Microglia are permanently in contact with astrocytes, neurons, and endothelial cells, constantly monitoring surrounding tissue for an ongoing infection or trauma via thin processes (45). In response to injury or infection, resting microglia undergo morphological alterations, change their transcriptional activity, and present antigens via the major histocompatibility complexes (MHC)-I and -II (46–48). Unlike MHC-I, which is constitutively expressed on all nucleated cells and platelets (49–52), MHC-II is specifically expressed on antigen-presenting cells such as microglia, astrocytes, monocytes, macrophages, dendritic cells, and granulocytes, and can be induced upon activation.

As brain-resident innate immune cells, microglia are typically the first responders to altered homeostasis within the CNS. Microglia activation increases the amount of nuclear factor ‘kappa-light-chain-enhancer’ of activated B-cells (NFKB) and NLR family pyrin domain-containing 3 (NLRP3), with subsequent upregulation of nicotinamide adenine dinucleotide phosphate (NADPH)-oxidase and cytokines such as interleukin-1 beta (IL-1B) and tumor necrosis factor alpha (TNF) (53, 54). Phagocytic activity and MHC-II expression are also enhanced (55, 56). An increase in activated microglia has been observed in both PD patients and healthy older people, indicating that various triggers might induce a proinflammatory shift with age (21, 57). Compared to age-matched controls, PD patients exhibited significantly elevated levels of proinflammatory cytokines such as IL-1B, IL-6, and TNF in CSF and blood (39, 58). Furthermore, ([¹¹C](R)-PK11195)-based positron emission tomography (PET) has revealed a higher density of activated microglia in the midbrain and putamen of early PD patients (59, 60), which correlated with decreased activity of dopamine transporter ligands ([¹¹C] CFT). Similar results were described in patients suffering from a REM sleep behavior disorder (60, 61). While the presence of chronic inflammatory processes in PD is now widely accepted, the underlying reason for neuroinflammation is still unclear. Due to its function as a damage-associated molecular pattern (DAMP) (53, 62, 63), pαSYN might trigger and maintain a proinflammatory shift of the immune system. In addition, DAMPs may be released from dying or damaged cells. Well-known DAMPs triggering an innate immune response upon interaction with pattern recognition receptors (PRRs) are IL-1α or mitochondrial reactive oxygen species (mROS), which activate NLRP3. Consecutive NLRP3 activation leads to increased IL-1B synthesis as a trigger for further innate immune responses (64). Initial microglia activation in PD (21, 65) may therefore result from PRR-mediated responses to DAMPs.

In addition to DAMPs, pathogen-associated molecular patterns such as viral RNA or the bacterial cell wall component lipopolysaccharide (LPS) may also maintain neuroinflammation. While a correlation between viral infection and PD has not strictly been established, preceding infections might still modulate the risk for developing PD (66). For example, there is evidence for toxin-related PD, which implies that toxins released by infectious agents might also be responsible for some cases of ‘sporadic’ PD. Other toxins such as the pesticide rotenone have neurotoxic and neuroinflammatory effects that may kill dopaminergic cells and thereby cause PD in people exposed to higher concentrations (67).

Possible links between PD and infection have become even more topical with the still-ongoing worldwide COVID-19 pandemic. While there is very little consolidated knowledge regarding possible interactions between the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and PD, there are some interesting observations on pathogenic coronaviruses and neuroinflammation or, more specifically, PD. Antibodies against coronavirus antigens have been found in the CSF of PD patients (68). In addition, intranasal infection of hACE2 transgenic mice with SARS-CoV-1 caused severe CNS infections that spread from the olfactory bulb to the basal ganglia and midbrain within approximately four days (69). Considering that COVID-19 patients often experience anosmia, SARS-CoV-2 is also likely to reach the CNS via the olfactory bulb. Disconcertingly, in vitro assays have shown that the nucleocapsid(N)-protein of SARS-CoV-2 accelerates pαSyn-aggregation by a factor of 10 (70). Thus, it is concerning that COVID-19-related inflammation may trigger or accelerate PD in patients at risk. Yet to date, there are only a few case studies describing the development of PD after SARS-CoV-2 infection (71, 72). Whether infection with SARS-CoV-2 impacts the development of PD remains to be explored.

Neuroinflammation in PD is not confined to the innate immune system; it also involves adaptive immune responses. Several findings indicate that various T cell subpopulations may contribute to PD pathophysiology (73).

Neuroinflammation within the substantia nigra (SN) is well-described in PD. Significant infiltration of both CD4⁺ and CD8⁺ T cells into the SN of PD patients has been described (42), particularly elevated levels of CD8⁺ T cells (74, 75). It is possible that CD8⁺ T cells have an important role in early PD, even before pαSYN can be detected in the SN, as relevant infiltration into the SN has been observed in very early PD that subsides with disease progression (74). Altered counts of CD4⁺ T cell populations in the blood of PD patients have also been reported, but remain controversial. Many studies describe a decrease in the overall CD4⁺ T cell population in PD patients (44, 76–78). However, both elevated (76) and reduced T_reg cell counts (79) were found in the blood of PD patients, although the studies used a relatively non-specific marker (CD4⁺CD25⁺) to identify T_reg cells. Increased numbers of CD4⁺FoxP3⁺ T_reg have been observed in the blood of PD (aged 46-80 years) and AD (62-82 years) patients and healthy older people (51-87 years) compared to healthy young controls (23-40 years) (80). An increase in T_reg cell activity with age was also seen, which was significantly more pronounced in PD and AD patients compared to young controls (80). In contrast, decreased levels of CD4⁺CD25^highCD127^low T_reg cells (44, 81) and reduced suppressive activity of T_reg cells has also been found in PD patients (77). Furthermore, peripherally-reduced T_reg cells and increased Th1 cell counts were found to correlate with the severity of motor dysfunction (assessed by Unified Parkinson’s Disease Rating Scale III [UPDRS III]) (44, 81). In addition, decreased Th2 cell numbers and a relative increase in Th1 cells were seen in PD patients (44). Despite these discrepancies, most studies highlight an increased peripheral immune response in PD, with a reduced amount of T_reg cells (CD4⁺CD25^highCD127^low) (44, 81) (Table 1).

These population and activity shifts of T cells, along with an increase in HLA-DR positive antigen-presenting microglia in PD patients (82), support the idea that CD4⁺ T cells might contribute to neurodegeneration in PD. Studies on the role of Th17 cells in PD have generally confirmed this concept. Addition of IL-17 to autologous cocultures between T cells and pluripotent stem cells (iPSC)-derived mid brain neurons (MBN) from PD patients also induced NFKB-dependent cell death. iPSC-derived MBN from PD patients further showed higher expression of IL-17 receptors, together with enhanced susceptibility towards death induction by recombinant IL-17 or autologous Th17 cells (73). Interestingly, Th17 cells are increased in PD patients, which is consistent with the results of previous studies (81, 83, 84). However, one study in PD patients reported a decrease in the Th17 cell count, along with unchanged levels of IL-17 (44). Based on these studies, Th17 cells and IL-17 are likely to favor the progression of PD, but their distinct role remains unresolved (85, 86). Consequently, a comparison of different analytical approaches might be needed to rule out possible methodological issues.

In addition to flow cytometric analyses, gene expression analyses and genome-wide association studies that investigate alterations in immune response pathways are playing an increasing role in PD research. Expression quantitative trait loci (eQTLs) are obtained by analyzing the genome and transcriptome of patients with correlating alleles and the expression level of transcripts. Based on CD4⁺ T cells and monocytes from a multi-ethnic cohort of 461 healthy individuals, susceptibility alleles for MS, rheumatoid arthritis, type 1 diabetes, and PD were found to be overrepresented in CD4⁺ T cells, while disease- and trait-associated cis-eQTLs associated with AD and PD were primarily overrepresented in monocytes (87). Further evidence for the relevance of the immune system in PD pathogenesis comes from the association of PD with an immune haplotype, including the DRB5*01 and DRB1*15:01 alleles, and from a non-coding polymorphism in the region that might increase MHC-II expression (88). Finally, a very recent study stratified PD patients by their T cell responsiveness to α-SYN as a proxy for an ongoing inflammatory autoimmune response (89). Gene expression analysis in peripheral memory T cells then revealed a clear PD-specific gene signature with transcriptomic markers for inflammation, oxidative stress, phosphorylation, autophagy of mitochondria, cholesterol metabolism, and chemokine signaling via CX3CR1, CCR5, and CCR1.

While T cells are the more prevalent adaptive immune cells in CSF (90), B and T cell immunity are closely intertwined (91). Although immunohistochemical analyses have found B cell-rich tertiary lymphoid structures in most brains from patients with MS, but not in the two PD patients investigated (92), there is ample evidence for the involvement of B cells, antibodies, and humoral immune effector mechanisms in PD (93–95). In a postmortem study of 13 patients with idiopathic PD and three with genetic PD (compared with 12 controls), all PD patients showed IgG (mostly IgG1) but no IgM binding to dopaminergic neurons, while Lewy bodies were strongly immunolabelled with IgG (94). Nearby activated microglia expressed the high-affinity activating IgG receptor FcγRI, which implies a significant role for antibody-dependent cell-mediated cytotoxicity. The inhibitory IgG receptor FcγRII was absent in all cases. Subsequent studies on B cell-related adaptive immune responses in PD support the involvement of B cells and humoral immune effector mechanisms (96–98). Remarkably, some patient sera also contained antibodies capable of neutralizing pαSYN ‘seeding’ (forming of αSYN aggregates) in vitro (99). Elevated levels of serum antibodies against αSYN have also been reported in a comparison of sera from PD patients with AD patients and controls (100). Of note, B cell numbers were found to be reduced in PD patients (96, 101), along with an increase in IgA levels (102) that correlated with non-motor symptoms (96). Moreover, cytokine expression of B cells was altered in PD (96). Cytokine-dependent interactions between B cells and Th1 and Th17 cells (103) imply that B cells may also modulate proinflammatory Th17 cells in PD patients (73). In addition, there are relevant interactions between the humoral immune response and the complement system, as described in a comprehensive review (104).

Given the paucity of studies focusing on the distinct function of B cells in PD or other neurodegenerative diseases (103, 105), a more profound understanding of B cell function might reveal new opportunities for disease modification. Again, the targeted transfer of knowledge from ‘classical’ neuroinflammatory disease would be useful (103).

Since proinflammatory changes have been consistently observed in PD, several studies analyzed whether suppressing inflammation might modify the disease course of PD. A population-based, case–control study found that patients receiving immunosuppressive therapy had a decreased risk of developing PD (106). Interestingly, the greatest risk reduction was conveyed by drugs affecting T cells such as the inosine monophosphate dehydrogenase inhibitors, azathioprine and mycophenolat mofetil (106), while a protective effect was revealed for corticosteroids. However, this finding might be biased by smoking behavior. Smoking is described as a possible protective factor for PD, and smokers frequently need corticosteroids due to pulmonary maladies (106, 107). Moreover, cigarette smoke induces chronic inflammation, which results in immune cell exhaustion and generally attenuates the function of many immune responses (108). Therefore, smoking can be considered an immune-inhibitory activity (109, 110). The potential effectiveness of anti-inflammatory treatment is further exemplified by a lowered risk for developing PD in patients receiving immunosuppressive therapy (111).

Accordingly, patients with inflammatory bowel disease (IBD) receiving an anti-TNF treatment had a 78% reduction in the incidence of PD compared to IBD patients who did not receive this therapy (112). However, the lack of direct comparison between the incidence of PD in treated IBD patients compared with non-IBD patients (112) means that the ability of anti-TNF treatment to reduce PD incidence in non-IBD patients remains unanswered. Other research has indicated that severity of IBD is inversely associated with the risk of developing PD (-15%) (113). However, although patients suffering from severe IBD are more likely to receive anti-inflammatory drugs, immune-modulatory treatment was not directly assessed. Accordingly, further studies are necessary to verify a positive or negative correlation between PD and anti-inflammatory treatment, both for IBD and for non-IBD patients.

One of the most frequently used anti-inflammatory drugs is ibuprofen, a reversible cyclooxygenase-2 (COX-2) inhibitor and non-steroidal anti-inflammatory drug (NSAID) (114). Intriguingly, ibuprofen users showed a lower risk for PD, whereas no protective effect could be demonstrated for acetylsalicylic acid (ASA) (114). In a prospective study of subjects without PD who were assessed at baseline for NSAID intake, 0.2% developed PD during a 6-year follow-up (115). A significant dose-response relationship between ibuprofen intake (tablets/week) and reduced PD risk was found. Again, this effect was not observed for ASA, other NSAIDs (e.g., indomethacin), or acetaminophen. Although all NSAIDs and ASA are known to inhibit COX-2, only ibuprofen was shown to have some impact on PD prevention. Pre-clinical findings from in vitro models indicate that ibuprofen prevents oxidative damage induced by LPS injection and better protects dopaminergic neurons against glutamate neurotoxicity than ASA (116, 117). The neuroprotective effects of ibuprofen have been confirmed in studies that considered confounder variables of PD, such as age, smoking, and caffeine intake (115, 118). However, the putative neuroprotective effect of ibuprofen in PD remains controversial; other research focusing on all NSAIDs rather than on ibuprofen alone found that NSAIDs had no neuroprotective effects (107). In this context, it is noteworthy that ibuprofen is known to reduce inflammation via inhibition of COX-2, but also via modulation of the angiotensin pathway by shifting the ACE/ACE-2 (angiotensin converting enzyme) receptor ratio towards ACE-2 (119, 120) (discussed in detail in Section 4.2).

Drugs such as β-agonists and β-antagonists have been shown to affect the incidence of PD in opposing directions, with β-agonists seeming to be somewhat protective (121, 122). However, further evidence is needed to confirm whether β-agonists might act directly on immune cells and are thus capable of modulating PD-related inflammation (122). It should be noted that β-agonists are often used in patients suffering from chronic obstructive pulmonary disease, a common consequence of frequent smoking. The fact that smoking was reported to lower the incidence of PD in one study represents an epidemiological bias (121).

Given that observational studies have yielded controversial data regarding the neuroprotective effects of different drugs and immunomodulators (123–128), randomized, controlled trials are needed to better decipher which drugs have distinct neuroprotective effects in PD.

Current clinical trials on anti-inflammatory therapy in PD focus on the treatment of glucose metabolism, improvement of oxidative stress, and regulation of gut microbiota.

Toxins such as MPTP, rotenone, and 6-OHDA act as DAMPs and initiate a strong innate immune response with microglia activation and subsequent neurodegeneration in the SN (170–172). This immune response resembles the microglia activation found in human brain autopsies (21, 173). In the 6-OHDA model, neurodegeneration is accompanied by a gradual repolarization of microglia from an anti-inflammatory M2 to a pro-inflammatory M1 phenotype (174). Cytokine production by M1 cells is intracellularly initiated by NFKB (53, 175, 176), which in turn induces interleukin and procaspase-1 transcription. Together with the inflammasome NLRP3, caspase-1 activates IL-1B. Moreover, other proinflammatory proteins such as iNOS and TNF are also released from M1 cells (53, 175, 176) and contribute to neurodegeneration in PD (177). Consequently, inhibition of NFKB protects MPTP mice from neurodegeneration (178). M2 macrophages, in contrast, contribute to neuroprotection and release neurotrophic factors (176) (Figure 1). Since TLR-4 knock-out reduced the number of MHC-II⁺ microglial cells and partially protected MPTP-treated mice from dopaminergic degeneration in the SN, MPTP triggers also neuroinflammation via TLR-4 (179). However, the canonical ligand for TLR-4 is LPS, which is commonly found on cell walls of gram-negative bacteria (180) In LPS-driven models, LPS is either systemically administered or focally injected into the SN (181). Since microglia activation after delivery of LPS or another of the aforementioned toxins results in the same downward cascades, all of these models lend themselves to studying innate immune responses (181–183). However, in mice lacking the mitophagy inducers PINK and Parkin, LPS triggers the release of mitochondria-derived vesicles (184, 185). This can prime CD8⁺ T cell responses against mitochondrial antigens (184, 185). External stressors such as LPS can thus also induce adaptive immune responses, in particular in transgenic models with impaired mitochondrial functions.

For MPTP, but not 6-OHDA, a Lewy body-like pathology has been found in non-human primates (186). An even more prominent Lewy body-like histopathology was observed in viral vector-based models overexpressing human mutations of αSYN (e.g., A53T) (187). Furthermore, viral vector-based models demonstrate a similar innate immune response, as seen in toxin-based models. pαSYN, which is believed to enter the cell via TLR-2 to unleash its toxic potential (53, 188, 189), thus apparently acts like a DAMP (190). Interestingly, orthotopic injection of an αSYN-encoding AAV into the SN only induced neurodegeneration in the presence of Fc gamma receptors (191), which implies an antibody-dependent activation of the complement cascade and/or antibody-dependent cellular cytotoxicity that may be mediated by microglia, for example (192). However, unlike astrocytes, microglia do not directly contribute to a proinflammatory environment after exposure to pαSYN (193). Nevertheless, progressive neurodegeneration can be induced by microglia-specific overexpression of pαSYN in the absence of pαSYN in dopaminergic cells (194). The critical role of CD4⁺ T cells was confirmed by showing that virus-based αSYN-overexpression can only induce microglia activation and neurodegeneration when MHC-II is present (195). Therefore, viral vector-based models require a complex interplay between different types of immune cells, which may also best reflect observations made in human PD.

Since animal research in PD has focused mostly on alterations within the CNS, our understanding of peripheral innate and adaptive immune processes in PD pathology is still limited. Recent studies have provided evidence that increased gut permeability and gut microbiota are likely to contribute to PD development (149, 196). This has resulted in PD becoming more and more accepted as a systemic disease (197). For instance, overexpression of human TNF (hTNF) in a mouse model leads to microglia activation in various regions of the brain (198, 199), while immunomodulation with infliximab could attenuate microglia activation (199). Since infliximab cannot penetrate the blood-brain barrier, this finding indicates that effective treatment of neuroinflammation could be achieved by modulating peripheral immune mechanisms.

Several animal models support the hypothesis that the adaptive immune response in PD is mainly T cell driven, whereas the role of B cells remains enigmatic (41, 42, 73, 75, 88, 200). Nevertheless, the mechanisms of T cells vary across different animal models (42, 201).

ASA is a standard drug for secondary prophylaxis of ischemic stroke and a potent non-competitive inhibitor of COX-2. Since COX-2 is expressed in many cell types, including platelets and microglia, it is involved in many physiological processes. Intriguingly, the protective role of COX-2 inhibitors on neuroprotection remains controversial. An early administration of high ASA doses showed a neuroprotective effect in MPTP mice, while other COX-2 inhibitors including ibuprofen did not (226). In contrast, the protective effect of ASA in PD patients is comparatively low, while ibuprofen is associated with decreased disease risk compared to other NSAIDs (114, 115). Furthermore, some studies have shown that NSAID intake has no effect on PD development, but instead correlates with an increased risk for PD (107, 227). It is important to note that none of these studies compared equivalent dosages. In MPTP mice, however, COX-2 is upregulated and known to be involved in the oxidation of dopamine, forming dopamine quinone (DAQ) (202). Interestingly, COX-2-deficient mice are protected against induction of PD, which was not explained by reduced inflammation but by prevention of DAQ formation (202, 228, 229). DAQ conjugates with GSH to form GSHDAQ. After conversion to 5cysDAQ, GSHDAQ inhibits mitochondrial complex I. In turn, this leads to decreased degradation of ROS and increased DAQ levels. Subsequently, this vicious circle results in microglia and astrocyte activation and a shift towards a proinflammatory state (175), accompanied by exhaustion of the pentose-phosphate pathway (PPP). However, there are multiple pathways that result in upregulation of COX-2 in brain neurons, including increased N-methyl-D-Aspartate (NMDA) receptor-dependent activity (230), cerebral ischemia, seizures, and diseases such as AD or PD (231). Another important function of COX-1 and COX-2 is the further processing of arachidonic acid. COX-1 induction leads to a substantial increase in thromboxane A2 (TXA₂) in human endothelial cells, while induction of COX-2 by IL-1B leads to a minor increase in TXA₂ but a substantial induction of the proinflammatory proteins prostacyclin (PGI₂) and prostaglandin E2 (PGE₂) (232).

PGI₂ and PGE₂ are physiologically involved in body temperature regulation. PGI₂ and TXA₂ have important roles in the maintenance of vascular homeostasis and can mainly be described as agonist and antagonist (233). Among the different PGE₂ receptors (EP1-4), EP2 contributes to neurodegeneration in the presence of pαSYN in PD (234). EP2-deficient microglia from mice showed increased clearance of pαSYN on brain slices from Lewy-body patients. Furthermore, EP2^-/- MPTP mice were spared from neurodegeneration in the SN (234). However, this mechanism only seems to apply for neurotoxicity associated with chronic inflammatory processes. In acute pathological processes such as brain ischemia or NMDA-induced toxicity, EP2 activation exerts protein kinase A-dependent protective effects (231). Interestingly, EP2 is also neuroprotective in the 6-OHDA primary cell culture (235). The cause of this dichotomous action of EP2 is not yet understood (231).

Beside possible neuroprotective effects, COX-2 inhibitors increase intestinal permeability that correlates with their potency to inhibit COX-2, thus ultimately leading to local inflammatory side effects (236, 237). Furthermore, since increased intestinal permeability is discussed as an important pathological axis of PD development, prolonged treatment of PD with potent NSAIDs is not advisable, which could explain the controversial results (149, 196).