Astrocytic Nrf2 Expression Protects Spinal Cord From Oxidative Stress in a Spinal Cord Injury Animal Model

Background Spinal cord injury (SCI) induces a multitude of deleterious processes, including neuroinammation and oxidative stress (OS) which contributed to neuronal damage and demyelination. Recent studies have suggested that increased formation of reactive oxygen species (ROS) and the consequent OS are critical events associated with SCI. However, there is still little information regarding the impact of these events on SCI. Astrocytes are key regulators of oxidative homeostasis in the CNS and astrocytic antioxidant responses promote the clearance of oxidants produced by neurons. Therefore, dysregulation of astrocyte physiology might largely contribute to oxidative damage. Nuclear factor erythroid 2-related factor 2 (Nrf2) is the main transcriptional regulator of cellular anti-oxidative stress responses. These experiments indicate an of in our contusion of to in vivo analysis, we measured the expression of four typical Nrf2-target genes including heme oxygenase-1 (HO-1), glutamate-cysteine ligase catalytic subunit (GCLC), NAD(P)H dehydrogenase [quinone] 1(NQO1) and thioredoxin reductase-1(TXNRD1) by RT-qPCR. These results showed an induction of the gene expression of all selected genes in epicenter after SCI (gures 2 A-D). To further investigate OS, we analyzed protein levels of HO-1 and NQO1 in SC using WB. E-G, of and NQO1 was signicantly as early post-OP and remained at comparable levels until These results indicate the critical role of the Nrf2-regulated anti-oxidant machinery in the early phase after


Abstract Background
Spinal cord injury (SCI) induces a multitude of deleterious processes, including neuroin ammation and oxidative stress (OS) which contributed to neuronal damage and demyelination. Recent studies have suggested that increased formation of reactive oxygen species (ROS) and the consequent OS are critical events associated with SCI. However, there is still little information regarding the impact of these events on SCI. Astrocytes are key regulators of oxidative homeostasis in the CNS and astrocytic antioxidant responses promote the clearance of oxidants produced by neurons. Therefore, dysregulation of astrocyte physiology might largely contribute to oxidative damage. Nuclear factor erythroid 2-related factor 2 (Nrf2) is the main transcriptional regulator of cellular anti-oxidative stress responses.

Methods
In the current study, using mice with a GFAP-speci c kelch-like ECH-associated protein 1 (Keap1)deletion, we induced a hyperactivation of Nrf2 in astrocytes and further its effects on SCI outcomes. SCIinduction was performed in mice using the In nite Horizon Spinal Cord Impactor with a force of 60 kdyn. To assess the quantitative pattern of Nrf2/ARE-activation, we included transgenic ARE-Luc mice. Data were analyzed with GraphPad Prism 8 (GraphPad Software Inc., San Diego, CA, USA). Brown-Forsythe test was performed to test for equal variances and normal distribution was tested with Shapiro-Wilk.

Results
In ARE-Luc mice, a signi cant induction of luciferase-activity was observed as early as 1 day post injury, indicating a functional role of Nrf2-activity at the epicenter of SCI. Further, SCI induced loss of neurons and oligodendrocytes, demyelination and in ammation in wild type mice. The loss of myelin and oligodendrocytes was clearly reduced in Keap1 KO mice. In addition, Keap-1 KO mice showed a signi cantly better locomotor function and lower neuroin ammation responses compared to wild type mice.

Conclusions
In summary, our in vivo bioluminescence data showed Nrf2-ARE activation during primary phase of SCI. Further, we found that cell speci c hyperactivation of Nrf2 was su cient to protect the spinal cord against injury which indicate a promising therapeutic approach for SCI-treatment.

Background
Spinal cord injury (SCI) characteristically causes life-long neurological consequences such as paralysis, loss of sensation and voluntary motor function downwards the site of injury which remains a major cause of permanent disability (1). SCI is divided into primary and secondary injuries in a pathophysiological classi cation; primary injury refers to the initial physical damage of the spinal cord (SC) caused by an indirect or direct external force, while the secondary injury is described by a series of physiological and pathological changes including neuroin ammation, neuronal apoptosis and excessive occurrence of oxidants such as reactive oxygen species (ROS) (2). However, oxidants are not necessarily harmful per se but do also play important roles at low exposition rates during redox-related signaling cascades by which they regulate and participate in physiological processes (thoroughly reviewed by Sies and Jones in 2020 (3)). The redox homeostasis under these circumstances should be understood as a range rather than a clearly de ned intracellular state in which the ratio of oxidants and antioxidants within the cell is in dynamic equilibrium with a pending electron ow, a condition that is called oxidative eustress (4)(5)(6). Nevertheless, once endogenous antioxidants are overwhelmed by an increased occurrence of oxidants, exceeding the limits of homeostatic conditions, a harmful imbalance between products of oxidation and antioxidant defenses results, known as oxidative distress (OS) (7). Besides, oxidative damage to proteins and DNA, these oxidants attack polyunsaturated fatty acids in the membrane lipids, change membrane potentials and eventually rupture membranes leading to the release of cell and organelle contents (8). Neurons are particularly prone to oxidative and electrophilic distress due to many factors, including a high content of polyunsaturated fatty acids, high rate of oxidative metabolic activity, intense production of reactive oxygen metabolites and relatively low antioxidant capacity (9). Although increased OS is an important hallmark of the secondary phase of SCI (10), there is still little information about the impact of this biochemical changes associated with SCI outcome.
Astrocytes are the most abundant glial cells in the central nervous system (CNS) and among the rst responders to SCI (11). After SCI, local environment undergoes profound biochemical and cellular changes that affect astrocytes to become reactive and include astrogliosis (12). Reactive astrogliosis is characterized by the proliferation and hypertrophy of astrocytes, which eventually leads to scar formation in epicenter of injury post SCI (13). Astrocytes play critical roles in the normal SC; they closely interact with neurons to provide structural, metabolic and trophic support and actively participate in the modulation of neuronal excitability and neurotransmission (14). In the presence of astrocytes, neurons are more resistant to OS induced by several compounds (15). Astrocytes carry out this function mainly through transcription factor nuclear factor erythroid 2-related factor 2 (Nrf2)-driven genes such as quinone oxidoreductase 1 (NQO1) and heme oxygenase-1 (HO-1) which are both preferentially activated in astrocytes but lesser in neurons (16). Therefore, functional alterations in astrocytes (such as aggravation of OS) can shape and alter their interaction with neurons.
Nrf2, belonging to the cap 'n' collar subfamily of basic region-leucine zipper transcription factors, is the main transcription factor involved in the regulatory cascades during oxidative eustress responsible for the regulation of a whole battery of cytoprotective and antioxidative proteins by binding to antioxidant response elements (ARE) in their promoter region (17,18). The activity of Nrf2 is negatively regulated by the cytosolic regulatory protein kelch-like ECH-associated protein 1 (Keap1) (19). Once, cells are exposed to OS or chemopreventive factors, Nrf2 is dissociated from Keap1, translocated into the nucleus, and induces the expression and up-regulation of downstream cytoprotective enzymes that attenuate tissue injury (20). Since, Nrf2 is rapidly degraded through interactions with Keap1, increased Nrf2 expression does not necessary lead to increased Nrf2 activity (21). Nrf2-inducers such as pharmacological activation of the Nrf2/ARE system is a pathway which leads to increased Nrf2-signaling. For example, treatment with dimethyl fumarate (DMF) results in bene cial effects in different CNS diseases, such as multiple sclerosis (22) and SCI (23). Although the mechanism of action is not completely understood, it has been shown that dimethyl fumarate (DMF) induces Nrf2-activity by disruption of the Nrf2-Keap1 interaction due to a modi cation of the Keap1 protein (24). However, all known Nrf2-inducers not only enhance Nrf2activity but also exert various side effects within the brain and other peripheral tissues (25). Therefore, genetic and cell type-speci c targeting of Keap1 and Nrf2 represents the most suitable method to investigate Nrf2-speci c effects. In the current study using GFAP-Cre::keap1 ox/ ox KO mouse, we induced hyper-activation of the Nrf2/ARE system in astrocytes in a SCI model and investigated the protective effects of astrocyte-speci c Nrf2/ARE-activation in the SC after SCI.

Animals
All experimental procedures and animal care were approved by the review board for the care of animal subjects of the district government (LANUV, Germany) and are reported in accordance with the ARRIVE guidelines. The mice were housed and handled in accordance with the guidelines of the Federation for European Laboratory Animal Science Associations (FELASA) under standard laboratory conditions.

Spinal cord injury
General anesthesia was initiated with iso urane (2-3 vol. %) in an anesthetic chamber. During surgery, iso urane (1,5-2 vol. %) was further administered via a face mask. Intraoperative analgesia was attained through injection of buprenorphine (0.05-1 mg/kg sc.) 30 min preoperatively. After the exposure of the spinal column (Th7-Th10), a laminectomy of Th8 was performed (Fig. S1A). A standardized injury of the SC at this level was induced by contusion (In nite Horizons Spinal Cord Impactor) with a force of 60 kdyn (28). During the surgery, body temperature was maintained at 36-37°C. Postoperative aftercare included rehydration (0,5 mL NaCl), postoperative analgesia and manual emptying of bladder twice a day until spontaneous urination returned. In the control group (sham), laminectomy without contusion of the SC was carried out to preclude possible falsi cations of the results caused by the mere surgical procedure. ARE activity ARE-activity measurement were performed as previously described (27). Brie y, after shaving the back of mice, luciferin solution was prepared and used at a nal concentration of 20 mg/mL. Totally, 0.2 mL luciferin solution was injected intraperitoneal. After injection, mice were anesthetized by inhaling 4% (v/v) iso urane at a ow rate of 2 L/min oxygen for approximately 1 min and then kept at 1.5% (v/v) and 1 L/min using a mouse face mask. To ensure an adequate distribution of Luciferin in the whole body, the measurements were conducted fteen minutes after the injection. For each measurement, one sham and one SCI mouse was imaged simultaneously for 5 min at high-resolution settings (binning: 2, f/stop: 1) with a eld of view of 7.5 cm. Luminescence signals were detected with an IVIS Lumina 100 Series Imaging device (Xenogen Corporation,vAlameda, CA), digitized and evaluated using the Living Image software 3.2. The signal intensities are expressed as average radiance (p/sec/cm²/sr) and visualized with the help of a false color scale. Spinal cord-speci c Nrf2/ARE activity was determined by correcting luminescence signals from pre-de ned region of interest (ROIs) using background ROI (Fig. S1B).

Functional activity
The locomotion de cits of mice after SCI were scored in an open eld according to Basso, Beattie, and Bresnahan (BBB) locomotion where hindlimb function could be assessed (Basso et al., 1995). This is a 22-point scale that systematically details hindlimb function of joint movements, stepping ability, the degree of ne control of coordinated stepping and trunk stability. Test sessions were 4 min in duration and mice were tested every day, from days 1 to 7 post-injury and were scored by two experienced raters which were blinded to experimental groups.

Tissue preparation
To obtain samples for molecular biology and biochemical examinations, mice were deeply anesthetized and transcardially perfused with ice cold phosphate-buffered saline (PBS), the whole SC was quickly removed, and the epicenter part was dissected, snap-frozen in liquid nitrogen, and stored at -80°C until further experiments. For immunohistochemistry, mice were transcardially perfused with ice cold PBS followed by a 3.7% paraformaldehyde solution (PFA, pH 7.4). For decalci cation, spinal columns were incubated in 20% ethylenediaminetetraacetic acid (EDTA) for 48 h at 37°C prior to para n embedding. To this end, tissue specimens were embedded in para n (Merck, Germany) and 5 µm para n sections were cut.

Gene expression studies
For RNA isolation, tissues were placed in homogenization tubes containing 1.4 mm ceramic beads. RNA was isolated by phenol-chloroform extraction using peqGold RNA TriFast (PeqLab, Germany). Therefore, samples were homogenized at 5.000 x g for 15 s. Total RNA amount and purity were determined with a Nanodrop 1000 photospectrometer (PeqLab, Germany) using A260/A280 as well as A260/230 ratios. cDNA was synthesized by reverse transcription using M-MLV reverse transcription (RT)-kit (Invitrogen, USA) and random hexanucleotide priming (Invitrogen, Germany). Gene expression levels were determined by real-time reverse transcription-PCR using SensiMix™ SYBR® & Fluorescein Kit (Bioline, Germany) in a CFX connect Real-Time PCR system device (Bio-Rad, Germany). Primer sequences, individual annealing temperatures (AT), expected amplicon length are shown in Table 1. Results were evaluated using Bio-Rad CFX manager (Bio-Rad, Germany) and were normalized to a reference gene index (RGI), which was calculated from cyclophilin A and HSP90 expression data. Ampli cation e ciencies were calculated by standard curve analyses using a 2-fold dilution series of pooled samples. The target gene expression was calculated using e ciency corrected ΔΔCq method, using the geometric mean of both reference genes.  Table 2) was applied for 2 h at room temperature. Signals were analyzed via chemiluminescence detection (Westar Supernova, XLS 3,0100, Cyanagen, Italia), visualized (Fusion Solo X, Vilber, Germany) and subjected to densitometry analysis using Image J. Results were normalized to the expression of GAPDH or b-actin as reference protein.
Immunohistochemistry and luxol fast blue staining For immunohistochemistry, 5 µm thick sections of SC were depara nized, rehydrated, and antigens were unmasked by heating in Tris/ EDTA (pH 9.0) buffer for 20 min. After blocking with 5% normal goat serum in PBS, the sections were incubated overnight at 4°C with the primary antibody diluted in blocking solution. Primary antibodies and dilutions used in the study are given in Table 2. After washing and blocking of endogenous peroxidase with 0.3% hydrogen peroxide, slides were incubated with the appropriate secondary antibody (1:50; see Table 2) diluted in 5% normal serum in PBS for 1 h (RT).
Afterwards   Brown-Forsythe test was performed to test for equal variances and normal distribution was tested with Shapiro-Wilk test. If necessary, data were transformed via Box-Cox-Y to achieve homoscedasticity. Oneway ANOVA followed by Dunnett post-hoc test or two-way ANOVA followed by Tukey post-hoc test was used for parametric data. Non-parametric data were analyzed with Kruskal-Wallis test followed by Dunn's multiple comparisons or Friedman test. All data are given as arithmetic means ± standard errors of the mean (SEM). p-values < 0.05 were considered signi cant. Detailed description of signi cance levels are given in the gure legends.

Results
Activation of the Nrf2/ARE pathway after SCI To investigate whether the Nrf2/ARE system plays a role in our experimental SCI animal model and to assess the quantitative pattern of its activation, we included transgenic ARE-Luciferase reporter gene mice. In this approach, due to the ARE-driven expression of the luciferase, the luminescence intensity directly correlates with Nrf2-activity. Using the XENOGEN Living Image system, the luminescence signals in SCI and sham mice were daily measured in vivo. As shown in gure 1, compared to sham control animals a signi cant increase of luciferase activity was observed in the epicenter of SC in SCI mice as early as 24 h post injury. The luminescence intensity gradually declined during the study but remained signi cantly higher for up to 4 days compared to sham controls ( gures 1A and B). Afterwards, the mean signal intensities of SCI mice were still higher than those of sham control animals but without reaching statistical signi cance. These experiments indicate an involvement of Nrf2-activity in our contusion model of SCI. Further, to con rm in vivo bioluminescence analysis, we measured the expression of four typical Nrf2-target genes including heme oxygenase-1 (HO-1), glutamate-cysteine ligase catalytic subunit (GCLC), NAD(P)H dehydrogenase [quinone] 1(NQO1) and thioredoxin reductase-1(TXNRD1) by RT-qPCR. These results showed an induction of the gene expression of all selected genes in epicenter after SCI ( gures 2 A-D). To further investigate OS, we analyzed protein levels of HO-1 and NQO1 in SC using WB. As shown in gures 2 E-G, protein expression of HO-1 and NQO1 was signi cantly induced as early as 12h post-OP and remained at comparable levels until day 7. These results indicate the critical role of the Nrf2-regulated anti-oxidant machinery in the early phase after SCI.

Nrf2 hyperactivation improves the functional recovery
In the next step, to investigate the effect of astrocytic Nrf2-activation on locomotor recovery after SCI, we performed the BBB scoring for 7 consecutive days after injury. During the rst 5 d after insult, there was no signi cant difference between WT and Keap1-KO mice ( gure 3 A) even if the mean score of Keap1-KO was slightly higher than the score of WT mice. At day 6 and 7 post injury, the Keap1-KO group showed a signi cantly higher BBB scoring compared to WT animals which indicates that increased Nrf2 activity improves the recovery of locomotor abilities in SCI mice. In a further attempt, we explored whether Nrf2target genes are regulated in Keap1-KO mice. The RT-qPCR data revealed that mRNA expression levels of NQO1 in Keap1-KO mice were signi cantly higher than in WT mice at 24 and 72 h post SCI ( gure 3 C), whereas, the expression level of HO-1 showed no signi cant changes in both mice strains (data are not shown).

Sustained elevated Nrf2/ARE pathway protects spinal cord tissue
We have shown that the activation of the Nrf2/ARE pathway in the epicenter of the SC is a characteristic feature of our SCI model. We, therefore, aimed at investigating whether a continuous hyperactivation of Nrf2 might prevent cellular pathological responses in traumatic SCI. Since astrocytes are critically involved in the maintaining of hemostasis during SCI, Keap1 was deleted speci cally in this cell type, representing a genetic model for sustained activation of the Nrf2/ARE pathway in astrocytes (29).
To demonstrate its effect on the prevention of demyelination, sagittal sections of SC were stained with LFB. In sham operated mice, LFB staining of myelin in the SC was mainly restricted to the white matter (yellow arrows gure 4 E). The microscopic observations revealed that at 7 d post SCI, LFB staining was strongly reduced with almost no staining visible in the dorsal column at the center and caudal of the injury (stars in gures 4 B and F). In addition, most of the ventral column also showed a partially reduction of staining in the epicenter. In sharp contrast, the loss of myelin was signi cantly prevented in Keap1-KO mice in both the central and caudal parts of SC ( gures 4 C and G). Since myelin basic protein (MBP) is one of the main protein known to be required for myelin assembly, we measured the amount of this protein in WT and Keap1-KO mice. Our results show that SCI signi cant reduced MBP in WT mice and that this effect is ameliorated in Keap1-KO animals, which also showed a decreased amount of MBP but signi cantly more than WT mice ( gures 4 I and J).
To further examine whether the suppression of demyelination is correlated with improved survival of oligodendrocytes in Keap1-KO mice, we conducted immunohistochemical detection with anti-Olig2 (oligodendrocytes) and anti-APC (mature oligodendrocytes) antibodies. The quanti cations of Olig2 immunoreactive oligodendrocytes showed that SCI induced a signi cant loss of Olig2-and APC-positive cells in the gray matter ( gures 5 B and F). While we found higher numbers of Olig2-positive cells in Keap1-KO animals compared to WT mice after SCI ( gures 5 B, C and D).To investigate whether increased astrocytic Nrf2-activity protected neurons against SCI-mediated damage, neuronal density was evaluated using immunohistochemistry against NeuN (a speci c neuronal protein). We measured a signi cant decline of NeuN-positive cells in WT mice after SCI and a similar reduction in Keap1-Ko mice Astrocyte reactivity in Keap1-KO mice A main characteristic feature of SCI is the accumulation of reactive astrocytes identi ed as astrogliosis. Therefore, our next approach was to investigate the activation of astrocytes in WT as well as Keap1-KO mice after SCI. As expected, SCI in WT mice increased the number of GFAP-( gures 7 B and D) and S100b-( gures 7 F and H) positive cells in the SC compared to sham treated animals ( gures 7 A and E). This induction in epicenter was comparable in both mouse strains and was not abolished in Keap1-KO mice ( gures 7 C and G). Additionally, mRNA and protein levels of GFAP were examined using RT-qPCR and WB, respectively. Although the number of GFAP-positive cells did not differ between genotypes, we observed a signi cant reduction of GFAP after 7 days at both the mRNA ( gure 7 I) as well as the protein ( gures 7 J and K) level in Keap1-KO mice compared to WT animals. This indicates that Nrf2 may rather affect the activation/reactivity of astrocytes than their proliferation.

Discussion
Spinal cord injury (SCI) has been shown to be associated with different kind of tissue damage (30). Besides others, oxidative stress has frequently been described to be involved during SCI pathogenesis (31). While oxidative eustress (low oxidant exposure) regulates a couple of intracellular redox-regulated processes to maintain redox-homeostasis (6), oxidative distress represents an excess of oxidants leading to tissue damage. The nuclear factor erythroid-2 related factor 2 (Nrf2) regulates a battery of genes that encode for anti-oxidative enzymes that are part of eustress-related signaling. We demonstrated that the Nrf2/ARE system was activated immediately in the course of SCI-induced redox signaling. However, signi cant tissue damage occurs in this model despite Nrf2 activation, suggesting that the physiological Nrf2 response is insu cient to adequately counteract the full extent of this damage (32).
Fu and colleagues just recently demonstrated that luteolin pre-treatment, a avonoid known to activate Nrf2 and to exert antioxidant properties, improved SCI-related burden in rats (33). Besides from luteolin there have been other Nrf2 activating substances been used in similar studies (34)(35)(36). However, all these studies share a common drawback that these compounds have a variety of other intracellular effects in addition to Nrf2 activation, which are not always completely known.
Astrocytes are prominent in the cellular response to SCI which maintain the neuronal structure and nutrition. Furthermore, it has been shown that astrocytes provide protection against different kind of damage, including oxidative stress (37). For this reason, a more detailed study of the role of Nrf2 speci cally in astrocytes during SCI, which has not been addressed in previous studies, would be of scienti c value.
Therefore, in addition to wildtype mice we used transgenic GFAP-mediated Keap1-KO mice to continuously activate Nrf2 in astrocytes. In contrast to before-mentioned studies, in which the treatment regiments induce a systemic Nrf2 activation, we were able to show that Nrf2 hyperactivation exclusively in astrocytes was absolutely su cient to overcome SCI-related outcomes. In particular, our study revealed that besides protecting against general tissue damage (including demyelination and oligodendrocyte loss), sustained Nrf2 activity also prevented the intracellular redox milieu to switch to oxidative distress.
Neuroin ammatory response is one of the main events which occurs within early hours from the onset of SCI, which contributes to neuronal death and further neurologic damages. The neuroin ammation is mainly characterized by the in ltration of in ammatory cells such as macrophages as well as the accumulation of in ammatory mediators like IL-1b and IL-6. In this study we observed that Keap-1 KO mice showed a signi cantly lower expression of pro-in ammatory cytokines and also SCI-induced microgliosis, suggesting that increased astrocytic activity of Nrf2 plays a critical role in counteracting the in ammation. Even though microgliosis was clearly reduced in Keap1-KO mice, astrocyte numbers did not alter. However, in the context of neuroin ammation not simply the amount of astrocytes but rather their reactivity is crucial for tissue damage (38). Because the expression of GFAP, one of the prime markers of astrocyte reactivity, was signi cantly reduced in Keap1-KO mice, it is reasonable to conclude that Keap1 depletion signi cantly reduces the astrocyte-triggered in ammatory response, despite unchanged astrocyte numbers. This is consistent with the results of one of our previous studies showing that astrocytic Keap1-KO reduced neuroin ammation and tissue damage in a mouse model of multiple sclerosis, although the number of astrocytes also did not differ from treated WT mice (27). Several evidences con rm that Nrf2 can limit the in ammatory response by inhibiting NF-kB activation through maintenance of redox homeostasis (39,40). This important protective function of Nrf2 is mainly mediated through its antioxidant and detoxifying enzymes, such as NQO1. Thus, enhancement of these antioxidant systems, as is the case in astrocytes of Keap1-KO mice, could lead to decreased production of pro-in ammatory cytokines. Therefore, enhancement of these antioxidant systems, as is the case in astrocytes of Keap1-KO mice, could lead to decreased production of pro-in ammatory cytokines. This is particularly also important because, among other functions, astrocytes are responsible for maintaining a suitable environment for neurons including redox homeostasis. To ensure this, they contribute signi cantly to the reduction of oxidative distress by counteracting excessive accumulation of ROS in their surrounding area (41). Since increased emergence of ROS not only leads to oxidative damage of biomolecules but also activates microglial cells, astrocytes with increased antioxidant defenses may exert a direct inhibitory in uence on microgliosis (41).

Conclusion
In summary, our in vivo bioluminescence data showed Nrf2-ARE activation during primary phase of SCI. Further, we were able to demonstrate that hyperactivation of Nrf2 improves functional recovery, prevents the neuroin ammation and demyelinating. Better understanding of underlying mechanisms of oxidative stress could help us to promote new therapeutic intervention in SCI patients. Availability of data and materials The datasets analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.