Discussion
Single cell analysis is a novel technology that can help to answer important questions related to DM1. The first question is whether all skeletal muscle cells are affected equally by the disease. In this effect, it is known that somatic mosaicism affects muscle tissue regarding the size of the CTG expansion (Thornton et al., 1994). In our study, we could demonstrate that the number of RNA foci number is very different among DM1 myoblasts and that the expression of DMPK andMBNL1 differs considerably at the single cell level. So, the myoblasts of the same DM1 muscle tissue are differently affected by the disease. A second important question is whether there are correlations between RNA foci and splicing alterations at the single cell level. In this regard, we found no correlation between RNA foci number and expression of the exon 7 exclusion isoform of MBNL1 . We also found no correlation between RNA foci and DMPK expression at the single cell level. However, we could not study important splicing alterations found in pools of DM1 cells, due to the inability to detect expression of those transcripts at the single cell level. Another important question is whether the average number of RNA foci per cell in a patient correlates with the patient’s muscle affectation. We found correlation between RNA foci mean/median and 6MWD. Thus, patients having myoblasts carrying more foci showed a more severe manifestation of the disease, with subsequent impairment in walking ability.
Foci accumulation is likely to be a dynamic process in patient-derived myoblasts. In this effect, we found heterogeneity in both presence and number of foci in patient-derived myoblasts. RNA foci vary from zero to seven, indicating a dynamic process of RNA foci accumulation being active in all cell lines. A random process of forming and degrading RNA foci has been described in DM1 cellular models (Querido et al., 2011). However, in our patient-derived myoblasts, the proportion of myoblasts with RNA foci that every patient was carrying was very variable. For example, P3 and P5 (i.e. , the less severely affected patients) had overall the highest proportion of myoblasts without foci (71% and 19%, respectively). In contrast, the most severely affected patient in our series, P4, was carrying 2 and 3 RNA foci in 25% and 48% of the myoblasts, respectively.
The RNA foci number was low in DM1 myoblasts (with 90% of the myoblasts showing 0 to 3 RNA foci), as previously described (Gudde et al., 2016). An in-depth analysis is needed to unravel the processes that contribute to the heterogeneity found in RNA foci load and what is the explanation for the fact that a cell carrying the CTG expansion might still have no RNA foci. From the point of view of a spliceopathy disease, getting rid of the RNA foci seems to be key in preserving muscle functionality. In previous studies, one of the main determinants of the RNA foci load was the CTG repeat size (Botta et al., 2008). Although we were unable to find such type of correlation in our study, the patient carrying the shortest CTG expansion had the lowest RNA foci load.
RNA foci load seems to be linked to muscle dysfunction in DM1. We found indeed correlations between 6MWD and foci mean/median. The correlations for RNA foci median and MIRS, biceps MRC and mRS scales were not significant. Many factors could have potentially contributed to our inability to find more correlations with the clinical data, but it must be noted that our study cohort was very small, largely due to the need to perform a muscle biopsy – an invasive and unpleasant procedure – in the patients as well as to research budget constraints. In addition, the heterogeneous manifestation of the disease and therefore the heterogeneous pool of patients must be kept in mind, as well as the fact that the majority of scales we used are categorical (i.e. , MIRS, MRC or mRS) which, in comparison to continuous scales (6MWT) are more challenging to use in correlation studies. In the first design of our study with DM1 experts, we decided to evaluate these scales and we did not consider using a test to measure biceps strength in a more quantitative way. 6MWD was the only physical measure that was related to RNA foci number, but caution must be taken when interpreting these results, since the data was obtained from biceps-derived myoblasts. Additional studies with a larger patients’ cohorts are needed to test the validity of these results and to analyze whether other correlations can be found with the categorical scales.
The splicing analysis was very challenging at single myoblast level since many transcripts could not be detected. The transcripts with higher expression levels in myoblasts were DMPK and the normalMBNL1 isoform. Although we were able to detect and analyze these two transcripts, this was only possible in less than 50% of the studied myoblasts. Regarding INSR and ATP2A1, we were able to detect the transcripts in myoblast cell pools. However, this was not translated to the single cell level, even though a customized qPCR method was used. A possible explanation for our inability to find the transcripts by qPCR is a diminished RNA integrity. At the start of the experiments, we wanted to use the automated system fluidigm, which would decrease the processing time and therefore preserve RNA integrity. However, due to incompatibility of the fluidigm chip with the RNA visualization, we had to rely on sorting cytometry, thereby increasing the time of cell manipulation and microscopic cell visualization, with a higher risk of compromising RNA integrity. One example of the difficulty of using the sorting cytometry is that we could only recover information from 40 to 70 myoblasts of the 120 myoblasts that were originally sorted.
In previous studies analyzing cell pools, RNA foci load correlated with splicing expression patterns (Botta et al., 2008). Yet, we were unable to replicate these results at the single cell level. Overall, we could only find that the MBNL1 normal isoform was more expressed in controls and a wide variety in the expression levels of DMPK was found among the participants (patients and controls). Previous studies analyzing DMPK expression levels in patients are contradictory, and it is not clear if they are affected by the disease (Furling et al., 2001, 2003; Gudde et al., 2016). For a deeper insight into DMPKexpression we tried to analyze DMPK expression by differentiating expanded from normal transcripts with a previously described polymorphism present in exon 10 of the DMPK gene (Korneluk, 1993). Unfortunately, this was not possible, as all of our patients were homozygous for the polymorphism. Other techniques, such as Northern Blot, which have been used for this purpose, were not applicable in this case, as we were working with single cells.
Expansion size can differ between muscle tissue and myoblast primary cultures derived from the muscle in question. As we expected and as other authors have explained before, DM1 cultures are submitted to mitotic drive (Khajavi et al., 2001). In two patients we were able to detect only one CTG expansion size, due to clonal expansion. Our results showed that CTG expansion size differs between samples, meaning that some showed greater difference compared to the muscle tissue than others. Moreover, myoblasts had a tendency towards the expansion of the progenitor allele, which has been also observed previously (Monckton et al., 1995; Wong et al., 1995). It is therefore recommended to size the expansion in DM1 cellular models since, in culture they differ from the patients’ tissue from which they are derived.
CTG expansion size in muscle correlated with age of disease onset in our patients. We used small pool PCR, a technique widely used for measuring CTG repeats (Gomes-Pereira et al., 2004). Previous relationships between CTG size in blood and age of disease onset have been reported (Cumming et al., 2019; Morales et al., 2012; Overend et al., 2019) and recently this correlation has been found applying the same technique in DNA isolated from saliva (Corrales et al., 2019). Sizing the CTG expansion in muscle is always difficult since muscle tissue has larger expansions than other tissues (Lavedan et al., 1993; Thornton et al., 1994). However, when analyzing these samples with small pool PCR, finding expansion size is easier since few molecules are amplified. By using this method, we could find that CTG muscle expansion size is also a good indicator of disease severity.
The single cell study we did here presented many challenges. For example, it can only be done with cells that can be sorted, meaning that differentiated cells such as myotubes or neurons, which are key cells to study in this disease, cannot be studied. In addition, although a quick FISH staining protocol was applied the process of staining and the time of visualization probably affects and compromises RNA material. An approach without staining may benefit splicing analysis, but in this case, one will lose the information regarding how many foci the myoblast cell contains. One way or another, with the current technical limitations of this analysis we lose precious molecular data. Finally, the diverse clinical presentation of DM1 patients and the currently available scales to evaluate muscle function further (most being categorical type of scales) complicates single cell study. The vast heterogeneity found in DM1 patients, both at a cellular and a symptomatic level, asks for a further refinement of the current single cell technologies available and further development of continuous scales to study muscle function. This will give us valuable insight into the diversity of DM1 pathology at a single cell level, and how this correlates to the clinical manifestations of the disease.
To our knowledge, this is the first study performed at a single cell level in DM1. In addition, we analyzed the highest number of myoblasts per patient compared to other studies and for the first time attempted to analyze alternative splicing defects in single cells. We determined that the number of RNA foci was heterogeneous in DM1 myoblasts and submitted to a dynamic process, allowing some DM1 myoblasts to be free of RNA foci, even though they were carrying the CTG expansion. This RNA foci load compromises muscle functionality evaluated through 6MWD. The heterogeneity found at the single cell level in the patients could be a key factor for treatment development and efficacy. Differences in alterations between cells could mean that different concentrations of future therapies would be needed. It will be key to find the equilibrium between treating all cells and avoiding toxic effects.