Our prior findings show that exposing microglial cells to the neurotoxicant Mn immediately following LPS priming (1 μg/mL for 3 h) activated the NLRP3 inflammasome inflammatory signaling pathway (8). To determine the optimal time of LPS exposure and to establish the neuroimmune paradigm in the current study, we first performed a time-course assessment of the inflammatory response following LPS priming using iNOS activity by Griess assay. LPS-primed MMC cells induced a maximal proinflammatory response at 12 h and gradually recovered to a near steady-state after 48 h of LPS priming ( Figure 1A ). Thus, we adopted this approach as our priming treatment paradigm, that is, LPS as a memory priming trigger and Mn as the secondary environmental insult after a 48-h interval that encompasses a triple wash and steady-state recovery ( Figure 1B ). A pilot RT-qPCR experiment with 48-h recovery time in PMG cells revealed a dramatic increase of IL-1β and iNOS following Mn exposure ( Figure S1 ), similar to the heightened secondary inflammatory response of microglial priming without recovery time. Note too that, although not statistically significant, the mean transcription levels of IL-1β and iNOS in LPS-primed only cells did not numerically return to the baseline during the recovery phase after LPS triple-washout ( Figures S1A, B ), thus providing evidence that microglial cells retain a heightened state after exposure to primary memory trigger. These results suggest that LPS-primed microglia can form microglial memory and become more sensitive to a secondary stimulus. Because the inflammatory responses can be possibly confounded by reduced cell viability due to a prolonged Mn exposure when extending the treatment paradigm, the recovery time of 24 h was used in our treatment paradigm ( Figure 1B ).

Next, we evaluated whether immune-regulated memory occurs as expected and how it affects neuroinflammation. Griess assay shows that Mn, as a secondary stimulus, induced an exaggerated inflammatory response in LPS-primed, 24-h-recovered MMC cells ( Figure 1C ). Cytokine assay confirms that Mn significantly potentiated the secreted levels of IL-6 and TNFα in primed MMCs ( Figures 1D–F ). Our RT-qPCR of LPS-primed PMG and MMC lysates shows Mn also exaggerated the inflammatory responses of Nos2, IL-1α, IL-1β, Nlrp3, IL-6, and Tnfα ( Figures 1G–L ) and Nos2, IL-1α, IL-12β, IL-10, and IL-6 ( Figures 1M–Q ), respectively. We also explored the possibility of using Mn as the primary memory trigger followed by LPS/Mn treatment. RT-qPCR ( Figure S2A ) and Griess ( Figure S2B ) assays showed that Mn priming can induce trained immunity with LPS being a more potent second insult as compared to Mn. Collectively, our LPS-trained immunity treatment paradigm shows that Mn exposure recalled the memory of the primary memory trigger, leading to an exaggerated inflammatory response.

Although the mechanisms underlying microglial priming and trained immunity have not yet been well defined, emerging evidence implies that immune regulation of trained immunity is deposited through epigenetic reprogramming and that microglial priming in the brain can be viewed as trained immunity (3, 7). It is well known that epigenetic reprogramming involving histone modification positively and negatively affects gene expression. Accumulating studies suggest that immune regulation-associated memory formed by environmental stimuli is stored by the specific deposition of H3K4me1, H3K4me3, and H3K27ac (7, 15). But whether Mn activates microglial immune regulation through deposition of epigenetic markers and the identity of which epigenetic marker(s) store(s) long-term microglial immune-regulated memory of their previous encounters are unknown. Our immunocytochemistry (ICC) analysis demonstrates increased H3K27ac accumulation per cell in Mn-exposed, LPS-primed MMCs ( Figure 2A ), and LPS- alone group induced increased H3K27ac deposition even higher in Mn exposure. Note that Mn treatment enlarged MMC cell bodies as is expected of activated microglia, thus making LPS-primed, Mn-treated MMCs appear slightly less intensely green than the less swollen LPS-trained MMCs (16, 17). Based on total H3K27ac intensity per cell, the LPS+Mn treatment induced more K27ac than LPS-alone per cell ( Figure 2B ), whose results were confirmed in PMGs ( Figure 2C ). We also investigated the effect of utilizing Mn as both the primary memory trigger and the secondary stimulus on H3K27ac deposition. Repeated Mn exposure also increased H3K27ac accumulation per cell but the effect was not significant ( Figures 2A, B ). To extend our findings, an in vivo study was performed, in which C57BL/6J mice were administered 100 mg/kg Mn by oral gavage for 30 days ( Figure 2D ). Western blot analysis shows increased deposition of H3K27ac in Mn-treated mouse striatal tissues when compared to saline-treated controls ( Figures 2E, F ). The MitoPark mouse model recapitulates the progressive nature and key pathophysiological features of PD, including most importantly microglial inflammation (18). According to our previous data (19), H3K27ac accumulation in substantia nigral tissues collected from both MitoPark mice and postmortem human PD brains was greater than that from control brains in Western blots ( Figures 2G, H ). Collectively, these findings suggest an important role and clinical relevance of H3K27ac in PD.

To study whether LPS induces priming-associated H3K27ac through p300, we used the p300/CBP pharmacological inhibitor GNE-049. Recent publications report that GNE-049, as a promising small-molecule therapy. Targets p300 and blocks H3K27ac at enhancers to efficiently suppress the growth of castration-resistant prostate cancers and hematological malignancies (20, 21). To reduce H3K27ac levels, MMC cells were co-treated with 1 µg/mL LPS and 500 nM GNE-049 for 3 h ( Figure 2I ). Compared to the LPS-priming- alone group, levels of H3K27ac and p300 decreased in the GNE-049/LPS co-treatment group as revealed by immunoblotting ( Figures 2J–M , respectively) and ICC images ( Figures 2N–Q , respectively).

Next, we investigated whether inhibiting p300/CBP by GNE-049 prevents Mn-potentiated H3K27ac deposition from LPS-primed microglial cells. MMCs were treated with 1 µg/mL LPS in the presence and absence of 500 nM GNE-049 for 3 h. The cells were then triple- washed, incubated 24 h to recover to normal state and treated with 100 µM Mn for another 24 h ( Figure 2R ). ICC analysis consistently demonstrates an exaggerated level of H3K27ac by Mn treatment in LPS-primed MMCs, whereas co-treatment with GNE-049 attenuated this potentiation effect ( Figures 2S, T ). Of note, GNE-049 treatment alone had no effect on H3K27ac levels. These results collectively indicate that H3K27ac possibly plays an important role in immune regulation.

Our data above identified that H3K27ac deposition serves as a primary trigger-induced, long-lived memory and regulates inflammatory responses. Our next step was to explore the downstream molecular mechanism underlying H3K27ac-mediated immune regulation. For this, RT-qPCR, ICC, and immunoblotting were performed to check multiple neuroinflammation-associated factors. GNE-049 co-treatment decreased iNOS transcript levels in RT-qPCR ( Figure 3A ) as well as its protein levels in ICC images ( Figures 3B, C ) and Western blot ( Figures 3D, E ).

Next, we knocked down p300 in MMC cell line to preliminarily confirm that LPS induces iNOS pathway-associated inflammatory factors through H3K27ac, whose results showed the decreased transcription expression of IL1β, Nos2, Nlrp3, and Tnfα ( Figure 3F ; Figures S3A–C ). Western blot also showed that the translation levels of p300, iNOS, and H3K27ac decreased ( Figure S3D ). Then we explored how GNE-049 protected microglia from a secondary Mn insult following a 24-h steady-state recovery interval. Thus, after the 3-h LPS primary stimulus, a triple washout, and a 24-h recovery time to return the inflammatory response to steady-state levels, we introduced a 24-h Mn (100 μM) insult. Interestingly, GNE-049 co-treatment significantly reduced iNOS expression in the Mn-exposed, LPS-primed MMCs ( Figure 3G ).

The dynamic changes of M1/M2 microglial phenotypes are critically involved in neuroinflammation, in which M2 polarization is associated with neuroprotective functions (22, 23). To check whether GNE-049 promotes M2 polarization, we tested M2 markers in GNE-049-treated microglial cells. GNE-049 co-treatment increased the expression of the M2 polarization marks IRF4 ( Figures 4A, B ) and ARG1 ( Figures 4C, D ).

For our treatment paradigm, we also briefly checked NLRP3 inflammasome activation in MMCs. Similar to microglial priming, LPS treatment increased NLRP3 levels, while GNE-049/LPS co-treatment almost abolished its expression ( Figure S4A ). GM-CSF has been reported as a trained immunity modulator involved in enhancing inflammatory responses (6, 24). GM-CSF expression was increased by LPS, but reduced by co-treating with GNE-049, suggesting the beneficial effects of GNE-049 on inflammatory responses to trained immunity ( Figures S4B, C ).

Mitochondrial stress is a key pathophysiological process of environmentally linked PD. Crosstalk has been reported between mitochondrial stress and inflammation (25, 26). In addition, we recently reported that mitochondrial dysfunction leads to epigenetic dysregulation in environmentally linked PD (19). As such, we examined the effects of GNE-049 on memory trigger-induced mitochondrial stress. MMCs co-treated with GNE-049/LPS displayed significantly decreased mitochondrial circularity via MitoTracker staining ( Figure 4E ), while MitoROS production markedly decreased as determined via MitoSox assay ( Figures 4F, G ). Moreover, the mitochondrial membrane potential increased in GNE-049 co-treated MMCs as revealed by an increased red/green ratio in the JC-1 assay ( Figures 4H, I ). These findings suggest that GNE-049 protects microglia from primary trigger-induced structural and functional derangement of mitochondria. Therefore, besides suppressing iNOS and NLRP3 factors, GNE-049 can also potentially protect microglial cells from inflammation by alleviating mitochondrial stress.

To sum up, blocking deposition of the epigenetic mark H3K27ac inhibited iNOS and its associated inflammatory factors. GNE-049 has beneficial pharmacological effects against LPS induced inflammation. In our studies, it inhibited the H3K27ac deposition that forms initial memory in microglia and consequently suppressed iNOS and NLRP3- associated inflammation.

Furthermore, we examined other trained immunity marks, that is, H3K4me3 and H3K4me1. Immunoreactivities of H3K4me1 and H3K4me3 increased in Mn-treated, LPS-primed PMGs, compared to LPS-primed alone cells, as displayed by ICC ( Figure S5 ). Then we used MitoParks to further explore the relevance of H3K4me3 and H3K4me1 with PD. We examined both the substantia nigra and olfactory bulb to study the regional distribution of H3K4me3 and H3K4me1, since olfactory dysfunction is one of the premotor symptoms reflecting early deposition of Lewy pathology in PD patients (27, 28). As confirmed by immunoblotting, H3K4me3 and H3K4me1 increased significantly in substantia nigral tissues of 16~20-week MitoParks relative to littermate controls ( Figures 5A–C ), while no significant change occurred in the olfactory bulb ( Figure S6 ). Next, regarding clinical relevance, H3K4me3 deposition increased significantly in the substantia nigra from postmortem human PD brains (n=4) when compared to age-matched healthy controls in Western blots of whole cells lysates ( Figures 5D, E ). The slightly higher mean signal density for H3K4me1 in postmortem human PD brain samples was not significant ( Figures 5F, G ). These results collectively indicate that H3K4me3 along with H3K4me1 is potentially an epigenetic mark for immune regulation.

Fetal bovine serum (FBS), RPMI, L-glutamine, penicillin, streptomycin, MitoTracker Red, MitoSOX Red stains, IRDye-tagged secondary antibodies, Hoechst nuclear stain, and other cell culture reagents were purchased from Invitrogen (Gaithersburg, MD). LPS (Escherichia coli 0111: B4, endotoxin content 6.6000000 EU/mg) and the Bradford protein assay kit were ordered from MilliporeSigma (Burlington, MA). The p300/H3K27ac inhibitor GNE-049 was purchased from MedChemExpress (Monmouth Junction, NJ). p300 siRNA was purchased from MilliporeSigma (Burlington, MA) for transfection (Cat. #EMU078861-50UG). The phalloidin antibody was purchased from Thermo Fisher Scientific, acetylation inhibitor cocktail from Santa Cruz, and protease and phosphatase inhibitor cocktail from Life Technologies.

The immortalized mouse microglial cell line (MMC) was donated by D.T. Golenbock (University of Massachusetts Medical School, Worcester, Massachusetts, USA) and characterized by our group (29). MMC cells were grown in Dulbecco’s modified eagle medium/nutrient mixture F-12 (DMEM/F-12) supplemented with 10% FBS, 1% sodium pyruvate, 1% glutamine, and 1% penicillin-streptomycin. Cell treatments were completed in 2% FBS-containing DMEM/F-12 medium. For the LPS time-response experiment ( Figure 1A ), time points include 0, 12, 24, 36, and 48 h.

For LPS- and Mn-primed treatments, the cells were treated with 1 µg/mL LPS or 100 µM Mn and incubated for 3 h. After the 3-h interval, the cells were triple-washed with full serum medium to completely remove the LPS/Mn. We then added 10% FBS-containing DMEM/F-12 medium to the cells, and they were incubated 24 h to allow the cells to recover to their normal state. Finally, 100 µM Mn or 1 µg/mL LPS was applied to the cells for another 24 h. After the final incubation period, cells were either collected for RT-qPCR or Western blot or applied to later processes.

For LPS/GNE-049 co-treatment experiments ( Figures 2E–M , 4 , 5 , and S5A ), MMCs were treated with 1 µg/mL LPS in the presence or absence of 500 nM GNE-049 for 3 h. Cells were washed to completely remove LPS and then harvested either immediately for experiments or given 48 h to recover to a steady state.

Some MMC flasks were applied to another treatment paradigm. Either 1 µg/mL LPS only or 1 µg/mL LPS and 500 nM GNE-049 was/were added to MMC cells for 3 h. After a triple-wash to completely remove LPS and GNE-049 and a 24-h incubation in 10% FBS-containing DMEM/F-12 medium, 100 µM Mn was applied to the cells for 24 h.

We followed the standard protocol of primary microglia isolation previously developed by Sarkar et al. (30). In brief, mixed glial cultures from P0 C57BL/6J mouse pups were obtained. Primary mouse microglia were isolated from mixed glial cultures by using magnetic beads and were maintained in similar media as used for MMC, except that 1% nonessential amino acids were added.

Cytokine levels were assessed via Luminex assays following the protocol previously utilized in our lab. Over a short interval, either PMG or MMC cells were treated in 96-well plates with 100–900 μL of 2% FBS-containing DMEM-F12 medium. After treatment, 40 μL of treatment medium was collected, and 40 μL of primary antibody conjugated to magnetic microspheres were added and incubated overnight at 4°C in a clear-bottom, black 96-well plate. Following this incubation, each well was triple-washed in a magnetic washer. The plate was then incubated for 1 h with the desired secondary antibodies. The samples then had streptavidin/phycoerythrin added to them and were incubated for another 30 min. Finally, a Bio-Plex reader was used to read the 96-well plates. A standard curve of all the cytokines was prepared using standard cytokines.

A 12-h light cycle in a climate-controlled mouse facility (22 ± 1°C) with food and water available ad libitum was used for all mice bred and maintained. MitoPark transgenic mice were a kind gift from Dr. Nils-Goran Larsson and were originally generated in his laboratory at the Max Planck Institute for Biology of Ageing by conditionally knocking out TFAM through control of the dopamine transporter (DAT) as previously described (31). All MitoPark mice used for this study were bred from the MitoPark breeding colony at Iowa State University (ISU). Equal numbers of male and female MitoParks were assigned to experiment groups comprising 16- to 20-week-old, age-matched C57BL/6 mice (TFAM^+/LoxP: Dat^+/+ from MitoPark litters or parental-strain litters) or MitoPark mice (Dat^+/Cre: TFAM^LoxP/LoxP), which were genotyped to confirm their identity and further characterized using VersaMax for monitoring locomotor activity and RotaRod for testing coordination of movement. For Mn animal studies, after acclimating for 3 days, C57BL/6 mice were orally gavaged daily with 30 mg/kg of MnCl₂ for 30 days, after which mice were sacrificed. Investigators involved with data collection and analysis were not blinded to group allocation of mice.

Freshly frozen substantia nigra tissue blocks and cryostat sections from the brains of confirmed post-mortem human PD patients and age-matched neurologically normal individuals were procured from the brain banks at the Miller School of Medicine, University of Miami, FL, and the Banner Sun Health Research Institute, AZ. All post-mortem human samples were obtained, stored, and distributed according to the applicable regulations and guidelines involving consent, protection of human subjects and donor anonymity as described previously (32). For immunoblot experiments, tissue homogenates from freshly frozen tissue blocks were prepared at a final concentration of 1 mg/mL total protein from which 35 μg of total protein was directly used. Alternatively, total histone was extracted from tissue blocks and 15-45 μg of total histone was used.

Cells were lysed with modified RIPA buffer, homogenized, sonicated, and finally centrifuged. The same procedure was utilized for processing animal tissues. Before being loaded onto Sodium Dodecyl Sulfate (SDS)-acrylamide gels, proteins were normalized based on their corresponding Bradford assay results. The standard curve utilized for this protein estimation was based on various serial dilutions of 1% BSA. In some cases, the total histone extracted was normalized by a Pierce BCA protein assay kit (Cat. #23225).

For protein separation, 20-40 μg of protein or 15-25 μg of total histone was loaded into the corresponding well of a 15% acrylamide gel. Gels were run at 103 V for 2.5 h at 4°C. Proteins were then transferred to a nitrocellulose membrane at 25 V overnight at 4°C. Following the transfer, membranes were blocked with LI-COR blocking buffer for 1 h and incubated in primary antibodies, according to the manufacturer’s protocol. Next, the membranes were washed with PBS-TWEEN 20 (0.05%) for 1 h and incubated in LI-COR IR secondary antibodies for 1 h on a Belly Dancer (Stovall Life Sciences, Greensboro, NC; setting 3, slow movement) at room temperature (RT). After washing with PBS-TWEEN 20 for 1 h, membranes were imaged by a LI-COR Odyssey scanner. The primary antibodies used were as follows: H3K27ac (Cell Signaling, 1:1000), p300 (Cell Signaling, 1:1000), iNOS (Santa Cruz, 1:750), H3 (Millipore, 1:1000) and β-actin (Sigma, 1:10000) as loading control. The secondary antibodies used were as follows: IR-800 conjugated goat anti-mouse IgG (LI-COR, 1:20000) and IR-700 conjugated goat anti-rabbit IgG (LI-COR, 1:20000).

The ICC protocol previously developed by Huang et al. (ref) was followed, in which the cells were first fixed with 4% paraformaldehyde (PFA), and then double-washed, blocked with blocking buffer, and incubated in primary antibodies. Following the primary antibody incubation (H3K27ac, 1:1000, Cell Signaling; p300, 1:1000, Cell Signaling; iNOS, 1:750, Santa Cruz; H3K4me1, 1:1000, Cell Signaling; H3K4me3, 1:1000, Cell Signaling), the Alexa dye-conjugated secondary antibody was added to the cells following the repeated PBS wash step. After the secondary incubation period finished, the cells were washed with filtered PBS and then left in Millipore water for approximately 1 min prior to being mounted. Cell coverslips were mounted onto microscope slides via Fluoromount aqueous mounting medium (Sigma) and then dried at RT while protected from light. The images were taken with an inverted fluorescence microscope Keyence BZ-X800 (Itasca, IL). Z-stack images of mitochondria were captured by confocal imaging. Quantification of fluorescence intensity was performed by automated BZ-X800 analyzer. For Figure 2A , cell body size enlarged upon treatment, therefore, automated BZ-X800 analyzer was set up to quantify the total fluorescence intensity of H3K27ac per cell rather than intensity peaks.

MitoROS generation was quantified using the MitoSOX red fluorescent indicator. MMCs were placed onto coverslips in a 24-well plate. After treatment, we added the MitoSOX probe (1 µM) to the cells. Fluorescence expressed by the generated mitochondrial superoxides was measured as per the manufacturer’s instructions.

The standard commercial protocol was followed for Invitrogen’s JC-1 mitochondrial potential sensor. In short, 2.0 μg/mL of JC-1 was diluted in 2% FBS-containing MMC media. This was then added to MMC cultures and incubated at 37°C for 20 min. Following this incubation, the cultures were gently triple-washed with PBS. Immediately after the washes, images were captured on the Keyence microscope before cells dried out. The ratio of red to green channel feedback was calculated based on the image analyzer’s results.

Following the standard commercial protocol for MitoTracker (Invitrogen), after the treatment paradigms described for cell culture treatments above, 300 μL of 166-nM CMXROS MitoTracker red dye diluted in 2% FBS-containing DMEM/F-12 medium was added to cell cultures and incubated at 37°C for 12 min. After gentle, triple washes with PBS, cells were fixed in 4% PFA for 30 min prior to following the steps for ICC and imaging.

RNA was isolated from the treated cells through the utilization of TRIzol reagent following the manufacturer’s protocol. Essentially, the cells were first lysed in 0.5 mL of TRIzol reagent during a 5-min incubation at RT. Then 0.1 mL of chloroform was added to the sample and the tubes were inverted to mix. The samples were then incubated for another 2 min and then centrifuged. Following centrifugation, the top clear layer was placed into a new tube containing 0.35 mL isopropanol. Another 150-min incubation at RT followed this and then centrifugation was performed again. After removing the supernatant, the RNA pellets were washed with 75% ethanol, air-dried, and finally re-suspended in warm ultra-pure RNase-free water. The sample nucleic acid concentration was then measured through a NanoDrop spectrophotometer and all samples were normalized to the lowest concentration by the addition of ultra-pure RNase-free water. The samples were then processed via high-capacity cDNA synthesis (kit 4368814, Applied Biosystems, Waltham, MA) following the manufacturer’s protocol. For normalization of each sample in qPCR, the 18S rRNA gene primer set (Cat. No. PPM57735E, Qiagen, Germantown, MD) was used. Based on manufacturer’s guidelines, dissociation curves and melting curves were run to ensure that single amplicon peaks were obtained without any non-specific amplicons. Fold change in gene expression was determined using the ΔΔC_t threshold cycle (C_t) method. Quantitative PCR was performed using RT²qPCR SYBR Green Master Mix (Agilent, Santa Clara, CA).

Microglial cells were plated in 96-well plates and treated in 2% FBS-containing DMEM/F-12 medium. Following treatment, 50 μL of medium was collected, and incubated with 50 μL of Griess reagent for 10 min. A plate reader was utilized to read the absorbance at 540 nm. Sodium nitrite solution was used for making the standard curve.

According to manufacture protocol, cytosolic extraction reagent 1 provided by the NE-PER kit (Thermo Fisher Scientific) and 0.5% Triton X-100 were added to samples for 10 min. After incubation, the pelletized nuclei were collected following centrifugation at 2000 x g for 5 min and then resuspended in 0.2 N HCl for overnight incubation at 4°C for complete dissolution of histones. The following morning, samples were centrifuged at maximum speed for 10 min to collect pellets.

All animal studies followed animal procedures approved by ISU’s (Ames, IA, USA) Institutional Animal Care and Use Committee (IACUC). For human samples, since the post-mortem human brain tissues were obtained de-identified from approved national brain banks, Institutional Review Board (IRB) approval from ISU was not required.

Data analysis was performed using GraphPad Prism 7.0. Normally distributed raw data were analyzed with either Student’s t-test (2-group comparisons) or one-way ANOVA (≥3-group comparisons) with Tukey post hoc test unless otherwise mentioned. Statistically significant differences were denoted as * p ≤ 0.05, ** p<0.01, and *** p<0.001.