Custom-Made Neural Stem Cells
In this column, we explore how the judicious use of neural stem cells (NSCs) has led to a research Holy Grail: the creation of research-ready, patient-specific neurons.
The Business of Pleasure and Pain
Elegant Knockdowns, DISC1, and Schizophrenia
It is ironic that an attempt to do a molecular end-run around a politically hot topic could result in an important breakthrough in the treatment of neurological disease with potentially strong implications for the psychiatric community. Ironic maybe, but true.
In this column, we explore how the judicious use of neural stem cells (NSCs) has led to a research Holy Grail: the creation of research-ready, patient-specific neurons. This technology did not use the famously controversial embryonic stem cells. These custom-made NSCs were created from politically neutral adult tissues (fibroblasts), which were originally isolated from an affected patient. With no embryo in sight, scientists genetically reprogrammed fibroblasts into stem cells, which were then induced to develop into NSCs. This is an extraordinary finding with many topics to be discussed here:
• The potential research utility for patient-specific neurons
• An explanation of how stem cells can be made from adult tissues
• A striking set of results that involve one of the most commonly inherited and lethal childhood neurological disorders: spinal muscular atrophy (SMA)
Research utility for NSCs
Of what possible utility could molecular investigations of a motor disorder have for the mental health community? Before getting into the specifics of the breakthrough, it might be useful to address a real-world psychiatric need, using depression and SSRIs as an example, to see where these data fit.
When we consider the molecular mechanisms of SSRI interactions, it is easy to resort to commonly taught ideas about interactions that involve a single synapse. Nothing could be further from the truth. The most comprehensive neurological view of SSRI interactions must take into account the participation of thousands of individual neurons strung together in coordinated, complex neural networks.
And not just serotonergic neurons. These cells are in contact with many other central nervous denizens, from adjacent glial cells to the extracellular matrix into which the cells are embedded. What do these circuits actually look like in patients who are vulnerable to depression? Is their architecture all that different from patients who do not exhibit this vulnerability? If there are differences, could they eventually predict drug efficacy? Could these differences only be detected by constructing parts of the circuit from scratch, or could they be observed at the level of a single cell?
The first step in answering these questions involves growing a custom-made batch of serotonergic neurons derived only from the affected patients, and then asking relevant structure/function questions. From attempting to understand molecular mechanisms of disease to testing the efficacy of potential medications, such patient-specific test beds would have a powerful research utility. Until recently, the creation of such tailor-made neural substrates had been an impossible goal.
While it will certainly be quite some time before we can grow entire parkinsonian dopaminergic pathways in a dish, it is now possible to create individual patient-specific neurons in culture. The technology comes from that end-run I mentioned earlier, through the use of a certain type of stem cell. It is to these interesting cellular substrates that we now turn.
Inducible stem cells
To say that embryonic stem cell research has been subject to heated political debate is an understatement. The bugaboo has been the source materials from which the stem cells would be isolated-human embryos-many left over from embryos generated in in vitro fertilization laboratories.
In 2006, researchers found a way to create stem cells that bypassed the need for human embryos. The original technique involved the introduction of 4 specific gene products into mature mouse fibroblasts. Surprisingly, this cocktail was found to reprogram adult stem cells and reverse-engineer them into pluripotent stem cells. Like embryonic stem cells, the altered stem cells had the ability to differentiate into any cell type. Eventually, a protocol was developed that did the same thing in human tissues. The cells were called iPSCs, short for induced pluripotent stem cells.
This was quite a breakthrough. No longer would researchers need to harvest cells from extant human embryos to do stem cell research. Skin cells would do. Scientists were soon able to regenerate-and then correct-molecular dysfunction in a mouse model of sickle cell anemia using this technology.
Could any of this work apply to humans, specifically to human neural tissue? Another successful round of experiments (with amyotrophic lateral sclerosis neurons) prompted researchers to study motor disease, ie, SMA.
Of those hereditary neurological disorders capable of causing death in pediatric populations, SMA is easily the most common. The disease is unique to humans and associated with 2 genes, SMN1 and SMN2. For reasons that are not well understood, the absence of the survival motor neuron (SMN) protein results in an alteration of the function of spinal motor neurons. The primary feature is muscle weakness and atrophy. Death occurs at infancy in the most severe forms of the disease, with symptoms generally presenting several weeks after birth. There are many other, nonlethal forms of the disorder, however, with a wide spectrum of symptoms that range from trivial motor effects to catastrophic impairment.
Why this variation? Both genes express in unaffected individuals, but the biological heavy lifting belongs to the SMN1 gene. Because of structural constraints, the expression pattern of the SMN2 gene normally results in only 10% of its protein being processed as a full-length (and functional) polypeptide; 90% of its protein output is truncated (and nonfunctional). That is okay, as long as the SMN1 gene is intact. But when SMN1 is mutated and silent, the disease condition results. Assuming there is a damaged SMN1, the severity of SMA varies according to the number of other SMN2 copies the infant may carry. The more copies of SMN2 gene, the greater the population of functional protein. This interaction explains in part why there can be so much varia-tion in the clinical presentation.
The great mystery is why SMN protein loss results in motor cell alterations that lead to the disease state in the first place. The protein is known to be essential for normal messenger RNA processing and is expressed throughout the body. Yet its absence most severely affects spinal motor neurons.
The most exacting way to attack this “black box” would be to isolate the motor neuron populations from the patient, then compare these populations with unaffected controls and look for differences, of which there are many. These include responses to various medications. It is well known that the application of valproic acid (an anticonvulsant and/or mood stabilizer) or tobramycin (an aminoglycoside) to cultured cells, for example, leads to changes in the expression patterns of both full-length SMN protein and truncated forms. What is the molecular basis of this unusual interaction? And could such differences be used as a “molecular flashlight” to ferret out other secrets regarding the SMN protein? Creating custom-made neurons-one population from an affected individual, another from an unaffected control-would certainly give a test bed capable of answering this question.
The data
Studying these 2 populations is precisely what a group of investigators did. The researchers isolated fibroblasts from an affected child and also from the child’s healthy unaffected mother.
The next step was to generate custom-made neurons. Several steps would be required (Figure). First, using the iPSC protocols I mentioned, the researchers would attempt to create stem cells from both child and parent sources. If that worked, the researchers would then try to induce these patient-derived stem cells into motor neurons-ones that would carry the same biological mechanisms observed in both the diseased and the healthy populations. If successful, the researchers would have their custom-made test beds. They could begin characterization studies; reactions to valproic acid and tobramycin would make obvious first choices to try.
The first step worked. The researchers were able to generate custom-made stem cells from both child and parent. The researchers then tackled the hard part: manufacturing spinal motor neurons from these stem cell populations. They certainly generated promising cellular populations. But the iPSC technolo-gies are new enough that a visual inspection of the generated cells might be necessary-but certainly not sufficient-to show the presence of motor neurons.
There are ways to gain greater reassurance. One way to assay the success of the protocol is to look for bona fide molecular markers of developing spinal neurons. It is known, for example, that extant motor neurons express the protein SMI032 and choline acetyltransferase. Did these induced cells express such proteins? The answer turned out to be yes, both for the affected child and for the unaffected parent. Developing cells in these populations possess transcription factors such as HOXB4, ISLET1, HB9, and OLIG2 as well. Did the induced populations express these markers? They did indeed. While not completely conclusive, it appeared that the researchers had generated patient-specific motor neurons from known affected and unaffected sources.
The next characterization experiments also yielded fruit. They were able to find that the child and parent neurons reacted very differently to the normally stimulating effects of valproic acid and tobramycin. The child’s cells showed elevated levels of SMN protein, both of the truncated form and full-length version. In addition, SMN-containing nuclear structures were altered. No such elevation occurred in the unaffected maternal line of cells.
These differences were significant for 2 reasons. First, it gave the investigators a toehold in their attempts to characterize at a more intimate level the differences between affected and unaffected cells. Second, the differences were discovered as reactions to known medications. The hope is that similar approaches could be used to test the efficacy of various medications before committing to human trials.
Conclusions
These data, full of promising implications as they are, need to be treated with some caution. First, the experimental cells are pure populations derived from stem cells. This hardly reflects the physical in vivo situation. The cells and matrix components that normally surround such cells in nature, including skeletal muscle tissues and even other neurons, are not present in these studies.
Another objection concerns the fidelity of the conversion process itself. The differentiation pattern seen in various molecular markers hinted that the investigators generated real live spinal motor neurons; however, one cannot a priori say they have in every way created a motor neuron that precisely mimics the real-world situation. These cells may lack many subtle molecular processes-and a few extra, equally subtle interactions-that could easily escape detection, at least by current technologies. Because subtle differences can profoundly influence intracellular molecular interactions, especially when we think about reactions to medications, this is a true concern.
The most exciting aspect of these studies comes from what the future holds. A great deal of speculation has gone into thinking about how to tailor medications to individual patients. That certainly is a psychiatric issue . . . I need not talk to this audience about the variable effects of, say, fluoxetine on clinical outcomes. We have visited this topic in past columns. The ability to create patient-specific cellular test beds may go a long way toward solving some of these problems. Indeed, clinics of the future might routinely screen to decide what medications their patients should receive-and in what concentrations.
There is much work to do. To date, none has been applied to neurological systems relevant to mental health professionals. Even given the cautions mentioned above, there is no reason why it couldn’t. That’s not bad for having to do with an end-run around a hostile, politically charged issue such as stem cell debates. Would that all ethical issues could be decided so cleanly, or with so much fruit.
