Using Human Stem Cells to Predict the Efficacy of Alzheimer’s Drugs

Scientists who work in the pharmaceutical industry have seen this time and time again: A candidate drug that works brilliantly in laboratory animals fails to work in human trials. So what’s up with this?

Now a research consortium from the University of Bonn and the biomedical company Life & Brain GmbH has shown that animal models of Alzheimer’s disease fail to recapitulate the results observed with cultured human nerve cells made from stem cells. Thus, they conclude that candidate Alzheimer’s disease drugs should be tested in human nerve cells rather than laboratory animals.

In the brains of patients with Alzheimer’s disease beta-amyloid protein deposits form that are deleterious to nerve cells. Scientists who work for drug companies are trying to find compounds that prevent the formation of these deposits. In laboratory mice that have a form of Alzheimer’s disease, over-the-counter drugs called NSAIDs (non-steroidal anti-inflammatory drugs), which include such population agents as aspirin, Tylenol, Advil, Nuprin and so on prevent the formation of beta-amyloid deposits. However in clinical trials, the NSAIDs royally flopped (see Jaturapatporn DIsaac MGMcCleery JTabet N. Cochrane Database Syst Rev. 2012 Feb 15;2:CD006378).

Professor Oliver Brüstle, the director of the Institute for Reconstructive Neurobiology at the University of Bonn and Chief Executive Officer of Life and Brain GmbH, said, “The reasons for these negative results have remained unclear for a long time.”

Jerome Mertens, a former member of Professor Brüstle’s research, and the lead author on this work, said, “Remarkably, these compounds were never tested directly on the actual target cells – the human neuron.”

The reason for this disparity is not difficult to understand because purified human neurons were very difficult to acquire. However, advances in stem cell biology have largely solved this problem, since patient-specific induced pluripotent stem cells can be grow in large numbers and differentiated into neurons in large numbers.

Using this technology, Brüstle and his collaborators from the University of Leuven in Belgium have made nerve cells from human patients. These cells were then used to test the ability of NSAIDs to prevent the formation of beta-amyloid deposits.

According to Philipp Koch, who led this study, “To predict the efficacy of Alzheimer drugs, such tests have to be performed directly on the affected human nerve cells.”

Nerve cells made from human induced pluripotent stem cells were completely resistant to NSAIDs. These drugs showed no ability to alter the biochemical mechanisms in these cells that eventually lead to the production of beta-amyloid.

Why then did they work in laboratory animals? Koch and his colleagues think that biochemical differences between laboratory mice and human cells allow the drugs to work in one but not in the other. In Koch’s words, “The results are simply not transferable.”

In the future, scientists hope to screen potential Alzheimer’s disease drugs with human cells made from the patient’s own cells.

“The development of a single drug takes an average of ten years,” said Brüstle. “By using patient-specific nerve cells as a test system, investments by pharmaceutical companies and the tedious search for urgently needed Alzheimer’s medications could be greatly streamlined.”

Understanding the Role of a Protein in Familial Alzheimer’s Disease

Lawrence Goldstein, director of the UC San Diego Stem Cell Program and a member of the Departments of Cellular and Molecular Medicine and Neurosciences, has an abiding interest in Alzheimer’s disease (AD).  To that end, he and his colleagues have used genetically engineered human induced pluripotent stem cells to determine the role a particular protein plays in the causation of familial AD.  Apparently, a simple loss-of-function model does not contribute to the inherited form of this disorder.  Goldstein hopes that his findings might be able to better explain the mechanisms behind AD and help drug makers design better drugs to treat this disease.

Familial AD is a subset of the larger group of conditions known as early-onset AD.  The vast majority of cases of AD are “sporadic” and do not have a precise known cause, even though age is a primary risk factor (an estimated 5.2 million Americans have AD).  Familial AD is causes by mutations in particular genes.  One of these genes, PS1, encodes a protein called “presenilin 1,” which acts as a protease (an enzyme that clips other proteins in half).  Presenilin 1 is the catalytic component of a protein complex called “gamma-secretase.”  Presenilin 1 forms a complex with three other proteins (Nicastrin, Aph1, Pen2) to form gamma-secretase, and this enzyme attacks specific proteins that are embedded in the cell membrane and clips them into smaller pieces.


By clipping these cell membrane proteins into smaller pieces, gamma-secretase helps the cell transport cellular material from one side of the cell membrane to the other side or form the outside of the cell to the inside.

One of the substrates of gamma-secretase is a protein called amyloid precursor protein (APP).  While the function of APP remains unknown, APP cleavage by the gamma-secretase produces small protein fragments known as amyloid beta.

A consensus among AD researchers is that the accumulation of specific forms of amyloid beta causes the formation of the amyloid plaques that kills off neurons and leads to the onset of AD.  The most abundant product of gamma-secretase cleavage of APP is a protein called “Aβ40.”  This protein is forty amino acids long and does not cause any brain damage.  However, a minority product of APP cleave by the gamma-secretase is “Aβ42,” which is 42 amino acids long and forms the amyloid plaques and neurofibillar tangles that are so characteristic of AD (see Scheuner, D., et al., Nat. Med. 2, 864–870).

According to Goldstein, most of the time, gamma-secretase clips APP without causing any problems, but some 20% of the time, the protein clips APP incorrectly and this results in the plaque-forming forms of amyloid beta.  Goldstein explained: “Our research demonstrates very precisely that mutations in PS1 double the frequency of bad cuts.”

To demonstrate this, Goldstein and his co-workers purchased human induced pluripotent stem cells and differentiated them into neurons.  These neurons contained different alleles (forms) of the PS1 gene, and some of these mutant forms of PS1 contained the types of mutations that cause familial AD.  Once PS1 allele in particular called PS1 ΔE9 increased the ratio of Aβ42 to Aβ40 dose-dependent manner.  Since the PS1 ΔE9 causes familial AD, this research elucidates precisely why it does so.

“We were able to investigate exactly how specific mutations and their frequency change the behavior of neurons.  We took finely engineered cells that we knew and understood and then looked how a single mutation causes changed in the molecular scissors and what happened next.”

Presenilin allele consequences

Goldstein further notes, “In some ways, this is a powerful technical demonstration of the promise of stem cells and genomics research in better understanding and ultimately treating AD.  We were able to identify and assign precise limits on how a mutations works in familial AD.  That’s an important step in advancing the science, in finding drugs and treatments that can slow, maybe reverse, the disease’s devastating effects.”

Drug Induces Hearing Restoration in Rodents

Fish and birds are able to regenerate their hearing after damage, but mammals are not able to do so, and hearing loss is irreversible in mammals like human beings. However, a new study has shown that the application of a particular drug can activate genes normally expressed during hair cell development. This work resulted from collaboration between researchers at Harvard Medical School, the Massachusetts Eye and Ear Infirmary, and Keio University School of Medicine in Japan. This finding is a first in the field or regenerative medicine.

Hair Cell Regeneration

In the cochlea, small cells known as hair cells convert sound waves into electrical signals that are interpreted by the brain into sounds. If these hair cells are damaged or destroyed by acoustic injury, then a permanent loss of hearing ensues. Such damage is treated with cochlear implants, which are surgically implanted devices that convert sounds to electrical signals.

“Cochlear implants are very successful and have helped a lot of people, but there’s a general feeling among clinicians, scientists, and patients that a biological repair would be preferable,” said Albert Edge, an otologist at Harvard University and the Massachusetts Eye and Ear Infirmary and lead author of the Neuron paper that reports these findings.

In previous work, Edge and his colleagues had shown that inhibiting the Notch signaling pathway was important for hair cells to form properly during fetal development (Jeon, S.J., Fujioka, M., Kim, S.C., and Edge, A.S.B. (2011). Notch signaling alters sensory or neuronal cell fate specification of inner ear stem cells. J. Neurosci. 31, 8351–8358). In their new study, Edge and his colleagues inhibited the Notch signaling pathway to determine, if such inhibition could initiate hair cell regeneration in adult mammals. They used a variety of approaches. In their first experiments, they used different inhibitors to determine their effects on isolated ear tissues. This allowed them to isolate one inhibitor in particular, the ɣ-secretase inhibitor LY411575, that led to increased expression of several molecular markers found in developing hair cells.


“It was quite a surprise,” said Edge. “We were very excited when we saw that a secretase inhibitor would have any effect at all in an adult animal.”

Next, Edge and his co-workers tested the inhibitor in mice that had hearing damage and reduced hair cell populations as a result of exposure to a loud noise. They tagged cells in the inner ear to follow their fate and discovered that the inhibitor, when applied to the inner ears of the mice, caused supporting cells to differentiate into replacement hair cells. These newly formed hair cells partially restored hearing at low sound frequencies, but not at higher frequencies. This effect lasted for at least three months.

This study examined the effect of the inhibitor when it was given one day after noise damage, which is a time when Notch signaling is naturally increased. This it is possible that a small window of time exists after an acoustic injury during which the drug is effective.

Edge concluded: “The improvement we saw is modest. So we’re now looking at variations of the approach and whether we can use the same drug to treat other types of hearing loss.”

See: Mizutari K, Fujioka M, Hosoya M, Bramhall N, et al. (2013) Notch Inhibition Induces Cochlear Hair Cell Regeneration and Recovery of Hearing after Acoustic Trauma. Neuron 77, 58-69.