Horizontal gene transfer: Learning from losers
The ruthless nature of Darwinian selection does not allow much room for carelessness. So, when a competitor eventually suffers a setback, the obvious response is to leave him in the dust and to snap relentlessly at the heels of anyone further ahead. Now, in eLife, Robert Cooper, Lev Tsimring and Jeff Hasty from the University of California, San Diego report how this may not always be the case (Cooper et al., 2017).
Some organisms, in particular bacteria, have the ability to transfer genetic material from nearby organisms, rather than just between parent and offspring. For many years it was thought that this process, also known as horizontal gene transfer, happened relatively rarely and mostly by accident. Moreover, it was thought that the recipients used the DNA mostly as a source of energy, though incorporation of foreign DNA could drive long-lasting evolutionary trends (Kurland et al., 2003; Nielsen et al., 2014; MacFadyen et al., 2001). In contrast, horizontal gene transfer is increasingly acknowledged as a routine way organisms acquire beneficial and adaptive genes, such as genes that confer resistance against antibiotics and antiseptics. One bacterium can, in effect, ‘learn’ important traits from other bacteria.
Some bacteria can speed up the process of horizontal gene transfer by killing their neighbors in order to get to their DNA. For example, the bacterium Streptococcus pneumoniae secretes toxins to kill sister cells or closely related bacteria, and then extracts their DNA (Steinmoen et al., 2002; Croucher et al., 2016; Wholey et al., 2016). Other bacteria, including the bacterium that causes cholera, use a contact-dependent killing mechanism known as the ‘type VI secretion system’ that involves delivering toxins directly into their victim (Borgeaud et al., 2015).
Cooper et al. placed two distantly related types of bacteria, Acinetobacter baylyi and Escherichia coli, on a surface containing nutrients over which they could grow. E. coli had been modified to produce a fluorescent protein in order to track its fate/whereabouts. The researchers observed that A. baylyi killed E. coli and extracted their DNA, which caused A. baylyi to become fluorescent. Moreover, mutant A. baylyi that had lost the ability to kill were still able to integrate the DNA released when their wild-type siblings killed E. coli. Cooper et al. showed that the extraction of DNA from E. coli and its integration into A. baylyi happened so often it could be observed in real time. A. baylyi that killed the E. coli and released its DNA increased the rate of gene transfer by several orders of magnitude compared to A. baylyi that had been stripped off the ability to kill. This raises the question: what can a bacterium learn from another bacterium that it has just vanquished?
Certain Acinetobacter bacteria thrive in hospital settings, where they can cause serious infections in compromised patients (Oberauner et al., 2013). These bacteria are also able to rapidly develop clinically-impactful antibiotic resistance (Durante-Mangoni et al., 2015). Cooper et al. showed that by killing nearby bacteria, A. baylyi could also acquire genes that made them resistant to antibiotics within hours. The researchers then tested two factors that could theoretically block horizontal gene transfer: enzymes that degrade DNA and irrelevant DNA. The enzymes reduced transfer rates, but the irrelevant DNA did not. Finally, a mathematical model was used to evaluate the experimental data and to simulate a range of conditions that could affect the gene transfer. The simulations showed that killing-enhanced gene transfer was more effective when A. baylyi outnumbered E. coli.
The work of Cooper et al. suggests that horizontal gene transfer rates, and as a particular case, the rate of antibiotic resistance aquisition, could be many orders of magnitude higher than previously thought, therefore challenging current antibiotic-resistance prevention strategies. Moreover, incorporating new genes and thus new abilities into a functioning genome is risky – even more so when the source genome has failed to survive. This risk is, however, minimized if the first bacterium ‘knows’ why the second one has died (that is, it has been killed by the first bacterium) and can, therefore, dismiss the cause of death as irrelevant.
Even under stressful conditions, such as exposure to antibiotics, A. baylyi will still encounter organisms that are able to grow. It will kill them, sample their genomes and may even locate those parts of the genome that enabled the other bacteria to survive (Sandegren and Andersson, 2009). However, it remains unclear to what extent A. baylyi samples the genome of the cells it has killed, and how it regulates the learning process to make it relatively ‘safe’ for themselves.
Bacteria face many selective pressures at any given moment, and most insights obtained through studying interactions between two species will not apply when several species are present (Østman and Adami, 2014; Guimarães et al., 2017). To understand why and how the bacteria learn as they do, we will need to explore DNA exchange under more realistic and multi-dimensional conditions. However, for the moment, Cooper et al. have furthered our understanding in exciting ways.
References
-
Emergence of colistin resistance without loss of fitness and virulence after prolonged colistin administration in a patient with extensively drug-resistant Acinetobacter baumanniiDiagnostic Microbiology and Infectious Disease 82:222–226.https://doi.org/10.1016/j.diagmicrobio.2015.03.013
-
Detecting rare gene transfer events in bacterial populationsFrontiers in Microbiology 4:415.https://doi.org/10.3389/fmicb.2013.00415
-
Recent Advances in the Theory and Application of Fitness Landscapes509–526, Predicting evolution and visualizing high-dimensional fitness landscapes, Recent Advances in the Theory and Application of Fitness Landscapes, Berlin, Heidelberg, Springer.
-
Bacterial gene amplification: implications for the evolution of antibiotic resistanceNature Reviews Microbiology 7:578–588.https://doi.org/10.1038/nrmicro2174
Article and author information
Author details
Publication history
Copyright
© 2017, Kirkup
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Metrics
-
- 2,467
- views
-
- 220
- downloads
-
- 1
- citations
Views, downloads and citations are aggregated across all versions of this paper published by eLife.
Download links
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Further reading
-
- Computational and Systems Biology
Manifold-learning is particularly useful to resolve the complex cellular state space from single-cell RNA sequences. While current manifold-learning methods provide insights into cell fate by inferring graph-based trajectory at cell level, challenges remain to retrieve interpretable biology underlying the diverse cellular states. Here, we described MGPfactXMBD, a model-based manifold-learning framework and capable to factorize complex development trajectories into independent bifurcation processes of gene sets, and thus enables trajectory inference based on relevant features. MGPfactXMBD offers a more nuanced understanding of the biological processes underlying cellular trajectories with potential determinants. When bench-tested across 239 datasets, MGPfactXMBD showed advantages in major quantity-control metrics, such as branch division accuracy and trajectory topology, outperforming most established methods. In real datasets, MGPfactXMBD recovered the critical pathways and cell types in microglia development with experimentally valid regulons and markers. Furthermore, MGPfactXMBD discovered evolutionary trajectories of tumor-associated CD8+ T cells and yielded new subtypes of CD8+ T cells with gene expression signatures significantly predictive of the responses to immune checkpoint inhibitor in independent cohorts. In summary, MGPfactXMBD offers a manifold-learning framework in scRNA-seq data which enables feature selection for specific biological processes and contributing to advance our understanding of biological determination of cell fate.
-
- Computational and Systems Biology
- Genetics and Genomics
Apart from ancestry, personal or environmental covariates may contribute to differences in polygenic score (PGS) performance. We analyzed the effects of covariate stratification and interaction on body mass index (BMI) PGS (PGSBMI) across four cohorts of European (N = 491,111) and African (N = 21,612) ancestry. Stratifying on binary covariates and quintiles for continuous covariates, 18/62 covariates had significant and replicable R2 differences among strata. Covariates with the largest differences included age, sex, blood lipids, physical activity, and alcohol consumption, with R2 being nearly double between best- and worst-performing quintiles for certain covariates. Twenty-eight covariates had significant PGSBMI–covariate interaction effects, modifying PGSBMI effects by nearly 20% per standard deviation change. We observed overlap between covariates that had significant R2 differences among strata and interaction effects – across all covariates, their main effects on BMI were correlated with their maximum R2 differences and interaction effects (0.56 and 0.58, respectively), suggesting high-PGSBMI individuals have highest R2 and increase in PGS effect. Using quantile regression, we show the effect of PGSBMI increases as BMI itself increases, and that these differences in effects are directly related to differences in R2 when stratifying by different covariates. Given significant and replicable evidence for context-specific PGSBMI performance and effects, we investigated ways to increase model performance taking into account nonlinear effects. Machine learning models (neural networks) increased relative model R2 (mean 23%) across datasets. Finally, creating PGSBMI directly from GxAge genome-wide association studies effects increased relative R2 by 7.8%. These results demonstrate that certain covariates, especially those most associated with BMI, significantly affect both PGSBMI performance and effects across diverse cohorts and ancestries, and we provide avenues to improve model performance that consider these effects.