This paper was retracted on 10 April 2014.

A human brain with Alzheimer’s disease cannot be studied in a dish, but scientists can now do the next best thing. In a paper in today’s Cell, researchers led by Asa Abeliovich at Columbia University, New York City, describe how they took skin cells from people with familial forms of AD and converted them directly into neurons. When Abeliovich and colleagues compared the AD neurons to neurons made from healthy people, they found differences in how the cells handled amyloid precursor protein (APP), shedding light on potential disease mechanisms. Induced human neurons will allow scientists to study the endogenous human pathology in vitro, complementing traditional approaches such as transgenic mouse models, Abeliovich told ARF. He suggested that induced neurons could help researchers glean clues not only to familial disease processes, but to the far more common sporadic cases as well.

“[This is] simply a remarkable and complete piece of work which will now set a standard for stem cell work in neurological disease. The standard of the characterization of the neuronal cultures is very high,” John Hardy at University College London, U.K., wrote to ARF. He was not involved in the work but is taking a similar approach in his own lab.

In the last five years, scientists have become intrigued by the potential of induced pluripotent stem (iPS) cells to provide neurons for disease modeling, drug screening, and cell replacement therapy (see, e.g., ARF series on iPS cells). A recent flurry of papers has now shown that human fibroblasts can be transformed directly into neurons, skipping the stem cell stage altogether (see ARF related news story on Pang et al., 2011; ARF related news story on Caiazzo et al., 2011; Pfisterer et al., 2011). Most such methods use transcription factors, but researchers led by Gerald Crabtree at Stanford University, Palo Alto, California, recently demonstrated that microRNA also can direct this conversion by altering chromatin-remodeling complexes (see Yoo et al., 2011). In a paper published online July 27 in Cell Stem Cell, researchers led by Stuart Lipton at the Sanford-Burnham Medical Research Institute, La Jolla, California, and Sheng Ding at the Gladstone Institute of Cardiovascular Disease, San Francisco, describe a similar method that uses microRNA plus transcription factors to turn fibroblasts into neurons.

Abeliovich and colleagues took as their starting point the direct conversion protocol developed by Marius Wernig and colleagues at Stanford University (see ARF related news story on Vierbuchen et al., 2010). Optimizing the protocol, joint first authors Liang Qiang, Ryousuke Fujita, Toru Yamashita, and Sergio Angulo found that a single lentiviral vector containing genes for the transcription factors Ascl1, Brn2, and Zic1 was sufficient to convert human, healthy control fibroblasts into neurons. By adding a second vector with the transcription factor Myt1l, they could achieve about 65 percent conversion efficiency. About 85 percent of the converted cells contained the neuronal marker MAP2, and showed the genetic profile and electrophysiological characteristics of neurons. To demonstrate that the neurons were functional in vivo, the researchers transplanted the cells into embryonic mouse brains and examined them at postnatal day seven. They found that the neurons were electrically active and appeared to have integrated into existing neuronal circuitry.

Since more than half the induced neurons were glutamatergic, Qiang and colleagues investigated whether they could use them to model AD pathology. The authors made neurons from the skin cells of three people with familial AD, who carried mutations in either presenilin-1 or -2, as well as from three healthy adults. When Qiang and colleagues compared the cells in vitro, they saw differences in how the AD cells handled APP. Neurons made from AD patients had less APP in the cell membrane, and more APP in early endocytic vesicles, than did neurons from healthy people. Neurons with mutant presenilin also had larger endosomes compared to normal neurons. This fits with other data implicating APP sorting in Alzheimer’s pathology (see, e.g., ARF related news story and ARF news story).

In addition, the AD neurons produced more Aβ and had a greater ratio of Aβ42/Aβ40 than the control cells. This could be a direct consequence of the sorting differences, since Aβ can be generated in endocytic compartments, but that remains to be proved, the authors note. Intriguingly, the original fibroblasts from AD patients had only a modest increase in Aβ ratio compared to control fibroblasts, and did not show the endocytic phenotype. This indicates that something specific about neurons leads to AD pathology. “Neurons handle APP differently from fibroblasts,” Abeliovich noted. The authors found they could restore normal endosomes by overexpressing wild-type presenilin in the AD neurons. By contrast, inhibiting presenilin’s γ-secretase function did not rescue the phenotype, but did cause neurons from healthy people to accumulate more APP-filled endosomes, mimicking the effect of the PS mutations. This suggests that at least some AD-associated presenilin mutations could be loss-of-function mutations, Abeliovich said, and that reduced γ-secretase activity can lead to an AD-like cell phenotype. A recent clinical trial of the γ-secretase inhibitor semagacestat was halted because people on the drug deteriorated more rapidly than those taking placebo (see ARF related news story).

The value of studying these human cells, Abeliovich told ARF, is that it allows researchers to investigate endogenous human mutations in their native environment. “We did not fiddle at all with the Alzheimer’s machinery,” he pointed out. By contrast, most mouse models involve overexpression of APP or presenilin, which may create distinct disease mechanisms. In addition, mice do not recapitulate many of the aspects of human AD, Abeliovich said. He is also excited about the possibility of using induced neurons to study sporadic AD, which makes up the majority of disease cases. “You cannot make a mouse model of sporadic AD,” Abeliovich pointed out, but you can look for common disease mechanisms in neurons made from people with the disease.

Ole Isacson at McLean Hospital in Belmont, Massachusetts, who studies iPS-based models for Parkinson’s disease, agrees that induced human neurons have tremendous value for research. “What we have found in Parkinson’s and Huntington’s disease is that these new systems are telling us much more about the disease than we imagined,” Isacson told ARF, adding that he is happy to see this approach coming to the Alzheimer’s field. Among other things, using cells made from individual patients allows researchers to investigate patient-specific disease mechanisms, Isacson said. He envisions an era of personalized medicine where patients could be directed to an appropriate therapy or clinical trial based on the phenotype that their induced neurons reveal.

Such manufactured neurons also hold tremendous potential for drug screening, Isacson said. A typical route for drug screens is to start with cell lines, then move to in-vivo rodent models, after which researchers hope the results translate to humans. By using human cells from the start, researchers, in effect, would be allowed to jump a step in the development process, Isacson pointed out, because scientists would find out right away if potential drugs were toxic to human neurons. Such approaches have already shown some success, he noted. Researchers led by Lorenz Studer at Sloan-Kettering Institute, New York City, used induced cells to identify a unique mechanism for the genetic disorder familial dysautonomia, as well as a drug that partially rescues the defect in culture (see Lee et al., 2009 and Lee and Studer, 2011).

Induced human neurons, whether made directly or via stem cells, could also provide the raw material for cell replacement therapy. Many hurdles remain before this could be realized, Abeliovich told ARF. The main question, he said, is whether such cells will integrate successfully into the brain and replace failing neurons. Other issues are technical, for example, finding ways to convert cells without introducing foreign genes into the DNA. Scientists have developed several non-integrating or removable transgene methods of producing iPS cells, although the conversion efficiencies are often quite low (see ARF coverage of adenoviral approaches, plasmids, transgene removal, and episomal vectors). A more efficient non-integrating reprogramming approach, described by researchers led by Noemi Fusaki at DNAVEC Corporation, Tsukuba, Japan, is to use RNA-based Sendai virus vectors (see Fusaki et al., 2009). In a paper published online this week in the Proceedings of the National Academy of Sciences, Fusaki and colleagues tweak this approach by introducing temperature-sensitive mutations into the vectors, allowing scientists to easily shut down viral replication and clear the viruses from induced cells.

The new reprogramming technologies raise the question of whether it is better to make neurons directly from fibroblasts, or to first produce iPS cells. Abeliovich and colleagues note that the direct conversion method is faster, simpler, and has a higher conversion efficiency, and, in addition, avoids the generation of potentially tumorigenic stem cells. Isacson, however, points out that iPS cells also have many advantages. Once an iPS clone is generated, it can be expanded indefinitely to produce as many cells as desired, and can also be frozen down and stored. The iPS cells have the potential to make diverse cell types, which is important because neurodegenerative diseases often affect numerous body systems as well as many neuronal subtypes. Isacson noted that neurons made from iPS cells have been shown to maintain their phenotype for years in vivo, but researchers have not yet demonstrated long-term stability in directly induced neurons. Stability for two or three weeks in vivo is insufficient proof, Isacson said, since human neurons take months to mature. “I am still in favor of [using] iPS cells, because I think it gives much more flexibility in terms of what you can do with the cells,” he added.—Madolyn Bowman Rogers


  1. This is a very nice paper showing direct conversion of skin fibroblasts to well-differentiated, functional glutamatergic human induced neurons (hIN). Use of this direct differentiation method with fibroblasts from PS1 and PS2 FAD subjects resulted in cells with increased Aβ42 (and aberrant APP intracellular localization), providing a proof of concept that this method can be used to derive cellular models expressing physiologic levels of pathogenic proteins in a neuronal cell context. These may be more useful model systems to study AD than those using undifferentiated cells or overexpressing pathogenetic proteins.

    View all comments by I. Paul Shapiro

Make a Comment

To make a comment you must login or register.


News Citations

  1. Falsifications Cast Shadow Over Stem Cell Field
  2. Where in the World Are the iPS Cells?
  3. Turning Human Fibroblasts Into Neurons; Making Safer Stem Cells
  4. Faster, Safer Ways to Cook Up Dopaminergic Neurons
  5. Research Brief: From Fibroblast to Neuron in One Easy Step
  6. Research Brief: Novel Sorting Protein May Affect Amyloid Production
  7. APP Sorting Protein May Link Alzheimer’s and Diabetes
  8. Lilly Halts IDENTITY Trials as Patients Worsen on Secretase Inhibitor
  9. Stem Cell Advance—A Safer, Inducible Pluripotent Cell?
  10. Newest Stem Cell Approaches Abandon Viruses, Tap Testes
  11. Without a Trace: iPS Cell Techniques Leave No Footprints
  12. Circuit Menders? Neurogenesis, Stem Cells Show Potential

Paper Citations

  1. . Induction of human neuronal cells by defined transcription factors. Nature. 2011 Aug 11;476(7359):220-3. PubMed.
  2. . Direct generation of functional dopaminergic neurons from mouse and human fibroblasts. Nature. 2011 Aug 11;476(7359):224-7. PubMed.
  3. . Direct conversion of human fibroblasts to dopaminergic neurons. Proc Natl Acad Sci U S A. 2011 Jun 21;108(25):10343-8. PubMed.
  4. . MicroRNA-mediated conversion of human fibroblasts to neurons. Nature. 2011 Aug 11;476(7359):228-31. PubMed.
  5. . Direct conversion of fibroblasts to functional neurons by defined factors. Nature. 2010 Feb 25;463(7284):1035-41. PubMed.
  6. . Modelling pathogenesis and treatment of familial dysautonomia using patient-specific iPSCs. Nature. 2009 Sep 17;461(7262):402-6. PubMed.
  7. . Modelling familial dysautonomia in human induced pluripotent stem cells. Philos Trans R Soc Lond B Biol Sci. 2011 Aug 12;366(1575):2286-96. PubMed.
  8. . Efficient induction of transgene-free human pluripotent stem cells using a vector based on Sendai virus, an RNA virus that does not integrate into the host genome. Proc Jpn Acad Ser B Phys Biol Sci. 2009;85(8):348-62. PubMed.

Further Reading

Primary Papers

  1. . Direct reprogramming of adult human fibroblasts to functional neurons under defined conditions. Cell Stem Cell. 2011 Aug 5;9(2):113-8. PubMed.
  2. . Directed conversion of Alzheimer's disease patient skin fibroblasts into functional neurons. Cell. 2011 Aug 5;146(3):359-71. PubMed. RETRACTED
  3. . Efficient generation of transgene-free human induced pluripotent stem cells (iPSCs) by temperature-sensitive Sendai virus vectors. Proc Natl Acad Sci U S A. 2011 Aug 23;108(34):14234-9. PubMed.