Fig 2: Inflammatory activation of HMC3 human microglia cell line by TNF-α and IFN-γ. (a) Flow cytometric analyses demonstrate the increased protein expression of inflammatory microglia biomarkers (IBA-1, CD40, MHC-II, CD68, CD14, and CD86) in the HMC3 cells treated with IFN-γ (0.3 µg/mL, 24 h) (red), but no alteration in the expression of the anti-inflammatory microglia biomarkers CD163 and CD206. TNF-α (0.3 µg/mL, 24 h) (blue) also increases the expression of inflammatory microglia biomarkers (IBA-1, CD40, and CD14). (b) q-PCR analysis revealed increased CD11b and TMEM119 expression by HMC3 microglia after addition of 0.3 µg/mL TNF-α or 0.3 µg/mL IFN-γ for 24 h. Fold change is normalized to untreated control. GAPDH was used as an internal control to normalize mRNA level of the gene of measurement in each sample (n = 5 independent experiments). Quantitative analysis (c) and representative fluorescence micrographs (d, displayed at ×10 and ×4 magnification) of the cell viability assay show that both TNF-α and IFN-γ induce cell death in a concentration-dependent manner (0, 0.01, 0.03, 0.1, 0.3 µg/mL) for 24 h in the HMC3 human microglial cells. Fluorescent green features are AO stains of live cells and fluorescent red features are PI stains of dying cells. *p < 0.1, **p < 0.01, ***p < 0.001, and NS (no significant difference).

Fig 3: Individual variation in myeloid cell subsets. (A) Among the CD45+ CD3– liver cells, the most abundant populations were lacking CD68. Around half of those were other myeloid cells, based on their expression of CD11b. Putative Kupffer cells, CD11b+ CD68+ were variable in abundance from 2% up to 40% of the total. Cells with CD68 but no CD11b were present in most samples but always 5% or less. (B) A frequent pattern of staining within the CD45+ CD11b+ CD68+ cells was that CD14-high, CD32-mid cells were in excess over CD14-mid, CD32-high cells. (C) Variation among 22 individual samples. The CD14-high, CD32-mid cells were most often between 60 and 100% of the total, and the CD12-mid, CD32-high cells correspondingly between 40% and none. The reproducibility of this imbalance was highly significant (p < 0.0001, paired Wilcoxon test). The data in A and C represent 22 separate patient samples.

Fig 4: Gene expression in Kupffer cell subsets. We evaluated the differences between FACS-sorted CD45+ CD3– CD68+ CD32-high and CD45+ CD3– CD68+ CD32-mid cells, using microfluidic qRT-PCR to measure 96 target genes. These cells were isolated from tissue samples sufficiently large, and with permissive anatomy to allow perfusion of the tissue with collagenase. This allowed us to evaluate tissue from three separate patients. (A) Genes were first stratified to identify those expressed more highly in either FACS-sorted sample than in total liver, then this subset stratified to determine which genes were different between the subsets. (B) We tested expression of two genes that encode elements of the CD32 antigen. Expression of FCGR2B was perfectly correlated with cell surface expression of CD32, but FCGR2A was not reliably predicted by surface staining. (C) Of four inflammasome component-encoding genes tested, three were consistently expressed more strongly in the CD32-mid subset. Since we could obtain only 3 sets of paired samples, these data were not subjected to a significance test.

Fig 5: Identification of human liver myeloid cells, Kupffer cells and Kupffer cell subsets within the viable single cells harvested from collagenase-digested human liver tissue. Cells were first gated using FSC-A and SSC-A to exclude sub-cellular debris, then sequentially using FSC-H vs. FSC-A and SSC-H vs. SSC-A to exclude doublets. A live-dead stain was used to exclude cells with damaged membranes. Subsequent analysis involved gating on CD45+ cells, then on CD3-negative cells, then on CD11b+ CD68+ cells. Within the CD11b+ CD68+ subset, we invariably found CD14-high CD32-mid cells and CD14-mid, CD32-high cells. Analyses based on this gating scheme were performed on 23 separate patient tissue samples.