Research Article: Molecular adaptation in Rubisco: Discriminating between convergent evolution and positive selection using mechanistic and classical codon models

Date Published: February 12, 2018

Publisher: Public Library of Science

Author(s): Sahar Parto, Nicolas Lartillot, Art F. Y. Poon.


Rubisco (Ribulose-1, 5-biphosphate carboxylase/oxygenase) is the most important enzyme on earth, catalyzing the first step of photosynthetic CO2 fixation. So, without it, there would be no storing of the sun’s energy in plants. Molecular adaptation of Rubisco to C4 photosynthetic pathway has attracted a lot of attention. C4 plants, which comprise less than 5% of land plants, have evolved more efficient photosynthesis compared to C3 plants. Interestingly, a large number of independent transitions from C3 to C4 phenotype have occurred. Each time, the Rubisco enzyme has been subject to similar changes in selective pressure, thus providing an excellent model for convergent evolution at the molecular level. Molecular adaptation is often identified with positive selection and is typically characterized by an elevated ratio of non-synonymous to synonymous substitution rate (dN/dS). However, convergent adaptation is expected to leave a different molecular signature, taking the form of repeated transitions toward identical or similar amino acids. Here, we used a previously introduced codon-based differential-selection model to detect and quantify consistent patterns of convergent adaptation in Rubisco in eudicots. We further contrasted our results with those obtained by classical codon models based on the estimation of dN/dS. We found that the two classes of models tend to select distinct, although overlapping, sets of positions. This discrepancy in the results illustrates the conceptual difference between these models while emphasizing the need to better discriminate between qualitatively different selective regimes, by using a broader class of codon models than those currently considered in molecular evolutionary studies.

Partial Text

Rubisco (Ribulose-1, 5-biphosphate carboxylase/oxygenase) is an enzyme that catalyzes the major step in carbon fixation in all photosynthetic organisms. It is the most abundant protein on earth [1], as it encompasses up to 50% of soluble proteins [2] and 20–30% of total nitrogen [3] in C3 leaves. During carbon fixation, Rubisco reacts with both CO2 and O2 as its substrate, with poor distinguishing ability. The carboxylase activity results in the incorporation of inorganic carbon into the metabolic C3 pathway, whereas the oxygenase activity boosts the photorespiration pathway. The latter prompts both energy consumption and CO2 loss.

Amaranthaceae is one of the plant families with the largest number of C4 species. This makes it a suitable case for Differential Selection (DS) analysis. Based on a multiple sequence alignment of rbcL genes and an annotated phylogenetic tree of Amaranthaceae, our DS model captures site-specific amino acid preferences as vectors of 20 fitness factors (for the 20 amino acids) under each condition. Then, contrasting for each position, the fitness factors estimated in the two conditions of interest (here, in the C3 and C4 regimes), allows us to identify positions for which the fitness of a specific amino acid has undergone a significant change, either upward or downward, associated with the transition between the C3 and the C4 photosynthetic regime (see Methods).

Rubisco has long been known to be under positive selection [19]. In addition, it has been shown that Rubisco has been evolved in different structural forms and functions [47]. It exemplifies a convergent evolution of enzyme properties in its phylogenetic pathway. One example of this convergent evolution happens between C3 and C4 plants through crossing the fitness landscape [20,23,48]. Therefore, the complex molecular evolutionary patterns displayed by the Rubisco gene in eudicots represent an interesting case-study for assessing and comparing current codon modeling strategies [20]. In this respect, our comparative analysis, by making an inventory of the amino acid-positions in rbcL sequences that are positively or differentially selected in C3 and C4 Amaranthaceae family, emphasizes the fundamental difference, in scope and meaning, between the two main classes of models currently considered in the literature: on one side, classic codon models based on the measure of the overall dN/dS, whose focus is primarily on positive selection; and on the other side, Differential Selection models, whose aim is instead, to detect convergent patterns of directional selection associated with repeated transitions between known evolutionary regimes. Our analysis also emphasizes that none of the models considered here, either omega-based models or mutation-selection approaches, offers a completely satisfactory explanation of the complex patterns of molecular evolution observed in Amaranthaceae, and probably also present in other species groups—thus suggesting that further developments are still needed on the front of phylogenetic codon models.




0 0 vote
Article Rating
Notify of
Inline Feedbacks
View all comments