Microenvironmental conditions and serum availability alter primary human macrophage NF-κB inflammatory response and function

Abstract

Macrophages are central to innate immunity and are routinely used in vitro to examine molecular mechanisms contributing to innate immune signaling. However, there is a lack of consensus within the field for optimal in vitro culturing methods, and it is not well understood whether differences in culture conditions produce incongruent outcomes. Here, we compared the effects of commonly used culture medium compositions on TLR4-mediated proinflammatory activity in primary human monocyte-derived macrophages (hMDMs) isolated from healthy blood donors. hMDMs were cultured in fetal bovine serum (FBS)–containing or FBS-free conditions in either Dulbecco's Modified Eagle Medium (DMEM), RPMI, or in Macrophage-Serum Free Medium (M-SFM). Lipopolysaccharide-mediated immune response was measured through nuclear factor κB activation and cytokine and chemokine secretion, which were muted in M-SFM cultures compared with DMEM and RPMI cultures. FBS supplementation increased total cytokine secretion in response to lipopolysaccharide but also showed higher baseline secretion, suggesting a proinflammatory phenotype. Moreover, M-SFM cultures exhibited less phagocytosis compared with DMEM and RPMI cultures. Morphologic analysis of unstimulated hMDMs revealed the highest cell area and length-to-width ratio in M-SFM compared with DMEM or RPMI cultures. FBS-free and M-SFM conditions produced distinct transcriptional profiles compared with media supplemented with FBS, most notably in cell cycle pathways and lipid homeostasis, respectively. Overall, DMEM and RPMI produce comparable morphologic and functional results, albeit with some small differences, while M-SFM produces a muted inflammatory response in macrophages. These data demonstrate that in vitro microenvironment drives differential inflammatory outcomes in human macrophages and is a critical component of experimental design in this cell type.

Serum availability alters human macrophage nuclear factor κB dynamics and inflammatory functions including cytokine secretion and phagocytic capacity in response to macrophage activation in vitro.

1. Introduction

Macrophages are highly plastic innate immune cells with vital roles in immunity and homeostasis within most tissues of the body. Tissue resident macrophages such as Kupffer cells (liver), alveolar macrophages (lungs). or osteoclasts (bone) are long lived and have key roles specific to their environment.^1–3 In contrast, monocyte-derived macrophages (MDMs) differentiate from circulating monocytes and infiltrate tissues to initiate an immune response and resolve an insult or injury.^3,4 In several tissues, infiltrating MDMs can acquire the phenotype of resident macrophages if there is a depletion in the tissue resident population.^3,5,6 While macrophage-like cell lines are available, the accessibility of monocytes from the blood and the relative similarity of MDMs to primary tissue macrophages make these cells a commonly used model for macrophage biology in vitro.⁷ Thus, understanding the impact of discrete culture systems on MDM immune processes is key to study design and maintaining translational value.

All in vitro models require establishment of an artificial environment with sources of nutrients such as type of culture media (RPMI, Dulbecco's Modified Eagle Medium [DMEM], or alternative), presence of serum (human and/or fetal bovine), antibiotics, and macrophage-defining cytokines such as macrophage colony-stimulating factor (M-CSF). Several groups have evaluated the impact of altering various components of the in vitro culture system, showing that the in vitro microenvironment impacts the function of MDMs and other myeloid cells.^8–10 In macrophage differentiation, M-CSF vs. granulocyte-macrophage colony-stimulating factor (GM-CSF) driven maturation promotes proinflammatory (GM-CSF) vs. anti-inflammatory (M-CSF) roles, as well as variance in antiviral response and signaling.^11,12 Human MDMs (hMDMs) cultured in DMEM, a medium reported to lack nonessential amino acids, show differences in phenotypic markers as measured by morphology, gene expression, and release of soluble proteins in comparison with other media formulations—RPMI, minimum essential medium, Iscove’s Modified Dulbecco’s Medium.¹³ However, there are limited studies specifically evaluating the effects of the culture medium, which is the most abundant component of the in vitro microenvironment but commonly varies between studies.

To address this, we differentiated human MDMs in 5 distinct media formulations and evaluated the effects on macrophage function and morphology. These formulations used the traditional base media DMEM and RPMI 1640 supplemented with heat-inactivated human AB serum. We tested both formulations supplemented with human AB serum and with or without heat-inactivated fetal bovine serum (FBS). We also tested a serum free formulation generated specifically for macrophage culture, Macrophage Serum Free Medium (M-SFM), which lacked supplementation with human AB serum and FBS. Our data confirmed that macrophages cultured in the two most common formulations DMEM (+FBS) and RPMI (+FBS) were similar transcriptionally by RNA sequencing but showed small differences in morphology and immune response. Our data also demonstrate changes mediated by FBS supplementation of primary macrophage differentiation,^14,15 as the cells differentiated in the absence of FBS show distinct morphological, transcriptomic, and inflammatory profiles relative to those differentiated in the presence of FBS. Our data also indicated that the macrophages cultured in M-SFM were distinct from those cultured in DMEM and RPMI, demonstrating a considerably smaller magnitude of response to inflammatory signals, showing lower magnitude of nuclear factor κB (NF-κB) activation and cytokine secretion in response to TLR4 stimulation. These cells also showed a distinct morphology and transcriptional profile that suggests dysregulated lipid homeostasis. Together, these data show clear but moderate differences in macrophage phenotype and responsiveness resulting from culture in various media and in the presence or absence of serum. Studies that investigate in vitro molecular mechanisms of innate immune activity may benefit from careful consideration of experimental conditions for human macrophage studies.

2. Methods

2.1. Reagents

DMEM, RPMI 1640, SFM, and penicillin/streptomyocin were from Invitrogen. HEPES solution from Fisher Scientific. Lipopolysaccharide (LPS) (from Escherichia coli 055:B5) was from Sigma-Aldrich. Human M-CSF was from PeproTech, reconstituted in distilled H₂O and stored at 4 °C for 1 mo. Fetal bovine serum from Corning and human AB serum from Gemini Bio Products.

2.2. Primary macrophage differentiation

Human peripheral blood mononuclear cells (PBMCs) were separated from blood obtained from de-identified healthy donors (New York Blood Center, Long Island City, New York; and BioreclamationIVT Collection Center, Miami, Florida). Blood collection was conducted in accordance with the protocols approved by the Institutional Review Board (IRB) at both the New York Blood Center and BioreclamationIVT. PBMCs are isolated using Ficoll-Paque (GE Healthcare) gradient centrifugation. Following isolation, monocyte percentage within PBMCs was quantified using a monocyte isolation kit (MACS; Miltenyi Biotec). AO/PI staining was used for cell count which provides a percentage of viable cells isolation—at both the PBMC and monocyte isolation steps. All donors used in these studies show at least 90% viability at both steps. PBMCs were plated at a density of 100,000 monocytes per cm² to obtain a pure culture of hMDMs via adherence isolation. Prior to resuspension in culture media, cells were washed only in Hank's Balanced Salt Solution to avoid any contact with media other than the specific media in which cells were cultured. PBMCs were cultured in either RPMI-1640 + L-glutamine (5% human AB serum, 10 mM HEPES, 1% penicillin/streptomycin) ± 10% FBS, DMEM + GlutaMAX (5% human AB serum, 10 mM HEPES, 1% penicillin/streptomycin) ± 10% FBS, or SFM (10 mM HEPES, 1% penicillin/streptomycin). All media included M-CSF (10 ng/mL) with mentioned supplemental components and specific details on each base media is shown in Table 1.^16,17 Cells were cultured for 3 d then washed with fresh media to remove nonadherent cells. Adherent cells were cultured another 3 d in fresh media containing M-CSF. After 6 d in culture maintained at 37 °C in a humidified incubator at 5% CO₂, cells are considered matured MDMs. Experiments were performed at day 6 or 7. All experiments with material isolated from human blood were performed using protocols approved by the Institutional Biosafety Committee at Drexel University College of Medicine. The Drexel University IRB reviewed these studies and determined them exempt from IRB review due to the exclusive use of de-identified human material.

Table 1.

Components of each base media.

DMEM				RPMI				M-SFM
Component	Molecular Weight	Concentration (mg/L)	mM	Component	Molecular Weight	Concentration (mg/L)	mM	Component
Amino acids				Amino acids				Amino acids
Glycine	75	30	0.4	Glycine	75	10	0.133333	Glycine
L-Alanyl-glutamine	217	862	3.97235	L-Arginine	174	200	1.149425	L-Cystine 2HCl
L-Arginine hydrochloride	211	84	0.398104	L-Asparagine	132	50	0.378788	L-Glutamate
L-Cystine 2HCl	313	63	0.201278	L-Aspartic acid	133	20	0.150376	Vitamins
L-Histidine hydrochloride-H2O	210	42	0.2	L-Cystine 2HCl	313	65	0.207668	Biotin
L-Isoleucine	131	105	0.801527	L-Glutamic acid	147	20	0.136054	Vitamin B12
L-Leucine	131	105	0.801527	L-Glutamine	146	300	2.054795	Ferric Nitrate (Fe(NO3)3″9H2O)
L-Lysine hydrochloride	183	146	0.797814	L-Histidine	155	15	0.096774	Other Components
L-Methionine	149	30	0.201342	L-Hydroxyproline	131	20	0.152672	Cholesterol
L-Phenylalanine	165	66	0.4	L-Isoleucine	131	50	0.381679	Human transferrin
L-Serine	105	42	0.4	L-Leucine	131	50	0.381679	Human serum albumin
L-Threonine	119	95	0.798319	L-Lysine hydrochloride	183	40	0.218579
L-Tryptophan	204	16	0.078431	L-Methionine	149	15	0.100671
L-Tyrosine disodium salt dihydrate	261	104	0.398467	L-Phenylalanine	165	15	0.090909
L-Valine	117	94	0.803419	L-Proline	115	20	0.173913
Vitamins				L-Serine	105	30	0.285714
Choline chloride	140	4	0.028571	L-Threonine	119	20	0.168067
D-Calcium pantothenate	477	4	0.008386	L-Tryptophan	204	5	0.02451
Folic acid	441	4	0.00907	L-Tyrosine disodium salt dihydrate	261	29	0.111111
Niacinamide	122	4	0.032787	L-Valine	117	20	0.17094
Pyridoxal hydrochloride	204	4	0.019608	Vitamins
Riboflavin	376	0.4	0.001064	Biotin	244	0.2	8.20E-04
Thiamine hydrochloride	337	4	0.011869	Choline chloride	140	3	0.021429
i-Inositol	180	7.2	0.04	D-Calcium pantothenate	477	0.25	5.24E-04
Inorganic salts				Folic Acid	441	1	0.002268
Calcium chloride (CaCl2) (anhydrous)	111	200	1.801802	Niacinamide	122	1	0.008197
Ferric nitrate (Fe(NO3)3″9H2O)	404	0.1	2.48E-04	Para-Aminobenzoic acid	137	1	0.007299
Magnesium sulfate (MgSO4) (anhydrous)	120	97.67	0.813917	Pyridoxine hydrochloride	206	1	0.004854
Potassium chloride (KCl)	75	400	5.333334	Riboflavin	376	0.2	5.32E-04
Sodium bicarbonate (NaHCO3)	84	3,700	44.04762	Thiamine hydrochloride	337	1	0.002967
Sodium chloride (NaCl)	58	6,400	110.3448	Vitamin B12	1,355	0.005	3.69E-06
Sodium phosphate monobasic (NaH2PO4-H2O)	138	125	0.905797	i-Inositol	180	35	0.194444
Other components				Inorganic salts
D-Glucose (dextrose)	180	4,500	25	Calcium nitrate (Ca(NO3)2 4H2O)	236	100	0.423729
Phenol red	376.4	15	0.039851	Magnesium Sulfate (MgSO4) (anhydrous)	120	48.84	0.407
				Potassium chloride (KCl)	75	400	5.333334
				Sodium bicarbonate (NaHCO3)	84	2,000	23.80953
				Sodium chloride (NaCl)	58	6,000	103.4483
				Sodium phosphate dibasic (Na2HPO4) (anhydrous)	142	800	5.633803
				Other components
				D-Glucose (dextrose)	180	2,000	11.11111
				Glutathione (reduced)	307	1	0.003257
				Phenol red	376.4	5	0.013284

2.3. High content imaging and analysis

For high content imaging studies, MDMs were cultured in Nunc MicroWell 96-well optical-bottom plates (Thermo Fisher Scientific) and treated with vehicle (H₂O) or LPS (1 ng/mL) at indicated timepoints for each assay. Cells were then fixed using 4% paraformaldehyde for 10 min and permeabilized with 0.1% Triton X-100 in phosphate-buffered saline (PBS) for 5 min. Cells were blocked for 30 min in 1% bovine serum albumin and 300 mM glycine in 0.1% Tween 20 in PBS. Primary and secondary antibody selection was assay dependent and outlined subsequently. When applicable, nuclei were counterstained with DAPI (0.2 μg/mL) and plasma membrane were stained with cell mask deep red (CMDR) (Thermo Fisher Scientific; 250 ng/mL).

To assess the nuclear translocation of NF-κB, hMDMs were treated with vehicle (H₂O) or a range of LPS (0.01 to 100 ng/mL) for 60 min, as previously determined.¹⁸ In addition to the steps outlined previously, cells were incubated at 4 °C overnight in rabbit monoclonal anti-NF-κB-p65 antibody (Cell Signaling Technology; 8242, 1:400) diluted in blocking solution. Alexa Fluor 488 goat anti-rabbit secondary antibody (Thermo Fisher Scientific; 1:500) was used for detection. Images were acquired on a Cell Insight CX7 High Content screening platform (CX7) at 20 × magnification with 4 wells per condition, acquiring approximately 1,000 to 1,500 cells per well. Images were acquired with a fixed exposure time and intrawell autofocusing with every field. Additional parameters are shown in Table 2. Images were analyzed using HCS studio software colocalization bio-application (Cellomics; Thermo Fisher Scientific). Binary masks were created around cytoplasmic (CMDR) and nuclear (DAPI) regions of interest (ROIs), and then the average intensity of NF-κB staining in each ROI for every cell is quantified. The NF-κB average intensity in the nuclear ROI is divided by the NF-κB average intensity in the cytoplasmic ROI in each cell to generate a nuclear colocalization ratio for each individual macrophage. This quantifies the relative amount of NF-κB in the nucleus, while also controlling for cell size and differences in total NF-κB between cells. The colocalization ratio is averaged across all wells per condition then used for statistical analysis.

Table 2.

Parameters used for each high content imaging bio-application in HCS studio software.

High Content Imaging Colocalization Analysis Parameters		High Content Imaging Morphology Analysis Parameters		High Content Imaging Target Activation Analysis Parameters		High Content Imaging Spot Detection Analysis Parameters
Condition	Value	Condition	Value	Condition	Value	Condition	Value
Ch1: SmoothFactor	1	Ch1: SmoothFactor	1	Ch1: SmoothFactor	1	Ch1: SmoothFactor	1
Ch1: Thresholding (Fixed)	500 to 700	Ch1: Thresholding (Fixed)	800	Ch1: Thresholding (Fixed)	300	Ch1: Thresholding (Fixed)	200
BackgroundCorrectionCh1	50	Object.BorderObject.Ch1	Y	BackgroundCorrectionCh1	28	BackgroundCorrectionCh1	255
Object.Ch1.Average Intensity.Ch1	0 to 6,000	BackgroundCorrectionCh1	Y	Object.Ch1.Average Intensity.Ch1	518.44 to 65,535	ObjectAreaCh1	1312.74 to 21262.58
Ch2: SmoothFactor	3	Ch1: Exposure Method	Fixed	Ch2: Exposure Parameters	Fixed (0.1)	Object.Ch1.Average Intensity.Ch1	0 to 65,535
Ch2: Thresholding (Auto or Fixed)	500 to 850	ObjectAreaCh1	1223.46 to 19288.87	Ch2: Thresholding (Fixed)	500 to 850	ObjectSegmentationCh1	−500
BackgroundCorrectionCh2	−255	ObjectShapeP2ACh1	0 to 1,000	BackgroundCorrectionCh2	50	SpotDetectRadiusCh1	12
Ch2: Segmentation (Intensity)	−300	ObjectShapeLWRCh1	0 to 1,000	Object.Average Intensity.Ch2	0 to 23809.66	Ch2: SmoothFactor	3
ObjectAreaCh2	53.52 to 1170.72	Object.Ch1.Average Intensity.Ch1	0 to 6,000	Ch3: SmoothFactor	2	Ch2: Thresholding (Fixed)	150
Object.Average Intensity.Ch2	0 to 4062.65	Ch2: Exposure Method	Fixed	Ch3: Thresholding (Fixed)	400	BackgroundCorrectionCh2	25
Ch3: SmoothFactor	2	Ch2: Thresholding (Fixed)	NA	Ch3: Segmentation (Intensity)	70	Ch2: Segmentation (Intensity)	−450
Ch3: Thresholding (Fixed)	400	Object.BorderObject.Ch2	Y	Object.Average Intensity.Ch3	0 to 65,535	SpotAreaCh2	1 to 625
BackgroundCorrectionCh3	−255	BackgroundCorrectionCh2	Y			Object.Average Intensity.Ch2	0 to 32,767
Ch3: Segmentation (Intensity)	−3000					CircModifierCh2	2
ObjectAreaCh3	22.09 to 8935.92					SpotDetectRadiusCh2	11
Object.Average Intensity.Ch3	0 to 65,535					Ch3: SmoothFactor	2
ROI.A.Mask Ch	Channel 1 (DAPI)					Ch3: Thresholding (Fixed)	200
ROI.A.Target_I	Channel 2 (NF-κB)					BackgroundCorrectionCh3	27
ROI.B.Mask Ch	Channel 3 (CMDR)					SpotAverageIntensity.Ch3	215.72 to 65,535
ROI.B.Target_I	Channel 2 (NF-κB)					CircModifierCh3	3
ROI.B.Exclude	Channel 1 (DAPI)					SpotDetectRadiusCh3	8
RejectBorderObjects	Y

To evaluate changes in morphology, images acquired for the translocation assay were analyzed using the specific parameters outlined in Table 2. This process uses the Cellomics Morphological Profiling bio-application, using the defined thresholding and segmentation parameters to create a cytoplasmic (CMDR) mask for each cell. Four wells were assessed for each condition. The CMDR intensity within the outlined mask is then used to define morphological parameters including area, length-to-width ratio (LWR) and perimeter to area ratio (P2A). Each morphological feature is averaged across all cells and wells from a particular condition, and then the averages from all conditions across all donors are compared by analysis of variance (ANOVA).

To evaluate changes in Ki-67 expression, macrophages were incubated in rabbit monoclonal anti-Ki-67 (MA514520, 1:250) diluted in blocking solution at 4 °C overnight. Alexa Fluor 488 goat anti-rabbit secondary antibody (Thermo Fisher Scientific; 1:1,000) was used for detection. Images were acquired with the CX7 with a fixed exposure time and intrawell autofocusing at every field, using additional parameters defined in Table 2. Images were analyzed using the HCS studio software Target Activation bio-application. This quantifies the relative amount of Ki-67+ nuclei within the population by creates nuclear mask (DAPI) ROI then determining the intensity of Ki-67 in each nuclear ROI. The percentage of Ki-67+ cells is determined for each well and averaged across all wells with the same treatment per donor and then compared between media by ANOVA.

2.4. Cytokine and chemokine analysis

Assessment of cytokine and chemokine secretion was performed using 2 distinct assays, both of which evaluated the same supernatants. To generate these supernatants, hMDMs were plated in 48-well plates at 9.5 × 10⁴ cells per well in all 5 media conditions and stimulated with the indicated concentrations of vehicle (H₂O) or LPS (1 ng/mL) for 24 h. Previous studies show maximal cytokine secretion from this hMDM model at 12 to 24 hours.¹⁸ Supernatants were collected and stored at −80 °C for batch processing. To analyze secretion of IL-6 and CXCL10 by AlphaLISA (Revvity), a high-throughput, sensitive, no-wash step immunoassay. Samples were processed according to the manufacturer's protocol as we have done previously. The limit of detection for these assays was 1 pg/mL for IL-6 and 3 pg/mL for CXCL10. To interrogate changes in a larger number of cytokines, supernatants were assessed using the multiplexed MSD V-Plex Human Neuroinflammation panel (K15210D-1; Meso Scale Diagnostics). Biomarkers analyzed using this assay included IL-1β, IL-2, IL-6, IL-10, IL-12p70, IL-13, TNF-α, IL-7, IL17A, VEGF-A, eotaxin, eotaxin-3, and IP-10 (CXCL10). All samples were run in duplicate according to the manufacturer's protocol. Plates were imaged on the MESO QuickPlex SQ 120MM and analyzed using MSD Discovery Workbench 4.0 Software (Meso Scale Diagnostics). Donors that showed analyte levels below the limit of detection were not included in statistical tests but are visually represented below dashed lines denoting the limit of detection in respective figures.

2.5. Phagocytosis assay

hMDMs were incubated with 0 to 50 μg/mL of pHrodo BioParticles Conjugates for Phagocytosis (Thermo Fisher Scientific) for 60 min in a 37 °C incubator in various media conditions. The bioparticles are conjugated to inactivated, unopsonized Staphylococcus aureus antigens and become fluorescent in acidified phagolysosomes in a pH-dependent manner. After 60 min of treatment with phagocytic beads, hMDMs were washed with 1× PBS then fixed with 4% paraformaldehyde. hMDMs are immediately permeabilized with 0.1% PBS-Triton X and stained with CMDR and DAPI as stated previously. Cells were imaged on the CX7 using the parameters in Table 2, with fixed exposure times and intrawell focusing, acquiring 56 fields per well at 40×, imaging a total of 3 wells (800 to 1,200 cells per well) per condition. Images were analyzed using the Cellomics Spot Detection bio-application, creating an ROI around each individual cell using a cytoplasmic mask (CMDR) to identify each cell body. Within each ROI, the number and size of each fluorescent puncta, signifying acidified phagolysosomes, are measured, and averaged across all wells for each condition. The averages across all donors are pooled and compared by ANOVA.

2.6. Quantitative real-time polymerase chain reaction

Total RNA was extracted from cells using Quick-DNA/RNA Microprep Plus Kit (Zymo), and RNA quantity and purity were determined by a NanoDropOne spectrophotometer (Nanodrop Technologies). RNA (1 μg) was used to synthesize complementary DNA (cDNA) using the high-capacity reverse transcriptase cDNA synthesis kit (Abcam). TLR4 and the housekeeping gene 18 s were amplified from cDNA by quantitative polymerase chain reaction on a QuantStudio 7 using gene-specific primers.

2.7. RNA sequencing

Samples were analyzed with respect to purity and potential degradation in the University of Nebraska Medical Center Genomics Core Facility using a Nanodrop (Thermo Fisher Scientific) instrument to measure absorbance, and potential degradation of the sample will be assessed by using an Advanced Analytical Technical Instrument Fragment Analyzer (AATI). Samples with A260/280 of 1.8 or above and with RQN scores >8.0 were utilized to prepare sequencing libraries.

2.8. Library generation and sequencing

Sequencing libraries were generated by the University of Nebraska Medical Center NGS Core beginning with 200 ng of total RNA from each sample using the NuGEN Universal Plus mRNA-Seq library kit from TECAN following the manufacturers recommended procedure (TECAN). Resultant libraries were assessed for size of insert by analysis of an aliquot of each library on a Bioanalyzer instrument (Agilent Technologies). Each library had a unique indexing identifier barcode allowing the individual libraries to be multiplexed together for efficient sequencing. Multiplexed libraries were sequenced on a single SP-200 flow cell of the NovaSeq6000 DNA Analyzer (Illumina) using a 2 × 100 bp paired end protocol to generate a total of approximately 16 to 20 million pairs of reads for each sample.

2.9. Statistical analysis

Prior to analysis, all data other than sequencing data were normalized to the mean of the control condition, and data were considered outliers and removed if they were more than 3 SDs outside the mean for the condition being analyzed. After normalization, data were evaluated for skewness and by using the D’Agostino-Pearson Omnibus Normality test to determine whether the data were parametric. Normally distributed data were analyzed using a paired t test, 1- or 2-way ANOVA, using repeated measures as appropriate. Nonparametric data were analyzed using a Wilcoxon test, Kruskal-Wallis test, or Friedman test as appropriate. For ANOVA analyses, post hoc analysis was performed using Tukey’s multiple comparisons for normally distributed data and Dunn's multiple comparisons test for nonparametric data. All data analysis was performed using GraphPad Prism 10.02 (GraphPad Software), with P < 0.05 was considered significant.

To analyze the data generated by bulk RNA sequencing, a read count matrix was generated using nf-core RNA sequencing pipeline version 3.10.1 with the FASTQ sequencing files as input. The reads are filtered and mapped to a customized reference genome which combines hg38 and NC_001802.1 using STAR aligner.¹⁹ The resulting count matrix was analyzed for the differential expression analysis using a custom Python Jupyter Notebook (Supplemental File 1). In brief, the analysis involved the following steps. First, each gene was z-scaled across all samples to normalize expression between genes. Second, this analysis requires simultaneous comparison of the effect of media (DMEM vs. RPMI), the presence or absence of FBS, and the effect of dopamine treatment along with their interactive effects; as such, a linear model was used to estimate these effects together. Statsmodels (v0.14.0)²⁰ was used to fit the equation expression ∼ donor + media*FBS + treatment. The effect-size and P value of each term was extracted. Third, P values were adjusted for multiple testing using the Benjamini-Hochberg (BH) procedure to control for a constant false discovery rate, and volcano plots were created using Seaborn (v0.12.2). Fourth, the genes with |z| > 0.5 and BH-adjusted P < 0.01 were submitted to the Enrichr Python API²¹ in 10/2024 to find enrichment in GO_Biological_Process_2023, Chromosome_Location, KEGG_2021_Human, TRANSFAC_and_JASPAR_PWMs, and The_Kinase_Library_2023 categories. Those with an adjusted P value <0.01 were considered significant.

2.10. Data sharing

RNA sequencing data from this study has been deposited to National Center for Biotechnology Information Gene Expression Omnibus and the National Center for Biotechnology Information Sequence Read Archive, accession number GSE281287.

3. Results

3.1. Culture media alters the activation of macrophage NF-κB

Macrophages respond to environmental stimuli via tightly controlled signaling pathways that regulate inflammatory output.^22,23 Prior studies have shown that serum can alter cell proliferation and immune responses, and this can be dependent on the type of serum to which cells are exposed.^24–26 To assess the effect of culture conditions on immune regulation, hMDMs were differentiated in 5 distinct culture media (Fig. 1). These included DMEM and RPMI supplemented with human AB serum and with or without FBS. Most of these media are commonly used to culture hMDMs, although studies examining differences between DMEM and RPMI are limited and culture without FBS is infrequent. M-SFM is a relatively new, proprietary formulation (Thermo Fisher Scientific) with limited publicly available information.²⁷ M-SFM shares common components with both DMEM and RPMI such as glutamate, glycine, and cysteine but also contains unique components such as cholesterol and human transferrin (Table 1).

Fig. 1.

Macrophage differentiation and experimental workflow. PBMCs were isolated from human donors via centrifugal isolation with Ficoll isolation. PBMCs were plated into experimental plates with culture media—DMEM (+FBS), RPMI (+FBS), M-SFM, DMEM (−FBS), or RPMI (−FBS)—in the presence of M-CSF to undergo adherence isolation and maturation of monocytes to macrophages for 6 to 7 d, with a media change on the third day. On day 6 or 7 in culture, matured hMDMs are left untreated or incubated with LPS or appropriate vehicle control. Following experiments, endpoint assays include immunofluorescent staining with high content imaging and analysis, cytokine profiling by AlphaLISA and Meso Scale, and transcriptional changes by RNA sequencing. Created in BioRender. Gaskill, P. (2025) https://BioRender.com/ph11kjf

We first evaluated the effect of culture media and/or serum supplementation on the innate immune response by quantifying the nuclear translocation of NF-κB. This transcription factor is a well-characterized component of the innate immune response ie triggered downstream of stimulation of TLR4 by LPS.²⁸ Activation of TLR4 by LPS enables the transmigration of the NF-κBp65 subunit to the nucleus, initiating the transcription of immune modulators such as IL-6 and CXCL10.^29,30 We quantified NF-κBp65 nuclear translocation by high content imaging in response to a 1 h treatment with LPS (0.01 ng/mL to 1,000 ng/mL). Representative images of the NF-κB response in hMDMs cultured in each media condition are shown (Fig. 2A), as well as a visual representation of the high content analysis of NF-κB nuclear translocation (Fig. 2B). The average intensity of each ROI was used to create the nuclear:cytoplasmic NF-κBp65 colocalization ratio, with larger ratios indicating more intense nuclear NF-κBp65 and suggesting more hMDM activation.

Fig. 2.

NF-κB translocation in response to TLR4 agonist is dose dependent with low sensitivity in serum-free cultured macrophages. Primary monocyte derived macrophages cultured in all 5 media conditions were stimulated with LPS (0.01 ng/mL to 1,000 ng/mL) for 1 h then fixed and immunostained. Images captured with CX7 Automated High Content Imager, 20 × objective at room temperature and analysis performed with HCS Studio Software using Colocalization Bio-application. Analysis parameters are shown in Table 2. (A) Representative images of macrophages differentiated in various media and response to LPS. Nuclei (DAPI), NF-κB p65, plasma membrane (CMDR). (B) Representative image of high content colocalization analysis used to determine the NF-κB colocalization ratio as measured by the intensity of NF-κB colocalized with nucleus (DAPI) divided by intensity NF-κB colocalized cytoplasm (CMDR). Analysis of images used HCS Studio and GraphPad Prism v10.0 to assess log dose curve E_max and EC₅₀ in (C) DMEM ± FBS, (D) RPMI ± FBS, and (E) DMEM (+FBS) vs. RPMI (+FBS). (F) DMEM + FBS vs. M-SFM and (G) RPMI + FBS vs. M-SFM pharmacodynamic curves. Data presented are the means ± SEM from 13 independent donors in DMEM (−FBS) and RPMI (−FBS), 15 independent donors in M-SFM, and 15 to 18 independent donors in RPMI (+FBS) and DMEM (+FBS). GraphPad Prism v10.0 was used to assess and graph log dose curves. E_max and EC₅₀ comparison between groups was performed with Mann-Whitney t test. *P < 0.05, **P < 0.001.

Treatment with LPS induced NF-κBp65 nuclear translocation in all media, but with different pharmacodynamics (Fig. 2C–G). We assessed both the maximal effect (E_max) of NF-κB nuclear translocation in response to LPS and the concentration of LPS producing 50% of the maximal effect (EC₅₀) of NF-κB activation in all culture conditions. In DMEM (−FBS) but not RPMI (−FBS)–cultured hMDMs there is a significantly lower E_max (DMEM [+FBS] Avg E_max = 6.8; DMEM [−FBS] Avg E_max = 3.7; P < 0.001, Mann-Whitney t test) for NF-κB activation in response to LPS (Fig. 2C, D) relative to media with FBS. This suggests that RPMI, but not DMEM, is able to sustain hMDM responses to LPS that elicit maximal NF-κB response regardless of FBS presence. Moreover, this suggests that in DMEM, FBS supplementation is important to elicit maximal LPS-induced NF-κB activity. The data also show that there is a rightward shift and greater EC₅₀ (RPMI [+FBS] Avg EC₅₀ = 0.03 ng/mL; RPMI [−FBS] Avg EC₅₀ = 0.12 ng/mL; P < 0.01, Mann-Whitney t test) in the dose curve in hMDM cells cultured in RPMI (−FBS) media (Fig. 2D). This indicates a decrease in the potency of LPS in this media, as more LPS is required to elicit the same level of NF-κB activity. While there were not significant differences in the NF-κB response in DMEM (+FBS) vs. RPMI (+FBS), hMDMs cultured in RPMI (+FBS) showed both a lower E_max and EC₅₀ (Fig. 2E), indicating subtle differences in LPS binding and NF-κB response in macrophages cultured in these 2 common media.

We also used M-SFM to assess the impact of culture conditions lacking both human AB serum and FBS on NF-κB activation in response to LPS in hMDMs. Culture in M-SFM dampened the NF-κB response to LPS, as the EC₅₀ of NF-κB activation in M-SFM–cultured macrophages is substantially greater (EC₅₀ > 100 ng/mL) than the tested range in hMDMs cultured in DMEM (+FBS) or RPMI (+FBS) (Fig. 2F, G). This difference in response between hMDMs cultured in M-SFM–based media vs. DMEM- and RPMI-based media suggests the importance of serum in eliciting LPS-induced NF-κB activation.

3.2. Culture media influences macrophage cytokine and chemokine secretion

We next evaluated the effect of culture media on cytokine secretion in LPS-treated (1 ng/mL) hMDMs. Following 24 h of treatment, we collected supernatant and screened for various secreted cytokines and chemokines, initially evaluating secretion of the commonly assessed cytokine IL-6. As anticipated, LPS stimulation significantly increased IL-6 secretion in RPMI (+FBS) and DMEM (+FBS) cultures, as well as DMEM (−FBS)–cultured cells. Notably, LPS did not increase IL-6 in hMDMs cultured in RPMI (−FBS) and M-SFM (Fig. 3A). As with NF-κB activation (Fig. 2), LPS treatment led to different magnitudes of response across culture conditions. The cells cultured in DMEM (+FBS) secreted the greatest amounts of cytokines in response to LPS, followed by cells in DMEM (−FBS), then RPMI (+FBS), RPMI (−FBS), and then M-SFM (Fig. 3A). Differences in the magnitude of cytokine secretion were consistent with the NF-κB response (Fig. 2C–G), suggesting that factors within culture media alter the macrophage inflammatory response dynamics and could influence downstream read-outs. Baseline secretion (secretion prior to any treatment) of IL-6 was lower in FBS(−) media compositions, including M-SFM media, which may explain the wider relative assay window between vehicle and LPS treatments.

Fig. 3.

M-SFM macrophages show low cytokine and chemokine secretion in response to TLR4 stimulation. Primary MDMs cultured in all 5 media conditions were stimulated with LPS (1 ng/mL) for 24 h and supernatants collected for AlphaLISA and Meso Scale assays. AlphaLISA data for (A) IL-6 secretion following LPS stimulation. Data are normalized to the mean of the vehicle control (media + H₂O)–treated cells of the respective media type. Significance was determined using Kruskal-Wallis test and post hoc analysis with Dunn's multiple comparisons. *P < 0.05, **P < 0.01, ***P < 0.001. Meso Scale data for (B) cytokines (IL-6, TNF-α, IL-17A, and IL-10) and (C) chemokines (CXCL10 and eotaxin). Statistical analyses used 2-way mixed effects analysis using Tukey's multiple comparison test for post hoc analysis. *P < 0.05, **P < 0.01, ***P < 0.001 ****P < 0.0001. The dashed line on graphs represents the limit of detection; samples below this limit were not included in the statistical analyses.

These data also indicated differences in the raw amount of hMDM-induced cytokine release that in response to LPS in DMEM-based vs. RPMI-based vs. M-SFM–based culture media. To determine whether these differences resulting were unique to IL-6, a multiplex assay (Meso Scale Discovery) was used to examine 13 cytokines, including IL-6 as well as CXCL10, TNF-α, IL-1β, IL-17A, IL-13, and eotaxin/CCL11, among others. These assays focused on the cytokine secretion specific differences in distinct culture media using the media formulations, DMEM (+FBS), RPMI (+FBS), or M-SFM. Across all measured cytokines, M-SFM–cultured hMDMs showed lower cytokine secretion at baseline and in response to LPS (Fig. 3B, C). LPS stimulation increased the secretion of cytokines IL6, TNF-α, and IL-10 in DMEM (+FBS) and RPMI (+FBS) but not in M-SFM macrophages (Fig. 3B). LPS also significantly increased the chemokines CXCL10 and eotaxin/CCL11 in DMEM (+FBS) cells but not in RPMI (+FBS) cells or M-SFM–cultured macrophages (Fig. 3C). LPS did not significantly increase IL-17A in any condition (Fig. 3B). For some cytokines, such as IL-1β, VEGF-A, IL-6, eotaxin-3, and IL-13, there were splits in the response resulting in 2 populations, one of which was often at or below the limit of detection (Fig. S1). This is likely a result of the heterogeneity inherent in primary human cells.³¹ Notably, these splits were not always consistent between media, with responses being more disparate in one media than another (Fig. 3B, C; Fig. S1). In addition to showing greater levels of cytokine secretion at baseline in the presence of FBS, these data also suggest differences in the raw amount of cytokine or chemokine that can be released in response to LPS in DMEM-based vs. RPMI-based vs. M-SFM–based culture media.

3.3. Macrophage show lower phagocytic capacity in the absence of serum

Macrophage phagocytosis is essential to the innate immune response.³² To determine how culture media impacts phagocytic activity, hMDMs differentiated in DMEM (+FBS), RPMI (+FBS), or M-SFM media were incubated with pH-sensitive pHrodo beads and then evaluated for phagocytotic capacity using high content imaging. The frequency of phagocytosis is measured by assessing the number of phagosomes using a spot count analysis—the number of spots (fluorescent green) within a defined plasma membrane boundary (CMDR). Preliminary analysis in DMEM (+FBS)– and RPMI (+FBS)–cultured hMDMs showed a dose-dependent uptake of beads that increased across concentration range of 1 to 50 μg/mL, with the largest differences in uptake between media observed between 2.5 to 7.5 ug/mL (Fig. S2). Therefore, subsequent studies focused on this concentration range. Representative images of hMDMs incubated with pHrodo beads for 1 h are shown (Fig. 4A). Spot count analysis identified the average number of phagosomes within hMDMs showed a dose-dependent increase in phagosome numbers in hMDMs cultured in all 3 media (Fig. 4B). Unlike the NF-κB activity and cytokine secretion, there were no differences in the magnitude of phagocytic activity in DMEM (+FBS) vs. RPMI (+FBS). However, M-SFM–cultured MDMs showed lower-level phagocytic uptake at all bead concentrations, with significantly lower uptake compared with DMEM (+FBS) at 2.5 ug/mL pHrodo beads (Fig. 4B, C). To account for possible phagosome fusion, the average phagosome area per macrophage was also evaluated using high content analysis. This showed similar results to the spot count analysis, demonstrating a dose-dependent increase in all media formulations, with M-SFM cells showing lower phagosome area overall with significant differences in phagosome area at 2.5 ug/mL pHrodo beads compared with DMEM (+FBS) (Fig. 4C). This suggests that in the absence of a serum source, (bovine or human) macrophages show lower phagocytic capacity as measured by uptake in M-SFM differentiated hMDMs.

Fig. 4.

Media affects magnitude of macrophage phagocytic activity. Primary MDMs were incubated with pH-sensitive, Staphylococcus aureus–coated pHrodo beads, using either 2.5, 5, or 7.5 μg/beads/mL. After 1 h at 37 °C, then cells were fixed, stained, and imaged with CX7 Automated High Content Imager, 20 × objective at room temperature. Analysis was performed with HCS Studio Software using the target activation Bio-applications. Parameters for image analysis are shown in Table 2. (A) Representative bright-field images of macrophages cultured in DMEM (+FBS), RPMI (+FBS), and M-SFM following phagocytosis assay show nuclei (DAPI), phagocytosed beads/phagosome, and plasma membrane (CMDR). High content analysis using the target activation bio-application shows (B) number of fluorescent phagosomes per macrophage as measured by fluorescent spot count per CMDR identified macrophage and (C) phagosome area per macrophage as measured by fluorescent spot area per CMDR identified cell boundary. Data were averaged across 3 wells per condition and normalized to the mean of vehicle condition. Significance determined using mixed-effects analysis using Tukey’s post hoc testing. *P < 0.05, **P < 0.01, ***P < 0.001.

3.4. Culture conditions influence primary macrophage morphological phenotype

Macrophages present a range of morphologies in response to distinct environmental cues and stimuli, including the type of culture dish, growth factors, or culture media.^33,34 At baseline, bright-field imaging shows 2 predominant hMDM morphologies, either spindled (blue arrows, Fig. 5A) or rounded (red chevron, Fig. 5A). Both RPMI (+FBS) and DMEM (+FBS) cultures appear to have more spindled cells, while cultures in M-SFM and RPMI (−FBS) show a mix of cell shapes (Fig. 5A). In DMEM (−FBS) cultures, macrophages appeared smaller and predominantly rounded. These morphological features were quantified by high content imaging, confirming the observation that hMDM area changes with distinct media conditions. Overall, M-SFM–cultured hMDMs had the largest average area and were significantly larger than DMEM (−FBS)–cultured cells (Fig. 5B). Quantification of roundness and shape using P2A (Fig. 5C) and LWR (Fig. 5D) also show media and serum-dependent differences. In both LWR and P2A analysis, RPMI (−FBS) cultures showed the lowest value, suggesting more rounded shape across the culture compared with RPMI (+FBS) and M-SFM. Interestingly, in DMEM-based cultures the addition of FBS did not change the LWR and P2A as in the RPMI-based cultures, although both DMEM (+FBS) and DMEM (−FBS) cultures showed lower ratios (i.e. rounder cells than RPMI [+FBS] and M-SFM). As macrophages move and change shape in response to stimuli, hMDMs from each media that were challenged with LPS (0.01 ng/mL to 100 ng/mL) for 1 h. Surprisingly, the trends in morphology did not change in response to LPS (Fig. S3). These data indicate that FBS supplementation can drive morphological differentiation in macrophages, particularly in RPMI cultures, and that culture in M-SFM consistently differentiates hMDMs into larger sizes.

Fig. 5.

Culture media impacts differentiated MDM morphology and cell cycle state. (A) Representative bright-field images of MDMs differentiated in all 5 media at baseline on days in vitro 6 or 7. Blue arrows denote spindled/elongated morphology and red chevrons highlight rounded morphology. High content quantification of morphological features at baseline includes (B) macrophage area (n = 10–18), (C) macrophage roundness or P2A (n = 10–16), and (D) macrophage elongation or LWR (n = 11 to 17). (E) Representative images of high content imaging to assess proliferation marker, Ki-67, expression on macrophages cultured in each media condition. (F) High content quantification of Ki-67+ cells (n = 8–12). Bright-field images captured with a Nikon NIS Elements microscope, 20 × objective at room temperature. Immunofluorescence images captured with CX7 Automated High Content Imager, 20×. Image analysis was performed using HCS Studio 2.0 using the Morphology and Target Activation Bio-applications. Parameters for image analysis are shown in Table 2. Data from all wells were pooled for analysis. Statistical analyses used Kruskal-Wallis test with Dunn's multiple comparisons for post hoc analysis. *P < 0.05, **P < 0.01, ***P < 0.001.

Macrophages are generally considered to be terminally differentiated,^35,36 though some data indicate a proliferative capacity in homeostatic and disease states.^37–40 To assess whether culture conditions alter hMDM proliferation, hMDMs were given fresh media at t = 0, then incubated for 3 h, fixed, stained, and analyzed for Ki-67, a marker often used to define proliferation. Representative images (Fig. 5E) and high content analysis (Fig. 5F) show that hMDM cultures in RPMI (+FBS) and DMEM (+FBS) contained approximately 8% Ki-67+ cells, while macrophages cultured in DMEM or RPMI without FBS contained <1% Ki-67+ cells. Cultures in M-SFM also contained around 8% Ki-67+ cells (Fig. 5F). As these results are only a snapshot in time, it is likely that more than just the 8% of cells staining for Ki-67 express this marker. These results indicate that in DMEM and RPMI cultures, FBS supplementation may increase cell proliferative activity in hMDMs, and that M-SFM can produce similar proportions of proliferating cells without the need for serum supplementation. As the DMEM (−FBS) and RPMI (−FBS) cultures contain human AB-serum while M-SFM does not, this suggests human AB serum supplementation is not driving these changes in Ki-67. Though Ki-67 alone is not a definitive marker of proliferation, it is commonly used as a surrogate marker for proliferation.^41–43 Thus, these data support and expand on prior studies showing that hMDMs have a proliferative capacity, and this finding may potentially correlate with the polarization state and inflammatory responses previously identified.⁴⁴

3.5. Serum supplementation alters transcriptomic profiles of primary human macrophages during differentiation

To further examine the impact of culture media, RNA from hMDMs (n = 5) cultured in all 5 media formulations was analyzed by RNA sequencing. Initial analyses examined the effect of FBS supplementation by assessing differential gene expression in cultures lacking FBS supplementation compared with the FBS containing media. Compared with hMDMs cultured in RPMI (+FBS) and DMEM (+FBS), multiple linear analysis (BH correction P < 0.01) shows that 1,210 genes were differentially expressed in hMDMs cultured in RPMI (−FBS), DMEM (−FBS) and M-SFM (Fig. 6A). We mapped these significant changes in gene expression using the Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) databases. Among the 21 GO terms most altered in hMDMs cultured in the absence of FBS compared with those in media supplemented with FBS were Regulation of Transcription by RNA Polymerase II (GO:0006357), Regulation of DNA-templated Transcription (GO:006355), Mitotic Spindle Checkpoint Signaling (GO:0071174), and Regulation of Mitotic Cell Cycle Phase Transition (GO:1901990) (Fig. 6B). The KEGG analysis showed that FBS supplementation dysregulated 26 pathways (Fig. 6C), many associated with cell cycle progression and regulation, including the FoxO Signaling Pathway (BH, adjusted P = 9.49 × 10⁻⁵), Cell Cycle (BH, adjusted P = 0.00005), Longevity Regulating Pathway (BH, adjusted P = 0.0005), and Cellular Senescence (BH, adjusted P = 0.0005) (Fig. 6C). These data support the observation that there are more Ki-67+ cells (i.e. more proliferation) in cultures of hMDMs in media containing FBS (Fig. 5F) relative to those lacking FBS. Dysregulated genes common in these pathways and important in macrophage inflammation include PLK1 (effect size = −1.672; P < 0.00077), associated with dampening NLRP3 inflammasome activation^45,46 and FOXO3 (effect size = 1.133; P < 0.00088), a longevity gene whose inactivation is associated with increased anti-inflammatory activity in macrophages⁴⁷ (Fig. 6D). These data suggest that FBS may dysregulate macrophage inflammation at a transcriptional level and could impact the response to immune challenge.

Fig. 6.

Differential gene expression in macrophages cultured in various media and serum environment. (A) Supplementation with FBS induced transcriptomic changes in hMDMs with 1,210 significantly dysregulated genes (effect size > |1| and BH false discovery rate < 0.01). Violin plot represents the relative gene expression and dashed lines represent thresholds. Gene set enrichment analysis by (B) GO and (C) KEGG pathways depicting critical biological processes and pathways, respectively, dysregulated in the presence of FBS. (D) Genes dysregulated with FBS that are implicated in macrophage inflammation and survival. The same analysis was performed to assess transcriptomic changes in M-SFM–cultured macrophages compared with DMEM (+FBS) which identified (E) 1,003 significantly dysregulated genes (effect size > |1| and BH false discovery rate < 0.01). The violin plot represents the relative gene expression and dashed lines represent thresholds. (F) GO Biological Processes and (G) KEGG pathways dysregulated in the presence of FBS. (H) Select dysregulated genes in hMDMs cultured in M-SFM that are important in macrophage cholesterol metabolism. RPMI compared with DMEM (+FBS) shows (I) 360 significantly dysregulated genes (effect size > |1| and BH false discovery rate < 0.01) and gene set enrichment analysis identifies dysregulated (J) Biological Processes and (K) KEGG pathways with (L) select genes associated with macrophage immune regulation.

We refined these analyses by comparing transcriptional changes in hMDMs cultured in DMEM (+FBS) and M-SFM, using DMEM (+FBS) as the reference media. Using a multiple linear regression (BH correction P < 0.01) we identified 1,003 significantly dysregulated genes, most of which were upregulated (Fig. 6E). We further evaluated known biological processes by mapping significantly dysregulated genes to GO and KEGG databases, identifying 18 significant GO terms (P < 0.05). Several significantly dysregulated terms in serum free cultures are associated with key steroid or lipid related pathways such as Cholesterol Biosynthetic Process (GO:0006695), Sterol Biosynthesis Process (GO:0016126), Cholesterol Metabolic Process (GO:0008203), and Lipid Biosynthetic Process (GO:0008610; Fig. 6F). This is supported by findings from the KEGG pathway analysis that identified significantly dysregulated pathways such as Steroid Biosynthesis (BH, P = 6.01 × 10⁻⁹); Synthesis and Degradation of Ketone Bodies (adjusted P = 0.0002); Peroxisome (BH, P = 0.007); and Valine, Leucine, and Isoleucine degradation (BH, P = 9.67 × 10⁻⁷; Fig. 6G). INSIG1 (effect size = 0.721), a gene associated with multiple of the identified pathways (Fig. 6H), encodes a protein important in the regulation of cholesterol metabolism and lipogenesis. Further, Insig1 interactions with other lipid associated proteins in the endoplasmic reticulum can alter macrophage cholesterol metabolism and impact type I IFN responses.⁴⁸ CYP51A1 (effect size = 0.795), which is significantly dysregulated in M-SFM macrophages, is also associated with cholesterol biosynthesis and regulates antimicrobial and antiviral responses in LPS stimulated macrophages (Fig. 6H).⁴⁹ These data suggest that a serum-free environment differentially regulate lipid metabolic pathways in macrophages in ways that are not observed in traditional media formulations and could explain our observation of low inflammatory response to LPS challenge in M-SFM–cultured hMDMs.

In a separate analysis, we compared 2 commonly used macrophage media formulations—DMEM and RPMI, again using DMEM (+FBS) as the reference media. This multiple linear analysis (BH correction P < 0.01) identified 360 genes differentially expressed between DMEM (+FBS) and RPMI (±FBS) (Fig. 6I). Gene set enrichment analyses showed 9 dysregulated terms in the GO biological processes and 2 dysregulated pathways in the KEGG analysis. Dysregulated terms included immune related terms such as Defense Response to Virus (GO:0051607), Negative Regulation of Immune Effector Process (GO:0002698) and terms related to protein regulation such as Positive Regulation of Ubiquitin Ligase Activity (GO:1904668) (Fig. 6J). The 2 dysregulated KEGG pathways are Pyrimidine Metabolism (BH, P = 0.007) and Arginine and Proline Metabolism (BH, P = 0.007), which involve key structural components of cellular processes for DNA and protein metabolism (Fig. 6K). Associated genes include UBE2C (effect size = −1.927), CDC20 (effect size = −1.655) and CSRP1 (effect size = −1.516; Fig. 6L). These findings suggest that transcriptionally, hMDMs differentiated and maintained in DMEM (+FBS) or RPMI (± FBS) show only minor differences; however, these may relate to the differences in NF-κB activation and cytokine secretion observed between hMDMs cultured in these formulations.

Macrophage polarization is a feature often used to categorize the immune response of macrophages,⁵⁰ so we examined how specific culture media compositions might impact macrophage polarization. Specifically, we examined whether media composition could bias cells to toward a “proinflammatory” M1 phenotype relative to a “wound-healing/anti-inflammatory” M2 phenotype.⁵¹ To do this, we identified overlaps in the differentially expressed genes in our analysis of hMDMs in various media and serum supplementation status with a previously published dataset of macrophage polarization associated genes significantly altered from the M0 state, identified by the Spiller Lab.⁵² This analysis showed significant overlap in genes altered in hMDMs cultured in media without FBS supplementation and genes significantly altered with M1 (P = 0.05) and M2A (P = 0.02) polarization from the previously identified dataset.⁵² Independent of FBS supplementation, RPMI cultured hMDMs share 150 significantly altered genes of the 3,518 genes altered in M1 macrophage polarization (P = 0.002). Data showed similar findings in DMEM cultured hMDMs irrespective of FBS supplementation. Interestingly, in M-SFM compared with DMEM (+FBS)–cultured hMDMs, there is an overlap of 137 genes of the 1,424 significantly altered genes associated with M2A polarization (P = 0.09). However, compared with RPMI (+FBS), M-SFM–cultured macrophages show slightly greater overlap with 147 altered genes of the 1,424 associated with M2A polarization (P = 0.01). Taken together, these findings suggest that in addition to known environmental stimuli that trigger macrophage polarization, the media formulation could in part contribute to the macrophage polarization state.

4. Discussion

This study examines the impact of culture media and serum supplementation on human macrophages, specifically evaluating how distinct culture microenvironments alter macrophage immune function, morphology, and transcriptional profile. The most common of these media formulations for macrophage culture are DMEM or RPMI supplemented with FBS and human serum.^53–55 These data show that in hMDMs the activation of NF-κB, secretion of cytokines, and phagocytic activity was not statistically different in DMEM (+FBS) and RPMI (+FBS). While the NF-κB response to LPS in hMDMs from these both culture compositions showed no difference in EC₅₀, there was a significant difference in E_max, suggesting that hMDMs cultured in DMEM (+FBS) can produce a much greater magnitude of NF-κB nuclear translocation, and potentially activity, in response to similar stimulation. Similarly, while there was not a clear trend across the cytokines and chemokines surveyed, but there were several cytokines in which the baseline or LPS-mediated secretion differed between media. And hMDMs differentiated in DMEM (+FBS) or RPMI (+FBS) also showed similar morphologies and little difference in transcriptomic profiles. These data suggest that, in many cases, data produced from macrophages in DMEM (+FBS) and RPMI (+FBS) may be comparable, but there are also likely to be specific assays and targets that differ in responsiveness.

However, there were significant differences in immune activation, morphology, and transcriptional profile in hMDMs cultured in the presence and absence of serum. Morphologically, relative to hMDMs cultured in DMEM (+FBS), DMEM (−FBS) hMDMs are smaller. Interestingly, while the lack of FBS in RPMI-based media did not affect cell size, it did significantly increase cell “roundness.” Compared with all other media, culture in M-SFM (no FBS or human AB serum) produced the largest cells with greatest LWR. Surprisingly, the cell size did not correlate with phagocytosis, as the cells cultured in M-SFM did not show greater phagocytic capacity than those in DMEM and RPMI cultures. Using DMEM (+FBS) as the baseline, a multiple linear regression model found that M-SFM macrophages showed 995 differentially expressed genes. Most of these genes and pathways were associated with lipid metabolism, such as cholesterol and steroid biosynthesis. This suggests significant differences between M-SFM and DMEM (+FBS) or RPMI (+FBS) in the function of key cellular pathways or organelles such as the endoplasmic reticulum, which is well connected to intracellular lipid metabolism.^56,57 Studies have linked lipid metabolism^58–60 as well as specific lipids and fatty acids to inflammatory and anti-inflammatory effects in macrophages,^61,62 potentially explaining, at least in part, the differential NF-κB response to LPS stimulation and low magnitude of cytokine secretion (DMEM + FBS > RPMI + FBS > M-SFM) and low phagocytic capacity (DMEM + FBS > RPMI + FBS > M-SFM) in these cells. It is also possible that the lower phagocytic capacity of M-SFM macrophages reflects a dysregulation of metabolic processes or lysosomal/endocytic function that has not previously been studied in cells cultured with this media.

Examination of these differences in macrophage gene expression, phenotype and function between media suggests that M-SFM–cultured macrophages could be useful for in vitro mechanistic studies of atherosclerosis and obesity, as the pathophysiology of these conditions involves macrophage exposure to greater exogenous lipid content with dysregulation of lipid and metabolic pathways.^63–66 M-SFM–cultured hMDMs may be a useful tool in studies of macrophages associated with a dampened or lower inflammatory activity. For example, at the maternal-fetal barrier, placental macrophages play a central role in maternal-fetal immunity and disease transmission, and those cells display similar characteristics to the M-SFM–cultured hMDMs, such as the potential for proliferation,⁴⁰ are larger in size, and have a lower inflammatory profile.^67,68 Alternatively, M-SFM–cultured cells could be useful in the study of senescence, as an association of macrophage phagocytic capacity with senescence has been identified.⁶⁹ However, these findings in murine macrophages would require human cell validation as several groups have shown differential responses to immune challenge in murine compared with human macrophages,^70–72 highlighting a key caveat and the necessity for further work in understanding hMDM phenotype and immune function.

Several lines of evidence indicate that macrophage polarization is a spectrum in which macrophages can exist with a change in phenotypes depending on the microenvironmental stimuli.⁵¹ In vitro, the M1/M2 dichotomy of macrophage polarization is widely used for phenotypic characterization of macrophages.⁵⁰ Our RNA sequencing studies suggest that media may impact macrophage polarization, showing that genes altered in M-SFM–cultured macrophages overlap with M2A gene sets and genes altered in RPMI (+FBS)–cultured cells overlap with M1 gene sets. These transcriptional profiles may also explain the low levels of phagocytosis in the M-SFM cultures and the relatively higher levels of phagocytosis in RPMI (+FBS) cultures. Morphologic analysis revealed a more “spindeloid” form in M-SFM cultures, which are also associated with an M2 phenotype in mammalian macrophages.^73,74 These connections need further verification, but an understanding of the contribution of media to the baseline polarization state could better inform study design and the in vitro immune responses of macrophages, while also providing information about the factors that could be used to manipulate macrophage phenotype.

Prior studies have shown that FBS is important in both immune and nonimmune cell responses^15,75–77; thus, we performed transcriptome analyses comparing all hMDMs cultured with FBS (DMEM [+FBS] and RPMI [+FBS]) with those cultured without FBS (DMEM [−FBS], RPMI [−FBS], and M-SFM). These analyses showed 1,210 differentially expressed genes, and pathway analysis identified that several altered pathways including cell cycle. The dysregulation of cell cycle pathways is also supported by our data, showing greater Ki-67 positivity (∼8% to 9%) in media supplemented with FBS, although M-SFM media also showed Ki-67 positivity at that level, suggesting that the presence of FBS is not the only factor promoting proliferation. Traditionally macrophages are considered terminally differentiated cells,³⁵ but these studies challenge this concept. Notably, the FBS used in these studies is from the same company, which removes concerns regarding prior findings that vary in company induces differential immune responses.²⁵ However, there were 3 different lots used in throughout this study based on lot availability and storage capacity. It is worth considering that the in vitro monoculture microenvironment may influence these highly plastic cells to display phenotypes and behaviors distinct to an in vivo or in situ environment, although the extent to which this occurs will need to be determined empirically.

A critical caveat to these studies is that there is a limit to the number of variables that can be tested and the lack of exploration of additional factors may limit the utility of the results. For example, while all macrophage donors used in this study were over 18 yr of age, age ranges were not specified, and prior data show that aging can also be a source of variation in macrophage immune response.^78,79 Further, time in culture could also be a factor in the variability in immune response, and protocols culturing hMDMs more than 6 to 7 d should consider possible changes in both gene expression and function induced by longer culture. Both M-CSF and GM-CSF are commonly used to differentiate monocytes into macrophages and have been shown to drive distinct immunophenotypes.^80,81 These studies only used M-CSF and did not specifically evaluate macrophages for differentiation into more inflammatory/M1 (GM-CSF–mediated) or homeostatic/M2 (M-CSF–mediated) phenotypes.^11,34,80,82 Prior studies also suggest other culture factors such as nonessential amino acids or binding proteins may alter macrophage phenotype.^13,83 The findings that M-SFM–cultured macrophages display a dampened inflammatory status across all measures could indicate that M-SFM diminishes the capacity to respond to LPS. This could be due to differences in activation of TLR4, although our data show that TLR4 is expressed at similar transcriptional levels in macrophages from all tested media formulations (Fig. S4). It may also indicate the need for LPS binding proteins for the LPS-TLR4 interaction, as the concentrations of LPS binding proteins likely differs in the presence and absence of serum. However, the altered TLR4 response does not explain the decrease in phagocytic activity seen in the M-SFM macrophages, suggesting that this media formulation could have a broader effect on immune function. Our sequencing analysis shows differences in metabolic requirements of the macrophages from different formulations, and cellular metabolism has been shown to regulate macrophage immune function.^60,84 Further studies are needed to examine these factors in concert to discretely assess their distinct and combined effects on the macrophage immune response. Finally, the variability inherent among human donors makes these studies challenging but also increases the physiological relevance and translatability of the data.

Together, these data demonstrate that the in vitro culture microenvironment, specifically media type and serum supplementation, is an active component of experimental design and should be considered along with the readouts and pathophysiology being studied. These studies identify several differences in immune activity between DMEM or RPMI 1640 supplemented with FBS, although morphologic or transcriptomic differences were minimal. Removal of FBS from these media dramatically reduces immune responsiveness, and culture in macrophage serum free media produces a macrophage with a distinct transcriptomic phenotype and much lower immune responsiveness. It is not clear that there is one specific media that best recapitulates human physiology in vitro, especially in a monoculture model; therefore, it is incumbent on researchers to design studies and compare the known variables while considering media as a potential factor.

Authors contributions

B.c. and P.J.G. contributed to the design and conception of the study. B.c., M.D., Y.A., and P.J.G. designed and analyzed the experiments and B.c., M.D., Y.A., L.S., and K.E. performed the experiments. B.M. helped with analysis of pharmacodynamics analysis and visualization of this data. K.E., M.N., and H.F. performed the RNA sequencing studies. W.D. devised and implemented the statistical analysis and visualization for the RNA sequencing study. All authors contributed to manuscript revision, read, and approved the final submitted version and P.J.G. was responsible for the final approval of the submitted version.

Supplementary material

Supplementary material is available at Journal of Leukocyte Biology online.

Funding

This work was supported by grants from the National Institutes of Drug Abuse, the National Institutes of Mental Health and the National Institutes of Allergy and Infectious Disease, DA057337 and DA058051 (P.J.G.), T32-MH079785 (supporting B.c.), F30AI179472 (B.c.), U01 DA053624 (H.F.), P30-MH092177 (supporting W.D.), as well as the Department of Pharmacology and Physiology at Drexel University College of Medicine.