Dynamics of monocyte-derived macrophage diversity in experimental myocardial infarction

Monocyte, Macrophage, Inflammation, Myocardial infarction, Single-cell RNA-seq

Time for primary review: 20 days

1. Introduction

Macrophages are critically involved in cardiac repair after myocardial infarction (MI), where they have a dual role as they can either promote tissue repair or precipitate myocardial damage.¹ In the infarcted heart, cardiac macrophages form a large group of cells with different ontogenies,² temporal dynamics and states, and delineating the contribution of each macrophage subtype to post-MI cardiac repair remains a major challenge.³ On one hand, resident tissue macrophages (RTMs) that self-renew independently of circulating monocyte input have cardioprotective functions.^4,5 On the other hand, recruited monocyte-derived macrophages that numerically predominate during the acute post-MI phase adopt more heterogeneous phenotypes, and can either drive tissue repair or damage.³ Fundamental work investigating the kinetics of monocyte/macrophage transitions in the infarcted heart based on flow cytometry or bulk transcriptomics has established the notion that functionally distinct subsets of monocytes and monocyte-derived cells are present in the heart with specific temporal dynamics.^6–8 More recent advances based on single-cell transcriptomic profiling have highlighted a substantial heterogeneity of monocyte-derived macrophages in various disease contexts.⁹ Although single-cell analysis has provided new insights into cardiac resident macrophage diversity⁴ or the Type I interferon (IFN) response in macrophages after ischaemic injury,^10,11 a detailed analysis of the monocyte-to-macrophage transition after acute MI has not been performed previously, and the identity of monocyte-derived macrophage subsets, their dynamics, and the mechanisms regulating their acquisition of discrete states associated with specific functional capacities remain to be precisely characterized. The recent introduction of multi-modal single-cell analysis, and in particular, the simultaneous measurement of transcript expression and cell-surface markers in single cells via cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq),¹² offers a unique opportunity to refine our understanding of monocyte-derived macrophage heterogeneity in the infarcted myocardium.

Here, we used CITE-seq,¹² combined with targeted depletion of circulating CCR2⁺ monocytes and fate-mapping analysis of tissue resident macrophages, to precisely characterize circulating and cardiac monocyte-derived cell states after experimental MI. We identified monocyte-derived macrophage subsets with discrete gene expression signatures and analysed their dynamics in the infarcted heart. Notably, we identified macrophages presenting a Trem2^hi lipid-associated gene expression signature acquired in the ischaemic heart, and reminiscent of a lipid-associated macrophage (LAM) signature found in other tissues in metabolic disease contexts. Altogether, our work presents a refined time-resolved map of monocyte/macrophage transitions after experimental MI, providing a valuable resource for further understanding the multifaceted roles of monocytes and macrophages in ischaemic heart repair.

2. Methods

Detailed experimental methods are available with Supplementary material online. Single-cell RNA-seq and CITE-seq data shown in this report can be browsed in a web-accessible interface: https://infection-atlas.org/Rizzo2022/.

All animal studies and numbers of animals used conform to the Directive 2010/63/EU of the European Parliament and have been approved by the appropriate local authorities (Regierung von Unterfranken, Würzburg, Germany, Akt.-Z. 55.2-DMS-2532-2-743 and 2-865). Illustrations were created using Servier Medical Art (smart.servier.com). In all experiments, animals were killed by cervical dislocation under isoflurane anaesthesia (induced by isoflurane inhalation 4.0%).

3. Results

3.1 Single-cell CITE-seq analysis of macrophages in the healthy and infarcted heart

Myocardial infarction is characterized by rapid and massive infiltration of inflammatory monocytes and a shift in the cardiac monocyte/macrophage landscape.^6,7 To obtain a high-dimensional characterization of monocyte/macrophage shifts in the ischaemic heart based on transcriptomic profiling and analysis of cell-surface marker expression, we performed CITE-seq analysis of cardiac CD45⁺ cells encompassing 45 epitopes defining all the main immune lineages and discrete subsets of innate and adaptive immune cells (see Section 2). We analysed CD45⁺ cells extracted from the heart of male wild-type C57BL6/J mice in the steady state (n = 5) and at 5 days after MI (n = 9), comprising a total of 13 805 cells (Figure 1A and B). We performed our main analysis at Day 5 after MI, as it represents a time point where differentiated macrophages are abundant, but active Ly6C^hi monocyte recruitment is still ongoing,⁷ which allows capturing unsynchronized cells at various stages of the monocyte-to-macrophage transition in a single data set. Immune cell lineages were identified using canonical surface markers of T cells (CD3), NK cells (NK1.1), γδT cells (γδTCR), B cells (CD19), dendritic cells (DCs: MHCII, CD11c) including XCR1⁺ cDC1, plasmacytoid DCs (SiglecH), neutrophils (Ly6G), F4/80⁺CD64⁺CX3CR1⁺Ly6C^lo macrophages comprising steady-state RTMs and macrophages appearing after MI (MI Mac), and monocytes (Ly6C⁺; Figure 1C, see Supplementary material online, Figure S1A). Cellular identities, including small clusters of contaminating endothelial cells and fibroblasts, were confirmed by examination of enriched transcripts (see Supplementary material online, Figure S1B). Monocytes/macrophages (n = 10 365 cells) were extracted and reclustered in silico to precisely delineate subpopulations and their distribution in the healthy and infarcted heart (Figure 1D and E). Based on differentially expressed cell-surface markers and transcripts (Figure 1F–H, see Supplementary material online, Figure S1C and D), we annotated 13 populations of monocytes/macrophages and their relative abundance in the steady-state and ischaemic heart (Figure 1D and E, see Supplementary material online, Figure S1E and F).

Figure 1

CITE-seq analysis of the monocyte/macrophage landscape in the steady-state and infarcted heart. For all graphs in this figure, cells were obtained from n = 5 mice without MI, and n = 9 mice with MI, pooled from two independent experiments (see Section 2). (A) Experimental design summary; (B) UMAP representation of transcriptome-based clustering of n = 13 805 total cardiac CD45⁺ cells; (C) CITE-seq signal for the indicated monocyte/macrophage surface markers projected onto the total CD45⁺ cells UMAP plot; (D) n = 10 365 cells corresponding to monocytes and macrophages (including proliferating macrophages) were extracted for clustering and UMAP dimensional reduction analysis with annotated cell clusters (left) and sample of origin colour coded on the UMAP plot (right); (E) average proportions of each cluster according to experimental condition; (F) surface markers and (G) transcript expression projected onto the UMAP plot for selected markers used to identify and annotate clusters; (H) expression of the indicated surface markers in each monocyte/macrophage cluster; (I and J) expression of Ccr2 projected on the UMAP plot of monocyte/macrophages (I) and shown across clusters (J). RTM, resident tissue macrophages; MI Mac, MI-associated macrophages; (p)DC, (plasmacytoid) dendritic cell; Endo, endothelial cells; Fibro, fibroblasts.

At the steady state, we found two RTM populations, the first one represented 23.40 ± 2.28% of all steady-state macrophages and was defined as TIMD4⁺LYVE1⁺MHCII^lowFolr2^hi, corresponding to previously described cardiac RTM defined as Lyve1⁺,¹³ TIMD4⁺,⁴ or more recently ‘TLF’ (TIMD4⁺LYVE1⁺Folr2⁺).¹⁴ Here, we termed this population RTM-TIMD4 (Figure 1D–F). The second RTM population was MHCII⁺ with low expression of RTM-TIMD4 surface markers and transcripts, called RTM-MHCII (60.33 ± 1.52% of all steady-state macrophages) (Figure 1D–H and see Supplementary material online, Figure S1C, E and F). We identified surface markers differentially expressed by RTM subsets: compared with MI-associated macrophages, RTM-TIMD4 and RTM-MHCII expressed higher surface MGL2 (Figure 1H). RTM-TIMD4 additionally expressed high CD163, and part of this cluster expressed VSIG4 (Figure 1H). While RTM-TIMD4 and RTM-MHCII were mostly negative for Ccr2 expression (Figure 1I and J), previous reports have identified a third cardiac RTM subset defined as CCR2⁺MHCII⁺,⁴ and it was recently proposed that three RTM subsets populate mouse tissues, namely ‘TLF’, MHCII⁺, and CCR2⁺ macrophages, each presenting a specific core gene expression signature conserved across organs.¹⁴ To better resolve steady-state populations of cardiac monocyte/macrophages, we reclustered cells from the steady state alone, which recovered similar populations (see Supplementary material online, Figure S2A–C). Ccr2 expression was observed in Ly6C^hi monocytes (which represent monocytes within the tissue, as ensured by exclusion of intravenous leucocytes during sorting, see Supplementary Methods). Ccr2 was also detected within the RTM-MHCII⁺ cluster, which is in line with the existence of resident CCR2⁺MHCII⁺ macrophages in the heart.^4,14 However, in our hands, CCR2⁺MHCII⁺ RTM did not form a separate cluster as proposed in Dick et al.^4,14 We manually gated Ccr2⁺ cells within the RTM-MHCII⁺ population (see Supplementary material online, Figure S2D and E) and analysed expression of ‘core signature genes’ of the three RTMs conserved across organs (including the heart) as defined in Dick et al.¹⁴ Within these three bona fide RTM alone, relative expression of the core signature genes appeared mostly consistent with the distribution recently proposed in Dick et al.¹⁴ When considering all steady-state cardiac populations, relative expression of the genes of the ‘CCR2⁺ RTM’ signature was highest in monocytes, excepted for MHCII encoding transcripts (see Supplementary material online, Figure S2F and G).

At Day 5 post-MI, the monocyte/macrophage compartment was dominated by Ly6C^hi monocytes (17.56 ± 6.58% of monocytes/macrophages at Day 5 post-MI), pro-inflammatory macrophages expressing MHCII and enriched in Il1b, Tnip3, Tlr2, or Tnfsf9 (MHCII⁺Il1b⁺; 14.51 ± 3.70% of monocytes/macrophages), and Type I IFN signature cells enriched for Isg15 (previously termed IFNIC: IFN-inducible cells¹⁰; Isg15^hi, 6.25 ± 1.91%; Figure 1D–G). We also delineated three clusters sharing enrichment for Trem2 (Trem2^hi; Figure 1D and G). A first Trem2^hi population had high expression of Spp1 encoding osteopontin (Trem2^hiSpp1^hi, 21.20 ± 4.63% of monocytes/macrophages), a second cluster was enriched for Gdf15 (Trem2^hiGdf15^hi, 20.74 ± 5.01% of monocytes/macrophages), while a third cluster was characterized by enrichment for Prdx1 (Trem2^hiPrdx1^hi; 2.51 ± 1.24% of monocytes/macrophages; Figure 1D and G). Low levels of Ly6C^low monocytes and a population defined as Fn1^hiLtc4s^hi were observed. Finally, we recovered a minor population characterized by mixed expression of tissue resident and MI-associated macrophage markers, found at similar levels in both experimental conditions, showing low transcript variety and a low RNA content (‘low RNA’ cluster, see Supplementary material online, Figure S1C and D). We interpret these cells as low-quality cells damaged during sample preparation, and did not consider them further in subsequent analyses.

As the data presented in Figure 1, results from computational integration of two independent CITE-seq experiments where samples were pooled using cell hashing¹⁵ (see Section 2), we ensured reproducibility of our observations by analysing both experiments independently with similar analysis parameters as those used initially in Figure 1. The result of the analysis is presented in Supplementary material online, Figure S3. In each separate experiment, similar populations of monocytes/macrophages were recovered, although some small populations (Trem2^hiPrdx1^hi and Ly6C^lo monocytes enriched in, e.g. Ace or Ear2) were not resolved at the clustering resolution used (see Supplementary material online, Figure S3A–J). We also observed variability in distribution of RTM to the RTM-TIMD4⁺ or RTM-MHCII⁺ clusters, which we attribute to ‘intermediate’ TIMD4^intMHCII^int cells, also observed by flow cytometry (see Supplementary material online, Figure S3L), distributing preferentially to the RTM-MHCII⁺ (Experiment 1) or RTM-TIMD4⁺ (Experiment 2) clusters in different experiments (see Supplementary material online, Figure S3B, G and K). Clustering analysis of each biological replicate independently also provided consistent results, although the analysis of much lower cell numbers in each single replicate overall yielded a lower clustering granularity (see Supplementary Analysis R Notebook).

3.2 Trem2^hi macrophage subsets have a LAM transcriptional signature conserved in humans

Trem2 expression is characteristic of a LAM (lipid-associated macrophage) signature previously defined by single-cell RNA-seq in metabolic disease contexts, such as in the obese adipose tissue,¹⁶ in atherosclerotic lesions,^17,18 and in the liver in experimental models of NAFLD¹⁹ and NASH.²⁰ At first glance, cardiac Trem2^hi macrophage populations appeared enriched for some characteristic LAM transcripts (e.g. Gpnmb, Spp1, Ctsd, Cd63, Psap; see Supplementary material online, Figure S1C). To further explore gene expression similarities across tissues, we first extracted a list of 66 commonly enriched transcripts (see Section 2) in Foamy/Trem2^hi macrophages in atherosclerotic lesions,¹⁸ and LAMs in the NASH liver,²⁰ two lipid driven pathologies, and used these to calculate a ‘LAM gene signature’ score in cardiac macrophages. The LAM signature was highest in Trem2^hiSpp1^hi and Trem2^hiPrdx1^hi macrophages, followed by the Trem2^hiGdf15^hi cluster (Figure 2A). Gene ontology analysis revealed putatively enriched biological processes relevant to a LAM state, for example, ‘positive regulation of macrophage foam cell formation’ in Trem2^hiSpp1^hi, and ‘positive regulation of cholesterol efflux’ in Trem2^hiGdf15^hi macrophages (Figure 2B). LAM-signature transcripts (e.g. Gpnmb, Pld3, Nceh1, Psap) showed a clear and specific enrichment in Trem2^hi macrophage populations (Figure 2C). When directly integrating scRNA-seq data obtained from the infarcted heart, NASH liver, and atherosclerotic aortas, Trem2^hi MI-associated cardiac macrophage populations clustered together with Foamy/Trem2^hi macrophages from atherosclerotic vessels¹⁸ and LAMs from NASH liver²⁰ (see Supplementary material online, Figure S4A–C). Gpnmb has previously been associated with the LAM signature^19–21 and appeared specifically enriched in Trem2^hi MI-associated cardiac macrophages with a LAM signature (Figure 2C). At Day 5 after MI, GPNMB expressing CD68⁺ macrophages were observed specifically in the infarcted area of the myocardium (Figure 2D, see Supplementary material online, Figure S5), indicating that LAM-signature macrophages accumulate in the infarcted area. TREM2 protein levels were increased in the heart of infarcted mice compared with sham controls at Day 5 (Figure 2E), further confirming cardiac accumulation of TREM2 expressing cells after MI.

Figure 2

Identification of Trem2^hi macrophages with a LAM signature in the ischaemic myocardium. (A) Sixty-six LAM-signature genes were extracted from the indicated data sets and a gene expression score was applied to cardiac macrophages (LAM-signature expression score), represented here as a violin plot. The atherosclerosis data set consists of a pool of n = 12 scRNA-seq data sets (see Zernecke et al.¹⁸); the NASH data set of one scRNA-seq library (Daemen et al.²⁰); the LAM-signature expression score was applied to cardiac macrophages (n = 10 365 cells; cells were obtained from n = 5 mice without MI, and n = 9 mice with MI, pooled from two independent experiments, see Figure 1, see Section 2). (B) Fold enrichment for the top 10 enriched Gene Ontology (GO) Biological Processes in the indicated cardiac macrophage cluster (all with adjusted P-value <0.05); macrophage clusters were determined in analysis shown in Figure 1D from n = 10 365 cells; cells were obtained from n = 5 mice without MI, and n = 9 mice with MI, pooled from two independent experiments. (C) Expression of selected LAM-signature transcripts projected on the UMAP plot of monocytes/macrophages as detailed in Figure 1D (n = 10 365 cells; cells were obtained from n = 5 mice without MI, and n = 9 mice with MI, pooled from two independent experiments). (D) Immunofluorescence labelling of CD68 and GPNMB in Day 5 infarcts with overview of a full myocardial section and high magnification images of the infarcted area and remote non-ischaemic myocardium (representative pictures shown for n = 1 heart). (E) Total levels of TREM2 detected by ELISA in extracts from the hearts of mice after sham operation (n = 5 mice) or permanent MI (Day 5; n = 6 mice). (F) Dot plot showing expression of the selected transcripts in the indicated cardiac macrophage clusters (all the transcripts shown significantly enriched in the relevant clusters with adjusted P-value <0.05); macrophage clusters were determined in analysis shown in Figure 1D from n = 10 365 cells; cells were obtained from n = 5 mice without MI, and n = 9 mice with MI, pooled from two independent experiments. NASH, non-alcoholic steatohepatitis.

The macrophage LAM signature is conserved across species in various disease contexts.^16,18,22 To evaluate whether the LAM signature was also found in the diseased human heart, we analysed recently published data from human ischaemic cardiomyopathy patients (n = 3 samples ischaemic left ventricle, 1 sample remote non-ischaemic left ventricle, and 1 sample control myocardium).²³ A population of macrophages enriched for characteristic genes of the LAM signature (e.g. TREM2, SPP1, CTSD, FABP5) was readily observed (see Supplementary material online, Figure S4D and E).

We further analysed the gene expression signature of the major Trem2^hiSpp1^hi and Trem2^hiGdf15^hi populations, relative to other abundant MI-associated macrophages, that is MHCII⁺Il1b⁺ and Isg15^hi clusters, with a focus on experimentally validated effectors (Figure 2F). MHCII⁺Il1b⁺ macrophages were enriched in transcripts associated with a pro-inflammatory and pathogenic profile (e.g. Axl,²⁴Nlrp3²⁵), while Isg15^hi highly expressed Il18²⁶ in addition to their typical Type I IFN response signature (Isg15, Irf7, Cxcl10).^10,11 In contrast, Trem2-enriched populations highly expressed genes involved in immune modulation, tissue repair and efferocytosis (Trem2^hiSpp1^hi: Hmox1,²⁷Arg1,²⁸Anxa1²⁹; Trem2^hiGdf15^hi: Igf1,³⁰Gdf15,³¹Mertk,³²Timp2,³³Apoe³⁴). Both Trem2^hi populations were enriched for expression of the profibrotic, LAM-signature transcript¹⁹Spp1 (with highest levels observed in Trem2^hiSpp1^hi). Trem2^hiSpp1^hi were additionally enriched for profibrotic Mmp14 (encoding MT1-MMP)³⁵ (Figure 2F). Altogether, our data show the accumulation of two populations of Trem2^hi macrophages in the ischaemic heart harbouring a LAM signature also identified in other organs and disease contexts and in the diseased human heart.

3.3 Dynamics of monocyte/macrophage populations in the infarcted heart

To resolve the kinetics of MI-associated macrophage accumulation, we integrated our data with existing scRNA-seq data sets of cardiac macrophages from the steady state and at 1, 3, 5, 7, and 11 days after MI^4,36,37 (Figure 3A and see Supplementary material online, Figure S6A and B), representing a total of 24 637 cells. Similar MI-associated populations were recovered, although the minute Trem2^hiPrdx1^hi population grouped with Trem2^hiSpp1^hi macrophages at this clustering resolution (see Supplementary material online, Figure S6C). We observed dynamic transitions within the monocyte/macrophage compartment (Figure 3A and B). Trem2^hiSpp1^hi cells peaked at Days 3–5 post-MI, while Trem2^hiGdf15^hi cells peaked at Days 5–7 (Figure 3A and B). Infiltration of MHCII⁺Ilb1⁺ and Isg15^hi macrophages also peaked at Days 3–7. At Days 7 and 11, partial recovery of RTM subsets was observed (Figure 3A and B). To better resituate our observations relative to previous analyses of monocyte/macrophage dynamics in the infarcted heart performed using flow cytometry,⁶ we analysed the expression of Ly6C, a monocyte surface marker, vs. expression of macrophage differentiation markers (CD64, CX3CR1, and F4/80) on monocytes and the major MI-associated macrophage populations (Figure 3C and D, see Supplementary material online, Figure S6D). Ly6C^hi monocytes where Ly6C^hiCD64^lowCX3CR1^lowF4/80^low and Trem2^hiGdf15^hi cells appeared as fully differentiated macrophages (Ly6C^lowCD64^hiCX3CR1^hiF4/80^hi), while Trem2^hiSpp1^hi had a Ly6C^intCD64^intCX3CR1^intF4/80^int profile suggestive of a monocyte-to-macrophage intermediate state (Figure 3C and D, see Supplementary material online, Figure S6D). Altogether, these data suggest that the Trem2^hiSpp1^hi population represents a monocyte-to-macrophage intermediate state while Trem2^hiGdf15^hi cells are differentiated macrophages, and that these populations sequentially peak in the infarcted heart.

Figure 3

Monocyte/macrophage dynamics in the infarcted heart. (A) The indicated MI data sets were integrated using Harmony and monocyte/macrophage population identified; the resulting clustering analysis and UMAP plot are shown split according to time point after MI with monocyte/macrophage clusters colour coded; n = 24 637 total cells from n = 6 independent scRNA-seq data sets. (B) Proportion of the indicated clusters within total monocyte/macrophage over the post-MI time continuum, calculated from the integrated data set. Each data point represents proportion of the indicated cluster at the indicated time point in independent scRNA-seq libraries (see also Supplementary material online, Figure S5). (C) CITE-seq signal for Ly6C vs. CD64 and (D) Ly6C vs. CX3CR1 in the indicated cell populations (analysis in C and D performed on monocyte/macrophage clusters determined in analysis shown in Figure 1D from n = 10 365 cells; cells were obtained from n = 5 mice without MI, and n = 9 mice with MI, pooled from two independent experiments).

3.4 Trem2^hi macrophage populations originate from recruited monocytes

Macrophage origin (i.e. self-renewing tissue resident macrophage vs. monocyte-derived macrophage) has been proposed as a major driver of cardiac macrophage function.³ We thus investigated the origin of the major MI-associated macrophage populations. High Ccr2 expression in MHCII⁺Ilb1⁺ and Isg15^hi clusters (Figure 1I and J), clearly indicates a monocytic origin. Trem2^hiSpp1^hi and Trem2^hiGdf15^hi macrophages expressed intermediate and low levels of Ccr2, respectively (Figure 1I), and lacked expression of surface markers associated with RTM subsets (CD163, TIMD4, LYVE1, VSIG4, MHCII; Figure 1), suggesting a monocytic origin. We hypothesized that Trem2^hi macrophages originate from Ly6C^hiCcr2⁺ monocytes that lose Ccr2 expression upon differentiation in the heart.

To test this hypothesis, we first depleted circulating Ly6C^hi monocytes in mice using the anti-CCR2 monoclonal antibody MC-21.³⁸ Twenty-four hours after a single injection, Ly6C^hi monocytes were efficiently depleted from the bloodstream as previously described³⁸ (see Supplementary material online, Figure S7A), and MI was induced. Monocyte depletion was maintained by daily i.p. injections of anti-CCR2 until Day 5 after MI (see Supplementary material online, Figure S7B), where cardiac CD45⁺ cells were sampled for CITE-seq analysis (Figure 4A–C). Analysis of CITE-seq data identified the main CD45⁺ cell populations and monocyte/macrophage subsets (Figure 4C) similar to those described above (Figure 1). While levels of RTM subsets were not affected by CCR2⁺ cell depletion, known circulating monocyte-derived populations (Ly6C^hi monocytes, MHCII⁺Il1b⁺ macrophages, Isg15^hi) were strongly reduced in the infarcted heart at Day 5. The Fn1^hiLtc4s^hi population was not affected, indicating that it is independent of CCR2⁺ monocyte recruitment. We furthermore noted a drastic reduction of Trem2^hiSpp1^hi and Trem2^hiGdf15^hi macrophages counts, indicating that accumulation of these cells is dependent on circulating monocyte infiltration (Figure 4D). Additionally, cDC1³⁹ and cDC2, proposed to be Ccr2 dependent in inflammatory contexts⁴⁰ were reduced (see Supplementary material online, Figure S7D). Cardiac neutrophils, T cells, NK cells, and pDC numbers were not affected by anti-CCR2 (see Supplementary material online, Figure S7C).

Figure 4

MI-associated macrophage populations originate from recruited CCR2⁺ monocytes. (A) Schematic representation of the experimental design; (B and C) UMAP representation of scRNA-seq analysis (n = 10 831 cells) with (B) sample of origin and (C) biological identity of cell clusters colour coded on the UMAP plot; (D) absolute counts of the indicated cell clusters (per mg of cardiac tissue); data shown in A–D were obtained from one experiment with n = 3 mice without MI, n = 4 mice at Day 5 after MI treated with isotype control; n = 5 mice at Day 5 after MI treated with anti-CCR2; (E) annotated UMAP plot of cells from Dick et al.⁴ (n = 5802 cells from n = 1 scRNA-seq data set from mice without MI and n = 1 scRNA-seq data sets from mice at 11 days after MI) extracted from integrated data analysis shown in Figure 3 and (F) identification of TdTomato⁺ fate mapped RTMs, cells ordered according to transcript detection, that is, cells with detectable transcripts moved to front of the plot; (G) experimental setup for CX3CR1-based fate mapping of tissue resident macrophages; (H) recombination controls in Ly6C^hi monocytes and microglia after the 4 weeks washout period; (I and J) fate mapping of cardiac macrophages before and at 7 days post-MI, pre-gated on live CD45⁺CD11b⁺F4/80^hiLy6C^low. Data shown in G–J were obtained in one experiment with n = 2 mice without MI and n = 3 mice at Day 7 after MI.

To further confirm that Trem2^hi macrophage populations derive from monocytes, we then analysed the expression of TdTomato transcripts that mark RTM in pulse-labelled Cx3cr1^CreERT2 mice in Dick et al.⁴ from our integrated analysis (Figure 4E and F). TdTomato transcripts mapped preferentially to tissue resident and proliferating macrophages, while only scattered cells were observed in Trem2^hi subsets and MHCII⁺Il1b⁺ cells (Figure 4E and F). We additionally performed fate-mapping analysis of RTM and MI-associated macrophages by flow cytometry in tamoxifen-pulsed Cx3cr1^creERT2-Rosa26^EYFP mice⁴¹ after a 4-week washout period (Figure 4G), when circulating Ly6C^hi monocytes had lost YFP expression while microglia were >90% YFP⁺ (Figure 4H). Only partial recombination was observed in long living macrophages such as TIMD4⁺MHCII⁻ RTM (40.35 ± 4.45%) in the steady state, which remained stable after MI (41.63 ± 7.65% YFP+ in TIMD4⁺MHCII⁻ RTM; Figure 4I and J). At 7 days after MI TIMD4⁻MHCII⁻ macrophages, most MI-associated macrophage populations including Trem2^hi subsets, as defined by CITE-seq (Figure 1), were overwhelmingly YFP^neg (93.47 ± 2.25%; Figure 4I and J). Although partial tamoxifen-induced recombination in RTM represents a limitation of our analysis, these data nevertheless suggest that pre-existing RTM do not represent a major source of MI-associated TIMD4⁻MHCII⁻ macrophages (Figure 4I and J). Altogether, our complementary approaches of circulating CCR2⁺ monocyte depletion, fate mapping of RTM in previously published scRNA-seq data, and fate mapping of RTM by flow cytometry indicate that Trem2-enriched, MI-associated macrophage subsets originate from recruited monocytes.

3.5 The macrophage Trem2^hi/LAM signature is acquired in the ischaemic heart

We next investigated whether MI-associated cardiac monocyte/macrophage states could be acquired remotely, notably via the production of ischaemic injury-associated atypical monocytes arising from emergency myelopoiesis,⁴² that would transit in the bloodstream before infiltrating the myocardium. We performed CITE-seq analysis of CD19⁻NK1.1⁻Ter119⁻CD11b⁺ cells from the blood of control mice, sham-operated animals at 1 and 3 days post-surgery, and mice with MI at 1 and 3 days. Cardiac cells from the ischaemic heart at 1 and 3 days post-MI were also included³⁷ (Figure 5A).

Figure 5

Monocytes/macrophages acquire the Trem2^hi LAM signature in the ischaemic heart. (A) Experimental design overview; (B) Ccr2 transcript expression and CITE-seq signal for the indicated surface markers in blood and heart CD19⁻NK1.1⁻Ter119⁻CD11b⁺ cells projected on the UMAP plot (n = 9848 cells); (C) tissue of origin of single cells corresponding to monocytes/macrophages/cDC2 (n = 3378 cells) projected on the UMAP plot; (D) clustering analysis and annotation of cell clusters; (E) pseudotime analysis of monocytes/macrophages in Monocle, cell identity colour-code identical to D (n = 3378 cells); (F) pseudotime analysis split according to tissue origin; (G) heatmap of pseudotime gene expression variation for selected genes on branches of the pseudotime tree (as indicated on E; n = 3378 cells; all indicated genes show statistically significant variation). All data shown in Figure 5 were obtained from n = 1 pooled cell preparation per experimental condition, grouped in a single (n = 1) multiplexed scRNA-seq data set.

We first investigated remote priming of blood monocytes by MI by separately reclustering cells obtained from the peripheral blood. This revealed expansion of injury-associated Ly6C^hi monocytes enriched for Chil3 (encoding Chitinase-like protein 3 also known as Ym1) and granulocyte marker genes (Lcn2, Prtn3) induced by MI, and also sham surgery at 1 day after MI (Chil3^hi monocytes, see Supplementary material online, Figure S8A–E). Bulk-sorted bone marrow Ly6C^hi monocytes also showed increased levels of Chil3, Lcn2, and Prtn3 at 3 days after MI (see Supplementary material online, Figure S8F). These results are consistent with the emergence of Ym1⁺ monocytes enriched in granulocyte marker transcripts induced by tissue injury⁴³ or during emergency myelopoiesis.⁴⁴ Although we noted a slight increase of the LAM-signature expression score in Chil3^hi monocytes compared with baseline Ly6C^hi monocytes, most LAM-signature transcripts were expressed at low levels and only two genes (Gapdh, Lpl) were significantly enriched in Chil3^hi monocytes compared with baseline Ly6C^hi monocytes (see Supplementary material online, Figure S8G).

To analyse the acquisition of the LAM signature by monocytes infiltrating the ischaemic heart, we analysed blood and cardiac CD115⁺Ly6G⁻ cells together (3378 cells; Figure 5). One cell cluster contained almost only cells extracted from the ischaemic heart and expressed Ccr2 indicating an infiltrating monocyte origin (Heart Mo/Mac, Figure 5A–D). Pseudotime ordering of monocytes/macrophages in Monocle⁴⁵ delineated two major trajectories (Figure 5E and F), with blood cells expectedly following a Ly6C^hi to Ly6C^low trajectory with the acquisition of Ly6C^low monocyte marker genes (Ace, Treml4, Nr4a1, Itgal; Figure 5G), while cells infiltrated in the heart acquired expression of Trem2 and transcripts associated with the LAM signature (Figure 5G). These data indicate that MI induces a peripheral shift of Ly6C^hi monocytes towards a Chil3^hi granulocyte-like state in the blood, while in the ischaemic heart monocyte-derived macrophages acquire the full Trem2^hi LAM signature.

3.6 Monocyte-to-Trem2^hi macrophage transition in the heart and potential inducers of the macrophage LAM signature

Based on our observations that (i) Trem2^hiSpp1^hi and Trem2^hiGdf15^hi macrophages sequentially peak in the heart, (ii) Trem2^hiSpp1^hi likely represents monocyte-to-macrophage differentiation intermediates, while Trem2^hiGdf15^hi is differentiated macrophages, (iii) Trem2^hiSpp1^hi and Trem2^hiGdf15^hi macrophages share features of a LAM gene expression signature, (iv) cells corresponding to Trem2^hiSpp1^hi and Trem2^hiGdf15^hi macrophages are not observed in the bloodstream (i.e. they acquire their gene expression signature in the injured tissue), and (v) Trem2^hiSpp1^hi and Trem2^hiGdf15^hi macrophages are lost upon depletion of circulating monocytes, we performed pseudotime analysis in Monocle,⁴⁵ following the assumption that Trem2^hiSpp1^hi and Trem2^hiGdf15^hi macrophages originate from Ly6C^hi monocytes, and might represent different states along a monocyte-to-macrophage differentiation trajectory (Figure 6A). This analysis yielded a putative trajectory with Ly6C^hi monocytes at its beginning, and Trem2^hiGdf15^hi macrophages at its end, with progressive loss of monocyte marker genes (Ccr2, Chil3, Ly6c2) and acquisition of macrophage differentiation markers (Adgre1, Mertk, Apoe), while Spp1 and genes involved in the differentiation of monocytes to alternatively activated macrophages (e.g. Hmox1, Arg1) were transiently increased (Figure 6B). These results suggest that Trem2^hiSpp1^hi cells are monocyte/macrophage intermediates that may represent a transitional state towards fully differentiated Trem2^hiGdf15^hi macrophages.

Monocyte transition to Trem2hi macrophages and potential inducers of the LAM signature. (A) Pseudotime trajectory analysis of Ly6Chi monocytes, Trem2hiSpp1hi and Trem2hiGdf15hi macrophages. (B) Expression of the indicated transcripts according to pseudotime and colour coded by cell population of origin. In A and B, cells belonging to the indicated clusters (n = 4633 total cells) were extracted from the analysis shown in Figure 1D (n = 10 365 cells obtained from n = 5 mice without MI, and n = 9 mice with MI, pooled from two independent experiments). (C) Expression of the indicated transcripts in mouse bone marrow–derived macrophages (BMDM) in control condition or after overnight exposure to apoptotic cells (Apo = apoptotic thymocyte at a 5:1 apoptotic cell:macrophage ratio). Each data point represents macrophages from one mouse assayed in technical duplicates, total n = 7 per condition, pooled from two independent experiments (*P < 0.05; **P < 0.01; ***P < 0.001). (D) Reanalysis of data from Murthy et al.48 with UMAP of in vitro human monocyte-derived macrophages differentiated in the presence of calcium, GM-CSF or M-CSF. (E) LAM-signature score projected on the UMAP plot. (F) Expression of the indicated LAM-signature transcripts induced by calcium (top) or GM-CSF (bottom).

Figure 6

Monocyte transition to Trem2^hⁱ macrophages and potential inducers of the LAM signature. (A) Pseudotime trajectory analysis of Ly6C^hi monocytes, Trem2^hiSpp1^hi and Trem2^hiGdf15^hi macrophages. (B) Expression of the indicated transcripts according to pseudotime and colour coded by cell population of origin. In A and B, cells belonging to the indicated clusters (n = 4633 total cells) were extracted from the analysis shown in Figure 1D (n = 10 365 cells obtained from n = 5 mice without MI, and n = 9 mice with MI, pooled from two independent experiments). (C) Expression of the indicated transcripts in mouse bone marrow–derived macrophages (BMDM) in control condition or after overnight exposure to apoptotic cells (Apo = apoptotic thymocyte at a 5:1 apoptotic cell:macrophage ratio). Each data point represents macrophages from one mouse assayed in technical duplicates, total n = 7 per condition, pooled from two independent experiments (*P < 0.05; **P < 0.01; ***P < 0.001). (D) Reanalysis of data from Murthy et al.⁴⁸ with UMAP of in vitro human monocyte-derived macrophages differentiated in the presence of calcium, GM-CSF or M-CSF. (E) LAM-signature score projected on the UMAP plot. (F) Expression of the indicated LAM-signature transcripts induced by calcium (top) or GM-CSF (bottom).

We next investigated potential inducers of the LAM signature. Based on the gene expression profile of cardiac Trem2^hi macrophages enriched for phagocytic genes and with low expression of pro-inflammatory cytokines, we hypothesized that efferocytosis, a major regulator of macrophage anti-inflammatory gene expression in the ischaemic heart,⁴⁶ might be involved in acquisition of the LAM signature. Expression of Spp1, Trem2, Gdf15, and Fabp5 was increased in mouse bone marrow–derived macrophages after exposure to apoptotic cells, alongside previously documented efferocytosis-induced genes such as Mertk, Abca1, or Il10^46,47 (Figure 6C). A recent report⁴⁸ showed that exposing human monocyte to calcium, which is released from necrotic cells to activate macrophages,⁴⁹ leads to their differentiation into macrophages producing high levels of SPP1, a marker of Trem2^hi macrophages. Reanalysing data from Murthy et al.,⁴⁸ we observed that exposure to calcium was associated with high expression of not only SPP1, but also many characteristic genes of the LAM signature such as TREM2, CTSD, and FABP5, while monocyte-derived macrophages exposed to GM-CSF also acquired some markers of the LAM signature (Figure 6D–F). Altogether, these data suggest that phagocytosis of dead cells, and exposure to damage associated molecular patterns released by dead cells, might be involved in macrophage acquisition of the Trem2^hi LAM signature.

4. Discussion

Using single-cell CITE-seq analysis of circulating and cardiac monocytes/macrophages, combined with fate mapping and CCR2⁺ monocyte depletion, we mapped monocyte/macrophage dynamics and characterized monocyte-derived macrophage heterogeneity in experimental MI. This allowed us to precisely delineate MI-associated monocyte-derived macrophage populations in the ischaemic heart, including Trem2^hi macrophage subsets with a tissue-acquired LAM gene expression signature.

We identified previously described populations of TIMD4⁺, MHCII⁺, and CCR2⁺ RTMs,^4,14 and could confirm CD163⁵⁰ or identify novel surface markers (MGL2, VSIG4) of these populations, that may be useful for cardiac macrophage characterization by flow cytometry or targeting of specific resident macrophage subsets. Consistent with previous reports, these populations almost entirely vanished from the infarcted heart immediately after MI.^4,51

Rapid shifts in monocyte/macrophage populations after acute MI had previously been characterized by flow cytometric analyses (e.g. Nahrendorf et al.⁶ or Weinberger et al.⁵²). Our CITE-seq and time-dependent scRNA-seq characterization of cardiac monocytes/macrophages provide a refined and accurate picture of these shifts and identify previously unrecognized MI-associated macrophage states. Notably, we confirm that bona fide Ly6C^low monocytes only infiltrate the heart in low numbers,⁵³ and identified a population of Fn1^hiLtc4s^hⁱ macrophages with a discrete gene expression signature and surface expression of some RTM markers (TIMD4 and VSIG4). The origin, localization, and function of these macrophages remain to be determined. The major populations of MI-associated macrophages comprised two pro-inflammatory populations (Isg15^hi and MHCII⁺Il1b⁺), and three Trem2^hi macrophage populations with low inflammatory gene expression. All the major MI-associated macrophage populations originate from infiltrating monocytes as determined by CCR2⁺ cell depletion and fate-mapping analysis.

Trem2^hi macrophages comprised three subpopulations (two large populations termed Trem2^hiSpp1^hi and Trem2^hiGdf15^hi, and a smaller Trem2^hiPrdx1^hi subset), and showed gene expression features reminiscent of a LAM signature previously observed in obese adipose tissue,¹⁶ atherosclerotic lesions,^17,18 and the liver in models of NAFLD¹⁹ and NASH.²⁰ Cardiac Trem2^hi LAM-signature macrophages were derived from monocytes, similar to previous observations in the liver,^19,20 and adipose tissue.¹⁶ However, the LAM signature is also shared by TREM2-dependent disease-associated microglia,⁵⁴ indicating that its acquisition does not depend on macrophage origin and is mostly driven by the local tissue environment. Features of the LAM transcriptional signature may not only be induced by pathological lipid loading but rather more generally induced in contexts of tissue damage. In line, our data indicate that Trem2^hi LAM-signature macrophage populations likely differentiate from monocytes in injured areas of the myocardium, as GPNMB, a highly specific marker of the LAM signature, was exclusively seen on CD68⁺ macrophages within the infarcted area of the heart, but not in the infarct border zone or remote non-affected tissue. Major transcriptional hubs involved in the regulation of lipid homeostasis are activated also in response to efferocytosis of dead cells, such as the liver-X-receptor pathway⁴⁷ which was recently implicated in microglial acquisition of the LAM gene expression in response to chronic phagocytic challenge.⁵⁵ This raises the possibility that macrophages with high efferocytosis activity, and lipid loading from dead cell engulfment, acquire a LAM signature. In line, exposure of bone marrow–derived macrophages to apoptotic cells in vitro induced expression of several characteristic LAM transcripts. More recently, the LAM signature was also observed in steady-state bile-duct macrophages, and LAM genes were induced by lipids during in vitro monocyte-to-macrophage differentiation,²¹ indicating that also homeostatic exposure to high lipid loads may drive acquisition of a LAM signature. Acquisition of the LAM signature might not only depend on lipid-related pathways, but also be driven by other microenvironmental cues. Exposure to calcium, a damage associated molecular pattern released by necrotic cells,⁴⁸ strongly induced expression of characteristic LAM markers in human monocyte-derived macrophages. Specific inflammatory cytokines might dictate the monocyte-to-macrophage transition in the heart, as recently shown for IFN-γ and GM-CSF in the context of neuro-inflammation.⁵⁶In vitro, our data indicate that some features of the LAM signature can be driven by GM-CSF. Importantly, macrophages with a LAM signature were observed in human ischaemic cardiomyopathy samples. While this represents a different pathological context compared with acute cardiac ischaemia in our mouse model, this provides proof-of-concept that the TREM2⁺ LAM signature can be observed in the disease human heart, consistent with the LAM signature also being observed in humans in atherosclerosis,¹⁸ adipose tissue,¹⁶ and in the steatotic human liver,²¹ indicating conservation of this macrophage state across species and disease contexts.

In the ischaemic mouse heart, features of the Trem2^hi LAM signature were shared by three macrophage subpopulations. A small population was highly enriched for Prdx1 and genes involved in iron handling (Ftl1, Fth1, Slc40a1, Slc48a1). Two other Trem2^hi populations represented the most abundant MI-associated macrophages and were characterized as Trem2^hiSpp1^hi and Trem2^hiGdf15^hi. Their sequential presence in the ischaemic heart, as well as expression patterns of Ccr2 and monocyte/macrophage differentiation markers, suggest that Trem2^hiSpp1^hi cells are a monocyte/macrophage intermediate giving rise to fully differentiated Trem2^hiGdf15^hi macrophages. However, our data do not exclude that pro-inflammatory macrophage populations (Isg15^hi and MHCII⁺Il1b⁺) could also shift towards a non-inflammatory Trem2^hiGdf15^hi state overtime after MI.

Our analysis of circulating monocytes indicates that ischaemic injury induces a shift towards monocytes enriched for the expression of Chil3 (encoding Ym1) and several granulocyte-associated genes (e.g. Prtn3, Lcn2). Recent reports proposed that mature monocytes arise from two distinct pathways in the steady state, with monocyte-dendritic progenitors and granulocyte-monocyte progenitors differentiating to monocytes with a ‘DC-like’ state or a ‘neutrophil-like’ state, respectively.⁵⁷ Our results are thus consistent with an MI-induced shift towards the production of ‘granulocyte-like’ monocytes, which appear similar to Ym1⁺Ly6C^hi monocytes that emerge after tissue injury.⁴³ Bulk transcriptome analysis of human monocytes sampled 48 h after acute MI showed upregulation of LCN2, a prototypical marker of the granulocyte-like signature, as well as other granulocyte-associated transcripts (IL1RN, CXCR1),⁵⁸ indicating that a similar shift in circulating monocyte transcriptome might occur in patients with MI. While our data indicate that the macrophage LAM signature is acquired in the ischaemic heart, and not directly induced in blood monocytes by MI, it remains to be determined whether injury-associated Chil3^hi monocytes are primed towards acquiring the LAM signature once infiltrated in the infarcted myocardium.

5. Conclusion

In conclusion, our work provides a novel high-resolution view of the heterogeneity and dynamics of monocyte/macrophage transitions during the acute post-MI inflammation phase and constitutes a valuable resource for further investigating how these cells may be harnessed and modulated to promote post-ischaemic heart repair.

Supplementary material

Supplementary material is available at Cardiovascular Research online.

Authors’ contributions

Performed experiments: G.R., J.G., M.Pi, E.V., A.R., S.R.B., P.A., T.K., N.D., A.P.A.L., K.S., C.C.; Analysed and interpreted data: G.R., J.G., M.Pi, E.V., A.R., S.R.B., P.A., T.K., N.D., A.P.A.L., K.S., C.H., J.S.S., A.Z., A.E.S., C.C.; Created the web-based interface: O.D.; Provided critical mouse lines, assays, and reagents: M.Pr., M.M., K.S., C.H.; Revised and edited the manuscript: J.S.S., M.Pr., M.M., C.H., A.Z.; Designed the study: G.R., A.E.S., C.C.; Wrote the manuscript: G.R., A.E.S., C.C.

Funding

This work was supported by the Interdisciplinary Center for Clinical Research (Interdisziplinäres Zentrum für Klinische Forschung, IZKF), University Hospital Würzburg (E-352 and A-384 to A.Z., E-353 to C.C.), the Deutsche Forschungsgemeinschaft (DFG; German Research Foundation, Projects 374031971-TRR 240, 324392634-TR221, ZE827/13-1, 14-1, project numbers 396923792 to A.Z., 432915089 to A.Z. and C.C., 458539578 and 471705758 to C.C.), the SFB1525 project number 453989101 (projects A1 and B3 to A.Z., project PS2 to A.E.S. and C.C., project A6 to C.C.), and the BMBF within the Comprehensive Heart Failure Centre Würzburg (BMBF 01EO1504 to C.C. and A.E.S.). A.E.S. is supported by the EMBO Young Investigator Program. This work was supported by grants from the Deutsche Forschungsgemeinschaft under Germany’s Excellence Strategy within the framework of the Munich Cluster for Systems Neurology (EXC 2145 SyNergy, ID 390857198 to C.H.) and the Helmholtz-Gemeinschaft Zukunftsthema ‘Immunology and Inflammation’ (ZT-0027 to C.H.).

Data availability

Single-cell RNA-sequencing data generated for this report has been deposited in Gene Expression Omnibus (GSE135310, GSE197441, GSE197853).

References

Frodermann

Nahrendorf

Macrophages and cardiovascular health

Physiol Rev

2018

;

2523

–

2569

Molawi

Wolf

Kandalla

Favret

Hagemeyer

Frenzel

Pinto

Klapproth

Henri

Malissen

Rodewald

Rosenthal

Bajenoff

Prinz

Jung

Sieweke

Progressive replacement of embryo-derived cardiac macrophages with age

J Exp Med

2014

;

211

2151

–

2158

Zaman

Hamidzada

Epelman

Exploring cardiac macrophage heterogeneity in the healthy and diseased myocardium

Curr Opin Immunol

2020

;

–

Dick

Macklin

Nejat

Momen

Clemente-Casares

Althagafi

Chen

Kantores

Hosseinzadeh

Aronoff

Wong

Zaman

Barbu

Besla

Lavine

Razani

Ginhoux

Husain

Cybulsky

Robbins

Epelman

Self-renewing resident cardiac macrophages limit adverse remodeling following myocardial infarction

Nat Immunol

2019

;

–

Bajpai

Bredemeyer

Zaitsev

Koenig

Lokshina

Mohan

Ivey

Hsiao

Weinheimer

Kovacs

Epelman

Artyomov

Kreisel

Lavine

Tissue resident CCR2− and CCR2+ cardiac macrophages differentially orchestrate monocyte recruitment and fate specification following myocardial injury

Circ Res

2019

;

124

263

–

278

Nahrendorf

Swirski

Aikawa

Stangenberg

Wurdinger

Figueiredo

Libby

Weissleder

Pittet

The healing myocardium sequentially mobilizes two monocyte subsets with divergent and complementary functions

J Exp Med

2007

;

204

3037

–

3047

Leuschner

Rauch

Ueno

Gorbatov

Marinelli

Lee

Dutta

Wei

Robbins

Iwamoto

Sena

Chudnovskiy

Panizzi

Keliher

Higgins

Libby

Moskowitz

Pittet

Swirski

Weissleder

Nahrendorf

Rapid monocyte kinetics in acute myocardial infarction are sustained by extramedullary monocytopoiesis

J Exp Med

2012

;

209

123

–

137

Mouton

DeLeon-Pennell

Rivera Gonzalez

Flynn

Freeman

Saucerman

Garrett

Harmancey

Lindsey

Mapping macrophage polarization over the myocardial infarction time continuum

Basic Res Cardiol

2018

;

113

Gessain

Bleriot

Ginhoux

Non-genetic heterogeneity of macrophages in diseases—a medical perspective

Front Cell Dev Biol

2020

;

613116

King

Aguirre

Sun

Roh

Jr,

Kohler

Arlauckas

Iwamoto

Savol

Sadreyev

Kelly

Fitzgibbons

Fitzgerald

Mitchison

Libby

Nahrendorf

Weissleder

IRF3 and type I interferons fuel a fatal response to myocardial infarction

Nat Med

2017

;

1481

–

1487

Calcagno

, Jr.,

Toomu

Zhang

Huang

Aguirre

Weissleder

Daniels

King

The myeloid type I interferon response to myocardial infarction begins in bone marrow and is regulated by Nrf2-activated macrophages

Sci Immunol

2020

;

eaaz1974

Stoeckius

Hafemeister

Stephenson

Houck-Loomis

Chattopadhyay

Swerdlow

Satija

Smibert

Simultaneous epitope and transcriptome measurement in single cells

Nat Methods

2017

;

865

–

868

Chakarov

Lim

Tan

Lim

See

Lum

Zhang

Foo

Nakamizo

Duan

Kong

Gentek

Balachander

Carbajo

Bleriot

Malleret

Tam

JKC

Baig

Shabeer

Toh

SES

Schlitzer

Larbi

Marichal

Malissen

Chen

Poidinger

Kabashima

Bajenoff

Angeli

Ginhoux

Two distinct interstitial macrophage populations coexist across tissues in specific subtissular niches

Science

2019

;

363

eaau0964

Dick

Wong

Hamidzada

Nejat

Nechanitzky

Vohra

Mueller

Zaman

Kantores

Aronoff

Momen

Nechanitzky

Ramachandran

Crome

Becher

Cybulsky

Billia

Keshavjee

Mital

Robbins

Mak

Epelman

Three tissue resident macrophage subsets coexist across organs with conserved origins and life cycles

Sci Immunol

2022

;

eabf7777

10.1101/2020.12.09.417535

Stoeckius

Zheng

Houck-Loomis

Hao

Yeung

Mauck

3rd,

Smibert

Satija

Cell Hashing with barcoded antibodies enables multiplexing and doublet detection for single cell genomics

Genome Biol

2018

;

224

Jaitin

Adlung

Thaiss

Weiner

Descamps

Lundgren

Bleriot

Liu

Deczkowska

Keren-Shaul

David

Zmora

Eldar

Lubezky

Shibolet

Hill

Lazar

Colonna

Ginhoux

Shapiro

Elinav

Amit

Lipid-associated macrophages control metabolic homeostasis in a Trem2-dependent manner

Cell

2019

;

178

686

–

98.e14

Cochain

Vafadarnejad

Arampatzi

Pelisek

Winkels

Ley

Wolf

Saliba

Zernecke

Single-cell RNA-Seq reveals the transcriptional landscape and heterogeneity of aortic macrophages in murine atherosclerosis

Circ Res

2018

;

122

1661

–

1674

Zernecke

Erhard

Weinberger

Schulz

Ley

Saliba

A-E

Cochain

Integrated scRNA-seq analysis identifies conserved transcriptomic features of mononuclear phagocytes in mouse and human atherosclerosis

bioRxiv

2020

. doi:

Crossref

Remmerie

Martens

Thone

Castoldi

Seurinck

Pavie

Roels

Vanneste

De Prijck

Vanhockerhout

Binte Abdul Latib

Devisscher

Hoorens

Bonnardel

Vandamme

Kremer

Borghgraef

Van Vlierberghe

Lippens

Pearce

Saeys

Scott

Osteopontin expression identifies a subset of recruited macrophages distinct from Kupffer cells in the fatty liver

Immunity

2020

;

641

–

57.e14

Daemen

Gainullina

Kalugotla

Chan

Beals

Liss

Klein

Feldstein

Finck

Artyomov

Schilling

Dynamic shifts in the composition of resident and recruited macrophages influence tissue remodeling in NASH

Cell Rep

2021

;

108626

Guilliams

Bonnardel

Haest

Vanderborght

Bujko

Martens

Thoné

Browaeys

De Ponti

Remmerie

Wagner

Vanneste

Zwicker

Vanhalewyn

Gonçalves

Lippens

Devriendt

Cox

Ferrero

Wittamer

Willaert

Kaptein

SJF

Neyts

Dallmeier

Geldhof

Casaert

Deplancke

ten Dijke

Hoorens

Vanlander

Berrevoet

Van Nieuwenhove

Saeys

Saelens

Van Vlierberghe

Devisscher

Scott

Spatial proteogenomics reveals distinct and evolutionarily-conserved hepatic macrophage niches

Cell

2022

;

185

379

–

396.e38

Ramachandran

Dobie

Wilson-Kanamori

Dora

Henderson

BEP

Luu

Portman

Matchett

Brice

Marwick

Taylor

Efremova

Vento-Tormo

Carragher

Kendall

Fallowfield

Harrison

Mole

Wigmore

Newsome

Weston

Iredale

Tacke

Pollard

Ponting

Marioni

Teichmann

Henderson

Resolving the fibrotic niche of human liver cirrhosis at single-cell level

Nature

2019

;

575

512

–

518

Rao

Wang

Guo

Wang

Chen

Yin

Chen

Zhang

Chen

Song

Resolving the intertwining of inflammation and fibrosis in human heart failure at single-cell level

Basic Res Cardiol

2021

;

116

DeBerge

Glinton

Subramanian

Wilsbacher

Rothlin

Tabas

Thorp

Macrophage AXL receptor tyrosine kinase inflames the heart after reperfused myocardial infarction

J Clin Invest

2021

;

131

e139576

Louwe

Olsen

Kaasboll

Yang

Fosshaug

Alfsnes

Ogaard

JDS

Rashidi

Skulberg

Yang

de Miranda Fonseca

Sharma

Aronsen

Schrumpf

Ahmed

Dahl

Nyman

Ueland

Melum

Halvorsen

Bjoras

Attramadal

Sjaastad

Aukrust

Yndestad

Absence of NLRP3 inflammasome in hematopoietic cells reduces adverse remodeling after experimental myocardial infarction

JACC Basic Transl Sci

2020

;

1210

–

1224

Xiao

Wang

J-J

Zhang

J-S

Shen

X-B

Zhang

C-C

J-M

Song

Wang

X-Y

H-Y

Deng

X-N

Z-J

Z-Z

Gao

Zhang

A-H

Feng

Zhang

Y-Y

IL-18 cleavage triggers cardiac inflammation and fibrosis upon beta-adrenergic insult

Eur Heart J

2018

;

–

Vijayan

Wagener

Immenschuh

The macrophage heme-heme oxygenase-1 system and its role in inflammation

Biochem Pharmacol

2018

;

153

159

–

167

Yang

Ming

Functions of arginase isoforms in macrophage inflammatory responses: impact on cardiovascular diseases and metabolic disorders

Front Immunol

2014

;

533

Ferraro

Leoni

Hinkel

Ormanns

Paulin

Ortega-Gomez

Viola

de Jong

Bongiovanni

Bozoglu

Maas

D'Amico

Kessler

Zeller

Hristov

Reutelingsperger

Sager

Doring

Nahrendorf

Kupatt

Soehnlein

Pro-angiogenic macrophage phenotype to promote myocardial repair

J Am Coll Cardiol

2019

;

2990

–

3002

Zaman

Hamidzada

Kantores

Wong

Dick

Wang

Momen

Aronoff

Lin

Razani

Mital

Billia

Lavine

Nejat

Epelman

Selective loss of resident macrophage-derived insulin-like growth factor-1 abolishes adaptive cardiac growth to stress

Immunity

2021

;

2057

–

2071.e6

Kempf

Zarbock

Widera

Butz

Stadtmann

Rossaint

Bolomini-Vittori

Korf-Klingebiel

Napp

Hansen

Kanwischer

Bavendiek

Beutel

Hapke

Sauer

Laudanna

Hogg

Vestweber

Wollert

GDF-15 is an inhibitor of leukocyte integrin activation required for survival after myocardial infarction in mice

Nat Med

2011

;

581

–

588

Wan

Yeap

Dehn

Terry

Novak

Zhang

Iwata

Han

Homma

Drosatos

Lomasney

Engman

Miller

Vaughan

Morrow

Kishore

Thorp

Enhanced efferocytosis of apoptotic cardiomyocytes through myeloid-epithelial-reproductive tyrosine kinase links acute inflammation resolution to cardiac repair after infarction

Circ Res

2013

;

113

1004

–

1012

Kandalam

Basu

Abraham

Wang

Soloway

Jaworski

Oudit

Kassiri

TIMP2 deficiency accelerates adverse post-myocardial infarction remodeling because of enhanced MT1-MMP activity despite lack of MMP2 activation

Circ Res

2010

;

106

796

–

808

Ali

Middleton

Pure

Rader

Apolipoprotein E suppresses the type I inflammatory response in vivo

Circ Res

2005

;

922

–

927

Alonso-Herranz

Sahun-Espanol

Paredes

Gonzalo

Gkontra

Nunez

Clemente

Cedenilla

Villalba-Orero

Inserte

Garcia-Dorado

Arroyo

Ricote

Macrophages promote endothelial-to-mesenchymal transition via MT1-MMP/TGFbeta1 after myocardial infarction

Elife

2020

;

e57920

Farbehi

Patrick

Dorison

Xaymardan

Janbandhu

Wystub-Lis

Nordon

Harvey

Single-cell expression profiling reveals dynamic flux of cardiac stromal, vascular and immune cells in health and injury

Elife

2019

;

e43882

Vafadarnejad

Rizzo

Krampert

Arampatzi

Arias-Loza

Nazzal

Rizakou

Knochenhauer

Reddy Bandi

Nugroho

Schulz

Roesch

Alayrac

Vilar

Silvestre

Zernecke

Saliba

Cochain

Dynamics of cardiac neutrophil diversity in murine myocardial infarction

Circ Res

2020

;

127

e232

–

e249

Mack

Cihak

Simonis

Luckow

Proudfoot

Plachy

Bruhl

Frink

Anders

Vielhauer

Pfirstinger

Stangassinger

Schlondorff

Expression and characterization of the chemokine receptors CCR2 and CCR5 in mice

J Immunol

2001

;

166

4697

–

4704

Schlitzer

Sivakamasundari

Chen

Sumatoh

Schreuder

Lum

Malleret

Zhang

Larbi

Zolezzi

Renia

Poidinger

Naik

Newell

Robson

Ginhoux

Identification of cDC1- and cDC2-committed DC progenitors reveals early lineage priming at the common DC progenitor stage in the bone marrow

Nat Immunol

2015

;

718

–

728

Bosteels

Neyt

Vanheerswynghels

van Helden

Sichien

Debeuf

De Prijck

Bosteels

Vandamme

Martens

Saeys

Louagie

Lesage

Williams

Tang

Mayer

Ronchese

Scott

Hammad

Guilliams

Lambrecht

Inflammatory type 2 cDCs acquire features of cDC1s and macrophages to orchestrate immunity to respiratory virus infection

Immunity

2020

;

1039

–

1056.e9

Goldmann

Wieghofer

Muller

Wolf

Varol

Yona

Brendecke

Kierdorf

Staszewski

Datta

Luedde

Heikenwalder

Jung

Prinz

A new type of microglia gene targeting shows TAK1 to be pivotal in CNS autoimmune inflammation

Nat Neurosci

2013

;

1618

–

1626

Schultze

Mass

Schlitzer

Emerging principles in myelopoiesis at homeostasis and during infection and inflammation

Immunity

2019

;

288

–

301

Ikeda

Asano

Kikuchi

Uchida

Ikegami

Takagi

Yotsumoto

Shibuya

Makino-Okamura

Fukuyama

Watanabe

Ohmuraya

Araki

Nishitai

Tanaka

Emergence of immunoregulatory Ym1(+)Ly6C(hi) monocytes during recovery phase of tissue injury

Sci Immunol

2018

;

eaat0207