On 2014 Aug 20, Claudiu Bandea commented:

Just a few days after I posted the comment above on the remarkable study by Sauvageau et al. published in eLife (Sauvageau M, 2013), in which I questioned the interpretation and validity of some of the results, Bassett et al. published in the same journal an article outlining a series of highly relevant "considerations" when investigating the biological functions of lncRNA (Bassett AR, 2014). Interestingly, Bassett et al. expressed similar concerns about the study by Sauvageau et al., which should incite a response from the authors and eLife reviewing editor, Danny Reinberg.

On 2014 Aug 08, Claudiu Bandea commented:

Multiple knockout mouse models reveal that some lincRNAs might be required for life and brain development

In a recent series of mini-essays on genome functionality and ‘junk DNA’ prompted by Ford Doolittle’s article “Is junk DNA bunk? A critique of ENCODE” (1), I used the article by Sauvageau et al. (2) as an example of poor use of words in communicating scientific observations , which could lead to confusion and misrepresentations (3). In fact, I went so far as to suggest that some of the results in this particular study might be problematic. However, these allegations might be wrong and, therefore, it is important to bring them to the attention of the authors here, so they have the opportunity to respond.

First, I would like to emphasize that the study by Sauvageau et al. is impressive. The problem, however, might be with the interpretation of the results and their integration into the existing body of knowledge, which are as important in advancing science as generating data, if not more. There might also be a problem with the absence of appropriate controls for the knockout mouse models, which questions the validity of some of the results.

Briefly, to investigate the functional relevance of long intergenic noncoding RNAs (lincRNAs), Sauvageau et al. developed a collection of 18 lincRNA knockout mouse strains in which the locus was maintained transcriptionally active. In order to select the best candidates for functional lincRNAs, the authors “have implemented a generalized and logical lincRNA candidate selection process that leverages a collection of cell-based functional assays, RNA-sequencing data and computational analyses….”

Remarkably, in their initial characterization of the 18 lincRNAs knockout strains, Sauvageau et al. found three strains that exhibited peri- or post-natal lethality and two additional strains with distinct developmental defects. Based on these findings the authors concluded that “This study demonstrates that lncRNAs play critical roles in vivo…” and entitled their article: “Multiple knockout mouse models reveal lincRNAs are required for life and brain development.”

However, both the conclusion and the title of the study by Sauvageau et al. can be misleading. Factually, the results reported in the article indicate that 5 out of 18 lncRNAs included in study appear to play critical roles in vivo and, therefore, it should have been entitled something like: “Multiple knockout mouse models reveal that some lincRNAs might be required for life and brain development.”

Moreover, based strictly on the results presented in the article, which showed that only 5 of the 18 lncRNAs were functional, and considering the strong bias introduced in the study by selecting lincRNAs that are most likely to be functional, a more appropriate scientific interpretation of the study would be that: “The results presented in this article indicate that most lncRNAs (i.e. 13 out of 18) do not appear to play critical roles in vivo….” It is possible that the authors will report in future articles results showing that some additional, or all the lincRNAs in their collection are functional; even then, the authors should be cautious about the broad interpretation of the results considering the bias of their collection toward putatively functional lincRNAs (therefore, the inclusion of a few random lincRNAs would have benefited the study).

Whether the assertion about the broad interpretation of the results, which might be considered an 'innocent' case of ‘hype in science’ (4), resonate with the authors or with other readers remains to be seen: however, the allegation that the phenotypes observed in their study might be associated with untargeted genome alterations introduced inadvertently during the procedure of generating the knockout mice, rather than with the deletion of specific lincRNAs, questions the validity of the study, at least for the time being and, therefore, it should be of concern.

Considering the well-recognized problems with generating knockout animal models (see, for example, Refs. 5-9), it is surprising that Sauvageau et al. did not discussed this issue in the context of their study, and that they did not include appropriate controls (e.g. duplicates) at least for some of the lincRNAs knockouts, preferably using mice with different genetic background. Although the data and observations presented by Sauvageau et al. might be correct, for the time being their validity remains questionable; therefore, it would make sense for the authors (and the editors) to add an explanatory note, both here at PubMed Commons and in the journal eLife.

References

(1) Doolittle WF. 2013. Is junk DNA bunk? A critique of ENCODE. Proc Natl Acad Sci USA., 110:5294-300. Doolittle WF, 2013

(2) Sauvageau M. et al., 2014. Multiple knockout mouse models reveal lincRNAs are required for life and brain development. eLife (DOI: 10.7554/eLife.01749) Sauvageau M, 2013

(3) Bandea CI. 2014. Everlasting confusion on ‘functional DNA’ and ‘junk DNA’ in the field of genome biology. PubMed Commons (National Library of Medicine; Bethesda, MD). Comment on: Doolittle WF, 2013

(4) Maderspacher F. 2014. Hype in Halifax. Curr Biol. 14;24(8):R298-301. Maderspacher F, 2014

(5) Wolfer DP, Crusio WE, Lipp HP. 2002. Knockout mice: simple solutions to the problems of genetic background and flanking genes. Trends Neurosci. 25:336-40. Wolfer DP, 2002

(6) Crusio WE1, Goldowitz D, Holmes A, Wolfer D. 2009. Standards for the publication of mouse mutant studies. Genes Brain Behav. 8(1):1-4. Crusio WE, 2009

(7) Eisener-Dorman AF, Lawrence DA, Bolivar VJ. 2009. Cautionary insights on knockout mouse studies: the gene or not the gene? Brain Behav Immun. 23:318-24. Eisener-Dorman AF, 2009

(8) Striebel JF, Race B, Pathmajeyan M, Rangel A, Chesebro B. 2013. Lack of influence of prion protein gene expression on kainate-induced seizures in mice: studies using congenic, coisogenic and transgenic strains. Neuroscience. 15;238:11-8. Striebel JF, 2013

(9) Nuvolone M et al. 2013. SIRPα polymorphisms, but not the prion protein, control phagocytosis of apoptotic cells. J Exp Med. 18;210:2539-52. Nuvolone M, 2013

On 2014 Aug 08, Claudiu Bandea commented:

Multiple knockout mouse models reveal that some lincRNAs might be required for life and brain development

In a recent series of mini-essays on genome functionality and ‘junk DNA’ prompted by Ford Doolittle’s article “Is junk DNA bunk? A critique of ENCODE” (1), I used the article by Sauvageau et al. (2) as an example of poor use of words in communicating scientific observations , which could lead to confusion and misrepresentations (3). In fact, I went so far as to suggest that some of the results in this particular study might be problematic. However, these allegations might be wrong and, therefore, it is important to bring them to the attention of the authors here, so they have the opportunity to respond.

First, I would like to emphasize that the study by Sauvageau et al. is impressive. The problem, however, might be with the interpretation of the results and their integration into the existing body of knowledge, which are as important in advancing science as generating data, if not more. There might also be a problem with the absence of appropriate controls for the knockout mouse models, which questions the validity of some of the results.

Briefly, to investigate the functional relevance of long intergenic noncoding RNAs (lincRNAs), Sauvageau et al. developed a collection of 18 lincRNA knockout mouse strains in which the locus was maintained transcriptionally active. In order to select the best candidates for functional lincRNAs, the authors “have implemented a generalized and logical lincRNA candidate selection process that leverages a collection of cell-based functional assays, RNA-sequencing data and computational analyses….”

Remarkably, in their initial characterization of the 18 lincRNAs knockout strains, Sauvageau et al. found three strains that exhibited peri- or post-natal lethality and two additional strains with distinct developmental defects. Based on these findings the authors concluded that “This study demonstrates that lncRNAs play critical roles in vivo…” and entitled their article: “Multiple knockout mouse models reveal lincRNAs are required for life and brain development.”

However, both the conclusion and the title of the study by Sauvageau et al. can be misleading. Factually, the results reported in the article indicate that 5 out of 18 lncRNAs included in study appear to play critical roles in vivo and, therefore, it should have been entitled something like: “Multiple knockout mouse models reveal that some lincRNAs might be required for life and brain development.”

Moreover, based strictly on the results presented in the article, which showed that only 5 of the 18 lncRNAs were functional, and considering the strong bias introduced in the study by selecting lincRNAs that are most likely to be functional, a more appropriate scientific interpretation of the study would be that: “The results presented in this article indicate that most lncRNAs (i.e. 13 out of 18) do not appear to play critical roles in vivo….” It is possible that the authors will report in future articles results showing that some additional, or all the lincRNAs in their collection are functional; even then, the authors should be cautious about the broad interpretation of the results considering the bias of their collection toward putatively functional lincRNAs (therefore, the inclusion of a few random lincRNAs would have benefited the study).

Whether the assertion about the broad interpretation of the results, which might be considered an 'innocent' case of ‘hype in science’ (4), resonate with the authors or with other readers remains to be seen: however, the allegation that the phenotypes observed in their study might be associated with untargeted genome alterations introduced inadvertently during the procedure of generating the knockout mice, rather than with the deletion of specific lincRNAs, questions the validity of the study, at least for the time being and, therefore, it should be of concern.

Considering the well-recognized problems with generating knockout animal models (see, for example, Refs. 5-9), it is surprising that Sauvageau et al. did not discussed this issue in the context of their study, and that they did not include appropriate controls (e.g. duplicates) at least for some of the lincRNAs knockouts, preferably using mice with different genetic background. Although the data and observations presented by Sauvageau et al. might be correct, for the time being their validity remains questionable; therefore, it would make sense for the authors (and the editors) to add an explanatory note, both here at PubMed Commons and in the journal eLife.

References

(1) Doolittle WF. 2013. Is junk DNA bunk? A critique of ENCODE. Proc Natl Acad Sci USA., 110:5294-300. Doolittle WF, 2013

(2) Sauvageau M. et al., 2014. Multiple knockout mouse models reveal lincRNAs are required for life and brain development. eLife (DOI: 10.7554/eLife.01749) Sauvageau M, 2013

(3) Bandea CI. 2014. Everlasting confusion on ‘functional DNA’ and ‘junk DNA’ in the field of genome biology. PubMed Commons (National Library of Medicine; Bethesda, MD). Comment on: Doolittle WF, 2013

(4) Maderspacher F. 2014. Hype in Halifax. Curr Biol. 14;24(8):R298-301. Maderspacher F, 2014

(5) Wolfer DP, Crusio WE, Lipp HP. 2002. Knockout mice: simple solutions to the problems of genetic background and flanking genes. Trends Neurosci. 25:336-40. Wolfer DP, 2002

(6) Crusio WE1, Goldowitz D, Holmes A, Wolfer D. 2009. Standards for the publication of mouse mutant studies. Genes Brain Behav. 8(1):1-4. Crusio WE, 2009

(7) Eisener-Dorman AF, Lawrence DA, Bolivar VJ. 2009. Cautionary insights on knockout mouse studies: the gene or not the gene? Brain Behav Immun. 23:318-24. Eisener-Dorman AF, 2009

(8) Striebel JF, Race B, Pathmajeyan M, Rangel A, Chesebro B. 2013. Lack of influence of prion protein gene expression on kainate-induced seizures in mice: studies using congenic, coisogenic and transgenic strains. Neuroscience. 15;238:11-8. Striebel JF, 2013

(9) Nuvolone M et al. 2013. SIRPα polymorphisms, but not the prion protein, control phagocytosis of apoptotic cells. J Exp Med. 18;210:2539-52. Nuvolone M, 2013