MafB/c-Maf Deficiency Enables Self-Renewal of Differentiated Functional Macrophages

See allHide authors and affiliations

Science  06 Nov 2009:
Vol. 326, Issue 5954, pp. 867-871
DOI: 10.1126/science.1176056


In metazoan organisms, terminal differentiation is generally tightly linked to cell cycle exit, whereas the undifferentiated state of pluripotent stem cells is associated with unlimited self-renewal. Here, we report that combined deficiency for the transcription factors MafB and c-Maf enables extended expansion of mature monocytes and macrophages in culture without loss of differentiated phenotype and function. Upon transplantation, the expanded cells are nontumorigenic and contribute to functional macrophage populations in vivo. Small hairpin RNA inactivation shows that continuous proliferation of MafB/c-Maf deficient macrophages requires concomitant up-regulation of two pluripotent stem cell–inducing factors, KLF4 and c-Myc. Our results indicate that MafB/c-MafB deficiency renders self-renewal compatible with terminal differentiation. It thus appears possible to amplify functional differentiated cells without malignant transformation or stem cell intermediates.

The nonproliferative state of terminally differentiated cells is assured by robust, often redundant mechanisms (1, 2), and in rare exceptions where fully mature cells can re-enter the cycle, proliferation remains transient and/or involves de-differentiation (3). It remains unknown what renders differentiated cells refractory to the same mitogen signals that stimulate the proliferation of their direct precursors. For example, the proliferative response of myelomonocytic progenitors to macrophage colony-stimulating factor (M-CSF) is lost upon differentiation to macrophages (4), despite the continued ability of these mature cells to sense the cytokine (5). Consequently, myeloid progenitor cells form colonies in M-CSF containing semisolid medium, whereas blood monocytes and tissue macrophages do not. Here, we have investigated whether this process involves the transcription factors MafB and c-Maf, which can both regulate M-CSF responsiveness (6, 7) and stimulate monocytic differentiation (810).

Intriguingly, we observed that in contrast to wild-type (WT) cells, MafB/c-Maf double deficient (Maf-DKO) blood leukocytes formed colonies in M-CSF containing medium at high efficiency (Fig. 1A). Several lines of evidence indicated that these colonies were initiated by mature monocytes rather than by other mature cell types or circulating progenitors. First, in a cytokine mix that can reveal rare circulating stem and progenitor cells (11), Maf-DKO leukocytes gave rise to the same low number of colonies as WT cells (fig. S1A). Furthermore, M-CSF colonies did not develop from lymphocytes or granulocytes (fig.S1C) but formed at very high frequency from purified mature Maf-DKO monocytes that expressed Mac-1, F4/80, and CD115 but were negative for c-kit (CD117), a marker of both primitive (11) and M-CSF–responsive macrophage/dendritic cell progenitors (12) (Fig. 1B and fig. S2). A high rate of colony formation was also observed for spleen and peritoneal macrophages (fig. S1B) as well as for purified CD117- Kupffer cells of the liver, which represent terminally differentiated tissue macrophages (Fig. 1B). These results indicated that, in contrast to WT, mature Maf-DKO blood monocytes and tissue macrophages could proliferate in response to M-CSF.

Fig. 1

Extended proliferation of mature Maf-DKO monocytes and macrophages. (A) Colony assays of WT or Maf-DKO blood leukocytes after 12 days in M-CSF methocult medium, showing culture dishes, a typical Maf-DKO colony (scale bar, 50 μm), and colony-forming unit (CFU) numbers. (B) Colony assay as in (A) from flow cytometry–sorted Mac-1+, F4/80+, CD117 blood monocytes (confirmed to be CD115+ after sorting) and F4/80high, Mac-1int, CD117 liver Kupffer cells. Upstream gating schemes are shown in fig. S2 (C) Flow cytometric analysis of BrdU incorporation and DNA content of monocytes [gated as in (B)] 4 hours after intravenous (IV) injection of 5 μg M-CSF and 1-hour BrdU labeling, showing representative profiles and the ratio of cycling to noncycling monocytes (WT n = 3, Maf-DKO n = 2). (D) Serial replating in methocult assay of monocyte-derived Maf-DKO cells washed out from M-CSF colony assays. (E) Growth curves of individual clones in liquid M-CSF culture. All colony assays show average CFU of replicate plates, all error bars indicate SEM, and (A), (B), and (D) are representative of at least two independent experiments.

Indeed, when we injected recombinant M-CSF directly into the circulation of mice, we observed that Maf-DKO, but not WT, blood monocytes had entered the cell cycle (Fig. 1C). The same observation was made when cells were stimulated with M-CSF ex vivo (fig. S1D). Although Maf-DKO monocytes or in vitro differentiated macrophages required significantly higher M-CSF concentrations for proliferation than bone marrow progenitors (fig. S3), their ability to divide was not restricted to a small number of cycles but continued in extended long-term culture. Maf-DKO monocyte–derived colonies could thus be serially replated in methocult assays at high efficiency and without loss of clonogenicity (Fig. 1D). This was intriguing, because even progenitors normally have only limited replating ability in this assay that is often employed to reveal the extended self-renewal capacity of transformed progenitors (13). Moreover, extended, possibly unlimited, expansion of Maf-DKO cells could be achieved in liquid culture. We have maintained Maf-DKO cells in continuous culture for more than 8 months without any signs of crisis. For three independent Maf-DKO populations, cell counts over 2 months revealed stable doubling times of 1.44 ± 0.05 days and theoretical amplification factors of 1011 to 1012 (fig. S4A). This was unlikely to be due to the outgrowth or selection of a small subpopulation; more than 80% of Maf-DKO cells gave rise to new colonies in replating assays (Fig. 1D), and individual colonies could be cloned and subcloned at 80 to 90% efficiency (fig. S4B) or undergo expansion in liquid culture with similar, unaltered growth curves (Fig. 1E). Together, the high cloning and recloning efficiency and the similar growth rates of individual clones indicate that a vast majority, possibly all, rather than a subpopulation of Maf-DKO cells have an extended proliferation capacity.

In specialized regenerative processes, differentiated cells can also reenter the cell cycle, but in these examples cells typically undergo de-differentiation and revert to an immature phenotype (3). By contrast, proliferating Maf-DKO cells maintained a differentiated phenotype and function. Like WT monocytes in M-CSF culture, Maf-DKO cells remained positive for the monocyte/macrophage surface markers FcγRII/III (CD16/32), Mac-1 (CD11b), F4/80, and CD115 (Fig. 2A); displayed a normal macrophage morphology (Fig. 2B); and were negative for progenitor markers CD117 and CD34 (Fig. 2A) or other lineage markers (fig. S5A), even after long-term expansion. Proliferating Maf-DKO cells also displayed a global gene expression profile highly similar to WT cells (Fig. 2C) and expressed a panel of characteristic monocyte/macrophage genes (Fig. 2D and fig. S5B). They also showed the same capacity as WT monocyte–derived macrophages to produce nitric oxide in response to lipopolysaccharide and interferon-γ (Fig. 2E) and to phagocytose latex beads (Fig. 2F). Cell cycle analysis further demonstrated that Maf-DKO cells in S and G2/M phase had the same amount of phagocytic activity as cells in G0/G1 phase, indicating that differentiated macrophage function is fully maintained through cell cycle progression in these cells (Fig. 2G). This was further confirmed with live bacteria, revealing that cycling (Ki67+) Maf-DKO cells could phagocytose large numbers of green fluorescent protein (GFP)–expressing Salmonella typhimurium (Fig. 2H). Interestingly, M-CSF–expanded monocyte-derived Maf-DKO cells retained the ability to acquire dendritic cell features when shifted to GM-CSF–containing medium (fig. S6A and B).

Fig. 2

Proliferating Maf-DKO cells have mature macrophage phenotype and function. (A) Flow cytometric analysis for monocyte/macrophage or progenitor markers and (B) Giemsa staining of 2-month monocyte-derived Maf-DKO macrophage cultures. (C) Comparison of global gene expression between WT and Maf-DKO bone marrow–derived macrophages by whole-genome microarray analysis, showing a scatter plot with a two-fold change corridor and the Pearson correlation coefficient (CC). (D) Semiquantitative reverse transcription polymerase chain reaction (RT-PCR) of characteristic monocyte/macrophage genes, (E) nitric oxide (NO) production after LPS/IFNγ stimulation, and (F) flow cytometric analysis of phagocytosed phycoerythrin (PE)–latex beads from monocyte-derived WT and Maf-DKO macrophages. Error bars indicate SEM. (G) Flow cytometric analysis of fluorescent bead phagocytosis by 2-month monocyte-derived Maf-DKO macrophage cultures in G1 or S/G2/M phase of the cell cycle. (H) Immunofluorescent staining for Ki67 (red) and nuclear 4′,6′-diamidino-2-phenylindole (blue) labeling of monocyte-derived Maf-DKO macrophages 1 hour after phagocytosis of GFP-expressing S. typhimurium (green). Cells outlined in white. Scale bar, 10μm. (A and B) and (D) to (H) are representative of at least two independent experiments.

To determine whether the extended proliferative capacity of Maf-DKO monocytes and macrophages was associated with tumorigenic transformation, we analyzed the long-term effects of MafB/c-Maf deficiency in vivo. Interestingly, bone marrow chimeras with a Maf-DKO hematopoietic system showed no sign of leukemia or myelo-proliferative disease for more than 1 year after reconstitution (fig. S7) or after hematopoietic stress induced by pharmacological hemato-suppression (fig. S8). Furthermore, monocyte-derived Maf-DKO macrophage cultures retained a normal number of chromosomes through long-term ex vivo expansion (Fig. 3A) and did not give rise to tumors upon transplantation into syngeneic or immunocompromised nude mice, irrespective of the injection route (Fig. 3, B and C), despite the cells’ ability to divide in vivo (fig. S9A). By comparison, under the same conditions the murine macrophage cell line J774.1 induced massive tumors within days and caused 100% mortality by 4 weeks (Fig. 3, B and C).

Fig. 3

Expanded Maf-DKO cells are not tumorigenic, but rather integrate as functional macrophages into host tissues in vivo. (A) Karyotype of long-term expanded monocyte-derived Maf-DKO macrophages. (B) Tumor-free survival of animals upon intravenous (IV), intraperitoneal (IP), or subcutaneous (SC) injection of 1 × 106 J774.1 or monocyte-derived Maf-DKO macrophages into syngeneic hosts. (C) Tumor development at 4 weeks and survival curve post-SC injection of 1 × 106 J774.1 or monocyte-derived Maf-DKO macrophages into nude mice (n = 3). Arrowheads indicate injection site. (D) Detection of transplanted carboxyfluorescein diacetate succinimidyl ester (CFSE)–labeled Ly5.2+ Maf-DKO macrophages by flow cytometric analysis of peritoneal exudate 6 days after IP injection and (E) by confocal immunofluorescence analysis of macrophage antigens on CFSE+ cells in spleen 3 days after IV injection. Scale bars, 50μm (top), 20μm (bottom). Immunofluorescence analysis of CFSE-labeled (F) and 5-(and-6)-(((4-chloromethyl)benzoyl)amino)tetramethylrhodamine (CMTMR)–labeled (G) Maf-DKO macrophages for F4/80 [(F) and (G)] and iNOS expression (G) 3 days after transplantation and 48 hours after IV inoculation with the red fluorescent protein dsRed-expressing S. typhimurium. Scale bar, 20 μm. (D) to (G) are representative of at least two independent experiments.

Rather than forming tumors, transplanted Maf-DKO cells showed homing and functional integration into normal macrophage populations of multiple tissues. Maf-DKO cells contributed to macrophages of the bone marrow, peritoneum, and red pulp and marginal zone of the spleen, and to Kupffer cells of the liver (Fig. 3, D and E, and fig. S9, B and C). Moreover, Maf-DKO macrophages in spleen and liver participated in the characteristic host response to infection with live S. typhimurium, which causes an invasive disease in mice similar to human typhoid fever (14). The transplanted cells phagocytosed bacteria (Fig. 3F), localized to typical discrete macrophage foci in infected organs (14), and activated expression of inducible nitric oxide synthase (iNOS) (Fig. 3G and fig. S10). Furthermore, some transplanted cells acquired CD11c expression in infected mice (fig. S6C). Together, these results indicate that expanded Maf-DKO cells are not transformed but give rise to functional macrophages and possibly monocyte-derived dendritic cells that integrate into the normal tissue architecture in vivo.

The p19ARF/p53 and p16/Rb cell cycle inhibitory pathways represent important control mechanisms of cellular proliferation, and their inactivation can extend the limited division number of mitotic progenitors in culture (15). However, neither mature leukocytes from p53−/− nor those from p19ARF/p16 (INK4a) double-deficient mice gave rise to M-CSF colonies (fig.S11), indicating that the observed self-renewal of differentiated Maf-DKO monocytes and macrophages must be based on distinct mechanisms.

Differentiated cells can be reprogrammed into pluripotent stem cells (iPS) by the four transcription factors Oct-4, Sox-2, KLF4, and c-Myc (16, 17), of which the latter two have been proposed to impart extended proliferation capacity (16) on the basis of their role in embryonic stem cell self-renewal (18, 19). In addition, KLF4 and c-Myc can also mediate monocytic differentiation (20) and proliferation (21, 22), respectively. We therefore investigated the role of these factors in the extended proliferative capacity of Maf-DKO monocytes and macrophages. Whereas expanded Maf-DKO macrophage cultures did not express the pluripotency-associated factors Sox-2, Oct-4, and nanog (fig. S12A), they showed a strong up-regulation of both KLF4 and c-Myc expression within 2 hours of M-CSF stimulation and maintained substantially higher expression levels than WT controls for the observation period of 72 hours (Fig. 4A). To determine the functional consequence of these changes, we generated small hairpin RNA (shRNA) retroviral vectors directed against KLF4 or c-Myc that could specifically reduce both endogenous and transfected target gene expression at the RNA and protein level (Fig. 4, B and C). When monocyte-derived Maf-DKO macrophages were infected with GFP-expressing retrovirus coding for no or control shRNA sequences, they gave rise to GFP+ colonies in methocult assays of the same size and morphology as uninfected cells. By contrast, cells infected with GFP retrovirus expressing either KLF4 or c-Myc shRNA gave rise to only small GFP+ cell clusters of less than 20 cells (Fig. 4D) that could not be propagated through serial replating (Fig. 4E). Internal controls of noninfected GFP colonies from the same plating showed identical morphology, frequency, and replating behavior under all conditions (Fig. 4D). Conversely, combined ectopic overexpression of c-Myc and KLF4 conferred serial replating capacity to in vitro differentiated WT macrophages (fig. S12B). Together, these results indicated that increased expression of both KLF4 and c-Myc are required and potentially sufficient for the extended self-renewal capacity of Maf-DKO macrophages.

Fig. 4

Elevated Klf4 and c-Myc expression are both required for extended proliferation of Maf-DKO macrophages. (A) Quantitative RT-PCR analysis for Klf4 and c-Myc after M-CSF stimulation of cytokine-starved WT and Maf-DKO macrophages or (B) in NIH3T3 cells expressing empty vector (EV), control (C), Klf4 (K1) or c-Myc (M1 and M2) shRNA, respectively. (C) KLF4 and c-Myc immunoblot of QT6 cells cotransfected with empty (–), Klf4, or c-Myc expression vectors and control, Klf4 (K1), or c-Myc (M1 and M2) shRNA, as indicated. (D) Representative images of infected (GFP+) and uninfected (GFP) colonies from the same M-CSF colony assays of monocyte-derived Maf-DKO macrophages transduced with the indicated shRNA-GFP retroviruses. (E) CFU counts for total (GFP+ and GFP) and GFP+ colonies upon serial replating of the same assays. All error bars indicate SEM and all panels are representative of at least two independent experiments.

Together, our results indicate that MafB/c-Maf deficiency dissociates cell cycle exit from terminal macrophage differentiation. Although specialized mononuclear phagocytes such as microglia and some dendritic cells can undergo transient proliferation in peripheral tissues (2325), their number of divisions is limited and it remains unresolved whether proliferation occurs in precursors or fully differentiated cells. Here, we show that MafB/c-Maf deficiency enables long-term expansion of differentiated, mature macrophages without loss of function or sensitivity to homeostatic control in vivo. Interestingly, this depends on the regulated activation of c-Myc and KLF4. Mechanistically, this may involve the loss of MafB/c-Maf mediated repression of Ets-1/2 (26) and PU.1 (7) (fig. S13), two myeloid activators of the c-Myc (4, 27) and KLF4 (20) promoters, respectively. KLF and c-Myc are two iPS-inducing factors that have been proposed to mediate self-renewal (16, 18, 19) but are not required for pluripotency (17). Thus, it appears that pluripotency and self-renewal are two stem cell characteristics that can be mechanistically dissociated. The nontumorigenicity of Maf-DKO macrophages is intriguing, given that individually both c-Myc and KLF4 can act as oncogenes (28, 29). In particular c-Myc can malignantly transform macrophages (21, 22) (fig. S12B) and induce tumors in iPS-derived mice (17). The controlled and joint up-regulation of c-Myc and KLF4 in Maf-DKO cells, however, may prevent malignancy (fig. S12B) by assuring a fine-tuned counterbalance of the factors’ partially antagonistic activities in cell cycle control (16, 28). Together, our results provide a first example for extended amplification of functional differentiated cells without passing through pluripotent or multipotent stem cell intermediates and may open up new perspectives for monocyte-based cellular therapies in infectious disease or tissue regeneration.

Supporting Online Material

Materials and Methods

Figs. S1 to S13

Tables S1 and S2


  • * These authors contributed equally to this work.

References and Notes

  1. We thank L. Chasson for immunofluorescence staining; E. Bertosio for leukocyte counts; C. Otto for communication of preliminary results; T. Benoukraf for microarray analysis; P. Grenot, M. Barad, N. Brun, and A. Zouine for cell sorting; L. Glimcher, A.-M. Schmitt-Verhulst, and F. Maina for c-Maf+/−, Ink4a−/−, and p53−/− mice; and J.-P. Gorvel and S. Meresse for FP-expressing S. typhimurium. Histological analyses were performed in the mouse functional genomics platform Réunion Interorganisme/Marseille-Nice Genopole with support from the Institut National du Cancer. Miame (minimum information about a microarray experiment)–compliant gene expression data have been deposited at the European Bioinformatics Institute ArrayExpress database under accession no. E-MEXP-2419. A.A. received fellowships from the Ministère d’Enseignement Supérieur et de la Recherche and the Société Française de la Hématologie (SFH), E.S. from the Leukemia and Lymphoma Society of America (5204-06), and S.S. from the Fondation de France (FdF) and SFH. We gratefully acknowledge grants from FdF (2004004150), the Association for International Cancer Research (05-0079), the Association pour la Recherche sur le Cancer (3857, 3422), and the Agence Biomedecine. M.S. directs an Equipe FRM (Fondation pour la Recherche Médicale, Deq 20071210559).
View Abstract

Navigate This Article