Thrombospondin-4 as potential cerebrospinal fluid biomarker for therapy response in pediatric spinal muscular atrophy

With a prevalence of 1:10,000, 5q-associated spinal muscular atrophy (hereinafter SMA) is the second most common bi-allelic disease and the most common neurodegenerative disorder in childhood [1]. SMA is characterized by the detrimental reduction of survival of motor neuron (SMN) protein leading predominantly to degeneration of alpha motor neurons (MN) and consequently to progressive muscle weakness and atrophy [1]. SMA is mainly caused by a homozygous deletion on chromosome 5 within the SMN1 gene [2]. The survival motor neuron 2 (SMN2) gene, which is a homologous copy of SMN1, is present in SMA in variable copies. Both genes encode the same protein, but SMN2 lacks exon 7, which leads to an unstable SMNΔ7 protein [3]. Given that SMN2 constantly produces approximately 10% functional SMN protein [4], SMN2 copy number in the majority of patients inversely correlates with disease severity and onset and thus in the past stratifies for pre-symptomatic treatment [5]. Clinically, different SMA-phenotypes are defined based on the age of manifestation and clinical severity in type 0–4. Under now available therapies, the phenotypic spectrum is changing and now classified due to patients motor ability in “non-sitter”, “sitter”, and “walker”. Although MNs are the most affected cells in SMA, their loss may not exclusively depend on the absence of SMN: retrograde signals originating from skeletal muscles and neuromuscular junctions (NMJs) may also represent crucial players in MN vulnerability and contribute to the overall clinical manifestation [6]. Indeed, in mouse models of SMA, one of the earliest events detected is NMJ-defects [7] and clinically, walking performance can identify ambulant SMA patients with NMJ-dysfunction [8]. Moreover, a recent study demonstrated that SMN controls NMJ-integrity through U7 snRNP [9]. In the light of these facts, it is plausible that therapeutic intervention with Pyridostigmine, a quaternary carbamate mainly used to treat NMJ-disease by indirectly increasing the concentration of acetylcholine at the NMJ, showed beneficial outcomes in SMA patients [10, 11]; however, the primary endpoint showed no efficacy. Also, treatment with Salbutamol, a β-adrenergic agonist with an impact on NMJ, showed beneficial outcomes in SMA patients [12, 13], however, this may result from the known impact of Salbutamol on SMN protein stability and turnover [14‐16]. Moreover, there is also an ongoing clinical trial with Amifampridine (3,4-Diaminopyridine) showing improvement in motor assessments [17].

Newly available therapies for SMA, specifically SMN2 splicing modifiers such as Nusinersen or Risdiplam and gene replacement therapy (Onasemnogene Abeparvovec), have dramatically changed survival and overall disease progression [18, 19]. Nusinersen for instance is an antisense oligonucleotide, which corrects the splicing of SMN2 pre-mRNA resulting in an increased production of functional SMN. An early initiation of treatment showed the strongest therapy response in SMA patients, but still the effect of this therapeutic intervention on motor functions varies between patients [20‐24]. To improve the effectiveness and applicability of current therapies targeting SMN abundances, a profound understanding of the different pathophysiological aspects is important, especially with regard to the impact of varying residual endogenous SMN levels. Consequently, based on the definition of appropriate biomarkers, it could be helpful to obtain molecular information on whether therapies also result in more effective synaptic transmission at the NMJ and consequently reduce retrograde MN vulnerability.

Although SMN2 copy number is valid as a biomarker of disease severity, alongside other infrequently used measures, it lacks a definitively reliable predictive value concerning ongoing disease activity. Particularly, it does not serve as an indicator of treatment response [25]. In the last years, neurofilaments (peripheral neurofilament light chain [pNF-L] and peripheral neurofilament heavy chain [pNF-H]) have been considered as biomarkers for SMA [26], but their clinical and analytical relevance is currently also being discussed [27]. Moreover, expression of the muscle-specific miRNA 206 has been linked to disease progression in a murine model of SMA [28]. However, therapeutic biomarkers with known relevance for neuromuscular transmission, for instance, based on known localizations and functional roles at the NMJ are still lacking.

Thrombospondin-4 (TSP4) is a member of the thrombospondin protein family, which represents a group of glycoproteins highly related to the extracellular matrix (ECM) [29]. Overall, thrombospondins participate in diverse biological functions, including cell adhesion and migration, cytoskeleton organization as well as cell–cell interactions and interactions between cells and the underlying matrix components [30‐33]. The thrombospondin family consists of five members (TSP1-5), according to their functional domains the members can be classified into two subgroups: TSP1/2 and TSP3/4/5 [31]. Expression pattern of each thrombospondin differs in developing and adult tissue [31]. TSP4 is expressed in the developing embryo in Xenopus, however, in adult human tissue, it is mainly expressed in the heart and skeletal muscles [34]. Moreover, TSP4 has been identified as a factor contributing to neuropathic pain [35] and promoting neurite outgrowth [36], respectively. Along this line, results of previous studies highlighted an important role of TSP4 at NMJ in mice [36].

The present study aimed to identify new CSF biomarkers for SMA. To this end, in an exploratory proteomics-based analytical approach, protein signatures were compared in CSF samples derived from genetically proven pediatric SMA patients. Doing so, a comparison between patient with and without manifestation of clinical symptoms was carried out. These patients were identified by newborn screening for SMA in Germany [37]. This approach led to the identification of TSP4 discriminating between these two clinically diverging groups of pediatric patients. To validate our proteomic findings and to investigate the potential of TSP4 to serve as a therapy marker, we next analyzed CSF samples derived from SMA patients before and after Nusinersen therapy by enzyme-linked immunosorbent assay (ELISA).

In this study, we utilized proteomic profiling aiming to unveil clinically relevant biomarkers in CSF derived from pediatric SMA patients. Our unbiased data-dependent acquisition approach identified nine proteins with significant diverging abundances including TSP4 as a protein presenting with altered levels between symptomatic and non-symptomatic therapy naïve children. Further confirmatory studies utilizing ELISA as an independent quantification approach show that TSP4 levels in CSF are reduced in therapy naïve pediatric SMA patients but not in adult patients. The robust quantification of TSP4 by ELISA highlights that this protein can also be studied by making use of analytical approaches, which are easily accessible in routine diagnostic settings, an important aspect regarding the need of easy-approachable patient stratification methods. Results of ELISA-based quantification studies moreover highlighted that the significant TSP4 decrease in CSF is not only present by the comparison of pediatric SMA patients with age-matched NDC but also by comparing these patients with pediatric patients suffering from other neurological diseases. Although this suggests a specificity of TSP4 as a CSF biomarker in children suffering from SMA, further studies on extended cohorts are needed to prove this hypothesis. With this requirement, however, it should be kept in mind that the occurrence of other MN diseases in childhood, especially in the first years of life, is extremely rare, making it difficult to investigate large cohorts of diverse genetic entities.

Further studies focusing on the effect of Nusinersen in the therapeutic intervention of pediatric SMA patients revealed an elevation of TSP4 CSF level upon treatment. This finding indicates that TSP4 may hold the potential to serve as a therapy marker for pediatric patients with SMA. Nusinersen is a non-systemic therapy that increases the expression of stable and functional SMN and thus strengthens the MN, which in turn is able to increasingly innervate the muscle again. A crucial process for the reinnervation is the “rebuilding” of synaptic sites [40]. TSP4 is an ECM protein and ECM components are known to be involved in neurite outgrowth, axonal pathfinding, and synapse formation [36]. In adult nervous system, TSP4 is expressed in certain neuronal populations and accumulates at the NMJ and at certain synapse-rich layers [36]. Of note, TSP4 has been localized to the NMJ also in mice and a protective role within the CNS was demonstrated [40, 41].

Thus, increased levels in clinically pre-symptomatic compared to symptomatic patients and TSP4 increase in CSF upon treatment might either reflect stage MN-damage associated with impaired release of secretory proteins based on apoptosis and MN repair/rescue based on Nusinersen treatment or elevated expression and thus secretion in terms of a rescue mechanism activated in the pre-symptomatic patients. However, further functional studies on animal models (therapy naïve versus treated) are needed to address the exact role of TSP4 in the molecular etiology of SMA. Nonetheless, our data combined with knowledge from literature suggest that TSP4—as a matricellular glycoprotein—may be involved not only in proper function of the CNS but also of the NMJs as a known main pathophysiological target of SMA [7, 8] with therapeutic relevance [10, 11]. Our results also support this assumption by the fact that decreased TSP4 levels in CSF elevate after pediatric patients were treated with Nusinersen which is targeting the expression of a stable SMN protein, improving MN function and thus restoring proper neuromuscular transmission. Consequently, we postulate that TSP4 levels in CSF may serve as a promising therapeutic biomarker of pathophysiological relevance in children suffering from SMA and hereby even enable to distinguish between clinically affected and non-affected patients in terms of a therapy marker.

Although, results of our quantification studies clearly indicate the potential of TSP4 to serve as a therapy marker, altered/restored levels in CSF upon treatment did not correlate with clinical outcome measures. Along this line, a biomarker study on 16 adults with SMA type 3 and 4 under Nusinersen treatment over 22 months reported on a significant decrease of pNF-H in CSF. However, this decrease was also not correlating with clinical outcome measures and a similar finding was obtained for Chitotriosidase-1 (CHIT1) levels. In contrast, a decrease of Chitinase-3-like protein 1 (YKL-40) strongly correlated with improvements in the revised upper limb module (RULM) [42]. These findings indicate that therapy markers in SMA do not necessarily correlate with clinical outcome measures even in adult patients in which—in contrast to children (see below “limitations of the study” section)—uniformed standardized motor tests such as RULM can be applied. However, it has to be taken into consideration that SMA belongs to the group of rare diseases and thus that significance in correlations might be limited by the number of patient-derived samples available for these studies.

Further studies on pediatric cases treated with other SMN-targeting drugs (such as Salbutamol [14‐16]) would be needed to draw final conclusions regarding the potential of TSP4 to serve as a generalized therapy marker. Moreover, more detailed biochemical studies are needed to decipher the exact molecular background of TSP4 increase upon restoration of appropriate SMN expression. Based on the functional role of TSP4 at synapses [35] including NMJs [36], its potential to serve as a disease relevant biomarker in SMA—even enabling patient stratification in terms of distinguishing between clinically non-affected and affected patients as well as type 1 therapy responders versus type 2 and 3 responders—is underlined by its pathophysiological meaning. Taking the informative potential of TSP4 only in pediatric but not adult SMA-patients into consideration, one might speculate that this molecular observation is based on a development-dependent expression of the protein.