The search of PubMed, Embase, and Web of Science databases (from inception up to January 2022) was performed by two authors (ZL and ZYS) independently. The search strategy involved selecting the most relevant Medical Subject Headings (MeSH) and title/abstract text keywords used in combination or alone: “Parkinson's disease,” “NK cell,” “natural killer cell,” “alpha-synucleinopathy,” and “dopaminergic nerve cell” (detailed search strategy is described in Supplementary Table 1).

To prove the eligibility of the studies, the following inclusion and exclusion criteria were formulated:

Inclusion criteria were as follows: (1) articles published in English, (2) original articles regarding animals and humans (peripheral lymphocyte analysis) and in vitro studies (cell cultivation) of natural killer cells where relevant, and (3) the research theme focused on “Parkinson's disease.” Exclusion criteria were as follows: (1) case report, case series, letter, poster, commentary, proceedings, laboratory science studies, review, and systematic review, (2) repeatedly published literature, unpublished data, or findings reported in abstract form only, and (3) literature of poor quality.

Data extraction from eligible articles was performed by two investigators (ZL and ZYS) independently. The following clinical information was extracted from the study: publication, region of the population, animal models of PD, number of patients and healthy controls, gender statistics, age at diagnosis, disease duration, disease severity, and the main results.

Quality assessment was carried out by two authors (ZL and ZYS) independently. Modified Collaborative Approach to Meta-Analysis and Review of Animal Data from Experimental Studies (CAMARADES) and Newcastle-Ottawa Scale (NOS) were used as quality assessment tools for the animal experimental studies and clinical studies separately. Ten items are included in the modified CAMARADES tool: sample size calculation, random allocation to treatment or control, allocation concealment, blinded assessment of outcome, appropriate animal model, use of anesthetic without significant intrinsic neuroprotective activity, statement of control of temperature, compliance with animal welfare regulations, peer-reviewed publication, and statement of potential conflict of interests. Seven items are included in the NOS tool: adequate case definition, representativeness of the cases, selection of controls, definition of controls, comparability of cases and controls on the basis of the design or analysis (study controls for age or any additional factors), ascertainment of exposure, and same method of ascertainment for cases and controls and non-response rate. The total scores of the two scales are 10 and 9, respectively. The final scores of each study are described in Supplementary Tables 2, 3. Any discrepancies were resolved through discussion. If an agreement could not be reached, the disagreements were resolved with the help of a third author.

We identified three studies on animals according to the inclusion criteria, and the main findings are summarized in Table 1. Two studies used a preformed fibril (PFF) α-syn injection model in C57BL/6J mice and transgenic mice that overexpressed human A53T α-syn mutant protein separately, and one study used a 6-hydroxydopamine (6-OHDA) injection model in Wistar rats. These two models are commonly used in the study of PD. For the PFF α-syn model, PFF α-syn acts as a template and induces the endogenous α-syn to accumulate into misfolded phosphorylated pathological aggregates in the substantia nigra (Earls et al., 2020). In the 6-OHDA model, 6-OHDA (a neurotoxin) is absorbed by dopaminergic neurons rich in monoamine oxidase in the substantia after injection, which can further convert into free radicals and cause damage to neurons; 6-OHDA does not induce the formation of α-syn aggregates or Lewy-like inclusions in PD (Grembecka et al., 2021).

The first study (Earls et al., 2019) showed that intrastriatal injection of PFF α-syn could induce Lewy body–like pathology, and the percentage of NK cells in the mononuclear immune cells at 5 months post injection (p.i.) was relatively increased in the CNS parenchyma, which was confirmed by the flow cytometry analysis. In addition, characterization of the leukocytes in the peripheral blood showed that the number of NK cells decreased, but the percentage of the NK cell population did not show significant changes.

In the second study, Earls et al. confirmed the presence of NK cells in the SNs of both postmortem PD patients and M83 Tg mice based on immunohistochemistry (IHC) markers (Earls et al., 2020). More importantly, obvious clinical motor deficits and an increased number of p-α-syn inclusions within the striatum, substantia nigra pars compacta (SNpc), and brainstem were observed in PFF α-syn–injected M83 Tg mice after systemic NK cell depletion, which indicated the protective role of NK cells in PD. Moreover, researchers carried out cell-based assays and found the bidirectional effects between extracellular α-syn and NK cells: NK92 cells and primary human NK cells could internalize the extracellular α-syn and scavenge it through the endosome and lysosome pathway, while extracellular α-syn aggregates attenuated NK cell cytotoxicity (NKCC) and IFN-γ secretion without causing any significant changes in NKG2A and NKG2D receptors on the cell surface. It is an interesting phenomenon that NK cells are not aberrantly activated.

For the third study (Grembecka et al., 2021), we extracted information regarding the comparison between the PD model and the blank controls. It seems that the number of NK (CD3-CD161a+) cells in PBMC did not show any significant difference between the 6-OHDA groups and the controls (without p-values recorded). Furthermore, they observed elevated peripheral NKCC levels after 6-OHDA microinjection into the SNpc, which means NK cells might be actively involved in the progression of PD.

Nine clinical studies explored the relationship between NK cells and sporadic Parkinson's disease according to the inclusion criteria, and the basic information and important findings of these studies are summarized in Tables 2, 3. Most of the included studies provided basic information regarding the subject number, gender, age at visit/onset of patients, disease duration, and H&Y/UPDRS-III (part III of the Unified Parkinson's Disease Rating Scale) assessment. NK cells are mainly characterized based on the cell number, cell cytotoxicity, and changes in the cell surface receptors.

Seven studies analyzed NK cell count in the peripheral blood. One of them found no differences between the patients and controls (Stevens et al., 2012), five of them reported a higher percentage of NK cells in the patients with PD (Mihara et al., 2008; Niwa et al., 2012; Cen et al., 2017; Sun et al., 2019; Huang et al., 2021), and in one study, Tian et al. divided NK cells into three clusters and reported that only C32 (CD56+, CD16+, CD57–, and CD28–) and C27 (CD56+, CD16+, CD57+, and CD28–) clusters increased (Tian et al., 2022). Furthermore, we noticed that the markers used to identify NK cells from the PBMC were quite different. CD16, a typical NK cell marker and also known as FCγRIII, is an activation receptor that mediates ADCC effects. CD56, known as nerve cell-associated adhesion molecule (NCAM), is considered as a functional indicator of human NK cells and separates the cells into precursor and mature populations based on its expression. In some studies, cells labeled by CD16 or CD56 were used as markers, while in others, those cells that expressed both CD16 and CD56 were recognized as NK cells. Mihara et al. and Niwa et al. reported the occurrence of the inhibitory NKG2A cells and the activating NKG2D cells, which indicated that the percentage of NKG2A+ cells among CD3-CD56+ NK cells was lower and the percentage of NKG2D+ cells in the peripheral lymphocytes was higher in the PD group (p < 0.05) (Mihara et al., 2008; Niwa et al., 2012).

Experiments and bioinformatics analysis were used to analyze the cytotoxicity in NK cells. Only two studies measured NKCC activity against the K562 target cells through the LDH cytotoxic assay and the calcein acetoxymethyl ester release assay, and both of them found no significant differences between the patients with PD and the controls (Mihara et al., 2008; Niwa et al., 2012). Additionally, the former study provided information that NKCC increased as the disease progressed in PD (Mihara et al., 2008). On the other hand, from the perspective of bioinformatics analysis, Huang et al. carried out the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis and revealed that NK cell-mediated cytotoxic pathways were enriched in PD based on the blood expression profile (Huang et al., 2021).

Since aging is a common predisposing factor for Parkinson's disease, stratification by age at disease onset needs to be considered (Pagano et al., 2016). Two studies analyzed the subset of NK cells in the early-onset Parkinson's disease (EOPD) and late-onset Parkinson's disease (LOPD). Cen et al. observed that the percentage of NK cells in peripheral blood did not differ significantly between the two groups (p = 0.067) (Cen et al., 2017). Tian et al. showed a more detailed comparison between early- or late-onset patients and the age-matched controls. For the less toxic NK cells, CD32 (CD56+, CD16+, CD57–, and CD28–) clusters were significantly higher in the EOPD/LOPD group, while CD27 (CD56+, CD16+, CD57+, and CD28–) clusters were significantly higher only in the LOPD group. Cytotoxic CD29 (CD56+, CD16+, CD57+, and CD28+) clusters were lower in the LOPD group when compared to the control group, with no statistical difference between the groups. Furthermore, it was found that C27 cell clusters increased as the disease was prolonged in LOPD (Tian et al., 2022).

In addition, the disease severity may also be associated with the NK cells. Two studies explored the relationship between NK cells and disease progression measured by UDPRS and Mini-Mental State Examination (MMSE) scales. According to Niwa et al., the percentage of CD16+ cells and CD56+ cells was not correlated with the UPDRS (p > 0.1) and UPDRS-III scores (p > 0.1) in the 29 patients diagnosed with sporadic PD (Niwa et al., 2012). However, an increase in C32 clusters was found to correspond to increased UPDRS and UPDRS-III scores but decreased MMSE scores in the patients with PD (Tian et al., 2022).

In addition, other markers of NK cells were found to have an impact on the clinical manifestations. Using the bioinformatics tools, Anderson et al. analyzed the interaction between polymorphic killer immunoglobulin-like receptors (KIR) and human leukocyte antigen (HLA) class I ligands in patients with PD (Anderson et al., 2020). KIR3DL1, a member of the KIR family, conducts inhibitory signals depending on two immunoreceptor tyrosine-based inhibition motifs (ITIM) and binds to HLA-B molecules containing the Bw4 phenotype. The researchers found that high expression of KIR3DL1 alleles in combination with HLA-Bw4/Bw4i reduced gait difficulties and rigidity in patients, while weak KIR3DL1/HLA interactions were found to be associated with a higher risk of rigidity. Thus, they speculated that high expression levels of inhibitory KIR to reduce NK cell-mediated inflammation may reduce the severity of symptoms and have a protective effect in PD.

Evidence showed that NK cells were found in the brains of patients with PD (Earls et al., 2020), and the percentage of NK cells increased in the CNS parenchyma prior to the dopaminergic neuronal degeneration (Earls et al., 2019), which indicated infiltration of NK cells in the CNS during PD progression. From the point of ethics and experimental convenience, NK cells in the peripheral blood were appropriate and easy to obtain and could also give us information about pathogenesis. However, analysis of the cell counts and cell cytotoxicity had produced some contradictory results in both clinical studies and animal studies.

Most groups reported increased NK cells (Mihara et al., 2008; Niwa et al., 2012; Cen et al., 2017; Sun et al., 2019; Huang et al., 2021; Tian et al., 2022) in patients, while some researchers showed that the number of NK cells did not change significantly when compared to the corresponding controls (Stevens et al., 2012; Earls et al., 2019; Grembecka et al., 2021). We speculated that cohort/model differences, PD subtype, disease duration, or time delay between the clinical measurements of the disease and the pathological changes are the possible influencing factors, and the mechanisms underlying altered immune profiles still need to be explored.

Moreover, peripheral blood mononuclear cells (PBMCs) were shown as a basic reference in most studies. PBMCs are cell groups that include blood cells with a round nucleus (i.e., lymphocytes, monocytes, NK cells, or dendritic cells) (Sen et al., 2017). Among the PBMCs, components other than NK cells may also be altered in the pathological process of PD. For example, changes in monocyte subpopulations were observed in the early and late PD (Harms et al., 2021; Tian et al., 2022). Therefore, it might be better to measure the absolute value of the NK cells for comparison rather than considering only the percentage of cells.

As for the NKCC, Huang et al. confirmed enriched NK cell-mediated cytotoxic pathways in the peripheral blood using Gene Set Enrichment Analysis (GSEA), supporting a disease-associated cytotoxic response of NK cells in PD (Huang et al., 2021). Also, NK cells may be actively involved in the acute injury of dopaminergic neurons, since increased peripheral NKCC was found in rats injected with 6-OHDA for several weeks (Grembecka et al., 2021). However, some studies revealed no significant changes in NKCC targeting the K562 cells lacking ligands of the inhibitory NKG2A receptors (Mihara et al., 2008; Niwa et al., 2012). Given the fact that alterations in the expression levels of NKG2A receptors on NK cells were inconsistent between different studies (Mihara et al., 2008; Niwa et al., 2012), a possible approach involving the measurement of NKCC activity would be an influencing factor in the final results.

Notably, researchers for the first time observed that extracellular α-syn aggregates are internalized and degraded in the NK cells, and a more intriguing finding is that NKCC decreased obviously in a dose-dependent manner after the interaction between α-syn aggregates and NK cells (Earls et al., 2020). It seemed that the internalization of α-syn modulated the cytotoxicity of NK cells in a non-activated way. Recent studies have demonstrated that microglia can ingest and degrade extracellular α-syn with high efficiency, but consequently, α-syn fibrils are likely to induce apoptosis of microglia by increased reactive oxygen species (ROS) production and mitochondrial network disintegration (Choi et al., 2020; Scheiblich et al., 2021). In this case, whether α-syn exhibits toxicity in the NK cells and how α-syn aggregates affect the NKCC after the phagocytosis are still unclear.

In addition, researchers also observed increased CD16 expression on mature NK cells and speculated that an enhanced ADCC effect was primed in patients with PD (Konstantin Nissen et al., 2022). Since information about antibodies and target cells is unknown, cytotoxicity through ADCC is not measured in an intuitive way, and hence more number of experiments are needed to prove this hypothesis. Therefore, some doubts remain about the measurement of NKCC. To further elucidate these findings, mechanisms and further validations are encouraged to identify the cytotoxicity of NK cells more accurately and comprehensively in the dynamic development of PD in patients.

Previous studies demonstrated that older age at onset was associated with a more severe motor phenotype and higher H&Y stage and UPDRS-III score, and similar results were observed by Cen et al. and Tian et el. (Szewczyk-Krolikowski et al., 2014; Pagano et al., 2016). Possibly, it can be explained by a greater dopaminergic dysfunction in patients with older age at onset, but the mechanisms are still unclear. Evidence suggests that genetic variation plays an important role in determining the age of onset of PD (Nalls et al., 2015). GWAS has implicated that SNCA, TMEM175, and GBA are associated with PD age at onset and are also implicated in α-syn aggregation pathways (Blauwendraat et al., 2019), which means that accumulation and clearance of α-syn are likely to be involved in the pathogenesis of age at onset. Besides, differences in the immune status exist in the patients with early- and late-onset PD (Tian et al., 2022). As for NK cells, functions alter with the differential expression of molecules on the cell surface in the relevant development stages and can be affected by various immune responses (Yu et al., 2013; Abel et al., 2018). It is possible that NK cells are related to age at onset in patients with PD.

Tian et al. analyzed peripheral mature NK cells marked by CD16 and CD56, and further distinguished cell maturity by expression of CD57 (a marker of terminal maturation) and CD28 (critical for co-stimulation of the cells to proliferate and produce IFN-γ) (Tian et al., 2022). Cytotoxicity of cells marked by CD57+CD28+ (CD29 clusters) is higher than those marked by CD57+CD28- (CD27 clusters) and CD57-CD28- (CD32 clusters). Interestingly, the researchers found higher CD27 and CD32 cell clusters and lower CD29 clusters in LOPD, while only higher CD32 clusters were noticed in EOPD, which suggested that age at onset might be associated with the development and toxicity alterations of NK cells in PD. Furthermore, considering that aging is one of the main risk factors for PD (Pagano et al., 2016; Poewe et al., 2017), it is crucial to control the variables such as age-matched healthy controls in the study design involving comparison between EOPD and LOPD.

NK cells are essential effectors in innate immunity, since NK cells can infiltrate the CNS during PD progression (Earls et al., 2020). However, whether there are pathophysiological reflections on the clinical motor and non-motor symptoms is an important issue for us to understand the role of NK cells in PD. Tian et al. observed less toxic NK cells increased with more severe motor and cognitive dysfunctions based on the evaluation of UPDRS-III and MMSE scales (Tian et al., 2022). Anderson et al. found that high expression of inhibitory KIR on NK cells might protect against severe motor symptoms in patients (Anderson et al., 2020). Additionally, in the PFF α-syn model, increased motor symptoms, exacerbated motor deficits, and increased p-α-syn inclusions were observed due to NK cell depletion (Earls et al., 2020). Therefore, we speculated that NK cells carry out killing and phagocytosis in a less toxic or non-hyperactivated way, which tend to play a protective role in the progression of PD.

However, previous studies have demonstrated that NK cells likely act as a double-edged sword in central nervous system disorders (Poli et al., 2013, 2018). On the one hand, they can induce the activation of neural cell death (Garofalo et al., 2020), while on the other hand, they have the ability to suppress CNS inflammation (Hao et al., 2010). In addition, immune dysfunction is quite complex in the development of PD. Pro-inflammatory or anti-inflammatory cytokines and cells, such as microglia, monocytes, T cells, B cells, and so on, are undoubtedly important participants in the pathological process and correlate with disease severity and disability in patients with PD (reviewed in detail in: Qin et al., 2016; Chen et al., 2018; Sabatino et al., 2019; Harms et al., 2021). However, only a few studies have explored their interactions with NK cells in PD till now, so the detrimental effect of NK cells is still uncertain and needs further investigation.

Drugs are the influential factors in the study design. Basu et al. reviewed that dopamine and its receptor agonists altered the function of NK cells, for example, administration of dopamine increased the killing ability of specific NK cells (Basu and Dasgupta, 2000). Thus, data collection may be better in newly diagnosed PD patients to reduce the effects of dopamine replacement therapies. Additionally, neurotransmitters and neuroendocrine factors, such as glucocorticoids, serotonin, and epinephrine, have been reported to affect the activities of NK cells through a complex communication between the immune system and the nervous system (Capellino et al., 2020).