1. Introduction
The existence and functioning of any organism can be seen to be solely
due to proteins in its cellular environment. Most of the functionalities
arise due to several interactions of proteins with various
macromolecular entities like nucleic acids, lipids, carbohydrates, etc.,
or with other proteins. Among these interactions, protein-protein
interactions (PPIs) are of primary importance as these interacting
complexes play crucial roles in several cellular processes like
replication, transcription, translation, regulation, signaling, etc..
These protein-protein complexes (PPCs) can be categorized into different
groups based on the proportion of interacting protomers or stability of
protomers or the lifetime of interactions into homo/hetero complexes or
obligate/non-obligate complexes or permanent/transient complexes,
respectively. The complexes where the protomers become unstable when
they are separated are obligates, while the complexes where the
interactors remain stable even though they are separated are
non-obligates. On the other hand, the complexes where the protomers
interact throughout their functional lifetime are permanent, while the
complexes where the protomers associate and dissociate temporarily are
transient complexes. In general, obligate complexes are permanent both
structurally and functionally, while non-obligate complexes are mostly
transient associations with a few permanent associations like
antibody-antigen complexes. One of the great examples of such
interaction could be the heterotrimeric G protein. G protein consists of
three subunits: α, β, and γ, where β and γ subunits interact throughout
their lifetime, making it a permanent interaction. Instead, the α
subunit interacts transiently to βγ complex when inactive and
dissociates when active, making it a transient interaction. Among these,
transient interactions are of utmost importance as these complexes are
crucial for various biological processes as they act as hubs in
protein-protein interaction networks (PPINs), are multi-specific, are
great drug targets and are involved in various cellular processes.
There are several studies that distinguish structural characteristics of
such interaction types amongst proteins and PPIs. Few such
physicochemical properties which discriminate permanent and transient
interactions are contact area, interface shape and size, number of
contacts, polarity, hydrophobicity, complementarity of the interface,
involvement of secondary structures at the interface, evolution of the
interface, etc. Based on these properties, several groups focused on
distinguishing these two types of PPIs. Few groups focused solely on the
physicochemical properties or interfacial properties to predict
permanent and transient PPIs. Some groups represented these
physicochemical properties into vectors for better prediction using
machine learning approaches, while some researchers used desolvation
energy explicitly to predict permanent and transient. Apart from these,
some used sequence features to predict permanent or transient and few
developed algorithms which do not require information about binding
partners for prediction.
After the first enzyme was solved, it was found that there are some
distinct lobes present within the protein. However, the term ‘domain’
was coined by Wetlaufer by defining these entities as structurally
independent regions within proteins. These domains are also referred to
as units of protein evolution. Several structure-based identification
and analysis of domains have been performed and organized as databases.
The vast functional diversity of a protein arises by combining such
domains into a single polypeptide chain, calling it a multi-domain
protein, and most proteins, even in a simple proteome, are multi-domain
proteins. The interactions amongst multi-domain proteins with other such
proteins are mainly carried out by a portion of the protein structure, a
protein domain, rather than the whole protein. The interactions between
domains are called domain-domain interactions (DDIs), and they generally
facilitate protein interactions. It is also observed that interacting
domain pairs tend to co-evolve with each other in an interaction in
order to maintain a better interaction. The domain pairs are also
consistent with their parent protein interactions. There are few studies
that take help of known structural DDIs to predict PPIs, whether these
are input sequences or structures. Deng and coworkers used maximum
likelihood approach to estimate the probabilities of domain pairs in
protein interactions to predict PPIs. A J Gonzalez and Li Liao used
fisher scores derived from the domain interaction profiles as features
to predict DDI using SVM, which can be used to predict PPI. Instead of
using generative methods of predicting PPI, Zhao et al. used
information of both PPI and non-PPI to infer DDI, which in turn can be
used again to predict PPI from the inferred DDIs. Similarly, Sprinzak
and Margalit used correlated sequence signatures in proteins to predict
DDI.
Often, functional characteristics of a multi-domain protein is dependent
on the arrangement of the domains in it and interactions among them,
which can be compared to arrangement of words to form meaningful
sentences in natural languages. Interactions among the domains
facilitate proper functioning of multi-domain proteins. These domains
are also known to be responsible for functional and evolutionary
relationship of proteins. The occurrence of multiple domains also
confers additional stability to individual domains and hence the whole
protein. Hence, there is a need to study inter-domain interactions,
mostly in monomeric proteins, for their resident time of interactions or
strength of interactions which could provide immense knowledge about the
functional and structural aspects of multi-domain proteins. However,
unlike studies differentiating PPIs into permanent and transient
interactions, there is no systematic and organized approach to classify
DDIs into permanent and transient interactions. Instead, there are a few
studies which investigate DDIs in a single polypeptide chain and regard
such interactions to be either permanent interactions or to have
characteristics intermediate between PPIs.
In this work, we extended the concept of permanent and transient
interactions to intra-protein inter-domain interactions and
characterized the underlying interaction types. Using a dataset of
monomeric two domain proteins whose domain definitions are taken from
SCOPe, we could identify such domain interactions to be either permanent
or transient. Permanent and transiently interacting domains are not much
different in terms of evolution of the interface, and the type of
functions they are involved in, when investigated human proteome only.
However, we found that these two types of DDI differ in their
physicochemical properties of their interface, dynamically correlated
motion of their residues, and preference for choosing its interacting
partner. This work would shed light on the principles of domain
interactions, prediction of domain orientation, and protein functioning
by these rules of domain interactions. Structurally, this study would
also help to understand the folding of multi-domain proteins correctly
in the near future.