Mobile DNA sequences called L1 contribute to the brain's genetic heterogeneity and may affect neuron function. The protein MeCP2, which is mutated in Rett syndrome, seems to regulate the activity of these genomic elements. See Letter p.443
In his struggle to rationalize the laws of quantum mechanics, Albert Einstein once wrote that God “does not play dice”. A similar struggle comes to mind when trying to understand the biological consequences of several reports from the labs of Gage and Muotri, including one appearing on page 443 of this issue (Muotri et al.1). These studies offer evidence for the occurrence of quasi-random genetic changes in neurons during development. The work raises provocative questions about whether 'playing dice' during brain development contributes to the differences that make each of us unique, and how such changes may be related to brain disorders.
The immune system is well known for using genetic recombination and random mutations2 to adapt antibody defences against invaders. The question is, does the nervous system use a similar approach to create neuronal diversity? For example, each olfactory neuron expresses only one of more than 1,000 possible olfactory-receptor genes. Permanent genetic changes are unlikely to underlie selection of olfactory-receptor genes for expression: mice generated from the genetic material of a neuron expressing a single olfactory receptor — by the technique of somatic-cell nuclear transfer — re-establish the full complement of olfactory receptors3,4. Such evidence, however, does not rule out the possibility of other permanent genetic changes occurring within the neuronal lineage.
Transposons, or 'jumping genes'5, can mediate one such mechanism, inducing changes in the DNA. In the human genome, the most common class of transposons is retrotransposons, which have the ability to amplify themselves. For instance, one type of retrotransposon, known as LINE-1 or L1, constitutes a staggering 17% of the human genome. However, only a small fraction of L1 is functionally intact and active, and the role of these retrotransposons in the human genome remains mysterious.
It has been proposed6,7 that L1 is involved in genome evolution, and there is evidence8 that retrotransposition can induce genetic changes responsible for human disease. Until recently, L1 activity was thought to be confined to the earliest stages of embryonic development and the germ-cell lineage. But this picture started to crumble a few years ago, when the Gage lab demonstrated9 active L1 retrotransposition in neuronal precursor cells. Subsequent mechanistic studies identified10 dual binding sites in the promoter region of L1 for the transcriptional regulators SOX2 and TCF/LEF, which are responsible for changes in the associated chromatin (DNA–protein complexes), and so for the switch between repression and activation of L1 expression (Fig. 1).
The intricate regulation of L1 activity was the starting point for the next chapter of this remarkable story, as told by Muotri and colleagues1. Previous work in cell lines11 had suggested that the protein MeCP2 can recruit the enzyme HDAC1 and that the two contribute to L1 repression. A role for MeCP2 in regulating L1 activity was particularly intriguing, because the gene encoding this protein is mutated in Rett syndrome (RTT) — an X-chromosome-linked disorder specific to the nervous system that is a leading genetic cause of mental retardation in girls.
Muotri et al. now use a broad armamentarium of approaches to investigate the link between MeCP2 and L1 retrotransposition. They present compelling in vitro and in vivo evidence that both promoter activity and L1 retrotransposition rates are significantly increased in neuronal precursors of mice that lack the Mecp2 gene. Indeed, the authors' detailed imaging studies demonstrate clear differences in L1 retrotransposition rates between normal mice and those lacking Mecp2 across several regions of the adult brain.
Muotri and co-workers also investigate the involvement of MeCP2 in regulating L1 retrotransposition in the cells of patients with RTT. For this, they reprogram the patients' skin fibroblast cells into patient-specific induced pluripotent stem cells (iPSCs), and detect increased L1 retrotransposition rates in neurons derived from the iPSCs of patients with RTT compared with neurons from control iPSCs. The authors further corroborate these data with post-mortem work, finding increased levels of L1 DNA content in the brain, as opposed to matched heart-tissue samples, of patients with RTT, and in brain samples from patients compared with those from age-matched controls.
These findings1 provide intriguing evidence for the modulation of L1 activity by MeCP2 during development of the nervous system. Is such modulation causally involved in RTT pathology? On the basis of the current knowledge of the disease, probably not.
There is no obvious correlation between the timing of L1 activity during embryonic development and the delayed disease onset in patients with RTT, which typically occurs 1–2 years after birth. It is also difficult to imagine how an increased rate of quasi-random genetic changes, even if biased towards genes expressed in neurons9, can lead to the highly reproducible disease symptoms observed in RTT. What's more, in Mecp2-mutant mice, a striking recovery from RTT symptoms occurs on re-expression of Mecp2, even in mature animals — a strategy that clearly does not affect early retrotransposition events. Therefore, changes in L1 activity might not cause RTT, but may contribute to variability among patients — beyond the well-known differences in mutation type and X-chromosome inactivation status.
Another intriguing issue relates to the use of RTT iPSCs in modelling human disease. After reprogramming to iPSCs, the inactive X chromosome of differentiated cells does not become active12. Therefore, all differentiated cells generated from a given RTT iPSC line are expected to show identical inactivation patterns of either their normal X chromosome or their MECP2-mutant X chromosome, rather than the random inactivation pattern observed in patients. The increased L1 activity in neuronal precursors derived from RTT iPSCs indicates that these cell lines showed inactivation of the normal X chromosome and may, therefore, model a more severe form of RTT.
The holy grail for defining the functional impact of L1 activation in neuronal development would be a method that can selectively switch on and off retrotransposition events using genetic or pharmacological tools. Given the sheer number and genetic complexity of transposable elements, this remains a daunting task, although targeting the reverse transcriptase enzyme or other mission-critical determinants of L1 activity may represent a potentially tractable approach. Until then, we are left to wonder whether playing dice during development of the central nervous system is indeed part of what makes each of us unique in both health and disease.
Muotri, A. R. et al. Nature 468, 443–446 (2010).
Tonegawa, S. Nature 302, 575–581 (1983).
Eggan, K. et al. Nature 428, 44–49 (2004).
Li, J., Ishii, T., Feinstein, P. & Mombaerts, P. Nature 428, 393–399 (2004).
McClintock, B. Proc. Natl Acad. Sci. USA 36, 344–355 (1950).
Xing, J. et al. Proc. Natl Acad. Sci. USA 103, 17608–17613 (2006).
Cordaux, R. & Batzer, M. A. Nature Rev. Genet. 10, 691–703 (2009).
Kazazian, H. H. Jr et al. Nature 332, 164–166 (1988).
Muotri, A. R. et al. Nature 435, 903–910 (2005).
Kuwabara, T. et al. Nature Neurosci. 12, 1097–1105 (2009).
Yu, F., Zingler, N., Schumann, G. & Strätling, W. H. Nucleic Acids Res. 29, 4493–4501 (2001).
Tchieu, J. et al. Cell Stem Cell 7, 329–342 (2010).
About this article
Stem Cell Reviews and Reports (2014)
Philosophical Transactions of the Royal Society B: Biological Sciences (2011)