Drug effects on developmentally regulated genes

As we have shown in Figure 5.1 that the changes of gene expression is not a continuous process and that different sets of genes are expressed at different time points of differentiation, we next aimed to investigate this more profoundly.

Recently, it has been suggested in murine embryonic stem cells that waves of gene expression during in vitro neurodevelopment can not only define the differentiation steps but also could represent time windows to examine developmental neurotoxic effects of chemicals (Zimmer et al. 2011a). To test if also human neurodevelopment in vitro is associated with waves of gene expression we performed a clustering analysis depending on the same regulation pattern of the PS followed by GO analysis (Fig. 5.3A).

The 104 PS up-regulated during the first 6 h of differentiation (Fig. 5.3A, developmental cluster a+) corresponded to genes that belong to rather general functional groups like

“positive regulation of cellular component”. GOs overrepresented in the 33 genes down-regulated after 6 h (developmental cluster a-) comprise GOs like “transmembrane receptor protein serine /threonine kinase signaling pathway” and “BMP signaling pathway”. After 4 days of differentiation 424 PS were up-regulated more than 2-fold compared to 6 h of differentiation (developmental cluster b+) and 440 were down-regulated (developmental cluster b-). The 45 GOs overrepresented among these middle up-regulated PS contain 13 related to neural development whereas the 6 GOs of the down-regulated PS contain 4 GOs related to circulatory system development. 164 PS at least two-fold higher at DoD6 than at DoD4 (developmental cluster c+) belong to 29 GOs that comprise 4 GOs related to neural development, which are also in cluster b+. Genes that are up-regulate only in this late phase, and therefore are good candidates for marker genes for late differentiation, comprise PAX6 and EMX2. PS that peak at 6 h of differentiation (71 regulated) or at DoD4 (20 up-regulated) do not result in significantly over-represented GOs (Fig. S7A).

Figure 5.2. : Drug effects on developmentally regulated genes. (A) Waves of gene expression during neural differentiation. Samples were prepared as in Fig. 1. Significantly regulated PS (p<0.01) were assigned to clusters according to the rules indicated above each curve. Average expression levels of all PS (their number n is indicated) belonging to the cluster are displayed. (B) Comparison of PS regulated by TSA to PS that are regulated during normal non-disturbed differentiation (as defined in Fig.1). The number of respective TSA-regulated PS at given time points, and with the indicated direction of regulation was used as 100%

reference point. (C) Overlap of developmental clusters increasing early, middle or late with TSA-regulated PS affected at 6 h (T1), 4 days (T2) or 6 days (T3). (D) Word clouds of overrepresented gene onthologies (GOs) affected after 6 h drug exposure. The character size in the word clouds is relative to the p-value of the

Taken together, most part of the neural differentiation seems to occur between 6 h of differentiation and DoD4. Additionally, after DoD4 neural differentiation proceeds (expression of genes functionally related to neural development further increases) and some genes are up-regulated only during this late phase. This wave like gene expression during human neural differentiation can be used to extract marker genes specifying the differentiation process in more detail than previously possible (Fig. S7B). In addition, this wave-like expression pattern was confirmed by qPCR analysis of some chosen marker genes that are relevant for neural differentiation or stem cell maintenance (Fig. S8).

Having identified the regulated genes specifically regulated during our time periods we next investigated if the genes regulated during normal differentiation are more susceptible to TSA.

The HDACi activity of TSA might have an impact on general chromatin structure by increasing the acetylation of histones. Generally actively transcribed genes are highly acetylated and probably have a more open chromatin structure. Therefore next we asked the question if actively transcribed genes are more susceptible to be altered by drug exposure. To this end, we first compared genes regulated by TSA (T2) to PS regulated in any developmental wave cluster. We found that only about 30% of the genes regulated by TSA (T2) (Down-regulated 32%, up-regulated 27%) are also regulated in any specific cluster during normal development (Fig. S9). In a next step, we compared PS altered by treatment to the clusters of genes that are regulated at the respective days of differentiation. We found that the overlap of clusters a, b and c with T1, T2, T3 increase from T1 to T3. After 6 h of differentiation most of the PS regulated by TSA are not the genes that are regulated under normal conditions as only 12% of the cluster a PS are also regulated by T1. At DoD4 more of the open active genes in cluster (b) are regulated also by TSA (33%). Finally, at DoD6 PS regulated by TSA seem to be a surrogate for open active genes as 93% of “open” genes (c) are regulated by TSA (Fig. 5.3B). This indicates that it is not the chromatin state in general that determines the susceptibility but more likely this could be because TSA specifically disturbs neural development and only in cluster b and c neural genes are up-regulated by TSA (as seen in Fig. 5.1C and 5.6A).

To further investigate if this could be the case, we compared genes regulated by TSA to genes regulated during development (D-genes, Fig. 5.1). Comparison of all D-genes to all PS up- or down-regulated by T1, T2 or T3 reveals that with increasing exposure and differentiation time, the overlap increases. A higher percentage of TSA-regulated genes is also regulated by

development. The overlap of PS down-regulated during development (D down) to PS regulated by TSA treatment (T up) increases with prolonged exposure from 22% of all PS up-regulated at T1, to 58% at T2 and 62% at T3. Also the overlap of PS up-up-regulated during development (D up) to PS down-regulated by TSA increases with prolonged exposure from 4% of all PS down-regulated at T1 to 21% at T2 and 27% at T3 (Fig. 5.3C). Again these findings foster the hypothesis that it is not the chromatin state in general that determines the susceptibility. However, it could also be that “open” genes are more likely to be regulated by TSA treatment and that the up-regulation of neural genes at DoD4 and DoD6 (D-genes) is decreased by TSA treatment.

However, amongst the genes down-regulated after 6 h by TSA 6 of 13 overrepresented GOs deal with chromatin modification and 7 with acetylation. All of these GOs are also overrepresented among genes down-regulated by VPA. Most importantly, in both cases genes related to chromatin modification, like ING5 or KAT5 and several PHF proteins, are down-regulated by both compounds. Thus, the primary effect of VPA and TSA is probably affecting the gene expression of epigenetic regulator genes and therefore might change the epigenetic landscape of the cells. Up-regulated GOs in both last cases comprise neuronal GOs like transmission of nerve impulse, synaptic transmission and neurological system process (Fig.

5.3D) but not these clear neuro-developmental GOs that we found for T2 and T3 (Fig. 5.1D).

Thus, the GO analysis revealed a first hint that after 6 h of TSA treatment epigenetic mechanisms could be involved to initialize the massive changes observed later after prolonged differentiation.