The most important properties of these proteins is that, not possessing an adequate amount of hydrophobic amino acids, are unable to create an apolar core in solution and therefore do not form a globule but are dynamically characterized by a great flexibility. The flexibility is thus the major structural feature of these proteins. It derives from the presence in the sequence of a large number of amino acids of small volume and hydrophilic which do not remain trapped in a rigid core and exert their function of entropic stabilization of the system favoring an open structure in solution. Without going into the problem of the actual structure physically present in a crowded cellular environment, the main observation is that these proteins are characterized by their high flexibility, which depends directly on the physical and chemical characteristics of their amino acids. This is a property common to all proteins but that they possess to a greater degree because of their particular composition. It’s well known that the paradigm of "structure-function" valid for all the globular proteins is more appropriately described by the paradigm of "structure-flexibility-function". In fact, no protein can be functionally active or stable without adequate flexibility.
So, up to this point disordered proteins seem to base their existence on those physical and chemical properties that drive the behavior of all the proteins in solution. They represent a particular situation depending on their composition with a predictable and suitable structural response in aqueous solution. So what is the big deal? Why classify them in a different structural class? The fact is that they show only very high flexibility, a structural feature shared with all the proteins. They can be studied and characterized structurally with many of the techniques that are used for the other proteins (CD, fluorescence, NMR, viscosity, etc.. etc..). Why to use predictive disorder methods when a prediction of local flexibility based on clear physical and chemical properties of amino acids or an inspection of sequence is decisive? I would like to hear from others on this reflection.
However, it is my opinion that their structural features should be discussed on the basis of the molecular mechanism underlying the functional behavior, because their biological action depends on the evolutionary adaptation that has tailored the flexibility to functional tasks involving the recognition of a specific molecular partner, among many possible, in a specific cellular compartment, among the many possible, in a specific moment of the cellular metabolism, among the many possible. It’s the coordination operated by the cell of all these events what makes different the so-called disordered proteins, not their structure. Everything appears as a tool, already used in other occasions by the evolution, to save labor, through the use of biological objects already existing.
Dear Giovanni,
I agree with all what you stated.
Some statements need some clarification though. The modern view of protein structure is that they combine features of solid and liquid physical phases. Therefore the old paradigm of sequence, to structure, to function is just simply defunct. If you want to read about it a bit more, look at the works of many people, but particularly Tawfik and maybe one of my papers (Burra et al. PNAS 2009). This paper showed that every protein has both characteristics with variable and unique compositional percentage that determines, or rather is in tune with the function. Your statement that classifications are not needed is not exactly helpful because, as you see in the literature, the physical characteristics of amino acids are not unique and absolute (take into account buffering and titration and these are not the only variables, for instance mobility of an amino acid depends on the context). Like in life (organism alive in one environment but dead in the other) everything depends on context. From works by Minor and Kim we know that the same sequences behaves completely differently in different contexts (their works showed beta or alpha characteristics for the same sequences in different environments). (Read also about ambivalent sequences. I, myself found at least several examples in which the same sequence had oppositely different structural characteristics in the same crystal structure. From the works of Frauenfelder we know that dynamics of proteins are coupled with dynamics of solvent (slaved dynamics). For discovery of a new glassy state that controls protein function you can look at another paper (Teeter et. al PNAS 2001). From the works of Kern we know that dynamics modulates the function (not imprinted in structure alone).
Therefore, it is of paramount interest to classify the behavior of particular fragments (or stretches of residues) whether they have ordered (solid-like) or disordered (liquid-like characteristics in the particular environment. Also, you confidently started from a physical point of view stating that if we only knew the thermodynamic quantities (H,S,G) we would be able to be happy predictors. This belief is widely spread among the classically trained researchers but think for a second, or two, about this life conundrum. Life and therefore proteins must be unstable. Otherwise we all would live forever defying laws of thermodynamics because that would require exponential use of energy (in whatever form you think about). Life and proteins are therefore complex systems that have limited time and space existence. Besides such a point of view is a reductionist point of view and neglects that complex systems usually exhibit chaotic behavior. Knowing the state of S or H for a liquid tells you literally nothing about the shape it is in. So the same applies for proteins. As in a chaotic system the only possible descriptor is the knowledge of the entire phase space which is equivalent to knowing the limit cycles and extend of space in which the chaotic behavior dominates. This is computationally intractable task. So classical thermodynamics has very little to do with what proteins are and therefore the old protein folding paradigm of sequence determining a structure and structure determining function is only the first and a very crude approximation. In eukaryotes only less than 20% of proteins would fold on its own which underlines the importance of a context and complex character of proteins.
So, in conclusion I think that you express mostly sound and appropriate views but somewhat internally inconsistent. If you think that function is determined by the environment and localization in the cell why you are confident that simple sequence characteristics provide this description to the protein. You are mixing up the reductionist and holistic points of view in a single statement. At the same token I agree with you that there is no need for reclassification of proteins as being intrinsically disordered or not. I proposed in my papers to determine the degree of liquid to solid characteristics and assigning them to the structural context. In essence I advocated a continuum description of proteins not a dichotomy. However, as a beautiful corollary to my previous statements, the whole field of intrinsically disordered proteins is a regular marketing trick on which scientist are making careers. Naming things makes you a name, and as such it is normal. Our duty though as growing scientists is to see these ploys and discard them as a vestige of all reductionists views that are failing us and impede progress.
Dear Boguslaw,
thanks for the holistic tone of your reply. This is what I expected someone did. Years and years of reductionist approach to the study of proteins are not easy to clean up from the minds of those who study them. Understanding a new logic and new aspects is complex, especially if they are not discussed in an informal way. The rise of the so-called “omics disciplines”, with a global vision of the scientific problems, both from the logical point of view that methodological, is leading to major changes and a decisive leap forward in our knowledge, after years of stagnation, but many researchers, particularly those of bio -medicine, are still anchored on the old deterministic view and do not understand that overcoming is obtained, at this stage, with a large and ultidisciplinary collaborative effort needed to spread up and understand the proper holistic view.
I hope that others will want to express their opinions on this topic.
If we define natively disordered proteins as proteins with low secondary structure that adopt multiple conformations upon binding to different ligands, there is practical reason for doing database searches for them. Empirically, natively disordered proteins are more likely to be signaling proteins, more likely to have a toxic effect when overexpressed, more likely to aggregate,have on average half shorter half lives than structured proteins, and other properties.
On an individual level, knowing a particular protein is unstructured is useful for planning future experiments and interpreting data. For example, the order in a NMR structure of a partially disordered ensemble is overestimated. In effect, the ordered conformations are selected out of the ensemble. Similarly, knowing if a protein is natively disordered gives insight into the difficulty of crystallization and into possible crystal packing effects.
Dear Jeffrey,
I appreciated your comment. It, however, is essentially phenomenological, namely, it describes obvious aspects of these proteins. The real question is, are to be considered different from globular proteins? or are an extreme case of the globular world (and thus the paradigm of structure-flexibility-function still exists). Or are they something else? So we must try to define them in some way, because we have some important techniques for rigid protein (globular) that do not work and it is difficult to speak about them without being able to see these proteins in our minds. It is a bit like the function of the probability distribution in the electronic quantum mechanics. It is a mathematical function (Ψ) therefore not easily representable, but if we generate a surface that encloses a certain probability (Ψ2), then we can see the electron in our mind with a very precise shape that reflects also some of its functions. This helps a lot. However, it is only a simplification because the essence is to understand whether they are or are not part of the world of the proteins that we know.
Dear Giovanni,
I agree with Jeffrey. Things that may be obvious for you are not necessarily obvious for people with a less biophysical background. I think these predictions are helpful, e.g. to design constructs for structural studies. They are also helpful to prepare researchers for "unsusual" behaviour of their protein of interest e.g. on size exclusion chromatography, SDS-PAGE or co-purification of cellular chaperones. Finally, these algorithms are also useful if you combine them with additional information. We work a lot with short peptide motifs that bind to globular domains. If you want to predict such a motifs, it is very helpful to look mainly in unstrucutred regions and not in globular domains.
Dear Giovanni,
The first problem with disordered proteins, i think, is their name. As you wrote, in a structural point of view, they are not so different, they could have stable structure, they have functional (coupled withs structure) states. I agree with you, that a functional classification would be much better to these proteins, especially from the point of view of their the molecular mechanism. If a new student come to our group, and i start to speak about IDP-s, the first thing i try to tell them, to forget the structural things about disordered proteins. They say "The name of the research group, is group of disordered proteins, so they shuld have no structure", "the first thing you can read about IDPs, that they have no structure", etc. Yes the strutural propensity was maybe the first thing we saw about them (more than 10 years ago), but now we know about them much much more. Of course i have the answers (or the questions back..) to the students, with some nice example. I always explain them the specific functions, related to disordered proteins. "Can you make a filter like the nuclear pore meshwork from spheres?", "How big globule do you need to transport the huge amount of calcium, as the DSPP do in osteogenesis?", "How can a globule interact with several hundred partners (maybe in several hundred structural ways), like p53 do?". So i always try to convince them, to look these proteins in a differnt way, if you see the globular proteins with your eyes, try to hear the IDPs with your ears, forget (almost) everything you learn about globular proteins, and try to approach this new field without any preconception. Of course you will find a lot similarities, but you will also be able to find speecific (different) things related to the so called disordered proteins.
Do they different from the globular proteins? It depends on the question you try to answer.
If you are interested in the specific residues responsible for the interaction, you can use conventional methods, like NMR, not a real difference. If you are just interested in the structure (structural ensemble), you may need some extra tool, like MD, and you cannot use some conventional method, like X-Ray, but the difference again is not significant. If you want to observe their function, you cannot measure enzymatic activity, but you can measure interaction parameters, not a real difference.
But what happened, if you try to make a mechanistic model about their interacion, about ther behavior? You have to "think different", considering their much higher flexibility, adaptability. What about the funcional elements in globular and disordered proteins? While you have domains (long peptide region, high conservation, etc) in ordered proteins, you will find motifs in disorderd proteins (usually less than 10 residues, only few residues are conserved). We have to search, manipulate, mutate these functional elements with different thinking. What about their evolution? It is definitly different, usually not the residue, but the property is conserved, or much more the property of the disorderd region is conserved. You cannot use the conventional substitution matrices (like BLOSUM) to find homologies. I think we can find many aspects, where they are significantly different from the GAPDH like proteins :).
I our lab, we use conventional biochemical and biophysical methods to characterize IDPs, like we should do it with a globular protein, but we always search (and usually find) something, which is clearly different compared to the ordered ones. We also use many bioinformatic approach, which are, i think, more customized to the disordered proteins.
So, i see many differences between ordered and disordered proteins, but it is not only depends on the results we have, it is also based on a different, "shifted" tinking.
I agree with Jeffrey and Gerrit. As someone working in short linear motif prediction, it is useful to sub-classify proteins in this way, even though any classification of this kind is likely to be a false dichotomy to some extent. In other circumstances, it is more useful to think of a propensity to form (dis)ordered states. Given that the interaction mechanisms and dynamics are likely to be different for globular domains versus "intrinsically unstructured" regions of proteins, they can indeed be considered as different entities in certain circumstances. Given that is it presumably the same set of biophysical rules that determine their propensity to disorder in any given environment, in other circumstances such a distinction does not make sense. It all depends what aspect of structure/function you are interested in at that moment.
Giovanni,
I do not quite understand what you're saying.
You write that intrinsically unstructured proteins "... can be studied and characterized structurally with many of the techniques that are used for the other proteins (CD, fluorescence, NMR, viscosity, etc.. etc..)"
I would think not. If they do not have any structure, the CD spectrum would not tell you anything. In a classical outdated world of structure-function paradigm, the ability of the protein to fulfill its function could be easily tested by CD - if the right structure wasn't there, the protein's activity would be gone as well.
If you have two proteins alpha-lactalbumin and beta-lactoglobulin you can easily tell one from another by CD (helical vs. sheet). In unstructured proteins you cannot tell one from another. In addition you cannot tell what function they fulfill. I mean both can have exactly the same random coil native state, although their functions could be different. However, in the structured proteins, particular structural features are often associated with the particular function.
In unstructured proteins you derive no useful information from their structural characteristics (or the lack of those). That is when "... prediction of local flexibility based on clear physical and chemical properties of amino acids or an inspection of sequence" become important - they give at least some indication of the potential function.
Yes, of course flexibility is extremely important for the function, so it is for the stability. However, optimization of both at the same time might not be possible for a specific activity in a specific cellular compartment. In some cases recognition of the binding partner depends on the structure formation in an unstructured protein, again making (un)structure prediction important.
Is that what you meant?
Dear Giovanni,
I really appreciate what you are trying to do. Language is the key.
In development of modern sciences we inherently have two elements: 1) descriptive, 2) quantitative. So, sharpness of language, or the naming conventions play a crucial roles in our progress. So let's all settle on naming proteins as proteins. Obviosly initial names are helpful but there is no additional quality in naming subclasses as helical, beta or natively disordered. We, for instance showed that there is only a miniscule difference in amino acid composition between the sequences that always fold, sometimes fold or never fold in the PDB representations. But there was a substantial difference which class participated in signaling by being prone to postransational modifications (Zhang et al Structure 2007). It is easy to imagine that some elements are folding under one set of conditions while a different part of the same sequence at another. This kind of knowledge is useful and needed. So packing everything to one bag and calling it natively disordered does not make sense. I understand Jefferey and Lajos at the same time. While simple phenomenology is useful at first as registering observations and designing simple experiments (like crystallizations) it does not address the main point about what quality and to what degree is different.
AND, this is the essence, in my opinion of the Giovanni's remarks and questions. Some folks clearly get offended by my remarks about marketing ploys, but if you look behind my remarks, dominance in science is just salesmanship. Some bad ideas persisted for millenia only because some powerful forces were behind them. So instead getting offended we should think how to design experiments to create predictive (quantitative) tools that would allow us to cut the crystallization construct at the right residues (for instance, identification of hinges Stec, FEBS L 2012) versus cutting a helix in half (which happens a lot) and creating additional disorder in the final structure.
By the way Jeff, disordered proteins cannot be defined as having low level of secondary structure. I am even neglecting the ensemble view strongly advocated by Hilser who treats everything as a partition function of conformers from completely unfolded to the completely folded. Multiple classes of transcriptional factor fold only on the targets so 'per se' they do not have even momentarily any "level" of the secondary structure. So as we see even the definitions are not that well established. For the sake of science we should do better.
Dear Boguslaw,
Science progresses by making models of reality rather than trying to consider all complexities in one go. Yes, those models simplifications that does not mean that they're not useful. Four example, we model humans as individuals, when in fact we are communities of many different cells and even species. In many cases this is a useful simplification to make. When you get cancer, however, or a problem related to your gut flora, the complex reality becomes more important.
Models are useful for understanding, designing experiments, and teaching. Good scientists will always have one eye on the model and one eye on the complexities of the reality behind it. If you drop either you're not going to succeed as a scientist. This is an important lesson for young scientists. Attacking useful simplifications in this way is therefore counter-productive and does both science and good scientists a disservice.
Finally, have you neglected environment in which these initial classifications were made. Even six years ago I was still in encountering many structural biologists obsessed with the structure-function paradigm, who didn't think that anything without a stable structure could possibly be important for function. Emphasising a different class of interaction or even whole proteins that may be dominantly acting through a different mechanism was/is therefore incredibly important. This is not just about marketing or dominance!
I appreciate the attempts to clarify thinking. It is certainly true that there is a continuum and any divisions we make are going to be fairly arbitrary. However, I think that the concept of intrinsically disorders proteins and regions is a very useful one and the model will serve us well for many years to come. After all, Newtonian physics is still useful in some situations and yet we all know that is a simplified model of reality. I should also point out that even naming proteins as proteins is not as simple as you suggest. What about glycoproteins? Ribosomes? Why not just call them organic biomolecules? But then what about those with active metal ions, etc? I think that you are in danger of trying to draw an arbitrary line in a place that suits YOUR perspective and erroneously projecting that onto others as some kind of universal truth, which it is not.
I would say first that classifications are not good or bad by themselves, but useful or not depending on the specific case or purpose.
For example, there is some controversy about "conformational selection" and "induced fit" in ligand binding to proteins. Some protein-ligand interactions seem to correspond to conformational selection, while in other cases induced fit may seem more appropriate. I would say that any protein-ligand interaction event, can be described, at some extent, as a combination of both, in some cases one of them predominating. Conformational selection and induced fit are just two limit cases regarding conformational changes and protein conformational equilibrium coupled to ligand binding: if the energy gap between the two conformational states connected by ligand binding (free protein and bound protein) is small conformational selection predominates, whereas if the energy gap is large, induced fit will do so. With structured vs. disordered proteins the same reasoning applies. Proteins are molecules with certain flexibility, heterogenously distributed through the molecule. Some regions in the molecules show higher propensity to be unfolded and other regions are more structurally rigid. Even a completely folded molecule may exhibit some regions with low structuring. And even a a disordered protein may exhibit some regions with residual structure.
We should address this issue considering that the conformation of a protein depends on the primary structure (sequence of aminoacids) and the context. For example, a given protein may be folded at a given pH, but unfolded at another pH. This pH-dependent conformational equilibrium is a reflection of the previous assertion: the context. Proton binding/dissociation at certain ionizable residues will shift the equilibrium towards certain conformational state(s). And the context in this case is the environment (solvent) with low or high proton concentration. We cannot describe the protein equilibrium without considering the protons (binding).
Therefore, in proteins we should always consider binding together with folding. Proteins in plain water as a solvent and at constant temperature and pressure will never undergo conformational changes. However, they interact with other molecules (ions, low molecular weight ligands, proteins, and other biomolecules) and those interactions will shift the conformational equilibrium toward certain conformational states, which, in turn will show different functions (activities). This is the underlying idea of the general allosteric model developed by Wyman decades ago for explaining protein function regulation by ligand binding (in general, all proteins can be considered allosteric since their conformational equilibrium is controlled by ligand binding). And for disordered proteins, binding is even more evident since the structural rearrangements towards a (partially) folded conformation are triggered by the interaction with a binding partner from an unstructured state. The main difference is that the conformational change upon ligand binding in structured proteins is relatively small, and in disordered proteins is dramatic.
The context for a protein is constituted by the environmental variables (temperature, pressure) and the composition (different molecules and their concentration) of the environment surrounding the protein molecule. In the absence of a given binding partner, all conformational states are possible, but only a fraction of them are probable (that is, only a fraction of the conformational states will be significantly populated). The protein will explore (populate) certain conformational states. In some proteins the ensemble of conformational states compatible (populated, probable) with the absence of ligand is reduced and well-defined (structured protein), while in other proteins is not (disordered protein). In the presence of the binding partner, again, all conformational states are possible, but only a fraction of them are probable (that is, only a fraction of the conformational states will be significantly populated). The protein will explore (populate) certain conformational states. For structured proteins, the ensemble of states compatible (populated, probable) with ligand binding can be considered a subset of those compatible (populated, probable) under the absence of ligand binding. For disordered proteins, cannot be considered a subset.
From my understanding, disordered proteins are highly dynamic in nature, and while techniques like NMR or CD can capture the dynamicity of the proteins, they fail to understand the 3d structural architecture of the disordered protein. The disorderedness can related to molecular chaos, suggesting a possibility of forming a variety of transient short lived complexes.. probably ternary complexes with several proteins. The chaos is responsible for the very "active" nature of these proteins making them readily available and perfect candidates for temporary PPIs. This could be the probably difference between disordered and ordered proteins. As for their functions, they can play a role beyond signaling proteins. For example, Microtubule Associated proteins (MAPs) such as MAPT (involved in diseases like Alzheimer) are inherently disordered in nature and are essential in stabilizing microtubule filaments. Thus disordered proteins have a defined structural role to play but the precise nature is yet to be understood clearly. One way could be by the help of "omics" method.
Thanks to Prof. Colonna for starting this debate on one of the most important topic in modern day proteomics and shedding some light on the fundamental aspects of protein structures.And also to Prof. Stec, who has provided a holistic view on this matter. With immense respect to every contributor , I as a naive contributor, would like to contribute a couple of things as per my understanding.
After going through all the posts and suggestions can we conclude that irrespective of naming terminology or categorization of intrinsically disordered proteins or ordered proteins(as this is not our aim), the ultimate functioning of protein depends on a blend of both the states(disordered & ordered)?
Like as highlighted by Dr. Adrian , the view on conformational selection suggests that proteins without structure obtain a conformation or populate a conformation upon binding to a specific partner, may it be a small ligand or small peptide/protein. Thereafter, on binding with the partner it moves towards a structured state and finally it shows its biological functions. importantly in this scenario, we are considering a holistic view on proteins, because we are taking into consideration firstly their flexibility or disordered state, their environment in which they seek biological partners, their ordered state and finally their function. As, ultimately fate or aim of any protein(disordered or ordered) is to show some biological function, therefore the blend of both the protein states propel them towards their ultimate function.For ordered or structured protein the paradigm of structurefunctions holds true ground and the studies can be based on this ground but for other category of proteins we need to have a fresh or different view.
So can we suggest that the new paradigm instead of Structure-Function can be termed as this(for disordered proteins) :
ProteinProtein Flexibility/Disordered state Protein Structure/Ordered state Protein Function.
Between the protein disordered state and protein structured state there exists a mechanism which prompts proteins to attain a structure(may be even partial) and that mechanism can be:
[Selection of Protein/Peptide binding partners in environment.(e.g. in cell)]
This is based conformational selection and conformational population.
I would request to have inputs and suggestions from all contributors and correct me If I am wrong in my understanding.
You might find this paper interesting: http://www.ncbi.nlm.nih.gov/pubmed/19260013
Tompa P et al (2009. Close encounters of the third kind: disordered domains and the interactions of proteins. Bioessays 31(3):328-35. doi: 10.1002/bies.200800151.
There is evidence that some for some protein( region)s, the disordered state IS the functional state. "The most common function associated with these domains is molecular recognition, during which they undergo a disorder-to-order transition, but in some cases they may remain disordered and function as entropic chains."
As an evolutionary biologist, I would also warn against assumptions like "As, ultimately fate or aim of any protein(disordered or ordered) is to show some biological function..." - there may well be (regions of) proteins that have essentially no (important/necessary) function and just represent evolutionary baggage. They may well interact with stuff but that is not necessarily why they exist. The question before "how does this attribute of a protein contribute to its function?" is always "does this attribute of a protein contribute to its function?" (and before that, "does this protein have a function?").
Even the lack of a well-defined conformation can be considered as a structural feature. The plasticity and high flexibility of a disordered protein is instrumental for its function. This protein, due to the interplay between entropic and enthalpic contibutions to the Gibbs energy of folding and binding, will interact with a binding partner by restructuring a certain region of the molecule. Because there will be an energetic penalty due to the folding, the affinity will be lower than if the protein was prefolded. Therefore, the high flexibility of a disordered protein allows its interaction with multiple partners with moderate to low affinity (transient complexes).
Finally, I would suggest that no distinction should be made between structured and disordered proteins. It is important to know if a given protein is disordered or not for certauin purposes (e.g. obtaininging crystals or expressing it soluble), but, in principle and in a broad sense, there is no difference regarding conformational equilibrium and binding or function: a completely disordered protein undergoing a large conformational change upon interaction with a binding partner and a completely structured protein undergoing a small conformational change upon interaction with a binding partner are two limit cases from a gradation of possibilities.
@Adrian, I would agree with all of this except the suggestion that no distinction be made, unless you mean no distinction that is maintained in all circumstances. For reasons that have been alluded to above, it is often useful to make this distinction in certain circumstances for practical and/or conceptual reasons. This is pretty standard practice for biology. You cannot consider all things in all situations and so simplifying assumptions/distinctions are sometimes the only way forward. The trick is to make sure that your simplifying assumptions/distinctions are fit for purpose - and not to extrapolate them beyond the domain/application/experiment/analysis for which they are appropriate.
@Richard, Yes, I agree with that. That is why I said that in some cirsumstances it is useful to know if a protein is intrinsically disordered or not (for example, whether or not initiating an X-ray crystallography project or performing NMR experiments). Likewise, it is important to know if a protein has no tryptophan residues or has many of them if fluorescence assays are planned for getting structural information. Classifications constitute a simplification that may not be appropriate in general, and may lead to erroneous conclusions or misinterpretations if not applied wisely.
What I wanted to stress is that, in a broad sense, intrinsically disordered proteins are not so different from intrinsically structured proteins.
I'd like to add that the computational techniques used to study IDPs are also different from the ones used for folded proteins. Not formally, if one starts from first-principle models, but in practice, the approximations valid for some systems are not valid for others. For example, using Elastic Network Models for IDPs makes no sense, and using a random-coil model for a folded proteins is also absurd.
Of course, the challenging situations are the intermediate ones, where ones needs to test which models or experimental techniques are applicable.
The waters are rather crowded with these disordered proteins, I would say... My opinion is that ALL proteins attain their 3D functional state due to the interaction with its surroundings. Let me explain this: globular proteins fold in presence of water, otherwise -- no hydrophobic effects would have been present. The disordered proteins go into functional 3D state only in the presence of ligands as well. So that the most important is to establish a proper way to describe protein-ligand (in broad sence) interactions, which results in target 3D structure, and which would be valid for both globular and disordered proteins. I do not know the recipe how to do that, though. Since you were talking phylosophy and methodology, this is my private opinion.
Another open question for myself is: what are the key variables or what the language should be used to describe disordered proteins? What is the exact question or problem under consideratin we are trying to solve? Obaining snapshots which look like some structures measured, or what? I would be very interesting to see the changes in radius of gyration distribution or mean-square radius of gyration vs temperature (or pH or other external control variable) for some globular vs some disordered proteins. The question is which polymer model to apply? Is Go-like scheme applicable?
My feeling is that we do not have clear understanding even at the level of problem formulation and need to invest time in it. If one looks for a phase diagram and the answer to the question WHY, it is necessary to speak in polymer physics terms. Otherwise one can go on and on speaking about paradigms, pathways, evolutionary optimization. etc., without approaching any at least qualitatively correct answer to the question WHY. The info about pathways, etc. should result from the correct description, not vice versa. Sorry for beeing rude.
@Artem Radius of gyration give very little information on the local structure of IDPs. See for example: http://www.pnas.org/content/101/34/12497.full.pdf I am not saying Rg measures are useless, but they need to be complemented with other techniques to characterize an IDP in solution.
@Artem: I'm sorry but I do not understand what you mean. Why what? (And why is your answer rude?) If you mean that we need to first decide why we are interested in defining "disordered" proteins/regions, I think that many here (including myself) would agree with you and have been saying just that - the globular/disordered distinction is useful in some situations, not in others.
To Ramon: Thanks for bringing the paper by Fitzkee and Rose to my attention. I read it and see no contradiction to what I have been saying -- vs temperature would be extremely helpful and would explain us many peculiarities of intrinsically disordered versus globular. The paper you mentioned didn't report any such a curve. On the other hand, complementary methods, such as rotational viscometry or spectrophotometry or CD would give us a hint, if any important secondary structure element transformation goes on after folding/unfolding event. Comparison with similar curves for a globular protein (say, CI2) could provide us with the important info on the system.
To Richard: I meant, that physics normally answers to the question "why", and explains the grounds, while pure biology gives the answer to "how" and shows what happens with nice figures and snapshots, that illustrate how do the things happen, without explaining. To some scientists such an opinion may sound rude :-)
The most of papers I have read on disordered proteins (for instance, Tompa P et al paper) were describing such proteins in terms of their biological function, evolutionary mechanisms that made such proteins important, etc, not attempting to explain or at least to formulate the problem in polymer physics way. It is interesting and informative, but doesn't anyhow answer "why".
Richard said: "I think that many here (including myself) would agree with you and have been saying just that - the globular/disordered distinction is useful in some situations, not in others."
--- My suggestion is to check if there is any difference in basic polymeric characteristics between what is called "globular" and "disordered" protein, before going further. Do (quasi)phase diagrams differ, for example?
And I do not argue against anyone's opinion, presented above, I am trying to complement the opinions by my own, personal view on state-of-the-art in the field.
I appreciate your comments,
Thank you!
I have a hopefully simpler question. Is it important to recognize that a disordered population of molecules may exist, under certain circumstances, even for intrinsically ordered proteins?
To Raffaele: till now we spoke here about a single macromolecule, or, if you wish, about a molecule from the dilute solution. Taken so, ANY protein, be it globular, or IDP, under certain external conditions (temperature, pressure, pH, or etc....) MUST be found in disordered, structureless and wildly fluctuating state. Questionable is if in practice we may observe IDP folding back to a structured conformation. Both disordered, called 'random coil' conformation or unfolded, and native, or globular structured conformation exists for any polymer with excluded volume interactions and attraction, it is just it may be inobservable due to say, water freezing or boiling.
I don't think that certain conditions exist under which IDPs must be found in a structureless state. This is true for structured proteins, which may ideally populate two limiting states, from whose free energy difference what we call the structural stability of a protein can be defined. I can only think of IDPs as intrinsically structureless in a physiological condition. This is what makes them strongly different from structurally organized proteins. On the basis of my experience, I am oriented to think of IDPs as proteins totally lacking one of these two limiting states, the structured one. From a thermodynamic point of view, this makes their hypothetical structured state infinitely unstable. It may be that in some conditions IDPs can assume a structured conformation. However, I am not sure that the structure-function paradigm may be applied to them after whatever modification. For example, couldn't the structural organization that IDPS apparently assume upon binding to some other molecule be a consequence of their plasticity in adapting to the structure of the bound molecule?
This is related to what has been commented before: In proteins, folding is context-dependent. And this context is constituted by the solvent, ions, small molecular-weight solutes, other proteins and biomacromolecules, etc. and the physical environmental variables (temperature, pressure,...). In all proteins, IDPs or "regular" structured proteins, folding or structuring occurs mediated by, and as a consequence of, the interactions within this context. It is not completely true that just the primary sequence of a protein determines the conformation of the protein: the same protein may exhibit different conformations depending of this context.
In some proteins you may have (at least) two well-defined states separated by an energy barrier, in others there is no significant energy barrier (down-hill proteins), and in others there is no structured conformation (in the absence of a corresponding partner).
The main difference between IDPs and "regular" proteins is the intrinsic stability or plasticity. In IDPs, the internal interaction network and the interactions with solvent and ions is not enough to achieve a folded structured state. Binding to another protein/ligand provides the driving force for folding (at least, partially). But in "regular" proteins occurs the same, but in a lesser extent (that is, the conformational change induced by binding connects more or less structured conformational states).
Thus, IDPs and "regular" proteins are limit cases of the same concept, and general, albeit somewhat vague, principles can be generalized to both classes of proteins. Therefore, I consider there is no "significant" difference between both classes of proteins, but it is important in some cases (e.g. from an experimental point of view) to make that distinction.
Intrinsically disordered proteins are highly flexible proteins in its conformational dynamics and it is their inherent character in order to perform functionality as if they achieve a folded state it is quite possible that they cannot perform the functionality. In comparison with globular proteins(stable proteins ) which are well studied and we already know order of these proteins well because of their simplified interaction between amino acids where as in disordered proteins its just we are not able to decode that order in complex interaction in residues.Complex (disordered) systems does have an order to work efficiently as these system will have a larger requirement of systematic order to perform multiple functionality by stratifying the tasks in order of necessity..These complex systems(disordered proteins ) are flexible to adapt,, co-operative and mutitasking and robust sometimes but simplified systems(stable proteins ) are monotonous, rigid and mostly can perform one or two task at time.
Scientific community is still searching for adequate resources and protocols on decoding complex interaction between residues such as cohesiveness, co-operation and adaptation levels in amino acid interactions.
my comments on the last observations are contained in an answer given to the question asked by Francesco Secundo about "What about the stability of Intrinsically Disordered Proteins (IDPs)?"
the address is below:
https://www.researchgate.net/post/What_about_the_stability_of_Intrinsically_Disordered_Proteins_IDP
to Rafaelle: what I am trying to say is that be it IDP or globular protein, the phase diagram, or, in other words, the general regularities in behavior should be similar. Depending on primary structure the stability can be shifted to one or another side. You might take a look at Uversky et al papers, where he shows the appearence of order in IDPs at changing external conditions. And I am not saying a single word about structure-function paradigm :-) Unless it has not been explained how and why are these proteins found disordered under the conditions the others are ordered, we can keep talking without knowing what do we talk about, I think. IDPs are made out of the same aminoacids, as the others, and operate inside the same intracellural environment, so my question is: What is so specific in their primary structure and interactions between monomers and solvent, making them disordered at normal conditions? I am writing a paper on the topic, by the way ;-)
To Artem: Please, take a look at Uversky's publications. He has tried to rationalize the differential features of IDPs compared to structured well-folded proteins.
A wholesome review on disordered proteins by Vladimir Uversky with very nice initiative to summarize research of previous quarter in a review
http://www.landesbioscience.com/journals/idp/2013IDP015.pdf
Thanks for your comments! I am aware of these publications. What I see important and am working on is the phase diagram of a model protein in water and explanation, why and how did it happen that IDPs are disordered. Uversky's ideas help, but in addition it is important to show that they work on a model and on the language of formulas and graphs -- exactly what I am doing. Thanks for your comments once more!
@Artem Badasyan - That's an interesting aspect of studying Co-operation and adaptations between amino-acids and modules (domains ) of proteins in different circumstances. such as how much each module (domain) or amino acid invest in performing function. As we know in proteins, only some residues can be considered as key residues other amino acids are needed for facilitating those amino acids to perform function
Dear Ankush, Artem, Adrian and Raffaele,
Unfortunately, my remarks disappeared (they can still be retrieved by requesting to "show all discussion") and they addressed all your concerns. What is slightly ironic that my comments disappeared but the opposition to my remarks by Richard Edwards survived.
The modern view of proteins is that they combine solid-like and liquid-like characteristics. This language is important because one can reasonably expect that the solid-like part shows some thermodynamic stability (energy minima) that determine the structure. The liquid-like portions are not expected to have a well defined structure determined by the minimization of any energy related functionals. Therefore, as I commented before, the language of IDP only obfuscates the issue, because it stresses the duality, or this, or this. The reality is far more complex. Practically, it means that there are proteins (or their parts, sometimes domains) that are folded most of the time, folded conditionally or sometimes, and never folded. Every protein has both components (solid-like, and liquid-like) in a characteristic and determined by evolution (to optimize the function) proportion. The simplest consequence is that every protein is metastable (conditionally stable) and never exists in an absolute energy minimum. If it were it would be dead as a dead log, and would not have capability of performing function (life). Therefore, the review by Uversky is very important but we have to reach out beyond this language and develop tools that Artem is trying to advocate. We have to understand all the elements of the conditionality in the structures and their connection to the function.
For instance, recently there was a very interesting work published. Proteins as glassy systems exhibit the so called Boson peak in their vibrational spectra (overabundance of low frequency modes). The author provided evidence that the presence of the Boson peak originates form a dynamic selection of the fluctuations that can be connected to the function. In another words proteins to function need to have a certain low energy modes of vibrations that in a long time regime might lead to their instability. This is the first clear example of applying heuristic understanding of what proteins really are that goes beyond this outdated paradigm of sequence to function relationship.
I commented on other threads that this paradigm needs to be revised as very few sources discuss orthogonality of these two stressing out only colinearity. It is true that high sequence homology leads in many cases to high degree of structural similarity. But the interesting cases come from studying a high degree of sequence homology and radically different structures or from completely unrelated homology and identical structures. So I would encourage to comment on the issues of ambivalent sequences (ambivalent to the context of structure) or these that never fold (a significant progress has been made to why). I also would like to turn attention to the issues of coupling the dynamic properties of proteins to the solvent so strongly advocated by Frauenfelder for so many years.
To Boguslav: I mostly agree with your comments, just have to remind, that proteins, being polymers, have some polymeric specificity. Solid-like and liquid-like analogy is not much useful in that respect, since polymeric systems have properties in between the two. Low energy modes of vibration may mean that to function, the protein must have large fluctuations. Such fluctuations are intrinsic properties of any polymer in coil state, including the protein. The globular state, instead, is poor in fluctuations. Clearly, to perfom any biological action, the protein needs to attach to or accomodate many low- and high-molecular weight compounds, therefore, the low energy modes show up. Hope I am not missing your point.
Regarding the solvent -- exactly! It is often forgotten. There is even a post alive now on Research Gate, where a scientist asks, why does he have to do simulations for a protein in water, not in vacuo. It is franky to read protein folding papers, that, on one hand, state that hydrophobic interactions lie in the heart of protein folding, and on the other hand, do not account for the water itself. The situation in the field has reached a catastrophy: more and more meaningless results appear due to the loose grounds that become "classical and well-known results". After the 90-ies and the breakthrough results of Bryngelson and Wolynes, Chan and Dill, Shakhnovich and Finkelstein and several others (sorry for not mentioning, I just wanted to say, at that time there were much less scientists in the field, publishing breakthrough papers), there is a stagnation. Although there are several nice attempts to explain the hydrophobicity, namely, to show how the 'fear to water' is arising due to the protein-water microscopic interactions, it is not widely accepted, instead, everyone blindly cites Kauzmann. Although Kauzmann didn't say that there is a new kind of physical interaction, the hydrophobicity is often considered as if it was true, thus leading to paradoxes and artifacts. Taking into account that the biophysics is a hot topic (though seems the 'nano' becomes more hot), a lot of free and commercial software that use these improper ideas become available durig the last 10-15 years. There is a generation of scientists grown on such improper tools, so that none doubts the correctness of the grounds anymore. This is the end of a folding story and a very sad tendency -- to be successful you have to use the software and must avoid digging in the grounds, since many others make their living basing on the same faulty ideas, hidden in the soft.
Artem,
Not all hope is lost yet and not everyone blindly cites Kausmann. Try to read for example:
Baldwin RL ( 2013) The new view of hydrophobic free energy. FEBS Lett. Apr 17;587(8):1062-6. doi: 10.1016/j.febslet.2013.01.006.
or
Alan Cooper (2000) Heat capacity of hydrogen-bonded networks:
an alternative view of protein folding thermodynamics. Biophysical Chemistry 85 25
or
George Rose's latest papers on hydrogen bonding as a driving force of folding.
You will probably enjoy this article by Ken Dill as well:
DOI 10.1007/s10955-011-0232-9
@Artem, I would not say that the solvent is forgotten. Solvent is considered either implicit or explicitly. You may use different methodologies at different levels of description of the inherent interatomic interactions. For the abundance of meaningless publications on Protein Folding you should not blame everybody, but the authors of those works.
I do not blame everybody, but instead share my personal feelings and concern after reading many papers on the topic. Wheather implicit or explicit, it should be clearly and explicitly reported in the paper, while it is rarely done.
To Marina: thanks for the citations! i have read the Dill's paper, as well as the one by Rose, as far as I remember. They are far from being widely accepted, though. Dill's approach is brilliant! Also, Debenedetti, Franceze and Stanley have reported very interesting results on the topic. These studies are far from being in the mainstream, though. A typical folding paper reads now like "we took a xxx field of a xxx simulation suite, added xxx filed to account for xxx and have run for xxx nanoseconds the simulation of an xxx protein from the xxx PDB entry''. Output of the simulations is reported in terms of averages of radius of gyration vs some variable. Who cares, that the average Rg is just zero, and only a mean-square average is not. Thanks!
Concerning 'hydrophobicity', in one of my last papers on protein stability, I have instead used the terms 'polar' and 'nonpolar' to identify two broad classes of interactions. I have also tried to show that the heath capacity change, which is usually seen as a signature of hydrophobicity, plays a minor role in protein stability. The paper raised some criticism and has got very few citations till now. See: PROTEINS: Structure, Function, and Bioinformatics 54:323-332 (2004); PROTEINS: Structure, Function, and Bioinformatics 64:789–791 (2006); and PROTEINS: Structure, Function, and Bioinformatics 64:792–794 (2006).
Of course, Rafaelle, no doubt! In the field of folding it is very hard to report anything that deviates large from the main ideas and 'paradigms' :-) Good luck, though!
I offer some comments, mostly borrowed from my PhD Thesis:
-In the field self-styled as 'Structural Biology' there is a pervasively dismissive tone carried by the descriptor 'unstructured.' This tone is unjustified because IDPs have adaptive roles in extant organisms.
-Primary structure is the same as molecular structure; all proteins have a molecular structure.
-Tertiary structure is the same as conformation; not all proteins have a single dominant conformation in the biological milieu.
-Both concepts can be honored if we refer to IDPs as polystructured. No protein is unstructured; some are polystructured.
EDIT1: in response to Richard Edwards' rhetorical, the question of the stability could be treated orthogonally to the question of the number of dominant energetic minima. Allosteric proteins and IDPs are both polystructured. I don't think it makes sense to worry about the intrinsic stability because the primary structure does encode the potential conformations (independently of whether they're observed) and the question of stability is intrinsically a function of the conditions.
As for the cost/benefit of introducing jargon, I agree with Edwards that it's difficult to assess the value in the abstract. If people converged on 'protein disorder' that would suit me just fine -- it's the still-popular alternatives that bother me; terms like 'unstructured' (prevalent and evinced in this thread) and 'natively unfolded' are problematic.
EDIT2: in response to Ewan Blanch, I find the concept of being 'natively unfolded' problematic for several reasons. 'Native' is problematic because the concept of 'native conditions' tries to coarse-grain all of the biologically relevant conditions into a single context. If a protein is unfolded in TBS, but folded when it binds to a biologically relevant (native!) binding partner, is it natively unfolded? 'Unfolded' is problematic because a) it disregards the possibility of transiently ordered species or the importance of being simultaneously compact and disordered, and b) it implies the nonexistence of a folded state as an exceptional property, when the large majority of sequences will not fold. A stable conformation is the exceptional behavior for a heteropolymer. We don't refer to the solution conformation of PEG or cellulose as 'unfolded,' do we?
That's a good point about primary structure, which is why I (and most people, I think) prefer the phrase "intrinsically disordered" rather than "unstructured". "Polystructured" is an interesting suggestion for an alternative. However, given that there is already a vibrant and active research community working (and publishing) on "protein disorder", I think this is already sufficiently well understood as a term and beyond replacement. I would also wonder whether the term "polystructure" could imply multiple *stable* conformations, rather than (or in addition to) multiple unstable structures. As a result, it would not get my vote as a replacement for disorder, although I can see circumstances where it might help as an additional descriptor.
Dear Ryan , thank you for your contribution which I find very interesting and realistic. There is a constant lack of interest, in people working on IDPs, for the molecular basis of the phenomena they study. In fact, they always try to highlight the functional aspects . The IDPs are trivially organic polymers with a composition that strongly favors the presence of charges. Their net charge is generally high . These simple molecular evidences classify these proteins primarily as polyampholytes rather than as polyelectrolytes. There is a growing literature on the conformational ensembles depicting these systems and has been also proposed a State Diagram to define their conformational states. I do not understand why this literature is not taken into account as it should. Of course no longer we can have a photographic vision with fixed structures of the protein over time, but a vision of many highly dynamic conformations that populate the system (although, narrowing the clusters of conformations most likely some simple visualization can occur) is much more close to the real physical state of these polymers. What prevents from using the “polyampholytic” approach, is, in my opinion, the strong globular-centric vision still diffuse in the researchers. Of course, this makes more complex the study of the physical basis of these systems. For example, the properties that are measured in solution are average statistical properties of the system and we cannot attribute a single structure to the population of objects under study. Moreover, it is not said that these molecules, formed by charged and polar residues, find the water a good solvent. This means that the interaction between the amino acid residues is energetically favored over water. The polymer physics explains that in such a case the molecule may collapse in ensembles formed by structures conformationally interconverting, and so on. I think the definition of “polystructured” is enlightening and opens a mental window on these molecules that remain proteins, but on which the evolution has made specific adjustments.
10 years ago I went to a meeting where several high profile crystallographers dismissed disordered proteins as 'junk protein', analogously to the similarly misused term of 'junk DNA'. I agree with several people here that research into intrinsically disordered/natively unfolded proteins was poorly served for many years by the old fashioned crystallographic view of what 'structure' was. Thankfully, most crystallographers I know now are much more appreciative of disordered protein structure and its biological importance. But some crystallographers do still seem to think that if you can't crystallise it then it's not important. With the right tools though we can now appreciate in much greater detail the structural complexity of disordered sequences, the roles of PPII helix, turns and transient strand and helical segments.
In terms of terminology I do have some concerns though. I agree with Richard Edwards, "polystructured" could easily imply multiple stable structural motifs rather than just interconverting ensembles of unstable elements. Perhaps rheomorphic, or plastic, better conveys the concept of malleability and conformational variability? I'm not sure what the problem with "natively unfolded" is, and I use that description sometimes myself. Admittedly "unfolded" can be a little ambiguous but in this context I think most people would consider that in terms of globular tertiary structure.
After about a year just now I think that the discussion is taking the right direction. The presence of multiple energy minima suggests that there is certainly dynamic interconversion between different conformations, although there may be some of them statistically more favored. Hence the name I have in mind is " dynamically polystructured ." However , the disorder, connected to a high net charge of the system, is present in many phenomena of molecular recognition. If you analyze the sequences of some membrane receptors (G-proteins) you can see that the disorder overlaps the charges. In these areas there is always great probability of encountering residues post- transationally modified, which means that by inserting , for example . , phosphates , we strongly alter the net charge of the system and, as consequence, we influence the population of the conformational ensembles, changing the charge fractions. The state diagram, suggested by Pappu, shows how, moving from a region to an other, the conformational state will change and most likely the function. What is not yet clear is that this structural framework derives from model peptides (simple polyampholytes ) in solution but we do not know yet, how these same peptides behave when they have constraints , that is, when they are covalently linked to the core of the receptor. I am waiting for other considerations.
To Giovanni Colonna: Dynamic interconversion between conformations is always there for any protein in disordered conformation (also known as 'coil' in polymer terminology). My understanding of the problem is the (still missing) explanation of why the IDPs, that are structurally same polypeptide chains as globular proteins, appear to be found in disordered, coil conformation at external conditions, where many other polypeptides (globular proteins) show stable 3D structure. Pappu offered a possible view onto the problem. What I do not understand in his picture is the explanation of the mechanism of folding that is valid for both IDPs and globular proteins. Also, most important is the turned out response of IDPs to, say, temperature increase -- they gain order, as opposed to globular proteins that lose order. What is the conceptual mechanism of folding after Pappu that explains the difference in behavior of polypeptide chains?
To Ewan:
> But some crystallographers do still seem to think that if you can't crystallise it then it's not important.
And they are right -- it is not a job of a crystallographer to explain the behavior of the systems they study, reporting the structure is their task :-) Crystallography is a (one among many) tool of structural research. Soft condensed matter physics, or, more precisely, polymer physics is the discipline for studies of polymer solutions. I kinda don't believe that conformation of the protein in crystals can serve as a basis to explain the rich behavior of proteins in vivo. We do not even know if protein is ever getting such a conformation in the cell.
To Giovanni Colonna:
> What prevents from using the “polyampholytic” approach, is, in my opinion, the strong globular-centric vision still diffuse in the researchers.
Hmmm... water provides strong screening, and if the polymer is weakly charged, due to water it may behave as uncharged, as far as I remember. This is not the case for polynucleotides, which have an electron charge per base pair. I remember the scaling studies of random polypeptide sizes (radius of gyration), that showed the behavior reminiscent that of uncharged chains, while for DNA the scaling agreed polyelectrolite picture. But have to check once more...
About the negative influence of the globular-centric view onto the progress in the field I agree, though.