|Home | About | Journals | Submit | Contact Us | Français|
The role of macrophages in modulating the systemic response to hypoxia and oxidative stress is emerging from basic biological processes, such as the regulation of red blood cell production, and from analysis of tumor progression, as a key factor determining whether cells survive, proliferate or differentiate under micro-environmental pressures. Our recent work identified a novel role for macrophages in promoting expansion of erythroid progenitors in vitro while confirming previous data that macrophages are not required for red cell enucleation. This work emerged from analyses of hypoxia and cell death in the Rb null fetal liver where we demonstrated that defects in erythropoietic islands were due to deterioration in the fetal liver microenvironment that disrupted heterotypic interactions of macrophages with erythroblasts and not to intrinsic defects in Rb null macrophages. The significance of these findings for the effect of hypoxia on macrophage interactions and activity during tumor progression is also discussed.
Analysis of the developmental phenotypes of genetically engineered mice has been an important approach to identifying physiologically relevant mechanisms regulated by tumor suppressor genes and oncogenes.1 Despite the complexity of the phenotype, mice lacking the function of the Rb tumor suppressor have proven to be a particularly informative system for those seeking to reveal the critical functions of pRb, both in embryonic development and in the development of cancer.2 The major challenge has been in determining the extent to which defects in tissue differentiation are cell autonomous or non-cell autonomous, as a consequence of abnormal placental development and micro-environmental effects. The observation that there are cell intrinsic defects in trophoblast proliferation and differentiation leading to placental malfunction, that in turn contributes extensively to the incidence of cell death in multiple tissues in the Rb null embryo was a major breakthrough in understanding the phenotype of Rb null mice.3,4 However, conditional targeting of Rb deletion to specific tissues and lineages now makes it apparent that cell intrinsic defects, such as the inability to exit cell cycle properly (the signature effect of Rb loss) exist in other tissues independent of placental defects.4-6 Perhaps more importantly however in terms of novel insight into the role of pRb as a tumor suppressor, is the emerging realization that Rb deficient cells are intrinsically defective in their response to extrinsic stresses, such as hypoxia and nutrient deprivation induced in the embryo by a malfunctioning placenta. For example, we showed recently that Rb null hepatocytes were sensitized to non-apoptotic cell death in ischemic regions of the developing fetal liver.7,8 Further exploration showed that RB deficient human tumor cell lines or Rb null primary fibroblasts were sensitized to non-apoptotic cell death induced in culture by nutrient deprivation and other stresses,7 indicating that Rb is required in a cell intrinsic manner for stress responses. Thus differentiation defects in the Rb null mouse appear to arise from both a failure to maintain the tissue microenvironment (extrinsic effects) and from defective responses to these extrinsic pressures, including failure to enforce stress-induced cell cycle checkpoints (intrinsic effects).
Over the past few years, the role of pRb in the erythroid system has been of particular interest as it has become apparent that Rb null erythroblasts exhibit defects in cell cycle exit and terminal differentiation that are not explained by defective placental development.4-6 Of note, it was reported that the association of differentiating erythroblasts with macrophages in so-called erythropoietic islands, was defective in Rb null fetal liver and that this contributed to red cell maturation defects.9 It was proposed that pRb promotes macrophage differentiation by opposing the inhibitory effect of Id2, a member of the helix-loop-helix family of transcriptional regulators, on the lineage commitment factor PU1, and that erythroblast differentiation failed in Rb null mice due to the resultant failure of macrophages to promote erythroblast island formation.9 However, PU1 deficient mice produce mature enucleated red cells despite developmental defects leading to the absence of mature macrophages and erythropoietic islands suggesting that de-regulated Id2 activity in Rb null macrophages did not explain defects in red cell enucleation. Furthermore, recent work now demonstrates that erythroblasts can enucleate independent of macrophages,8,10 that deletion of Rb in macrophages is not sufficient to induce erythropoiesis defects,11 that Rb null erythroblasts are defective for enucleation irrespective of macrophage function,4,8 and that Rb null erythropoietic islands are disrupted due to the deteriorating state of the fetal liver micro-environment.8 Interestingly, germline deletion of Id2 compensated for Rb loss in multiple developing mouse tissues, such that there was no cell death in either the nervous system or fetal liver of compound Rb-/-;Id2-/- mice, that now survived till birth.12 Given the non-cell autonomous nature of cell death in the Rb null embryo and that Id2 blocks trophoblast differentiation,13 an alternative explanation for the apparent rescue of Rb null mice to term may be that loss of Id2 promotes normal placental development, thereby restoring normal nutrient and oxygen transport to the developing embryo. This aspect of the phenotype of Rb-/-;Id2-/- mice awaits examination.
Our results showed that defects in macrophage-erythroblast interactions were associated with increased hypoxia and necrotic cell death in the Rb null fetal liver, and furthermore, we demonstrated that the association of macrophages with erythroblasts in vitro in reconstituted erythropoietic islands was specifically disrupted by growth under hypoxic conditions.8 Macrophages exhibit pleiotropic responses to hypoxia including increased migration against a hypoxic gradient, heightened invasive properties (possibly the consequence of altered adhesion properties), induction of anaerobic glycolysis, chemokine receptor expression and increased survival properties.14 Furthermore, anoxia or ischemia and reperfusion associated with sustained or intermittent hypoxia induces necrotic cell death in affected cell types, including cardiomyocytes, neurons and tumors cells during myocardial infarction, cerebral stroke and cancer.15 Macrophages are recruited to such regions of necrosis by pro-inflammatory signals produced by dying cells.16 One such pro-inflammatory molecule is HMG-B1, a chromatin component that is normally tightly bound within the minor groove of genomic DNA in healthy cells and even more tightly bound to chromatin in apoptotic cells, but under conditions of necrosis is released into the cytoplasm and extracellular milieu.17 HMG-B1 interacts with RAGE and Toll-like receptors on the surface of macrophages to promote their migration and activation, including secretion of TNF-alpha and nitric oxide.16 HIF-1α deficient macrophages are defective for inflammatory responses to hypoxia implicating HIF target genes in these processes,18 although whether effects of hypoxia are additive or synergistic with those of necrosis is not clear. Nevertheless, macrophage activation by hypoxia and/or necrosis is key to clearance of cellular debris and adaptive immune responses to tissue injury.19
Increased macrophage infiltration to affected regions of the Rb null fetal liver raised the possibility that previously reported effects of macrophage malfunction in the Rb null fetal liver9 were non-cell autonomous due to deterioration of the fetal liver microenvironment. Our data confirmed this showing that while macrophages exhibited normal cell-cell contacts and morphology in healthy areas of the liver, there were increased numbers of macrophages and reduced cell contacts with erythroblasts in necrotic regions of the liver.8 Furthermore, we showed that Rb null macrophages in dying parts of the liver were actively phagocytosing cellular debris indicative of a normal functional response to pro-inflammatory signals. Finally, we failed to observe reduced macrophage-specific gene expression in the Rb null fetal liver compared to wild-type but detected marked induction of hypoxia-inducible genes, including hexokinase 2, BNip3, Vegf and heme oxygenase.8 In this respect, our data appeared to contradict that observed by others,9 but given that these former studies were carried out by comparing cultured wild-type and Rb null MEFs at undetermined passage number, whereas our work was done with fetal liver harvested directly from the mouse, we postulate that there may be non-physiological reasons for why previous studies observed decreased macrophage-specific gene expression9 and we did not.8
To assess whether hypoxia, such as we observe in the mid-gestational Rb null fetal liver, explained altered erythropoietic island formation in distal regions of the liver, we examined the effect of culture at 1% oxygen on islands formed in vitro, and observed increased release of attached erythroblasts from adherent macrophages compared to growth at 21% oxygen. These results are consistent with the conclusion that erythropoietic island formation becomes defective in Rb null fetal liver as the tissue becomes increasingly hypoxic and the microenvironment deteriorates.
Importantly, we show that defects in enucleation persist in Rb null erythroblasts when plated with or without wild-type or Rb null macrophages, or with conditionally deleted Rb null macrophages.8 These results are consistent with erythroid maturation defects and enucleation in particular being a cell intrinsic defect, (as reported previously in refs. 5 and 6). However, if macrophages are not required for enucleation, what role do macrophages play in erythropoietic islands?
Various functions have been attributed to macrophages in the context of erythroblastic islands including regulation of erythroblast proliferation, viability and differentiation through provision of growth factors, cell-cell contacts and of course, getting rid of extruded red cell nuclei.20 Recently, it has emerged from work showing that erythroid defects in DNase II deficient mice were due to an auto-immune reaction to naked DNA, that although macrophages are required to engulf extruded nuclei, they are not required for the enucleation process itself10 and our work is consistent with these findings.8 Interestingly, we noted that while macrophages associated with erythroblast islands are dispersed throughout the liver, phagocytes that engulf red cell nuclei were predominantly located close to liver sinusoids and this prompted us to ask whether there might be two classes of macrophage involved in red cell maturation: the central macrophage present in erythroblast islands that may act to regulate the balance between proliferation, differentiation and survival in less mature erythroblasts, and peripheral phagocytic macrophages that engulfs nuclei extruded by mature erythroblasts.
Given that macrophages in the context of erythropoietic islands do not appear to be required for red cell enucleation,8,10 speculation arises as to what role central macrophages may play in erythroid proliferation and differentiation. While macrophage coculture did not promote red cell enucleation, we observed that it increased retention of erythroid progenitors in vitro. When cKit+CD71+TER119- erythroid progenitors from E13.5 fetal liver are plated in methylcellulose media, they undergo two to five cell divisions before exiting cell cycle and terminally differentiating (including enucleation).6 The small globinized colonies that arise after two days for mouse and after give days for human are known as “colony-forming unit erythroid” or CFU-E.6 We observed that fetal liver erythroblasts plated in liquid culture in the absence of macrophages lost the potential to form CFU-E colonies when plated 48 hours later in semi-solid media whereas those erythroblasts that were plated in association with macrophages retained the ability to form CFU-E colonies.8 These results suggest that macrophages are promoting the expansion of CFU-E progenitors, possibly by supporting increased proliferation, reduced cell death or by preventing their differentiation.
What selective advantage to the organism is there in the regulation of macrophage-erythroblast interactions by hypoxia? Regulation of erythropoietin expression by the kidney in response to changes in blood oxygen tension plays a major role in modulating the systemic production and survival of red blood cells. We are suggesting that in addition to regulating erythropoiesis rates through erythropoietin synthesis, hypoxia also promotes stress erythropoiesis by inducing release of immature erythroblasts from associated macrophages in erythropoietic islands in the bone marrow.8
Although, these two effects of hypoxia on erythropoiesis appear to be counter-productive, we have proposed a model (Fig. 1) in which release of erythroblasts by macrophages in the bone marrow promotes increased migration of immature erythroblasts to the spleen. Under conditions of stress, including hemolytic anemia and following adoptive transplant, the spleen becomes the major site of erythropoiesis and provides an environment that allows for rapid erythroid expansion and differentiation.21 Interestingly, we have noted that erythropoietic islands are less apparent in the spleen than in the bone marrow, under either normal or stressed conditions suggesting that the ability of the spleen to promote erythropoiesis in anemic mice may be linked to the absence of restraining macrophage interactions. Thus, by promoting release of erythroid progenitors from the bone marrow, we propose that hypoxia is increasing the rate of systemic red cell production at a time when oxygen transport to peripheral tissues needs to be maximized. Increased serum levels of erythropoietin in response to hypoxia likely plays a critical role in promoting survival of circulating erythroblasts and further promotes their expansion in the spleen, as well as the bone marrow. Thus, our model (Fig. 1) points to a critical role for macrophages in erythroid homeostasis by maintaining a pool of immature erythroblasts in the bone marrow ready for mobilization under stress conditions.
Finally, our model predicts that there are adhesion molecules on the surface of the red cell and macrophage, whose expression and/or interaction is sensitive to oxygen tension. A number of molecules have been identified as mediating erythroblast-macrophage adhesion, including α5-integrin and ICAM-4,22 or α4-integrin and VCAM-1,20 and it will be interesting to determine how their levels or interactions are modulated in response to hypoxia.
Hypoxia develops in growing tumors as they outstrip their ability to take up oxygen and nutrients from their environment by diffusion.23,24 Increased tumor hypoxia and necrosis is predictive of a poor therapeutic response25 and consequently understanding the effects of hypoxia on tumor progression is important to developing better treatment outcomes.26 The effects of hypoxia are largely mediated by the HIF transcription factors that induce expression of genes that promote glycolysis, adhesion, angiogenesis and survival.23,24 HIF also modulates cellular proliferation and promotes genome instability through interactions with the c-Myc oncoprotein at the promoters of genes such as p21Waf-1 and the mismatch repair genes MSH2 and MSH6.27 There is also growing significance attached to the role of tumor hypoxia in promoting tumor metastasis, although the underlying mechanisms are still incompletely understood. Lysyl oxidase (LOX) is a HIF target gene that was recently reported to play an important role in promoting breast cancer metastasis in response to hypoxia, possibly by establishing the pre-metastatic niche.28 The role of other HIF targets in tumor metastasis is not known. However, given the role of tumor-associated macrophages in promoting tumor progression and metastasis,29 it is reasonable to suggest that hypoxia may also promote metastasis by recruiting macrophages to the necrotic regions of tumors.
Tumor-associated macrophages (TAMs) go on to promote tumor progression through secretion of growth factors (such as EGF that also functions as a chemotactic agent for tumor cells), angiogenic factors (such as VEGF) and matrix degrading proteases (such as MMP9).14,16,29 TAMs are also proposed to play a role in suppressing adaptive immune responses to the tumor, through expression of arginase and nitric oxide synthase that induce T cell dysfunction.16 Intriguingly, more than 70% of human breast cancers over-express the macrophage growth factor CSF-1 and its up-regulation is particularly prevalent in invasive carcinomas.30 Importantly, some studies have suggested TAMs are essential for breast tumor metastasis.29 CSF-1 produced by tumor cells acts in a paracrine manner as a chemo-attractant for TAMs, and to prevent anti-tumor activity associated with macrophages, for example by inhibiting their differentiation to mature tumor-antigen presenting dendritic cells.30 Thus, the inflammatory response to necrosis may be key to explaining the role of hypoxia in promoting the malignant progression of certain types of cancer.19
While hypoxia and necrosis appear influential in recruiting macrophages to tumors, and macrophages secrete factors that influence tumor cell growth, the question arises whether hypoxia also regulates macrophage adhesion to tumor cells and how this may influence tumor cell dissemination. Tumor cells can be visualized in association with perivascular macrophages as they intravasate,31 indicating that macrophages are directly involved in the metastatic process by acting as both the vehicle and by generating the portal through which tumor cells escape into the circulation, as depicted in Figure 2. Macrophages interact with tumor cells and the tumor vasculature through heterotypic interactions mediated by ICAM-1 and other cell adhesion molecules. Given our data showing the sensitivity of erythroblast/macrophage interactions to hypoxia, it is interesting to speculate on whether the interaction of tumor cells with macrophages is also regulated by hypoxia (Fig. 2). We show that hypoxia disrupts macrophage association with erythroblasts and if this also influences tumor cell interactions with macrophages, it would argue that macrophages preferentially adhere to tumor cells that are close to the periphery of the tumor and/or close to associated blood vessels. This is indeed what has been observed with a specific role attributed to perivascular macrophages in promoting tumor cell intravasation.31
The role of macrophages in regulating the microenvironment of primary cells or tumor cells is emerging as a key factor in determining responses to stresses, such as hypoxia and nutrient deprivation. Not only do macrophages get rid of cellular debris by phagocytosis but through heterotypic interactions appear to instruct the cellular response of resident cells, in terms of their proliferation, adhesion and survival properties, to such stresses, The societal and stress-responsive nature of these heterotypic interactions is clearly important to both normal development and tumor development, and their investigation will undoubtedly provide continuing insight to the basic biological consequences and therapeutic potential of disrupting such interactions.