Macromolecular complexes
and interactions
 Macromolecular complexes
 Structure of complexes
 Prediction of 3D structures of complexes
 Analysis of macromolecular complexes
Outline
2Macromolecular complexes and interactions
 Types of biologically relevant complexes
 Protein – small molecule 
 Protein – protein
 Protein – nucleic acids
 Nucleic acids – small molecule 
Biological relevance
3Macromolecular complexes
Macromolecular
complexes
Biological relevance
 Many proteins are formed by two or more polypeptide
chains (protomers) interacting with each other
 Protein-protein and protein-nucleic acid interactions have
central importance for virtually every process in a living cell
(molecular recognition)
 Regulation
 Transport
 Signal transduction
 Genetic activity (transcription, translation, replication, repair, ...)
 ...
Macromolecular complexes
 Oligomerization
 Native interactions between proteins in native conditions
 Aggregation
 Interactions between native proteins at extreme conditions
 Interactions between misfolded/partially folded proteins  disease
Protein-protein complexes
5Macromolecular complexes – protein-protein complexes
 Obligate complexes
 Protomers (individual polypeptides) do not function as independent
structures, only when associated
 Examples: GABA receptors, ATP synthase,
many ion channels, ribosome, etc.
 Non-obligate complexes
 Protomers can exist and be functional as independent structures
 Examples: hemoglobin, beta-2 adrenergic receptor, insulin
receptor, etc.
Protein-protein complexes
6Macromolecular complexes – protein-protein complexes
GABAB receptor
 Oligomerization is common
 More than 35 % of proteins in a cell are oligomers
 Tetramer is the average oligomeric state of
proteins in E. coli
 Homo-oligomers – the most common
 Some proteins exists solely in the oligomeric state
 Oligomers are often symmetric
 Oligomerization interfaces are complementary
 Oligomerization is favored by evolution
Protein oligomerization
7Macromolecular complexes – protein-protein complexes
 Why do proteins form oligomers?
Advantages of oligomerization
8Macromolecular complexes – protein-protein complexes
 Morphological function
 More complex structures are often required for multiple functions
 Cooperative function
 Allostery
 Multivalent binding
 Enhanced stability
 Smaller surface area
 More interactions
 … (ex. Translation error control)
Advantages of oligomerization
9Macromolecular complexes – protein-protein complexes
 Characteristics of oligomeric interface
 Large surface area (> 1400 Å2)
 Tendency to circular and planar shape (not for obligates)
 Some residues protrude from the surface
 More non-polar residues (about 2/3) than in other parts of surface
 More polar residues (about 1/5) than in protein cores
 About 1 H-bond per 200 Å2
 Hot-spot residues
 Responsible for most of the oligomeric interactions
 More evolutionary conserved than other surface residues
 Frequently polar residues, located about the center of the interface
Oligomerization interface
10Macromolecular complexes – protein-protein complexes
 Protein-nucleic acid interactions
 Non-specific – electrostatic interactions with negative charge on
the backbone of nucleic acid -> Lys and Arg residues
 Specific – recognition of particular nucleotide sequences
 Major groove – B-DNA
 Minor groove – A-DNA or A-RNA
 Single strand RNA
 Typical interfaces/motifs
 DNA binding proteins
 RNA binding proteins
Protein-nucleic acids complexes
11Macromolecular complexes – protein-nucleic acids complexes
 DNA binding proteins
 Helix-turn-helix
 Zinc finger
Protein-nucleic acids complexes
12Macromolecular complexes – protein-nucleic acids complexes
 RNA binding proteins
 Recognition is often also governed by particular structures of RNA
 Many motifs employed
Protein-nucleic acids complexes
13
RNA recognition motif K-homology domain Pumilio repeat domain
Macromolecular complexes – protein-nucleic acids complexes
 Quaternary structure in PDB database
 Complex or crystallization artifact?
Structure of complexes
14Structure of complexes
 Asymmetric unit (ASU)
 Macromolecular structures from X-ray crystallography deposited to
PDB as a single asymmetric unit
 The smallest portion of a crystal structure to which symmetry
operations can be applied in order to generate the unit cell
 Unit cell (crystal unit)
 The basic unit of a crystal that, when repeated in three dimensions,
can generate the entire crystal
Quaternary structure in PDB database
15Structure of complexes – quaternary structure in PDB database
Quaternary structure in PDB database
16Structure of complexes – quaternary structure in PDB database
 Crystal contacts
 Intermolecular contacts solely due to protein crystallization
 Causes artifacts of crystallization
 Crystal packing - complicates identification of native quaternary
structure
Crystalline environment
17Structure of complexes – quaternary structure in PDB database
ASU
Crystal U
 Artifacts of crystallization
 Concerns about conformation of some surface regions
 Often loops or side chains are affected
 Can complicate the evaluation of the effects of mutations
Crystalline environment
18Structure of complexes – quaternary structure in PDB database
 Biological unit
 The functional form of a protein in nature
 Also called: functional unit, biological assembly, quaternary structure
 Can depend on the environment, post-translational modifications
of proteins and their mutations
Quaternary structure in PDB database
19Structure of complexes – quaternary structure in PDB database
Homotetramer
hemoglobin
 Biological unit can consist of:
 Multiple copies of the ASU
 One copy of the ASU
 A portion of the ASU
Biological versus asymmetric unit
20Structure of complexes – quaternary structure in PDB database
ASU Biol. U
 Large assemblies
 Viral capsid
 Filamentous bacteriophage PF1
Biological versus asymmetric unit
21Structure of complexes – quaternary structure in PDB database
ASU Biol. U
ASU Biol. U
 Problem
 Most proteins in the PDB have three or more crystal contacts that
sum up to 30% of the protein solvent accessible surface area
 How to recognize biologically relevant contacts from crystal one?
Complex or artifact?
22Structure of complexes – complex or artifact?
 Experimental knowledge of oligomeric state helps with
identifying of the structure of native complex
 Search literature
 Experimental methods
 Gel filtration, static or dynamic light scattering, analytical
ultracentrifugation, native electrophoresis, …
 How to get the structure of a biological unit?
 Author-specified assembly
 Databases
 Predictive tools
Complex or artifact?
23Structure of complexes – complex or artifact?
 REMARK 350 in headers of PDB file
 Contains symmetry operations to reconstruct biological unit, but…
 Verify author-proposed biological unit by other means
 Sometimes the specific oligomers were not known at the time
the ASU was published
 Some authors may have failed to specify the biological unit
even when it was known
 Rarely, the specified biological unit might be incorrect
 Employed by
 RCSB PDB and other tools
Author-specified assembly
24Structure of complexes – complex or artifact?
 RCSB PDB
 Generates a PDB file in which all protein chains are as separate
models  complicates visualization and analysis
Author-specified assembly
25Structure of complexes – complex or artifact?
 PyMOL
 Generate > Symmetry mates  to visualize nearest partners
 You can select some and combine them in a PDB file
Crystal lattice
26Structure of complexes – complex or artifact?
Prediction of 3D structure of complexes
 How can we predict macromolecular complexes?
Prediction of 3D structure of complexes 27
Prediction of 3D structure of complexes
 Homology-based methods
 Machine learning-based threading
 Macromolecular docking
Prediction of 3D structure of complexes 28
Homology based methods
 The model of a protein complex is built based on a similar
protein complex with a known 3D structure
 Assumes that the interaction information can be
extrapolated from one complex structure to close homologs
of interacting proteins
 Close homologs (≥ 40% sequence identity) almost always interact in
the same way (if they interact with the same partner)
 Sequence similarity is only rarely associated with a
similarity in interactions
 Limited applicability (low number of templates)
Prediction of 3D structure of complexes – homology based methods 29
Homology based methods
 HOMCOS (Homology Modeling of Complex Structure)
 https://homcos.pdbj.org/
 Predicts 3D structure of homodimers and heterodimers by homology
modeling
 Optionally, identifies potentially interacting proteins
 Steps:
1. BLAST search to identify homologous templates in the latest
representative dataset of heterodimer (homodimer) structures
2. Evaluation of the model validity by the combination of sequence
similarity and knowledge-based contact potential energy
3. Generation of a script for building full atomic model by MODELLER
Prediction of 3D structure of complexes – homology based methods 30
Homology based methods
Prediction of 3D structure of complexes – homology based methods 31
Homology based methods
Prediction of 3D structure of complexes – homology based methods 32
Machine learning-based
 AlphaFold-Multimer
 Predicts 3D structure of multimers; similar to AlphaFold
Prediction of 3D structure of complexes – homology based methods 33
Experimental
AlphaFold-multimerRMSD (Ca) = 0.81 Å
Macromolecular docking
 Prediction of the best bound state for given 3D structures of
two or more macromolecules
 Difficult task
 Large search space - many potential ways in which macromolecules
can interact
 Flexibility of the macromolecular surface and conformational
changes upon binding
 Can be facilitated by prior knowledge
 Ex: known binding site → significant restriction of the search space
 Distance constraints on some residues
Prediction of 3D structure of complexes – macromolecular docking 34
Macromolecular docking
 Macromolecule representation
 Search algorithm
 Scoring function
Prediction of 3D structure of complexes – macromolecular docking 35
Macromolecule representation
 Representation of the macromolecular surface (applicable
to both receptor and ligand)
 Geometrical descriptors of shape (set of spheres, surface normals,
vectors radiating from the center of the molecule,...)
 Discretization of space: grid representation
Prediction of 3D structure of complexes – macromolecular docking 36
Macromolecule representation
 Macromolecule flexibility
 Fully rigid approximation
 Soft docking – employs tolerant “soft” potential scoring functions to
simulate plasticity of otherwise rigid molecule
 Explicit side-chain flexibility – optimization of residues by rotating
part of their structure or rotation of whole side-chains using
predefined rotamer libraries
 Docking to molecular ensemble of protein structure – composed
from multiple crystal structures, from NMR structure determination
or from trajectory produced by MD simulation
Prediction of 3D structure of complexes – macromolecular docking 37
Macromolecule representation
 Macromolecule flexibility
 Rigid body docking – basic model that considers the two
macromolecules as two rigid solid bodies
 Semiflexible docking – one of the molecules is rigid, and one is
flexible (typically the smaller one)
 Flexible docking – both molecules are considered flexible
Prediction of 3D structure of complexes – macromolecular docking 38
Macromolecular docking - search
 Generally based on the idea of complementarity between
the interacting molecules (geometric, electrostatic or
hydrophobic contacts)
 The main problem is the dimension of the conformational
space to be explored:
 Rigid docking: 6D (hard)
 Flexible docking: 6D + Nfb (impossible!)
 Information on the rough location of the binding surface
(experimental or predicted) → reduction of the search space
Prediction of 3D structure of complexes – macromolecular docking 39
Macromolecular docking - search
 Exhaustive search
 Full search of the conformational space: try every possible relative
orientation of the two molecules
 Computationally very expensive – 6 degrees of freedom for rigid
molecules (translations + rotations)
 Grid approaches
Prediction of 3D structure of complexes – macromolecular docking 40
Macromolecular docking - search
 Stochastic methods
 Monte Carlo
 Genetic algorithms
 Brownian dynamics
 ...
Prediction of 3D structure of complexes – macromolecular docking 41
Macromolecular docking - scoring
 Scoring functions
 Evaluation of a large number of putative solutions generated by the
search algorithms
 Methods often use a two-stage ranking
1. Approximate and fast-to-compute function – used to eliminate very
unlikely solutions
2. More accurate function – used to select the best among the
remaining solutions
Prediction of 3D structure of complexes – macromolecular docking 42
Macromolecular docking - scoring
 Scoring functions
 Empirical
 Knowledge-based
 Force field-based
 Clustering-based – the presence of many similar solutions is taken as
an indication of correctness (all solutions are clustered, and the size
of each cluster is used as a scoring parameter)
Prediction of 3D structure of complexes – macromolecular docking 43
 Good scores – a combination of several parameters:
 Low free energy or pseudo-energy based on force field functions
 Large buried surface area
 Good geometric complementarity
 Many H-bonds
 Good charge complementarity
 Polar/polar contacts favored
 Polar/non-polar contacts are disfavored
 Many similar solutions (large clusters)
 ...
Prediction of 3D structure of complexes – macromolecular docking
Macromolecular docking - scoring
44
Macromolecular docking - programs
Prediction of 3D structure of complexes – macromolecular docking 45
Macromolecular docking - programs
 ClusPro 2.0
 http://cluspro.bu.edu/
 Performs a global soft rigid-body search using PIPER docking
program; employs knowledge-based potential
 The top 1,000 structures are retained and clustered to isolate highly
populated low-energy binding modes
 A special mode for prediction of molecular assemblies of
homo-oligomers
Prediction of 3D structure of complexes – macromolecular docking 46
Macromolecular docking - programs
 PatchDock
 http://bioinfo3d.cs.tau.ac.il/PatchDock/index.html
 Performs a geometry-based search for docking transformations that
yield good molecular shape complementarity (driven by local feature
matching rather than brute force searching of the 6D space):
1. The molecular surface is divided into concave, convex and flat patches
2. Complementary patches are matched → candidate transformations
3. Evaluation of each docking candidate by a scoring function considering
both geometric fit and atomic desolvation energy
4. Clustering of the candidate solutions to discard redundant solutions
 Results can be redirected to FireDock for refinement and re-scoring
Prediction of 3D structure of complexes – macromolecular docking 47
Macromolecular docking - programs
 PatchDock
Prediction of 3D structure of complexes – macromolecular docking 48
Macromolecular docking - programs
 FireDock
 http://bioinfo3d.cs.tau.ac.il/FireDock/index.html
 Refines and re-scores solutions produced by fast rigid-body docking
algorithms
 Optimizes the binding of each candidate by allowing flexibility in the
side-chains and adjustments of the relative orientation of the
molecules
 Scoring of the refined candidates is based on softened van der Waals
interactions, atomic contact energy, electrostatic, and additional
binding free energy estimations
Prediction of 3D structure of complexes – macromolecular docking 49
Analysis of macromolecular complexes
 Binding energy
 Macromolecular interface
 Interaction hot spots
Analysis of macromolecular complexes 50
Binding energy
 FastContact
 http://structure.pitt.edu/servers/fastcontact/
 Rapidly estimates the electrostatic and desolvation components of
the binding free energy between two proteins
 Additionally, evaluates the van der Waals interactions using
CHARMM and reports contribution of individual residues and pairs
of residues to the free energy → highlight the interaction hot spots
Analysis of macromolecular complexes – binding energy 51
Macromolecular interface
 The region where two protein chains or protein and nucleic
acid chain come into contact
 Can be identified by the analysis of the 3D structure of the
macromolecular complex
Analysis of macromolecular complexes – interface analysis 52
Interface analysis
 Provides information about basic features of macromolecular
complexes interactions (e.g., shape complementarity,
chemical complementarity,...)
 Provides information about interface residues
 Acquired information is useful for a wide range of applications
 Design of mutants for experimental verification of the interactions
 Development of drugs targeting macromolecular interactions
 Understanding the mechanism of the molecular recognition
 Computational prediction of interfaces and complex 3D structures
 ...
Analysis of macromolecular complexes – interface analysis 53
Interface analysis
 Most common approaches for the definition of interfaces:
 Methods based on the distance between interacting residues
 Methods based on the change in the solvent accessible surface area
(ASA) upon complex formation
 Computational geometry methods (using Voronoi diagrams)
 All three approaches provide very similar results
Analysis of macromolecular complexes – interface analysis 54
Interface analysis - databases
 PDBsum (Pictorial database of 3D structures in the Protein
Data Bank)
 http://www.ebi.ac.uk/pdbsum/
 Provides numerous structural analyses for all PDB structures and
AlphaFold DB (human proteins), including information about
protein-protein and protein-nucleic acid interfaces
 Protein-protein interactions – schematic diagrams of all proteinprotein
interfaces and corresponding residue-residue interactions
 Protein-nucleic acid interactions – schematic diagrams of proteinnucleic
acid interactions generated by NUCPLOT
Analysis of macromolecular complexes – interface analysis 55
Interface analysis - databases
 PDBsum
Analysis of macromolecular complexes – interface analysis 56
Interface analysis - databases
 PDBsum
Analysis of macromolecular complexes – interface analysis 57
Interface analysis - tools
 Analyze interface of a given macromolecular complex
 PISA (Protein Interfaces, Surfaces and Assemblies)
 MolSurfer
 Contact Map WebViewer
 PIC (Protein Interaction Calculator)
 …
Analysis of macromolecular complexes – interface analysis 58
Interface analysis - tools
 PISA (Protein Interfaces, Surfaces and Assemblies)
 www.pdbe.org/pisa
 An interactive tool for the exploration of macromolecular interfaces
(protein, DNA/RNA and ligands), prediction of probable quaternary
structures, database searches of structurally similar interfaces and
assemblies
 Overview and detailed characteristics of all interfaces found within
a given structure (including those generated by symmetry
operations)
 Provides interface area, ΔiG, potential hydrogen bonds and salt
bridges, interface residues and atoms, ...
Analysis of macromolecular complexes – interface analysis 59
Interface analysis - tools
 MolSurfer
 http://projects.villa-bosch.de/dbase/molsurfer/index.html
 Visualization of 2D projections of protein-protein and proteinnucleic
acid interfaces as maps showing a distribution of interface
properties (atomic and residue hydrophobicity, electrostatic
potential, surface-surface distances, atomic distances,...)
 2D maps are linked with the 3D view of a macromolecular complex
 Facilitates the study of intermolecular interaction properties and
steric complementarity between macromolecules
Analysis of macromolecular complexes – interface analysis 60
Interface analysis - tools
 MolSurfer
Analysis of macromolecular complexes – interface analysis 61
Interface analysis - tools
 Contact Map WebViewer
 http://cmweb.enzim.hu/
 Represents residue-residue contacts within a protein or between
proteins in a complex in the form of a contact map
 PIC (Protein Interaction Calculator)
 http://pic.mbu.iisc.ernet.in/
 Identifies various interactions within a protein
or between proteins in a complex
Analysis of macromolecular complexes – interface analysis 62
Interaction hotspots
 Hot spots: the residues contributing the most to the binding
free energy of the complex
 Knowledge of hot spots has important implications to:
 Understand the principles of protein interactions (an important step
to understand recognition and binding processes)
 Design of mutants for experimental verification of the interactions
 Development of drugs targeting macromolecular interactions
 ...
Analysis of macromolecular complexes – interaction hotspots 63
Interaction hotspots
 Hot spots are usually conserved and appear to be clustered
in tightly packed regions in the center of the interface
 Experimental identification by alanine scanning mutagenesis
 if a residue has a significant drop in binding affinity when
mutated to alanine it is labeled as a hot spot
 Experimental identification of hot spots is costly and
cumbersome → the computational predictions of hot spots
can help!
Analysis of macromolecular complexes – interaction hotspots 64
Prediction of hotspots - tools
 Most of the available methods are based on the 3D structure
of the complex
 Knowledge-based methods
 Combination of several physicochemical features
 Evolutionary conservation, ASA, residue propensity, structural
location, hydrophobicity,...)
 Energy-based methods
 Calculation of the change in the binding free energy (∆∆Gbind) of the
complex upon in silico modification of a given residue to alanine
Analysis of macromolecular complexes – interaction hotspots 65
Prediction of hotspots - tools
 Robetta
 http://old.robetta.org/alascansubmit.jsp
 Energy-based method
 Performs in silico alanine scanning mutagenesis of protein-protein or
protein-DNA interface residues
1. The side chain of each interface residue is mutated to methyl
2. All side chains within 5 Å radius sphere of the mutated residue are
repacked; the rest of the protein remains unchanged
3. For each mutant, ∆∆Gbind is calculated (residues with predicted
∆∆Gbind ≥ +1 kcal/mol = hot spot)
Analysis of macromolecular complexes – interaction hotspots 66
Prediction of hotspots - tools
 Robetta
Analysis of macromolecular complexes – interaction hotspots 67
Prediction of hotspots - tools
 KFC2 (Knowledge-based FADE and Contacts)
 https://mitchell-web.ornl.gov/KFC_Server/
 Knowledge-based method utilizing machine learning
 Predicts hot spots in protein-protein interfaces by recognizing
features of important binding contacts – solvent accessibility, residue
position within the interface, packing density, residue size, flexibility
and hydrophobicity of residues around the target residue
 Optionally, user can provide data to improve the prediction (ConSurf
conservation scores, Rosetta alanine scanning results or
experimental data)
68
Prediction of hotspots - tools
 KFC2 (Knowledge-based FADE and Contacts)
69
References I
 Liljas, A. et al. (2009). Textbook Of Structural Biology, World Scientific Publishing
Company, Singapore.
 Goodsell, D. S. & Olson, A. J. (2000) Structural symmetry and protein function. Annual
Review of Biophysics and Biomolecular Structure 29: 105-153.
 Demachenko, A. P. (2001). Recognition between flexible protein molecules: induced and
assisted folding. Journal of Molecular Recognition 14: 42-61.
 Ali, M. H. & Imperiali, B. (2005) Protein oligomerization: How and why. Bioorganic &
Medicinal Chemistry 13: 5013-5020.
 Jahn, T. R. & Radford, S. E. (2008) Folding versus aggregation: Polypeptide conformations
on competing pathways. Archives of Biochemistry and Biophysics 469: 100-117.
 Csermely, P. et al. (2010) Induced fit, conformational selection and independent dynamic
segments: an extended view of binding events. Trends in Biochemical Sciences 35: 539-
546.
References 70
References II
 Bujnicki, J. (2009). Prediction of Protein Structures, Functions, and Interactions, John
Wiley & Sons, Ltd., Chichester, p. 302.
 Tramontano, A. (2005). The Ten Most Wanted Solutions in Protein Bioinformatics, CRC
Press UK, London, p. 186.
 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.
 Ezkurdia, I., et al. (2009). Progress and challenges in predicting protein-protein interaction
sites. Briefings in bioinformatics 10: 233-246.
 Fernández‐Recio, J. (2011). Prediction of protein binding sites and hot spots.
Computational molecular science 6: 680-698.
 Szilagyi, A., et al. (2005). Prediction of physical protein–protein interactions. Physical
biology 2: S1-S16.
 Moreira, I. S., et al. (2010). Protein-protein docking dealing with the unknown. Journal of
computational chemistry 31:317-342
References 71