1. Genetics and Genomics
  2. Medicine
Download icon

SMA-miRs (miR-181a-5p, -324-5p, and -451a) are overexpressed in spinal muscular atrophy skeletal muscle and serum samples

  1. Emanuela Abiusi
  2. Paola Infante
  3. Cinzia Cagnoli
  4. Ludovica Lospinoso Severini
  5. Marika Pane
  6. Giorgia Coratti
  7. Maria Carmela Pera
  8. Adele D'Amico
  9. Federica Diano
  10. Agnese Novelli
  11. Serena Spartano
  12. Stefania Fiori
  13. Giovanni Baranello
  14. Isabella Moroni
  15. Marina Mora
  16. Maria Barbara Pasanisi
  17. Krizia Pocino
  18. Loredana Le Pera
  19. Davide D'Amico
  20. Lorena Travaglini
  21. Francesco Ria
  22. Claudio Bruno
  23. Denise Locatelli
  24. Enrico Silvio Bertini
  25. Lucia Ovidia Morandi
  26. Eugenio Mercuri
  27. Lucia Di Marcotullio
  28. Francesco Danilo Tiziano  Is a corresponding author
  1. Department of Life Sciences and Public Health, Section of Genomic Medicine, Università cattolica del Sacro Cuore, Italy
  2. Center For Life Nano Science@Sapienza, Istituto Italiano di Tecnologia; Department of Molecular Medicine, Università degli Studi di Roma "La Sapienza", Roma, Italy, Italy
  3. Clinical and Experimental Epileptology Unit, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milano, Italy, Italy
  4. Department of Molecular Medicine, Università degli Studi di Roma "La Sapienza", Italy
  5. Pediatric Neurology, Department of Woman and Child Health and Public Health, Fondazione Policlinico Universitario A. Gemelli IRCCS-Università Cattolica del Sacro Cuore, Italy
  6. Centro Clinico Nemo, Fondazione Policlinico Universitario A. Gemelli IRCCS-Università Cattolica del Sacro Cuore, Italy
  7. Unit of Neuromuscular and Neurodegenerative Disorders, Dept. Neurosciences, Bambino Gesu' Children's Hospital IRCCS, Italy
  8. Department of Pediatric Neuroscience, Fondazione IRCCS Istituto Neurologico Carlo Besta, Italy
  9. Neuromuscular Diseases and Neuroimmunology Unit, Fondazione IRCCS Istituto Neurologico Carlo Besta, Italy
  10. Department of Medical and Surgical Sciences, Fondazione Policlinico Universitario Agostino Gemelli IRCCS, Italy
  11. Bioenergetics and Molecular Biotechnologies (IBIOM), CNR-Institute of Biomembranes, Italy
  12. CNR-Institute of Molecular Biology and Pathology (IBPM), Italy
  13. Amazentis SA, EPFL Innovation Park, Switzerland
  14. Department of Translational Medicine and Surgery, Section of General Pathology, Università Cattolica del Sacro Cuore, Italy
  15. Fondazione Policlinico Universitario A. Gemelli - IRCCS, Italy
  16. Center of Translational and Experimental Myology, IRCCS Istituto Giannina Gaslini, Italy
  17. Laboratory affiliated to Istituto Pasteur Italia-Fondazione Cenci Bolognetti, Department of Molecular Medicine, Sapienza University, Italy
  18. Unit of Medical Genetics, Department of Laboratory science and Infectious Diseases, Fondazione Policlinico Universitario IRCCS “A. Gemelli”, Italy
Research Article
  • Cited 0
  • Views 359
  • Annotations
Cite this article as: eLife 2021;10:e68054 doi: 10.7554/eLife.68054

Abstract

Background:

Spinal muscular atrophy (SMA) is a neuromuscular disorder characterized by the degeneration of the second motor neuron. The phenotype ranges from very severe to very mild forms. All patients have the homozygous loss of the SMN1 gene and a variable number of SMN2 (generally 2–4 copies), inversely related to the severity. The amazing results of the available treatments have made compelling the need of prognostic biomarkers to predict the progression trajectories of patients. Besides the SMN2 products, few other biomarkers have been evaluated so far, including some miRs.

Methods:

We performed whole miRNome analysis of muscle samples of patients and controls (14 biopsies and 9 cultures). The levels of muscle differentially expressed miRs were evaluated in serum samples (51 patients and 37 controls) and integrated with SMN2 copies, SMN2 full-length transcript levels in blood and age (SMA-score).

Results:

Over 100 miRs were differentially expressed in SMA muscle; 3 of them (hsa-miR-181a-5p, -324-5p, -451a; SMA-miRs) were significantly upregulated in the serum of patients. The severity predicted by the SMA-score was related to that of the clinical classification at a correlation coefficient of 0.87 (p<10-5).

Conclusions:

miRNome analyses suggest the primary involvement of skeletal muscle in SMA pathogenesis. The SMA-miRs are likely actively released in the blood flow; their function and target cells require to be elucidated. The accuracy of the SMA-score needs to be verified in replicative studies: if confirmed, its use could be crucial for the routine prognostic assessment, also in presymptomatic patients.

Funding:

Telethon Italia (grant #GGP12116).

Introduction

Spinal muscular atrophy (SMA) is an autosomal-recessive neuromuscular disorder, characterized by the degeneration of the α-motor neurons of the ventral horns of the spinal cord. The severity of the infantile forms is classically ranked into types I–III, according to the age of onset and the maximum motor milestone achieved (Mercuri et al., 2020). According to the conventional classification, the onset of the condition in type I is below 6 months of age and patients do not acquire the sitting position; in type II, symptoms occur within 18 months and children do not acquire autonomous ambulation. Type III is the most variable phenotype; developmental phases are comparable to that of the general population, the onset is over 18 months. Type III is subclassified into -a and -b, based on onset below or over 3 years of age (Zerres and Rudnik-Schöneborn, 1995). An additional form, with onset over 18 years of age, is usually reported as type IV: the outcome is variable and the degree of disability is usually mild. Indeed, SMA phenotype is better depicted as a continuous spectrum: for this reason, patients are preferably stratified according to a decimal classification, available for types I and II only (Dubowitz, 1995; Main et al., 2003).

Irrespective of the phenotypic severity, SMA patients have the same genetic defect: the homozygous loss of the SMN1 gene, located in 5q13 (Lefebvre et al., 1995). In the same region, a hypomorphic allele of SMN1 is present (SMN2), which produces insufficient levels of the SMN protein. Due to an alternative splicing, SMN2 is mainly transcribed into an isoform lacking exon 7 (SMN-del7) and then translated into an unstable protein (Mercuri et al., 2020). The number of SMN2 genes is variable in patients, generally 2–4; SMN2 copy number is the only consistent phenotypic modifier known to date, grossly and inversely related to disease severity (Calucho et al., 2018). The efficiency of exon 7 inclusion in SMN2 mRNA can be enhanced by two relatively rare variants (rs121909192 and rs1454173648), more commonly found in the less severely affected patients (Vezain et al., 2010; Wu et al., 2017).

The molecular pathophysiology of SMA is largely unknown: even though the SMN protein is ubiquitously expressed and has a housekeeping function in splicing regulation, the second motor neuron is the main target cell of the disease (Beattie and Kolb, 2018). However, a growing bulk of evidence supports the pathogenic role of skeletal muscle or even of the whole motor arch (Bricceno et al., 2014; Martínez-Hernández et al., 2014; Ripolone et al., 2015).

Physiological and pathological degeneration of skeletal muscle in SMA has been intensively investigated, also focusing on the role of microRNAs (miRs or miRNAs), 20-22mers non-coding RNAs. miRs are involved in most cellular processes: these regulate gene expression through mRNA degradation or translational inhibition, mainly by binding cis-regulatory elements present in the 3′UTR of mRNAs (Bartel, 2009; Vidigal and Ventura, 2015). Among miRs, the myomiRs play a pivotal role in regulating myogenesis and muscle degeneration (including hsa-miR-1, miR-206, miR-133a, and miR-133b) (Horak et al., 2016).

miRs have been intensively studied also as potential biomarkers in several conditions, including SMA (Kariyawasam et al., 2019). This topic has become even more relevant after the development and registration of the first effective treatments, which have led to a revolutionary change of perspective for patients (Mercuri et al., 2020). The results of presymptomatic treatments (Mercuri et al., 2020) are prompting the development of newborn screening programs, already started in some countries worldwide (Dangouloff et al., 2020), and are making compelling the need of prognostic biomarkers to predict the progression trajectories of patients identified at birth. Besides the SMN2 products, either transcripts or protein, (Tiziano et al., 2013; Tiziano et al., 2019; Crawford et al., 2012), few SMN-independent biomarkers have been evaluated so far, with discordant results; these include the SMA-MAP, creatinine, neurofilament dosage, and a few miRs (Kariyawasam et al., 2019; Kobayashi et al., 2013; Alves et al., 2020; Darras et al., 2019).

In the present study, we have first used whole miRNome sequencing of muscle specimens and cultures from SMA patients and controls to identify a specific signature of the disease and investigate on the muscle involvement in the pathogenic process. Subsequently, the levels of deregulated miRs have been evaluated in serum samples of patients and controls as SMN-independent biomarkers.

We identified three deregulated miRs (miR-181a-5p, miR-324-5p, miR-451a: SMA-miRs) that have been integrated in a composite score (SMA-score), including SMN2 full-length (SMN2-fl) transcript levels, SMN2 copy number and age at sampling, markedly improving the phenotypic predictive value of SMN2 copy number assessment alone.

Methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Gene (Homo sapiens)SMN1GenBankHGNC:HGNC:11,117
Gene (Homo sapiens)SMN2GenBankHGNC:HGNC:11,118
miR (Homo sapiens)hsa-miR-181a-5pmiRBaseMIMAT0000256
miR (Homo sapiens)hsa-miR-324-5pmiRBaseMIMAT0000761
miR (Homo sapiens)hsa-miR-451amiRBaseMIMAT0001631
Strain, strain background (Mus musculus)SMNΔ7 miceFVB.Cg-Grm7Tg(SMN2)89Ahmb Smn1tm1Msd Tg(SMN2*delta7)4,299Ahmb/JJackson LaboratoryStock number: 005025Hum Mol Genet 14(6):845–57, 2005
Genetic reagent (Homo sapiens)miRCURY LNA microRNA mimic: hsa-miR-181a-5p, hsa-miR-324-5p, hsa-miR-451aExiqon50–100–200 nM
Genetic reagent (Homo sapiens)miRCURY LNA microRNA antogomiR: hsa-miR-181a-5p, hsa-miR-324-5p, hsa-miR-451aExiqon0.1 nM
Cell line (Homo sapiens)Primary myoblastsItalian Telethon Network of Genetic Biobanks6756, 6760, 6762,
6816, 7147, 8823,
8655, 8537
Biological sample (Homo sapiens)Muscular biopsiesItalian Telethon Network of Genetic Biobanks10370, 10351, 8023,
4688, 10543, 10583,
7669, 5944, 5824,
5717, 6760, 6438,
6082, 5379, 7689,
5842, 9814
Sequence-based reagentSee Supplementary file 3IDT (Integrated DNA Technologies)
Commercial assay or kitTruSeq Small RNA Sample Preparation kitIlluminaTruSeq Small RNA
Library Prep Kit –
RS-200-0024
Commercial assay or kitmiRCURY RNA Isolation Kit – BiofluidsExiqon
Commercial assay or kitUniversal cDNA synthesis kit IIExiqon
Commercial assay or kitPick-&-Mix miRNA PCR Panel 96-wellExiqon
Commercial assay or kitE.Z.N.A PX Blood RNA KitOmega bio-tekSKU: R1057-01
Commercial assay or kitHigh Capacity cDNA Reverse Transcription KitThermo Fisher ScientificCatalog number: 4368814
Software, algorithmIllumina Genome AnalyzerIllumina
Software, algorithmRealTime StatMinerVersion 4.1
Software, algorithmStatgraphics Centurion XV softwareStatPoint Inc
Software, algorithmSPSS 18.0 softwareSPSSRRID:SCR_002865

Samples

Muscle biopsies from seven SMA patients (three SMA I, two SMA II, and two SMA III) were obtained from the Italian Telethon Network of Genetic Biobanks, held at Istituto Neurologico Carlo Besta in Milan. The characteristics of subjects are specified in Supplementary file 1. The selection criteria of specimens were as follows: (1) unique site of sampling (femoral quadriceps) and (2) first stage of disease (defined as onset of the first clinical signs that prompted the diagnostic workflow) to minimize the presence of fibrosis. Seven muscle biopsies were selected as controls among morphologically normal samples of age-matched subjects who underwent muscular biopsy for neonatal hypotonia (for type I) or hyper-CKemia (for types II and III).

Muscle cell cultures (five from SMA patients and four from controls) were obtained from the Telethon Biobank or set in-house, as previously reported (Zanotti et al., 2007). Four patients (two type I and two type II) had the homozygous deletion of SMN1; the latter was a type I subject, compound heterozygote for the c.439_443delGAAGT (p.Glu147SerfsTer2) variant.

51 SMA patients were enrolled (3 type I, 21 type II, 26 type III, 1 type IV; Supplementary file 2). 47 DNA samples were available for SMN2 copy number assessment and SNP (rs121909192 and rs1454173648) genotyping; whole blood samples were collected from 45 patients for total RNA extraction, and 51 serum samples were available for small RNA extraction.

Serum samples from 37 controls from the general population were analyzed: the selection of these subjects was performed according to a pairing criterion for age and sex with patients. Samples were obtained upon anonymization from the discards of the routine analysis laboratories of Fondazione Policlinico Gemelli and Ospedale Pediatrico Bambino Gesù; only information on sex and age was retained. From controls, no DNA or total RNA were extracted. The two groups were homogenous for sex (28 and 19 females, respectively) and age (mean 17.3 ± 19.2 and 11.9 ± 12.4 years, respectively, p>0.05).

This study was approved by the Ethics Committee of Fondazione Policlinico Universitario IRCCS “A. Gemelli” (Prot # 25188/19, ID: 2614).

For SMA mice, the original breeding pairs of SMNΔ7 mice were purchased from Jackson Laboratory (stock number 005025). The colony was maintained by interbreeding carrier mice, and the offspring were genotyped by PCR assays on tail DNA according to the protocols provided by Jackson Laboratory. According to the ARRIVE guidelines, procedures were carried out to minimize discomfort and pain in compliance with national (D.L. 116 Suppl 40/1992 and D.L. 26/2014) and international guidelines and laws (2010/63/EU Legislation for the protection of animals used for scientific purposes). The experimental protocols were approved by the Ethics Committee of the Fondazione IRCCS Istituto Neurologico C. Besta and by the Italian Ministry of Health (protocol numbers: 962/2016-PR and 1039/2020-PR).

Cell cultures

Myoblasts (passages from 5 to 15) were cultured in high-glucose Dulbecco Modified Eagle’s Medium (DMEM), 20% fetal bovine serum (FBS), 100 u/ml of penicillin, 100 mg/ml of streptomycin, 2 mM L-glutamine, 10 ng/ml of epidermal growth factor (EGF), and 10 µg/ml of bovine insulin in 5% CO2 atmosphere. For myotube differentiation, we have used a standard protocol of serum deprivation (5% FBS) for 2 weeks.

SH-SY5Y (human neuroblastoma) were already present in-house for previous studies and were tested for mycoplasma contamination. We verified the identity of SH-SY5Y cell line by evaluation of neuronal phenotype, following differentiation with 10 pM retinoic acid for 7 days. Cells were cultured in 1:1 DMEM/Ham’s F12 nutrient medium with 20% FBS, 100 u/ml of penicillin, 100 mg/ml of streptomycin, and 2 mM L-glutamine.

Patients

The patients included in the present study were in routine clinical follow-up in the four participating Italian referral neuromuscular centers (Fondazione Policlinico Universitario IRCCS “A. Gemelli,” Istituto Neurologico IRCCS “Carlo Besta,” Ospedale Pediatrico Bambino Gesù, Istituto Giannina Gaslini). Subjects were evaluated by expert neurologists/pediatric neurologists/physiotherapists: SMA type was first attributed to each patient, according to the usual classification (types I–IV). Three of us (MP, GC, EM) were requested to assign each patient to a SMA subtype, based on the clinical data, according to the decimal classification for types I and II (Zerres and Rudnik-Schöneborn, 1995; Dubowitz, 1995). For type III, due to the lack of a decimal classification, we arbitrarily assigned the value 3 to type IIIa and 3.5 to type IIIb.

Whole miRNome sequencing

Total RNA from muscle biopsies and myoblast/myotube cultures was extracted by TRIzol Reagent (Life Technologies) as specified in the manufacturer’s protocol. Libraries were obtained through the TruSeq Small RNA Sample Preparation kit (Illumina). Next-generation sequencing (NGS) miRNome analysis was performed by Illumina Genome Analyzer (GAIIX) platform.

For miRNome analysis, we have used the following pipeline. The sequencing raw data (.bcl files) were processed by the Illumina Casava software (v1.8.0) to convert the data into fastq files (raw data of miRSeq have been deposited at NCBI-SRA database; BioProject PRJNA748014). The fastq files (31-base single-end reads) were first filtered by quality using the FASTX-toolkit (fastq_quality_filter: -q28 p50) and then trimmed to remove the adapter from their 3′ end (TrimGalore tool). Only reads longer than 15 bases were retained and mapped on the miR-precursor sequences annotated in the miRBase repository (v19). The Bowtie2 algorithm was used for the alignment, allowing no more than two mismatches. Quantification, TMM normalization, and significant differential expression test of the known mature miRs were performed using the edgeR package (v2.4.1). Only miRs with >1 count per million (cpm) in at least one condition and in a minimum number of samples (depending on group size) were retained. Multiplicity correction was performed by applying the Benjamini–Hochberg method on the p-values to control the false discovery rate (FDR). The significantly up- and downregulated miRs were selected at FDR < 0.05.

Molecular biomarkers

Genomic DNA was extracted from whole blood by conventional salting out method. SMN2 copy number and RNA analyses were carried out as previously reported (Tiziano et al., 2010). The presence of the exon 7 splicing modifier variants (rs121909192 and rs1454173648; Vezain et al., 2010; Wu et al., 2017) was assessed by Sanger sequencing with the R111 primer (Lefebvre et al., 1995), following PCR amplification with primers R111 and C1120 (Lefebvre et al., 1995).

Whole blood was collected in PAXgene blood RNA tubes (BD Biosciences) and total RNA extracted by the E.Z.N.A. PX Blood RNA Kit (Omega bio-tek), according to the manufacturer’s protocol. RT-PCR was performed by the High Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). SMN2 transcript levels were assessed as previously reported (Tiziano et al., 2010).

miRs were extracted from serum by miRCURY RNA Isolation Kit – Biofluids (Exiqon). RT-PCR was performed by Universal cDNA synthesis kit II (Exiqon). For both protocols, manufacturer’s instructions were followed.

Commercial relative qPCR assays were purchased at Exiqon (Pick-&-Mix miRNA PCR Panel 96-well); UniSp6 was used as calibrator (miRCURY LNA primers). miRs with Ct <34 were considered as expressed.

For absolute qPCR assays, specific forward primers were designed based on the target mature miR sequence reported in http://www.mirbase.org. The reverse primer was shared by all assays and was complementary to the tag of the Universal cDNA synthesis kit II. The melting temperature (Tm) of primers was established by the OligoAnalyzer 3.1 tool (available at the Integrated DNA Technologies website, http://www.idtdna.com). In case of forward oligo Tm <60°C, the latter was optimized by adding a (GACT)n tail at the 5′ end. For external standard construction, we proceeded as previously described (Tiziano et al., 2013). Amplicons for cloning were obtained by PCR-filling of two partially overlapping sequences (Supplementary file 3).

In silico, in vitro, and in vivo experiments

miR-181a-5p, miR-324-5p, miR-451a mimics and scramble, and the respective antagomiRs were purchased at Exiqon (miRCURY LNA microRNA mimicS/antogomiR). For in vivo experiments, antagomiRs were resuspended in artificial cerebrospinal fluid (aCSF: NaCl 119 mM, NaHCO3 26.2 mM, KCl 2.5 mM, NaH2PO4, 1 mM, MgCl2 1.3 mM, glucose 10 mM).

SH-SY5Y cells were transfected with three different concentrations (50–100–200 nM) of miR-181a-5p, miR-324-5p, miR-451a mimics or scramble. Transient transfections were performed through Lipofectamine 2000 (Invitrogen), according to the manufacturer’s protocols. Both total and small RNAs were extracted by the miRvana Paris RNA extraction kit (Ambion). RT-PCR was performed by High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) or Universal cDNA synthesis kit II (Exiqon) for mRNAs or miRs, respectively. miR-181a-5p, miR324-5p, miR-451a, and SMN2 transcript (SMN-fl, SMN-del7) levels were quantified as described above.

At postnatal day 1 (P1), SMA-like pups (Smn−/−, hSMN2+/+, SMNΔ7+/+) were cryo-anesthetized and injected with 5 μl of 0.1 nmol of each specific antagomiR, into the cerebral lateral ventricle. Injections were performed with a pulled capillary needle under the guidance of a transilluminator as reported (Glascock et al., 2011) All the litters were culled so that each litter contained six siblings, daily weighted, and controlled. miRwalk 3.0 (Dweep et al., 2014) was used to identify miRs binding the 3′-UTR of the SMN2 genes.

Sequencing analysis of SMA-miR genes

We have amplified the genomic regions of interest by PCR in a final volume of 12.5 µl using the 2X GoTaq Hot Start Colorless Mastermix (Promega) and 0.4 µM of each primer pair (Supplementary file 3). The amplification cycle was : 95°C 5′; (95°C 45′′; 60°C 45′′; 72°C 30′′) × 35; 72°C 5′; 4°C. Thereafter, following purification of PCR products by ExoSap-IT (USB Corporation), sequencing reactions were performed by the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) and purified by the BigDye XTerminator Purification Kit (Applied Biosystems). DNA sequencing was performed by capillary electrophoresis using the ABI-Prism 3130 instrument (Applied Biosystems). Electropherograms were analyzed with the Sequencing Analysis Software 6 (Applied Biosystems).

Statistical analysis

Relative qPCR data were analyzed using RealTime StatMinerVersion 4.1 software and Benjamini–Hochberg FDR method. Grubbs’ test was employed to exclude the outliers. miR levels in patients and controls were compared by non-parametric tests (Wilcoxon test); miRs were identified as significantly differentially expressed at FDR <0.05.

Absolute qPCR data were analyzed by Statgraphics Centurion XV software (StatPoint Inc). miR levels in patients and controls were compared by Mann–Whitney U test, setting α value at 0.05. While in relative qPCR a large number of miRs were analyzed simultaneously, in absolute qPCR experiments the levels of each miR were evaluated separately; for this reason, the FDR threshold was not applied. To rule out possible false-positive results, we opted for increasing the number of samples in each group only for significant miRs (p<0.05) or showing a trend of significance.

For clinical and molecular correlations, continuous variables were compared by linear regression models. Multiple regression analysis was used to correlate clinical severity and molecular parameters (miR levels, SMN2 transcripts, SMN2 copy number), setting SMA type as dependent variable. Receiver operating characteristic (ROC) curves were constructed by SPSS 18.0 software; the cutoff value of single miR or their sum was identified according to the highest values of sensitivity and specificity.

The survival analysis in SMA-like mice was made with SPSS 18.0 software; treated and untreated mice were compared by Kaplan–Meier survival curves; differences in survival were estimated by the log-rank test.

For all tests, p≤0.05 was considered significant.

Results

miRNome profile suggests a primitive muscular defect in SMA patients

The analysis of the whole miRNome of muscle biopsies showed a distinct clusterization of patient and control samples (Figure 1A). Similar findings were obtained also for myoblast and myotube cultures (Figure 1B–C). Globally, miR production was preserved; at α ≤ 0.05, 99, 20, and 19 miRs were differentially expressed in SMA biopsies, myoblast, and myotube cultures, respectively (Supplementary file 4).

Heatmaps obtained by the whole miRNome analysis of muscle biopsies (A), myoblasts (B), and myotubes (C) of spinal muscular atrophy (SMA) patients and controls; patient and control samples display a separate clusterization.

99, 20, and 19 miRs were found deregulated in SMA in muscle biopsies, myoblasts, and myotubes, respectively; (D) Venn’s diagram showing the five miRNAs shared among the three groups, three between myoblasts and myotubes, two between myoblasts and biopsies, and three between myotubes and biopsies.

The three groups of samples shared five upregulated miRs (hsa-miR-1, -133a, -133b, -204-5p, -208b, Figure 1D), mostly belonging to myomiRs (except for hsa-miR-204-5p). Three differentially expressed miRs were in common between myoblasts and myotubes (hsa-miR-206, -483-5p, and -4697-3p); two were shared between myoblasts and biopsies (hsa-miR-146a-5p and -184), and three between myotubes and biopsies (hsa-miR-378a-3p, -378f, -501-5p) (Figure 1D); hsa-miR-378a-3p and -378f had opposite trend (upregulated in myotubes, downregulated in biopsies).

Serum levels of miR-181a-5p, miR-324-5p, and miR-451a are candidate biomarkers for SMA

Based on the data above, we determined serum levels of miRs that were differentially expressed in muscle samples of patients to identify potential SMN-independent biomarkers for SMA. The validation pipeline is schematized in Figure 2; the results are summarized in Supplementary file 5. Briefly, as a first-tier test we determined the levels of the 99 deregulated miRs in serum samples from 10 patients (one SMA I, 9 SMA II; median age 1.8 years; five females) and 10 age-matched controls. This subgroup of patients has been selected by homogeneity of age and severity. We included 11 additional miRs that were identified in SMA patients in other studies (Kye et al., 2014; Valsecchi et al., 2015; Murdocca et al., 2016; Catapano et al., 2016; Wertz et al., 2016; O’Hern et al., 2017; Sison et al., 2017; Kaifer et al., 2019; Bonanno et al., 2020; Haramati et al., 2010; Gonçalves et al., 2018; Kirby and McCarthy, 2013) or with key function in skeletal muscle. For 74 miRs, qPCR assays were commercially available: the 24 miRs that were differentially expressed were validated in a larger cohort. Globally, we developed in-house absolute qPCR assays for 60 miRs: 24 for the miRs to be validated from the first-tier test, and 36 for the remaining.

Validation pipeline of miRNAs identified by whole miRNome analysis in serum samples of patients and controls.

‘Others’ indicates miRs that were identified in other studies or with key function in skeletal muscle.

The validation step has been performed in 51 patients (3 SMA I, 21 SMA II, 26 SMA III, 1 SMA IV) and 37 age- and sex-matched controls (Supplementary file 2). Most miRs were undetectable in serum samples or did not display different expression levels in the two groups. Three miRs were significantly upregulated in SMA patients: miR-181a-5p, miR-324-5p, and miR-451a (SMA-miRs; Mann–Whitney U test, p=4.3 * 10–4; 0.02; 0.004, respectively; Figure 3A–C). To rule out that the observed differential expression could be biased by RNA quality/quantity, we performed linear correlation analysis among the three miRs: miR-181a-5p and miR-451a levels were related in both patients and controls, while miR-324-5p levels were independent from the two others (Figure 3—figure supplement 1).

Figure 3 with 4 supplements see all
The SMA-miRs (miR-181a-5p [A], miR-324-5p [B] and miR-451a [C]) were significantly upregulated in serum samples of spinal muscular atrophy (SMA) patients (p=4.3 * 10–4; 0.02; 0.004, respectively).

Receiver operating characteristic (ROC) curves showed that the quantification of SMA-miRs has 80% sensitivity and 75% specificity in distinguishing patients from controls (D). Correlation of miR-324-5p with SMA type (E): the levels in SMA II and SMA III patients were significantly increased compared to those of SMA I patients (p=0.03 and 0.04, respectively).

We evaluated sensitivity and specificity of the quantification of the SMA-miRs by using the ROC curves: the highest predictive value was found for miR-181a-5p, with 75% and 61% of sensitivity and specificity, respectively (cutoff 70.5 molecules/µl of serum, p<10–4, Figure 3D). To evaluate whether the combination of the SMA-miRs could be more predictive than the levels of the single miR, we constructed the ROC curves of the sum and found an increase in both sensitivity and specificity, up to 80 and 75%, respectively (cutoff 380 molecules/µl of serum, p<10–4, Figure 3D).

Globally, the SMA-miRs did not show any correlation with age or sex (p>0.05, Figure 3—figure supplements 1 and 2); in patient samples, miR-181a-5p levels were significantly increased in females compared to males (p=0.024, Figure 3—figure supplement 2).

Regarding the correlation between SMA-miR levels and the severity of the disease, only the levels of miR-324-5p were significantly decreased in type I compared to types II and III (p=0.03 and 0.04, respectively; Figure 3E). Finally, we compared miR levels in type II and III patients with three SMN2 copies. The difference in miR levels was not significant; however, the two groups were significantly different by age (p=0.0092; Figure 3—figure supplement 3A and B).

SMA-miRs do not modulate SMN transcript levels

To evaluate whether SMA-miRs could modify SMN1 or SMN2 expression levels, we performed transient transfections of SH-SY5Y neuroblastoma cells with commercial mimics or scramble. The transfection of the single mimics, at the final concentration of 50 or 100 nM, led to an increase of SMA-miR levels from 0 to 1 mol/ng of total RNA in untreated cells, up to 60–100 or 160–200 mol/ng, respectively (Figure 3—figure supplement 4A). Despite the huge increase in SMA-miR levels, SMN1/SMN2 transcripts remained unchanged, except for the SMNdel7 isoform in cells treated with miR-324-5p, which was reduced by 50% independently of the mimic concentration (Figure 3—figure supplement 4B). Since in untreated cells SMA-miR levels were almost undetectable, we did not perform experiments with the antagomiRs.

SMA-miRs, in the absence of SMN-modifying treatments, do not improve the survival of SMA-like mice

To test the hypothesis of the retrograde effect of the secretion of SMA-miRs on spinal cord cells, we evaluated the survival of SMNΔ7 mice in the absence of any modification of SMN levels. We first treated at P1 five affected mice by intrathecal injection of each specific SMA-miR antagonist. Anti-miR-324-5p and anti-miR-451a did not affect mice survival and were not further studied, although the first led to a significant transient increase in body weight (between P7 and P10, p=0.002; Figure 4—figure supplement 1). Conversely, since anti-miR-181a-5p significantly improved the survival, we increased the cohort of treated animals (n = 36), which were compared with untreated animals (n = 71), and with those treated with the scramble (n = 31). The overall survival remained unchanged (p>0.05, Figure 4).

Figure 4 with 1 supplement see all
Survival curves of SMNΔ7-mice treated with intrathecal injection of anti-miR-181a-5p (n = 36) and untreated (n = 71); the overall survival remained unchanged (p>0.05).

SMA-miRs do not display gene variants in patients

We tested whether sequence variation in genes encoding the SMA-miRs could be related to SMA phenotype. We did not identify any variant in 38 samples of patients affected from different forms of SMA (3 type I, 17 type II, 18 type III).

Hsa-miR-9, -19a-3p, -23a-3p, -146a, and -206 were not differentially expressed in SMA serum samples

Some miRs were previously reported to be differentially expressed in serum samples of SMA patients (hsa-miR-9, -19a-3p, -23a-3p, -132, -146a, -183, -206, -431) (Kye et al., 2014; Valsecchi et al., 2015; Murdocca et al., 2016; Catapano et al., 2016; Wertz et al., 2016; O’Hern et al., 2017; Sison et al., 2017; Kaifer et al., 2019; Bonanno et al., 2020; Haramati et al., 2010). We evaluated whether these miRs were differentially expressed also in our samples (results are schematized in Table 1). miR-9, -132, -183, -206, and -431 were not differentially expressed in muscle samples, whereas miR-19a-3p and -146a were upregulated; miR-23a-3p was downregulated. miR-19a-3p, -146a, and -23a-3p were upregulated in serum samples of patients when evaluated by r-qPCR; these preliminary data were not confirmed by a-qPCR. Also miR-206 and -9 were not differentially expressed by a-qPCR. At the time of study design, miR-132, -183, and -431 were not identified yet and thus have not been tested.

Table 1
miRs differentially expressed in spinal muscular atrophy (SMA), as reported in previous studies.
hsa-miRmiRNomeRelative qPCRAbsolute qPCR*Reference
miR-19a-3pUpregulatedUpregulated<10 molecules/µl serum; p=0.42Haramati et al., 2010; Gonçalves et al., 2018
miR-23a-3pDownregulatedUpregulated150–200 molecules/µl serum; 10 patients/controls analyzed; p=0.62Kaifer et al., 2019
miR-206NonsignificantUpregulated50–100 molecules/µl serum; 15 patients/controls analyzed; p=0.24Valsecchi et al., 2015; Catapano et al., 2016; Bonanno et al., 2020
miR-9NonsignificantNot tested<10 molecules/µl serum; p=0.30Catapano et al., 2016
miR-132NonsignificantNot testedNot testedCatapano et al., 2016
miR-146aUpregulatedUpregulated<5 molecules/µl serum; p=0.10Sison et al., 2017
4miR-431NonsignificantNot testedNot testedWertz et al., 2016
miR-183NonsignificantNot testedNot testedKye et al., 2014
  1. *

    p-Values refer to the significance of comparison of the miR levels in patients and controls by Mann–Whitney U-test. p-values < 0.05 were considered significant.

The SMA-score: phenotypic severity can be predicted by combining SMN2 copy number, SMN2-fl, miR-181a-5p, miR-324-5p, miR-451a, and age

We evaluated whether SMA-miR levels in serum could improve the accuracy of phenotype prediction with respect to the available molecular biomarkers. SMN2 copy number determination alone provided a moderate accuracy when related to SMA type (R2 = 52.45%, n = 41, p<10–5; Figure 5—figure supplement 1A); the use of the decimal classification of SMA raised the R2 up to 67.04% (n = 39, p<10–5; Figure 5—figure supplement 1B). None of the patients had the rs121909192 or rs1454173648 variants. Considering also age at sampling in a multiple regression model, R2 raised to 61.58% (R2 = 53.16%, n = 22, p=0.0005, in patients < 6 years). When including also SMN2-fl levels and the sum of SMA-miRs as covariates, R2 further raised up to 67.04% (n = 40, p<10–5); more importantly, when considering only patients < 6 years, R2 raised to 72.17% (n = 21, p=0.0001). The equations describing the multivariate models were as follows.

  • All ages:

  • SMA type = 0.2473 + 0.0013 * age (months) + 0.0013 * SMN2-fl + 0.5417 * #SMN2 +0.00002 * SMA-miRs.

  • Age <6 years:

  • SMA type = –0.0700 + 0.0175 * age (months) + 0.0021 * SMN2-fl + 0.4398 * #SMN2 +0.00001 * SMA-miRs.

We compared by linear regression models the SMA-scores obtained with the two equations above in patients < 6 years; the correlation coefficient was 0.90, and R2 80.31 (p<10–5, n = 21, Figure 5—figure supplement 1D).

Then, we related the SMA subtype for each patient as from the equations above, with the decimal classification obtained from the blind evaluation. We found a correlation coefficient of 0.87 (R2 = 75.77%, n = 38, p<10–5) for the whole group and 0.87 (R2 = 77.14%, n = 21, p<10–5) for patients < 6 years (Figure 5A and B). When evaluating patients with three SMN2 copies only, the decimal classification and the SMA-score were significantly related (R2 = 30.04, p=0.008, n = 21, Figure 5—figure supplement 1C).

Figure 5 with 1 supplement see all
The spinal muscular atrophy score (SMA-score) predicts the phenotypic severity in SMA patients.

Correlation between the SMA-score and the clinical decimal SMA subtype in the whole cohort (A) and aged <6 years (B). Red circles are individual samples, the blue line indicates the expected distribution, the green line indicates the 95% confidence interval, and the black lines are the prediction interval.

Discussion

The landscape of SMA has been so revolutionized over the last few years by the availability of effective treatments that the solution of past issues has become more urgent and novel ones have come to the surface. Firstly, the usual clinical classification is unsatisfactory for several reasons: (1) the treatment of patients has uncovered novel phenotypes that do not fall in any of the classical forms (Mercuri et al., 2020); (2) the spreading of newborn screening programs is changing the diagnosis of SMA into that of subjects with a genetic defect who might or not develop signs of the condition. Secondly, the identification of prognostic response and predictive biomarkers has become even more urgent since (1) the available outcome measures may not be sensitive enough to detect slight improvements that may still be clinically relevant and (2) the molecular effect of treatments that target CNS only cannot be evaluated peripherally.

The only genomic biomarker with clinical relevance is the determination of SMN2 copy number, alongside two alternative splicing-modulating variants (rs121909192 and rs1454173648; Vezain et al., 2010; Wu et al., 2017) even though some other modifier genes have been reported (Kariyawasam et al., 2019). In this study, we have exploited an unbiased approach to identify deregulated muscular miRs that could be dosed in serum samples as candidate biomarkers for SMA; at the same time, high-throughput data might provide hints on the possible pathogenic role of skeletal muscle. miRs and neurodegeneration processes are tightly related: the depletion of DICER in mice leads to a phenotype resembling SMA (Gonçalves et al., 2018); DROSHA is downregulated in motor neurons of a SMA model (Haramati et al., 2010). In this latter study, the depletion of the bouquet of expressed miRs was interpreted as due to a global deregulation of miR biogenesis. In our study, less than 10% of miRs (100/1115 annotated miRs) were significantly deregulated in patients’ muscle biopsies. This discrepancy might be related to the obvious difference in the tissues analyzed. More importantly, while we used human tissues, the motor neurons studied by Goncalves et al. were from the Taiwanese murine model of SMA that displays very low SMN levels, incomparable with those found in patients.

The involvement of skeletal muscle in SMA is unquestionable; to discern whether active or passive, we used the dual approach of muscle cell cultures and biopsies. While alterations found in vivo only could have been either primitive or secondary to the denervation/atrophy processes, the alterations found also in cultured cells point to a primary defect related to SMN deficiency.

In cell cultures, the number of differentially expressed miRs was smaller than in muscle samples. This finding could be related to the higher variability of cultured cells compared to the in vivo specimens. Interestingly, of the five miRs shared in all groups of samples (hsa-miR-1, -133a, -133b, -204-5p, -208b all upregulated), four belonged to the myomiRs: miR-1 and miR-133 family levels inversely correlate with myogenic differentiation/proliferation status (Kirby and McCarthy, 2013); miR-208b overexpression favors proliferation to differentiation during skeletal muscle development (Fu et al., 2020). Globally, these data suggest a primary muscular defect in SMA, determining a delay in the differentiation path from myoblasts to maturely innervated myofibers. The maturation defect of SMA skeletal muscle has been historically reported in morphological studies ahead of the identification of SMN1 (Fidziańska et al., 1990) and has been confirmed in more recent studies on human (Martínez-Hernández et al., 2014) and murine tissues (Bricceno et al., 2014). Additionally, Houzelle et al. have very recently shown that hsa-miR-204-5p and -133b levels inversely correlate with the mitochondrial activity in skeletal muscle (Houzelle et al., 2020), suggesting that these miRs could be involved in the mitochondrial depletion observed in SMA (Ripolone et al., 2015). How miR modulation occurs is unknown: SMN is not directly binding miRs even though the SMN complex can assemble with these small RNAs (Chen and Chen, 2019); conversely, in silico predictions (Dweep et al., 2014) indicate that miR-133a, -133b, and -204-5p bind the 3′-UTR of SMN2, and thus might downregulate SMN transcript/protein levels.

In this study, we used the unbiased dataset above to identify candidate SMN-independent biomarkers. We first tested >100 miRs in a small cohort of patients and controls to get the shortlist of potential biomarkers. Most miRs were not detectable in serum, indicating the integrity of the sarcolemma in SMA, differently from other neuromuscular conditions such as Duchenne muscular dystrophy (Cacchiarelli et al., 2011). Only three miRs were overexpressed in serum of patients (miR-181a-5p, miR-324-5p, and miR-451a, the SMA-miRs), suggesting that these might be actively secreted by skeletal muscle. We can hypothesize that these miRs, which do not modulate SMN2, may exert a loco-regional effect in muscle and have, at the same time, an at-distance-effect following secretion.

Regarding the possible pathophysiological function of the SMA-miRs, to our knowledge, few data are available on the role of miR-324-5p in skeletal muscle: a single study reported the upregulation in human CD56+ myoblasts, during differentiation (Dmitriev et al., 2013), again supporting the immaturity of SMA skeletal muscle. The putative role of the secreted miR-324-5p remains unknown: Sun et al., 2019 reported an essential function of this miR in synapsis formation. Intriguingly, we observed that even if the intrathecal administration of the specific antagomir did not improve the survival of SMNΔ7 mice, however, it induced a significant transient increase in body weight (Figure 4—figure supplement 1).

More information is available regarding the role of miR-451a: an increase was found in aged muscle (Mercken et al., 2013), whereas the downregulation occurred during endurance training (Zacharewicz et al., 2013), suggesting that miR-451a levels may be inversely related to muscle mass. However, in our cohort, this miR was upregulated in patients independently of the degree of the atrophic process. Some studies have reported the simultaneous modulation of miR-451a and miR-181a-5p: in acute exercise both miRs are upregulated, whereas in aging these displayed an opposite trend (Zacharewicz et al., 2013), suggesting that the modulation in SMA is independent of the reduced mobility of patients; interestingly, we found that the expression levels of the two miRs are related (Figure 3—figure supplement 1). miR-181a-5p displays the more intriguing profile in terms of putative involvement in SMA: Ouyang et al., 2012 reported a worsening of brain injury following the increase of miR levels in a mouse model of stroke, whereas the depletion accelerated the recovery. More importantly, Benigni et al., 2016 found an increase in miR-181a levels in CSF of amyotrophic lateral sclerosis patients. We argue that miR-181a-5p might be a part of a retrograde signaling system from skeletal muscle, which may accelerate motor neuron loss in SMA. To test this hypothesis, we treated SMNΔ7 mice with both mimic and anti-miR-181a-5p without increasing SMN levels. miR-181a-5p modulation alone was not sufficient to significantly improve the survival of affected mice even if some changes in Kaplan–Meier curves were observed in a subset of animals (Figure 4).

If confirmed in other studies with more therapeutic or pathogenic purposes, our data suggest that systemic therapeutic approaches increasing SMN levels also in skeletal muscle may provide additional benefits to SMA patients, and that miR-181a-5p (and/or miR324-5p) modulation might be a potential target for combinatorial treatments in addition to SMN modulation.

To the best of our knowledge, the miRs we have identified in the present study have not been described in SMA so far. Previously, other miRs have been found differentially expressed in animal models or in serum samples of patients (miR-9, -19a-3p, -23a-3p, -132, -146a, -183, -206, -431) (Kye et al., 2014; Valsecchi et al., 2015; Murdocca et al., 2016; Catapano et al., 2016; Wertz et al., 2016; O’Hern et al., 2017; Sison et al., 2017; Kaifer et al., 2019; Bonanno et al., 2020; Haramati et al., 2010). Of these (Table 1), three (miR-132, -183, -431) were not differentially expressed in muscle biopsies and had not been described at the time of study design, thus have not been tested here; among the others, three were almost undetectable in serum samples of our cohorts (miR-9, -19a-3p, -146a) and two were not differentially expressed (miR-23a-3p, -206). The discrepancy between our and previous data may be ascribed to the different technical approaches used: as in the case of SMN2 mRNAs (Tiziano et al., 2010), the results of absolute qPCR, differently from the relative approaches, are not affected by the expression levels of endogenous genes and calibrators. Moreover, in the case of low-copy number transcripts, the relative approaches may magnify even small differences in expression levels, which may be statistically significant but of doubtful biological meaning.

One of the most relevant results of our study is the development of the SMA-score. The global spreading of newborn screening programs for SMA has made compelling the identification of tools with good predictive power of the clinical severity (Dangouloff et al., 2020). So far, the stop-or-go to the treatment of presymptomatic patients is uniquely based on SMN2 copy number assessment, which is roughly predictive of the clinical severity in individual patients. Even more critical are patients with three SMN2 copies whose severity may range from type I to type III (1.9 to 3b in our cohort).

Two SMN2 variants (rs121909192 and rs1454173648) modulate the inclusion efficiency of exon 7 into mature mRNA, (Vezain et al., 2010; Wu et al., 2017), but these variants are relatively rare in patients. Additional SMN2 variants have been very recently described (Blasco-Pérez et al., 2021); however, the frequency and the functional effect of these variants have not been elucidated yet. For these reasons, more functional and dynamic molecular markers could reasonably improve the prediction of the severity. The inclusion of serum SMA-miRs, whole-blood SMN2-fl levels, and age has markedly increased the accuracy of the severity prediction of SMN2 copy number alone, from about 52% to about 75%. The age effect might be related to the physiological modulation of SMN2-fl levels over time, as previously reported (Crawford et al., 2012). In patients with three SMN2 copies, the SMA-score was significantly related to the decimal classification of patients, even if with about 30% strength (Figure 5—figure supplement 1C). The increase in the population size might improve these results.

The main drawback of our results is related to the cross-sectional nature of the present study. While we have previously shown that SMN2-fl levels are stable in untreated patients for >1 year (Tiziano et al., 2019), longitudinal data on SMA-miR stability are lacking. In any case, repeated samplings could not be feasible in our study since almost all patients (except for the few type I subjects) were treated with SMN-modifying compounds (such as salbutamol; Tiziano et al., 2019; Tiziano et al., 2010; Angelozzi et al., 2008) or new experimental treatments. Also, the effect of SMN-modifying treatments on SMA-miRs is unknown. The collection of longitudinal data would be highly desirable, namely in presymptomatic patients, identified in our and other newborn screening projects (Dangouloff et al., 2020). If the accuracy will be confirmed in replicative studies, the SMA-score might be included in the clinical routine, as part of the prognostic process, once newborn testing will be universally available.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting files. Raw sequencing data are available at NCBI-SRA database; BioProject PRJNA748014.

References

  1. Book
    1. Dweep H
    2. Gretz N
    3. Sticht C
    (2014) RNA mapping
    In: Alvarez ML, Nourbakhsh M, editors. Methods in Molecular Biology, Vol. 1182. New York, NY: Springer. pp. 289–305.
    https://doi.org/10.1007/978-1-4939-1062-5_25

Decision letter

  1. Christopher Cardozo
    Reviewing Editor
  2. Mone Zaidi
    Senior Editor; Icahn School of Medicine at Mount Sinai, United States
  3. Christopher Cardozo
    Reviewer
  4. Zachary Graham
    Reviewer

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

The most interesting aspects of the study identification of potential biomarkers that could be applied to treatment decisions at a time when therapies for SMA are now available. If confirmed, the findings would be of broad interest for therapeutics and for greater understanding of the role of miRs in SMA pathogenesis.

Decision letter after peer review:

Thank you for submitting your article "SMA-miRs (miR-181a-5p, -324-5p, and -451a) are overexpressed in spinal muscular atrophy skeletal muscle and serum samples" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Christopher Cardozo as Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Mone Zaidi as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Zachary Graham (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

General comments:

1. Data that is listed as 'available on request' or 'data not shown', should be provided in a Supplementary file either with the journal or a 3rd party repository (figshare or similar). In this age of almost unlimited file space, all data reported should be directly presented to the reader.

2. Figures should be redone with open boxes so that all of the data points can be easily seen.

3. This Italian cohort of authors should be commended for their preparation of the manuscript in English. However, there are many sections (such as paragraph 4 of the introduction) that would be greatly improved by a careful proofread by a native English speaker.

4. One abbreviation of microRNA, either miR or miRNA, should be used throughout the paper.

5. Background information on what miRNA are and their biological affect would be useful.

6. It is unclear what 51 serum/RNA/DNA samples means. Does it simply mean the authors collected 51 samples, then isolated RNA and DNA from it? Or were samples received with both RNA and DNA already isolated?

7. Exact p values should be listed uniformly throughout the paper (some figures have asterisks or section have p<0.05).

Specific Comments

Abstract

8. Under results, add that the SMA-score improved prediction to > 80% for those under age 6.

9. Under Discussion – it seems premature to speculate that the role of the differentially altered miRs was in programs of satellite cell differentiation. Keep in mind that the muscle biopsies sampled primarily muscle fibers and that satellite cells in those biopsies were quiescent, rather than the active proliferating form they take as myoblasts. Is it not also possible that these miRs target other key genes in mature myofibers? In addition, the paper really did not delve into role(s) of SMA in muscle per se, and did not try to link SMA1/2 levels to expression of the miRs identified. Thus, to me at least, any comment about function of SMA in muscle seems inappropriate for the abstract for this paper.

Introduction

10. The authors do not describe what SMN1/2 encodes, what its biological function is, and why genetic disruption of its expression is so detrimental to muscle function. A brief paragraph providing some general background on SMN would be appreciated for readers not familiar with SMA. Additional notes of onset, severity and phenotype differences between SMA I-III would also be useful to provide some additional background for the disease.

11. Paragraph #2 of the introduction should be removed or worked into other portions of the paper.

Methods

12. If possible, a more clear definition of what the authors mean by 'first stages' using, for example, clinical staging criteria, would improve the manuscript.

13. What passage numbers were use for cell cultures? How were the primary myoblasts isolated and stored (i.e. was CD56+ screening used)? These are important for people that may wish to try to replicate your study.

14. Please list the exact local ethics committee (hospital or university IRB or equivalent) and approval number.

15. Under 'Samples', add a reason that those controls were selected for the studies. Under discussion, review whether selection of these controls could be a confound.

16. Please provide a method for isolating/freezing myoblasts or cite your prior work if you have published these techniques.

17. Under 'Patients', were the authors involved in assigning SMA types blinded to the patient's diagnosis?

18. Under 'Whole miRome Sequencing', what filters were used (e.g., exclude miRs under 100 counts, etc). Are the raw read data available? Were they deposited in NIH GEO or other databases?

19. Under 'Statistical Tests' please clarify if an FDR threshold was applied for non-parametric tests (Mann-Whitney U-test). Also, please expand on how the multivariate analysis was performed in sufficient detail that others could use the sama analysis technique.

Results and Discussion

20. p values reported for miR-324 and 451 in Results section for Figure 3 do not match those listed on the figures for Figure 3 (.02 and.004 in Results section,.05 and.001 on the figures).

21. A significant weakness of the manuscript is that the experiments with injection of miRs into ventricles of the brain did not include controls to confirm that the miRs influenced mRNA or protein levels in the central nervous system. The most logical place to look for this disease of lower motor neurons would be spinal cord. Also unclear is why the miR were delivered to the central nervous system rather than skeletal muscle where they arose. Conceptually, any effects of the miRs identified in muscle and the circulation would be most likely to be evident in muscle or, possibly, at the nerve terminal; alternatively, circulating miRs could influence gene expression at any location in the body. I do not know if miRs cross the blood brain barrier.

22. It is quite difficult to figure out whether muscle or serum levels of miRs were used in the multivariable regression. The text of the results and discussion should be revised accordingly.

23. Relative abundance is one key parameter to consider when evaluating biological significance of miR expression. A comment regarding which of the differentially expressed miRs are highly expressed would be appropriate. The myoMirs are generally highly expressed in muscle and would be expected to be included among the highly expressed miRs in this dataset.

Reviewer #1 (Recommendations for the authors):

In this manuscript, the authors aimed to define a set of micro-RNAs (miRs) that were differentially expressed in patients with spinal muscular atrophy (SMA), to determine whether any of these miRs had roles in the disease, to validate prior data on altered miR expression in muscle of SMA patients with mutations of SMA 1, and to use information on miR levels in serum as a biomarker to improve the accuracy with which disease severity was predicted.

Strengths of the study include examination of miRs as a potential disease modifier in muscle biopsy samples and serum from a relatively large number of participants studied for relatively rare condition, properly controlled analysis to identify differentially expressed miRs, well controlled cell cultures studie of effects of those miRs on SMA 1 expression and an interesting algorithm for increasing disease severity prediction.

Experiments testing effects of injection into cerebral ventricles showed no effect of miRs on survival of SMA δ-7 mice, but did not report levels of these miRs in spinal cord or effects of the injection on known targets in spinal cord tissues thus limiting interpretation of the data.

If independently validated using a separate cohort, these findings may improve counseling and personalized treatment of patients with SMA.

The manuscript would be considerably improved if the following points were addressed.

General comments:

The authors refer to multiple sets of data that has not been included in the manuscript. These data should be added as supplemental material.

Figures should be redone with open boxes so that all of the data points can be easily seen.

The manuscript would benefit from careful editing for English, typographical errors and punctuation.

Overall, while the data from the manuscript are very interesting, it is a bit unfocused and includes experiments that address several different questions such as which miRs are differentially expressed in muscle, whether these changes are observed in serum, whether changing expression of these miRs in the nervous system influences disease progression, and whether circulating miRs could be a clinically useful biomarker for disease severity.

Specific Comments

Abstract

Under results, add that the SMA-score improved prediction to > 80% for those under age 6.

Under Discussion – it seems premature to speculate that the role of the differentially altered miRs was in programs of satellite cell differentiation. Keep in mind that the muscle biopsies sampled primarily muscle fibers and that satellite cells in those biopsies were quiescent, rather than the active proliferating form they take as myoblasts. Is it not also possible that these miRs target other key genes in mature myofibers? In addition, the paper really did not delve into role(s) of SMA in muscle per se, and did not try to link SMA1/2 levels to expression of the miRs identified. Thus, to me at least, any comment about function of SMA in muscle seems inappropriate for the abstract for this paper.

Methods

Under 'Samples', add a reason that those controls were selected for the studies. Under discussion, review whether selection of these controls could be a confound.

Please provide a method for isolating/freezing myoblasts or cite your prior work if you have published these techniques.

Under 'Patients', were the authors involved in assigning SMA types blinded to the patient's diagnosis?

Under 'Whole miRome Sequencing', what filters were used (e.g., exclude miRs under 100 counts, etc). Are the raw read data available? Were they deposited in NIH GEO or other databases?

Under 'Statistical Tests' please clarify if an FDR threshold was applied for non-parametric tests (Mann-Whitney U-test).

Results and Discussion

A significant weakness of the manuscript is that the experiments with injection of miRs into ventricles of the brain did not include controls to confirm that the miRs influenced mRNA or protein levels in the central nervous system. The most logical place to look for this disease of lower motor neurons would be spinal cord. Also unclear is why the miR were delivered to the central nervous system rather than skeletal muscle where they arose. Conceptually, any effects of the miRs identified in muscle and the circulation would be most likely to be evident in muscle or, possibly, at the nerve terminal; alternatively, circulating miRs could influence gene expression at any location in the body. I do not know if miRs cross the blood brain barrier.

It is quite difficult to figure out whether muscle or serum levels of miRs were used in the multivariable regression. The text of the results and discussion should be revised accordingly.

Relative abundance is one key parameter to consider when evaluating biological significance of miR expression. A comment regarding which of the differentially expressed miRs are highly expressed would be appropriate. The myoMirs are generally highly expressed in muscle and would be expected to be included among the highly expressed miRs in this dataset.

Reviewer #2 (Recommendations for the authors):

Abiusi et al., describe a novel pipeline for predicting severity of infantile spinal muscular atrophy (SMA). They use muscle biopsies and primary cell cultures as well as serum from individuals with SMA. For additional studies, they use an SMA mouse model with miRNA anti-sense injections. They conclude their new model was able to improve prediction of phenotypic severity when related to patient subtype and stratified by age. While the paper has a solid base, I think there are areas that need to be clarified, strengthened or carefully considered. I thank the authors for doing this work and putting together the manuscript. I hope the following will help solidify and improve aspects of the paper.

Strengths

– SMA is a deleterious genetic disease and the ability to improve prediction of severity would be an important clinical outcome.

– Use of miRNA sequencing to find biomarkers is a solid experimental approach.

– Coupling biomarkers with genetic outcomes is a logical predictive strategy.

– Follow-up testing on previously proposed biomarkers.

Weaknesses

– There is a disparity in FDRs used for various miRNA analyses. Reasons for differences in the accepted FDR should be explained and also may have affected downstream analyses. FDRs listed for miRNome was said to be selected at p<0.01 in the methods, with muscle having 99, myoblasts having 20 and myotubes having 19 DE miRNA. However, using their stated FDR of p<0.01, two of their selected predictors, miR-181 and miR-324 do not meet that threshold for being differentially expressed. Then there is a switch to accepting an FDR of p<0.05. Uniformity of data analyses is crucial for biomarker studies and this is a major oversight.

– For the first tier round of qPCR follow up, how and why where those individuals selected out of the complete cohort? Having 1 SMA1 and 9 SMA II seems like a very oddly weighted group breakdown.

– How was type I error limited for the absolute qPCR? FDRs of relative qPCR were stated to be an FDR but nothing is mentioned for absolute qPCR. This is important considering 50 assays were completed for these outcomes.

– A table showing all the outcomes of the relative and absolute qPCR would provide a great deal of information and transparency. For example, the miRNA selected as 'predictors' for SMA were upregulated, but biomarkers can also be down-regulated.

– The link of SMN being depleted in muscle having primary effects is intriguing, but should be interpreted with caution. The importance of having a tonic motor connection with a healthy motor neuron cannot be underestimated. Changes in differentiation factors in proliferating myoblasts and forming myotubes related to myomiR are often seen in other pathological conditions that result from denervation (such as Parkinson's, age-related denervation and ALS). Additionally, only these factors were considered and not any phenotypic measurements of differentiating myotubes. The data are intriguing, but directed and focused studies are needed to investigate how muscle may or may not be regulated by global SMN1 reductions.

– There is no description of when the biopsies of the SMA patients were taken. As they say they were taken from the quadriceps during the first stages of the disease, this could be quite variable depending on fiber type of the muscle group the biopsy was taken from, time diagnosis was confirmed and the age at the time of biopsy. Clarification of these points, as well as for the primary tissue cultures, is important.

– It is not listed where the serum from individuals with SMA came from. Please explain if these were obtained from a repository similar to the muscle tissues or if separate process was conducted.

– The disparity in group sizes from type I to type II and III is unfortunate. It is understood the difference in disease disparity causes this, but it is a limitation in the search for prognostic biomarkers to have such unbalanced group differences.

– Assignment of patients to the SMA subgroup is vague and likely difficult to understand for non-clinicians. Explicit descriptions of how each individual used the clinical data available to stratify patients would be appreciated.

– The authors do not describe what SMN1/2 encodes, what its biological function is, and why genetic disruption of its expression is so detrimental to muscle function. A brief paragraph providing some general background on SMN would be appreciated for readers not familiar with SMA. Additional notes of onset, severity and phenotype differences between SMA I-III would also be useful to provide some additional background for the disease.

– Data that is listed as 'available on request' or 'data not shown', should be provided in a Supplementary file either with the journal or 3rd party repository (figshare or similar). In this age of almost unlimited file space, all data reported should be directly presented to the reader.

– Paragraph #2 of the introduction should be removed or worked into other portions of the paper.

– This Italian cohort of authors should be commended for their preparation of the manuscript in English. However, there are many sections (such as paragraph 4 of the introduction) that would be greatly improved by a careful proofread by a native English speaker.

– One abbreviation of microRNA, either miR or miRNA, should be used throughout the paper.

– Background information on what miRNA are and their biological affect would be useful.

– It is unclear what 51 serum/RNA/DNA samples means. Does it simply mean the authors collected 51 samples, then isolated RNA and DNA from it? Or were samples received with both RNA and DNA already isolated?

– Exact p values should be listed uniformly throughout the paper (some figures have asterisks or section have p<0.05).

– What passage numbers were use for cell cultures? How were the primary myoblasts isolated and stored (i.e. was CD56+ screening used)? These are important for people that may wish to try to replicate your study.

– Please list the exact local ethics committee (hospital or university IRB or equivalent) and approval number.

– p values reported for miR-324 and 451 in Results section for Figure 3 do not match those listed on the figures for Figure 3 (.02 and.004 in Results section,.05 and.001 on the figures).

–Related to figures, improved opacity to allow the reader to see the individual data points would be appreciated.

Reviewer #3 (Recommendations for the authors):

The manuscript by Abiusi E, Infante P and co-workers describes a study to discover and validate miRNAs in muscle of mice models of SMA and biopsies and cultures of SMA patients. Furthermore, the authors study their presence in serum of 51 patients and report an attempt to correlate results in serum miRNAs with SMN2 copy number, FL transcripts and age of the SMA patients under study. The resulting correlation is named SMA-score.

The authors managed to develop an appropriate streamline to detect and validate candidate miRNAs. Data of specific miRNAs is interesting but at present there is lack of longitudinal analysis or trajectories to ascribe them a value as dynamic biomarkers. These issues should be investigated to ascribe a potential role of miRNAs as biomarkers and to further validate the utility of the SMA-score.

SMN2 analysis of these patients was appropriate including the variants rs121909192 and rs1454173648 as positive modifiers (that were discarded in the cases under study) strengthen the accuracy of this variable to include into the score. Further investigation of the entire sequence of the SMN2 of patients may provide additional information about prediction of severity and would be a next step to help improving accuracy of SMA prediction and possibly will be included in the score.

1) Title. I would edit the title to reflect a more realistic scenario of this investigation that is the discovering and validation of miRNAs. The application of the SMA-score is in my opinion preliminary (see below further comments).

2) Abstract. Edits are also necessary in the abstract. i.e. there is incomplete information given that a part of these studies were performed in mice models; wording: "evaluate its controversial pathogenic role" is not addressed in this work and the conclusion that "may provide additional benefit to SMA patients" is not new; also "the SMA-score could be crucial for the prognostic assessment of pre-symptomatic patients" is rather speculative given the lack of data in pre-symptomatic patients regarding the candidates miRNAs and the score application. In summary, the abstract should be more focused in the issues of the study.

3) Methods. The authors should clarify the meaning of "first stages", does mean that all the samples were taken at the beginning of manifestations? Perhaps a Table (could be also supplementary) can be included with a summary of the samples studied with additional information of first stages.

4) Discovering and data of specific miRNAs is interesting and the analysis is adequate. The lack of longitudinal analysis or trajectories to ascribe them a value as dynamic biomarkers should be commented.

5) The cohort of type I patients is small in my opinion to establish a correlation (only four patients). But it could be interesting to try a correlation between the values of type II versus type III cases to determine if patients with three copies acquire (type III) or not (type II) the ability for autonomous walking.

6) There are four patients in the list of type II and III disease with two SMN2 copies. In absence of positive modifier variants (rs121909192 and rs1454173648 also known as NM_017411.3:c.859G>C and NM_017411.3:c.835-44A>G)(both nomenclatures can be included for clarity in the paper) these "outliers" may influence the final score? i.e. PT11 2.70 vs. PT 38 2.12 >6 years (according to data set). Values should be with period not commas, the same applies to the equation.

7) What explanation is feasible to explain the correlation differences in > or < 6 years? Can the authors clarify scores >6 years or < 6 years from the same patient?

8) Discussion: neurofilament should be referenced (i.e. Darras BT, et al., Ann Clin Transl Neurol. 2019 Apr 17;6(5):932-944. doi: 10.1002/acn3.779. PMID: 31139691; PMCID: PMC6530526.

9) The authors provide additional evidence that muscle is involved in SMA but as they said in Discussion, there is some heterogeneity in the samples. The statement "Taken together, these data suggest that systemic therapeutic approaches increasing SMN levels also in skeletal muscle tissues may provide additional benefits to SMA patients" remains speculative according to what is discussed below this statement about miR-324-5p, miR-451a and miR-181a-5p. I would try to elaborate a statement after the discussion of the role in muscle of the miRNAs detected with a more realistic conclusion.

10) "More information is (are) available".

11) Please clarify some aspects of the multivariate analysis performed. Selection of variables that are introduced in the multivariate analysis can be done by a full model (the more variables you enter, the better R2 you will have…) or selection of only those variables with significant value in the bivariate linear regression (one by one). The authors state: "Continuous variables were compared by linear correlations. Multivariate analysis was used to correlate severity and molecular parameters (miRNA levels, SMN2 transcripts, SMN2 copy number)". Which variables were significant if compared one by one?

12) Please clarify in Table 1 the meaning of P in the column of absolute qPCR.

13) The statement "The inclusion of SMA-miRs, age and SMN2-fl levels has almost doubled the accuracy of the severity prediction of SMN2 copy number alone, from about 43% to 82%." should be better discussed. Data of calculations on SMN2 alone (according to the authors apparently is 43%) should be provided because is confusing. Looking at the data set Table, in type III there are 13 with 4 copies (54%), 9 with 3 copies (37%) and 2 with 2 copies (8%). In a total of 18 type II patients the score was calculated and 15 (83%) had 3 SMN2 copies as expected in these type of patients. Adding the 4 type I patients the total number of patients with calculated score is 46 (just serum was available in 51 but the score was performed in 46, please mention in the text).

14) I found 7/24 (30%) cases with type III disease with an insufficient SMA-score (less than 2.5 assuming in type II disease according to the score). It is possible to discuss these results?

15) As the authors mentioned in the last paragraphs, data in newborns (control and SMA) and trajectories with longitudinal data and evolution of the miRNAS should be investigated to ascribe a potential role as a biomarker and to validate the utility of the score.

16) Even though the variants rs121909192 and rs1454173648 were discarded, a sentence may be included mentioning that investigation of the entire sequence of the SMN2 of patients (perhaps with a reference) may provide additional information about prediction of severity and would be a next step to help improving accuracy of prediction.

https://doi.org/10.7554/eLife.68054.sa1

Author response

Essential revisions:

General comments:

1. Data that is listed as 'available on request' or 'data not shown', should be provided in a Supplementary file either with the journal or a 3rd party repository (figshare or similar). In this age of almost unlimited file space, all data reported should be directly presented to the reader.

Data previously indicated as “not shown” have been included in the manuscript, mainly as supplementary information (see supplementary figures 1-6 and tables).

2. Figures should be redone with open boxes so that all of the data points can be easily seen.

As requested, figures have been re-drawn. In order to better appreciate the whole distribution of miRNA levels in patients and controls, we have changed the Y-axis scale of figure 3 from linear to exponential. The same approach has been used for the novel supplementary figures 2 and 3.

3. This Italian cohort of authors should be commended for their preparation of the manuscript in English. However, there are many sections (such as paragraph 4 of the introduction) that would be greatly improved by a careful proofread by a native English speaker.

We have extensively revised the manuscript both for the form and possible typos.

4. One abbreviation of microRNA, either miR or miRNA, should be used throughout the paper.

As requested, we opted for miR as a unique abbreviation for miRNA.

5. Background information on what miRNA are and their biological affect would be useful.

As requested, we added some sentences and references regarding this matter.

6. It is unclear what 51 serum/RNA/DNA samples means. Does it simply mean the authors collected 51 samples, then isolated RNA and DNA from it? Or were samples received with both RNA and DNA already isolated?

We have specified the number of each specimen available in the Method section.

7. Exact p values should be listed uniformly throughout the paper (some figures have asterisks or section have p<0.05).

We have uniformly listed the p values throughout the paper.

Specific Comments

Abstract

8. Under results, add that the SMA-score improved prediction to > 80% for those under age 6.

9. Under Discussion – it seems premature to speculate that the role of the differentially altered miRs was in programs of satellite cell differentiation. Keep in mind that the muscle biopsies sampled primarily muscle fibers and that satellite cells in those biopsies were quiescent, rather than the active proliferating form they take as myoblasts. Is it not also possible that these miRs target other key genes in mature myofibers? In addition, the paper really did not delve into role(s) of SMA in muscle per se, and did not try to link SMA1/2 levels to expression of the miRs identified. Thus, to me at least, any comment about function of SMA in muscle seems inappropriate for the abstract for this paper.

The abstract has been almost completely rephrased.

Introduction

10. The authors do not describe what SMN1/2 encodes, what its biological function is, and why genetic disruption of its expression is so detrimental to muscle function. A brief paragraph providing some general background on SMN would be appreciated for readers not familiar with SMA. Additional notes of onset, severity and phenotype differences between SMA I-III would also be useful to provide some additional background for the disease.

This point has been addressed in the Introduction section. We agree with the reviewer that further information on SMN function in muscle would be appreciable but, unfortunately, most detrimental mechanisms induced by SMN reduction are unknown, in particular in skeletal muscle but also in motor neurons.

11. Paragraph #2 of the introduction should be removed or worked into other portions of the paper.

This point has been addressed in the Introduction section.

Methods

12. If possible, a more clear definition of what the authors mean by 'first stages' using, for example, clinical staging criteria, would improve the manuscript.

This point has been addressed, by adding Supplementary table 1. “First stages of disease” were defined as “as onset of the first clinical signs that prompted the diagnostic workflow”. Unfortunately, to the best of our knowledge, no clinical staging criteria are available, besides the usual classification of SMA.

13. What passage numbers were use for cell cultures? How were the primary myoblasts isolated and stored (i.e. was CD56+ screening used)? These are important for people that may wish to try to replicate your study.

Passages of cell cultures were between 5 and 15. For isolation of myoblasts, we used CD56 selection, as indicated in the ref Zanotti et al., 2007. These aspects were specified in the Methods.

14. Please list the exact local ethics committee (hospital or university IRB or equivalent) and approval number.

This information has been specified as requested in the Methods section.

15. Under 'Samples', add a reason that those controls were selected for the studies. Under discussion, review whether selection of these controls could be a confound.

This point has been addressed in the Methods section.

16. Please provide a method for isolating/freezing myoblasts or cite your prior work if you have published these techniques.

As reported above (point 13.) we have included our ref Zanotti et al., where experimental details were specified.

17. Under 'Patients', were the authors involved in assigning SMA types blinded to the patient's diagnosis?

Data on the clinical performance of patients, evaluated by functional scale scores and motor ability, are sufficient to assign a decimal SMA subtype, even if the gross classification (type I-III) is unknown. In this sense, the utility of molecular tests is very limited, due to the finding of the same genotype at the SMN locus, in patients with markedly different phenotypes.

18. Under 'Whole miRome Sequencing', what filters were used (e.g., exclude miRs under 100 counts, etc). Are the raw read data available? Were they deposited in NIH GEO or other databases?

The pipeline analysis has been specified in Methods. Raw data of miRSeq have been deposited at NCBI-SRA database (BioProject PRJNA748014). The significantly up- and down-regulated microRNAs were selected at False Discovery Rate (FDR) <0.05

19. Under 'Statistical Tests' please clarify if an FDR threshold was applied for non-parametric tests (Mann-Whitney U-test). Also, please expand on how the multivariate analysis was performed in sufficient detail that others could use the sama analysis technique.

Due to the analytical workflow we used for absolute qPCR, i.e. a progressive increase in the number of samples tested, in case of first tier significant results, we established an α-value at 0.05, without applying an FDR threshold. The rationale behind this strategy was that the number of miRNAs that were selected for the second tier absolute qPCR was small, thus minimizing the risk of false positive results, due to the number of variables. This aspect has been detailed in Methods section.

Results and Discussion

20. p values reported for miR-324 and 451 in Results section for Figure 3 do not match those listed on the figures for Figure 3 (.02 and.004 in Results section,.05 and.001 on the figures).

Figure 3 has been replaced with a new version, as requested. Indeed, the proper p-values are those indicated in the text. Those indicated in the figures were originated in a previous version of the graphs. We apologize.

21. A significant weakness of the manuscript is that the experiments with injection of miRs into ventricles of the brain did not include controls to confirm that the miRs influenced mRNA or protein levels in the central nervous system. The most logical place to look for this disease of lower motor neurons would be spinal cord.

We agree with the reviewer that the putative mode of action of the SMA-miRs/antagomiRs has not been explored at all in our study. These aspects were, in our opinion, out of focus with respect to the main aim of our project that was essentially aimed at identifying SMN-independent biomarkers. For similar reasons, we did not check whether miR level modulation could affect SMN levels in CNS or namely in spinal cord. Additionally, two lines of evidences were in support of this strategy: (1) as shown in the present study, SMA-miRs do not modulate SMN levels in human cells of neuronal origin; (2) since the survival of treated mice is not affected by antagomiRs, we think unlikely a direct effect on SMN levels that, based on preclinical and clinical data, would have led to a positive effect on mice phenotype.

Also unclear is why the miR were delivered to the central nervous system rather than skeletal muscle where they arose. Conceptually, any effects of the miRs identified in muscle and the circulation would be most likely to be evident in muscle or, possibly, at the nerve terminal; alternatively, circulating miRs could influence gene expression at any location in the body. I do not know if miRs cross the blood brain barrier.

With our preliminary experiments (but more detailed studies on miR/antagomiR system are ongoing) we aimed at evaluating a specific hypothesis, that is the retrograde effect of SMA-miRs on CNS. We agree with the reviewer that it is conceivable that, based on the finding of secreted miRs in serum, these molecules are likely exert a systemic effect: to reduce possible confounding variables, we decided to restrict miR modulation to a specific target tissue, namely the CNS, rather than evaluating the loco-regional effect in skeletal muscle. However, several studies report the ability of extra-cellular vesicles to cross the blood brain barrier and to deliver their content to the CNS (see for example Izco et al., Neuroscientist 2021 Feb 3; 1073858421990001. doi: 10.1177/1073858421990001).

22. It is quite difficult to figure out whether muscle or serum levels of miRs were used in the multivariable regression. The text of the results and discussion should be revised accordingly.

This point has been addressed as requested

23. Relative abundance is one key parameter to consider when evaluating biological significance of miR expression. A comment regarding which of the differentially expressed miRs are highly expressed would be appropriate. The myoMirs are generally highly expressed in muscle and would be expected to be included among the highly expressed miRs in this dataset.

To address this point of the reviewer, we have included in Supplementary Table 4 also logCPM of miRNA levels in muscle samples, as obtained by NGS. Since miRNA levels in serum were determined by absolute qPCR, the relative abundance of the different miRs is reported as no. of molecules/μl of serum in figures and tables. As hypothesized by the reviewer, myo-miRs were abundantly expressed in muscle samples, at significantly higher levels in patients. Intriguingly, these miRNAs were not differentially expressed in serum, suggesting that, differently from the SMA-miRs, myo-miRs are not actively secreted.

Reviewer #1 (Recommendations for the authors):

General comments:

The authors refer to multiple sets of data that has not been included in the manuscript. These data should be added as supplemental material.

The following materials have been included as supplementary:

Supplementary file 1: demographic and clinical characteristics of patients who underwent to biopsy

Supplementary file 2: molecular and clinical details of the single patients

Supplementary file 4: detailed list of differential miRs, including the logCPM

Additionally, the raw data of NGS runs have been uploaded in a public repository (NCBI-SRA database, BioProject PRJNA748014)

Figures should be redone with open boxes so that all of the data points can be easily seen.

This point has been addressed as requested. In order to magnify the distribution of miR levels in our cohort, we have replaced the linear scale with an exponential.

The manuscript would benefit from careful editing for English, typographical errors and punctuation.

This point has been addressed as requested.

Overall, while the data from the manuscript are very interesting, it is a bit unfocused and includes experiments that address several different questions such as which miRs are differentially expressed in muscle, whether these changes are observed in serum, whether changing expression of these miRs in the nervous system influences disease progression, and whether circulating miRs could be a clinically useful biomarker for disease severity.

Specific Comments

Abstract

Under results, add that the SMA-score improved prediction to > 80% for those under age 6.

Under Discussion – it seems premature to speculate that the role of the differentially altered miRs was in programs of satellite cell differentiation. Keep in mind that the muscle biopsies sampled primarily muscle fibers and that satellite cells in those biopsies were quiescent, rather than the active proliferating form they take as myoblasts. Is it not also possible that these miRs target other key genes in mature myofibers? In addition, the paper really did not delve into role(s) of SMA in muscle per se, and did not try to link SMA1/2 levels to expression of the miRs identified. Thus, to me at least, any comment about function of SMA in muscle seems inappropriate for the abstract for this paper.

See the response to specific comments #9.

Methods

Under 'Samples', add a reason that those controls were selected for the studies. Under discussion, review whether selection of these controls could be a confound.

This point has been addressed as requested in the Methods section.

Please provide a method for isolating/freezing myoblasts or cite your prior work if you have published these techniques.

See response to item 16 in the Specific comments.

Under 'Patients', were the authors involved in assigning SMA types blinded to the patient's diagnosis?

See response to item 17 in the Specific comments.

Under 'Whole miRome Sequencing', what filters were used (e.g., exclude miRs under 100 counts, etc). Are the raw read data available? Were they deposited in NIH GEO or other databases?

Under 'Statistical Tests' please clarify if an FDR threshold was applied for non-parametric tests (Mann-Whitney U-test).

See response to item 19 in the Specific comments.

Results and Discussion

A significant weakness of the manuscript is that the experiments with injection of miRs into ventricles of the brain did not include controls to confirm that the miRs influenced mRNA or protein levels in the central nervous system. The most logical place to look for this disease of lower motor neurons would be spinal cord. Also unclear is why the miR were delivered to the central nervous system rather than skeletal muscle where they arose. Conceptually, any effects of the miRs identified in muscle and the circulation would be most likely to be evident in muscle or, possibly, at the nerve terminal; alternatively, circulating miRs could influence gene expression at any location in the body. I do not know if miRs cross the blood brain barrier.

See response to item 21 in the Specific comments.

It is quite difficult to figure out whether muscle or serum levels of miRs were used in the multivariable regression. The text of the results and discussion should be revised accordingly.

See response to item 22 in the Specific comments.

Relative abundance is one key parameter to consider when evaluating biological significance of miR expression. A comment regarding which of the differentially expressed miRs are highly expressed would be appropriate. The myoMirs are generally highly expressed in muscle and would be expected to be included among the highly expressed miRs in this dataset.

See response to item 23 in the Specific comments.

Reviewer #2 (Recommendations for the authors):

Abiusi et al., describe a novel pipeline for predicting severity of infantile spinal muscular atrophy (SMA). They use muscle biopsies and primary cell cultures as well as serum from individuals with SMA. For additional studies, they use an SMA mouse model with miRNA anti-sense injections. They conclude their new model was able to improve prediction of phenotypic severity when related to patient subtype and stratified by age. While the paper has a solid base, I think there are areas that need to be clarified, strengthened or carefully considered. I thank the authors for doing this work and putting together the manuscript. I hope the following will help solidify and improve aspects of the paper.

Strengths

– SMA is a deleterious genetic disease and the ability to improve prediction of severity would be an important clinical outcome.

– Use of miRNA sequencing to find biomarkers is a solid experimental approach.

– Coupling biomarkers with genetic outcomes is a logical predictive strategy.

– Follow-up testing on previously proposed biomarkers.

Weaknesses

– There is a disparity in FDRs used for various miRNA analyses. Reasons for differences in the accepted FDR should be explained and also may have affected downstream analyses. FDRs listed for miRNome was said to be selected at p<0.01 in the methods, with muscle having 99, myoblasts having 20 and myotubes having 19 DE miRNA. However, using their stated FDR of p<0.01, two of their selected predictors, miR-181 and miR-324 do not meet that threshold for being differentially expressed. Then there is a switch to accepting an FDR of p<0.05. Uniformity of data analyses is crucial for biomarker studies and this is a major oversight.

We are very grateful to the reviewer for noticing our mistake, due to a typo. The FDR for all analyses was selected <0.05. The text has been modified accordingly in the Methods section.

– For the first tier round of qPCR follow up, how and why where those individuals selected out of the complete cohort? Having 1 SMA1 and 9 SMA II seems like a very oddly weighted group breakdown.

Thank you very much for the constructive comment, we specified this information in the Results. At that stage, due to the lack of preliminary data, we opted for the more homogenous sub-cohort of patients, in terms of age and severity, to reduce potential biases in miR quantification. The type I patient was the only available at the time.

– How was type I error limited for the absolute qPCR? FDRs of relative qPCR were stated to be an FDR but nothing is mentioned for absolute qPCR. This is important considering 50 assays were completed for these outcomes.

See response to Specific comments #19.

– A table showing all the outcomes of the relative and absolute qPCR would provide a great deal of information and transparency. For example, the miRNA selected as 'predictors' for SMA were upregulated, but biomarkers can also be down-regulated.

As requested we have included a supplementary table (#5) reporting the results of the single miRNAs in relative and/or absolute qPCR.

– The link of SMN being depleted in muscle having primary effects is intriguing, but should be interpreted with caution. The importance of having a tonic motor connection with a healthy motor neuron cannot be underestimated. Changes in differentiation factors in proliferating myoblasts and forming myotubes related to myomiR are often seen in other pathological conditions that result from denervation (such as Parkinson's, age-related denervation and ALS). Additionally, only these factors were considered and not any phenotypic measurements of differentiating myotubes. The data are intriguing, but directed and focused studies are needed to investigate how muscle may or may not be regulated by global SMN1 reductions.

Thank you very much for the comment. We fully agree that our data alone do not allow drawing any conclusion on the pathogenic role of skeletal muscle in SMA, but this was not the primary aim of the study that has been designed for biomarker discovery. However, besides data obtained in murine models or those in human whole muscle samples, some in vitro studies on human muscle cell cultures have demonstrated primitive alterations of patients’ satellite cells, including the inability of SMA myotubes to sustain motor neuron innervation, when co-cultured with wild type motor neurons (see for example Braun et al., 1995; Guettier-Sigrist et al., 2002).

– There is no description of when the biopsies of the SMA patients were taken. As they say they were taken from the quadriceps during the first stages of the disease, this could be quite variable depending on fiber type of the muscle group the biopsy was taken from, time diagnosis was confirmed and the age at the time of biopsy. Clarification of these points, as well as for the primary tissue cultures, is important.

See response to item 12 in the Specific comments.

– It is not listed where the serum from individuals with SMA came from. Please explain if these were obtained from a repository similar to the muscle tissues or if separate process was conducted.

We agree with the reviewer. Patients were enrolled ad hoc in the participating neuromuscular centers. These patients are or were in routine clinical follow-up. This aspect has been specified in the Methods section.

– The disparity in group sizes from type I to type II and III is unfortunate. It is understood the difference in disease disparity causes this, but it is a limitation in the search for prognostic biomarkers to have such unbalanced group differences.

We fully agree with the reviewer. However, it is unlikely to succeed in balancing the sub-populations, in such wide phenotypic spectrum of disease, in terms of age of onset, severity and relative abundance of the different forms. Additionally, the total number of type I patients in our study has been limited also by the difficulties in sampling adequate amounts of blood from such young and/or hypotonic patients. Indeed, we had to exclude some other patients due to the lack of serum samples.

– Assignment of patients to the SMA subgroup is vague and likely difficult to understand for non-clinicians. Explicit descriptions of how each individual used the clinical data available to stratify patients would be appreciated.

In the Introduction, we specified some details on how SMA patients are classified, based on the clinical severity. Since this classification has been published several years ago, we preferred not to detail the single items used for scoring patients. The refs have been specified in Introduction and Methods section.

– The authors do not describe what SMN1/2 encodes, what its biological function is, and why genetic disruption of its expression is so detrimental to muscle function. A brief paragraph providing some general background on SMN would be appreciated for readers not familiar with SMA. Additional notes of onset, severity and phenotype differences between SMA I-III would also be useful to provide some additional background for the disease.

See response to Specific comments #10. As requested, we added some details on SMN.

– Data that is listed as 'available on request' or 'data not shown', should be provided in a Supplementary file either with the journal or 3rd party repository (figshare or similar). In this age of almost unlimited file space, all data reported should be directly presented to the reader.

See response to item 1 in the General Comments.

– Paragraph #2 of the introduction should be removed or worked into other portions of the paper.

– This Italian cohort of authors should be commended for their preparation of the manuscript in English. However, there are many sections (such as paragraph 4 of the introduction) that would be greatly improved by a careful proofread by a native English speaker.

See response to item 3 in the General Comments.

– One abbreviation of microRNA, either miR or miRNA, should be used throughout the paper.

See response to item 4 in the General Comments.

– Background information on what miRNA are and their biological affect would be useful.

See response to item 5 in the General Comments.

– It is unclear what 51 serum/RNA/DNA samples means. Does it simply mean the authors collected 51 samples, then isolated RNA and DNA from it? Or were samples received with both RNA and DNA already isolated?

See response to item 6 in the General Comments.

– Exact p values should be listed uniformly throughout the paper (some figures have asterisks or section have p<0.05).

See response to item 7 in the General Comments.

– What passage numbers were use for cell cultures? How were the primary myoblasts isolated and stored (i.e. was CD56+ screening used)? These are important for people that may wish to try to replicate your study.

See response to item 13 in the General Comments.

– Please list the exact local ethics committee (hospital or university IRB or equivalent) and approval number.

See response to item 14 in the General Comments.

– p values reported for miR-324 and 451 in Results section for Figure 3 do not match those listed on the figures for Figure 3 (.02 and.004 in Results section,.05 and.001 on the figures).

See response to item 20 in the General Comments.

–Related to figures, improved opacity to allow the reader to see the individual data points would be appreciated.

See response to item 2 in the General Comments.

Reviewer #3 (Recommendations for the authors):

1) Title. I would edit the title to reflect a more realistic scenario of this investigation that is the discovering and validation of miRNAs. The application of the SMA-score is in my opinion preliminary (see below further comments).

The title has been rephrased as requested.

2) Abstract. Edits are also necessary in the abstract. i.e. there is incomplete information given that a part of these studies were performed in mice models; wording: "evaluate its controversial pathogenic role" is not addressed in this work and the conclusion that "may provide additional benefit to SMA patients" is not new; also "the SMA-score could be crucial for the prognostic assessment of pre-symptomatic patients" is rather speculative given the lack of data in pre-symptomatic patients regarding the candidates miRNAs and the score application. In summary, the abstract should be more focused in the issues of the study.

The abstract has been rephrased as requested.

3) Methods. The authors should clarify the meaning of "first stages", does mean that all the samples were taken at the beginning of manifestations? Perhaps a Table (could be also supplementary) can be included with a summary of the samples studied with additional information of first stages.

As requested, a supplementary table has been added (#1). See response to item # 12 of the Specific comments.

4) Discovering and data of specific miRNAs is interesting and the analysis is adequate. The lack of longitudinal analysis or trajectories to ascribe them a value as dynamic biomarkers should be commented.

See response to item # 8 of the minor comments.

5) The cohort of type I patients is small in my opinion to establish a correlation (only four patients). But it could be interesting to try a correlation between the values of type II versus type III cases to determine if patients with three copies acquire (type III) or not (type II) the ability for autonomous walking.

We agree with the reviewer that the type I cohort is very small. However, due to the difficult sampling and to the amount of biological specimen required for the different analyses reported in this study, we had to exclude 4 additional patients due to the lack of serum or RNA samples. Additionally, the spreading of experimental treatments has complicated the enrollment of further naïve patients.

We re-evaluated with the referral clinicians the whole cohort and concluded that one of the patients has been erroneously attributed to type I while being type II. This aspect has been emended in table and figures, and the equations have been concordantly corrected. These modifications have led to a further improvement of the correlation between the SMA-score and the SMA decimal classification.

As suggested by the reviewer, we compared SMA-miRs’ levels in type II and III patients with 3 SMN2 copies (Figure 3 – figure supplement 3A). The difference was not significant for any miR or the sum. However, the age at sampling was significantly higher in type III (Figure 3 – figure supplement 3B). In our opinion it would be more appropriate to compare age-matched groups to draw any conclusion.

6) There are four patients in the list of type II and III disease with two SMN2 copies. In absence of positive modifier variants (rs121909192 and rs1454173648 also known as NM_017411.3:c.859G>C and NM_017411.3:c.835-44A>G)(both nomenclatures can be included for clarity in the paper) these "outliers" may influence the final score? i.e. PT11 2.70 vs. PT 38 2.12 >6 years (according to data set). Values should be with period not commas, the same applies to the equation.

We have included an additional supplementary table (#2) with individual data for each patient, which has integrated and replaced the previous file “patient_database”. Of the mentioned type II patients, one was twin-sister of a type I male that were previously reported [Pane et al., Neuromuscular Disorders 27 (2017) 890–893]. The 2 other patients were confirmed, both in terms of clinical severity and SMN2 copy number/absence of splicing variants. Regarding the type III patient, due to homonymy, the number of SMN2 copies was erroneously attributed. Indeed, DNA sample of this patient was not available. These data have been amended in figures, table, and equations.

7) What explanation is feasible to explain the correlation differences in > or < 6 years? Can the authors clarify scores >6 years or < 6 years from the same patient?

In supplementary table 2, we have indicated the SMA score obtained with both equations for patients below 6 years. Indeed, the correlation between the scores is very high, since the correlation coefficient is 0.90 and R2 is 80.31 (p<10-5, n=21, supplementary Figure 5D). These aspects have been pointed in the manuscript.

8) Discussion: neurofilament should be referenced (i.e. Darras BT, et al., Ann Clin Transl Neurol. 2019 Apr 17;6(5):932-944. doi: 10.1002/acn3.779. PMID: 31139691; PMCID: PMC6530526.

See response to item #7 in minor comments.

9) The authors provide additional evidence that muscle is involved in SMA but as they said in Discussion, there is some heterogeneity in the samples. The statement "Taken together, these data suggest that systemic therapeutic approaches increasing SMN levels also in skeletal muscle tissues may provide additional benefits to SMA patients" remains speculative according to what is discussed below this statement about miR-324-5p, miR-451a and miR-181a-5p. I would try to elaborate a statement after the discussion of the role in muscle of the miRNAs detected with a more realistic conclusion.

Thank you very much for the comment. We have rephrased the Discussion accordingly.

10) "More information is (are) available".

Thank you very much for the comment. We have corrected the typo.

11) Please clarify some aspects of the multivariate analysis performed. Selection of variables that are introduced in the multivariate analysis can be done by a full model (the more variables you enter, the better R2 you will have…) or selection of only those variables with significant value in the bivariate linear regression (one by one). The authors state: "Continuous variables were compared by linear correlations. Multivariate analysis was used to correlate severity and molecular parameters (miRNA levels, SMN2 transcripts, SMN2 copy number)". Which variables were significant if compared one by one?

Indeed, even if it has not been reported, in the present study, besides SMN2-fl levels, we have also evaluated SMN-del7 and SMN-total transcripts. The inclusion of these variables in the multivariate analyses led to a reduction in both significance and strength of association, since p-value increased and R2 decreased. Thus, we do not agree that by simply adding variables the R2 value increases. Similarly, the single miRNA performed worse than the miR-sum. As reported in the Results, the strongest contribution to the model was not surprisingly provided by SMN2 copy number and SMN2-fl levels.

Besides the contribution of the single variables in the definition of the equation, in our opinion, the correlation of two completely independent variables (decimal SMA classification and SMA-score) is unequivocally strong, up to levels that have never been raised so far in SMA biomarker research. Nonetheless, we fully agree that these data need to be validated in independent cohort, as well as, longitudinal evaluation of SMA-miR levels is required. We remind that we have shown in our previous studies that longitudinal levels of SMN2-fl transcripts are very stable over more than one year (Tiziano et al., 2019). This aspect has been pointed in the Discussion.

12) Please clarify in Table 1 the meaning of P in the column of absolute qPCR.

The p-value has been obtained by comparing miR levels in patients and controls by Mann-Whitney U-test. This aspect has been specified in the table legend and in Methods.

13) The statement "The inclusion of SMA-miRs, age and SMN2-fl levels has almost doubled the accuracy of the severity prediction of SMN2 copy number alone, from about 43% to 82%." should be better discussed. Data of calculations on SMN2 alone (according to the authors apparently is 43%) should be provided because is confusing.

The correlation between SMA Type and SMN2 copy number has been evaluated by a linear regression model (R2=52.45%, p<10-5, n=41; Figure 5 – figure supplement 1A). This has been specified in the Results and in the statistical analysis section of the Methods. Additionally, we have related SMN2 copies with the decimal SMA classification that has raised the R2 up to 67.00% (Figure 5 – figure supplement 1B). While SMN2 copy number alone provides sufficient predictability in the case of patients with 2 or 4 copies, the relevance in patients with 3 copies is very limited. Indeed, in our cohort, the SMA decimal classification ranges in these patients ranges from 1.9 to 3b. The correlation of the SMA score with the decimal classification in this subgroup of patients (n=21), is significant with a correlation coefficient of 0.55 (p=0.008, R2=30.00, Figure 5 – figure supplement 1C).

Looking at the data set Table, in type III there are 13 with 4 copies (54%), 9 with 3 copies (37%) and 2 with 2 copies (8%). In a total of 18 type II patients the score was calculated and 15 (83%) had 3 SMN2 copies as expected in these type of patients. Adding the 4 type I patients the total number of patients with calculated score is 46 (just serum was available in 51 but the score was performed in 46, please mention in the text).

The different biological samples were not available for all patients. As requested, this has been specified in the Methods. Additionally, for each statistical test, the number of samples available has been detailed.

14) I found 7/24 (30%) cases with type III disease with an insufficient SMA-score (less than 2.5 assuming in type II disease according to the score). It is possible to discuss these results?

As reported in supplementary file 2, in the updated version of the equation, the number of type III with low SMA-score is of 1 patient (#26, SMA-score=2.37). This is likely due to the SMN2-fl levels, which appear to be quite low.

15) As the authors mentioned in the last paragraphs, data in newborns (control and SMA) and trajectories with longitudinal data and evolution of the miRNAS should be investigated to ascribe a potential role as a biomarker and to validate the utility of the score.

We agree with the reviewer. Longitudinal data are mandatory for any biomarker study.

16) Even though the variants rs121909192 and rs1454173648 were discarded, a sentence may be included mentioning that investigation of the entire sequence of the SMN2 of patients (perhaps with a reference) may provide additional information about prediction of severity and would be a next step to help improving accuracy of prediction.

As requested, a sentence and a reference have been added.

https://doi.org/10.7554/eLife.68054.sa2

Article and author information

Author details

  1. Emanuela Abiusi

    Department of Life Sciences and Public Health, Section of Genomic Medicine, Università cattolica del Sacro Cuore, Roma, Italy
    Contribution
    Data curation, Investigation, Writing - original draft
    Contributed equally with
    Paola Infante
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9028-012X
  2. Paola Infante

    Center For Life Nano Science@Sapienza, Istituto Italiano di Tecnologia; Department of Molecular Medicine, Università degli Studi di Roma "La Sapienza", Roma, Italy, Roma, Italy
    Contribution
    Investigation, Methodology, Data curation
    Contributed equally with
    Emanuela Abiusi
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0682-3916
  3. Cinzia Cagnoli

    Clinical and Experimental Epileptology Unit, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milano, Italy, Roma, Italy
    Contribution
    Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6863-6687
  4. Ludovica Lospinoso Severini

    Department of Molecular Medicine, Università degli Studi di Roma "La Sapienza", Roma, Italy
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1032-5496
  5. Marika Pane

    1. Pediatric Neurology, Department of Woman and Child Health and Public Health, Fondazione Policlinico Universitario A. Gemelli IRCCS-Università Cattolica del Sacro Cuore, Rome, Italy
    2. Centro Clinico Nemo, Fondazione Policlinico Universitario A. Gemelli IRCCS-Università Cattolica del Sacro Cuore, Roma, Italy
    Contribution
    Investigation, Supervision
    Competing interests
    No competing interests declared
  6. Giorgia Coratti

    Centro Clinico Nemo, Fondazione Policlinico Universitario A. Gemelli IRCCS-Università Cattolica del Sacro Cuore, Roma, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  7. Maria Carmela Pera

    Centro Clinico Nemo, Fondazione Policlinico Universitario A. Gemelli IRCCS-Università Cattolica del Sacro Cuore, Roma, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  8. Adele D'Amico

    Unit of Neuromuscular and Neurodegenerative Disorders, Dept. Neurosciences, Bambino Gesu' Children's Hospital IRCCS, Roma, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  9. Federica Diano

    Department of Life Sciences and Public Health, Section of Genomic Medicine, Università cattolica del Sacro Cuore, Roma, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  10. Agnese Novelli

    Department of Life Sciences and Public Health, Section of Genomic Medicine, Università cattolica del Sacro Cuore, Roma, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  11. Serena Spartano

    Department of Life Sciences and Public Health, Section of Genomic Medicine, Università cattolica del Sacro Cuore, Roma, Italy
    Contribution
    Investigation, Software
    Competing interests
    No competing interests declared
  12. Stefania Fiori

    Department of Life Sciences and Public Health, Section of Genomic Medicine, Università cattolica del Sacro Cuore, Roma, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  13. Giovanni Baranello

    Department of Pediatric Neuroscience, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milano, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  14. Isabella Moroni

    Neuromuscular Diseases and Neuroimmunology Unit, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milan, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  15. Marina Mora

    Neuromuscular Diseases and Neuroimmunology Unit, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milan, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  16. Maria Barbara Pasanisi

    Neuromuscular Diseases and Neuroimmunology Unit, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milan, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  17. Krizia Pocino

    Department of Medical and Surgical Sciences, Fondazione Policlinico Universitario Agostino Gemelli IRCCS, Roma, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  18. Loredana Le Pera

    1. Bioenergetics and Molecular Biotechnologies (IBIOM), CNR-Institute of Biomembranes, Bari, Italy
    2. CNR-Institute of Molecular Biology and Pathology (IBPM), Rome, Italy
    Contribution
    Investigation, Software
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0076-9878
  19. Davide D'Amico

    Amazentis SA, EPFL Innovation Park, Losanne, Switzerland
    Contribution
    Investigation, Software
    Competing interests
    Davide D'Amico is affiliated with Amazentis SA. The author has no financial interests to declare. At the time of study developement, Dr. D'Amico had an academic affiliation
  20. Lorena Travaglini

    Unit of Neuromuscular and Neurodegenerative Disorders, Dept. Neurosciences, Bambino Gesu' Children's Hospital IRCCS, Roma, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  21. Francesco Ria

    1. Department of Translational Medicine and Surgery, Section of General Pathology, Università Cattolica del Sacro Cuore, Roma, Italy
    2. Fondazione Policlinico Universitario A. Gemelli - IRCCS, Rome, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  22. Claudio Bruno

    Center of Translational and Experimental Myology, IRCCS Istituto Giannina Gaslini, Genova, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  23. Denise Locatelli

    Clinical and Experimental Epileptology Unit, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milano, Italy, Roma, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  24. Enrico Silvio Bertini

    Unit of Neuromuscular and Neurodegenerative Disorders, Dept. Neurosciences, Bambino Gesu' Children's Hospital IRCCS, Roma, Italy
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  25. Lucia Ovidia Morandi

    Neuromuscular Diseases and Neuroimmunology Unit, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milan, Italy
    Contribution
    Investigation, Supervision
    Competing interests
    No competing interests declared
  26. Eugenio Mercuri

    1. Pediatric Neurology, Department of Woman and Child Health and Public Health, Fondazione Policlinico Universitario A. Gemelli IRCCS-Università Cattolica del Sacro Cuore, Rome, Italy
    2. Centro Clinico Nemo, Fondazione Policlinico Universitario A. Gemelli IRCCS-Università Cattolica del Sacro Cuore, Roma, Italy
    Contribution
    Investigation, Supervision
    Competing interests
    No competing interests declared
  27. Lucia Di Marcotullio

    1. Department of Molecular Medicine, Università degli Studi di Roma "La Sapienza", Roma, Italy
    2. Laboratory affiliated to Istituto Pasteur Italia-Fondazione Cenci Bolognetti, Department of Molecular Medicine, Sapienza University, Rome, Italy
    Contribution
    Project administration, Supervision, Writing – review and editing, Conceptualization, Data curation, Funding acquisition
    Contributed equally with
    Francesco Danilo Tiziano
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0274-7178
  28. Francesco Danilo Tiziano

    1. Department of Life Sciences and Public Health, Section of Genomic Medicine, Università cattolica del Sacro Cuore, Roma, Italy
    2. Unit of Medical Genetics, Department of Laboratory science and Infectious Diseases, Fondazione Policlinico Universitario IRCCS “A. Gemelli”, Rome, Italy
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Project administration, Supervision, Writing – review and editing
    Contributed equally with
    Lucia Di Marcotullio
    For correspondence
    francescodanilo.tiziano@unicatt.it
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5545-6158

Funding

Fondazione Telethon (GGP12116)

  • Francesco Danilo Tiziano
  • Lucia Di Marcotullio

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We are very grateful to Famiglie SMA and to single patients and families for the continuous support. This study was granted by Fondazione Telethon Italia (#GGP12116) to FDT and LDM. CC was supported by Girotondo/ONLUS. LLS was supported by a FIRC-AIRC fellowship for Italy. LLP was supported by ELIXIR IIB (https://elixir-europe.org/), the Italian Node of ELIXIR – the European Research Infrastructure for Life Science Data.

Ethics

Human subjects: Informed consent was obtained from patients for genetic analyses. The study was approved by the local Ethics Committee.

According to the ARRIVE guidelines, procedures were carried out to minimize discomfort and pain, in compliance with National (D.L. 116 Suppl 40/1992 and D.L. 26/2014) and International guidelines and laws (2010/63/EU Legislation for the protection of animals used for scientific purposes). The experimental protocols were approved by the Ethics Committee of the Fondazione IRCCS Istituto Neurologico C. Besta and by the Italian Ministry of Health (protocol numbers: 962/2016-PR and 1039/2020-PR).

Senior Editor

  1. Mone Zaidi, Icahn School of Medicine at Mount Sinai, United States

Reviewing Editor

  1. Christopher Cardozo

Reviewers

  1. Christopher Cardozo
  2. Zachary Graham

Publication history

  1. Received: March 3, 2021
  2. Accepted: September 13, 2021
  3. Accepted Manuscript published: September 20, 2021 (version 1)
  4. Accepted Manuscript updated: September 23, 2021 (version 2)
  5. Version of Record published: October 1, 2021 (version 3)

Copyright

© 2021, Abiusi et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 359
    Page views
  • 84
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Evolutionary Biology
    2. Genetics and Genomics
    Gabriela Santos-Rodriguez et al.
    Research Article Updated

    Many primate genes produce circular RNAs (circRNAs). However, the extent of circRNA conservation between closely related species remains unclear. By comparing tissue-specific transcriptomes across over 70 million years of primate evolution, we identify that within 3 million years circRNA expression profiles diverged such that they are more related to species identity than organ type. However, our analysis also revealed a subset of circRNAs with conserved neural expression across tens of millions of years of evolution. By comparing to species-specific circRNAs, we identified that the downstream intron of the conserved circRNAs display a dramatic lengthening during evolution due to the insertion of novel retrotransposons. Our work provides comparative analyses of the mechanisms promoting circRNAs to generate increased transcriptomic complexity in primates.

    1. Evolutionary Biology
    2. Genetics and Genomics
    Franziska Gruhl et al.
    Research Article Updated

    Circular RNAs (circRNAs) are found across eukaryotes and can function in post-transcriptional gene regulation. Their biogenesis through a circle-forming backsplicing reaction is facilitated by reverse-complementary repetitive sequences promoting pre-mRNA folding. Orthologous genes from which circRNAs arise, overall contain more strongly conserved splice sites and exons than other genes, yet it remains unclear to what extent this conservation reflects purifying selection acting on the circRNAs themselves. Our analyses of circRNA repertoires from five species representing three mammalian lineages (marsupials, eutherians: rodents, primates) reveal that surprisingly few circRNAs arise from orthologous exonic loci across all species. Even the circRNAs from orthologous loci are associated with young, recently active and species-specific transposable elements, rather than with common, ancient transposon integration events. These observations suggest that many circRNAs emerged convergently during evolution – as a byproduct of splicing in orthologs prone to transposon insertion. Overall, our findings argue against widespread functional circRNA conservation.