NRF2 Loss Accentuates Parkinsonian Pathology and Behavioral Dysfunction in Human α-Synuclein Overexpressing Mice

Nuclear factor (erythroid-derived 2)-like 2 (NRF2) is a central regulator of cellular stress responses and its transcriptional activation promotes multiple cellular defense and survival mechanisms. The loss of NRF2 has been shown to increase oxidative and proteotoxic stress, two key pathological features of neurodegenerative disorders such as Parkinson’s disease (PD). Moreover, compromised redox homeostasis and protein quality control can cause the accumulation of pathogenic proteins, including alpha-synuclein (α-Syn) which plays a key role in PD. However, despite this link, the precise mechanisms by which NRF2 may regulate PD pathology is not clear. In this study, we generated a humanized mouse model to study the importance of NRF2 in the context of α-Syn-driven neuropathology in PD. Specifically, we developed NRF2 knockout and wild-type mice that overexpress human α-Syn (hα-Syn+/Nrf2-/- and hα-Syn+/Nrf2+/+ respectively) and tested changes in their behavior through nest building, challenging beam, and open field tests at three months of age. Cellular and molecular alterations in α-Syn, including phosphorylation and subsequent oligomerization, as well as changes in oxidative stress, inflammation, and autophagy were also assessed across multiple brain regions. It was observed that although monomeric α-Syn levels did not change, compared to their wild-type counterparts, hα-Syn+/Nrf2-/- mice exhibited increased phosphorylation and oligomerization of α-Syn. This was associated with a loss of tyrosine hydroxylase expressing dopaminergic neurons in the substantia nigra, and more pronounced behavioral deficits reminiscent of early-stage PD, in the hα-Syn+/Nrf2-/- mice. Furthermore, hα-Syn+/Nrf2-/- mice showed significantly amplified oxidative stress, greater expression of inflammatory markers, and signs of increased autophagic burden, especially in the midbrain, striatum and cortical brain regions. These results support an important role for NRF2, early in PD progression. More broadly, it indicates NRF2 biology as fundamental to PD pathogenesis and suggests that targeting NRF2 activation may delay the onset and progression of PD.

specific modifications to the KEAP1-NRF2 complex occur, that allow NRF2 to stabilize and subsequently migrate to the nucleus to stimulate multiple cellular defense genes. In fact, NRF2 can regulate the expression of a vast number of genes bearing an antioxidant response element (ARE), and activate targets involved in nearly every facet of cell signaling and metabolism, including redox regulation, proteostasis, DNA repair, prevention of apoptosis, drug/xenobiotic metabolism, iron homeostasis, lipid and carbohydrate catabolism, and mitochondrial function [3]. Based on its control of a host of transcriptional responses, loss of proper NRF2 function can result in maladaptive effects across several pathological contexts [6]. For instance, the decreased expression of NRF2 is associated with many of the key pathogenic hallmarks of aging and age-related pathologies, including increased oxidative stress and inflammation, disruption of proteostasis, genomic instability, and mitochondrial dysfunction, among others [7].
Alterations in NRF2 signaling have been linked to neurodegeneration in Parkinson's disease (PD) [8][9][10]. PD is the second most common age-related neurodegenerative disorder and is characterized by chronic progressive motor deficits and neurocognitive decline [11]. Neuropathologically, the pathognomonic features of PD include the loss of dopaminergic (DA) neurons in the substantia nigra pars compacta (SNpc), and Lewy pathology, which is the abnormal aggregation of the protein alpha-synuclein (α-Syn) in degenerating neurons [12][13][14]. In its native form α-Syn has diverse neural functions, however when misfolded, it forms intracellular inclusions within the cell body (Lewy bodies) and processes (Lewy neurites) which cause cellular toxicity [15,16]. This abnormal α-Syn accumulation is associated with changes such as oxidative stress, inflammation and altered proteostasis, which are other key aspects of PD [17][18][19][20][21]. In this regard, it has been shown that NRF2 mitigates neurodegenerative phenotypes in cellular and animal models of PD, including acute virally induced α-Syn models [22][23][24]. Conversely, acute α-Syn-mediated PD pathology is heightened in the absence of NRF2 [25]. Interestingly, it has also been noted that NRF2 is primarily cytosolic in nigral DA neurons of healthy individuals as compared to age-matched PD patients where it is found in the nucleus [10]. Moreover, a number of NRF2 target genes, including NADPH quinone oxidoreductase 1 (NQO1) and heme oxygenase-1 (HMOX1), have been shown to be elevated in PD brains [8,[26][27][28]. These studies support the notion that NRF2 status can dictate the pathogenic effects of α-Syn and influence PD pathogenesis. Nevertheless, the precise role of NRF2 loss in endogenous development of α-Syn pathology in PD, and its associated cellular, molecular, and behavioral effects remains elusive.
Here, we use a genetic approach to address this question. Specifically, we generated novel mice by cross breeding previously established Nrf2 knockout (Nrf2 -/-) or wild-type (Nrf2 +/+ ) mice [29], with mice overexpressing full length human wild-type α-Syn under the Thy-1 promoter (Thy1-α-Syn mice) [30]. The Thy1-α-Syn mice recapitulate many progressive features of sporadic PD seen in patients with the disease [30]. Using the newly established hα-Syn/Nrf2 mouse model, we comprehensively examined the involvement of NRF2 in the development of PD pathology and behavioral deficits. We found that NRF2 loss significantly worsens sensorimotor function, increases α-Syn aggregation, oxidative stress, inflammation, and autophagic burden, and induces DA neuron degeneration, as early as 3 mos of age. To our knowledge, these data are the first to explicate NRF2's involvement in promoting α-Syn pathology and its functional effects in a translationally relevant representation of PD onset and progression.

Animals
Hemizygous human Thy1-α-Syn mice (hα-Syn + ; C57Bl6/DBA background; founder breeder pairs were originally provided by Dr Marie-Francoise Chesselet, UCLA) [30], Nrf2 knockout (Nrf2 -/-) [29], and wild-type (Nrf2 +/+ ) mice were maintained in the University of Arizona Animal Care Facility. Mice were handled according to the rules and regulations of the NIH and Institutional Guidelines on the Care and Use of Laboratory Animals. All experimental protocols were approved by the University of Arizona Institutional Animal Care and Use Committee. The mice were housed under a reverse 12-hour light-dark cycle condition with food and water available ad libitum. The hemizygous human Thy1-α-Syn mice were crossed with Nrf2 -/mice to generate hα-Syn -/Nrf2 +/and hα-Syn + /Nrf2 +/mice. Heterozygous hα-Syn + /Nrf2 +/were subsequently cross bred to generate four genotypes: hα-Syn -/Nrf2 +/+ , hα-Syn -/Nrf2 -/-, hα-Syn + /Nrf2 +/+ and hα-Syn + /Nrf2 -/-. The breeding strategy is depicted in Figure 1. Male littermates were used in the study, and the genotypes of all mice were verified with PCR analysis of tail DNA.

Experimental Design
All four genotypes of mice produced were maintained until 3 mos of age at which point they were subjected to behavioral tests. Subsequently, animals were sacrificed at ~3.5 mos of age and brains processed for histological and Aging and Disease • Volume 12, Number 4, July 2021 966 molecular studies. The number of animals per group were as follows: hα-Syn -/Nrf2 +/+ (n = 16), hα-Syn -/Nrf2 -/-(n = 14), hα-Syn + /Nrf2 +/+ (n = 17) and hα-Syn + /Nrf2 -/-(n = 16). For histological studies, mice were sacrificed using sodium pentobarbital (60 mg/kg) and perfusion with 4% paraformaldehyde (PFA). Subsequently, brains were extracted and post-fixed in 4% PFA, sunk through a 30% sucrose solution, and sectioned (30 μm) in the coronal plane on a freezing sliding microtome. For molecular studies, animals were sacrificed using sodium pentobarbital (60 mg/kg), brains extracted and subjected to microdissection on ice to isolate tissues from the midbrain, striatum, cortex and hippocampus. The dissected tissues were then snap frozen in liquid nitrogen for later molecular analyses.

Behavioral analyses
Nest Building Task -Nest building is a natural motor behavior requiring the use of orofacial and forelimb movements that can be used to assess PD relevant sensorimotor function in rodents [31]. This task tests the ability of the animal to retrieve and use woven cotton pads (nestlets) placed in their cage's feeding bin. Once the nestlets are grasped and placed into the cage, the animals pull the nesting material apart with their forelimbs and teeth, breakdown the cotton in their mouths, and incorporate it into their bedding. Thus forelimb dexterity and fine motor behavior can be measured with this task. Briefly, mice are transferred to individual testing cages approximately 1 hr before the dark phase and weighed nestlets are placed in the feeder bin of the cage (5 nestlets, ~12g/cage). The amount of nestlets that were retrieved (nestlet pulldown) and broken down (nestlet usage) for nest building was measured every 12 hours over a 72 hour period (12,24,36,48,60 and 72 hours) by weighing the unused material in the feeder and inside the cage [31]. Data were presented as percent of nestlet pulldown and percent used as normalized to control (hα-Syn -/Nrf2 +/+ ).
Challenging Beam Task -Mice were subjected to a challenging beam task to test motor coordination and agility [31,32]. A wooden beam consisting of four continuous sections (25 cm in length each, 1m total length), which incrementally narrows from a starting width of 3.5 cm to an ending width of 0.5 cm, was used. The widest segment of the beam served as a loading platform for the animals and the narrowest end connected into the home cage. Mice received 2 days of training before testing. During the training phase, animals were brought to the behavior room at the start of their dark cycle and acclimated to the testing room for ~2 hrs each day. The animal was placed on the beam at its widest point and trained to move across the beam towards the narrowest part and into the home cage. Testing started on day 3 when mice underwent 5 unassisted trials on the beam. To increase difficulty, a mesh grid (1 cm 2 grid) of corresponding width was placed over the beam surface leaving an ∼1 cm space between the grid and the beam surface to allow adequate visualization of the mice's paws as they walked. Each trial was videotaped, then viewed and rated in slow motion for foot slip errors, number of steps made by each animal to cross the beam, and time taken to traverse the beam, by an investigator blind to the mouse genotype. A foot slip error was counted when, during a forward movement, a limb (forelimb or hindlimb) slid through the grid and was visible between the grid and the beam surface. By scoring each limb slip individually, the severity of the error could be measured. The trial ended when an animal fell off the beam, reached the maximum allowed time (60 sec), or traversed the full distance [33]. Errors per step, time to traverse, and number of steps were calculated as the average across all five trials.
Open field task -The open field test was used to determine spontaneous exploratory activity of the mice [34].  [35].

Western Blotting
Brain tissue from the four specified regions was homogenized using the TissueLyser II (QIAGEN, Germantown, MD) in 10 vol. (w/v) of cold RIPA buffer (25 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS) supplemented with 1 mM phenylmethylsulfonyl fluoride (PMSF) and a phosphatase inhibitor cocktail (PIC). Tissues were then sonicated for 10 s to generate total cell lysates. For isolating soluble and insoluble protein fractions, the total cell lysates were divided into Triton X-100 soluble and insoluble fractions by adding Triton X-100 (final concentration 1%) and incubating for 30 min on ice followed by centrifugation (15,000×g, 60 min, 4°C). The supernatant was designated as the Triton X-100 soluble fraction. The pellet was dissolved in lysis buffer containing 2% SDS and sonicated for 10 sec. This Aging and Disease • Volume 12, Number 4, July 2021 967 fraction was designated to be Triton X-100 insoluble. The protein concentration in the different lysates was determined via the bicinchoninic acid method (BCA, Thermo/Pierce, Waltham, MA). Lysates were then boiled, sonicated, and resolved by SDS-PAGE, and membranes were subjected to appropriate antibodies at 4°C for overnight. Then membranes were incubated with anti-mouse or anti-rabbit horseradish peroxidase (HRP) conjugated secondary antibodies (1:3000, Sigma Aldrich, St. Louis, MO) for 1 hr. All immunoblot images were taken using the Azure Biosystems c600 (Azure cSeries Advanced Imaging Systems, Dublin, CA). Relative densitometry analysis of western blots was performed using the ImageJ Program (National Institutes of Health, http://rsb.info.nih.gov/ij). For the densitometric quantification of α-Syn oligomers, all oligomeric bands were quantified and normalized to monomeric α-Syn.

Stereology and Quantification
Stereology -Stereological probes were applied using a Zeiss Imager M2 microscope (Carl Zeiss, Jena, Germany) equipped with StereoInvestigator software (v2019.1.3; MBF Bioscience, VT, USA) according to previously published methods [37,38]. Using the optical fractionator probe, Tyrosine Hydroxylase (TH) positive cells were counted in sections 480 μm apart using a grid size of 170 X 100 μm and counting frame size of 75 X 75 μm. Contours were drawn around the region of interest at 10X magnification, and cells were counted under a 63X oil immersion objective. Guard zones were set at 2 μm each at the top and bottom of the section, and the counting frame was lowered at 1-2 μm interludes and each cell in focus was marked. The Gundersen method for calculating the coefficient of error was used to estimate the accuracy of the optical fractionator results. Coefficients obtained were generally less than 0.1.
Other cellular quantifications -For the quantification of TH immunoreactivity in the striatum, 10x images were captured and the optical density (OD) of both hemispheres was measured using Image J software (NIH Image, Bethesda, MD) in 30 μm thick coronal sections (1:12 series, 4 sections per animal, n=6 animals/group). Quantification of Iba1 immunoreactivity was conducted on 40x confocal images (30 μm thick coronal sections, 1:6 series, 4 sections per animal, n=3 animals/group). The mean Iba1 fluorescence intensity was measured by Image J. The number of TH + /phospho-Syn + cells were counted in 40x images (30 μm thick coronal sections, 1:6 series, 4 sections per animal, n=3 animals/group). Data was expressed as mean ± SEM of the total number of cells obtained across the sections counted in each experimental group.

Electron Paramagnetic Resonance (EPR)
EPR was performed as described previously [39]. Briefly, brain tissues were dissected (midbrain, striatum, hippocampus and cortex), and incubated with spin trap in the presence of metal chelators (200 µM cyclic hydroxylamine 1-hydroxy-. 3-methoxycarbonyl-2,2,5,5tetramethylpyrrolidine [CMH], 25 µM deferoxamine [DF], and 5 µM diethyldithiocarbamate [DETC] in filtered 20 mM KREBS-HEPES buffer) (Noxygen) for 30 min prior to measurement. Buffer was then collected, and changes in CMH oxidation were measured for 15 min using the e-scanM Multipurpose Bench-top EPR system (Noxygen Science and Transfer Diagnostics GmbH). The relative production of reactive oxygen species was represented as the nanomolar concentration of oxidized spin trap divided by the time of trap incubation normalized to the total milligrams of tissue, then data were normalized to control.

Transmission Electron Microscopy
Brains were extracted and four regions, namely the midbrain, striatum, hippocampus and cortex, were microdissected. Tissue slices for transmission (TEM) electron microscopy were fixed in 2.5% glutaraldehyde + 2% PFA in 0.1 M piperazine-N,N′-bis (2-ethanesulfonic acid) or PIPES buffer (pH 7.4) for 1 hr at RT or at 4 o C overnight. Samples were post-fixed in 1% osmium tetroxide in PIPES buffer for 1 hr following a wash in 0.05 M PIPES buffer + 0.05 M Glycine and 2 X 10 min washes in 0.1 M PIPES. Following two more 10 min washes in deionized water (DIW), samples for TEM were block stained in 2% aqueous uranyl acetate, washed in DIW, dehydrated through a graded series of alcohols, infiltrated with 1:1 alcohol and Spurr's resin overnight, and then embedded in 100% Spurr's resin overnight at 60°C. Sections were viewed using a FEI Tecnai Spirit electron microscope (FEI Company, Hillsboro, OR) operated at 100 kV. 8-bit TIFF images were captured through an AMT 4 Mpixel camera. Morphometric measurements were conducted in digital images using Image J (NIH) software. Number of autophagic vacuoles per cell profile (16-20 cell profiles) across a minimum of 10 micrographs per animal were counted using standard criteria [20,40]. Vesicles were classified as autophagic vacuoles (autophagosomes or autolysosomes) if they met two or more of the following criteria: for autophagosomes they should have a completely or partially visible double membrane, absence of ribosomes in the outer membrane, identifiable organelles or parts of organelles in their lumen, and luminal density comparable to the surrounding cytosol; for autolysosomes they should have similar size but with less than 40% of membrane visible as double, cargo in the lumen is not identifiable as particular organelles and/or amorphous material, and luminal density below that of the surrounding cytosol.

Light and Confocal Microscopy
A Zeiss M2 Imager microscope (Carl Zeiss, Jena, Germany) connected to an AxioCam Mrc digital camera was used for brightfield microscopy (α-Syn, phospho-α-Syn and TH). Images were collected through a 63X lens for the α-Syn and phospho-α-Syn images, and a 20X lens for the TH images, using Microlucida software (v2019.1.3; MBF Bioscience, VT, USA). Fluorescence analysis was performed for α-Syn and Iba1 using a Leica SP5-II confocal microscope (Leica Microsystems, Wetzlar, Germany). 40X and 60X lenses were used, and Z sectioning was performed at 1-2 μm intervals in order to verify the co-localization of markers. Image extraction and analysis was conducted via the Leica LAS software (LAS AF 2.7.3.9723).

Statistical analysis
GraphPad Prism 8 software (San Diego, CA) was used for statistical analyses. For comparing two groups, unpaired t tests were used. For comparisons between three or more groups, one-way analysis of variance (ANOVA) followed by Tukey's post-hoc test for multiple comparisons between treatment groups was conducted. For analyzing the nest building behavior, a two-way repeated measures ANOVA was performed using the Sigma Plot 14 (San Jose, CA) software. Differences were accepted as significant at p < 0.05. Statistical details of each experiment are provided within the relevant result and legend sections.

Loss of NRF2 enhances behavioral deficits and induces TH neuron loss in the hα-Syn overexpressing mice.
To determine the effects of NRF2 loss on motor and other spontaneous behaviors, all four strains of mice were subjected to nest building, challenging beam, and open field tasks at 3 mos of age. In the nest building task, pieces of cotton nesting material (nestlets) are placed in the feeding bin at the top of the cage, and the amount of nestlet pulled down and used for nest building can be assessed ( Fig. 2A). As expected, mice expressing hα-Syn (hα-Syn + /Nrf2 +/+ and hα-Syn + /Nrf2 -/-) both pulled down (  Supplementary Table 1). However, hα-Syn overexpressing mice which lack NRF2 (hα-Syn + /Nrf2 -/-) showed significantly reduced abilities to both pull down and use nestlets compared to hα-Syn overexpressing mice with NRF2 (hα-Syn + /Nrf2 +/+ ) (Fig.  2B, C). Next, we assessed the ability of the mice to cross an increasingly narrow beam (Fig. 2D) by measuring the time taken and steps utilized to traverse the beam, as well as the number of foot slip errors. Interestingly, hα-Syn + /Nrf2 +/+ mice exhibited a slower traverse time than the ha-Syncontrols but hα-Syn + /Nrf2 -/showed quicker traversal across the beam ( Fig. 2E; P = 0.009, F3, 60 = 5.305, One-way ANOVA). Moreover, hα-Syn + /Nrf2 -/mice also showed a higher number of errors per step than the hα-Syn + /NRF2 +/+ mice ( Fig. 2F; P = 0.0519, F3, 58 = 33.09, One-way ANOVA). There were no significant differences in the number of steps utilized to cross the beam across all four groups (Fig. 2G). These data infer that loss of NRF2 in the hα-Syn overexpressing mice results in a more error prone hyperactive (and potentially anxious) phenotype. To further confirm a change in both locomotor function and anxiety-like behavior, an open field test, which determines spontaneous movements and entries/time spent in a central area of the field was used (Fig. 2H). While all four strains spent a similar amount of time moving in the open field arena (Fig. 2I), both hα-Syn + /Nrf2 +/+ and hα-Syn + /Nrf2 -/mice entered the central area more than the hα-Syn -/Nrf2 +/+ and hα-Syn -/Nrf2 -/mice ( Fig. 2J; P = 0.0011, F3, 60 = 10.86, One-way ANOVA). This suggested increased novelty seeking behavior and exploration-related hyperactivity in the hα-Syn + mice, compared to hα-Synmice. However, hα-Syn + /Nrf2 -/mice spent significantly less time exploring the central field than the hα-Syn + /Nrf2 +/+ mice (Fig. 2K, Supplementary Table 1; P = 0.0002, F3, 60 = 8.647, Oneway ANOVA) signifying greater anxiety-like behavior in the novel environment (unprotected central area of the field) in these animals. Overall, these findings indicate that a loss of NRF2 exacerbates motor dysfunction and induces affective changes in mice overexpressing hα-Syn. Given the heightened motor deficits seen in the hα-Syn + /Nrf2 -/mice, we examined Tyrosine Hydroxylase (TH) immunoreactivity in the substantia nigra (SN) and striatum to assess the nigrostriatal pathway. The Thy1-α-Syn mice typically do not exhibit loss of TH + midbrain DA neurons in the SN [30]. However, a loss of TH immunoreactivity was evident within the lateral tier of the SN in the hα-Syn + /Nrf2 -/animals compared to hα-Syn + /Nrf2 +/+ controls (Fig. 3C-F).

Oligomerization of α-Syn is promoted by genetic ablation of NRF2.
The oligomerization and eventual formation of insoluble α-Syn aggregates is a common pathological hallmark associated with neuronal cell dysfunction and death during PD. To determine if the behavioral deficits observed in the 3 mos old hα-Syn + /Nrf2 -/mice were associated with α-Syn oligomerization and aggregation, different PD-relevant brain regions were assessed for monomeric and oligomeric α-Syn species levels using western blotting of soluble and insoluble protein fractions. Specifically, four brain regions were studied; the midbrain (MB), striatum (ST), cortex, and hippocampus (HC). The expression of monomeric α-Syn was similar between hα-Syn overexpressing mice that either possessed or lacked NRF2, both within the soluble and insoluble fractions across the four brain regions (Fig. 4A-E). However, the expression of insoluble oligomeric α-Syn species were notably increased in the MB and ST of hα-Syn + /Nrf2 -/mice compared to hα-Syn + /Nrf2 +/+ mice (Fig. 4A, B). . There was no difference in the expression levels of soluble or insoluble α-Syn in the HC between the two genotypes ( Fig. 4D-F). Thus, loss of NRF2 causes greater α-Syn oligomerization and formation of insoluble aggregates in the MB, ST, and cortex, but not in the HC, of the 3 mos old hα-Syn overexpressing mice.

α-Syn overexpressing mice develop higher oxidative stress levels in the absence of NRF2.
As NRF2 is known to regulate numerous aspects of the antioxidant response, the effect of genetic ablation of NRF2 on oxidative stress was studied. 4-hydroxynonenal (4-HNE), a reactive lipid species that accumulates in the brain during PD progression, can form adducts with protein cysteine residues and serve as an indicator of increased oxidative modification of proteins. Overall, both the hα-Syn overexpressing mouse strains, hα-Syn + /Nrf2 +/+ and hα-Syn + /Nrf2 -/mice, showed an increase in 4-HNE protein adduct formation in the MB, ST, cortex, and HC compared to the non hα-Syn overexpressing mouse strains, hα-Syn -/Nrf2 +/+ and hα-Syn -/Nrf2 -/- (Fig. 6A-D).

Loss of NRF2 induces autophagic changes in hα-Syn overexpressing mice.
Autophagic dysfunction is a key feature of PD that is known to drive the accumulation of toxic α-Syn aggregates. NRF2 has also been shown to regulate key components of the autophagy machinery [6,22]. Thus, we evaluated autophagy by assessing protein levels of two standard autophagy markers, LC3-II and SQSTM1/p62 through western blot. LC3-II is generated by the conjugation of cytosolic LC3-I to phosphatidylethanolamine (PE) on the surface of nascent autophagosomes (thus it is an autophagosome marker), and p62 is involved in binding and recruiting ubiquitinated cargo for autophagy-mediated degradation (p62 itself is an autophagy substrate). LC3-II levels were significantly higher in the MB, ST, and cortex of the hα-Syn + /Nrf2 -/mice than hα-Syn + /Nrf2 +/+ animals ( No specific differences in LC3-II were noted in the hippocampus between the four mice groups (Fig. 8D-F). In terms of p62 expression, there were no significant differences between the four groups in any of the brain regions ( Fig. 8A-B & E-F). Nevertheless, qRT-PCR indicated that p62 transcript levels were higher in hα-Syn + /Nrf2 -/mice than hα-Syn + /Nrf2 +/+ controls ( Fig. 8G; MB -P = 0.056, F3, 12 = 20.36; ST -P = 0.006, F3, 12 = 22.91, One-way ANOVA). These data, along with the LC3-II changes, indicated increased basal autophagy in the hα-Syn + /Nrf2 -/mice. To further understand these autophagic changes, we conducted transmission electron microscopy on tissues from the MB, ST and cortex. Here, it was observed that hα-Syn + /Nrf2 -/mice had more autophagic vesicles (arrows) in these 3 regions than hα-Syn + /Nrf2 +/+ mice ( Fig. 8H; quantification in Fig. 8I-K) (MB -P <0.0001, unpaired t-test, t = 11.18, df = 37; ST -P <0.0001, unpaired t-test, t = 11.44, df = 37; Cortex -P = 0.003, unpaired t-test, t = 3.2, df = 30). Typical double membraned autophagosomes and autolysosomes (red arrows in Fig. 8H [b, d, f]; high magnification images of an autophagosome and autolysosome are in Fig. 8H [g &  h]), as well as what appeared to be dense residual bodies (yellow arrows), multilamellar bodies (MLBs, blue arrows), and multivesicular bodies (MVBs, black arrows), which are all vesicular compartments of the autophagic pathway, were noted in the hα-Syn + /NRF2 -/mice. These data indicate that depletion of NRF2 results in a compensatory increase in basal autophagy, plausibly to cope with the greater load of insoluble α-Syn aggregates.

DISCUSSION
In essence, our study generates novel data that reveal that an early lack of NRF2 promotes α-Syn proteinopathy and other hallmark PD pathologies such as DA neuron degeneration, oxidative stress, inflammation and altered autophagy. Importantly, the study also determines that NRF2 loss exacerbates motor deficits and induces anxiety, underlining the functional relevance of NRF2 activity in PD. These results emphasize a vital role for NRF2 in the pathogenesis of PD.
Our data show that the lack of NRF2 worsens motor function and induces anxiety in the α-Syn overexpressing mice. Specifically, deficits in the nest building, challenging beam, and open field tasks were seen. The nest building task involves a combination of rearing and complex fine motor skills to grasp and pull-down cotton from the feeder bin. Moreover, on the floor of the cage, animals require the use of orofacial and forelimb movements to tear apart the nesting material and incorporate it into their bedding. It was found that both the amount of cotton retrieved from the bin, as well as the amount of cotton shredded for nest building, was lower for the NRF2 knockout mice at all tested time points, with significant differences from control noted between 36 and 72 hrs. This indicated that NRF2 loss promoted impairments in both fine and gross motor function in the animals. In the challenging beam task, interestingly, the hα-Syn + /Nrf2 -/mice traversed the beam faster but committed more foot slip errors than the hα-Syn + /Nrf2 +/+ animals, suggesting reduced motor coordination and a hyperactive phenotype. Assessment in the open field showed that although both the NRF2 knockout and wildtype animals overexpressing α-Syn entered the central box equally, the NRF2 knockout mice spent less time in the box. Mice typically exhibit 'thigmotaxis' (hugging the walls) relative to exploring the central, more unprotected area of the open field [41]. Thus, the increased central entries suggest higher novelty seeking behavior and exploration-related hyperactivity in the hα-Syn + mice, compared to hα-Synanimals. However, the reduced time spent in the center (more thigmotaxis) in the hα-Syn+ mice lacking NRF2, compared to those with NRF2, may indicate strong emotional responses to the novel environment in these animals (that is greater anxiety-like behavior in the exposed central area of the field). All in all, these data indicate that a deficit in NRF2 promotes motor and affective changes reminiscent of symptoms in individuals at early stages of PD [42][43][44][45], and provide the first evidence for a direct role for NRF2 in the expression of PD-relevant behaviors.
We studied the distribution and expression of α-Syn in four brain regions, specifically the striatum, midbrain, cortex and hippocampus. It was found that the lack of the NRF2 gene does not significantly alter monomeric α-Syn expression, however, levels of phosphorylated α-Syn were increased in the striatum, midbrain and cortex. An elevation in oligomeric and insoluble α-Syn species was also determined in these regions. These data infer that NRF2 loss enhances α-Syn phosphorylation, which causes oligomerization and eventual conversion of α-Syn into insoluble aggregates in the hα-Syn overexpressing mice. These findings align with a study by Lastres-Becker et al., which showed that virally induced expression of α-Syn in the ventral midbrain of NRF2 knockout mice resulted in the increased immunohistochemical expression of phosphorylated α-Syn and greater α-Syn aggregation in the neurites of dopaminergic neurons [25]. Nevertheless, in contrast, our studies provide a detailed qualitative and quantitative analysis of different α-Syn isoforms across multiple brain regions in a translationally relevant model where there is continual endogenous production of human α-Syn. Moreover, our studies link the molecular alterations in α-Syn to behavioral (motor and emotional) changes in the hα-Syn + /Nrf2 -/mice. Additionally, connected to the synuclein alterations, we also noted higher basal autophagy in the hα-Syn + /Nrf2 -/mice. These data suggest that the buildup of α-Syn aggregates due to the enhanced modification of monomers and formation of oligomers in the brains of hα-Syn + /NRF2 -/mice activated autophagy and may overwhelm clearance through the autophagosome-lysosome machinery leading to neuronal toxicity [46]. Supporting this notion, a notable reduction in DA neuron numbers was observed in the SN of hα-Syn + /NRF2 -/mice. Moreover, TH-positive neurons in the SN with high phospho-α-Syn expression appeared to have lower TH immunoreactivity in the hα-Syn + /Nrf2 -/mice, suggesting that α-Syn aggregation could be contributing to TH neuron dysfunction.
Elevated oxidative stress and neuroinflammation were observed in the NRF2 deficient hα-Syn overexpressing animals. In particular, higher oxidative stress levels (4-HNE adducts and ROS) were found in the midbrain and striatum of hα-Syn + /NRF2 -/mice compared to controls, with associated pro-inflammatory changes involving increased COX2 and iNOS levels in the midbrain, striatum and cortex. Additionally, more microglia were also observed especially in α-Syn rich regions in mice lacking NRF2. In this context, it has long been recognized that activated microglia surround Lewy bodies in postmortem PD brains [47][48][49][50]. α-Syn, which is a major constituent of Lewy bodies, has been shown to bind to and activate microglia through toll-like receptors (TLRs) that mediate innate immunity [30,51,52]. Such activated microglia not only represent an immunological reaction to ongoing damage by phagocytosing damaged cells and debris, but can also emit harmful ROS and pro-inflammatory cytokines [47]. In turn, these oxidative stress and pro-inflammatory changes can further aggravate α-Syn misfolding and aggregation [46,53,54]. Interestingly, the Lastres-Becker et al., study also reported increased gliosis and inflammation in NRF2 knockout mice in response to acute viral expression of α-Syn as well as impaired phagocytotic ability of microglia in NRF2 deficient animals [25]. Given NRF2's classical role in modulating the antioxidant and anti-inflammatory response, the stimulation of oxidative stress and inflammation in the hα-Syn + /Nrf2 -/is not completely unexpected [7,55]. However, the significantly higher intensity of these changes in hα-Syn + /Nrf2 -/as compared to their wild-type counterparts and hα-Syn -/Nrf2 -/animals suggests an important role for NRF2 in controlling α-Syn related chronic inflammation and excessive production of damaging free radical species, further solidifying the importance of maintaining proper NRF2 levels even during the earliest stages of neurodegeneration. This finding is also of particular interest as it indicates that premature loss of NRF2 could result in a more rapid and severe progression of the disease.
Interestingly, the hippocampus was the least affected region in our studies. It has been shown in the Thy1-hα-Syn mice that hippocampal changes in terms of αsynuclein expression and cognitive alterations are consistently seen only after 4-5 mos of age [30]. Addressing such regional differences due to NRF2 loss with age will be important to investigate in future studies. With regards to cell-specific effects, it has been reported that astrocytes express high levels of NRF2 and may be the major cell type for NRF2 activation during neurodegeneration. For example, in the basal ganglia of 1-methyl-4-phenyl-1,2,3,6 tetrahydropyridine hydrochloride (MPTP)-treated mice, astrocytes are the main cell type to upregulate NRF2 in response to sulforaphane (an NRF2 inducer) treatment [56]. Moreover, NQO1, peroxiredoxin 6, and HO-1, all NRF2 targets, are strongly expressed in astrocytes in SN of human post-mortem PD brains [8,56,57]. Additionally, it has been shown that the viral overexpression of NRF2 in astrocytes renders mice resistant to MPTP toxicity and mutant α-Synuclein [24,58], and that the transplantation of NRF2 overexpressing astrocytes protects against 6hydroxydopamine (6-OHDA) neurotoxicity [59]. Nevertheless, on the other hand, it is also known that nuclear NRF2 expression is largely restricted to neurons in the human post-mortem brain SN of PD patients [10], and HO-1 is also increased [8]. Furthermore, increased NRF2 expression seen in response to dimethyl fumarate treatment (NRF2 inducer) in MPTP-treated mice is partially localized to midbrain neurons [60]. Overall, these findings highlight a key involvement of astrocytes, and also yet unknown interactions between neurons and astrocytes that may be crucial to NRF2's role in PD pathogenesis. Such cell-specific effects related to NRF2 will be vital to understand.
In summary, this study reveals the early loss of NRF2 as a critical driver of α-Syn-mediated pathology and cellular dysfunction that leads to behavioral deficits similar to those commonly observed in PD patients. Indeed, NRF2 may be crucial to PD pathogenesis through its effects on core neurodegenerative processes such as oxidative stress, protein quality control and inflammation that ultimately induce cell death. By extension, a deeper understanding of NRF2-based mechanisms will support the development of much needed therapeutics for mitigating the harmful neuropathological features across all stages of this devastating degenerative disease.