Direct conversion of adult Huntington’s disease patient fibroblasts into medium spiny neurons recapitulates hallmark phenotypes such as cell death, in contrast to models that lack epigenetic markers of aging. This successful ‘disease-in-a-dish’ highlights the benefits of capturing age in an adult-onset disorder model.
A 1945 bottle of Château Mouton-Rothschild has matured over time into an amazing wine because of a complex chemical reaction occurring among sugars, acids and phenolic compounds. In this issue of Nature Neuroscience, Victor et al.1 describe an effect of aging that elicits specific pathological phenotypes in cells derived from patients with Huntington’s disease (HD). In doing so, they have created an improved model of this devastating neurological disorder.
HD is caused by the presence of 36 or more CAG trinucleotide repeats in the huntingtin (HTT) gene1, with symptoms including involuntary movements, motor dysfunction, cognitive impairment, psychiatric abnormalities and eventual death2. The number of repeats is correlated with age of onset and severity2. Striatal medium spiny neurons (MSNs) display mutant HTT protein aggregation and are selectively vulnerable to death, although at later stages of disease the cortex and other brain regions are also affected. While there are juvenile-onset cases of HD associated with very long repeat lengths (>60), in most cases this disease is primarily an adult-onset disorder for which there are no current treatments. In vitro models of HD could facilitate the discovery of therapies, but these have been challenging to develop using human tissues. Disease-in-a-dish modeling of adult neurodegenerative diseases often generate neurons from patient-derived induced pluripotent stem cells (iPSCs)3, which have been reprogrammed to an embryonic stem cell-like ‘young’ state4. In effect, reprogramming wipes clean most of the epigenetic marks of aging. Differentiation protocols that mimic human development are used to differentiate iPSCs into MSNs. iPSC models generated from juvenile-onset HD with more than 60 CAG repeats display strong CAG-length-dependent phenotypes. However, perhaps not surprisingly, many pathways disrupted in these models are associated with development3. In contrast, most iPSC-derived lines with under 60 CAG repeats have no major observable phenotypes, such as aggregate formation or cell death3, indicating that modeling an adult-onset disorder might require the maintenance of signatures of aging.
Direct reprogramming of adult fibroblasts into neurons, accomplished using transcription factors expressed during neurogenesis and/or specific microRNAs (miR9/9* and miR124)5, provided a breakthrough method for producing neurons from aged tissues6. With the addition of MSN-specific transcription factors (Ctip2, DLX1 and DLX2), these microRNAs were even able to directly convert human fibroblasts specifically to a MSN fate7. Unlike iPSC-derived cells, directly reprogrammed cells retain aging-related signatures, such as age-specific transcriptional profiles (including a decrease of RanBP17 expression), nucleocytoplasmic compartmentalization8 and an aged DNA methylation epigenetic clock9. This begets the question: what impact would the maintenance of these signatures of aging have on adult-onset disease modeling? Victor et al.1 now report the direct reprogramming of control and HD patient fibroblasts into MSNs1. Remarkably, HD neurons generated in this manner displayed both inclusion-body formation and cell death, mirroring what is seen in the striatum of individuals with HD.
Victor et al.1 collected HD fibroblasts, with a range of 40–180 CAG repeats, as well as control fibroblasts, and directly converted them into MSNs using miR9/9*-124 with Ctip2, DLX1, DLX2 and Myt1L (miR9/9*-124 + CDM). After 30 days, regardless of the CAG length, 70–80% of fibroblasts became MSNs. The MSNs from symptomatic HD patients in the adult onset range (40–47 CAGs) displayed characteristic phenotypes seen in HD postmortem MSNs, such as HTT aggregates, DNA damage, mitochondrial and metabolic dysfunction, and cell death (Fig. 1). Notably, HD-MSNs differentiated from iPSC-derived fibroblasts with 40 CAGs did not display the phenotypes observed in the directly converted aged HD-MSNs, suggesting that reprogramming erases essential aging characteristics required for the death of the MSNs. In addition, directly reprogrammed presymptomatic HD patient fibroblasts showed a less severe phenotype than their aged counterparts, while still having similar levels of mutant HTT aggregates.
To address the neuronal type-specificity of this effect, the authors next directly converted HD fibroblasts into cortical neurons using miR9/9*-124, NeuroD2, ASCL1 and MYT1L. While HTT aggregates were observed, the DNA damage and cell death phenotypes were lost, suggesting that these were specific to the aged MSN populations, supporting the early vulnerability of these neurons in HD. Finally, Victor et al.1 showed that mRNA for SP9, a transcription factor required for the development of MSNs, is downregulated in the directly converted HD-MSNs and that SP9 overexpression can rescue the HD-MSNs from cell death.
The key finding of this article is that using aged fibroblasts as the starting source for neurons provided hallmark phenotypes of HD such as HTT aggregate formation and cell death in HD-MSN samples without requiring stressors such as proteasome inhibition for induction or specialized single-cell tracking equipment for detection3. These results corroborate a previous report that showed HTT aggregates and cell death in MSN-like neurons directly converted from HD fibroblasts using short hairpin RNA knockdown of PTB110. However, in the earlier study, as with prior reports of disease-in-a-dish models for HD, the patient fibroblasts used for direct conversion were from juvenile-onset patients, with 68 and 86 repeats. In contrast, Victor et al.1 used patient fibroblasts with repeat lengths in the 40s and still showed pronounced phenotypes in these adult-onset HD-MSNs. It will be interesting to observe whether the number of CAG repeats and/or disease-onset information correlates with the severity of the HD-MSN cell death phenotypes and whether reprogramming to iPSCs reverses it, as this would be a powerful validation of the assay3.
A potential limitation of direct conversion is the consistent availability of patient primary tissue. The limited expansion potential of fibroblasts due to senescence over time11 and their genomic instability in culture may hinder large-scale ‘omics’ studies requiring abundant tissue and collaborative projects with cell distribution across laboratories. In contrast, iPSCs provide a limitless source of tissue that can be differentiated into the cell type of interest. While Victor et al.1 showed that reprogrammed iPSCs cannot faithfully model aged HD phenotypes, an aging strategy may be able to circumvent this limitation. For instance, overexpression of the ‘aging’ gene encoding progerin has been used in an iPSC-based in vitro model of Parkinson’s disease12. Ideally, transcriptomic data from directly converted neurons that were responsible for the toxic effect of mutant HTT could elucidate key aging-related genes or epigenetic signatures required for cell death and thus suggest a more defined transcriptomic or epigenetic ‘roadmap’ for inducing in vitro aging and disease-specific phenotypes.
One concern in using these directly converted HD-MSNs is their rapid degeneration in culture, with MSN generation at day 30 and onset of death at day 35. While the authors found that treatment with KU600019 significantly reduced cell death, the limited time window in which to influence disease progression may minimize this cellular platform’s utility for drug screening. Another concern using aged HD-MSNs is, ironically, that the involvement of early development is overlooked. This is critical given that both rodent and patient-derived iPSC models of HD have shown that development is important to HD manifestation3,13. As such, iPSC models that replay development may still be required to fully understand this disorder. Fortunately, the growing field of MSN generation can provide various in vitro models to use based on the specific question.
Still to be addressed is the finding that the MSN death phenotypes occurred in the absence of other neuronal cells (such as cortical cells or interneurons) or support cells (such as astrocytes or microglia). This suggests an MSN-autonomous mechanism of mutant HTT toxicity, at least in this in vitro system. Direct conversion was used here to generate cortical neurons and has been successfully used to generate astrocytes14. Given that support cells and other neuronal subtypes play a role in HD15, future studies could co-culture HD-MSNs with other directly converted cell types to assess protective or deleterious effects.
Even after years of HD research, certain disease aspects remain unresolved, including the specific role of mutant HTT aggregation and the selective vulnerability of MSNs, and no US Food and Drug Association–approved therapeutic exists. This is partly because animal models are limited in their ability to recapitulate the human disease. Victor et al.1 have recapitulated key phenotypes of HD in MSNs directly converted from aged adult-onset HD fibroblasts of both preonset and symptomatic patients. This represents a significant breakthrough for the field of adult disease modeling and should provide a powerful platform for drug screening on a human cell type relevant to the disease. These findings now need to be reproduced by other laboratories, and hopefully this in vitro model will be able to complement animal models to fully understand the pathogenesis of HD, ultimately leading to a successful therapy.
Victor, M. B. et al. Nat. Neurosci. https://doi.org/10.1038/s41593-018-0075-7 (2018).
The Huntington’s Disease Collaborative Research Group. Cell 72, 971–983 (1993).
Mattis, V. B. & Svendsen, C. N. Brain Res. 1656, 76–87 (2017).
Takahashi, K. et al. Cell 131, 861–872 (2007).
Yoo, A. S. et al. Nature 476, 228–231 (2011).
Vierbuchen, T. et al. Nature 463, 1035–1041 (2010).
Victor, M. B. et al. Neuron 84, 311–323 (2014).
Mertens, J. et al. Cell Stem Cell 17, 705–718 (2015).
Huh, C. J. et al. eLife https://doi.org/10.7554/eLife.18648.001 (2016).
Liu, Y. et al. PLoS One 9, e109621 (2014).
Hayflick, L. & Moorhead, P. S. Exp. Cell Res. 25, 585–621 (1961).
Miller, J. D. et al. Cell Stem Cell 13, 691–705 (2013).
Connor, B. Stem Cells https://doi.org/10.1002/stem.2747 (2017).
Caiazzo, M. et al. Stem Cell Reports 4, 25–36 (2015).
Ehrlich, M. E. Neurotherapeutics 9, 270–284 (2012).
The authors declare no competing financial interests.
About this article
Cite this article
Mattis, V.B., Svendsen, C.N. Huntington modeling improves with age. Nat Neurosci 21, 301–303 (2018). https://doi.org/10.1038/s41593-018-0086-4
Frontiers in Neuroscience (2018)