Distinct and separable roles for EZH2 in neurogenic astroglia

1) We are not sure if the microarray/ChIP-Seq analysis adds much to the paper. It seems that at least some in vivo validation of a few key targets in their system would strengthen this part of the paper. As it stands, it is simply a list of genes with correlated functions. If these data are retained, the construction of the manuscript would benefit from presenting the microarray analysis earlier and then investigating the mechanism of action of EZH2 focusing on Ink4a/Arf and Olig2.

We agree with the sentiment of the reviewers. Since this analysis is not a primary focus of this paper, we have converted the microarray/ChIP-seq data into a figure supplement and have substantially reduced the text related to these findings (now summarized in one short paragraph). We believe that the manuscript is now overall more concise and appropriately more focused on the mechanism involving Ink4a/Arf and Olig2.

2) The analysis of early post-natal brain is interesting, but the rationale for doing these experiments is weak. The adult mouse SVZ is not necessarily a model for developing human LV neurogenesis and the authors should be more cautious here.

The reviewers appropriately point out that there are important differences between the mouse and human SVZ, and we have accordingly revised the text. For instance, we have removed our previous comparisons between findings of the human and mouse SVZ. The Results and Discussion sections describing the human postnatal brain analyses have been greatly streamlined. By making such revisions to the text, we believe that our descriptions of these results are now more circumspect.

3) Similarly, it is not logical that the authors repeatedly refer to adult neural stem cells as a subtype of astrocyte, when the whole point of the study is to set about demonstrating how the NSC differ from bona fide astrocytes. The reasons presented in the paper for referring to adult NSC as a subtype of astrocyte do not seem very compelling and the findings of the present study in fact make this claim even less compelling. Based on available functional and molecular evidence, including the findings presented in this study, adult NSC seem to have far more in common with developmental radial glial NSC than they do with mature astrocytes. Mature differentiated astrocytes have many specific functions in the CNS, such as maintenance of ion and water homeostasis, or interactions with synapses in grey matter, and many more. There is no evidence that adult NSC exert any of these astrocyte functions. If such evidence exists, the authors should refer to it. What reason do the authors have for not referring to adult NSC simply as adult radial glia or as a subtype of radial glia? Is this not more logical, based even on the findings of the present study? What purpose does it serve to call adult NSC a subtype of astrocyte? To do so seems only to confuse the differences between functionally distinct cell types. Astrocytes are a principal cell type of the CNS. In no other organ system is a stem cell referred to as a sub-type of a tissue specific principal cell. Better reasons for the repeated claim of adult NSC as an astrocyte subtype should be presented, if indeed they exist.

Astrocytes were originally named for their star-like morphology, and the heterogeneity of this glial cell population has been underappreciated (Zhang and Barres, 2010). We believe that our work illustrates how epigenetic regulators such as EZH2 may contribute to this cellular diversity. As pointed out by researchers in the astrocyte research community, while ion and water homeostasis and synapse regulation are indeed important roles for astrocytes, it is now recognized that some astrocytes can have other roles including the function as neural stem cells (Zhang and Barres, 2010; Wang and Bordey, 2008; Robel et al., 2011). Importantly, distinct subtypes of this glial cell population likely contribute to the diversity of astrocyte functions.

Adult SVZ NSCs have not been referred to as radial glia due to their non-radial orientation, non-bipolar cellular morphology, and lack of contact with the pial surface of the brain. Instead, NSCs in the SVZ have been defined as astrocytes by a number of different criteria. Briefly, SVZ astrocytes have many morphological and ultrastructural similarities to mature astrocytes (Doetsch et al., 1997), have the presence of glycogen granules (Peretto et al., 1999), and have endfeet contact with blood vessels (Mirzadeh, et al., 2008; Shen et al., 2008). There are also similarities in their electrophysiological properties (Filippov et al., 2003, Fukuda et al., 2003). Furthermore, SVZ astrocytes express multiple genes associated with astrocyte identity and function including GFAP, GLAST, AQP4, Connexin 30, and GLT1. Of note, AQP4 is a water channel protein that is key to the regulation of water homeostasis, and this protein is expressed prominently by SVZ astrocytes and is dynamically regulated during neurogenesis (Cavazzin et al., 2006). GLAST and GLT1 are glial glutamate transporters that are involved in the regulating the extracellular concentrations of glutamate and are expressed in GFAP + cells in the SVZ (Liu et al., 2006).

4) In Figure 1, the authors should explain why Olig2 and Ezh2 are co-expressed, when later in the paper they show that Ezh2 represses Olig2. This might confuse readers who are not experts in gene regulation, etc.

We agree that these results might be confusing to readers who are not familiar with epigenetic mechanisms of transcriptional regulation. We have therefore added text to the Discussion related to this interesting finding:

“In addition, how EZH2-mediated repression is targeted to specific loci is still not well understood. In a subpopulation of SVZ cells, EZH2 was co-expressed with OLIG2 (Figure 1E), suggesting that while EZH2 is present in these cells, its repressive activities have not been localized to the Olig2 locus. Future work may reveal how other factors, possibly including long non-coding RNAs (Simon and Kingston, 2013), may serve to dynamically target EZH2 to specific loci at different states of the neurogenic lineage.”

5) In Figure 2, they should confirm the presence of SVZ populations (i.e., B, C, A cells) in the Ezh2-cKO mouse. Given that the hGFAP-Cre is active relatively early, it is important to confirm that there are no developmental defects in these mice.

We have performed new analysis to address this reviewer comment that we have added as Figure 2–figure supplement 3. As we reported in the previous manuscript, there was a 4-fold decreased in the number of DCX + type A cell neuroblasts hGFAP-cre;Ezh2^F/F mice. However, as we added in the text:

”This decrease was not due to a developmental defect in the SVZ, as we did not find any significant differences in the type C cell (DLX2+, DCX-negative) population nor a deficit in the type B cell (GFAP+, Nestin+) population in hGFAP-cre;Ezh2^F/F mice (Figure–figure supplement 3)”.

Thus, we do not believe that developmental defects primarily account for the observed reduction in SVZ neurogenesis.

6) For Figure 3, they should include the figure supplement 1B/C in the main figure. Having the ARF/Ink4a expression and ChIP-Seq data in the body of the paper will help reader with the rationale for this key experiment.

We agree with this comment and have moved Figure 3–figure supplement 1B–C to Figure 3A–C with the other parts moved to Figure 3D–H. References in the text have been changed accordingly. We thank the reviewers for this helpful and clarifying suggestion.

7) The Olig2-shRNAi experiment in vitro is nice. It is important to do this experiment in vivo; inject Olig2-shRNAi virus into the SVZ of P16-null;Ezh2-null mice and assess whether neurogenesis is restored.

We agree that an in vivo demonstration of the rescue of the neurogenesis defect observed in Ezh2-deleted cells by Olig2-shRNA virus would be of interest. Unfortunately, although the experiment is conceptually quite straightforward, from past experience, we estimate that this experiment would take more than six months to perform based on the following reasons:

Each hGFAP-cre;Ezh2^F/F;Ink4a/Arf^-/- mouse has 5 transgenic or mutant alleles, and this is further complicated by the fact that the hGFAP-cre transgene cannot be reliably crossed in from the female. Thus, this complex genotype is quite time-consuming to generate.

Currently, we have very limited numbers of animals that can serve as the parents of the required crosses. It will then take a few more successful breedings to generate the numbers of mice (4–6 per group) usually required for stereotactic injection experiments. Then, we would need to wait 60 more days to perform the shRNA injection in the adult SVZ. We thus anticipate that it would take four or more months to even attempt this experiment once.

Furthermore, in our experience, these types of experiments (stereotactic injections into complex genetic backgrounds) require several experimental groups and repetitions to generate reliable results.

In addition, because stereotactic injections of viruses does not selectively infect the neural stem cell (NSC) population (in fact, mostly migratory young neurons are infected), the analysis will also require birthdate analysis and a differentiation time course, which may require even greater numbers of mice (e.g., 4–6 animals 2–4 days after injection and also 4–6 animals after 1 or 2 weeks after injection).

In summary, if this experiment were to be included, we believe that the issues outlined above would cause a significant delay in the publication of this manuscript.

8) Loss of Olig2 results in increased astrocyte production (Zhou, et al. Cell 2002). In the Olig2-shRNAi studies, they should assess astrocyte production as well.

In Zhou et al. (Cell 2002), the authors show that loss of Olig2 in oligodendrocyte progenitors of the embryonic spinal cord increases astrocyte production. In our studies, we showed that Olig2 knockdown of aberrant Olig2 expression in Ezh2-deleted SVZ cells increases neurogenesis. We have performed new experiments to address the reviewers’ question about astrocyte differentiation in our experimental system and added it to the text.

“Ezh2^Δ/Δ;Ink4a/Arf^-/- SVZ NSCs with Olig2-knockdown produced nearly 3–fold more Tuj1+ cells after 3 d of differentiation as compared to cells infected with the control vector with no significant increase of GFAP-positive cells (Figure 4H; Figure 4–Figure Supplement 2E and 2F).”

9) The co-staining in Figure 6 is confusing. In the paper they make the point that Ezh2 is downregulated once the neuroblasts reach the OB and differentiate in to neurons. This is correlated with their observation that Ezh2 expression is generally downregulated in the human after neurons have been produced; I would streamline this part of the paper to simply make that point.

We have worked to clarify this section of the manuscript. We recognize that the adult mouse SVZ is not necessarily a model of postnatal human SVZ neurogenesis, and so we have simplified the Results section as suggested by the reviewers.

10) EZH2 is expressed in most (all?) cell types where it plays an active role in regulating cellular programs. The observed pattern of expression with exclusive presence of EZH2 in neurogenic niches thus likely represents an effect of a threshold of detection by the antibody. The wording of the manuscript should cautiously convey this point rather than assuming that cells outside of SVZ are devoid of EZH2 expression.

This is an important point, and we have changed our wording accordingly. For instance, the first subheading of the Results was changed from “EZH2 expression in the postnatal brain is progressively restricted to SVZ astroglia and their neurogenic lineage,” to “Robust EZH2 expression in the postnatal brain is retained by SVZ astroglia and their neurogenic lineage.” Other wording changes clearly indicate the IHC methods used for these findings.