Report

The Transcription Factor Gata6 Links Tissue Macrophage Phenotype and Proliferative Renewal

See allHide authors and affiliations

Science  09 May 2014:
Vol. 344, Issue 6184, pp. 645-648
DOI: 10.1126/science.1251414

Gata6 Controls Peritoneal Macs

Macrophages seed tissues throughout the body and are shaped both phenotypically and functionally by the microenvironment they inhabit. Despite such heterogeneity, most tissue macrophages self-renew by local proliferation. How this is regulated, however, is unclear. Rosas et al. (p. 645, published online 24 April) used gene expression analysis to show that the transcription factor Gata6 is specifically expressed in peritoneal macrophages. Gata6 was critical for maintaining the transcriptional signature of peritoneal macrophages and for their proliferative renewal during homeostasis and under inflammatory conditions.

Abstract

Tissue-resident macrophages are heterogeneous as a consequence of anatomical niche–specific functions. Many populations self-renew independently of bone marrow in the adult, but the molecular mechanisms of this are poorly understood. We determined a transcriptional profile for the major self-renewing population of peritoneal macrophages in mice. These cells specifically expressed the transcription factor Gata6. Selective deficiency of Gata6 in myeloid cells caused substantial alterations in the transcriptome of peritoneal macrophages. Gata6 deficiency also resulted in dysregulated peritoneal macrophage proliferative renewal during homeostasis and in response to inflammation, which was associated with delays in the resolution of inflammation. Our investigations reveal that the tissue macrophage phenotype is under discrete tissue-selective transcriptional control and that this is fundamentally linked to the regulation of their proliferation renewal.

Tissue-resident macrophages play fundamental roles specific to their microanatomical niche, ranging from dedicated homeostatic functions to immune surveillance (1). Such heterogeneity predicts that discrete transcriptional controls probably exist in specific macrophage populations that determine both their particular phenotypes and tissue-specific functions.

Many resident macrophages self-renew by local proliferation [(1) and references therein]. This is initiated after seeding of macrophages into tissues during development and their expansion during the neonatal period (1). Under specific conditions, these tissue-resident macrophages may also be derived from blood monocytes (1). Classic F4/80highCD11bhigh peritoneal-resident macrophages fit this model (27), and they proliferate above homeostatic levels in response to inflammation (6). Proliferation of human macrophages has also been observed in several contexts [reviewed in (1)]. However, the factors controlling these processes remain ill-defined. We hypothesized that discrete transcriptional controls would govern both the specific phenotype of tissue macrophages and their proliferative renewal in a select tissue microenvironment.

We performed a transcriptional analysis of murine monocyte-like cells during acute peritonitis (figs. S1 to S3 and tables S1 to S3). Our approach analyzed populations specifically enriched in tissue-resident macrophages (6), allowing definition of a tissue macrophage–restricted transcriptional profile, which was associated with homeostatic and metabolic processes (cluster 15; fig. S2 and table S3, A to C). Gata6 was selectively expressed in peritoneal macrophages when compared to both in vitro–generated bone marrow–derived macrophages and, when isolated during inflammation, contemporary monocyte-derived (8) macrophages (fig. S1, D, F, and G). Perhaps best known for its essential requirement in the development of the heart, gut, and liver (911), the role of Gata6 in macrophages is unknown.

We crossed conditional knockout (KO) Gata6tm2.1Sad/J mice (12) with Lysozyme M (Lyz2) Cre-recombinase knockin mice (Lyz2Cre, official strain name B6.129P2-Lyz2tm1(cre)Ifo/J) (13) to generate mice with a myeloid deficiency of Gata6 (Gata6-KOmye mice) (14). Lyz2Cre mediates recombination in approximately 95% of peritoneal macrophages (13). Flow cytometric analysis of peritoneal cells from Gata6-KOmye mice compared to their wild-type (WT) littermates indicated a gross change in the characteristic F4/80highCD11bhigh phenotype, with the majority (~95%) of classic peritoneal macrophages exhibiting reduced F4/80 and CD11b expression (Fig. 1A). Further analysis of peritoneal myeloid cells (CD11b+CD19), indicated that although the F4/80low macrophages exhibited relatively normal expression of Tim4 [a marker expressed by the majority of peritoneal resident macrophages and found in this study to be selectively expressed by these cells during acute peritonitis; fig. S1 and (6)], there was a reduction in their numbers and an increase in eosinophils and MHCIIhighF4/80low macrophages/dendritic cells (Fig. 1, B and C). However, there were no substantive alterations in the numbers of peritoneal lymphocytes (fig. S4A) or peripheral blood cells (fig. S4, B and C).

Fig. 1 Selective myeloid cell alterations in the peritoneum of mice with myeloid Gata6 deficiency.

(A) Representative flow cytometric and immunofluorescent assessment of peritoneal-resident macrophages from WT and Gata6-KOmye mice. F4/80high (arrowhead) and F4/80low (arrows) macrophages are indicated. Fluorescent images were captured with a ×40 objective lens, the scale bar is indicated, and the images are representative of four mice per group (fig. S9). (B) Representative flow cytometric analysis of the peritoneal myeloid cell (CD11b+CD19) composition of the Gata6-WT and Gata6-KOmye mice. Percentages indicate typical proportions of the cell types of all peritoneal cells. (C) Quantification of peritoneal myeloid cells in the Gata6-WT (black bars, n = 9♂/7♀) and Gata6-KOmye mice (white bars, n = 5♂/5♀) analyzed by flow cytometry in (A) and (B) above. Data represent the mean ± SEM of mice pooled from two independent experiments and were analyzed by two-way analysis of variance (ANOVA) MØ, macrophage; Res, tissue-resident; Eos, eosinophil; DC, dendritic cell; Int, interaction statistic; Gata6, Gata6 effects; Sex, sex effects.

We established a panel of lentiviral vectors (fig. S5A and table S4) with which we achieved selective high expression of transgenes in peritoneal-resident macrophages in vivo (fig. S5, A and B). Lentiviral delivery of Cre to the peritoneal-resident macrophages of adult Gata6tm2.1Sad/J mice resulted in alteration of their phenotype, including lower F4/80 expression (fig. S5C). This confirmed that Gata6 was important for phenotype maintenance in the adult. We also excluded a role for Cre toxicity (15) (fig. S5, C to E).

We assessed the importance of Gata6 as a regulator of the characteristic peritoneal macrophage phenotype by microarray analysis of macrophages from WT and Gata6-KOmye mice (14) (Fig. 2A and table S5, A and B). Analysis of peritoneal macrophage–specific transcripts indicated that there was a significant overrepresentation of probe sets that were down-regulated in the absence of Gata6 (Fig. 2B). The array data were validated by examination of surface receptors whose mRNA was altered (Fig. 2, C to E). An additional study (16) identified genes specific to peritoneal macrophages when compared to other tissues, and there was a similar overrepresentation of genes from this list that were down-regulated in the absence of Gata6 (fig. S6 and table S6). Using both data sets, in addition to Gata6, we identified a gene list that could be considered peritoneal macrophage–specific by both criteria (within and between tissues), of which 60% of genes were down-regulated in the absence of Gata6 (fig. S6). This confirmed Gata6 as a major regulator of the peritoneal macrophage phenotype. Peritoneal macrophage–selective transcripts were not the only transcripts that were altered in the absence of Gata6, however, indicating a broader impact on phenotype (Fig. 2, A and B). Consistent with a role in peritoneal phenotype specialization, enforced Gata6 expression in bone marrow–derived macrophages promoted their peritoneal retention and altered their phenotype toward that of peritoneal resident macrophages (fig. S7). Although we have not addressed this, we would anticipate that Gata6 would also be up-regulated in bone marrow–derived cells recruited to the peritoneum when replacing the tissue-resident pool; for example, as can occur in irradiation chimeras (17). Peritoneal macrophages in WT mice are capable of renewal without monocytic input (6, 7), particularly under homeostatic conditions. In accordance with this, genes associated with the regulation of cell proliferation (Gene Ontology term GO:0042127) were also altered by Gata6 deficiency (table S5A), including Cdkn2b, Csf1, Igf1, Tgfb2, Tgfbr2, and Bmpr1a. Moreover, Gata6-deficient peritoneal macrophages exhibited increased basal proliferation as compared to macrophages in WT mice (Fig. 3, A and B). As with WT cells (8), the proliferation of Gata6-deficient peritoneal macrophages in vivo is dependent on the cytokine macrophage colony-stimulating factor (M-CSF). We observed marked polyploidy in Gata6-deficient cells (Fig. 3, A and C, and fig. S8). Polyploid Gata6-deficient cells were multinucleate, and this could reflect failed cytokinesis (18) or the creation of a fusogenic phenotype by the marked alteration in membrane-associated molecules (an enrichment of Gene Ontology term GO:0005886, Benjamini P = 0.000007). We took advantage of the existence of F4/80high peritoneal macrophages in the mice (Fig. 1A), which we confirmed had escaped Cre-mediated Gata6 deletion and were phenotypically normal (fig. S9). Within the Gata6-KOmye mice, we observed a significantly lower level of proliferation of F4/80high WT cells as compared to F4/80low KO macrophages (Fig. 3D), which was comparable to that observed in WT mice. Similar results were obtained by lentiviral-mediated Cre delivery into peritoneal macrophages of adult conditional-KO mice (Fig. 3E). These studies demonstrated a cell-intrinsic role for Gata6 in limiting basal proliferation. Although the mechanisms controlling this response are unclear, it is likely that Gata6 influences proliferation through both direct and indirect impacts on the cellular phenotype of peritoneal macrophages. Given the similarities between tissue-resident peritoneal and pleural macrophages, we examined Gata6 expression and found it comparable at both sites (Fig. 3F). Similar to peritoneal macrophages, the pleural macrophages of Gata6-KOmye mice were predominantly F4/80low and exhibited a cell-intrinsic increase in proliferation and polyploidy when compared to the contemporary F4/80high pleural macrophages from the same microenvironment (Fig. 3G and fig. S8D).

Fig. 2 Gata6 is fundamental to the peritoneal-resident macrophage phenotype.

(A) Volcano plot showing the differential gene expression between peritoneal macrophages from Gata6-KOmye and WT mice. Significantly twofold down-regulated (green) and up-regulated (magenta) probe sets are indicated. (B) Same volcano plot as (A), overlaid (orange) with the 215 peritoneal macrophage–selective cluster 15 (Cl. 15) probe sets [figs. S1 and S2 (14)], which were significantly (below) disproportionately down-regulated in the absence of Gata6. (C to E) Representation [(C) and (E)] and quantification [(D) and (E)] of flow cytometric validation of the array data from (A). Data (analyzed by t test) represent the difference in median fluorescent intensity (ΔMFI) between receptor-specific and isotype-control antibodies (mean ± SEM) of individual mice (n = 4) from one of two experiments (solid bars denote WT and hatched bars denote Gata6-KOmye mice).

Fig. 3 Dysregulated peritoneal macrophage proliferation in the absence of Gata6.

(A to C) Representative density plots (A) gated on resident peritoneal macrophages (Fig. 1A) showing proliferation (SG2M) and polyploidy, which were quantified (B) and visualized [arrowheads in (C)], respectively. Data in (A) and (B) are derived from one of two independent experiments [Gata6-KOmye, n = 5 mice; heterozygous Gata6-KOmye (Het), n = 4 mice; WT, n = 3 mice] represented as mean ± SEM and analyzed by one-way ANOVA (P value as indicated) with Bonferroni post tests. Immune fluorescence is representative of five mice. (D) Examination of proliferative differences between the majority F4/80low (white circles) KO and the WT F4/80high (black circles) macrophages (fig. S9) within the same Gata6-KOmye mice. Lines denote paired samples from the same mice (n = 9), which were pooled from two similar experiments and analyzed by paired t test. (E) The impact of Gata6 deletion on proliferation was examined 7 days after delivery of Cre-expressing lentiviruses to Gata6tm2.1Sad/J mice intraperitoneally. The proportion of cells in the SG2M phases of the cell cycle were compared between F4/80lowCre+ (white circles) and F4/80highCre (black circles) macrophages. Data are represented as three independent experiments (Exp), with lines denoting paired samples from the same mice and analyzed as indicated. (F) Gata6 mRNA expression compared by quantitative polymerase chain reaction between peritoneal and pleural leukocytes. The data show mean ± SEM from one of two independent experiments in 129S6 mice (n = ≥3 per group), normalized to reflect the number of resident macrophages. (G) Similar analysis to that in (D), except using pleural macrophages. Data from two similar experiments were pooled and analyzed by a paired t test. *P < 0.05, **P < 0.01, ***P < 0.001.

During acute inflammation, peritoneal macrophage numbers first decrease. This is followed by the M-CSF–dependent (Il4ra-independent) proliferation of surviving macrophages (6, 8). We induced acute peritonitis with 2 × 106 zymosan particles and observed increases in the numbers of neutrophils and eosinophils in both WT and Gata6-KOmye mice (fig. S10). The initial decrease in the number of Tim4+ macrophages was seen in both genotypes; however, whereas the Tim4+ macrophages were mostly restored to preinflammation levels in WT mice, this did not occur in the Gata6-KOmye mice 48 hours after challenge (Fig. 4A). In contrast to the inflammation-enhanced proliferative response in WT mice, the proliferation of Gata6-KOmye macrophages remained unaltered (Fig. 4B). Because alterations in inflammation or basal conditions between WT and Gata6-KOmye mice could affect the proliferative recovery of the tissue macrophages, we compared the proliferative state of the Tim4+F4/80high WT macrophages to that of the Tim4+F4/80low KO cells within the same Gata6-KOmye mice (Fig. 4, C and D). Unlike the KO macrophages, the F4/80high WT cells in the same environment responded to inflammation with elevated proliferation, confirming that the role of Gata6 was cell-intrinsic and not a consequence of secondary and/or environmental factors (Fig. 4, C and D). The mechanism underlying these phenotypic alterations is likely to be complex. A bioinformatic analysis indicated a high probability that multiple transcriptional networks were influenced by Gata6 deficiency (table S7). Thus, the altered phenotype imposed by the loss of Gata6 activity may arise from both direct Gata6 signaling and indirect responses mediated downstream of Gata6. Gata6 can therefore alter both cell proliferation and the phenotypic specialization of macrophages within the resident tissue. As validation of our approach, we selected Map3k8, which exhibits significantly altered expression in the absence of Gata6 (fig. S11, A and B). We anticipated that Map3k8 may be involved in the proliferation of peritoneal macrophages, because we had found that this process has an absolute requirement for M-CSF (8). Lentiviral short hairpin RNA–mediated Map3k8 knockdown resulted in significantly reduced proliferation during inflammatory resolution (fig. S11, C to E). Alterations in macrophage phenotype and restoration could affect inflammatory resolution; for example, leading to delayed neutrophil clearance, so we initiated inflammation with a higher dose of zymosan (2 × 107 particles), where the initiation of inflammation is less macrophage-dependent (3, 1921) (Fig. 4E). Compared to WT animals, Gata6-KOmye mice had substantially lower numbers of recoverable Tim4+ macrophages, slightly increased levels of Tim4 macrophages, and increased neutrophil numbers during the resolution of inflammation (Fig. 4E).

Fig. 4 Impaired proliferative recovery of peritoneal macrophages during inflammation in the absence of Gata6.

(A and B) Quantification of the numbers (A) and proliferation (B) of Tim4+ macrophages at the indicated times after intraperitoneal zymosan injection (2 × 106 particles). Data were pooled from two independent experiments with Gata6-WT (black bars) and Gata6-KOmye (white bars) mice (n = 5 to 11 per group). (C) Comparison of the SG2M and G1 stages of the cell cycle in the Gata6-KO (Tim4+F4/80low, white bars) and WT (Tim4+F4/80high, black bars) macrophages within the Gata6-KOmye mice from (A) and (B). (D) Verification of the proliferative alterations shown in (C) by in vivo incorporation of 5-ethynyl-2'-deoxyuridine (EdU) (n = 8 mice, pooled from two independent experiments). (E) Analysis of inflammatory parameters in the resolution phase of a higher-dose zymosan (2 × 107 particles) peritonitis model comparing cell counts from WT (black bars) and Gata6-KOmye (white bars) mice [data pooled from three independent experiments (n = 11 to 13 per group)]. All data in this figure represent mean ± SEM of individual mice and were examined by ANOVA (as indicated) with pairing as appropriate and Bonferroni post tests, except (D), which was analyzed by paired t test. *P < 0.05, **P < 0.01, ***P < 0.001.

Here we have identified Gata6 as a master controller of the peritoneal macrophage–specific phenotype. This phenotype is intrinsically linked to the regulation of proliferation. Our observations demonstrate transcriptional control of tissue-resident macrophage proliferative renewal and have implications for the study of tissue macrophages and tissue physiology in general. They indicate that not only do resident macrophages acquire a specialized phenotype adapted for a specific microenvironment, but that this is integral to the systems that preserve regulated self-renewal. Ex vivo, peritoneal macrophages alter their phenotype (22, 23), including an apparent absence of proliferation (24). We also observed a down-regulation of Gata6 in ex vivo cultures, which in itself alludes to the presence of local Gata6 induction within the tissue. Further afield, these observations dictate that to understand the master controllers and interaction of any resident macrophage population within its tissue, a full context-specific characterization of these cells will be required. It can be anticipated that many of the individual downstream pathways used by tissue-resident macrophages to interact with their environment may be common between different sites. The development of viable Gata6-deficient peritoneal macrophages provides an opportunity to dissect the functional interaction between tissue-resident macrophages and their tissue in a highly tractable system to aid in the identification of approaches to promote tissue homeostasis, the resolution of inflammation, and host defense.

Supplementary Materials

www.sciencemag.org/content/344/6184/645/suppl/DC1

Materials and Methods

Figs. S1 to S11

Tables S1 to S7

References (2537)

References and Notes

  1. Information on materials and methods is available as supplementary material on Science Online.
  2. Acknowledgments: We thank the staff of our animal facilities for the care of the animals. We also thank C. Pepper, J. Fisher, C. Watkins, and M. Musson for their help with cell purifications and Affymetrix analysis. P.R.T conceived and designed the project and wrote the manuscript; M.R., L.C.D, P.J.G, C.-T.L., B.K., T.C.S., and P.R.T designed and conducted the experiments; and all authors contributed to the analysis and interpretation of the data. The data presented in this manuscript are tabulated in the main paper and in the supplementary materials. Microarray data has been deposited in the Gene Expression Omnibus (accession nos. GSE28621 and GSE47049). P.R.T. is a Medical Research Council (MRC) UK Senior Non-Clinical Fellow (grant G0601617). This work was also supported by an MRC project grant (MR/J002151/1). V.O.D. is supported by a Wellcome Trust Programme Grant. L.C.D. is an MRC Doctoral Training Grant recipient, Cardiff University 125 for 125 scholar, and MRC Centenary Award holder. All animal work was conducted in accordance with institutional and UK Home Office guidelines. The authors declare no conflicts of interest.
View Abstract

Navigate This Article