How can we investigate a disease affecting neurons, which cannot be isolated from patients for analysis? As the study of one neurological disorder shows, a first step might be to make patient-specific neurons.
Spinal muscular atrophy (SMA) is among the most common inherited neurological disorders that can cause death in childhood. Although the most severe form of the condition is known to be due to mutations in both copies of the SMN1 gene, an understanding of how SMA develops has been slow. On page 277 of this issue, Ebert et al.1 describe an elegant method in which skin cells taken from a patient with SMA were used to generate neurons of the same genetic make-up and characteristic features as the neurons affected in this disorder. The approach will not only lead to a better understanding of SMA, but should also aid the testing and development of drugs for its treatment.
For genetic reasons, SMA is unique to humans. Whereas rodents, flies and worms have only one version of the gene that is mutated in SMA, the human genome has two — SMN1 and SMN2. These two genes differ in three nucleotides, one of which affects the genes' expression. Consequently, whereas SMN1 expression predominantly produces the full-length SMN protein, SMN2 expression results in only 10% occurring as full-length protein, with the rest being a truncated, non-functional protein. The severity of SMA in humans varies according to the number of copies of SMN2 the patient carries — the more copies of SMN2 there are, the better the patient can cope with the SMA-associated SMN1 mutation. As mice carry only one copy of the Smn gene, its deletion results in early embryonic death2. Furthermore, if SMN expression is reduced in cells not normally affected by the disease, the cells die3. Therefore, a mouse model of SMA can be produced only by artificially introducing the human SMN2 gene, or mutant or truncated versions of the human SMN1 gene, into mice lacking the Smn gene.
Another problem in elucidating the mechanism underlying SMA stems from the fact that, although the SMN protein is ubiquitously expressed and is essential for the assembly of the machinery for processing messenger RNA, its absence in SMA predominantly affects spinal motor neurons. And clearly, unlike blood cells or skin fibroblast cells, for example, human neurons cannot simply be isolated for analysis.
A breakthrough came in 2006, when it was found that, by introducing four specific gene transcription factors, mouse fibroblasts could be reprogrammed to become induced pluripotent stem (iPS) cells, which have the ability to differentiate into any cell type4. Progress in stem-cell research has since been impressive, allowing previously unresolved problems in biomedicine to be addressed. For example, a mutation in the gene for haemoglobin was corrected in reprogrammed fibroblasts to treat sickle-cell anaemia in a mouse model5. Moreover, patient-specific iPS cells have been generated from the cells of patients with other disorders, including amyotrophic lateral sclerosis6, a neurodegenerative disease that also affects motor neurons.
Ebert et al.1 isolated fibroblasts from a child with SMA and from the child's healthy mother, and reprogrammed each set of cells to become iPS cells. From the iPS cells, they then derived motor neurons of the same genetic make-up as that of each donor. The authors found that, compared with the motor-neuron-like cells derived from the mother, those of the child showed the abnormalities typical of SMA motor neurons — a finding that makes the iPS cell line an invaluable tool for studying this disease.
Can this approach1 be used to develop therapies? To investigate this possibility, the authors treated their iPS-cell-derived motor neurons with valproic acid or tobramycin, two drugs that increase the expression of both full-length and truncated versions of the SMN protein. In the cells derived from the patient with SMA, they observed increased production of full-length SMN and alterations in SMN-containing nuclear structures. By contrast, the equivalent cells from the mother did not show the same increase. This observation is indeed exciting, as it suggests that the same approach could be used to test the efficacy of drug candidates before their use in patients. As is often the case, however, the situation is not so simple.
Ebert and colleagues' motor-neuron-like cells indeed express markers of bona fide developing spinal motor neurons, such as the transcription factors OLIG2, ISLET1, HB9 and HOXB4, as well as the SMI-32 protein and the enzyme choline acetyltransferase. But what remains unclear is whether, in vivo, the cells surrounding the motor neurons and making contact with them — including other neurons, skeletal-muscle cells and support cells called glia — signal to the motor neurons to mediate their differentiation and the development of disease-specific features.
Take amyotrophic lateral sclerosis, for example. Experience with a mouse model of this disease, in which expression of a mutant form of the enzyme SOD-1 is increased, has shown7 that the disease process is not limited to intrinsic defects in neurons, and that other cell types — in particular glial cells called astrocytes — also contribute. The same phenomenon was observed in vitro in motor neurons generated from iPS cell lines from patients with amyotrophic lateral sclerosis8. These motor neurons are sensitive to the toxic effects of astrocytes and other glial cells that carry a mutation in the SOD-1 gene8.
Determining the contribution of cells other than motor neurons to the development of SMA is therefore important, not least because, even in the most severe form of the disorder, symptoms do not usually appear until a few weeks after birth — long after these neurons have begun to express their specific markers, developed axonal processes and made contact with skeletal-muscle cells. The availability of iPS cell lines should allow researchers to elucidate the contribution of other cells to motor-neuron differentiation as the neurons become affected in SMA, and to identify signals that act later in the development of embryonic motor neurons in the spinal cord.
The availability of protocols for inducing iPS cells to differentiate into motor neurons — using factors such as retinoic acid, the sonic hedgehog protein, GDNF and BDNF — is a first step towards resolving these issues. Moreover, additional strategies for tracking the differentiation of other subcellular structures9,10 in iPS-cell-derived motor neurons should help to determine why the disease is specific to motor neurons, with other types of neurons being unaffected.