Long-Term Hyperglycemia Induces α-Synuclein Aggregation and Dopaminergic Neuronal Loss in Parkinson's Disease Mouse Model


 Background: Growing evidence suggest the association between Parkinson’s disease (PD) and diabetes mellitus (DM). On a cellular level, it was proven that long-term elevated levels of glucose might lead to nigrostriatal degeneration in PD models. However, the underlying mechanism is still unclear. Previously, we have elucidated the potential of type 2 diabetes mellitus (T2DM) in facilitating PD progression, involving aggregation of both alpha-synuclein (α-syn) and islet amyloid polypeptide (IAPP) in the pancreatic and brain tissues. However, due to the complicated effect of insulin resistance on PD onset, the actual mechanism of hyperglycemia-induced dopaminergic degeneration remains unknown. Methods: In the present study, we employed the type 1 diabetes mellitus (T1DM) model induced by streptozotocin (STZ) injection, to investigate the direct effect of elevated blood glucose level on nigrostriatal degeneration. Results: We found that STZ treatment induced more severe pathological alterations in the pancreatic islets and T1DM symptoms in α-syn-overexpression mice than that in wild type (WT) mice, one month and three months after STZ injections. Behavioral tests evaluating motor performance confirmed the nigrostriatal degeneration. Furthermore, we observed a marked decrease in dopaminergic profiles and an increase of α-syn accumulation and Serine 129 (S129) phosphorylation in STZ-treated α-syn mice compared with vehicle-treated mice. At last, we observed more severe neuroinflammation in the brain of the STZ-treated α-syn mice.Conclusion: Our results solidify the potential link between DM and PD, providing insights on how hyperglycemia induces nigrostriatal degeneration and contributes to pathogenetic mechanisms in PD.


Background
Parkinson's disease (PD) is the second most common neurodegenerative disease after Alzheimer's disease (AD) in human, presenting mainly motor discoordination accompanied by various non-motor symptoms, such as cognitive impairments [1]. Progressive loss of dopaminergic neurons in the substantia nigra pars compacta (SNpc) and the formation of cytoplasmic inclusions, called Lewy Bodies (LBs), are the two major neuropathological hallmarks of PD [2]. LBs are primarily composed of misfolded α-syn, which plays a crucial role in PD pathogenesis [3]. The etiology of PD is not yet clear. Most PD cases are sporadic, associated with different contributing risk factors such as aging and neurotoxins, etc. [4].
Available evidence suggests that certain metabolic diseases may also serve as risk factors for PD, such as diabetes mellitus (DM) [5]. DM is a group of complex metabolic diseases characterized by impaired glucose metabolism and subsequent hyperglycemia resulting from defects in insulin secretion and e ciency [6]. Based on the concepts of being "insulin-sensitive" and "insulin-insensitive" [7], DM can be divided into type 1 diabetes mellitus (T1DM) with insu cient insulin secretion due to progressive destruction of pancreatic β cells, and type 2 diabetes mellitus (T2DM) with the lack of appropriate insulin response [8][9][10][11].
PD and DM share similarities in multiple aspects, including epidemiology, etiology and pathogenesis.
First, epidemiologically, DM presents as a risk factor for PD. Both PD and DM are of multifactorial origins with high prevalence. Diabetes is among the most prevalent chronic diseases, and researchers predicted a near doubling of patients with DM by 2030 compared to gures obtained in 2000 [12]. PD is a disease with an increasing prevalence in the elderlies, and the projected prevalence cases for PD are predicted to increase during the next 40 years [13]. Accumulating evidence suggests that DM constitutes a risk factor associated with increased neurodegenerative diseases [5]. The elevated risk of developing cognitive abnormalities in individuals with impaired glucose metabolism is well documented [14,15].
Epidemiological studies indicate that patients with preexisting DM have an increased incidence of developing PD and often display parkinsonian symptoms [16][17][18][19][20]. Secondly, PD and DM have shared potential contributing factors and overlapping pathology. Environmental exposure, genetic susceptibility and lifestyle factors [21][22][23] are well-known PD and DM causal factors. Clinically, some diabetic patients exhibit pathologies related to striatal dopaminergic dysfunction [11,24]. Furthermore, DM and PD also share common etiopathogenic features [15,20], such as dopaminergic neuronal loss in SNpc and α-syn aggregation in the pancreatic β cells in diabetic patients [25].
Although a growing body of epidemiological and clinical data favor the association between DM and PD [26], to date, the molecular mechanisms by which elevated levels of blood glucose associates with the development of PD are still unclear. Experimental studies support the toxic role of hyperglycemia in the central nervous system (CNS) [27,28]. Our previous study has suggested the possible role of protein coaggregation during the interplay of the two diseases in a non-human primate model [29]. However, considering the in uence of insulin resistance in amyloid protein aggregation [30], the mechanisms of hyperglycemia induced-neurodegeneration remain obscure. Of all the shared mechanisms of DM and PD, neuroin ammation has unneglectable importance. First, diabetes mediates many of its effects through in ammation. T2DM can present oxidative stress and peripheral in ammation [31]. Diabetic patients often show mild chronic in ammation, re ected by the elevation of peripheral blood cytokines such as interleukin-6 (IL-6), tumor necrosis factor-alpha (TNF-α) and interleukin-1β (IL-1β) [32,33]. Secondly, systemic in ammatory status is associated with the interruption of blood-brain barrier (BBB) integrity.
Corrupted BBB may subsequently cause leukocyte in ltration, diffusion of cytokines and entry of toxins into the CNS [34,35], therefore trigger neuroin ammation and activate microglia [36]. At last, prolonged microglial activation is proven to be a major factor in driving dopaminergic degeneration in PD [37]. From DM to systematic in ammation, which thereafter induces neuroin ammation and then promotes the onset and progress of PD, a lot of evidence is shown to explain the mechanism of hyperglycemia-induced neurodegeneration. As a matter of fact, many studies highlight the importance of DM-induced neuroin ammation and its link to neurodegenerative diseases, such as AD [38].
Therefore, to study the effect of hyperglycemia on the nigrostriatal pathway and better understand the mechanistic connection between DM and PD, we have employed. the streptozotocin (STZ) induced-T1DM model in the background of both wild type (WT) and α-syn transgenic mice. After characterizing the T1DM phenotypes, we investigated PD related features in the mice regarding motor dysfunction, dopaminergic neuron and terminal loss, and α-syn aggregation. We found that intraperitoneal injection (i.p.) of STZ exacerbated degeneration of dopaminergic neurons in the α-syn-overexpression mice, which were accompanied by an increase in α-syn aggregation and phosphorylation. Moreover, we observed markedly increased neuroin ammation in the nigrostriatal systems of STZ-injected α-syn mice, suggesting the potential mechanistic role of neuroin ammation connecting hyperglycemia and the subsequent PD-related alternations.
Brie y, the transgenic mice express human WT full-length α-syn fused with green uorescent protein (GFP) under mouse α-syn gene promoter, generated by pronuclear inoculation of bacterial arti cial chromosome (BAC) to C57BL/6 background. Mice were housed (4 to 6 animals per cage) under standard laboratory conditions (a strict 12/12-h light-dark cycle with food and water supplied ad libitum) in a temperature-controlled (25°C) facility. All housing, breeding and animal experiment procedures were approved by the Research Ethics Committee of China Medical University and performed according to the international guidelines.

Induction of T1DM with STZ
Both BAC-α-syn-GFP mice and C57BL/6 mice were given low-dose STZ (Sigma-Aldrich S0130) for 5 consecutive days [42]. Food deprivation was performed 16 h prior to STZ treatment. STZ compound was dissolved in 0.1 mol/L sodium citrate buffer (pH 4.5) and injected i.p. with 70 mg/kg of body weight within 15 min of dissolution, once per day. The vehicle group received an equal volume of citrate buffer without STZ correspondingly.

Evaluation of blood glucose and body weight
Body weights and blood glucose were measured on day 0, 7, 14, 28, 49 and 84 after STZ injection. Blood was collected from tail veins for glucose determinations, using an electronic scale (YuWell, China) following fasting for 12 h (Fig. 1A). The mice were then re-fed. Animals with random blood glucose value >150 mg/dL (8.3 mmol/L) were de ned as successfully induced diabetic mice.

Behavioral tests
Pole Test. Pole test was used to assess the motor coordination and balance of mice before sacri ce. Based on previously published methods [43], a steel pipe with the length of 55 cm and diameter of 1 cm was tightly wrapped in white antiskid gauze and xed on plastic foam base. A spherical protrusion with the diameter of 2 cm was placed on top of the pipe as attachment points for mice. During the experiment, mice were put on the spherical protruding point with heads upward. The time of the heads to turn down (T-turn) and the total time (T-total) of mice to climb down the pole was recorded. Average value of 5 repeated tests (minimum 3-min interval) was used for statistical analyses.
Open Field Test. Open eld test was used to assess anxiety and spontaneous activity of mice as previously described [44]. Brie y, mice were placed in the test room for an hour to adapt to the surroundings. Mice were put in the center of an open eld (90-cm long, 90-cm wide, and 40-cm high, divided into 9 squares) and monitored for 15 min by an overhead digital camera connected to an automated video-tracking system (Harvard Apparatus, model: SMART® v 3.0). To avoid interference from the odors left by the last mouse, the inner wall and bottom surface of the eld was thoroughly wiped with 75% ethyl alcohol in between each measurement. The mean speed in the total zone, the total distance, distance in the center zone and distance in the periphery zone of mice were recorded.

Sample Collection and Processing
Mice were deeply anaesthetized with sodium pentobarbital (Apoteksbolaget, Sweden) and sacri ced by trans-cardinal perfusion with 4% paraformaldehyde (PFA). Brains were rapidly removed, kept in 4% PFA solution overnight and stored in 30% sucrose solution. 30-µm-thick coronal sections were cut and kept in anti-freeze medium (30% sucrose containing 20% ethylene glycol and 0.05% sodium azide in 0.1 M PBS) at -20°C. Pancreatic tissues were xed in 4% PFA at 4°C for 24 h. Following dehydration with an ascending gradient of ethanol and xylene, tissues were embedded in para n and sectioned into 5-µmthick slices encompassing the islets, using a rotary microtome (RM2016; Leica Microsystems GmbH). Fresh brain tissues were also collected and stored at -80°C for Western blot assessment.

Immunohistochemistry
After washing, antigen retrieval of the brain sections was performed in citrate acid buffer (pH 6.0) at 80°C for 30 min. Sections were then quenched with peroxidase solution (3% H2O2 in 10% methanol) for 15 min before blocking in 5% normal goat or house serum diluted with 0.5% TritonX-100-PBS for 1 h. Primary antibody incubation (Table S1) was performed at 4°C overnight followed by washing and incubation with biotinylated secondary antibodies. After 30-min ABC (Vector Laboratories, Cat. # VEPK-6100) incubation at room temperature, the sections were developed with 0.03% diaminobenzidine (DAB) in 0.05 M Tris-HCl (pH 7.5) for 2 min. The sections were then dehydrated and mounted with neutral balsam for microscopic analyses. Para n-embedded pancreatic sections were rst depara nized with xylene and decreasing concentrations of ethanol before subjected to antigen retrieval. The quanti cation of immunohistochemistry was performed using the Image J software (version 1.34.3.67; National Institutes of Health) analyses.

Western blot
Mouse brain substantia nigra (SN) tissues were fully immersed in ice-cold lysis buffer (Beyotime) containing 1% protease inhibitor cocktail and phenylmethylsulfonyl uoride (PMSF) (Beyotime) and homogenized by ultrasonic crashing. 10 µg of total protein was loaded from each sample. SDS-PAGE were performed for 75 min at 120 V followed by transfer onto polyvinylidene di uoride membranes (PVDF) (Millipore) for 120 min at 200 mA. After blocked in 5% skimmed milk (Millipore) in TBS with 0.05% Tween-20 (TBST), the membranes were incubated with primary antibodies overnight at 4°C and horseradish peroxidase (HRP) conjugated antibodies (ECL Rabbit IgG, #NA934; ECL Mouse IgG, #NA931; GE Life Sciences) at room temperature for 1 h. Detection was enhanced by ECL kits (#180-5001, Tanon) and developed with chemiluminescence imaging analysis system (Tanon 5500). Bands were analyzed using Image J software.

Quanti cation
Stereological cell counting. Unbiased quanti cation of the total number of tyrosine hydroxylase (TH) and serine 129 phosphorylated α-syn (pS129-α-syn) positive neurons in SN were calculated according to the optical fractionator principle [45], using a stereological system including 2-dimensional anatomical mapping and cell quanti cation program (Stereo Investigator, MBF Bioscience, USA) coupled to a color bright eld microscope (Nikon M 570E). A full series of sections (6 sections) per mouse was used to count. The virtual outlines of the right and left SN were drawn in a manner consistent with the description provided by previous article [46]. Stereo Investigator's Virtual Slice module was used to draw interesting counting regions at low magni cation. Starting at a random eld, the number of stained neurons were counted under high magni cation (100X objective).
The total numbers of TH-positive and pS129-α-syn-positive neurons in the SN were calculated according to the following equation: N= 1/SSF × 1/ASF × 1/TSF × Q − N stands for the number of the target neurons. SSF represents the section sampling fraction. ASF means the area of sampling fraction and HSF is the thickness of sampling fraction. TSF was calculated as dissector height/mean thickness and Q − was the number of neurons counted. For this investigation, SSF=6 and ASF=60%. The Gundersen coe cient of error (CE) was used to determine su ciency of the samples (CE < 0.1).
The counting point of a neuron is appointed where the nuclear membrane was in sharp relief while focusing up and down in the z-axis, when the estimated center of the nuclei was within the counting frame and contour lines simultaneously, without touching the counting-frame line or contour line in the xy-axis.
Quanti cation of islet β-cell area. The area of islet β-cell area was quanti ed with insulin staining. Five different islet structures in at least three mice in each group were included. The area measurements of the whole islet and insulin positive tissues was determined by the Image J software (version 1.34.3.67; National Institutes of Health) analyses. The average ratio of insulin positive area in the whole islet (insulin+/islet area%) was employed to evaluate the damage of pancreatic tissues.
Quanti cation of microglia. For the microglia quanti cation, eight optical elds of high magni cation (40X objective) were taken from each SN. The number of microglia was counted using the manual cell counting marker function of the Image J software. Morphological grading of microglia was performed using a standard published previously [47]. Brie y, microglia were graded based on the length and thickness of their processes, the characteristics of their cell body and the look of the nucleus, into four stages. The uorescence co-localization between the major histocompatibility complex Class II (MHC class II) and ionized calcium binding adapter molecule 1 (Iba-1) in the striatum were quanti ed using the Pearson's Correction with the Fiji plugin Coloc 2 [40].

Statistical analysis
Statistical analysis was performed with GraphPad Prism 8 software. Three-way analysis of variance (ANOVA) and two-way ANOVA with Dunnett-t's multiple and Tukey's multiple comparisons were employed. All values were presented as the mean ± standard error of the mean (SEM). P values and statistical methods are described in gure legends.

Body weight loss in STZ-injected mice
To evaluate the effects of STZ in inducing hyperglycemia in mice, we routinely monitored the body weight and blood glucose level. 40% (8 of 40) of the BAC-α-syn-GFP mice died due to the STZ treatments, among which three mice died within the rst month after injection, and ve mice died during the following two months. In contrast, only one mouse died after vehicle injection (Fig. 1B). All STZ-injected mice developed signi cant loss of body weight during the rst two weeks post-injection. Although there was an increasing trend in all groups during the following months, the body weights of the mice with i.p. STZ (WT+STZ, α-syn+STZ) were signi cantly lower than those with vehicle injection (WT-vehicle, α-syn-vehicle) (Fig. 1C). The signi cantly lower body weight of STZ injected groups suggested a disrupted metabolic status.

STZ treatment induced more severe hyperglycemia in α-syn-overexpression mice compared to WT ones
In the STZ treated groups, blood glucose levels increased in both WT and BAC-α-syn-GFP mice from the rst week after injection and remained at high levels (Fig. 1D). In the α-syn+STZ group, the blood glucose levels ranged from 16.3-21.0 mmol/L, while the WT+STZ group exhibited blood glucose of 10.4 to 15.28 mmol/L. Interestingly, the blood glucose level of the α-syn+STZ group was signi cantly higher than WT+STZ in the whole experimental period (three months) (Fig. 1D). The data indicated that we successfully created the T1DM diabetic mouse model and that STZ induced more severe hyperglycemia in the α-syn-overexpression mice than in the WT ones. On the other hand, the blood glucose level remained stable in all vehicle-treated groups, with WT-vehicle mice showing 7.3-8.7 mmol/L and α-synvehicle mice showing 8.3-9.3 mmol/L.
3.3. STZ treatment exacerbated more severe structural damage of islets in α-syn-overexpression mice than in WT mice A signi cant decrease in the insulin-positive pro le was observed in STZ-injected mouse pancreases, with the average ratio between insulin-positive area and the total islet area decreased by 31.42% and 21.7% in WT mice and 39.94% and 37.89% in BAC-α-syn-GFP mice, one and three months after injection, respectively ( Fig. 2A-B). Neither WT nor BAC-α-syn-GFP injected with vehicle showed decreased insulin immunoreactivity. Notably, the ratio between insulin-positive area and the islet area in pancreases of α-syn+STZ mice reduced signi cantly more than that of WT+STZ mice by 21.65% at three months postinjection (Fig. 2B), indicating a more severe pancreatic β cell mass loss in α-syn mice. The pathological alterations in the pancreases suggested that STZ-treated mice exhibit features reminiscent of T1DM[48]. More importantly, under the same conditions, the more severe loss of islet β-cells occurs in α-syn mice.

Long-term hyperglycemia-induced motor de cits in αsyn-overexpression mice
We performed the pole test and open eld test to assess the motor performance of the STZ-injected mice to study whether and to which extent nigrostriatal degeneration might have taken place. In the pole test, both T-turn and T-total of mice in the α-syn+STZ group were signi cantly longer than those of the WT+STZ and vehicle-injected groups, three months after STZ injections, suggesting that hyperglycemia aggravate more dramatic motor de cits in the α-syn mice than that in the vehicle-treated mice. Moreover, it seemed that the appearance and severity of altered behaviors are associated with the length of hyperglycemia. One-month long hyperglycemia appeared not su cient to initiate the PD like behavioral phenotypes in WT mice (Fig. 3A-B).
We also measured the spontaneous locomotor and exploratory activity of the mice in the open eld test. At one month after injection, BAC-α-syn-GFP mice exhibited no signi cant alterations in spontaneous movement, regardless of STZ injection or not. However, we found that the mean movement speed in α-syn+STZ mice was noticeably slower than in WT-vehicle mice three months after injection (Fig. 3C). Meanwhile, the total traveled distance (including the distance in the center and periphery) was reduced compared to WT-Vehicle mice (Fig. 3D-F). Furthermore, we found that the α-syn+STZ mice traveled a shorter distance than WT+STZ mice in the central area of the open eld at three months after injection (Fig. 3E), indicating that hyperglycemia increase the susceptibility of BAC-α-syn-GFP mice to develop motor impairment (more signi cant at three months after injection), which might be attributable to dopaminergic damage.

Long-term hyperglycemia-induced dopaminergic neuronal degeneration in α-syn-overexpression mice
To investigate whether hyperglycemia has any impacts on the nigrostriatal dopaminergic system, we speci cally examined the expression of dopaminergic neuronal markers, tyrosine hydroxylase (TH), the rate-limiting enzyme of catecholamine synthesis, and dopamine transporter (DAT). We observed signi cant reductions of TH and DAT in the SN of α-syn+STZ mice three months post-injection, suggesting the degeneration of dopaminergic neurons (Fig. 4A-C). No difference in the expression levels of either TH or DAT in the SN of STZ-injected mice compared to vehicle-injected groups at one month after injection was observed (Fig. 4A-C). The results were further con rmed by immunohistochemical and stereological analyses, focusing on discrete dopaminergic neurons in the ventral mesencephalon, including the SNpc and the ventral tegmental area (VTA) (Fig. S1, Fig. 4D). Mice within the α-syn+STZ group displayed a signi cant loss of dopaminergic neurons in the SNpc compared to WT+STZ and α-synvehicle groups three months after induction of hyperglycemia (Fig. 4D-E). Stereological analyses also showed time-dependent neurodegeneration in α-syn+STZ groups comparing one month and three months post-injection (Fig. 4F). Furthermore, in the striatum, the decreased expression levels of TH and DAT were also observed in STZ-treated mice three months after injection (Fig. S2). The results were consistent with what we obtained in SN, indicating that long-term hyperglycemia causes damage to both dopaminergic neurons and terminals in α-syn mice.

Hyperglycemia increased α-syn phosphorylation in the nigrostriatal system
As a consequence of α-syn overexpression, the BAC-α-syn-GFP mice develop α-syn phosphorylation and aggregation in an age-dependent manner [39]. To determine whether hyperglycemia can worsen the α-syn pathology, we studied α-syn aggregation and phosphorylation in the STZ-injected mouse brains. Overall, we found signi cantly higher α-syn expression and phosphorylation level [49] in the SN of the BAC-α-syn-GFP mice than WT mice in both STZ-and vehicle-injected groups. Moreover, α-syn level and phosphorylation increased in α-syn-overexpression mice along time after induction of hyperglycemia, suggesting a link between hyperglycemia and the severity of α-syn pathology (Fig. 5A-C).
Immunohistochemical analyses also showed that phosphorylated α-syn positive signals were only present in the BAC-α-syn-GFP SN, while the WT mice were lack of positive signals (Fig. 5E). Notably, in stereological quanti cation, we observed two different types of pS129-α-syn positive pro les (Fig. 5D), one with only pS129-α-syn positive immunoreactivity but lack of visible inclusions and the other containing pS129-α-syn-positive inclusions. The proportion of neurons containing pS129-α-syn-positive inclusions signi cantly increased three months after injection in α-syn+STZ mice compared to the α-synvehicle ones (Fig. 5F-H). These results indicated that hyperglycemia exacerbates α-syn aggregation and phosphorylation.

Hyperglycemia-induced severe neuroin ammation in αsyn-overexpression mice
To investigate the potential mechanism contributing to the pathological interaction between DM and PD, we examined mouse brain tissues with microglial (Iba-1) and astroglial (glial brillary acidic protein, GFAP) markers.
By immunoblotting (Fig. 6A), we observed a signi cant increase in Iba-1 level in the SN of α-syn+STZ mice compared to the WT-vehicle mice (Fig. 6B), indicating potential microglial activation.
Immunohistochemical analyses showed a signi cant increase of Iba-1 positive cells in α-syn+STZ groups (both one and three months after injection) in SN (Fig. 6D-E), indicating the hyperglycemia in α-syn mice induced severe neuroin ammation. No difference in Iba-1 immunoreactivity was observed in the SN in vehicle-injected groups. Morphological alterations of microglia were further analyzed. The activation degrees of microglia were divided into four stages, including I) microglia presenting small round cell body with long thin processes, II) cells with a dense soma and processes remaining thin but longer, III) cell bodies appearing enlarged and irregular in shape, with processes becoming thicker and shorter, IV) cell bodies becoming swollen and dark, merging with thickened and branched processes [47,50] (Fig. 6F). Quantitative analyses of microglial morphology in the SN region con rmed hyperglycemia-induced microgliosis in the STZ injected BAC-α-syn-GFP mice. Stage IV cells (Stage IV were deemed highly activated microglia, showing amoeba-like morphology) constituted over 35% and 37% of the microglia population at one and three months after injection in the α-syn+STZ group, signi cantly higher than that of α-syn-vehicle groups (22% and 25%) at the same time point (Fig. 6G). We observed no difference in the expression levels of GFAP in the SN of STZ-injected mice compared to vehicle-injected groups at both one and three months after hyperglycemia induction (Fig. 6C).
We further examined the functional alterations of activated microglia. Microglia expressing MHC class II in the striatum were signi cantly increased in the α-syn+STZ group (Fig. 7A). In addition, the colocalization of MHC class II and Iba-1 was more robust in the striatum in the α-syn+STZ group than the αsyn-vehicle and WT groups (Fig. 7B, Fig. S3). The increased expression of MHC class II in microglia indicated an activated status, wherein α-syn-overexpression mice with hyperglycemia exhibited evident neuroin ammation.

Discussion
Although the exact etiology of PD is still to be uncovered, there exist many risk factors facilitating the onset and progress of the disease. One major risk factor of PD is the impairment of glucose metabolism [26], which may occur years before the rst motor symptoms appear, suggesting the unneglectable role of hyperglycemia in initiating PD. In this study, we used a direct hyperglycemic model by inducing pancreatic islet damage using STZ to observe whether and how the elevated blood glucose affects the degeneration process of a PD (α-syn overexpression) mouse model. We treated 3-month-old BAC-α-syn-GFP mice (an α-syn transgenic mouse model), providing experimental reference value for the early diagnostic and preventing strategies of PD. 4.1. T2DM vs. T1DM-models to study the direct effect of hyperglycemia on the progression of PD Despite some discrepancies, there have been a number of previously published papers suggesting the association between DM and PD [17,20,51]. Among these reports, studies on the relationship between T2DM and PD account for the majority. Indeed, the incidence rate of T2DM is much higher than that of T1DM and it shares similarities with PD in the aspects of chronic disease progress and dominant ratio in the elderlies [11,52]. Nevertheless, insulin resistance in T2DM affects the regulation of dopaminergic transmission and maintenance of synapses in the CNS [53]. The brains of PD patients with T2DM exhibited a process analogous to peripheral insulin resistance, suggesting potential common molecular pathways. In animal models, both high-fat diets treated and leptin receptor homozygous de ciency (db/db) mice exhibited increased vulnerability to nigrostriatal neurodegeneration [51], indicating the relation between T2DM and PD again. Our previous studies on T2DM cynomolgus monkeys also indicated that T2DM might facilitate PD onset and progress by interfering with the pathological protein aggregation [29].
However, T2DM itself has complicated etiology and pathology. Both insulin insu ciency and de ciency can occur at the same time [54]. In addition, insulin resistance on its own is challenging to elucidate [53].
Therefore, it would have been rather narrow to own the T2DM induced PD disease progression to sole hyperglycemia. Despite hyperglycemia, it has been widely proven that T2DM is often accompanied by aberrations in lipid metabolism and other related metabolic abnormalities, making the disease background hard to elucidate. Diabetic dyslipidemia is often present among patients with T2DM (72-85%), while patients with T1DM usually do not have such dyslipidemia (<10%) [55,56]. Simultaneously, such aberrations of lipid metabolism can also be observed in T2DM animal models [29,57]. Both hyperglycemia and dyslipidemia are associated with PD. Therefore, the contribution of hyperglycemia in a T2DM background remains unclear in inducing neurodegeneration. In summary, T2DM is possibly related mainly to PD, and a T2DM model may be substantially suitable for studying metabolic disturbance as an early risk factor for PD. However, investigation of a direct effect of hyperglycemia on nigrostriatal degeneration would demand a straightforward model harvesting mainly elevated blood glucose levels in a relatively short time.
Compared to T2DM models, the T1DM model with direct damage to insulin secretion certainly possesses a purer background. Although studies focusing on the effect of hyperglycemia caused by T1DM on neurodegenerative diseases are very limited [51,58], cognitive impairments and structural alterations in the brain are common in adults with T1DM [59], suggesting the close relation between T1DM and neurodegeneration. In a prospective study, patients with T1DM showed a substantial reduction in limb exibility and spontaneous movements, associated with brain dysfunction [60]. The direct damage of insulin-producing β-cell provides T1DM with hyperglycemia as its main metabolic pathology [61]. The effects of hyperglycemia-induced brain function de cits come from an immediate result of the increased glucose levels [62]. Therefore, in our work, we chose the T1DM model to emphasize the direct effect of hyperglycemia on nigral neuron and PD protein pathology.

The mechanisms of hyperglycemia-induced PD progression
In the present study, we have proposed neuroin ammation as the major mechanistic process from hyperglycemia to PD, linking T1DM and PD on a cellular level. In T1DM, how does hyperglycemia cause neuroin ammation? High blood glucose levels can saturate mitochondrial respiration in endothelial cells, astrocytes and pericytes, which in turn promotes reactive oxygen species (ROS) production and oxidative stress[63, 64]. When a high glucose level is maintained for a long time, enhanced ROS and in ammatory cytokines may interrupt the integrity of BBB [33]. The integrity of the BBB is essential to maintain the brain environment homeostasis by limiting the pass-through of peripheral immune cells and toxins space [65].
Upon the penetration of peripheral proin ammatory factors, microglia are activated, and neuroin ammation occurs. Neuroin ammation has been commonly recognized as a crucial feature of PD pathogenesis [66]. Therefore, a chain of events as described above constitute the whole mechanistic process of hyperglycemia-induced neuronal damages. Although in our study, we chose only neuroin ammation, which is only one point on the mechanistic chain, to elucidate the molecular and cellular events from T1DM to PD, the underlying cellular changes such as increased oxidative stress, overproduction of ROS, mitochondrial dysfunction and leakage on the BBB are all present. Further researches on the exact point of action of high levels of blood glucose can be very informative.

Hyperglycemia-potential risk factor but not direct initiator in PD pathogenesis
In our study, we did not observe α-syn aggregation in diabetic WT mice. This is consistent with the behavioral results and the TH downregulation. STZ-induced T1DM model we utilized is an acute islet injury model rather than a slowly progressive diabetic model, making it less likely to recapitulate the slow, progressive aggregation of α-syn in WT mice. Hyperglycemia on its own fails to induce the protein pathology of PD. Previous studies have shown T2DM triggers α-syn accumulation, aggregation and phosphorylation in pancreatic β cells and the brain[28, 29,67]. Impaired insulin signaling in T2DM causes neuronal insulin resistance in patients, which promotes α-syn accumulations [37,68]. In the T2DM background, the joint effects of multiple metabolic de ciencies, including hyperglycemia and insulin resistance, lead to pathological protein changes. Hyperglycemia alone, at least in the short term, is not able to initiate neurodegeneration. However, hyperglycemia still serves as a risk factor to PD from the protein pathology point of view. In our PD mice, hyperglycemia can accelerate the already existing PD pathology. α-Syn accumulation during ageing showed more severe aggregation and phosphorylation after STZ injection than the vehicle groups.

Conclusion
In conclusion, our nding showed that hyperglycemia aggravates PD progression. STZ-induced T1DM mice exhibited signi cant depletion of dopaminergic pro les, increased α-syn accumulation and phosphorylation in the nigrostriatal system and de ciency in motor performances. In addition, we observed severe neuroin ammation in the α-syn+STZ mice and concluded neuroin ammation as one of the mechanisms contributing to hyperglycemia-induced nigrostriatal degeneration. Overall, our study bridges metabolic disorders and neurodegeneration from a cellular mechanism point of view, providing referential value to investigating the etiology of PD and age-related metabolic and neuronal disorders.