Monitoring structural changes in nucleic acids with single residue spatial and millisecond time resolution by quantitative hydroxyl radical footprinting

Systematic Fe(II)-EDTA Method of Dose-Dependent Hydroxyl Radical Generation for Protein Oxidative Footprinting

Correlating the structure and dynamics of proteins with biological function is critical to understanding normal and dysfunctional cellular mechanisms. We describe a quantitative method of hydroxyl radical generation via Fe(II)-ethylenediaminetetraacetic acid (EDTA)-catalyzed Fenton chemistry that provides ready access to protein oxidative footprinting using equipment commonly found in research and process control laboratories. Robust and reproducible dose-dependent oxidation of protein samples is observed and quantitated by mass spectrometry with as fine a single residue resolution. An oxidation analysis of lysozyme provides a readily accessible benchmark for our method. The efficacy of our oxidation method is demonstrated by mapping the interface of a RAS-monobody complex, the surface of the NIST mAb, and the interface between PRC2 complex components. These studies are executed using standard laboratory tools and a few pennies of reagents; the mass spectrometry analysis can be streamlined to map the protein structure with single amino acid residue resolution.

Fungal mycotoxin penisuloxazin A, a novel C-terminal Hsp90 inhibitor and characteristics of its analogues on Hsp90 function related to binding sites

Hsp90 is a promising drug target for cancer therapy. However, toxicity and moderate effect are limitations of current inhibitors owing to broad protein degradation. The fungal mycotoxin penisuloxazin A (PNSA) belongs to a new epipolythiodiketopiperazines (ETPs) possessing a rare 3H-spiro[benzofuran-2,2'-piperazine] ring system. PNSA bound to cysteine residues C572/C598 of CT-Hsp90 with disulfide bonds and inhibits Hsp90 activity, resulting in apoptosis and growth inhibition of HCT116 cells in vitro and in vivo. We identified that analogues PEN-A and HDN-1 bound to C572/C597 and C572 of CT-Hsp90α respectively, with binding pattern very similar to PNSA. These ETPs exhibited different effects on ATPase activity, dimerization formation and selectivity on client protein of Hsp90, indicating client recognition of Hsp90 can be exactly regulated by different sites of Hsp90. Our findings not only offer new chemotypes for anticancer drug development, but also help to better understand biological function of Hsp90 for exploring inhibitor with some client protein bias.

Chemical Generation of Hydroxyl Radical for Oxidative 'Footprinting'

BACKGROUND

For almost four decades, hydroxyl radical chemically generated by Fenton chemistry has been a mainstay for the oxidative 'footprinting' of macromolecules.

OBJECTIVE

In this article, we start by reviewing the application of chemical generation of hydroxyl radical to the development of oxidative footprinting of DNA and RNA and the subsequent application of the method to oxidative footprinting of proteins. We next discuss a novel strategy for generating hydroxyl radicals by Fenton chemistry that immobilizes catalytic iron on a solid surface (Pyrite Shrink Wrap laminate) for the application of nucleic acid and protein footprinting.

METHOD

Pyrite Shrink-Wrap Laminate is fabricated by depositing pyrite (Fe-S2, aka 'fool's gold') nanocrystals onto thermolabile plastic (Shrinky Dink). The laminate can be thermoformed into a microtiter plate format into which samples are deposited for oxidation.

RESULTS

We demonstrate the utility of the Pyrite Shrink-Wrap Laminate for the chemical generation of hydroxyl radicals by mapping the surface of the T-cell co-stimulatory protein Programmed Death - 1 (PD-1) and the interface of the complex with its ligand PD-L1.

CONCLUSION

We have developed and validated an affordable and reliable benchtop method of hydroxyl radical generation that will broaden the application of protein oxidative footprinting. Due to the minimal equipment required to implement this method, it should be easily adaptable by many laboratories with access to mass spectrometry.

Exploiting post-transcriptional regulation to probe RNA structures in vivo via fluorescence

While RNA structures have been extensively characterized in vitro, very few techniques exist to probe RNA structures inside cells. Here, we have exploited mechanisms of post-transcriptional regulation to synthesize fluorescence-based probes that assay RNA structures in vivo. Our probing system involves the co-expression of two constructs: (i) a target RNA and (ii) a reporter containing a probe complementary to a region in the target RNA attached to an RBS-sequestering hairpin and fused to a sequence encoding the green fluorescent protein (GFP). When a region of the target RNA is accessible, the area can interact with its complementary probe, resulting in fluorescence. By using this system, we observed varied patterns of structural accessibility along the length of the Tetrahymena group I intron. We performed in vivo DMS footprinting which, along with previous footprinting studies, helped to explain our probing results. Additionally, this novel approach represents a valuable tool to differentiate between RNA variants and to detect structural changes caused by subtle mutations. Our results capture some differences from traditional footprinting assays that could suggest that probing in vivo via oligonucleotide hybridization facilitates the detection of folding intermediates. Importantly, our data indicate that intracellular oligonucleotide probing can be a powerful complement to existing RNA structural probing methods.

Sizing up long non-coding RNAs: do lncRNAs have secondary and tertiary structure?

Long noncoding RNAs (lncRNAs) play a key role in many important areas of epigenetics, stem cell biology, cancer, signaling and brain function. This emerging class of RNAs constitutes a large fraction of the transcriptome, with thousands of new lncRNAs reported each year. The molecular mechanisms of these RNAs are not well understood. Currently, very little structural data exist. We review the available lncRNA sequence and secondary structure data. Since almost no tertiary information is available for lncRNAs, we review crystallographic structures for other RNA systems and discuss the possibilities for lncRNAs in the context of existing constraints.

Identification of activated cryptic 5' splice sites using structure profiles and odds measure

The activation of cryptic 5′ splice sites (5′ SSs) is often related to human hereditary diseases. The DNA-based mutation screening strategies are commonly used to recognize the cryptic 5′ SSs, because features of the local DNA sequence can influence the choice of cryptic 5′ SSs. To improve the identification of the cryptic 5′ SSs, we developed a structure-based method, named SPO (structure profiles and odds measure), which combines two parameters, the structural feature derived from hydroxyl radical cleavage pattern and odds measure, to assess the likelihood of a cryptic 5′ SS activation in competing with its paired authentic 5′ SS. Compared to the current tools for identifying activated cryptic 5′ SSs, the SPO algorithm achieves higher prediction accuracy than the other methods, including MaxEnt, MDD, Markov model, weight matrix model, Shapiro and Senapathy matrix, R_i and ΔG. In addition, the predicted ΔSPO scores from the SPO algorithm exhibited a greater degree of correlation with the strength of cryptic 5′ SS activation than that measured from the other seven methods. In conclusion, the SPO algorithm provides an optimal identification of cryptic 5′ SSs, can be applied in designing mutagenesis experiments for various splicing events and may be helpful to investigate the relationship between structural variants and human hereditary diseases.

The formation of a potential spring in the ribosome

Time-dependent chemical modification and cleavage results have provided intriguing insights into structural changes that occur in the distal loop of helix 11 in 16S ribosomal RNA (rRNA). Located distant from the decoding region, between proteins S17 and S20, the results of this study suggest that this region of rRNA may act as a buffer or a spring between these two proteins during protein biosynthesis. During the assembly process, protein S17 apparently produces the major structural deformations in this region, causing it to be folded in a spring-like structure. Base C264 in this region shows erratic behavior during assembly and also shows time-dependent enhancement when elongation factor G with GTP is added to 70S ribosomes. Evidence is presented to suggest that this region of rRNA may be used to allow relative motion to occur between proteins S17 and S20 during translocation.

RNA molecules with conserved catalytic cores but variable peripheries fold along unique energetically optimized pathways

Functional and kinetic constraints must be efficiently balanced during the folding process of all biopolymers. To understand how homologous RNA molecules with different global architectures fold into a common core structure we determined, under identical conditions, the folding mechanisms of three phylogenetically divergent group I intron ribozymes. These ribozymes share a conserved functional core defined by topologically equivalent tertiary motifs but differ in their primary sequence, size, and structural complexity. Time-resolved hydroxyl radical probing of the backbone solvent accessible surface and catalytic activity measurements integrated with structural-kinetic modeling reveal that each ribozyme adopts a unique strategy to attain the conserved functional fold. The folding rates are not dictated by the size or the overall structural complexity, but rather by the strength of the constituent tertiary motifs which, in turn, govern the structure, stability, and lifetime of the folding intermediates. A fundamental general principle of RNA folding emerges from this study: The dominant folding flux always proceeds through an optimally structured kinetic intermediate that has sufficient stability to act as a nucleating scaffold while retaining enough conformational freedom to avoid kinetic trapping. Our results also suggest a potential role of naturally selected peripheral A-minor interactions in balancing RNA structural stability with folding efficiency.

Sharing and archiving nucleic acid structure mapping data

Nucleic acids are particularly amenable to structural characterization using chemical and enzymatic probes. Each individual structure mapping experiment reveals specific information about the structure and/or dynamics of the nucleic acid. Currently, there is no simple approach for making these data publically available in a standardized format. We therefore developed a standard for reporting the results of single nucleotide resolution nucleic acid structure mapping experiments, or SNRNASMs. We propose a schema for sharing nucleic acid chemical probing data that uses generic public servers for storing, retrieving, and searching the data. We have also developed a consistent nomenclature (ontology) within the Ontology of Biomedical Investigations (OBI), which provides unique identifiers (termed persistent URLs, or PURLs) for classifying the data. Links to standardized data sets shared using our proposed format along with a tutorial and links to templates can be found at http://snrnasm.bio.unc.edu.

Time resolved SAXS and RNA folding

Small angle X-ray scattering provides low resolution structural information about macromolecules in solution. When coupled with rapid mixing methods, SAXS reports time-dependent conformational changes of RNA induced by the addition of Mg(2+) to trigger folding. Thus time-resolved SAXS provides unique information about the global or overall structures of transient intermediates populated during folding. Notably, SAXS provides information about the earliest folding events, which can evade detection by other methods.

Nonhierarchical ribonucleoprotein assembly suggests a strain-propagation model for protein-facilitated RNA folding

Proteins play diverse and critical roles in cellular ribonucleoproteins (RNPs) including promoting formation of and stabilizing active RNA conformations. Yet, the conformational changes required to convert large RNAs into active RNPs have proven difficult to characterize fully. Here we use high-resolution approaches to monitor both local nucleotide flexibility and solvent accessibility for nearly all nucleotides in the bI3 group I intron RNP in four assembly states: the free RNA, maturase-bound RNA, Mrs1-bound RNA, and the complete six-component holocomplex. The free RNA is misfolded relative to the secondary structure required for splicing. The maturase and Mrs1 proteins each stabilized long-range tertiary interactions, but neither protein alone induced folding into the functional secondary structure. In contrast, simultaneous binding by both proteins results in large secondary structure rearrangements in the RNA and yielded the catalytically active group I intron structure. Secondary and tertiary folding of the RNA component of the bI3 RNP are thus not independent: RNA folding is strongly nonhierarchical. These results emphasize that protein-mediated stabilization of RNA tertiary interactions functions to pull the secondary structure into an energetically disfavored, but functional, conformation and emphasize a new role for facilitator proteins in RNP assembly.

Evaluation of the information content of RNA structure mapping data for secondary structure prediction

Structure mapping experiments (using probes such as dimethyl sulfate [DMS], kethoxal, and T1 and V1 RNases) are used to determine the secondary structures of RNA molecules. The process is iterative, combining the results of several probes with constrained minimum free-energy calculations to produce a model of the structure. We aim to evaluate whether particular probes provide more structural information, and specifically, how noise in the data affects the predictions. Our approach involves generating "decoy" RNA structures (using the sFold Boltzmann sampling procedure) and evaluating whether we are able to identify the correct structure from this ensemble of structures. We show that with perfect information, we are always able to identify the optimal structure for five RNAs of known structure. We then collected orthogonal structure mapping data (DMS and RNase T1 digest) under several solution conditions using our high-throughput capillary automated footprinting analysis (CAFA) technique on two group I introns of known structure. Analysis of these data reveals the error rates in the data under optimal (low salt) and suboptimal solution conditions (high MgCl(2)). We show that despite these errors, our computational approach is less sensitive to experimental noise than traditional constraint-based structure prediction algorithms. Finally, we propose a novel approach for visualizing the interaction of chemical and enzymatic mapping data with RNA structure. We project the data onto the first two dimensions of a multidimensional scaling of the sFold-generated decoy structures. We are able to directly visualize the structural information content of structure mapping data and reconcile multiple data sets.

Advances in RNA structure analysis by chemical probing

RNA is arguably the most versatile biological macromolecule because of its ability both to encode and to manipulate genetic information. The diverse roles of RNA depend on its ability to fold back on itself to form biologically functional structures that bind small molecule and large protein ligands, to change conformation, and to affect the cellular regulatory state. These features of RNA biology can be structurally interrogated using chemical mapping experiments. The usefulness and applications of RNA chemical probing technologies have expanded dramatically over the past five years because of several critical advances. These innovations include new sequence-independent RNA chemistries, algorithmic tools for high-throughput analysis of complex data sets composed of thousands of measurements, new approaches for interpreting chemical probing data for both secondary and tertiary structure prediction, facile methods for following time-dependent processes, and the willingness of individual research groups to tackle increasingly bold problems in RNA structural biology.

The Mrs1 splicing factor binds the bI3 group I intron at each of two tetraloop-receptor motifs

Most large ribozymes require protein cofactors in order to function efficiently. The yeast mitochondrial bI3 group I intron requires two proteins for efficient splicing, Mrs1 and the bI3 maturase. Mrs1 has evolved from DNA junction resolvases to function as an RNA cofactor for at least two group I introns; however, the RNA binding site and the mechanism by which Mrs1 facilitates splicing were unknown. Here we use high-throughput RNA structure analysis to show that Mrs1 binds a ubiquitous RNA tertiary structure motif, the GNRA tetraloop-receptor interaction, at two sites in the bI3 RNA. Mrs1 also interacts at similar tetraloop-receptor elements, as well as other structures, in the self-folding Azoarcus group I intron and in the RNase P enzyme. Thus, Mrs1 recognizes general features found in the tetraloop-receptor motif. Identification of the two Mrs1 binding sites now makes it possible to create a model of the complete six-component bI3 ribonucleoprotein. All protein cofactors bind at the periphery of the RNA such that every long-range RNA tertiary interaction is stabilized by protein binding, involving either Mrs1 or the bI3 maturase. This work emphasizes the strong evolutionary pressure to bolster RNA tertiary structure with RNA-binding interactions as seen in the ribosome, spliceosome, and other large RNA machines.

Rapid quantification and analysis of kinetic •OH radical footprinting data using SAFA

The use of highly reactive chemical species to probe the structure and dynamics of nucleic acids is greatly simplified by software that enables rapid quantification of the gel images that result from these experiments. Semiautomated footprinting analysis (SAFA) allows a user to quickly and reproducibly quantify a chemical footprinting gel image through a series of steps that rectify, assign, and integrate the relative band intensities. The output of this procedure is raw band intensities that report on the relative reactivity of each nucleotide with the chemical probe. We describe here how to obtain these raw band intensities using SAFA and the subsequent normalization and analysis procedures required to process these data. In particular, we focus on analyzing time-resolved hydroxyl radical ((•)OH) data, which we use to monitor the kinetics of folding of a large RNA (the L-21 T. thermophila group I intron). Exposing the RNA to bursts of (•)OH radicals at specific time points during the folding process monitors the time progress of the reaction. Specifically, we identify protected (nucleotides that become inaccessible to the (•)OH radical probe when folded) and invariant (nucleotides with constant accessibility to the (•)OH probe) residues that we use for monitoring and normalization of the data. With this analysis, we obtain time-progress curves from which we determine kinetic rates of folding. We also report on a data visualization tool implemented in SAFA that allows users to map data onto a secondary structure diagram.

Hydroxyl-radical footprinting to probe equilibrium changes in RNA tertiary structure

Hydroxyl-radical footprinting utilizes the ability of a highly reactive species to nonspecifically cleave the solvent accessible regions of a nucleic acid backbone. Thus, changes in a nucleic acids structure can be probed either as a function of time or of a reagent's concentration. When combined with techniques that allow single nucleotide resolution of the resulting fragments, footprinting experiments provide richly detailed information about local changes in tertiary structure of a nucleic acid accompanying its folding or ligand binding. In this chapter, we present two protocols of equilibrium hydroxyl-radical footprinting based on peroxidative and oxidative Fenton chemistry and discuss how to adjust the Fenton reagent concentrations for a specific experimental condition. We also discuss the choice of the techniques to separate the reaction products and specifics of the data analysis for equilibrium footprinting experiments. Protocols addressing the use of peroxidative Fenton chemistry for time-resolved studies have been published [Schlatterer and Brenowitz, 2009. Methods; Shcherbakova and Brenowitz, 2008. Nat. Protoc.3(2), 288-302; Shcherbakova et al., 2006. Nucleic Acids Res.34(6), e48; Shcherbakova et al., 2007. Methods Cell Biol.84, 589-615].

Complementing global measures of RNA folding with local reports of backbone solvent accessibility by time resolved hydroxyl radical footprinting

A variety of analytical techniques are used to probe the mechanisms by which RNA molecules fold to discrete three dimensional structures. Methods such as small angle X-ray scattering (SAXS) report global properties like overall size and shape of the RNA. Other methods such as chemical or enzymatic mapping (footprinting) report properties with resolution as fine as single nucleotide. The hydroxyl radical (*OH) is a footprinting probe which cleaves the oligonucleotide backbone independently of sequence and thus is a valuable reporter of backbone solvent accessibility. Combinations of global and local measures of folding reactions are uniquely able to distinguish specific from nonspecific processes. This article highlights the application of *OH footprinting as a complement to SAXS for kinetics analysis of RNA folding. We illustrate this combination of techniques using a study of the role played by the stiffness of a hinge in determining the rate limiting step of a Mg(2+)-mediated RNA folding reaction.

Time-resolved RNA SHAPE chemistry: quantitative RNA structure analysis in one-second snapshots and at single-nucleotide resolution

RNA selective 2'-hydroxyl acylation analyzed by primer extension (SHAPE) chemistry exploits the discovery that conformationally dynamic nucleotides preferentially adopt configurations that facilitate reaction between the 2'-OH group and a hydroxyl-selective electrophile, such as benzoyl cyanide (BzCN), to form a 2'-O-adduct. BzCN is ideally suited for quantitative, time-resolved analysis of RNA folding and ribonucleoprotein (RNP) assembly mechanisms because this reagent both reacts with flexible RNA nucleotides and also undergoes auto-inactivating hydrolysis with a half-life of 0.25 s at 37 degrees C. RNA folding is initiated by addition of Mg(2+) or protein, or other change in solution conditions, and nucleotide resolution structural images are obtained by adding aliquots of the evolving reaction to BzCN and then 'waiting' for 1 second. Sites of the 2'-O-adduct formation are subsequently scored as stops to primer extension using reverse transcriptase. This time-resolved SHAPE protocol makes it possible to obtain 1-second structural snapshots in time-resolved kinetic studies for RNAs of arbitrary length and complexity in a straightforward and concise experiment.

A mechanism for S-adenosyl methionine assisted formation of a riboswitch conformation: a small molecule with a strong arm

The S-adenosylmethionine-1 (SAM-I) riboswitch mediates expression of proteins involved in sulfur metabolism via formation of alternative conformations in response to binding by SAM. Models for kinetic trapping of the RNA in the bound conformation require annealing of nonadjacent mRNA segments during a transcriptional pause. The entropic cost required to bring nonadjacent segments together should slow the folding process. To address this paradox, we performed molecular dynamics simulations on the SAM-I riboswitch aptamer domain with and without SAM, starting with the X-ray coordinates of the SAM-bound RNA. Individual trajectories are 200 ns, among the longest reported for an RNA of this size. We applied principle component analysis (PCA) to explore the global dynamics differences between these two trajectories. We observed a conformational switch between a stacked and nonstacked state of a nonadjacent dinucleotide in the presence of SAM. In the absence of SAM the coordination between a bound magnesium ion and the phosphate of A9, one of the nucleotides involved in the dinucleotide stack, is destabilized. An electrostatic potential map reveals a 'hot spot' at the Mg binding site in the presence of SAM. These results suggest that SAM binding helps to position J1/2 in a manner that is favorable for P1 helix formation.

A rate-limiting conformational step in the catalytic pathway of the glmS ribozyme

The glmS ribozyme is a conserved riboswitch in numerous Gram-positive bacteria and is located upstream of the glucosamine-6-phosphate (GlcN6P) synthetase reading frame. Binding of GlcN6P activates site-specific self-cleavage of the glmS mRNA, resulting in the downregulation of glmS gene expression. Unlike other riboswitches, the glmS ribozyme does not undergo structural rearrangement upon metabolite binding, indicating that the metabolite binding pocket is preformed in the absence of ligand. This observation led us to test if individual steps in the reaction pathway could be dissected by initiating the cleavage reaction before or after Mg(2+)-dependent folding. Here we show that self-cleavage reactions initiated with simultaneous addition of Mg(2+) and GlcN6P are slow (3 min(-1)) compared to reactions initiated by addition of GlcN6P to glmS RNA that has been prefolded in Mg(2+)-containing buffer (72 min(-1)). These data indicate that some level of Mg(2+)-dependent folding is rate-limiting for catalysis. Reactions initiated by addition of GlcN6P to the prefolded ribozyme also resulted in a 30-fold increase in the apparent ligand K(d) compared to those of reactions initiated by a global folding step. Time-resolved hydroxyl-radical footprinting was employed to determine if global tertiary structure formation is the rate-limiting step. The results of these experiments provided evidence for fast and largely concerted folding of the global tertiary structure (>13 min(-1)). This indicates that the rate-limiting step that we have identified either is a slow folding step between the fast initial folding and ligand binding events or represents the rate of escape from a nativelike folding trap.

Time-resolved RNA SHAPE chemistry

Selective 2'-hydroxyl acylation analyzed by primer extension (SHAPE) chemistry yields quantitative RNA secondary and tertiary structure information at single nucleotide resolution. SHAPE takes advantage of the discovery that the nucleophilic reactivity of the ribose 2'-hydroxyl group is modulated by local nucleotide flexibility in the RNA backbone. Flexible nucleotides are reactive toward hydroxyl-selective electrophiles, whereas constrained nucleotides are unreactive. Initial versions of SHAPE chemistry, which employ isatoic anhydride derivatives that react on the minute time scale, are emerging as the ideal technology for monitoring equilibrium structures of RNA in a wide variety of biological environments. Here, we extend SHAPE chemistry to a benzoyl cyanide scaffold to make possible facile time-resolved kinetic studies of RNA in approximately 1 s snapshots. We then use SHAPE chemistry to follow the time-dependent folding of an RNase P specificity domain RNA. Tertiary interactions form in two distinct steps with local tertiary contacts forming an order of magnitude faster than long-range interactions. Rate-determining tertiary folding requires minutes despite that no non-native interactions must be disrupted to form the native structure. Instead, overall folding is limited by simultaneous formation of interactions approximately 55 A distant in the RNA. Time-resolved SHAPE holds broad potential for understanding structural biogenesis and the conformational interconversions essential to the functions of complex RNA molecules at single nucleotide resolution.

[Biochemical methods for the analysis of DNA-protein interactions]

Investigation of DNA-protein interactions is fundamental to understand the mechanism underlying a variety of life processes. In this article, various types of biochemical methods in DNA-protein interaction study in vivo and in vitro at the level of DNA, protein, and the complex, respectively were briefly reviewed. Traditional assays including Nitrocellulose filter-binding assay, Footprinting, EMSA, and Southwestern blotting were summarized. In addition, chromatin immunoprecipitation techniques including nChIP, xChIP, and ChIP-on-chip, which were widely used in epigenetics, were particularly introduced.

High-throughput SHAPE and hydroxyl radical analysis of RNA structure and ribonucleoprotein assembly

RNA folds to form complex structures vital to many cellular functions. Proteins facilitate RNA folding at both the secondary and tertiary structure levels. An absolute prerequisite for understanding RNA folding and ribonucleoprotein (RNP) assembly reactions is a complete understanding of the RNA structure at each stage of the folding or assembly process. Here we provide a guide for comprehensive and high-throughput analysis of RNA secondary and tertiary structure using SHAPE and hydroxyl radical footprinting. As an example of the strong and sometimes surprising conclusions that can emerge from high-throughput analysis of RNA folding and RNP assembly, we summarize the structure of the bI3 group I intron RNA in four distinct states. Dramatic structural rearrangements occur in both secondary and tertiary structure as the RNA folds from the free state to the active, six-component, RNP complex. As high-throughput and high-resolution approaches are applied broadly to large protein-RNA complexes, other proteins previously viewed as making simple contributions to RNA folding are also likely to be found to exert multifaceted, long-range, cooperative, and nonadditive effects on RNA folding. These protein-induced contributions add another level of control, and potential regulatory function, in RNP complexes.