Current mechanistic scenarios in hematopoietic stem/progenitor cell mobilization

The localization, as well as survival and proliferation/differentiation, of adult hematopoietic cells within bone marrow (BM) is vested on their relationships with microenvironmental cells and extracellular matrix. This relationship is highly dynamic, as it also allows their acute expansion or accelerated migration out of BM to meet physiologic demands. At steady state, proliferating hematopoietic cells in various stages of differentiation are confined within specialized BM “niches,” whereas terminally differentiated, mature cells leave the BM and migrate into blood. In addition, a small proportion of morphologically unrecognizable primitive stem cells regularly escapes the BM and circulates throughout life. Thus, in the peripheral blood (PB), cells at 2 extremes of the developmental spectrum of hematopoiesis continuously circulate. The mechanisms that allow these cells to leave the BM spaces are not clear. More importantly, the circulating pool of stem cells can increase in significant numbers by their enforced migration from BM using several empiric interventions. Enforced emigration of hematopoietic cells or “mobilization” has been observed in all species examined up to now, suggesting preservation of similar mechanisms of stem cell trafficking across species. Despite the significance and the great clinical impact of mobilization, deciphering the mechanisms of mobilization has defied significant efforts in that direction. In the last 3 years or so, several important insights have emerged, from both murine and human studies. It appears that during mobilization the bone marrow becomes the playground of a complex interplay between cytokines/chemokines, potent proteases, and adhesion molecules. Although each one of these players can independently initiate or intercept the mobilization cascade at some point, their pathways become interdependent. These new developments, as they shape our expanded understanding of the molecular mechanisms involved in mobilization, have been summarized in several recent reviews.^1-9  This perspective is intended to bring to light controversies in the data presented and the remaining gaps in our knowledge.

Granulocyte colony-stimulating factor (G-CSF)–induced mobilization affords great efficiency and enjoys broad clinical application. However, pondering the mechanism(s) involved has not been an easy task. Given the rather slow kinetics of mobilization with G-CSF (peak at 5-6 days) and the broad spectrum of target cells affected by G-CSF (both hematopoietic and nonhematopoietic), it was predicted that mobilization by G-CSF would be a complex cascade not to be easily dissected. Recently, however, thanks to availability of several genetic mouse models, many useful insights were obtained. The absence of response to interleukin-8 (IL-8) or G-CSF in G-CSF receptor (G-CSFR)–deficient mice solidified the fact that a sufficient pool of functional neutrophils with a normal response to G-CSF or IL-8 is critical for mobilization.^4,7,10  Transplantation experiments with G-CSFR^–/– or +/+ cells, separately or mixed in several ratios, provided additional enlightening data. In contrast to the absence of mobilization of G-CSFR^–/– cells transplanted alone into +/+ recipients, G-CSFR^–/– cells transplanted in the company of normal +/+ cells were mobilized after G-CSF treatment.⁷  These insightful data introduced the important concept that either a transcellularly mediated response or a diffusible factor is likely responsible for mobilization after G-CSF treatment. Direct in vivo studies have shown that neutrophils activated after G-CSF treatment release the contents of either specific (ie, matrix metalloproteinase 9 [MMP-9] or lactoferrin) or azurophilic (elastase, cathepsin G [CG], or proteinase 3) granules, and increase shedding of l-selectin or up-regulation of CD11b.¹¹  However, the mode of action of these proteases, their specific in vivo substrates, and their interplay with other molecules to bring about mobilization were only recently explored. Lévesque et al showed that following G-CSF mobilization, serine proteases, especially neutrophil elastase, accumulate within the bone marrow environment, and their substrates included molecules previously implicated in mobilization such as vascular cell adhesion molecule 1 (VCAM-1), c-kit, CXC chemokine receptor 4 (CXCR4), and its ligand stromal-derived factor 1 (SDF-1).^12-14  Nevertheless, the roster of other target molecules for these omnivorous proteases is quite broad and could theoretically include other, unidentified important molecules. For example, cleavage of vascular endothelial (VE)–Cadherin by proteases may play a critical, or only a facilitating role.¹⁵  Also, of interest is the inclusion of G-CSF itself as a target.¹⁶  This effect, as well as putative G-CSF consumption by an expanded pool of receptor-bearing cells, may shed light on the previously unexplained drop in mobilization efficiency after day 6 or so, despite continuation of G-CSF treatment. In addition to serine proteases, metalloproteases or dipeptyl peptidase IV (DPPIV/CD26) are also known to cleave some of the same target molecules, including SDF-1.¹⁷  The studies, therefore, suggested that the BM microenvironment with its tight cell-cell and cell-matrix interactions and membrane presentation of proteases becomes a favorable environment for their effective proteolytic activity, likely evading the action of their inhibitors.

Although several molecules previously implicated in mobilization were cleaved by proteases (ie, VCAM-1, CXCR4/SDF-1, kit), it was not clear to what extent each one contributed to G-CSF–induced mobilization, or whether a concerted action of several proteases with different target molecules was needed. Several subsequent studies attempted to address these issues in genetic mouse models deficient in serine proteases, or in MMP-9, or DPP1. G-CSF treatment of mice with combined elastase and cathepsin-G deficiency, or mice deficient in DPP1, lacking functional activation of many neutrophil proteases, provided further surprises. Each of these mice responded to G-CSF–induced mobilization to the same extent as control mice.¹⁸  To test whether other overlapping protease activities contributed to the response, MMP-9–deficient mice, or DPP 1–deficient mice were treated with a broad spectrum of MMP inhibitors (MM1270) and G-CSF. Despite the expectation that combinations of deficiencies would greatly attenuate the response, G-CSF–induced mobilization was not impaired.¹⁸  Compensation by other proteases in these deficient animals was offered as an explanation. Additional surprising observations followed. CXCR2-deficient mice were not only unable to increase MMP-9 activity from neutrophils following ligation by Groβ or IL-8, as expected, but they also failed to respond to G-CSF by increased numbers of circulating stem cells (although neutrophil numbers did increase), implicating MMP-9 deficiency as the culprit in G-CSF mobilization⁶  (Table 1). The response of these mice, however, differed from those with G-CSFR deficiency, as the CXCR2^–/– mice in contrast to G-CSFR^–/– mice were capable of responding to cyclophosphamide treatment,⁶  thus dissociating the effect of cyclophosphamide from MMP-9 activity, or the G-CSF effect.

A common characteristic of all the protease-deficient mice responding to G-CSF was a reduction of SDF-1α protein levels within the bone marrow,¹⁸  suggesting that reduction in SDF-1 may be the common denominator in G-CSF–mediated response in these mice. Whether this fall in SDF-1 within bone marrow is due to the action of another protease, such as CD26, which is physically associated with CXCR4 receptor and has been previously shown to cleave SDF-1,¹⁷  is presently unclear. However, CD26^–/– mice do have an attenuated response to G-CSF¹⁹  (Table 1). The emerging picture from the studies with protease-deficient mice is that the contribution of a single protease may not be critical; however, the requirement for a coalition of proteases within bone marrow is also speculative at present. Additionally, if MMP-9 is solely responsible for IL-8– or Groβ-induced mobilization, what are its targets? Is it the extracellular matrix components, or is it cell-bound molecules? Is CD26 responsible for the reduction in SDF-1 in mice with protease deficiencies, or are we dealing with a non–protease-dependent modulation of SDF-1? If a decrease in SDF-1 is necessary, is it dependent on the number and functional properties of neutrophils to bring about down-regulation (or up-regulation, depending on the study) of CXCR4? Thus, whether the effect is largely directed at the receptor or the SDF-1 levels within BM or downstream inhibition of SDF-1/CXCR4 signaling has not been clear from the published studies. (See further discussion in the next section.)

Mobilization has been achieved thus far with the use of several chemokines, such as IL-8, monocyte chemoattractant protein-1 macrophage inflammatory protein 1α or 1β, Groβ, SDF-1, or others (reviewed in Pelus et al⁶ ). A hallmark of the chemokine response is its fast kinetics with peak levels achieved from 30 minutes to the first few hours, in contrast to days required for cytokine response. Although this difference was cited to indicate independent mechanisms involved in cytokines versus chemokines, this reasoning may require revision. Mechanisms of chemokine mobilization have been explored only with the use of IL-8, Groβ, or SDF-1. The IL-8 and Groβ responses appear to be dependent on adequate numbers and normal function of circulating neutrophils, and they are associated with increase of MMP-9, presumably released by mature neutrophils.^20,27  Use of antineutrophil antibodies or of MMP-9 inhibitors greatly attenuated the response.²⁰  Corroborating evidence was obtained when CXCR2 knock-out mice showed little or no response to Groβ or IL-8 ligands.⁶  Also, G-CSFR^–/– mice responded neither to IL-8, nor to Groβ, presumably because of low numbers (G-CSFR^–/– mice are neutropenic) and/or unresponsiveness of G-CSFR^–/– neutrophils.

Biologic responses to SDF-1 have been extensively studied and recently reviewed.^5,8  Studies with SDF-1 or CXCR4 knock-out animals revealed important roles of the CXCR4/SDF-1 pathway in the active retention of hematopoietic cells within bone marrow. Furthermore, a number of subsequent studies suggested important roles in homing and stem cell mobilization, and SDF-1 has emerged as the most potent chemoattractant of hematopoietic stem/progenitor cells.²⁸  Its role in hematopoietic stem/progenitor mobilization was suspected by earlier studies showing decreased migration of G-CSF–mobilized cells to SDF-1.²⁹  Subsequent experimental data have bolstered this concept. Mobilization was seen using Met SDF-1 (an SDF-1 analog), which is refractory to cleavage/degradation and results in prolonged desensitization of CXCR4.²³  Moreover, adenovirus-driven SDF-1 with release of high circulating levels of SDF-1 or administration of a peptide-agonist of SDF-1³⁰  resulted in mobilization, presumably because of a change in SDF-1 gradient between BM and PB.³¹  Further observations suggested that small molecule inhibitors of CXCR4, which inhibit its binding to SDF-1 and disrupt SDF-1–dependent signaling, showed significant mobilization both in mice and humans.²⁴  Nevertheless, some conflicting observations also surfaced, suggesting that activation of CXCR4/SDF-1 signaling is required for mobilization, and that anti-CXCR4 antibodies, or a small molecule antagonist of CXCR4 that inhibits SDF-1 signaling (TC14012), inhibited G-CSF–induced mobilization.³²  On the other hand, mobilization through the use of Met SDF-1,²⁹  SDF-1 intrakine,³³  or CXCR4 antagonists,²⁴  or following Pertussis toxin (Ptx) treatment,³⁴  is likely mediated through down-regulation of the CXCR4 receptor and disruption of signaling on hematopoietic cells. Further, truncation of SDF-1 by serine proteases or MMP-9, as demonstrated earlier after G-CSF, will also inhibit SDF-1–dependent signaling in hematopoietic cells, rather than change the BM to PB SDF-1 gradient. The preponderance of evidence thus suggests that inhibition of CXCR4/SDF-1 signaling is a dominant pathway in G-CSF–induced mobilization. That being the case, it is hard to interpret the absence of G-CSF response in CXCR2^–/– animals⁶  (unresponsive to Groβ or IL-8 ligands), or the synergistic effects of CXCR4 antagonists in boosting G-CSF mobilization.²⁴  Non–SDF-1–dependent and Ptx-insensitive transmigration of cells shown to be promoted by BM endothelial cells³⁵  or liberation within BM of lipid mediators influencing cell motility³⁶  may also be participating.

The extent to which SDF-mediated mobilization is implemented through cross talk of SDF-1 with integrins, or tyrosine kinase receptors, or with small GTPases³⁷  involved in cell migration is also unclear. In this context, the study of patients with Warts Hypogammaglobulinemia Immunodeficiency Myelokathexis (WHIM) syndrome with a mutant CXCR4 receptor and diminished neutrophil egress from BM,³⁸  or of patients with Wiskott-Aldrich Syndrome (WAS), or mice lacking WAS protein and displaying adhesion/migration abnormalities and decreased cdc42 activation³⁹  will furnish additional insight.

Numerous studies thus far have examined the composite antigenic and functional phenotype of mobilized progenitor cells and compared them to steady-state BM cells. (Fu and Liesveld²  and Kronenwett et al³  [and references therein]). Among the most consistent differences irrespective of the mobilization stimulus and its kinetics are the absence of cycling and a decreased expression of very late activation antigen 4 (VLA-4) and kit in mobilized cells.⁴⁰  Baseline circulating stem/progenitor cells are also noncycling, and this feature has led to the conclusion that quiescent cells may exit the BM preferentially. More important than the expression level, however, is the reduced functional state of VLA-4 receptors.³  As the same functional changes were also observed in BM after treatment, they invited the speculation that this phenotype promotes transmigration. Whatever the predominant mechanism is in G-CSF–induced mobilization, chemokine-driven or not, it is well accepted that migratory responses are mediated through integrin participation. Thus, one could envision that integrin (functional) down-regulation is a common step at the final stages of transmigration through the endothelial sinuses. Such a proposition is in line with VLA-4 involvement in firm retention of normal⁴¹  or leukemic cells within the bone marrow.⁴²  The fact that inducible ablation of alpha4 integrins in adults leads to hematopoietic progenitor egress in mice⁴³  (Table 2) further reinforces this concept. Like the down-regulation of alpha4 integrin, additional hallmarks of transmigrated cells are the down-regulation of CXCR4 receptor⁴⁷  and hyporesponsiveness to SDF-1.²⁹  This could suggest that transmigration can be efficiently achieved only in cells that have down-regulated either their alpha4 integrin or CXCR4/SDF-1 function. It is important to emphasize that down-regulation of VLA-4 or CXCR4 receptor responses by themselves may not be sufficient to induce mobilization in all cells and additional complex interactions and cooperative signaling with other pathways involved in transendothelial migration (ie, small GTPases) may be necessary for the cells to exit the BM.

The detailed molecular pathways involved in the egress of progenitor cells from bone marrow to blood at steady-state hematopoiesis are currently unclear. Furthermore, it is also unsettled whether the same mechanisms are operable during the enforced emigration or mobilization of stem/progenitor cells from BM following several treatments. If the basal levels of circulating stem cells are the result of a continuous remodeling within the BM leading to their random release in circulation, then mobilization could be viewed as an exaggeration of normal processes, that is, increased bone resorption or other protease-dependent remodeling within the marrow environment. Alternatively, the fact that certain molecules are involved only in stress hematopoiesis and not under steady-state hematopoiesis would support the notion that during mobilization novel pathways assume importance not evident at steady state. Supporting this view is the evidence that mobilization defects have been observed in mice that do not show any changes in circulating progenitors at baseline (Table 2). Moreover, there are strain-dependent differences in mice both for baseline levels and for mobilization efficiency, and these are thought to be controlled by undetermined loci in chromosomes 2 and 11 and likely others⁴⁸  or possibly related to hematopoietic stem cell turnover and their response to cytokines.^49,50 

Additional questions relevant to the turnover and the fate of cells mobilized under basal conditions warrant further studies. The fact that hematopoietic stem cells have been isolated from many other tissues besides BM would indicate that circulating cells are in a dynamic equilibrium with these other tissues. Is the pattern and their survival in these other tissues altered after mobilization? Information on these issues is just beginning to emerge.^51,52 

Irrespective of the mechanisms, and the fate and the turnover rate of mobilized cells, it should be emphasized that both at baseline and after mobilization only a small proportion of cells is mobilized compared with the pool of cells already remaining in BM. Is this dictated by the presence of these cells in certain anatomic locations in BM or by their collective phenotype, which facilitates their egress? In addition, progenitor cells, with only few exceptions, leave the BM in the company of a much higher number of mature cells. How much does one population influence the other in mobilization (transcellular or humoral influence?), and is this different in the various mobilization schemes? Certainly, if changes in stromal cells and endothelial cells play a critical role in mobilization, this could favor the egress of both types of cells.

Finally, mobilization induced by growth factors other than G-CSF, that is, kit ligand or Flt-3 ligand, and their mechanisms are poorly understood and have remained unexplored. Assuming that mobilization is an active process, signaling pathways required are unclear, but emerging evidence incriminates several signaling molecules, that is, small GTPases or Src family kinases (SFKs) (Table 2). Their role, however, is most likely complex, so that combined deficiencies or cross talk with other signaling pathways may be required.

From the clinical standpoint, mobilization from BM to PB has become the most important aspect of stem/progenitor cell migratory behavior. However, our understanding of this important process is incomplete. Much of the recent information aimed at uncovering mechanisms of mobilization, especially G-CSF–induced mobilization, was generated by focusing attention to events within BM, and by studying genetically deficient mouse models. Recent experiments suggest that mobilization sets in motion an intramarrow proteolytic machinery with participation of diverse cell-bound or free proteases largely liberated by mature cells. This has provided a framework implicating multiple target molecules on hematopoietic cells or stromal cells and their matrix (adhesion receptors, chemokines and their receptors, or signaling molecules, etc). Depending on the stimulus applied, several distinct pathways can initiate mobilization. However, there is a broad interdependency between the pathways initiating and/or amplifying mobilization, underscoring the complexities involved. For example, proteases may affect each other's function (elastase affecting MMP-9 or MMP-9 affecting SDF-1, and vice versa), may target cytokine/chemokines receptors (CXCR4, kit) as well as their ligands (SDF-1, kit-ligand), or may additionally modulate integrins and endothelial structural components. It is evident, therefore, that there are large avenues open for future exploration. Nevertheless, one thing is clear: unveiling the regulatory pathways in mobilization will not only reap clinical benefits, but greatly enhance our basic understanding of the concept of stem cell “niche” pivotal in the retention and development of hematopoietic cells within bone marrow.

To all authors whose work has not been mentioned because of space constraints, I offer my apologies. The expert secretarial help of Marta Hill is gratefully acknowledged.