Physiological Migration of Hematopoietic Stem and Progenitor Cells

See allHide authors and affiliations

Science  30 Nov 2001:
Vol. 294, Issue 5548, pp. 1933-1936
DOI: 10.1126/science.1064081


Hematopoietic stem cells (HSCs) reside predominantly in bone marrow, but low numbers of HSCs are also found in peripheral blood. We examined the fate of blood-borne HSCs using genetically marked parabiotic mice, which are surgically conjoined and share a common circulation. Parabionts rapidly established stable, functional cross engraftment of partner-derived HSCs and maintained partner-derived hematopoiesis after surgical separation. Determination of the residence time of injected blood-borne progenitor cells suggests that circulating HSCs/progenitors are cleared quickly from the blood. These data demonstrate that HSCs rapidly and constitutively migrate through the blood and play a physiological role in, at least, the functional reengraftment of unconditioned bone marrow.

Bone marrow (BM) transplantation is a common procedure wherein suspensions of cells are harvested from marrow cavities and intravenously injected into patients or animals whose blood-forming system is compromised by radio- and/or chemotherapy. The donor cells colonize the ablated marrow and reconstitute blood-forming capacity (1). It is not clear, a priori, why BM transplantation should work. Why should hematopoietic stem and progenitor cells isolated from resting BM accurately home to hematopoietic BM niches? Is this simply a random event or are the cells taking advantage of a normal physiological process—blood to BM homing—that exists in the absence of BM ablation? To investigate these questions, we examined the behavior of the hematopoietic stem and progenitor cells that normally exist in the blood.

Although most adult HSCs reside in BM and marrow is the principal site of adult hematopoiesis, it has been known for nearly 40 years that hematopoietic precursors are also present in the peripheral blood of normal animals (2–4). However, whether blood-borne HSCs play a role in BM or extramedullary hematopoiesis has remained unclear. To evaluate the physiological relevance of circulating HSCs, we generated pairs of parabiotic mice, in which partners differed at the CD45 locus (CD45.1 and CD45.2) (5–7). Cross circulation was established by day 3 after surgical joining, and blood chimerism reached ∼50% by days 7 to 10 (Fig. 1). Thereafter, blood chimerism remained stable, with no significant difference between chimerism of CD45.1 and CD45.2 partners (8).

Figure 1

Rapid establishment of cross circulation after parabiosis. Peripheral blood chimerism of parabiotic mice was determined by CD45 allotype analysis. Data were obtained by flow cytometry and are plotted as the percentage of partner-derived blood leukocytes for a representative pair. ▪, CD45.1 cells in CD45.2 partner; ○, CD45.2 cells in CD45.1 partner.

To test for engraftment of unconditioned BM by circulating HSCs, we separated established pairs of parabiotic mice for 7, 10, or 22 weeks before analysis of their blood and BM for chimerism. BM was analyzed by flow cytometry for the presence of partner-derived long-term (LT) HSCs, defined by the surface phenotype Thy-1.1loLineageSca-1+c-Kit+(9–13), and blood was analyzed for the presence of partner-derived mature blood cells (Fig. 2 and Table 1). After 7 weeks of parabiosis, surgical separation, and an additional 7 weeks as individuals, nine of nine surviving mice from five pairs had chimeric granulocyte compartments (Table 1, experiment 1). Because granulocytes are short lived [life-span ∼1 day (14)] and maturation of these cells from nonrenewing progenitor cells requires less than 4 weeks (15), partner-derived granulocytes present 7 weeks after separation must have been generated from BM engrafting LT-HSCs, which give rise to granulocytes for life, or from short-term HSCs (ST-HSCs), which give rise to granulocytes for only ∼6 to 8 weeks (9).

Figure 2

Cross engraftment of partner-derived LT-HSCs in BM of parabiotic mice. Mice were joined for 7 weeks and separated 10 weeks before analysis. Contour and dot plots show staining profiles and gating of BM cells for CD45.1 and CD45.2 LT-HSC. Boxes and irregular polygons indicate gates—the left-hand plot shows whole BM, the center plot shows only Thy-1.1loLineage cells, and the right-hand plot shows only Thy-1.1loLineageSca-1+c-Kit+cells (LT-HSCs). Numbers within plots represent the percentages of cells in the plot falling in gated areas.

Table 1

Stable engraftment of BM of unconditioned partners by LT-HSCs after 7 weeks of parabiosis. Pairs of mice were parabiosed for 7, 16, and 39 weeks in experiments 1, 2, and 3, respectively. In experiment 1, pairs were surgically separated for 7, 10, or 22 weeks before analysis. Data are presented as percentage of partner-derived granulocytes in peripheral blood (PB) or as percentage of partner-derived LT-HSC in BM. No statistically significant differences in granulocyte or LT-HSC chimerism of CD45.1 versus CD45.2 mice were observed.

View this table:

Ten or 22 weeks after separation, previously parabiosed mice were killed and analyzed for chimerism of BM LT-HSCs (Fig. 2 and Table 1). In five of five surviving mice from three pairs analyzed after 22 weeks of separation, all mice had partner-derived peripheral blood granulocytes, and all mice had partner-derived BM LT-HSCs in about the same proportions as the granulocytes (r 2 = 0.983) (16). In one pair (pair 5), analyzed after 10 weeks of separation, the CD45.1, but not the CD45.2, partner was chimeric for LT-HSCs, indicating that the partner-derived granulocytes observed at 7 weeks in the CD45.2 animal were descendants of engrafting ST-HSCs.

In mice that have received granulocyte colony-stimulating factor (G-CSF) or cyclophosphamide (Cy) plus G-CSF to mobilize HSCs and progenitors into blood, the spleen becomes a major homing site of mobilized cells (11, 17).However, the spleen is only a low-efficiency site of hematopoiesis (18), and so, in parabiotic pairs 2 and 3 (Table 1), we splenectomized one partner before parabiosis to attempt to enhance our ability to detect partner-derived marrow LT-HSC engraftment and hematopoiesis (19). The splenectomized CD45.1 partner of pair 2 died before analysis, but the surviving CD45.2 partner had high levels of granulocyte and BM LT-HSC chimerism 22 weeks after separation (15% and 20%, respectively). In pair 3, both the splenectomized CD45.2 mouse and its CD45.1 partner exhibited peripheral blood granulocyte chimerism (0.3% and 5.2%, respectively). The relatively high levels of chimerism in partners of splenectomized mice are consistent with previous reports describing increased numbers of circulating progenitors after splenectomy (20). Furthermore, these data indicate that the spleen is not required as a source of circulating HSCs for BM cross engraftment.

Slightly increased chimerism of BM LT-HSCs was observed after longer intervals of parabiosis. Whereas mice joined for 16 weeks (experiment 2) exhibited LT-HSC chimerism about equivalent to that of animals joined for only 7 weeks, two animals from a pair joined for 39 weeks showed 8.1 and 8.8% BM LT-HSC chimerism (Table 1, experiment 3).

Numerous cytokines and mediators of inflammation can increase the number of circulating progenitors and HSCs in the mouse [for review, see (21)]. To control for the possibility that enhanced mobilization of LT-HSCs could be induced by parabiosis surgery or by the state of parabiosis, we compared the LT-HSC content of blood obtained from parabiotic pairs with that of blood from unmanipulated animals (22, 23) (Web fig. 1) (24). No significant differences were observed in the levels of donor cell engraftment of animals competitively repopulated with parabiont blood (collected 24 hours, 72 hours, or 10 days after surgery) or with control blood. These data strongly suggest that enhanced LT-HSC mobilization is not required for BM cross engraftment, although we cannot exclude a potential effect of parabiosis on the kinetics of HSC exchange between marrow and blood.

To estimate the magnitude of hematopoietic stem and progenitor cell flux through the blood, we determined the rate of disappearance from circulation of intravenously transferred day 12 colony-forming units–spleen (CFU-S12) (25). In the BM, CFU-S12 colonies derive from LT-HSCs, ST-HSCs, multipotent progenitors (MPPs), common myeloid progenitors (CMPs), and megakaryocyte-erythroid progenitors (MEPs) (9,26–28). In blood, the majority of CFU-S12activity is contained within the Sca-1 fraction of blood cells, which is relatively enriched for myeloid committed progenitors (CMPs and MEPs) and depleted of Sca-1+ HSCs; however, a low level of CFU-S12 activity is also detected in the Sca-1+ fraction of blood (Web fig. 2) (24).

To assay the blood residence time of multipotent and oligopotent progenitors and HSCs, we obtained lineage-depleted (Lin) peripheral blood cells (29) from 10 primary donor mice and injected them intravenously into anesthetized primary recipients, whose blood was sampled through the tail vein before injection (prebleed) and at two time points (30 s and 6 min) after cell transfer. To measure CFU-S12 clearance, we injected collected blood from the primary recipient into lethally irradiated secondary recipient mice and counted spleen colonies in secondary recipients 12 days later. 92.8% of CFU-S12 activity contained in peripheral blood of donor mice was cleared from the circulation within 30 s, and 99.6% was cleared within 6 min (Fig. 3) (30) (see also Web fig. 3) (24). These data suggest rapid flux of HSCs/progenitor cells in the circulation of normal animals.

Figure 3

Rapid clearance of peripheral blood CFU-S12 activity. Lineage-depleted peripheral blood cells collected from 10 primary donor mice were injected intravenously through the retroorbital sinus into anesthetized primary recipients, whose blood was sampled through the tail vein before injection (“prebleed”) and at two time points after injection (30 s and 6 min). Blood collected from the primary recipient was injected intravenously into lethally irradiated secondary recipient mice, and spleen colonies in secondary recipients were counted 12 days later. As a control, 5% of the lineage-depleted peripheral blood cells from the primary donor were injected directly into irradiated recipients (“direct transfer”), bypassing the primary recipient, so that the starting CFU-S12 content of the primary donors' blood could be determined. Data are presented as the mean number (±SD) of colonies per μl of transferred blood. n = number of secondary recipients transplanted in each group. Differences between “direct transfer” and each of the other experimental groups were statistically significant (P < 0.05).

On the basis of phenotypic and functional analyses, we estimate that there are 100 to 400 LT-HSCs in the blood of a mouse (17) (Web fig. 1) (24). This pool of blood-borne HSCs is relatively stable in frequency (Web fig. 4) (24), indicating that the flux of HSCs into the blood from BM and other tissues and the flux of HSCs out of the blood must be essentially the same. Thus, the rapid clearance from circulation that we observed for injected stem/progenitor cells may imply that large numbers of HSC pass through the bloodstream each day. However, our data do not exclude the possibility that a substantial component of this clearance rate may relate to an exchange of cells between a freely circulating state and a “marginated” state, in which HSCs are effectively removed from circulation but remain in the vascular space through adherence to lumenal endothelial surfaces. Marginated HSCs could immediately extravasate into tissue or, alternatively, could release from the vessel wall to reenter the freely circulating pool. In this circumstance, each blood-borne HSC likely transits into and out of circulation several times, thereby reducing the absolute number of individual HSCs in the circulating pool. Thus, the rapid clearance of blood-borne HSCs from the free circulation does not necessarily imply eqivalent migration into and out of the marrow, and the marrow to blood influx rate of HSCs, as well as the fraction of blood-borne HSCs that are derived from marrow versus extramedullary tissues, must still be determined.

Although a complete description of the potential functions of blood-borne HSCs and oligolineage hematopoietic progenitors will require further investigation, it is clear from the above data that at least some of these cells are capable of reengrafting BM. This possibility has been suggested by Quesenberry and colleagues, who reported high levels of donor BM chimerism after intravenous injection of large numbers of BM cells into unconditioned hosts (31, 32). Our findings in parabionts considerably extend these observations, indicating that HSC engraftment of unconditioned BM does not depend on infusion of massive numbers of cells into the bloodstream and that cell turnover among BM HSC niches in normal animals is likely greater than previously appreciated. HSC migration may be important for redistribution of LT-HSCs within the BM compartment and may help to maintain hematopoietic homeostasis by insuring that HSC BM niches are not left unoccupied after HSC death or differentiation. Migratory HSCs may also function as an immediately available pool that can be rapidly recruited for extramedullary hematopoiesis after catastrophic blood loss.

The recent demonstration that small numbers of highly enriched BM-derived LT-HSCs give rise to hepatocytes in a liver injury model (33) suggests an additional potential function of blood-borne HSCs as a source of pluripotent or multipotent stem cells available for the repair of nonhematopoietic, as well as hematopoietic, tissues. Likewise, the constitutive transit of functional HSCs in the circulation of normal animals suggests caution in the interpretation of recent data regarding the generation of blood cells from nonhematopoietic tissues (34–36), as these tissues are likely to be contaminated with circulating HSCs. Even at low frequency, highly potent, itinerent HSCs could contribute substantially to hematopoietic readouts in assays designed to evaluate the “transdifferentiation” potential of nonhematopoietic tissues.

Finally, the existence of a physiological process by which HSCs and progenitors travel from BM to blood and back to functional niches in BM provides a rationale for the success of BM transplantation; the injected cells likely home to BM niches in irradiated recipients because the mechanisms underlying this migration preexist in unmanipulated animals. These data also suggest a modification of current thinking regarding mechanisms of induced HSC and progenitor cell mobilization, as “mobilizing” agents may function not only by increasing HSC emigration from BM but also by inhibiting normal reentry of these cells into tissues from the blood. Given the potential importance of the constitutively circulating pool of HSCs, further investigation of blood-borne HSCs phenotype, function, and evolutionary significance is of substantial interest.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: irv{at}


View Abstract

Navigate This Article