The ecdysone receptor (ECR-USP) has undergone a dramatic phase of mutations during the emergence of the Mecopterida superorder
[4–6]. This acceleration of evolutionary rate is correlated with structural and functional changes of the LBP of USP and of the heterodimer interface between ECR and USP
[7, 8]. These results suggested that the maintenance of the ecdysone pathway was achieved, at least in part, through molecular adaptation of the USP LBD. We thus tested the possibility that this domain was under positive selection during the emergence of Mecopterida.
Regions of USP LBD under positive selection during Mecopterida emergence
Our previous results suggested that the main functional shifts affecting USP LBD of insects occurred during the emergence of Mecopterida, approximately 280–300 million years ago (early Permian). We therefore used branch-site models to detect changes in selective constraints along this branch. Importantly, we also analysed two more recent branches within the Mecopterida clade in order to decipher the successive state of evolutionary forces acting on USP LBD. Indeed, the interpretation of changes in evolutionary rates must take into account all the biological complexity of the protein, such as the dynamic of the process or the distinction between functional modules. Using this approach, we were able to detect positive selection at amino acids located on the LBP and on the heterodimerisation interface of USP LBD.
The episode of positive selection is correlated with the acquisition of a large LBP filled by fatty acids. In the basal insect Locusta migratoria (Orthoptera), USP is able to bind 9-cis retinoic acid, like homologous vertebrate RXR, which natural ligands are fatty acids
. In Bemisia (Hemiptera) and Tribolium (Coleoptera), USP is unable to be activated by RXR ligands and acts as a constitutively silent partner of ECR. In fact, in these species, residues of the loops L6 and L11 fill the deep unliganded pocket and, therefore, USP lacks a proper LBP
[7, 28]. All the four residues that underwent strong positive selection during the emergence of Mecopterida are located at the bottom of the deep pocket, which is similar to the LBP of RXR. These residues make contacts with the tails of the phospholipid in Mecopterida structures. Later during the diversification of Mecopterida, the same positions became under purifying selection. These results suggest that the Mecopterida pocket was positively selected for a new function, which then became highly constrained and remained largely unchanged until now. What was this innovation? We can hypothesize that the presence of phospholipids allowed the stabilisation of the new Mecopterida LBD. Indeed, this structure is characterised by the acquisition of a new loop L1-3 that locks the LBD in a conformation similar to the one seen for antagonist-bound nuclear receptors
[25, 26, 29]. In that perspective, phospholipids would be analogous to pharmacological chaperone, which are small molecules that help to maintain the correct folding of mutant proteins
. This interpretation is in agreement with a previous proposition, based on an information-theoretic approach, according to which the ligand-contacting residues perform a dual role: a functional role in ligand recognition and a structural role as core residues
. Thus, the structure of the LBP of USP was highly plastic during insect evolution, adopting three of the four states that are known for nuclear receptors: nutritional sensor (basal insects), real orphan (Hemiptera, Coleoptera) and receptor with a constitutive activity (Mecopterida). The first evolutionary transition involved a loss of function, from a sensor to an orphan receptor
. Our results highlight the second transition, from an orphan receptor to a constitutive receptor, which could be seen as a gain of function.
We also detected positive selection in the dimerisation interface during the emergence of Mecopterida. This episode of positive selection is correlated with a loss of ability to homodimerise for both ECR and USP. Non-Mecopterida ECRs are able to homodimerise and are active without USP
[42–44]. By contrast, even in presence of their ligand, the ECR LBD monomers of Heliothis and Drosophila are very unstable in solution and need USP for solubilisation and stabilisation
[45, 46]. Similarly, non-Mecopterida USPs are able to form homodimers, whereas Mecopterida USPs are not able to do so
[7, 42, 47, 48]. It seems therefore that Mecopterida USP acts as a chaperone-like partner for ECR stabilisation. We have shown previously that, during the emergence of Mecopterida, a relaxation of evolutionary constraints occurred at a site that affects homodimerisation but not heterodimerisation
[8, 49]. This residue is the glutamine Gln457 in helix H10, for which we provide here evidence for strong positive selection on the branch leading to Mecopterida (Figure
5A). This result confirms our first hypothesis proposing that substitution of this site located in the core of the dimerisation surface may have induced a loss of homodimerisation for USP. Positive selection was found at those three sites in an independent study
Importantly, if the homodimerisation ability has been lost in Mecopterida, on the contrary, the heterodimerisation between ECR and USP has been conserved by all arthropods and was even reinforced in Mecopterida. How can we explain the maintenance of the functional interaction between ECR and USP, despite the challenging burst of mutations that occurred in the stem lineage of Mecopterida? The heterodimerisation between ECR and USP depends largely on a core of residues localised in helices H9 and H10 of the LBDs. While the sequence of this interface is well conserved in all insects, in Mecopterida, the loop L8-9 of USP interacts with the helix H7 of ECR, creating a symmetric and larger heterodimerisation surface
. This novel surface is a consequence of the Mecopterida specific structural divergences of USP, where the loop L1-3 and the presence of a phospholipid, which is in tight contact with L1-3 and H3, induce a 15°-torsion of a sub-domain of USP LBD towards ECR. If the maintenance of heterodimerisation depends on this new surface, then positive selection should be detected at this specific enlargement on the stem lineage of Mecopterida. We have confirmed this prediction for two sites of USP: Arg426 in helix H9 and Arg399 in helix H7 (Figure
5A). The arginine Arg426 is involved in the Mecopterida specific enlargement of the heterodimerisation surface, while the Arg399 was predicted to form an electrostatic interaction with ECR in the ancestral Mecopterida model
. In addition, we detected positive selection at the site Asp452, which is located in the core interface and makes contacts with two residues of the helix H9 of ECR. It is thus possible that this substitution played a role in the reinforcement of the ECR-USP heterodimerisation. If we assume that the reinforcement of ECR-USP heterodimerisation is a selectively favoured phenotype, then every amino acid replacement causing a slight increase in interaction would be selectively favoured. However, once binding is optimum, no further changes will be favoured. Thus, a phase of purifying selection would follow an initial phase of positive selection. We have confirmed this expectation by showing that the same four positions described above became under purifying selection during the diversification of Mecopterida. In conclusion, the dimerisation interface of USP has experienced modifications of selective constraints during the emergence of Mecopterida, followed by a phase of constrained evolution.
Although our results suggested that coevolution has maintained the interaction between ECR and USP, we could not detect any significant modification of evolutionary constraints in the dimerisation surface of ECR
. Recent random-sites model analyses confirmed the absence of positive selection in ECR
. It seems therefore that the helix H7 of ECR, which is implicated in the new Mecopterida interface with USP, was able to adapt to its changing partner in a manner independent of selection. In order to explain the evolution of this domain, it may be necessary to explore the role of molecular drive, which includes all the DNA turnover mechanisms that change the genetic composition of a population, independently of selection and drift
. There is indeed a renewed interest in the evolutionary forces, other than natural selection or drift, that can induce co-evolution between functionally interacting genetic systems in order to maintain essential functions
[15, 52]. However, interestingly, six residues with elevated ω were identified in the core interface of Mecopterida ECR, suggesting that the dimerisation core of ECR, like that of USP/RXR, may also be under relaxed constraint in the Mecopterida clade
Within the family of nuclear receptors, evidence for positive selection has been reported in a few other cases, such as PXR (NR1I2), CAR (NR1I3) and ERα (NR3A1)
[33, 53, 54]. For example, positive selection was detected at amino acids involved in ligand binding of the frog PXRs, which have lost broad specificity for ligands and gained efficient activation by endogenous amphibian-specific benzoates
. A recent study has shown that RXR, the USP homolog, was under positive selection immediately following the first round (1R) and second round (2R) of whole genome duplications in vertebrates
. Most of the positive selected sites were located either in the LBD or in the N-terminal region, which contain activation domains (AF-1 and AF-2) and phosphorylation sites that control RXR expression. During this evolutionary transition, it seems that positive selection did not affect the LBP or the dimerisation interface of RXR (except for one site of helix H9 on the post-duplication branch leading to RXRγ). The authors thus conclude: “The presence of positive selection/accelerated rates in the paralogs, coupled with the evidences of altered expression of paralogous genes explains the preservation of the additional RXR copies following genome duplication”
. It would be interesting to analyse the evolution of the N-terminal region of USP and ECR, but the structure and the function of these highly divergent domains are largely unknown in insects.
Episodic evolution of USP
In conclusion, our results allow to define three phases of evolution for the LBD of USP in Mecopterida. We can now propose a model to explain this episodic mode of evolution. We have previously proposed that the ancestral state of USP LBD was characterised by the absence of a LBP and the ability to form homodimers as well as heterodimers with ECR
[7, 8]. During the emergence of Mecopterida, an unexplained acceleration of evolutionary rate was probably responsible for the insertion of 11 residues in the loop L1-3. This insertion is the primordial event in our hypothesis, because it changed the structure of the LBD in two ways. First, it increased the size of the heterodimer interface. Second, it restored a LBP. However, this new pocket is very large and filled with phospholipids, which is unusual for a nuclear receptor of the NR2B group. These modifications allowed the exploration of new functions, as revealed by the detection of positive selection at these two regions during the first phase of Mecopterida evolution. The new LBP probably allowed the maintenance of a correct folding for the whole LBD, while the new extended surface of dimerisation was associated with a reinforcement of the obligatory partnership between ECR and USP, at the expense of homodimerisation. Therefore, the correlation between positive selection and functional innovations suggests that the USP LBD underwent molecular adaptation during the emergence of Mecopterida. One limit of this kind of analysis is the difficulty to distinguish between positive selection and relaxation of negative selection
. However, the availability of crystal structures and functional data provide a priori biological hypotheses regarding the amino acids upon which positive selection is expected. Once the new functions are effective, no further changes will be favoured. Thus, the initial phase of positive selection should be followed by a second phase of purifying selection. In agreement with this hypothesis, we found that the new functional domains were under strong negative selection during the subdivision of Mecopterida between Amphiesmenoptera and Antliophora. This second phase of evolution explains the conservation of the new surfaces among Mecopterida species. Finally, we could characterise a third phase of evolution during the subdivision of Diptera into Brachycera and Nematocera. On this branch, USP LBD evolved mainly under a relaxation of negative selection, with neutrality at some sites. However, the new Mecopterida functional domains remained highly constrained (except the tyrosine Y372 in the LBP of Diptera). The alternation of selective and conservative periods during the transformation of USP LBD in insects provides an example of the episodic model of protein evolution, a major mechanism recently revealed by application of new methods at the genomic level
[12, 56, 57].