|Home | About | Journals | Submit | Contact Us | Français|
Bone marrow (BM) resident macrophages (Mϕs) regulate hematopoietic stem cell (HSC) mobilization, however their impact on HSC function has not been investigated. We demonstrate that depletion of BM resident Mϕs increases HSC proliferation as well as the pool of quiescent HSCs. At the same time, during bacterial infection where BM resident Mϕs are selectively increased we observe a decrease in HSC numbers. Moreover, strategies that deplete or reduce Mϕs during infection prevent HSC loss and rescue HSC function. We previously found that the transient loss of HSCs during infection is interferon-gamma (IFNγ)-dependent. We now demonstrate that IFNγ signaling specifically in Mϕs is critical for both the diminished HSC pool and maintenance of BM resident Mϕs during infection. In addition to the IFNγ-dependent loss of BM HSC and progenitor cells (HSPCs) during infection, IFNγ reduced circulating HSPC numbers. Importantly, under infection conditions AMD3100 or G-CSF-induced stem cell mobilization was impaired. Taken together our data show that IFNγ acts on Mϕs, which are a negative regulator of the HSC pool, to drive the loss in BM and peripheral HSCs during infection. Our findings demonstrate that modulating BM resident Mϕ numbers can impact HSC function in vivo, which may be therapeutically useful for hematologic conditions and refinement of HSC transplantation protocols.
Hematopoietic stem cells (HSCs) reside within the BM in a quiescent state, however, under stress conditions such as infection, HSCs can undergo proliferation and mobilize from the BM. HSC proliferation can result in differentiation and/or self-renewal to keep up with the cellular demands of infection while preserving the HSC pool. HSC fate is regulated, in part, by the BM microenvironment, which consists of niche cells including osteoblasts (Obs), mesenchymal stromal cells (MSCs), and endothelial cells. Niche cells express adhesion factors critical for anchoring HSCs in the BM [1–3] and also for maintaining HSC quiescence , which preserves self-renewal and long-term repopulation capacity .
Depletion of phagocytic cells, including BM resident Mϕs, with clodronate-encapsulated liposomes induces mobilization, which is thought to be due to diminished Obs, reduced Nestin+ MSCs, and impaired production of HSPC retention factors [6, 7]. Similarly, mobilization of HSPCs by granulocyte-colony stimulating factor (G-CSF) correlated with reduced BM-resident Mϕs numbers [8, 9], and depended upon G-CSF signaling in monocyte-Mϕ lineage cells . During infection, inflammatory factors can inhibit stromal niche cell function [11–13], however, the impact that infection has on resident Mϕs, and consequently HSPC mobilization, has not been investigated. Moreover, the impact of resident Mϕs on HSC fate, including self-renewal and differentiation, is unclear.
Using a mouse model of human monocytic ehrlichiosis (HME), we previously demonstrated that IFNγ drives the activation of dormant HSCs (Lin− c-Kit+ CD150+ CD48−) resulting in a transient loss in the HSC pool . HME is caused by the obligate intracellular bacterium Ehrlichia chaffeensis, and is modeled in mice using infection with E. muris (Em; ). In this infection model, robust CD4 T-cell-dependent IFNγ production in the BM  drives a loss of HSCs and profound anemia and thrombocytopenia [17, 18], which is also observed in HME patients . While HSC activation may benefit the host by increasing immune cell output during acute insults [16, 19, 20], prolonged HSC activation during chronic inflammatory conditions can result in BM failure [21–24]. Therefore, defining the mechanisms by which IFNγ directs the loss of HSCs may reveal potential therapies for patients with BM failure.
Based on recent observations that BM resident phagocytes maintain HSC niches we sought to investigate the impact of infection on these populations and evaluate their role in modulating HSC function. We found that phagocytic cells, most likely BM resident Mϕs, regulate the size and quiescence of the HSC pool under both homeostatic and infection conditions. During infection, BM Mϕ numbers were increased, which correlated with a reduced pool of HSCs in the BM and impaired HSPC mobilization. Mechanistically, diminished HSCs and impaired mobilization during infection was dependent upon IFNγ signaling in Mϕs. Our data suggest that Mϕs regulate mobilization during infection by diminishing the pool of HSCs and HSPCs in the BM, rather than increasing retention of HSPCs. Together, these findings refine our understanding of how IFNγ modulates HSC function and identify resident Mϕs as negative regulators of the HSC pool size at steady state and during infection.
C57BL/6 mice and the following gene-targeted strains were obtained from Jackson Laboratory (Bar Harbor, ME): C57BL/6-TG(CAG-EGFP), IFNγR1-deficient (B6.129S7-Ifngr1tm1Agt/J), and a CD45 congenic mouse strain (B6.SJL-Ptprca Pepcb/BoyJ). CD45.1 × CD45.2 mice were generated by crossing C57BL/6 (CD45.2) mice with Pepcb/BoyJ (CD45.1) mice. The MIIG (Mϕ-insensitive to IFNγ) strain was previously described  and a gift of Dr. Michael Jordan. All mice were bred and housed in the Animal Research Facility at Albany Medical College under microisolator conditions.
Female mice, at 6–8 weeks of age, were infected with 5 × 104 copies of Em via intraperitoneal injection. Bacteria was obtained from infected mouse splenocytes, as previously described .
PBS or 10 µg rIFNγ (PeproTech, Rocky Hill NJ) was administered to mice via retroorbital injection and BM was harvested 24 hours post-injection. PBS or 250µg/kg G-CSF (PeproTech, Rocky Hill NJ) was administered subcutaneously for 5 consecutive days and BM and blood was harvested 1 hour after the final injection.
250µl of PBS-encapsulated liposomes or clodronate-encapsulated liposomes (ClodronateLiposomes.com) was administered to mice via retroorbital injection every other day for three days. BM was harvested 4 hours after the last injection. During infection, PBS- or clodronate-encapsulated liposomes were administered on day 4 and day 6 post-infection and BM was harvested on day 11 post-infection.
BM was flushed from one femur and tibia and filtered through a 70 um mesh filter as previously described . Spleens were homogenized by crushing between frosted slides. RBC lysis was performed on single cell suspensions with ammonium chloride Tris buffer. Blood cells were obtained from whole blood using Lympholyte™-Mammal per the manufacturers instructions (Cedarlane, Burlington, NC).
Blood or spleen single-cell suspensions were plated at 4.0×105 or 2.0 × 105 per 35-mm tissue culture dish, in duplicate, in methocellulose media (MethoCult™ GF M3434, Stem Cell Technologies, Vancouver, BC, Canada). After incubation for 8 days at 37°C in 5% CO2 total myeloid colonies were counted under a light microscope.
Single-cell suspensions were plated, washed and stained with appropriate antibodies. The antibodies used for flow cytometry included the following: biotin-conjugated lineage markers specific for B220/CD45R (clone RA3-B62), CD3 (17A2), CD11b (M1/70), Ter119 (TER-119), Gr-1 (RB6-8C5), 7AAD (eBioscience), F4/80 (CI:A31), Ly6G (IA8), Ly6C (HK1.4), CD11b (M1/70), CD115 (AFS98), CD68 (FA-11), cKit (2B8), Sca-1 (D7), CD150 (TC150-12F12.2), CD48 (HM48.1), CD169 (3D6-112 AbD Serotec). Cells were analyzed on an LSR II (BD Biosciences) equipped with Diva software and analyzed using FlowJo software (TreeStar, Ashland, OR).
Mice were administered 5-bromo-2-deoxyuridine (BrdU) via intraperitoneal injection and BM was harvested 4 hours post-injection. Cells were surface stained followed by fixation/permeabalization (BD Cytofix/Cytoperm kit). Intracellular staining was performed for cell cycle analysis using Ki-67 (M-19; Santa Cruz) and DAPI was added 15 minutes prior to analysis. For BrdU staining, after fixation/permeabalization cells were incubated with DNAseI (Sigma) followed by staining for anti-BrdU antibody.
C57BL/6 or Pepboy (CD45.1) mice were lethally irradiated (950 RADs, administered in 2 doses, 4 hours apart). For steady state experiments, irradiated mice received a total of 5 × 106 BM cells derived from WT or MIIG (2.5 × 106 cells; CD45.1/2) and WT (2.5 × 106 cells; CD45.2) mice. For MIIG mouse infection experiments, irradiated mice received 2.5 × 104 sort-purified BM LK+ cells derived from Em-infected WT or MIIG (CD45.1/2) mice and 3.5 × 105 protective BM cells from WT (CD45.2) mice. For clodronate-liposome and infection experiments, irradiated mice received an equal mixture (2.5 × 104) of sort-purified BM LK+ cells derived from Em-infected mice treated with PBS-liposome (CD45.2) or clodronate-liposome (GFP:CD45.2) and 3.5 × 105 protective BM cells from WT (CD45.1) mice. Blood was screened every four weeks for up to 16 weeks.
Data was analyzed using Prism software.
BM resident Mϕs are key regulators of the HSC niche and their depletion results in mobilization [6, 7], however their role in regulating HSC function in the BM is unknown. Two populations of BM-resident Mϕs have been characterized as playing a role within the HSC niche: CD11b+F4/80+Ly6G+ Mϕs , referred to here as CD11b+ Mϕs, and CD11blo/−Ly6C−Ly6G−F4/80+CD115int. Mϕs , referred to as CD11blo/− Mϕs (Figure 1A). Both Mϕ populations express BM resident surface markers ([6, 26–28]; Figure S1A), and both populations are depleted or reduced upon administration of two-doses of clodronate-encapsulated liposomes or five consecutive doses of G-CSF, respectively (Figure 1B–C; [6, 9, 28]). We first wanted to examine the effect that Mϕ depletion (clodronate) or reduction (G-CSF) had on HSC numbers in the BM. HSCs were identified as LK+CD150+CD48− cells, similarly to previous studies [29, 30]. Because Sca-1 is dispensible for identifying HSCs , we excluded this marker from our analysis in the event that Mϕ cell death caused inflammation, which could arbitrarily induce Sca-1 on progenitors [32, 33]. However, LK+CD150+CD48− HSCs highly expressed Sca-1 when compared to all Lin−cKit+ (LK+) progenitor cells (Figure 1D). Two days after clodronate-liposome treatment we observed a significant increase in BM HSCs (Figure 1E). This appeared to be due to selective depletion of Mϕs on day 2 post-clodronate-liposome treatment, as BM monocytes and neutrophils were not depleted and were actually increased at this time point (Figure S1B). We also noted that the magnitude of the increase in HSC numbers was proportional to the degree of Mϕ depletion, as only a slight increase in BM HSCs was observed after Mϕ reduction with G-CSF. These data suggested that Mϕs limit the HSC pool at steady state, however, it is also possible that HSPC mobilization contributed to the less striking expansion of BM HSCs in G-CSF treated mice relative to the clodronate-liposome treated mice.
G-CSF has previously been shown to increase the BM HSC pool through self-renewing divisions [5, 34], thus to determine if HSC proliferation contributed to the expansion of the HSC pool after Mϕ depletion we examined BrdU incorporation by HSCs on day 1 and 2 post-clodronate-liposome administration (Figure 1F). Although we did not observe an increase in the frequency of HSCs on day 1 post-clodronate administration, there was a significant increase the number of HSCs incorporating BrdU, indicating that Mϕ depletion induced HSC proliferation. Furthermore, by day 2 post-clodronate we observed a significant increase in both BrdU+ and BrdU− HSCs, suggesting that HSCs were undergoing self-renewing divisions (Figure 1G). In support of this idea, the expanded HSC cell population contained an increased frequency and number of quiescent HSCs in G0 of the cell cycle (Ki67loDAPIlo) and cycling HSCs (Ki67+DAPI+) after Mϕ depletion (Figure S1C). Although we observed an increase in phenotypic HSCs after clodronate-liposome administration, we observed reduced reconstitution of whole BM from clodronate treated mice when compared to BM from PBS treated mice during competitive transplantation (Figure S1D). It is possible that HSC engraftment is compromised after Mϕ depletion due to increased proliferation at this time point . In addition, BM HSCs from Mϕ-depleted mice lost surface expression of a critical homing and engraftment protein, VLA-4 (Figure S1E) [35, 36]. Altogether, our data demonstrate that reduced BM resident Mϕs correlate with an expansion of phenotypic HSCs in the BM.
The observation that Mϕ depletion caused an increase in the number of phenotypic HSC prompted us to examine Mϕs in the BM during E. muris (Em) infection where we previously reported a loss in HSCs . During infection we observed an increased frequency of BM Mϕs, and despite marked BM hypocellularity , CD11blo/− Mϕ numbers remained stable and a significant increase in CD11b+ Mϕs was observed (Figure 2A). We noted an inverse correlation between BM Mϕ and HSC numbers. The decline in HSCs was most prominent on day 11 post-infection when Mϕ numbers were at their peak, suggesting that Mϕs may contribute to the loss in BM HSCs. To test this, Mϕs were reduced or depleted during infection (Figure 2B). Indeed, Mϕ reduction (via administration of G-CSF) and depletion of Mϕs by clodronate rescued HSC numbers during Em infection (Figure 2C and D). Our data suggest that Mϕ depletion alone accounted for rescuing HSC numbers, as monocyte and neutrophil frequencies remained stable when compared to PBS-liposome control mice during infection (Figure 2E). To determine if the phenotypic change in HSC numbers reflected a functional difference we performed competitive repopulation transplantations. Em can be detected in Lineage+ cells in the BM, therefore, to avoid transferring infection to lethally irradiated recipients, we enriched for HSPCs by sorting Lineage−cKit+ (LK+) cells. LK+ cells were sorted from PBS- or clodronate-liposome treated mice during infection and competitively transplanted in lethally irradiated recipient mice (Figure 2F). Upon screening the transplanted mice at 4, 8, 12, and 16 weeks post transfer we found significantly more donor-derived white blood cells (WBCs) from clodronate-treated mice (GFP+ cells) relative to cells from PBS-treated mice (GFP− cells), and significantly more BM HSCs at 16 weeks post-transplantation (Figure 2G–H). Thus, our data support a novel role for Mϕs in limiting HSC numbers and function during bacterial infection.
The infection-induced increase in BM Mϕ numbers coincided with the peak of IFNγ expression  and the loss in HSCs , which was previously shown to be IFNγ-dependent. Thus, we predicted that the increase in CD11b+ Mϕs required IFNγ. We found that this was the case as CD11b+ Mϕs were not increased, but rather decreased significantly, during Em infection in IFNγR-deficient mice (Figure 3A). IFNγ was also important in maintaining CD11blo/− Mϕ numbers during infection as this population decreased in IFNγR-deficient mice (Figure 3B). To determine if IFNγ acted directly on Mϕs, we examined “Mϕs insensitive to IFNγ “ (MIIG) mice that express a mutant IFNγR under the CD68 promoter rendering Mϕ-lineage cells unresponsive to IFNγ . Similar to IFNγR-deficient mice, CD11b+ Mϕs were not increased and CD11blo/− Mϕs were decreased in Em infected MIIG mice. Therefore, IFNγ signaling in Mϕs is responsible for their increase/maintenance during Em infection. Additionally, recombinant IFNγ (rIFNγ) alone, in the absence of infection, increased CD11b+ Mϕs in littermate control (LC) mice (Figure 3C). The increase in Mϕs was again dependent on the ability of Mϕs to respond to IFNγ, as a difference in CD11b+ Mϕ numbers was not observed in MIIG mice. Similar to infected WT mice, we did not observe a change in CD11blo/− Mϕ numbers in LC mice after rIFNγ treatment (Figure 3D). In contrast to infection, however, we did not observe a loss in CD11blo/− Mϕs after rIFNγ, which suggests that infection induces additional factors that promote the loss of CD11blo/− Mϕs in the absence of IFNγ. Our findings indicate that IFNγ is required to maintain or increase Mϕ numbers, particularly during infection.
Depletion of BM resident Mϕs increases HSPCs in the periphery [6, 7], thus we wanted to investigate whether IFNγ impaired HSPC mobilization in the context of infection when IFNγ increased BM Mϕs. We were unable to identify CD150+CD48− HSCs in the blood or spleen of Em infected mice suggesting that mobilization does not contribute to the loss of HSCs in the BM during infection (data not shown). However, consistent with the idea that Mϕs prevent increased circulating HSPC, we observed a significant decrease in blood progenitor cells (LK+), on day 11 post-infection (Figure 4A) when BM Mϕs were increased. Additionally, circulating HSPCs were not observed at earlier time points (data not shown). In contrast to WT mice, a significant increase in blood and spleen LK+ cells and splenic colony forming units (CFUs) was observed in Em-infected IFNγR-deficient mice (Figure 4A–D). Thus, impaired mobilization required IFNγ and correlated with the infection-induced increase in BM resident Mϕs. We did not observe a significant difference in blood LK+ cells in MIIG mice relative to LC mice (Figure 4E), however infected MIIG mice exhibited increased splenic HSPCs and CFUs (Figure 4F–G). The observation that MIIG mice did not have increased blood LK+ cells may be explained by increased recruitment of progenitors into peripheral tissues as endothelial cells and other cell types can respond to IFNγ in MIIG mice.
To examine the possibility that increased bacterial burden in IFNγR-deficient mice caused an increase in mobilization we examined the relationship between peripheral HSPCs and bacterial burden. We failed to detect a correlation between bacterial burden and peripheral HSPC numbers (Figure S2) demonstrating that mobilization is not dependent upon bacterial burden. Thus, IFNγ signaling in Mϕs impairs mobilization during Em infection, possibly through maintaining the population of resident Mϕs.
Although previous data implicated retention in the BM as the mechanism behind Mϕ-depletion-induced HSPC mobilization [6, 7], we found that impaired mobilization during infection was not due to increased retention in the BM. In fact, the infection-induced decrease in blood LK+ cells correlated with reduced LK+ cells in the BM (Figure S3). In addition, AMD3100 (CXCR4 antagonist) and G-CSF-induced mobilization was diminished during infection (Figure S3A). Thus, reduced mobilization during infection is due to an overall decrease in the HSPC pool size, rather than increased retention in the BM.
The fact that IFNγ impaired HSPC mobilization during infection is consistent with the finding that IFNγ increased BM Mϕs, an important niche cell. However, this observation is somewhat surprising as it was previously shown that rIFNγ induced mobilization as evidenced by the accumulation of phenotypic HSCs (LSK+ CD150+) in the spleen . Thus, we tested the impact of rIFNγ on mobilization in the absence of infection. While we observed increased splenic LSK+CD150+ cells upon rIFNγ administration (Figure S4A), this increase most likely reflects a change in Sca-1 expression as we also noted an overall increase in Sca-1 expression on splenic LK+ cells (Figure S4B–C). As IFNγ can cause Sca-1 expression on many cell types, we excluded Sca-1 from our analysis, which revealed a decrease in the frequency of splenic LK+ cells (Figure S4D). Moreover, we observed a functional decrease in progenitor activity in the spleen and blood (Figure S4E–F). Thus, our data incisively demonstrate that IFNγ does not induce HSPC mobilization, but actually impairs mobilization.
The observation that IFNγ is critical for both the increase in BM Mϕs and the loss in BM HSCs suggests that the IFNγ-dependent loss in HSCs is dependent upon Mϕs. In line with this, the absence of IFNγ signaling in Mϕs prevented the loss of BM HSCs and HSPCs during infection (Figure 5A–C). Upon transplantation of sort-purified LK+ cells from Em-infected MIIG or LC mice (Figure 5D) we found an increase in donor-derived WBC frequencies in lethally irradiated mice that received LK+ cells from infected MIIG mice relative to infected LC mice (Figure 5E). In addition, there was a trend towards increased MIIG-derived donor HSCs in the BM sixteen weeks post-transfer (Figure 5F). Increased numbers of HSCs among LK+ cells in the infected MIIG mice was likely responsible for our observations, however we can not rule out the possibility that HSCs from infected MIIG mice had enhanced engraftment or self-renewal ability.
HSC quiescence directly correlates with increased self-renewal and reconstitution potential after transplantation , thus we examined HSC cell cycle during infection. During infection, where we observed increased reconstitution of MIIG HSPCs relative to LCs, we also observed an increased frequency and number of G0 (Ki-67loDAPIlo) HSCs in MIIG mice when compared to LCs (Figure 5G). These data support the conclusion that IFNγ signaling in Mϕs reduces the quiescent pool of HSCs during infection.
IFNγ can act directly on BM HSCs to reduce their numbers in vitro  and during viral infection in vivo . However, our observation that Mϕs were required for the IFNγ-dependent loss of HSC function in vivo is supported by our finding that IFNγ did not act directly on HSCs to drive their loss in mixed BM chimeric mice (Figure S5A). We observed a similar proportion of WT to IFNγR-deficient HSCs in infected mice when compared to uninfected controls (Figure S5B). In addition, we observed increased sensitivity to IFNγ by BM Mϕs as compared to sort-purified HSCs, short-term repopulating HSCs (LSK+CD150+CD48+) and multipotent progenitors (LSK+CD150−CD48+) (Figure S5C–D). Thus, whereas HSCs can respond to IFNγ in vitro, other cell types including Mϕs are more sensitive to IFNγ and represent likely targets of IFNγ in vivo.
Although we did not observe a significant difference in HSC numbers (LK+CD150+CD48−) between uninfected MIIG and LC mice, we were curious if basal IFNγ signaling in Mϕs affected HSC quiescence. Thus we performed cell cycle analysis on HSCs from uninfected MIIG and LC mice. We observed a higher frequency and number of HSCs in G0 in MIIG mice relative to LC mice and a significant reduction of HSCs in G1 of the cell cycle (Figure 6A–B). These data suggest that even at steady state, IFNγ signaling in Mϕs can impact HSC quiescence. To determine if the increase in G0 HSCs resulted in improved HSC function we tested the ability of whole BM derived from MIIG and LC mice to competitively reconstitute the hematopoietic compartment of irradiated mice (Figure 6C). We observed a slightly higher percent of donor WBCs and BM HSCs in lethally-irradiated mice that received MIIG BM relative to LC BM (Figure 6D–E). Thus, in addition to infection, the ability of Mϕs to respond to IFNγ regulates the population of quiescent HSCs at steady state.
These studies define a novel role for BM resident Mϕs in regulating the HSC pool size in vivo, and refine our understanding of how IFNγ impacts hematopoiesis and HSC function. Mϕ depletion was previously shown to increase circulating HSPCs at steady state [6, 7]. Our data extend this finding to show that increased Mϕs restrict the BM and peripheral HSPC pool size by reducing BM HSCs and HSPCs in an infection model (Figure 7). The inverse relationship between Mϕs and HSCs was further demonstrated in transplantation assays where increased Mϕs correlated with diminished HSC function. To our knowledge, the observation that IFNγ promotes increased Mϕs during infection is the first example of a physiologically relevant condition in which BM resident Mϕs are increased and directly impact HSC function. The mechanism by which IFNγ drives the increase in resident Mϕs is an area of current investigation.
Our finding that IFNγ signaling in Mϕs promoted the loss of HSC function builds upon the established role of IFNγ as a suppressor of HSC function. In addition to Em infection, IFNγ reduced HSC reconstitution ability during Mycobacterium avium infection . Even at steady state, HSCs from IFNγ-deficient mice engraft with greater efficiency relative to HSCs from wild type mice . MIIG mice contained more HSCs in G0 of the cell cycle relative to LC mice, suggesting that IFNγ signaling in Mϕs reduces HSC engraftment by actively suppressing their quiescence. Thus, the increased engraftment of HSCs derived from an IFNγ-deficient environment is likely Mϕ-dependent.
Similar to our findings, IFNγ was reported to negatively impact the HSC pool size during viral infection with lymphocytic choriomeningitis virus (LCMV) . In the case of LCMV infection IFNγ was not responsible for the loss of HSCs, rather it acted directly on HSCs to suppress self-renewal, thus delaying HSC recovery in the BM. This difference highlights the unique contexts of viral and bacterial infections where expression of IFNγ and other cytokines may be distinct or have different kinetics. LCMV induces the expression of both type I and II IFNs [38, 39], and the presence of type I IFNs during early LCMV infection may sensitize HSCs to IFNγ. Type I IFNs are not elicited by Em infection , which could explain the observational differences.
The ability of Mϕs to reduce HSC numbers during infection complements previous work demonstrating Mϕ-dependent suppression of HSC expansion in vitro [40–42]. Our data is also consistent with in vivo data demonstrating that cyclophosphamide/G-CSF, two agents that reduce BM Mϕs [8, 9], expands the BM HSC pool [5, 34]. This expansion, similar to our findings with clodronate-liposomes, suggests that reducing Mϕ numbers increases self-renewal. However, it is also possible that previously active HSCs (LSK+CD48+CD150+) contributed to the increase in the quiescent HSC population after Mϕ depletion. Importantly, the HSC expansion was most prominent on day 2 post-clodronate when compared to day 7 post-clodronate, indicating that Mϕ depletion causes a transient expansion in the BM HSC pool.
Notably, abrogating IFNγ signaling in Mϕs at steady state increased HSC reconstitution whereas Mϕ depletion reduced HSC reconstitution at steady state. A potential explanation for these seemingly contradictory findings is that different forms of proliferation may be occurring in these experimental models. It is possible that tonic IFNγ signaling in Mϕs promotes differentiating divisions, while Mϕ depletion removes these differentiation cues thus allowing for self-renewing divisions. Both scenarios would result in proliferation and ultimately reduce the ability of HSCs to home and engraft in a transplantation setting . The difference in HSC function may also be due to the fact that MIIG mice were not experimentally manipulated prior to BM transplantation, while transplantation in Mϕ-depleted mice took place two days post-clodronate administration. Similar to Mϕ-depletion, G-CSF administration, which reduces Mϕs, also reduced BM reconstitution [9, 43]. Interestingly, however, BM reconstitution was actually increased two weeks post-G-CSF treatment when compared to untreated control mice . These findings suggest that although reducing Mϕs initially reduces HSC function, allowing the HSC pool to recover after treatment may be necessary to assess an increase in HSC function and self-renewal. Examination of the BM compartment after extended Mϕ depletion may reveal the true impact that Mϕs have on HSC function at homeostasis.
While the precise mechanisms whereby Mϕs regulate HSC function are not clear, there are several possibilities. Mϕs communicate with Obs, MSCs, and endothelial cells [6, 7, 45], thus these niche cells may serve as an intermediary between Mϕs and HSCs in determining HSC fate both at steady state and under inflammatory conditions. Resident Mϕs may also directly act on HSCs to regulate their function as a rare population of α-smooth muscle actin+ monocyte-Mϕs was shown to regulate HSPC function after irradiation in a contact-dependent manner . Thus Mϕs may modulate the BM niches where HSCs reside, or directly interact with HSCs.
Similar to Mϕ depletion, neutrophil depletion also increases HSCs, possibly through a density feedback mechanism to replenish this short-lived cell type . Thus it is possible that, in addition to Mϕs, transient depletion of other phagocytic cells may promote HSC expansion to enhance myeloid cell replenishment. BM Mϕs efferocytose senescent neutrophils [48, 49], thus it is also possible that Mϕ depletion disrupts neutrophil homeostasis and expands HSCs in response to emergency granulopoiesis. In either case, our work highlights the importance of Mϕs in suppressing HSC expansion at steady state and emphasizes their negative impact on the HSC pool during infection.
Mϕ-mediated regulation of HSC proliferation is likely critical for maintaining normal hematopoiesis. At homeostasis, limiting excessive HSC proliferation may reduce the likelihood of HSCs to acquire mutations or become exhausted. As similar mechanisms regulate HSC and leukemic stem cell (LSC) function within the BM , Mϕs may also suppress LSC proliferation. Under infection conditions, Mϕ-mediated suppression of HSC self-renewal may promote a shift toward differentiation allowing for the rapid production of mature effector cells. Increased monocyte production, for example, is critical for protection against intracellular pathogens including Em  and Listeria monocytogenes , and this process coincides with a decrease in quiescent HSCs and is dependent on IFNγ . Thus, it is likely that Mϕs drive a shift from HSC self-renewal to differentiation in order to shape the immune response during infection (Figure 7). Determining additional factors that modulate Mϕ numbers may provide a better understanding of how hematopoietic stress responses are governed.
HSPC mobilization is thought to protect the host by enhancing rapid production of mature effector cells in situ [52, 53]. However, our data suggest that the effect of mobilization is context dependent as IFNγ is a protective cytokine during Em infection that increases Mϕs and consequently impairs HSPC mobilization. Mobilization has been reported to occur during Escherichia coli infection in a G-CSF-dependent manner . Interestingly, G-CSF reduces BM Mϕ numbers and causes HSPC mobilization [8, 9], thus it is likely that infection-induced mobilization depends upon the type of pathogen present and differential modulation of BM Mϕs numbers.
The impact that Mϕs have on the BM HSC pool suggests that targeting this cell type may be an effective treatment strategy for hematologic disorders. Depletion or reduction of BM Mϕs may eliminate cytopenias associated with chronic inflammatory conditions such as infection and aplastic anemia. In fact, we found that IFNγ, which is highly expressed in the BM of Fanconi anemia patients , drives an increase in Mϕs and consequently reduces the HSC pool. Our study is also relevant to mobilization strategies for harvesting peripheral HSPCs as Mϕ-dependent restriction of the BM HSC pool size directly correlated with reduced peripheral HSPCs. Thus, reducing Mϕ numbers or Mϕ-specific factors will likely improve the efficiency of clinically induced HSPC mobilization, particularly in patients with chronic inflammation or infection. Together, our data define a novel role for BM resident Mϕs as a negative regulator of the HSC pool size at steady state and under infectious conditions, and highlight a key role for IFNγ in increasing and maintaining the population of BM resident Mϕs.
Supplemental Figure 1. Expression of resident Mϕ-specific surface proteins on BM myeloid cells and HSC cell cycle and function after Mϕ depletion. (A) Expression of surface markers VCAM-1 (top panel) and CD169 (bottom panel) on CD11b+Mϕs, CD11blo/−Mϕs, monocytes (CD11b+Ly6Chigh), and neutrophils (CD11b+F4/80−Ly6G+) in the BM. (B) The percent and number of monocytes and neutrophils in the BM on day 2 post PBS- (white bar) or clodronate-liposome (black bar) administration in mice. (C) The percent and absolute number of BM HSCs in G0 (white bar), G1 (gray bar), and S-M-G2 (black bar) of the cell cycle 2 days post PBS- or clodronate-liposome treatment is shown. (D) Lethally irradiated mice were competitively transplanted with equal amounts (2.5 × 106) of whole BM from mice treated with PBS- (GFP−;CD45.2; white bar) or clodronate-liposomes (GFP+;CD45.2; black bar) and the percent of donor-derived WBCs is shown at 4, 8, 12, and 16 weeks post transplantation. (E) Graph represents the mean fluorescent intensity of VLA-4 on the surface of BM HSCs on day 2-post PBS- or clodronate-liposome treatment in mice. The mean±SEM is shown. Two-tailed student’s t-test was used to compare between control and treatment groups. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.
Supplemental Figure 2. Mobilization of HSPCs in infected mice does not correlate with bacterial burden. DNA was isolated from 1 × 106 spleen cells using DNAzol and quantitative real-time PCR was used to detect Em copy number with an input of 50ng DNA using forward and reverse primers for the dsb gene. Em bacterial burden (dsb gene copy number; y-axis) plotted against blood HSPCs (x-axis) for each individual WT (left panel) or IFNγR−/− (right panel) mouse.
Supplemental Figure 3. Infection impairs HSPC mobilization after administration of AMD3100 or G-CSF. Em-infected C57BL/6 mice and uninfected controls were subcutaneously injected with PBS, G-CSF (250µg/kg) starting on day 4 post-infection, or AMD3100 (5mg/kg) (kind gift of Dr. Michael Jordan) starting on day 7 post-infection (dpi). Mice were sacrificed on day 8 post-infection. (A) Graphs represent the number of LK+ cells in the blood and (B) BM in control (white bars) and Em-infected (black bars) mice. The mean±SEM is shown. Two-tailed student’s t-test was used to compare between control and Em-infected groups. *p<0.05.
Supplemental Figure 4. Administration of recombinant IFNγ does not induce HSPC mobilization. PBS or rIFNγ (10µg) was intravenously administered to C57BL/6 mice, and spleens and blood were collected 24hr post-injection. (A) Graph represents the absolute frequency of splenic KLS+ (cKit+Lin−Sca-1+) CD150+ cells is shown. (B) Representative histogram of the mean fluorescent intensity of Sca-1 on LK+ cells in the spleen after PBS (light line) or rIFNγ (dark line) administration. (C) Graphs represent the Sca-1 MFI on splenic LK+ cells. (D) Percent of splenic LK+ cells after PBS or rIFNγ administration. (E) CFUs from the spleen and (F) blood are shown. The mean±SEM is shown. Two-tailed student’s t-test was used to compare between control and treatment groups. *p<0.05, **p<0.01, ***p<0.001.
Supplemental Figure 5. Responsiveness of HSCs to IFNγ in vivo and in vitro. CD45 congenic mice were lethally irradiated and received a total of 5 × 106 BM cells derived from WT (2.5×106 cells) and IFNγR−/− (2.5×106 cells) mice. Peripheral blood from reconstituted mice was screened for chimerism at 4 weeks and mice were infected with Em at 6 weeks post-reconstitution. (A) BM was collected from control and Em-infected mixed WT (CD45.1): IFNγR−/− (CD45.2) BM chimeric mice on day 11 post-infection. Flow plot represents CD45 surface expression on BM LK+CD150+CD48− cells. (B) The percent of WT (white bars) and IFNγR−/− (black bars) HSCs is shown. Data represent one experiment repeated at least two times with 3–4 mice/group. (C) Histograms represent intracellular pSTAT1 staining from sort-purified HSCs (CD150+CD48−), ST-HSCs (CD150+CD48+), MPPs (CD150−CD48+), monocytes (Ly6ChiCD11b+), and Mϕs (F4/80+CD11b+) after PBS (open peak) or rIFNγ (grey filled peak) stimulation. Sort-purified BM cells were incubated with PBS or 200ng rIFNγ for 15 minutes at 37°C. Cells were immediately fixed/permeabalized (BD PhosFlow) and stained with an anti-STAT1 (pY701) antibody. (D) Graph represents the fold change in pSTAT1 of rIFNγ stimulated cells over PBS stimulated cells. Data represents pooled data from three separate sorts.
This work was supported by a National Blood Foundation Research Grant to K.C.M., R01 GM105949 to K.C.M., and R01 HL091769 to M.B.J..
Authorship ContributionsA.M and K.C.M. conception and design, data analysis and interpretation, and manuscript writing. A.M., Y.Z., M.J., V.T. provision of study material, collection of data, data analysis and interpretation. K.C.M and M.B.J. financial support. M.B.J. conception and design.
Conflict of Interest Disclosure
The authors declare no conflict of interest.