Background
Pseudogenes are defective copies of genes that evolve neutrally. Pseudogenes originating from protein-coding genes lack the ability to code for proteins and bear features of coding sequence decay, such as:
An intriguing and a basic aspect that remains yet unknown is whether sequence length plays any role in the evolution of pseudogenes. If so, is such an effect common to both basic categories of pseudogenes (
Results and Discussion
The proportion of protein-coding genes that produced nonprocessed pseudogenes was found to increase with parental gene length (Fig.
Percentage of protein-coding genes producing nonprocessed pseudogenes in the various length categories
Percentage of protein-coding genes producing nonprocessed pseudogenes in the various length categories. (a) For human and mouse combined, (b) for human, and (c) for mouse.
Percentage of parental proteins (human+mouse) associated with processed pseudogenes (retropseudogenes) in the various length categories
Percentage of parental proteins (human+mouse) associated with processed pseudogenes (retropseudogenes) in the various length categories.
These trends are explainable as follows. Under a neutral evolutionary scenario, longer sequences are more likely to accumulate deleterious mutations than shorter ones. This seems to be the case in nonprocessed pseudogenes. A similar effect has been noticed in protein-coding genes associated with hereditary diseases
Conclusion
This study demonstrates that the occurrence of pseudogenized gene copies is a function of gene length. Parental genes encoding for proteins longer than 1000 aa are 6 times more likely to produce nonprocessed pseudogenes than processed ones.
Methods
The annotations of pseudogenes were obtained from pseudogene.org
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
ANK performed the analyses. ANK and PMH interpreted the results and wrote the paper.
Reviewers' comments
Reviewer's report 1
Dr. Dan Graur
Department of Biology and Biochemistry, University of Houston, USA
Accepted for publication with some stylistic suggestions (not for publication).
Reviewer's report 2
Dr. Craig Nelson (nominated by Dr J Peter Gogarten, University of Connecticut).
Molecular & Cell Biology, University of Connecticut, USA.
In this study the authors describe a relationship between the protein coding length of a gene and number of RNA and DNA mediated duplicate pseudogenes derived from each gene. They find that long genes tend to produce fewer RNA-generated pseudogenes than do shorter genes.
The data presented appears sound and the core finding valid. I recommend accepting for publication following minor revision.
Several suggestions follow for possible improvements to the manuscript.
Major suggestions:
1) Both RNA-mediated and DNA-mediated duplication events give rise to duplicate genes that may become pseudogenized over time. Referring to DNA-mediated events as duplications and RNA-mediated events as something other than duplications does not reflect this fact. I urge the authors to change the way this is presented in the text.
Unlike gene copies that arise from DNA-mediate duplication, copies that arise following a RNA-mediated duplication or retrotransposition are essentially functionless at the time of birth, because they do not retain the parental promoter regions for their immediate transcription. Hence, in this study, we considered retrotransposition as a distinct event generating retrotransposed pseudogenes (retropseudogenes). Only some of the retrotransposed copies are able to recruit new promoters over the time to become functional.
2) "Processed" and "Non-processed" are not intuitive terms for those outside the field and, while these terms are correct, I suggest that the authors adopt more descriptive terms like RNA-mediated and DNA-mediated duplications, and/or retrotransposed pseudogenes.
3) No clear distinction I made between the duplication event and the pseudogenization event. Some discussion about which of these events are detected and analyzed by the authors and what impact this might have of the core finding would be welcome.
4) In Materials and Methods section, more specific data sources and preprocessing methods should be specified. For example, which Ensembl release was used, and was any filter for pseudogenes and protein coding genes applied? The numbers of pseudogenes and protein-coding genes in the paper are quite different from the pseudogene data from Pseudogene.org and the protein-coding genes from Ensembl.org. For example, the protein-coding genes listed in Ensembl (release 55) are around 22,000 but the number in the text is 46,689.
The figure 46,689 is for the number of human proteins (< = 2000 aa). We have now included cases with sequence length >2000 aa and have corrected the number in the text accordingly.
5) Is the trend same with human and mouse genomes analyzed separately? Any reason to put them together?
Minor suggestions:
1) From the figures, it is not easy to see that the longer parental genes (>1000 AAs) are 6 times more prone to produce non-processed pseudogenes than processed. Figure or Table might help the cause.
2) In Fig
corresponding pseudogenes?
3) In the discussion the authors mention: "Because the probability of interruption in
the transcription of parent genes and subsequent reverse transcription during a retrotransposition event is higher for longer genes than for shorter ones". It might be worth mentioning here that transcript abundance also has a large effect on this probability and that transcript abundance, gene length, and the abundance of retrotransposed pseudogenes are all correlated.