Skip to main content

Response to “Variable directionality of gene expression changes across generations does not constitute negative evidence of epigenetic inheritance” Sharma, A. Environmental Epigenetics, 2015, 1-5


Abhay Sharma brings two arguments in favor of transgenerational epigenetic inheritance (TGEI) in mammals when criticizing our work. He uses probability calculations and finds that the probability of obtaining the number of common changes in the in utero-exposed prospermatogonia and the same cells in the next generation is significant in our study. He also compares our results to other published datasets and concludes that the probability for the observed overlap between independent studies is significant. We disagree with both arguments of Sharma and show here that his meta-analysis and statistical calculations are not correct.

Sharma [1] directly criticized our paper [2] in the first issue of Environmental Epigenetics in an attempt to defend the prevailing dogma of transgenerational epigenetic inheritance (TGEI) of environmentally induced changes in mammals [3, 4]. The journal (Michael Skinner, Editor in Chief) has not given us a chance to respond.

In the paper under dispute [2], we failed to find evidence supporting TGEI at the level of transcription and DNA methylation. We hypothesized that if an epigenetic aberration was triggered in the male fetal germ cells by environmental insult, it could be inherited by the sperm and might prevail in the fetal germ cells of the next, unexposed generation if not corrected by epigenome reprogramming. Part of our work involved assessing gene transcription in the fetal germ cells of the fetuses exposed in utero to endocrine disruptors and in fetal germ cells of the next generation (Table 4 in [2]). Any change would be considered inherited if it was affecting gene transcription similarly in the subsequent generations. Using rigorous statistics (1.5-fold change and FDR P <0.05) to interpret our RNA microarray results, we found an extremely low number of significant changes in the exposed germ cells (seven and two probes total for di(2-ethylhexyl)phthalate [DEHP] and vinclozolin [VZ] exposures, respectively) but none of these prevailed between generations. When we relaxed our statistical stringency (1.5-fold change and unadjusted P <0.05), we found one and eight probe overlaps between generations for DEHP and VZ, respectively. Every one of these changes occurred in the opposite direction in the subsequent generation. These findings using generally acceptable statistical rigor are consistent with the explanation that germline epigenetic reprogramming corrects the environment-induced mistakes in gene regulation.

In his first argument, Sharma selectively interpreted our results by focusing only on the next part of our analysis, when we further reduced the statistical stringency (1.05-fold change, P <0.05). He calculated the hypergeometric probabilities for the common hits between generations and reported them to reach statistical significance. At that extremely low level of statistical stringency we counted 284 and 325 genes that exhibited changes in subsequent generations for DEHP and VZ, respectively. Among these, only 77/325 and 30/284 occurred in the same direction (upregulated in two generations or downregulated in two generations). We suggest that the differences that are found in the opposite direction [2] may indicate a slight overcompensation in the erasure process. We disagree with Sharma’s opinion that a change that occurs in the opposite direction in subsequent generations constitutes an inherited epigenetic aberration. Further, a decrease of a gene transcript in one generation and an increase of the same transcript in the next generation cannot account for the same phenotype in the subsequent generations. Therefore, molecular changes occurring in the opposite direction in subsequent generations should not be combined in support of TGEI. Importantly, when we calculated the hypergeometric distribution probabilities of the number of changes that occurred in the same direction, those numbers were not significant: P(x > = 77) = 1 for DEHP and P(x > = 30) = 1 for VZ (Table 1).

Table 1 Hypergeometric probability calculations for the common changes defined using extremely low stringency (1.05-fold, P <0.05) detected in [2]

Sharma argues [1]: “expression change in the opposite direction across generations does not necessarily compromise the positive evidence of transgenerational effects because phenotypic variability across generations is known in epigenetic inheritance. For example, transgenerational weakening and strengthening of phenotypes have been reported in several studies in mammals.” He also writes: “Moreover, gene expression changes in opposite direction across generations have also been reported in studies pertaining to effects of environmental agents in animals.” We would like to point out that the papers cited by Sharma report results regarding: (1) a single generation; (2) intergenerational changes that occur in the same direction; (3) transgenerational phenotypic (not molecular) changes that occur in the same direction between three generations; (4) are non-mammalian examples; or (5) concern miRNA. Specifically, none of these cited papers reports a TGEI phenotype that can be explained by a molecular change in the opposite direction between the generation exposed as germ cells and the next, unexposed, generation (or in the exposed germ cell and germ cell of the next generation). Brieno-Enriquez [5] reports that the embryonic phenotype, primordial germ cell (PGC) number defect at 13.5 dpc and apoptosis at 13.5 dpc, occur in F1 and F2 PGCs but not in F3 PGCs. The Let-7/LIN28/BLIMP1 pathway is misregulated in F1, F2, and F3 PGCs. The direction of change is the same for each component between F1 and F2 PGCs, exactly when the phenotypic changes that can be anticipated from misregulated BLIMP1 are actually observed in embryonic germ cells. The direction of change is opposite for certain components in the F3 PGCs, when the phenotype is no longer observed.

For his second argument [1], Sharma compares our data [2] to previously published rat and mouse datasets using hypergeometric probability calculations. He writes [1]: “In this analysis of mouse and rat gene expression data from diverse studies, the entire set of human genes listed in the Gene ontology was used for normalization, to find overlap between previous gene sets and genes identified by Iqbal and colleagues. The premise of this analysis was that a higher-fold enrichment with greater statistical significance in VZ-VZ comparison relative to VZ-DEPH comparison would indicate TGEI. Importantly, a higher enrichment with greater significance is observed for VZ-VZ overlap than VZ-DEPH overlap.” To the contrary of his expectations, a lack of probability of inter-study overlap is apparent in Figure 2 of Sharma [1], where significance values should accompany the columns according to the figure legend as they do in [6], but only the columns pertaining to all 36 datasets combined are labeled with statistical value. Using the same gene lists Sharma used [1] (and one should use with caution, see below), we also counted the common hits and calculated the hypergeometric probabilities of finding common hits between our study [2] and previous studies (Supplement 2 in [1], References to source data therein). As displayed in Table 2, two comparisons are marginally significant but the probability for finding an overlap pertaining to all 36 datasets combined is not statistically significant. We note that probabilities were calculated without considering adjustments for multiple testing. If we accounted for that, these probabilities would be even larger.

Table 2 Hypergeometric probability calculations for common VZ-induced changes between our study [2] and previous work considered by Sharma [1]

There are several problems with Sharma’s interpretations [1]: (1) he did not acknowledge the lack of common hits at the rigorous statistical cutoff value but only focused on the findings using extremely low statistical stringency; (2) he made the assumption that the changes occurring in the opposite direction between studies could be considered common changes; (3) he calculated the probability of obtaining the exact number of successes in the sample (not the probability of obtaining at least that many successes); (4) he did not apply the same statistical cutoff values between studies; (5) he used the number of known human genes as reference, but this analysis should be limited to what can be detected by the specific array; (6) he considered studies that did not actually search for transgenerational changes in germ cells; and (7) none of his study-to study comparisons reached significant probability values in his Figure 2, yet he concluded finding more significant overlap for VZ-VZ than VZ-DEHP comparisons.

Based on the results of our study [2] and our reassessment of Sharma’s calculations we cannot provide evidence in support of TGEI in mammals.


  1. Sharma A. Variable directionality of gene expression changes across generations does not constitute negative evidence of epigenetic inheritance. Environ Epigenet. 2015;1:1–5.

    Article  Google Scholar 

  2. Iqbal K, Tran DA, Li AX, Warden C, Bai AY, Singh P, et al. Deleterious effects of endocrine disruptors are corrected in the mammalian germline by epigenome reprogramming. Genome Biol. 2015;16:59.

    Article  PubMed  PubMed Central  Google Scholar 

  3. Anway MD, Cupp AS, Uzumcu M, Skinner MK. Epigenetic transgenerational actions of endocrine disruptors and male fertility. Science. 2005;308:1466–9.

    Article  CAS  PubMed  Google Scholar 

  4. Skinner MK. Endocrine disruptors in 2015: Epigenetic transgenerational inheritance. Nat Rev Endocrinol. 2016;12:68–70.

    CAS  PubMed  Google Scholar 

  5. Brieno-Enriquez MA, Garcia-Lopez J, Cardenas DB, Guibert S, Cleroux E, Ded L, et al. Exposure to endocrine disruptor induces transgenerational epigenetic deregulation of microRNAs in primordial germ cells. PLoS One. 2015;10:e0124296.

    Article  PubMed  PubMed Central  Google Scholar 

  6. Sharma A. Novel transcriptome data analysis implicates circulating microRNAs in epigenetic inheritance in mammals. Gene. 2014;538:366–72.

    Article  CAS  PubMed  Google Scholar 

Download references


The author thanks Dr. Mary E Winn and Zach Madaj (VARI Bioinformatics and Biostatistics Core) for their guidance in statistical data analysis. The initial work in the author’s laboratory was supported by a grant from the NIEHS (grant R01 ES015185) and from the NIGMS (grant RO1GM064378) to PES.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Piroska E. Szabó.

Additional information

Competing interests

The author declares that no competing interest exists.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Szabó, P.E. Response to “Variable directionality of gene expression changes across generations does not constitute negative evidence of epigenetic inheritance” Sharma, A. Environmental Epigenetics, 2015, 1-5. Genome Biol 17, 105 (2016).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: