Hematopoietic Stem Cell Quiescence Maintained by p21cip1/waf1

See allHide authors and affiliations

Science  10 Mar 2000:
Vol. 287, Issue 5459, pp. 1804-1808
DOI: 10.1126/science.287.5459.1804


Relative quiescence is a defining characteristic of hematopoietic stem cells, while their progeny have dramatic proliferative ability and inexorably move toward terminal differentiation. The quiescence of stem cells has been conjectured to be of critical biologic importance in protecting the stem cell compartment, which we directly assessed using mice engineered to be deficient in the G1 checkpoint regulator, cyclin-dependent kinase inhibitor, p21cip1/waf1 (p21). In the absence of p21, hematopoietic stem cell proliferation and absolute number were increased under normal homeostatic conditions. Exposing the animals to cell cycle–specific myelotoxic injury resulted in premature death due to hematopoietic cell depletion. Further, self-renewal of primitive cells was impaired in serially transplanted bone marrow from p21−/− mice, leading to hematopoietic failure. Therefore, p21 is the molecular switch governing the entry of stem cells into the cell cycle, and in its absence, increased cell cycling leads to stem cell exhaustion. Under conditions of stress, restricted cell cycling is crucial to prevent premature stem cell depletion and hematopoietic death.

High levels of production of mature blood cells are needed to replace their rapid turnover, yet it has been hypothesized that the proliferative activity of hematopoietic stem cells (HSCs) is highly restricted in order to prevent susceptibility to myelotoxic insult or consumption of the regenerative cell pool (1–3). Once cells embark on a path of high proliferation, they appear to have longevity limited to 1 to 3 months (4). It has therefore been hypothesized that hematopoietic tissue is organized so that stem cells are relatively quiescent, but their more differentiated offspring have extremely robust proliferative potential (5). The precise role of stem cell quiescence in the maintenance of the stem cell pool is unclear, and the molecular mechanisms governing it are not well defined.

Although it has been hypothesized that the quiescence of stem cells is protective, it poses practical problems for bone marrow transplantation (BMT) and stem cell gene therapy. Current methods for expanding the number of stem cells often involve the use of recombinant cytokines. However, these molecules have pro-differentiative as well as proliferative effects, and expansion often occurs at the expense of multipotentiality. Alternative strategies include the disruption of the dominant antiproliferative tone, which must be mediated proximately by molecular checkpoints in the cell cycle machinery. Cyclin-dependent kinase inhibitors (CKIs) participate in the sequential activation and inactivation of cyclin-dependent kinases, processes that are central to progression through the cell cycle (6,7). We and others have postulated that perturbation of these regulatory circuits may result in changes in stem cell proliferation, and indirect evidence has been generated through demonstration that antisense p27Kip1 augments retroviral transduction of primitive hematopoietic cell populations (8). Targeted disruption of the gene encoding p21cip1/waf1 (hereafter p21) in mice has resulted in cells that are impaired in their ability to achieve cell cycle arrest after irradiation (9, 10), and antisense p21 has been shown to release human mesenchymal cells from G0(11). Therefore, p21 plays a role in at least some cell types in the transition out of the cell cycle and maintenance in G0. However, in hematopoiesis, levels of p21 have not been shown to be increased in CD34+ cells (12,13), and p21−/− mice have not been noted to have an altered hematologic profile (9, 10). Further, bone marrow progenitor cells from p21−/− mice paradoxically have decreased proliferative ratios in response to cytokines (14, 15). Yet we noted high levels of p21 mRNA when we assessed the quiescent stem cell–like fraction of bone marrow mononuclear cells. We therefore hypothesized that p21 plays distinct roles in subcompartments of the hematopoietic cascade, augmenting progenitor cell proliferation while inhibiting stem cell proliferation.

The direct impact of p21 on the stem cell compartment was assessed using mice engineered to be deficient in p21 (16). The cell cycling status of stem cells was determined using the RNA dye pyronin Y (PY) as a measure of quiescence among the lineage negative (lin) (17) and Hoechst 33342 (Ho) low–staining bone marrow cells (18). Cells from p21−/− animals consistently demonstrated a smaller fraction in the PY low portion of the continuum (Fig. 1) (P = 0.005,n = 6), which suggests that p21 does function to impede the entry of stem cells into the active cell cycle. Independently, rhodamine (Rho), a mitochondrial dye, and Ho were used to define the population of cells with low levels of metabolic activity and exclusion of Ho, corresponding to a quiescent stem cell pool (19). The Rholow/Holow population of linSca-1+ cells was also smaller in the p21−/−animals (P = 0.07, n = 3), confirming this observation.

Figure 1

Distribution of G0 versus G1 in the lin bone marrow mononuclear cell population defines an increased cycling fraction in p21−/− mice. Mouse bone marrow cells were stained with lineage antibodies, PY (RNA dye), and Ho (DNA dye). Lin cells were gated by means of a stringent parameter. Cells residing in G0 appear at the bottom of the G0/G1 peak, and G1 cells are the upper part as indicated (A). The average G0% in lin Holow cells from six experiments is shown in the graph (B). Data represent the mean ± SE, n = 6, P= 0.005. One or two littermates of each genotype were analyzed in each experiment.

To further define this issue, we injected −/− or +/+ mice with 200 mg of the antimetabolite 5-fluorouracil (5-FU) per kilogram of body weight (200 mg/kg) to selectively kill cycling cells (20,21). Marrow was harvested 1 day after 5-FU injection, and long-term co-culture or cobblestone area–forming cell (CAFC) assays were performed. These assays linearly correlate with in vivo repopulating potential (22, 23) and were used here as a stem cell assay instead of competitive repopulation assays, given the lack of a congenic mouse strain with a 129/SV background. A significant reduction in CAFCs was noted after a pulse of 5-FU in the −/− group as compared with the +/+ group controls (60.5% versus 10.8%, P = 0.0019) (Fig. 2A).

Figure 2

Response of p21−/− mice to 5-FU treatment in vivo demonstrates a higher cycling status and increased sensitivity to toxic injury. (A) CAFC reduction after a 5-FU pulse. A single i.v. injection of 5-FU at the dose of 200 mg/kg was performed, and cells for LTC with limiting dilution were obtained 1 day after the injection. CAFCs were counted at week 5. The yaxis values = [(CAFCs from untreated mice – CAFCs from 5-FU–treated mice)/(CAFCs from untreated mice)] × 100%. Data represent the mean from three independent experiments. Three littermates from each genotype were used in each experiment, and three to five limiting dilutions were applied for each sample. Student'st test was used to analyze the data (n = 3, P = 0.0019). (B) Survival outcome after sequential 5-FU treatment. 5-FU was administered i.p. weekly at a dose of 150 mg/kg, and the survival rates of the groups were defined. Results were analyzed with a log-rank nonparametric test and expressed as Kaplan-Meier Survival curves (n = 10, P = 0.0054).

When the animals were given 5-FU weekly as a challenge to assess the relative restriction on the cell cycle entry of primitive cells, the survival percentage in the −/− group was much lower than in littermate +/+ controls (10% versus 70% in 1 month, P = 0.0054) (Fig. 2B). To exclude the possible influence of toxicity to other tissues, we repopulated the hematopoietic system of lethally irradiated +/+ hosts with either +/+ or −/− bone marrow cells. One month after transplantation, we challenged the reconstituted animals with an identical protocol of sequential 5-FU treatment. A similar relative survival pattern was observed in mice carrying the −/− hematopoietic system, demonstrating markedly increased mortality as compared with those with a +/+ hematopoietic system (P= 0.0089). Therefore, death was due to hematopoietic and not other tissue sensitivity to the antimetabolite treatment. Thus, p21 restricts the entry of stem cells into the cell cycle and protects hematopoietic cells from destruction by cell cycle–dependent myelotoxic agents.

We next sought to determine whether the lack of p21 resulted in an increase in stem cell number in the basal state or in a decline due to more rapid depletion. The relative number of stem cells present in wild-type versus p21−/− mice was directly measured by limit-dilution CAFC assays. A significant increase in primitive cells in the p21−/− animals (Table 1, P = 0.0393, n = 7) was noted. Thus, p21 provides a dominant negative effect, which is sufficient to inhibit stem cell cycling. In the absence of p21, the inhibition is alleviated, leading to an expansion of the primitive cell pool under resting conditions. In contrast, no significant differences in colony-forming cells (CFCs), bone marrow cellularity, or white blood cells were noted (Tables 2 through 4), implying that p21 has a differentiation stage–specific function in HSCs. The paradoxic pro-proliferative effect of p21 in more mature progenitors observed by others may balance the inhibitory influence of p21 on stem cells (14, 15). This apparent dichotomy may reflect the complex biochemical role p21 has been noted to play as either a requisite participant in the formation of the cyclin–cyclin-dependent kinase complex that is necessary for the movement of the cell through late G1 into S, or as a CKI, inhibiting entry into S phase (24). We speculate that p21 plays a central role in determining the known differences in sensitivity to proliferative stimuli between stem cells and progenitor cells and that the opposing effects of p21 are selectively manifest based on the differentiation status of the hematopoietic cell.

Table 1

Comparison of CAFCs scored at week 5 between p21+/+ and p21−/− mice (per 105bone marrow mononuclear cells) demonstrates increased numbers of stem cells. Each pair was pooled from two or three −/− or +/+ littermate mice in each experiment. Each data point was generated from three to five limiting dilutions, and data were analyzed with the pairedt test.

View this table:
Table 2

Comparison of CFCs between p21+/+ and p21−/− mice (per 104 bone marrow mononuclear cells) indicates no difference in progenitors. Data represent colony-forming ability at day 10. Each pair was pooled from two or three −/− or +/+ littermate mice in each experiment. Each data point was generated from at least four replicates, and data were analyzed with the paired t test.

View this table:
Table 3

Comparison of total mononuclear cell number per bone marrow harvest between p21+/+ and p21−/− mice indicates no difference in cellularity. Each data point represents the mean from one to three –/– or +/+ littermate mice in each experiment. The total cell number (×107 per femur pair) was counted from each harvest, and data were analyzed using the paired t test.

View this table:
Table 4

Comparison of blood cell counts between p21+/+ and p21–/– mice (n = 10, mean ± SD) indicates no significant difference in mature cell populations. Blood was collected by tail bleeding. All the blood counts were performed and analyzed using the t test for two samples with the same variance. WBCs, white blood cells; RBCs, red blood cells; PLTs, platelets.

View this table:

The expansion of stem cells under normal homeostatic conditions may or may not reflect a capacity to self-renew under conditions of stress. The cytokine milieu dramatically changes during stress, including the elaboration of cytokines such as granulocyte-macrophage colony-stimulating factor (GM-CSF), granulocyte-CSF (G-CSF), and interleukin-3 (IL-3) with strong pro-differentiative properties. We hypothesized that the outcome of enhanced proliferation of the stem cell compartment under such conditions will markedly differ from normal homeostasis, and we directly assessed stem cell self-renewal capability using a serial transplantation approach (25–28). Bone marrow from 10 male animals of each genotype was individually transplanted into lethally irradiated female mice. Two to 4 months after engraftment, 1 × 106 to 2 × 106 bone marrow mononuclear cells from the transplanted recipients were used as donor cells for a lethally irradiated host, and the same procedure was repeated sequentially. Recipient animals began to die after the third serial transplant, and marked differential survival in the group was noted (Fig. 3A). No −/− transplanted animals survived after the fifth transplant, whereas the +/+ transplanted animals had a 50% survival 1 month after that transplant. To confirm the paucity of stem cells in −/− transplanted mice, we used two different doses of cells from the fourth transplant to rescue lethally irradiated hosts. The two irradiation protection experiments at different doses confirmed the significantly poorer ability of cells from the −/− group to rescue irradiated mice (Fig. 3B). We observed an approximately 100% contribution from the original donor p21−/− or p21+/+ cells in hosts examined after each transplant by means of semiquantitative Y chromosome–specific (Sry) polymerase chain reaction (PCR) and p21 genotyping PCR (Fig. 3C).

Figure 3

Animal survival after serial BMT demonstrates reduced self-renewal of hematopoietic potential. Male mice were used as marrow donors. Female recipient mice were lethally irradiated with 10 Gy of whole-body irradiation at 5.96 Gy/min. Two million nucleated cells were injected intravenously into the lateral tail veins of warmed recipient mice. Recipient mice were monitored daily for survival for more than 1 month. The mice were killed after 2 to 4 months, and bone marrow cells were prepared from those mice and injected into new female irradiated recipients. This process was repeated an additional four times. (A) Cumulative survival after serial BMT. Each group included 10 mice initially. The donor marrow from the previous transplant was injected into a new recipient individually, and therefore the actual recipient number was reduced during the serial transplantation. The ratio between the actual number of surviving animals at each BMT and the total number at first BMT is plotted as survival % (y axis value). Numbers above the bars indicate % survival. (B) Radiation protection of the marrow from the fourth BMT. Five × 105mononuclear cells from the fourth BMT mice were transplanted into the lethally irradiated recipients described as above, and survival data were analyzed using a log-rank nonparametric test and expressed as Kaplan-Meier Survival curves (n = 6 for each group,P = 0.002). Similar results were obtained at lower doses (105) of donor cells (n = 10 for each group, P = 0.008, curve not shown). (C) Donor contribution monitored by PCR. The contribution of the original donor cells was monitored by a PCR-based semiquantitative analysis for a Y chromosome–specific sequence (Sry), using an aliquot of marrow sample from each transplant. DNA was prepared from donor cells collected at the fourth transplant, and 200 ng was used for the PCR analysis. Two percent agrose gel was used to display the PCR products. The left gel shows the positive controls, which were mixed with male and female DNA at the ratios indicated. The complete contribution from donor cells was further confirmed by p21 genotyping PCR, shown in the right gel. Similar results were obtained from the first, second, and third transplants (data not shown in figure).

To evaluate whether the transplantation data could be affected by altered homing of stem cells in the absence of p21, we directly measured the localization of ex vivo fluorescently labeled p21−/− and p21+/+ bone marrow cells with carboxy fluorescein diacetate succinimidyl diester (CFSE) (29) after transplantation. The fraction of mononuclear cells or of lin, Sca-1+ cells homing to either bone marrow or the spleen was the same for the −/− and +/+ mice (Fig. 4).

Figure 4

p21−/− stem cell depletion is not due to altered bone marrow homing. Donor bone marrow cells were stained with the cytoplasmic dye CFSE and intravenously injected into lethally irradiated mice. Bone marrow and the spleen were harvested 9 hours after injection, and nucleated cells were stained with Sca-1 and lin antibodies and analyzed by flow cytometry. Two or three littermates of each genotype were analyzed in each experiment. Data shown are for bone marrow cells from one of two experiments with similar results.

These functional in vivo parameters of stem cell function were corroborated with quantitative in vitro measures of function of the primitive cell compartment. CAFCs scored at week 5 from −/− mice were completely exhausted after the third transplant, whereas detectable CAFCs were still noted in the +/+ group (Fig. 5A). Although an absence of CAFCs after 4 weeks was observed in both the +/+ and −/− groups from the fourth transplant, early CAFCs (scored at week 2 and 3) reflecting short-term repopulating cells (22, 23), demonstrated a significant difference (Fig. 5B). The CAFCs in −/− transplant recipients dampened to zero after 2 weeks, whereas the CAFCs in +/+ transplant recipients remained detectable at 3 weeks. The levels of CAFCs in −/− mice were also significantly lower than in the +/+ group [all P values <0.05 (Fig. 5B)].

Figure 5

CAFCs over the course of serial BMT confirm stem cell exhaustion. LTC with limiting dilution was performed on the donor cells of each transplant to quantify the frequencies of hematopoietic progenitors and stem cells. Normal untransplanted marrow was used as a control to ensure the quality of the stroma and the comparability of the experiments at different times. Data are represented as the mean ± SD and graphed as log scales on they axis. All P values are less than 0.05 (−/− versus +/+). (A) CAFCs at week 5 from the first and third transplants. (B) CAFCs at the indicated weeks from the fourth transplant.

Stem cell quiescence is therefore critical for both protection from myelotoxic injury and preservation of the stem cell pool under conditions of stress. p21−/− animals exhibit increased stem cell numbers under homeostatic conditions, without the ability to appropriately self-renew when transplanted. The distinction in stem cell generation between these different conditions may have several potential explanations. Telomere shortening with the demands of transplantation leading to cell death is possible (30) as is enhanced apoptosis in the −/− state (31), although the latter would be expected to effect basal as well as posttransplant cells. Differentiation of HSCs may be directly inhibited by p21, thereby enhancing differentiation in the −/− animals, but a pro-differentiative role for p21 has been described in all but terminally differentiated cells (13, 32). Rather, we propose a model in which the dominant inhibitory tone of p21 on cell cycle entry blunts the response to inflammatory cytokines, maintaining cells in G0 thereby precluding accomplishment of the differentiation program. Simply conceived, without the entry into the cell cycle, the differentiative effects of cytokines cannot proceed. This effect on the cell cycle, evident in regulating the size of the stem cell pool under normal homeostasis, is crucial in protecting stem cells from being consumed under conditions of stress. Therefore, cell cycle control is itself a critical determinant of stem cell pool persistence in vivo.

The ability of a single molecule, p21, to dictate stem cell pool kinetics in vivo suggests that p21 may be an important target molecule in efforts to manipulate stem cell proliferation ex vivo. Relieving p21-enforced inhibition of the cell cycle in the absence of pro-differentiative cytokines ex vivo may permit direct analysis of whether stem cell proliferation is necessarily linked to differentiation and will assess the validity of the concept of stem cell expansion.

  • * To whom correspondence should be addressed at Massachusetts General Hospital, 149 13th Street, Room 5212, Boston, MA 02129, USA. E-mail: scadden.david{at}


View Abstract

Navigate This Article