Does Nrf2 Gene Transfer Facilitate Recovery After Contusion Spinal Cord Injury?

Introduction

Traumatic spinal cord injury (SCI) results in permanent loss of motor, sensory, and autonomic function due to the loss of neurons and axonal degeneration. Currently, there is no adequate therapy to repair the damaged SC mainly because of the complex SCI pathophysiology, consisting of the primary injury followed by spreading of secondary tissue damage. The secondary SCI phase, leading to further loss of tissue and function, occurs in response to a number of cytotoxic and degradative molecules, and consists of several interrelated processes, such as local vascular disturbances, excitotoxicity, free radical production and lipid peroxidation, necrotic and apoptotic cell death, and dysregulation of ion homeostasis (7). Therefore, one of the most desired goals in SCI therapy is the inhibition of secondary injury and promotion of functional recovery. Considering the important role of reactive oxygen species (ROS) and oxidative damage in the pathology and outcome of the secondary SCI phase, search for novel approaches directed to oxidative stress (OS) attenuation in SCI is of great importance.

Nuclear factor (erythroid-derived 2)-like 2 (Nrf2) is the “master regulator” of the antioxidant response, modulating the expression of antioxidant, phase II, and other cytoprotective genes (3). In nonstressed conditions, Nrf2 is constitutively ubiquitinated and thus inhibited through the Keap1/Cullin3 ubiquitin ligase complex. Upon exposure to electrophiles or OS, Keap1 is modified at its regulatory cysteine residues and inactivated. Keap1 inactivation enhances Nrf2 nuclear accumulation and antioxidant response element (ARE)-dependent transcription of genes encoding numerous protective enzymes and scavengers, including hemeoxygenase-1 (HO-1), NAD(P)H:(quinine acceptor) oxidoreductase 1 (NQO1), and glutamate cysteine ligase (GCL) (1). The Nrf2-ARE pathway has been shown to be activated in the SC after mouse compression SCI and rat contusion SCI (6, 9). Moreover, Nrf2 deficiency exacerbates the neurologic deficit and inflammation after compression SCI (6). As OS reflects an imbalance between the oxidants produced by the cells and the antioxidant gene products, a promising alternative would be to induce Nrf2 activation instead of the administration of single antioxidants to combat OS (3). Recently, sulforaphane, a natural compound with various potentially beneficial effects, including activation of Nrf2, was shown to provide protection in SCI (5, 9).

The introduction of transcription factors to the central nervous system with virus vectors enables induction or repression of genes. This is especially important in situations requiring the interplay of several factors sharing a common regulatory pathway. Therefore, in OS conditions the simultaneous induction of several antioxidant genes may be a better approach than transfer of individual antioxidant genes. Lentiviral vectors are commonly used to achieve stable gene expression in vitro and in vivo (8). Recently, we successfully used this approach to deliver human Nrf2 into the mouse hippocampus, a brain area affected in Alzheimer disease (AD). This resulted in an efficient and sustained Nrf2 expression in the hippocampus and improved cognitive functions of transgenic AD mice (4).

In this work, using the most clinically relevant mouse SCI model, we have determined whether Nrf2 in addition to combating OS participates in other pathophysiological mechanisms leading to secondary injury after SCI. Moreover, we have tested whether enhancing Nrf2 expression in the SC by gene transfer leads to improved functional recovery.

Induction of the Nrf2-ARE Pathway After Mouse Contusion SCI

There is some evidence that the Nrf2-ARE pathway becomes activated within the first 24 h after SCI (6, 9). Because it remains unknown how long this activation persists, we studied the time-course expression of the Nrf2 gene (Nfe2l2) and its main target genes Hmox1, Nqo1, Gclc, and Gclm at six different time points after injury. We detected significant elevation of Nfe2l2 expression starting as late as 3 days postinjury (dpi) (p<0.001) with a peak at 7 dpi (Fig. 1A). Hmox1 was significantly upregulated already at 12 h after SCI (p<0.001) and together with Nfe2l2 peaked at 7 dpi, at which time it was induced by 23-fold over controls. Nfe2l2 and Hmox1 elevation persisted for up to 42 dpi (Fig. 1B). Thus, we observed a delayed induction of Nfe2l2 expression peaking not earlier than 1 week after injury and persisting high at least up to 6 weeks when compared to controls. This upregulation of Nfe2l2 was associated with even earlier induction of Hmox1, which showed a sharp peak simultaneously with Nfe2l2 at 1 week, followed by clear reduction at 3 weeks but remaining elevated at least until 6 weeks. The very early induction of Hmox1 at 12 and 24 h may be explained by action of other transcription factors controlling Hmox1. For example, nuclear factor kappa-B (NF-κB) or activating protein-1 both regulate Hmox1 expression and are upregulated by a wide variety of prooxidant and proinflammatory stimuli.

OPEN IN VIEWER

FIG. 1.

Temporal expression of the Nrf2-ARE pathway genes following mouse contusion SCI. The expression levels of (A) Nfe2l2 and its target genes (B) Hmox1, (C) Nqo1, (D) Gclm, and (E) Gclc were assessed by quantitative real-time PCR at 12, 24 h, 3, 7, 21, and 42 days after mouse contusion SCI, n=10 for each group. mRNA expression levels are expressed as mean fold change normalized to rRNA±SEM. *p<0.05, **p<0.01, ***p<0.001. Nrf2, nuclear factor (erythroid-derived 2)-like 2; ARE, antioxidant response element; SCI, spinal cord injury; PCR, polymerase chain reaction.

The expression patterns of other Nrf2 target genes were different. Nqo1 is another important Nrf2 inducible gene and a key enzyme in the detoxication of reactive quinines. Although Nqo1 is considered to be one of the most consistently and robustly inducible genes among other cytoporotective enzymes activated through the Nrf2-ARE pathway, in our experiment it was significantly downregulated from 1 dpi (p<0.001) to 3 dpi (p<0.01), and only by 7 dpi reached the level of sham mice. Nqo1 became significantly elevated at 42 dpi (p<0.001) (Fig. 1C). Thus, in our model the pattern of Nqo1 expression was different from previous observations (9) where early upregulation of Nqo1 was demonstrated.

Gclm and Gclc, the two glutamate-cysteine ligase subunits, in contrast started with upregulation and shifted to downregulation several days after the injury. Interestingly, Gclm, but not Gclc was upregulated 24 h after SCI. Starting from 3 to 7 dpi both Gclc and Gclm were significantly downregulated until 21 and 42 dpi, respectively (Fig. 1D, E).

Together with previous studies, (9) our results suggest that while the immediate activation of the Nrf2-ARE pathway maybe crucial for acute protection after the SCI, a late phase activation of some target genes of the Nrf2-ARE pathway may further contribute to recovery processes. However, we cannot exclude the possibility that the activation of Nrf2-ARE pathways in cells outside of the lesion area or even in periphery plays a role in Nrf2-mediated beneficial responses. In addition, as Nrf2 regulates the expression of numerous genes, it is quite possible that target genes other than those investigated in the current study may contribute to the beneficial effects. The latter hypothesis is supported by our observation that mRNA levels of Nrf2 remained elevated at least up to 42 days postinjury.

Nrf2-Deficient Mice Exhibit Impaired Outcome After Contusion SCI

Previous studies on SCI have shown that Nrf2 deficiency is accompanied by elevated NF-κB activity, and increased production of pro-inflammatory cytokines, eventually resulting in increased neuronal death and deterioration of hindlimb function in models of mouse SC compression (5, 6). To assess the effect of Nrf2 deficiency on the recovery of motor function after contusion SCI, we monitored locomotor performance of wild-type (WT) and Nrf2-deficient mice (Nrf2^−/−) for 4 weeks after injury using the Basso mouse scale (BMS) (2). All the injured mice became almost completely paraplegic on the first day after the injury and then gradually displayed partial recovery of locomotor behavior (Fig. 2). Already at 7 dpi the locomotor performance of Nrf2^−/− mice was significantly impaired when compared to WT mice (p<0.05). This reduction in the performance of the Nrf2-deficient mice continued at 14, 21, and 28 days after injury. At 4 weeks after injury, the Nrf2^−/− mice attained an average BMS score of 2.8±0.2, which corresponds to extensive ankle movements or plantar placement of the paw, but not plantar stepping. The WT mice obtained an average BMS score of 4.7±0.6, which corresponds to occasional or frequent plantar stepping without coordination or frequent or consistent plantar stepping with coordination, but with the paw rotation. These results suggest that Nrf2 deficiency worsens the recovery of hindlimb motor function and propose that Nrf2 is essential for functional recovery after contusion SCI.

OPEN IN VIEWER

FIG. 2.

Impaired locomotor recovery in Nrf2^−/− mice after contusion SCI. Hindlimb motor function of WT and Nrf2^−/− mice was evaluated 1, 7, 14, 21, and 28 days after SCI by BMS scoring. The data are shown as mean±SEM. n=12 for WT with injury, n=11 for Nrf2^−/− with injury, n=10 for WT sham, and Nrf2^−/− sham. *p<0.05, **p<0.01, ***p<0.001. BMS, Basso mouse scale; WT, wild-type.

To assess the effect of Nrf2 deficiency on the morphology of the SC, we next examined the transverse SC area at the different levels around the injury epicenter on the sections stained with Luxol fast blue (LFB). The transverse SC area was reduced in the Nrf2^−/− mice 28 days after SCI (Fig. 3A). Because it is known that the functional recovery after SCI correlates with the white matter sparing in the injury epicenter (2), we assessed the degree of white matter sparing in the contused SC 28 days after injury. LFB staining revealed that the myelinated areas were less preserved rostrally to the lesion epicenter in Nrf2^−/− mice (Fig. 3B, C). This suggests that contusion SCI reduces not only the SC surface area but also white matter size in Nrf2-deficient mice. As white matter sparing influences locomotor recovery after SCI (2), the detected reduction in myelin preservation is most likely a reason why Nrf2^−/− mice demonstrated impaired motor function recovery. This is consistent with previous findings demonstrating that Nrf2^−/− mice exhibit greater demyelination than WT mice in experimental autoimmune encephalomyelitis. Moreover, administration of sulforaphane increased white sparing after rat SCI (9). Therefore, activation of Nrf2 appears to be required for proper remyelination after SCI.

OPEN IN VIEWER

FIG. 3.

Reduced SC area, decreased white matter sparing, and increased astrogliosis in injured Nrf2^−/− mice. Quantitative analysis of the SC area (A), spared white matter (B) assessed by LFB staining and astrogliosis by GFAP staining (E) measured in axial sections at different regions 28 days after SCI. (C) Representative LFB-stained images from the lesion epicenter 28 days after SCI. (D) Representative GFAP-stained images 1 mm caudally to the lesion epicenter. 0 μm indicates the lesion epicenter. Data are shown as mean area±SEM. n=12 for WT mice with injury and n=11 for Nrf2^−/− mice with injury, *p<0.05, **p<0.01, ***p<0.001. GFAP, glial fibrillary acidic protein; LFB, Luxol fast blue.

Next, we assessed whether Nrf2^−/− mice exhibited aggravated astrogliosis. We used glial fibrillary acidic protein (GFAP) histochemistry to assess the degree of astrogliosis at the injury epicenter, and 1 mm caudally and 1 mm rostrally to the injury epicenter. Astrogliosis was 2.2-folds higher in Nrf2^−/− mice following SCI caudally to the injury epicenter (Fig. 3D, E). As astrocytes are one of the major contributors to cytokine and chemokine synthesis following SCI, activation of Nrf2 may limit astrogliosis and therefore inflammation.

Gene expression levels of Nrf2 target genes were assessed 3 and 7 days after SCI in WT and Nrf2^−/− SC. Hmox1 and microsomal glutathione S-transferase 1 (mGST1) gene (Mgst1) expression were increased 3 and 7 dpi in both WT and Nrf2^−/− mice (Fig. 4A, B). Interestingly, at 7 dpi Nrf2^−/− mice exhibited significantly higher Hmox1 expression (p<0.001) when compared with WT mice (Fig. 4B). Mgst1 was highly upregulated in WT when compared with Nrf2^−/− both 3 and 7 dpi, but this difference was not statistically significant (Fig. 4A, B). Nqo1 levels were lower (p<0.05) in Nrf2^−/− mice compared with WT mice (Fig. 4A, B). Importantly, we observed that impaired motor recovery in Nrf2^−/− mice is not mediated by reduced expression of such important Nrf2-target genes as Hmox1 or Mgst1, but instead is associated with diminished expression of Nqo1, prior to and at the time of the onset of the motor impairment.

OPEN IN VIEWER

FIG. 4.

Nrf2-target gene expression profiles following SCI in Nrf2^−/− mice. The relative expression of Hmox1, Mgst1, and Nqo1 was measured by RT-PCR in WT and Nrf2^−/− mice (A) 3 days and (B) 7 days after SCI. Data are shown as mean±SEM. n=8–14. *p<0.05, ***p<0.001.

Apoptosis is the main mechanism of cell death following SCI and the role of OS in promoting apoptosis is well established. To our knowledge, the role of Nrf2 in regulating apoptosis in vivo has not been studied. We observed that expression of pro-apoptotic caspase 3 gene (Casp3) was induced at 3 dpi (p<0.05) (data not shown) and 7 dpi (p<0.001) in both studied genotypes (Fig. 5A). In contrast, the anti-apoptotic B-cell lymphoma 2 (BCL2) gene (Bcl2) was significantly downregulated at 7 dpi in both WT and Nrf2^−/− mice (p<0.05) (data not shown). Importantly, the expression of the anti-apoptotic BCL2-extra large (BCL2-xl) gene (Bcl2l1) was significantly reduced at 7 dpi (p<0.05) only in Nrf2^−/−mice (Fig. 5B). As induction of Bcl2l1, an anti-apoptotic gene, was prevented in Nrf2^−/−, it is well possible that the aggravated damage after contusion SCI in Nrf2-deficient mice is also, at least in part, due to increased apoptotic death of SC cells, including oligodendrocytes.

OPEN IN VIEWER

FIG. 5.

Expression profiles of select apoptotic, growth factor and inflammation genes are altered following SCI in NrP2^−/− mice. Gene (A–E) and protein expression (F) profiles following SCI in Nrf2^−/− mice. The relative expression of (A) Casp3 and (B) Bcl2l1, (E) Tnf was measured by RT-PCR and (F) protein levels of IL6 were measured by CBA in WT and Nrf2^−/− mice 7 days after SCI. Relative expression of (C) Ngf and (D) Gdnf was measured by RT-PCR in WT and Nrf2^−/− mice 3 days after SCI. Data are shown as mean±SEM. n=8–14. *p<0.05, **p<0.01, ***p<0.001. CBA, Cytometric Bead Assay; IL6, interleukin 6.

Growth factors modulate neuronal survival, neurite outgrowth, synaptic plasticity, and neurotransmission. Exogenous administration of growth factors has been proposed as a potential therapy for SCI, but whether Nrf2 regulates endogenous growth factor expression has not been described. We observed an induction of nerve growth factor (NGF) gene (Ngf) and glial cell-derived neurotrophic factor (GDNF) gene (Gdnf) at 3 dpi (Fig. 5C, D) and 7 dpi in both WT and Nrf2^−/− (p<0.001) mice. The upregulation of Ngf at 3 dpi was significantly delayed in Nrf2^−/− mice compared with WT (p<0.01) (Fig. 5C). In contrast to Ngf and Gdnf, the expression of brain-derived neurotropic factor (BDNF) gene (Bdnf) decreased at 3 dpi (p<0.001) and 7 dpi (p<0.05) in both genotypes (data not shown). So, the third group of genes with altered expression in Nrf2-deficient mice after contusion SCI was growth factors, as upregulation of Ngf was reduced in Nrf2^−/− mice when compared with WT. At the same time we detected a clear trend toward delayed Gdnf upregulation in Nrf2-deficient mice.

Finally, we studied the expression and protein levels of number of pro- and anti-inflammatory cytokines. The expression levels of tumor necrosis factor-alpha (TNFα) gene (Tnf), interleukin-1 beta (IL1β) gene (Il1b), interleukin-6 (IL6) gene (Il6), and interleukin-10 (IL10) gene (Il10) were increased at 3 dpi in both WT and Nrf2^−/− mice when compared with sham mice (p<0.001) (data not shown). At 7 dpi the level of Il1b and Il6 gene expression in animals with SCI reached the level of sham mice, while Tnf and Il10 remained significantly upregulated (p<0.001). At this time point the expression of Tnf was higher in Nrf2^−/− mice (p<0.001) when compared with WT mice (Fig. 5E). The Cytometric Bead Assay revealed a significant elevation in IL6 (p<0.001) and decline in IL2, IL17α (p<0.001) protein levels in the SCs of both genotypes in comparison with sham mice. Of this, Nrf2^−/− mice exhibited a significant increase in IL6 level (p<0.001) in comparison with WT mice (Fig. 5F). We noticed that from the investigated cytokines the expression of pro-inflammatory Tnf and Il6 was strongly upregulated in Nrf2-deficient mice at the time when the impaired recovery of these mice from SCI became evident. It is well known that upregulation of pro-inflammatory cytokines substantially contributes to secondary injury following SCI and it is believed that Nrf2 attenuates pro-inflammatory cytokines by inactivation of NF-κB. Indeed, the absence of Nrf2 resulted in greatly increased NF-κB activity in the acute phase of SCI. Moreover, sulforaphane treatment blocked the expression of pro-inflammatory cytokines via influencing the NF-κB signaling pathway (5, 9). As astrogliosis was apparently starting to increase in the SCI tissue during that time, eventually resulting in twofold expression of GFAP, it is possible that the increased inflammatory response in Nrf2-deficient mice is secondary to activation of astroglia, the major cell type contributing to inflammation after CNS injury and expressing Nrf2.

Collectively, these results suggest new aspects of Nrf2 action and indicate that Nrf2 is essential for the normal recovery from contusion SCI. Therefore, we hypothesized that enhancing Nrf2 expression by gene transfer of Nrf2 gene directly into the injured SC may be a promising strategy in SCI treatment.

Stable and Efficient Lentivirus-Mediated Gene Delivery to SC Neurons and Astrocytes

To investigate the efficiency of lentivirus-mediated gene transfer, a lentiviral vector carrying green fluorescent protein (LV-GFP) was injected into SCs of laminectomized mice. The pattern of GFP fluorescence was assessed 7 days later. GFP fluorescence was mainly localized to the grey matter (Fig. 6A).

OPEN IN VIEWER

FIG. 6.

Neurons and astrocytes but not microglia are efficiently transduced with LV-GFP in the spinal cord. Representative images of SCs of laminectomized mice subjected to GFP gene transfer (A). SC section stained with nuclear DAPI stain. Neurons (B) and astrocytes (C) are transduced with LV-GFP. Confocal microscope images from SC sections of laminecomized mice injected with LV-GFP stained with neuronal marker NeuN (A) and astrocytic marker GFAP (B) 7 days after injection. Arrows show colocalization of GFP fluorescence with cell-specific markers. LV-GFP, lentiviral vector carrying green fluorescent protein. To see this illustration in color, the reader is referred to the Web version of this article at www.liebertpub.com/ars

Staining of SC sections from GFP-transduced mice with neuronal marker NeuN and astrocyte marker GFAP revealed that both neurons and astrocytes were transduced by the viral vector (Fig. 6B, C). There were no GFP-positive microglia, as assessed by immunostaining for CD45 (data not shown).

To investigate the stability of lentivirus-mediated transfer, the expression of the transgene was assessed in SCs of mice 28 days after virus delivery. Intraspinal delivery of the lentiviral vector carrying human Nrf2 (LV-Nrf2) resulted in stable long-term expression; 28 days after LV-Nrf2 injection the expression of human Nrf2 was 350-fold higher when compared with background signal in LV-GFP group (data not shown).

Nrf2 Overexpression in Contused SC Protects from Lentivirus Toxicity and Improves Behavioral Recovery Compared with Lenti-GFP, but not Vehicle

To assess the effect of lentivirus-mediated delivery of Nrf2 on behavioral recovery we next compared the BMS scores of injured mice injected with LV-Nrf2 with BMS scores of mice injected with either LV-GFP or the vehicle, phosphate-buffered saline (PBS) (Fig. 7A). Starting from 14 days after the injection of viral vector, the locomotor performance of mice injected with either LV-Nrf2 or LV-GFP was significantly worse when compared with mice injected with PBS (p<0.001). At 28 dpi mice in the control group treated with PBS in average exhibited plantar placement of the paw, dorsal stepping, or occasional plantar stepping. At the same time we did not observe a difference between BMS scores of mice injected with LV-GFP and mice injected with LV-Nrf2. In contrast to PBS-treated mice, mice treated with any lentiviral vector on average demonstrated only extensive ankle movements.

OPEN IN VIEWER

FIG. 7.

LV-Nrf2 and LV-GFP worsen locomotor recovery compared with PBS. Thirty-three percent increase in injected virus leads to Nrf2 protection from lentiviral toxicity, but does not improve locomotor recovery compared with PBS. Hindlimb motor function of injured (A, C) and uninjured mice (B, D) injected with PBS, LV-GFP, or LV-Nrf2 2 μl/spot (A, B) or 3 μl/spot (C, D) was evaluated 1, 7, 14, 21, and 28 days after SCI or laminectomy by BMS scoring. The data are shown as mean±SEM. n=9–15 for injured groups, n=5–6 for uninjured groups. *p<0.05, **p<0.01, ***p<0.001. LV-Nrf2, lentiviral vector carrying human Nrf2; PBS, phosphate-buffered saline.

Despite the significant reduction in BMS score at 14 dpi between injured mice treated with PBS and mice treated with LV-GFP or LV-Nrf2 the size of the SC lesion was not affected, as measured by magnetic resonance imaging (MRI) (data not shown).

There were no significant differences between the BMS scores of noninjected sham mice or uninjured mice injected with PBS (data not shown). Uninjured mice injected with LV-Nrf2 exhibited significantly impaired locomotor performance (p<0.001) starting from 7 dpi when compared with noninjected shams and shams injected with PBS (Fig. 7B). The BMS scores of sham mice injected with LV-GFP were significantly lower than in all other studied groups.

Further, we increased the amount of viral vector used for injection by 33%. In contrast to the previous experiment, already at 14 dpi the locomotor performance of the injured mice treated either with PBS or LV-Nrf2 was significantly better in comparison with injured mice treated with LV-GFP (p<0.001) (Fig. 7C). However, we did not observe a significant difference between BMS scores of injured mice injected with PBS and injured mice injected with LV-Nrf2 during the whole observation period.

Similar to BMS scores obtained from injured mice, we did not detect differences between the BMS scores of uninjured mice injected with PBS or LV-Nrf2, whereas sham mice injected with LV-GFP exhibited significantly impaired locomotor performance starting from 14 dpi (Fig. 7D).

We suppose that both LV-GFP and LV-Nrf2 delivered to the SC in current titers are toxic and therefore worsen hindlimb recovery after contusion SCI. Increase in the amount of introduced viral vector encoding Nrf2, although apparently leading to the protection against lentiviral toxicity, may not be sufficient to improve hindlimb recovery compared to the vehicle.

Differential Gene Expression Profile Following Gene Transfer to the Injured SC

To understand the mechanisms that underlie the functional changes after the introduction of Nrf2 or GFP into the SC after SCI, we compared the expression of Nrf2 target genes and apoptosis-related genes, and pro- and anti-inflammatory cytokines and growth factors 3 days after transduction. Human Nfe2l2 mRNA expression was 1300-fold higher in LV-Nrf2-injected animals when compared with background signal in other studied groups that do not express human Nfe2l2 (p<0.001) (data not shown). Mouse Nfe2l2 and Hmox1 mRNA were upregulated in all injured groups, not depending on treatment (p<0.001) (Fig. 8A, B), whereas Nqo1 was higher in LV-Nrf2-treated group in comparison with other treated groups (p<0.05 for LV-GFP; p<0.01 for PBS) (Fig. 8C). In contrast, Mgst1 was significantly upregulated in PBS- (p<0.05) and LV-GFP (p<0.001)- treated groups in comparison with noninjured mice (Fig. 8D).

OPEN IN VIEWER

FIG. 8.

Nrf2-target gene expression profiles after injection of PBS, LV-GFP, or LV-Nrf2 into injured SC. Relative expression of mouse (A) Nfe2l2, (B) Hmox1, (C) Nqo1, and (D) Mgst1 in comparison with uninjured shams 3 days after viral delivery, *p<0.05, **p<0.01, ***p<0.001.

Interestingly, the expression of pro-inflammatory cytokines genes such as Tnf, Il1b, and Il6 was greatly increased in both LV-GFP- and LV-Nrf2-treated groups in comparison with uninjured mice and mice treated with PBS (Fig. 9A–C). We did not observe difference between cytokine expression in LV-Nrf2 and LV-GFP injected groups.

OPEN IN VIEWER

FIG. 9.

Cytokine, pro-apoptotic gene and growth factor expression profiles after injection of PBS, LV-GFP, or LV-Nrf2 into injured SC. Relative expression of (A) Tnf, (B) Il6, (C) Il1b, (D) Casp 3, (E) Ngf, and (F) Gdnf in comparison with uninjured shams 3 dpi after viral delivery, *p<0.05, **p<0.01, ***p<0.001.

Casp3 expression was increased in all treated groups in comparison with uninjured mice (p<0.01 for PBS; p<0.001 for LV-Nrf2 and LV-GFP) (Fig. 9D). The expression of anti-apoptotic Bcl2l1 and Bcl2 was not altered (data not shown). In addition, the induction of Ngf and Gdnf was observed in the LV-GFP and LV-Nrf2 treated groups in comparison with uninjured mice and mice treated with PBS (Fig. 9E, F). There was no difference in Ngf expression in SCs transduced with LV-Nrf2 and LV-GFP, whereas Gdnf expression in the group treated with LV-Nrf2 was significantly higher (p<0.05).

Concluding Remarks and Future Directions

Exploiting the hypothesis of enhanced Nrf2 expression in injured SC we were able to achieve strong long-term overexpression of transgene throughout the grey matter, including both neurons and astrocytes. Moreover, we observed that Nrf2 overexpression induced the expression of Nqo1 and Gdnf. However, it did not result in significant improvement of functional recovery. In Nrf2^−/−mice the expression of Nrf2 is absent since the first days of embryonic development in all SC cell types, whereas the overexpression of Nrf2 was introduced only right after the contusion, and it was limited only to neurons and astrocytes, excluding microglia. Since we did not study transduction of oligodendrocytes or other cells in the periphery, it is possible that the lack of Nrf2 overexpression on the outcome of SCI is due to late timing or cell type-specific restriction of expression. Assuming that in the previous studies sulforaphane targeted the Nrf2 pathway (5, 9) and as a small molecule reached this target in all cell types present in the SC, our hypotheses are in line with previous reports on the beneficial effects of this drug in the rat model of SCI. In future studies more thorough attention and effort should be placed on transduction of oligodendrocytes with the Nrf2 gene. Also, the role of those beneficial effects of sulforaphane that are unrelated to Nrf2 would be helpful to identify.

As an important side observation, we observed that the lentivirus used for overexpression studies in current titers is toxic to the SC, possibly by increasing inflammation (TNFα, IL6, and IL1β), but without extending the volume of injured tissue or changing the balance between pro-apoptotic and anti-apoptotic genes. It is of particular interest that lentivirus-mediated gene transfer, even at same titers that was previously successfully used in the hippocampus of a mouse model of Alzheimer's disease (4), resulted in severe paralysis of intact, healthy mice. Delivery of an increased volume (by 33%) of lenti-Nrf2 into the SC of naïve or contused mice significantly reduced the toxicity of viral vector, but still was not sufficient to improve hindlimb recovery compared to vehicle. Therefore, our results suggest that the SC may be exceptionally sensitive to lentivirus or any virus-mediated gene transfer, by so far unidentified mechanism. The results also underline the importance of the optimal titer of the virus vector used to gain functional improvement to avoid the toxicity associated with the gene transfer approach.

Notes