You are here

| Direct Reprogramming

Of Mice and Men: Direct Conversion of fibroblasts to neurons



By Carla Mellough

It is just 5 years since Takahashi and Yamanaka1 demonstrated that both embryonic and adult mouse somatic cells could be reprogrammed back to a pluripotent state, creating induced pluripotent stem cells (iPSCs). In the short time since this achievement the equivalent has been demonstrated across multiple human somatic cell types, which has paved the way for a new era of disease modelling and a focus towards the generation of autologous transplantable tissues by differentiating patient-specific iPSCs into mature cell types for cell replacement. The availability of disease- and indeed patient-specific iPSC disease models have enabled rapid advances in other scientific fields, the most notable of which may be genetics and gene therapy. Further to this, not only does the list of reprogrammable somatic cells expand with time, more recent studies have demonstrated that the iPSC stage can be completely bypassed by the direct conversion of somatic cells into other mature somatic cell types. One of these studies, published in January this year, reported that the expression of three neuronal transcription factors (Ascl1, Brn2, and Myt1l) in mouse fibroblasts could efficiently convert these cells into induced neurons (iN).2 Four months later, a study published in PNAS by Pfisterer et al.3 demonstrated that, using the same strategy, the same direct lineage conversion is also achievable in human.

The authors used tissue from fetuses from 5.5 to 7 weeks post-conception to generate a primary human embryonic fibroblast (hEF) culture, taking care to discard all cells with neurogenic potential. To initiate conversion to a neuronal lineage, fibroblasts were transduced with doxycycline-responsive lentiviral particles encoding Ascl1, Brn2, and Myt1l and then grown under neural induction conditions. Twelve days later many cells exhibited a neuronal morphology and expressed the neural marker β–III-tubulin, the cells acquiring more elaborate neuronal morphology with time in culture - an effect which was not seen in control non-transduced fibroblast cultures. The conversion efficiency of human iN (hiN) production from hEFs (passage 2) was an impressive 16% but this dropped to 9% (at best) when passage 3 hEFs were used. Over 90% of the hiNs generated expressed MAP2 and synaptophysin, yet patch clamp recordings at early stages showed no electrophysiological activity. However by day 32 of culture electrophysiologically active hiN cells were present which exhibited action potentials and the normal range of neuronal resting membrane potentials.

Pfisterer et al.3 then tested whether the continual expression of Ascl1, Brn2, and Myt1l was required for hiN induction. They found that the removal of doxycycline from days 3 or 7 onwards had no effect, whereas only sporadic hiN cells were observed if doxycyline was removed on day 0, indicating that transient expression of Ascl1, Brn2, and Myt1l is sufficient to generate hiNs which can continue to mature under neural media conditions. Further, while neither Brn2a nor Myt1l alone were able to generate hiNs, Ascl1 in isolation could do so without affecting induction efficiency, although cells displayed underdeveloped neuronal morphology compared to those derived following three factor reprogramming. When introduced in pairs, Ascl1 and Brn2 were as effective in producing hiNs as all three factors together, yet these cells failed to mature into electrophysiologically active neurons.

To exclude the possibility that hiNs were derivatives of residual neural crest cells from primary hEF cultures (even though stringent measures were taken to avoid this the authors report <0.1% of hEF cultures at passage 2 were contaminating neural crest cells), Pfisterer et al. also used two long-term propagated commercially available hEF lines (HFL1 and hFF). Transduction of the human fetal lung fibroblast (HFL1) and human foreskin fibroblast (hFF) cell lines with Ascl1, Brn2, and Myt1l generated hiNs with similar morphology and typical neuronal membrane electrophysiology to that of hiNs derived from primary hEF cultures, including the firing of action potentials in response to depolarizing current injection. Yet whilst the efficiency of hiN production from the fetal HFL1 line was similar to primary hEF cultures, the postnatal hFF line displayed reduced efficiency, almost half that of fetal fibroblast cultures, suggesting that mature adult fibroblasts may be more difficult to reprogram.

In accordance with results from the mouse study of the same nature,2 hiNs demonstrated both excitatory and inhibitory neurotransmitter phenotypes but cholinergic, serotonergic, or dopaminergic phenotypes were rare. To determine whether hiNs could be directed specifically towards a particular neuronal subtype, the authors chose 10 key genes implicated in midbrain patterning and dopaminergic specification (En1, Foxa2, Gli1, Lmx1a, Lmx1b, Msx1, Nurr1, Otx2, Pax2 and Pax5) and generated a pool of lentiviruses containing all 10 genes (LentiDA) which they used to transduce the human fibroblasts alongside the three conversion factors Ascl1, Brn2, and Myt1l. This approach generated a small (<1% of hiNs) but reproducible population of dopaminergic neurons, demonstrating that fate instruction can enhance the production of the desired neuronal phenotype. Further research demonstrated that Lmx1a and FoxA2 used alongside the three conversion factors performed just as well as using the entire LentiDA pool. HFL1 cells transduced with Ascl1, Brn2, and Myt1l and then Lmx1a and FoxA2 from day three of induction yielded around 10% (in the best case 25%) of hiNs expressing the dopaminergic neurotransmitter tyrosine hydroxylase (TH). The resulting cells showed dopaminergic neuronal morphology and could generate action potentials either spontaneously or by depolarizing current by day 28 of induction, typical of midbrain dopaminergic neurons. In addition, TH+ cells expressed the second enzyme in dopamine synthesis (Aromatic L-amino acid decarboxylase) and Nurr1, an orphan nuclear receptor expressed by midbrain dopaminergic neurons.

The ability to directly convert human dopaminergic neurons from fibroblasts provides an important additional system for the study of motor diseases or Parkinson’s disease or for the generation of replacement dopaminergic neurons for cell therapy which, like iPSC-generated neural cell types, are excluded from the ethical concerns surrounding the use of human embryonic stem cells (hESCs). Compared to iPSC reprogramming, direct lineage conversion has more rapid reprogramming dynamics, higher reprogramming efficiency and lower potential risk of tumor formation, making them an excellent potential source of transplantable tissue. Just one month after Pfisterer et al.3 published their results, another group reported that a different set of factors (in this case Mash1, Nurr1 and Lmx1a) can also enable hiN dopaminergic specification, indicating that the activation of different (although closely related) transcriptional cascades can bring about the same end result.4 Later this year another group has reported that terminally differentiated hepatocytes can also be converted into functional iN cells, revealing that direct lineage reprogramming is possible across the different germ layers.5

One imagines that the investigation of the long term survival and reinnervation capability of directly converted hiN cells is already underway, and the results of this work will be greatly anticipated. Whether bypassing the pluripotent stage will enable transplanted hiNs to integrate with less immunoreactivity within the host (a recent study6 reported the surprising result that some mouse iPSCs are immunogenic and in fact that autologous iPSCs transplanted into matched mice are more immunogenic than matched hESCs) remains to be seen, but will no doubt render hiNs more clinically relevant as a result, assuming that viral- and integration-free conversion systems can be developed. The achievements of this study also pave the way for the directed differentiation of other neuronal subtypes for the wider study of neurological disease and generation of transplantable cells to reconstitute degenerate or dysfunctional neural circuitry.

An impressive Letter to Nature published online last week from Kriks et al.7 reports that by utilising a novel method of dopaminergic differentiation from hESCs that mimics midbrain development in vivo, greater dopaminergic engraftment efficiency than what has been shown previously can be achieved. In this Letter, the long term engraftment and functionality of hESC-derived dopaminergic neurons is reported in the mouse, rat and Parkinsonian monkey, and the results indicate that attaining the appropriate ontogenic stage of dopaminergic cell differentiation is key – in this case, as soon as the cells exit the cell cycle and become postmitotic. Similar impressive results have also been achieved in a mouse model of retinal degeneration using postmitotic photoreceptor precursors,8 implying that determination of the optimal developmental stage at which to capture transplantable tissues is an important consideration for the success of ongoing transplant studies. Taking into account the incremental advances we are seeing from all of the studies discussed herein, it does seem that we are making steady progress towards understanding how best to produce scalable functional transplantable dopaminergic neurons and encourage their long-term functional engraftment to try and alleviate the symptoms of many currently incurable degenerative conditions.



1. Takahashi K, Yamanaka S. (2006) Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell 126:663–676.

2. Vierbuchen T, et al. (2010) Direct conversion of fibroblasts to functional neurons by defined factors. Nature 463:1035–1041.

3. Pfisterer et al. (2011) Direct conversion of human fibroblasts to dopaminergic neurons. PNAS 108(25):10343-8.

4. Caiazzo et al. (2011) Direct generation of functional dopaminergic neurons from mouse and human fibroblasts. Nature 476(7359):224-7.

5. Marro et al. (2011) Direct Lineage Conversion of Terminally Differentiated Hepatocytes to Functional Neurons. Cell Stem Cell 9(4):374-382.

6. Zhao et al. (2011) Immunogenicity of Induced Pluripotent Stem Cells. Nature 474:212–215.

7. Kriks et al. (2011) Dopamine neurons derived from human ES cells efficiently engraft in animal models of Parkinson’s disease. Nature doi:10.1038/nature10648.

8. MacLaren et al. (2006) Retinal repair by transplantation of photoreceptor precursors. Nature 444, 203-207

Also see related article from Nature