Research ArticleALS

A human-derived antibody targets misfolded SOD1 and ameliorates motor symptoms in mouse models of amyotrophic lateral sclerosis

See allHide authors and affiliations

Science Translational Medicine  05 Dec 2018:
Vol. 10, Issue 470, eaah3924
DOI: 10.1126/scitranslmed.aah3924

Wellderly antibodies to combat ALS

Amyotrophic lateral sclerosis (ALS) is characterized by the presence of abnormal forms of proteins such as superoxide dismutase 1 (SOD1) that accumulate in neurons, which then degenerate and die. Maier et al. have now developed a human antibody derived from a cohort of healthy elderly individuals that specifically recognizes abnormal SOD1 in postmortem spinal cord tissue from patients with ALS. When administered to animal models of ALS, the antibody delayed the onset of motor impairments, extended the survival of the animals, and reduced the degeneration of neurons and accumulation of SOD1 aggregates. These results support the further evaluation of this human antibody for the potential treatment of ALS involving misfolding of SOD1.

Abstract

Mutations in the gene encoding superoxide dismutase 1 (SOD1) lead to misfolding and aggregation of SOD1 and cause familial amyotrophic lateral sclerosis (FALS). However, the implications of wild-type SOD1 misfolding in sporadic forms of ALS (SALS) remain unclear. By screening human memory B cells from a large cohort of healthy elderly subjects, we generated a recombinant human monoclonal antibody (α-miSOD1) that selectively bound to misfolded SOD1, but not to physiological SOD1 dimers. On postmortem spinal cord sections from 121 patients with ALS, α-miSOD1 antibody identified misfolded SOD1 in a majority of cases, regardless of their SOD1 genotype. In contrast, the α-miSOD1 antibody did not bind to its epitope in most of the 41 postmortem spinal cord sections from non-neurological control (NNC) patients. In transgenic mice overexpressing disease-causing human SOD1G37R or SOD1G93A mutations, treatment with the α-miSOD1 antibody delayed the onset of motor symptoms, extended survival by up to 2 months, and reduced aggregation of misfolded SOD1 and motor neuron degeneration. These effects were obtained whether α-miSOD1 antibody treatment was administered by direct brain infusion or peripheral administration. These results support the further development of α-miSOD1 antibody as a candidate treatment for ALS involving misfolding of SOD1.

INTRODUCTION

Amyotrophic lateral sclerosis (ALS) is characterized by adult-onset motor neuron degeneration starting focally and spreading contiguously through the upper and lower motor neuron system, suggestive of an actively propagating process (1). The exact pathological mechanisms that cause ALS remain unclear. The majority of cases are sporadic (SALS), with no apparent familial predisposition. An estimated 5 to 10% of cases are familial (FALS) and are caused by mutations in one of several genes encoding C9ORF72, superoxide dismutase 1 (SOD1), TAR-DNA binding protein–43 (TDP-43), and fused in sarcoma (FUS) (2). SOD1 is a ubiquitously expressed antioxidant enzyme with important cellular functions in the detoxification of superoxide anion radicals. Mutations in SOD1 confer a toxic gain of function with cytosolic inclusions containing mutated and possibly also wild-type SOD1 in both patients with ALS (3) and transgenic animal models overexpressing human SOD1 (48).

Accumulating evidence suggests that disease progression may be driven by the spreading of SOD1 pathology in a prion-like manner, resulting in spatiotemporal transmission of SOD1 misfolding and aggregation (3, 9, 10). Recent experimental findings in transgenic animal models of ALS recapitulate the disease progression along the spinal cord and brainstem observed in patients with ALS and implicate the propagation of misfolded SOD1 containing seeds as a potential mechanism (11, 12). Cell-to-cell spreading of pathology may involve both intracellular and extracellular pools of SOD1 aggregates that are accessible to the immune system. Both active immunotherapy targeting SOD1 and intracerebral antibodies against misfolded SOD1 have been shown to ameliorate disease phenotypes in transgenic mouse models overexpressing ALS-causing SOD1 mutations (1316).

We hypothesized that neo-epitopes generated upon spontaneous misfolding of SOD1 could trigger humoral immune responses, leading to the formation of B cell memory. To test this hypothesis, we screened the repertoires of human memory B cells from a library of healthy elderly human subjects for reactivity against misfolded SOD1. We describe here the recombinant cloning, the biochemical properties, and the pharmacological effects of α-miSOD1, a recombinant human monoclonal antibody with selective high-affinity binding to misfolded SOD1 but not to physiological SOD1 dimers. On human postmortem spinal cord tissue sections, α-miSOD1 bound to aggregated SOD1 in neurons of SOD1 mutation carriers with ALS. This antibody also detected misfolded SOD1 in human postmortem spinal cord tissue sections from a large fraction of patients with SALS. Administered either peripherally or directly to the central nervous system (CNS), a chimeric version of α-miSOD1 rescued motor deficits, attenuated loss of spinal cord motor neurons, and prolonged overall survival in three different mouse models of ALS.

RESULTS

A recombinant human-derived antibody is selective for misfolded human SOD1

By screening human memory B cell repertoires of healthy elderly subjects, we generated recombinant high-affinity antibodies that selectively bound to misfolded and aggregated SOD1 (17, 18), but not to physiological SOD1 dimers isolated from human erythrocytes. Antibody α-miSOD1 bound with subnanomolar EC50 (half maximal effective concentration) to denatured SOD1 and oxidized aggregated SOD1 preparations with >1000-fold selectivity over natively folded SOD1 dimers (Fig. 1A). In contrast, a rabbit monoclonal pan-SOD1 antibody displayed equivalent binding to physiological dimers and to misfolded SOD1 conformations, confirming equivalent coating of the preparations (fig. S1A).

Fig. 1 Human α-miSOD1 antibody targets an epitope in pathologically misfolded SOD1.

Human α-miSOD1 antibody targets an epitope in pathologically misfolded SOD1 that is exposed in motor neurons of patients with FALS and SALS. (A) In a direct enzyme-linked immunosorbent assay (ELISA), α-miSOD1 bound with high affinity to denatured (EC50, ~70 pM) and oxidized misfolded (EC50, ~600 pM) human SOD1, while showing only minimal binding to the physiological SOD1 dimers. (B) In a direct ELISA, α-miSOD1 bound with similar affinity to the short bovine serum albumin (BSA)–coupled SOD1 76-82 peptide containing the minimal α-miSOD1 epitope (SOD1 73-83 peptide; EC50, ~30 pM; SOD1 76-82 peptide; EC50, ~13 pM) compared to denatured (EC50, ~33 pM) human SOD1. (C) α-miSOD1 detected misfolded SOD1 in the ventral horn of lumbar spinal cord sections in the majority of patients with SALS and patients with FALS carrying mutant SOD1 as well as selected patients with FALS carrying non-SOD1 mutations or carrying C9ORF72 mutations compared to non-neurological controls (NNC; scale bar, 50 μm). (D to G) Quantification of the area positive for α-miSOD1 binding within the cell soma of ventral horn motor neurons (intraneuronal) is shown. (D) Shown is staining of spinal cord sections from patients with SALS (n = 98), FALS non-SOD1 patients (n = 20), and FALS SOD1 patients (n = 3) relative to the staining of spinal cord sections from NNC patients (n = 41). A reduction in neuron number (E) and neuron size (F) was observed in the SALS and FALS non-SOD1 mutant group. The total parenchymal-positive area for α-miSOD1 binding (G) was increased in all groups of patients with ALS. Kruskal-Wallis, Dunn’s multiple comparison test, **P < 0.01, ****P < 0.0001. OD450 nm, optical density at 450 nm.

In addition, α-miSOD1 antibody bound with high affinity to SOD1 carrying ALS-linked mutations even in the absence of a denaturation step. SOD1 carrying the ALS mutations G37R, A4V, G93A, or G85R was bound by α-miSOD1 with EC50 values comparable to that for denatured wild-type SOD1 isolated from human erythrocytes (fig. S1C). In contrast, no binding of α-miSOD1 was detected for native recombinant wild-type SOD1; a pan-SOD1 antibody displayed equivalent binding to all SOD1 preparations examined (fig. S1D). The selectivity of α-miSOD1 was confirmed by biolayer interferometry (fig. S1, E to H) and dot-blot analyses (fig. S1I). High-affinity binding to soluble misfolded SOD1 without binding to the native SOD1 dimer was observed. No off-target binding of α-miSOD1 was observed for unrelated aggregating proteins including tau, α-synuclein, TDP-43, Aβ, transthyretin, and islet amyloid polypeptide (fig. S1J).

The α-miSOD1 epitope in SOD1 was mapped to amino acids K76-V82, corresponding to a region in loop IV of native SOD1 (fig. S1, K and M). A corresponding peptide was targeted with an EC50 comparable to that of denatured full-length human SOD1 (Fig. 1B). Alanine replacement of individual amino acids identified K76, D77, E79, and V82 as essential residues for α-miSOD1 binding (fig. S1N). No binding was detected to mouse SOD1 (fig. S1O) consistent with a K76A exchange in the mouse ortholog of SOD1.

α-miSOD1 detects pathologically misfolded SOD1 in human postmortem ALS spinal cord tissue sections

On postmortem lumbar spinal cord sections from patients with FALS carrying the SOD1 A4V mutation, α-miSOD1 antibody bound to skein-like inclusions and numerous swollen axons in ventral horn motor neurons (Fig. 1C). In a set of 98 SALS spinal cord tissue samples collected by six independent European and U.S. brain banks, α-miSOD1 detected diffuse and granular cytoplasmic and neuritic inclusions in ventral horn motor neurons (Fig. 1, C to G). The proportion of immunoreactive motor neurons varied from only a few to all motor neurons (Fig. 1C and figs. S2 to S6) and was often observed in Clarke’s column neurons, which also degenerate in ALS. A similar staining pattern was observed in several FALS spinal cord tissue sections, where the patients carried C9ORF72 hexanucleotide repeat expansions or unknown genetic mutations (FALS non-SOD1; Fig. 1C and figs. S2, S3, and S5). No or only weak immunoreactivity was observed in motor neurons of most of the 41 spinal cord tissue samples from NNC patients (Fig. 1C and figs. S2 to S6).

A quantitative analysis of the area occupied by α-miSOD1 staining within the soma of ventral horn motor neurons (intraneuronal staining; Fig. 1D) revealed a significant 2.7-fold increase in SALS spinal cord specimens compared to those from NNC (P < 0.0001). This suggested increased accumulation of misfolded SOD1 in a subset of patients with SALS in the absence of SOD1 mutations. Significantly increased α-miSOD1 immunoreactivity was found in four of the five subanalyses conducted on tissues provided by independent brain banks (P < 0.05; table S2). Patients with FALS carrying non-SOD1 mutations including those carrying mutations in C9ORF72 (10) and FUS (1) or carrying unknown genetic mutations (9) showed a nonsignificant 1.7-fold increase in α-miSOD1–positive soma immunostaining compared to NNC (P = 0.15) (Fig. 1D). A similar increase in α-miSOD1–positive soma immunostaining was observed in postmortem spinal cord specimens from three patients with FALS carrying SOD1 mutations (FALS SOD1) (Fig. 1D). This apparently low-percentage area of immunoreactivity in FALS SOD1 specimens was consistent with the dense skein-like aggregate pathology occupying only parts of the motor neuron soma in these specimens.

As expected, significantly lower numbers of spinal cord motor neurons and a trend toward a smaller average neuron size were observed in postmortem spinal cord tissue samples from patients with SALS and those with FALS carrying non-SOD1 mutations compared to those from NNC (Fig. 1, E and F). A receiver operating characteristics (ROC) analysis of neuronal immunoreactivity revealed that the fold change compared to controls had an area under the ROC curve of 0.79 (95% confidence interval, 0.71 to 0.86; P < 0.0001, fig. S7A). If a twofold increase was selected as a threshold corresponding to sensitivity 90% (77 to 97), a specificity of 51% (41 to 61) was calculated. Quantification of the α-miSOD1–positive area of the entire ventral horn revealed a 2.5-fold increase in SALS spinal cord samples (P < 0.01) and a trend for a 1.6-fold increase in FALS non-SOD1 specimens (P = 0.051) compared to samples from NNC. This suggested that the accumulation of misfolded SOD1 species could extend beyond the soma of motor neurons (Fig. 1G) into neuronal processes. No prominent overlap of α-miSOD1 immunoreactivity was detected with glial fibrillary acidic protein (GFAP)–positive astroglia or Iba1-positive microglia using double immunohistochemical analysis; there was only occasional association of α-miSOD1 with staining for glial markers (fig. S7, H and I).

α-miSOD1 antibody selectively targets pathologically misfolded SOD1 in mouse models of ALS

In lumbar spinal cord sections from transgenic mice carrying SOD1G93A or SOD1G37R human mutations, α-miSOD1 prominently stained SOD1 pathology including intracellular dispersed inclusions, diffuse cytoplasmic structures, larger extracellular SOD1 aggregates, and vacuolar structures (Fig. 2, A and B). α-miSOD1–positive dense aggregates were detectable on spinal cord sections from presymptomatic SOD1G93A transgenic mice at 30 days of age. These α-miSOD1–positive dense aggregates increased with age as motor impairments progressed, with additional staining appearing in vacuolized axonal structures at end stage (Fig. 2B, arrows). No staining was detectable in spinal cord sections from age-matched wild-type mice (Fig. 2B).

Fig. 2 α-miSOD1 selectively targets pathologically misfolded SOD1 in the spinal cords of SODG93A and SODG37R transgenic mice.

(A) Shown is SOD1 pathology in the spinal cords of SODG93A and SOD1G37R transgenic mice in the terminal stages of disease stained with α-miSOD1 antibody. (B) α-miSOD1 immunoreactivity was detectable in the spinal cords of SOD1G93A transgenic mice at 30 days of age. These α-miSOD1–positive dense aggregates increased with age as motor impairments progressed, with additional staining appearing in vacuolized axonal structures (black arrows) at the terminal stage of disease (scale bar, 100 μm). No staining was detectable in the spinal cords from age-matched wild-type mice. (C) Immunoprecipitation (IP) with α-miSOD1 antibody captured SOD1 specifically in spinal cord but not in kidney homogenates of SODG93A and SOD1G37R transgenic mice. A pan-SOD1 antibody captured SOD1 in both tissues. (D) α-miSOD1 immunoprecipitated SOD1 in spinal cord homogenates from 160-day-old SODG93A mice in the terminal stages of disease and, to a lesser extent, in 40-day-old SODG93A animals; no SOD1 was precipitated from kidney homogenates. The pan-SOD1 antibody (Sigma-Aldrich S2147) captured SOD1 in all three homogenates tested, whereas the isotype control antibody did not pull down SOD1. (E) α-miSOD1 detected NP40 detergent–insoluble SOD1 aggregates that accumulated in an age-dependent and tissue-specific manner as revealed by filter retardation assays. WB, Western blot.

The selectivity of α-miSOD1 antibody for pathologically misfolded SOD1 in spinal cord tissue sections from SOD1G93A and SOD1G37R transgenic mice was further assessed by immunoprecipitation from tissue extracts. α-miSOD1 efficiently precipitated SOD1 from spinal cord homogenates from symptomatic SOD1G93A and SOD1G37R transgenic mice (Fig. 2C, second lane), but not from kidney homogenates with comparable amounts of total SOD1 (19). In contrast, a pan-SOD1 antibody precipitated comparable amounts of SOD1 from all homogenates (Fig. 2C, third lane). Misfolded SOD1 was detectable by α-miSOD1 immunoprecipitation in spinal cord homogenates from presymptomatic SOD1G93A transgenic mice as early as 40 days of age (Fig. 2D). α-miSOD1 also recognized large, NP40 detergent–insoluble SOD1 aggregates, which accumulated in an age-dependent and tissue-specific manner as revealed by filter retardation assays (Fig. 2E). As expected, such aggregates were most abundant in spinal cord homogenates from end-stage SOD1G93A transgenic mice (Fig. 2E, bottom part), with lower amounts detectable in brain but not in liver or kidney homogenates. Western blotting with α-miSOD1 antibody after SDS–polyacrylamide gel electrophoresis of spinal cord homogenates from SOD1G93A transgenic mice, patients with SALS, or human NNC showed detection of a single band running at 16 kDa, corresponding to the denatured SOD1 monomer (fig. S7F and G). This demonstrated the target specificity of α-miSOD1 antibody.

CNS infusion of α-miSOD1 into SOD1G93A transgenic mice shows beneficial effects

The effects of chronic intracerebroventricular infusion of α-miSOD1 were assessed in SOD1G93A transgenic mice, with antibody delivered by osmotic minipumps starting at 60 days of age and continuing until the end stage of disease. To avoid mouse antihuman antibody responses upon chronic dosing, a murine immunoglobulin G2a (IgG2a) chimeric derivative, chα-miSOD1, was used. In this aggressive mouse model of ALS with >17-fold overexpression of the SOD1G93A transgene (7), mitochondrial morphological alterations were described as early as 7 days of age (20), followed by denervation of neuromuscular junctions and the first pathological gait changes at day 50, whereas impairments in classical motor performance tests manifested later around day 90 (21).

High-speed video gait analysis revealed significant improvements in gait abnormalities in SOD1G93A mice treated with chα-miSOD1 antibody at 80 days of age compared to vehicle-treated littermates (P < 0.05). chα-miSOD1–treated animals showed an improved upright gait and a steeper movement pattern of the hindlimb as indicated by wider maximum (Fig. 3A; P < 0.05) and minimum angles (fig. S7J; P < 0.05) during maximum protraction between the iliac crest and the ankle of the hindlimb at 80, 90, and 100 days of age. Furthermore, the mean horizontal position and maximum protraction of the hindlimb toe were improved following chα-miSOD1 infusion at 80, 90, 100, and 140 days (Fig. 3B and fig. S7K), with smaller mean toe distances of frontlimbs and hindlimbs at 80, 90, and 100 days (fig. S7L).

Fig. 3 Intracerebroventricular treatment of SOD1G93A mice with chα-miSOD1 antibody.

Intracerebroventricular treatment of SOD1G93A transgenic mice with mouse chimeric chα-miSOD1 antibody increased survival, delayed muscle atrophy, and ameliorated gait symptoms. (A and B) High-speed video gait analysis showed that gait disturbances due to muscle atrophy in the lower body of SOD1G93A transgenic mice were ameliorated by continuous intracerebroventricular infusion of chα-miSOD1 antibody 20 days after the start of treatment. This was reflected by a wider maximum angle between iliac crest and ankle during maximum protraction (A) and a more forward mean horizontal position of the hindlimb toe (B). (C) Treatment of SOD1G93A transgenic mice with chα-miSOD1 antibody from 60 days of age onward increased median survival by 7 days (P < 0.05, log-rank Mantel-Cox test). (D) Delayed muscle atrophy led to a higher body weight in chα-miSOD1-treated SOD1G93A transgenic mice. (E and F) Misfolded SOD1 in the ventral horn of the lumbar spinal cord was reduced at the end stage of disease by treatment with chα-miSOD1 antibody (scale bar, 400 μm; n = 14 for the vehicle-treated group and n = 10 for the chα-miSOD1–treated group). (G to J) Treatment with chα-miSOD1 antibody resulted in a trend toward lower numbers of Iba1-positive microglia (brown, H and I) and an increased percentage of microglia associated with SOD1 aggregates (blue, asterisks, H and J) compared to vehicle-treated control animals. (K) Drug concentrations in blood plasma and spinal cord homogenates are shown 2, 5, 10, and 20 days after 100 mg/kg ip injection of human α-miSOD1 antibody in 3-month-old SOD1G93A transgenic mice (n = 4 for each subgroup). (L to N) Human α-miSOD1 in vivo binding detects structures in the lumbar spinal cord containing high concentrations of misfolded SOD1 such as vacuolized axons and supernumerary axonal structures (black arrows), which were labeled in independent α-miSOD1 (M and N) but not in isotype control antibody–treated (L) SOD1G93A transgenic animals. *P < 0.05, t test.

Treatment with chα-miSOD1 antibody resulted in an increase in median survival of SOD1G93A transgenic mice by 7 days compared to vehicle-treated control littermates (167 versus 160 days; P < 0.05, Fig. 3C) and attenuated body weight loss in the final phase of disease (Fig. 3D). Histological analyses of spinal cord tissues at end stage revealed a significant reduction in SOD1 aggregate load by 51% in the ventral horn of the lumbar spinal cord of chα-miSOD1–treated animals compared to vehicle-treated control littermates (Fig. 3, E and F; P < 0.05). This was associated with a trend toward a lower overall number of Iba1-positive microglia (Fig. 3, G to I). In contrast, the fraction of microglia associated or overlapping with SOD1 aggregates with predominantly extracellular localization was increased in chα-miSOD1–treated animals compared to vehicle-treated control mice (Fig. 3, G, H, and J).

Peripherally administered α-miSOD1 binds to misfolded SOD1 in SOD1G93A transgenic mice

Penetration of human α-miSOD1 antibody into the CNS was assessed after a single intraperitoneal administration of antibody (100 mg/kg) to SOD1G93A transgenic mice. Two days after injection, the antibody concentration in plasma reached 239 ± 71 μg/ml and dropped by 66% 20 days after injection, corresponding to an estimated elimination half-life of 10 to 12 days. Concentrations of α-miSOD1 in the mouse spinal cord peaked at 423 ± 42 ng/g 5 days after injection, suggesting a target-mediated retention of α-miSOD1 antibody in the spinal cord of SOD1G93A transgenic mice (Fig. 3K). The corresponding spinal cord compared to plasma antibody concentration increased from 0.1% at day 2 to 0.4% at day 20 after injection and was higher than the 0.1% frequently reported for systemically administered antibodies (22). Immunohistochemical analysis of mouse spinal cord sections 2 days after antibody injection revealed in vivo binding of α-miSOD1 to structures containing large amounts of misfolded SOD1, such as vacuolized and supernumerary axons; no such staining was observed in animals treated with isotype control antibody (Fig. 3, L to N).

Peripherally administered chα-miSOD1 antibody improves survival in slowly progressing SOD1G37R transgenic mice

We next investigated the therapeutic potential of peripherally administered chα-miSOD1 antibody to transgenic SOD1G37R mice that exhibited slowly progressing motor deficits (8). The mice were treated with weekly intraperitoneal injections of chα-miSOD1 antibody (10 mg/kg), vehicle control, or an isotype-matched antibody targeting an epitope within amino acids 110 to 120 of SOD1, corresponding to loop VI of the native SOD1 structure (fig. S1M). SOD1G37R transgenic mice displayed a less-aggressive phenotype compared to SOD1G93A mice, with a fourfold transgene overexpression and development of the first signs of impaired motor coordination on the rotarod test around 400 days of age, with a median survival time of close to 500 days. Peripheral administration of chα-miSOD1 antibody delayed disease onset and ameliorated the severity of motor impairments observed during disease progression in SOD1G37R transgenic mice. Onset of symptoms, defined as a continuous drop in rotarod performance, was delayed by 49 days in treated mice compared to vehicle controls (Fig. 4A; 437 ± 8 days for vehicle; 486 ± 5 days for chα-miSOD1–treated mice; P < 0.001). Median survival of chα-miSOD1–treated mice increased by 59 to 538 days compared to a median survival of 479 days for vehicle-treated littermates (P < 0.001, Fig. 4B). chα-miSOD1–treated mice displayed higher muscular strength in the paw grip endurance (PaGE) test from 456 days of age onward and performed significantly better than vehicle-treated control animals on the rotarod test from 472 days of age onward (P < 0.001, Fig. 4, C and D). Neither of these disease parameters was significantly altered by treatment with the isotype-matched antibody compared to the vehicle-treated group.

Fig. 4 Weekly intraperiteoneal injections of chα-miSOD1 increase survival of SOD1G37R transgenic mice.

Treatment with weekly intraperitoneal injections of mouse chimeric chα-miSOD1 antibody from 90 days of age onward increased survival of SOD1G37R transgenic mice and ameliorated motor symptoms, muscle atrophy, and loss of motor neurons. (A) chα-miSOD1 treatment delayed the median onset of motor symptoms on the rotarod test by 49 days compared to vehicle-treated animals or by 55 days compared to isotype control antibody–treated animals (mean disease onset, chα-miSOD1–treated mice, 486 days; vehicle-treated mice, 437 days; isotype antibody–treated mice, 431 days; P < 0.001, log-rank Mantel-Cox test). (B) Median survival of chα-miSOD1–treated mice increased by 59 days compared to vehicle-treated control mice or by 36 days compared to isotype antibody–treated control animals (median life span of 538 days for chα-miSOD1–treated mice, 479 days for vehicle-treated mice, and 502 days for isotype antibody–treated mice; P < 0.001, log-rank Mantel-Cox test). (C and D) Treatment with chα-miSOD1 antibody delayed the onset of motor symptoms on the rotarod test (C) and in the PaGE test (D) compared to isotype antibody–treated control or vehicle-treated control mice. (E) Number of surviving motor neurons in the lumbar spinal cord was increased by more than twofold in chα-miSOD1–treated mice compared to isotype antibody–treated control or vehicle-treated control mice. (F and G) Treatment with chα-miSOD1 antibody increased gastrocnemius muscle (F) and flexor digitorum longus muscle weight (G) at the disease end point in chα-miSOD1–treated mice compared to both control groups. *P < 0.05, **P < 0.01, ***P < 0.001 [analysis of variance (ANOVA), Bonferroni’s multiple comparison test; t test at individual time points for (C) and (D)].

Quantification of neuron counts at the end stage of disease revealed a more than twofold increase in remaining neurons in the ventral horn of chα-miSOD1–treated mice compared to animals receiving vehicle or control isotype antibody injections (Fig. 4E). In addition, a significant increase of 20% in the weight of the hindlimb gastrocnemius muscle (Fig. 4F; P < 0.05) and 10% in the flexor digitorum longus muscle (Fig. 4G; P < 0.05) was observed compared to vehicle-treated animals. Mean antibody concentrations in plasma at sacrifice were 135 ± 15 μg/ml for chα-miSOD1 and 98 ± 70 μg/ml for the control isotype antibody, indicating robust and comparable plasma concentrations for both antibodies measured 1 to 7 days after the last intraperitoneal dosing.

Peripherally administered chα-miSOD1 antibody boosts survival at low doses in fast-progressing SOD1G93A transgenic mice

To assess the dose-response relationship of chα-miSOD1 antibody treatment, weekly intraperitoneal doses of chα-miSOD1 (3, 10, or 30 mg/kg) were administered in rapidly progressing SOD1G93A transgenic mice starting at 1 month of age. chα-miSOD1 treatment significantly delayed the median disease onset by 5 days for all three doses tested compared to vehicle-treated control mice (vehicle, 113 days; chα-miSOD1 groups, 118 days; Fig. 5A; P < 0.05 for 3 and 10 mg/kg, P < 0.01 for the 30 mg/kg dose). The median life span of chα-miSOD1 antibody–treated SOD1G93A mice was extended by 6, 10, or 5 days for the 3, 10, and 30 mg/kg dose groups, respectively (vehicle, 155 days; chα-miSOD1 groups, 161, 165, and 160 days; Fig. 5B; P < 0.01 for the 3 mg/kg dose and P < 0.001 for the 10 and 30 mg/kg dose). Motor improvements in chα-miSOD1–treated animals were observed on the rotarod test and in the paw grip task with significant differences (P < 0.05) between chα-miSOD1–treated and vehicle-treated animals at several time points during the longitudinal testing starting at 100 days of age.

Fig. 5 Weekly intraperitoneal dosing of SOD1G93A transgenic mice with chα-miSOD1 antibody delays disease onset.

Shown is delayed disease onset, extension of survival, and reduced muscle atrophy after weekly intraperitoneal dosing with 3, 10, or 30 mg/kg chα-miSOD1 antibody in SOD1G93A transgenic mice. (A) A delay in disease onset (peak body weight) was observed with all three doses tested (P < 0.05 for 3 and 10 mg/kg, P < 0.01 for 30 mg/kg, log-rank Mantel-Cox test). (B) Extension of survival was observed with all three doses tested (P < 0.01 for 3 mg/kg and P < 0.001 for 10 and 30 mg/kg, log-rank Mantel-Cox test). (C) Attenuation of muscle atrophy in animals with end-stage disease was observed after chα-miSOD1 treatment at all three doses compared to vehicle-treated control mice (Kruskal-Wallis, Dunn’s post hoc, **P < 0.01; *P < 0.05, t test, n = 25 to 55 per group). (D) Antibody concentrations in plasma correlated linearly with dose after weekly intraperitoneal dosing with chα-miSOD1 antibody. (E) No effect on survival was seen in SOD1G93A transgenic mice after treatment with an aglycosylated version of chα-miSOD1 (30 mg/kg) called chα-miSOD1–agly (n = 18 to 25 per group).

Muscle weight analyses revealed an increase in the weight of remaining gastrocnemius muscle by 13, 17, or 18% in the 3, 10, or 30 mg/kg chα-miSOD1 dose groups, respectively, compared to the vehicle-treated mice (Fig. 5C). Antibody concentrations were determined in the terminal bleed 1 to 7 days after the last intraperitoneal dosing. They were found to be 16 ± 4, 114 ± 11, and 427 ± 28 μg/ml for the 3, 10, and 30 mg/kg chα-miSOD1 dose groups and correlated linearly with dose (Fig. 5D). Plasma antibody concentrations in the 10 mg/kg group were in a similar range to that observed in SOD1G37R antibody–treated animals. No increase in misfolded SOD1 in plasma was observed upon intraperitoneal dosing with chα-miSOD1 antibody as detected by an ELISA specific for misfolded SOD1 using α-miSOD1 as a capture antibody and a pan-SOD1 antibody as the detector.

Antibody effector functions are required for chα-miSOD1 treatment effects

To investigate the contribution of immune effector function to the beneficial effects of chα-miSOD1 antibody, we engineered an aglycosylated chα-miSOD1 antibody variant (chα-miSOD1–agly) with impaired effector functions but with equivalent binding affinity and selectivity toward misfolded SOD1 (23). Treatment with chα-miSOD–agly (30 mg/kg) did not improve survival of SOD1G93A transgenic mice compared to vehicle-treated animals (Fig. 5E) when the identical treatment regimen and end points were used as in the SOD1G93A transgenic mouse efficacy study using fully glycosylated chα-miSOD1. Plasma antibody titers were comparable to those of fully glycosylated chα-miSOD1 antibody (Fig. 5D; 403 ± 54 μg/ml versus 427 ± 29 μg/ml), thus excluding differences in antibody concentration as a confounding factor.

chα-miSOD1 antibody reduces SOD1 pathology and neuroinflammation in B6SJL-SOD1G93A transgenic mice

To assess the effects of chα-miSOD1 antibody treatment at a defined time point before disease end stage, a study was conducted in B6SJL-SOD1G93A transgenic mice with a prespecified end point for analysis of 120 days of age. B6SJL-SOD1G93A transgenic mice on a mixed B6SJL background displayed an even faster disease course compared to the corresponding C57BL/6 mouse strain. Three months of weekly chα-miSOD1 (30 mg/kg) antibody given intraperitoneally resulted in a 66% reduction of misfolded SOD1 by immunohistochemical quantification and a 25% reduction of SOD1 aggregates quantified by the filter retardation assay (fig. S7, N and O). The reductions in SOD1 pathology were accompanied by a 37 and 43% reduction in microgliosis and astrogliosis, respectively, revealed by Mac-2 and GFAP staining (fig. S7, P and Q). These improvements in pathology were associated with a delayed disease onset in α-miSOD1–treated animals compared to vehicle-treated controls (fig. S7M). The age of disease onset correlated with the amount of SOD1 aggregates as determined by the filter retardation assay (fig. S7R), further supporting this association.

DISCUSSION

Neurodegenerative disorders are characterized by neuropathological lesions in specific areas of the CNS resulting from the misfolding, aggregation, and accumulation of endogenous proteins such as Aβ and tau in Alzheimer’s disease, α-synuclein in Parkinson’s disease, and TDP-43, C9ORF72-derived dipeptide repeat proteins, or SOD1 in ALS/frontotemporal lobar degeneration spectrum diseases (24). Here, we established in a healthy elderly donor population the occurrence of specific adaptive immune responses characterized by the presence of B cell memory against pathological conformations of SOD1. The Ig heavy- and light-chain variable sequences of corresponding antibodies were cloned from individual B cells, and human monoclonal antibodies were engineered and recombinantly expressed to include glycosylated IgG1 constant domains. We generated a panel of 20 recombinant antibodies, the majority of which bound with high affinity to epitopes exposed specifically on misfolded SOD1 and presented poor binding affinity to physiological SOD1 dimers. The presence of such high-affinity antibodies that had undergone extensive somatic hypermutation suggested that selective adaptive immune responses can be triggered by neo-epitopes exposed as a consequence of misfolding and aggregation of wild-type SOD1, for example, following unconventional secretion of SOD1 under cellular stress (25).

The following structural considerations support the observed selectivity of α-miSOD1 antibody for misfolded SOD1. In the high-resolution crystal structure of native SOD1 (Protein Data Bank ID # 2V0A0; fig. S8) (26), the loop harboring the α-miSOD1 binding epitope with the residues K76 to V82 is stretched over the protein surface with many of the side chains engaged in contacts with other residues, precluding interactions with an antibody in the native conformation (fig. S8). Several residues (R80, H81, and V82) are largely shielded from solvent with poor accessibility for binding of antibody. Although some residues have multiple side-chain conformations, the loop has very low temperature factors indicating low mobility (e.g., K76, 12.6 Å2; D77, 11.9 Å2; E79, 12.9 Å2; and V82, 9.9 Å2) and contains three important hydrogen bonds implicated in the stabilization of the SOD1 structure (27).

Our analyses of >100 healthy elderly subjects revealed the presence of B cell memory against misfolded SOD1 in a majority of donors from this cohort. This observation suggested that misfolding of SOD1 and the subsequent humoral immune response are frequent events in the elderly. A study investigating serum antibody titers against altered conformations of SOD1 revealed a correlation between serum titers and survival of patients with SALS, suggesting a protective function (28).

One high-affinity antibody candidate generated in this study, α-miSOD1, bound with subnanomolar EC50 selectively to preparations of misfolded human wild-type SOD1 but with >1000-fold lower affinity to physiological SOD1 dimers. In a comprehensive target analysis using human postmortem spinal cord sections, α-miSOD1 immunostaining revealed misfolded SOD1 in motor neurons in a majority of ALS cases with minimal immunoreactivity in postmortem spinal cord sections from NNC subjects. α-miSOD1 immunoreactivity was prominent not only in patients with FALS with SOD1 mutations but also in patients with SALS and those with FALS carrying C9ORF72 hexanucleotide repeat expansions or other mutations. This suggested that α-miSOD1 detected misfolded or aggregated SOD1 species that formed even in the absence of SOD1 mutations. In contrast, a pan-SOD1 antibody revealed a broad immunoreactivity throughout the entire ventral horn parenchyma with no obvious differences between SALS and non-neurological human control subjects, consistent with the ubiquitous expression of the physiological SOD1 dimers. Elevated misfolded SOD1 was detected in four of the subanalyses conducted with spinal cord tissues provided by different brain banks, whereas one subanalysis with generally very low immunoreactivity did not reveal a difference in α-miSOD1 immunoreactivity (table S2 and figs. S2 to S6). This may be due to a low abundance of misfolded SOD1 in this patient cohort or, alternatively, differences in tissue fixation, processing, or storage that could lead to masking of the target epitope detected by α-miSOD1. This is consistent with a recent study reporting misfolded SOD1 concentrations below the detection limit using tissue from this cohort of sporadic or FALS non-SOD1 cases after staining with a set of antibodies against misfolded human SOD1 generated in mice or rabbits (29). In our analyses, no correlation between the postmortem delay before fixation of tissue and the observed α-miSOD1 immunoreactivity was found for the autopsy material used in our study, excluding this as a major confounding factor.

The extent and the pathophysiological impact of SOD1 misfolding and aggregation in patients with ALS without SOD1 gene mutations are a focus of intense study. Consistent with the prevalent detection of non-native SOD1 species by α-miSOD1 across the ALS spectrum, a panel of rabbit antibodies raised against linear epitopes of misfolded SOD1 was reported to detect granular SOD1-immunoreactive inclusions in astrocytes, microglia, oligodendrocytes, and motor neurons in postmortem spinal cord tissue from patients with SALS and FALS (5, 30). Selected mouse monoclonal antibodies raised against a conformation of mutant SOD1G93A or the inaccessible regions in natively folded SOD1 revealed misfolded and protease-sensitive wild-type SOD1 in a subset of all patients with SALS examined (3, 4). In contrast, other mouse monoclonal antibodies raised against conformational epitopes specific for non-native SOD1 revealed predominant staining in patients with FALS harboring SOD1 mutations but not in patients with SALS (31, 32). These observed differences may in part be due to the exact SOD1 binding epitopes targeted with different exposures across the diverse spectrum of species and strains of misfolded and aggregated SOD1 (11, 12, 33). This is consistent with our own findings where different recombinant antibodies generated from healthy elderly subjects displayed different degrees of reactivity to misfolded human SOD1 in SALS spinal cord tissue, with α-miSOD1 antibody being the most prominent and most selective binder.

Wild-type and mutant SOD1 have been shown to display similar aggregation kinetics and toxicity: Both can dissociate into a common aggregation-prone monomeric intermediate (17) and follow a similar aggregation pattern, in which the rate-limiting factor is the formation of stable seeds that accelerate further aggregate formation (34). It was also shown that oxidized wild-type SOD1 acquires the same properties as mutant SOD1 and is similarly toxic to cultured motor neurons (35). Misfolded wild-type SOD1 can interfere with cellular functions, leading to inhibition of fast axonal transport, endoplasmic reticulum stress, and mitochondrial damage (4, 3638). Consistent with a pathogenic function of misfolded wild-type SOD1 in SALS, overexpression of wild-type human SOD1 in mice results in an accumulation of misfolded SOD1 in spinal motor neurons and induces an ALS-like phenotype with early mortality (6). Furthermore, pathological TDP-43 and FUS were shown to kindle the misfolding and cell-to-cell spreading of wild-type SOD1 (39), suggesting a central role for SOD1 misfolding in ALS. Recently, Trist et al. (40) presented evidence of catalytically dysfunctional, misfolded conformations of soluble and aggregated wild-type SOD1 in degenerating regions of the Parkinson’s disease brain. Further studies will be required to determine the exact contribution of misfolded SOD1 to neuronal degeneration in Parkinson’s disease.

Chronic α-miSOD1 antibody treatment ameliorated motor symptoms and improved survival in three independent transgenic mouse models overexpressing human mutant SOD1, with a very rapid (SOD1G93A) or more slowly progressing (SOD1G37R) disease course. α-miSOD1 antibody treatment ameliorated disease progression and the decline of motor function and increased survival in the three mouse models. Even when animals were euthanized at the same end stage of disease, that is, at an older median age in α-miSOD1–treated mice compared to controls, marked improvements in SOD1 aggregate load, muscle weight, and neuronal survival were observed after chronic α-miSOD1 treatment. The treatment effects were most prominent in the slower-progressing SOD1G37R mice when compared to the aggressively progressing SOD1G93A mice, in line with previous studies using active immunization to target SOD1 (1316). These differences may be driven, in part, not only by the respective severity of the SOD1 mutation but also by the transgene overexpression of >17-fold in SOD1G93A mice compared to endogenous SOD1 and about 4-fold in SOD1G37R transgenic mice (7, 8). As the pathogenicity of mutant SOD1 is proportional to its dose, it seems likely that the effect size of α-miSOD1 treatment may be considerably larger in human ALS where SOD1 is expressed at normal levels.

Treatment with α-miSOD1 was efficacious after continuous intracerebroventricular infusion and after peripheral administration of doses as low as 3 mg/kg weekly ip, demonstrating that relatively low concentrations of α-miSOD1 were sufficient to exert therapeutic effects. This is consistent with our findings with aducanumab, a recombinant human-derived antibody selective for aggregated Aβ and generated using the same technology. In mice overexpressing amyloid precursor protein, aducanumab showed a minimal effective dose of 3 mg/kg upon weekly systemic administration consistent with the clinical doses, leading to dose- and time-dependent reductions in brain amyloid load and evidence of slowing of clinical decline in prodromal patients or those with mild Alzheimer’s disease (41).

The therapeutic effects of α-miSOD1 treatment may be mediated by different nonexclusive mechanisms. These include the direct neutralization of extracellular toxic SOD1 conformers, interference with prion-like spreading of SOD1 aggregates released from dying cells or via exosome-dependent and exosome-independent pathways from living cells (42), trans-synaptic propagation (11), macropinocytotic uptake of aggregates (3, 12, 38, 43), or targeting for degradation of intracellular misfolded SOD1 species after cellular uptake of antibodies (44, 45). Our results demonstrate that α-miSOD1 can penetrate the blood–spinal cord barrier and a functional glycosylated Fc portion is required for full therapeutic efficacy, suggesting that FcγR-mediated recruitment of microglia and phagocytosis of toxic SOD1 aggregates may be a central mechanism of action. Similar findings were reported for other antibody-based experimental therapies targeting extracellular or intracellular aggregating CNS proteins (41, 46). CNS penetration by antibodies may be facilitated by impairments in the blood–spinal cord barrier as has been described in patients with ALS and animal models (47, 48).

In addition to the immune effector functions, our results revealed that the precise epitope targeted in misfolded SOD1 may be a further critical determinant for therapeutic efficacy. A human antibody targeting a loop VI epitope failed to confer any therapeutic benefit in SOD1G37R transgenic mice compared to vehicle-treated controls despite its selective binding to in vitro aggregated SOD1 preparations with an affinity comparable to that of α-miSOD1 for human mutant SOD1G37R. In contrast, α-miSOD1, which targets an epitope within loop IV of SOD1, extended survival of SOD1G37R transgenic mice by 2 months and slowed the degeneration of motor neurons and hindlimb muscles. Loop IV harbors the Zn2+ binding site within SOD1. This loop and its zinc ion stabilize the protein structure and protect SOD1 against conformational changes that increase the exposure of hydrophobic sections of the protein (49). In the absence of metals and disulfide bonds, loop IV loses its structure, resulting in a disruption of the dimer interface facilitating the formation of additional β-strands, oligomerization of SOD1 (50), and protein interactions, leading to endoplasmic reticulum stress (51). The loop IV epitope targeted by α-miSOD1 is exposed in motor neurons in postmortem spinal cord sections from patients with ALS but not in those of NNC. Immunohistochemical staining with antibody against the loop VI epitope revealed a more uniform staining throughout the neuropil that was similar between patients with ALS and controls. The observed differences in therapeutic effects are unlikely to be accounted for by antibody exposure as the mean steady-state plasma concentrations of both isotype-matched antibodies were comparable. A recent immunization study supports the importance of the selected epitope for immunotherapy because targeting a more N-terminal SOD1 epitope increased accumulation of misfolded SOD1 and consequently induced adverse effects in an ALS mouse model (6, 52).

There are some limitations to our study including the use of transgenic animal models that highly overexpressed mutant SOD1. In addition, α-miSOD1’s effects on neuromuscular junctions were not assessed. Potential differences in the protocols for human tissue collection, processing, and fixation among contributing brain banks may have affected the detection of misfolded SOD1. Furthermore, our analysis was limited to postmortem spinal cord samples and immunohistochemical detection of misfolded SOD1. Alternative techniques will be needed to study the role of misfolded SOD1 during disease onset and progression and for the development of corresponding disease markers.

Together, these data establish in healthy elderly populations the presence of B cell memory antibodies directed against misfolded wild-type SOD1. α-miSOD1, a recombinant human-derived antibody that showed selective high-affinity targeting of pathologically misfolded SOD1, ameliorated motor symptoms and improved survival in three mouse models of ALS, supporting its further development as a potential treatment for patients with ALS with misfolded and aggregated SOD1.

MATERIALS AND METHODS

Study design

This study was designed to derive from a de-identified memory B cell library from healthy elderly subjects a human antibody that selectively targeted with high-affinity misfolded and aggregated SOD1. The study then set out to assess the therapeutic potential of this antibody, α-miSOD1, in three different murine ALS models and to determine the prevalence of misfolded SOD1 species recognized by the α-miSOD1 antibody in patients with FALS and those with SALS. To assess the binding properties of α-miSOD1, a range of biochemical and immunohistochemical assays were used. The use of human biospecimens was approved by the Ethics Committee of the Canton of Zurich, and written informed consent was obtained from all patients who provided blood. Paraffin-embedded spinal cord tissues from patients with SALS (n = 98), FALS non-SOD1 patients (n = 20), FALS SOD1 patients (n = 3), and NNC patients (n = 41) were obtained from the ALS Clinic, St. Gallen, Switzerland; Netherlands Brain Bank, Amsterdam, Netherlands; Institute for Neurodegenerative Diseases, MGH, Boston, MA; Newcastle Brain Tissue Resource, Newcastle University, UK; Neuropathology, Academisch Medisch Centrum (AMC), Amsterdam, Netherlands; and University of California, San Diego (UCSD), San Diego, CA. Staining and image analysis were performed in a blinded fashion and in a random order.

All in vivo experiments were approved by the veterinarian authorities of the Canton of Zurich (animal license numbers: 152/2010, 237/2010, and 82/2014) in compliance with Swiss law (“455.163 Tierversuchsverordnung,” 2010). SOD1G37R [low-expressor line 29, B6.Cg-Tg(SOD1*G37R)29Dpr/J] and SOD1G93A [high copy number; B6.Cg-Tg(SOD1*G93A)1Gur/J] were obtained from the Jackson Laboratory, Maine, USA. B6SJL.Cg-Tg(SOD1*G93A)1Gur/J was obtained from ALS Therapy Development Institute (ALSTDI), Massachusetts, USA. All transgenic mice were held under specific pathogen–free conditions as described previously (21). Animals with more than 1.3-fold difference in transgene copy number were excluded. To assess the therapeutic efficacy of a mouse chimeric derivative of human α-miSOD1, chα-miSOD1, transgenic mice were treated either intracerebroventricularly or intraperitoneally. Mice were allocated to the different treatment groups by weight, gender, and littermate randomization. Group sizes were determined according to the guidelines “Working with ALS Mice, Guidelines for preclinical testing and colony management” (http://jackson.jax.org/rs/444-BUH-304/images/Working_with_ALS_Mice.pdf). Probability of disease onset was defined as the age of the animal reaching its peak body weight or the age after which a continuous drop in rotarod performance was observed during behavior testing. End-stage phenotype was defined as a loss of righting reflex, the inability of the animal to stand within 15 s after it was laid on its side. Mice that reached predefined non–ALS-phenotype–related ethical humane end points had to be euthanized before the end of the study and were excluded from the study analysis. Animal caretakers and investigators conducting the preclinical efficacy studies and investigators conducting the assessment of outcomes were blinded to the treatment allocation.

Antibody generation

Human SOD1 antibodies were derived from a de-identified blood lymphocyte library collected from healthy elderly subjects according to (41). In brief, memory B cells, isolated from peripheral blood lymphocyte preparations by anti-CD22–mediated sorting, were cultured on γ-irradiated human peripheral blood mononuclear cell feeder layers. Supernatants from isolated human memory B cells were screened for their ability to bind SOD1 preparations. Positive hits were subjected to complementary DNA cloning of IgG heavy-chain and κ or λ light-chain variable region sequences and subcloned in expression constructs using Ig-framework–specific primers for human variable heavy- and light-chain families in combination with human J-H segment-specific primers. α-miSOD1 was engineered to incorporate glycosylated human IgG1 heavy-chain and human λ light-chain constant domain sequences. A murine chimeric IgG2a/λ version of α-miSOD1 (chα-miSOD1) was generated for use in chronic efficacy studies in SOD1 transgenic mice. An aglycosylated variant of α-miSOD1 (chα-miSOD1–agly), incorporating a single point mutation (N297Q, using standard EU numbering), which eliminates N-glycosylation of the Fc region and severely reduces FcγR binding (23), was generated to test for Fc-related activities. Recombinant antibodies were transiently expressed in Chinese hamster ovary cells, purified using standard protein A affinity chromatography and desalted to phosphate-buffered saline (PBS) buffer. Endotoxin levels were confirmed to be <10 endotoxin units/ml.

Transgenic mouse studies

To test the therapeutic efficacy of central nervous application of chα-miSOD1, female SOD1G93A transgenic mice (7) were continuously treated intracerebroventricularly starting at 60 days of age. A brain infusion cannula (ALZET, Cupertino, CA, USA) connected to an osmotic minipump (model 1004, ALZET) filled with either chα-miSOD1 (n = 15, 0.92 mg/ml, equals 0.1 mg kg−1 day−1) or PBS vehicle control (n = 16) was implanted stereotaxically into the left ventricle using fentanyl/midazolam/medetomidin for anesthesia and naloxone/flumazenill/atipamezol and buprenorphine as an antidote and postoperative analgesic. Video gait analysis was performed at 80, 90, 100, 115, 130, and 140 days of age according to Preisig et al. (21), and pumps were surgically exchanged every 28 days using inhalation anesthesia (3.5% Sevoflurane) until end-stage phenotype was reached.

To assess the therapeutic effects of systemic (intraperitoneal) application of chα-miSOD1, gender, body weight, and litter balanced groups of slow-progressing SOD1G37R mice (8) were treated by weekly intraperitoneal injections of either chα-miSOD1 (10 mg/kg, n = 20), isotype-matched antibody LVI (10 mg/kg, n = 23), or PBS vehicle control (n = 20) from 12 weeks of age until end-stage phenotype. Motor performance was tested biweekly on the rotarod and PaGE.

To test the dose-dependent efficacy of systemic (intraperitoneal) application of chα-miSOD1, gender, body weight, and litter balanced groups of fast-progressing SOD1G93A tg mice were treated with weekly intraperitoneal injections of 3 mg/kg (n = 32), 10 mg/kg (n = 32), or 30 mg/kg (n = 25) chα-miSOD1 or PBS vehicle control (n = 55) starting at 28 days (range, 24 to 32 days) of age until end-stage phenotype. Body weight was measured twice per week, and motor performance on rotarod and PaGE were tested weekly. Systemic treatment of chα-miSOD1–agly (30 mg/kg, n = 25) or PBS control (n = 18) in SOD1G93A tg mice was performed accordingly.

For the defined time point study, gender and litter balanced groups of B6SJL SOD1G93A mice (12) were treated by weekly intraperitoneal injections of chα-miSOD1 (30 mg/kg, n = 21) or vehicle control (n = 21) starting from 30 days of age until 120 days of age using the tissue of 14 mice of each group for biochemical analysis and 7 mice of each group for histological end points. Phenotype scoring was performed according to (53).

For the spinal cord penetration and target engagement study, symptomatic 3.5- to 4.5-month-old SOD1G93A were injected once intraperitoneally with human α-miSOD1 (100 mg/kg) or isotype control antibody and perfused 2, 5, 10, or 20 days after the injection (n = 4 for each subgroup). Spinal cord tissue was homogenized in 10 volumes (10 ml/g of wet tissue) of a solution containing 50 mM NaCl and 0.2% diethylamine with protease inhibitors and sonicated for 15 to 20 s on ice, and human IgG levels were measured using a sandwich human IgG ELISA (41).

Motor performance assessment by video gait analysis, rotarod, and PaGE

The high-speed video analysis system MotoRater (TSE Systems, Bad Homburg, Germany) was used to evaluate the gait pattern of intracerebroventricularly treated SOD1G93A tg animals according to Preisig et al. (21). Motor coordination was measured using a rotarod apparatus (Ugo Basile, Monvalle, Italy). Animals were placed on a rotating cylinder that accelerates from 4 to 40 rpm within 180 s, and latency to fall was recorded. Muscle strength endurance was tested in the PaGE. Mice were suspended upside down on a steel mesh grid for a maximum of 120 s, and latency to fall was recorded.

SOD1 preparations

SOD1 purified from human erythrocytes (Sigma-Aldrich, Buchs, Switzerland) was dissolved overnight at 4°C in PBS (2.68 mM KCl, 1.47 mM KH2PO4, 137 mM NaCl, and 8.1 mM Na2HPO4, pH 7.0) and centrifuged at 20,000g to remove potential SOD1 aggregates, and the supernatant was used as native SOD1 preparation. Denatured SOD1 was prepared according to Zetterström et al. (18). In brief, native SOD1 was denatured with 3.5 M guanidinium chloride and 25 mM EDTA for 4 hours at 22°C, followed by dialysis against PBS containing 5 mM EDTA. Oxidized SOD1 was prepared according to Rakhit et al. (17). In brief, 10 μM human SOD1 was incubated in 10 mM tris acetate buffer (pH 7.0), containing 4 mM ascorbic acid and 0.2 mM CuCl2 for 48 hours at 37 °C. For some experiments, recombinant human SOD1 (Biomol, Germany), murine SOD1 (EMELCA Bioscience), or different BSA-coupled synthetic peptides around the α-miSOD1 epitope (73-GGPKDEERHVG-83) were used without any further denaturing steps.

Binding analysis of α-miSOD1 on human autopsy spinal cord sections

Paraffin-embedded spinal cord tissues from patients with ALS and NNC were obtained from the following sources: ALS Clinic, Cantonal Hospital St. Gallen (KSSG), Switzerland (20 SALS and 1 NNC); Netherlands Brain Bank, Amsterdam, Netherlands (11 NNC); Institute for Neurodegenerative Diseases, Massachusetts General Hospital (MGH), Boston, MA, USA (17 SALS); Newcastle Brain Tissue Resource, Newcastle University, Newcastle upon Tyne, UK (18 SALS, 3 FALS, and 11 NNC); Neuropathology, Academisch Medisch Centrum (AMC), Amsterdam, Netherlands (26 SALS, 14 FALS, and 10 NNC); and University of San Diego (UCSD), San Diego, CA (17 SALS, 6 FALS, and 8 NNC) (table S1 and S2). Staining and image analysis were performed in a blinded fashion and random order on two independent staining runs from each subset of 5-μm paraffin sections from patients with ALS and corresponding NNC subjects except for the University of San Diego/Ludwig Cancer Institute, San Diego, CA, samples, where the quantification was based on a single staining run (table S2). Sections were pretreated by cooking in citrate buffer (pH 6) and microwave irradiation for 12 min [according to (30)]. After peroxidase inactivation in 3% H2O2 in methanol and blocking in 5% horse serum/5% goat serum/4% BSA in PBS, the sections were incubated with 25 or 50 nM chimeric human-derived α-miSOD1 antibody overnight. Detection was performed with biotinylated donkey–anti-mouse secondary antibody (Jackson ImmunoResearch; 1:250) and the Vectastain ABC kit (Vector Laboratories) with diaminobenzidine (Pierce). Stained sections were imaged using a dotSlide system (Olympus), and regions of interest (ROIs) included the complete ventral horn for “total misfolded SOD1” or the outline of the somata of all individual motor neurons in the ventral horn for “intraneuronal misfolded SOD1” excluding staining in corpora amylacea structures. ImageJ image processing software (www.imagej.net) was used to quantify the percent area positive for misfolded SOD1 immunoreactivity in the ROIs using a fixed color threshold for the automated analysis of all independent staining runs. Data from the nine different staining runs were pooled by normalization to the corresponding NNC, resulting in a fold change compared to NNC. Additional materials and methods data are provided in the Supplementary Materials.

Statistics

Statistical analyses were performed with GraphPad Prism. Statistical tests and P values are reported in the text and figure legends. Data are represented as average ± SEM.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/10/470/eaah3924/DC1

Materials and Methods

Fig. S1. α-miSOD1 is selective for misfolded human SOD1.

Fig. S2. α-miSOD1 immunohistochemical analysis of human spinal cord (Newcastle brain bank).

Fig. S3. α-miSOD1 immunohistochemical analysis of human spinal cord (AMC brain bank).

Fig. S4. α-miSOD1 immunohistochemical analysis of human spinal cord (KSSG and Netherlands brain banks).

Fig. S5. α-miSOD1 immunohistochemical analysis of human spinal cord (UCSD brain bank).

Fig. S6. α-miSOD1 immunohistochemical analysis of human spinal cord (MGH and Netherlands brain banks).

Fig. S7. α-miSOD1 immunoreactivity is associated with glial markers, and α-miSOD1 treatment reduces neuroinflammation.

Fig. S8. α-miSOD1 binding epitope in the native human SOD1 structure.

Table S1. Average patient demographics.

Table S2. Subset analysis of misfolded SOD1 histology in patients with SALS versus NNC.

Data file S1. Raw data (Excel file).

REFERENCES AND NOTES

Acknowledgments: We thank N. Glassl, A. Hagenbuch, I. Combaluzier, M. Kalveram, M. Krüger, A. Jeske, B. Ries, and S. Imobersteg for technical assistance. We thank J. Ravits (UCSD), D. Troost and M. Casula (AMC, Amsterdam), M. Frosch, C.S. Kubik Laboratory for Neuropathology (MGH, Boston), and Newcastle Brain Tissue Resource, UK, for providing postmortem human spinal cord tissue sections. Funding: This project was funded in part by grant no. 51NF40-144620 of the Swiss National Science Foundation (SNSF). Author contributions: M.M., T.W., F.M., R.M.N., and J.G. conceived the study. M.M., T.W., F.G.V., F.M., F.W., M.W., S.P., T.S., R.M.N., C.H., and J.G. contributed to the experimental design and interpretation of data and critically revised the manuscript. F.M., M.M., F.W., and T.W. contributed to antibody cloning and characterization. M.M., T.W., F.W., F.G.V., D.P., J.M., L.K., and C.S. contributed to the in vivo studies and their data analysis. M.M., T.W., and J.G. wrote the manuscript. Competing interests: M.M., T.W., F.M., J.M., C.H., R.M.N., and J.G. are coinventors on patent number WO/2012/080518 entitled “Human anti-SOD1 antibodies.” M.M., F.W., F.M., C.H., J.G., and R.M.N. are shareholders and employees of Neurimmune AG. The other authors declare no competing interests. Data and materials availability: All data associated with this study are in the paper or the Supplementary Materials. Antibodies have been licensed to AL-S Pharma and academic researchers can contact M. Salzmann at Michael.Salzmann{at}al-spharma.com to request the antibodies.
View Abstract

Navigate This Article