© 2009 Nature America, Inc. All rights reserved.
news and views expression correlated with early adversity, but not with the stress of suicide per se. Turning to the potential epigenetic mechanisms underlying these changes in glucocorticoid receptor expression, McGowan et al.1 found that there are lower levels of exon 1F methylation in suicide victims with a history of being abused than in nonabused suicide victims or controls. In in vitro studies, the methylation of exon 1F markedly inhibited NGFI-A–inducible gene expression. Also, in vitro methylation of exon 1F, in the pattern observed in the hippocampi of the abuse victims, inhibited NGFI-A activation of gene expression and binding of NGFI-A to the promoter, as would have been predicted by the rodent model. These postmortem findings have now brought a mechanism for long-term effects of early experience to the level of human biology. This extension of an animal model of epigenetic regulation is not only elegant, but also timely; investigations of other animal models are beginning to suggest that a variety of behavioral patterns, such as social defeat stress in rodents and pharmacologic interventions, produce sustained effects on gene expression via covalent modification of DNA or histones12.
Together, the emerging studies of epigenetic mechanisms may suggest new therapeutics approaches to treating neuropsychiatric disorders. Given the challenges of developing convincing animal models of neuropsychiatric disorders, the extension of a biologically important mechanism to humans may help motivate treatment development. Studies of the epigenetic modification of the NR3C1 gene from animal models to the postmortem human hippocampus1,4–6 illustrate the power of pursing a problem in depth, especially when the mechanisms being studied influence expression of a biologically important gene. Studies focused on known genes must, however, be complemented by unbiased searches for other genes that are regulated by salient experiences, illness and pharmacologic agents13. McGowan et al.1 do not make excessively reductive claims for the influence of the NR3C1 gene on behavior. In the long run, however, an understanding of how experience and other factors interact with the sequence information contained in individual genomes to produce long-term patterns of behavior will require more. It will require the depth illustrated by the line of research
c ulminating in the work of McGowan et al.1, as well as far greater breadth, and then the means to understand the dizzying number of genegene and gene-environment interactions that underlie who we are and what we do. COMPETING INTERESTS STATEMENT The authors declare competing financial interests: details accompany the full-text HTML version of the paper at http://www.nature.com/natureneuroscience/.
1. McGowan, P.O. et al. Nat. Neurosci. 12, 342–348 (2009). 2. Hyman, S.E. Nature 455, 890–893 (2008). 3. Kendler, K.S. et al. Am. J. Psychiatry 163, 109–114 (2006). 4. Meaney, M.J. Annu. Rev. Neurosci. 24, 1161–1192 (2001). 5. Meaney, M.J. & Szyf, M. Trends Neurosci. 28, 456–465 (2005). 6. Weaver, I.C.G. et al. Nat. Neurosci. 7, 847–854 (2004). 7. Heim, C. et al. J. Am. Med. Assoc. 284, 592–597 (2000). 8. Heim, C. et al. Psychoneuroendocrinology 33, 693–710 (2008). 9. Dube, S.R. et al. J. Am. Med. Assoc. 286, 3089–3096 (2001). 10. Caspi, A. et al. Science 301, 386–389 (2003). 11. Kendler, K.S. et al. Arch. Gen. Psychiatry 62, 529–535 (2005). 12. Tsankova, N.M. et al. Nat. Neurosci. 9, 519–525 (2006). 13. Alter, M.D. et al. PLoS One 3, e3344 (2008).
Notch keeps ependymal cells in line Chunmei Zhao, Hoonkyo Suh & Fred H Gage The ependymal cells lining the lateral ventricles are not stem cells, but a study now shows that they can be activated to generate neuroblasts in a stroke model, and mature olfactory bulb neurons when Notch signaling is disrupted. The identity of neural stem cells (NSCs) responsible for lifelong neurogenesis in the rodent olfactory bulb has been hotly debated over the past decade. New olfactory bulb neurons originate from a region adjacent to the lateral ventricles, consisting of a single layer of ependymal cells, and the adjoining subventricular zone (SVZ). Ependymal cells were once credible candidates for NSC status, but soon moved to the background as more and more studies showed that they are largely quiescent, and that an astroglialike population of cells in the SVZ are the NSCs responsible for neurogenesis in the olfactory bulb1–4. A new study by Carlén et al. in this issue puts ependymal cells back in the spotlight with the discovery that they
The authors are at the Laboratory of Genetics, Salk Institute for Biological Studies, 10010 North Torrey Pines Road, La Jolla, California 92037, USA. e-mail:
[email protected]
may turn into neural cells after ischemia or disruption of Notch signaling5 (Fig. 1). First, the authors revisited the question of whether ependymal cells possess NSC properties. For this purpose, they used CMVCre adenoviral and FoxJ1-Cre lentiviral vectors to trace the putative progeny of ependymal cells in Rosa26 reporter mice. FoxJ1 is only expressed in cells with motile cilia, and the FoxJ1-Cre lentivirus allows specific targeting of ependymal cells. With each virus, the only recombined cells detected were ependymal cells themselves, suggesting that these cells are terminally differentiated and do not give rise to other cell types (Fig. 1b). The authors also showed that ependymal cells do not express any markers of proliferative cells, further evidence for their quiescence. A surprise came, however, when they examined the fate of ependymal cells in a stroke model (Fig. 1c). The authors first injected FoxJ1-Cre viruses into the lateral ventricle, then induced stroke on the other side
nature neuroscience volume 12 | number 3 | MARCH 2009
of the brain one week after virus injection. In contrast to results under normal conditions, the authors frequently found recombined nonependymal cells located away from the ependymal cell layer. These cells stained for markers of neuroblasts (doublecortin; DCX) or astrocytes (glial fibrillary acidic protein; GFAP). To exclude the possibility of residual viral activity at the time of stroke induction, the authors electroporated in an inducible cre under the control of the Foxj1 promoter. They induced cre shortly after electroporation and induced stroke 3 weeks later. Again, they observed recombined nonependymal cells. It is important to note, however, that expression of a single molecular marker is insufficient for the identification of a cell type. For example, GFAP is expressed by both mature astrocytes and NSCs in the SVZ. In addition, the intricate nature of stroke might cause ectopic DCX or GFAP expression irrespective of cell fate. Nevertheless, these studies suggest that, under injury conditions, ependymal
243
news and views b
c LV
E
SVZ
d E
OB LV
DCX+
SVZ
E E Uninjured; or constitutively active Notch + stroke
© 2009 Nature America, Inc. All rights reserved.
E
E
DCX+
Death
GFAP+ Stroke
E Rbpj –/–
Mature neurons in the olfactory bulb
Kim Caesar
a
Figure 1 Ependymal cells switch to a different fate after stroke. (a) New neurons are added to the rodent olfactory bulb throughout their lifetime. These neurons are thought to originate from neural stem cells in the SVZ of the lateral ventricles, an area adjacent to the ependymal cell layer that lines the ventricles. (b) Under normal conditions, ependymal cells are quiescent and do not give rise to any other cell types. Notch signaling is responsible for the maintenance of ependymal cells. (c) When stroke is induced in mice, some ependymal cells detach from the single layer. The progeny of these detached cells are mitotic or express markers for neural cells such as doublecortin (DCX) and glial fibrillary acidic protein (GFAP). These ectopic neuroblasts do not survive to become mature neurons. The downregulation of Notch signaling may be responsible for this response as overexpression of an active form of Notch ligands (b) prevents these responses. (d) Complete disruption of Notch signaling in ependymal cells mimics, to a certain degree, the phenotypes seen in the stroke model. Ependymal cells also detach, and most become neuroblasts that later migrate and become mature neurons in the olfactory bulb. LV, lateral ventricle; SVZ, subventricular zone; OB, olfactory bulb; E, ependymal cells.
cells indeed have the potential to give rise to nonependymal cells. In addition to recombined nonependymal cells, the authors observed recombined mitotic cells (revealed by bromodeoxyuridine incorporation or Ki67 immunoreactivity), which were never observed in control mice. These cells were all located in the SVZ or rostral migratory stream but not in the ependymal layer. In addition, a severe loss of ependymal cells was observed in the injured brains. These observations indicate that the ependymal cells may have changed fate before entering into cell cycle and that the maintenance of ependymal cells is disrupted by stroke. Because Notch signaling is required for the maintenance of neuronal progenitors during development, the authors asked whether the Notch pathway could be responsible for the maintenance of ependymal cells in the adult brain. Indeed, Notch signaling in ependymal cells was considerably decreased after stroke. Expression of a constitutively active Notch ligand was able to prevent ependymal cell loss in the stroke model (Fig. 1b). Conversely, disruption of the Notch pathway, by removing the essential Notch downstream component RBP-J, resulted in phenotypes similar to those seen in the stroke model, such as ectopic recombined cells, mitotic recombined cells and almost complete loss of ependymal cells (Fig. 1d). Interestingly, ependymal cells could give rise to mature neurons in the olfactory bulb when Notch signaling was disrupted. In contrast, no mature neurons were found in the olfactory bulb or the injury site in the stroke model (Fig. 1c, d). These data establish a new role for Notch signaling in the maintenance of ependymal cells.
244
This study appears to settle the question of whether ependymal cells are NSCs: the authors show that, under physiological conditions, ependymal cells are quiescent and do not have the potential to give rise to other cells types. However, the findings do contradict previous observations by the same group1 and two recent independent studies, in which adenovirus-labeled cells6 or CD133+ ependymal cells7 labeled by electroporation were found to give rise to neurons in the olfactory bulb. In a further discrepancy, both the Frisen1,5 and the Alvarez-Buylla2,6 groups have obtained inconsistent results regarding the contribution of adenovirus-labeled cells to olfactory neurons. The recent study by Mirzadeh, Alvarez-Buylla and colleagues6 showed that GFAP+ cells in the SVZ have an apical surface that contacts the ventricles and suggested that GFAP+ cells can be directly targeted by adenoviruses. The authors observed labeled olfactory neurons after ventricular injection of GFAP-Cre adenovirus. However, they did not show that GFAP+ cells were directly infected by the virus, and the possibility remains that infected ependymal cells give rise to GFAP+ cells in which Cre recombinase is expressed. It may be possible that the contribution of ependymal cells to olfactory neurogenesis is minimal and the current strategy with adenovirus labeling may not be robust enough to detect such an infrequent event because all studies examined the hemisphere contralateral to the virus injection site. Therefore, the question is still open of whether a small population of ependymal cells has the potential to generate other cell types under normal conditions.
A specific mouse model may be necessary to ultimately address this question. For example, a transgenic mouse with an inducible Foxj1-cre would allow for more specific and robust tracing of ependymal cells. Despite the remaining uncertainty about the potential of ependymal cells under normal conditions, this study5 clearly shows that ependymal cells respond to stroke and can change fate in this context. The authors also show that although ependymal cells could change fate and enter the cell cycle, they could not self-replenish and therefore do not qualify as NSCs under this circumstance. More importantly, the authors establish that Notch signaling is required for the maintenance of ependymal cells and that alterations in the Notch pathway are responsible for the fate change and subsequent loss of ependymal cells after stroke. The physiological relevance of the response of ependymal cells to stroke remains to be explored, considering that (i) the number of ependymally derived neuroblasts is far fewer than the number of SVZ-derived neuroblasts and (ii) there is severe loss of ependymal cells in this stroke model. Nevertheless, this study adds ependymal cells and possibly other non-NSCs as targets for future therapeutic application if new methods can be developed to prevent ependymal cell loss and to promote the survival and integration of newborn cells for brain repair. The study by Carlén and colleagues5 in fact supports two emerging concepts. First, non-NSCs can be recruited into the initial steps of brain repair8. Second, there is growing evidence for the potential of neural cells to be usefully changed by molecular
volume 12 | number 3 | MARCH 2009 nature neuroscience
news and views anipulations9. In light of recent advances in m the field of induced pluripotent stem cells10,11, the change of ependymal cell fate after injury or by molecular manipulation in vivo adds to the hope for future development of therapies for brain repair, especially in the aging brain, where the number of original neural stem cells drops substantially12.
1. Johansson, C.B. et al. Cell 96, 25–34 (1999). 2. Doetsch, F., Caille, I., Lim, D.A., Garcia-Verdugo, J.M. & Alvarez-Buylla, A. Cell 97, 703–716 (1999). 3. Consiglio, A. et al. Proc. Natl. Acad. Sci. USA 101, 14835–14840 (2004). 4. Capela, A. & Temple, S. Neuron 35, 865–875 (2002). 5. Carlén et al. Nat. Neurosci. 12, 259–267 (2009). 6. Mirzadeh, Z., Merkle, F.T., Soriano-Navarro, M., Garcia-Verdugo, J.M. & Alvarez-Buylla, A. Cell Stem Cell 3, 265–278 (2008).
7. Coskun, V. et al. Proc. Natl. Acad. Sci. USA 105, 1026–1031 (2008). 8. Kohyama, J. et al. Proc. Natl. Acad. Sci. USA 105, 18012–18017 (2008). 9. Jessberger, S., Toni, N., Clemenson, G.D. Jr., Ray, J. & Gage, F.H. Nat. Neurosci. 11, 888–893 (2008). 10. Yu, J. et al. Science 318, 1917–1920 (2007). 11. Takahashi, K. et al. Cell 131, 861–872 (2007). 12. Kuhn, H.G., Dickinson-Anson, H. & Gage, F.H. J. Neurosci. 16, 2027–2033 (1996).
I can see what you see Kendrick N Kay & Jack L Gallant
© 2009 Nature America, Inc. All rights reserved.
Previous studies have attempted to decode functional imaging data to infer the perceptual state of an observer, but the level of detail has been limited. A new decoding study reconstructs accurate pictures of what an observer has seen. Can we decode a person’s brain activity to determine what that person is perceiving? Interest in this question has recently surged as a result of the success and popularity of applying multivariate classification techniques to functional magnetic resonance imaging (fMRI) data1. Standard fMRI analyses average activity across all voxels in a given region of interest and then correlate this activity with stimulus or task conditions. In contrast, classification techniques harness the entire pattern of activity observed across multiple voxels to predict which stimulus or task condition the subject is in. Classification techniques are limited, however, because they can only distinguish among a handful of predetermined states; for example, whether the subject saw a face or a house. Is it possible to overcome this limitation and obtain more detailed information about a person’s mental state? Recent fMRI studies2,3 have advanced beyond classification by using brain activity measurements to identify, out of a set of potential images, the specific image that the subject saw. One study4 even showed that it is possible to reconstruct the actual image that was seen, rather than simply choosing the image from a known set. However, the resolution and accuracy of the reconstructions in this early study were somewhat low. A new study by Miyawaki et al.5 uses sophisticated decoding techniques to achieve high-quality image reconstructions (Fig. 1). Miyawaki et al.5 began their experiment by constructing contrast-defined images4; these images consisted of a 10 × 10 grid,
The authors are at the Department of Psychology, University of California at Berkeley, 3210 Tolman Hall #1650, Berkeley, California 94720, USA. e-mail:
[email protected]
Contrast-defined image
fMRI signals
Reconstructed image
Figure 1 Schematic of visual image reconstruction performed by Miyawaki et al.5. Flickering checkerboard patterns arranged on a 10 × 10 grid were shown to each subject while fMRI signals were recorded from early visual cortex. The recorded signals were then used to reconstruct the images that the subjects had seen.
in which each element was either gray (zero contrast) or filled with a flickering checkerboard pattern (full contrast). The authors presented a large number of these contrast-defined images to each subject while simultaneously recording fMRI signals from early visual areas (V1, V2 and V3). Next, they developed a reconstruction model and fit it to their data. In the first stage of the model, the authors used linear combinations of voxel responses to predict the amount of contrast in local regions of the stimulus. This technique works well because individual voxels in early visual areas reliably signal the amount of contrast in their spatial receptive fields2,4,6. In the second stage, they combined the predicted contrasts for the various stimulus regions into a single image that represents the estimated pattern of contrast the subject saw. Finally, the authors tested their reconstruction model using separate data that was reserved for this purpose. Reconstruction accuracy was quantified by correlating reconstructed images with the actual images seen by each subject. An interesting aspect of the reconstruction model used by Miyawaki et al.5 is that images
nature neuroscience volume 12 | number 3 | MARCH 2009
are represented in terms of many overlapping regions that occur at different scales. The use of overlapping regions complicates reconstruction because it requires the estimation of the relative contributions of the various regions to the final reconstructed image. On the other hand, the use of multiple scales enhances reconstruction performance because voxels convey contrast information at different scales. For example, peripheral voxels have larger spatial receptive fields than foveal voxels2,6 and therefore convey relatively more information about the contrast of large stimulus regions. By decoding contrast information at different scales, the reconstruction model maximizes the amount of information that is extracted from each voxel. The work of Miyawaki et al.5 constitutes the latest development in a long series of visual decoding studies that have emerged over the years. Although reconstruction is qualitatively different from identification and classification, all decoding studies are similar in that they establish a systematic mapping between visual stimuli and brain activity (Fig. 2). In some studies5,7–10, the directionality of the mapping is from brain activity to the stimulus, and decoding is achieved by simply evaluating the mapping. In other
245
