This web page was produced as an assignment for an undergraduate course at Davidson College
Review of Paper 1:
Deep RNA Sequencing Reveals Novel Cardiac Transcriptomic Signatures for Physiological and Pathological Hypertrophy (Song et al. 2012)
Away with the microarrays...
Image courtesy of Wikimedia Commons
Summary & Opinion
This was an interesting article to read directly after learning about microarrays in class because, at the heart of it, this article highlights the benefits of the RNA-seq technique and examines why it determines differential gene expression with “high sensitivity, accuracy, and reproducibility” in comparison to the traditional microarrays (Song et al. 2012). For clarifications sake, RNA-Seq is a technique where all mRNA from a certain cell is replicated and coded and then compared to the host’s genome while microarrays use probes that hosts mRNA binds to on a chip, which is then processed. The secondary function of the paper was to try to generate data on the differences between genes expressed, their regulation pathways, and potential functions in two heart conditions (physiological hypertrophy (PHH) and pathological hypertrophy (PAH)) using said technique. The scientists generated a large amount of data presented through the entire article that not only addressed the questions above but also highlighted the key advantages of RNA-Seq over microarrays. The most important attributes include its higher sensitivity to amount of differential genes detected, the ability to determine the magnitude that a certain gene was transcribed (no use of ratios), and the ability to determine alternative splicing. Such characteristics were not just assumed but rather shown through their generation of data that was also relevant in investigating what altered gene expression led to the generation of PAH versus PHH. At the end, the article provided a concise overview of the important differential genes and their regulation that they determined and provided the scientific community with an impressive amount of data that could be further experimented for better understanding of the disorders. The final conclusion was that RNA-Seq was a more accurate technique that could provide more data than a microarray and therefore should become more widely accepted and standardized within the scientific community.
Personally, I though that this article provided compelling evidence that RNA-Seq was in fact a much better technique. It highlighted how RNA-Seq could: one, generate data on the actual intensity of gene product by doing away with ratios, and two, be used to detect alternate splicing. These two attributes are the major drawbacks of microRNA analysis and thus I thought it was impressive that they could display this while providing meaningful insights to the transcriptomic differences in PAH and PHH. After this class, I have certainly learned not to take things at face value but the information was very compelling and well documented. In the results of RNA-Seq were confirmed as being accurate by using RT-PCR and other techniques however I felt that Figure 3 did not accurately portray this. In fact, I found it in some ways misleading and I second-guessed for a second the accuracy of the results. However, the further evidence does show that the RNA-Seq results are confirmed using older techniques that are familiar and trusted in the scientific community
The article was very long but textually easy to follow. There were very few times where I got lost in the scientific jargon, which is a compliment to the authors on their ability to concisely explain everything. Furthermore, when I did the material and methods at the back of the paper was useful in trying to deduce the techniques and specifics that they referenced to. Figure-wise, I thought the authors could have improved their article solely by putting relevant figures near or on the page in which they discussed it. Spacing is always and issue, but I found it hard to really understand the data while having to flip through three pages to find the figure and then flip back to read about it again. Furthermore, some of the Figures were a little confusing to read and I did not like how the heatmap color changed from red to blue between Figure 2 and 4.
There was a lot of data thrown at the reader, and in that way it was hard to absorb details on particular genes and their functions. However, I conjecture that the Hong et al. could have presented ten more pages on their findings from such a broad search and thus I think they did an excellent job of giving enough information to illustrate their point as well as give data to the scientific community to continue research. All in all, I thought it was a good article.
Analysis of Figures
A. Panel A is a visual representation of the expressed genes shared by all the mRNA reads for each individual control and experimental group. The use of a Venn diagram shows the number of genes detected by each individual mRNA read with the overlap of the individual circles displaying which genes were detected multiple times. This was necessary to show that the gene expression was not unique to the individual mouse but rather shared between the group and thus worth further examining.
B. Panel B examines which genes were differentially expressed between Sham(control) vs. TAC (PAH) and Sedentary vs. Exercise (PHH), put more simply as what genes were unique to hypertrophy. It was further determined if the genes were different due to upregulation (increased product) or downregulation (decreased product). Of the genes unique to hypertrophy (2,045 of them), Hong et al. then determined those that were being transcribed in the same or opposite direction in PHH versus PAH. Results from PHH and PAH mice were compared and results showed significantly differentiated regulation between PAH and PHH Figure 2
A. Hong et al. utilized Ingenuity Pathway Analysis database to generate a “top scoring network” or the most likely way that the genes probably interact for the 417 that were upregulated in PAH to an extent that was not observed in any other group of mice. The results provided this network where the most abundant genes were cell cycle regulators, forkhead box, and polo like kinase 1 or the genes circled in blue that were further and more extensively examined in the paper. All of the red nodes are the genes that were upregulated and their shapes indicate what type of product the gene produced. I am going to assume the shade of red indicates the amount of increased expression but I wish the authors had included this in the legend or provided the scale.
B. This panel provides histograms or visual interpretations of the expression of different isoforms of these three genes (Plk1, Foxm1, E2f1) expressed in the four groups of mice studied. The blue blocked lines at the bottom display the exons and the length the vertical lines shows increased expression. At each of these exons, the PAH mice longer lines providing further evidence that these genes are being transcribed more often.
C.Hong et al. went on to exam the 1,000 bp upstream of the 417 genes looking for similar transcription factor motifs. Motifs for FOXM1 and PU.1 were identified as seen in the figure where they display the sequence and its prevalence based on letter size. They showed these motifs were specific to PAH through a heat map that again using RPKM value to display that its expression was much more prevalent in the TAC or PAH mice.
D. Panel D further investigates the FoxM1 motif by looking at its targets, or genes that it potentially turns. These genes were coupled with previous evidence from other studies. The amount of gene expressed was again shown in a heatmap using RPKM values and higher values were seen in PAH mice.
Hong et al. looked at previous studies to determine what/how differences alternative splicing between PAH and PHH proteins may have affected function. They determined three major categories of changes in protein: domain gain or loss, activity change, and localization.
In general, all panels of Figure 3 were used to show the experimental evidence that confirmed the accuracy of the RNA-Seq in determining alternative splicing through detecting exon exclusion and inclusion. Ten exon variants for eight genes were used as samples to then be validated through RT-PCR assays, which would then be compared with the results from the RNA-Seq.
A. Here is another histogram similar to the one Figure 2.B. This histogram displays the different reads of eight genes. The blue lines with boxes at the bottom of each histogram shows the different exon combinations for different isoforms that can be formed from the genes. The asterisks indicate at which exon there is differential expression or alternative splicing at each gene that was examined. Sometimes differences were only observed between the Sham and TAC phenotypes while others were observed across all four groups of mice (SHAM, TAc, Sedentary, and Exercise).
B. Panel B shows bar graphs (one for each gene examined) as representation of the distribution or amount of the different isoforms found between whichever mice groups showed variation of exons in Panel A for that gene. The “amount” again was in values of RPKM values. The top 6 graphs showing the differences between Sham and TAC are good visualizations with small error bars of exon variation. However the bottom two where all four groups are included, there are very few discernable trends and large error bars pointing to the fact that the data here may have been more inaccurate.
C.Panel C shows the RT-PCR results, specifically probing the different exons(labeled by number that corresponds to numbered exons on the histogram) of the eight genes where they varied in the mice groups. Unfortunately there is no reference bar which would be beneficial in determining specific differences. Visually there are cases where one can see a clear distinction in expression however there are many where it is not extremely obvious. In general this panel provides some evidence that the results are legitimate.
This final figure relates the data back to the differences between PAH and PHH gene expression, connecting the results with the broader picture. This graph which looks very similar to the classic microarray results groups pathways that were enriched or had at least five differentially expressed genes or five exon variants between PAH and PHH phenotypes. The different pathways are grouped by similar function and represent muscle contraction and metabolism; immune function and cell cycle; autoimmunity; cell signaling and finally pathways related to cardiac disease respectively (I-V). The intensity of differential expression is determined by RPKM value and visually shown through color heat map. How the genes are differentially expressed (whether it be through up or down regulation or alternatively splicing) is shown in the different columns. The results are pretty staggering showing differential expression in all categories. This graph accurately shows the extent of the variation across the genome in the two disorders and the potential pathways to be further investigated.
1) Song HK, Hong S-E, Kim T, Kim DH (2012) Deep RNA Sequencing Reveals Novel Cardiac Transcriptomic Signatures for Physiological and Pathological Hypertrophy. PLoS ONE 7(4): 1-13. Acessed 26 Feb. 2013. <http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0035552>
Katie Gwathmey's Home Page
Email Questions or Comments to firstname.lastname@example.org
© Copyright 2013 Department of Biology, Davidson College, Davidson, NC 28035