Administration of Adipose-Derived Stem Cells Extracellular Vesicles in a Murine Model of Spinal Muscular Atrophy: Effects of a New Potential Therapeutic Strategy

Major Points and Findings:

This animal study tests the extracellular vesicles (EVs) released by adipose-derived mesenchymal stem cells. The vesicles were injected into the brain ventricles of newborn mice with a severe form of spinal muscular atrophy. The stem cells themselves were not injected. No human beings took part, and the study gives no information on what would happen in a child.

We include it because the stem cell literature in SMA is extremely thin. It is one of the few recent laboratory studies to test a mesenchymal stem cell product in a living SMA model and not only in cell culture. The authors also explain why they chose vesicles over cells. Engraftment of transplanted mesenchymal stem cells in the central nervous system is, in their words, only a small percentage, and the therapeutic effect of those cells appears to come from what they secrete and not from their survival in the tissue.

Aim:

To find out whether extracellular vesicles isolated from mouse adipose-derived mesenchymal stem cells (ASC-EVs) can slow disease progression in SMNΔ7 mice, and to establish whether such an approach could later be combined with the approved SMN-raising drugs. The authors set out to test an SMN-independent strategy, meaning one that does not work by increasing survival motor neuron protein.

Methods:

Adipose stem cells were isolated from the inguinal fat of five healthy adult C57Bl6/J mice and grown to passages 14 to 18. Before harvest the cells were deprived of serum for 48 hours so that no vesicles from the culture serum could contaminate the preparation. Vesicles were then isolated with a commercial kit and used fresh.

The vesicle preparation was characterised before use. Nanoparticle tracking gave a concentration of 2.38 × 10⁸ particles per mL, with a size mode of 113.6 nm and a mean of 181.7 nm. Electron microscopy showed round lipid bilayer vesicles of 50 to 150 nm. Western blotting confirmed the vesicle markers CD9 and HSP70. A previous proteomic study by the same group found no full-length SMN protein inside these vesicles, which is why the authors call the approach SMN-independent. The cargo does contain insulin-like growth factor 1, which acts through the PI3K-Akt pathway.

SMNΔ7 mice (Jackson stock 005025) were bred to give SMA and wild-type pups. In total 43 SMA and 15 wild-type mice were used. SMA pups were randomly assigned to vesicles (n = 16) or PBS (n = 27), and untreated wild-type pups (n = 15) served as the healthy reference. Each treated pup received 0.5 µg of ASC-EVs in 2 µL, injected into a cerebral ventricle on postnatal day 3 and again on day 6. The injections were given under hypothermia anaesthesia lasting 3 to 5 minutes, using a neonatal stereotaxic frame and a Hamilton microsyringe. Controls received 2 µL of PBS by the same route.

Body weight and three neonatal motor tests (tail suspension, righting reflex, negative geotaxis) were recorded from day 2 or 4 to day 10. All animals were killed on day 10 for tissue analysis. This comprised stereological motor neuron counts in the lumbar spinal cord, cleaved Caspase-3 as a marker of cell death, GFAP and IBA-1 for astrocyte and microglial activation, muscle fibre size in gastrocnemius and quadriceps, and neuromuscular junction innervation. There was no survival endpoint, since the study ended at day 10 for every animal.

Results:

Motor behaviour: Body weight did not differ between treated and untreated SMA pups, although the treated group was still gaining weight at day 10 while the control group had plateaued at day 8. Treated pups scored better on tail suspension at day 8 (p = 0.0251) and righted themselves faster at day 8 (p = 0.0167), and the difference in negative geotaxis at day 10 was larger (p below 0.0001). None of the groups reached wild-type performance.

Motor neurons: The clearest effect was on motor neuron counts. Lumbar motor neuron density was 2106.87 ± 96.47 cells per mm³ in treated mice against 1438.80 ± 73.09 in PBS controls (p = 0.0010). Healthy wild-type mice had 4604.97 ± 173.24, so the treatment slowed the loss and came nowhere near preventing it. Motor neurons positive for cleaved Caspase-3 fell from 21.48% to 10.59%.

Inflammation: Astrocyte activation (GFAP signal) fell from 9.06% to 6.16% (p = 0.0040). Total IBA-1 signal, the measure of microglial activation, did not differ between treated, untreated and healthy animals. Only the proportion of resting, ramified microglia differed, and only on a direct two-group comparison.

Muscle: The muscle results were the weakest. In the gastrocnemius there was no difference in fibre area or diameter. In the quadriceps the difference reached significance only when the two SMA groups were compared directly by t-test (fibre area p = 0.0188). It was not significant in the comparison across all three groups. Neuromuscular junction innervation improved by a few percentage points in both muscles, and none of those differences were statistically significant. The authors state that at the peripheral level they could not show a significant impact.

Conclusions:

Twice-repeated injection of adipose stem cell vesicles into the ventricles of newborn SMA mice protected lumbar motor neurons, reduced apoptosis and astrocyte activation, and produced modest gains on motor tests. Muscle and neuromuscular junction effects were partial at best. The authors propose the vesicles as a possible add-on to the approved SMN-raising drugs. They do not suggest them as a replacement. The mechanism remains to be clarified, including where the vesicles go after injection.

Two limitations are relevant for families reading this report. The intracerebroventricular route used here is invasive and the authors themselves call it controversial, suggesting intranasal delivery as a future alternative. The study also has no survival data and no comparison against nusinersen, risdiplam or gene therapy, so it says nothing about how this would compare with treatments that are already approved.

Background Information:

The paper contains an internal inconsistency. The abstract and methods describe the SMNΔ7 mouse as a model of severe SMA. The conclusions section calls it a model of SMA type II. The SMNΔ7 mouse is conventionally used as a severe model, and this summary follows the methods.

A correction to this paper was published on 29 April 2024 (10.1186/s13287-024-03744-x). It only restores a note that the final two authors contributed equally, and changes nothing in the data.

The discussion makes a wider argument. The approved drugs raise SMN protein, but reduced SMN affects many tissues beyond motor neurons, including muscle, heart, blood vessels and the immune system. SMN-independent approaches are aimed at that gap. Whether cells, or vesicles derived from cells, can fill any part of it in human beings is a question this study does not answer.