Why does binding of proteins to DNA or proteins to proteins not necessarily spell function?

Asymmetrical diversification of the receptor-ligand interaction controlling self-incompatibility in Arabidopsis

How two-component genetic systems accumulate evolutionary novelty and diversify in the course of evolution is a fundamental problem in evolutionary systems biology. In the Brassicaceae, self-incompatibility (SI) is a spectacular example of a diversified allelic series in which numerous highly diverged receptor-ligand combinations are segregating in natural populations. However, the evolutionary mechanisms by which new SI specificities arise have remained elusive. Using in planta ancestral protein reconstruction, we demonstrate that two allelic variants segregating as distinct receptor-ligand combinations diverged through an asymmetrical process whereby one variant has retained the same recognition specificity as their (now extinct) putative ancestor, while the other has functionally diverged and now represents a novel specificity no longer recognized by the ancestor. Examination of the structural determinants of the shift in binding specificity suggests that qualitative rather than quantitative changes of the interaction are an important source of evolutionary novelty in this highly diversified receptor-ligand system.

Predicting conformational ensembles and genome-wide transcription factor binding sites from DNA sequences

DNA shape is emerging as an important determinant of transcription factor binding beyond just the DNA sequence. The only tool for large scale DNA shape estimates, DNAshape was derived from Monte-Carlo simulations and predicts four broad and static DNA shape features, Propeller twist, Helical twist, Minor groove width and Roll. The contributions of other shape features e.g. Shift, Slide and Opening cannot be evaluated using DNAshape. Here, we report a novel method DynaSeq, which predicts molecular dynamics-derived ensembles of a more exhaustive set of DNA shape features. We compared the DNAshape and DynaSeq predictions for the common features and applied both to predict the genome-wide binding sites of 1312 TFs available from protein interaction quantification (PIQ) data. The results indicate a good agreement between the two methods for the common shape features and point to advantages in using DynaSeq. Predictive models employing ensembles from individual conformational parameters revealed that base-pair opening - known to be important in strand separation - was the best predictor of transcription factor-binding sites (TFBS) followed by features employed by DNAshape. Of note, TFBS could be predicted not only from the features at the target motif sites, but also from those as far as 200 nucleotides away from the motif.

Allosteric therapies for lung cancer

Allostery is a regulation at a distance by conveying information from one site to another and an intrinsic property of dynamic proteins. Allostery plays an essential role in receptor trafficking, signal transmission, controlled catalysis, gene turn on/off, or cell apoptosis. Allosteric mutations are considered as one of causes responsible for cancer development, leading to "allosteric diseases" by stabilizing an active or inactive conformation or changing the dynamic distribution of preexisting propagation pathways. The present article mainly focuses on the potential of allosteric therapies for lung cancer. Allosteric drugs may have several advantages over traditional drugs. The epidermal growth factor receptor mutations and signaling pathways downstream (such as PI3K/AKT/mTOR and RAS/RAF/MEK/ERK pathways) were suggested to play a key role in lung cancer and considered as targets of allosteric therapy. Some allosteric inhibitors for lung cancer-specific targets and a series of preclinical trials of allosteric inhibitors for lung cancer have been developed and reported. We expect that allosteric therapies will gain more attentions to develop combinatorial strategies for lung cancer and metastasis.

Protein Ensembles: How Does Nature Harness Thermodynamic Fluctuations for Life? The Diverse Functional Roles of Conformational Ensembles in the Cell

All soluble proteins populate conformational ensembles that together constitute the native state. Their fluctuations in water are intrinsic thermodynamic phenomena, and the distributions of the states on the energy landscape are determined by statistical thermodynamics; however, they are optimized to perform their biological functions. In this review we briefly describe advances in free energy landscape studies of protein conformational ensembles. Experimental (nuclear magnetic resonance, small-angle X-ray scattering, single-molecule spectroscopy, and cryo-electron microscopy) and computational (replica-exchange molecular dynamics, metadynamics, and Markov state models) approaches have made great progress in recent years. These address the challenging characterization of the highly flexible and heterogeneous protein ensembles. We focus on structural aspects of protein conformational distributions, from collective motions of single- and multi-domain proteins, intrinsically disordered proteins, to multiprotein complexes. Importantly, we highlight recent studies that illustrate functional adjustment of protein conformational ensembles in the crowded cellular environment. We center on the role of the ensemble in recognition of small- and macro-molecules (protein and RNA/DNA) and emphasize emerging concepts of protein dynamics in enzyme catalysis. Overall, protein ensembles link fundamental physicochemical principles and protein behavior and the cellular network and its regulation.

Dynamic multiprotein assemblies shape the spatial structure of cell signaling

Cell signaling underlies critical cellular decisions. Coordination, efficiency as well as fail-safe mechanisms are key elements. How the cell ensures that these hallmarks are at play are important questions. Cell signaling is often viewed as taking place through discrete and cross-talking pathways; oftentimes these are modularized to emphasize distinct functions. While simple, convenient and clear, such models largely neglect the spatial structure of cell signaling; they also convey inter-modular (or inter-protein) spatial separation that may not exist. Here our thesis is that cell signaling is shaped by a network of multiprotein assemblies. While pre-organized, the assemblies and network are loose and dynamic. They contain transiently-associated multiprotein complexes which are often mediated by scaffolding proteins. They are also typically anchored in the membrane, and their continuum may span the cell. IQGAP1 scaffolding protein which binds proteins including Raf, calmodulin, Mek, Erk, actin, and tens more, with actin shaping B-cell (and likely other) membrane-anchored nanoclusters and allosterically polymerizing in dynamic cytoskeleton formation, and Raf anchoring in the membrane along with Ras, provides a striking example. The multivalent network of dynamic proteins and lipids, with specific interactions forming and breaking, can be viewed as endowing gel-like properties. Collectively, this reasons that efficient, productive and reliable cell signaling takes place primarily through transient, preorganized and cooperative protein-protein interactions spanning the cell rather than stochastic, diffusion-controlled processes.

A combination of computational and experimental approaches identifies DNA sequence constraints associated with target site binding specificity of the transcription factor CSL

Regulation of transcription is fundamental to development and physiology, and occurs through binding of transcription factors to specific DNA sequences in the genome. CSL (CBF1/Suppressor of Hairless/LAG-1), a core component of the Notch signaling pathway, is one such transcription factor that acts in concert with co-activators or co-repressors to control the activity of associated target genes. One fundamental question is how CSL can recognize and select among different DNA sequences available in vivo and whether variations between selected sequences can influence its function. We have therefore investigated CSL-DNA recognition using computational approaches to analyze the energetics of CSL bound to different DNAs and tested the in silico predictions with in vitro and in vivo assays. Our results reveal novel aspects of CSL binding that may help explain the range of binding observed in vivo. In addition, using molecular dynamics simulations, we show that domain-domain correlations within CSL differ significantly depending on the DNA sequence bound, suggesting that different DNA sequences may directly influence CSL function. Taken together, our results, based on computational chemistry approaches, provide valuable insights into transcription factor-DNA binding, in this particular case increasing our understanding of CSL-DNA interactions and how these may impact on its transcriptional control.

Multiple conformational selection and induced fit events take place in allosteric propagation

The fact that we observe a single conformational selection event during binding does not necessarily mean that only a single conformational selection event takes place, even though this is the common assumption. Here we suggest that conformational selection takes place not once in a given binding/allosteric event, but at every step along the allosteric pathway. This view generalizes conformational selection and makes it applicable also to other allosteric events, such as post-translational modifications (PTMs) and photon absorption. Similar to binding, at each step along a propagation pathway, conformational selection is coupled with induced fit which optimizes the interactions. Thus, as in binding, the allosteric effects induced by PTMs and light relate not only to population shift; but to conformational selection as well. Conformational selection and population shift take place conjointly.

Dynamic conformational switching in the chemokine ligand is essential for G-protein-coupled receptor activation

Chemokines mediate diverse functions from organogenesis to mobilizing leucocytes, and are unusual agonists for class-A GPCRs (G-protein-coupled receptors) because of their large size and multi-domain structure. The current model for receptor activation, which involves interactions between chemokine N-loop and receptor N-terminal residues (Site-I) and between chemokine N-terminal and receptor extracellular loop/transmembrane residues (Site-II), fails to describe differences in ligand/receptor selectivity and the activation of multiple signalling pathways. In the present study, we show in neutrophil-activating chemokine CXCL8 that the highly conserved GP (glycine-proline) motif located distal to both N-terminal and N-loop residues couples Site-I and Site-II interactions. GP mutants showed large differences from native-like to complete loss of function that could not be correlated with the specific mutation, receptor affinity or subtype, or a specific signalling pathway. NMR studies indicated that the GP motif does not influence Site-I interactions, but molecular dynamics simulations suggested that this motif dictates substates of the CXCL8 conformational ensemble. We conclude that the GP motif enables diverse receptor functions by controlling cross-talk between Site-I and Site-II, and further propose that the repertoire of chemokine functions is best described by a conformational ensemble model in which a network of long-range coupled indirect interactions mediate receptor activity.

Dipeptide analysis of p53 mutations and evolution of p53 family proteins

p53 gain-of-function mutations are similar to driver mutations in cancer genes, with both promoting tumorigenesis. Most previous studies focused on residues lost by mutations, providing information related to a dominantly-negative effect. However, to understand gain-of-function mutations, it is also important to investigate what are the distributions of residues gained by mutations. We compile available p53/p63/p73 protein sequences and construct a non-redundant dataset. We analyze the amino acid and dipeptide composition of p53/p63/p73 proteins across evolution and compare them with the gain/loss of amino acids and dipeptides in human p53 following cancer-related somatic mutations. We find that the ratios of amino acids gained via somatic mutations during evolution to those lost through p53 cancer mutations correlate with the ratios found in single nucleotide polymorphisms in the human proteome. The dipeptide mutational gain/loss ratios are inversely correlated with those observed over p53 evolution but tend to follow the increasing p63/p73-like dipeptide propensities. We successfully simulated the p53 cancer mutation spectrum using the dipeptide composition across the p53 family accounting for the likelihood of mutations in p53 codons. The results revealed that the p53 mutation spectrum is dominated not only by p53 evolution but also by reversal of evolution to a certain degree. This article is part of a Special Issue entitled: Computational Proteomics, Systems Biology & Clinical Implications. Guest Editor: Yudong Cai.

The spatial structure of cell signaling systems

The spatial structure of the cell is highly organized at all levels: from small complexes and assemblies, to local nano- and microclusters, to global, micrometer scales across and between cells. We suggest that this multiscale spatial cell organization also organizes signaling and coordinates cellular behavior. We propose a new view of the spatial structure of cell signaling systems. This new view describes cell signaling in terms of dynamic allosteric interactions within and among distinct, spatially organized transient clusters. The clusters vary over time and space and are on length scales from nanometers to micrometers. When considered across these length scales, primary factors in the spatial organization are cell membrane domains and the actin cytoskeleton, both also highly dynamic. A key challenge is to understand the interplay across these multiple scales, link it to the physicochemical basis of the conformational behavior of single molecules and ultimately relate it to cellular function. Overall, our premise is that at these scales, cell signaling should be thought of not primarily as a sequence of diffusion-controlled molecular collisions, but instead transient, allostery-driven cluster re-forming interactions.

The underappreciated role of allostery in the cellular network

Allosteric propagation results in communication between distinct sites in the protein structure; it also encodes specific effects on cellular pathways, and in this way it shapes cellular response. One example of long-range effects is binding of morphogens to cell surface receptors, which initiates a cascade of protein interactions that leads to genome activation and specific cellular action. Allosteric propagation results from combinations of multiple factors, takes place through dynamic shifts of conformational ensembles, and affects the equilibria of macromolecular interactions. Here, we (a) emphasize the well-known yet still underappreciated role of allostery in conveying explicit signals across large multimolecular assemblies and distances to specify cellular action; (b) stress the need for quantitation of the allosteric effects; and finally, (c) propose that each specific combination of allosteric effectors along the pathway spells a distinct function. The challenges are colossal; the inspiring reward will be predicting function, misfunction, and outcomes of drug regimes.

Structured crowding and its effects on enzyme catalysis

Macromolecular crowding decreases the diffusion rate, shifts the equilibrium of protein-protein and protein-substrate interactions, and changes protein conformational dynamics. Collectively, these effects contribute to enzyme catalysis. Here we describe how crowding may bias the conformational change and dynamics of enzyme populations and in this way affect catalysis. Crowding effects have been studied using artificial crowding agents and in vivo-like environments. These studies revealed a correlation between protein dynamics and function in the crowded environment. We suggest that crowded environments be classified into uniform crowding and structured crowding. Uniform crowding represents random crowding conditions created by synthetic particles with a narrow size distribution. Structured crowding refers to the highly coordinated cellular environment, where proteins and other macromolecules are clustered and organized. In structured crowded environments the perturbation of protein thermal stability may be lower; however, it may still be able to modulate functions effectively and dynamically. Dynamic, allosteric enzymes could be more sensitive to cellular perturbations if their free energy landscape is flatter around the native state; on the other hand, if their free energy landscape is rougher, with high kinetic barriers separating deep minima, they could be more robust. Above all, cells are structured; and this holds both for the cytosol and for the membrane environment. The crowded environment is organized, which limits the search, and the crowders are not necessarily inert. More likely, they too transmit allosteric effects, and as such play important functional roles. Overall, structured cellular crowding may lead to higher enzyme efficiency and specificity.

Selective molecular recognition in amyloid growth and transmission and cross-species barriers

Mutual conformational selection and population shift followed by minor induced-fit optimization is the key mechanism in biomolecular recognition, and monomers and small oligomers binding to amyloid seeds in fibril growth is a molecular recognition event. Here, we describe amyloid aggregation, preferred species, cross-species barriers and transmission within the broad framework of molecular recognition. Cross-seeding of amyloid species is governed by conformational selection of compatible (complementary) states. If the dominant conformations of two species are similar, they can cross-seed each other; on the other hand, if they are sufficiently different, they will grow into different fibrils, reflecting species barriers. Such a scenario has recently been observed for the tau protein, which has four repeats. While a construct consisting of repeats 1, 3 and 4 can serve as a seed for the entire four-repeat tau segment, the inverse does not hold. On the other hand, the tau protein repeats with the characteristic U-turn shape can cross-seed Alzheimer's amyloid β and, similarly, the islet amyloid polypeptide. Within this framework, we suggest that the so-called "central dogma" of amyloid formation, where aggregation takes place through nonspecific backbone hydrogen bonding interactions, which are common to all peptides and proteins, is a simple reflection of the heterogeneous, polymorphic free-energy landscape of amyloid species. Here, we review available data and make some propositions addressing this key problem. In particular, we argue that recent theoretical and experimental observations support the key role of selective molecular recognition in amyloidosis and in determining cross-species barriers and transmission.

Operator sequence alters gene expression independently of transcription factor occupancy in bacteria

A canonical quantitative view of transcriptional regulation holds that the only role of operator sequence is to set the probability of transcription factor binding, with operator occupancy determining the level of gene expression. In this work, we test this idea by characterizing repression in vivo and the binding of RNA polymerase in vitro in experiments where operators of various sequences were placed either upstream or downstream from the promoter in Escherichia coli. Surprisingly, we find that operators with a weaker binding affinity can yield higher repression levels than stronger operators. Repressor bound to upstream operators modulates promoter escape, and the magnitude of this modulation is not correlated with the repressor-operator binding affinity. This suggests that operator sequences may modulate transcription by altering the nature of the interaction of the bound transcription factor with the transcriptional machinery, implying a new layer of sequence dependence that must be confronted in the quantitative understanding of gene expression.

Understanding biomolecular motion, recognition, and allostery by use of conformational ensembles

We review the role conformational ensembles can play in the analysis of biomolecular dynamics, molecular recognition, and allostery. We introduce currently available methods for generating ensembles of biomolecules and illustrate their application with relevant examples from the literature. We show how, for binding, conformational ensembles provide a way of distinguishing the competing models of induced fit and conformational selection. For allostery we review the classic models and show how conformational ensembles can play a role in unravelling the intricate pathways of communication that enable allostery to occur. Finally, we discuss the limitations of conformational ensembles and highlight some potential applications for the future.

Gene-specific transcription activation via long-range allosteric shape-shifting

How is specificity transmitted over long distances at the molecular level? REs (regulatory elements) are often far from transcription start sites. In the present review we discuss possible mechanisms to explain how information from specific REs is conveyed to the basal transcription machinery through TFs (transcription factors) and the Mediator complex. We hypothesize that this occurs through allosteric pathways: binding of a TF to a RE results in changes in the AD (activation domain) of the TF, which binds to Mediator and alters the distribution of the Mediator conformations, thereby affecting transcription initiation/activation. We argue that Mediator is formed by highly disordered proteins with large densely packed interfaces that make efficient long-range signal propagation possible. We suggest two possible general mechanisms for Mediator action: one in which Mediator influences PIC (pre-initiation complex) assembly and transcription initiation, and another in which Mediator exerts its effect on the already assembled but stalled transcription complex. We summarize (i) relevant information from the literature about Mediator composition, organization and structure; (ii) Mediator interaction partners and their effect on Mediator conformation, function and correlation to the RNA Pol II (polymerase II) CTD (C-terminal domain) phosphorylation; and (iii) propose that different allosteric signal propagation pathways in Mediator relate to PIC assembly and polymerase activation of the stalled transcription complex. The emerging picture provides for the first time a mechanistic view of allosteric signalling from the RE sequence to transcription activation, and an insight into how gene specificity and signal transmission can take place in transcription initiation.

Do protein-protein interaction databases identify moonlighting proteins?

One of the most striking results of the human (and mammalian) genomes is the low number of protein-coding genes. To-date, the main molecular mechanism to increase the number of different protein isoforms and functions is alternative splicing. However, a less-known way to increase the number of protein functions is the existence of multifunctional, multitask, or "moonlighting", proteins. By and large, moonlighting proteins are experimentally disclosed by serendipity. Proteomics is becoming one of the very active areas of biomedical research, which permits researchers to identify previously unseen connections among proteins and pathways. In principle, protein-protein interaction (PPI) databases should contain information on moonlighting proteins and could provide suggestions to further analysis in order to prove the multifunctionality. As far as we know, nobody has verified whether PPI databases actually disclose moonlighting proteins. In the present work we check whether well-established moonlighting proteins present in PPI databases connect with their known partners and, therefore, a careful inspection of these databases could help to suggest their different functions. The results of our research suggest that PPI databases could be a valuable tool to suggest multifunctionality.

The role of response elements organization in transcription factor selectivity: the IFN-β enhanceosome example

What is the mechanism through which transcription factors (TFs) assemble specifically along the enhancer DNA? The IFN-β enhanceosome provides a good model system: it is small; its components' crystal structures are available; and there are biochemical and cellular data. In the IFN-β enhanceosome, there are few protein-protein interactions even though consecutive DNA response elements (REs) overlap. Our molecular dynamics (MD) simulations on different motif combinations from the enhanceosome illustrate that cooperativity is achieved via unique organization of the REs: specific binding of one TF can enhance the binding of another TF to a neighboring RE and restrict others, through overlap of REs; the order of the REs can determine which complexes will form; and the alternation of consensus and non-consensus REs can regulate binding specificity by optimizing the interactions among partners. Our observations offer an explanation of how specificity and cooperativity can be attained despite the limited interactions between neighboring TFs on the enhancer DNA. To date, when addressing selective TF binding, attention has largely focused on RE sequences. Yet, the order of the REs on the DNA and the length of the spacers between them can be a key factor in specific combinatorial assembly of the TFs on the enhancer and thus in function. Our results emphasize cooperativity via RE binding sites organization.

Side-chain conformational changes of transcription factor PhoB upon DNA binding: a population-shift mechanism

Using molecular dynamics (MD) simulations and analyses of NMR relaxation order parameters, we investigated conformational changes of side chains in hydrophobic cores upon DNA binding for the DNA binding/transactivation domain of the transcription factor PhoB, in which backbone conformational changes upon DNA binding are small. The simulation results correlated well with experimental order parameters for the backbone and side-chain methyl groups, showing that the order parameters generally represent positional fluctuations of the backbone and side-chain methyl groups. However, topological effects of the side chains on the order parameters were also found and could be eliminated using normalized order parameters for each amino acid type. Consistent with the NMR experiments, the normalized order parameters from the MD simulations showed that the side chains in one of the two hydrophobic cores (the soft core) were highly flexible in comparison with those in the other hydrophobic core (the hard core) before DNA binding and that the flexibility of the hydrophobic cores, particularly of the soft core, was reduced upon DNA binding. Principal component analysis of methyl group configurations revealed strikingly different side-chain dynamics for the soft and hard cores. In the hard core, side-chain configurations were simply distributed around one or two average configurations. In contrast, the side chains in the soft core dynamically varied their configurations in an equilibrium ensemble that included binding configurations as minor components before DNA binding. DNA binding led to a restriction of the side-chain dynamics and a shift in the equilibrium toward binding configurations, in clear correspondence with a population-shift model.

Allostery and population shift in drug discovery

Proteins can exist in a large number of conformations around their native states that can be characterized by an energy landscape. The landscape illustrates individual valleys, which are the conformational substates. From the functional standpoint, there are two key points: first, all functionally relevant substates pre-exist; and second, the landscape is dynamic and the relative populations of the substates will change following allosteric events. Allosteric events perturb the structure, and the energetic strain propagates and shifts the population. This can lead to changes in the shapes and properties of target binding sites. Here we present an overview of dynamic conformational ensembles focusing on allosteric events in signaling. We propose that combining equilibrium fluctuation concepts with genomic screens could help drug discovery.

From large-scale assays to mechanistic insights: computational analysis of interactions

The activity in the living cell is carried out by a myriad network of interactions between macromolecules. These include interactions between proteins that form a functional complex, a protein modifying another protein in a transient interaction, a transcription factor that binds a specific DNA locus triggering a change in chromatin or transcription, and so on. Characterization of these interactions in terms of timing, context, and function is crucial for understanding how cells carry out basic biological processes. The recent years have led to the introduction of many assays for probing these interactions in a systematic and large-scale manner. However, there is a large gap between assay results and understanding of biological systems. The challenge for computational methods is to bridge this gap by combining results of different assays and introducing statistical methodologies. In this review we discuss recent advances in approaches dealing with these challenges, and key directions for the future.

Enzyme dynamics point to stepwise conformational selection in catalysis

Recent data increasingly reveal that conformational dynamics are indispensable to enzyme function throughout the catalytic cycle, in substrate recruiting, chemical transformation, and product release. Conformational transitions may involve conformational selection and induced fit, which can be viewed as a special case in the catalytic network. NMR, X-ray crystallography, single-molecule FRET, and simulations clearly demonstrate that the free enzyme dynamics already encompass all the conformations necessary for substrate binding, preorganization, transition-state stabilization, and product release. Conformational selection and substate population shift at each step of the catalytic turnover can accommodate enzyme specificity and efficiency. Within such a framework, entropy can have a larger role in conformational dynamics than in direct energy transfer in dynamically promoted catalysis.

Transcription factor binding variation in the evolution of gene regulation

Transcription factor interactions with DNA are one of the primary mechanisms by which expression is modulated, yet their evolution remains poorly understood. Chromatin immunoprecipitation followed by microarray (ChIP-chip) or sequencing (ChIP-Seq) has revolutionized the study of protein-DNA interactions. However, only recently has attention focused on determining to what extent these regulatory interactions vary between species across entire genomes. A series of recent studies have compared in vivo binding data across a range of evolutionary distances. Binding events diverge rapidly, indicating gene regulation is an evolutionarily flexible process.

Lysine120 interactions with p53 response elements can allosterically direct p53 organization

p53 can serve as a paradigm in studies aiming to figure out how allosteric perturbations in transcription factors (TFs) triggered by small changes in DNA response element (RE) sequences, can spell selectivity in co-factor recruitment. p53-REs are 20-base pair (bp) DNA segments specifying diverse functions. They may be located near the transcription start sites or thousands of bps away in the genome. Their number has been estimated to be in the thousands, and they all share a common motif. A key question is then how does the p53 protein recognize a particular p53-RE sequence among all the similar ones? Here, representative p53-REs regulating diverse functions including cell cycle arrest, DNA repair, and apoptosis were simulated in explicit solvent. Among the major interactions between p53 and its REs involving Lys120, Arg280 and Arg248, the bps interacting with Lys120 vary while the interacting partners of other residues are less so. We observe that each p53-RE quarter site sequence has a unique pattern of interactions with p53 Lys120. The allosteric, DNA sequence-induced conformational and dynamic changes of the altered Lys120 interactions are amplified by the perturbation of other p53-DNA interactions. The combined subtle RE sequence-specific allosteric effects propagate in the p53 and in the DNA. The resulting amplified allosteric effects far away are reflected in changes in the overall p53 organization and in the p53 surface topology and residue fluctuations which play key roles in selective co-factor recruitment. As such, these observations suggest how similar p53-RE sequences can spell the preferred co-factor binding, which is the key to the selective gene transactivation and consequently different functional effects.