Defining the proteomic landscape of cultured macrophages and their polarization continuum

Abstract

Macrophages have previously been characterized based on phenotypical and functional differences into suggested simplified subtypes of MØ, M1, M2a and M2c. These macrophage subtypes can be generated in a well-established primary monocyte culture model that produces cells expressing accepted subtype surface markers. To determine how these subtypes retain functional similarities and better understand their formation, we generated all four subtypes from the same donors. Comparative whole-cell proteomics confirmed that four distinct macrophage subtypes could be induced from the same donor material, with > 50% of 5435 identified proteins being significantly altered in abundance between subtypes. Functional assessment highlighted that these distinct protein expression profiles are primed to enable specific cell functions, indicating that this shifting proteome is predictive of meaningful changes in cell characteristics. Importantly, the 2552 proteins remained consistent in abundance across all macrophage subtypes examined, demonstrating maintenance of a stable core proteome that likely enables swift polarity changes. We next explored the cross-polarization capabilities of preactivated M1 macrophages treated with dexamethasone. Importantly, these treated cells undergo a partial repolarization toward the M2c surface markers but still retain the M1 functional phenotype. Our investigation of polarized macrophage subtypes therefore provides evidence of a sliding scale of macrophage functionality, with these data sets providing a valuable benchmark resource for further studies of macrophage polarity, with relevance for cell therapy development and drug discovery.

Keywords: dexamethasone; macrophage; macrophage subtype; polarization; proteomics.

Figure 1

Human cultured polarized macrophage subtypes differ in cell surface marker expression and morphology. (a) Representative cytospins at day 7 from each of the subtypes, MØ control, M1, M2a and M2c stained with May–Grünwald's and Giemsa's stain. Scale bars are 20 μm; images are representative of N = 3. (b) Flow cytometry analysis summary of cell surface expression for key macrophage markers at day 7 of ex vivo culture across four different subtypes; CD14, CD169, CD163, CD16, CD86, CD80, CD209 and CD206. N = 3 with a minimum of 10 000 events per sample. (c) Median fluorescence intensity of flow cytometry markers tested in b. All flow cytometry experiments show the mean and standard deviation for the data (N = 3, ns P ≥ 0.05), with a minimum of 10 000 events per sample and significance tested using a one‐way ANOVA followed by a Tukey's multiple comparison test, comparing each subtype. (d) Summary of cell surface marker expression in the four cultured macrophage subtypes assessed by flow cytometry. Expression presented as a scale. *P < 0.05; ***P < 0.001.

Figure 2

Semiquantitative tandem mass tagged (TMT) proteomics highlights the significant proteomic changes underlying macrophage polarization. (a) Experimental design for TMT proteomic analysis. CD14⁺ cells were isolated from donated human peripheral blood before being stimulated towards differentiation to M1, M2a, M2c or MØ (control) subtypes. (b) Heatmap visualization of the TMT proteomic analysis generated from log₂ fold changes in expression of whole‐cell lysates with subcluster descriptions. Lower expression is represented in blue, while higher expression is in red; false discovery rate set at 5%. Log₂ fold change values were clustered with hierarchical Euclidean distance clustering. (c) Chord plot summarizing the proportion of upregulated and downregulated proteins for each of the M1, M2a and M2c macrophage subtypes with values representing this for each subtype. (d) Principal component analysis of the four macrophage subtypes in the total proteomic data set. Dex, dexamethasone; FC, fold change; IFN, interferon; IL, interleukin; NT, not treated; PBMC, peripheral blood mononuclear cell; TNF, tumor necrosis factor.

Figure 3

Semiquantitative TMT proteomics elucidates specific protein changes in polarized macrophage populations. (a) The top five enriched meta‐categories per macrophage subtype according to the Enrichr BioPlanet 2019 enrichment module. (b) Bubble plot from results in a. (c) Log₂ fold change heatmap of subset‐specific overexpressed proteins encompassed in the general “Primary Metabolic Process” (totaling 85 proteins, with MØ as a baseline). Gene ontology terms enriched in each protein cluster are displayed to the right. Lower expression is shown in blue and higher expression in red. (d) Mean (± standard error of the mean) log₂ fold change of differentially expressed tricarboxylic acid cycle and pentose phosphate–associated pathway proteins, grouped by polarized macrophage subset specificity. TNF, tumor necrosis factor.

Figure 4

The macrophage “surfaceome” is dynamically altered upon polarization. (a) Expression of CD marker proteins (previously assessed in Figure 1b, c) in the tandem mass tagged (TMT) surfaceome analysis after additional filtering for surface proteins using the Cell Surface Protein Atlas in silico database (accessed June 2020). Expression is shown in Log₂ FC from each subtype against MØ control as a baseline. (b) Heatmap visualization of the surfaceome of M1, M2a and M2c subtypes with the MØ control as a baseline. Descriptions of the subclusters are also included. Lower expression is represented in blue, while higher expression is in red. Log₂ fold change values were clustered via hierarchical Euclidean distance clustering. (c) Venn diagram of the consistently upregulated and downregulated (log₂ fold change ≤1 and ≥1) proteins between three subtypes using the MØ control macrophages as a baseline for comparison. (d) Table illustrating results from Figure 3b; presented are the proteins either significantly upregulated or downregulated for each of the three subtypes when expressed as a log₂ fold change (log₂ fold change ≤ 1 and ≥ 1) against the MØ control. (e) Heatmap visualization of the “myeloid cell activation involved in immune response” gene ontology category term in the surface‐filtered TMT data set. N = 3 for macrophage subtypes and N = 2 for MØ control; false discovery rate was set at 5%.*P < 0.05; **P < 0.01; ***P < 0.001.

Figure 5

Protein expression of macrophage subtypes is primed for specific functionality. (a) Expression of proteins of the “Oxidative stress” GO category term in the gene set enrichment analysis (GSEA)–filtered whole‐cell tandem mass tagged (TMT) data set using the GO Cellular Component 2021 module in Enrichr, filtered for macrophage‐labeled proteins from RNA‐seq data within the ARCHS4 database. (b) Respiratory burst formation in response to phorbol 12‐myristate 13‐acetate (PMA) stimulation measured for each of the four macrophage subtypes for 100 min. Error bars represent the standard deviation. (c) Area under the curve analysis for b, where a Kruskal–Wallis test (**P < 0.01) followed by Dunn's multiple comparison test was performed (N = 3, **P < 0.01 and *P < 0.05). (d) Expression of proteins of the “Cytoskeleton” GO category term in the GSEA‐filtered whole‐cell TMT data set using the GO Cellular Component 2021 module in Enrichr, filtered for macrophage‐labeled proteins from RNA‐seq data within the ARCHS4 database. (a, d) Blue denotes decreased expression (log₂ fold change < 0), red denotes decreased expression (log₂ fold change > 0) and white denotes no change (log₂ fold change = 0) in comparison to MØ control. N = 3 for macrophage subtypes and N = 2 for MØ control; false discovery rate (FDR) was set at 5%. (e) Scatter plot with mean and standard deviation of the mean speed of macrophages in each Incucyte imaging field per 20‐min time frame of the four macrophage subtypes. The Kruskal–Wallis test (****P < 0.0001) followed by Dunn's multiple comparison test was performed on 75 fields of view (25 fields of view per donor, N = 3, ****P < 0.0001). (f) Scatter plot with mean and standard deviation of the mean Euclidean distance macrophages cover in each imaging field per 20‐min time frame of the four macrophage subtypes. The Kruskal–Wallis test (****P < 0.0001) followed by Dunn's multiple comparison test was performed on 75 fields of view (25 fields of view per donor, N = 3, ****P < 0.0001). ns, not significant.

Figure 6

In vitro polarized macrophages are able to undergo partial repolarization through the maintenance of a core set of commonly expressed proteins. (a) The top 11 enriched meta‐categories per macrophage subtype according to the Enrichr BioPlanet 2019 enrichment module of commonly expressed proteins within the tandem mass tagged proteomic data set generated from log₂ fold changes in expression of whole‐cell lysates compared with MØ control, where log₂ fold change is ≥−1 and ≤1. (b) The top 10 enriched meta‐categories per macrophage subtype according to the Enrichr BioPlanet 2019 enrichment module of “Immune system” proteins identified in Figure 5a. (c) Bubble plot from results in a, where red bubbles denote the “Immune system” meta‐category. (d) Bubble plot from results in b. (e) Flow cytometry analysis of dexamethasone treatment for 48 h on M1‐induced macrophages to determine plasticity of inflammatory macrophages. Subtype‐specific markers were analyzed: CD14, CD169, CD163, CD16, CD86, CD80, CD209 and CD206. Percentage positive of population is shown. Bar shows mean, with each point representing individual samples (N = 6, ns P > 0.05), with a minimum of 10 000 events per sample; significance was tested using a two‐tailed t‐test. (f) Heatmap visualization of proteomic comparison showing significantly upregulated or downregulated proteins when comparing dexamethasone‐treated samples against M1 as a baseline. N = 2 for M1 dexamethasone treated and MØ control; false discovery rate was set at 5%. ns, not significant; *P < 0.05.