Sunday, 1 June 2014

Evolution of RubisCO enzyme under structural constraints

(Reason of this post: I had to make a summary of my recent work for an application. So, here it is!)

Stability-activity trade-off constrains the adaptive evolution of RubisCO
Studer RA, Christin PA, Williams MA, Orengo CA.
Proc Natl Acad Sci U S A. 2014 Feb 11;111(6):2223-8. doi: 10.1073/pnas.1310811111.
Epub 2014 Jan 27. PMID:  24469821 Free PMC Article

Importance of the RubisCO enzyme
The ribulose-1,5-bisphosphate carboxylase/oxygenase [EC:] ]is the central enzyme in photosynthesis and one of the most abundant proteins in the world. The biological unit is a complex composed by eight large subunits and eight small subunits (Fig. 1A-B). Its catalytic activity is to fix the CO2 in the Calvin cycle and is performed by the large subunit (Fig. 1C). This reaction is extremely slow, with up to 3 molecules fixed per second. The most prevalent system in plants is the C3 photosynthesis pathway. In flowering plants (angiosperms), some lineages have evolved to a C4 photosynthesis pathway. In this pathway, the RubisCO is twice faster, with the ability to fix up to 6 molecules per second. A striking fact is that the emergence of this C4 photosynthesis pathway occurred in many divergent lineages, in a convergent manner. A major problem in the RubisCO is its dual affinity for CO2 and O2, which can lead to undesired photorespiration (fixation of O2), instead of photosynthesis (fixation of CO2). As this problem increases with the activity and can be costly to the plant, C4 plant lineages have also developed cellular mechanisms to concentrate the CO2 around the RubisCO, which prevent photorespiration.

Figure 1: Structural view of the RubisCO enzyme. A) Front view, with the large subunits in blue/yellow and the small subunits in purple; B) Top view, with the central solvent channel visible; C) Front view, with the two large subunits (in ribbons) forming the catalytic dimer; D) Sites detected under functional divergence.

Theoretical concepts in the evolution of proteins
Proteins are long chains of amino acids that tend to organise in the 3D structural space. Generally, proteins fold in a way to maximise the numbers of favourable atomic contacts, and to reduce the global free energy (ΔG, in kcal mol-1). However, enzymes need some degree of freedom, in order to move between different conformation, i.e. allowing the opening and closing of the catalytic pocket. Similarly, active sites are unfavourable in term of stability. This is why proteins are said to be marginally stable. The importance of the stability effect is seen when the replacement of an amino acid to another one occurs. Most amino acid replacements are very likely to disrupt the stability of the protein, and thus only a very few subset of amino acid changes is tolerated during evolution (Fig 2). However, it may happen that the change to another function, or the optimisation of a current function, can shift the stability towards its neutral area. This model is called stability-activity tradeoffs. These evolutionary events can be preceded by capacitive stabilising mutations and/or followed by compensating stabilising mutations.

Fig 2: The evolution of protein stability as a constrained “random walk” through sequence space. Protein sequences are represented as circles (yellow circles indicate sequences that are selectively neutral; red circles indicate those that have deleterious effects). Missense mutations are shown as the connecting labelled arrows. The series from 1 to 6 represents a trajectory of fixations through sequence space. The series of mutations from 1 to 3 represents a neutral “meandering” through sequence space. The adaptive fixation 4, which is advantageous despite its effects on stability and aggregation, induces a strong selection pressure for the compensating mutation 5 to restore stability to the neutral zone (reproduced from DePristo et al. 2005, Nat Rev Genet. 6(9):678-87).

Biological question
The RubisCO enzyme provides a perfect framework to study how enzymes evolved under structural constraints. In this study, I wanted to determine what are the residues responsible for the increase in catalytic activity, where are they located in the 3D structure and how they alter the stability of the complex.

Twelve amino acids are likely to be responsible for functional divergence
The phylogenetic-based algorithm TDG09 (Tamuri AU et al. 2009, PLoS Comput Biol 5(11): e1000564) aims to identify shift in selective pressures between groups of amino acids. Twelve sites have been identified (at 1% confidence) to be under strong selective pressure in RubisCO from C4 plant lineages compared to RubisCO from C4 plant lineages. None of these sites are part of the active site (which is, due to its importance, 100% conserved), but they are at different key positions in the 3D structure (Fig 1D), such as in the contact interface between subunit or in the opening/closing loop.

Reconstruction of ancestral sequences and structures
The intermediate sequences of RubisCO have been reconstructed under maximum likelihood, based on the phylogenetic trees of the 240 plant lineages. These sequences have been used to reconstruct the ancestral 3D structures by homology modelling. We have obtained a very high accuracy in each step, thanks to the extreme conservation of the RubisCO sequences (>90% of identity and no insertion/deletion, despite millions years of evolution) and the high quality of the crystal structure, which serves as template (1.35Å). We then were able to describe precisely all mutations that occurred at a particular time point, especially the contribution to the stability of these mutations. The stability effect of these substitutions (ΔΔG, in kcal mol-1) has been estimated by FoldX and the result has been mapped on the phylogenetic (Fig. 3).

Figure 3: Mapping of stability effect on the phylogenetics tree. This is an extract of the full phylogenetic tree, which has been built using the 240 sequences analysed in this study. C3 plants are in light green and C4 plants are in dark green. Blue slices indicate the percentage of stabilising mutations, while red slices indicate the Mutations L270I, A281S and A328S are frequently found in C4 and are destabilising. A328S is very close to the loop opening the catalytic pocket. A destabilising residue at this position could accelerate both the opening and closing of that loop. The mutation M309I is also frequently observed in C4, but has no effect in term of stability. Tree visualisation made with EvolView.

Stability-activity trade-off constrains the adaptive evolution of RubisCO
The comparison of the different C4 lineages led to the conclusion that the evolution of RubisCO to new environmental constraints (the C3->C4 transition) is constrained by stability-activity trade-offs (Fig. 4). Statistically speaking, there is a significant excess of destabilising mutations observed during the C3-C4 transition (p-value = 0.0080) and a significant excess of stabilising (compensatory) mutations observed right after the C3->C4 transition (p-value < 0.0001). While not statistically significant, we also observed an accumulation of slightly stabilizing mutations (which create the capacity to tolerate the functionally destabilizing mutations) by a long period before the C3->C4 transition

Figure 4: Frequency plot of mutations according to the positions relative to the transition C3->C4. Branches are annotated relative to the node where the transition C3->C4 occurred (position=0). Negative nodes are prior to the transition and positive nodes are after the transition. Interestingly, there is a peak a destabilising mutations (in orange) on the branches where the adaptation (transition C3->C4) occurred, followed by peak a stabilising mutations (in blue) on the posterior branches. This suggests that some mutations change the function (and destabilise the structure) and other mutations follow to compensate for this loss of stability.

Concluding remarks
These results demonstrated that the evolution of an enzyme, here the RubisCO, can be under strong structural constraints and that adaptive mutations are balanced between stabilising and destabilising effects. This shows that stability-activity trade-offs found in laboratory experiments (i.e. Bloom & Arnold 2009, PNAS 106:9995) have direct counterparts in the past 120 million years of plant evolution. A follow up of this project would be to extend this analytical framework to other enzymatic families where functional divergence is observed, and to apply this knowledge to directed-evolution experiment.

No comments:

Post a Comment