Month: May 2013

In this paper Richard E. Lenski and colleague are showing an example of how efficient adaptation by natural selection is. During 20 year they have been growing twelve populations of Escherichia coli in glucose medium containing also abundant citrate, this famous long-term evolution experiment called LTEE allowed the evolution to run for 40’000 generations in controlled laboratory conditions. Surprisingly after 31’000 generations some mutants appeared that were able to use citrate, as a source of carbon instead of glucose, the inability to use citrate is one of the frame that define E.coli as a species. Researcher Zachary D. Blount dissected all the mechanism in using whole genome re-sequencing of 29 clones along the population history (Fig.1). Three steps are analyzed in details, as they were necessary to permit the evolution of Cit+ trait, potentiation, actualization and refinement.

The phylogeny (Fig.1) highlight two clades that coevolved and were maintain with the new Cit+ clade around generation 20’000 (Fig 1.). Moreover after the evolution of the citrate clade around 36’000, bacteria that have Cit+ phenotype have a SNP that is causing a defect in methyl directed mismatch DNA repair such that mutants accumulate SNPs much faster than any other clades. (Fig.1 inset).

To analyze the three steps leading to the evolution of this new trait, bacteria are ideal organism as they can be frozen and revive, with this capacity replay experiments were run and a proportion of new Cit+ mutants were re-discovered in the three clades (Fig.1) with a significantly higher proportion for the third  clade.

1) The first step of potentiation is represented by the clones of the three clades at generation 20’000 with different level of potentiation implying multiple potentiating mutation. It’s the genetic background that evolve and allow a function to be accessible by mutation. Phylogeny tells us that at least two mutations occur, one before that the three clades diverge and another one in Clade 3. The replay experiments showed that potentiation play an important role for the rise of the new function. This mechanism of potentiation can be explained in two different ways, first epistasis between the background and the mutation, this interaction will express the phenotype and second the background physically promoted the final mutation. In testing this they were not able to show evidence for one or the other hypothesis. Little clue tip the balance in favor of the epistasis hypothesis. But the arguments here are not enough convincing therefore the author focused on identifying the mutational basis of actualization.

2) Actualization is the step where weak mutants appear and is able phenotypically to use citrate; it’s also the most tractable step. The inability to transport citrate is the reason why E.coli cannot use it as a source of carbon in oxic environment. Thus in the Cit+ clones and only on these clones they found that they all have at least two tandem copies of a citrate fermentation operon (cit) that contain a citrate/succinate antiporter ,ciT (Fig. 2)

The explanation proposed by the author to the arising of this new trait is that the second citT of figure 2b is using a rnk promoter that is downstream the old citT but now due to the amplification it’s found upstream of the new citT, thus being able to be transcribe in a fusion protein Rnk-CitG-CitT.

To test this hypothesis they first tested the ability of rnk-citT module to spport citT expression in oxic environment. They constructed three low copy plasmids with rnk, citT and rnk-citT. After transformation of each plasmid in three strains, one ancestral (REL606), one potentiated clone of clade 3 (ZDB30) and one weak mutant (ZDB172) they found out that they was expression (in red in Fig. 3) of the rnk-citT module in the two potentiated and weak mutant clones. These indicate that with the rnk-citT transformation there is citT expression under oxic conditions for potentiated clones.

This lead to another question; does the transformation of this module into clones give a Cit+ phenotype? To test this they simply inserted a rnk promoter upstream of the natural copy of citT found in the genome (Fig. 4). The construct (ZDB595 in blue) made with the potentiated clone ZDB30 and the insertion of rnk promoter show in figure 4 a weak Cit+ phenotype with an increase growth compared to the red curve (potentiated clone ZDB30) and other curve that are Cit+ phenotypes and have better growth on citrate.

3) Finally refinement is the increase in efficiency of the new function by probably additional mutations. Here the main mutations that occur in the early Cit+ mutants are the increase in the number of copy of the cit operon (Fig.5a) with stabilization on later generations to three copies. To test if the number of copy influence the Cit+ phenotype, they cloned the module rnk promoter and citT into a multi-copy plasmid and transformed it into the potentiated strain ZDB30 (in red Fig. 5c). The new build strain ZDB612 (in green Fig.5c) is strongly Cit+ and is growing on citrate as a Cit+ clone (CZB152 in black), which is a good clue in favor of the refinement of Cit+ phenotype by the multiplication of the cit operon.

The neo-functionalization created by gene duplication was already observed in comparative studies, but this long-term experiment is a nice opportunity to study the history of the citrate transporter gene duplication and promoter capture that give rise to a new adaptive function. This example shows also the importance of the evolution of the potentiated genetic background that will allow a last mutation to give rise to the function.

Blount, Z., Barrick, J., Davidson, C., & Lenski, R. (2012). Genomic analysis of a key innovation in an experimental Escherichia coli population Nature, 489 (7417), 513-518 DOI: 10.1038/nature11514

 

1. How genomes evolve in natural populations? is a question that, despite to be a long-standing search for geneticists, recent molecular genomic approaches may help to understand. Their evolution among natural populations may be shaped by forces derived from different processes, such as as mutation, neutral evolution, negative or positive selection and demographic changes (Nosil & Feder 2011). In addition, the study of population divergences and species formation has an important temporal context. That is, the point in time during the evolutionary trajectory of the set of populations or related species we aim to study, can display different patterns. For instance, during initial stages the level of genome differentiation may be low. Then, as divergence proceeds (by intrinsic or extrinsic forces) this level is expected to increase in the genome, in either a clustered or an interspersed way. Further, hybridization between differentiated populations may produce the secondary infiltration of genomic regions of one population into the other. The study of the genomic complexity in natural populations may shed light into the population dynamics and the relevant processes during species evolution.

2. Staubach et al. (2012) used natural populations of house mice (Mus musculus) subspecies (M. m. domesticus which does not refer to a domesticated subspecies and M. m. musculus) to detect pattern of selection and introgression at a genomic scale. A large distribution across the world on varied habitats has characterized the past few thousand years of house mouse evolution. Two natural populations each subspecies were used. The proposed model of demographic history is depicted in figure 1 on the article [ ! ]. Mice were collected from 4 populations in Germany, France, Czech Republic and Kazakhstan, respectively. Eleven unrelated wild individuals per population were genotyped using Affymetrix Mouse Diversity Array for a total approx. 470,000 high quality SNPs. SNP calling was less reliable for M.m. musculus and lower polymorphic sites may reflect ascertainment bias towards M. m. domesticus [ + more about ascertainment bias see http://www.ncbi.nlm.nih.gov/books/NBK9792/]. Neighbor-joining trees based on the SNP set evidenced the population clustering of individuals (Ger,Fra),(Cze,Kaz). Some genomic regions, i.e. 6 MB in chr6, revealed an introgression pattern in M.m. musculus Kaz by M.m. domesticus haplotypes (Figure 1).

[ ! ] The demographic model assumes equal population sizes for the ancestral subspecies (N0) and for actual populations (N1), which could be not the case and events of bottleneck, range expansions, and effective population size (Ne) differences between populations might distort the discussed patterns.

Figure 1. Neighbor joining trees of population samples, based on Manhattan distances. The scale bars refer to the number of substitutions per SNP genotyped. (A) For all autosomal SNPs in the study, grouped by sample and rooted by outgroup species. (B) For all haplotypes of a 6MB region on chromosome 6. Some haplotypes of Kaz group with M. m. domesticus due to introgression.

 3. The estimates of linkage disequilibrium (LD) provided an indicator of the size of haplotypes physical linked, which when recognizing selected and diverged loci, may represent the expected size of genomic islands of differentiation. As we move away from the selected loci we expect a decay in LD [ + ]. Here, a linkage of SNPs beyond 100 kb is rare (Figure 2). Kaz and Cze populations showed the strongest and slower decay patterns, respectively. The first could be related with the high genetic diversity found in this population, and the latter pattern probably due to the introgressed regions. However, the removal of the introgressed regions does not eliminate completely the difference of Cze and the other populations [ ! ].

[ + ] LD plot presented here uses r2, defined as the correlation coefficient of the SNP frequencies. A value of 1 is called Perfect LD. r2 is affected by recombination rate and Ne.

[ ! ] The slow r2 decay in Cze after removal of introgression may indicate further effects of either small population sizes or the level of genetic diversity in this population, or both. This could have been discussed in the study by including details from previous reported microsatellite screens and considering additional demographic models.

Figure 2. Pairwise LD (r²) between markers over distance for each of the populations in the study.

 4. Selective sweep scans were based on the XP-CLR statistic [ + ], Rsb, delta-DAF and Fst. Thresholds (through simulations and constraints) were defined to visualize the distribution of potential selective sweep signatures across autosomes. Detected peaks seem to be scattered across chromosomes, few of them clustered and particular to specific populations (i.e. chr 6 presented three clusters of peaks for Kaz population, Figure 3). Summary statistics showed striking differences in the number of significant regions when using one or other method. Several loci were identified under potentially recent positive selection with non-significant GO category enrichment.

[ + ] In basics, XP-CLR statistic or “cross-population composite likelihood ratio test” uses population differentiation as a means to detect regions under selection and hence is particularly robust against ascertainment bias. It works with biallelic multilocus data, searching for regions in the genome, where the change in allele frequency at the locus occurred too quickly (as assessed by the size of the affected region) to be due to random drift. Details of this method are presented in Chen et al. (2010).

Figure 3. Distribution of selective sweep signatures as determined by the XP-CLR statistic for the two population comparisons across all autosomes.

 5. Introgression was evaluated by comparison of focal vs two potential source populations. Each comparison includes the population of the same subspecies as one possible source, and the two populations from the other subspecies as alternative source. Large introgressed haplotypes were detected in Cze from M. m. domesticus, these regions were distributed across all chromosomes. Staubach et al. decided to compare the introgression patterns with a neutral migration-introgression model by simulating dataset with different population sizes and migration rates. The results suggested that the observed frequency of introgressed haplotypes would require higher migration rates than in a neutral model (4Nm=1). However, with these migration rates the expected fraction of introgressed genome should be higher than the observed for the Ger, Fra and Kaz populations. The authors suggested and emphasized that the large fraction of introgressed haplotypes is subject to forces that lead to a faster increase in frequency than expected by chance (i.e. a deterministic or selected process) [ ! ]. Incomplete lineage sorting was discarded by simulations without migration. Some of the introgressed regions were visualized as colored-blocks using UCSC browser.

[ ! ] Despite the the authors refrained from an even deeper exploration of the demographic parameter space, we discussed that the oversimplified demographic model used may neglect relevant differences in populations sizes or migration rate asymmetries. The mosaic nature of introgression has been reviewed in different systems by Arnold & Noland (2009). Here, the role of positive selection in shaping the genomes of house mouse populations, via introgressed haplotypes, is not completely understood.

 Finally, the differentiation between house mouse subspecies evidenced a stage of genomic divergence similar to closely related species pairs (check the previously discussed Flycatcher’s genome paper). These two articles evidenced that genomic divergences between two sister species (completely separated lineages) and between subspecies (closely related populations) seems do not to involve few genomic regions or single loci, but differentiation across all chromosomes.

 References

Arnold, M. & Noland, M. (2009). Adaptation by introgression. Journal of Biology 2009, 8:82 (http://jbiol.com/content/8/9/82)

Nosil, P. & Feder, J. (2012). Genomic divergence during speciation: causes and consequences. Phil. Trans. R. Soc. B. vol. 367 (1587), 332-342.

Chen, H., Patterson, N. Reich, D. (2010) Population differentiation as a test for selective sweeps. Genome Res. 20 (3): 393-402.

Staubach F, Lorenc A, Messer PW, Tang K, Petrov DA, & Tautz D (2012). Genome patterns of selection and introgression of haplotypes in natural populations of the house mouse (Mus musculus). PLoS genetics, 8 (8) PMID: 22956910