Protein kinetics of superoxide dismutase‐1 in familial and sporadic amyotrophic lateral sclerosis

Abstract Objective Accumulation of misfolded superoxide dismutase‐1 (SOD1) is a pathological hallmark of SOD1‐related amyotrophic lateral sclerosis (ALS) and is observed in sporadic ALS where its role in pathogenesis is controversial. Understanding in vivo protein kinetics may clarify how SOD1 influences neurodegeneration and inform optimal dosing for therapies that lower SOD1 transcripts. Methods We employed stable isotope labeling paired with mass spectrometry to evaluate in vivo protein kinetics and concentration of soluble SOD1 in cerebrospinal fluid (CSF) of SOD1 mutation carriers, sporadic ALS participants and controls. A deaminated SOD1 peptide, SDGPVKV, that correlates with protein stability was also measured. Results In participants with heterozygous SOD1 A5V mutations, known to cause rapidly progressive ALS, mutant SOD1 protein exhibited ~twofold faster turnover and ~ 16‐fold lower concentration compared to wild‐type SOD1 protein. SDGPVKV levels were increased in SOD1 A5V carriers relative to controls. Thus, SOD1 mutations impact protein kinetics and stability. We applied this approach to sporadic ALS participants and found that SOD1 turnover, concentration, and SDGPVKV levels are not significantly different compared to controls. Interpretation These results highlight the ability of stable isotope labeling approaches and peptide deamidation to discern the influence of disease mutations on protein kinetics and stability and support implementation of this method to optimize clinical trial design of gene and molecular therapies for neurological disorders. Trial Registration Clinicaltrials.gov: NCT03449212.


Introduction
Amyotrophic lateral sclerosis (ALS) is a neurodegenerative syndrome defined by motor neuron loss in the motor cortex and spinal cord that typically results in death in 3 to 5 years. Although 90% of ALS cases are considered sporadic, about 10% are due to genetic etiology. Nearly 20% of familial ALS (fALS) is caused by toxic gain-offunction mutations in Cu/Zn superoxide dismutase-1 (SOD1). SOD1 consists of 154 amino acids and mutations in over 100 residues have been described to cause ALS though the pathogenicity and penetrance of individual mutations can vary significantly. 1,2 Misfolded SOD1 aggregates accumulate in motor neurons of patients with SOD1-related ALS, 3 and numerous mutations are known to promote SOD1 protein turnover in cell culture and ALS mouse models. [4][5][6][7] Lower abundance of SOD1 mutant protein in erythrocytes is associated with shorter survival suggesting that protein instability may influence disease progression. 8 Wild-type SOD1 is also known to misfold under certain cellular conditions 9 so a pathogenic role for SOD1 in sporadic ALS (sALS) has been hypothesized. Although some immunohistochemical studies identified SOD1 inclusions in post-mortem sALS brain and spinal cord, 10,11 other studies were unable to reproduce these results. [12][13][14] However, in depth biochemical analysis of post-mortem familial and sporadic ALS tissue suggests that unstable, mismetallated and alternatively post-translationally modified SOD1 accumulates in ventral spinal cord even in non-SOD1-linked cases. 15 These studies implicate accumulation of misfolded mutant (and possibly wild-type) SOD1 in ALS pathogenesis. Unraveling factors that modify SOD1 protein kinetics may clarify mechanisms that influence ALS disease pathogenesis.
SOD1-lowering therapeutics have emerged as a promising strategy for treating patients with SOD1-associated ALS. Tofersen an antisense oligonucleotide (ASO) targeting SOD1 has completed a Phase III trial and continues to be provided through expanded access. 16,17 In addition, a strategy involving SOD1 targeting microRNA (miRNA) delivered via adeno-associated virus (AAV) 18 was cleared for Phase I/II trial. Both agents target SOD1 mRNA and are predicted to inhibit new synthesis of SOD1.
The ability to measure SOD1 synthesis and clearance in vivo in humans may provide better understanding of the pathogenesis of SOD1 in sporadic ALS and earlier assessments of target engagement in the setting of ASO or other mRNA-targeted therapeutics. 19 Stable isotope labeling kinetics (SILK) is a technique that enables the study of protein turnover in vivo. 20,21 This approach involves labeling participants with a non-radioactive, stable isotope-labeled tracer (e.g., [U-13 C 6 ]L-leucine) and measuring its incorporation into newly synthesized proteins using quantitative mass spectrometry. SILK has been successfully employed to examine the dynamics of key proteins implicated in Alzheimer's disease including amyloidb, tau, and ApoE in human CSF. [22][23][24][25][26] Using SILK, we previously demonstrated that human CSF SOD1 is a long-lived protein with a half-life of 25 days. In SOD1 G93A transgenic rats, SOD1 turnover was similar in CSF compared to spinal cord indicating that CSF SOD1 turnover likely reflects SOD1 kinetics in spinal cord. 6 Furthermore, decreasing SOD1 production in the brain and spinal cord of rodents leads to parallel reduction in CSF SOD1 19 indicating that CSF SOD1 can be used as a proxy for SOD1 expressed in the central nervous system.
Here, we measure in vivo SOD1 protein kinetics in fALS, sALS, and healthy control participants. We show that mutant SOD1 protein turnover is faster compared to wild-type protein in SOD1 A5V mutation carriers, providing in vivo evidence that certain SOD1 mutations can alter protein stability. In addition, SOD1 A5V protein is expressed at~6% of the level of wild-type SOD1. To determine whether altered SOD1 metabolism could play a role in non-SOD1 ALS, we compared the half-life of CSF SOD1 protein in sALS and C9ORF72 hexanucleotide repeat expansion carriers (C9HRE) to controls and found no differences suggesting that SOD1 turnover is not altered in non-SOD1 ALS. These results elucidate the pathological contexts in which altered SOD1 kinetics impacts ALS pathogenesis and paves the way for use of this method to examine the impact of targeted therapeutics on long-lived proteins.

Cohort recruitment
The study was approved by the Washington University Human Studies Committee (WU IRB# 201207043) and Massachusetts General Brigham Human Research Committee (MGH IRB# 2016P000249). Informed consent was obtained from all participants prior to study inclusion. All research participants had an initial screening visit consisting of a neurological examination and phlebotomy for complete blood count, complete metabolic panel, prothrombin time, and partial thromboplastin time.
Inclusion criteria for controls included age > 18 years and no evidence of ALS by history or examination. ALS participants were eligible if they were > 18 years of age, diagnosed with possible, probable, or definite ALS by El Escorial criteria, and able to tolerate lumbar puncture. Exclusion criteria included dependence on invasive ventilation, contraindication to lumbar puncture (i.e., bleeding disorder, allergy to local anesthetics, evidence of increased intracranial pressure, or skin infection near lumbar puncture site), connective tissue disease, dermatologic disease, adherence to special diet (i.e., glutenfree), pregnancy, or lab values exceeding twofold upper limit of normal.
Participants with ALS underwent evaluation of ALS functional rating score (ALSFRS-R) and slow vital capacity at each visit. Blood samples from ALS participants were sent to Prevention Genetics for testing to determine presence of mutations in SOD1 or C9orf72. Participants with sALS were screened and found negative for mutations in C9orf72 and SOD1.

Stable isotope labeling
Two methods of 13 C 6 -leucine labeling were employed in participants, a 10 day oral leucine diet as previously described 6 followed by a transition to 16 h intravenous 13 C 6 -leucine infusion to ease participant burden and improve labelling consistency. Participants in the orallabeled leucine cohort received either a controlled leucine diet (2000 mg/ day) prepared by dieticians in the Washington University Research Kitchen or normal diet prepared at home and directed to keep a diet diary for 10 days. The 2000 mg provided in the controlled diet is lower than a typical diet (6000-14,000 mg) but higher than the 1200 mg/ day leucine requirement advised. Labeled U-[ 13 C 6 ]L-leucine (Cambridge Isotope Laboratories, CLM-2261) was provided as 330 mg of 13 C 6 -leucine dissolved in 120 mL Kool-Aid and given three times a day.
Intravenous 13 C 6 -leucine administration was performed using protocols similar to tau SILK studies. 24 Participants recruited from both study sites were admitted to the Clinical Research Unit (CRU) at Washington University, provided a low-leucine diet (500-700 mg leucine on day of infusion) and administered 13 C 6 -leucine for 16 h at a rate of 4 mg/kg/h. Subjects recruited from MGH returned to MGH for subsequent blood and CSF collections.
Preparation of 13 C 6 -leucine-labeled SOD1 standard curve 13 C 6 -labeled SOD1 enrichment standards were prepared by growing HEK293T cells that constitutively express SOD1 in RPMI-1640 media supplemented with varying ratios of labeled/unlabeled media as previously described. 6 Generation of 15 N-and 14 N-recombinant wild-type and mutant SOD1 A hSOD1 cDNA construct with a N-terminal GST tag was subcloned into pGEX4T-1 backbone. Plasmids were propagated in XL-1 Blue E. coli cells and isolated using Qiagen MIDI prep kits (Qiagen). Site-directed mutagenesis (QuikChange II, Agilent) was performed to introduce the p.A5V mutation into pGEX4T-1-hSOD1. Plasmids were transformed into Rosetta 2 E. Coli (Novagen) to generate recombinant human SOD1 protein.
Bacterial cultures were grown overnight at 37°C in 15 N-Celtone complete medium or unlabeled Celtone complete medium (Cambridge Isotope Laboratories), and expression of 15 N-and 14 N-SOD1 recombinant proteins was induced with 1 mM IPTG. Bacterial lysates were incubated with 600 lL glutathione sepharose 4B beads (GE Healthcare) for 30 min. hSOD1 protein was released from GST tag by incubating with 1 unit biotinylated thrombin (Thrombin Cleavage Capture Kit, EMD Millipore) at room temperature for 20 h, followed by addition of 30 lL strepavidin agarose for 30 min to remove thrombin. Supernatant containing hSOD1 protein was collected, quantified by BCA assay, and stored as aliquots at À80°C.
Analysis of 13 C 6 -leucine-labeled SOD1, 13 C 6leucine-labeled total protein, and plasmafree 13 C 6 -leucine by mass spectrometry Mouse monoclonal anti-SOD1 antibodies (Sigma-Aldrich, S2147) were purified and conjugated to M-270 Epoxy Dynabeads (Invitrogen) at a concentration of 25 lg antibody per 1 mg beads as previously described. 6 To enable protein quantification, 100 ng of 15 N SOD1 was added to samples. For analysis of CSF from SOD1 mutation carriers and mutant SOD1 concentration standards, 50 ng of SOD1 and 50 ng of corresponding mutant SOD1 were spiked into samples.
Soluble SOD1 was immunoprecipitated from 1 mL of human CSF collected from participants and HEK293T enrichment standard curve lysates (100 lg protein) using 50 lL of anti-SOD1 crosslinked Dynabeads. The beads were washed with 25 mM ammonium bicarbonate buffer (AmBic, pH 8.0) (Sigma) twice, and the beads were eluted with 100 lL formic acid (Sigma). Immunoprecipitation, PBS washes, and elution steps were automated and performed in 96-well format using purification apparatus equipped with magnet head (Kingfisher Flex). The eluent was lyophilized via speedvac (Labconco CentriVap), and isolated SOD1 was resuspended in 25 mM Ambic (pH 8.0) before undergoing reduction with 2.5 mM dithiothreitol (DTT) (Sigma) for 30 min at 37°C followed by alkylation with 7.7 mM iodoacetamide (Sigma) for 30 min at room temperature. LysC (250 ng) was added and allowed to incubate for 4 h followed by incubation with 250 ng Trypsin (Promega) for 16 h at 37°C. Samples were purified using TopTip C18 tip columns and eluted with 66% acetonitrile/ 0.1% formic acid (Fisher Chemical). Samples were lyophilized and resuspended in 50 lL 2.5% acetonitrile/ 1% formic acid prior to liquid chromatography-mass spectrometry. A TSQ Altis (ThermoFisher) was utilized for all measurements with the exception of SOD1 A5V analyses which were performed on Lumos Orbitrap (ThermoFisher).
Tracer-to-tracee ratios (TTR) were obtained by comparing the area under the curve of 13 C 6 -leucine and 12 C 6leucine signal for leucine-containing SOD1 peptides derived from LysC/ Trypsin digestion and converted to mole fraction labeled (MFL) using the equation, MFL = TTR/ (1 + TTR). One to three transition ions for each peptide were selected for acquisition based on strong signal correlation.
Plasma-free 13 C 6 -leucine TTR (which represents the precursor pool enrichment for SOD1 protein synthesis) was measured as tert-butyldimethylsilyl (tBDMS) derivatives by electron impact ionization gas chromatographymass spectrometry (GC/ MS) as previously described. 24 CSF total protein 13 C 6 -leucine isotopic enrichment was measured as previously described. 6 Compartmental modeling of protein kinetics A compartmental model was used to assess the kinetics of CSF SOD1 and total protein. A forcing function of the plasma leucine 13 C 6 -leucine time course (a linear interpolation of sampled time points from t = 0 through 18 h connected to a sum of 3 exponentials function that defined plasma leucine kinetics from 18 h to the last CSF time point) was used as an input into a system of two serial compartments with identical fractional turnover rates to represent either CSF SOD1 or total protein. Some direct input of newly synthesized protein into the second of the two compartments was required to optimally fit the protein enrichment time courses. Parallel arms of the model implicitly tracked the change in labeled and unlabeled proteins from an initial steady state consisting of only unlabeled protein as the MFL of plasma leucine changed over time, and an adjustable scaling factor was used to account for in vivo isotopic dilution and/or residual errors in the accuracy of SOD1 peptide enrichment measurements. The whole system residence time (days) and fractional turnover rate (FTR, pools/day) were determined using the SAAM modeling program (The Epsilon Group, Charlottesville, VA). Production rates (ng/mL/ day) were calculated as the product of FTR and concentration. Half-life was calculated as ln [2]/ FTR.

SDGPVKV peptide analysis
The SDGPVKV peptide was measured from 50 lL aliquots of CSF by Clarus Analytical as previously described. 27 Statistics Statistical analyses were performed using GraphPad Prism 7.0 (GraphPad). Quantitative data are represented as mean AE SD. Intergroup comparisons of mean values were conducted using two-tailed t-tests assuming normal distribution of the data. Statistical analysis of SDGPVKV levels was conducted using one-way ANOVA. P-values <0.05 were considered significant. Spearman correlation analyses were used to determine correlation between SOD1 kinetic and clinical parameters.

Clinical cohort
We examined six controls, four sALS, and one participant with a SOD1 mutation (p. A5V) who underwent 10 day oral administration of 13 C 6 -leucine (Table 1). We also infused 13 C 6 -leucine intravenously over 16 h in a parallel cohort comprised of five controls, 11 sALS, two C9HRE carriers, and three SOD1-mutation carriers (p. A5V) ( Table 2). Controls recruited into the study were healthy and did not have neurological illness by history or examination. Three of the four SOD1 A5V carriers were asymptomatic. Detailed kinetic analysis was not available from one SOD1 A5V ALS participant (ALS13) who was only able to complete lumbar punctures at days 8, 15, and 29 post-13 C 6leucine infusion before withdrawal from the study.
Reduced levels of mutant protein can occur due to increased clearance via increased protein degradation, accumulation in aggregates, decreased production, or reduced cellular release into CSF. The fractional turnover rate is increased by~twofold for peptide 1 A5V relative to peptide 1 WT and peptide 5 suggesting that mutant protein turns over more rapidly (Table 3). In participants harboring SOD1 A5V , peptide 1 A5V half-life was shorter compared to native peptide 1 WT , showing that SOD1 A5V protein is turned over~twofold faster than SOD1 WT (Table 3, Fig. 2B-E). The turnover of peptide 1 WT was similar to peptide 5, a C-terminal SOD1 peptide, in SOD1 A5V participants as well as control participants (Table 3). Participant ALS13 exhibited divergence in SOD1 A5V and SOD1 WT peptide behavior similar to other SOD1 A5V participants but did not undergo kinetic modeling due to an abbreviated time course (Fig. 2E). Furthermore, the production rate of peptide 1 A5V is~15-18% that of peptide 1 WT and peptide 5 indicating that SOD1 A5V protein has a lower rate of production and/or cellular release compared to SOD1 WT (Table 3). Thus, the~16-fold reduced levels of mutant protein are due to~twofold faster turnover of SOD1 A5V as well as significantly reduced production or release into CSF.

SOD1 kinetics are not altered in sALS cohort
To determine whether altered SOD1 metabolism plays a role in non-SOD1 ALS, we compared the kinetics of CSF SOD1 protein in ALS to controls. Plasma enrichment of 13 C 6 -leucine was similar between control and sALS cohorts for either infused or oral tracer (Fig. 3A,D) and reflect the available labeling pool. At peak,~1-2% of SOD1 protein (Fig. 3B,E) and CSF total protein (Fig. 3C,F) was isotopically labeled in control, sALS, and C9HRE participants using both labeling protocols. The concentration of SOD1 was quantified and found to be similar between sALS and controls (sALS [SOD1] = 63.0 AE 30.9 ng/mL, n = 15; control [SOD1] = 67.7 AE 16.0 ng/mL, n = 12, p = 0.639) (Fig. 3G).
Within the 13 C 6 -leucine infused cohort, SOD1 half-life was similar in sALS, C9HRE participants, and controls ( Table 3). SOD1 half-life in oral-labeled sALS and control participants was also not significantly different ( Table 3).
The half-life of the total CSF protein pool was twofold to fourfold faster compared to SOD1 consistent with prior studies, 6 indicating that CSF SOD1 is a relatively longlived protein (Table 3). Together, these data indicate that the kinetics and concentration of soluble CSF SOD1 are not altered in sALS.

Correlations between sALS clinical characteristics and SOD1 half-life
Correlations between SOD1 kinetic parameters and sALS clinical characteristics including age, body mass index (BMI), rate of ALSFRS-R decline, and disease duration were examined for the infused 13 C 6 -leucine cohort. SOD1 kinetic parameters examined were SOD1 half-life, total protein half-life, SOD1 production rate, and SOD1 concentration. SOD1 half-life correlated with SVC decline (Spearman's r = 0.736, p = 0.013) but no other clinical characteristics. In addition, ALSFRS-R decline correlated with SOD1 concentration (Spearman's r = 0.673, p = 0.039). However, correlations of ALSFRS-R decline with SOD1 production rate (Spearman's r = 0.588, p = 0.081) and total protein half-life (Spearman's r = À0.576, p = 0.088) did not reach significance. Notably, the rate of ALSFRS-R decline in this cohort was relatively slow (mean ALSFRS-R decline = 0.60 AE 0.72 points per month) and only included one fast progressor (ALSFRS-R decline = 2.54 per month), limiting interpretation of these findings.

Deamidated SOD1 peptide, SDGPVKV, is elevated in SOD1 mutation carriers but not in controls or sALS
An endogenous SOD1 peptide, SDGPVKV, was previously identified as a product of intracellular SOD1 degradation whose levels correlated with in vivo protein stability and degradation. 27 SDGPVKV is a deaminated product of the SNGPVKV sequence (amino acids 25-31) of SOD1 (Fig. 4A). Using in silico tools, the sequence was predicted to derive from 20S proteasomal degradation and shown to undergo post-proteolytic asparagine deamidation at N 26 . Asparagine deamidation is promoted by the oxidative environment of neurodegenerative conditions and associated with protein misfolding. SDGPVKV levels are increased in spinal cords of SOD1 G93A mice and were found to be elevated in the CSF of SOD1 mutation carriers compared to controls in a manner that correlates to SOD1 stability changes conferred by individual mutations.

Conclusions
Protein aggregation and impaired proteostasis are pathogenic mechanisms that underlie numerous neurodegenerative disorders. SILK is a method that offers a window to examine in vivo turnover (synthesis and clearance) of key proteins and how these processes are affected in neurodegenerative diseases characterized by misfolding and accumulation of pathogenic proteins. Moreover, measures of protein kinetics can inform optimal dosing schedules for therapies in clinical trials and provide improved biomarkers for assessing target engagement.
We utilized SILK methods to measure in vivo turnover of CSF SOD1 protein in SOD1 A5V mutation carriers. SOD1 A5V mutations comprise~50% of SOD1-ALS cases in North America and cause a form of the disease with rapid progression and average survival of 1.4 years. We demonstrate that mutant protein turnover is rapid compared to corresponding wild-type protein, demonstrating for the first time a divergence in kinetics between mutant and wild-type protein alleles in an inherited neurodegenerative disease. This finding corroborates in vitro and SOD1 rodent model studies that demonstrate faster turnover of mutant protein. 4,6,7 Three of the SOD1 A5V participants were asymptomatic, so SOD1 kinetic changes are evident prior to clinical onset. This observation mirrors findings from SOD1 G85R mutant mice showing that SOD1 G85R protein turns over faster than SOD1 WT even prior to development of clinical symptoms. 7 Aggregates in SOD1 G85R mice have also been shown to develop before onset of clinical symptoms and are among the first pathological signs of disease. 28 Thus, altered SOD1 kinetics may signify propensity to develop disease suggesting that this approach may be able to identify variants of unknown significance in SOD1 that are likely pathogenic. Although the turnover of SOD1 A5V is~twofold faster, the concentration of CSF SOD1 A5V protein is~16-fold lower than SOD1 WT . Faster turnover does not solely account for observed differences in mutant and wild-type protein providing support that mutant protein is not only unstable but undergoes reduced synthesis and/or cellular release into CSF. Prior studies that examined erythrocytes from SOD1 mutation carriers similarly found mutant protein was reduced or undetectable for a subset of variants, including SOD1 A5V . 8 Erythrocytes are enucleated cells devoid of protein synthesis, a property likely to highlight instability of mutant SOD1. The accelerated turnover and reduced stability of SOD1 A5V protein is consistent with a model in which mutated protein misfolds and is cleared via proteostatic mechanisms or incorporated within aggregates.
To further assess the pathogenicity of SOD1 A5V and examine stability of SOD1 in sALS, we measured an endogenous deamidated SOD1 peptide, SDGPVKV, previously shown to reflect the degree of misfolding conferred by distinct mutations. 27 SDGPVKV levels were markedly elevated in symptomatic and asymptomatic SOD1 A5V suggesting a link between the high propensity of the protein to misfold and aggressive phenotype of this mutation. Further clinico-pathologic studies correlating CSF SDGPVKV to SOD1 aggregate burden and disease progression in SOD1 mutation carriers are warranted to define the nature of this biomarker. Nearly 150 mutations in SOD1 have been described. Pathogenic SOD1 mutations are hypothesized to cause destabilization, misfolding, and aberrant toxic aggregation of mutant and wild-type protein. Multiple in vitro and in vivo studies have characterized the chemically and structurally diverse consequences of SOD1 mutations. 29 Although factors including aggregation potential, protein stability, and potential for mutant-to-wild-type heterodimerization have been associated with disease progression, 8,30-32 they are not altogether predictive of clinical phenotype. Given the variety of variants in SOD1, approaches are needed to ascertain disease-causing mutations to facilitate identification of SOD1 mutation carriers for clinical trials of SOD1-lowering therapies. Expanding the array of SOD1 mutations examined on the basis of kinetic behavior will further clarify determinants of pathogenicity. Correlation of kinetic parameters to SOD1 inclusion burden will also be beneficial. However, determining inclusion burden is challenging as PET-based modalities that employ aggregate-binding radiotracers have not been developed for SOD1 aggregates and postmortem analysis may be temporally dissociated from relevant measures. Nonetheless, the methods developed to measure the concentration as well as turnover of mutant and wild-type SOD1 provide a versatile platform for characterizing pathogenicity of SOD1 variants in vivo.
The involvement of SOD1 aggregates in sALS is controversial as some studies have identified SOD1 inclusions in brain and spinal cord from sALS patients by immunohistochemistry. 10,11 However, other studies were unable to recapitulate these findings. [12][13][14] Recent biochemical and proteomic analysis of post-mortem tissue suggests that disordered SOD1 conformers associated with mismetallation, altered post-translational modifications, and loss of enzymatic activity accumulate in the ventral spinal cord of SOD1-linked and non-SOD1 linked ALS. 15 Misfolding of native SOD1 in sALS would suggest a common disease mechanism between familial and sporadic forms of the disease that could be targeted by SOD1-lowering therapies. Our results indicate that wild-type SOD1 turnover is not altered in the CSF of non-SOD1 ALS compared to controls. We also determined that CSF SOD1 levels are similar between sALS and controls consistent with studies that employed immunoassay-based methods. 33,34 Notably, the levels of deaminated SOD1 peptide in sALS did not differ from controls suggesting that the stability of SOD1 may not be principally affected in sporadic cases. Together, these findings indicate that turnover of soluble SOD1 is not significantly altered in non-SOD1-ALS. Nearly 80% of CSF proteins derive from blood while only~20% are neuronally derived. 35 Although the proportion of blood-derived SOD1 in CSF is unknown, its presence may obscure the contribution of disordered SOD1 from spinal cord. Given this caveat, we are unable to exclude a contribution of native SOD1 to sALS pathogenesis based on these studies. Several limitations complicate interpretation of these results. First, the sample size of SOD1 mutation carriers, non-SOD1 ALS, and control cohorts was relatively small. Second, we observed more rapid turnover of the CSF total protein pool in orally labeled compared to infused cohorts so these groups were analyzed separately. This suggests that 13 C 6 -leucine labeling paradigms may label distinct subsets of proteins with different kinetic profiles, or that the compartmental model that uses plasma leucine enrichment is unable to resolve local (in situ) tracer recycling issues that affect the accuracy of assessing protein FTR. Third, the SILK method involves immunoprecipitation of SOD1 using antibodies to native SOD1 and may not fully capture modified SOD1 species such as misfolded SOD1. However, we were able to quantitatively measure mutant SOD1 protein and discern rapid turnover of SOD1 A5V . In addition, employing antibodies to native SOD1 allows application of the assay to pharmacodynamic studies of SOD1-lowering therapies that target wild-type SOD1. Assay modifications that allow specific targeting of misfolded SOD1 or PTMs associated with disordered SOD1 conformers may improve sensitivity for detecting altered protein turnover. Fourth, the factors that dictate release of SOD1 into CSF are not well understood,  and further studies are needed to elucidate these mechanisms including the role of passive (i.e., neurodegeneration) and/or active processes (i.e., secretory release) in release, contribution of glymphatic clearance, and cellular source of CSF SOD1. SOD1-lowering strategies that reduce SOD1 mRNA and inhibit synthesis of new SOD1 protein have emerged as promising therapeutic approaches for SOD1-ALS. In the Phase III trial of tofersen in SOD1-ALS, CSF SOD1 protein declined by~30% at 12 weeks in the high-dose group. 16 Given the long half-life of SOD1 of~30 days, reductions in total CSF SOD1 protein would be expected to lag ASO-mediated inhibition of SOD1 mRNA and reduced protein synthesis. Notably, SOD1 G93A rats treated with SOD1-lowering ASO have decreased rates of wildtype SOD1 protein synthesis that were observed prior to decreases in total CSF SOD1 protein. 19 The ability to characterize kinetic behavior of mutant SOD1 protein advances use of this method to clarify impact of ASO treatment on mutant protein concentration, the primary toxic therapeutic target, and improve clinical trial design. We propose that applying protein kinetic analysis to clinical trials has the potential to establish earlier pharmacodynamic biomarkers of target engagement and guide optimal timing and dosing for mRNA-directed therapeutics.

Author Contributions
CVL designed and conducted SOD1 immunoprecipitation and plasma leucine GC/MS experiments, acquired clinical data, analyzed data, and wrote the manuscript. MDI assisted with SOD1 immunoprecipitation experiments, processed human biofluid samples, conducted and analyzed plasma leucine GC-MS data, and edited the manuscript. WS helped develop SOD1 SILK methods, contributed intellectual guidance, and edited the manuscript, JB provided intellectual guidance and acquired mass spectrometric data. JJB, HH, PA, LM, MD, and IAB provided research coordination and acquired participant demographic data, BT acquired clinical data. TK ascertained quality control of clinical specimens. KN provided site coordination of the clinical study at MGH. RB acquired clinical data. BWP and RJB provided intellectual guidance and edited the manuscript and RJB supervised the SOD1 measures including method development and edited the manuscript. TMM supervised the study, provided intellectual guidance, provided financial support, and edited the manuscript.