Comparative genotypic and phenotypic analysis of human peripheral blood monocytes and surrogate monocyte-like cell lines commonly used in metabolic disease research

Abstract

Monocyte-like cell lines (MCLCs), including THP-1, HL-60 and U-937 cells, are used routinely as surrogates for isolated human peripheral blood mononuclear cells (PBMCs). To systematically evaluate these immortalised cells and PBMCs as model systems to study inflammation relevant to the pathogenesis of type II diabetes and immuno-metabolism, we compared mRNA expression of inflammation-relevant genes, cell surface expression of cluster of differentiation (CD) markers, and chemotactic responses to inflammatory stimuli. Messenger RNA expression analysis suggested most genes were present at similar levels across all undifferentiated cells, though notably, IDO1, which encodes for indoleamine 2,3-dioxygenase and catabolises tryptophan to kynureninase (shown to be elevated in serum from diabetic patients), was not expressed in any PMA-treated MCLC, but present in GM-CSF-treated PBMCs. There was little overall difference in the pattern of expression of CD markers across all cells, though absolute expression levels varied considerably and the correlation between MCLCs and PBMCs was improved upon MCLC differentiation. Functionally, THP-1 and PBMCs migrated in response to chemoattractants in a transwell assay, with varying sensitivity to MCP-1, MIP-1α and LTB-4. However, despite similar gene and CD expression profiles, U-937 cells were functionally impaired as no migration was observed to any chemoattractant. Our analysis reveals that the MCLCs examined only partly replicate the genotypic and phenotypic properties of human PBMCs. To overcome such issues a universal differentiation protocol should be implemented for these cell lines, similar to those already used with isolated monocytes. Although not perfect, in our hands the THP-1 cells represent the closest, simplified surrogate model of PBMCs for study of inflammatory cell migration.

Fig 1. Hierarchical clustered heat map of mean relative expression values for undifferentiated and differentiated CD14⁺ human peripheral blood mononuclear cells (PBMC, M(GC) and M(GC)LPS/IFNγ) and monocyte-like cell lines (MCLCs ± PMA) including THP-1, U-937 and HL-60s.

PBMCs were differentiated using 10 ng/mL granulocyte-macrophage colony stimulating factor (GM-CSF) for 6 days to give M(GC) and activated using 100 ng/mL LPS and 20 μg/mL IFNγ for 24 h to generate M(GC)LPS/IFNγ. MCLCs were differentiated using 16 ng/mL phorbol-12-myristate-13-acetate (PMA) for 48 h. Data represent n = 3–10; raw data, including the relative expression ± SEM used to generate this heat map, are shown in S2 and S3 Tables. All qPCR data were normalized against both housekeeper genes GAPDH and ACTB2 and the values calculated as described in the Methods. Relative expression values were then calculated versus the relevant non-differentiated cell type. Where no expression was detected a value of 0 was used. Thirty-five genes that encode for inflammatory chemokines, cytokines, adipokines and their relevant receptors were selected as they are associated with inflammation or have been implicated in the development and/or progression of obesity-induced insulin resistance. In addition, small subsets of genes encoding for regulatory factors and enzymatic processes that have been implicated in the pathogenesis of T2DM were profiled. Differential gene expression data were analysed using R version 3.4.1 and hierarchical clustering was performed using complete linkage method with Euclidean distance measure.

Fig 2. Relative expression analysis of chemokines, cytokines, receptors, enzymes, and regulatory factors for undifferentiated, differentiated and polarized human peripheral blood monocytes (PBMCs; A and E; white, grey and black bars respectively), and monocyte-like cell lines (MCLCs) including U-937 (B and F; red bars), THP-1 (C and G; blue bars), and HL-60 (D and H; yellow bars).

PBMCs were differentiated using 10 ng/mL granulocyte-macrophage colony stimulating factor (GM-CSF) for 6 days to give M(GC) and activated using 100 ng/mL LPS and 20 μg/mL IFNγ for 24 h to generate M(GC)LPS/IFNγ. MCLCs were differentiated using 16 ng/mL phorbol 12-myristate 13-acetate (PMA) for 48 h. Grouped data are shown as mean ± SEM (n = 3–10). Relative expression data including the mean ± SEM are shown in S3 Table. Genes not detected were marked as n/d. Statistical significance was determined by one-way ANOVA with Tukey’s multiple comparison test compared to undifferentiated cells for the PBMCs, or by Student’s t-test for the MCLCs, with *P<0.05, **P<0.01, ***P<0.001 and ****P<0.0001 deemed significant.

Fig 3. Principal component analysis (PCA) of the mean relative expression values (n = 3–10) grouped by either cell type (A) or gene family (B) shown in Fig 1 and (C) Venn diagram using the relative expression values of genes displaying a significant up-regulation or down-regulation in expression, after GM-CSF or PMA treatment for PBMCs, U-937, THP-1, and HL-60 cells.

Connecting arrows in the PCA analysis represent changes from undifferentiated to differentiated state. Data used to generate (C) are summarized in Table 1.

Fig 4. Effects of differentiation treatment on cluster of differentiation (CD) markers expression on human peripheral blood monocytes (PBMCs; grey bars) and monocyte-like cell lines (MCLCs; U-937; red bars and THP-1; blue bars) and concentration-response curves to proinflammatory chemoattractants using a transwell migration assay using PBMCs (grey circles), U-937 (red circles) and THP-1 (blue circles) cells.

Data are expressed as % cells expressing (%PE+ population) for (A) CD11c, (B) CD163, (C) CD14, (D) CD68, (E) CD80, (F) CD86 and (G) HLA Class II, or as the relative level of expression (mean PE) for (H) CD11c, (I) CD163, (J) CD14, (K) CD68, (L) CD80, (M) CD86 and (N) HLA Class II. Grouped data are shown as mean ± SEM (n = 3–5). Data are expressed on a Log2 scale as the chemotactic index compared to basal levels following 3–4 h treatment with the various chemoattractants using (O) MCP-1, (P) fMLP, (Q) LTB4, (R) MIP1α and (S) 10% FBS. Grouped data are shown ± SEM (n = 3–22). Statistical significance was determined by one-way ANOVA with Tukey’s multiple comparisons test compared to undifferentiated cells for the CD markers, and two-way ANOVA with Dunnett’s multiple comparison tests compared to vehicle control for chemotaxis, with *P<0.05, **P<0.01, ***P<0.001 and ****P<0.0001 deemed significant.