G-CSF [55].

In contrast, an earlier report demonstrated that while G-CSF increases osteoclast number and bone resorption in both BALB/c mice and humans, the increase in osteoclasts did not occur until 10–15 days or 6–8 days, respectively, after treatment with G-CSF [56], a finding also observed by others using similar systems [32,57]. Since G-CSF mobilization is typically evaluated after 4 to 5 days, the importance of osteoclasts to G-CSF induced mobilization remains unclear. Furthermore, treatment of mice with bisphosphonates, which inhibit osteoclast activity and/or number, prior to G-CSF administration does not result in impaired mobilization [49**,56], and in fact, in one case, bisphosphonate treatment increased mobilization [49**]. Of note, the endosteal bone surface, particularly underneath resorbing osteoclasts, is a significant source of extracellular calcium and studies by Adams et al. demonstrated that HSC expression of calcium sensing receptors mediates chemoattraction to soluble Ca²⁺ [58]. Calcium sensing receptor knockout mice had reduced marrow HSC content and increased peripheral blood HSC, perhaps suggesting that increased G-CSF mobilization seen in zoldronate treated mice resulted from reduced calcium retention signaling within the niche. While increased osteoclast activity can clearly induce mobilization, their role in G-CSF-mediated mobilization is not sufficiently defined and may not be a primary mechanism of mobilization. Figure 1 summarizes some of the key cellular components involved in HSC retention and subsequent mobilization following G-CSF treatment.

Figure 1.

Schematic representation of HSC niche retention and G-CSF induced mobilization. At steady state (left panel), HSC reside in bone marrow niches, closely associated with cells such as osteoblasts or nestin⁺ MSC that express SDF-1α and other supportive factors. Resident macrophages provide a positive supporting factor maintaining osteoblast and MSC activity. During G-CSF mobilization (right panel) G-CSF acts directly on the resident macrophages, causing them to transit away from the endosteal niche and reduces their number, and alters SNS signaling. This results in an attenuation of osteoblast function and a characteristic flattening, and reduction of supportive factors like SDF-1α in the marrow, with an increase in plasma SDF-1α. The role of both activated osteoclasts and neutrophils still remains unclear.

Hypoxia harbors hematopoiesis

Hematopoietic stem cells reside in hypoxic niches within the bone marrow [59–61], and HSC that reside in hypoxic niches have greater hematopoietic repopulating ability [62*]. Stabilization of the transcription factor hypoxia inducible factor 1-α (HIF-1α) is a known physiological response to hypoxia and HIF-1α has been shown to up-regulate erythropoietin (EPO) production [63]; numerous cell proliferation and survival genes [64-66]; the angiogenic growth factor VEGF [67]; and other genes. The hypoxic niche may maintain HIF-1α activity thereby maintaining HSC [68], a hypothesis supported by the fact that hypoxic conditions expand human HSC [69] and HPC populations [70–72] in vitro. In response to G-CSF, HIF-1α expression is increased [73*] and both the hypoxic environment and HIF-1α expand within the marrow compartment [74], with increased VEGF production; however, marrow vascular density and permeability are not increased [62]. In addition, HIF-1α is stabilized in peripheral blood HSC acquired from G-CSF mobilized donors [75]. HIF-1α also increases SDF-1α production [76] and CXCR4 expression [77], suggesting that hypoxia may be a physiological regulator of this important signaling axis. Of note, one report suggests that decreased osteoblast SDF-1α expression and reduced CXCR4 expression on bone marrow cells can be facilitated by osteoclasts activated by the hypoxia-mimetic CoCl₂ in vitro [54]. Recently, a population of marrow-derived very small embryonic-like stem cells was reported to mobilize as a result of intermittent hypoxia, coincident with an increase in plasma SDF-1α [78*], further suggesting that hypoxia may regulate HSC retention/trafficking.

HIF-1α also prevents hematopoietic cell damage caused by reactive oxygen species (ROS) [79*], suggesting that the hypoxic niche helps maintain HSC lifespan. In slight contrast, another report demonstrated that enhanced c-Met activity promotes mobilization by activating mTOR and increasing ROS production in HSC and HPC [80**], while inhibition of mTOR with rapamycin reduced HSC mobilization [80**,81], suggesting that a small amount of ROS may be necessary for optimal mobilization. Knockout of the thioredoxin-interacting protein gene results in increased mobilization under stress conditions [82*], suggesting a role for oxygen tension and ROS in mobilization.

Nervous Niche

The marrow microenvironment is highly innervated [83], and catecholamine signaling can alter numerous physiological processes of immune cells [84]. A seminal study by Katayama et al. demonstrated that G-CSF mobilization was reduced in chemically sympathectomized mice; mice treated with the β-blocker propanolol; or mice genetically deficient in the gene for dopamine β-hydroxylase (Dbh), an enzyme that converts dopamine into norepinephrine, demonstrating that mobilization requires peripheral β₂-adrenergic signals [30]. This study also demonstrated that G-CSF attenuated osteoblast function, via the sympathetic nervous system (SNS), resulting in osteoblasts having a marked flattened appearance. Intriguingly, β₂-adrenergic signaling can regulate osteoclast differentiation [85] and nestin⁺ MSC are highly innervated [31], perhaps suggesting that SNS signaling alters niche cell components through multiple mechanisms. In addition to the niche, human CD34⁺ cells express β₂-adrenergic and dopamine receptors that are upregulated by G-CSF [86], and neurotransmitters serve as direct chemoattractants to hematopoietic cells [86] and increase CXCR4 expression [87]. Epinephrine treatment also results in mobilization [86].

Signaling from β₃-adrenergic receptors regulate circadian oscillations mediating norepinephrine release, CXCR4 expression, and SDF-1α production, leading to rhythmic egress of hematopoietic cells from the niche [88,89]. Both β₂ and β₃-adrenergic signals cooperate to regulate G-CSF-induced mobilization, although knockout of both only partially blocks mobilization [90*] and does not abrogate mobilization by Clo-lip [50], suggesting multiple/parallel pathways regulating mobilization. β₂-adrenergic signaling also up-regulates osteoblast vitamin D receptors (VDR) and it was recently demonstrated that expression of VDR is necessary for the G-CSF-induced suppression of osteoblast function and that HSC mobilization is reduced in VDR knockout mice [91*]. VDR is also regulated by circadian rhythms [92], demonstrating further complexity in interconnected mobilization mechanisms.

Fatty acids for the future

Studies exploring the role of SNS signaling in G-CSF-mediated mobilization began in mice defective in ceramide galactosyltransferase (CGT), an enzyme that synthesizes galactocerebrosides, major lipid components of myelin sheaths and are important for proper nerve conduction [30]. Mice deficient in the related enzyme galactocerbrosidase (GALC) have a defective niche in which transplanted HSC fail to engraft [93*]. Sphingosine-1-phosphate (S1P) and ceramide, which affect a wide variety of cellular processes, are part of the GALC metabolic pathway [94]. HSC and HPC express the S1P receptor S1P1 [95], and S1P1 signaling alters CXCR4/SDF-1α signaling and chemotaxis [95–97]. S1P directs trafficking of immature B cells [98*] and trafficking of HSC and HPC from blood, bone marrow and lymph tissues [1]. A recent report by Ratajczak et al. suggests that plasma S1P increases following G-CSF administration and directs peripheral chemoattraction, facilitating mobilization [99**]. An increase in S1P in peripheral blood coordinate with a decrease in marrow was recently reported, suggesting the formation of an S1P gradient driving mobilization [100], or perhaps even a dual lipid action mediate by both S1P and ceramide concentrations in peripheral blood and bone, respectively [101]. If S1P signaling through S1P1 regulates G-CSF mobilization, then co-treatment with an S1P1 antagonist, like FTY720, would be expected to reduce G-CSF mobilization. This was indeed reported in one case [100], however others report no decreases in G-CSF mobilization in FTY720-treated mice [1,102]. Further studies evaluating the role of S1P and ceramide will aid in determining their functional importance in hematopoietic mobilization, and whether agents modifying S1P1 signaling, like the S1P agonist SEW2871, can be utilized as an adjunct therapeutic agent in mobilization.

In addition to the sphingolipids, other bioactive lipids have the capacity to regulate hematopoiesis, including eicosanoids. We recently reported that prostaglandin E₂ (PGE₂) regulates CXCR4 expression on HSC and HPC and facilitates their chemoattraction to SDF-1α and homing [103]. Like many lipid systems, homeostasis is maintained by a balance of signaling amongst eicosanoids with several of the eicosanoids acting in opposing manners. We also showed that agonism of cannabinoid receptors, receptors for the eicosanoid-related endocannabinoids, act in an opposing fashion to PGE₂ signaling, decreasing adhesion molecule expression and CXCR4, and enhancing G-CSF mobilization [104**]. Two separate reports by Jiang et al. have also demonstrated that cannabinoid signaling mediates mobilization, and that endocannabinoids are expressed in bone marrow and increase HPC migration and proliferation in vitro [105**,106*]. Another report demonstrated that cannabinoids mobilize myeloid-derived suppressor cells, with a possible role for endogenous G-CSF production [107*]. It is interesting to note that cannabinoid receptors are physically associated with β₂-adrenergic receptors and modulate adrenergic signaling [108*,109*], possibly suggesting a SNS/cannabinoid mechanistic link. With the abundance of FDA approved compounds modulating eicosanoid signaling, additional studies exploring their role in mobilization are likely to lead to new therapeutic strategies.

Conclusion

While significant progress has been made in defining the mechanisms mediating mobilization by G-CSF, a complete understanding remains elusive. Do osteoblasts cells comprise the important hematopoietic niche suppressed by G-CSF administration, or is it innervated nestin⁺ MSC, or a heterogeneous population of cells forming several unique niches? Experimental animal models designed to explore mobilization have significantly advanced the field, but manipulation of a specific systems while sufficient to mimic or block mobilization by G-CSF, may not tell the whole story. While both genetic alteration of the SNS or macrophages can mediate mobilization, neither regulates the niche exclusively during mobilization and other parallel or intersecting pathways are likely involved. Combination therapies manipulating multiple mechanistic pathways should be explored to not only maximally mobilize hematopoietic cells for transplant, but possibly to mobilize other potential therapeutically efficacious cell populations.

Key Points.

• G-CSF reduces resident monocytes/macrophages, causing niche suppression and mobilization.

• The niche consists of a heterogeneous population of cells that can coordinately be altered during mobilization.

• Sphingosine-1-phosphate, endocannabinoids and other bioactive lipids mediate HSC retention and trafficking to and from the bone marrow.

• The mechanisms regulating G-CS- induced mobilization alter multiple, parallel mechanistic pathways, suggesting combinatorial therapies should be explored.