THP-1 cells were seeded at increasing densities in 96-well black clear-bottom imaging microplates (Corning #353219) in RPMI medium with HEPES and GlutaMAX (Gibco #72400-021) supplemented with 10% heat-inactivated (30 minutes at 56°C) fetal bovine serum (FBS; SERANA #S-FBS-SA-015) and 1% penicillin-streptomycin (Gibco #15070-063). Seeding densities were 5,000, 12,000, 21,000, 37,000, 53,000, 69,000, or 87,000 cells/well, denoted as x 10³ cells/mm² (0.16, 0.38, 0.66, 1.16, 1.66, 2.16, or 2.72 x 10³ cells/mm², respectively) that were rounded to one decimal digit to improve readability (0.2, 0.4, 0.7, 1.2, 1.7, 2.2, or 2.7 x 10³ cells/mm², respectively). THP-1 cells were differentiated into MΦ by exposure to 2000 nM phorbol 12-myristate 13-acetate (PMA; Sigma #P1585) for 48 hours at 37°C, 5% CO₂ after which they were rested for 24 hours in fresh culture medium before performing the functional assays.

Peripheral blood mononuclear cells (PBMCs) were isolated from leukocyte reduction system cones, a by-product of thrombopheresis of routine blood donations from healthy volunteers collected at the University Hospital RWTH Aachen, Germany, by density centrifugation with Lymphoprep™ (STEMCELL Technologies #07861). Isolated PBMCs were cryopreserved for later use in one-third freezing medium containing 75% FBS and 25% dimethyl sulfoxide (DMSO; Merck #102950). CD14 MicroBeads (Miltenyi #130-096-052) were used to positively select CD14+ monocytes, following the manufacturer’s protocol. Monocytes of 6 donors were pooled before seeding at the respective densities in 96-well black clear-bottom imaging microplates, or kept separate and seeded in 384-well plates (Greiner Bio-One #781866) in culture medium containing RPMI medium with HEPES and GlutaMAX supplemented with 10% heat-inactivated FBS and 1% penicillin-streptomycin. Seeding densities were 30,000, 64,000, 98,000, 132,000, 166,000, or 200,000 cells/well, denoted as x 10³ cells/mm² (0.94, 2.0, 3.06, 4.13, 5.19, or 6.25 x 10³ cells/mm², respectively) that were rounded to integers to improve readability (1, 2, 3, 4, 5, or 6 x 10³ cells/mm², respectively). At 384-well format, cells were plated at 13,000 cells/well. Monocytes were differentiated into MΦ with 100 ng/ml recombinant human macrophage colony-stimulating factor (M-CSF) (ImmunoTools #11343113) for 7 days at 37°C, 5% CO₂ with one medium change.

MΦ functions were assessed using the “MacroScreen” high-content analysis (HCA) platform developed in-house, a semi-automated microscale (96- to 384-well format) assay platform to perform an expanding range of fluorescence-based functional assays (18). All MacroScreen assays have been benchmarked against conventional mesoscale assays. Images were taken using the BD Pathway 855 automated fluorescent microscope (BD Biosciences) by taking 9 images per well with a 10x Olympus 0.40 NA objective (96-well format) or 20x Olympus 0.75 NA objective (384-well format). All experiments were performed in n=3-6 replicates. Images were analyzed with CellProfiler software version 4.0.4 (19) by creating a digital segmentation mask for each cell based on its nuclear staining signal using the IdentifyPrimaryObjects module followed by the ExpandOrShrinkObject module. Percentage of positive cells was measured using the MeasureObjectIntensity module followed by the ClassifyObjects module, which determines the amount of positive cells relative to the total number of segmented objects in the image. Distance to the closest neighboring cell was measured per image, for each cell, using the MeasureObjectNeighbors module in CellProfiler 4.0.4 (19).

Cells were incubated with 25 μl/ml pHrodo™ Red Zymosan Bioparticles™ (ThermoFisher Scientific #P35364) per well in culture medium for 1 hour at 37°C, 5% CO₂. Zymosan particles taken up by cells are reflected by tetramethyl rhodamine isothicyanate (TRITC) fluorescence signal. Nuclei were stained with Hoechst 33342 (Sigma #B2261) in culture medium for 10 minutes at 37°C, 5% CO₂ which was replaced by PBS before imaging.

Low-density lipoprotein (LDL) was isolated from serum of healthy volunteers via density centrifugation and oxidized using CuSO₄ as described previously (20). Cells were incubated with 8 μg/ml oxidized LDL (oxLDL) pre-mixed with 2 μg/ml TopFluor^® Cholesterol (Avanti Polar Lipid #810255P) in culture medium for 3 hours at 37°C, 5% CO₂, reflected by fluorescein isothiocyanate (FITC) fluorescence signal. Nuclei were stained with Hoechst 33342 in culture medium for 10 minutes at 37°C, 5% CO₂ which was replaced by PBS before imaging.

Cells were primed with 50 ng/ml LPS from E. coli (Invivogen #tlrl-eblps) for 3 hours at 37°C, 5% CO₂. 10 μM nigericin (Invivogen #tlrl-nig) for 1 hour at 37°C, 5% CO₂ served as second signal for inflammasome activation. Fc receptor was blocked with Fc receptor binding inhibitor antibody (Invitrogen #14-9161-73) before cells were fixed using 2% PFA with 5 mM EDTA in PBS and permeabilized using 5% FBS and 0.5% Triton X-100 in PBS for 20 minutes on ice. Next, the intracellular adapter protein apoptosis associated speck-like protein containing a CARD (ASC) was stained using PE-conjugated anti-ASC antibody (clone HASC-71; BioLegend #653904) overnight at 4°C. Nuclei were stained with Hoechst 33342 in PBS for 10 minutes on ice, and cells were washed with and imaged in PBS afterwards.

Mitochondrial stress was induced using 1200 nM staurosporine (Sigma #S4400) for 1 hour at 37°C, 5% CO₂. Mitochondria and nuclei were stained simultaneously with 250 nM MitoTracker™ Deep Red FM (ThermoFisher Scientific #M22426) and Hoechst 33342, respectively, in culture medium for 30 minutes at 37°C, 5% CO₂ before imaging in PBS. Alexa 594 fluorescence signal reflects mitochondrial staining dependent on membrane potential.

Cells were incubated with 10 μM 5’ethynyl-2’-deoxyuridine (EdU; ThermoFisher Scientific #A10044) for 2 hours at 37°C, 5% CO₂. Next, cells were fixated with 3.7% PFA in PBS for 15 minutes and washed twice with PBS before permeabilization with 0.1% Triton X-100 in PBS for 15 minutes. After another wash with PBS, Click-iT reaction cocktail (ThermoFisher Scientific #C10269) including Alexa Fluor 594 azide (ThermoFisher Scientific, #A10270) prepared according to the manufacturer’s instructions was added. After 30 minutes, cells were washed with PBS and nuclei stained with Hoechst 33342 for 15 minutes. Cells were imaged in 1:5 KI quencher (1M KI in 10 mM KH₂PO₄) in PBS.

THP-1-derived MΦ were stimulated with 50 ng/ml LPS from E. coli for 6 hours at 37°C, 5% CO₂. Supernatant was collected and cytokine levels of IL-1β, IL-6, IL-8, IL-10, IL-12p70, and TNF-α were measured in a custom V-plex human cytokine ELISA (MSD Meso Scale Diagnostics) according to the manufacturer’s protocol. Measured levels of IL-12p70 were close to or below the standard curve (0.1169 pg/ml), thus these data are not presented here.

We calculated the mean of the replicates for each feature of HCA functional data and cytokine data at the same density, resulting in a density-feature matrix with 7 rows (densities) and 10 columns (features/functions). The functional profile of THP-1 MΦ for the 7 different densities was analyzed in an integrated manner by two-dimensional Principal Component Analysis (PCA), where the relative contribution of each individual function to the first (PC1) and second principal component (PC2) axes was visualized outside the X and Y axes.

Data are expressed as mean ± SEM, unless stated otherwise. Normal distribution was assessed by Shapiro-Wilkes normality test, and equality of variances by Brown-Forsythe test. For normally distributed data with equal variance, significance was assessed by one-way ANOVA followed by Tukey’s multiple comparisons test at a significance level of p <0.05. For normally distributed data with unequal variance, significance was assessed by Brown-Forsythe ANOVA and Dunnett’s T3 multiple comparisons test, and Kruskal-Wallis test with Dunn’s multiple comparisons test was used for not normally distributed data. All statistical analyses were performed using GraphPad Prism 8 software.

We assessed the functional phenotype of THP-1-derived MΦ seeded at 7 different densities ranging from 0.2 to 2.7 x 10³ cells/mm² using the MacroScreen platform, an HCA platform based on several fluorescent imaging-based functional assays. For clarity’s sake, we presented the p-values of all group comparisons for all assays in a significance matrix ( Supplementary Table 1 ). As expected, the number of nuclei per fluorescent image increased with density in all high-content functional assays ( Supplementary Figures 1A–E ). No changes in the pH of the culture medium were observed. Whereas the phagocytic activity of THP-1 MΦ was found to increase with higher density ( Figure 1A ), uptake of oxLDL decreased ( Figure 1B ). MΦ seeded at 0.2 x 10³ cells/mm², the lowest density in the tested density range, showed the highest inflammasome activation in response to LPS and nigericin ( Figure 1C ). Moreover, at lower densities, MΦ were found to be more resistant to staurosporine-induced mitochondrial stress ( Figure 1D ). Proliferation measured by EdU incorporation increased with density ( Figure 1E ). Secretion of the cytokines IL-10, IL-1β, IL-6, IL-8, and TNF-α in response to LPS stimulation was measured using multiplex ELISA, and was found to decrease with density, after an initial sharp increase between the two lowest densities included in the density range (0.2 and 0.4 x 10³ cells/mm²) ( Figures 1F–J ). PCA allowed separation of THP-1 MΦ seeded at different densities based on their functional profiles ( Figure 1K ). A large proportion of variation (61.27%) could be explained by PC1 which loadings included mitochondrial stress, lipid uptake, proliferation and phagocytosis, suggesting these functions to be mostly influenced by density. Variation along PC2 (largely attributable to inflammasome activation and cytokine secretion) almost exclusively manifested at the lowest density (0.2 x 10³ cells/mm²). Together, these data demonstrate the profound and variable impact of culture density on several THP-1 MΦ functions. Execution of certain functions may depend on exceeding a lower density threshold.

We also investigated whether culture density influences the functional phenotype of human MDM, pooled from 6 healthy donors. A significance matrix with the p-values of all group comparisons can be found in Supplementary Table 2 . As seen in THP-1 MΦ, the number of detected nuclei increased with increasing seeding density ( Supplementary Figures 1F–I ). Phagocytic activity increased in the lower densities but decreased again at seeding densities exceeding 3 x 10³ cells/mm², possibly reflecting limited substrate ( Figure 2A ). Using a constant substrate-to-cell ratio, increased phagocytosis also at higher densities could be observed ( Supplementary Figure 2 ). OxLDL uptake between MDM seeded at different densities was similar, contrary to THP-1 MΦ ( Figure 2B ). With increasing density, MDM demonstrated more inflammasome activation ( Figure 2C ) and were more susceptible to mitochondrial stress ( Figure 2D ). Overall, our findings show that density also affects primary MΦ and leads to changes in several functions. In MDM, these changes were less pronounced and less consistent compared to THP-1 MΦ, suggesting that the observed impact of density on MΦ function may be particularly relevant for cell lines.

We observed that THP-1 MΦ formed colonies in culture, especially if cultured at higher densities. Thus, we set out to study if at least part of the density-dependent functional changes may be attributed to changes in the colony-forming growth pattern of THP-1 cells. We assessed the association between the functional outcome of an index cell with its distance to the closest neighboring cell. Index cells that were close to neighboring cells had higher proliferation rates compared to MΦ in more scarcely populated niches ( Figure 3A ). However, at very low seeding density (0.2 x 10³ cells/mm²), MΦ proliferation was low regardless of the index cell’s distance to neighboring cells, suggesting proliferation to be linked to colony formation. Similarly, we found that at a seeding density of 0.2 x 10³ cells/mm², MΦ with neighboring cells in close proximity had taken up fewer fluorescent beads compared to equally dense MΦ seeded at higher densities ( Figure 3B ), a finding that was less pronounced for the uptake of oxLDL ( Figure 3C ). Interestingly, a similar proximity effect on phagocytosis and lipid uptake in MDM ( Figures 3D, E ), in which we did not observe colony formation in culture, was not apparent. These findings suggest that MΦ only exert certain functions beyond a local cell density threshold.

Inter-donor variability has previously been proposed to compromise reproducibility of in vitro experiments with MDM (21, 22). Therefore, we looked at functions of MDM from 6 healthy donors (A-F) plated at the same density (13,000 cells/well). A significance matrix with the p-values of all group comparisons can be found in Supplementary Table 3 . Differences in the number of nuclei ( Figure 4A ), phagocytosis ( Figure 4B ), oxLDL uptake ( Figure 4C ), inflammasome activation ( Figure 4D ) and mitochondrial stress ( Figure 4E ) between donors were apparent. Interestingly, MDM from donors A and B, which showed higher phagocytic activity and oxLDL uptake, also had a higher number of nuclei compared to the other donors. This suggests that functional differences between MDM obtained from different donors may well be related to inter-donor differences in density, possibly resulting from differences in adherence and survival capacity.