(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



Dimeric allostery mechanism of the plant circadian clock photoreceptor ZEITLUPE

['Francesco Trozzi', 'Department Of Chemistry', 'Center For Research Computing', 'Center For Drug Discovery', 'Design', 'Delivery', 'Southern Methodist University', 'Dallas', 'Texas', 'United States Of America']

Date: 2021-09

In Arabidopsis thaliana, the Light-Oxygen-Voltage (LOV) domain containing protein ZEITLUPE (ZTL) integrates light quality, intensity, and duration into regulation of the circadian clock. Recent structural and biochemical studies of ZTL indicate that the protein diverges from other members of the LOV superfamily in its allosteric mechanism, and that the divergent allosteric mechanism hinges upon conservation of two signaling residues G46 and V48 that alter dynamic motions of a Gln residue implicated in signal transduction in all LOV proteins. Here, we delineate the allosteric mechanism of ZTL via an integrated computational approach that employs atomistic simulations of wild type and allosteric variants of ZTL in the functional dark and light states, together with Markov state and supervised machine learning classification models. This approach has unveiled key factors of the ZTL allosteric mechanisms, and identified specific interactions and residues implicated in functional allosteric changes. The final results reveal atomic level insights into allosteric mechanisms of ZTL function that operate via a non-trivial combination of population-shift and dynamics-driven allosteric pathways.

From growth to flowering, many aspects of plants development are regulated by daily and seasonal changes in light intensity and duration. These circadian clocks employ light sensing proteins to adjust their functional states responding to different wavelengths and light intensities that result from daily and seasonal oscillations in our environment. The mechanistic insights into how the absorption of light causes changes in their functional states remain elusive for many circadian clock proteins. ZEITLUPE (ZTL) is a circadian clock protein found in Arabidopsis thaliana. Previous experimental studies demonstrated that ZTL functions as a dimer, which could adopt either parallel or anti-parallel orientations depending on the presence of blue-light or darkness, respectively. In this study, the authors investigated the allosteric mechanism of ZTL in both parallel and anti-parallel dimeric states through computational means. Two double mutants with significant functional deviation, G46S:G80R and V48I:G80R, were also studied as comparison. As the results, the authors provided mechanistic and dynamic models with atomistic details underlying the functions of this circadian clock protein.

Funding: The authors acknowledge funding sources, including NIH research grant R15GM122013 to P.T. and NIH research grant R15GM109282 to B.D.Z. https://www.nih.gov/ The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Copyright: © 2021 Trozzi et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

The remainder of the paper is organized as follows. First, it is presented how Gln154 responds to the change in state of the flavin and the impact of G46S and V48I variants on Gln154 dynamics. Then how the different conformations that Gln154 adopts impact the interactions between residues in the functional N- and C-termini are investigated. Several intramolecular interactions that could be crucial to propagate the signal throughout the protein are also investigated. Particularly, how these intramolecular interactions affect the population of functional metastable states on the conformational landscape of the protein is evaluated. Towards these goals we adopted the dimensionality reduction technique tICA and Markov state modeling. Lastly, with the use of Machine Learning we elucidated characteristic structural changes that correlate with different metastable states involved in ZTL allosteric process.

The ensemble of computational tools employed in this study enabled us to validate and integrate otherwise inaccessible information to the experimental investigations of ZTL allostery. The MD simulations support the stability of dimer complexes, and also reveal the propagation of the allosteric perturbation starting from the flavin photoreceptor to the overall protein structure.

Herein, we employ atomistic simulations that integrate Markov State modeling, and supervised machine learning classification models to elucidate a dynamic allosteric process gating ZTL function. These results confirm a complex and dynamic ZTL landscape that links sequence motifs within the Aβ and Iβ strands to an insertion between the E/F helices (E-F loop), which drives the reorientation of the ZTL LOV-dimer. The results provide keen insight into the role of Gly46 and Val48 mutations in disrupting ZTL allostery that have implications for plant physiology, and broader impacts to LOV allostery.

To surpass this limitation and obtain a dynamical description of ZTL allostery, exploring the conformations that lie in between the starting and end points in the allosteric process to ultimately extrapolate the structural changes during this process is necessary. In recent years, computational approaches to advance a quantitative characterization of allosteric mechanism in proteins have been developed. Molecular dynamics (MD) simulations are a useful tool to explore the conformational landscape of a protein by providing the evolution of a system over time at an atomistic level.[ 26 – 28 ] Combined with MD simulations, Markov state model (MSM) approaches can provide connectivity maps of states on the free energy landscape, estimate the effect of allosteric perturbations on the conformational equilibrium, and rigorously describe kinetics of allosteric transitions.[ 19 , 29 – 34 ]

Examination of G46S and G46A structures revealed coupling between C-terminal Gln dynamics, the residue identity at position 46, and global conformational changes.[ 16 , 23 ] G46S and G46A structures mimic a light-state like orientation of the active site Gln that is coupled to the movement of a conserved Phe residue (Phe156), C-terminal salt-bridge formation, and the movement of the Ncap. Specifically, rotation of Gln154 towards the buried conformation disrupts contacts to the N/Ccap to induce a 180° rotation about the LOV-dimer interface ( Fig 1 ). However, these observations relied on static structures of allosteric variants that may block observation of dynamic motions necessary for light/dark regulation of ZTL structure, and may not necessarily reflect allostery in WT proteins. Further, the use of allosteric variants to trap specific ZTL configurations only allows for snapshots of starting and end points in an allosteric process, and not the dynamic motions gating the altered mechanisms of signal transduction, thereby limiting understanding of how the ZTL allosteric landscape diverges from other LOV proteins.

Recent structural and biochemical studies of the Arabidopsis thaliana circadian clock photoreceptor ZEITLUPE (ZTL) have called the consensus LOV signaling mechanism into question, and have suggested that LOV allostery may be more fluid across the LOV superfamily.[ 16 , 23 ] Structures of ZTL proteins in the dark- and light-states indicate divergent signaling elements compared to other LOV proteins: i) The conserved Gln residue (Gln154) adopts a dynamic and heterogeneous population rotating between buried and exposed conformations in the dark. Subsequent photoactivation biases the Gln population towards the buried conformation seen in all other LOV-structures to induce downstream signal transduction[ 16 , 23 ]. ii) Q154L variants are tolerated in ZTL and are naturally present in ZTL family members[ 23 ]. iii) Evolutionary analysis revealed that the altered signaling pathway was dependent on conservation of two residues in the Aβ-strand, Gly46 and Val48, that are co-conserved in all ZTL proteins[ 16 ]. Further, examination of other LOV structures with Gly residues at the position equivalent to Gly46 revealed altered Gln dynamics in these proteins, thereby highlighting conservation of allosteric motions linking the C-terminal Gln, to N-terminal conformational changes dependent on the residue identity at position 46 in ZTL[ 23 – 25 ].

Structural and computational approaches have been used to study LOV signal transduction in a number of systems.[ 11 – 20 ] These approaches have identified a mechanism of signal transduction dependent on three conserved elements: i) UVA/Blue-light triggers formation of a covalent adduct between a flavin cofactor (Riboflavin/FMN/FAD), and a conserved Cysteine residue resulting in sp 3 hybridization of the flavin C4a position and protonation of the flavin N5 position[ 2 , 15 ]; ii) N5 protonation causes a conserved Gln residue, coplanar to the isoalloxazine ring (buried conformation), to flip to optimize H-bonding interactions to N5, and alter H-bonding interactions with elements in the Aβ-strand[ 11 , 12 ]; iii) The Gln flip and altered H-bonding interactions induce conformational changes within a central five-stranded β-sheet and/or N/C-terminal extensions to the core LOV domain (Ncap/Ccap)[ 13 , 16 ]. The resulting consensus mechanism indicates that the conserved Gln residue is indispensable for LOV-signaling, and N5 protonation is both necessary and sufficient for LOV allostery.[ 11 , 21 , 22 ]

Light-Oxygen-Voltage (LOV) domains are widespread in nature, where they couple sensing of UVA/Blue light into regulation of diverse modes of biological activity.[ 1 – 3 ] Central to their function is an innate ability to couple cofactor chemistry to dynamic allosteric changes in protein conformation to regulate either protein-protein interactions, or activity of signal-transduction domains N- or C-terminal to the LOV domain.[ 2 ] The ability of LOV domains to regulate diverse signaling elements has led to their widespread use in optogenetic tools,[ 4 – 6 ] and has identified them as model proteins for studying allostery, the change of the protein structural conformation or states distribution due to a perturbation of a non-functional secondary site.[ 7 – 10 ]

Results

To elucidate a model of ZTL allostery, we performed a series of MD simulations on ZTL structures for either wild-type (WT) or allosteric variant forms of the isolated LOV domain. Specifically, we simulated the WT dark-state in an anti-parallel conformation (PDB ID: 5SVG), the light and dark anti-parallel conformations of the V48I:G80R variant that disrupts light-driven allostery (PDB ID 5SVV and 5SVW, respectively), WT and G46S:G80R (PDB ID 6WLP) variants in a parallel conformation that mimic the light-state conformation of Q154, and transient structures for both the anti-parallel and parallel conformations. All systems subjected to the simulations in this study are listed in Table 1. Initial focus was placed on resolving several important limitations of allosteric variant structures. Namely, we anticipated to verify two key aspects of ZTL signaling independent of allosteric variants or crystal contact restraints: 1) Verify stability of the ZTL LOV parallel dimer in WT light-state proteins (Protein Stability Analysis in S1 Text); 2) Examine Q154 conformational states and dynamics in WT and allosteric variants of ZTL LOV dimers.

Role of Gln154 in the function of ZTL dimer ZTL function within the circadian clock is dictated by daily changes in light intensity. The consensus model of LOV signaling involves a light-driven Gln-flip in response to changes in flavin N5 protonation.[11,16,17,24] In most LOV structures, the conserved Gln residue occupies a buried conformation coplanar to the flavin isoalloxazine ring. In the dark state, N5 is unprotonated, with the active site Gln forming H-bonds to the N5 and O4 positions via its amide functionality.[16,17,24] Upon illumination, N5 becomes protonated, leading to 180° rotation of the active site Gln to allow H-bond formation between the Gln-carbonyl group and N5-H.[16,17,24] The result is a disruption of contacts with residues in Aβ strand promoting conformational changes in N/C-terminal extensions to the LOV core.[11,16] Structures of ZTL indicate an alternative mechanism, where in the dark state Gln154 is dynamically adopting heterogeneous conformations between exposed (perpendicular to the isoalloxazine ring) and buried conformations. Light-activation was then predicted to drive rotation of Gln154 towards the buried conformation leading to downstream signal transduction.[16] To characterize the protein structural and dynamical profile related to the Gln154 switch, we focus on the analysis of Gln154 orientations corresponding to different FMN states in WT and allosteric variants. Based on the hydrogen bonds between Gln154 and FMN, we identified three main conformations: exposed, buried-I, and buried-II. The exposed conformation is identified when the only interaction presented between the FMN and Gln154 is a hydrogen bond between the side chain amino group and the O4 position of the FMN (Fig 2B). The buried-I conformation is formed when a hydrogen bond between the Gln154 side chain amino group and the N5 of the FMN is present (Fig 2C), in addition to the hydrogen bond in the exposed conformation. The buried-II conformation is formed when the carbonyl moiety of the Gln154 side chain forms a hydrogen bond with the protonated N5 of the FMN (Fig 2D). To quantify which conformations of Gln154 are sampled in different ZTL structures and states, we used a combination of the chi1, chi2, and chi3 dihedral angles (S4 Fig), which can uniquely identify different Gln154 conformations. The representative values for each conformation are provided in S1 Table. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Conformational analysis of Gln154. a) Overlap of different Gln154 conformations. Representative Gln154 structures of b) Exposed conformation, c) Buried-I conformation, d) Buried-II conformation, e) Exposed-II conformation, f) Extended conformation. Hydrogen bonds formed by Gln154 shown in dashed line. g) Percentage of different conformations adopted by Gln154 during the simulations for ZTL structures in the dark state. h) Percentage of different conformations adopted by Gln154 during the simulations for ZTL structures in the light state. https://doi.org/10.1371/journal.pcbi.1009168.g002 To quantify which conformations Gln154 explores in different ZTL structures and states, we performed a similarity analysis. For each frame, the set of dihedrals are compared to the representative dihedrals of the Gln154 conformation described above. The conformation is assigned based on similarity (Fig 2G and 2H). Consistent with the WT structures, these three conformations differentiate dark- and light-state configurations of WT ZTL. Specifically, in the dark state, Gln154 is heterogeneous and samples both exposed and buried-I conformations with a preference for the buried-I conformation (Fig 2B and 2C). Photoactivation leads to a strong bias toward the buried-II conformation observed in the light-state ZTL and other LOV structures, supporting the ordering of Gln154 locus. In contrast to WT MD simulations, exposed, buried-I, and buried-II conformations are insufficient to describe the conformational states of Gln154 in allosteric variants G46S:G80R and V48I:G80R. Specifically, in V48I:G80R a significantly exposed conformation (chi2 angle of approximately -177°; Exposed-II conformation) with an additional stabilizing H-bond formed with Asn123 is the predominant orientation (Figs 2E and S5). The exposed-II conformation is consistent with V48I:G80R structures, which are trapped in a predominantly dark-state conformation and functions as dominant-dark proteins in vivo.[16] In G46S:G80R, a different additional conformation is present, in which the amino group of the Gln154 side-chain forms a H-bond with the O4 position of FMN in addition to the H-bond in the buried-II conformation (Fig 2F). We refer to this latter state as an extended conformation. The Gln conformational analysis provided insight into both the ZTL mechanism, and variances between Gln dynamics in MD simulations and the static structures. First, as noted above, dihedral-based conformation classification demonstrates a dynamic dark-state Gln154 conformation that flips between buried-I and exposed conformations. A dynamic dark-state Gln conformation is consistent with WT ZTL structures.[16,23] Light-activation leads to Gln154 adopting a buried-II conformation for over 80% of the trajectories (Fig 2H), confirming light-driven ordering of Gln154 in a buried conformation. In contrast to WT proteins, G46S:G80R Gln conformational dynamics deviates from those observed in static crystal structures. Experimental studies showed that in G46S:G80R and G46A:G80R structures Gln154 largely occupied a single conformation where the Gln154 side chain was rotated toward the buried-I conformation. In contrast, MD simulations indicated that G46S:G80R could sample the exposed conformation, and could further sample a unique extended conformation containing H-bonds to both the N5 and O4 positions. These divergent observations can provide some insight into prior in vitro functional assays of G46S containing variants. Even though G46S variants demonstrate light-state like activity, exposure to light can still enhance light-driven complex formation with GIGANTEA.[23] These aspects can be explained by the Gln conformational analysis, specifically, the ability to form both Buried-II and Extended conformations. Both conformations result in Gln154 being largely co-planar with FMN, and prevent sampling of dark-state like exposed conformation leading to an enhanced light-state response. Overall, these results complicate the Gln-flip mechanism. Although ZTL and LOV proteins retain a light-state signaling buried-II conformation, ZTL light-state activation largely results in ordering of the Gln154 side chain as opposed to a simple Gln-flip mechanism observed in other LOV proteins. Further, side chain identities at the G46 and V48 positions in Aβ strand can significantly perturb the Gln154 landscape to disrupt allosteric signal transduction. Although providing insight into the proximal light-driven signaling events, these observations lead to two conundrums regarding reconciling aspects of downstream signal transduction in ZTL: i) If Gln154 H-bonds are still driving Gln154 ordering and signal transduction, how do ZTL proteins signal in a Q154L background; ii) If G46S proteins are still sampling an exposed Gln conformation, why do they prefer the light-state parallel LOV dimer. To answer these two questions, we performed detailed analyses of Gln conformational effects on N/C-termini conformational changes dictating dimer reorganization and light/dark-functionality.

Gln154 effect on ZTL conformational flexibility The investigation of the structural changes in the N- and C-termini granted only a partial picture of the possible downstream effects of ZTL light-dependent Gln154 conformational gating. In fact, shifts in the population distributions on the conformational landscape without appreciable structural transformations can contribute to the allosteric protein functional changes.[36–49] To show how alteration in Gln154 conformational sampling in WT ZTL and allosteric variants impacts the ZTL structural landscape, we performed time-structure Independent Component analysis (tICA) by featurization of the protein conformation using all the Cα pairwise distances. The resulting 2-dimensional plots represent the impact of light- and dark-states and allosteric variants on the ZTL conformational space explored. The two components corresponding to the slowest-relaxing degrees of freedom, which associate with functionally relevant motions, distinguish the native dark- and light-state structures (Fig 7). Specifically, the native dark-state (Red, dark_wt_anti), samples considerably more conformational space than the native light-state (Maroon, light_wt_para), consistent with ordering following adduct formation (Fig 7A). To verify the effect of adduct formation and Gln154 conformation on ZTL dynamics, we further analyzed the conformational space explored in the transient structures, which mimics the initial landscape sampling immediately following adduct formation (transient-light) and scission (transient-dark). Consistent with disorder-order transitions, the transient dark state (tr_dark_wt_anti) stretches from the light state region to the dark native region. Further, the transient light state (tr_light_wt_anti in Fig 7C) displays dynamical behavior similar to the light state. This is indicated by its overlapping with the native light state structures in the tICA plot and by significantly less conformational space explored. These results confirm that light drives a reversible disorder-order transition that is directly coupled to adduct formation and Gln154 dynamics (Fig 7). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 7. Dimensionality reduction analysis using tICA method. a) All structures. b) Dark states including the dark native anti-parallel and the transient dark states. c) Light states including the light native parallel and the transient light states. https://doi.org/10.1371/journal.pcbi.1009168.g007 Examination of how the conformational landscape is perturbed by G46 and V48 allosteric variants provides insight into how subtle local alteration in side-chain identity can impact dynamics-driven allostery. Specifically, the conformational space sampled by the V48I:G80R mutant in the dark state is more restricted than the WT-dark state and largely overlaps with the anti-parallel transient states (dark_v48i_g80r in Fig 7). These results are consistent with restricting the Gln154 conformation to a primarily exposed conformation. Similarly, the light-state conformations are more restricted, leading to only minimal overlap between V48I:G80R and the native light parallel simulations (Fig 7C). Thus, introduction of V48I restricts dynamic motions gating interconversion between light- and dark-configurations. In contrast, G46S variants lock the dynamics away from the dark state structures (dark_g46s_g80r and light_g46s_g80r in Fig 7), supporting light-state-like functionality in the absence of the flavin C4a adduct. In summary, tICA was used to analyze the effect of mutations and light state on the overall protein dynamics. In particular, the dark state displays higher flexibility, while the light state seems to have limited protein motions. Although the tICA plots show overlap between parallel and anti-parallel structures, it is important to point out that in our simulations the inter-conversion between the parallel and anti-parallel dimers was not fully observed. Rather, this analysis captured dynamic similarities between different systems, which were strongly influenced by monomeric atomic motions (S7 Fig). We observed a correlation between the protein flexibility deduced by the tICA projections and the Gln154 conformational dynamics. This indicates the relationship between Gln154 and the protein dynamics, whereby an ability to interconvert between Gln154 conformations is essential to drive light-dependent allosteric conversions in ZTL structure. To better understand how these dynamic motions kinetically gate conformational changes we exploited the reduced dimensional space generated using the tICA to generate a Markov state model.

Identification of functional stable states in ZTL conformational landscape Markov State Model (MSM) can help to discretize the protein conformational landscape into functional metastable states and obtain a kinetically relevant picture of ZTL allosteric process. To achieve this goal, we followed the following protocol: i) k-means micro-clustering into microstates, ii) building of a MSM, and iii) Perron-cluster cluster analysis (PCCA) for transition-based macro-clustering. The details for each step can be found in the Materials and Methods section. Different metastable states lie in different free energy basins illustrating the effectiveness of PCCA in separating kinetically separated states (S8B Fig). As shown in Fig 8, the probability of a structure to remain in its original state is higher than the probability for it to transition to other states. This indicates that each macrostate is in a minimum on the free energy surface, and that the kinetic barriers prevent the system from switching among macrostates frequently. The occupation of these macrostates from different ZTL simulations is plotted in S9A Fig. This analysis allowed the division of the projected conformational space into four areas: Dark Native, Light Native, Transient and Transient Parallel (S9B Fig), based on the structures that predominantly occupy each macrostate. The effect of the light-state adduct on the protein transitions is investigated by plotting the transition for light and dark structures separately (Fig 8B and 8C). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 8. Transition probabilities among different macrostates in the Markov state model. a) Transition probabilities among macrostates using all trajectories; b) Transition probabilities based on Dark state trajectories; c) Transition probabilities based on Light state trajectories; d) Representative structures for different macrostates obtained from each cluster center. https://doi.org/10.1371/journal.pcbi.1009168.g008 The plot of the transition probabilities among different metastable states (Fig 8) shows that the states contained in the transient area (States 3, 5, 7, 8 and 9) have higher probability to transit to the states occupied by the native conformations. This confirms the role of the transient states as intermediates between the stable native dark and native light states. When the FMN cofactor is modeled in the light state, the conformations of the protein shift from dark state-like State 2 to the intermediates between the State 2 and the light-state like State 5. This shift of conformations mimics the biological process of the protein changing its functional dynamics and structures upon change of light conditions.[13,15] As shown in Fig 8B, the native dark state structures cover a large area, exploring most metastable states except for States 4 and 5. This shows that the absence of the photo-induced covalent bond between FMN-Cys82 correlates with the high conformational flexibility. On the contrary, the distribution of the light states is limited in a narrow area (Fig 8C), suggesting that the photo-induced covalent bond between FMN-Cys82 correlates with a low conformational flexibility. Fig 8D illustrates representative structures for the 11 metastable states identified in the MSM. Interestingly, there is a partial rotation of the two monomers in State 11, suggesting a rotation of the monomers towards the anti-parallel conformation. This state is populated solely by parallel transition dark structures, whose RMSD was found to significantly fluctuate during the MD simulations (Protein Stability Analysis in S1 Text). The combination of these two observations could provide an ulterior dynamical confirmation of the light influence to ZTL dimer orientation. However, from the visual inspection, we were not able to find other clear structural differences among different metastable states. For this reason, we employed a machine learning based classification model to identify other structural differences that may play a significant role in ZTL allostery.

[END]

[1] Url: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1009168

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.