Menu

Close

The transcriptional legacy of developmental stochasticity

Sara Ballouz, Maria T. Pena, Frank M. Knight, Linda B. Adams, Jesse A. Gillis

Preprint posted on December 12, 2019 https://www.biorxiv.org/content/10.1101/2019.12.11.873265v1

Developmental stochasticity gives rise to transcriptional signatures that mark identity in armadillo quadruplets

Selected by Sergio Menchero

Categories: genomics

Background

The phenotype of an individual is largely determined by genetic contribution and the influence of the environment. Stochastic variability of transcription, particularly during embryonic development, is also thought to affect phenotypic variation but it has always been a difficult component to isolate and study. In this work, the authors use an exceptional feature of the nine-banded armadillo (Dasypus novemcinctus) to study this developmental stochasticity. Armadillos have developed a unique reproductive strategy to produce litters of identical quadruplets. This system allows the control of genetic and environmental contributions to focus on the role of transcriptional stochasticity. Collecting blood samples and measuring gene expression at three different time points per armadillo quadruplet, the authors investigate the influence of transcriptional changes, and whether they may account for identity signatures among siblings.

 

Main findings:

The authors aimed to identify transcriptional differences in armadillo quadruplets – maintained throughout the three time points – that were sufficient to distinguish individuals among siblings. For that, the authors developed a machine-learning method that helped them identify differential transcripts consistently over time as indicators of identity.

Random X-chromosome inactivation signature. Like all mammals, female armadillos undergo X-chromosome inactivation to silence one of their X-chromosomes and balance the gene dosage content with males. This process is random when it first happens and, in armadillos, it actually takes place after the embryos have split. Therefore, each sibling will have a unique signature. Using their machine-learning method, the authors verified that imbalance of allelic ratios of X-linked genes are predictive of an individual within a quadruplet. Interestingly, this is not something exclusive of the X chromosome because they also observed allelic imbalances in autosomes which were predictive of individuality.

Figure 1 (Figure 2C in the preprint). X-inactivation estimates from RNA-seq data in one of the quadruplets analysed shows variation in skewing. Made available under a CC-BY-NC-ND 4.0 International license.

 

Gene expression differences as mark of identity. As opposed to allelic ratios, the authors then investigated the impact of gene expression. They took X-linked genes and autosome-linked genes separately to avoid biases in the analysis. Interestingly, the genes that acted as predictors of individuality were not differentially expressed in all the armadillo cohorts analysed. Those genes were always expressed in all quadruplets, indicating that identity signatures were not due to genes that are uniquely expressed in specific individuals. The functions of predictor genes were enriched in lifestyle stimuli such as cardiac muscle growth.

Individual signatures in human twins. In order to see if similar functional perturbations driving identity in armadillos could be observed in humans, the authors analysed data from human identical twins. Because the post-natal environment was not controlled as in armadillos, differences in gene expression could not be completely linked to transcriptional stochasticity. In fact, differences in gene expression in humans were higher than in armadillos. Differentially expressed genes were not associated with known eQTLs (genomic loci that affect expression levels), suggesting that those differences were not due to genetic variations. An interesting aspect is that genes that showed variation in gene expression in armadillos and humans are frequently perturbed in differential expression studies and could be associated with stimulus-responsive regulation.

Figure 2 (Figure 4C in the preprint): Modeling signatures of identity as observed in controlled and variable environments. Made available under a CC-BY-NC-ND 4.0 International license.

 

General comments and why I chose this preprint

A non-heritable variability that complements the effect of the genotype and the environment for every phenotype has been appreciated and discussed in multiple organisms and contexts [1]. I really liked the authors’ strategy of using armadillo quadruplets to study the effect of this component that has always been difficult to isolate from the effect of the environment. This is a beautiful example of how one of the most important things when addressing a question is to choose the model system to work with and why, although challenging, using a non-classical model organism can be the key.

The authors have done a very important effort in this work that opens many questions and predicts interesting directions in future individual stochasticity studies.

 

References

[1] Honegger and de Bivort (2018). Stochasticity, Individuality and Behavior. Curr Biol, 28 (1): R8-R12.

 

Questions to the authors

The authors see that genes that were differentially expressed and gave an identity among siblings were not always the same between armadillo quadruplets. Do the authors think the genes that provide individuality are completely random? Or do they think there may be some genes that are more prone to have differences in expression levels due to a less strict regulation?

Although the authors talk about “developmental stochasticity”, the analyses are performed on blood sample data and thus, some genes they can detect may be different from genes expressed during embryo development. Do the authors expect a different or broader set of genes to account for individual signature during embryo development?

Do the authors expect different functions for those genes which provide individuality during a specific period of time versus those genes whose differential gene expression are maintained during a longer time window?

Tags: armadillo, developmental stochasticity, individual signature, transcriptional variability

Posted on: 23rd January 2020 , updated on: 24th January 2020

doi: https://doi.org/10.1242/prelights.16454

Read preprint (3 votes)




  • Author's response

    Jesse A. Gillis shared

    1. The authors see that genes that were differentially expressed and gave an identity among siblings were not always the same between armadillo quadruplets. Do the authors think the genes that provide individuality are completely random? Or do they think there may be some genes that are more prone to have differences in expression levels due to a less strict regulation?

    No, I don’t think they are completely random but our data do suggest that the effects are quite broad.  My internal model is probably that a great many genes are implicated to varying degrees.  Which specific genes rise to prominent within an individual would then involve a lot of randomness, but not sample at all uniformly across the genome.

    I am not sure if genes would show more or less of an effect if regulation were less strict.  In order to be “identifying” the gene’s expression must be quite tightly controlled (or set) but not dynamically.  I’m not sure one would want to call that less strict regulation.  Lax early and then strict thereafter is probably the easiest model.  On the other hand rather than thinking of it as lax, I think we are just running to the limits of what is controllable (for expression) or what needs to be (for allelic variation).

     

    2. Although the authors talk about “developmental stochasticity”, the analyses are performed on blood sample data and thus, some genes they can detect may be different from genes expressed during embryo development. Do the authors expect a different or broader set of genes to account for individual signature during embryo development?

    Yes – broader.  We use the distribution of allelic imbalances to estimate the approximate size of the cell population at the time expression was set.  In the case of X-inactivation, this looks to happen early enough that our results are likely general across tissues.  For the vast majority of our results (which focus on effects outside the X), the identified results are likely at least somewhat tissue specific and do not account for all signatures of individuality during embyro development.

     

    3. Do the authors expect different functions for those genes which provide individuality during a specific period of time versus those genes whose differential gene expression are maintained during a longer time window?

    Our focus is almost solely on the genes whose differential expression (or allele specific expression) looks to be very long term.  That is not to say that there may not be upstream events contributing to individuality in a transient way, just that is not what we were investigating directly.  It might be tempting to speculate about stochasticity having a greater or lesser impact depending on the degree of developmental constraint but I think it would be hard to have strong expectations at this point.

    Have your say

    Your email address will not be published. Required fields are marked *

    This site uses Akismet to reduce spam. Learn how your comment data is processed.

    Sign up to customise the site to your preferences and to receive alerts

    Register here
    Close