The BV2 mouse microglial cell line was obtained from Banca Cellule ICLC, Genova, Italy. Lipopolysaccharide (LPS) 500X was obtained from Invitrogen (Waltham, MA). Murine IL-4 was obtained from Peprotech Inc. (Rocky Hill, NJ). NOX2 inhibitors apocynin and GSK2795039 (GSK) were purchased from Millipore Sigma (Burlington, MA) and MedChemExpress (Monmouth Junction, NJ), respectively. Primary antibodies against phospho-p47 ^phox (Ser 370), PER2, and BMAL1, and goat anti-rabbit secondary antibody were obtained from Invitrogen. Phosphate buffered saline (PBS) and trypsin-EDTA for cell culture were obtained from Gibco, ThermoFisher Scientific (Waltham, MA). The Amido Black stain was obtained from ThermoFisher Scientific.

BV2 cells were cultured in RPMI-1640 medium (ThermoFisher Scientific) with L-glutamine (2 mM) and supplemented with 10% heat-inactivated FBS (ThermoFisher Scientific). All circadian studies used a serum-shock method to synchronize the cells (22). The cells were plated in normal growth media (RPMI + 10% FBS) at a density of 10⁵ cells/mL in 6-well plates and allowed to reach 90% confluence. Culture media was then replaced with starvation media (RPMI) for 24 h. Following media starvation, the cells were subjected to serum-shock using 50% FBS for 2 h to synchronize their clock (22, 37–39). Post-serum shock, normal growth media was used, and the cells were allowed to recover for 16 h from the serum-shock and attain homeostasis. Samples were collected every 2 h for 24 h, starting at 16 h post serum-shock (Hours Post Serum Shock 16 or HPS16), and processed for RT-qPCR and Western blotting.

Bone marrow was extracted from the tibias and femurs of 3-6 month old male Per2:Luc (C57BL/6J) mice. The bone marrow progenitor cells plated on 35 mm cell culture plates, were differentiated with DMEM supplemented with M-CSF and 10%FBS as mentioned in a previous protocol (22, 39). After 7 days, the differentiated macrophages were synchronized by a 24 h starve in serum-free media followed by a 2 h serum-shock with 50% FBS as mentioned previously (22, 39). Post serum-shock synchronization, cultures were sealed with grease and glass cover slips and luminescence was measured over a period of 5 days using LumiCycle32 (Actimetrics, Wilmette, IL) with cells plated in Leibovitz media containing Luciferin and 10% FBS (22, 39).

The High Pure RNA Isolation Kit from Roche Molecular Systems, Inc. (Branchburg, NJ) was used for RNA extraction and purification. Purified RNA was reverse-transcribed using the High-Capacity cDNA Reverse Transcription Kit (ThermoFisher Scientific). mRNA abundance was determined by quantitative PCR using TaqMan Gene Expression Master Mix (Applied Biosystems, Waltham, MA) and the following pre-designed TaqMan primer assays: Per2 (Assay ID: Mm00478099_m1), Bmal1 (Assay ID: Mm00500223_m1), Hprt1 (Assay ID: Mm03024075_m1), p47^phox (Assay ID: Mm00447921_m1) and gp91^phox (Assay ID: Mm01287743_m1). Hprt1 was used as the reference gene for all RT-qPCR experiments. Data obtained from RT-qPCR was then input into the ECHO software (see Supplementary Information ) for analysis of oscillation of gene expression (40).

Western blotting was used to investigate the oscillation of PER2 and BMAL1 clock proteins, and for quantification of phosphorylated-p47 ^phox levels in BV2 microglia. Time course samples were collected from synchronized BV2 cells as described above. For measurement of phospho-p47 ^phox levels, BV2 cells were plated in 6-well plates at a density of 10⁵ cells/mL and allowed to reach 90% confluence. At confluence, cells were treated with LPS (1 μg/mL), or LPS (1 μg/mL) + apocynin (100 μM), or LPS (1 μg/mL) + GSK2795039 (25 μM) for 24 h. BV2 cells with no additives were used as a control. For all samples, cells were washed with ice-cold PBS and lysed with RIPA lysis buffer (Millipore Sigma, Burlington, MA) containing the Halt Protease Inhibitor Cocktail (ThermoFischer Scientific). Protein concentrations were measured using the Pierce BCA Protein Assay Kit (Pierce, Waltham, MA). Samples were loaded at 25 μg of total protein per lane, resolved by SDS-PAGE (BioRad, Hercules, CA), and transferred onto nitrocellulose membranes (Bio-Rad). Following blocking with 5% skim milk in PBST (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄, pH 7.4, 0.2% Tween-20) for 2 h at room temperature, the membranes were incubated with primary antibodies against PER2 (1:1000, Invitrogen, Waltham, MA), BMAL1 (1:1000, Invitrogen), and phospho-p47 ^phox (Ser 370) (1:1000, Invitrogen) at 4⁰C in 1% skim milk in PBST overnight. Post-overnight incubation, the membranes were washed six times with PBST and incubated with secondary antibody (1:10,000) for 1 h. The membranes were then washed with PBST, and the protein bands were detected using SuperSignal West Pico Chemiluminescent Substrate (Pierce). Amido black staining (see below) was used to normalize protein loading. Protein bands were detected using the BioRad ChemiDoc XRS+ Imager (BioRad). Quantification of detected protein bands was performed with the ImageLab 6.0.1 software (Life Science, Waltham, MA).

Amido Black staining was used to measure total protein loaded in a lane as the internal loading-control for Western blots. Following detection of protein bands, membranes were washed three times in DI water. The membranes were then stained for 1 min using a staining solution containing 0.1% amido black reagent in 10% acetic acid solution. The membranes were then washed twice for 1 min with a de-staining solution containing 5% acetic acid. Post de-staining, the membranes were washed two times for 10 min each in DI water. The membranes were then air-dried and visualized using a Bio-Rad ChemiDoc XRS+ Imager. Quantification of amido black stained membranes was done using the ImageLab 6.0.1 software.

ECHO version 4.1 was used to analyze all transcriptional and protein data obtained from the time-course experiments (40). All data was free run with the ‘smooth data’ and ‘linear detrend’ options selected. For transcriptional data, the ‘normalize data’ option was also selected. Data obtained from ECHO analysis was then transferred and re-plotted on GraphPad Prism 7 (GraphPad Software, CA, USA). Details about the ECHO software and ECHO data for all transcriptome and proteome oscillations are available in the Additional File 6: Table S1 .

ROS released by BV2 in the presence of LPS (1 μg/mL) and NOX2 inhibitors apocynin (100 μM) and GSK2795039 (25 μM) was measured using two different assays. The DCFDA/H2DCFDA Cellular ROS Assay Kit (Abcam, MA) was used to measure the levels of intracellular superoxide produced and the Amplex Red Hydrogen Peroxide/Peroxidase Assay kit (Invitrogen) was used to measure the levels of extracellular hydrogen peroxide produced in BV2 cells. Cells were plated in 96-well black-walled flat-bottom plates at a density of 10⁵ cells/mL. The cells were incubated with LPS with and without a NOX2 inhibitor for 2 h in the assay media and fluorescence was measured using a SpectraMax plate reader (Molecular Devices, San Jose, CA). Fluorescence values were normalized to cell number measured using the Presto Blue Cell Viability Assay (ThermoFischer Scientific) as described below.

BV2 cells were seeded in 96-well flat-bottom, black-walled plates at a density of 10⁴ cells/well and allowed to reach 90% confluence. The cells were then treated with media containing different additives. The media was then removed from each well and replaced with 90 μL fresh media and 10 μL PrestoBlue reagent (ThermoFisher Scientific) and incubated at 37°C and 5% CO₂ for 10 min. Fluorescence was measured at Ex/Em = 560/590 nm and cell numbers were correlated against a standard curve.

A sandwich enzyme-linked immunosorbent assay (ELISA) was used to quantify cytokines produced by BV2 cells and primary macrophages. The cells were plated in 96-well plates at a density of 10⁵ cells/mL and allowed to reach 90% confluence. The cells were then incubated for 24 h with media containing the following combination of additives: LPS (1 μg/mL); LPS (1 μg/mL) + apocynin (100 μM); apocynin (100 μM); LPS (1 μg/mL) + GSK (25 μM); and GSK (25 μM). To understand the influence of IL-4 addition on cytokine production, cells were incubated with the following combination of additives: IL-4 (20 ng/mL); LPS (1 μg/mL); IL-4 (20 ng/mL) + LPS (1 μg/mL); and IL-4 (20 ng/mL) for 24 h followed by LPS (1 μg/mL) for 24 h. Untreated BV2 cells were used as a control. Post-incubation, the supernatants were collected and levels of two pro-inflammatory cytokines, TNF-α and IL-6, were measured using the DuoSet ELISA Assay kit (R&D Systems, MN, USA). ELISA data was analyzed using GraphPad Prism 7.

Data for RNA-sequencing and proteomics in BMDMs were acquired and analyzed as mentioned previously (22).

One-way ANOVA was used for biological replicates. For each data point, triplicate biological replicates, each with triplicate technical replicates, were performed. Circadian rhythmicity was determined using ECHO analysis software. Statistical analyses were performed using GraphPad Prism 7 or Excel software (Microsoft).

Per2 and Bmal1 are the best studied representatives of the positive and negative arm of the circadian transcriptional-translational feedback loop with predictable relationship (19, 20). They have been extensively studied in BV2 microglia, and hence, were employed as indicators of clock activity (41–43). To investigate whether the clock was functional in the BV2 microglial cell line, Per2 and Bmal1 gene expression were quantified in naïve (resting) BV2 microglial cells. To assess transcriptional levels of both clock genes, samples were collected starting at 16 h post-serum shock (Hours Post Serum Shock 16 or HPS16), every 2 h for 24 h ( Figure 1A ). Total RNA was isolated from each sample, and Per2 and Bmal1 transcript levels were measured using RT-qPCR. The Extended Circadian Harmonic Oscillator (ECHO) application (40) was used to identify oscillations in the transcriptional data and determine whether the gene products oscillated with a circadian period. The ECHO algorithm uses an extended solution of the fixed amplitude oscillator that incorporates the amplitude change coefficient and provides detailed information on the nature of circadian oscillations (40). ECHO analysis indicated that both Per2 (ECHO period = 16 h, ECHO p-value = 8.31 x 10^-13) ( Figure 1B ) and Bmal1 (ECHO period = 16 h, ECHO p-value = 1.11 x 10^-10) ( Figure 1C ) displayed oscillations at the transcriptional level, although these oscillations are shorter than those considered to be circadian. Both Per2 and Bmal1 mRNA oscillations had peaks at around HPS20 and HPS36 post serum shock and a trough around HPS28, paralleling what has been seen in other studies of BV2 microglia (41, 42). To further analyze the circadian clock components in BV2 microglia, Western blotting was performed using anti-PER2 and anti-BMAL1 antibodies over circadian time to quantify protein levels ( Figure 1D ). Both PER2 ( Figure 1E ) (ECHO period = 16 h, ECHO p-value = 2.7 x 10^-8) and BMAL1 ( Figure 1F ) (ECHO period = 16 h, ECHO p-value =1.3 x 10^-4) displayed significant oscillations according to ECHO with periods and phases that corresponded to what was noted for these genes at the transcript level.

A functional circadian clock in bone-marrow derived mouse macrophages is receptive to the inflammatory state of the cells (31). Specifically, the activation of bone-marrow derived macrophages into a pro-inflammatory phenotype using LPS, TNF-α, or IFN-γ suppressed the oscillation of core clock components Per2 and Nr1d1, whereas anti-inflammatory activation using IL-4 enhanced their expression (31). To assess whether a similar result is observed in microglia, we subjected BV2 cells to both pro- and anti-inflammatory activation. The bacterial endotoxin LPS, and the cytokine IL-4 have been successfully used in previous work and are known to be the strongest polarizing stimulants to activate BV2 microglia. Hence, LPS and IL-4 were used to activate BV2 microglia into pro- and anti-inflammatory states, respectively (44–52). Therefore, to evaluate the influence of LPS exposure on the circadian clock in BV2 microglia, BV2 cells were synchronized using serum-shock, and LPS (1 μg/mL) was added to the growth media post serum-shock. Time course samples for RT-qPCR analysis were collected every 2 h for 24 h starting at HPS16. We found that LPS exposure suppressed the oscillation of the clock genes Per2 and Bmal1 in BV2 microglia ( Figure 2A ). This loss of rhythm in the expression of the positive and negative arm components of the circadian feedback loop indicates that the pro-inflammatory activation of BV2 microglia has a suppressive effect on the circadian system, paralleling what was observed in primary macrophages (31).

The effect of the activation of BV2 microglia into the anti-inflammatory phenotype was assessed by exposing the BV2 cell line to IL-4 (20 ng/mL) after the BV2 cells were synchronized using serum-shock. As with LPS exposure, time course samples for RT-qPCR analysis were collected every 2 h for 24 h starting at HPS16. Unlike for LPS addition, the expression of Per2 (ECHO period = 16 h, ECHO p-value = 1.04 x 10^-15) and Bmal1 (ECHO period = 16 h, ECHO p-value = 2.62 x. 10^-13) remained oscillatory ( Figure 2B ) following exposure to IL-4. However, the characteristics of the oscillation of Per2 and Bmal1 changed in response to the addition of IL-4. Compared to the resting state, upon IL-4 exposure both Per2 and Bmal1 underwent a phase inversion, with a peak around HPS24 to 28 and HPS28 to 32 respectively in the IL-4 activated microglia, as opposed to the peak at around HPS20 for both the genes in the resting state (compare Figures 1 , 2 ). This change in the oscillatory behavior of IL-4 activated BV2 cells vs. resting cells, however, does not abrogate the retention of a daily oscillation in the anti-inflammatory state.

A hallmark of a cellular pro-inflammatory state is the presence of oxidative stress, as reflected in the production of ROS (9, 53). NOX2 is a major contributor to ROS production leading to oxidative stress (11, 15, 32, 54–57). With the known connection between the clock and inflammation, we hypothesized the circadian clock may control Nox2 gene expression in BV2 microglia. NOX2 is comprised of six subunits, with gp91 ^phox and p22 ^phox serving as integral membrane proteins that together form the large heterodimeric subunit flavocytochrome b₅₅₈ (cytb₅₅₈) (58), and p40 ^phox , p47 ^phox , p67 ^phox , and Rac together forming the cytosolic subunits (55). Activation of NOX2 occurs though a complex series of protein-protein interactions and translocation (59, 60) driven by phosphorylation of p47 ^phox , which results in attachment of phospho-p47 ^phox to the p22 ^phox component of cytb₅₅₈, thus forming an active enzyme complex. For this study, we focused on two essential sub-units of NOX2; p47^phox and gp91^phox (11).

To assess whether gene expression of p47^phox and gp91^phox was under circadian control in BV2 microglia, time course samples were collected every 2 h for 24 h post-HPS16 from serum-shock synchronized BV2 cells and RT-qPCR was used to measure the expression of p47^phox and gp91^phox . In the resting state, both p47^phox (ECHO period = 16 h, ECHO p-value = 1.7 x 10^-11) and gp91^phox (ECHO period = 16 h, ECHO p-value = 1.3 x 10^-17) displayed significant oscillations, matching the oscillations in Per2 and Bmal1 ( Figure 3A ). ECHO plots for both subunits showed peaks around HPS20-24 and HPS36-40, and a trough around HPS28-32, paralleling what was seen for Per2 and Bmal1 in the resting state in BV2 cells.

It is known that NOX2 is activated upon addition of LPS (61). Given what we found relating to the effect of LPS on the clock in BV2 microglia, we next assessed the effect of LPS on NOX2 component oscillations. The addition of 1 μg/mL LPS after serum synchronization resulted in the loss of circadian oscillations in both p47^phox and gp91^phox transcript levels ( Figure 3B ). Conversely, the addition of 20 ng/mL IL-4 after serum synchronization did not result in a loss of circadian oscillations for either p47^phox (ECHO period = 16 h, ECHO p-value = 4.98 x 10^-7) or gp91^phox (ECHO period = 16 h, ECHO p-value = 1.57 x 10^-6) ( Figure 3C ). Interestingly, there was again a shift in phase between the resting and IL-4 activated states, paralleling what was seen in Per2 and Bmal1 after the addition of IL-4. p47^phox expression underwent a phase inversion, with a peak at HPS28-32, as compared to a peak at HPS20 in the resting state. Thus, the two key NOX2 subunits are under circadian regulation in BV2 microglia.

Given the oscillation of NOX2 and the loss of circadian gene expression both in clock and Nox2 genes in the pro-inflammatory state, we hypothesized there was a role for NOX2 in clock-driven inflammation. To validate this role for NOX2 in circadian-driven inflammation, we next assessed the effect of NOX2 inhibitors on oscillations in the core clock gene transcripts as well as NOX2 gene expression. To do so, we employed two NOX2 inhibitors to suppress NOX2 function – apocynin (4-hydroxy-3-methoxy-acetophenone) (62–64) and GSK2795039 (GSK, N-(1-Isopropyl-3-(1-methylindolin-6-yl)-1H-pyrrolo[2,3-b]pyridin-4-yl)-1-methyl-1H-pyrazole-3-sulfonamide) (65). Apocynin is known to block the translocation of p47 ^phox to the membrane and its interaction with membrane-bound p22 ^phox , thereby preventing the formation of the active enzyme complex (66). Apocynin in not NOX2 specific but is used indiscriminately as a broad NOX inhibitor (67). GSK is a competitive reversible inhibitor of NADPH binding to cytb₅₅₈ in vitro and in vivo (65), and is selective for NOX2 as compared to the other NOX isoforms, xanthine oxidase, and endothelial nitric oxide synthase (65).

As a proxy for the activity of NOX2 in the presence of LPS and inhibitors, we determined ROS levels via Amplex Red (extracellular) and DCFDA (intracellular) assays. For both assays, additives such as LPS, apocynin or GSK were introduced to serum-shock synchronized BV2 at HPS20 and ROS was measured upon exposure of the BV2 cells to LPS only, LPS with either apocynin or GSK, or no additive (control) for 2 h. In the presence of LPS, both extra- and intracellular ROS levels in BV2 microglia increased relative to the control ( Figures 4A, B ). In the absence of LPS, neither apocynin nor GSK influenced the levels of ROS produced ( Figures 4A, B ). However, in the presence of both LPS and either apocynin or GSK, the levels of both extra- and intracellular ROS were significantly reduced relative to the presence of LPS alone, and were even significantly lower than the control ( Figures 4A, B ). These results suggest that the inhibition of NOX2 may prevent ROS production in BV2 microglia as compared to even basal levels as well as in the presence of LPS.

A key step in the activation of NOX2 is the phosphorylation of the p47 ^phox subunit (55). To confirm that inhibition of ROS levels by apocynin was NOX2 dependent, even in the presence of LPS, Western blots were obtained to quantify levels of phospho-p47 ^phox upon LPS and apocynin addition ( Figure 4C ). Given that traditionally used loading controls, such as GAPDH and β-Actin, are under the control of the circadian clock, Amido Black staining was used as an internal loading control to measure total protein loaded in a lane and normalize protein levels (68). In the presence of LPS, there was a 40% increase in phospho-p47 ^phox levels. In the presence of LPS + apocynin, however, levels of phospho-p47 ^phox fell to approximately 15% below that of the control. These results correlated with extracellular ROS levels ( Figures 4A, B ). The reduction in phospho-p47^phox levels in the presence of LPS + apocynin suggests that apocynin might regulate ROS production upstream of p47 ^phox phosphorylation (69–71). Interestingly, when we repeated the experiment with LPS + GSK, we found that GSK did not affect phospho-p47 ^phox levels ( Figure 4C ), which was not unexpected as GSK inhibits the NOX2 cytochrome downstream of the phospho-p47 ^phox membrane translocation. Overall, these results confirm effective reduction in ROS levels due to microglial pro-inflammatory activation through the inhibition of NOX2 in the presence of apocynin or GSK, suggesting that NOX2 is the primary source of ROS in BV2 microglia.

Having established that BV2 cells express an active NOX2 that is inhibited by apocynin and GSK, we next studied the effect of NOX2 inhibition on clock gene expression to validate a role for NOX2 in clock-driven inflammation. Serum-shock synchronized BV2 cells were exposed to LPS (1 μg/mL) in the presence of apocynin (100 μM) or GSK (25 μM). Time course samples were collected every 2 h for 24 h starting at HPS16. Addition of apocynin under LPS activation rescued the oscillations in Per2 (ECHO period = 16 h, ECHO p-value = 1.17 x 10^-6) and gp91^phox (ECHO period = 16 h, ECHO p-value = 1.03 x 10^-12), which showed oscillations consistent with naïve BV2 microglia ( Supplementary Figures 1A, D ). Conversely, Bmal1 and p47^phox expression levels no longer showed significant oscillations ( Supplementary Figures 1B, C ). However, the addition of LPS + GSK rescued oscillations in all four genes: Per2 (ECHO period = 16 h, ECHO p-value = 2.06 x 10^-12), Bmal1(ECHO period = 16 h, ECHO p-value = 1.02 x 10^-17), p47^phox (ECHO period = 16 h, ECHO p-value = 5.46 x 10^-5) and gp91^phox (ECHO period = 16 h, ECHO p-value = 1.48 x 10^-10), displaying oscillations consistent with naïve BV2 macrophages at the transcriptional level ( Figure 5 ). This retention of oscillations under NOX2 inhibition, particularly with GSK, suggests that NOX2 might play a circadian regulatory role in BV2 microglia by facilitating the transition into a pro-inflammatory phenotype.

To validate that the results obtained in BV2 microglia were consistent in primary cells, we next studied the effect of NOX2 inhibition on the circadian clock in bone marrow-derived macrophages (BMDMs). We searched our previous quantitative RNA-sequencing and proteomic analysis datasets (22) to assess the oscillation of p47^phox and gp91^phox in BMDMs. At the transcriptional level, Per2 (ECHO period = 23.1 h, ECHO p-value = 3.89 x 10^-20) and Bmal1 (ECHO period = 29.3 h, ECHO p-value = 3.96 x 10^-20) showed significant oscillation along with p47^phox (ECHO period = 19.5 h, ECHO p-value = 1.37 x 10^-4), though gp91^phox had an oscillation beyond the circadian range (ECHO period = 36 h, ECHO p-value = 6.82 x 10^-8) ( Supplementary Figure 2A ). However, analysis of the proteomic dataset revealed that along with PER2 protein levels ( Figure 6A ), BMAL1 (ECHO period = 28.9 h, ECHO p-value = 1.32 x 10^-5), and gp91^phox (ECHO period = 19.7 h, ECHO p-value = 6.76 x 10^-5) oscillated with a circadian period, suggesting the post-transcriptional circadian regulation of gp91^phox, which is common in murine macrophages ( Supplementary Figure 2B ) (22).

To assess the role of NOX2 inhibition on the circadian clock in BMDMs, we used luciferase luminescence measurements to track the levels of PER2 in real time. For this, Per2::Luc mice, in which a Luc gene is fused in-frame to the 3’ end of the endogenous mPer2 gene, was used (72). This targeted reporter results in the expression of PER2::LUC bioluminescence similar to endogenous Per2 expression, rendering it as a real-time reporter of circadian dynamics (72). Bone marrow progenitor cells extracted from Per2:Luc C57BL/6J mice were differentiated with recombinant M-CSF into BMDMs and serum-shock synchronized, as per a previously established protocol (22, 39). PER2::LUC bioluminescence in BMDMs was assayed over several days using a LumiCycle32 with Leibovitz media containing Luciferin (22, 39). Samples treated with neither LPS nor NOX2 inhibitors added were used as controls to confirm that PER2 protein levels oscillated with a circadian period in Per2:Luc BMDMs ( Figures 6A, B ). We next treated these BMDMs with either LPS (1 μg/mL) or IL-4 (20 ng/mL) post-serum shock, as we did with the BV2 microglia, to elicit pro- and anti-inflammatory activation in the BMDMs respectively. We also analyzed the effect of LPS with NOX2 inhibitors (apocynin (100 μM) and GSK (25 μM)) on PER2 levels in BMDMs. When LPS and/or NOX2 inhibitors were added to the BMDMs post serum-shock synchronization, LPS resulted in a reduction of PER2 amplitude compared to the control sample ( Figure 6A ). When apocynin was added in parallel with LPS, there was a similar decrease in the amplitude of the PER2 oscillation. However, samples in which GSK was added in parallel with LPS maintained the amplitude in PER2 oscillations as compared to untreated BMDMs ( Figure 6A ). Conversely, IL-4 addition resulted in an increase in the amplitude of the PER2 oscillation as compared to the control sample.

Unlike what we had carried out with BV2 microglia, previous work on the response of PER2 oscillations in BMDMs to immunological challenge was performed by adding the treatment prior to serum shock (31). Therefore, to mimic this approach, we repeated our assay of PER2 oscillations in BMDMs by treating the cells with immune insults prior to serum shock synchronization. Like what was seen in BV2 microglia and in post serum shock exposure, exposure to IL-4 did not affect PER2 oscillations as compared to the control sample ( Figure 6B ). When added prior to serum synchronization, LPS addition resulted in the reduction of PER2 levels overall, as well as a reduction in the amplitude of the PER2 oscillation, as has been previously reported ( Figure 6B ) (31). When apocynin was added in combination with LPS prior to serum synchronization, the reduction in PER2 levels and amplitude of oscillation was conserved ( Figure 6B ). However, when GSK was added in combination with LPS prior to serum synchronization, PER2 levels and oscillations were similar to that of the control ( Figure 6B ). The ability of GSK to rescue the levels and amplitude of oscillation of PER2 in BMDMs during LPS activation is consistent with our results in BV2 microglia, and further suggests that NOX2 might play a regulatory role in the circadianly-regulated transition of cells from the monocyte lineage into a pro-inflammatory state.

To further profile the role of NOX2 in the pro-inflammatory activation of BV2 microglia and BMDMs, we quantified the expression of two common inflammatory cytokines, TNF-α and IL-6, in response to LPS treatment with and without NOX2 inhibitors. Serum-synchronized BV2 cells were stimulated using LPS (1 μg/mL) with and without apocynin (100 μM) or GSK (25 μM) at HPS20. We chose HPS20 as a time point both to allow sufficient time for the cells to recover from the serum-shock, to avoid artifactual gene expression that occurs immediately following serum-shock (37), and to overlap with the peak time in Per2 oscillation as observed in inactivated BV2 microglia ( Figure 1B ). After 24 h incubation starting at HPS20, the supernatant was collected from BV2 microglia and TNF-α and IL-6 levels were measured using ELISA ( Figures 7A, B ). BV2 microglia treated with LPS demonstrated a significant increase of both TNF-α and IL-6 compared to untreated cells. BV2 microglia treated with only apocynin, or GSK showed no significant change in TNF-α and IL-6 levels compared to untreated cells. However, when apocynin or GSK was added individually in parallel with LPS treatment, there was a significant reduction in TNF-α and IL-6 levels relative to samples with LPS treatment alone, with the greatest reduction in cytokine levels observed in the GSK-treated cells.

We next compared NOX2 inhibition to the anti-inflammatory effect on cytokine production in BV2 microglia. To this end, we assessed the effect of IL-4 activation on TNF-α and IL-6 production in BV2 microglia ( Figures 7C, D ). Serum synchronized BV2 cells were incubated with IL-4 (20 ng/mL) for 24 h starting at HPS20. After a 24 hr incubation with IL-4, the media was replaced with fresh media containing LPS (1 μg/mL). In addition, we treated serum synchronized BV2 cells with LPS and IL-4 simultaneously or individually and incubated the cells for 24 h starting at HPS20. While TNF-α levels showed a slight, albeit significant, drop in levels in the presence of IL-4 ( Figure 7C ), no significant change in IL-6 levels was observed ( Figure 7D ). Co-treatment of LPS and IL-4 resulted in significant reduction of TNF-α levels compared to samples that had only LPS, IL-6 levels remained unchanged ( Figure 7D ). Interestingly, LPS activation of BV2 cells post-24 h incubation with IL-4 resulted in a significant drop of both TNF-α and IL-6 ( Figures 7C, D ).

We repeated these experiments in BMDMs to study the effect of NOX2 inhibition on pro-inflammatory activation in primary cells. Serum shock synchronized BMDMs were stimulated using LPS (1 μg/mL) with and without apocynin (100 μM) or GSK (25 μM) at HPS20. After a 24 h incubation, the supernatant was collected and TNF-α and IL-6 levels were measured using ELISA ( Figures 8A, B ). BMDMs treated with LPS demonstrated a significant increase in TNF-α and IL-6 levels compared to the untreated control ( Figures 8A, B ). Concordant with what was seen in BV2 cells, NOX2 inhibition by either apocynin or GSK during LPS treatment resulted in a significant reduction in both TNF-α and IL-6 levels as compared to BMDMs exposed only to LPS. While LPS + GSK showed a more significant drop in TNF-α and IL-6 levels than LPS + apocynin in BV2 microglia, such a difference was not observed in the BMDMs. Compared to BMDMs treated with LPS, addition of IL-4 along with LPS resulted in lower levels of IL-6 ( Figure 8B ). Interestingly, addition of apocynin or GSK along with LPS resulted in the secretion of significant amounts of IL-4 in BMDMs compared to naïve BMDMs or BMDMs exposed only to LPS ( Figure 8C ). Comparison of ELISA results from IL-4 activation ( Figures 7C, D ) and NOX2 inhibition ( Figures 7A, B ) in BV2 microglia and that of mouse BMDMs ( Figure 8 ) suggests that inhibiting NOX2 leads to a microglial response that is similar to that of an anti-inflammatory activation. This is consistent with the inhibition of NOX2 using GSK, wherein the circadian clock retained its oscillation for all genes of interest. These results taken together might indicate that NOX2 inhibition leads to an anti-inflammatory response from both microglia and macrophages, which results in the conservation of circadian oscillations even upon pro-inflammatory activation.