Elevated α5 integrin expression on myeloid cells in motor areas in amyotrophic lateral sclerosis is a therapeutic target

α5 Integrin Expression Is Associated with Severe Pathology and Myeloid Cells with Distinct Proinflammatory Properties in SOD1^G93A ALS Mice.

Our recent study using CyTOF as a high-throughput and high-resolution, single-cell proteomic screening platform identified elevated α5 integrin on myeloid cells, but not on T or B cells in experimental autoimmune encephalomyelitis (EAE), a neuroinflammatory model that shares some features with multiple sclerosis (22). In the current study, we extended these analyses and evaluated α5 integrin expression in 78,293 single cells, including 21,250 CD45⁺/CD11b⁺ myeloid cells from CNS of SOD1^G93A ALS mice, using an unbiased computational approach. Unsupervised clustering with T-stochastic neighbor embedding (t-SNE) revealed subpopulations of CD45⁺/CD11b⁺ CNS myeloid cells expressing α5 integrin in the late stages of motor neuron disease (defined by severe paralysis at 140 d) compared to disease onset (95 d) (Fig. 1A). Immunohistological analysis of the spinal cord, the primary site of pathology in SOD1^G93A mice, confirmed increased expression of α5 integrin at disease end-stage in SOD1^G93A mice compared to age-matched wild-type (WT) mice (Fig. 1B). Consistent with our single-cell CyTOF results showing that alpha5 integrin was expressed by myeloid cells, α5 integrin staining was mostly colocalized with ionized calcium-binding adaptor molecule 1 (Iba1), a pan-macrophage marker expressed by microglial cells in the CNS, suggesting that they are the main cell type expressing this integrin in the late stage of this disease model (Fig. 1B). Quantification of microglial cells (DAPI⁺, Iba1⁺) expressing α5 integrin in the ventral horn of the spinal cord revealed α5 integrin was significantly increased in SOD1^G93A end-stage mice (25.6 ± 1.8%) compared with C57Bl6/J control mice (0.06 ± 0.03%) (Fig. 1C). We recently showed the pathophysiological importance of peripheral macrophages in the sciatic nerve of ALS SOD1^G93A mice and patients (8). Therefore, we compared α5 integrin expression in the sciatic nerve of SOD1^G93A mice with that of WT mice. The expression of α5 integrin in the sciatic nerve was increased in SOD1^G93A mice, predominantly at the disease end-stage (Fig. 1D). Costaining of α5 integrin with a cocktail of myeloid cell markers (CD11b, CD68, and F4/80) revealed increased α5 integrin expression in sciatic nerve macrophages (49.6 ± 7.6%) at end stage compared to control mouse sciatic nerve macrophages (11.07 ± 10%) (Fig. 1E). To determine whether the myeloid cells expressing α5 integrin in the spinal cord and sciatic nerve of SOD1^G93A mice had proinflammatory properties, which may play a role in the pathophysiology of ALS, we analyzed cytokine expression by CyTOF at disease end-stage (SI Appendix, Tables S1 and S2). In the spinal cord, α5 integrin–positive myeloid cells produced significantly higher levels of all cytokines measured––TNF-α, GM-CSF, IL-6, IFN-α, TGF-β, and IL-10 compared to α5 integrin–negative or low expressing microglia (Fig. 1F). TNF-α, a highly proinflammatory cytokine, was significantly up-regulated (fivefold increase) in α5 integrin–positive CNS myeloid cells (Fig. 1F). Similarly, sciatic nerve macrophages produced higher levels of cytokines compared to α5 integrin–negative macrophages (Fig. 1G). IL-6 and TGF-β were the most prominent cytokines produced by α5 integrin–positive macrophages (Fig. 1G). To determine whether our findings were consistent with results based upon different methodologies, we mined previously published transcriptomic datasets of SOD1^G93A microglial cells from two studies (8, 11) and confirmed increased α5 integrin expression in microglial cells (SI Appendix, Fig. S1 A and B). These findings suggest that elevated α5 integrin on myeloid cells may play a role in the pathology of ALS.

Increased Expression of α5 Integrin in the Ventral Horn and Motor Tracts in ALS with Different Clinical Presentations and Disease Durations.

To further explore the significance of α5 integrin in ALS, we investigated the presence of α5 integrin and colocation with microglia in ALS patients with a range of clinical presentations and disease durations. We studied autopsy tissues from 132 ALS patients. Most had abnormal TDP-43 aggregates, including patients with sporadic ALS and c9ALS, as well as ALS patients with rare genetic variants. The latter included familial ALS due to mutations in TARDBP and ALS associated with mutations in TBK1 or SOD1.

Since adult-onset ALS is characterized by selective loss of motor neurons, especially the large α-motor neurons in the ventral horn of the spinal cord, we assessed neuronophagia with Iba1 immunohistochemistry in 132 ALS patients. Markedly increased Iba1-positive microglia were detected in 99 cases (75%) using a semi-quantitative approach (0 = none, 1 = mild, 2 = moderate, and 3 = marked). Immunoreactive microglia had morphological features of activated microglia, with amoeboid rather than ramified morphology. There were also activated microglia in the lateral and anterior corticospinal tracts. Neuronophagia was assessed in anterior horns as an indicator of active motor neuron disease. Neuronophagia was defined as clusters of microglial cells in the ventral horn often with an “empty cell bed” where a degenerating neuron had resided. Neuronophagia was detected in cervicothoracic, thoracic, and lumbar sections of the spinal cord in 18 of 132 patients, irrespective of etiology (sporadic, C9ORF72, SOD1, TARDBP, and TBK1) (SI Appendix, Table S3). We compared ALS patients with active neuronophagia to individuals without motor neuron disease (N = 10).

We found higher expression of α5 integrin in ALS compared to normal controls in the lateral corticospinal tracts and ventral horns, irrespective of the etiology of ALS or the disease duration (Fig. 2 A–C). The expression of α5 integrin was detected in cells surrounding motor neurons in the ventral horn (Fig. 2A). In addition, endothelial cells in spinal cord vessels and meninges of ALS patients had higher expression of α5 integrin compared to normal controls (Fig. 2 D–I). Importantly, α5 integrin was increased not only in cells adjacent to ventral horn neurons (Fig. 2H), but also in “empty cell beds” and areas of gliosis in the anterior horn (Fig. 2 J–M). A subset of ALS patients with severe motor cortex involvement had increased α5 integrin immunoreactivity in the parenchyma and blood vessels, sometimes associated with foamy macrophages, a microglial morphology associated with a phagocytosis (Fig. 2 N and O).

Similar selective involvement of motor tracts was also evident in nerve roots from ALS patients. Expression of α5 integrin was detected in anterior roots, which contain motor neuron axons, but not in posterior roots and dorsal root ganglion where sensory neuronal axons originate (Fig. 2 P–T).

We further confirmed elevated α5 integrin expression in ALS by mining previously published transcriptomic data from spinal cord ventral horns of postmortem ALS human donors (23). This analysis confirmed that α5 integrin expression is significantly increased in the spinal cord of ALS patients compared to controls (SI Appendix, Fig. S1C).

Our CyTOF analysis and the gene expression analyses from other groups have shown that microglia and sciatic nerve macrophages express elevated levels of α5 integrin in the SOD^G93A ALS mouse model. Thus, we further evaluated the expression of Iba1 (microglial marker) and its correlation with α5 integrin in the spinal cord of ALS patients and patients without MND. Consistent with previous studies, Iba1 was significantly increased in the ALS spinal cord compared to normal controls (Fig. 2U). Foamy macrophages and α5 integrin–positive cells were increased in the same areas in ALS (Fig. 2 V–X). Together, detailed analysis of α5 integrin staining in ALS revealed prominent expression of α5 integrin by microglia in affected motor regions. These findings were not observed in normal controls.

α5 Integrin Expression Is Increased Irrespective of the ALS Subtype and Duration of the Disease.

Immunohistological assessment showed a significant increase of α5 integrin expression in ALS patients compared to normal controls (Fig. 3A). More importantly, a subanalysis across ALS etiologies confirmed increased expression of α5 integrin in the spinal cord compared to controls irrespective of ALS subtype (Fig. 3A). Furthermore, α5 integrin was significantly increased in the ALS spinal cord regardless of age of death or disease duration (Fig. 3B). We next used CyTOF to examine α5 integrin–positive myeloid cells in an additional cohort of ALS patients (SI Appendix, Fig. S2A and Table S5). α5 integrin– positive cells represented more than 40% of the myeloid cells (CD11b+) in ALS patient cervical spinal cord (48.5 ± 10.8%, n = 3) and motor cortex (42.8 ± 4.9%, n = 3), without significant differences between the two regions (SI Appendix, Fig. S2 A and B). Additional CyTOF analysis revealed that α5 integrin–positive cells expressed higher levels of HLA-DR and CD74, critical for antigen presentation, in addition to CD11c and CD9 compared to α5 integrin–negative myeloid cells (SI Appendix, Fig. S2C). Interestingly, these markers are part of the signature of disease-associated microglia (DAM), a subgroup of microglial cells primarily associated with neurodegeneration (24, 25). Addressing whether the clinical treatments that ALS patients received might have mitigated α5 integrin expression, we performed an analysis after subgrouping patients into two treatment groups to compare patients that received no treatment or riluzole only (labeled “Standard” in Fig. 3C), to patients that received drugs such as prednisone, TNF-α, edaravone, antisense oligonucleotide, or mesenchymal stem cell therapy (labeled “Modulatory” in Fig. 3C). There was no significant difference in α5 integrin in the spinal cord between the two groups, highlighting that α5 integrin is not adequately targeted by current modulatory treatment strategies.

α5 Integrin Demonstrates Specificity to ALS Pathology Compared to Another Druggable Integrin.

To address the specificity and the distribution pattern of α5 integrin in ALS compared to other integrins, we stained the ALS spinal cord for α5 integrin and compared findings to another druggable integrin, α4. In 1992, we identified α4 integrin as the key molecule involved in homing of leukocytes to inflamed regions of the brain (26). An antibody to α4 integrin called natalizumab has been approved as a therapy for multiple sclerosis (26). Recently, a study demonstrated that treating SOD1^G93A and TDP43^A315T mice, another ALS model, with natalizumab prolonged survival in these models (27). Since natalizumab blocks homing of all leukocytes, it has been shown to be associated with progressive multifocal leukoencephalopathy in JC virus–seropositive patients by preventing lymphocytes from adhering to the endothelium of the blood–brain barrier, thereby reducing their migration from the blood into the CNS and suppressing T cell–mediated immune responses in the brain (28). To determine whether α5 integrin might present a better druggable target in ALS compared to the α4 integrin, we compared spinal cord expression of α5 integrin versus α4 integrin in ALS (SI Appendix, Fig. S3 A and B). In contrast to α5 integrin, α4 integrin was not expressed by foamy macrophages in ALS (SI Appendix, Fig. S3B). Myelin staining with Luxol fast blue suggested that α4 integrin was expressed in the myelin of ALS patients (SI Appendix, Fig. S3A). ALS spinal cord tracts demonstrating myelin pallor had fragmented α4 integrin myelin profiles in the ventral horn (SI Appendix, Fig. S3 A and B) and in the lateral corticospinal tract (SI Appendix, Fig. S3B), including cells morphologically consistent with oligodendrocytes (SI Appendix, Fig. S3B, arrowheads). Lack of α4 integrin immunoreactivity in microglia in ALS patients indicates that α4 integrin is unlikely to contribute to neuronophagia or tissue destruction in ALS.

Anti-α5 Integrin Has Therapeutic Potential by Improving Motor Symptoms and Increasing Survival in a Preclinical Model of ALS.

To determine the therapeutic potential of blocking α5 integrin in ALS, we explored the potential effect of modulating α5 integrin expression in the SOD1^G93A ALS mouse model. Despite the discovery of numerous additional genes related to the genetic forms of ALS disease, SOD1^G93A mice are still the most dependable model in developing an ALS-like phenotype. This model can be used to measure the outcome of potential ALS treatments by assessment of parameters of disease progression, including age at onset, lifespan, disease duration, inflammation, and motor performance. In a pilot study, we treated a small cohort of SOD1^G93A mice with anti-α5 integrin antibody, MFR5, which blocks adhesion of α5 integrin–expressing cells. A group of 10 SOD1^G93A mice received intraperitoneal (IP) injections of MFR5 two times per week and 10 mice received the isotype control antibody (Fig. 4A). Blocking α5 integrin in SOD1^G93A mice extended the survival of mice (169.9 ± 3.5 d) compared to isotype controls (160.7 ± 3.4 d) (Fig. 4B). These data show that α5 integrin plays a role in the pathology of ALS and identifies a previously unreported therapeutic target. To determine the impact of blocking α5 integrin on improving motor symptoms and to determine whether our pilot study was reproducible when performed in another laboratory (a useful standard in experiments including animal models), we tested the effect of long-term treatment of a larger cohort of SOD1^G93A and control mice (N = 60) at QPS Neuropharmacology in Austria. Prior to the validation cohort study, we performed pharmacokinetic (PK) and receptor occupancy (RO) studies in a group of control mice (C57Bl/6) to understand whether the dosing regimen of biweekly injections of anti-α5 integrin antibody, as performed in the preliminary study, was sufficient to block α5 integrin in the peripheral circulation and CNS. After a single IP injection at 5 mg/kg, the anti-α5 integrin (MFR5) antibody was measured in blood and cerebrospinal fluid (CSF), and receptor occupancy was determined on blood cells (SI Appendix, Fig. S4 A and B). The results of the PK study show that serum anti-α5 integrin antibody reached Cmax at 4 h at approximately 100 µg/mL, and at 72 h was at 31.5 µg/mL (SI Appendix, Fig. S4A). Anti-α5 integrin antibody concentrations in CSF at 24 and 72 h after the single injection were approximately 1,000-fold lower (0.1%) than what was measured in serum at the same time points (SI Appendix, Fig. S4A), indicating that low levels of this antibody penetrate the blood–brain barrier, which is typical of what is seen with IgG antibodies in general (2). The results of the RO study show that after a 5-mg/kg IP injection of the anti-α5 integrin antibody, monocytes and NK cells have greater than 90% RO out to 72 h (SI Appendix, Fig. S4 B and C). Notably, the anti-α5 integrin antibody was not significantly detected in B and T cells, which is consistent with the lack of α5 integrin expression in these cell types, and consistent with what we reported previously (22). The PK and RO data taken together suggest that biweekly injections of the anti-α5 integrin antibody at 5 mg/kg, as was used in the pilot study, are sufficient to achieve high levels of exposure in peripheral circulation and low levels of exposure in the CNS. Thus, the biweekly dosing scheme was continued for the larger validation study performed at QPS Neuropharmacology.

Survival, behavioral, and histological outcomes of treatment were measured in SOD1^G93A mice that received either the anti-α5 integrin antibody (MFR5; n = 30) or an isotype control antibody (n = 30) by IP biweekly injections starting at 8 wk until disease end-stage or until 20 wk (late symptomatic stage) when tissue was harvested (n = 6 per group). In-depth analysis of the disease course of anti-α5 integrin–treated ALS mice revealed a similar age of disease onset, but a delay of the symptomatic stage and a longer survival compared to isotype-treated ALS mice (Fig. 4 C–E), consistent with microglia/macrophages playing a major role in ALS disease progression but not in its initiation (3, 4). These results show that α5 integrin expression is strongly associated with disease progression in SOD1^G93A mice (Fig. 4).

To assess the effect of anti-α5 integrin treatment on motor phenotype, we performed several behavioral analyses. The beam walking test, a measure of coordination and balance, revealed significantly fewer slips (Fig. 4F and SI Appendix, Fig. S4D) and reduced latency to traverse the beam (Fig. 4G) in the group treated with anti-α5 integrin antibody compared to the isotype-treated group. Similarly, Vercelli scoring, used to assess motor phenotype in ALS mice (29), showed a trend toward better scores for α5 integrin–treated mice, with a significant difference at week 18 (corresponding to early disease stage) compared to isotype-treated mice (SI Appendix, Fig. S4E). Finally, a general evaluation of ALS mouse phenotype and health assessed daily, from week 16 (corresponding to early disease stage and considered as day 1 in the SI Appendix graph) until disease end stage, revealed significantly lower scores during the symptomatic phase of the disease, indicative of better general health and reduced symptoms, in α5 integrin–treated ALS mice compared to isotype-treated mice (Fig. 4H and SI Appendix, Fig. S4H).

In the current study, we have shown using CyTOF, transcriptomic profiling, and immunohistochemistry that microglia are the main immune cells expressing α5 integrin in ALS. Microglia are known to become activated in response to motor neuron degeneration (Fig. 2).

Therefore, we analyzed the effect of anti-α5 integrin treatment on microglial cells and Iba1 immunoreactivity, in the spinal cords of SOD1^G93A mice (Fig. 4 I–L). While the percent of microglial cells expressing α5 integrin (Fig. 4J) and the number of microglial cells (Fig. 4K) remained unchanged, the Iba1 immunoreactivity was significantly decreased in α5 integrin–treated SOD1^G93A mice compared to isotype-treated mice (Fig. 4 I and L). Note that the immunoreactive area measurement considers both the surface occupied by the staining and its intensity. As microglial activation is associated with larger cell body size and an increase of Iba1 intensity, a decrease in both parameters may indicate lower microglial reactivity in the anti-α5 integrin–treated mouse cohort compared to isotype-treated mice. The findings suggest that mitigation of an α5 proinflammatory microglial response may contribute to increased SOD1^G93A mouse survival.

Mice.

For the purpose of this study, SOD1 transgenic mouse lines were used: hSOD1^G93A mice [B6.Cg Tg (SOD1*G93A)1Gur/J; #004435, Jackson Laboratory]. Mice carrying the human SOD1 transgene were identified by PCR screening of tail DNA. C57bl/6J (#000664, Jackson Laboratory) nontransgenic mice were used as controls for histology experiments. Mouse experiments were carried out in three different facilities: Stanford University for mice used in Fig. 1 A and F, Upper, Fig. 4 A and B; Oregon Health and Science University for mice used in Fig. 1 B–E and F, Lower; and QPS (Austria GmbH) for mice used in Fig. 4 C–L. Mouse experiments at Stanford University were approved by the Institutional Animal Care and Use Committee at Stanford University and performed in compliance with HIH guidelines. The QPS (Austria GmbH) animal facility is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). All the procedures in this study comply with the animal care and welfare committee and are in accordance with standard operation procedures. Additional mouse experiments for histology analysis were approved by and performed in compliance with the NIH guidelines of the Institutional Animal Care and Use Committee at Oregon Health and Science University. In all facilities, the mice were housed in individual ventilated cages on standardized rodent bedding. Each cage contained a maximum of five mice. The room temperature was maintained at 20 to 24 °C and the relative humidity was maintained between 45 and 65%. The animals were housed under a constant light cycle (12 h light/dark). Dried, pelleted standard rodent chow and normal tap water were available to the animals ad libitum.

Treatment.

Behavioral Assessment.

Vercelli score and beam walk test were performed for all animals to assess the motor behavior of all animals. All animals were examined using the Vercelli score as a reference standard for seven times (week 8—baseline, 10, 12, 14, 16, 18, and 20). Beam walk test (two beams tested) was performed once at an age of 18 wk. The behavioral testing was done on nondosing days. Clinical score was assessed daily starting at 16 wk old. For more details, see SI Appendix, Materials and Methods.

Mouse Single-Cell Isolation.

Mouse CyTOF Experiment.

Antibody testing, panel design and antibody conjugation, and cell staining were performed as described in our previous publication (22). For more details, see SI Appendix, Materials and Methods.

Human Postmortem Single-Cell Isolation.

Fresh human brain tissues were provided from the donor to the Oregon Brain Bank (Oregon, USA) following the consent of next of kin and deidentification of tissues as stipulated by the IRB to conform to the requirements of nonhuman subject research on human postmortem tissue (SI Appendix, Table S5). Immediately following the autopsy, the motor cortex and cervical spinal cord tissues were dissociated using Dounce homogenization in cold 0.1 M HBSS (Gibco) on ice. The homogenized tissue was filtered through a 70-μm cell strainer (Falcon) and centrifuged at 750 g for 7 min at 4 °C. Pelleted cells were then resuspended in 5 mL of 0.1 M HBSS + 2 mL of RPMI (Gibco) + 3 mL of isotonic Percoll [90% Percoll (GE Healthcare) + 10% 1 M HBSS], then a 2-mL layer of 70% Percoll was slowly layered to the bottom of the tube using a 3.5-inch spinal needle (BD Biosciences). Cells were then centrifuged at 500 g for 15 min at RT without break. Single cells were isolated from the interface between the 30% and 70% Percoll layers. Isolated cells were then washed twice with 0.1 M PBS (Gibco) with 2% heat-inactivated fetal bovine serum (FBS) (Atlas Biologicals). The cells were fixed with 0.2 μm-filtered 1.6% paraformaldehyde (electron microscopy), shaken for 10 min at RT, and then washed twice with 0.1 M PBS supplemented with 2% FBS. Fixed cells were stored at −20 °C until use.

Human Postmortem Tissue CyTOF Analysis.

Frozen isolated microglia were thawed on ice, washed twice at 500 g for 5 min at 4 °C using 0.1 M Maxpar PBS (Fluidigm) supplemented with 2% FBS, refiltered through a 70-μm cell strainer (Falcon), and counted under a microscope. Cells were pelleted down and washed twice with 1 mL Maxpar Cell Staining Buffer (CSB) (Fluidigm) at 500 g for 5 min at 4 °C. The cells were then resuspended in 50 μL CSB, and a surface antibody cocktail (SI Appendix, Table S6) created in 50 μL CSB was added. The cells were incubated in antibodies for 45 min at RT on a shaker. After incubation, the cells were washed twice in CSB and spun down at 500 g for 5 min at 4 °C. After staining, the cells rested for 15 min at 4 °C, and 500 μL cold methanol (Fisher Chemical) was added for 20 min at 4 °C. Once permeabilized, the cells were washed twice with CSB, pelleted down at 500 g for 5 min at 4 °C, and resuspended in 100 μL CSB. The cells were then incubated overnight at 4 °C in a solution of 0.1 μM 191/193Ir DNA intercalator (Fluidigm) and 0.2 μm-filtered 1.6% paraformaldehyde diluted in 0.1 M Maxpar PBS. Prior to CyTOF acquisition, the cells were washed once with cell staining media and then two times with double-deionized water (ddH₂0). EQ Four Element Calibration Beads (Fluidigm) containing known concentrations of 140/142Ce, 151/153Eu, 165Ho, and 175/176Lu were added at 1:10 to each sample. The cells were analyzed on a CyTOF Helios™ (Standard BioTools) at an event rate of 200 to 300 cells per second. All mass cytometry files were normalized together using the mass cytometry data normalization algorithm. Stained cells were analyzed on a Helios mass cytometer (Fluidigm) at an event rate between 50 and 200 cells per second. Sample normalization was performed using normalization beads. Further analysis was performed using Cell Engine (https://cellengine.com).

Tissue Collection for Histology.

Histology.

Immunofluorescence.

Spinal cord sections were incubated in a blocking solution containing 0.1 M PBS, 0.3% Triton X-100 (PBST), and 3% BSA (Sigma Aldrich) for 1 h at room temperature. Spinal cord transverse cryosections were incubated overnight at room temperature with a rabbit anti-Iba1 (1:500, Wako Chemicals) and a rat anti-α5 integrin (1:400, BioLegend) in PBST. Anti-Iba1 antibody was revealed with a goat anti-rabbit Alexa Fluor-594 secondary antibody (Life Technologies), anti-α5 integrin antibody was revealed with a goat anti-rat Alexa Fluor-488 secondary antibody (Life Technologies) and sections were stained for DAPI before mounting the slides. Sciatic nerves were incubated in a blocking solution containing 0.1 M PBS, 0.3% Triton X-100 (PBST), and 3% BSA (Sigma Aldrich) for 1 h at room temperature. Sciatic nerves were sequentially stained first with a rat anti-α5 integrin antibody (1:400, BioLegend) overnight at room temperature and revealed with a goat anti-rat Alexa Fluor-488 secondary antibody (Life Technologies). On the second day, sciatic nerves were incubated with a cocktail of rat anti-CD11b (1:400, BD Biosciences), rat anti-CD68 (1:400, AbD Serotec), and rat anti-F4/80 (1:100, AbD Serotec) antibodies overnight at room temperature and revealed with a goat anti-rat Alexa Fluor-594 secondary antibody (Life Technologies).

Image Acquisition and Quantification.

Image of spinal cords and sciatic nerve were obtained using a Zeiss Apotome.2 on an Axio Imager with Zen v3.5 software. All image analyses were performed using Fiji software (ImageJ).

Disease Stage Analysis.

Disease time points were defined as follows. The time of disease onset was retrospectively determined as the time when mice reached peak body weight. The time of early symptomatic stage was defined as the age at which the animals had lost 10% of their maximal weight. Mice were checked daily when they approached end stage, which was defined by paralysis so severe that the animal was unable to right itself within 20 s when placed on its side. Mice of each group were weighed weekly. Kaplan–Meier analysis was used for survival curves. Log-rank tests were used for statistical analysis.

Pharmacokinetics.

Male C57Bl/6J mice (Jackson Laboratories) were habituated at least for 1 wk prior to study start. Only animals in apparently good health condition were included in the study. All animals were randomly assigned to starting groups (cohort). Injections were initiated at 8 wk of age. Ultra-LEAF™ Purified anti-mouse α5 integrin Antibody Clone: 5H10-27 (MFR5) was purchased from BioLegend, was diluted in PBS (Lonza; 1X w/o Ca++, Mg++) to 1 mg/mL concentration, and administered once by IP route at a volume of 10 μL/g body weight at a dose of 5 mg/kg. Twenty mice were divided into five cohorts of four mice each such that each mouse was only bled once in vivo and once terminally. This number of mice sufficed to cover blood draws that were collected at timepoints over a 2-wk period at 1, 2, 4, 8, 24, 72, 168, 240, and 336 h. For more details, see SI Appendix, Materials and Methods.

Receptor Occupancy.

C57Bl/6 mice were dosed IP at 5 mg/kg of α5 integrin antibody. Blood was collected at predose, 1, 4, and 24 h, and terminal collections were performed at 3 and 7 d, from three mice per time point. Control blood was collected at the same time points from mice that were injected with vehicle (saline) without α5 integrin antibody. For more details, see SI Appendix, Materials and Methods.

Human ALS Cases.

Inclusion criteria were ALS as a both clinical and neuropathologic diagnosis. We screened 635 spinal cord levels from 132 ALS patients with sections stained for hematoxylin and eosin (H&E), a myelin stain [Luxol fast blue–periodic acid Schiff (LFB–PAS)], and immunohistochemistry for ionized calcium-binding adaptor molecule 1 (Iba1, 1:3,000, Wako Chemicals USA), a widely used microglial/macrophage marker. Neuropathologic evidence of active neuronophagia was defined as clusters of macrophages in the ventral horn often with an “empty cell bed” where a degenerating neuron had resided. The cohort assessed in the current study consisted of 96 patients with sporadic ALS (negative for the following genetic mutations: C9ORF72, SOD1, TBK1, TARDBP), 29 patients with C9ORF72 (C9-ALS), 5 patients with SOD1 mutation (SOD1-ALS), 1 ALS patient with a TARDBP mutation, and 1 ALS patient with a TBK1 mutation. Sporadic ALS and C9ORF72-ALS cases had motor neuron pathology as well as neuronal and glial inclusions with TDP-43 immunohistochemistry, while ALS-SOD1 had neuronal aggregates with immunohistochemistry for SOD-1 or SEDI (38) (SI Appendix, Table S3).

Immunohistochemistry of Human Formalin-Fixed Paraffin-Embedded Tissue.

The human tissue was stained with antibodies and conditions as previously described (39) (SI Appendix, Table S7). Briefly, immunohistochemistry was performed on 5 µm-thick sections. Paraffin sections were dewaxed in xylene for 3 × 5 min washes, rehydrated in a graded series of ethanol for 3 × 2 min, and washed in distilled dH₂0. Normal goat serum in tris-buffered saline with Tween® (TBST) was used to block unspecific antibody binding prior to incubation with the primary antibodies. All stains were processed on the DAKO AutostainerPlus (DAKO) using the DAKO Envision⁺ HRP detection system. Two sets of integrin antibodies were further tested (Absolute Antibody and LSBio) to determine reliability and reproducibility prior to staining the cohort favoring LSBio antibody for formalin-fixed human tissues.

Digital Image Analyses and Quantification.

Whole-slide images were scanned on an Aperio ScanScope slide scanner (Aperio Technologies) producing high-resolution images. Image quantifications were performed using a digital positive pixel algorithm optimized for each histiocyte-associated stain (Iba1, α5 integrin) expressing densities as number of strong positive pixels per mm² in patients and control cases. All annotations, digital algorithm processing, and data exports were done in a blinded fashion.

Statistics.