Adeno-associated virus-mediated delivery of a recombinant single-chain antibody against misfolded superoxide dismutase for treatment of amyotrophic lateral sclerosis.

There is emerging evidence that the misfolding of superoxide dismutase 1 (SOD1) may represent a common pathogenic event in both familial and sporadic amyotrophic lateral sclerosis (ALS). To reduce the burden of misfolded SOD1 species in the nervous system, we have tested a novel therapeutic approach based on adeno-associated virus (AAV)-mediated tonic expression of a DNA construct encoding a secretable single-chain fragment variable (scFv) antibody composed of the variable heavy and light chain regions of a monoclonal antibody (D3H5) binding specifically to misfolded SOD1. A single intrathecal injection of the AAV encoding the single-chain antibody in SOD1(G93A) mice at 45 days of age resulted in sustained expression of single-chain antibodies in the spinal cord, and it delayed disease onset and extension of life span by up to 28%, in direct correlation with scFv titers in the spinal cord. The treatment caused attenuation of neuronal stress signals and reduction in levels of misfolded SOD1 in the spinal cord of SOD1(G93A) mice. From these results, we propose that an immunotherapy based on intrathecal inoculation of AAV encoding a secretable scFv against misfolded SOD1 should be considered as potential treatment for ALS, especially for individuals carrying SOD1 mutations.

Introduction

Amyotrophic lateral sclerosis (ALS) is an adult-onset neurodegenerative disorder characterized by the selective loss of upper and lower motor neurons.[1] Approximately 20% of familial ALS cases are caused by mutations in the Cu/Zn superoxide dismutase 1 (SOD1).[2,3,4] Although the mechanism by which SOD1 mutations cause selective degeneration of motor neurons is not fully understood, many lines of evidence suggest that the toxicity of mutant SOD1 is related to its propensity to misfold and to aggregate.[5,6] Furthermore, some studies suggest a possible involvement of SOD1 abnormalities in sporadic ALS cases with no SOD1 mutations.[7,8,9,10,11] For instance, oxidation of wild-type (WT) SOD1 generates misfolded proteins that may acquire the binding and toxic properties of mutant SOD1.[7,10]

The finding that mutant SOD1 can be secreted and evidence of toxicity of extracellular mutant SOD1[12] provided a rationale for testing immunization approaches for ALS treatment. An active immunization approach with recombinant mutant or WT SOD1 as immunogen was found to delay disease onset and to increase life span of SOD1G37R mice and SOD1G93A mice expressing moderate levels of mutant SOD1.[13,14] Similar results have been obtained with active immunization using an antigenic peptide that targets the dimer interface of SOD1 using SOD1G37R or SOD1G93A mice.[15] However, because of potential adverse effects of immune responses to active vaccination approaches, passive immunization strategies appear more appropriate for future human ALS clinical trials. Some monoclonal antibodies recognizing the misfolded forms of SOD1 have been tested in SOD1G93A mice.[16] Intracerebroventricular injection of one of those monoclonal antibodies in SOD1G93A mice, named the D3H5 antibody, caused reduction in levels of misfolded SOD1 in the spinal cord and prolonged the life span of SOD1G93A mice in relation to duration of treatment. The monoclonal D3H5 antibody was shown to react against various human SOD1 mutants besides SOD1 G93A, including SOD1 G37R, G127X, G85R, and D90A.[16] In addition, the D3H5 antibody also detected WT SOD1 after treatment with metal chelators that induce protein misfolding.[16] So, the D3H5 antibody acts as a probe for SOD1 misfolding whether it is caused by mutations or other alterations such as copper or zinc depletion. The activity of D3H5 antibody against central nervous system (CNS) tissue from sporadic cases of ALS remains to be investigated.

Interestingly, intracerebroventricular injection of the variable Fab fragment of the same anti-SOD1 antibody (D3H5) also slowed down disease in SOD1G93A mice, raising the possibility to engineer a single-chain fragment of variable regions from this antibody to neutralize the toxicity of misfolded SOD1. Such single-chain fragment variable (scFv) antibody should offer some advantages such as small size and low immunogenicity. Moreover, scFv antibodies can be used in gene delivery systems. Recombinant adeno-associated viruses (AAVs) are presently vehicles of choice for gene transfer in the nervous system.[17] AAV vectors provide stable and safe gene expression with minimal immune responses and broad cell type tropism. In recent years, AAV has been used successfully for gene delivery in treatment of human genetic disorders, especially in retinal disease.[18] When injected into the cerebrospinal fluid (CSF), AAV vectors were reported to confer widespread and sustained transgene expression in the CNS.[19]

Here, we report the generation an AAV vector encoding a secretable scFv antibody (AAV-scFv) to target misfolded SOD1. A single intrathecal injection of this AAV viral vector in adult SOD1G93A mice led to sustained production of secretable scFv antibodies in the spinal cord, and it significantly delayed disease onset and mortality. This therapeutic approach may be applicable to ALS cases with SOD1 mutations and perhaps to subset of sporadic ALS cases given that misfolded and aggregated SOD1 species have been detected in sporadic ALS[7,89,10,11]

Results

Generation of a scFv from the D3H5 hybridoma

Based on our finding that the Fc region of mAbD3H5 was dispensable for extending the survival of transgenic SOD1G93A mice when delivered into the CSF,[16] we planned to test whether tonic delivery of the binding domain of the antibody via virus-mediated gene transfer in the spinal cord would be effective in blocking the toxicity of misfolded human SOD1. Thus, a scFv-encoding plasmid was generated by cloning and assembly of the coding sequences of the variable immunoglobulin heavy and light chains of mAbD3H5 into a single open-reading frame encoding a hybrid protein bearing both variable immunoglobulin regions attached by a short flexible peptide linker. The DNA sequences and deduced amino acid sequences of the cloned hypervariable regions of mAbD3H5 light and heavy chains are shown in .

The scFv DNA construct was subcloned into an expression vector under the control of cytomegalovirus (CMV) promoter that provided a murine immunoglobulin (Ig) κ-secretory signal and a human c-myc epitope to facilitate detection. We analyzed the expression of scFvD3H5 after DNA transfection in HEK293T (human embryonic kidney cell line) cells by western blotting and enzyme-linked immunosorbent assay (ELISA). The scFv was detected in the 1% Triton cell lysate and in the conditioned media after transient transfection in HEK293T cells (). A 28-kDa band was detected by immunoblotting after sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using an antibody against the human c-myc tag in the recombinant scFv protein. To test whether the scFvD3H5 antibody was able to detect misfolded SOD1, we used conditioned media from cells transiently transfected with the D3H5scFv encoding plasmid as a primary antibody source for immunoblotting detection after SDS-PAGE of spinal cord lysate from SOD1G93A mice. The scFvD3H5 in the media detected specifically SOD1 species by immunoblotting as revealed by an anti-myc antibody, whereas no signal was seen with conditioned medium from mock-transfected cells (). In addition, when conditioned media containing scFvD3H5 antibodies was loaded on ELISA plate coated with recombinant human SOD1, the bound scFvD3H5 protein could be detected using HRP conjugated anti-myc ().

AAV-mediated transduction of scFv anti-SOD1 in the spinal cord of SOD1G93A mice

As described in Methods, we then generated an AAV viral vector encoding the secretable scFvD3H5 protein. To test the therapeutic effect of scFvD3H5 in ALS pathogenesis, single intrathecal injection of 3 × 109 particles of the AAV 2/1 encoding scFvD3H5 was carried out in SOD1G93A mice at 45 days of age. At 5 weeks after injection, spinal cord sections were examined by immunostaining using an anti-myc antibody for detection of scFvD3H5. Robust scFvD3H5 immunostaining was observed in the ventral spinal cord of lumbar and sacral regions. The scFvD3H5 antibody was detectable over an extensive portion of the spinal cord. Double immunostaining was also performed using anti-myc tag and anti-choline acetyltransferase (ChAT) antibodies. The myc tag immunoreactivity was detected in some of the ChAT-positive cells in ventral horn spinal cord of SOD1G93A mice injected with AAV-scFvD3H5 (,). No staining was detected in vehicle-injected SOD1G93A mice. The myc ta0 days or more after AAV injection), indicating a stable transduction of scFvD3H5 by AAV-infected cells in SOD1G93A mice (,). The presence of scFvD3H5 antibody in the spinal cord was also confirmed by immunoprecipitation of scFvD3H5 antibody from total spinal cord lysate using a monoclonal anti-c-myc as described previously[20] and then by detection of immunoprecipitated protein on immunoblot after SDS-PAGE with a second polyclonal anti-c-myc by immunoblotting. The immunoblots further confirmed the presence of scFvD3H5 in the spinal cord at 5 weeks postinjection of AAV-scFvD3H5 in SOD1G93A mice ().

One potential concern associated with AAV-mediated gene delivery is the host immune response. To investigate any possible immune response of AAV-scFvD3H5 into mice, we took advantage of a TLR2-LUC-AcGFP (Toll like receptor 2–luciferase Aequorea coerulescens green fluorescent protein) mouse model where transcriptional activation of TLR2 as reporter of inflammation can be visualized in live animals using biophotonic molecular imaging.[21] Single intraperitoneal injection of lipopolysaccharide (LPS) at a rate of 5 mg/kg in mice induced a strong induction of bioluminescense signal in spinal cord after 24 hours (Supplementary Figure S1A–C). We injected intrathecally 3 × 109 particles of the AAV 2/1 encoding scFvD3H5 or same amount of saline in TLR2-LUC-AcGFP mice. Live imaging of the spinal cord of injected mice was carried out by up to 3 weeks. Our imaging analysis did not reveal induction of inflammatory response after intrathecal injection of AAV-scFvD3H5, control virus AAV-scFvD1.3 (AAV encoding lysozyme specific single chain fragment D1.3) in TLR2-LUC-AcGFP transgenic mice (Supplementary Figure S1D–J).

Single intrathecal injection of AAV-scFvD3H5 delayed disease onset and increased life span of SOD1G93A mice

We examined the effects on life span of SOD1G93A mice of a single intrathecal injection of AAV-scFvD3H5 and vehicle at 45 days of age. AAV-scFvD3H5–injected mice had a significant extension of life span by an average of 16 days (P value of 0.0001, median survival of 159 days) when compared with SOD1G93A mice injected with vehicle and AAVscFvD1.3 (median survival of 143 days; ). No adverse effect or premature death occurred in normal nontransgenic mice injected with AAV-scFvD3H5. Moreover, the administration of AAV-scFvD3H5 significantly delayed the onset of disease as assessed by rotarod performance tests, reflex score, and body weight, compared with control virus AAV-scFvD1.3– or vehicle-injected SOD1G93A mice (–). Rotarod tests demonstrated that AAV-scFvD3H5–treated SOD1G93A animals maintained their ability to coordinate their movement for a longer period than both the control group animals from 110 days of age onward until end stage (P < 0.05 from 110 days of age onward; n = 10 animals/group) (). In addition, reflex score measurements demonstrated that AAV-scFvD3H5–treated animals maintained significantly (P < 0.05) higher reflex of hind limb from 106 to 135 days of age when compared with those animals having received AAVscFvD1.3 or vehicle injection (). Body weight measurements revealed a significantly slower loss of body weight in AAV-scFvD3H5–treated transgenic mice from 113 to 127 days (P < 0.05) (). The mean delay in the combined assessments for rotarod test and body weight was 15 days. In all motor function tests, there was no difference between vehicle-treated controls and AAV-scFvD1.3–treated SOD1G93A mice.

Correlation between scFvD3H5 titer and longevity of SOD1G93A mice

Remarkably, a single intrathecal injection of AAV-scFvD3H5 increased the life span of SOD1G93A mice by up to 40 days (up to 28%). However, there was a large variation in the survival of transgenic mice injected with AAV-scFvD3H5, which is likely due to technical difficulty in achieving consistent viral delivery via intrathecal injection in mice. This led us to assess the antibody titer of the scFvD3H5 protein in the spinal cord at end stage of disease in these mice. Again, we took advantage of the presence of the human c-myc tag on the protein to design ELISA test for dosing scFvD3H5 proteins in spinal cord lysates of treated SOD1G93A mice. We found that the survival time of SOD1G93A mice directly correlated with the titer of scFvD3H5 protein in spinal cord extract at end stage of disease (r = 0.8312; P < 0.0001) (). The mice that survived for the longest time showed higher titers of scFvD3H5 in their spinal cord tissues.

SOD1G93A–mediated disease is associated with onset of early neuronal stress

Synthesis of mutant SOD1 within motor neurons has been found to be a primary determinant of early neuronal stress, disease onset, and motor neuron injury.[22] To visualize these events from live animals and to analyze the effects of treatment of SOD1G93A with scFvD3H5 on neuronal stress in real time, we took advantage of the GAP-43-luc/gfp reporter mice recently generated and validated in our laboratory.[23] Recent studies using laser capture dissection and gene array analysis revealed that GAP-43 is upregulated in the motor neurons of SOD1G93A mice in the presymptomatic and early symptomatic stage of disease[24], suggesting that GAP-43 may represent a good biomarker to visualize early neuronal stress/damage in ALS mice. Double-transgenic GAP-43-luc/gfp;SOD1G93A mice were generated by crossing heterozygous mice carrying the mutant SOD1G93A transgene with the heterozygous GAP-43-luc/gfp mice coexpressing reporter transgene, luciferase (luc) and green fluorescent protein (gfp), driven by the murine GAP-43 promoter. In this mouse model, a GAP-43 upregulation (luciferase expression detectable as a bioluminescence/photon emission and gfp expression detectable by confocal microscopy) can be followed longitudinally in live animals using bioluminescence/biophotonic imaging and a high-sensitivity/high-resolution charge-coupled device camera (). The quantitative analysis of biophotonic signals revealed significant early and presymptomatic upregulation of the GAP-43 signals starting at 8 weeks/56 days and reaching the peak at 11 weeks/77 days of age in the spinal cord of SOD1G93A mutant mice when compared with WT littermates (Supplementary Figure S2A). Fluorescence microscopy of gfp and anti-NeuN has been carried out to identify neurons as being the cells expressing induced GAP-43-luc/gfp transgene (Supplementary Figure S2B). Furthermore, as previously reported,[22] the ATF3 is upregulated in spinal motor neurons of SOD1G93A mice (). The detection of ATF3 in gfp-positive motor neurons in the spinal cord sections of GAP-43-luc/gfp/SOD1G93A mice further confirmed the validity of the GAP-43-luc/gfp transgene for live analysis of neuronal stress in this mouse model ().

AAV-scFvD3H5 treatment reduces neuronal stress

Series of live imaging experiments and quantification of signals in GAP-43-luc/gfp;SOD1G93A mice injected with AAV-scFvD3H5 or vehicle revealed that scFvD3H5 expression resulted in a significantly weaker bioluminescence signal in the spinal cord at 5 and 6 weeks after injection (11 and 12 weeks of age) compared with vehicle-treated double-transgenic mice (,). The reduction of neuronal stress in mice was further confirmed by examination of spinal cord sections from AAV-scFvD3H5–treated and untreated SOD1G93A mice after immunofluorescence processing to detect the presence of cyclic AMP-dependent transcription factor activating transcription factor 3(ATF3), a marker for neuronal stress[25] (). Thus, the ATF3-like immunofluorescence was much weaker in the spinal neurons of scFvD3H5-treated SOD1G93A mice at 6 weeks after injection as compared with that of untreated mice. The combined data from GAP-43-luc/gfp transgene and ATF3 expression signals demonstrate that treatment with AAV-scFvD3H5 reduced the neuronal stress associated with ALS pathogenesis in SOD1G93A mice.

Alleviation of gliosis and reduction in burden of misfolded SOD1

In addition to motor neuron loss, gliosis in the spinal cord is a prominent pathological feature in ALS patients and in rodent models of ALS.[26,27] Reactive astrocytes and microglia are characterized by an upregulation of Iba1 (Ionized calcium binding adaptor molecule 1) and of GFAP (glial fibrillary acidic protein) in the disease process.[26] We therefore examined the effect of scFvD3H5 treatment on microglial and astroglial activation in SOD1G93A mice at 120 days of age. As shown in ,, Iba 1 immunofluorescence in spinal cord samples from scFvD3H5-treated mice was 29% lower than in control samples (significant difference, P = 0.0438). Similarly, astrocyte reactivity as determined by GFAP immunodetection was 38% weaker in SOD1G93A mice injected with AAV-scFvD3H5, also a statistically significant difference (P = 0.0369) (,).

To examine the effect of scFvD3H5 treatment on the levels of misfolded SOD1 species in affected tissues, we carried out immunoprecipitation of misfolded SOD1 from total spinal cord extract at postnatal day 120, using the anti-misfolded SOD1-specific B8H10 antibody[16] followed by SDS-PAGE and immunoblotting using a polyclonal anti-SOD1 antibody. The immunoblots revealed a 19% reduction (P = 0.0293) in the levels of misfolded SOD1 species immunoprecipitated by the B8H10 antibody from spinal cord extracts of AAV-scFvD3H5–injected SOD1G93A mice in comparison with the levels detected in vehicle-injected SOD1G93A mice (,). In addition, comparison of immunoreactivity scores of spinal cord sections measured by fluorescence microscopy revealed marked reduction of anti-misfolded SOD1 antibody signals (B8H10 and C4F6) from samples of AAV-scFvD3H5–injected mice when compared with controls (n = 3; P = 0.01) (,). From these results, we conclude that the AAV-mediated delivery of scFvD3H5 antibodies succeeded in reducing the levels of misfolded SOD1.

Attenuation of motor neuron loss in SOD1G93A mice injected with AAV-scFvD3H5

Motor neuron loss correlates with disease severity in transgenic mice expressing mutant SOD1.[2] Therefore, we investigated the effect of the scFvD3H5 treatment on motor neuron survival in SOD1G93A mice by quantitative evaluation of the number of large Nissl-positive neurons in the ventral horn region of lumber spinal cord sections (). At 120 days of age, the SOD1G93A mice injected at 45 days with the AAV-scFvD3H5 vector exhibited 28% more motor neurons in the lumbar spinal cord than the vehicle-injected SOD1G93A mice (38.67 ± 0.67 per hemi section in AAV-scFvD3H5–treated mice versus 27.67 ± 2.19 in vehicle-treated mice, P = 0.0086) ().

Progressive axonal degeneration is also a hallmark of mutant SOD1 expression in transgenic mice.[28] Transverse sections from the L5 ventral root of AAV-scFvD3H5– and vehicle-injected SOD1G93A mice at 120 days of age were collected and analyzed for the number of myelinated axons (). In sections from AAV-scFvD3H5–treated SOD1G93A mice, 30% more myelinated axons were detected when compared with sections from vehicle-treated SOD1G93A mice () (582.3 ± 57.75 versus 407.0 ± 10.12) in AAV-scFvD3H5–treated and vehicle-treated mice, respectively P = 0.0403).

Discussion

Here, we report the testing of a novel gene therapy approach for ALS based on AAV delivery of recombinant secretable scFv antibody specific to misfolded SOD1 species. This secretable scFv antibody was derived as described below from a hybridoma cell line expressing a monoclonal antibody called D3H5, which is specific to misfolded SOD1.[16] A single intrathecal injection of AAV-scFvD3H5 in SOD1G93A mice at 45 days of age yielded sustained expression of scFv molecules in the spinal cord. The treatment extended survival by an average of 16 days and by up to 40 days in direct correlation with the levels of antibody detected in tissue. It is also noteworthy that the approach succeeded in delaying disease onset and in attenuating gliosis.

Recent studies reported that intrathecal injection of AAV vectors yielded strong levels of biodistribution and motor neuron transduction in the spinal cord.[19] Here we tested this method to achieve high level and sustained production of secretable scFv antibodies in the spinal cord. Our results show that 5 weeks after single intrathecal injection of AAV-scFvD3H5 into SOD1G93A mice, there was widespread immunodetection of scFv antibodies throughout the spinal cord including within ChAT-positive motor neurons (). The expression of scFv antibodies was persistent until end stage of disease, as revealed by anti-myc immunofluorescence and ELISA (). We noticed variation in scFv antibody titers in the spinal cord from one mouse to another (), phenomenon due in part to inherent technical difficulty to inject intrathecally in a small animal constant amount of AAV vectors. Nonetheless, the data revealed that longevity of SOD1G93A mice correlated directly with the scFv antibody titer measured by ELISA in whole spinal cord lysates at end stage of disease. It is noteworthy that this immunotherapy approach via AAV delivery system extended survival of SOD1G93A mice by up to 40 days. This would rank among the best therapeutic interventions accomplished so far in this mouse model of ALS with high copy number of mutant SOD1 transgene. There is also a recent report that AAV9 delivery of short hairpin RNA in SOD1G93A mice to suppress SOD1 expression was similarly effective in conferring protection.[29] It should be noted that knocking down SOD1 levels in astrocytes from sporadic ALS was also found to attenuate astrocyte-mediated toxicity to motor neurons.[30] Nonetheless, an immunotherapeutic approach to target misfolded SOD1 species is more selective than short hairpin RNA approach[29] or an antisense oligonucleotide therapy, which aims to reduce messenger RNA levels encoding both mutant SOD1 and WT SOD1.[31]

The success of this therapeutic strategy lies in the fact that scFv antibodies are targeting misfolded SOD1 species, the source of toxicity in the disease.[32,33,34] Immunoprecipitation experiments from spinal cord lysates at end stage of disease mice confirmed that treatment led to a reduction in amount of misfolded SOD1 species (). The results also suggest a direct beneficial effect of scFv immunotherapy antibody on motor neurons. Bioimaging analysis of GAP-43/luc/gfp; SOD1G93A mice treated with AAV-scFvD3H5 revealed reduction of neuronal stress at presymptomatic stage (). Symptoms monitoring also indicated a postponing of disease onset in SOD1G93A mice injected with AAV-scFvD3H5 (–). Such alleviation of motor neuron damage by scFvD3H5 antibody associated with delayed disease onset is consistent with previous studies on genetic ablation of mutant SOD1 within various cell types, which concluded that toxicity of mutant SOD1 exerted within motor neurons constitute a primary determinant of disease onset.[35]

A common characteristic of ALS is the occurrence of a neuroinflammatory reaction consisting of activated glial cells, mainly microglia and astrocytes. Previous studies have demonstrated that activated microglia induces the production of neurotoxic factors such as superoxide, nitric oxide, and proinflammatory cytokines, whereas reactive astrocytes express inflammatory markers such as iNOS.[36,37,38] It is now well established that the surrounding cells are active players in motor neuron dysfunction and death in mouse model of ALS as well as in familial ALS andsporadic ALS.[39,40] Here, the scFv treatment succeeded in attenuating the activation of microglia and astrocytes in the spinal cord of SOD1G93A mice (,), an effect that may contribute to neuroprotection.

These results suggest that AAV-based delivery of scFv antibodies against misfolded SOD1 species should be considered for treatment of ALS patients bearing SOD1 mutations. Unlike other drugs that act on deleterious pathways or cell survival, such immunotherapy-based approach can target specifically misfolded SOD1 species, the primary cause of toxicity in the disease. Moreover, there is evidence that aggregated forms of SOD1 can seed misfolding and aggregation of native WT SOD1 protein[41,42,43] Thus, the beneficial effects of antibodies might come not only from clearance of pathogenic SOD1 molecules but also from neutralization of toxic epitope exposed by misfolded SOD1 or from interference in formation and propagation of misfolded SOD1 species.[43] Moreover, another advantage of the AAV delivery approach is that it can achieve high level and sustained expression of secreted scFv antibodies without repeated injections of antibodies. Finally, the small size of scFv antibodies lacking Fc fragment allows them to thoroughly penetrate CNS tissues without adverse immune system response. It should be noted that secreted scFv antibodies may confer beneficial effects not only through binding of extracellular misfolded SOD1 but also through binding of intracellular SOD1 species as a previous study provided evidence of internalization of antibodies within motor neurons.[16] Here, scFv antibodies were detected in subsets of motor neurons ().

An immunotherapy based on intrathecal inoculation of AAV to transduce in the CNS secretable scFv antibodies against misfolded SOD1 would be applicable for treatment of human ALS. Recombinant AAV vectors are vehicles of choice for gene transfer in the nervous system.[17,44] AAV can provide stable and safe gene expression with minimal immune responses. In recent years, AAV delivery of genes has been used with success in treatment of human genetic disorders, especially in retinal disease.[18,45] When injected into the CSF, AAV vectors can confer widespread and sustained transgene expression in the CNS.[17,19] Intrathecal injection in human ALS patients would require less AAV vectors than systemic injection and it would target production of scFv antibodies to the CNS, an immunoprivileged location. In the gene therapy approach described here, the question of what CNS cell types express the scFv antibody is not of crucial importance because the scFv protein has been engineered with an immunoglobulin secretion signal. The crucial point in this approach is the amount of scFv antibodies being secreted in the spinal cord milieu to maximize the benefits. As shown in , there is a direct correlation between the amount of scFv antibodies in the spinal cord milieu and therapeutic effects. Thus, the transduction of secretable scFv molecules in various CNS cell types would be advantageous to achieve high levels of scFv antibodies in the milieu. Here, we have targeted through intrathecal injection expression of scFv antibodies to the CNS. There is no evidence that a peripheral sink for misfolded SOD1 would be beneficial as in Alzheimer's disease. An active immunization approach in SOD1G93A mice failed to delay onset of disease and to delay survival,[15] whereas passive immunization based on intracerebroventricular infusion of anti-misfolded SOD1 antibody[16] or AAV-mediated delivery of scFv antibodies as shown here succeeded in conferred therapeutic effects.

An AAV-based immunotherapy to target misfolded SOD1 would be applicable for treatment of ALS patients carrying SOD1 mutations. A prophylactic treatment would even be conceivable for individuals carrying ALS-linked SOD1 mutations since the approach is capable to delay the onset of disease. Moreover, potential application of such immunotherapy to subsets of sporadic ALS cases cannot be excluded at this time as recent studies suggest the existence of misfolded/aggregated forms of SOD1 in sporadic ALS cases with no SOD1 mutations.[10,11,46,47]

Methods

Messenger RNA was isolated from D3H5 hybridoma cell line[16] using a messenger RNA isolation kit (Qiagen, Chatsworth, CA). cDNA was synthesized using Superscript First Strand, catalog no. 12371-019 (Invitrogen,Carlsbad, CA) with oligodT priming according to the manufacturer's instructions. The variable regions of heavy chain (VH) and light chain (Vκ) were amplified separately from first-strand cDNA by using a mixture of universal polymerase chain reaction (PCR) primers and Platinum Pfx DNA polymerase (Invitrogen). The PCR products for heavy chain and light chain were cut with restriction enzymes PstI/BstEII and SacI/XhoI, respectively and agarose gel-purified. The cDNA inserts corresponding to VL and VH were cloned into the pBZUT7 vector and sequenced using M13 forward primer 5′-GTAAAACGACGCCCAG-3′ and reverse primer 5′-CAGGAAACAGCTATGAC-3′. The VH and VL domains were assembled and linked together by PCR to yield the full-length scFv gene. The scFv gene that was constructed in a VH-linker-VL format together with a standard flexible 20-amino acid linker (Gly4Ser)3 was then subcloned upstream of the Myc-tagged Psw1 scFVD1.3 Tag1 expression vector to generate scFvD3H5. Each scFv contained a murine immunoglobulin (Ig) κ-secretory signal for efficient secretion and a human c-myc epitope to facilitate detection ().

HEK293 cells were cultured in Dulbecco's modified Eagle's medium. The plasmids were transiently transfected with Lipofectamine (Invitrogen) into cells. After 48 hours, cells were harvested, and culture media was collected, centrifuged at 15,000g for 15 minutes at 4 °C, and the proteins in the media and cells were separated by 14% SDS-PAGE. After electrotransfer to polyvinylidene difluoride membranes (Immobilon-P, Millipore, Bedford, MA), scFvs were visualized by an HRP-conjugated anti-myc antibody (Invitrogen).

The D3H5 monoclonal antibody has been shown to be specific only for misfolded SOD1, and it does not detect WT SOD1.[16] Western blot analysis and ELISA were performed to evaluate the immunoreactivity of scFvD3H5 to SOD1. 14% SDS-PAGE was carried out with spinal cord lysates of SOD1G93A mice. Culture media from HEK293 cells transfected with scFv construct was used as a primary antibody source. After electrotransfer to polyvinylidene difluoride membranes, scFvs were visualized with an anti-myc antibody (Invitrogen). For ELISA, plates were coated with 3 µg/ml of recombinant SOD1G93A protein. The harvested culture media was added at different dilutions to the coated plates and incubated at 4 °C overnight. On the next day, wells were washed and incubated with anti-myc-HRP antibody for 1 hour at room temperature to detect the scFv.

The AAV serotype-2/1 vector (AAV2 terminal repeats in AAV1 capsids) encoding misfolded SOD1-scFv was prepared by standard method.[48] Briefly, to generate an AAV viral vector capable of conferring the capacity to produce and secrete the mAbD3H5 scFv upon infection, the fragment encompassing the scFv expression cassette in pScFvD3H5 was excised using HindIII and EcoRV and cloned into the plasmid BluescriptII KS(+) (Stratagene, Canada). It was then recloned as a SalI/NotI fragment into the XhoI/NotI digested plasmid AAV-CMV-GFP[49] replacing the EGFP-encoding sequence and creating the pscAAV-D3H5 plasmid to be used in the production of AAV recombinant viruses. For the production of scAAV recombinant viruses, the 293T cell line was used. Cells were cultured in Dulbecco's modified Eagle's medium (Gibco, Canada) supplemented with 10% normal bovine serum, 100 U/ml of penicillin G, and 100 µg/ml of streptomycin. Plasmids used for cotransfection of 293T cells were an ITR-containing plasmid derived from scAAV-CMV-GFP containing a scFv-encoding transgene (AAV-scFv or AAV-D1.3) and plasmids pXR1 and pxx-6 as packaging and helper plasmids, respectively.[48,49]

Transgenic mice overexpressing SOD1G93A mutations (B6SJL-TgN_ [SOD1-G93A] _1 Gur) were purchased from the Jackson Laboratory (Bar Harbor, ME) and were genotyped in accordance with Jackson Laboratory protocols. To confirm that the transgene copy number of SOD1G93A was not altered in the mice used for this study, we evaluated genomic SOD1 levels by quantitative real-time PCR using genomic DNA isolated from tail tissue. Analysis of the mouse housekeeping gene encoding glyceraldehyde-3-phosphate dehydrogenase was used for normalization purposes. Oligoprimer pairs (used at concentration of 300 nm) were designed by GeneTool 2.0 software (Biotools, Edmonton, AB), and their specificity was verified by blast in the GenBank database. Standard cycling conditions were used.

Mice were maintained heterozygous in the C57BL/6 background. The transgenic GAP-43-luc/gfp reporter mice were generated as described previously.[23] These mice were crossed with the SOD1G93A transgenic mice (C57/BL6, Jackson labs) to generate double-transgenic GAP-43-luc/gfp/ SOD1G93A mice. To avoid the effects of genetic background, all experiments were performed on age-matched littermates. Double-transgenic mice were genotyped according to the following procedure. The presence of GAP-43-luc/gfp transgene was assessed by PCR of the luciferase reporter gene with the following primers: 5′-GGCGCAGTAGGCAAGGTGGT and 5′-CAGCAGGATGCTCTCCAGTTC as described previously.[21] The transgenic TLR2-LUC-AcGFP mice were generated and genotyped as described previously.[21] The presence of the SOD1-G93A mutant transgene was assessed by PCR as previously described.[50] The use and maintenance of the mice described in this article was performed in accordance to the Guide of Care and Use of Experimental Animals of the Canadian Council on Animal Care.

In vivo As previously described, images were gathered using IVIS 200 Imaging System (Xenogen, Alameda, CA).[21] Twenty minutes before imaging session, the mice received intraperitoneal injection of D-luciferine, a luciferase substrate (150 mg/kg; Xenogen) dissolved in 0.9% saline. The mice were then anesthetized with 2% isoflurane in 100% oxygen at a flow rate of 2 l/ min and placed in the heated, light-tight imaging chamber. Images were collected using high-sensitivity charge-coupled device camera with wavelengths ranging from 300 to 600 nm. Exposition time for imaging was 1 minute using different field of views and F/1 lens aperture. The bioluminescence emission was normalized and displayed in physical units of surface radiance, photons per second per centimeter squared per steradian (photons/second/cm2/sr). The light output was quantified by determining the total number of photons emitted per second using the Living Image 2.5 acquisition and imaging software (Xenogen). Region of interest measurements on the images were used to convert surface radiance (photons/s/cm2/sr) to source flux or total flux of photons expressed in photons/second. Three-dimensional images were created using diffuse luminescent imaging tomography algorithms to reconstruct for the position, geometry, and strength of the internal light sources. The modifiable parameters were analyzed across the wavelengths, source spectrum, and tissue properties (Living Image 3D Analysis Software, Xenogen).

Intrathecal injection of AAV-scFvD3H5 vector of vehicle in SOD1G93A mice or GAP-43-luc/gfp/SOD1G93A mice was carried out at postnatal day 45. A total of 3 × 109 particles were injected in a 10 µl volume by a Hamilton syringe, which was slipped under the dura, and the vector was slowly released into the CSF. The syringe was removed after 1 minute to minimize CSF and vector leakage. For all surgeries, mice were anesthetized with 2% isoflurane and posttreated with subcutaneous buprenorphine (0.05 mg/kg) for pain. As a control, age-matched SOD1G93A mice received either same amount of vehicle or AAV vector encoding a lysozyme-specific single-chain fragment D1.3 (AAV-scFvD1.3) (n = 20 for all groups). After surgery, animals were housed in cage with free access to food and water till end point.

The onset of weight loss was determined as the time when mice started to exhibit a decline of body weight after reaching a peak value. The survival was defined as the age when the animal could not stand on feet within 30 seconds when placed on its side. Measurements of body weight, hind-limb reflex, and rotarod performance were used to score the clinical effects of SOD1G93A mice. The extensibility and postural reflex of the hind limbs when mice were held up with their tails were scored as described previously.[12] The SOD1G93A reflex score and body weight were measured every 2 days, beginning at 90 days. Scoring was performed in a blind manner by animal technicians who had no information about the genotype but had experience in grading SOD1G93A mice paralysis. Analysis of SOD1G93A mice disease progression was performed with an accelerated rotarod, starting at 4 rpm with a 0.25 rpm/second acceleration, and time was noted when the mice fell off the roll. Three trials were done per animal, and the mean value was calculated for statistics and graphs. Rotarod tests for SOD1G93A mice were performed once a week.

The levels of scFvD3H5 in the spinal cord lysate were determined by ELISA. Spinal cord tissues were homogenized in lysis buffer (137 × 10−3 mol/l NaCl, 20 × 10−3 mol/l Tris (pH 8.0), 1% NP40, and 10% glycerol), supplemented with Protease Inhibitor Cocktail Tablets Complete Mini (Roche Applied Science, Laval, Quebec), ultrasonicated, and then centrifuged at 4 °C for 15 min at 8,000g. The supernatant was diluted in bicarbonate/carbonate coating buffer, and 100 µl of that was used to coat the 96-well ELISA plate (Peprotech, Dollard des Ormeaux, Quebec) overnight at 4 °C, followed by washing and blocking overnight again. The plates were then incubated sequentially with biotinylated 9E10 to c-myc (Abcam, Cambridge, MA) overnight, washed followed by incubation in streptividin-HRP (BD Pharmingen, Canada) for 1 hour at room temperature, and washed again followed by incubation in tetramethylbenzidine solution for color development. The reaction was stopped with hydrochloric acid, and the absorbance at 450 nm was read. Spinal cord lysates from vehicle-injected mice were used as a control.

At the end point, the spinal cord was dissected out, rapidly frozen in liquid nitrogen and stored at −80 °C for ELISA, immunoprecipitation, and western blot analysis. Whole protein lysates from spinal cord were extracted by homogenization of the tissues in TNG-T lysis buffer (50 mmol/l Tris-HCl pH: 7.4; 100 mmol/l NaCl; 10% glycerol; and 1% Triton X) and centrifugation for 20 minutes at 9,000g at 4 °C. The soluble protein was quantified by the Lowry method. Immunoprecipitation of misfolded SOD1 was performed as previously described.[16] Briefly, beads were coated with mouse monoclonal anti-misfolded SOD1 antibody B8H10 for immunoprecipitating misfolded SOD1 and incubated overnight with whole spinal cord lysate, washed, and fractioned on 14% SDS-PAGE. Immunoprecipitation of spinal cord extracts expressing scFvD3H5 was performed by immunoprecipitating with anti c-myc tag as described earlier.[20]

Mice were anesthetized by intraperitoneal injection of 4% chloral hydrate and perfused intracardially with phosphate-buffered saline, followed by 4% paraformaldehyde, pH 7.4. Spinals cords were removed, postfixed for 1 hour, frozen in Tissue-Tek OCT embedding compound (Sakura Finetek, Torrance, CA), permeabilized in 0.25% Triton-X for 10 minutes, blocked in 3% normal goat serum for 1 hour, and then incubated with the monoclonal antibodies for 16 hours at 20 °C. Sections were then incubated overnight at 20 °C using primary antibodies, such as 1:100 mouse monoclonal anti-myc 9E10 (Abcam), 1:250 rabbit ChAT (Millipore), 1:500 rabbit polyclonal anti-GFAP (Dako, Carpinteria, CA), 1:500 rabbit anti-Iba1 (Wako Chemicals, Richmond, VA), 1:50 rabbit polyclonal ATF3 (Santa Cruz Biotechnology, Santa Cruz, CA), 1:500 mouse monoclonal NEU N (Millipore, Temecula, CA), and 1:100 mouse monoclonal B8H10 and C4F6 (Medimabs, Montreal, Canada).Visualization was made by incubating the slides with Alexa-Fluors 488 or 594 rabbit antimouse secondary antibody (Invitrogen). Dissected dorsal root ganglia were postfixed in a solution of 3% glutaraldehyde for a period of 48 hours, washed in phosphate-buffered saline, treated with 1% osmium tetra oxide for 2 hours, and dehydrated through graded alcohol solutions. Before Epon plastic embedding, dorsal root ganglia were further dissected to ensure that all ventral root axons would be sampled at a distance of 3 mm from the dorsal root ganglia cell body. Semi-thin cross-sections were stained with toluidine blue, rinsed, and cover slipped. To quantify the immunoreactivity score of immunofluorescent sections, we measured the optical densities of each staining with ImageJ software (NIH, Bethesda, MD).

Data were analyzed using Prism 5.0 software (GraphPad Software, LaJolla, CA). Behavioral data were computed by performing two-way ANOVAs (except when specified) followed by Bonferroni posttests and survival data using Mantel-Cox log-rank tests. Ventral root axon counts and immunoreactivity scores for Iba1, GFAP, and anti-misfolded SOD1-specific antibody were compared using two-tailed Student's t-tests. Data are expressed as mean ± SEM. P < 0.05 was considered statistically significant.

Figure S1. Real-time imaging of TLR2 induction after LPS/AAV-scFvD3H5 or AAV-svFvD1.3 injection in TLR2-LUC-AcGFP transgenic mice. Figure S2. Longitudinal quantsitative analysis of total photon GAP-43 signal/bioluminescence in GAP-43-luc/gfp/SOD1G93A double-transgenic mice and WT mice in spinal cord.