In this 10-month longitudinal study, the HT-NGS transcriptomic analysis performed on adult SMA patients showed several features possibly related to the nusinersen administration that, in our view, deserve some attention.

The first note concerns the estimation of SMN2 copies analyzed early in the T0-T10 interval. Data showed that in patients with some clinical improvement in one or both the disability scales (SMA_01, SMA_02, SMA_03 and SMA_06), the SMN1/2 ratio slightly increased over the time. On the other hand, in the stable SMA_09, the ratio decreased, suggesting that the treatment was not as beneficial as in the other patients, while in SMA_05 (whose prognosis was rapidly fatal) we may hypothesize that the increase in SMN2 expression was not sufficient to counteract the progress of the pathology. Some influencing factors should be considered for explaining the latter finding, above all the age at nusinersen administration (73 years old in SMA_05) that is of course closely connected to the disease duration; and of course, 10 months is a short interval for conclusive statement on treatment responsiveness. Nevertheless, as far as we know, this is the first study reporting the SMN2/SMN1 expression changes during nusinersen in adult SMA. If confirmed in a larger cohort, this data may lead to recommend the estimation of SMN2/SMN1 expression as early indicator for nusinersen efficacy in adult SMA, which may be a valuable support for the clinical practice, considering the administration route of this therapy.

It is also worth mentioning that 38 genes that resulted DE in SMA patients at T0 compared to HC were no longer different at T10. As an example, the upregulated TRADD and JUND were “normalized” at T10, so we may speculate that nusinersen was able to impact, e.g., the IL-17 signaling pathway in which both the genes are involved, which is in line with the potential of this therapy to mitigate the neuroinflammatory features of SMA (Nuzzo et al. 2023a). Another interesting tip comes from the involvement of genes encoding for zinc-finger proteins that are known potential modifiers of SMA (Kannan et al. 2020). Although these data need functional confirmation in the long-term treatment and larger samples, we believe that they add some novel insights in the molecular evaluation of the effect of nusinersen, as well as they suggest that peculiar molecular pathway may drive the lack of its efficacy, i.e., the upregulation of the Integrin-mediated signaling pathway in SMA_05, which may be a negative prognostic signature (Delers et al. 2022).

As the second goal of the investigation, we found several intriguing results that seem to characterize the molecular profile of adult SMA patients compared to age-matched HCs. Among the significant miRNAs, we confirmed the dysregulation of miR-146a(-5p) (Sison et al. 2017), miR-324-5p (Abiusi et al. 2021), and miR-423(-5p) (Zaharieva et al. 2022) in our SMA subjects. Of interest, miR-146a-5p targeted SMN1, the miRNAs resulted significantly overexpressed while the gene was downregulated. To our view, this finding is in line with a recent report discussing the experimental role of astrocyte-produced miR-146a in the motor neuron loss that characterized SMA. In particular, the study revealed a significant increase of miR-146a in SMNΔ7 mouse spinal cord; furthermore, when iPSC-derived motor neurons were treated with synthesized miR-146a, it seemed to induce significant motor loss, whereas this process was blocked by miR-146a inhibitor (Sison et al. 2017). The authors hypothesized that the mechanism upregulating mir-146a may be trough GATA transcription factors, mainly GATA6, which was found highly expressed in SMA mouse and human samples and correlated with the disease severity, or via NOTCH2 impact (Yang et al. 2016). It has been also demonstrated that NFkB interact synergistically with GATA6, and this interplay leads to activation of miR-146a (Boopathi et al. 2013), which is consistent with the upregulation of NFkB observed in the SMA iPSC-derived astrocytes (Sison et al. 2017).

Neither of these regulations clearly explain the astrocyte malfunctions observed in SMA pathology, nor they showed a direct effect on SMN1. Although without experimental validation, we believe that our extensive analysis added few tips in this view; in fact, the heatmap of the most significant pathways involved by the 13 DE miRNAs (enclosing miR-146a-5p), which targeted validated DE genes in SMA versus HCs (Table 5), confirmed the involvement of both Notch and NFkB signaling (Fig. 3). Most important, the bioinformatics analysis showed that miR-146a significantly targeted SMN1. Functional validations need to follow to verify this pathogenic mechanism.

On the other hand, in our analysis, some genes were targeted by several miRNAs, and their roles emerged as particularly interesting in the scenario of NDs like SMA. Among the others, IFNLR1 was significantly downregulated in SMA as they were targeted by overexpressed miR-151a-3p, miR-125a-5p, miR-328-3p, and miR-3940-3p. Although only the first 2 pairs were validated results by literature—the remaining being predicted by our bioinformatics analysis—this data suggest that IFNLR1 may in fact be relevant in SMA pathogenesis. Of note, literature data reported that IFN-λ receptor (IFNLR1) deficiency was associated to significantly impact of the immune cells’ activation and to the skin and kidneys damage without effects on autoantibody production (Goel et al. 2020), suggesting that an “immune interference” should be considered in SMA. Several recent evidence pointed in fact towards an immune dysregulation in SMA, as SMN seem to have a central role for the healthy development of the lymphoid system (Deguise et al. 2017), as well as specific drugs like nusinersen have been reported to improve some features more likely related to neuroinflammation (Bonanno et al. 2022; Nuzzo et al. 2023b).

Indeed, the functional bioinformatic analysis performed in our study pointed to a significant downregulation of the immune system during SMA (Fig. 4A–C) represented by enriched terms concerning, e.g., MHC class II protein complex and binding, auto/immune diseases (thyroid, type-1 diabetes, graft-versus host), intestinal immune network for IgA production. It is also noteworthy that one of the molecular networks that resulted from the functional analysis (Fig. 5) enclose HLA-DRB1 and HLA-DRB5 that have been confirmed implicated in MS and other autoimmune diseases (Irizar et al. 2012; Agliardi et al. 2023) as well as in AD and other NDs (Hampel et al. 2020). In depth analysis still needs to be done also in this direction, taken for overturned the past assumption that SMA is just a motor neuron disease (Yeo and Darras 2020).

We are aware that the study suffers for some limitations, as the small sample size and the heterogeneity of the recruited patients, whose age at first nusinersen administration ranged from 19 to 73 years old, which leads to a very wide changes in the disease duration and clinical disabilities, with subsequent differences of the individual transcriptomic profile (Schaum et al. 2020; Rutledge et al. 2022). Unfortunately, considering the way of administration and the evidence that nusinersen is rapidly effective mainly in infant SMA (Finkel et al. 2017; De Vivo et al. 2019), only few adult patients agreed to be treated, so we were not able to minimize these variables. On the other hand, although a recent study on children and adolescent SMA patients did find significant DE miRNAs after the first 6 month of nusinersen therapy, which is consistent with the dramatic improvement of clinical disability (Zaharieva et al. 2022), it is reasonable to believe that in our adult cohort a 10-month interval might be a quite short time for uncovering significant transcriptomic changes as for clinical improvement, if any. A longer observation with clinical scales for adult subjects that will be able to provide more significant information (e.g., about changes in daily activities) should help in this purpose (Maggi et al. 2020); however, we must disclose that during this observation another specific drug (Risdiplam) became available as oral therapy authorized in Italy also for the adult SMA. Since most of our recruited subjects expressed their will to shift the therapy, given the complexity of the nusinersen administration, it would be rather difficult to extend this transcriptomic analysis to further time points.

Nevertheless, we believe that combined miRNAs/mRNAs expression analysis may be crucial for a more comprehensive approach also in rare genetic diseases like SMA, in which the causative mutation is known but so far, the whole pathogenic mechanism still needs some clarification.

Looking at the whole picture, an interesting network was in fact depicted from our resulting data (Fig. 5) in which some hub-genes seem to exert crucial roles, as they had central position in the connection between the nodes. As an example, we found that MRLP41, encoding for the protein BMRP with pro-apoptotic activity (by binding Bcl-2) (Malladi et al. 2011), was related to 9 others (TICAM2, ATP5F1D, TRAPPC5, SCAND1, ALKBH7, TMEM160, TMED7-TICAM2, REX1BD, RPL17-C18orf32) with a high bottleneck coefficient suggesting a fundamental role of this gene in the SMA molecular network.

Other information derives from the net composed by 5 TFs (KLF2, CEBPB, CEBPD, JUND, and JUNB), all upregulated and closely implicated in NDs like AD, ALS, SCA, and PD (Evert et al. 2006; Wu et al. 2013; Doxakis 2020; Sun et al. 2022; He et al. 2023). Since depletion of KLF2 (the encoding gene being one of the top genes with highest MCC score measuring its centrality in the net) was reported to cause enhanced apoptosis NGF-mediated (Dutta et al. 2011), while overexpression of the two TFs CEBPB and CEBPD significantly characterized ALS (Sun et al. 2022), and JUND is one of the “allegedly restored” gene by nusinersen in our study, it is reasonable to hypothesize that the interaction between them—possibly mediated by miRNAs like miR-532-3p—could be critically involved also in SMA.

Investigation on larger cohorts together with functional validations should be performed to possibly confirm the role of these novel molecular hotspots toward addressing more therapeutic efforts.