Skip main navigation

Dynamics of Cardiac Neutrophil Diversity in Murine Myocardial Infarction

Originally published Research. 2020;127:e232–e249



After myocardial infarction, neutrophils rapidly and massively infiltrate the heart, where they promote both tissue healing and damage.


To characterize the dynamics of circulating and cardiac neutrophil diversity after infarction.

Methods and results:

We employed single-cell transcriptomics combined with cell surface epitope detection by sequencing to investigate temporal neutrophil diversity in the blood and heart after murine myocardial infarction. At day 1, 3, and 5 after infarction, cardiac Ly6G+ (lymphocyte antigen 6G) neutrophils could be delineated into 6 distinct clusters with specific time-dependent patterning and proportions. At day 1, neutrophils were characterized by a gene expression profile proximal to bone marrow neutrophils (Cd177, Lcn2, Fpr1), and putative activity of transcriptional regulators involved in hypoxic response (Hif1a) and emergency granulopoiesis (Cebpb). At 3 and 5 days, 2 major subsets of Siglecfhi (enriched for eg, Icam1 and Tnf) and Siglecflow (Slpi, Ifitm1) neutrophils were found. Cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq) analysis in blood and heart revealed that while circulating neutrophils undergo a process of aging characterized by loss of surface CD62L and upregulation of Cxcr4, heart infiltrating neutrophils acquired a unique SiglecFhi signature. SiglecFhi neutrophils were absent from the bone marrow and spleen, indicating local acquisition of the SiglecFhi signature. Reducing the influx of blood neutrophils by anti-Ly6G treatment increased proportions of cardiac SiglecFhi neutrophils, suggesting accumulation of locally aged neutrophils. Computational analysis of ligand/receptor interactions revealed putative pathways mediating neutrophil to macrophage communication in the myocardium. Finally, SiglecFhi neutrophils were also found in atherosclerotic vessels, revealing that they arise across distinct contexts of cardiovascular inflammation.


Altogether, our data provide a time-resolved census of neutrophil diversity and gene expression dynamics in the mouse blood and ischemic heart at the single-cell level, and reveal a process of local tissue specification of neutrophils in the ischemic heart characterized by the acquisition of a SiglecFhi signature.


After acute myocardial infarction (MI), cardiac ischemic injury triggers a rapid and massive influx of neutrophils in the heart.1–4 Neutrophils have long been considered deleterious in tissue injury contexts given their characteristic functional features such as production of proinflammatory cytokines, release of neutrophil extracellular traps and production of reactive oxygen species (ROS). However, recent evidence indicates that neutrophils may also have essential functions in tissue healing, for example, by promoting angiogenesis.5

In the post-MI heart, neutrophils indeed have contrasted functions promoting both tissue repair and damage,4 and a recent report notably proposed a protective role via modulation of macrophage function.6 Furthermore, a process of temporal neutrophil polarization in the ischemic heart has been suggested, with N1 polarized pro-inflammatory neutrophils infiltrating the heart early after MI (day1), while at days 5 and 7, the proportion of N2 polarized anti-inflammatory neutrophils increases.7 However, the gene expression analysis performed in this report was limited to measurements of few selected markers commonly associated with in vitro M1/M2-polarized macrophages, an arbitrary dichotomy that may be of limited relevance to neutrophils in cardiac inflammation in vivo,8,9 emphasizing the need for a more comprehensive characterization of neutrophil subsets and their dynamics in the ischemic heart. The recent development of single-cell RNA-sequencing (scRNA-seq) in high throughput allows investigating the transcriptome of individual immune cells in disease models and has recently been employed to uncover novel immune cell transcriptional states in MI10,11 or atherosclerosis.12,13 Oligonucleotide-barcoded antibody labeling of cells before their processing for scRNA-seq further allows a multimodal measurement of the expression of cell surface epitopes in addition to transcript expression14,15 and the multiplexing of several biological samples in a single scRNA-seq library.16 These technological advances improve the precision of immune cell subset identification and allow comparing cells from different biological samples with minimal intersample technical bias and cellular doublets.

Here, we employed single-cell transcriptomics combined with cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq)15 of circulating and cardiac neutrophils in a murine model of MI. We investigated the dynamics of neutrophil heterogeneity, and how they adapt to infiltration in the ischemic heart. Combining CITE-seq analysis with flow cytometry validation experiments and in vivo modulation of circulating neutrophil levels, we describe the temporal diversity of neutrophil states in the infarcted heart, and show a local transition of neutrophils toward a SiglecFhi state within the ischemic cardiac tissue.


Detailed methods are available in the Data Supplement.

Data Availability

Data sets that were generated for this report have been deposited in Gene Expression Omnibus (GSE135310). The additional data of atherosclerotic aorta scRNA-seq have been deposited in Gene Expression Omnibus together with data from reference by Cochain et al13 (GSE97310). Other data and analytical methods are available from the corresponding authors on request.

Single-cell gene expression data sets can be browsed in a web-accessible interface:

Figure 1: Neutrophils alone:

Neutrophils and monocytes/macrophages:

Figure 3: Blood and heart neutrophils:


Time-Dependent Transcriptional Heterogeneity of Neutrophils in the Infarcted Heart

Neutrophil infiltration in the infarcted mouse heart peaks around day 1 (41.1±10.63% of CD45+ cells, P=0.0010 versus day 0) to 3 (33.13±5.19% of CD45+ cells, P=0.016 versus day 0) post-MI in mice and then quickly resolves with only minimal neutrophils remaining at days 5 or 7 post-MI (13.35±0.8% and 4.22±1.37% of CD45+, respectively; Online Figure IA, and Nahrendorf et al1 and Cochain et al2). To analyze neutrophil transcriptional heterogeneity during the acute post-MI phase, we performed a multiplexed time series of scRNA-seq analyses combining cell surface epitope labeling and transcriptomics (cell hashing16 and CITE-Seq,15 Methods). MI was induced by permanent ligation of the left anterior descendant coronary artery in male C57BL6/J mice. Surgeries were consistently performed between Zeitgeber Time 2 (ZT2, 2 hours after lights on) and ZT6 to minimize the impact of circadian oscillations on neutrophil gene expression patterns17 (see Methods). Cell suspensions from the heart of mice at 1, 3, and 5 days after MI were labeled with a viability dye and anti-CD11b antibodies coupled to distinct fluorochromes, and viable CD11b+ cells were sorted simultaneously (Figure 1A). Blood-borne neutrophils were excluded from analysis via intravenous injection of anti-CD45.2-APC before sacrifice.13 As expected, we observed 2 major populations of CD64+Ly6G (lymphocyte antigen 6G) monocytes/macrophages (including Ly6C+ [lymphocyte antigen 6C] monocytes) and CD64Ly6G+ neutrophils that also expressed characteristic transcripts of monocytes/macrophages (Fcgr1, Ly6c2) and granulocytes (S100a8), respectively (Figure 1B). Data were demultiplexed in Seurat v3 18,19 (Methods), and the hashing antibody signal allowed backtracking the time point of origin of each single cell (Online Figure IB and IC). We focused our analysis on cells corresponding to neutrophils (monocyte/macrophage data have been analyzed in details in an accompanying report20), representing a total of 1334 individual neutrophils after excluding low-quality cells and doublets (Figure 1C and 1D; Online Figure IB and IC), in which a median of 1365 expressed genes per cell were detected (Online Figure ID). As previously observed by Zilionis et al,21 the number of detected transcripts was much lower in neutrophils compared with other myeloid cells (a median of 4685 genes/cell was detected in monocyte/macrophages).

Figure 1.

Figure 1. Single-cell RNA (scRNA)-seq reveals time-dependent neutrophil transcriptional heterogeneity after myocardial infarction (MI).A, Flowchart of the experimental design (hashtag antibody [HT]); (B) cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq) signal for the indicated surface markers and expression levels (log-transformed expression) of the indicated transcripts projected onto the uniform manifold approximation and projection (UMAP) representation of scRNA-seq gene expression data of the total CD11b+ data set; C and D, UMAP representation of gene expression data in 1334 neutrophils with (C) time point of origin of individual cells color coded and (D) Seurat cluster assignment projected onto the UMAP plot; (E) proportion of each cluster among total neutrophils at each time point; (F) heatmap of the top 5 marker genes (sorted by Log2 fold change) associated with the 6 neutrophil clusters (scaled average expression). G, activity of the indicated regulons (A.U.C. as measured using SCENIC [single-cell regulatory network interference] and clustering) projected onto the UMAP plot. Note that (A) and (B) are shared with Rizzo et al20 and shown here again for clarity. Ly6C indicates lymphocyte antigen 6C; and Ly6G, lymphocyte antigen 6G.

We compared gene expression patterns across all neutrophils at all time points and could delineate 6 transcriptionally distinct cell clusters (Figure 1C through 1F) showing a time-dependent appearance in the ischemic heart (Figure 1C through 1E). As the number of cells sampled for each time point (day 1: 232 cells; day 3: 912 cells; day 5: 190 cells) did not reflect the real levels of neutrophils in the ischemic heart (Online Figure IA), we calculated the proportion that each cluster represented at the different time points (Figure 1E). The vast majority of neutrophils at day 1 (72.8%) segregated in cluster Neutro4 (Cxcl3, Lcn2, Osm, Cd177, Ccl6, Sell, Fpr1), a profile almost absent at later time points (<3% at days 3 and 5). At day 3, Neutro1 cells (43.2% of total neutrophils; Tnf, Icam1, Il23a, Gpr84) were predominant. Cluster Neutro2 (Slpi, Ifitm1, Wfdc17, Asprv1) was present at all time points and its levels gradually increased from 20.7% of cells at day 1 to 34.2% at day 5. Cluster Neutro3 also increased to reach its highest level at day 5 (38.9% of all neutrophils; Rps19, Ltc4s, Nr4a2). Neutro5 represented a minor cluster (3.0%, 14.7% and 4.7% of neutrophils at days 1, 3, and 5, respectively) with a type I interferon response signature (Isg15, Rsad2, Ifit1). Finally, cluster Neutro6 was characterized by highly specific expression of some transcripts (Psap, Slc26a11, Gdf15), and represented 0.0%, 2.1%, and 3.4% of all neutrophils at days 1, 3, and 5 post-MI, respectively (Figure 1C through 1F). Neutro1 and Neutro3 were characterized by a significantly increased proportion of transcripts encoding ribosomal proteins (eg, Rps19, Rpl32; Figure 1F and Online Figure IE, P<0.05 versus all other clusters, Online Table I), with Neutro3 showing the highest level (P<0.0001 versus Neutro1, Online Figure IE, Online Table I). When performing clustering analysis at a lower resolution, Neutro1 and Neutro3 grouped as a single population (not shown), and were further characterized by shared enrichment for expression of, for example, Siglecf, Ppia, or Gngt2 (Online Figure IF). As neutrophils are considered short-lived cells that can be rapidly mobilized from the bone marrow (BM) in response to tissue injury,22,23 we applied a BM proximity score to neutrophils based on the expression of a set of genes previously identified in transcriptomic analysis of neutrophil at different developmental stages24,25 (see Methods). Cluster Neutro4 (day 1 specific) had the highest BM proximity score (P<0.0001 versus all other clusters, Online Table I), followed by Neutro2 (P<0.01 versus all other clusters, Online Table I; Online Figure IE). Recent studies have reported clustering bias in scRNA-seq analysis of immune cells related to enzymatic digestion induced expression of immediate early genes.26–28 Based on the expression of 18 immediate early genes (Methods), we applied an immediate early genes score to each individual cell, which showed no substantial differences across clusters (Online Figure IE).

To validate our findings, we generated and analyzed an independent scRNA-seq experiment where total CD45+ cells were sorted from the heart before and at 1, 3, 5, and 7 days after infarction (Online Figure IIA and IIB), containing 425 cells corresponding to neutrophils in which a median of 870 genes were detected (Online Figure IIC). Three major populations with time-dependent patterning were observed, and termed Day 1, Day3/7-A, and Day3/7-B based on their time of appearance in the ischemic heart (Online Figure IID through IIF). The proportion of ribosomal proteins encoding genes in neutrophils was higher in the Day3/7-A and Day3/7-B clusters compared with Day1 cells with the Day3/7B cluster showing the highest level (P<0.0001 for all comparisons, Online Figure IIE, Online Table I). Day 1 neutrophils had a higher BM proximity score, followed by Day3/7-A (P<0.0001 for all comparisons; Online Figure II E, Online Table I). Day 1 neutrophils were enriched for, for example, Cd177, Cxcl3, Lcn2, and Fpr1 and are thus proximal to cluster Neutro4 in the Figure 1 data set (Online Figure IIG). Day3/7-A were enriched for eg, Ifitm1, S100a11, or Slpi (proximal to Neutro2 in Figure 1), while Day3/7-B were enriched for Siglecf, Icam1, Tnf or Ppia (proximal to Neutro1/3 in Figure 1) (Online Figure II G). Only 6 neutrophils were retrieved from the control condition, consistent with the steady state heart containing only minute amounts of neutrophils. Although they all mapped to cluster Day3/7-B, they did not show enrichment for Siglecf, Icam1, or Tnf (Online Figure IIH).

We also reanalyzed a previously published scRNA-seq data set of cardiac CD45+ cells collected 4 days after MI29 (Online Figure IIIA and IIIB). We could detect 2 major populations of granulocytes (Online Figure IIIA and IIIB): one was enriched for Siglecf, Tnf, and Icam1 expression and had a low BM proximity score, while the second was enriched for Slpi, S100a11, and Ifitm1 and had a higher BM proximity score (P=2.06×1012; Online Figure IIIC and IIID). Ribosomal protein encoding genes were not measured in this data set, as they had been bioinformatically removed from the author-provided gene expression matrix 29 (final median genes detected/cell in total CD45+ cells=500; Online Figure IIIE). These clusters respectively correspond to Neutro1/3 and Neutro2 observed at days 3 and 5 in our main data set (Figure 1). Despite technical limitations (low numbers of cells and genes detected per cell, possible variation caused by batch effects in pooled libraries), results obtained from these 2 additional data sets are consistent with conclusions obtained from our main data set, as the major time-dependent clusters were recovered.

We then sought to identify transcriptional regulators that may modulate gene expression in neutrophils and underlie their temporal heterogeneity and employed SCENIC (single-cell regulatory network interference).30 SCENIC infers activity of gene regulatory networks (regulons) based on co-expression of transcription factors and their putative target genes.30 In cluster Neutro4, we observed preferential activity of regulons corresponding to, for example, Hif1a, Xbp1, and Cebpb. Regulons related to type I interferon response had enriched activity in Neutro5 (eg, Irf7; Figure 1G). Some regulons showed preferential activity in Neutro1 and Neutro3 (eg, Srebf2, Nfkb1, Nfkb2, Bclaf1; Figure 1G).

Altogether, our scRNA-seq analysis of cardiac neutrophils indicate that 3 major time-dependent neutrophil states can be found in the ischemic heart: (1) at day 1, cells with high gene expression proximity to BM neutrophils, high expression of specific inflammatory cytokines and chemokines (eg, Osm, Cxcl3) and activity of regulons involved in hypoxic response (Hif1a) and emergency granulopoiesis (Cebpb); (2) at day 3 onward, cells characterized by an intermediate BM proximity score and expression of a characteristic set of genes (S100a11, Ifitm1, Slpi); (3) at day 3 onward, cells characterized by a low BM proximity score, high proportion of ribosomal protein encoding genes, and enrichment for Siglecf, Tnf, or Icam1. In addition, we observed in our main data set (Figure 1) a minor neutrophil cluster with a characteristic type I interferon response (cluster 5), and a very minor cluster 6.

SiglecFhi Neutrophils Time-Dependently Populate the Infarcted Heart

Neutrophils from days 3 and 5 post-MI (in particular, cells from clusters Neutro1 and Neutro3) displayed a clear enrichment in the expression of Siglecf compared with expression at 1 day after MI (Figure 2A; Online Figure IF). We investigated the presence of SiglecF expressing neutrophils in the infarcted heart by flow cytometry. As the main switch in neutrophil Siglecf expression was evident between days 1 and 3 post-MI when neutrophils were the most abundant in the heart (Figure 2A; Online Figure IA), we focused our analysis on these time points. In mechanically dissociated cardiac tissue (Methods), Ly6G+SiglecFhi cells represented 52.08±3.78% of Ly6G+ neutrophils at day 3 post-MI, whereas only 2.15±0.27% of Ly6G+ neutrophils expressed SiglecF at day 1 (Figure 2B through 2D; Online Figure IVA). Ly6GnegSiglecFhi cells were identified as eosinophils based on SSC (side scatter)/

FSC (forward scatter) properties (Figure 2C). Ly6G+SiglecFlow and Ly6G+SiglecFhi cells were SSCint, supporting the notion that these cells are neutrophils (Figure 2C). No significant infiltration of Ly6G+SiglecFhi neutrophils was found in the control heart or in the heart of sham operated mice at 1 and 3 days after surgery, and in the remote nonischemic myocardium of mice at 1 and 3 days after MI (Online Figure IVB and IVC). We observed similar time dependent patterning of SiglecFhi neutrophils presence in the heart at day 3, but not day 1, after ischemia/reperfusion injury (Online Figure IVD). Immunofluorescence analysis of MI tissue cryosections corroborated our flow cytometry analysis, with abundant SiglecF+ cells found in the infarcted area at day 3, but not day 1, after MI (Figure 2E). We then evaluated whether SiglecFhi and SiglecFlow neutrophils differed in cardinal neutrophil functions, phagocytosis and reactive oxygen species (ROS) production. At day 3 post-MI, SiglecFhi neutrophils displayed a higher ability to phagocytose Ecoli-derived bioparticles (Figure 2F), and higher ROS-production as measured by DHR123 (dihydrorhodamine 123) labeling compared with SiglecFlow neutrophils (Figure 2G). Day 1 total neutrophils displayed a similar level of phagocytosis and ROS-production compared with day 3 SiglecFhi neutrophils (Figure 2F and 2G). Altogether, this indicates that SiglecFhi neutrophils have a higher phagocytic and ROS-production capacity than their SiglecFlow counterparts at day 3 post-MI. Of note, a similar time-dependent presence of SiglecFhi neutrophils was also observed in the heart of female mice at day 3 but not day 1 post-MI, and day 3 SiglecFhi neutrophils also showed higher phagocytosis and ROS-production compared with SiglecFlow neutrophils, indicating that the main features of temporal neutrophil heterogeneity in the infarcted heart are not sex dependent (Online Figure VA through VF). At day 5 after MI in male mice, total neutrophil infiltrate was greatly reduced consistent with resolution of the acute inflammatory phase (Online Figure IVE), and SiglecFhi neutrophils still represented half (51.01±7.89%) of total neutrophils (Online Figure IVE and IVF).

Figure 2.

Figure 2. SiglecFhi neutrophils populate the heart at day 3 post-myocardial infarction (MI) and differentially express aging and activation markers.A, Siglecf log normalized expression in single cells projected onto the uniform manifold approximation and projection (UMAP) plot (split according to time point of origin); (B) SiglecF vs Ly6G (lymphocyte antigen 6G) flow cytometry plots of cardiac cells gated on viable CD45+CD11b+ at 1 and 3 d after MI; (C) FSC (forward scatter)-A/SSC (side scatter)-A signal in gated Ly6GSiglecFhi eosinophils, Ly6G+SiglecFhi, and Ly6G+SiglecFlow neutrophils; (D) neutrophils per milligram of cardiac tissue, proportion of neutrophil among CD45+ leukocytes, and proportion of SiglecFhi cells among Ly6G+ neutrophils in the heart at 1 and 3 d after MI; (E) immunofluorescence staining for SiglecF in cryosections of hearts in the infarcted area at 1 and 3 d after MI, ×200, scale bar 100 µm (isotype control staining was performed on day 3 post-MI hearts); (F) phagocytosis of Ecoli fluorescent bioparticles in neutrophil subsets and (G) reactive oxygen species production (geometric Mean of DHR123 fluorescence); (H) log normalized expression levels of the indicated transcripts projected onto the UMAP plot; (I) representative flow cytometry histogram; and (J) quantitative analysis of geometric mean fluorescence intensity for the indicated markers in neutrophil subsets at 1 and 3 d after MI. Statistical analysis: Mann-Whitney U test (D), Kruskall-Wallis test with Dunn multiple comparison test (F, G, J); indicated P values adjusted for multiple comparisons. All flow cytometry analyses representative of 2 independent experiments. APC indicates allophycocyanin; CXCR2, C-X-C chemokine receptor type 4; CXCR4, C-X-C chemokine receptor type 2; DHR123, dihydrorhodamine 123; FMO, fluorescence minus one; ICAM, intercellular adhesion molecule; and PE, phycoerythrin.

Acquisition of SiglecF Parallels Acquisition of Aging and Activation Markers

Besides Siglecf, several markers previously associated with neutrophil aging or activation23,31,32 showed differential transcript levels according to time or clusters (Figure 2H). The aging/activation markers Itga4 (encoding CD49d), Icam1 (ICAM1 [intercellular adhesion molecule-1]/CD54), and Cxcr4 levels increased over time, while the marker of young neutrophils Sell (encoding CD62L) decreased. Cxcr2 transcript levels decreased at the later time points. We analyzed expression of these markers by flow cytometry. Compared with total day 1 neutrophils and day 3 SiglecFlow neutrophils, SiglecFhi neutrophils had higher ICAM1 and lower CD62L surface expression, and had higher CD49d and CXCR4 (C-X-C chemokine receptor type 4) surface levels compared with day 1 neutrophils. All neutrophils highly expressed CXCR2, and its surface levels were slightly increased in SiglecFhi versus SiglecFlow neutrophils at day 3 (Figure 2I and 2J). At day 5, a similar pattern was observed for expression of ICAM1 and CD49d, with higher expression on SiglecFhi neutrophils (Online Figure IVG and IVH, expression of other epitopes was not measured in this experiment).

Combined with our previous analysis of neutrophil BM proximity, this characterization of neutrophil aging features at the transcript and cell surface marker level suggest that neutrophils in cluster Neutro4 (Figure 1), preponderant at day 1, have features of young neutrophils (ICAM1lowCD49dlowCXCR4lowCD62Lhi), while at days 3 and 5, 2 major populations of old SiglecFhi (ICAM1hiCD49dhiCXCR4hiCD62Llow) and young SiglecFlow neutrophils (ICAM1lowCD49dlowCXCR4lowCD62Lhi) are found that likely correspond to clusters Neutro1/3 and Neutro2, respectively.

The SiglecFhi Neutrophil State Is Acquired via a local tissue specification process

To investigate whether SiglecFhi neutrophils transited via the blood circulation before infiltrating the heart, we performed CITE-seq analysis of viable CD19NK1.1Ter119CD11b+ cells from blood of control mice, sham-operated animals at 1 and 3 days post-surgery, and mice with MI at 1 and 3 days. Cardiac neutrophils from 1 to 3 days post-MI hearts were also analyzed. Samples from all conditions were analyzed together in a cell hashing-based multiplexed experiment to reduce potential batch effects (see Methods and study by Rizzo et al20 for detailed experimental design).

Neutrophils were identified as Ly6G+CD115 cells, showed low levels of Ly6C, and variable levels of CD62L (Figure 3A). Sample demultiplexing allowed identifying experimental condition and tissue of origin (Figure 3B, Online Figure VIA). Based on CITE-seq analysis and transcriptome profiling, we identified 3 major populations in the blood (Figure 3C through 3F): young neutrophils (Blood Young cluster: CD62Lhi, low Cxcr4), old blood neutrophils (Blood Old cluster: CD62Llow, high Cxcr4), type I IFN (interferon) response neutrophils (Neu-IFN), and a minor cluster (<1% of neutrophils in all samples) likely representing neutrophil/platelet conjugates (Neutro/Plt: Pf4, Cd9, Ppbp). Blood Young neutrophils had a higher BM proximity score compared with blood Old neutrophils and Neu-IFN (P<0.0001 for both comparisons, Online Table I), while blood Old neutrophils had higher proportion of ribosomal protein encoding genes (Figure 3D, P<0.0001 versus all other circulating neutrophils, Online Table I). Type I interferon response neutrophils (Neu-IFN) were found at substantial levels already in the steady state blood (23.8% of all neutrophils), consistent with a recent report,33 and their proportion did not increase after sham or MI surgery (Figure 3F), indicating that they are not induced by ischemic injury. Of note, clustering at higher resolution revealed subpopulations with weakly defined gene expression signatures within blood Young neutrophils and type I IFN response neutrophils (Online Figure VIB through VIE). Although these populations were present in all samples, MI or sham surgery at day 1 induced expansion of a subtype of young blood neutrophils with higher BM proximity score (Online Figure VIB through VIE), consistent with transient systemic inflammation induced by MI but also sham surgery, as previously reported,34 and mobilization of BM neutrophils.

Figure 3.

Figure 3. Cell surface epitope detection by sequencing (CITE-seq) analysis of blood to heart neutrophil transitions.A, Uniform manifold approximation and projection (UMAP) plot of single-cell gene expression data of viable Ter119CD3B220CD11b+ cells sorted from the blood and ischemic heart of mice with projection of the CITE-seq signal for the indicated cell surface markers; dashed red line identifies Ly6G+CD115 neutrophils; (B) UMAP plot of the neutrophil data subset with identification of individual cell tissue of origin; (C) neutrophil populations projected onto the UMAP plot (Plt: platelet); (D) bone marrow (BM) proximity score and percent Rpl/Rps genes in the neutrophil clusters (statistics for all multiple comparisons in Online Table I); (E) dot plot showing expression of selected significant marker transcripts of neutrophil populations; (F) proportion of neutrophil populations according to experimental condition and tissue of origin; (G) Siglecf transcript expression and SiglecF CITE-seq signal projected on the neutrophil UMAP plot (top) and displayed as a violin plot (bottom); (H) Violin plot of ICAM1 (intercellular adhesion molecule 1) CITE-seq signal in the neutrophil populations. Note that (A) is shared with reference by Rizzo et al20 and shown here again for clarity. IFN indicates interferon; Ly6C, lymphocyte antigen 6C; and Ly6G, lymphocyte antigen 6G.

We identified 2 heart specific neutrophil populations, with CITE-seq signal for surface SiglecF corroborating our previous findings: at day 1, the predominant heart neutrophil cluster was SiglecFneg (Heart-SiglecFneg cluster), while at day 3, a SiglecF+ population was observed (Heart-SiglecF+ cluster; Figure 3F and 3G). Heart-SiglecFneg cluster had a transcriptome proximal to the day 1 specific cluster (Neutro4, Figure 1), while Heart-SiglecF+ showed clear similarities to Siglecf enriched neutrophils found at days 3 and 5 (Neutro1 and Neutro3, Figure 1) we had observed in our first heart data set (Figure 3E). Of note, while only 47.3% of cells from the Heart-SiglecF+ cluster had detectable Siglecf transcripts, above 70% of these cells were SiglecF+, indicating that the Siglecf transcript may not be fully recovered in droplet-based scRNA-seq (Figure 3G). Although 5.4% to 9.4% of cells in the blood neutrophil clusters had detectable Siglecf transcripts, the proportion of these cells expressing surface SiglecF was actually lower (0.9%–3.3%; Figure 3G). While the CITE-seq signal for ICAM1 was generally poor on neutrophils, it appeared higher in the Heart-SiglecF+ cluster (Figure 3H). Flow cytometry analysis corroborated the absence of substantial levels of SiglecFhi neutrophils not only in the blood, but also in the BM or spleen of MI and sham-operated mice at 1 and 3 days post-surgery (Online Figure VIIA and VIIB). To verify absence of intracellular SiglecF protein in neutrophils that would not be detected by CITE-seq or surface labeling, we labeled neutrophils after permeabilization, but could not detect a SiglecFhi neutrophil population in the BM or blood in control mice and at 1 and 3 days after MI, indicating that acquisition of the SiglecFhi state is specific to the injured heart (Online Figure VIIC). ICAM1 expression on blood, spleen, and BM neutrophils also did not change after MI (Online Figure VIID). Altogether, these data show that the SiglecFhi state is acquired within the ischemic heart tissue, and not at upstream sites, and represents a tissue differentiated neutrophil state.

Relation of Heart-Infiltrating Neutrophils to Blood, BM, and Spleen Neutrophils

To better understand the relationship between blood neutrophils and their heart infiltrating counterparts, we integrated our heart-specific data set (Figure 1) with the combined heart and blood data set (Figure 3) in Seurat v319 (Figure 4A and 4B). Of note, heart-infiltrating cells from the Figure 1 data set had been collected after exclusion of circulating leukocytes (based on intravenous injection of a fluorescently labeled anti-CD45.2 antibody before sacrifice, Methods), ensuring that these cells were truly infiltrated in the heart. Uniform manifold approximation and projection (UMAP) representation of the integrated data outlined 2 major group of cells according to tissue of origin (Figure 4B and 4C), in which we examined the coordinates of heart infiltrating neutrophil clusters (Figure 4C and 4D). The majority of the Neutro2 cluster (78.9%) mapped to coordinates indicating a blood gene expression signature (Figure 4D), while the day 1 post-MI specific Neutro4 (76.6%), and the Siglecf enriched clusters Neutro1 (96.1%) and Neutro3 (92.9%), mapped to coordinates indicating a heart gene expression signature (Figure 4D). These results indicate that the Neutro2 cluster represents neutrophils freshly infiltrated from the circulation. As cells corresponding to the BM-proximal cardiac Neutro4 cluster were scarcely observed in the blood, we analyzed whether neutrophils with a similar transcriptional signature could be found in hematopoietic organs and analyzed single-cell RNA-seq data of steady-state BM cells from the Tabula Muris35 (Online Figure VIIIA and VIIIB) and from steady-state splenocytes (Online Figure VIIIC through VIIIF). Both in the BM and in the spleen, we observed neutrophils enriched for transcripts characteristic of Neutro4 (notably, Cd177 or Fpr1).

Figure 4.

Figure 4. Neutrophil differentiation trajectories in the blood and ischemic heart.A, Overview of the data analysis strategy: the data sets presented in Figure 1 and Figure 3 were integrated in Seurat; (B) uniform manifold approximation and projection (UMAP) plot of the resulting integrated data set split according to experiment of origin (top) and tissue of origin (bottom) to (C) determine where cells with a genes expression signature of blood and heart infiltrating neutrophils mapped in the UMAP dimensionally reduced gene expression space; (D) projection of the heart specific clusters from the Figure 1 data set in the integrated data set; (E) pseudotime ordering of neutrophils in Monocle split according to cardiac neutrophil cluster; and (F) heatmap of gene expression variation of the indicated transcripts according to pseudotime (Figure 1; heart only data set); (G) RNA-velocity analysis in the Figure 1 heart only data set; (H and I) pseudotime analysis of blood and heart neutrophils (Figure 3 data set) split according to (H) tissue of origin and (I) cluster; (J) expression of the indicated transcripts projected onto the pseudotime tree. IFN indicates interferon.

Trajectory Inference Analysis Reveals Neutrophil Tissue Specification Trajectories

Altogether, our observations clearly indicate that SiglecFlow circulating neutrophils infiltrate the heart, where they become SiglecFhi and acquire a specific gene expression profile in a process of tissue specification. Our data further indicate that neutrophils observed at day 1 in the heart have a transcriptome proximal to medullar neutrophils. Based on these assumptions, we performed trajectory inference analysis of cardiac neutrophil tissue specification using pseudotemporal ordering of single cells in Monocle.36 In the heart-only data, the BM-proximal Neutro4 cluster was at the start of the pseudotime ordering, followed by the Siglecfneg Neutro2 cluster, while Siglecfhi clusters (Neutro1 and Neutro3) were at its end (Figure 4E). Neutro5 (Type I IFN signature) did not map to particular pseudotimes, consistent with the type I IFN signature being orthogonal to the neutrophil tissue specification process.37 Gene expression in pseudotime showed decreasing expression of, for example, Cd177, Lcn2, Retnlg or Sell, and acquisition of Icam1, Tnf, Rps19, or Siglecf (Figure 4F). In this data set, RNA-velocity analysis38,39 showed a global directionality from Neutro4 to Neutro2 to Neutro1/3 (Figure 4G). We also performed pseudotime analysis of the blood to heart transition. This analysis further suggested that acquisition of the SiglecFhi state represents a local aging trajectory distinct from neutrophil aging in the blood, while cardiac specific SiglecFneg neutrophils from the day 1 time point segregated in a branch of the pseudotime tree characterized by high Cd177 expression (Figure 4H through 4J).

Anti-Ly6G Antibodies Only Partially Deplete Heart Neutrophils and Induce a Local Shift Toward the SiglecFhi Phenotype

We next set out to time-dependently deplete neutrophils to evaluate the functional consequences of their time-dependent heterogeneity. Mice were treated with repeated injections of anti-Ly6G antibodies, a widely employed strategy previously used to investigate the role of neutrophils in post-MI cardiac repair.6 To unequivocally validate neutrophil depletion, we first devised a strategy of Ly6G-independent detection of blood and cardiac neutrophils to overcome masking of the Ly6G epitope induced by depleting antibodies40 and established that gating neutrophils as CD11b+CD64negCXCR2+ in the heart, and CD11b+CD115negCXCR2+ in the blood, was as efficient and specific as gating neutrophils on Ly6G expression (Online Figure IXA and IXB). In the BM, CXCR2-based gating was less accurate, likely because of varying expression levels of CXCR2 in neutrophils at different stages of maturation (Online Figure IXC). Nevertheless, Ly6G+Ly6Cint neutrophils were adequately captured within CD11b+CD115neg cells (Online Figure IXC and IXD). For consistency with previously published data, we employed the same depletion strategy as Horckmans et al,6 where mice received daily 50 µg intraperitoneal doses of the anti-Ly6G monoclonal antibody clone 1A8 starting 1 day before the induction of MI. We sacrificed mice at day 3 post-MI and analyzed neutrophil proportions in the blood and heart (Figure 5A). While we could not observe any CD115negLy6G+ or CD64negLy6G+ cells in the blood and heart of anti-Ly6G treated mice, respectively, CD64negCXCR2+ neutrophils were detected and especially abundant in the ischemic heart, indicating that measuring neutrophils levels based on Ly6G expression drastically overestimates depletion efficiency in anti-Ly6G-treated mice. The proportion and absolute counts of CXCR2 gated neutrophils was reduced by ≈90% in the blood (P=0.0007; Figure 5B and 5C) but only ≈50% in the heart of anti-Ly6G treated mice (Figure 5D through 5F). Depletion in the BM appeared inefficient (Online Figure IXC and IXD). Interestingly, anti-Ly6G induced a significant shift toward higher SiglecFhi neutrophil proportions in the ischemic heart (P=0.0027; Figure 5G). Counts of infiltrated SiglecFhi and SiglecFlow neutrophils were reduced by 30% (P=0.02), and 72% (P=0.0027), respectively (Figure 5H). We used a similar design with sacrifice of the mice at day 1 after MI and also observed a stronger reduction in neutrophil levels in the blood (−73.3%, P=0.0079) than in the heart (−38.2%, P=0.0079; Figure 5I). Similar to previous observations, at day 1 post-MI neutrophils were SiglecFneg in both groups (not shown). Levels of mature CXCR2+ neutrophils were not significantly affected in the BM of anti-Ly6G treated mice 1 day after MI (Figure 5I), in line with the notion that anti-Ly6G antibodies do not efficiently deplete rapidly renewing BM neutrophils.40 Altogether, these results show that anti-Ly6G treatment failed to entirely counter initial neutrophil influx in the heart after MI, but rather efficiently reduced blood neutrophil levels at the time points analyzed, likely leading to reduced influx of fresh SiglecFlow neutrophils to the infarcted heart, and a shift toward locally aged SiglecFhi cells. These results corroborate the notion of SiglecFhi neutrophils representing a locally differentiated state arising from SiglecFlow neutrophils.

Figure 5.

Figure 5. Anti-Ly6G (lymphocyte antigen 6G) treatment induces partial cardiac neutrophil depletion and a shift toward SiglecFhi neutrophils in the heart.A, Summary of the anti-Ly6G treatment regimen; (B) representative flow cytometry plots of CD115 vs Ly6G and CD115 vs CXCR2 (C-X-C chemokine receptor type 2) in the blood (pregated on viable CD45+CD11b+ cells); (C) proportions within CD45+ leukocytes and absolute counts of CXCR2-gated neutrophils in the blood; (D) representative flow cytometry plots of CD64 vs Ly6G and CD64 vs CXCR2 in the heart (pregated on viable CD45+CD11b+ cells); (E) proportions within CD45+ leukocytes and (F) absolute counts of CXCR2-gated neutrophils in the heart; (G) proportions of SiglecFhi cells among total cardiac neutrophils; and (H) counts of SiglecFhi and SiglecFlow neutrophils in the heart. I, Proportions of CXCR2 gated neutrophils in the bone marrow (BM), blood, and heart at 1 d after myocardial infarction (MI) in mice undergoing a similar treatment regimen. All statistical analyses: Mann-Whitney U test. All analyses representative of 2 independent experiments, except (F) and (H): 1 experiment.

NicheNet Analysis of Neutrophil to Macrophage Communication Pathways

It has been previously proposed that neutrophils may impact postischemic cardiac remodeling by inducing a pro-repair phenotype in macrophages.6 To computationally investigate putative communication pathways between neutrophils and macrophages, we employed NicheNet, a recently developed software that predicts ligand/receptor intercellular communication and its effect on gene expression in target cells.41 In the total CD11b+ cells data set (including not only Ly6G+ neutrophils but also Ly6ClowCD64hi macrophages; see Figure 1A and 1B), we defined neutrophils as sender cells (expressing ligands) and macrophages as receiver cells (expressing receptors; Figure 6A). NicheNet identified several experimentally validated ligand-receptor pairs (eg, Csf1/Csf1r; Vegfa/Nrp1 and Nrp2; Tnf and various receptors; Figure 6B). Ligand activity ranking pinpointed Il1b, Tnf and Csf1 as the neutrophil expressed ligands with the highest putative impact on macrophage gene expression (Figure 6C), and numerous target genes in macrophages that may be induced by neutrophil-expressed ligands (Figure 6D). Some ligands showed enrichment in time-specific neutrophil clusters (eg, Tnf in cluster Neutro1/Neutro3 predominant at days 3 and 5 post-MI, C3 and Ccl4 in cluster Neutro4 predominant at day 1), indicating time-dependent neutrophil/macrophage cross-talk (Figure 6E). Some ligands/receptors pairs showed reciprocally segregated expression of ligands in neutrophils (C3, Sema4d, and Csf1) and their receptors in macrophages (C3ar1, Plxnb2, and Csf1r, respectively; Figure 6F).

Figure 6.

Figure 6. Analysis of neutrophil to macrophage communication pathways in the infarcted heart.A, Uniform manifold approximation and projection (UMAP) plot of total cardiac CD11b+ cells from days 1, 3, and 5 after myocardial infarction (MI) with identification of major cell lineages; (B) NicheNet ligand-receptor interaction matrix (restricted to bona fide ligand-receptor interactions documented in literature and publicly available databases) as analyzed with neutrophils as sender cells and macrophages as receiver cells; (C) NicheNet ligand activity (Pearson correlation coefficient); (D) NicheNet ligand-target matrix; (E) violin plot showing log normalized expression of the top 18 neutrophil ligands in the 6 cardiac neutrophil clusters (see Figure 1); (F) expression of the indicated ligands and their respective receptors projected onto the UMAP plot of total CD11b+ cells. Note that (A) is shared with reference by Rizzo et al20 and shown here again for clarity. cDC2 indicates type 2 classical dendritic cells; IFNIC, interferon inducible cells; and Ly6C, lymphocyte antigen 6C.

scRNA-Seq of Atherosclerotic Aortas Reveals Two Distinct Neutrophil Subsets

To evaluate whether similar neutrophil diversity could be observed in a distinct and rather chronic cardiovascular disease context, we analyzed scRNA-seq data of CD45+ cells isolated from control (375 cells) and atherosclerotic (1723 cells) aortas of low densitiy lipoprotein receptor deficient (Ldlr/−) mice (corresponding to a previously published data set combined with novel data, see study by Cochain et al,13 and Methods; Figure 7A). Two distinct neutrophil clusters were observed, with the Neutro-1 cluster originating preferentially from the atherosclerotic samples (34 out of 37 cells), while Neutro-2 originated equally from control (13 cells) and diseased (12 cells) aortas (Figure 7A). Cells expressing Siglecf, Icam1, Cxcr4, or Mrpl52 were observed in Neutro-1, while Neutro-2 cells expressed Retnlg, Wfdc21 or Mmp8 (Figure 7B). We next performed flow cytometry analysis of vascular tissue neutrophils, with exclusion of potential circulating neutrophil contamination by labeling them via intravenous CD45.2 antibody injection (see Methods). This revealed increased proportions of neutrophils in the aorta and aortic sinus of atherosclerotic Ldlr/− mice compared with chow fed control (Figure 7C and 7D). While no SiglecF expressing neutrophils were observed in control vessels, 22.4% and 38.2% of neutrophils were SiglecFhi in atherosclerotic aortas and aortic sinuses, respectively (Figure 7C and 7D). Altogether, these results demonstrate that locally aged SiglecFhi neutrophils also emerge in chronic vascular inflammation in atherosclerosis.

Figure 7.

Figure 7. Single-cell RNA (scRNA)-seq of control and atherosclerotic aortas reveals 2 distinct neutrophil subsets. scRNA-seq was performed in a total of 2098 CD45+ cells from control (375 cells) and atherosclerotic (1723 cells) aortas and major immune cell lineages were identified. A, Uniform manifold approximation and projection (UMAP) representation of scRNA-seq gene expression data and clustering analysis, with a focus on neutrophils (62 cells), and their experimental condition of origin in control and atherosclerotic aortas from Ldlr−/− mice; (B) expression of the indicated transcripts projected onto the UMAP plot; (C) representative flow cytometry plots for detection of CD11b+Ly6G+ (pregated on tissue viable total CD45+ leukocytes) neutrophils and SiglecF expression on neutrophils extracted from the aorta or aortic sinus of control (Chow) or high fat diet fed (HFD) Ldlr−/ mice; (D) proportions of neutrophils among total CD45+ leukocytes (left), and of SiglecFhi cells among neutrophils (right) in the aorta and aortic sinus of chow or HFD fed Ldlr−/− mice; statistical analysis: Mann-Whitney U test. Flow cytometry data from 1 experiment. CXCR4 indicates C-X-C chemokine receptor type 4; ICAM1, intercellular adhesion molecule 1; and SiglecF, Sialic acid-binding Ig-like lectin F.


Here, we demonstrate the time-dependent presence of distinct neutrophil states defined by discrete transcriptional and cell surface protein expression profiles in the infarcted mouse heart. By analyzing the blood-to-heart neutrophil transition at the single-cell level, we further show that neutrophils follow distinct aging trajectories in the circulation and in the ischemic heart and acquire a tissue-restricted SiglecFhi state in the infarcted myocardium via a local process of tissue specification.

At 24 hours after the onset of MI, neutrophils with a transcriptional state reminiscent of newly produced medullar and splenic neutrophils (Cd177, Fpr1, Mmp8, Lcn2) infiltrated the heart. At 3 days onward, we observed 2 main neutrophil populations in the heart, characterized by a SiglecFhi and a SiglecFlow state. Integration of scRNA-seq data of neutrophils obtained (1) specifically from the heart and (2) simultaneously from the heart and blood indicated that the SiglecFlow state represents young blood neutrophils (CD62Lhi, Slpi, Wfdc21, Retnlg). We observed an additional subset of neutrophils characterized by a typical type I interferon response (Isg15, Irf7). Although these neutrophils were abundant in the blood, only low levels were observed in the ischemic heart. Importantly, these neutrophils were observed already in the steady state, consistent with a report by Xie et al.33 Type I IFN response neutrophils did not seem to be induced by MI, in contrast to a recently proposed emergence of type I IFN response neutrophils induced by cardiac ischemia.37 Using SCENIC inference analysis,30 we identified putative activity of specific transcriptional regulators in the neutrophil subsets such as Cebpb, Hif1a, or Srebf2. Whether these transcription factors regulate neutrophil acquisition of specific gene expression profiles and function will need to be addressed in future studies.

Aging of neutrophils in the circulation and its impact on neutrophil functional capacities and trafficking has been extensively documented.23,42,43 Analysis of neutrophils from the blood and heart by CITE-seq and flow cytometry allowed investigating compartment specific aging and specification of neutrophils. While cardiac SiglecFhi neutrophils shared some characteristic of the circulating aged neutrophils signature (low CD62L, high CXCR4, expression of specific transcripts, enrichment of ribosomal protein encoding genes), they appeared to follow a tissue restricted specification pathway characterized by acquisition of surface SiglecF and ICAM1, and enrichment for specific transcripts (Snrpe, Cox7a2). Differentiation trajectory inference by RNA-velocity38 and pseudotime ordering in Monocle36 corroborated the notion that neutrophils undergo a specific aging process within the ischemic myocardium, and that it differs from that occurring in the circulation. However, events of gene expression regulation and transitions occurring at upstream sites may to some extent be transferred to the inflamed heart upon neutrophil recruitment. Thus, new methods able to capture the true immediate dynamics of transcription such as single-cell thiol-(SH)-linked alkylation of RNA for scSLAM-seq (metabolic labeling sequencing)44 may be better suited to investigate in situ phenotypic modulation of neutrophils in the ischemic heart.

The various cardiac neutrophil subsets appeared endowed with distinct functional capacities. At day 3 post-MI, SiglecFhi neutrophils showed higher phagocytosis and ROS-production than their SiglecFlow counterparts. Higher ROS-production by SiglecFhi neutrophils is consistent with previous observations in lung cancer.45 Day 1 neutrophils showed a phagocytic and ROS-production ability similar to day 3 SiglecFhi neutrophils. The functional implication of the local acquisition of surface SiglecF remains to be elucidated. As SiglecF ligation has been proposed to induce apoptosis in eosinophils46,47 and its upregulation in neutrophils coincides with resolution of their infiltration in the ischemic heart, acquisition of SiglecF expression in neutrophils could be associated with increased neutrophil apoptosis and would constitute an active process promoting the resolution of inflammation. Whether SiglecF expression in murine neutrophils in disease states is of any relevance regarding neutrophil biology in other species is unclear. SiglecF is considered a murine functional paralog of human Siglec-8.48 In humans, Siglec-8 is expressed on circulating eosinophils, and to a lesser extent on mast cells and basophils.48 Regardless of these considerations, SiglecF expression could constitute a valuable tool to track neutrophil specification within diseased tissues in murine models. Cardiac SiglecFhi neutrophils also expressed high surface ICAM1. ICAM1 expression can be induced by neutrophil activators (lipopolysaccharide, TNFα [tumor necrosis factor α], zymosan) and can drive effector functions such as ROS production and phagocytosis, placing it as a functional marker of neutrophil activation.31 Consistent with these observations, we noted higher phagocytosis and ROS-production in SiglecFhiICAM1hi compared with SiglecFlowICAM1low neutrophils at day 3 post-MI.

Nevertheless, the functional consequences of neutrophil temporal heterogeneity for the cardiac repair process remain to be determined. We attempted to address this issue using a widely employed model of neutrophil depletion based on intraperitoneal injection of the 1A8 monoclonal antibody.40 However, using a Ly6G-independent neutrophil detection strategy, we determined that this failed to fully restrict the initial neutrophil influx into the heart but instead induced a shift toward higher proportion of cardiac SiglecFhi neutrophils. As blood neutrophils appeared to be depleted rather efficiently, this further corroborates the notion of SiglecFhi neutrophils being locally aged. These results also suggest that the functional effects of anti-Ly6G antibodies on cardiac repair in vivo may not only be related to reduced neutrophil levels as previously suggested 6, but might also be caused by a phenotypic shift in neutrophil populations.40 Elucidating the role of neutrophil subsets in the heart will require using more efficient depletion models such as Ly6G-DTR (diphtheria toxin receptor) mice, or targeting of specific pathways using Ly6g-cre or S100a8-cre mice.49 Importantly, the main features of time-dependent neutrophil heterogeneity in the ischemic heart were also observed in female mice. However, uncovering fine sex-specific differences in neutrophil phenotype and function50 during ischemic heart repair will require more direct and comprehensive comparisons in future studies.

Neutrophil temporal heterogeneity in the heart was also associated with increased expression of transcripts that may orchestrate the inflammatory response. In particular, we observed increased expression of Tnf, which may act on numerous immune and nonimmune cells (fibroblasts, cardiomyocytes, endothelial cells) in the ischemic heart, as well as genes that may mediate crosstalk of neutrophils with other immune cell types such as IL23R-expressing T cells (Il23a).51 Using the NicheNet software41 to predict ligand/receptor interaction and their impact on target cells, we identified pathways by which neutrophils may affect macrophage phenotype, such as C3/C3ar1, Sema4d/Plxnd2, and Csf1/Csf1r. Of note, ligand/receptor analysis in scRNA-seq data is inherently limited by the efficiency of transcript detection, and may not detect all potential interactions. For instance, neutrophil-derived Lcn2 (NGAL [neutrophil gelatinase-associated lipocalin]) has previously been proposed to drive macrophage reparative functions in the heart but transcripts for its putative receptor (encoded by Slc22a17) were poorly detected in macrophages (not shown).

A recent scRNA-seq study in human and mouse lung cancer has evidenced conserved neutrophil transcriptomic signatures across species in disease.21 Notably, neutrophils with a strong type I interferon response were found across species.21 Although the cancerous lung and the ischemic myocardium represent 2 distinct environments, the study by Zilionis et al21 provides proof of concept that disease-associated neutrophil gene expression signatures can be conserved from mice to humans and raises the possibility that neutrophils similar to the ones we observed in our study may also populate the human ischemic heart. Elucidating the conservation of neutrophil states between the murine and human ischemic heart will be of critical importance to estimate the relevance of basic and preclinical research on acute inflammatory processes after MI in murine experimental models.

In murine atherosclerotic aortas, we observed 2 broad neutrophil clusters with gene expression features reminiscent of the major cardiac clusters (ie, SiglecFhi and SiglecFlow), suggesting that these populations may be conserved in other cardiovascular inflammation contexts. The functional role of neutrophil heterogeneity in atherosclerosis and potential tissue-specific aspects of neutrophil diversity will need to be addressed in future studies.

In summary, our work provides a high-resolution, dynamic census of blood and cardiac neutrophil heterogeneity after MI and describes the process of neutrophil tissue specification in the ischemic heart. This data set may constitute a valuable resource for further investigating the functional implications of neutrophil temporal heterogeneity in the infarcted myocardium, and how it may affect ischemic heart repair.

Nonstandard Abbreviations and Acronyms


bone marrow


cellular indexing of transcriptomes and epitopes by sequencing


C-X-C chemokine receptor type 2


C-X-C chemokine receptor type 4


diphtheria toxin receptor


intercellular adhesion molecule 1




lymphocyte antigen 6C


lymphocyte antigen 6G


myocardial Infarction


neutrophil gelatinase-associated lipocalin


reactive oxygen species


single-cell RNA-sequencing


sialic acid-binding Ig-like lectin F


tumor necrosis factor α



Supplemental Materials

Expanded Materials & Methods

Online Figures I–IX

Online Table I


*E.V., G.R., and L.K. contributed equally to this article.

The Data Supplement is available with this article at

For Sources of Funding and Disclosures, see page e247.

Correspondence to: Clément Cochain, PhD, Comprehensive Heart Failure Center Wuerzburg, University Hospital Wuerzburg, Wuerzburg 97078, Germany, Email
Antoine-Emmanuel Saliba, PhD, Helmholtz Institute for RNA-based Infection Research (HIRI), Helmholtz-Center for Infection Research (HZI), Wuerzburg 97080, Germany, Email


  • 1. Nahrendorf M, Swirski FK, Aikawa E, Stangenberg L, Wurdinger T, Figueiredo JL, Libby P, Weissleder R, Pittet MJ. The healing myocardium sequentially mobilizes two monocyte subsets with divergent and complementary functions.J Exp Med. 2007; 204:3037–3047. doi: 10.1084/jem.20070885CrossrefMedlineGoogle Scholar
  • 2. Cochain C, Auvynet C, Poupel L, Vilar J, Dumeau E, Richart A, Récalde A, Zouggari Y, Yin KY, Bruneval P, et al. The chemokine decoy receptor D6 prevents excessive inflammation and adverse ventricular remodeling after myocardial infarction.Arterioscler Thromb Vasc Biol. 2012; 32:2206–2213. doi: 10.1161/ATVBAHA.112.254409LinkGoogle Scholar
  • 3. Zlatanova I, Pinto C, Silvestre JS. Immune modulation of cardiac repair and regeneration: the art of mending broken hearts.Front Cardiovasc Med. 2016; 3:40. doi: 10.3389/fcvm.2016.00040CrossrefMedlineGoogle Scholar
  • 4. Puhl SL, Steffens S. Neutrophils in post-myocardial infarction inflammation: damage vs. resolution?Front Cardiovasc Med. 2019; 6:25. doi: 10.3389/fcvm.2019.00025CrossrefMedlineGoogle Scholar
  • 5. Phillipson M, Kubes P. The healing power of neutrophils.Trends Immunol. 2019; 40:635–647. doi: 10.1016/ Scholar
  • 6. Horckmans M, Ring L, Duchene J, Santovito D, Schloss MJ, Drechsler M, Weber C, Soehnlein O, Steffens S. Neutrophils orchestrate post-myocardial infarction healing by polarizing macrophages towards a reparative phenotype.Eur Heart J. 2017; 38:187–197. doi: 10.1093/eurheartj/ehw002MedlineGoogle Scholar
  • 7. Ma Y, Yabluchanskiy A, Iyer RP, Cannon PL, Flynn ER, Jung M, Henry J, Cates CA, Deleon-Pennell KY, Lindsey ML. Temporal neutrophil polarization following myocardial infarction.Cardiovasc Res. 2016; 110:51–61. doi: 10.1093/cvr/cvw024CrossrefMedlineGoogle Scholar
  • 8. Nahrendorf M, Swirski FK. Abandoning M1/M2 for a network model of macrophage function.Circ Res. 2016; 119:414–417. doi: 10.1161/CIRCRESAHA.116.309194LinkGoogle Scholar
  • 9. Bassler K, Schulte-Schrepping J, Warnat-Herresthal S, Aschenbrenner AC, Schultze JL. The myeloid cell compartment-cell by cell.Annu Rev Immunol. 2019; 37:269–293. doi: 10.1146/annurev-immunol-042718-041728CrossrefMedlineGoogle Scholar
  • 10. Dick SA, Macklin JA, Nejat S, Momen A, Clemente-Casares X, Althagafi MG, Chen J, Kantores C, Hosseinzadeh S, Aronoff L, et al. Self-renewing resident cardiac macrophages limit adverse remodeling following myocardial infarction.Nat Immunol. 2019; 20:29–39. doi: 10.1038/s41590-018-0272-2CrossrefMedlineGoogle Scholar
  • 11. Farbehi N, Patrick R, Dorison A, Xaymardan M, Janbandhu V, Wystub-Lis K, Ho JW, Nordon RE, Harvey RP.Single-cell expression profiling reveals dynamic flux of cardiac stromal, vascular and immune cells in health and injury.Elife. 2019; 8:e43882. doi: 10.7554/eLife.43882CrossrefMedlineGoogle Scholar
  • 12. Kim K, Shim D, Lee JS, Zaitsev K, Williams JW, Kim KW, Jang MY, Seok Jang H, Yun TJ, Lee SH, et al. Transcriptome analysis reveals nonfoamy rather than foamy plaque macrophages are proinflammatory in atherosclerotic murine models.Circ Res. 2018; 123:1127–1142. doi: 10.1161/CIRCRESAHA.118.312804LinkGoogle Scholar
  • 13. Cochain C, Vafadarnejad E, Arampatzi P, Pelisek J, Winkels H, Ley K, Wolf D, Saliba AE, Zernecke A. Single-cell RNA-seq reveals the transcriptional landscape and heterogeneity of aortic macrophages in murine atherosclerosis.Circ Res. 2018; 122:1661–1674. doi: 10.1161/CIRCRESAHA.117.312509LinkGoogle Scholar
  • 14. Peterson VM, Zhang KX, Kumar N, Wong J, Li L, Wilson DC, Moore R, McClanahan TK, Sadekova S, Klappenbach JA. Multiplexed quantification of proteins and transcripts in single cells.Nat Biotechnol. 2017; 35:936–939. doi: 10.1038/nbt.3973CrossrefMedlineGoogle Scholar
  • 15. Stoeckius M, Hafemeister C, Stephenson W, Houck-Loomis B, Chattopadhyay PK, Swerdlow H, Satija R, Smibert P. Simultaneous epitope and transcriptome measurement in single cells.Nat Methods. 2017; 14:865–868. doi: 10.1038/nmeth.4380CrossrefMedlineGoogle Scholar
  • 16. Stoeckius M, Zheng S, Houck-Loomis B, Hao S, Yeung BZ, Mauck WM, Smibert P, Satija R. Cell hashing with barcoded antibodies enables multiplexing and doublet detection for single cell genomics.Genome Biol. 2018; 19:224. doi: 10.1186/s13059-018-1603-1CrossrefMedlineGoogle Scholar
  • 17. Schloss MJ, Horckmans M, Nitz K, Duchene J, Drechsler M, Bidzhekov K, Scheiermann C, Weber C, Soehnlein O, Steffens S. The time-of-day of myocardial infarction onset affects healing through oscillations in cardiac neutrophil recruitment.EMBO Mol Med. 2016; 8:937–948. doi: 10.15252/emmm.201506083CrossrefMedlineGoogle Scholar
  • 18. Satija R, Farrell JA, Gennert D, Schier AF, Regev A. Spatial reconstruction of single-cell gene expression data.Nat Biotechnol. 2015; 33:495–502. doi: 10.1038/nbt.3192CrossrefMedlineGoogle Scholar
  • 19. Stuart T, Butler A, Hoffman P, Hafemeister C, Papalexi E, Mauck WM, Hao Y, Stoeckius M, Smibert P, Satija R. Comprehensive integration of single-cell data.Cell. 2019; 177:1888–1902.e21. doi: 10.1016/j.cell.2019.05.031CrossrefMedlineGoogle Scholar
  • 20. Rizzo G, Vafadarnejad E, Arampatzi P, Silvestre JS, Zernecke A, Saliba AE, Cochain C. Single-cell transcriptomic profiling maps monocyte/macrophage transitions after myocardial infarction in mice.bioRxiv. 2020. Scholar
  • 21. Zilionis R, Engblom C, Pfirschke C, Savova V, Zemmour D, Saatcioglu HD, Krishnan I, Maroni G, Meyerovitz CV, Kerwin CM, et al. Single-cell transcriptomics of human and mouse lung cancers reveals conserved myeloid populations across individuals and species.Immunity. 2019; 50:1317–1334.e10. doi: 10.1016/j.immuni.2019.03.009CrossrefMedlineGoogle Scholar
  • 22. Ng LG, Ostuni R, Hidalgo A. Heterogeneity of neutrophils.Nat Rev Immunol. 2019; 19:255–265. doi: 10.1038/s41577-019-0141-8CrossrefMedlineGoogle Scholar
  • 23. Hidalgo A, Chilvers ER, Summers C, Koenderman L. The neutrophil life cycle.Trends Immunol. 2019; 40:584–597. doi: 10.1016/ Scholar
  • 24. Evrard M, Kwok IWH, Chong SZ, Teng KWW, Becht E, Chen J, Sieow JL, Penny HL, Ching GC, Devi S, et al. Developmental analysis of bone marrow neutrophils reveals populations specialized in expansion, trafficking, and effector functions.Immunity. 2018; 48:364–379.e8. doi: 10.1016/j.immuni.2018.02.002CrossrefMedlineGoogle Scholar
  • 25. Ericson JA, Duffau P, Yasuda K, Ortiz-Lopez A, Rothamel K, Rifkin IR, Monach PA; ImmGen Consortium. Gene expression during the generation and activation of mouse neutrophils: implication of novel functional and regulatory pathways.PLoS One. 2014; 9:e108553. doi: 10.1371/journal.pone.0108553CrossrefMedlineGoogle Scholar
  • 26. Van Hove H, Martens L, Scheyltjens I, De Vlaminck K, Pombo Antunes AR, De Prijck S, Vandamme N, De Schepper S, Van Isterdael G, Scott CL, et al. A single-cell atlas of mouse brain macrophages reveals unique transcriptional identities shaped by ontogeny and tissue environment.Nat Neurosci. 2019; 22:1021–1035. doi: 10.1038/s41593-019-0393-4CrossrefMedlineGoogle Scholar
  • 27. van den Brink SC, Sage F, Vértesy Á, Spanjaard B, Peterson-Maduro J, Baron CS, Robin C, van Oudenaarden A. Single-cell sequencing reveals dissociation-induced gene expression in tissue subpopulations.Nat Methods. 2017; 14:935–936. doi: 10.1038/nmeth.4437CrossrefMedlineGoogle Scholar
  • 28. O’Flanagan CH, Campbell KR, Zhang AW, Kabeer F, Lim JLP, Biele J, Eirew P, Lai D, McPherson A, Kong E, et al; CRUK IMAXT Grand Challenge Team. Dissociation of solid tumor tissues with cold active protease for single-cell RNA-seq minimizes conserved collagenase-associated stress responses.Genome Biol. 2019; 20:210. doi: 10.1186/s13059-019-1830-0CrossrefMedlineGoogle Scholar
  • 29. King KR, Aguirre AD, Ye YX, Sun Y, Roh JD, Ng RP, Kohler RH, Arlauckas SP, Iwamoto Y, Savol A, et al. IRF3 and type I interferons fuel a fatal response to myocardial infarction.Nat Med. 2017; 23:1481–1487. doi: 10.1038/nm.4428CrossrefMedlineGoogle Scholar
  • 30. Aibar S, González-Blas CB, Moerman T, Huynh-Thu VA, Imrichova H, Hulselmans G, Rambow F, Marine JC, Geurts P, Aerts J, et al. SCENIC: single-cell regulatory network inference and clustering.Nat Methods. 2017; 14:1083–1086. doi: 10.1038/nmeth.4463CrossrefMedlineGoogle Scholar
  • 31. Woodfin A, Beyrau M, Voisin MB, Ma B, Whiteford JR, Hordijk PL, Hogg N, Nourshargh S. ICAM-1-expressing neutrophils exhibit enhanced effector functions in murine models of endotoxemia.Blood. 2016; 127:898–907. doi: 10.1182/blood-2015-08-664995CrossrefMedlineGoogle Scholar
  • 32. Sunderkötter C, Nikolic T, Dillon MJ, Van Rooijen N, Stehling M, Drevets DA, Leenen PJ. Subpopulations of mouse blood monocytes differ in maturation stage and inflammatory response.J Immunol. 2004; 172:4410–4417. doi: 10.4049/jimmunol.172.7.4410CrossrefMedlineGoogle Scholar
  • 33. Xie X, Shi Q, Wu P, Zhang X, Kambara H, Su J, Yu H, Park SY, Guo R, Ren Q, et al. Single-cell transcriptome profiling reveals neutrophil heterogeneity in homeostasis and infection.Nat Immunol. 2020; 21:1119–1133. doi: 10.1038/s41590-020-0736-zCrossrefMedlineGoogle Scholar
  • 34. Iyer RP, de Castro Brás LE, Cannon PL, Ma Y, DeLeon-Pennell KY, Jung M, Flynn ER, Henry JB, Bratton DR, White JA, et al. Defining the sham environment for post-myocardial infarction studies in mice.Am J Physiol Heart Circ Physiol. 2016; 311:H822–H836. doi: 10.1152/ajpheart.00067.2016CrossrefMedlineGoogle Scholar
  • 35. Tabula Muris Consortium; Single-cell transcriptomics of 20 mouse organs creates a Tabula Muris.Nature. 2018; 562:367–372. doi: 10.1038/s41586-018-0590-4CrossrefMedlineGoogle Scholar
  • 36. Trapnell C, Cacchiarelli D, Grimsby J, Pokharel P, Li S, Morse M, Lennon NJ, Livak KJ, Mikkelsen TS, Rinn JL. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells.Nat Biotechnol. 2014; 32:381–386. doi: 10.1038/nbt.2859CrossrefMedlineGoogle Scholar
  • 37. Calcagno DM, Ng RP, Toomu A, Zhang C, Huang K, Aguirre AD, Weissleder R, Daniels LB, Fu Z, King KR. Type I interferon responses to ischemic injury begin in the bone marrow of mice and humans and depend on Tet2, Nrf2, and Irf3.bioRxiv. 2020;765404. Scholar
  • 38. La Manno G, Soldatov R, Zeisel A, Braun E, Hochgerner H, Petukhov V, Lidschreiber K, Kastriti ME, Lönnerberg P, Furlan A, et al. RNA velocity of single cells.Nature. 2018; 560:494–498. doi: 10.1038/s41586-018-0414-6CrossrefMedlineGoogle Scholar
  • 39. Wolf FA, Angerer P, Theis FJ. SCANPY: large-scale single-cell gene expression data analysis.Genome Biol. 2018; 19:15. doi: 10.1186/s13059-017-1382-0CrossrefMedlineGoogle Scholar
  • 40. Boivin G, Faget J, Ancey PB, Gkasti A, Mussard J, Engblom C, Pfirschke C, Contat C, Pascual J, Vazquez J, et al. Durable and controlled depletion of neutrophils in mice.Nat Commun. 2020; 11:2762. doi: 10.1038/s41467-020-16596-9CrossrefMedlineGoogle Scholar
  • 41. Browaeys R, Saelens W, Saeys Y. NicheNet: modeling intercellular communication by linking ligands to target genes.Nat Methods. 2020; 17:159–162. doi: 10.1038/s41592-019-0667-5CrossrefMedlineGoogle Scholar
  • 42. Adrover JM, Del Fresno C, Crainiciuc G, Cuartero MI, Casanova-Acebes M, Weiss LA, Huerga-Encabo H, Silvestre-Roig C, Rossaint J, Cossío I, et al. A neutrophil timer coordinates immune defense and vascular protection.Immunity. 2019; 50:390–402.e10. doi: 10.1016/j.immuni.2019.01.002CrossrefMedlineGoogle Scholar
  • 43. Adrover JM, Aroca-Crevillén A, Crainiciuc G, Ostos F, Rojas-Vega Y, Rubio-Ponce A, Cilloniz C, Bonzón-Kulichenko E, Calvo E, Rico D, et al. Programmed ‘disarming’ of the neutrophil proteome reduces the magnitude of inflammation.Nat Immunol. 2020; 21:135–144. doi: 10.1038/s41590-019-0571-2CrossrefMedlineGoogle Scholar
  • 44. Erhard F, Baptista MAP, Krammer T, Hennig T, Lange M, Arampatzi P, Jürges CS, Theis FJ, Saliba AE, Dölken L. scSLAM-seq reveals core features of transcription dynamics in single cells.Nature. 2019; 571:419–423. doi: 10.1038/s41586-019-1369-yCrossrefMedlineGoogle Scholar
  • 45. Engblom C, Pfirschke C, Zilionis R, Da Silva Martins J, Bos SA, Courties G, Rickelt S, Severe N, Baryawno N, Faget J, et al. Osteoblasts remotely supply lung tumors with cancer-promoting SiglecF(high) neutrophils.Science. 2017; 358:eaal5081. doi: 10.1126/science.aal5081CrossrefMedlineGoogle Scholar
  • 46. Mao H, Kano G, Hudson SA, Brummet M, Zimmermann N, Zhu Z, Bochner BS. Mechanisms of Siglec-F-induced eosinophil apoptosis: a role for caspases but not for SHP-1, Src kinases, NADPH oxidase or reactive oxygen.PLoS One. 2013; 8:e68143. doi: 10.1371/journal.pone.0068143CrossrefMedlineGoogle Scholar
  • 47. Kiwamoto T, Katoh T, Evans CM, Janssen WJ, Brummet ME, Hudson SA, Zhu Z, Tiemeyer M, Bochner BS. Endogenous airway mucins carry glycans that bind Siglec-F and induce eosinophil apoptosis.J Allergy Clin Immunol. 2015; 135:1329–1340.e9. doi: 10.1016/j.jaci.2014.10.027CrossrefMedlineGoogle Scholar
  • 48. O’Sullivan JA, Carroll DJ, Bochner BS. Glycobiology of eosinophilic inflammation: contributions of siglecs, glycans, and other glycan-binding proteins.Front Med (Lausanne). 2017; 4:116. doi: 10.3389/fmed.2017.00116CrossrefMedlineGoogle Scholar
  • 49. Hasenberg A, Hasenberg M, Männ L, Neumann F, Borkenstein L, Stecher M, Kraus A, Engel DR, Klingberg A, Seddigh P, et al. Catchup: a mouse model for imaging-based tracking and modulation of neutrophil granulocytes.Nat Methods. 2015; 12:445–452. doi: 10.1038/nmeth.3322CrossrefMedlineGoogle Scholar
  • 50. Gupta S, Nakabo S, Blanco LP, O’Neil LJ, Wigerblad G, Goel RR, Mistry P, Jiang K, Carmona-Rivera C, Chan DW, et al. Sex differences in neutrophil biology modulate response to type I interferons and immunometabolism.Proc Natl Acad Sci U S A. 2020; 117:16481–16491. doi: 10.1073/pnas.2003603117CrossrefMedlineGoogle Scholar
  • 51. Yan X, Shichita T, Katsumata Y, Matsuhashi T, Ito H, Ito K, Anzai A, Endo J, Tamura Y, Kimura K, et al. Deleterious effect of the IL-23/IL-17A axis and γδT cells on left ventricular remodeling after myocardial infarction.J Am Heart Assoc. 2012; 1:e004408. doi: 10.1161/JAHA.112.004408LinkGoogle Scholar

Novelty and Significance

What Is Known?

  • Neutrophils massively infiltrate the heart after myocardial infarction, where they can promote both tissue repair and damage processes.

  • Previous reports have suggested that cardiac neutrophils follow a temporal polarization pattern after myocardial infarction, switching from a proinflammatory N1 to an anti-inflammatory N2 profile.

What New Information Does This Article Contribute?

  • We demonstrate a previously unrecognized complexity in the temporal cardiac neutrophil heterogeneity in experimental myocardial infarction, and the time-dependent appearance of a SiglecFhi neutrophil subset in the ischemic heart.

  • Neutrophils acquire the SiglecFhi state specifically in the ischemic heart, and SiglecFhi neutrophils are characterized by increased effector functions (phagocytosis, reactive oxygen species production).

  • SiglecFhi neutrophils are found in murine atherosclerotic vessels, indicating that they also arise in chronic vascular inflammation.

Neutrophils are the first circulating innate immune cells to massively infiltrate the heart after myocardial infarction, where their role is unclear as they have functional capacities promoting both tissue healing and damage. Previous studies have suggested the existence of functionally heterogeneous neutrophil subsets in the ischemic heart, and a temporal proinflammatory N1 to anti-inflammatory N2 neutrophil polarization switch. Using single-cell RNA-sequencing analyses of neutrophils combined with surface epitope detection in a mouse model of myocardial infarction, we demonstrate a previously unrecognized complexity in neutrophil subset dynamics. In the acute post-myocardial infarction phase (1 day), the heart contained young neutrophils with a gene expression profile proximal to bone marrow neutrophils. At day 3 onward, 2 major neutrophil subsets were found: SiglecFlow neutrophils resembling circulating blood neutrophils, and a SiglecFhi neutrophil subset found specifically in the heart but not in the blood, bone marrow or spleen. Our data suggest that the SiglecFhi state is acquired locally in the ischemic heart, and is characterized by increased effector functions (phagocytosis, reactive oxygen species production). SiglecFhi neutrophils were also found in atherosclerotic vascular tissue, suggesting a role for this novel innate immune cell subset both in acute and chronic cardiovascular inflammation.


eLetters should relate to an article recently published in the journal and are not a forum for providing unpublished data. Comments are reviewed for appropriate use of tone and language. Comments are not peer-reviewed. Acceptable comments are posted to the journal website only. Comments are not published in an issue and are not indexed in PubMed. Comments should be no longer than 500 words and will only be posted online. References are limited to 10. Authors of the article cited in the comment will be invited to reply, as appropriate.

Comments and feedback on AHA/ASA Scientific Statements and Guidelines should be directed to the AHA/ASA Manuscript Oversight Committee via its Correspondence page.