Membrane Induced Folding of a cationic ß-hairpin

We recently developed a minimalistic coarse-grained model for lipids and peptides by using the Drude oscillator approach to model the  electrostatic component of Hydrogen bonds. With our models, we were able to simulate membrane induced folding of a cationic ß-hairpin, SVS-1 on POPS model bilayer.

Here is a short animation:

MSO8AAFC3CC173Colors: Hydrophobic tails: green; Serine: cyan; Ester: purple; Phosphate: tan;Peptide backbone: black;LYS: crimson and VAL side chain: pink. All bead sizes exaggerated for clarity. 

You can see the peptide fold and freely diffuse on the surface of the membrane.

Another cool thing is that we find membrane-induced peptide folding to be driven by both (a) cooperativity in peptide self interaction (something that is expected for folding) and (b) cooperativity in membrane-peptide interaction (which is pretty cool). For more, check out our latest article.

Water Transport in Membranes

Here is a movie I made today, it shows water transport through a membrane inserted peptide. All the colored beads represent lipid head groups, the peptide backbone is in black, the orange beads represent LYS side chains and the cyan bead is a water. You can see the water molecule move from one LYS residue to another and slowly to the other leaflet.

It is nothing new and processes like these have been observed before, but I think it is very cool to watch the animation.

Also, sizes are exaggerated for clearer depiction.

pore1.gif

aMD: a possible solution to sampling issues in bilayer simulations

Liquid-liquid phase co-existence in lipids have been observed since 1970s. There is a wealth of experimental data, however, deciphering the molecular details has been a challenge.

Ever since “Lipid raft” hypothesis gained prominence, many computationalists are trying to understand the content and structure of ordered and disordered phases of lipid mixtures.

Why is this relevant?
Several proteins (GPI-anchored proteins, cytochrome C) are known to partition into ordered phases. Also, the ordered phases are known to regulate the activity of certain proteins. There are studies that suggest the order/disorder of plasma membranes can be used to detect apoptosis (Detection of apoptosis through the lipid order of the outer plasma membrane leaflet).

However, here is the  problem:
The lateral diffusion coefficient of lipids is in the order of 10^-7 cm^2/s, which means, in a 100ns simulation (which might take weeks), a lipid molecule will cover an area of 12nm^2 (which is much lesser than the size of a typical “lipid raft”).

What is the solution?

  • Run really, really long simulations on really, really powerful machines (eg: this super-cool study on characterizing the ordered phase (7 micro seconds on Anton) –  Page on acs.org)
  • Identify a good enough sampling technique that will help sample lipid mixing degrees of freedom.


Sometime last month, I came across this paper: Enhanced Lipid Diffusion and Mixing in Accelerated Molecular Dynamics.

What is accelerated molecular dynamics (aMD)?
aMD lets you set a threshold, and adds a continuous (non-negative) bias potential to energy minima below the set threshold. So, the energy barriers above the threshold remains unaffected, while the ones above the set threshold are greatly reduced, which allows the system to sample many more conformations.

In the above figure: E is the set threshold energy . V(r) is the potential of the system concerned and alpha is an “acceleration factor”, which controls the smoothness of the landscape.

So, whenever, the potential energy falls below the set threshold, the energy is modified or biased:


The amount of bias is determined by:

How does one un-bias the potential?
Well, the authors claim that running a conventional MD simulation after sampling, helps the system relax back to the unbiased state.

Why do I think this is cool?
With this technique, the authors show a 250% increase in lipid diffusion. Even though their system size is really small and there could be size effects, the technique looks very promising.

 

Data from the paper: (Panel a) : conventional MD simulation (Panel b) data from aMD

Reference:

Accelerated Molecular Dynamics

Cooperativity of Hydrogen Bonding in Amyloids

Last year,  I wrote a post on polyglutamine (polyQ) aggregation. **

** PolyQ tract is a series of consecutive glutamine residues present in a protein. It is associated with trinucleotide repeat disorder (a genetic disorder), such as Huntington’s disease (HD). In the case of  HD, people with more than 36 repeats of Q/glutamine, have a mutant form of the Huntingtin protein (mHt). We don’t fully understand the nature or behavior of this mutant form, however, mHt is known to form protein aggregates, rather than folding into functional forms. These aggregates accumulate with time and eventually interfere with cell function and intercellular communication.

Here is the post: What makes polyglutamine aggregates so toxic?

Quick summary: polyglutamine stretches aggregate due to (a) water being a poor solvent for polyQ and (b) the ability of both the backbone and side chains to form hydrogen bonds, thus contributing energetically to the stability of the aggregate.

Writing that post made me think about the relevance of hydrogen bonding in biological systems. So, this post is about the influence of H-bonding on the energetics of protein folding.



Role of Hydrogen Bonds (H-Bonds):
Hydrogen bonding is one of the most crucial inter and intra molecular interaction in biological systems. Whether it is interaction with solvent or protein folding, energetics is largely driven by hydrogen bonds. The directional nature of these interactions also gives rise to a multitude of spectroscopic properties. This paper published in 2000, explored the cooperative nature of hydrogen bonds in peptide systems, thus suggesting that the strength of H-bond should increase (asymptotically) with the extent of H-bond network. This does seems a little intuitive, especially if you are considering the folding of a helix, where the barrier is usually the formation of the first few turns.

Role of H-bonds in amyloids:

  • Formation of protein fibrils (stacks of beta sheets) from monomers is a characteristic of a big class of diseases known as protein aggregation diseases [more on protein aggregation  diseases].  Structures of some amyloid fibers have been identified [1], and the process of amyloid formation has been extensively studies [2-4].
  • One of the more popular theory is the formation of a nucleus seed for amyloid fibril formation.

Figure 1: cartoon representation of the aggregation process from [5]

 

  • There have been studies in the last few years, suggesting that the typical nucleus size is 3 to 4 peptides. [6]
  • As this is the minimal size that would not dissociate quickly due to slower diffusion.
  • One of the more important finding is the role of H-bonds and its cooperativity in this nucleus size of 3-4 peptides.
  • A 2006 study, explored this cooperative H-bond effect in a prion protein, using classical electrostatics and quantum DFT calculations.[7]
  • They were able to show that the strength or contribution of H-bonds between peptides increases nonlinearly up to 4 peptides, and then levels off. Thus suggesting the cooperative nature of H-bonds within β sheets of a fibril. 


 

Figure 2: (a) one layer of two peptides, (b) 3 peptides stacked one below the other, (c) energy per monomer in a fibril (d) binding energy of a layer to a preexisting fibril. [7]

 

  • From figure 2d, you can clearly see the leveling off of energy beyond a fiber length of four monomers.
  • This effect has also been validated in a polyQ system, where the cooperative effect is shown to have an effect of the geometry of the aggregate [8]


To summarize:

  • Hydrogen bonding, a partially covalent interaction plays a very significant role in determining energetics of protein folding and aggregation.
  • The directional nature of the interaction makes modeling the hydrogen bonding energy landscape computationally challenging.
  • Empirical molecular mechanics (MM) force fields have much less accuracy and QM electronic structure calculations cannot be adopted to biological systems.
  • With the advent of polarizable MM forcefields, there might be hope for  more consistency with both electronic structure calculations and other experimental data.

 

References:
[1] Structures for amyloid fibrils.
[2]On the nucleation and growth of amyloid beta-protein fibrils: detec…
[3] Simulations as analytical tools to understand protein aggregation a…
[4]  Interpreting the aggregation kinetics of amyloid peptides.
[5]Page on cell.com
[6]Page on nih.gov
[7]Page on nih.gov
[8]Hydrogen Bonding Cooperativity in polyQ β-Sheets from First Principle Calculations

Role of Water in Protein Binding Interfaces

We could never expect to understand the chemistry of proteins completely without understanding aqueous solutions and the structure of water. – Walter Kauzmann   [1]
One of the goals in computational biochemistry is to accurately model molecular interactions to make better predictions for experimentalists to test. Since biological reactions occur in an aqueous solvent, understanding the role of water (and environment in general) has been central to progress in the field. Furthermore, water is intimately involved in both thermal and cold denaturation of proteins [see: Sai Janani Ganesan’s answer to How does cold denaturation of proteins happen? ] and hence a key player in understanding protein stability.
However, interactions between a protein molecule and its solvent is highly nuanced. The properties and fluctuations of hydration water molecules proximal to the protein surface (or interfacial waters) is vastly different from the bulk, and is known to affect protein motion [2] .  The relationship between this interfacial region and protein structure, dynamics and function is still very much an open question.
Understanding the role of water will be particularly valuable in binding studies  (and ergo drug design) due to the desolvation effect (removal of interacting water) that occurs when two protein molecules (or a protein and a ligand molecule) interact. Which consequently affects hydrogen bond strengths, salt bridges, electrostatic interactions, hydrophobic effect etc.
Figure 1: Binding of ligand changes hydrogen bond strengths (weak to strong) in a protein.
A recent study from Shoichet’s group demonstrates how water molecules in and around a cationic ligand binding site do not cause generic effects on binding [3] . The study compares electrostatic driven ligand binding to a buried cavity of a mutant Cytochrome c Peroxidase (CcP) protein, with an open cavity variant created by loop deletion. Ligands that are known to bind to the buried cavity were tested on the open cavity.  Ligands that (a) maintain ionic interactions with the site and form (b) a water interaction network were seen to have increased affinity to the open cavity.
Figure 2: Crystallographic pose for the ligand  3-fluorocatechol  in buried (gray) and open(pink) cavity. Ordered water in red. Ligand seen interacting favorably with water and protein.  [4]
While, ligands that could not favorably interact with both the waters and the site, changed orientation to preferentially interact with water. **
Figure 3: Crystallographic pose for the ligand 2-amino-5-methylthiazole  in buried (gray) and open(pink) cavity. Ordered water in red. Ligand  and protein seen interacting favorably with water.  [5]

Thus the role of water is complex and difficult to predict, and to adequately quantify these effects, structural, dynamical and thermodynamic information is necessary.  Structural effects are made more complex by the diversity in water-mediated interactions  [6] . In general, removing water from a binding site has  a thermodynamically favorable effect, due to an entropic gain when interfacial water is released to bulk solvent.  However, water mediated interaction can also lead to enthalpic gains resulting from new H-bonds. It is this fine balance that makes these interactions difficult to estimate, and more advances in both structural and thermodynamic characterization of such interactions is needed for improvements in drug design.

** As an MD-person, I think it might help to use MD before docking and after X-ray crystallography to make sure there are no artifacts in water interactions.

Footnotes:

[1] Page on nih.gov

[2] Slaving: Solvent fluctuations dominate protein dynamics and functions

[3] Roles for Ordered and Bulk Solvent in Ligand Recognition and Docking in Two Related Cavities

[4] Roles for Ordered and Bulk Solvent in Ligand Recognition and Docking in Two Related Cavities

[5] Roles for Ordered and Bulk Solvent in Ligand Recognition and Docking in Two Related Cavities

[6] Page on cell.com

Why We Need To Study PTMs

I heard Philip Selenko give a great talk a couple of years back  [1]  and have since been (very slowly) getting acquainted with the literature on post-translational modifications (PTMs) in proteins.  Last year I wrote about the role of PTMs in protein folding and how we are now starting to look at disordered regions in proteins differently [ Sai Janani Ganesan’s answer to How do post-translational modifications affect protein folding? ]. PTMs also occur in highly structured regions of the protein. There are over 400 different types of PTMs  [2] , each with the potential to drastically change the conformational space of the protein and hence its function. The high diversity of  PTMs and their reversible nature make them a crucial part of understanding protein function, signaling pathways, allostery, binding and even protein energy landscapes.

Among the hundreds of different types of PTMs, phosphorylation (On serine, threonine and tyrosine) is one of the most well studied (mass spectrometry(MS)-based proteomics is now a pretty large field  [3] ).  Although a complete list of phosphosites is not yet available, the central question remains how to link the known PTM sites to conformational changes and therefore function.  Conservation of PTM is one  way to identify functionally relevant sites (this is not to say all functionally relevant sites are conserved) and hence understand protein regulation and their role in protein interaction network.  For example,  kinases have preferences for certain specific residues near the target phosphorylated site, and identifying the conservation of such sequences can be used to predict regulated sites  [4] .


A recent article from the Sali lab [5] touches on some aspects of correlating (conserved) PTMs with function. The study (s) uses  MS to identify phosphorylation sites in Xenopus laevis, (b) compares the obtained data with information from 13 other species to identify conserved sites, (c) uses predictive analysis to estimate conserved kinase-protein interactions for a set of cell-cycle kinases across species, and correlate degree of conservation with known kinase-protein regulatory interactions  [6] . They also model phosphosites to gain structural insights.
Some of their findings are very cool, and seem almost intuitive as you read along:
  • Only 39.8% of phosphosites were found to be conserved in one or more species. (The data on identified phosphosites is largely incomplete and hence must be kept in mind while interpreting all data related PTM studies. )
  • The fraction of sites with known function increased with the level of conservation across species, thus suggesting that conserved sites are more likely to have function.
  • For example, a phosphosite in the activation loop of GSK3B is one of the more conserved sites across species. Similarly, the conserved site in NDP Kinase A is located near the active site.
Figure 1: Example comparative models with highly conserved phosphorylation sites. The phosphorylation site is highlighted in red. For the NDP kinase A, the structure represents the homo-oligomeric complex. One of the subunits is indicated in blue, with the phosphosite position in red and the substrate in the ball-and-stick representation. [from footnote 6]
  • About 20% of the phosphosites identified appeared to be less solvent exposed, although intuitively, adding a phosphate should make the protein more exposed. The authors suggest that conformational flexibility might play a role, as it is well known that PTMs can change function by altering the conformational space of the protein. If that is the case, then we can use structural information to identify PTMs that can regulate protein conformation. However, if conformational flexibility is playing a role, these regions could also be poorly modeled. As an MD-person, I really think it is a good idea to integrate these structural results with multiple MD studies to get a more complete understanding.
Figure 2: An explanation for the sites that appear to be less solvent exposed. [from footnote 6]

We still need a whole lot of experimental data before we draw any major conclusions. Considering the fact that PTMs play such a major role in our understanding of molecular processes, I think more work should be done on correlating identified PTM sites (obtained under distinct conditions) to conformational changes and function.
Footnotes:

[1] Welcome to the Selenko Lab

[2] The Universal Protein Resource (UniProt) in 2010

[3] Status of Large-scale Analysis of Post-translational Modifications by Mass Spectrometry

[4] Deciphering a global network of functionally associated post-translational modifications

[5] Andrej Sali Lab

[6] Prediction of Functionally Important Phospho-Regulatory Events in  Xenopus laevis  Oocytes