This web page was produced as an assignment for an undergraduate course at Davidson College.

Garrett Smith's Genomics Web Assignment 2

A Review of Song, et al's "Deep RNA Sequencing Reveals Novel Cardiac Transcriptomic Signatures for Physiological and Pathological Hypertrophy"

(Note: summary highlights for certain divisions of this review are presented in in bold. Further elaborations of the study and interpretations of figures are presented in normal type).

Summary

    This study by Song, et al. (2012) addressed the differential expression of RNA between cells of hearts (i.e. the heart cells' transcriptomes) having experienced either physiological hypertrophy (PHH) or pathological hypertrophy (PAH). Both of these types of hypertrophy lead to similar morphological changes in cardiac tissue; however, PHH shows beneficial contributions to cardiac function, whereas PAH is associated with progressive declines in cardiac function. This study thus sought a deeper understanding of the molecular differences in the two hypertrophy processes, with the hope of potentially highlighting certain key molecular players in PAH pathology.

    This study expanded upon studies of transcription patterns in these cells found using microarray – a technique with certain limitations such as low sensitivity – by utilizing RNA sequencing (RNA-seq), a newer technique found to be superior to more traditional microarray techniques in sensitivity, accuracy, and reproducibility. RNA-seq, also called whole transcriptome shotgun sequencing (Iacobucci, et al., 2012), involves sequencing of cDNAs obtained from RNAs using high-throughput sequencing technologies to precisely discern the magnitude of expression of certain RNAs and thus genes (Nagalakshmi, et al., 2010).

    Cardiac tissue from four groups of mice were analyzed in the study. In the exercise group, mice were made to swim for progressively longer periods over four weeks. Mice in the transaortic constriction (TAC) group underwent surgery involving ligation of the transverse aortic arch (Cha, et al., 2008), and sham mice went through the same surgical preparation as the TAC group with the exception of aortic ligation. Sedentary mice did not undergo any forced exercise or operation of any kind. RNA sequencing was performed utilizing all heart tissue from a given animal.

    Analysis of differentially expressed genes (DEGs) between PAH and PHH cardiac cells showed that genes and signaling pathways involved in PHH were highly distinct from PAH, with PAH cells showing a marked increase in DEGs relative to PHH cells. The results from RNA-seq as used in the study, were found mirror the results of RT-PCR, substantiating the utility of RNA-seq, but comparisons of RNA-seq to microarray results revealed significant fdifferences between the two, with microarray being much less sensitive to the variety of genes found through RNA-seq. Additionally RNA-seq was able to identify variations in alternative splicing on top of DEG, using the same single approach - a feat that would be challenging for microarrays.
    The transcription factor FOXM1 was implicated in the PAH pathogenesis, as was found through various manifestations of its associations with other genes also found to be linked to the PAH transcriptomic profile. Patterns in alternative splicing (AS) of RNAs were found to differ between the two hypertrophied tissue types as well, which may have resulted from differential splicing factor expression. Groupings of DEGs and alternatively spliced genes by function revealed interesting qualitative trends as well, revealing, for example, that genes upregulated in PAH tended to be involved in immune function and cell cycles.
    Overall these findings provide a rather comprehensive overview of differential transcriptomics of PAH and PHH – information predicted to be of potential use in understanding the mechanisms of each of these types of cardiac hypertrophies, potentially leading to new avenues for disease management or promotion of greater cardiovascular health.

Personal Opinion

    I found this report to be an excellent example of the utility of RNA sequencing as a more sensitive means of identifying DEGs between two tissue types. The RNA-seq technique was reported to be more sensitive than and consistent with microarray, a technology assumed reliable in studies of genomics, just as previous studies found with RNA-seq (Nagalakshmi, et al. 2010; Song, et al., 2012), and demonstration of the technique's utility compared to that of preexisting techniques seemed well supported. In most cases analyses were both thorough and multifaceted (both qualitative and quantitative on several levels), inspecting patterns ranging from genes' promoter segments to their alternative splicing. A more thorough set of comparisons to older techniques (as was given in figure 3C), however, could have been conducted or at least referenced to more clearly verify RNA-seq's utility relative to more trusted techniques, as only 8 genes were reported to have been verified by both techniques.
     I imagine the same RNA-seq approach can and will be used on a larger scale to identify molecular differences between normal and diseased (or disease susceptible) tissue types for other conditions in which morphological differences in normal versus diseased tissue are not obvious (Chohn's disease, for example). The fact that RNA-seq could be used to sequence both differentially expressed genes and alternatively spliced genes using the same assay was particularly noteworthy, especially considering how accurate and complete the RNA-seq methods was in comparison to older techniques such as microarray. I also imagine that the limitations addressed in the study for the RNA-seq approach (limited available software, bioinformatic algorithms, and standardization) were not particularly substantial in that they will likely resolve in time as this technique becomes more widely practiced and a greater infrastructure develops around the use of RNA-seq. The RNA-seq technique, which, as reported, can even detect de novo transcripts and long, non-coding RNAs (unlike most microarrays) may show promise for completely supplanting microarray where applicable, providing an elaborate yet elegant approach to understanding and comparing the genomics both within and between systems.

    There were a number of mostly minor points of concern I had regarding the reporting of some information through the figures, however, listed below in the order in which they appeared in the article:

It was a bit disorienting for figure 1 to begin a third row of the flow chart that did not clearly link it with the far right side of the upper rows. It would have been helpful for the figure to have been arranged as a single continuous flowchart. If arranged vertically, this chart could have probably used about as much space as figure 4.
Figures 2B and 3A do not include labels for their charts' y-axes, and though the "degree of expression" as the figure caption describes the vertical measures can be used to surmise the meaning of the data portrayed, I feel it would have helped to include a more explicit description of what this figure axis was portraying.
The table left a few items vague as well. It is not obvious as to whether the "function" column refers to the function of the normal splice variant in control cases (presumably so) or the induced function of the splice variant in one of the hypertrophy states, or both. "Normal gene function" may have been preferable here. Other aspects of the chart generally fall into understanding once patterns are observed and when relevant terms (e.g. "domain") are defined. Given that the table was constructed almost entirely from previous studies' data, and only a subset of its data was directly relevant for other figures (specifically figure 3 which only references a few of the AS genes from the table, and includes mostly AS genes uncovered in the study itself) it also seemed that the table could have been made supplementary.
Using Figure 3 and the methods, I could not discern how the 8 genes selected for further analysis and verification for alternative splicing were selected. The paper did not reveal a particular reason for choosing these particular genes from the subset of 470 and 387 genes containing alternative exons in PAH and PHH respectively. This may mean the genes were chosen randomly, the authors neglected to either report their full findings for all of the genes, or a rationale for selecting these genes existed but was not reported.
Also in figure 3, I was left curious as to why all four experimental groups were not included in every histogram in B, as they were in the final two histograms; perhaps these data were not notable to the authors, but this is not clear. Additionally an inconsistency is also apparent in the continuity of these figures' labels; the histogram corresponding to the gene Cxxc5 has labels for TAC and sham, but the adjacent RT-PCR blot contains labels for sedentary and exercise. This is opposite of the pattern for other rows, where TAC and sham were labeled similarly in the histograms and blots. The figure caption notes that Cxxc5 variants were identified in PHH, but I believe this information was just inexplicably excluded from the histogram.
Figure 4 makes what I thought was a minor error in not explicitly noting whether "multilateral pathway analysis" (a term I could not find used in any other published paper) refers to analysis of KEGG (Kyoto Encyclopedia of Genes and Genomes) information. Though I believe that it likely does, this led to some confusion during interpretation.
FOXM1 was implicated in a number of figures to be involved in the pathogenesis of PAH, but an unsatisfying discussion was provided regarding follow up studies that might be performed to manipulate this gene or its protein to alter the outcome of PAH.

Overall these criticisms were relatively minor, not greatly obscuring the central message of the article: that despite morphological similarities in two different tissues – given in the case study between PAH and PHH cardiac tissues – differences in functional outcomes (i.e. health) may be largely explained through differential gene expression, and RNA-seq provides a thorough and accurate way of elucidating these genetic differences relative to previously available techniques. The progression of ideas throughout the study is reflected in the progression of its figures and tables. Click below to visit the first figure and see an overview of genetic differences identified between cardiac tissue of PAH and PHH models.

Proceed to Figure 1

Figure 2

Table 1

Figure 3

Figure 4

References

Iacobucci, I., et al. (2012). Application of the whole-transcriptome shotgun sequencing approach to the study of Philadelphia-positive acute lymphoblastic leukemia. Blood Cancer Journal, 2(3), e61.

Nagalakshmi, U., Waern, K. & Snyder, M. (2010). RNA-Seq: a method for comprehensive transcriptome analysis. Current Protocols in Molecular Biology, Chapter 4: Unit 4.11.1-13.

Song, H. K., Hong, S. E., Kim, T., Kim D. H., et al. (2012). Deep RNA sequencing reveals novel cardiac transcriptomic signatures for physiological and pathological hypertrophy. PLoS One, 7, e35552.

Return to Garrett's Home Page

Genomics Page Biology Home Page

Email Questions or Comments to gasmith@davidson.edu.