Critical Analysis Of The Paper “Peroxidasin Is A Novel Target Of The Redox-Sensitive Transcription Factor Nrf2” BY HAMMER AND Mavri-Damelin
“Peroxidasin is a novel target of the redox-sensitive transcription factor Nrf2” by Hammer and Mavri-Damelin uncovers a molecular mechanism underlying the regulation of Peroxidasin (PXDN). Specifically, the paper identifies nuclear factor erythroid 2-related factor 2 (Nrf2), a major antioxidant response transcription factor, as a potential candidate to study because PXDN expression is modified by H2O2. The authors analyzed PXDN expression levels through western blotting and modification by H2O2. Additionally, they performed chromatin immunoprecipitation experiments to identify a Nrf2 binding region at the PXDN promoter and concluded with luciferase reporter assays to find a direct link between Nrf2-PXDN promoter binding and expression. The paper hopes that its findings can improve the understanding of PXDN in the context of pathophysiology and physiology. The paper, however, is not without its flaws. It could be improved by supplanting the use of bar graphs and error bars, removing unnecessary claims, and improving western blot figures.
The authors’ use of bar graphs and error bars hinders the presentation of the data. This can, firstly, be observed by the scarcity of explicitly reported statistical values in the figures or in the text. As such, the reader must analyze the bar graphs to estimate treatment means and variances. Given that the bar graphs from Fig. 1A, 2A, 3A, and 4A have a y-axis with 50 units of increment, the feat of ascertaining an exact value for these experimental results is difficult. If the authors would have explicitly reported the mean and standard deviation for each treatment, the results would be clearer.
Moreover, because the authors had small sizes for some of their treatments, the possibility of replacing bar graphs and the use of error bars in favour of plotting each treatment’s samples as dots (dot plot) could enhance the paper’s results. In this way, there would be absolute clarity in analyzing the distribution of treatment samples because individual samples could be observed. Error bars generalize the distribution of samples as a consistent variance around the mean which equates to lost information about the true distribution for treatments. When bar graphs and error bars are used, it is uncertain to which extent the experiment is reproducible. This can be seen in many of the figures where error bars appear to overlap between the different experimental treatments. In figures such as this, it is unclear whether samples had a consistent, and statistically significant result from sample to sample, or whether some samples were statistically insignificant, but the significance of the result was due to sample variance which generated a statistically significant mean. Using a bar graph generalizes the results and hides pertinent information.
The authors claim that PXDN expression increases upon H2O2 treatment in a time-dependent manner, yet the interpretation of this statement is ambiguous. True, for the 3 h H2O2 treatment, PXDN expression does increase compared to the negative control in a significant manner. For the 18 h H2O2 treatment, it too, increases but only compared to the negative control and not compared to the 3 h treatment in a significant manner. Even still, the authors claim that they found that after 3 h and 18 h of H2O2 treatment, PXDN expression increased from 40% to 48%, respectively. This is unjustified given that their analysis of these claims only show t-tests against the negative control. Also, in each of Fig. 1A and 2A, PXDN distributions at the 3 h and 18 h time periods are similar. Perhaps the authors could have just omitted the 18 h treatment’s results or made their claims clearer.
In the paper’s current state, the claim does not add to their rhetoric given its lack of validity. Better figures could have also been used to present western blot results. Because western blot results were presented by only showing the specific bands for Nrf2, PXDN, NQO1, and β-actin, opposed to presenting the blots in their entirety, information is lost. Particularly, in Fig. 3A, the positive control, NQO1, appears to have banding above the bands corresponding to it. However, because the image is cut off and there is no further explanation of this observation in the paper, the reader is only left with their own assumptions on the identity of this band. When comparing this figure’s band to the corresponding bands for Fig. 1A, 2A, and 4A, the NQO1 images in these figures do not have the apparent band above NQO1. Additionally, the top banding of Fig. 3A’s NQO1 western blot does not appear to be the bottom of any of the other presented bands – further leading to confusion about the band’s identity. If, however, the authors presented the entire western blot images, as opposed to cutting off sections of it, less would be left to the imagination and the paper’s results would be clearer. On the contrary, the reader can only trust that the author’s results are withstanding.
The paper provides a molecular explanation for the regulation of PXDN by Nrf2 and possible connections to pathophysiology and physiology. Despite the potential for its findings, the paper would have been enhanced had the results been unequivocal. Explicit presentation of results opposed to the use of bar graphs and error bars, removal of unnecessary information, and presentation of entire western blots would be a start. Nevertheless, their findings will benefit the understanding of PXDN in future research.
