Circadian Control of the Response of Macrophages to Plasmodium Spp.–Infected Red Blood Cells Open Access

Malaria is a life-threatening infectious disease responsible for >200 million cases and nearly half a million deaths in the world each year (1). It is caused by the protozoan Plasmodium spp. and transmitted by the bite of Anopheles spp. mosquitos. Within the mammalian host, the parasite differentiates into mature forms able to infect RBCs during the later stages of the disease. This process culminates with malaria’s paroxysms, characterized by fever, shaking chills, muscle aches, and other features (2), which are the result of the activation of innate immune cells of the monocyte lineage, mainly macrophages. Strong activation of macrophages leads to secretion of proinflammatory cytokines that are key factors for the disease severity outcomes, including enhancement of adhesion molecules expression, sequestration of infected RBCs in lungs and brain, renal impairment, and disruption of the blood–brain barrier. Notably, 24-h rhythms (or multiples of 24 h) have been observed in symptoms (malaria paroxysms). The peak in paroxysms also coincides with the period of rupture (or lysis) of infected RBCs and high levels of proinflammatory cytokines in the bloodstream (3, 4).

Circadian (∼24-h) rhythms have been reported in all aspects of physiology and are controlled by circadian clocks found in all tissues (5–7). These clocks rely on clock genes, acting via transcription/translation feedback loops. This mechanism leads to the daily rhythm in the levels of transcripts and proteins encoded by the clock genes (8). Moreover, the clock mechanism regulates in a rhythmic fashion a large proportion of the cell’s transcriptome and proteome, which underlies the circadian regulation of cell functions (9–11)

All immune cell types harbor an intrinsic clock controlling various immune functions (12–16). In macrophages, >15% of the macrophage transcriptome is under circadian regulation (9, 17). Consequently, a number of functions in macrophages/monocytes present circadian rhythms such as phagocytosis (9, 18, 19), metabolism (9), reactive oxygen species (ROS) generation (20), cell trafficking (21–23), cytokine response to LPS stimulation (17, 24), and more.

The extensive circadian regulation of immune functions, and in particular of macrophage/monocyte functions, indicates that the response of the immune system to parasitic infections and the course of these diseases could also be under circadian control (25). We previously showed that Leishmania major infection is regulated by the host circadian clock, such that parasite development, inflammation, and recruitment of innate immune cells at the site of infection, as well as the production of cytokines and chemokines, were all varying according to the time of infection (26). Interestingly, circadian rhythms were also found at the cellular level, upon infection of bone marrow–derived macrophages (BMDMs) by Leishmania parasites, and these rhythms were lost in cells lacking a functional circadian clock (lacking the essential clock gene Bmal1) (26, 27).

It is unknown whether such a circadian regulation exists for macrophages internalizing Plasmodium spp.–infected RBCs. Given the strong contribution of the innate immune system to malaria severity (28), it is important to investigate whether a host’s circadian clocks, in particular those of macrophages, influence the magnitude of the response to the parasite. To address this, we designed a new ex vivo procedure consisting in the presentation of lysed Plasmodium-infected RBCs to macrophages. This allowed us to mimic the induction of inflammation that occurs upon the daily rupture of the RBCs in infected individuals. We detected circadian rhythms in levels of cytokines and chemokines, and in the generation of ROS following treatment with infected RBCs. This was paralleled by major time-dependent changes in the response of the cellular proteome and phosphoproteome to stimulation.

Six-week-old male C57BL/6NCrl mice (strain code 027) were purchased from Charles River Laboratories (Saint-Constant, QC, Canada). PER2::LUCIFERASE (PER2::LUC) knock-in mice (29) were purchased from The Jackson Laboratory (strain no. 006852). Mice were housed at the animal facility of the Douglas Mental Health University Institute, in a pathogen-free environment. Animal use was in accordance with the guidelines of the Canadian Council of Animal Care and was approved by the Douglas Institute Facility Animal Care Committee.

BMDMs were generated by culturing bone marrow from C57BL/6NCrl mice or PER2::LUC mice for 7 d in macrophage medium, composed of RPMI 1640 1× medium (Multicell, catalog no. 350-000-CL) with 10% Corning Nu-Serum IV growth medium supplement (catalog no. 355104), 1% Life Technologies penicillin-streptomycin (catalog no. 15070063), and 30% L cell conditioned medium, following a published protocol (30). Briefly, following euthanasia, legs were placed in 20 ml of macrophage medium. Bone marrow was aspirated from bone, centrifuged, resuspended with 1 ml of erythrocyte lysis buffer (BioLegend, catalog no. 420301), and incubated for 3 min. Cells were centrifuged and 0.7 × 10⁶ cells/well were plated in a six-well plate containing 2 ml of macrophage medium (day 0). Cells were incubated at 37°C, and on day 4, medium was replaced with fresh macrophage medium. On day 6, the cell circadian rhythms were synchronized with a serum shock (31): 2 ml/well 50% horse serum (Life Technologies, catalog no. 16050122)/50% macrophage medium was added, cells were incubated at 37°C for 2 h, washed with Dulbecco’s PBS (DPBS) (MilliporeSigma, catalog no. D8537), and placed in macrophage medium.

For bioluminescence rhythms recordings, BMDMs were prepared from PER2::LUC mice. Cell rhythms were synchronized, after which medium was replaced with 1 ml/well recording medium containing RPMI 1640 without phenol red (Life Technologies, catalog no. 11835) with 10% Nu-Serum IV growth medium supplement (Life Technologies, catalog no. 355104), 1% penicillin-streptomycin (Life Technologies, catalog no. 15140122) and 0.1 mM d-luciferin (Sigma-Aldrich, catalog no. 50227). Plates were sealed with a sterile glass coverslip and bioluminescence was recorded using LumiCycle equipment (Actimetrics). Data were exported as counts per second from the LumiCycle software and are represented in a graphical format using MS Excel.

C57BL/6NCrl mice were injected i.p. at Zeitgeber time (ZT)6 with 10⁶Plasmodium berghei, strain ANKA 676m1cl1 (MRA-868, BEI Resources) (PbA)–infected RBCs (iRBCs) 5 d after receiving daily i.p. injections of 4 mg of iron dextran (MilliporeSigma, catalog no. D8517). After infection, mice received additional daily injections of 4 mg of iron dextran up to day 5 postinfection. Infected blood was collected at ZT6 (i.e., 6 h after lights on) by cardiac puncture (terminal) when parasitemia reached 37% (day 18 postinfection). Blood was also collected from a noninfected mouse (control) at ZT6. Cells were separated by centrifugation. The plasma (supernatant) was discarded and the pellet containing RBCs was resuspended in sterile DPBS. Cells were counted and lysed with an ultrasonic sonicator for 15 s. Lysates were kept at −80°C for future stimulation assays. All experiments described in this study were performed using the same iRBC preparation.

Sixteen, 22, 28, or 34 h after BMDM synchronization, recombinant murine IFN-γ (PeproTech, catalog no. 315-05) was added to the wells (final concentration of 10 U/ml) to induce cell maturation and M1 polarization. Thirty minutes later, cells received either macrophage medium (negative control), from Escherichia coli O111:B4 (MilliporeSigma, catalog no. L2630) at a final concentration of 100 ng/ml (positive control), or 10⁶ intact or lysed iRBCs. For RNA assays, cells were washed after 3 h, collected, lysed in TRIzol (Invitrogen, catalog no. 15596026), and kept at −80°C. For cytokine/chemokine measurements, supernatants were collected 24 h after treatment and shipped for protein quantification (Mouse Cytokine Proinflammatory Focused 10-Plex Discovery Assay array) by Eve Technologies (Calgary, AB, Canada).

RNA was extracted using TRIzol according to the manufacturer’s instructions. Reverse transcription was performed using a high-capacity cDNA reverse transcription kit (Applied Biosystems, catalog no. 4368814) according to the manufacturer’s instructions. Real-time quantitative PCR was done using iTaq Universal SYBR Green supermix (Bio-Rad, catalog no. 1725120). The reaction (final volume 10 μl) was run using 0.4 ng of cDNA (6 μl), 10 μM forward primer (0.5 μl), 10 μM reverse primer (0.5 μl), and 5 μl SYBR Green. Samples were run at the QuantiStudio 6 flex system (Applied Biosystems). Gapdh was used as a housekeeping gene. Gene expression was determined using the comparative quantification method (2^−ΔΔCt method). The following primers were used: Il6 (forward, 5′-ACGGCCTTCCCTACTTCACA-3′, reverse, 5′-TGCCATTGCACAACTCTTTTCTC-3′) (32), Il1β (forward, 5′-TGCCACCTTTTGACAGTGATG-3′, reverse, 5′-GTGCTGCTGCGAGATTTGAA-3′) (33), Tnfα (forward, 5′-CGTCAGCCGATTTGCTATCT-3′, reverse, 5′-CGGACTCCGCAAAGTCTAAG-3′) (34), and Gapdh (forward, 5′-GGGTCGGTGTGGAACGATTTG-3′, reverse, 5′-TGCCGTGAGTGGAGTCATACTG-3′) (35).

BMDMs were plated (0.2 × 10⁶/well, because wells are smaller than in the other experiments) in a MilliCell EZ Slide four-well glass chamber (MilliporeSigma, catalog no. PEZGS0416) on day 5 postextraction and had their rhythms synchronized the following day. After IFN-γ preboost and stimulation with medium, LPS (100 ng/ml), 10⁶ lysed iRBCs, or RBCs for 30 min, CellROX Deep Red (Thermo Fisher Scientific, catalog no. C10422) was added to a final concentration of 5 μM along with MitoSpy Orange CMTMRos (chloromethyl-tetramethylrosamine; BioLegend, catalog no. 424803) at the final concentration of 500 nM. After 30 min, cells were washed with DPBS and fixed in 4% paraformaldehyde for 10 min, washed, and incubated with DAPI solution (1:10,000 in DPBS, MilliporeSigma, catalog no. D9542) for 2 min, and washed. A coverslip was placed on the slides and cells were imaged at ×400 using an Olympus FV1200 confocal microscope. Analysis was performed using ImageJ version 2.1.0/1.53c.

BMDMs were generated as previously described and plated as 0.7 × 10⁶ cells/well in a six-well plate. Cells were synchronized and preboosted with IFN-γ (30 min) followed by the addition of 10⁶ lysed iRBCs or RBCs for 2 h, at 22 or 34 h postsynchronization. Cells were washed, detached, pooled (20 × 10⁶ cells/sample), and centrifuged. Pellets were flash-frozen in liquid nitrogen and stored at −80°C prior to lysis. Cells were lysed in protein extraction buffer containing 5% SDS/100 mM Tris-HCl (pH 8.5) supplemented with PhosSTOP phosphatase and cOmplete protease inhibitors (both from Roche). Samples were heated at 95°C for 10 min and subjected to probe-based ultrasonication (Thermo Fisher Scientific sonic dismembrator). Samples were then clarified by centrifugation at 21,000 × g for 5 min. Protein disulfide bonds were reduced in 20 mM Tris(2-carboxyethyl)phosphine for 30 min at 60°C, and alkylated in 25 mM iodoacetamide for 30 min at room temperature. Proteins (200 µg) were proteolytically digested with trypsin (Sigma-Aldrich, T1426) at a 1:20 enzyme-to-substrate ratio overnight at 37°C using S-Trap mini cartridges (ProtiFi). Peptides were eluted from S-Trap micro cartridges sequentially with 50 mM ammonium bicarbonate, 0.1% formic acid, and 50% acetonitrile, respectively. Approximately 5% of the sample was reserved for total proteome profiling, whereas the remainder was vacuum concentrated to dryness, and reconstituted in 80% acetonitrile with 0.1% trifluoroacetic acid prior to automated phosphopeptide enrichment using an Agilent Bravo liquid handling system equipped with an AssayMAP head, and AssayMAP Fe(III)-NTA immobilized metal affinity chromatography enrichment cartridges. Peptide-containing eluates lyophilized to dryness and an equivalent of 750 ng of peptide or the complete phosphopeptide-enriched sample were loaded onto Evotips (Evosep) according to the manufacturer’s instructions prior to liquid chromatography–tandem mass spectrometry (LC-MS/MS).

Samples were analyzed by data-dependent acquisition using an Evosep One LC system coupled to a Thermo Fisher Scientific Q Exactive Plus mass spectrometer. Peptides were separated using the 15 samples per day extended method, with an EV1137 analytical column (Evosep). MS acquisition was conducted in the data-dependent acquisition mode, based on the top 15 most intense precursor ions (+2 to +4 charge state). Full MS scans were acquired at 70K resolution from 350 to 1500 m/z (AGC 1E6, 50 ms maximum injection time), and MS2 spectra were collected at 17.5K resolution (AGC 2E4, 64 ms maximum injection time) using a normalized collision energy of 28. Dynamic exclusion was set to 40. MS/MS data were analyzed using Proteome Discoverer version 2.5 (Thermo Fisher Scientific), and database searching was performed with the Sequest HT node, against a mouse reference proteome FASTA file containing only reviewed canonical sequences downloaded from UniProt (v2022-06-14). Proteins were quantified by precursor-based label-free quantitation using unique peptides, scaled according to total peptide abundance, and missing values were imputed using the low abundance resampling method within Proteome Discoverer. Phosphosite localization confidence was scored using the ptmRS module (36) within PD 2.5. Proteomic and phosphoproteomic datasets have been deposited to the ProteomeXchange Consortium via the PRIDE (37) partner repository with the dataset identifier PXD049209. Ingenuity Pathway Analysis (IPA) (Qiagen) was performed on the differentially expressed proteins and phosphoproteins using the default application settings.

Statistical analyses were performed using GraphPad Prism version 9.3.1. A one-way ANOVA was used to assess one factor, followed by a Tukey multiple comparison post hoc test. Rhythm analysis was performed using a nonlinear regression (curve-fit) cosinor analysis where a cosine wave equation, that is, y = B + {A × cos[2 × π(x − Ps)/24] }, where A indicates amplitude, B indicates baseline, and Ps indicates phase shift, with a fixed period of 24 h, was fit. Significance was calculated based on F value (observed R², sample size, and number of predictors). For proteomics/phosphoproteomics, protein expression ratios were calculated based on protein abundance, and p values were calculated using ANOVA (individual proteins) within Proteome Discoverer 2.5. Statistical difference was assumed when p < 0.05.

To investigate the response of macrophages upon stimulation with iRBCs, BMDMs, preboosted with IFN-γ, were presented with intact or lysed iRBCs. The iRBCs were from an iron dextran–treated mouse infected with P. berghei ANKA. Iron dextran treatment prevents lethality from cerebral malaria, such that a high parasitemia can be obtained (Fig. 1A, 1B) (38). Il6 gene induction was not detected upon stimulation of cells with noninfected RBCs or with intact iRBCs (Fig. 1C). Il6 expression was induced upon stimulation with lysed iRBCs (Fig. 1C), to a similar extent as stimulation with LPS.

We then used this cell culture assay to investigate whether the response of BMDMs to lysed iRBCs varies in a time-dependent manner. The circadian clocks of macrophages were synchronized using a serum shock (31), and the cells were then stimulated with IFN-γ followed by addition of lysed iRBCs (or RBCs) 16, 22, 28, or 34 h after clock synchronization (Fig. 2A). Clock synchronization was done in different series of wells 6 h apart from each other over 24 h, which allowed us to stimulate (with IFN-γ and then lysed iRBCs) cells at the same “clock time” and using the same batch of parasites, while they were at different phases of their circadian clock. Thus, the magnitude of the immune response generated afterward indicates dependency on the macrophage circadian clock. Clock synchronization was verified by bioluminescence recordings of BMDMs from PER2::LUC mice (Fig. 2B). Projected circadian times were estimated based on the known approximate phase of PER2 expression rhythm in rodent peripheral clocks (39).

In cells collected 3 h after iRBC stimulation, the induction of genes encoding IL-6, IL-1β, and TNF-α (Fig. 2C) showed time dependency. Protein analysis of cell supernatants collected 24 h after stimulation with iRBCs/noninfected RBCs, resulted in rhythmic production of IL-1β (cosinor p = 0.015, ANOVA p = 0.0014) and IL-12p70 (cosinor p = 0.011, ANOVA p = 0.0003) (Fig. 3). For MCP-1 (cosinor p = 0.088, ANOVA p = 0.0008) and IL-10 (cosinor p = 0.1810, ANOVA p < 0.0001), there was an effect of time (ANOVA), but the cosine fit was not significant. A trend for an effect of time and for cosine fit was found for IFN-γ (cosinor p = 0.077, ANOVA p = 0.0850). No rhythmicity or effect of time was found for TNF-α (cosinor p = 0.3602, ANOVA p = 0.2618), IL-6 (cosinor p = 0.1712, ANOVA p = 0.1679), and GM-CSF (cosinor p = 0.6574, ANOVA p = 0.7762) (Fig. 3). As expected, in both RNA and protein measurements, there was no detectable immune activation when cells were stimulated with noninfected RBCs (Figs. 2C, 3).

Next, as a first step to investigate intracellular signaling events involved in the response of macrophages to Plasmodium spp., we assessed the generation of ROS by using a fluorogenic dye. A 30-min incubation of BMDMs with iRBCs induced ROS production, an event not triggered when cells were treated with noninfected RBCs (Fig. 4A, 4B). ROS generation was then assessed in clock-synchronized BMDMs and stimulated at different time points during 24 h (Fig. 4C). Time-dependent differences in ROS generation were found: stimulation of cells 22 h after synchronization led to significantly higher levels of ROS when compared with the 16 and 28 h stimulation time points (Fig. 4C).

For an in-depth investigation of the intracellular networks, in particular signaling pathways, affected by iRBCs in macrophages in a circadian time-dependent manner, we used LC-MS/MS to analyze the whole proteome of BMDMs stimulated at 22 or 34 h postsynchronization (Fig. 5). Approximately 3500 proteins were quantified, among which 425 were found to be differentially expressed in at least one of the four comparisons (iRBCs versus RBCs at 22 h, iRBCs versus RBCs at 34 h, iRBCs at 22 h versus 34 h, RBCs at 22 h versus 34 h) (Fig. 6A). The complete list of detected proteins and the results of the comparisons between groups can be found in dataset 1 of Supplemental Table 1. Circadian clock proteins were not detected, but this is not unexpected, as the low abundance of these proteins often makes them undetectable in proteomic screens (40). Notably, in the pool of proteins differentially expressed in iRBC-treated samples at 22 versus 34 h, we detected 100 proteins with higher levels at 34 h and 57 proteins with higher levels at 22 h (Fig. 6B, third panel).

Interestingly, when we analyzed this list of 157 proteins affected by the time of treatment (in the cells treated with iRBCs) (Fig. 6B, third panel, dataset 1 of Supplemental Table 1), we found that most proteins were involved in stress response, metabolism, cell adhesion, and cell cycle. In this list, we identified three main signaling pathways: PI3K-Akt-mTOR, type II IFN, and MAPK signaling pathways. Examples of proteins belonging to these pathways and that are differentially regulated are Itgb5, Col1a1, Rap1a, KRAS (PI3K-Akt-mTOR pathway), IL-1β, CXCL9 (type II IFN pathway), and Mapk3k20 (MAPK pathway).

The same samples were also used for quantitative phosphoproteomics. We detected 4600 phosphopeptides, among which 827 were differentially expressed in at least one of the four comparisons (Fig. 7A). The complete list of detected phosphopeptides and the results of the comparisons between groups can be found in dataset 2 of Supplemental Table 1. Among the 741 differentially regulated phosphosites (out of the 827, 86 phosphoproteins were excluded from the analysis as they lacked information regarding the posttranslation modifications in master proteins), more than half were on serine residues, and the remainder were on threonines and tyrosines (Fig. 7B), which represents a normal distribution. Interestingly, there was a big shift in the impact of iRBC stimulation on phosphorylation events when comparing the 22- and 34-h samples (Fig. 7B): at the first time point, many more phosphopeptides were induced (n = 106) than reduced (n = 58) by the treatment, whereas at 34 h, only 52 phosphopeptides were induced, while 210 were downregulated by the treatment. On cells stimulated with iRBCs at either of the two time points, 396 phosphopeptides were differentially regulated according to the time of stimulation (234 sites higher at 22 h, 162 sites higher at 34 h).

Additionally, analysis of the protein and phosphopeptide changes upon iRBC treatment revealed interesting time-dependent differences (Fig. 8, Tables I and II). In particular, the MAPK and inflammation-related signaling pathways were differentially regulated (Fig. 8). Moreover, we have identified a list of 16 phosphopeptides upregulated by treatment with iRBCs only at the 22-h time point; that is, these showed increased expression in iRBC-treated cells at 22 h in comparison with RBC-treated cells at 22 h, but were not differentially expressed between iRBC-treated cells and RBC-treated cells at 34 h (Table I). Conversely, 24 phosphopeptides were found to be upregulated by treatment with iRBCs only at the 34 h time point (Table II). To complement this, IPA revealed >30 signaling pathways predicted to be positively or negatively regulated in cells treated with iRBCs between the 22 and 34 h time points in both proteome and phosphoproteome datasets (Supplemental Tablel 2).

We developed a new ex vivo culture model to study the response of macrophages to iRBC-derived factors. This model allows one to mimic the inflammatory stage of malaria that occurs upon rupture of the iRBCs, which typically takes place in a synchronized fashion. Because of this, we reasoned that time of day matters in terms of the magnitude of the macrophage response, and the ex vivo model allows to focus the investigation on this stage of the disease, without other factors that exist in vivo. We observed that the responses to Plasmodium spp.–infected RBCs is dependent on the nature of the material presented to cells (intact versus lysed RBCs), with an absence of proinflammatory responses upon presentation of intact cells. Using this approach, we reported that the response of macrophages to Plasmodium spp.–derived molecules, more specifically lysed-infected RBCs, is affected by the circadian clock in these cells: clear time-dependent variations were detected in cytokine/chemokine production, in ROS generation, and in the cell proteome and phosphoproteome.

During malaria pathogenesis, innate immunity, in particular involving macrophages/monocytes, plays a crucial role at recognizing Plasmodium spp.–derived particles, resulting from the burst of iRBCs, and at initiating a cascade of proinflammatory responses. Several studies have demonstrated strong immunomodulatory effects of such molecules, in particular hemozoin, through the activation of a variety of signaling pathways, such as the NF-κB (41, 42), MAPK (42, 43), and NLRP3 inflammasome (44) pathways.

Given that not only one, but several molecules are known to be released from infected RBCs during lysis, we chose to stimulate macrophages with lysates of P. berghei ANKA–infected RBCs, as a more realistic proxy to study circadian rhythms in immunity during malaria pathogenesis, compared with studies performed with isolated compounds, such as hemozoin (45). Our stimulation model consisted of a mix of infected and noninfected RBCs, collected from a mouse with a parasitemia of 37% (iRBCs). Interestingly, a macrophage immune response was detected upon presentation of lysed infected material, but not its nonlysed form; this is reminiscent of previous studies, where the effect of lysed iRBCs on dendritic cell maturation or the effect of iRBC microparticles on macrophage activation was studied (46, 47).

After confirming that our lysed iRBC model led to a proinflammatory response, we asked whether the magnitude of this response was influenced by the phase of the macrophage clock at the time of iRBC stimulation. Indeed, the circadian clock in immune cells was reported to regulate activation and production of proinflammatory cytokines by macrophages (17, 24). Not all cytokines/chemokines showed rhythmic expression. For example, TNF-α and IL-6 showed no changes in protein expression across time points. Aligned with this, in the proteomics data, TNF-α was induced by iRBC treatment, but with no significant effect of time of treatment. In contrast, production of IL-1β was highly rhythmic in the cytokine array data, which was further supported by the proteomic screen, where IL-1β was detected as one of the proteins affected by iRBC treatment in a time-dependent manner, with a significantly higher upregulation after treatment at 22 h postsynchronization when compared with 34 h. IFN-γ, another cytokine involved in malaria rhythmic fevers (48), was upregulated specifically at 34 h in the pool of iRBC-stimulated samples in the proteomic data. Similarly, in the cytokine panel data, a trend for a 24-h rhythm in IFN-γ was detected, with highest levels upon stimulation at 34 h.

These rhythms in proinflammatory cytokines corroborate with the paroxysms in malaria, as a result of rhythms in the burst of infected RBCs (49). Moreover, in a murine air pouch model to study the host response to malaria, increased levels of proinflammatory cytokines (IL-6, IL-1α, IL-1β) and chemokines (MCP-1, MIP-1α, MIP-1β) were detected following hemozoin challenge (50). To compare our results with the available in vivo data on circadian rhythms in macrophage functions, we have aimed to estimate a projected circadian time of our cells. To do this, we used the phase of PER2::LUC bioluminescence rhythms, which tracks PER2 protein expression: a peak at 22 h postsynchronization would therefore be equivalent to early night in vivo, a time of peak PER2 expression in many peripheral tissues in mice (Fig. 2B). Our BMDMs secreted more IL-1β protein following a stimulation done at 22 h postsynchronization. This is consistent with a peak of IL-1β rhythmicity at ZT12 (beginning of the night) in the mouse peritoneal cavity (51). As for Il1β mRNA, we observed a peak at 34 h postsynchronization (equivalent to beginning of the subjective day in the projected circadian times), which is consistent with the reported peak of Il1β mRNA (ZT6) in mouse peritoneal macrophages (18). This approach can be used to correlate our findings with malaria paroxysms. For instance, it is known that the onset of fevers in P. vivax–infected patients occurs around 4:00 pm (middle/end of the active phase) (52). The dynamics of these fevers follow the synchronized schizont rupture and TNF-α release (49), starting ∼1–2 h after this event (3). Interestingly, this time-dependent set of events is conserved in Plasmodium chabaudi–infected mice (52) where schizont rupture and the peak of TNF-α occurs at the end of the active phase. In our data, all three major proinflammatory cytokines involved in malaria response (TNFα, IL-1β, and IL-6) were induced in response to iRBC stimulation, and all showed a peak of mRNA expression at 28–34 h postsynchronization (equivalent to the end of the active phase/beginning of the rest phase). Moreover, the secretion of IL-1β, MCP-1, and IL-12p70 also peaked during an animal’s active phase. This suggests that proinflammatory responses are increasingly triggered by the encounter of macrophages with iRBCs when this is performed at the end of the active phase, matching the fever dynamics in malaria patients.

Activation of macrophages by iRBCs during malaria pathogenesis requires TLR9 expression, as this receptor recognizes malarial DNA (46, 54). Interestingly, and supporting the evidence of higher immune responses against malaria during the host’s active phase, a study identified that TLR9 mRNA is rhythmically expressed in many cells (including macrophages) with a peak during the animal’s active phase. Moreover, by using a TLR9-dependent model, higher levels of IL-6, IL-12p40, and MCP-1 were induced at ZT19 when compared with ZT7 (55). Importantly, although many cytokines/chemokines detected in our experiments showed a rhythm of their response to iRBC stimulation, distinct phases were observed: whereas IL-1β, IL-12p70, and MCP-1 peaked at 22 h postsynchronization, IL-10 and IFN-γ peaked at 34 h. Interestingly, the anti-phasic pattern of proinflammatory (IL-1β, MCP-1) versus anti-inflammatory (IL-10) proteins detected in our study is aligned with the findings observed in the P. chabaudi model in mice, as well as in Plasmodium falciparum–infected human subjects (53). Altogether, a balance between proinflammatory and anti-inflammatory proteins occurs during malaria pathogenesis, and not only do our data support this, but they also indicate a circadian regulation of these events in time.

After having shown a rhythmicity of cytokine induction, we turned to upstream signaling events. ROS production was highly affected by the time of stimulation. ROS induction is known to be important during hemozoin presentation, and it is directly linked to the activation of NLRP3 inflammasome pathway, as well as IL-1β production (56). In our study, the highest levels of both IL-1β protein and ROS were reached after stimulation at 22 h postsynchronization. It was shown that the macrophage circadian clock controls ROS and IL-1β production through direct interaction of the clock transcription factor BMAL1 with nuclear factor erythroid 2–related factor 2 (NRF2) (20). Also, the clock protein REV-ERBα can bind to Nlrp3 promoter and repress its transcription (57), showing how the NLRP3 inflammasome pathway is regulated by the circadian clock. Proteins such as Ras-related protein Rap-1, G protein subunit γ14 (GBγ), proto-oncogene c-Src (Src), and Rho-associated coiled-coil containing protein kinase 2 (ROCK2), which were upregulated by iRBCs at 22 h (compared with RBCs) in our proteomic and phosphoproteomic datasets (Fig. 8), are known members of the chemokine signaling pathway. One of the downstream events of chemokine signaling pathway activation in phagocytes is the induction of intracellular ROS production (58, 59). Interestingly, hemozoin treatment of monocytes was shown to induce IL-1β production via Src kinases (44).

Proteomics analysis of samples treated with iRBCs also revealed, at 22 h, an upregulation of chemokines, including CXCL2 and CXCL9, as well as transcription factors involved in immune signaling pathways, such as IRF3. These downstream events are linked to an activation of NF-κB and JNK and p38 signaling pathways at 22 h postsynchronization (Fig. 8). Interestingly, phosphorylation of p38 and JNK1/2 was reported in monocytes stimulated with hemozoin, as well as phosphorylation of MAPK-activated protein kinase 2 (MAPKAPK-2), a substrate of p38 (60). The NF-κB pathway is involved in the hemozoin response, with phosphorylation of IκBα and p50/p65 nuclear translocation (42), and this pathway can be regulated by the circadian clock (through REV-ERBα repression) (57). Furthermore, our IPA analysis additionally confirmed an enrichment of the acute phase response signaling pathway (with proteins such as IL-1β, MAP2K3, MAP2K4, and others) in cells stimulated with iRBCs at 22 versus 34 h. Similar to our findings, a study using macrophages stimulated with LPS also detected increased phosphorylation of MAPK family members, including Mapk14 and Mapk3 (61). Moreover, at 34 h postsynchronization, MKK4, a JNK activator, was upregulated in iRBC- versus RBC-treated cells. Altogether, these results showed that JNK and the p38 signaling pathway are modulated by iRBCs in a time-dependent manner.

Furthermore, in our list of phosphoproteins upregulated by treatment (iRBCs versus RBCs) only at 22 h but not at 34 h, several Rho GTPases were present. Consistent with this, we have also detected “Fcγ receptor-mediated phagocytosis” as one of the top three enriched pathways in the IPA analysis of the cellular phosphoproteome. The role of Rho GTPases in phagocytosis (endocytic pathway) is well known (62). Accordingly, phagocytosis by macrophages has been shown to be circadian clock controlled (9, 18, 19).

It is known that the macrophage phosphoproteome can be strongly shaped by intracellular immune activation. For instance, 25% of the phosphorylation sites in macrophage proteins were regulated by LPS-triggered TLR4 activation (61). In a context of time-dependent effects, in the samples treated with iRBCs, at the 22 h time point, we observed 2-fold more phosphopeptides upregulated by the iRBC treatment (n = 106) than at the 34 h time point (n = 52), whereas there were many more downregulated phosphopeptides at 34 h (n = 210) compared with 22 h (n = 58). This indicates a strong effect of the time of treatment on downstream signaling events in macrophages. A recent study reported the temporal dynamics of the cellular proteome and phosphoproteome in murine muscle tissue when samples were collected at ZT0 or ZT12. The authors observed time-dependent differences in protein levels and posttranslational modifications mainly regarding stress responses. Similar to our findings, the study identified phosphorylation enrichment of pathways involved in MAPK signaling and autophagy at ZT12 (63). Other examples of a circadian control of the phosphoproteome were published for other mammalian tissues (64–66).

In summary, we designed a new ex vivo macrophage stimulation model that could be useful to other researchers interested in the inflammatory response to infected RBC molecules. We used this protocol to show that stimulation of macrophages at different circadian time points with P. berghei ANKA–derived molecules results in different levels of their inflammatory. To our knowledge this is the first study to look at the effect of macrophage endogenous circadian rhythms on the response to Plasmodium spp.–derived molecules. The use of an experimental model mimicking the inflammatory events in the malaria RBC stage suggests that the circadian rhythms we have observed might be of relevance to understanding the factors influencing the human disease.

The authors have no financial conflicts of interest.

We thank the members of the Cermakian and Olivier laboratories for discussions, knowledge sharing, technical training, and reagents. We also thank Marie-Ève Cloutier for experimental assistance. Furthermore, we thank the staff of the Immunophenotyping Platform at the McGill University Health Centre, the Molecular and Cellular Microscopy Platform, and the Animal Facility at the Douglas Research Centre, as well as their respective staffs.