Cerebrospinal fluid samples of SMA patients

CSF samples of 35 adult patients (age 16–65 years; demographic data of patients are provided in Supplementary Tab. 1 online resource) and of 14 pediatric cases (age 8–12 years; demographic data of patients are provided in Supplementary Tab. 2 online resource) with genetically proven 5q- SMA (type 1, 2, or 3, respectively), as well as controls, were analyzed. Moreover, serum samples derived from 8 adult patients (age 16–65 years) 3 and as well as 8 adult controls were analyzed in addition to 14 pediatric cases (age 1–6 years) with 5q-SMA (SMA type 1 to 3). All patients or their caregivers gave written informed consent. Control CSF was obtained from non-SMA patients (diagnostic procedure to exclude CNS disease; non-disease controls) and from non-SMA patients with inflammatory (n = 11 pediatric and n = 3 adult) and non-inflammatory CNS diseases (n = 6 pediatric and n = 2 adult) serving as disease controls. Serum control samples were obtained from healthy donors (non-disease controls) or patients suffering from other neuromuscular conditions (disease controls with genetically confirmed VWA1-related neuromyopathy (adult; n = 6), BICD2-related spinal muscular atrophy with lower extremity predominance (SMALED with adult onset; n = 6), and CHRNE-related congenital myasthenic syndrome (pediatric; n = 7)). SMA-patients responding to therapeutic intervention with nusinersen (were distinguished from such not responding according to the Scores on the Children’s Hospital of Philadelphia Infant Test of Neuromuscular Disorders (CHOP INTEND). Ranging from 0 to 64, with higher scores indicating better motor function; a CHOP INTEND non-response was defined as a failure of an increase of at least 4 points from baseline. The Hammersmith Infant Neurological Examination (HINE) is a neurological examination consisting of 26 items, each scored on a scale of 0 to 3, designed for evaluating infants between 2 and 24 months of age. Motor scores of adult patients were measured using the Hammersmith Functional Motor Scale-Expanded (HFMSE) score or the Revised Upper Limp Module (RULM). Treatment response was classified with an increase of 3 or more motor points. The time point of 180 days post-treatment was chosen as it reflects the ending of the nusinersen “loading phase” and the beginning of the “maintenance phase”, thus providing the full nusinersen effect for each patient and providing robust information regarding the value of therapy markers. Study approval was obtained from the University Duisburg-Essen ethics committee (approval number 18-8285-BO).

Proteomic profiling on human CSFSample preparation

Human CSF samples were stored on -80 °C until further processing. A total of 50 μl of each sample (n = 16; 2 samples from non-SMA affected individuals, 4 samples from SMA patients not responding to treatment with nusinersen and 10 samples from SMA patients responding to treatment with nusinersen) was denatured with Biognosys’ Denature Buffer. Reduction and alkylation were carried out by adding Biognosys’ Reduction/Alkylation Solution for 1 h at 37 °C. Digestion was then performed with 0.5 μg trypsin (Promega) per sample overnight at 37 °C.

Clean-up for mass spectrometry

After digestion, samples were desalted using a BioPureSPEMIDI C18 spin plate (The Nest Group) following the manufacturer’s instructions and dried using a SpeedVac system. Peptides were then dissolved in 20 μl of 1% acetonitrile containing 0.1% formic acid (FA). Prior to mass spectrometric measurements, iRT calibration peptides from Biognosys were added to all samples prepared in this manner. Peptide concentrations were determined using a BCA assay (Thermo Fisher).

HPRP fractionation

To perform high-pH fractionation of peptides, equal volumes of digested samples were pooled according to time point (pool 1: baseline, pool 2: 6 months). Ammonium hydroxide was added to achieve a pH > 10. Fractionation was performed using a DionexUltiMate3000 RS pump (ThermoScientific) on an Acquity UPLC CSH C18 1.7 μm, 2.1 × 150 mm column (Waters). The gradient was 1% to 40% solvent B in 20 min, the solvents were A: 20 mM ammonium format in water, B: acetonitrile. Fractions were collected every 30 s and sequentially combined into 6 fraction pools. These were allowed to dry and dissolved in 17 μl solvent A. Before mass spectrometric analyses, they were spiked with iRT calibration peptides from Biognosys. Peptide concentrations were determined using a UV/Vis spectrometer (SPECTROstarNano, BMG Labtech).

Shotgun LC–MS/MS for spectral library generation

For DDA LC–MS/MS analyses, 2 µg of peptides per sample or fraction were loaded onto a self-packed C18 column (Dr. Maisch ReproSilPur, 1.9 μm particle size, 120 Å pore size; 75 μm inner diameter, 50 cm length, New Objective) in a Thermo Scientific Easy nLC1200 nano-liquid chromatography system. This system was coupled to a Thermo Scientific Q Exactive mass spectrometer with a standard nano-electrospray source. The following were used as solvents: A: 1% acetonitrile in water containing 0.1% FA; B: 15% water in acetonitrile containing 0.1% FA. The following method was used as the nonlinear gradient: 1–52% solvent B in 120 min, followed by 52–90% B in 10 s, 90% B for 10 min, 90–1% B in 10 s, and 1% B for 5 min. The MS method used was a modified TOP12 method by [15]. Full MS covered the m/z range of 350–1650 with a resolution of 70 000 (AGC target value was set to 3e6) and was followed by MS/MS scans with a resolution of 17 500 (AGC target value was 5e5). The isolation width of the MS/MS acquisition precursor was 2 m/z, while the normalized collision energy was set at 25 (10% staged collision energy) and the default charge state was 2 + .

HRM mass spectrometry acquisition

Exactly as for the DDA LC–MS/MS measurements, 2 μg of peptides per sample were loaded onto the in-house packed C18 column in a Thermo Scientific Easy nLC1200 nano-liquid chromatography system for the DIA measurements. The system was coupled to a Thermo Scientific Q Exactive HF mass spectrometer equipped with a standard nano-electrospray source. The LC solvents were A: 1% acetonitrile in water containing 0.1% FA; B: 15% water in acetonitrile containing 0.1% FA. The nonlinear LC gradient was 1–55% solvent B in 120 min, followed by 55–90% B in 10 s, 90% B for 10 min, 90–1% B in 10 s, and 1% B for 5 min. The DIA method included a one full range scan followed by 22 DIA windows.

Database search of LC–MS/MS data and spectral library generation

Shotgun and HRM mass spectrometry data were analysed using SpectroMine software (Biognosys). The false discovery rate was set to 1% at the peptide and protein levels. Data were validated against a human UniProt protein database (Homo sapiens, 2019-–07-01). The search preferences were set as follows: 2 missed cleavages were allowed. Modifications were set as variable for N-term acetylation, methionine oxidation, deamidation (NQ), carbamylation (KR)). To create a CSF resource spectral library, shotgun data were obtained from commercially available healthy individual CSF and searched using the above settings. Shotgun, HRM, and resource search data were used to create a hybrid spectral library (library ID: 2452_R03) in SpectroMine software.

HRM data analysis

HRM mass spectrometric data were analysed using Spectronaut Pulsar software (Biognosys) with the hybrid spectral library generated in this project. The false discovery rate on protein and peptide level was set to 1%, data was filtered using row-based extraction. The HRM measurements analyzed with Spectronaut were normalized using local regression normalization [2].

Data analysis

Data analysis and plotting was performed in R. Distance in heat maps was calculated using the “manhattan” method, the clustering using “ward.D” for both axis. Principal component analysis was conducted in R using prcomp and a modified ggbiplot function for plotting, and partial least squares discriminant analysis was performed using mixOMICS package.

Enzyme-linked immunosorbent assay (ELISA)

The level of LARGE1 were measured in CSF and serum samples derived from adult and pediatric SMA patients and matching controls by making use of the ELISA technique according to the manufacturer’s protocol (#MBS281249, MyBioScource, San Diego, CA, USA).

Animals

SMN-deficient mouse model FVB.Cg-Smn1tm1Hung Tg(SMN2)2Hung/J (Jackson #005058), reflecting later-onset SMA, homozygote for the murine SMN1 knockout, and the insert of human SMN2 (4 copies) were purchased from Jackson Laboratory (Bar Habor, ME, USA) and bred in the Animal Research Lab of the University Hospital Essen. Male and female mice were used for spinal cord tissue (P10, P20, P28, P42, and P52) harvesting and muscle tissue (Musculus tibialis anterior (TA)) at postnatal days P10, P28, and P52. Age-matched male FVB/N mice (wild-type, wt) served as control. At P42, SMA mice already show loss of spinal motor neurons [18].

All animals were kept on a 12/12 h light/dark cycle with water and standard food pellets available ad libitum.

All experiments were conducted under the animal welfare guidelines of the University Duisburg Essen. Furthermore, the use of the SMA mouse model was approved by the State Agency for Nature, Environment and Consumer Protection (LANUV) in North Rhine-Westphalia (reference number 81–02.04.2020.A335).

Preparation of murine spinal cord tissue and tibialis anterior as well as human quadriceps muscle slices

Lumbar spinal cord tissue of late-onset SMA and wild-type mice were snap-frozen in liquid nitrogen and stored at − 80 °C until usage. Spinal cord cryosections of 20 µm were prepared. Every fifth section of each spinal cord is placed on one independent microscopy slide.

The murine TA was removed and immediately snap-frozen in a mold of Tissue-Tek® O.C.T.™ Compound on dry ice in isopropanol. Afterwards, muscles were stored at − 80 °C until usage. Muscle cryosections of 12 µM were prepared for immunostaining studies. For immunostaining studies on quadriceps muscle derived from pediatric SMA type 3 patients (muscle biopsies were collected for diagnostic purposes), 10 µM cryosections were used.

Immunostaining of spinal cord tissue slices and muscle biopsy specimen

Murine lumbar spinal cord or TA sections were fixed in 4% Paraformaldehyde, washed, permeabilized (PBS, 0.1 v/w Triton X-100), and blocked (PBS, 5% bovine albumin serum). Primary antibodies for motor neurons (anti-SMI-32, mouse, 1:400, #801,701, BioLegends, San Diego, CA, USA), LARGE1 (anti-LARGE1, rabbit, 1:500, #PA5-78,393, Thermo Fisher Scientific, Taufkirchen, Germany), Golgi marker GM130 (anti-GM130, mouse, 1:500, #AB169276, Abcam, Cambridge, United Kingdom) and BiP (GRP78) (anti-BiP BD, mouse, 1:500, Transduction Laboratories #610,978) were diluted in blocking solution and incubated at 4 °C overnight. Sections were washed, and secondary antibodies (goat anti-rabbit, goat anti-mouse, 1:300, Dianova, Hamburg, Germany) and DAPI (1:1000, Sigma-Aldrich, Taufkirchen, Germany) were diluted in blocking solution. Sections were incubated for 1.5 h at room temperature.

Images were obtained using a Zeiss Axio Observer.Z1 Apotome (Zeiss, Jena, Germany) fluorescence microscope and Zeiss Zen software to determine the relative protein levels. All microscope settings such as laser intensity, exposure time, or contrast were kept the same to analyze LARGE1, GM130 and BiP immunoreactivities.

Immunoreactivity of LARGE1 in the entire ventral horn or SMI-32 positive spinal motor neurons as well as in the TA sections and immunoreactivity of GM130 and BiP was measured using Image J software (NIH, Bethesda, MD, USA). LARGE1, GM130 or BiP immunoreactivity was measured and normalized against the background of each slice.

The relative protein level of LARGE1 or GM130 was calculated by normalizing the intensity of SMA tissue against wild-type tissue.

Colocalization of LARGE1 and GM130 as well as LARGE1 and BiP was determined using Image J software plug-in “Coloc 2” and calculation of Pearson coefficient.

Western blot analysis

We performed Western blot analysis to substantiate the evaluated protein level by immunostaining. Therefore, the spinal cord tissue of SMA or wt mice were homogenized in RIPA buffer containing a protease inhibitor cocktail (Roche, Germany). The amount of protein in those lysates was determined by a bicinchoninic acid (BCA) protein assay.

Ten micrograms of protein were applied to 4–15% TGX Stain-Free gels (Biorad, Germany), and proteins were transferred to 0.2 µm nitrocellulose membranes using a semi-dry blotting technique. Images of membranes were taken for total-protein evaluation. Afterward, the membranes were incubated in fast-blocking solution (Biorad, Germany) under gentle agitation for 10 min at RT. Then, the membranes were incubated with primary antibodies (in blocking solution) against LARGE1 (anti-rabbit, 1:10,000) at 4 °C overnight. Primary antibody against β-actin (hFAB,1:10,000 #12,004,164 BioRad, Germany; anti mouse 1:10,000 #ab8226 Abcam, United Kingdom) as well as GAPDH (anti-mouse, 1:10,000, # MA5-15,738, Thermo-Fisher) was used as an additional loading control.

After washing, the membranes were incubated with anti-rabbit horseradish peroxidase-coupled secondary antibody for 1.5 h at RT. Immunoreactivity was detected using an enhanced chemiluminescence substrate and a Western blot imaging system (Biorad, Germany).

Analysis of Western blot signals was performed using Biorad imaging software. First, the signal of each protein and actin lane was measured. Then, the protein signal of each lane was normalized to its total protein value (PonceauS). Finally, the calculated protein level of SMA mice was further normalized to the value detected in age-matched wt mice, and additionally the wt mice were normalized.

All Western blots and total protein staining used for mean value calculations are provided in the supplementary material (Supplementary Figs. 7–8 online resource).