REBUTTAL TO THE EXPRESSION OF CONCERN PUBLISHED BY PLOS ONE
BY:
Mary H. Whiteley,

Corresponding author: Whiteley, Mary H., Jerold S. Bell, and Debby A. Rothman. “Novel Allelic Variants in the Canine Cyclooxgenase-2 (Cox-2) Promoter Are Associated with Renal Dysplasia in Dogs.” Ed. Michael Pazin. PLoS ONE 6.2 (2011)
THIS EXPRESSION OF CONCERN IS NOT IN PROPER FORM ACCORDING TO THE GUIDELINES FOR COMMENTS ON THE PLOS ONE WEBSITE:

“Unsupported assertions or statements should be avoided. Comments must be evidence-based, not authority-based.”

NONE of these comments are evidence-based, as no references are or data are provided. As such they merely represent a difference of opinion.

COMMENT 1. The description of the alleles in the article is inadequate.

Response:
This is vague and no supporting information for this statement is presented. The DNA sequence of each allele is listed in the paper. This should adequately describe these alleles. As stated clearly in the text, these alleles are insertions or deletions of SP1 transcription factor binding sites, and their relative position to the start site of transcription is listed. The description of the alleles is given in Fig S1.

COMMENT 2. There are concerns over the study design employed to study the association; a single-gene association study based on Cox-2 or a genome-wide association study have been recommended as more appropriate approaches to study this association.

Response:

Again, there are no justifications or scientific references given to support this comment. There is nothing in the scientific literature to substantiate that GWAS is more appropriate than a candidate gene analysis, especially when the phenotype is well defined (1). Candidate gene analysis relies on previous biological data. In this case, the mouse Cox-2 knockout and knockdown phenotypes support this as a candidate gene (2,3,4).

GWAS requires identification affected and non-affected individuals within a population as well as subjects that are not related like a different breed ( SEE comment 3) to this population (5). The mode of inheritance of renal dysplasia (RD) is dominant with incomplete penetrance. Dogs that are biopsy negative have been shown to produce biopsy positive offspring (6,7). Inclusion of biopsy negative individuals that carry a disease causing mutation does not allow for proper stratification of the population in a GWAS (5). By definition common disease causing alleles must have low penetrance (8). Bovee concluded that the prevalence of Shih tzus with a kidney defect was approximately 85%, but the actual numbers of dogs that die from renal failure is low (5-10%) (6). This is consistent with the RD alleles being common but of low penetrance.

While GWAS has uncovered low penetrance common alleles many of these studies cannot be replicated frequent false positive associations are found. This missing heritability is thought to be in part because of the limitations in the power of GWAS (9,10,11). There remain many challenges in replicating GWAS results from low penetrance common alleles (12). In summary, a GWAS approach is particularly unsuitable for the RD study contrary to comment 2.

Single gene association studies are likewise limited the issues such as non-replication and population stratification as in GWAS discussed above (13).

A linkage analysis by researchers using the same Lhasa apso family samples used in this study resulted in an inaccurate association of a marker with this disease. See: http://www.gompalhasaapso...

No scientific peer-reviewed has been published about this linkage marker, however, the above table reveals the flaws in this approach, and excludes this as a viable approach to this research as suggested in comment 2.

COMMENT 3. The validity of the control population employed in the study is compromised as it involved a different breed.

Response:
This is not true and is unsubstantiated. Examples of using different breeds as controls are given in references 14,15.

COMMENT 4.There are concerns about the strength of the evidence shown to support an association between the Cox-2 variant and the dogs' phenotypes, as the evidence from other breeds suggests that this may be a neutral DNA variant.

Response:
There is no evidence presented as to why this could be a neutral variant. This comment is speculative in nature.
Recently I published a functional evaluation of these RD alleles (16) that demonstrates that these are NOT neutral. The data summarized below is more than adequate to define an association between these alleles and the disease.
Data supporting that the RD alleles are causal for disease:

1. The presence of one or more copies of an RD allele predisposes the canine Cox-2 gene to aberrant methylation. Aberrant methylation of promoter regions results in down-regulation of gene expression. This has been extensively studied for well over a decade (17,18, 19 and 100s more). Methylation of genomic regions near transcription start sites, CpG island, CpG shore, and first exon is strongly associated with gene repression (17). The RD alleles represent exactly this case: they are near the start of transcription, and in a CpG island, and are near the first exon.

2. The wild type Cox-2 gene is never methylated and this is of particularly important in the case of an RD heterozygote. The wild type allele in this case serves as an internal control for aberrant methylation.

3. This epigenetic modification MUST have occurred early in development as methylated DNA is found in a variety of cell types.

4. Aberrant methylation was found only in biopsy positive samples, but not in biopsy negative samples. wt/wt controls were never methylated. Thus methylation is correlated with the disease.

5. Aberrant methylation occurred in different genetic backgrounds (breeds), signaling a universal mechanism of action.

6. Normally, the CpG island in the human Cox-2 promoter is unmethylated. However, hypermethylation of the human Cox-2 CpG island has been identified in some types of cancer [20,21], in which the level of Cox-2 repression increased with higher levels of methylation.

7. Another example of a developmental gene involving aberrant methylation is the yellow Agouti mouse (Avy), in which coat color is highly variable, and is caused by varying degrees of methylation (22). The phenotype ranges from yellow (unmethylated) to varying degrees of pseudoagouti phenotypes that are caused by varying degrees of methylation. Another example of a single gene disorder in the mouse strain AxinFu produces a variable phenotype via variable methylation during development (23). This strengthens the argument that the RD alleles of Cox-2, a gene that is important for kidney development, could be single-locus disease causing alleles. Since the Avy phenotype is caused via epigenetic modification is influenced by environmental factors (22) the RD alleles could also be influenced in this way, as suggested in reference 16.
This Agouti mouse strain is caused by a particle (IAP) murine retrotransposon upstream of the transcription start site of the Agouti gene that causes ectopic expression of Agouti. The variable phenotype in this case is inversely proportional to the degree of methylation at 6 CpG sites in the IAP.

FINAL THOUGHTS:

Given the strengths of the arguments above, it is my opinion that this expression of concern needs to be retracted by PLOS ONE.
This is necessary to maintain the accuracy of the scientific literature.
Mary Whiteley
whiteley@dogenes.com

References
1. Picut CA, Lewis RM (1987) Microscopic features of canine renal dysplasia Vet
Pathol 24: 156–63.


2. Dinchuk JE, Car BD, Focht RJ, Johnston JJ, Jaffee D, et al. (1995) Renal
abnormalities and an altered inflammatory response in mice lacking cyclooxygenase
II Nature 23: 406–409.

3. Morham SG Langenbach R, Loftin CD, Tiano F, Vouloumanos N, et al. (1995)
Prostaglandin synthase 2 gene disruption causes severe renal pathology in the
mouse, Cell 83: 473–482.

4. Seta F, Chung AD, Turner PV, Mewburn JD, Yu Y, et al. (2009) Renal and
cardiovascular characterization of COX-2 knockdown mice Am J Physiol Regul
Integr Comp Physiol 296: R1751–1760.

5. Liu L, Zhang D, Liu H, Arendt C. Robust methods for population stratification in genome wide association studies. BMC Bioinformatics. 2013;14:132.


6. Bovee KC (2003) ‘‘Renal Dysplasia in Shih Tzu Dogs’’. Proceedings of the 28th
World Congress of the World Small Animal Veterinary Association.

7. Whiteley M.H., Bell J.S. & Rothman D.A. 2011. Novel allelic variants in the canine cyclooxygenase-2 (Cox-2) promoter are associated with renal dysplasia in dogs. PlosOne 7:e16684.



8. Bush WS, Moore JH. Chapter 11: Genome-Wide Association Studies. Lewitter F, Kann M, eds. PLoS Computational Biology. 2012;8(12):e1002822. doi:10.1371/journal.pcbi.1002822.

9. Galvan, Antonella, John P.A. Ioannidis, and Tommaso A. Dragani. “Beyond Genome-Wide Association Studies: Genetic Heterogeneity and Individual Predisposition to Cancer.” Trends in genetics : TIG 26.3 (2010): 132–141.

10. Brookfield JF. Q&A: Promise and pitfalls of genome-wide association studies. BMC Biology. 2010;8:41.

11. Cooper DN, Krawczak M, Polychronakos C, Tyler-Smith C, Kehrer-Sawatzki H. Where genotype is not predictive of phenotype: towards an understanding of the molecular basis of reduced penetrance in human inherited disease. Human Genetics. 2013;132(10):1077-1130


12. Kraft P, Zeggini E, Ioannidis JPA. Replication in genome-wide association studies. Statistical science : a review journal of the Institute of Mathematical Statistics. 2009;24(4):561-573.

13. Thomas DC, Witte JS. Point: population stratification: a problem for case-control studies of candidate-gene associations? Cancer Epidemiol Biomarkers Prev. 2002 Jun;11(6):505-12.



14. Drögemüller M, Jagannathan V, Becker D, et al. A Mutation in the FAM83G Gene in Dogs with Hereditary Footpad Hyperkeratosis (HFH). Barsh GS, ed. PLoS Genetics. 2014;10(5).


15. Jagannathan, Vidhya et al. “A Mutation in the SUV39H2 Gene in Labrador Retrievers with Hereditary Nasal Parakeratosis (HNPK) Provides Insights into the Epigenetics of Keratinocyte Differentiation.” Ed. Matthew T. Webster. PLoS Genetics 9.10 (2013):

16. Whiteley, Mary H. “Allelic Variation in the Canine Cox-2 Promoter Causes Hypermethylation of the Canine Cox-2 Promoter in Clinical Cases of Renal Dysplasia.” Clinical Epigenetics 6.1 (2014):7

17. Lee CJ, Evans J, Kim K, Chae H, Kim S. Determining the effect of DNA methylation on gene expression in cancer cells. Methods Mol Biol. 2014;1101:161-78. doi: 10.1007/978-1-62703-721-1_9.

18. Burton ER, Gaffar A, Lee SJ, et al. Down-regulation of filamin A interacting protein 1-like is associated with promoter methylation and induces an invasive phenotype in ovarian cancer. Molecular cancer research : MCR. 2011;9(8):1126-1138. doi:10.1158/1541-7786.MCR-11-0162.

19. Morenos L, Chatterton Z, Ng JL, et al. Hypermethylation and down-regulation of DLEU2 in paediatric acute myeloid leukaemia independent of embedded tumour suppressor miR-15a/16-1. Molecular Cancer. 2014;13:123. doi:10.1186/1476-4598-13-123.

20.Toyota M, Shen L, Ohe-Toyota M, Hamilton SR, Sinicrope FA, Issa JP:
Aberrant methylation of the cyclooxygenase 2 CpG island in colorectal
tumors. Cancer Res 2000, 60:4044–4048.

21. Hur K, Song SH, Lee HS, Ho Kim W, Bang YJ, Yang HK: Aberrant
methylation of the specific CpG island portion regulates
cyclooxygenase-2 gene expression in human gastric carcinomas.

22. Dolinoy DC. The agouti mouse model: an epigenetic biosensor for nutritional and environmental alterations on the fetal epigenome. Nutrition reviews. 2008;66(Suppl 1):S7-11. doi:10.1111/j.1753-4887.2008.00056.x.

23. Rakyan VK, Chong S, Champ ME, et al. Transgenerational inheritance of epigenetic states at the murine AxinFu allele occurs after maternal and paternal transmission. Proceedings of the National Academy of Sciences of the United States of America. 2003;100(5):2538-2543. doi:10.1073/pnas.0436776100.