Integrative Profiling of Amyotrophic Lateral Sclerosis Lymphoblasts Identifies Unique Metabolic and Mitochondrial Disease Fingerprints

Amyotrophic lateral sclerosis (ALS) is a devastating neurodegenerative disease with a rapid progression and no effective treatment. Metabolic and mitochondrial alterations in peripheral tissues of ALS patients may present diagnostic and therapeutic interest. We aimed to identify mitochondrial fingerprints in lymphoblast from ALS patients harboring SOD1 mutations (mutSOD1) or with unidentified mutations (undSOD1), compared with age-/sex-matched controls. Three groups of lymphoblasts, from mutSOD1 or undSOD1 ALS patients and age-/sex-matched controls, were obtained from Coriell Biobank and divided into 3 age-/sex-matched cohorts. Mitochondria-associated metabolic pathways were analyzed using Seahorse MitoStress and ATP Rate assays, complemented with metabolic phenotype microarrays, metabolite levels, gene expression, and protein expression and activity. Pooled (all cohorts) and paired (intra-cohort) analyses were performed by using bioinformatic tools, and the features with higher information gain values were selected and used for principal component analysis and Naïve Bayes classification. Considering the group as a target, the features that contributed to better segregation of control, undSOD1, and mutSOD1 were found to be the protein levels of Tfam and glycolytic ATP production rate. Metabolic phenotypic profiles in lymphoblasts from ALS patients with mutSOD1 and undSOD1 revealed unique age-dependent different substrate oxidation profiles. For most parameters, different patterns of variation in experimental endpoints in lymphoblasts were found between cohorts, which may be due to the age or sex of the donor. In the present work, we investigated several metabolic and mitochondrial hallmarks in lymphoblasts from each donor, and although a high heterogeneity of results was found, we identified specific metabolic and mitochondrial fingerprints, especially protein levels of Tfam and glycolytic ATP production rate, that may have a diagnostic and therapeutic interest.


Introduction
Amyotrophic lateral sclerosis (ALS) disease, the most common motor neuron disorder [1][2][3] is characterized by motor neuron degeneration, progressive muscle atrophy, and paralysis that ultimately leads to death [4], without diseasemodifying treatments available [5][6][7][8]. On average, the disease affects people between 51 and 66 years of age, with a survival time, from symptoms onset to death, between 24 and 50 months [1]. Approximately 90-95% of ALS cases are sporadic (sALS), with unknown etiology and no prior family history [9]. The remaining patients develop familial ALS (fALS) [9], which is inherited and associated with an earlier age of onset [5,10]. Growing evidence demonstrates that fALS and sALS can be triggered by common pathological mechanisms, presenting similar clinical features [9][10][11]. Several molecular mechanisms play a potential role in the disease development, including altered RNA metabolism, glutamate excitotoxicity, protein misfolding and aggregation, endoplasmic reticulum stress, disrupted protein trafficking, defective axonal transport, oxidative stress, inflammation, and mitochondrial dysfunction [2,3,[12][13][14]. Which events are causative of the disease, rather than a consequence, is still unknown. In particular, mitochondrial dysfunction is considered to play a fundamental role in ALS progression, and multiple mitochondrial alterations have been described not only in the central nervous system (CNS) of ALS patients but also in peripheral tissues, such as skeletal muscle [15,16], liver [17], and lymphocytes [18]. Indeed, the disruption of mitochondrial structure, dynamics, bioenergetics, and calcium buffering has been reported in ALS patients, in vitro and in vivo ALS models [9,19]. A considerable number of genes were discovered to be disease-modifying [20]. One of the most studied genes in ALS research, implicated in both forms of ALS, is SOD1. Around 220 mutations have been identified so far as modifiers of the SOD1 Cu/Zn superoxide dismutase 1 (SOD1) amino acid sequence (https:// alsod. ac. uk/ output/ gene. php/ SOD1, accessed 29 June 2022), affecting its normal function [21][22][23][24][25], SOD1 is a metalloprotein responsible for eliminating superoxide anion radicals, localized mainly in the cytosol but also found in the nucleus, peroxisomes, and mitochondria [23]. Alterations in mitochondrial respiration have been extensively described in mutant SOD1 (mutSOD1) mouse models [26][27][28][29][30][31] and in ALS patients with mutSOD1 [27,32]. Although there is evidence of SOD1 accumulation in mitochondria [26,[33][34][35][36][37], the real function of this enzyme in this organelle remains unclear. Most studies suggest that wild-type and mutSOD1 accumulate mainly in the mitochondrial intermembrane space (IMS) [26,[33][34][35]37], although mutSOD1 was also shown to be localized on the outer mitochondrial membrane (OMM) [38,39], associated with proteins such as B cell lymphoma 2 (Bcl-2) [39] and voltage-dependent anion-selective channel (VDAC) [40], or accumulate in the mitochondrial matrix [41]. Contradictory results can be found in the literature concerning the effect of mutSOD1 on the mitochondrial electron transport chain (ETC) activity. Increased complex I activity in the frontal cortex [32] and in the motor and parietal cortices [27] were described in patients carrying mutSOD1 A4V . In these patients, complex II-III activity was also increased in the motor cortex and cerebellum and in the motor and parietal cortices, also occurring in the cerebellum from a single patient with mutSOD1 Ill 3 T , as well as in the motor cortex and cerebellum of ALS patients with no mutSOD1 identified to date [27]. Thus, it is possible that the increased ETC activity represents a compensatory mechanism resulting from oxidative damage to the IMM, which can result in the uncoupling of oxidative phosphorylation [42]. In postmortem spinal cord tissues from sALS patients, mitochondrial impairment related to the decrease in complexes I, II, III, and IV activities was shown to be associated with a reduction in mtDNA content and citrate synthase activity [43]. Conversely, no alteration in ETC activity was found in crude mitochondrial preparations of the motor and parietal cortices and cerebellum [27]. These contradictory findings can be explained by a more pronounced impairment of ETC function in the spinal cord when compared to brain tissue, by a small number of samples analyzed in both studies, and by the heterogeneity of sALS patients [44]. Altered ETC activities were also detected in non-neuronal tissues, such as reduced complexes I and IV activities in skeletal muscle [43,45] and reduced complex I activity, together with lower ATP levels, in lymphocytes [46]. Increased mitochondrial membrane potential (ΔΨm) in fibroblasts from sALS patients [47,48] may indicate a compensatory mechanism to rescue inefficient ATP synthesis, although further studies are needed to confirm these variations. Importantly, although there is strong evidence of a general mitochondrial dysfunction in ALS, whether this represents a cause or a consequence of other intracellular toxic mechanisms remains to be elucidated [9]. In sum, metabolism plays a crucial role in ALS pathogenesis, and mitochondrial dysfunction is likely to be one of the earliest pathophysiological events in ALS [9]. Further research is needed to allow patient stratification and find fingerprints and therapeutic targets for ALS based on mitochondrial metabolism. Because ALS is a multisystemic disease, lymphoblasts have been increasingly used to study metabolic characterization in ALS since they share molecular mechanisms, especially mitochondrial dysfunction, with degenerated neurons [49]. Lymphoblasts enable the analysis of multiple variables from the same biological source, allowing identifying internal biological correlations and testing personalized therapeutics. In the present study, by performing a multidimensional analysis of multiple end-points collected in the same individual cells, we identified mitochondrial fingerprints in lymphoblasts from ALS patients harboring SOD1 mutations [(mutSOD1); two patients carrying mutSOD1 A4V and one carrying mutSOD I113T ] or with still unidentified SOD1 mutations (undSOD1), compared to their age-and sex-matched controls. This study describes the potential role of circulating mitochondrial disruption markers for patient stratification and identifies metabolic targets for developing novel disease-modifying drug candidates.

Cell Selection/Experimental Design
Lymphoblast cell lines, established by Epstein-Barr transformation, from 3 patients with mutSOD1 and 3 patients without identified SOD1 mutations (undSOD1), and 3 controls of the same age and sex were obtained from the cell line repository at Coriell Institute for Medical Research, USA (www. corie ll. org). The groups were divided into 3 cohorts, each containing 1 sample from healthy control, 1 sample from one undSOD1 patient, and 1 sample from one mutSOD1 patient: C1-females 46 years old; C2-males 46 years old; C3-males 26/27 years old ( Table 1). The redox profiles of these cells were recently characterized by our group [50].

Cell culture
Lymphoblasts were grown in RPMI 1640 medium from Sigma-Aldrich (Saint Louis, USA, R6504) supplemented with 2 g/L sodium bicarbonate, 15% (v/v) fetal bovine serum, and 0.5% (v/v) of penicillin-streptomycin plus amphotericin B in T25 or T75 flasks, in an upright position, at 37 °C in a humidified atmosphere of 5% CO 2 . The cells were kept in the culture at a density of 0.4 to 1.2 × 10 6 cells/mL and counted using a Bio-Rad® TC20 Automated Cell Counter (Bio-Rad, Hercules, CA, USA). For cell counting, 10 μL of each cellular suspension was diluted in 0.4% Trypan blue solution from Gibco (Hampton, USA, 15,250-061) in a ratio 1:1. Every 2-3 days, the cultures were diluted and re-fed with fresh medium according to the rate of cell growth and the required number of cells needed for the experiments, and once per week, the medium was renewed entirely after cell centrifugation. Before all experiments, lymphoblasts were plated at 0.4 or 0.7 × 10 6 cells/mL in RPMI medium and incubated for 48 h or 24 h, respectively, to secure a cell density of around 1 × 10 6 cells/mL in each assay. Considering that doubling times can affect cell phenotype and responses, cells were discarded after 30 population doublings from the initial culture provided by Coriell, and new cultures were started from aliquots frozen at low cell passage.

Real-time Cell Metabolic Analysis
Cellular oxygen consumption and extracellular acidification were measured at 37 °C using an XFe96 Extracellular Flux Analyzer (Agilent Technologies, Inc. Santa Clara, CA, USA). Cells were seeded in XF96 cell culture microplates, pre-coated with 0.1 mg/mL of poly-d-lysine hydrobromide. For the coating, we added to each well 30 µL of a solution (0.1 mg/mL) of poly-d-lysine hydrobromide (Sigma-Aldrich, Ref. P7280, Lot SLBW6178) and let it rest for 2 h at 37 °C. After, we rinsed each well with Milli-Q water and placed the plate at 4 °C, in sterile conditions, to use the next day.

Lymphoblast Seeding in Poly-D-lysine Pre-coated XF96 Cell Culture Microplates
On the day of the assay, lymphoblasts were centrifuged at 259 × g for 4 min. The supernatant was discarded, and cells were resuspended in 1 mL of respective assay media. Cell density was determined using a TC20™ automated cell counter (Bio-Rad Laboratories, Inc. CA, USA). Cells were diluted in assay media at a 3.2 × 10 6 cells/mL density, and 50 µL of the cell suspension was added to each well. For the background wells, 50 µL of assay media was added. Following, 125 µL of assay media was added to all the wells, and the plate was centrifuged at 200 × g for 1 min with no brake. After centrifugation, we confirmed cell adherence to the well bottom and placed the plate in a non-CO 2 incubator at 37 °C for 1 h before the assay. The assay used to evaluate mitochondrial bioenergetics (known as Mitochondrial Stress Test) consisted of 3 serial injections of oligomycin, FCCP, and rotenone/antimycin A. Selected optimized concentrations of oligomycin, FCCP, and rotenone/antimycin A were respectively 3 µM, 0.25 µM, and 1 µM. Normalization was performed by the number of cells seeded in each well. The mitochondrial bioenergetics-related parameters were analyzed using the Agilent Seahorse XF Cell Mitostress test report generator software (version 2.6.0).

Real-time ATP Production Rate Assay
Real-time ATP production rate assay was performed in serum-free RPMI medium ( Real-time ATP production rate assay consisted of a sequential injection of oligomycin (3 µM) and a mixture of rotenone/antimycin A (1 µM of each inhibitor). Normalization was performed by the number of the cells seeded. The XF real-time ATP rate assay parameters were calculated using the Agilent Seahorse XF real-time ATP rate report generator software (version 2.6.0).

Metabolic Microarrays
The metabolic microarrays were performed using the Biolog MitoPlate S-1 system (Biolog, Hayward, CA, USA), which contained triplicate wells for 31 different substrates of mitochondrial or glycolytic metabolism dried and pre-coated into the wells (rows A-B cytoplasmic, rows C-H mitochondrial) plus three negative wells (no substrate).
On the day of the assay, 30 µL of assay mix composed by Biolog mitochondrial assay solution (MAS) 2x (#72,303, Lot 0,722,191), Redox dye MC 2 × (#74,353, Lot 198,019), and saponin 100 µg/mL (SAE0073, Sigma Aldrich) were added to each well of MitoPlate, and incubated for 1 h, at 37 °C. Lymphoblasts were centrifuged at 259 × g for 4 min, resuspended in 1 mL of Biolog MAS 1 × , and cell density was determined using a TC20™ automated cell counter (Bio-Rad Laboratories, Inc. CA, USA). Lymphoblasts were resuspended at 5.3 × 10 6 cells/mL in Biolog MAS 1 × , and 30 µL of the cell suspension was added to each well. The plates were sealed and then incubated in an OmniLog phenotypic microarray system for 6 h for kinetic analysis, with 5-min intervals. The maximal rate (measured in "OmniLog Units per hour" [51]) was calculated in the linear phase (first 4 h), using the data analysis® 1.7 software (Biolog Hayward, CA, USA).

Analysis of Gene Expression
Total RNA was extracted from dry pellets containing ~ 5 × 10 6 cells using RNeasy mini-kit (Qiagen, Düsseldorf, Germany), following the manufacturer's protocols, and quantified using a Nanodrop 2000 (ThermoScientific, Waltham, MA, USA), confirming that the A260/280 ratio was higher than 1.9. RNA was converted into cDNA using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, USA), following the manufacturer's instructions. RT-PCR was performed using the SsoFast EvaGreen Supermix, in a CFX96 real-time PCR system (Bio-Rad, Hercules, CA, USA), with the primers defined in Table 2, at 500 nM. Amplification of 12.5 ng of cDNA was performed with an initial cycle of 30 s at 95.0 °C, followed by 40 cycles of 5 s at 95 °C plus 5 s at the annealing temperature (Ta) shown in Table 2. At the end of each cycle, EvaGreen fluorescence was recorded to enable the determination of Cq. After amplification, the melting temperatures of the PCR products were determined by performing melting curves. For each set of primers, amplification efficiency was assessed using tenfold dilutions of a pool of all samples, and no template and no transcriptase controls were run. Relative normalized expression was determined by the CFX96 Manager software (v. 3.0; Bio-Rad), using TBP, YWHAZ, PUM1, and B2M as reference genes and divided by the mean of the pool of controls.

Analysis of mtDNA Copy Number by Quantitative Real-time PCR
Lymphoblast cell lines were collected and centrifuged at 259 × g for 5 min. The pellets were washed in 5 mL of PBS, and the suspension was centrifuged at 259 × g for 5 min. The resulting pellets were stored at -80 °C until DNA extraction. Total DNA was extracted from dry pellets containing ~ 5 × 10 6 cells using the QIAamp DNA mini-kit (Qiagen, Düsseldorf, Germany), following the manufacturer's protocols, and quantified using a Nanodrop 2000 (Thermo-Scientific, Waltham, MA, USA). DNA was sonicated for 10 min to avoid dilution bias, and RT-PCR was performed using the SsoFast EvaGreen Supermix, in a CFX96 real-time PCR system (Bio-Rad, Hercules, CA, USA), with the primers defined in Table 2, at 500 nM. Amplification of 25 ng DNA was performed with an initial cycle of 2 min at 98 °C, followed by 40 cycles of 5 s at 98 °C plus 5 s at 60 °C. At the end of each cycle, EvaGreen fluorescence was recorded to enable the determination of Cq. For quality control, no template controls were run, and the melting temperature of the PCR products was determined after amplification by performing melting curves. To assess amplification efficiency and inter-run calibration, ten-fold dilutions of one sample were run in all plates. mtDNA copy number was determined in each sample by the relative quantities of the mitochondrial gene ND5 and the B2M single-copy nuclear gene, using the CFX96 Manager software (v. 3.0; Bio-Rad).

Mitochondrial Transmembrane Potential
The Δψ m was assessed by measuring tetramethylrhodamine methyl ester perchlorate (TMRM, #T668, Invitrogen) fluorescence by flow cytometry. TMRM is a lipophilic cationic dye that accumulates within mitochondria in inverse proportion to Δψm according to the Nernst equation [52]. For that purpose, 3 × 10 6 cells in a suspension of each lymphoblast cell line were centrifuged at 259 × g, at room temperature, for 5 min. The supernatant was then discarded, and the pellet resuspended in 900 μL of phosphate-buffered saline (PBS, containing 137 mM of NaCl, 2.7 mM of KCl, 1.4 mM of K 2 HPO 4 , and 4.3 mM of KH 2 PO 4 , at pH 7.4) and volume equally distributed into 3 microtubes. Afterwards, cells of the first microtube were kept at 37 °C, for 15 min, in the dark (control condition for fluorescence background), while cells in the other two microtubes were incubated with TMRM (at a final concentration of 100 nM) in the same conditions as the control microtube. At the end of this incubation period, FCCP (at a final concentration of 1 µM) was added to cells  Ta   NDUFA9  NM_005002  GCC TAT CGA TGG GTA GCA AGAG  TGA GTT CCA GTG GTG TTG CC  60  SDHA  NM_004168  CGG GTC CAT CCA TCG CAT AAG  TAT ATG CCT GTA GGG TGG AAC TGA A  60  UQCRC2  NM_003366  TTC AGC AAT TTA GGA ACC ACCC  GGT CAC ACT TAA TTT GCC ACCAA  60  CYCS  NM_018947  CGT TGA AAA GGG AGG CAA GC  TCC CCA GAT GAT GCC TTT GTTC  60  COX4i1  NM_001861  GAG AAA GTC GAG TTG TAT CGCA  GCT TCT GCC ACA TGA TAA CGA  60  ATP5G  NM_001002027  GGC TAA AGC TGG GAG ACT GAAA  GTG GGA AGT TGC TGT AGG AAGG  60  TFAM  NM_003201  GTT TCT CCG AAG CAT GTG  GGT AAA TAC ACA AAA CTG AAGG  60  G6PD NM_000402

Western Blotting
Total cellular extracts were obtained from a volume of cell suspension containing 30 × 10 6 cells of each sample, which was centrifuged at 259 × g, at room temperature, for 5 min, and the resulting pellet was washed once with PBS.

Intracellular ATP Levels
Intracellular ATP levels were measured using CellTiter-Glo Luminescent Cell Viability Assay (G7571, Promega) following the manufacturer's instructions. Cells were seeded at a concentration of 160,000 cells in white opaque-bottom, 96-well plates, with a final volume of 50 μL per well and treated or not with 3 µM oligomycin or 50 mM 2-Deoxyglucose (2-DG). After 3 h, 50 μL CellTiter-Glo® Reagent (CellTiter-Glo Buffer + CellTiter-Glo Substrate) was added to the cells. Contents were mixed for 2 min on an orbital shaker to induce cell lysis and, after 10 min of incubation at 22 °C, the luminescence signal was monitored in a Biotek Cytation 3 spectrophotometer (BioTek Instruments Inc., USA). The luminescence signal was proportional to the amount of ATP present in the solution [55], interpolated into the ATP standard curve prepared following the manufacturer's instructions.

Lactate levels
Extracellular lactate levels were determined colorimetrically using the l-Lactate Assay Kit (LC2389, Randox), as recommended by the manufacturer. In this assay, lactate is oxidized by lactate oxidase to generate pyruvate and H 2 O 2 , which reacts with 4-aminoantipyrine and N-ethyl-N-(2-hydroxy-3-sulphopropyl)-m-toluidine to produce a purple product proportional to the amount of lactate. Lymphoblasts were seeded at a concentration of 3.2 × 10 6 cells/mL in 12-well plates, with a final volume of 1 mL per well and treated with 3 µM of oligomycin. After 24 h, the cell suspension was centrifuged at 300 × g for 5 min at 4 °C, and the supernatant was collected. The kit reagent was added to the supernatant and incubated 5 min, at 37 °C. The absorbance was measured at 550 nm in a Biotek Cytation 3 spectrophotometer (BioTek Instruments Inc., USA). A lactate standard curve was performed following the manufacturer's instructions.

Lactate dehydrogenase activity
LDH activity was spectrophotometrically measured using pyruvate and NADH as substrates, according to [56]. Total cellular extracts were obtained from a volume of cell suspension containing 30 × 10 6 cells of each sample, centrifuged at 259 × g, at room temperature, for 5 min. The supernatant was discharged, and the resulting pellet was washed in PBS with 0.5% (v/v) Triton X-100. The samples were submitted to 3 cycles of freezing in liquid nitrogen and then centrifuged at 12 000 × g, for 10 min. The protein content of each sample was then determined by the bicinchoninic acid method (BCA; 4,4′-dicarboxi-2,2′-biquinoline) [57]. Briefly, 0.5 μg of each sample was incubated in the reaction buffer containing 25 mM of Tris-NaCl, pH = 8, and 1.464 mM of monosodic pyruvate, at 37 °C, and enzymatic activity was followed by the decrease in the absorbance at 340 nm due to the oxidation of NADH, upon the addition of 0.0732 mM of NADH, for 5 min in a VICTOR Multilabel Plate Reader (Perkin Elmer Inc.,Waltham, MA, USA). LDH activity was determined by using an ε 340 nm = 0.63 mM −1 mm −1 by subtracting the basal and the reaction slopes (i.e., in the absence and presence of NADH). Results were expressed as %/(control).

Hexokinase Activity
Hexokinase activity was spectrophotometrically measured based on the reduction of NADP + to NADPH, which occurs during the phosphorylation of glucose into glucose-6-phosphate [58]. The cellular extracts were prepared as described in the determination of LDH activity. In this assay, 20 μg of each sample was incubated in reaction buffer containing 37.94 mM of Tris-NaCl pH = 8, 10 mM of MgCl 2 , 1.1 mM of ATP, 1.2 mM of NADP + , and 2 U/mL of glucose 6-phosphate dehydrogenase. The enzymatic activity was measured by following at 340 nm the reduction of NADP + at 37 °C, after adding 216 mM of glucose for 5 min in a VIC-TOR Multilabel Plate Reader (Perkin Elmer, Waltham, MA, USA). Hexokinase activity was determined using an ε 340 nm = 0.63 mM −1 mm −1 , by subtracting the basal and the reaction slopes (i.e., in the absence and presence of glucose). Results were expressed as %/(control).

Computational Data Analysis
Orange 3.27.1 [59] was used for the computational data analysis and visualization. Features were ranked regarding entropy based on the information gained between each individual feature and the targets. Five targets were considered to be relevant for the analysis: group (control, undSOD1, mutSOD1), disease (control, ALS), mutation (no mutSOD1, mutSOD1), sex (male, female), and age (26/27 years old, 46 years old). The top 3 features and any additional attributes with identical information gain values were considered for their potential to distinguish samples according to the targets. Linear projections were generated for the selected features after applying principal component analysis (PCA) for dimensionality reduction. As the selected features provided a good visual separation between the experimental classes, we used them to train a Naïve Bayes classifier model, using leave-one-out cross-validation. Performance evaluation was based on confusion matrices, representing the num-

Statistics
Paired and pooled statistical analyses were performed using GraphPad Prism 8.02 software (GraphPad Software, Inc., San Diego, California, USA). For pooled analysis, the data were organized by the experimental group, aggregating all controls, all undSOD1, and all mutSOD1 donors, resulting in 3 data points contributing to each bar, one from each cohort. In the paired analysis, variability is mainly due to technical reasons, since no biological variability should be present, as the cells come from the same donor. Thus, the replicates in the paired analysis correspond to different measurements performed independently in different cell populations from the same donor. This variability can be due to cell culture artifacts that should always be considered, although we always try to keep the conditions as constant as possible. In contrast, in the pooled analysis, the main factors of variability are of biological origin, between the different donors. Thus, for each donor, the average value obtained for each individual was considered. Data are represented in bar graphs (mean ± SEM) with dot plots for the number of experiments indicated in figure legends. Statistical significance was set at p < 0.05 and determined by the Kruskal-Wallis method, followed by Dunn's multiple comparisons test.

Results
ALS is a rare and heterogeneous disease, and for this reason, within the limitations of the samples available at the Coriell Institute, we chose to gather a diverse, rather than a homogeneous, sample. However, as there may be other confounding factors associated with this decision, we matched the samples by age and sex and performed an analysis inside these more homogenous cohorts, each containing 1 sample from healthy control, 1 sample from one undSOD1 patient, and 1 sample from one mutSOD1 patient. These cohorts were identified as C1-females 46 years old; C2-males 46 years old; C3-males 26/27 years. Notwithstanding, our final goal was to obtain a general view of ALS metabolic profile, and thus, we also analyzed the samples as a pool of 3 independent individuals in each group, aggregating all controls, all undSOD1, and all mutSOD1 donors.

Mitochondrial Respiration and Glycolytic Fluxes
The mitochondrial function and metabolic profile of lymphoblasts from ALS patients and age and sex matching controls were evaluated by measuring the oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) using the Agilent-Seahorse XFe96 analyzer. Modulators of mitochondrial respiration (3 µM of oligomycin, 0.25 µM of FCCP, and 1 µM of rotenone/antimycin A) were sequentially added to determine basal respiration, oxygen consumption associated with ATP synthesis, maximal respiration, and oxygen consumption associated with proton leak. The ECAR was simultaneously evaluated. From the individual analysis of each cohort, we observed that although no significant differences were observed regarding basal respiration in the C1 cohort (lymphoblasts from 46-year-old females) (Fig. 1A), for the same group an increase in the proton leak-associated OCR and spare respiratory capacity in the ALS patient with unknown SOD1 mutation (undSOD1), comparatively to the matched control (Fig. 1B, C) was observed. Furthermore, an increase in baseline and stressed (after oligomycin and FCCP addition) ECAR was also observed in undSOD1, when compared with the control (Fig. 1D, E), suggesting that lymphoblasts from the ALS patient with undSOD1 were metabolically more active. Inside the same cohort, comparing lymphoblast from undSOD1 with lymphoblast from the ALS patient with mutSOD1, the latter showed lower basal respiration (Fig. 1A), ATP production-driven OCR (Fig. 1F), maximal respiration (Fig. 1G), non-mitochondrial respiration (Fig. 1H), and bioenergetic health index (Fig. 1I), appearing to be metabolically more glycolytic (Fig. 1J).
When pooling the samples from all cohorts, we did not observe statistically significant differences in the analyzed parameters, including in the energy map (Fig. 1M). High data variability and a low number of samples in each group are the possible reasons for the lack of statistical differences when performing this type of analysis, despite the individual differences found in terms of disease condition, sex, and age of the individual.

ATP-Generating Pathways
After metabolic flux evaluations, we next measured the contribution of glycolysis and oxidative phosphorylation to ATP production. The cellular ATP production rate was quantified using Agilent Seahorse XF technology by sequentially inhibiting the ATP synthase with oligomycin and the mitochondrial ETC with rotenone and antimycin A. Lymphoblasts from undSOD1 ALS patients presented a higher ATP production rate derived from glycolysis ( Fig. 2A). In C1, the mitochondrial contribution to ATP production rate in mutSOD1 lymphoblasts was lower than in undSOD1 (Fig. 2B). When calculating the ATP rate index, the rate of mitochondrial vs. glycolytic-produced ATP, there was an overall decrease in the ALS groups when comparing with the respective controls, although only statistically significant for the mutSOD1 C1 cohort (Fig. 2C). Furthermore, in C1, the contribution of glycolysis for ATP production (Fig. 2D) were higher in mut-SOD1 than in controls, without significant alterations in LDH activity (Fig. 2E). The results observed in C1 are in agreement with the increased hexokinase activity (Fig. 2F), higher ATP levels without (Fig. 2G), and with oligomycin (Fig. 2H), and increased lactate levels (Fig. 2I). The data suggest that the glycolytic pathway has a high contribution to ATP production in lymphoblasts isolated from undSOD1 ALS patients. In fact, in the presence of 50 mM of 2-DG, a 66.7% reduction in ATP levels was observed in lymphoblasts from undSOD1 patients ( Fig. 2J; in C1 and C2). On the other hand, lymphoblasts from mutSOD1 patients presented a lower total ATP production rate than undSOD1, being statistically significant in C2 (Fig. 2K), which matches their quiescent metabolic phenotype (Fig. 1J, K). C2 mutSOD1 lymphoblasts presented higher lactate levels in the absence (Fig. 2I) or presence of oligomycin (Fig. 2L), showing a higher glycolytic capacity in this group of patients. Overall, the short-term incubation with oligomycin had no significant effects on ATP levels in any of the cohorts studied (Fig. 2H), suggesting that glycolysis is still capable of Fig. 1 Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) parameters in lymphoblasts from ALS patients with mutant SOD1 (mutSOD1) and unknown mutation (undSOD1). Cells were plated at a density of 160,000 cells/well, as described in "Materials and Methods." A Mitochondrial Stress Test providing OCR-and ECAR-associated parameters was performed using the Seahorse XFe96 Extracellular Flux Analyzer. OCR and ECAR parameters, as well as energy maps showing the metabolic potential of cells. Data were plotted for all cohorts (pool), C1-female 46 years old, C2-male 46 years old, and C3-male 26/27 years old. A pooled analysis considered the average value obtained for each donor of a given experimental group. OCR parameters and ECAR parameters evaluated: (A) basal respiration, (B) proton leakassociated OCR, (C) spare respiratory capacity, (D) baseline ECAR, (E) stressed ECAR, (F) ATP production-linked OCR, (G) maximal respiration and (H) non-mitochondrial respiration, (I) bioenergetic health index (BHI) assesses the cellular bioenergetic health in lymphoblast from ALS patients with mutSOD1 and undSOD1 comparatively to controls, based on the OCR, according to the following equation: BHI = (spare respiratory capacity × ATP production-linked OCR)/(proton leak OCR × non-mitochondrial Respiration). (J-M) Energy maps evidence the metabolic potential of cells under basal or stress conditions. Data are the mean ± SEM from the number of independent experiments represented as individual data points in each bar. *p < 0.05, **p < 0.01, ***p < 0.01 compared to respective control. + p < 0.05, + + p < 0.01 compared to respective undSOD1 lymphoblasts. Data are available at: https:// doi. org/ 10. 6084/ m9. figsh are. 19229 190 sustaining ATP production during the time frame of the experiment in this type of cells, again supported by the corresponding increase in extracellular lactate. No significant alterations were observed in transcripts for glucose-6-phosphate dehydrogenase (G6PD, Fig. 2M), pyruvate dehydrogenase E1-alpha subunit (PDHA1, Fig. 2N), and pyruvate dehydrogenase kinase 1 (PDK1, Fig. 2O) in lymphoblasts from ALS patients of all cohorts when compared to lymphoblasts from controls.

Mitochondrial Transcripts, Biogenesis, and Membrane Potential
To further characterize how mutSOD1 and undSOD1 lymphoblasts differ in terms of mitochondrial biogenesis, we measured mitochondria-relevant transcripts and protein amount by qPCR and Western blotting, respectively, and mitochondrial membrane potential by flow cytometry. We demonstrated that transcripts of mitochondrial oxidative phosphorylation subunits in lymphoblasts from ALS patients of all cohorts were not significantly altered when compared to lymphoblasts from controls ( Fig. 3A-F). However, transcripts for mitochondrial complex I-related subunit NDUFA9 (Fig. 3A) were increased in lymphoblasts from C1 and C3 mutSOD1 patients relative to the matched-control counterparts. There were also increases in transcripts for cytochrome c (CYCS, Fig. 3B) in lymphoblasts from C3 mutSOD1 and cytochrome c oxidase subunit IV isoform 1 (COX4i1, Fig. 3C) in lymphoblasts from C1 mutSOD1 patient and C3 undSOD1 patient, compared to the respective controls. ATP5G1 transcripts were also increased in lymphoblasts from the C3 undSOD1 patient (Fig. 3D) relatively to control. No significant changes were observed between lymphoblasts from ALS patients and age and sex-matched controls regarding transcripts for SDHA (Fig. 3E) or UQCRC2 (Fig. 3F) or mitochondrial potential (Fig. 3G). Concerning mitochondrial biogenesis markers, our results showed that mitochondrial DNA copy number (Fig. 3H) and Tfam transcripts (Fig. 3I) did not change in lymphoblasts of ALS patients from all cohorts or for each cohort C1, C2, and C3, apart from a significant increase in mtDNA copy number in mutSOD1 lymphoblasts from C2, when compared to undSOD1 lymphoblasts (Fig. 3H). However, the protein levels of Tfam (Fig. 3J) were significantly decreased in lymphoblasts from undSOD1 patients when pooling samples or from individual C1 and C3 when compared to the respective controls, while the PGC1α and TOM20 protein levels did not significantly change between lymphoblasts from ALS patients and controls (Fig. 3K, L).

Mitochondrial Oxidation of TCA Cycle Substrates
To analyze the disruption of mitochondrial metabolism in lymphoblasts from ALS patients with mutSOD1 and und-SOD1, we performed a phenotypic metabolic analysis of TCA cycle substrates' oxidation using the Biolog Mitoplate S1 system (Fig. S1 and Fig. 4). Considering the pooled analysis of all cohorts, we observed that the rate of oxidation of TCA cycle substrates between lymphoblasts from ALS patients and controls was not significantly altered (Fig. 4). Interestingly, by analyzing cohorts individually, our results showed that undSOD1 lymphoblasts from C1 presented lower oxidation rates of cis-aconitic acid (Fig. 4A), isocitric acid (Fig. 4B), α-ketoglutarate (Fig. 4C), l-malic acid (Fig. 4D), l-glutamic acid (Fig. 4E), and l-glutamine (Fig. 4F) compared to control, while mutSOD1 lymphoblasts from C1 presented lower oxidation rates of pyruvic acid in the presence of l-malic acid (Fig. 4G), succinic acid (Fig. 4H), and fumaric acid (Fig. 4I) when compared with the respective matched-controls. undSOD1 lymphoblasts in C2 had lower oxidation rates of pyruvic acid (Fig. 4J), l-glutamic acid (Fig. 4E), l-glutamine (Fig. 4F), and alanine-glutamine (Fig. 4K) with a higher rate of utilization of l-malic acid (Fig. 4D) compared to control. Curiously, undSOD1 lymphoblasts in C3 presented higher oxidation rates of isocitric acid (Fig. 4B) and l-glutamine (Fig. 4F), while mutSOD1 lymphoblasts showed a higher rate of oxidation of pyruvic acid (Fig. 4J), citric acid (Fig. 4L), and l-glutamic acid (Fig. 4E), relative to control. The results show that, among other relevant alterations, the undSOD1 C1 sample presented an absent or very low capacity to oxidize glutamine and related substrates.

Mitochondrial Oxidation of Fatty Acids, Amino Acids, and Cytosolic Substrates
Next, we analyzed the alterations of mitochondrial metabolism of fatty acids, amino acids and cytosolic substrates in lymphoblasts from ALS patients with mutSOD1 and undSOD1, using the same phenotypic metabolic analysis (Fig. S1 and Fig. 5).
Concerning the oxidation rates of fatty acids, amino acids, and cytosolic substrates, among other compounds, in all cohorts, no significant variations were observed between lymphoblasts from ALS patients and controls (Fig. 5). Analyzing cohorts individually, we observed that undSOD1 lymphoblasts from C1 presented a lower rate of oxidation of octanoyl l-carnitine (+ l-malic acid) (Fig. 5A), while undSOD1 lymphoblasts from C2 Fig. 2 Pathways associated with ATP production in lymphoblasts from ALS patients with mutant SOD1 (mutSOD1) or unknown mutation (undSOD1) and respective controls. Cells were plated at a density of 160,000 cells/well, as described in "Materials and Methods." Oxygen consumption rate (OCR) and Extracellular Acidification Rate were monitored over time using the Seahorse XFe96 Extracellular  Fig. 3 Mitochondrial respiratory chain subunits, biogenesis markers, and mitochondrial membrane potential in lymphoblasts from ALS patients with mutant SOD1 (mutSOD1) or unknown mutation (undSOD1) and respective controls. Transcripts for (A-F) mitochondrial oxidative phosphorylation subunits and (I) mitochondrial biogenesis-related protein TFAM. Data were plotted for all cohorts (pool), C1-female 46 years old, C2-male 46 years old, and C3male 26/27 years old. Pooled analysis considered the average value obtained for each donor of a given experimental group. Gene expression was normalized to the geometric mean of TBP, YWHAZ, PUM1, and B2M. The mitochondrial potential (G) was measured by flow cytometry by subtracting the mean values of TMRM fluorescence in FCCP-treated lymphoblast from the mean values of TMRM fluorescence. Mitochondrial DNA copy number was normalized to ND5 and B2m levels (H). Western blot was used to semi-quantify protein expression levels of TFAM, PGC1α, and TOM20 and protein normalization was performed by Ponceau staining assay. Data are the mean ± SEM from the number of independent experiments represented as individual data points in each bar. *p < 0.05 and **p < 0.01 compared to respective control. + p < 0.05 compared to respective undSOD1 lymphoblasts. Data are available at: https:// doi. org/ 10. 6084/ m9. figsh are. 19229 214 presented a higher rate of oxidation of octanoyl l-carnitine (+ l-malic acid) (Fig. 5A) and palmitoyl l-carnitine (+ l-malic acid) (Fig. 5B), compared to control. Regarding the individual analysis of cytosolic substrates by cohorts, our results showed that undSOD1 lymphoblasts from C1 presented a higher rate of oxidation of d-glucose-6-PO 4 (Fig. 5C), while undSOD2 lymphoblasts from C2 presented higher rate oxidation of d-glucose-1-PO 4 (Fig. 5D) and a trend towards an increase in the rate of d-glucose-6-PO 4 oxidation (Fig. 5C), compared to respective controls. undSOD1 lymphoblasts also presented increased oxidation rates for palmitoyl l-carnitine (Fig. 5B). mut-SOD1 lymphoblasts from C2 presented a lower utilization capacity of d-gluconate-6-PO 4 (Fig. 5E), while mutSOD1 lymphoblasts from C3 presented a higher oxidation rate of d-glucose-1-PO 4 (Fig. 5D) d-glucose-6-PO 4 (Fig. 5C),   Fig. 4 Mitochondrial oxidation of TCA cycle substrates in lymphoblasts from ALS patients with mutant SOD1 (mutSOD1) or unknown mutation (undSOD1) and respective controls. Cells were plated at a density of 160,000 cells/well, and the rate of oxidation of a panel of TCA cycle substrates was assessed using the Biolog Mitoplate S1 assay. Data were obtained for all cohorts (pool), C1-female 46 years old, C2-male 46 years old, and C3-male 26/27 years old. Pooled analysis considered the average value obtained for each donor of a given experimental group. Rates of oxidation of (A) cis-aconitic acid, (B) isocitric acid, (C) α-ketoglutaric acid, (D) l-malic acid, (E) l-glutamic acid, (F) l-glutamine, (G) pyruvic acid (+ l-malic acid), (H) succinic acid, (I) fumaric acid, (J) pyruvic acid, (K) alanine-glutamine (Ala-Gln), and (L) citric acid. Data are the mean ± SEM from the number of independent experiments represented as individual data points in each bar. *p < 0.05 and **p < 0.01 compared to respective control. Data are available at: https:// doi. org/ 10. 6084/ m9. figsh are. 19229 229 d-gluconate-6-PO 4 (Fig. 5E), and l-lactic acid (Fig. 5F), relative to control. Analyzing the rate of oxidation of amino acids and other substrates specifically by cohorts, our results showed a lower rate of oxidation of d-l-βhydroxybutyric acid (Fig. 5G) and l-leucine (+ l-malic acid) (Fig. 5H) in undSOD1lymphoblasts from C1, as well as a higher oxidative capacity of tryptamine (Fig. 5I) and serine (Fig. 5J) in mutSOD1 lymphoblasts from C1 compared to undSOD1 and control, respectively. Interestingly, we observed that undSOD1 lymphoblasts in C3 presented a higher rate of oxidation of serine (Fig. 5J) and l-ornithine (Fig. 5K), and mutSOD1 lymphoblasts presented a higher rate of oxidation of tryptamine (Fig. 5I) relatively to control. No differences between cohorts were found when measuring oxidation rates of glucose (Fig. 5L), ɑ-glycerol-PO4 (Fig. 5M), glycogen (Fig. 5N, despite a tendency for a decrease in the oxidation rate in the C1 undSOD1 individual), or α-ketobutyric acid (Fig. 5O). When considering the oxidation rate of ɣ-aminobutyric acid (plus l-malic acid), there was a large difference in the C1 cohort, in which both ALS individuals showed a large decrease, a significant difference in the mutSOD1 patient (Fig. 5P).

Integrative Analysis
To integrate all the results and understand which alterations are more characteristic of ALS lymphoblasts metabolism and mitochondrial function fingerprinting, we performed multidimensional data analysis. Five factors (targets) were considered to be relevant for the analysis: group (control, undSOD1, mutSOD1), disease (control, ALS [undSOD1,  [control,undSOD1], mutSOD1), sex (male, female), and age (26-27 years old, 46 years old). The measured features were ranked regarding the information gain between each feature and the targets. The top 3 features, and any additional attributes with identical information gain values (Fig. 6A), were considered for their potential to distinguish samples according to the targets. Principal component analysis was applied for further dimensionality reduction, generating linear projections (Fig. 6B). The performance of a Naïve Bayes classifier model for target classification was also evaluated.
Considering group as a target, the features that contributed to a better segregation of control, undSOD1 and mut-SOD1 were found to be the protein levels of Tfam (Tfam_p) and glycolytic ATP production rate, followed by maximal and non-mitochondrial respirations, transcripts coding for cytochrome c (CYCs_mRNA), glucose-6-phosphate dehydrogenase (G6PD_mRNA) or pyruvate dehydrogenase A1 (PDHA1_mRNA), mitochondrial alanine-glutamine (Ala-Gln) oxidation rates, glycolysis (percent of glycolytic ATP production), and intracellular ATP levels after treatment with 2-DG. The Naïve Bayes classifier was able to predict the samples belonging to each group perfectly.
Considering disease as a target, Tfam_p had the highest information gain, followed by maximal respiration, mitochondrial oxidation of serine, extracellular lactate levels before and after treatment with oligomycin, LDH activity, glycolytic ATP production rate, and glycolysis (percent of glycolytic ATP production). Although, visually, the separation between ALS and no ALS samples was apparently very good, the Naïve Bayes classifier failed to classify mutSOD1 from C1 and C2 as ALS, correctly predicting all the control and undSOD1 samples.
When considering mutation as a target, there were 9 features with the highest information gain, namely basal, proton leak-associated and non-mitochondrial respirations, baseline and stressed ECAR, glycolytic ATP production rate, protein levels of Tfam, transcripts coding for PDHA1, and intracellular ATP levels after treatment with 2-DG. The Naïve Bayes classifier was able to predict the samples belonging to mutation or non-mutation samples perfectly.
Regarding sex, 4 features had a high information gain, perfectly segregating female and male samples. These features included ATP production-associated respiration, mtDNA copy number, protein levels of PGC1α, and transcripts encoding G6PD. The Naïve Bayes classifier was able to predict the samples belonging to male or female samples perfectly.
As for age, 10 features enabled a good separation between 26/27 years old and 46 years old samples, including spare respiratory capacity, bioenergetic health index (BHI), transcripts coding for UQCRC2, ATP5G1, and Tfam, mtDNA copy number, mitochondrial oxidation of α-ketobutyric acid, total and mitochondrial ATP production rates and XF ATP Rate index. The Naïve Bayes classifier failed to classify the control and mutSOD1 from C2 as 46 years old, having correctly predicted all the other samples.

Discussion
The present work identified unique mitochondrial and metabolic fingerprints in lymphoblasts from ALS patients with (mutSOD1) or without (undSOD1) known SOD1 mutations, compared with age-and sex-matched controls performing both pooled and paired comparisons.
Alterations in mitochondrial metabolism have been described in different ALS models. Mitochondrial dysfunction in ALS does not seem to be restricted to motor neurons, presenting also a systemic nature. In fact, metabolic changes associated with ALS pathology have been studied using other cells or tissue types obtained from patients, including fibroblasts [48,[60][61][62][63][64], skeletal muscle biopsies [65], and peripheral blood mononuclear cells (PBMCs) [66]. The latter are interesting models since they are easy to collect in a mildly invasive manner. Abnormalities of circulating blood cells may be a helpful prognostic or therapeutic biomarker in ALS.
Our results were highly heterogeneous between cohorts of different age and sex, which is not surprising due to the intrinsic biological heterogeneity of ALS, which presents different rates of progression, patterns of spread, and various sites of onset between patients [67], In addition, some heterogeneity introduced by the lymphoblast transformation process cannot be excluded. There are also various genetic mutations associated with ALS [68]. Thus, heterogeneity among the metabolism of circulating blood cells is also expected between patients, and it is reasonable to postulate that a single treatment strategy will not fit all patients, being imperative to identify small responder groups within larger patient populations [69,70].
Similarly to fibroblasts, PBMCs can be used for cell reprogramming into cerebral organoids to study mitochondrial health in a more complex neuronal-like environment [71]. Preserved mitochondrial genetics, function, and treatment responses were found across PBMCs to iPSCs to cerebral organoids [71], suggesting that the mitochondrial physiology of PMBCs can recapitulate neuronal mitochondrial properties from the donor. Although it is possible to measure multiple mitochondrial parameters even from cryopreserved PBMCs [72], the amount of samples obtained from a given patient is very limited. This is critical because ALS is a relatively rare and rapidly progressing fatal disease, qualifying as an orphan disease (ORPHA:803), limiting the number of patient samples obtained for analysis. An alternative is the use of lymphoblasts, immortalized PBMCs that can be obtained from biobanks such as the Coriell Institute. These cells proliferate in culture, allowing the measurement of multiple functional cellular parameters at different time points, and have also been previously used for studying mitochondrial function in other neurodegenerative diseases [73][74][75]. These cells can also be manipulated and challenged in vitro to highlight subclinical diseaseretarded metabolic alterations. In ALS, mtSOD1 lymphoblasts were previously used to demonstrate the existence of hyper oxidized SOD1 with toxic properties in patientderived cells and allowed the identification of common SOD1-dependent toxicity between mutant SOD1-linked familial ALS and a subset of sALS [76]. In addition, our group has recently shown that the same ALS lymphoblasts used in the present work showed distinct redox profiles, when compared to control cells [50]. These results highlighted how lymphoblasts can provide opportunities for biomarker development, sub-classifying ALS, and designing more disease-modifying therapies [76]. Moreover, increased ROS levels were reported in lymphoblasts from familial ALS patients with SOD1 mutations compared with sporadic ALS and normal controls (spouses of ALS patients), although the ROS generation was not directly correlated with SOD1 activity [76]. In the present work, we performed a multi-dimensional analysis on several metabolic and mitochondrial hallmarks obtained from lymphoblasts from each ALS patient and, although a high heterogeneity of metabolic data was found, we identified specific metabolic and mitochondrial alterations that may have a diagnostic and therapeutic interest.

Metabolic Hallmarks of ALS
To find metabolic hallmarks of ALS, we compared lymphoblasts from ALS patients with matched controls. No common changes between mutSOD1 vs control and undSOD1 vs control were evident from our data, following conventional data analysis. However, multidimensional analysis considering "Disease" (ALS versus no ALS) as a target factor identified a possible pattern, mainly involving decreased Tfam protein levels (the feature with highest information gain) in lymphoblasts from individuals with ALS, regardless of SOD1 mutation, compared to controls. Tfam is a nuclear-encoded mitochondrial transcription factor involved in mtDNA genome maintenance, transcription, and replication [77]. Changes in Tfam expression have been found in several neurodegenerative diseases [78]. In sALS, a decrease in Tfam expression was also observed in PBMC obtained from patients [66]. ALS spinal neurons were found to present varied and reduced mtDNA gene copy numbers and increased mtDNA gene deletions [79], suggesting that the therapeutic use of human recombinant Tfam could be beneficial in ALS. In agreement, Tfam overexpression was shown to protect motor neurons in mutSOD1 ALS mice [80].
In our work, both groups of ALS lymphoblasts showed markers of hypermetabolism. In the pooled analysis, und-SOD1 lymphoblasts were characterized by increased glycolytic ATP rate and decreased Tfam protein levels compared to matched controls. Moreover, although not statistically significant in the pooled analysis, all mutSOD1 cells showed increased extracellular lactate levels after challenge with oligomycin, compared to the respective controls. The data is consistent with the higher metabolic plasticity of mut-SOD1 under mitochondrial stress (except in the youngest cohort) visible in the Seahorse Extracellular Flux Analyzer energy maps. It was previously shown that muscle atrophy and reduced physical activity in ALS coexist with a paradoxical increase in energy expenditure (hypermetabolism) [81]. Hypermetabolism in ALS patients is described as an early and persistent phenomenon [82], associated with more significant functional decline and shorter survival [83].

Specific Hallmarks of ALS Associated with the SOD1 Mutation (mutSOD1 vs undSOD1)
To analyze the impact of mutSOD1 in ALS metabolic phenotype, we focused on the differences between mutSOD1 and undSOD1 lymphoblasts.
It is important to note that, in ALS, more than 220 mutations have been identified so far across the coding and noncoding regions of the SOD1 gene (https:// alsod. ac. uk/ output/ gene. php/ SOD1, accessed 29 June 2022) [25,84,85]. The influence of these mutations on dismutase activity is considerably variable, having been associated with decreased [86], maintained [87,88], or increased [86,89] enzymatic activity, compared to wild-type SOD1. In fact, we have previously observed that mutSOD1 lymphoblasts from C1 and C3 (bearing the Ala4Val mutation) showed a decrease in SOD1 activity (significant) and protein levels (non-significant), which were not observed in C2 (bearing the Ile113Thr mutation) [50].
Although no evident changes were found in the pooled analysis, paired analysis evidenced lower maximal respiration in mutSOD1 samples from all cohorts compared to the undSOD1 samples. Similar alterations were previously found in mutSOD1 models, including lower maximal respiration and reserve capacity in SOD1G93A mouse fibroblasts, compared to wild-type controls, with no variations in basal or ATP-linked respiration and no change in proton leak respiration [90]. A decrease in spare respiratory capacity was also found in fibroblasts from mutSOD1 patients compared to controls [62]. Moreover, a shift of energy generation from oxidative phosphorylation to glycolysis was observed in fibroblasts from mutSOD1 (I113T) patients [62]. Although glycolytic flux was upregulated in mutSOD1 fibroblasts, these cells did not increase their glycolytic capacity upon challenge with oligomycin, suggesting that they were already using their maximal glycolytic capacity [62]. Human mutSOD1 was previously shown to cause oxidative phosphorylation dysfunction in mitochondria of transgenic mice [26]. However, the decrease in maximal respiration observed in mutSOD1 compared to undSOD1 lymphoblasts seems to be due to a compensatory (although non-statistically significant) increase in maximal respiration in undSOD1 (but not in mutSOD1) compared to controls. This difference between undSOD1 and mutSOD1 lymphoblasts is likely related to the mitochondrial localization of mutSOD1 [33,34]. Metabolic reprogramming was also observed in skin fibroblasts obtained from ALS patients and motor neurons from presymptomatic SOD1G93A mice, and conserved changes were found in protein translation, folding and assembly, tRNA aminoacylation, and cell adhesion processes [60].
A previous study aimed at identifying and distinguishing two subgroups of ALS patients from control subjects, using mitochondria from skin fibroblasts obtained from sALS, fALS, and control donors, found different mitochondrial bioenergetic profiles, despite the small number of samples used and the high variability in the measured parameters [61]. Among the variables analyzed in this study, fALS samples (3 with SOD1 and 1 with C9Orf72 mutations) were characterized by a high maximal respiration rate in the presence of substrates for complexes I and II under permeabilized conditions. Under these artificial conditions, potential metabolic differences found in intact cells, such as those we identified in the present paper, may not be identified, instead translating into an apparent increased mitochondrial respiratory capacity. High levels of the complex I NDUFB8 subunit were also found in mitochondria from fALS fibroblasts. In turn, mitochondria from sALS fibroblasts (with no mutations identified) were characterized by a high flux control coefficient (C i ) for complexes I and IV. Mitochondria from control fibroblasts were mainly characterized by higher levels of SOD1 protein compared to their ALS counterparts [61].

Impact of Age and Sex on Lymphoblast Metabolism
Since ALS is a rare disease, our experimental strategy overcomes the limitation of a starting low number of lymphoblasts samples by using an integrated multi-parametric analysis. Despite the complexity of simultaneous assays and the low statistical power to distinguish age and sex effects, our study showed clear disease-related differences. For example, samples from the youngest cohort (C3) presented different metabolic behaviors than the other two cohorts. Interestingly, in skin fibroblasts from donors of different ages, aging has been associated with respiratory function decline, mtDNA mutations, oxidative stress, and altered gene expression [91]. However, it is important to note that changes observed in the younger ALS patients can either be due to their younger age or to the earlier onset of the disease. In agreement with our above-described findings, age-related changes in bioenergetic properties and mitochondrial morphology were also found in fibroblasts from sALS patients, compared to fibroblasts from control donors [92]. A metabolic profiling study using Biolog plates in fibroblasts and induced neuronal progenitor cell-derived iAstrocytes from C9orf72, TDP-43, SOD1, and sporadic ALS patients showed several metabolic alterations in patients, namely an age-dependent decrease in glycogen metabolism and in NADH production from different substrates [64]. NADH production resulting from glycogen metabolism in fibroblasts from familial cases was negatively correlated with end-points of disease severity including the age of onset and the age of death, while glucose-1-phosphate and D-fructose-6-phosphate negatively correlated with disease duration in the familial cohort [64]. This suggests that disrupted cellular metabolism is relevant and can be used to predict ALS progression.
Multidimensional analysis performed in this work using sex as the target variable identified 6 variables that allowed the distinction of male and female samples, including ATP production, mtDNA copy number, the levels of PGC1α, and transcripts for the metabolic enzyme glucose-6-phosphate dehydrogenase. Sex-associated differences in mitochondrial function were also previously found in human PBMCs [93], highlighting the need for proper sex and age-matched controls for this type of study. Interestingly, glucose 6-phosphate dehydrogenase levels and activity were previously found to partially explain the impact of donor sex, age, and ethnicity on red blood cell antioxidant metabolism [94].

Conclusions
The results from this study revealed metabolic and mitochondrial parameters in lymphoblasts from ALS patients that can be promising biomarkers for disease stratification. This work goes well beyond what has been already described in the literature, mostly in fibroblasts, identifying age-and sex-related markers of metabolic, including mitochondrial, dysfunction in ALS patients obtained from a minimally invasive source of cells. The features found in our multidimensional analyses can also be used to help develop new effective metabolic-based therapies to delay the progression of ALS.