The ratio of M1 to M2 microglia in the striatum determines the severity of L‐Dopa‐induced dyskinesias

L‐Dopa, while treating motor symptoms of Parkinson's disease, can lead to debilitating L‐Dopa‐induced dyskinesias, limiting its use. To investigate the causative relationship between neuro‐inflammation and dyskinesias, we assessed if striatal M1 and M2 microglia numbers correlated with dyskinesia severity and whether the anti‐inflammatories, minocycline and indomethacin, reverse these numbers and mitigate against dyskinesia. In 6‐OHDA lesioned mice, we used stereology to assess numbers of striatal M1 and M2 microglia populations in non‐lesioned (naïve) and lesioned mice that either received no L‐Dopa (PD), remained non‐dyskinetic even after L‐Dopa (non‐LID) or became dyskinetic after L‐Dopa treatment (LID). We also assessed the effect of minocycline/indomethacin treatment on striatal M1 and M2 microglia and its anti‐dyskinetic potential via AIMs scoring. We report that L‐Dopa treatment leading to LIDs exacerbates activated microglia numbers beyond that associated with the PD state; the severity of LIDs is strongly correlated to the ratio of the striatal M1 to M2 microglial numbers; in non‐dyskinetic mice, there is no M1/M2 microglia ratio increase above that seen in PD mice; and reducing M1/M2 microglia ratio using anti‐inflammatories is anti‐dyskinetic. Parkinson's disease is associated with increased inflammation, but this is insufficient to underpin dyskinesia. Given that L‐Dopa‐treated non‐LID mice show the same ratio of M1/M2 microglia as PD mice that received no L‐Dopa, and, given minocycline/indomethacin reduces both the ratio of M1/M2 microglia and dyskinesia severity, our data suggest the increased microglial M1/M2 ratio that occurs following L‐Dopa treatment is a contributing cause of dyskinesias.


| INTRODUC TI ON
L-Dopa remains the most effective symptomatic treatment for Parkinson's disease (PD).However, long-term use can lead to the development of hyperkinetic involuntary movements known as L-Dopa-induced dyskinesia (LID) in most PD patients within 10 years of initiating L-Dopa therapy (Manson et al., 2012).While this debilitating side effect limits the L-Dopa dose that can be administered as the disease progresses and therefore limits the effectiveness of L-Dopa, roughly 6-22% of PD patients never develop LID (Ahlskog & Muenter, 2001;Hauser et al., 2007;Hely et al., 2005;López et al., 2010;Tran et al., 2018).Notably, as observed in humans, a similar proportion of 6-OHDA lesioned rodents also fail to develop LID, despite prolonged L-Dopa treatment in dopamine (DA) neuron-depleted animals (Cenci & Lundblad, 2007).Thus far, research has revealed changes in synaptic plasticity (Picconi et al., 2003) and dendritic spine remodeling (Zhang et al., 2013) between the LID and non-LID phenotype in DA neuron lesioned mice, but the mechanism of LIDs remains elusive.If the drivers underlying the change in phenotype can be uncovered, novel therapeutic approaches that prevent LID may be developed.
It is now well-acknowledged that chronic neuroinflammation occurs concurrently with PD pathology and may contribute to DA neuron degeneration (Gelders et al., 2018;McGeer et al., 1988;Qian et al., 2010).Conceivably, as recently suggested, neuroinflammation may also be mechanistically linked to the development of LIDs (Morissette et al., 2022;Mulas et al., 2016;Rentsch et al., 2020).In particular, Morissette et al. found a positive correlation between dyskinesia scores and all (IBA1+) and activated (CD68+) microglia in the putamen of MPTP lesioned monkeys and Mulas et al. demonstrated an increase in activated microglia only in rats that received a L-Dopa regime leading to LID (Morissette et al., 2022;Mulas et al., 2016).Furthermore, we (Rentsch et al., 2020) and others (Barnum et al., 2008;Boi et al., 2019;Bortolanza et al., 2014) have shown anti-inflammatory treatment regimens decrease LIDs concurrently with reducing glial activation and pro-inflammatory cytokine expression in PD rodent models.However, a clear understanding of the links between PD, LIDs and specific changes in activated microglial subpopulations remains elusive.
Microglia polarization has traditionally been classified into classical (M1) and alternative (M2) activation categories (Orihuela et al., 2016;Saijo & Glass, 2011).While this simplification likely does not reflect the true complexity and dynamic nature of microglia (Morris et al., 2013), it remains an initial useful classification.As such, M1 microglia release pro-inflammatory cytokines such as tumor necrosis factor-α (TNF-α) and interleukin-1β (IL-1β) and increase the expression of CD86 and iNos, while M2 microglia up-regulate the surface expression of markers like CD206 and Arg1 and express anti-inflammatory cytokines such as IL-10 (Frakes et al., 2014;Kigerl et al., 2009;Tang & Le, 2015).
We now report the novel finding that an increase in M1 to M2 microglia ratio following L-Dopa treatment in 6-OHDA lesioned mice is a critical predictor of LID severity.Furthermore, lesioned mice that receive L-Dopa but do not show LIDs (non-LID group) show the same (lower) ratio of M1/M2 microglia as seen in lesioned mice that received no L-Dopa (PD group).Finally, minocycline plus indomethacin reduces both the ratio of M1/M2 microglia and dyskinesia severity.These data provide evidence for a causal relationship between the increased M1/M2 ratio that can occur following L-Dopa treatment and the onset of dyskinesias.

| Lesion confirmation
An extensive lesion was confirmed as previously described (Rentsch et al., 2019) in all lesioned mice either by quantifying TH+ and NeuN+ (Figure S1) cells in the SNpc using stereology or measuring TH protein expression in striatal tissue (Figure S1c).
As such, axial, limb and orolingual AIMs were evaluated on a scale of 1-4 based on the severity (amplitude scale) and the amount of time they are present (basic scale).A "global" AIM score was then produced by multiplying basic score and amplitude score.A mouse was classified as non-LID if it displayed none or mild and occasional AIMs (basic scale grade 0-1 and amplitude scale 1-2 on each AIM subtype at each timepoint, up to a total AIM score of 20), whereas mice classified as LID had displayed moderate to severe AIMs (basic and severity scale ≥2 on at least two of the three AIM subtypes at least 2 scoring timepoints).

| Anti-inflammatory treatment
Based on the pre-test AIM scores, animals were equally divided into treatment groups, so that the average of the pre-test AIM scores of both groups were identical.An optimal anti-inflammatory (AI) treatment regime has previously been optimized.Accordingly, a minocycline slow-release pellet (50 mg; Innovative Research of America cat.no.B-169) was implanted subcutaneously and animals received daily injections of indomethacin (1 mg/kg; in Tween 80:DMSO:saline = 1:1:18 i.p; Sigma cat.no.I7378) and minocycline (25 mg/kg; in Tween 80:DMSO:saline = 1:1:18 i.p; Sigma cat.no.M9511) 30 min before L-Dopa.

| Stereology
We quantified numbers of microglia in the entire striatum and TH+ or NeuN+ neurons in the SNpc using stereology with the optical fractionator method (Table S2) and Stereo Investigator 7 software (MBF Bioscience).Coefficient of error attributable to the sampling was calculated according to Gundersen and Jensen (Gundersen & Jensen, 1987).Errors ≤0.10 were acceptable.Data are presented as number of cells per striatal volume (both measures determined using Stereo Investigator) for a more accurate assessment, as striatal volume differs between subjects dependent on the start (1 in 12) of the serial section.Stereological counts of different cell markers were conducted simultaneously in triple stained tissue and colocalization analysis based on their X, Y, Z coordinates.

| Capillary western blotting
Striatal tissue homogenates and protein quantification were performed as previously described (Stayte et al., 2017).Briefly, tissue was homogenized by sonication in 100 μL RIPA buffer (Thermo Fisher cat.no J61529-AK) supplemented with a protease inhibitor cocktail (1:100, Sigma-Aldrich cat.no.S8830) and centrifuged at 16 000 g for 15 min at 4°C.The pellet was discarded and the supernatant stored at −80°C until further use.Quantification of proteins was performed using the capillary automated Wes system (ProteinSimple, SM-W004, DM-003, DM-002), according to manufacturer's instructions (Table S1).Data were analyzed using the Compass software and peak area measurements were obtained for the protein of interest and normalized to the biological loading control.

| Bead-based-immune assay
Using the same striatal tissue sample as for the protein quantifications, TNFα, IL-1β and IL-10 cytokine levels were quantitatively measured by the BD cytometric bead array mouse enhanced Kit (BD Bioscience cat.no.558299, 560 232, 558 300).The operations were performed according to the manufacturer's instructions utilizing a LSR-II flow cytometer (Becton Dickinson) with FACSDiva software and FCAP array software.

| Statistics
All statistical analyses were performed using Prism 6 (GraphPad).Shapiro-Wilk tests were performed on all data sets to assess normality, before analyzing data either with parametric or non-parametric tests.For normally distributed data, differences between means were assessed, as appropriate, by one-or two-way ANOVA with or without repeated measures, followed by Bonferroni post hoc analysis.All data are presented as mean ± standard error of the mean (SEM).For all statistical tests, p ≤ 0.05 was assumed to be significant.Animal numbers were based on previous studies of microglia quantification after anti-inflammatory treatments (Rentsch et al., 2020), where a power analysis, conducted using G*Power, using significance criterion of α = 0.05 and power = 0.80, the minimum sample size needed with the calculated effect size of f = 0.687 was N = 28.Thus, in the current study we deemed an n = 7-8 per group as adequate.

| RE SULTS
To determine disease stage-dependent changes in striatal microglia populations, we collected tissue from mice with different phenotypes (Figure 1a).At study onset, one subset of mice was arbitrarily deemed as the "naïve" group (no 6-OHDA, no L-Dopa).The remaining mice underwent 6-OHDA lesion surgery.
Three weeks post-lesioning, one subset was arbitrarily termed the "PD" group (6-OHDA, no L-Dopa), and the other subsets received repeated L-Dopa/Benserazide injections over 3 weeks.

| L-Dopa exacerbates microglia activation in PD
Neuroinflammation is characterized by the proliferation, migration and activation of non-neuronal cells such as microglia.To investigate phenotype-dependent changes on proliferation, we quantified striatal IBA1 (microglia/macrophage marker) and Tmem119 (microglia-specific marker (Bennett et al., 2016)) positive cells.
While the absolute numbers of microglia had not changed, we next asked if there may be changes in the numbers of cells expressing CD68, a lysosome marker of activated microglia/macrophages.We found a significant difference in CD68+ (F (3,27) = 32.57,p < 0.001; Figure 1d) cells suggesting a significant change in the number of activated microglia/macrophages.Bonferroni post hoc analysis confirmed a 6-OHDA lesion (p < 0.001) dependent increase in CD68+ microglia and intriguingly showed that L-Dopa treatment (LID p < 0.05; non-LID p < 0.01) resulted in a further increase in activated microglia/macrophages when compared to the PD state.
We next sought to differentiate microglia from macrophages, by identifying colocalized markers based on their X, Y, Z coordinates within the stereology software.Accordingly, IBA1+/Tmem119+/ CD68-identifies homeostatic microglia and IBA1+/Tmem119+/ CD68+ identifies activated microglia.IBA1+/Tmem119-/CD68cells are homeostatic macrophages and IBA1+/Tmem119-/CD68+ cells are activated macrophages (Figure 1e).Based on these definitions, our analysis revealed that the striatal macrophage population accounts for less than 10% of all IBA1+ cells and that observed changes in cell phenotype between groups are predominantly because of microglia adapting an activated phenotype.Macrophage population and activation remained largely unchanged.
As might be expected from the above analyses, we found the increase in activated microglia in the PD, LID and non-LID state coincides with a reduction in homeostatic microglia, suggesting that the 6-OHDA lesion triggers local microglia to shift towards an activated state (Two-way ANOVA; interaction: F (9,108) = 23.09,p < 0.001; phenotype: F (3,108) = 3.893, p < 0.05; cell subtype: F (3,108) = 505.5,p < 0.001; Figure 1e).
Overall, these results show for the first time that L-Dopa treatment, regardless of leading to a LID or non-LID phenotype, does not change microglial numbers but does cause an increase in activated microglia population in the striatum, beyond that associated with a PD phenotype alone.
As the striatal macrophage population was negligible, and there were no obvious changes in macrophage activation, we consider any IBA1+ cell to be microglia from here onwards.

| M1 and M2 microglia subpopulations characterize LID phenotype
In the broadest sense, microglial states are categorized as falling under either a pro-inflammatory (M1) or anti-inflammatory (M2) state (Kigerl et al., 2009;Orihuela et al., 2016;Tang & Le, 2015).As we saw an increase in activated (CD68+) microglia in LID mice, we next sought to extend our findings by determining if LIDs were associated with a change of pro-inflammatory (M1) and anti-inflammatory (M2) microglia.
To answer this question, we counted microglial cells in the striatum expressing CD86 and/or CD206, as these are accepted markers of M1 and M2 states, respectively (Frakes et al., 2014;

| Pharmacologically decreasing the M1/M2 microglia ratio alleviates LID
To test a possible causative association of the M1/M2 ratio with LIDs, we investigated the anti-dyskinetic potential of the known anti-inflammatory (AI) treatment regime of minocycline and indomethacin (Figure 3a).Co-administration of both treatments as a cocktail has been shown to potentiate their anti-inflammatory efficacy (Abu-Ghefreh & Masocha, 2010) and is expected to reduce M1+ microglia in vivo (Kobayashi et al., 2013).

| Relationship between microglial M1/M2 ratio and LIDs
Given our observation that the ratio of M1/M2 microglia increases after treatment with L-Dopa only in mice that express LIDs and that AI treatment-induced reductions in the M1/M2 microglia ratio results in a reduction of LID severity, we next investigated if the ratio of M1/M2 microglia can predict AIMs score.We therefore assessed if a correlation exists between these two measures across all groups of mice assessed in this study.As shown in Figure 5a

| NF-κ B activation but not inflammatory cytokines correlate with LID severity
The phosphorylation of NF-κB at Ser536 triggers several cellular responses, including release of classical inflammatory cytokines that are critical to the innate response of activated microglia (Hoesel & Schmid, 2013;Liu et al., 2017).We therefore used capillary western blotting to measure the proportion of phosphorylated NF-κB (Figure S3) and bead-based ELISA to assess expression of classical inflammatory cytokines TNFα, IL-1β and IL-10 (Figure S4) in naïve, PD, LID, non-LID and AI-treated groups of mice.We found that NF-κB activity is exclusively up-regulated in mice with a LID phenotype and suppressed in non-LID or AI-treated mice.By contrast, there was no conclusive relationship between cytokine expression and AIMs severity.This suggests that classical cytokine-driven pathways may not alone account for LIDs onset, but rather is a consequence of more complex actions and interactions of M1 and M2 microglia.
This human phenomenon can be closely modeled in rodents, as approximately 20% of 6-OHDA lesioned mice also fail to develop LIDs after repeated L-Dopa administrations (Cenci & Lundblad, 2007).
The reasons are unclear.Accordingly, our combined findings that L-Dopa treatment leading to LIDs exacerbates activated microglia numbers beyond that associated with the PD state and that the microglial M1/M2 ratio determines vulnerability to LIDs in mice have implications for understanding the human condition and accelerating therapeutic advancements (Figure 5d).

| L-Dopa worsens striatal microgliosis
A multitude of studies has demonstrated the anti-dyskinetic potential of anti-inflammatory treatments, and recent studies (Boi et al., 2019;Mulas et al., 2016) suggest that the underlying striatal microgliosis is indeed L-Dopa dependent rather than a remnant of the initial lesion.As such, studies performed in monkeys and rats, demonstrated that L-Dopa treatment leading to LID exacerbated striatal microgliosis, as assessed via western blotting for IBA1 and CD68 (Morissette et al., 2022) or densitometry of CD11b, OX42 and TNFα (Boi et al., 2019;Mulas et al., 2016).Here, using stereological quantification, we confirm a substantial increase in the amount of activated microglia in the striatum between parkinsonian and L-Dopa-treated mice.Collectively, the consistent finding of a L-Dopa-dependent increased microglia activation across studies, conducted in different species, using different microglia markers and methodologies, demonstrates the robustness of this finding.

| LIDs are associated with an increase in the striatal microglial M1/M2 ratio
While others have shown a correlation between IBA1/CD68 and LID scores (Morissette et al., 2022), we show for the first time that striatal M1/M2 microglia population ratio correlates to AIMs score and does not simply follow on from chronic L-Dopa administration after a confirmed, and extensive, DA neuron lesion.Importantly the M1/ M2 ratio in non-LID mice (lesioned mice that received L-Dopa but did not develop LIDs) is equivalent to that seen in PD mice (lesioned mice with no L-Dopa treatment).Therefore, inflammation following the lesion itself is not the basis of the dyskinetic phenotype; rather, the increased ratio of M1/M2 occurs in some mice following L-Dopa.
Why some mice but not others show this response to L-Dopa is unclear, but it is unlikely to be genetically determined, given the genetic homogeneity of the mice in this study, other than as may occur through epigenetic changes.

| Minocycline and indomethacin treatment reduce M1/M2 ratio and LIDs
If an increase in the M1/M2 microglia ratio is casual for LID development, then reducing this ratio pharmacologically will rescue LIDs.Minocycline and indomethacin are well-known antiinflammatories and minocycline in particular has been suggested to reduce M1 (Kobayashi et al., 2013) and increase M2 (Ahmed et al., 2017;Miao et al., 2018) microglia populations.We found that indeed, minocycline and indomethacin treatment reduced the M1/M2 ratio and reduced LIDs.Our combined observations that (1) in non-LID mice there is no increase in M1/M2 ratio above that seen in PD and (2) reducing M1/M2 microglia ratio using minocycline and indomethacin results in a reduction of LID, together supports the contention that this ratio is causative and that microglia polarization has a direct consequence on developing a dyskinetic phenotype.This has important implications for the development of clinical treatment, as it suggests that only the L-Dopa-associated increase in M1/M2 ratio needs to be reduced to achieve an antidyskinetic effect.

| Multiple neuroinflammatory pathways contribute to LID development
In line with previous research, we found that a dyskinetic phenotype is associated with an increase in NF-κB activity and pro-inflammatory cytokine expression (Barnum et al., 2008;Boi et al., 2019;dos-Santos-Pereira et al., 2016;Mulas et al., 2016;Yan et al., 2021).
However, while treatment with minocycline and indomethacin decreased NF-κB activity this did not translate into a reduction of proinflammatory cytokines.Interestingly, previous research using anti-inflammatory treatment regimens such as corticosterone (Barnum et al., 2008)  (CD40, BAFF), chemokines and protein aggregates (Liu et al., 2017).
Furthermore, in order to rule out technical limitation of the beadbased-immune-assay used in the current study, future work should replicate these findings using traditional ELISA assays or measure TNF-α containing microglia via densitometry as has been performed previously.

| An undetermined number of microglia subtypes
We note a group of microglia co-expressing M1 and M2 markers, termed by others intermediate, mixed, wandering or M1½ microglia (Erkenstam et al., 2016;Jurga et al., 2020;Zheng et al., 2019;Zhou et al., 2017) in vivo.Importantly, in the current study, we do not believe that the narrative of M1 being pro-inflammatory and M2 being antiinflammatory adequately describes the complexity of microglia in the CNS.Accordingly, we are not saying there is less inflammation in PD and more in LIDs; only that the ratio of M1/M2 is important.

| Future studies
As suggested by us (Morris et al., 2013) and others (Hammond et al., 2018), the dichotomous view of microglia as either "good" or "bad" might be an oversimplification and does not acknowledge the variety of tasks microglia serve in the brain simultaneously.
Therefore, using protein markers to identify M1, M2 and M1½ phenotype provides a snapshot, which needs to be interpreted within context that there will likely be many subtypes beyond M1 and M2 classifications.In support of this, single-cell technologies and mass cytometry have revealed numerous microglial subtypes with uniquely multifaceted molecular profiles (Geirsdottir et al., 2019;Hammond et al., 2019;Jordão et al., 2019;Stratoulias et al., 2019).
Accordingly, while our findings provide intriguing indication that microglial subtypes define LID severity, there might be further subtypes of these cells that ultimately even more specifically determine the dyskinetic phenotype.

| CON CLUS ION
Our findings suggest that L-Dopa treatment leading to LIDs exacerbates activated microglia numbers beyond that associated with the

A
total of 150 (n = 25 per group) male and female C57BL/6j mice (RRID:IMSR_JAX:000664; max 5 animals per cage) aged 7-11 weeks were obtained from Australian BioResources and kept on a 12 h light/dark cycle with access to food and water ad libitum.All animal experiments were performed with Garvan Institute and St. Vincent's Hospital Animal Ethics Committee approval (Approval numbers 18/37 and 20/10) in accordance with National Health and Medical Research Council animal experimentation guidelines and the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes (2013).All surgeries were performed under ketamine (8.7 mg/mL; Mavlab) and xylazil (2 mg/mL; Troy Laboratories) anesthesia.The experimenter was always blinded to group and outcome assignment and tissue collection and processing was performed in appropriate blocks.
Kigerl et al., 2009;Kobayashi et al., 2013;Tang & Le, 2015).One-way ANOVA analysis demonstrated a significant phenotype-dependent change of M1+ cell numbers (F (3,27) = 23.85,p < 0.001; Figure2a), with post hoc analysis confirming an exacerbation of M1+ microglia in LID (p < 0.001) which surpasses the increase associated with the lesion alone (PD mice) (p < 0.001).Intriguingly, however, the group that received L-Dopa but did not express LIDs (non-LID mice) had significantly fewer M1+ microglia than the LID group (p < 0.001), suggesting that not L-Dopa treatment per se, but rather the development of dyskinetic behavior is associated with microglia shifting towards a pro-inflammatory M1 state.Given these findings that increased M1+ microglia were associated with LID but not non-LID mice, we next assessed if there was a correlation between M1+ microglia population and LID severity.Indeed, we observed a strong positive correlation (Pearson r = 0.8154, p < 0.001; Figure2d) between AIMs score and CD86+ population, suggesting M1+ microglia population may drive LID development.Our findings for the M2+ microglia numbers were also intriguing.Specifically, we observed a significant increase in M2+ microglia numbers (F (3,27) = 4.84, p < 0.01; Figure2b) in non-LID mice (p < 0.05) but not in LID mice (p = 0.7272) when compared to the naïve state, emphasizing that the mechanism preventing animals from developing LID is associated with more anti-inflammatory M2 microglia.Indeed, a negative correlation (Pearson r = −0.4409,p < 0.05; Figure2d) between AIMs score and M2+ population suggests that increased M2 microglial numbers are indicative of lower LID severity.Colocalization analysis of M1 and M2 markers further confirmed the findings above, by highlighting that in a direct comparison, the LID phenotype harbors more M1 microglia, while the non-LID phenotype has more M2 microglia (two-way ANOVA; interaction: F (9,108) = 11.94,p < 0.001; phenotype: F (3,108) = 0.9329, p = 0.9329; cell subtype: F (3,108) = 475.3,p < 0.001; Figure2e).Additionally, it identified a group of microglia expressing both CD206 and CD86 markers, referred to as "M1½microglia"(Erkenstam et al., 2016;Jurga et al., 2020;Zheng et al., 2019;Zhou et al., 2017), suggesting the presence of a subgroup of microglia able to conduct different functions simultaneously.Given the severity of LID behavior positively correlated with M1+ and negatively correlated with M2+ microglia populations, we next created the M1/M2 ratio to investigate if this ratio can predict LID development.One-way ANOVA analysis demonstrated a significant effect between M1/M2 ratio and phenotype (F (3,27) = 9.528, p < 0.001; Figure2c).Post hoc analysis revealed the M1/M2 ratio is exclusively increased in mice with a LID phenotype, suggesting the M1/M2 ratio as a LID defining feature.As such, the M1/M2 ratio increase is directly linked to dyskinesia severity per se, and not to either chronic L-Dopa administration or the presence of the underlying 6-OHDA lesion.
, we found a strong positive correlation (Pearson r = 0.6821, p < 0.001; Figure 5a) between AIMs score and M1/M2 ratio across our whole dataset.Importantly, this effect was not because of differences in 6-OHDA lesion severity, as the remaining DA neuron population in the SNpc correlated with neither the M1/M2 microglia ratio (Pearson r = 0.2194, p = 0.2442; Figure 5b), nor LID severity (Pearson r = 0.007734, p = 0.9671; Figure 5c).

F
Treatment with minocycline and indomethacin results in an anti-dyskinetic effect.(a) Experimental timeline.Treatment of DA neuron lesioned mice with minocycline (slow-release pellet of 50 mg and daily 25 mg/kg i.p injections) and indomethacin (1 mg/kg) resulted in significant reduction in global AIMs score (b) on day 7 and 10 of AI treatment, by reducing limb and axial AIMs in particular and being most effective at the 80 and 100 min timepoint during each monitoring session on (d, e) day 7 and (f, g) day 10 (n = 17 per group).Western blotting analysis revealed (c) reduced FosB expression in the ipsilateral striatum when compared to the control (n = 8 per group).All values represent the mean ± standard error of the mean (SEM).*p < 0.05, **p < 0.01, ***p < 0.001.
or Capsazepine + Cannabidiol (dos-Santos-Pereira et al., 2016) demonstrated anti-dyskinetic potential alongside an exclusive decrease of either TNF-α or IL-1β, respectively.These differences are notable.Together the studies confirm the value of anti-inflammatory treatments for suppressing microglial activation and restoring the balance of M1/M2 microglia, but suggest that they are not exerting actions via an effect on TNF-α or IL-1β in isolation.There are clearly as yet undefined actions of microglia in causing LIDs that are reversible with anti-inflammatories.Accordingly, future studies should investigate what leads to the increased activation of NF-κB in LID, by measuring an increased range of cytokines (IL-12, IL-23, Interferons) as well as other NF-κB non-canonical stimulators , are increased in the PD and LID phenotype.While the functional relevance and nomenclature of this type of microglia are not yet defined, research has suggested the possibility of two-directional polarization in a single microglial cell.While this is against the classical conception of a defined polarization status of microglia, it highlights the dynamic nature of microglia in the brain F I G U R E 4 Minocycline and indomethacin treatment reduce the M1/M2 microglia ratio.Minocycline and indomethacin had no effect on total (a) microglia/macrophage (IBA1+) and (b) microglia-specific (IBA1+/Tmem119+) populations.However, (c) activated glial (IBA1+/ CD68+) and (d) in particularly the activated microglia (IBA1+/Tmem119+/CD68+) rather than the activated macrophage population (IBA1+/ Tmem119-/CD68+) were significantly decreased in the treatment group compared to control.Minocycline and indomethacin decreased (f) M1 (IBA1+/CD86+) microglia subpopulation, while not having an effect on (g) M2 (IBA1+/CD206+) or (i) intermediate (IBA1+/CD86+/ CD206+) microglia population.(h) Overall minocycline and indomethacin treatment reduced M1/M2 microglia ratio.(j) Representative images.(n = 7-8 per group).All values represent the mean ± standard error of the mean (SEM).**p < 0.01, ***p < 0.001 compared to control group.Scale bar represents 25 μm.Arrows point towards CD86+ microglia cells.

F
I G U R E 5 M1/M2 microglia ratio in the striatum is a predictor of LID severity.(a) AIMs score positively correlated with the M1/M2 ratio in our full data set including LID, non-LID, control and minocycline/indomethacin-treated mice.There was no correlation between lesion severity, as measured by TH+ neurons in the SNpc, between either (b) M1/M2 microglia ratio or (c) AIMs score.(N = 31 including n = 7-8 per group from LID, non-LID, control and AI-treated mice) (d) Cartoon illustration summarizing our findings that M1 and M2 microglia are increased in DA neuron lesioned (PD) versus non-lesioned mice.While the total number of microglia does not change, the ratio of M1+ to M2+ expressing microglia is up-regulated in mice expressing LIDs compared to mice that received L-Dopa but did not express LIDs or in mice treated with minocycline and indomethacin.