SMN Circular RNAs as Potential Biomarkers for the Therapeutic Response to Nusinersen in Spinal Muscular Atrophy Patients

Background/State of the art SMA represents the most frequent genetic cause of death in infancy with an incidence of ~1 in 10,000 live births . It is an autosomal recessive, neurodegenerative disorder caused by homozygous deletion or point mutation in the SMN1 gene, which encodes for SMN protein. Reduced SMN levels cause loss of motor neurons in the spinal cord and lead to progressive muscle weakness and atrophy in SMA patients. Notably, humans retain the almost identical SMN2 gene, as the complete absence of any form of SMN is embryonically lethal. Nevertheless, although SMN2 encodes a virtually identical protein, its expression levels are not sufficient to fully restore SMN activity because a translationally silent C-to-T transition at position +6 in exon 7 determines skipping of this exon from the mature mRNA. The resulting SMN7 protein displays a shorter C-terminal region and is highly unstable. At the moment Nusinersen is the only FDA-approved drug for treatment of SMA. It is an antisense oligonucleotide (ASO) that corrects SMN2 exon 7 splicing by masking a splicing silencer (ISS-N1) in intron 7 . Although intrathecal administration of Nusinersen ameliorates SMA phenotypes in patients, leading to substantial improvements in motor function, not all SMA patients respond equally well to this therapy. Thus, identification and development of combined and personalized therapies that take into account clinical variability observed between patients of the same genotype is necessary to improve clinical management of SMA. Due to their biochemical properties (stability, accumulation during aging, abundant secretion in body fluids) circular RNAs are now emerging as excellent and promising biomarkers for diagnosis and prognosis of numerous pathologies, such as cancer. Thousands of circRNAs have been discovered in eukaryotic cells and their expression is often regulated in a cell-type and stage-specific manner. Although the majority of circRNAs still lack functional annotations, recent observations revealed that they play some potentially important roles in gene regulation. Circular RNAs (circRNAs) originate from a back-splicing process of the pre-mRNA (i.e., the covalent joining of a downstream splice donor site with an upstream splice acceptor site from the same gene). Since canonical splicing and back-splicing utilize the same pre-mRNA and are both operated by the spliceosome, they likely compete with each other. The main mechanism favoring pre-mRNA circularization is the presence of repetitive sequences in inverted orientation, and in particular inverted Alu repeats.

Alu repeats belong to the primate-specific Short Interspersed Elements (SINE) family of retrotransposons. They are ~300 nucleotides long and account for up to 11% of the annotated human genome. Their presence in coding genes can modulate gene expression at multiple layers, including transcription, splicing, export and translation, thus profoundly impacting on their "host" gene regulation. SMN rank among the top human genes for Alu density, many of which are present in inverted orientation. Strikingly, Alus drive extensive and alternative circularization of the SMN pre-mRNA, thus negatively impacting on the protein-coding potential of the locus. Thus, based on the preliminary results, the propose will be to investigate whether cell-free SMN circRNAs circulating in body fluids of SMA patients represent biomarkers exhibiting diagnostic and prognostic power to predict the clinical response to Nusinersen therapy.

Hypothesis and significance:

The studies proposed in this project aim at investigating whether SMN circRNAs represent valuable biomarkers to predict the efficacy of Nusinersen in SMA patients that exhibit different degrees of disease severity. In particular, our work may help to better stratify the numerous SMA patients which do not respond as expected to treatment and/or for whom the SMN2 gene copy number does not correlate with phenotypic severity. Since extensive production of SMN circular transcripts, instead of the linear transcripts, in some patients could reduce the efficacy of Nusinersen, which selectively acts on the linear transcript, our study may also highlight the need for combined therapy to also limit pre- mRNA circularization in such patients. In the near future, the development of "personalized" therapies could circumvent such "individual" differences in the regulation of SMN expression, thus improving the clinical response of SMA patients to Nusinersen.

Preliminary data:

Whole genome comparison of Alu density highlighted SMN among the top-ranking human genes. Inspection of Alu position and orientation in SMN revealed the presence of several inter-intronic inverted repeat Alu elements (IRAlus) that could mediate circRNA biogenesis. Our bioinformatics analysis (data not shown) highlighted numerous IRAlus in intron 1 and 6, the longest introns in SMN, thus potentially favoring back-splicing of exon 6 with exon 2a. Indeed, RT-PCR using divergent primers in exon 2b and exon 6 followed by sequencing identified a circRNA in both human SMA fibroblasts (GM03813) and HEK293T cells. Since back-splicing events occur at the expense of canonical splicing, was hypothesized a competition between the biogenesis of circular and linear transcripts that could impact on SMN protein expression. In line with the hypothesis, loss of DHX9, a helicase that represses circRNA biogenesis, significantly increased the expression of SMN circRNAs and reduced the expression of the SMN protein. Furthermore, the administration of ASO E8, that masks a cryptic 5\&39; SS in exon 8 that is used for biogenesis of two circRNAs, but not for linear SMN splicing, selectively reduced the target SMN circRNAs and rescued SMN protein expression to ~30% of the efficacy of the ISSN1 ASO. These results indicate that SMN circRNAs competes with the protein-coding potential of the locus.

Specific Aims 1:

Identification of cell-free SMN circRNAs in body fluids of SMA patients as potential biomarkers before and after Nusinersen treatment

Specific Aims 2:

Evaluation of prognostic power of circulating SMN circRNAs as predictors of the clinical response of SMA patients to Nusinersen

Specific Aims 3:

Identification of human intronic polymorphisms present within inverted Alu elements and their impact on the efficacy of Nusinersen treatment

Experimental Design Aim 1:

SMA patients will be enrolled based on genetic documentation of a homozygous deletion or mutation in the SMN1 gene and they will be grouped in SMA type-I (1.1, 1.5 and 1.9), SMA type-II (from 2.3 to 2.9) and SMA type-III (3A and 3B) based on SMN2 gene copy number and onset of clinical symptoms. The plan will be to enroll ten patients/group, however, the exact number of patients for each group will be based on the level of heterogeneity detected.

Testing the expression of cell-free SMN circRNAs isolated from plasma, urine and cerebrospinal fluid of SMA patients and evaluating the correlation between SMN circRNA concentration and disease severity. In parallel, evaluating the expression level of SMN2 linear transcripts in peripheral blood mononucleated cells (PBMCs) for each patient to assess whether linear SMN2 mRNA expression is inversely related to SMN circRNA biogenesis and whether this parameter accounts for SMN expression and disease severity independently of SMN2 copy number.

The evaluation of the concentration of SMN circRNAs varies over time in Nusinersen-treated SMA patients by measuring their concentration in body fluids before and after 6 months of drug administration, according to standard therapeutic regimen in use in our Unit. Briefly, Nusinersen will be administered with intrathecal injection of initial loading doses (12mg/5ml) at days 1, 15, 30, 60, followed by less frequent maintenance doses (every 4 months). This dosing regimen is designed to provide an appropriate level of medication that is maintained between each dose.

In parallel to the in vivo study, testing the effects of Nusinersen on SMN circRNAs biogenesis in pre-clinical models of SMA. Human type-I SMA fibroblasts (GM03813, GM09766 and GM00232; Corriell Institute), which are already available in our laboratory, will be transfected with ISSN1 ASO, that mimics the effect of Nusinersen treatment, and SMN circRNAs expression will be analyzed by qPCR.

Experimental Design Aim 2:

Although approximately half of SMA patients show substantial improvements in motor function following treatment, Nusinersen seems to exert more modest effects in others. In light of this, SMA patients with same eligibility criteria (SMN2 copy number, onset and clinical symptoms) will be stratified into three subgroups based on the concentration of circulating cell-free SMN circRNAs (high, intermediate and low) and their responsiveness to the Nusinersen will be evaluated after the 5th injection of the drug (6 months after the first administration). Evaluation criteria of treatment efficacy consist in HINE (Hammersmith Infant Neurological Examination) and CHOP INTEND (Children Hospital Of Philadelphia Infant Test of Neuromuscular Disorder) for SMA type-I, in neurological examination and HFMSE (Hammersmith Functional Motor Scale-Expandend) for type-II and in neurological examination, HFMSE and 6MWT (6 Minute Walk Test) for type-III patients. This study could also benefit of retrospective analyses already performed by our Unit.

Experimental Design Aim 3:

The inverse correlation between the number of SMN2 copy and disease severity is not observed in all SMA patients (i.e. type-III cases having 2 copies of SMN2 or type-I cases with 3). Moreover, not all SMA patients respond similarly to Nusinersen therapy (i.e. discordant siblings with the same copy number). On the basis of the competition between linear and circular processing of SMN pre-mRNA, evaluating whether a differential SMN circularization index (understood as propensity of SMN pre-mRNA to circulate) could underlie these observed discrepancies. To this aim, it will perform Next Generation Sequencing analysis (Genosplice, Paris, France) of the entire SMN2 gene from genomic DNA of ~100 patients in which the SMN2 gene copy number does not correlate with phenotypic severity or whose response to Nusinersen treatment is below expectation. In particular, it will focus on intronic SMN2 polymorphisms harbored in inverted Alu elements, which could affect the circularization index of SMN2 pre-mRNA.

Furthermore, in order to determine whether the selected intronic polymorphisms impact on SMN circRNAs biogenesis it will perform in vitro mutagenesis experiments. The mutants will be transfected in human SMA fibroblasts or HEK293T and their circularization efficiency will be tested by qPCR for specific back-splice junctions.

Methodologies and statistical analyses:

The RNA from plasma of SMA patients will be extracted by using miRNeasy Serum/Plasma Kit by Qiagen (cat. no. 217184) and after treated with RNase R to degrade linear RNAs while preserving circRNAs. Kit by Qiagen is also recommended for the extraction of RNA from other body fluids, such as urine. The expression of SMN circRNAs will be evaluated by qPCR analyses using primers specific for the SMN circular splice junctions, previously identified (see Preliminary Results). As an internal control, 100 ng total RNA of C. elegans will artificially add into each body fluid specimen before RNA extraction.

The quantitative data will be expressed as the mean ± standard deviation (SD). Two-tailed Student's t test and analysis of variance (ANOVA) followed by Bonferroni's multiple comparison post-test will be performed using Prism 6.0 software (GraphPad Software).

Expected outcomes:

Aim1: determine the expression and concentration of circulating cell-free SMN circRNAs in body fluids of SMA patients and identify the specific body fluid where they are more abundant and on which most of the following experiments will be performed.

Furthermore,evaluating whether there is a direct correlation between the degree of severity of SMA pathology and the concentration of SMN circRNAs.

This study will also shed light on the possible dual role of this molecule on linear and circular SMN2 transcript expression, an important issue to test now considering that the majority of patients are already receiving Nusinersen therapy and in the near future the recruitment of drug-naive patients might be nearly impossible.

Aim2: determine whether SMA patients with a high concentration of cell-free SMN circRNAs have a reduced responsiveness to Nusinersen treatment, as this molecule selectively acts on the SMN2 linear transcription.

Aim3: identify human intronic polymorphisms, within inverted Alu elements, which interfere with SMN circRNAs biogenesis and have an impact on SMA phenotype. Interestingly, part of the identified mutations could underlie the genotype-phenotype mismatch or the failure from some SMA patients to respond to Nusinersen treatment.

Risk analysis, possible problems and solutions:

A critical point of the project is represented by the possibility that SMN circRNAs are difficult to detect in the body fluids of SMA patients. Anyway, a large and growing body of studies show the presence of numerous and abundant circular RNAs in plasma of patients, indicating their role as excellent biomarkers for diagnosis (Li et al., Cell Research, 2015). Thus, it is likely that export of circRNAs is a common and widespread process which also involves SMN circRNAs. Moreover, techniques to detect RNA in liquid biopsies are continuously improving, making the possibility to be unable to detect SMN circRNAs less likely. The remaining part of the project does not present particular risks neither regarding the recruitment of SMA patients, which are already enrolled in the Pera Unit, nor regarding the molecular and cellular techniques that are routinely used by the Pagliarini Unit. Genomic sequencing (Aim 3) will be performed by GenoSplice, a leader in next generation sequencing analyses, which has already collaborated with us in bioinformatics analyses. Nevertheless, in the worst scenario, if our studies demonstrate that SMN circRNAs are not good biomarkers, our project will lead to the identification of SMN2 gene variants associated with the milder and more severe SMA phenotype, providing a useful tool to predict the efficacy of the Nusinersen treatment.

Significance and innovation Due to their biochemical properties, circulating SMN circRNAs could represent promising biomarkers to predict the onset and the progression of disease, as well as the clinical response to Nusinersen treatment. Their concentration within body fluids might represent a valid parameter to stratify SMA patients that exhibit different responses to Nusinersen in spite of similar or identical genetic diagnosis. Furthermore, our studies could highlight the need of robust combinatorial 'SMN-plus' therapies for subset of patients, which may be required to generate long-lasting responses in SMA patients.

Our project also aims at identifying hidden intronic variants of SMN2 which may affect the final amount of SMN2 transcripts produced and determine the degree of severity of the SMA phenotype. Such hidden mutations would allow to more accurately predict the SMA phenotype from genotype and, thus, paving the way for the development of "personalized" therapies that account for "individual" differences.