Diesel exhaust particles induce polarization state-dependent functional and transcriptional changes in human monocyte-derived macrophages

Abstract

Macrophage populations exist on a spectrum between the proinflammatory M1 and proresolution M2 states and have demonstrated the ability to reprogram between them after exposure to opposing polarization stimuli. Particulate matter (PM) has been repeatedly linked to worsening morbidity and mortality following respiratory infections and has been demonstrated to modify macrophage function and polarization. The purpose of this study was to determine whether diesel exhaust particles (DEP), a key component of airborne PM, would demonstrate polarization state-dependent effects on human monocyte-derived macrophages (hMDMs) and whether DEP would modify macrophage reprogramming. CD14+CD16− monocytes were isolated from the blood of healthy human volunteers and differentiated into macrophages with macrophage colony-stimulating factor (M-CSF). Resulting macrophages were left unpolarized or polarized into the proresolution M2 state before being exposed to DEP, M1-polarizing conditions (IFN-γ and LPS), or both and tested for phagocytic function, secretory profile, gene expression patterns, and bioenergetic properties. Contrary to previous reports, we observed a mixed M1/M2 phenotype in reprogrammed M2 cells when considering the broader range of functional readouts. In addition, we determined that DEP exposure dampens phagocytic function in all polarization states while modifying bioenergetic properties in M1 macrophages preferentially. Together, these data suggest that DEP exposure of reprogrammed M2 macrophages results in a highly inflammatory, highly energetic subpopulation of macrophages that may contribute to the poor health outcomes following PM exposure during respiratory infections.

NEW & NOTEWORTHY We determined that reprogramming M2 macrophages in the presence of diesel exhaust particles (DEP) results in a highly inflammatory mixed M1/M2 phenotype. We also demonstrated that M1 macrophages are particularly vulnerable to particulate matter (PM) exposure as seen by dampened phagocytic function and modified bioenergetics. Our study suggests that PM causes reprogrammed M2 macrophages to become a highly energetic, highly secretory subpopulation of macrophages that may contribute to negative health outcomes observed in humans after PM exposure.

INTRODUCTION

Exposure to airborne particulate matter (PM) has been repeatedly linked to increased morbidity and mortality in humans (1–4). PM exposure has been shown to exacerbate existing pulmonary diseases including chronic obstructive pulmonary disease (COPD) (5, 6) and asthma (5, 6) and has been linked to the development of acute pulmonary infections (7, 8). Interestingly, epidemiological data have demonstrated a lag period between exposure to PM and increased hospitalization rates due to pulmonary infections (9–11). In addition, the lag period between PM exposure and disease severity in pulmonary diseases has been suggested to be longer than that of cardiovascular diseases (12–15), suggesting that PM may have a complex mechanism of action in the context of respiratory infection. One potential mechanism linking this lag period to delayed negative respiratory health outcomes following PM exposure is modification of critical innate immune cells, such as macrophages.

Macrophages represent a key cell type necessary for normal pulmonary function and health because of their various roles in development, homeostasis, inflammation, and tissue repair (reviewed in Ref. 16). After exposure to various extracellular stimuli, macrophages undergo a process referred to as polarization, where they undergo a phenotypic shift to proinflammatory or proresolutory states (reviewed in Ref. 17). Although macrophages are frequently characterized as classically activated (M1) or alternatively activated (M2), increasing evidence suggests they exist on a continuum between these states depending on exposure to extracellular cues (17–19). Considering the extremes of the polarization spectrum, M1 macrophages represent a proinflammatory state characterized by pathogen phagocytosis and secretion of inflammatory cytokines and chemokines (19, 20). In contrast, M2 macrophages represent a proresolution state characterized by endocytosis and clearance of dead and dying cells and secretion of cytokines and chemokines necessary for immune regulation and tissue repair (19, 20). Macrophages exposed to PM demonstrate increased proinflammatory cytokine secretion (21, 22), reduced oxidative burst (21, 23), and reduced phagocytosis of both bacteria (24, 25) and viruses (26, 27), suggesting that PM exposure affects macrophage function.

In the context of pulmonary immunity, two lineages of macrophages are commonly considered: alveolar and monocyte derived. Alveolar macrophages (AMs) descend from the fetal liver and populate the lung early in life (28–30), where they demonstrate strong self-renewal properties with minimal contribution from circulating monocytes (30–32). In contrast, monocyte-derived macrophages (MDMs) stem from the bone marrow and are recruited to the lung after exposure to inflammatory stimuli (33, 34). During homeostasis, AMs are the major macrophage population in the lung, responsible for patrolling the air space and initiating the immune response following infection (30, 35). In contrast, after exposure to inflammatory stimuli, MDMs are rapidly recruited to the lung and become the majority macrophage population, gradually declining in number and percentage with time (36, 37) while AMs proliferate to replace lost cells (31, 36, 38). Whereas mature myeloid cells such as MDMs possess a limited life span (39, 40) and therefore a limited ability to cause persistent changes in subsequent immune responses (41), MDMs remain a major population in the lung and serve important roles throughout infection, inflammatory resolution, and wound repair (36, 42–44). As such, modifications of MDM polarization states could have profound effects on respiratory disease progression.

PM exposure has been shown to enhance M1 polarization of macrophages while suppressing M2 polarization (45–47). Previous work has also demonstrated that polarized macrophages can shift from one polarization state to another after exposure to new extracellular stimuli (48–50), though M1 macrophages have demonstrated resistance to reprogramming to M2 states (48). This process, referred to as reprogramming (51, 52), has emerged as a potential treatment for numerous immune-related disorders (52, 53), but it is unclear how improper macrophage reprogramming may impact disease progression in a functioning immune response. Research to date has suggested that macrophages retain little “memory” of their initial polarization state (49, 50), instead expressing key markers indicative of their final polarization condition, similar to those of directly polarized cells. However, these studies were conducted with murine bone marrow-derived macrophages which have clear differences in gene expression patterns compared with human monocyte-derived macrophages (54). We sought to determine whether human peripheral blood monocyte-derived macrophages would demonstrate polarization state-dependent responses to a model pollutant, diesel exhaust particles (DEP), a major component of airborne PM. We further sought to characterize the reprogramming of M2 macrophages following exposure to M1-polarizing stimuli in combination with DEP to determine whether M2 macrophages would exhibit M1, M2, or mixed M1/M2 characteristics after these exposures and to what extent DEP may modify this reprogramming phenotype.

METHODS

Study Subjects

Healthy, nonsmoking adults underwent venous blood draw after giving written informed consent. Subject exclusion criteria included current nicotine use, allergy symptoms, asthma, current illness, pregnancy, and/or nursing. All studies were approved by the University of North Carolina at Chapel Hill School of Medicine Institutional Review Board (IRB no. 11-1363). Studies were conceptualized, performed, and analyzed before disclosure of subject demographic information to minimize potential subjective bias. Key demographics of study participants are included in Table 1.

Table 1.

Study subject demographics

Assay	Sex	Age, yr	BMI, kg/m2	Race
Phagocytosis	1M/4F	31.12 ± 7.65	25.54 ± 5.09	White (5/5)
Cytotoxicity	1M/4F	31.12 ± 7.65	25.54 ± 5.09	White (5/5)
Necroptosis	5F	32.56 ± 6.93	24.38 ± 5.1	White (5/5)
Mesoscale	5F	32.56 ± 6.93	24.38 ± 5.1	White (5/5)
Fluidigm	5F	32.56 ± 6.93	24.38 ± 5.1	White (5/5)
Seahorse	5F	29.2 ± 7.16	23.7 ± 5.46	White (5/5)

Values are means ± SD. BMI, body mass index.

Monocyte Isolation

Venous blood was collected in BD Vacutainer tubes with K₂EDTA (BD no. 367855). Peripheral blood mononuclear cells (PBMCs) were collected through density centrifugation using Ficoll-Paque Plus (Cytivia Life Sciences no. 17-1440). PBMCs were washed three times with Dulbecco’s PBS (DPBS; Gibco no. 14190), and CD14+CD16− monocytes were isolated with magnetic bead negative selection with the EasyStep Human Monocyte Isolation Kit according to the manufacturer’s protocol (StemCell Technologies no. 19359). Resulting cell populations have previously been demonstrated to average ∼90% CD14+ monocytes by flow cytometry (55).

Macrophage Differentiation and Polarization

CD14+CD16− monocytes were seeded in designated tissue culture-treated multiwell plates. Sixty thousand cells were seeded in 96-well plates (Costar 3603) for phagocytosis and cytotoxicity assessments. Two hundred thousand cells were seeded in 48-well plates (Coster 3548) to collect cell culture supernatants and RNA lysates. Sixty thousand cells were seeded in Seahorse XF24 cell culture microplates (Agilent 100777-004) for cellular bioenergetic measurements. Monocytes were differentiated into macrophages for 6 days with base medium consisting of RPMI-1640 (Gibco no. 11875) with 10% fetal bovine serum (FBS; Sigma-Aldrich no. F2442), 1% penicillin-streptomycin (100 U/mL final concentration; Gibco no. 15140), 2.5 mM l-glutamine (Gibco no. 25030), and 40 ng/mL macrophage colony-stimulating factor (M-CSF; R&D Systems no. 216-MC). Medium was replaced with fresh base medium on day 4. On day 6, macrophages were left unpolarized (M0) by replacing culture medium with fresh base medium or were polarized to M2 states with base medium supplemented with 20 ng/mL interleukin-4 (IL-4; R&D Systems no. 204-IL) for 24 h.

Macrophage Treatments

On day 7, M0 and M2 macrophages were exposed to one of four indicated treatments for 24 h: vehicle, 20 ng/mL lipopolysaccharide (LPS; Sigma-Aldrich no. L4391) and 20 ng/mL interferon-gamma (IFN-γ; R&D Systems no. 285-IF) (M1-polarizing conditions), 25 μg/cm² diesel exhaust particles (DEP), or coexposure to M1-polarizing conditions and DEP. DEP were generated by Sagai and colleagues (56) and were suspended in MEM (Gibco no. 11095) to a concentration of 5 mg/mL. The suspension was sonicated with a Fisher Sonic Dismembrator model 500, diluted to a final concentration of 1 mg/mL with MEM, snap frozen in liquid nitrogen, and stored at −80°C until use (57). Particles were thawed and suspended in treatment medium to a final concentration of 25 μg/cm² of growth area. Particle characteristics have been previously described (57). Vehicle treatment medium consisted of base medium supplemented with a volume of MEM equal to the volume of the DEP suspension added to the DEP treatment group medium.

Macrophage Phagocytosis

Phagocytosis of opsonized Staphylococcus aureus pHrodo Red BioParticles was assessed 24 h after indicated treatments. Briefly, S. aureus BioParticles (Invitrogen, A10010) were suspended in DPBS, pH 7.6, and sonicated for 5 min with a Branson B200 water bath sonicator (Branson Ultrasonics no. 15–337-21). BioParticles were mixed with S. aureus BioParticle opsonizing reagent (Invitrogen, S2860) and incubated for 1 h, mixing every 15 min. The particle mixture was washed with DPBS, pH 7.6, and suspended in monocyte base medium, pH 7.6. Treatment solutions were removed and replaced with fresh monocyte base medium, and the particle mixture was added to each well and incubated for 2 h at 37°C. Fluorescence of each well was quantified with a CLARIOstar plate reader (BMG Labtech) at 557/597 nm [excitation(Ex)/emission (Em)]. Background fluorescence was corrected by subtracting fluorescence of cell-free wells incubated with an equivalent volume of BioParticles and monocyte base medium.

Cytotoxicity Measurements

The CellTox Green Cytotoxicity Assay (Promega no. G8742) was performed according to manufacturer’s instructions to determine cytotoxicity. Briefly, after 24-h treatment and 2-h phagocytosis assessment, medium was replaced with fresh base medium. Lysis buffer was added to untreated hMDM wells, and cells were incubated for 15 min at 37°C to induce maximum cytotoxicity. Equal volumes of 2× CellTox Green dye in assay buffer were added to each well and incubated for 15 min at 37°C. Fluorescence was quantified with a CLARIOstar plate reader (BMG Labtech) at 487/535 nm (Ex/Em), and data were normalized by subtracting mean fluorescence intensity (FI) of 1× CellTox Green dye in base medium from FI of each well. Cytotoxicity is presented as percentage of average lysis well FI.

Cytokine and Chemokine Secretion Measurement

After 24-h treatments, medium was collected and frozen at −80°C until all samples were collected. Samples were thawed at 4°C and centrifuged at 1,000 g for 10 min to remove cellular debris. Protein concentrations were measured with the V-PLEX Human Cytokine 30-Plex (Mesoscale Discovery K15054D) and read on the MESO QuickPlex SQ 120 (Mesoscale Discovery) according to manufacturer’s instructions. IL-4 and IFN-γ were excluded from analysis because of their presence in polarization conditions. Values below the calculated lowest limit of detection (LLOD) were set to one-half the calculated LLOD.

Necroptotic Gene Expression

After 24-h treatments and collection of supernatants, cells were washed with ice-cold 1× DPBS (Gibco) and RNA was isolated with the PureLink RNA Mini Kit (Invitrogen no. 1283018A) according to manufacturer’s instructions. RNA was reverse transcribed to cDNA as previously described (58). Real-time quantitative PCR was performed with a QuantStudio 3 Real-Time PCR System (Applied Biosystems no. A28567) with the resulting cDNA, Applied Biosystems TaqMan Universal Master Mix II with UNG (Applied Biosystems no. 4440038), and indicated TaqMan assays as follows: RIPK3 Hs00179132_m1, MLKL Hs04188505_m1, and ZBP1 Hs00229199_m1. Log2 changes in gene expression were calculated by the $2^{\mathrm{-}\Delta \Delta {\mathrm{C}}_{\mathrm{t}}}$ method (where C_t is threshold cycle) (59) with ACTB (Applied Biosystems no. 4326315E) or SDHA, RPL13A, HPRT1, and TBP as the housekeeping genes and vehicle-exposed M0 hMDMs as the reference group.

Fluidigm Gene Expression Measurement

After 24-h treatments and collection of supernatants, cells were washed with ice-cold 1× DPBS (Gibco) and RNA was isolated with the PureLink RNA Mini Kit (Invitrogen) according to manufacturer’s instructions. RNA was reverse transcribed to cDNA as previously described (58) and frozen at −20°C until analysis. Fluidigm integrated fluidic circuit (IFC) was run by the University of North Carolina (UNC) Center for Gastrointestinal Biology and Disease Advanced Analytics Core. TaqMan assays were selected based on previous reports describing changes in their expression based on polarization state and are presented in Table 2. Six housekeeping genes (ACTB, HMBS, HPRT1, RPL13A, SDHA, and TBP) were selected based on previous reports of their expression stability in leukocytes (66), and the four most stably expressed genes (SDHA, RPL13A, HPRT1, and TBP) were determined in our assay with NormFinder software (67). Log₂ changes in gene expression were calculated by the $2^{\mathrm{-}\Delta \Delta {\mathrm{C}}_{\mathrm{t}}}$ method (48) with SDHA, RPL13A, HPRT1, and TBP as the housekeeping genes and vehicle-exposed M0 hMDMs as the reference group.

Table 2.

Fluidigm gene targets, TaqMan assay IDs, and known polarization state links

Gene ID	TaqMan ID	M1/M2 Linked	Citation
ACLY	Hs00982738_m1	M1	(60)
ACOD1	Hs00985781_m1	M1	(60)
SHPK (CARKL)	Hs00950008_m1	M2	(18, 60)
CD163	Hs00174705_m1	M2	(60, 61)
CD36	Hs00354519_m1	M2	(62)
CD47	Hs00179953_m1	M2	(52)
CD80	Hs01045161_m1	M1	(60)
CD86	Hs01567026_m1	M1	(60, 61)
CIITA	Hs00172106_m1	M1	(60)
CTCF	Hs00902016_m1	M2	(63)
E2F1	Hs00153451_m1	M2	(64)
FAS	Hs00236330_m1	M1	(62)
GATA3	Hs00231122_m1	M2	(60)
GPR32	Hs00265986_s1	M2	(52, 65)
HIF1A	Hs00153153_m1	M1	(60)
IDH1	Hs04966975_g1	M2	(60)
IFNG	Hs00989291_m1	M1	(18, 60–62)
IL10	Hs00961622_m1	M2	(60, 61)
IL12A (IL12p35)	Hs01073447_m1	M1	(18, 60, 61)
IL12B (IL12p40)	Hs01011518_m1	M1	(18, 60–62)
IL23A	Hs00372324_m1	M1	(18)
IL6	Hs00174131_m1	M1	(18, 60–62)
CXCL8 (IL8)	Hs00174103_m1	M1	(61)
LDHA	Hs01378790_g1	M1	(60)
CCL13 (MCP-4)	Hs00234646_m1	M2	(62)
CCL22 (MDC)	Hs01574247_m1	M2	(18)
CCL18 (MIP-4)	Hs00268113_m1	M2	(62)
MMP9	Hs00957562_m1	M2	(60)
MRC1 (CD206)	Hs00267207_m1	M2	(18, 60–62)
MYC	Hs00905030_m1	M2	(64)
NOS2	Hs01075529_m1	M1	(18, 60, 61)
PDK1	Hs01561847_m1	M1	(60)
PFKFB3	Hs00998698_m1	M1	(60, 62)
PKM	Hs00761782_s1	M1	(60)
PPARG	Hs01115513_m1	M2	(18, 64)
PTGS2	Hs00153133_m1	M1	(61)
SLC25A1	Hs01105608_g1	M1	(60)
CCL17 (TARC)	Hs00171074_m1	M2	(18)
TLR4	Hs00152939_m1	M1	(18)
TNF (TNFa)	Hs00174128_m1	M1	(18, 61, 62)
TNFRSF1A	Hs01042313_m1	M1	(18)
VEGFA	Hs00900055_m1	M2	(60)
ACTB	Hs01060665_g1	Housekeeping	(40)
HMBS	Hs00609296_g1	Housekeeping	(40)
HPRT1	Hs02800695_m1	Housekeeping	(40)
RPL13A	Hs04194366_g1	Housekeeping	(40)
SDHA	Hs00188166_m1	Housekeeping	(40)
TBP	Hs00427620_m1	Housekeeping	(40)

Cellular Bioenergetics Measurements

After 24-h treatments, hMDMs were assayed for bioenergetic properties with a modified Seahorse Extracellular Flux Cell Mito Stress Test (Agilent) as described previously (68). Briefly, treatment medium was removed, wells were washed once with Seahorse XF RPMI, pH 7.4 (Agilent no. 103576), supplemented with 2 mM l-glutamine (Gibco) (Seahorse medium), and fresh Seahorse medium was added to each well. Cells were placed in a humidified non-CO₂ incubator for ∼30 min before the assay start to allow for CO₂ outgassing. Assay injections were as follows: port A: 10 mM glucose; port B: 1 mM oligomycin; port C: 1.25 mM FCCP; port D: 0.5 mM rotenone and 0.5 mM antimycin A. Per manufacturer’s instructions, mix-wait-measure times were set as 3 min–2 min–3 min. Assay medium was removed immediately after the assay, and cells were stained with 0.01 mg/mL Hoechst 33342 (Invitrogen no. H3570) in DPBS (Gibco) for 10 min. Fluorescence in each well was quantified with a CLARIOstar plate reader (BMG Labtech). Data were normalized in Seahorse Wave Desktop Software 2.6.3 (Agilent), using cell seeding density (60,000 cells) and Hoechst 33342 fluorescence intensity in each well. Mitochondrial and glycolytic parameters were calculated as previously described (68).

Data Analysis and Visualization

Phagocytosis, cytotoxicity, necroptotic gene expression, and Seahorse data were analyzed and visualized with GraphPad Prism 9 software, whereas cytokine/chemokine secretion and Fluidigm gene expression were analyzed and visualized with R studios. Briefly, data were assessed by a repeated-measure (RM) two-way ANOVA with Fisher least significant difference (Fisher LSD) post hoc test comparing within initial polarization groups and within treatment groups, with P < 0.05 being considered statistically significant. Although the Fisher LSD does not correct for multiple comparisons and therefore risks elevating the type I error rate (false positive), the Fisher LSD protects against inflation of the type II error rate (false negative) and is therefore considered to be appropriate for the exploratory nature of our study design (69, 70). To balance this risk of greater type I error rates, results of statistical tests for all comparisons are included in the Supplemental Tables or Supplemental Figures for reference. Principal component analysis (PCA) of cytokine/chemokine secretion and Fluidigm gene expression levels was generated with R studios packages factoextra (71) and ggfortify (72). Row-scaled Z-score heat maps of cytokine/chemokine secretion and Fluidigm gene expression levels were generated with R studios package pheatmap (73). Means and SEs of analyte concentrations and mRNA expressions are presented in Supplemental Tables S4 and S5, respectively. Input data, Supplemental Figures, Supplemental Tables, and R code used for these analyses are available at https://doi.org/10.15139/S3/PV9RQY.

RESULTS

Coexposure to Diesel Exhaust Particles and M1-Polarizing Conditions Reduces Phagocytic Function Independent of Initial Polarization State

To test the effects of diesel exhaust particle (DEP) exposure on macrophage function, peripheral blood mononuclear cells from healthy volunteers were enriched for CD14+CD16− monocytes and differentiated into macrophages (human monocyte-derived macrophages, hMDMs). Macrophages were polarized to indicated conditions for 24 h before being exposed to vehicle, DEP, M1-polarizing conditions (LPS and IFN-γ), or M1-polarizing conditions and DEP for an additional 24 h before assessment (Fig. 1). Exposure to M1-polarizing conditions alone caused a significant reduction in phagocytosis of S. aureus bioparticles in M0 and M2 hMDMs (Fig. 2A). DEP exposure caused a similar reduction in phagocytosis in M0 and M2 hMDMs compared with M1-polarizing condition exposure (Fig. 2A). Coexposure to M1-polarizing conditions and DEP caused an even further reduction in phagocytosis in both M0 and M2 hMDMs, which was significantly lower than either individual condition (Fig. 2A). Each treatment condition induced an increase in cytotoxicity compared with vehicle-treated cells (Fig. 2B). Despite a significant difference between phagocytosis in M0 hMDMs exposed to M1-polarizing conditions and DEP, no difference in cytotoxicity was observed between M1-polarizing conditions and DEP in either M0 or M2 hMDMs (Fig. 2B). As seen in phagocytosis, coexposure to M1-polarizing conditions and DEP caused a greater increase in cytotoxicity compared with vehicle or individual exposure conditions in both M0 and M2 hMDMs (Fig. 2B). Results of individual statistical comparisons are available in Supplemental Table S1.

Figure 1.

Graphical representation of experimental methods. CD14+CD16− monocytes are isolated from the blood of healthy human volunteers and differentiated into macrophages for 6 days. Resulting macrophages are left unpolarized (M0) or polarized to M2 states for 24 h. M0 and M2 macrophages are exposed to indicated treatments for an additional 24 h before assessment of functional readouts. DEP, diesel exhaust particles; M-CSF, macrophage colony-stimulating factor. Image created with a licensed version of BioRender.com.

Figure 2.

M2 human monocyte-derived macrophages (hMDMs) demonstrate M1 properties after 24-h exposure to inflammatory conditions. A: Staphylococcus aureus BioParticle phagocytosis by polarization state. B: cytotoxicity by polarization state. Data represent means ± SE; n = 5 biological replicates per polarization state per treatment. Repeated measures (RM) 2-way ANOVA with Fisher’s least significant difference (LSD). DEP, diesel exhaust particles; FI, fluorescence intensity. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Macrophages Undergo Necroptotic Cell Death after Exposure to M1-Polarizing Conditions

We sought to further examine the mechanisms behind our observed changes in cell death. Recent evidence indicates that M1 macrophages are susceptible to necroptotic cell death (74, 75), a regulated form of necrotic cell death linked to prolonged inflammation due to release of danger-associated molecular patterns (DAMPs) (76), elevated inflammatory cytokine secretion (77), and reduced efferocytotic clearance of dead and dying cells (78). We measured changes in expression of receptor interacting serine/threonine kinase 3 (RIPK3), Mixed Lineage Kinase Domain Like Pseudokinase (MLKL), and Z-DNA Binding Protein 1 (ZBP1), three known markers of the necroptotic cell death pathway (75, 79). RIPK3 (Fig. 3A) and MLKL (Fig. 3B) expressions in vehicle-exposed M2 cells were demonstrated to be significantly lower than in vehicle-exposed M0 cells, suggesting that M2 cells may be resistant to necroptotic-linked cell death at baseline. In contrast, hMDMs exposed to M1-polarizing conditions or coexposed with DEP demonstrated significantly increased RIPK3, MLKL, and ZBP1 expression in both M0 and M2 cells (Fig. 3), although coexposed M2 cells demonstrated reduced MLKL expression compared with coexposed M0 cells (Fig. 3B). Despite causing a clear reduction in phagocytosis and a corresponding increase in cytotoxicity, DEP exposure alone caused no change in RIPK3, MLKL, or ZBP1 expression in M0 hMDMs (Fig. 3). Although DEP-treated M2 hMDMs demonstrated a significant increase in expression of both RIPK3 and MLKL compared with vehicle-exposed M2 cells, expression of both genes was still significantly lower than that observed in M0 cells exposed to DEP (Fig. 3, A and B). Similar results were observed when using four housekeeping genes (SDHA, RPL13A, HPRT1, and TBP) to normalize gene expression levels rather than solely ACTB (Supplemental Fig. S1). Study subject demographics from Supplemental Fig. S1 are presented in Supplemental Table S2. Results of individual statistical comparisons are available in Supplemental Table S3.

Figure 3.

Monocyte-derived macrophages exhibit upregulation of key necroptosis genes after 24-h exposure to M1 conditions. Log₂ fold change of RIPK3 (A), MLKL (B), and ZBP1 (C) mRNA from human monocyte-derived macrophages (hMDMs) exposed to indicated treatments for 24 h. Data represent means ± SE; n = 5 biological replicates per polarization state per treatment. Repeated measures (RM) 2-way ANOVA with Fisher’s least significant difference (LSD). DEP, diesel exhaust particles. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Macrophage Cytokine Secretion Patterns Are Defined by Initial Polarization State

As necroptotic cell death was suggested to be elevated in M1 macrophages and was demonstrated to differ between M0 and M2 cells after DEP exposure, we sought to further characterize mediator release by macrophages following the exposure regimens. Principal component analysis (PCA) clustering suggested that inflammatory mediator secretion patterns are linked to the initial polarization state of hMDMs (Fig. 4A). Although M0→M1 and M2→M1 macrophages demonstrated responses similar to each other compared with vehicle-exposed cells, M2→M1 macrophages showed increased secretion of numerous inflammatory mediators at baseline (Fig. 4B) and after coexposure to DEP (Fig. 4C). This increased secretion is characterized by both greater level of secretion compared with M0→M1 cells as well as several uniquely secreted by M2→M1 cells (Supplemental Fig. S2 and Supplemental Table S4). Interestingly, although coexposure with DEP did not change the overall secretion pattern of either M0→M1 or M2→M1 cells as demonstrated by PCA clustering (Fig. 4A), DEP coexposure did change the measured concentration of several mediators, suggesting that DEP exposure may impact the magnitude of inflammatory mediator secretion in M1 macrophages (Supplemental Fig. S2 and Supplemental Table S4).

Figure 4.

M2 human monocyte-derived macrophages (hMDMs) demonstrate distinct secretory pattern compared with directly polarized M1 cells. A: principal component analysis (PCA) clustering of hMDMs of indicated polarization states. B and C: row-scaled Z-score heatmaps describing secretion profile of vehicle (B)- and diesel exhaust particle (DEP) (C)-exposed hMDMs of indicated polarization states. n = 5 biological replicates per polarization state per treatment.

Macrophage Gene Expression Patterns Are Defined by Final Exposure Conditions

Because of differences in trends between inflammatory mediator responses, phagocytic function, and markers of necroptotic-linked cell death, we examined gene expression patterns of hMDMs after our exposure paradigm to further characterize macrophages, using a next-generation targeted gene expression array. In contrast to mediator secretion and in agreement with phagocytosis and cell death patterns, PCA clustering suggests that initial polarization state had a minimal effect on gene expression patterns following exposures and that final exposure conditions drive changes in gene expression patterns (Fig. 5A). Directly polarized M1 and reprogrammed M2→M1 cells demonstrated similar gene expression patterns after vehicle (Fig. 5B) and DEP coexposure (Fig. 5C), further supporting that initial polarization state has a lower impact on gene expression profiles than final exposure condition. Although significant differences in expression of individual genes can be seen (Supplemental Fig. S3 and Supplemental Table S5), overall gene expression patterns are similar between initial polarization conditions and cluster based on final exposure condition.

Figure 5.

M2 human monocyte-derived macrophages (hMDMs) demonstrate M1-like gene expression patterns after reprogramming. A: principal component analysis (PCA) clustering of indicated polarization states. B and C: row-scaled Z-score heatmaps describing gene expression profile of vehicle (B)- and diesel exhaust particle (DEP) (C)-exposed hMDMs of indicated polarization states. n = 5 biological replicates per polarization state per treatment.

Reprogrammed Macrophages Demonstrate Unique Bioenergetic Properties

We next sought to determine whether reprogrammed macrophages would exhibit different bioenergetic properties compared with directly polarized cells, using the Seahorse Extracellular Flux Modified Cell Mito Stress Tests (68). M2 cells demonstrated significantly increased basal respiration (Fig. 6A) compared with M0 cells. M2 cells exposed to DEP continued to demonstrate significantly increased basal respiration (Fig. 6A), whereas M2 cell coexposure to M1-inducing conditions and DEP demonstrated significantly increased basal (Fig. 6A) and maximum (Fig. 6B) respiration compared with M0 cells given the same coexposure. Interestingly, M0 and M2 cells coexposed to M1-inducing conditions and DEP demonstrated significantly higher basal respiration compared with individual exposures (Fig. 6A), whereas maximum respiration was only increased in coexposed M2 cells compared with M2→M1 cells (Fig. 6B). No difference in spare respiratory capacity was observed between M0 and M2 cell exposures (Fig. 6C). However, both cell types exposed to M1-polarizing conditions or coexposed with DEP demonstrated negligible spare respiratory capacity (Fig. 6C), suggesting that both M0→M1 and M2→M1 cells respire at near-maximum rates at baseline and possess a marginal ability to increase respiratory rate to react to increased energy demands.

Figure 6.

M2 human monocyte-derived macrophages (hMDMs) demonstrate modified cellular bioenergetics after diesel exhaust particle (DEP) and reprogramming conditions. Oxygen consumption rates (OCRs; A–C) and extracellular acidification rates (ECARs; D–F) of hMDMs after indicated treatments. Data represent means ± SE; n = 5 biological replicates per polarization state per treatment. Repeated measures (RM) 2-way ANOVA with Fisher’s least significant difference (LSD). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

In contrast to respiratory patterns, M0→M1 and M2→M1 cells demonstrated increased glycolysis compared with their corresponding vehicle-exposed cells (Fig. 6D). Glycolytic function demonstrated similar trends to respiratory function, as reprogrammed M2→M1 cells had a trending increase in glycolysis (Fig. 6D) and a significant increase in glycolytic capacity (Fig. 6E) compared with directly polarized M1 cells. Interestingly, cells coexposed to M1-polarizing conditions and DEP lost the observed increase in glycolysis (Fig. 6D) and glycolytic capacity (Fig. 6E), suggesting that M1 cells are uniquely susceptible to DEP-induced bioenergetic disruption. Although M2 cells exposed to vehicle or DEP demonstrated significantly higher glycolytic reserves compared with M0 cells receiving the same treatment, no difference in glycolytic reserve was observed between M0 and M2 cell exposure to either M1-polarizing conditions or coexposure to M1-polarizing conditions and DEP (Fig. 6F). These data further suggest that both directly polarized M1 and reprogrammed M2→M1 cells possess minimal abilities to increase energy generation over baseline with or without DEP coexposure. Results of individual statistical comparisons are available in Supplemental Table S6.

DISCUSSION

Airborne particulate matter (PM) has been linked to negative health outcomes following pulmonary infection (1–4), suggesting that PM-induced modification of host defenses likely occurs. Macrophages represent a key innate immune cell impacted by PM exposure because of their interactions with inhaled particles during endocytic clearance (reviewed in Ref. 80). Macrophages have demonstrated functional defects following PM exposure including reduced phagocytosis (24–27), dampened oxidative burst (21, 23), and elevated proinflammatory cytokine secretion (21, 22) following PM exposure. Macrophages can polarize to the proresolution M2 state or the proinflammatory M1 state after exposure to various stimuli (19, 20). As previously described, we used interleukin-4 (IL-4) to induce an M2 polarization state and M1 macrophages were generated with coexposure to lipopolysaccharide (LPS) and interferon-gamma (IFN-γ) (55). We found that human monocyte-derived macrophages (hMDMs) are capable of reprogramming from M2 to M1 conditions after exposure to proinflammatory stimuli. We further determined that, contrary to previous reports (49, 50), reprogrammed M2→M1 macrophages demonstrate a mixed M1/M2 phenotype, considering multiple facets of macrophage function. Finally, we demonstrate that exposure to diesel exhaust particles (DEP) induces functional defects in all polarization states, with a particular effect on M1 macrophages based on initial polarization state.

Reprogrammed M2→M1 macrophages demonstrate clear similarities to directly polarized M1 cells (M0→M1), considering phagocytic ability. Both M0→M1 and M2→M1 cells demonstrated significantly reduced phagocytic abilities compared with vehicle-exposed M0 and M2 cells (Fig. 2A). DEP exposure alone induced a significant reduction in phagocytic function in both M0 and M2 cells (Fig. 2A). Interestingly, both M0→M1 and M2→M1 cells coexposed to DEP demonstrated a much greater reduction in phagocytic ability compared with the individual exposures (Fig. 2A). These findings suggest that DEP exposure can diminish phagocytic function in all macrophage polarization states, but M1 macrophages may be more vulnerable to DEP-induced phagocytic defects. The baseline reduction in reprogrammed M1 macrophage phagocytosis is surprising because of previous findings indicating that M1 hMDMs possess phagocytic abilities similar to M0 and M2 hMDMs (55). M1 macrophages significantly upregulate expression of Fc-gamma receptor 1 (FcyR1, CD64) (55, 81, 82), implying that they would be more capable of phagocytosing opsonized particles like our S. aureus BioParticles (62). In addition, LPS and IFN-γ treatments alone have been shown to have opposing effects on phagocytic ability in macrophages (81–83). IFN-γ-polarized M1 cells have been demonstrated to have reduced phagocytosis despite upregulation of FcyR1, suggesting that other cellular properties likely influence phagocytic function in the context of opsonized particles (81).

One potential mechanism driving our reduction in M0→M1 and M2→M1 phagocytosis is our observed elevation in necroptotic cell death. Necroptosis is a regulated form of lytic cell death characterized by proinflammatory cytokine secretion (77), release of danger-associated molecular patterns (DAMPs) (76), and resistance to efferocytotic clearance (78). We observed a much greater induction of the key necroptotic gene receptor interacting serine/threonine kinase 3 (RIPK3), mixed lineage kinase domain like pseudokinase (MLKL), and Z-DNA binding protein 1 (ZBP1) after exposure to M1-polarizing stimuli compared with DEP exposure (Fig. 3). Previous reports have demonstrated that M1 macrophages are more prone to undergo necroptotic cell death compared with unpolarized and M2 macrophages (74, 75). Whereas IFN-γ, LPS, and IL-4 exposure induced similar levels of RIPK3 protein in bone marrow-derived macrophages (BMDMs), IFN-γ- and LPS-exposed macrophages demonstrated greater MLKL and ZBP1 protein levels compared with IL-4 treatment (75). Although DEP exposure alone did not induce an increase in necroptotic-linked genes in M0 cells, DEP-exposed M2 cells did demonstrate a significant increase in RIPK3 and MLKL expression compared with vehicle-exposed M2 cells (Fig. 3, A and B). However, their expression was significantly lower than that in M0 cells exposed to DEP, suggesting that our observed increase in expression was likely due to the lower baseline expression levels. This finding suggests that DEP exposure alone does not induce appreciable necroptotic cell death in M0 or M2 hMDMs. In contrast, we observed that coexposure to M1-polarizing conditions and DEP induced significantly increased MLKL and ZBP1 mRNA in both M0→M1 and M2→M1 hMDMs compared with cells exposed to M1-polarizing conditions alone (Fig. 3, B and C). As MLKL is responsible for disruption of the plasma membrane leading to cellular lysis (84), its upregulation is particularly likely to correspond with increased necroptotic cell death. These findings suggest that DEP likely enhances necroptotic cell death specifically in M1 cells.

As necroptotic cells have been demonstrated to induce cytokine production in a cell-autonomous manner (77), we sought to determine whether cytokine and chemokine secretion profiles would differ between M0→M1 and M2→M1 cells and whether DEP exposure would modify this secretion profile. We determined that M2→M1 cells demonstrated elevated proinflammatory cytokine and chemokine secretion to an even greater degree than M0→M1 cells (Fig. 4B; Supplemental Fig. S2 and Supplemental Table S4). We further determined that DEP exposure had a minimal effect on the overall secretion profile as demonstrated by PCA clustering patterns (Fig. 4A) but that coexposure modified the secretion of several mediators compared with those exposed to M1-polarizing conditions alone (Supplemental Fig. S2 and Supplemental Table S4). However, we observed significantly elevated concentrations of numerous mediators in both coexposed groups compared with their corresponding vehicle exposures, suggesting that they retain a proinflammatory secretory profile after DEP coexposure (Supplemental Fig. S2 and Supplemental Table S4). This finding suggests that the initial polarization state, rather than final exposure conditions, determines the inflammatory secretion profile of macrophages. The implication of elevated cytokine and chemokine secretion in M2→M1 macrophages is important, as elevated bronchoalveolar lavage (BAL) cytokine levels following bacterial and viral infection are associated with poorer health outcomes (85–88). Importantly, other cell types such as epithelial cells are known to secrete numerous proinflammatory cytokines and chemokines such as interleukin-6 (IL-6) and interleukin-8 (IL-8) after exposure to bacteria (89), viruses (90), and DEP (91). These mediators are heavily implicated in disease severity because of their role in the neutrophil response (92, 93) which induces significant tissue injury after recruitment (94, 95). However, macrophage and epithelial cell cocultures have been demonstrated to secrete elevated levels of proinflammatory cytokines and chemokines after PM exposure compared with monocultures (60, 96, 97), suggesting that our reported cytokine and chemokine levels could be lower than those generated from similar exposures in a coculture model. As such, reprogrammed macrophages may represent a hyperinflammatory subset of macrophages that may contribute to excessive inflammation and poorer prognosis following pulmonary infection.

Despite these findings, we determined that final exposure condition, rather than initial polarization state, drives the broader gene expression profile of M0→M1 and M2→M1 macrophages (Fig. 5). DEP exposure induced broad and similar changes in gene expression patterns in M0→M1 and M2→M1 macrophages (Fig. 5, A and C). Our tested genes spanned key cell surface receptors, functional enzymes, cytokines, and transcription factors that have been previously implicated in M1 or M2 polarization (see methods). As hundreds of genes are differentially expressed based on polarization state (61, 63, 64), we risk over- or underestimating changes in gene expression patterns in our cells. In addition, M1 cells demonstrate a greater change in gene expression pattern than M2 cells after polarization (65), suggesting that future investigations should focus on changes in M1-linked gene expression patterns. Epigenetic modifications have also been implicated in macrophage polarization (98–101); however, numerous studies have demonstrated macrophage plasticity, allowing them to transition from one polarization state to another after exposure to new stimuli (48–50). As such, it is unclear to what extent reprogramming modifies epigenetic patterns that may be established after initial polarization or how durable initial epigenetic programming may be in macrophages (98, 102). Future studies may be required to determine the extent and durability of macrophage epigenetic modifications in polarization and reprogramming. Finally, we did not investigate changes in any microRNAs (miRNAs) which can regulate the expression of other genes at the posttranscriptional level (reviewed in Ref. 103). Differential expression of miRNAs has been demonstrated between M1 and M2 polarized macrophages (104–106), further suggesting that the present study does not provide an exhaustive view of global gene expression patterns in M0→M1 and M2→M1 macrophages, presenting a limitation of this study. Still, our findings suggest that reprogrammed M2→M1 macrophages do possess M1-like gene expression patterns as opposed to M2-like patterns. Interestingly, M2→M1 macrophages may demonstrate unique responses to DEP exposure as demonstrated by changes in expression of individual genes (Supplemental Fig. S3 and Supplemental Table S5) and treatment-specific PCA clustering profiles (Supplemental Fig. S4). However, further work will be required to determine the extent of changes in gene expression patterns in reprogrammed M2→M1 macrophages with and without DEP coexposure.

Bioenergetic properties are major regulators of macrophage function. M1 polarization is associated with glycolysis (107, 108), whereas M2 polarization is associated with mitochondrial oxidative phosphorylation (109, 110). A shift toward glycolytic function in M1 macrophages is vital for the production of rapid bursts of energy and reactive species necessary for pathogen phagocytosis (111, 112) and destruction (113, 114), and inhibition of glycolysis has been demonstrated to reduce proinflammatory cytokine production (113, 115), phagocytosis (111, 112), and pathogen killing (113, 114). Modification of bioenergetic pathways has been directly linked to changes in macrophage polarization (101, 116), reinforcing the close link between energy generation and cellular function. However, it remains unclear to what extent the local environment drives changes in macrophage functions inducing bioenergetic changes versus bioenergetic changes inducing broader functional modification (101, 117). We were interested in determining whether reprogrammed M2→M1 macrophages would demonstrate abnormal bioenergetic properties and whether exposure to DEP would differentially affect bioenergetic properties based on polarization state. Similar to our previous study (55), M2 macrophages demonstrated higher baseline and maximum respiratory potential compared to M0 and M0→M1 macrophages (Fig. 6, A and B). In addition, M0→M1 macrophages demonstrated elevated baseline and maximum glycolysis compared with M0 and M2 macrophages (Fig. 6, D and E). We identified a complex bioenergetic profile in M2→M1 macrophages characterized by M2-like respiratory and M1-like glycolytic properties (Fig. 6). Interestingly, M2→M1 macrophages demonstrated greater overall bioenergetic properties characterized by significantly elevated maximum respiratory (Fig. 6B) and maximum glycolytic (Fig. 6E) rates as well as elevated, but nonstatistically significant, increases in basal respiratory (Fig. 6A) and basal glycolytic rates (Fig. 6D) compared with M0→M1 cells. Surprisingly, DEP exposure had a minimal effect on respiratory or glycolytic function in either M0 or M2 macrophages, despite previous reports suggesting that PM exposure modifies mitochondrial respiration (118, 119). In contrast, DEP coexposure induced greater basal respiratory rates compared with single treatments in both M0→M1 and M2→M1 cells (Fig. 6A). DEP coexposure also suppressed the elevated basal (Fig. 6D) and maximum (Fig. 6E) glycolytic function of both M0→M1 and M2→M1 hMDMs, suggesting that DEP preferentially shifts M1 macrophages from glycolysis to oxidative phosphorylation without major effects on M0 or M2 hMDM bioenergetics. Finally, M2→M1 macrophages demonstrated greater basal and maximum respiratory function compared with M0→M1 cells with and without DEP coexposure. Together, these findings demonstrate that M2→M1 macrophages demonstrate mixed M1/M2 bioenergetic properties characterized by elevated mitochondrial oxidative phosphorylation and glycolysis. These data further suggest that M1 macrophages are uniquely vulnerable to DEP-induced bioenergetic dysfunction characterized by a shift toward mitochondrial oxidative phosphorylation at the expense of glycolytic function, which has been heavily linked to antimicrobial function.

Although our use of 25 μg/cm² DEP is comparable to previous in vitro studies (45–47, 120, 121), it must be considered that the applied concentrations of DEP used in the studies described here are far higher than those seen in typical human exposures. Computational methods have estimated deposition fractions of particles with an aerodynamic diameter of <2.5 μm (PM_2.5) from 30% to 90% in the total respiratory tract depending on the specific particle diameter (122–124). Although the total deposited dose over a 24-h period will depend on the ambient PM concentrations, individuals in highly polluted regions would experience lower average deposition concentrations, assuming an adult human lung surface area of ∼80 m² (125) and a breathing rate of 0.54 m³/h (126). However, points of bifurcation within the airway are known to produce highly turbulent air flow which has been modeled to cause increased local particle deposition (127–131). Depending on particle size and inspiratory flow rate, these local enhancement factors (ratio of local to average particle deposition) can be as high as 2,400 (129–131), suggesting that higher particle concentrations may be appropriate for short-term in vitro studies to better stimulate worst-case exposure scenarios. Regardless, our findings regarding modifications of necroptotic cell death (Fig. 3), secreted mediator profiles (Fig. 4), and cellular bioenergetics (Fig. 6) in M0→M1 and M2→M1 macrophages exist independent of our DEP exposures, suggesting that our findings are relevant in situations involving PM exposures as well as in broader human health conditions.

In conclusion, we demonstrate that M2→M1 reprogrammed hMDMs demonstrate a complex, mixed M1/M2 phenotype in contrast to previous reports suggesting that reprogrammed macrophages are indistinguishable from directly polarized cells. Our results demonstrate that M2→M1 macrophages demonstrate phagocytic function, necroptotic death rates, and gene expression profiles similar to M0→M1 cells but possess a greater potential for secretion of proinflammatory cytokines and chemokines, as well as a highly energetic mixed M1/M2 bioenergetic profile. DEP exposure induced similar changes in phagocytic function, necroptotic cell death rates, and gene expression profiles in M0→M1 and M2→M1 macrophages. DEP coexposure shifted M1 macrophage bioenergetics from glycolysis to oxidative phosphorylation regardless of initial polarization state, but M2→M1 macrophages maintained their overall higher respiratory rates. Our results suggest that reprogrammed M2→M1 macrophages represent a uniquely hypersecretory, highly energetic population of macrophages in the context of pulmonary infection and DEP exposure. We further demonstrated that DEP exposure greatly diminishes the antimicrobial functions of M1 macrophages regardless of initial polarization state but that reprogrammed M2→M1 macrophages maintain a highly energetic and proinflammatory profile after coexposure. These findings suggest that reprogrammed M2→M1 macrophages may play an important role in the negative health outcomes following pulmonary infection and particulate matter exposure because of their highly inflammatory yet poorly antimicrobial nature.

DATA AVAILABILITY

Data will be made available upon reasonable request.

SUPPLEMENTAL MATERIAL

Supplemental Tables S1–S6, Supplemental Figs. S1–S4, and Supplemental R Code: https://doi.org/10.15139/S3/PV9RQY.

GRANTS

This research was funded by National Institute of Environmental Health Sciences (NIEHS) T32 ES007126 and R01 ES031173. The UNC Center for Gastrointestinal Biology and Disease Advanced Analytics Core, which performed the Fluidigm IFC presented in this paper, was funded by National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) P30 DK034987.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

T.S. and I.J. conceived and designed research; T.S. performed experiments; T.S. analyzed data; T.S. interpreted results of experiments; T.S. prepared figures; T.S. drafted manuscript; I.J. edited and revised manuscript; I.J. approved final version of manuscript.