Analysis of protein structures
Outline
❑ Residue solvent accessibility
❑ Protein solubility
❑ Molecular interactions
❑ Functional sites
▪ Binding sites
▪ Transport pathways
Analysis of protein structures 2
Residue solvent accessibility
❑ Solvent accessible surface area
Residue solvent accessibility 3
What is it?
Why do we care?
Residue solvent accessibility
❑ Solvent accessible surface area (ASA, SASA or SAS, in Å2)
→ It quantifies the extent to which a residue in a protein
structure is accessible to the solvent
❑ Typically calculated by rolling a spherical probe of a
particular radius over a protein surface and summing the
area that can be accessed by this probe on each residue
Residue solvent accessibility 4
➔
Residue solvent accessibility
❑ Solvent accessible surface area (ASA, SASA or SAS, in Å2)
❑ Solvent excluded surface (SES) – also known as molecular
surface, or Connolly surface area
Water radius  1.4 Å
5Residue solvent accessibility
VdW
VdW = Van der Waals radius
Residue solvent accessibility
❑ Solvent accessible surface area (ASA, SASA or SAS, in Å2)
❑ Solvent excluded surface (SES) – also known as molecular
surface, or Connolly surface area – usually represented in
“surface” visualization
6Residue solvent accessibility
SASA SES

Residue solvent accessibility
❑ Relative accessible surface area (rASA)
▪ Ratio of the actual accessible area of a given residue
rASA = ASA / ASAMAX
▪ Enables comparison of accessibility of different amino acids (e.g.,
long extended vs. spherical amino acids)
❑ Simplified two state description
▪ Buried vs. exposed residues
▪ Threshold for differentiating surface residues vs. buried is not well
defined (usually rASA = 15–25 %)
▪ rASA < threshold => buried
rASA ≥ threshold => exposed
7Residue solvent accessibility
Residue solvent accessibility – programs
❑ POLYVIEW-2D (PDB) / SABLE (sequence)
▪ https://polyview.cchmc.org/ / https://sable.cchmc.org/
▪ Visualization tool for structural and functional annotations of
proteins, including solvent accessibility
▪ Residue SASA calculated by DSSP and transformed to rASA
9Residue solvent accessibility
Protein solubility
❑ Definition: concentration of protein in saturated solution that
is in equilibrium with solid phase
❑ For proteins expressed in the lab: multiple factors
❑ Hydrophilic/hydrophobic balance of the solvent-exposed residues
❑ Aggregation-prone regions (APRs) – mainly hydrophobic residues
prone to form beta-structures
❑ Protein expressibility in the cells
Protein solubility 10
Cross-beta spines of amyloid fibrils
Protein solubility
❑ SoluProt
▪ https://loschmidt.chemi.muni.cz/soluprot/
▪ Soluble expression of protein sequences in E.coli
▪ Based on machine learning
Input Output
11Protein solubility
Protein solubility
❑ Aggrescan3D
▪ http://biocomp.chem.uw.edu.pl/A3D2/
▪ Predicts the aggregation propensities by identifying APRs
▪ Can introduce mutations and predict the impact on stability and
aggregation-propensity
▪ Can account for protein flexibility (“dynamic mode”)
Mutations
12Protein solubility
Protein solubility
❑ AggreProt
▪ https://loschmidt.chemi.muni.cz/aggreprot/
▪ Identifies APRs in sequence
▪ ML-based tool trained on (non)amyloidogenic hexapeptides
▪ Structure information used to define ASA to discard buried regions
Protein solubility
Molecular interactions
❑ Intra-molecular – within the same protein structure
❑ Inter-molecular – between different proteins in assemblies
❑ Essential to understand the molecular basis for function
and stability of proteins and their complexes
14Molecular interactions
Remember?...
Types of interactions
❑ Charge-charge (ionic) interactions
▪ Present in charged residues; ex. salt bridges
❑ Hydrogen bonds (H-bonds)
▪ Donor and acceptor atoms sharing a hydrogen atom
❑ Aromatic (π-π) interactions
▪ Attractive interaction between aromatic rings
❑ Van der Waals (vdW) interactions
▪ Between any two atoms; more important for non-polar residues
❑ Hydrophobic interactions
▪ Entropic origin; important for non-polar/hydrophobic residues
Molecular interactions 15
Types of interactions
❑ Disulfide bonds (cysteine bridges)
❑ Cation-π interactions
▪ Electrostatic interaction of a positively charged residue (Lys or Arg)
with an aromatic residue (Phe, Trp, or Tyr)
Lys
Trp
16Molecular interactions
Aromatic ring
Cation +
2 Cys:
Polar interactions
❑ Arginine interactions
❑ Cation-π: positively charged Arg interacts with aromatic rings
❑ Arginine-arginine stacking: two Arg form parallel “aromatic” stacking
Arg:
Guanidinium
group:  charge
Molecular interactions
Molecular interactions – how to identify?
❑ Criteria for recognizing various types of interactions
▪ Atom types/functional group
▪ Geometric rules (distances, angles)
▪ Energetics (physicochemical rules)
▪ Contact surface area between atoms
18Molecular interactions
If SASATotal < SASAA + SASAB
 Interaction
Molecular interactions – programs
❑ CMView
❑ https://www.bioinformatics.org/cmview/
▪ Represents residue-residue contacts within a protein or between
proteins in a complex in the form of a contact map
▪ 3D visualization using PyMol
19Molecular interactions
Molecular interactions – programs
❑ ProteinTools - A Toolkit to Analyze Protein Structures
▪ https://proteintools.uni-bayreuth.de/
▪ Identifies various types of interactions: hydrophobic clusters,
electrostatic interactions (salt bridges and charge segregation),
hydrogen bond networks, contact maps
Molecular interactions
Molecular interactions – programs
❑ ProteinTools - A Toolkit to Analyze Protein Structures
Molecular interactions
Molecular interactions – programs
❑ ESBRI (Evaluating the Salt BRIdges in Proteins )
▪ http://bioinformatica.isa.cnr.it/ESBRI/introduction.html
▪ Analysis of salt bridges interactions (ionic interaction + H-bond)
▪ Checks if at least one Asp or Glu side-chain carboxyl oxygen atom
(O ) and one side-chain nitrogen atom of Arg, Lys or His (NH ) are
within a distance ≤ 4.0 Å
24Molecular interactions
Functional sites
Functional sites 25
Examples?
Why are they
important?
Functional sites
❑ Binding sites
▪ Binding sites for small molecules
▪ Binding sites for macromolecules
❑ Transport pathways
▪ Tunnels
▪ Channels
Functional sites 26
Binding sites
❑ Sites on the protein that provides the complementarity for
the bound molecule (ligand)
▪ Binding site – its function is molecular recognition
▪ Active/catalytic site– its function is to promote chemical catalysis
(break/formation of covalent bonds) – special case of the binding site
❑ Binding involves the formation of non-covalent interactions
between the protein and the bound molecule
❑ Bound molecule – small molecule or macromolecule
❑ Binding is usually very specific – complementarity in shape
and charge distribution between the site and bound molecule
Functional sites → binding sites 27
Binding sites
Complementarity in shape and charge distribution
between the active site and substrate
28Functional sites → binding sites
Binding sites for small molecules
❑ Usually: internal cavities, surface pockets or clefts
▪ Concave regions
▪ Provide microenvironment different from that of the bulk solvent
(e.g., many residues with negative charge → very strong
electrostatic field enabling binding of highly charged ligands)
▪ Often identifiable by a simple examination of the protein structure
❑ Highly conserved by evolution
❑ Low desolvation energy
❑ Characteristic physicochemical properties
Functional sites → binding sites → binding sites for small molecules 29
Binding sites for small molecules
Active site
pocket on the
protein surface
Active site
cavity buried
inside the
protein core
Functional sites → binding sites → binding sites for small molecules 30
▪ residues influencing binding of ligands, transition-state
stabilization or product release
LigandActive site pocket
Binding sites for small molecules
31Functional sites → binding sites → binding sites for small molecules
❑ Can be very ligand-specific
▪ residues influencing binding of ligands, transition-state
stabilization or product release
Residue to be mutated
Binding sites for small molecules
32Functional sites → binding sites → binding sites for small molecules
❑ Can be very ligand-specific
▪ residues influencing binding of ligands, transition-state
stabilization or product release
Binding sites for small molecules
Before mutation
33Functional sites → binding sites → binding sites for small molecules
❑ Can be very ligand-specific
▪ residues influencing binding of ligands, transition-state
stabilization or product release
Binding sites for small molecules
34Functional sites → binding sites → binding sites for small molecules
❑ Can be very ligand-specific
Before mutation
▪ residues influencing binding of ligands, transition-state
stabilization or product release
After mutation
Binding sites for small molecules
35Functional sites → binding sites → binding sites for small molecules
❑ Can be very ligand-specific
▪ residues influencing binding of ligands, transition-state
stabilization or product release
No longer a good fit!
Binding sites for small molecules
36Functional sites → binding sites → binding sites for small molecules
❑ Can be very ligand-specific
After mutation
Binding sites for small molecules
❑ Approaches to identify binding sites:
❑ Evolutionary conservation
❑ Physical detection of “pockets”
▪ Geometry based methods
▪ Energy based methods
❑ Knowledge-based
▪ Machine learning-based methods
▪ Template-based methods
▪ Microenvironment-based methods
37Functional sites → binding sites → binding sites for small molecules
Evolutionary conservation
❑ Residues important for protein function or stability tend to
be highly conserved over evolution
❑ Residue conservation in a set of related proteins can be
derived from a multiple sequence alignment (MSA)
❑ Mapping of conservation on structure can reveal patches of
conserved surface residues – potential binding sites
❑ Protein interior usually more conserved than surface – not
suitable for prediction of buried cavities
❑ Not very specific – better to combine with other features
38Functional sites → binding sites → binding sites for small molecules
Evolutionary conservation
Map of evolutionary data
on the structure
Phylogenetic analysis
Conservation scoring
39Functional sites → binding sites → binding sites for small molecules
Evolutionary conservation
❑ ConSurf
▪ http://consurf.tau.ac.il/
▪ Estimates the level of evolutionary conservation of individual
positions in protein and maps this information onto its 3D structure
▪ Conservation score is derived based on the site-specific evolutionary
rates calculated for each position by Rate4Site software
▪ ConSurfDB – pre-calculated conservation scores for all structures in
wwPDB
40Functional sites → binding sites → binding sites for small molecules
Evolutionary conservation
❑ ConSurf
41Functional sites → binding sites → binding sites for small molecules
Physical detection of “pockets”
❑ Analyze the protein surface for pockets (clefts, cavities)
❑ Geometry-based methods
▪ Define favorable cleft regions based on steric assessments
❑ Energy-based methods
▪ Define favorable cleft regions based on energetic evaluations
42Functional sites → binding sites → binding sites for small molecules
Geometry-based methods
❑ Computed Atlas of Surface Topography of proteins (CASTp)
▪ http://sts.bioe.uic.edu/castp
▪ Uses computational geometry methods including Delaunay
triangulation, alpha shape and discrete flow theory
▪ Measures the volume and surface area of each pocket and cavity
using the ASA model and molecular surface (Connolly) model
Delaunay
triangulation
Alpha shape
Voronoi
diagram
43Functional sites → binding sites → binding sites for small molecules
Geometry-based methods
❑ Computed Atlas of Surface Topography of proteins (CASTp)
▪ http://sts.bioe.uic.edu/castp
44Functional sites → binding sites → binding sites for small molecules
Energy-based methods
❑ Pockets are defined by energetic criteria
❑ Evaluate the interaction energy between the protein and a
molecular fragment – probe (e.g., a methyl, hydroxyl, amine,
etc.) to locate energetically favorable binding sites
❑ Can be combined with other methods to assess the
ligandability (ability of a cavity to bind ligands)
46Functional sites → binding sites → binding sites for small molecules
Note: druggability is referred to the likelihood of
finding orally bioavailable small molecules that
bind to a particular target in a disease-modifying
way.
Ligandability is a requirement but not sufficient
condition for druggability.
Energy-based methods
❑ Cavity Plus
▪ http://www.pkumdl.cn/cavityplus
▪ Applies Cavity program to detect the potential binding sites and
rank them with ligandability and druggability scores
▪ Extracts pharmacophore features within the cavities
47Functional sites → binding sites → binding sites for small molecules
Energy-based methods
❑ Cavity Plus
48Functional sites → binding sites → binding sites for small molecules
Machine learning-based method
❑ P2rank
▪ https://prankweb.cz/
▪ Volume calculation
▪ Molecular docking using AutoDock Vina (future…?)
Functional sites → binding sites → binding sites for small molecules
Knowledge-based: binding site similarity
❑ Prediction of binding sites is based on the similarity with
other (known) binding sites
❑ Template-based methods
▪ Binding sites are represented by 3D templates
▪ Based on similarity between homologous proteins
❑ Microenvironment-based methods
▪ Based on description of local environment,
such as type of residues, their distances, solvent
accessibility and physicochemical properties
50Functional sites → binding sites → binding sites for small molecules
Template-based methods
❑ Definition and construction of 3D templates of features
▪ Local structural motifs, patterns and descriptors that characterize the
binding sites (e.g., functional groups, shape, solvent accessibility, etc.)
▪ Capture the essence of the binding sites in the protein
▪ Usually apply constraints on atom types and occasionally sequential
relationships
❑ Search a database for structures using template as a query
▪ Identification of structures with a given binding site
❑ Compare the query structure against a 3D template database
▪ Identification of potential binding sites in the query structure
51Functional sites → binding sites → binding sites for small molecules
Template-based methods
❑ PINTS (Patterns In Non-homologous Tertiary Structures)
▪ http://www.russelllab.org/cgi-bin/tools/pints.pl
▪ To compare a protein structure against a database of 3D patterns
(templates), as well as 3D templates against a database of protein
structures
▪ Additionally allows comparison of two structures
▪ The 3D template database includes ligand-binding sites and SITE
annotations from PDB files
Functional sites → binding sites → binding sites for small molecules 52
Template-based methods
❑ ProFunc (Prediction of protein function from 3D structure)
▪ http://www.ebi.ac.uk/thornton-srv/databases/profunc/
▪ Aims to identify the most likely function of a protein from its 3D
structure
▪ Uses several methods, including fold matching, residue
conservation, surface cleft analysis, and functional 3D templates
(templates for enzyme active sites, ligand-binding templates, DNAbinding
templates, reverse template comparison vs. structures in
wwPDB)
Functional sites → binding sites → binding sites for small molecules 53
Template-based methods
❑ Mechanism and Catalytic Site Atlas
▪ https://www.ebi.ac.uk/thornton-srv/m-csa/
▪ Database that provides information about the active sites, catalytic
residues and reaction mechanisms in enzymes with experimentally
determined 3D structure
▪ Defines catalytic residues as the residues directly involved in some
aspect of the enzymatic reaction
▪ Provides 3D templates for catalytic sites in the database
54Functional sites → binding sites → binding sites for small molecules
Binding sites for macromolecules
Functional sites → binding sites → binding sites for macromolecules 55
What’s different?
Binding sites for macromolecules
❑ Typically protruding loops, large surface clefts but also flat
binding sites – flatter than binding sites for small molecules
▪ Recognition of a macromolecule involves interactions over a large
continuous surface area or several discrete binding regions
▪ Difficult to identify by a simple examination of the protein structure
❑ High evolutionary conservation
❑ Low desolvation energy
❑ Characteristic physicochemical properties
❑ DNA binding sites have characteristic motifs and positive
charged electrostatic patches
Functional sites → binding sites → binding sites for macromolecules 56
Binding sites for macromolecules
protein-protein
complex
protein-DNA
complex
57Functional sites → binding sites → binding sites for macromolecules
Binding sites for macromolecules
❑ Approaches to identify binding sites
▪ Evolutionary conservation
▪ Knowledge-based
❑ Meta-servers (tools that combine several methods)
58Functional sites → binding sites → binding sites for macromolecules
Evolutionary conservation methods
❑ Same principles as for binding sites of small molecules
(see above)
❑ WHISCY
▪ https://wenmr.science.uu.nl/whiscy/
▪ Predicts protein-protein interface using conservation and structural
information (interface propensities for each residue at the surface
are used to adjust the score)
59Functional sites → binding sites → binding sites for macromolecules
Knowledge-based methods
❑ Combine multiple interface features
▪ Conservation
▪ Residue propensity for being at protein-protein interfaces
(hydrophobic, aromatic, and charged residues are more likely)
▪ Physicochemical properties
▪ Structural properties
❑ Use known binding sites for parameterization or training →
empirical scoring functions and machine learning methods
60Functional sites → binding sites → binding sites for macromolecules
Knowledge-based methods
❑ CONS-PPISP (Consensus Protein-Protein Interaction Site Predictor)
▪ http://pipe.scs.fsu.edu/ppisp.html
▪ Utilizes machine learning to predict protein binding sites
▪ Trained on position-specific sequence profiles and solvent
accessibilities of each residue and its spatial neighbors
❑ Patch Finder Plus
▪ http://pfp.technion.ac.il/
▪ Utilizes machine learning primarily to find DNA binding regions
▪ Identifies the largest positive electrostatic patch on a protein surface
– combination of residue frequency, composition and conservation,
surface concavity, accessible area and H-bond potential
61Functional sites → binding sites → binding sites for macromolecules
Meta-servers
❑ Combine multiple methods to improve prediction accuracy
❑ META-PPISP (Protein Protein Interaction Site Predictor)
▪ http://pipe.scs.fsu.edu/meta-ppisp.html
▪ Combines cons-PPISP, ProMate and PINUP
❑ PI2PE (Protein Interface/Interior Prediction Engine)
▪ http://pipe.scs.fsu.edu/
▪ Pipeline to use five different predictors including cons-PPISP, metaPPISP
and DISPLAR
62Functional sites → binding sites → binding sites for macromolecules
Transport pathways
Functional sites → transport pathways 63
What are these?
Examples?
Transport pathways
❑ Mediate transport of ions and small molecules in proteins –
an essential role in functioning of large variety of proteins
▪ Channels/pores – transport of substances across membranes
▪ Tunnels – exchange of ligands between buried active/binding site
cavities and the bulk solvent
▪ Intramolecular tunnels – transport of reaction intermediates
between two distinct active sites in bifunctional enzymes
❑ The permeability to different substances depends on their
size (radii), shape (length and curvature), amino acid
composition (physicochemical properties) and dynamics
Functional sites → transport pathways 64
▪ Bottleneck – the narrowest part of the
tunnel/channel; it has critical importance for
the selectivity
Transport pathways & voids
65Functional sites → transport pathways
Channel
Cavity
Pocket/
cleft/groove
tunnel
Bottleneck
Protein
channel
Enzyme
tunnels
Active site
pocket
Ligand
Transport pathways
Tunnel
66Functional sites → transport pathways
❑ Dependence on the residues
▪ residues influencing binding of ligands, transition-state
stabilization or product release
Transport pathways
Tunnel
Active site
pocket
Ligand
Residue on the bottleneck
(Valine)
67Functional sites → transport pathways
❑ Dependence on the residues
▪ residues influencing binding of ligands, transition-state
stabilization or product release
Transport pathways
Leucine
Wider tunnel
68Functional sites → transport pathways
❑ Dependence on the residues
▪ residues influencing binding of ligands, transition-state
stabilization or product release
Transport pathways
Closed tunnel
Tryptophan
69Functional sites → transport pathways
❑ Dependence on the residues
▪ residues influencing binding of ligands, transition-state
stabilization or product release
Transport pathways
70Functional sites → transport pathways
❑ Dependence on protein dynamics
Time (ns)
Bottleneck
radius(Å)
Prediction of transport pathways
❑ Identification of overall voids in proteins
❑ Identification of tunnels
❑ Identification of channels
71Functional sites → transport pathways
Identification of overall voids
❑ Methods that aim to accurately represent all types of voids
in a protein structure, including channels, tunnels, surface
clefts, pockets as well as internal cavities
❑ Usually provide very limited information on tunnel and
channel characteristics – the identified voids have to be
separated from each other
❑ Geometry-based methods for pocket detection
▪ HOLLOW – http://hollow.sourceforge.net/
▪ 3V – http://3vee.molmovdb.org/
▪ fPocket, LIGSITEcsc
, PASS, CASTp, SURFNET, POCASA …
72Functional sites → transport pathways
Identification of tunnels
❑ Methods that calculate tunnels connecting occluded cavities
with the surrounding bulk solvent
❑ Identify the pathways from a cavity to the protein surface
❑ Voronoi diagrams described by the skeleton of voids
between atoms to find all theoretically possible pathways
connecting the starting point with the bulk solvent
❑ Diagrams of optimal pathways using Dijkstra’s algorithm,
based on criteria defined by a cost function
❑ The probe size defines the lowest radius threshold
❑ Tunnel geometry is approximated by a sequence of spheres
73Functional sites → transport pathways
Identification of tunnels
Voronoi diagram Common limitation: the tools identify
two spherical tunnels instead of one
asymmetric tunnel
74Functional sites → transport pathways
Probe size: the minimum radius specified for the
tunnel search
Allowed pathway according to the selected probe
Disallowed pathways
Atom
Tunnel
mouth
Tunnel
origin
Identification of tunnels - programs
❑ CAVER 3.0
▪ http://caver.cz/
▪ Command-line stand-alone and PyMOL plugin
▪ GUI with CAVER Analyst 2
▪ For static structures and dynamic ensembles
❑ CAVER Web
▪ http://loschmidt.chemi.muni.cz/caverweb/
▪ Interactive guide-through web server
▪ Optimized protocol for detection of biologically relevant tunnels
❑ MOLE 2.0
▪ http://mole.upol.cz/
75Functional sites → transport pathways
Identification of tunnels - programs
76Functional sites → transport pathways
Identification of tunnels - programs
❑ CAVER Analyst
Functional sites → transport pathways
Identification of channels
❑ Methods that calculate channels (or pores) penetrating
throughout the proteins
❑ Not suitable to identify tunnels leading from occluded
cavities
❑ Usually analyze just one channel per structure
❑ Usually need information about approximate position and
direction of the channel (channel axis) – user-provided or
automatically identified
78Functional sites → transport pathways
Identification of channels - programs
❑ POREWALKER
▪ http://www.ebi.ac.uk/thornton-srv/software/PoreWalker/
▪ Identifies channel axis by heuristic iterative approach (based on the
axes of transmembrane secondary structures)
▪ Protein is divided into equally-spaced slices perpendicular to the
axis; the largest spheres fitting the channel are identified
79Functional sites → transport pathways
Identification of channels - programs
❑ POREWALKER
80Functional sites → transport pathways
References
❑ Gu, J. & Bourne, P. E. (2009). Structural Bioinformatics, 2nd Edition, Wiley-Blackwell, Hoboken, p. 1067.
❑ Laurie, A. T. & Jackson, R. (2006). Methods for the prediction of protein-ligand binding sites for structurebased
drug design and virtual ligand screening. Current Protein and Peptide Science 7: 395-406.
❑ Campbell, S. J. et al. (2003). Ligand binding: functional site location, similarity and docking. Current
opinion in structural biology 13: 389-395.
❑ Xin, F. & Radivojac, P. (2011). Computational methods for identification of functional residues in protein
structures. Current protein and peptide science 12: 456-469.
❑ Leis, S. et al. (2010). In silico prediction of binding sites on proteins. Current medicinal chemistry 17:
1550-1562.
❑ Fernández‐Recio, J. (2011). Prediction of protein binding sites and hot spots. Computational molecular
science 6: 680-698.
❑ Tuncbag, N., et al. (2009). A survey of available tools and web servers for analysis of protein-protein
interactions and interfaces. Briefings in bioinformatics 10: 217-232.
❑ Brezovsky, J. et al. (2012). Software tools for identification, visualization and analysis of protein tunnels
and channels. Biotechnology advances. In press: doi:10.1016/j.biotechadv.2012.02.002
81References