1.
INTRODUCTION:
Due to amino acid residue interaction, many biological
functions and processes occur. These interactions could be represented
graphically using vertices (amino acids) and edges (interactions) which
collectively form an amino acid residue interaction networks. They provide
better visualization in understanding these highly clustered small world
protein-protein residue interaction networks(1). In such networks, edges between two amino acid
residues are considered only when the distance between them is 8 A° or less, or
else the two residues are considered to be disconnected or at an infinite
distance from each other. Within the network analysis, there are various means
of measuring the centrality of a node which eventually helps in determining the
relative importance of that node within the network. One of them being closeness
centrality. Closeness centrality of a node is the inverse of the average farness
of that node from all the other nodes in the network(2).
According to facts, residues with high
closeness centrality value play a vital role in transmitting information to
other amino acid residues in the protein structure, thus fulfill essential
roles in the network’s communication. Therefore, concluding from the study conducted
by Fujihashi in 2006, it was seen that active site residues found in clefts or
cavities on enzymes were observed to have high closeness centrality value. However,
binding site residues with flatter shapes on specific proteins, don’t have high
closeness centrality score. This study recommended that the efficiency of
closeness centrality in identifying functionally important residues is related
to the shape of the site (2).
Thus closeness centrality can be used to identify those cavities or clefts that
contain residues that have significant protein function. Statically significant
central residues that are also conserved are called centrally conserved
residues. From the previous detailed analysis of the centrally conserved
residues and their location in the network, it is known that most of them are
found clustered in surface cavities or clefts. However, the parameter of
closeness centrality makes sure that these cavities or clefts only contain
residues essential for protein function.
In antibodies,
the 6 CDRs, three each on the light and heavy chain form a large cleft for
antigen binding(3). Moreover, the only function of the variable domain
is to identify the specific epitope and bind to it to form a reversible
antigen-antibody complex(4). After noting the facts mentioned above, it can be
said that the only functionally significant parts of the variable domain are
the CDR regions(5). Thus their closeness centrality values were expected
to be the highest(6). However, on progressing in our study by calculating
the Z scores of all the residues in the antibodies of the dataset, the expected
output was not seen in the result.
2 METHODS:
2.1
Dataset
compilation:
To identify functionally important residues in Human
Antibody Light Chain Variable Domain (HALCVD) and to determine their role in
antigen-antibody interaction, a dataset of 28 nonredundant antibody sequences was
created. Since the CDR regions are incorporated only in the variable domains of
both the chains of the antibody and because our study was localized to light
chains only, we retrieved the sequences of the variable domains of the light
chains of all the human antibodies under study. Following are the steps that went
into achieving the same.
2.1.1 HALCVD sequence dataset creation:
PDB ID’s of 120
human antibody sequences, both free and bound, were retrieved from the ABG
database and SAbDAb respectively(7). Each PDB ID was then searched in the PDB database
individually to collect the sequence of only the variable domain of the light
chain in a FASTA format(8). This dataset created with the help of the annotation
from Structural Classification of Protein (SCOP), which helps differentiate
between the sequence of the variable and constant domains on the same chain.
2.1.2 Non
redundant HALCVD sequence compilation:
It
was essential for our study that we had a dataset that included only
non-redundant HALCVD sequences. Thus
protein sequences that exceeded certain similarity thresholds had to remove.
Since the CDR regions are the source of hypervariability and antibody and
comprise of approximately 30% of the variable domain, the threshold was set at
70% similarity. Further, HALCVD
sequences with similarity higher than 70% were
removed. Thus a dataset of 28 nonredundant HALCVD sequences was
generated with the help of the program – CD hit Suite(9).
2.2 CDR region marking:
Sequences for
CDR L1, CDR L2, and CDR L3 regions of all the antibodies in the dataset were
individually found out using SAbDab’s non-redundant CDR database. A document created
denoting the location of these CDR regions on each HALCVD sequence of the
dataset by highlighting the CDR L1 (red), CDR L2 (blue) and CDR L3 (green)
sequences for all the antibody light chains under study. For example:
>1DCL:L
PSALTQPPSASGSLGQSVTISCTGTSSNVGGYNYVSWYQQHAGKAPKVIIYEVNKRPSGVPDRFSGSKSGNTASLTVSGLQAEDEADYYCSSYEGSDNFVFGTGTKVTVLG
2.3 Multiple sequence alignment:
Multiple
Sequence Alignment (MSA) was conducted with the help of Multiple Alignment
using Fast Fourier Transform (MAFFT) program, to find the conserved residues in
the HALCVD sequences all the antibodies of the dataset. The result of the MSA was modified to show CRD L1,
CDR L2 and CDR L3 sequences for all antibodies. This result was later utilized
to highlight centrally conserved residues (statistically significant central
residues that are also conserved) as shown in Figure 4.
2.4 Network representation of antibody
structures:
Each antibody
structure in the dataset was represented as a network, where amino acids were
nodes or vertices, and the interaction or contact between them formed the
edges. In such networks, edges between two amino acid residues were considered
only when the distance between them was 8 A° or less, or else the two residues
were considered to be disconnected or at an infinite distance from each other.
2.5 Closeness centrality study:
Within the network
analysis, there are various means of measuring the centrality of a node which
eventually helps in determining the relative importance of that node within the
network. One of them being closeness centrality. Closeness centrality of a node
is the inverse of the average farness of that node from all the other nodes in
the network.
Therefore, closeness
centrality value Ci for an amino acid residue i can be represented
as:
Where D (k, i)
is the shortest distance between residue i and residue k and n is the total
number of amino acid residues in the sequence(2).
With the help of
a Program, the closeness centrality values (Ci) of all the residues
of the antibodies in the dataset were determined, one antibody at a time. These
were then used by the same program to determine the Z scores of all the
residues in a given antibody. The formula represents z score:
Where, 
is the Z score of residue i, 
is the closeness centrality value of
residue i, 
is the closeness centrality average value
for all residues of an antibody and 
is the standard deviation.
The Z scores for
all the residues were obtained, one antibody at a time. These were then used to
divide residues of each antibody into two categories:
• Residues with Z scores>=1 or
Statistically Significant Central residues
• Residues with Z scores<=-1 or
Statistically Significant Peripheral residues
The same is
represented in Table 1 which shows statistically significant central and
peripheral residues of one HALCVD sequence of the database.
Since the
statistically significant central residues in all the HALCVD sequences were now
known, the MSA output was modified to show the statistically significant
central residues that were also conserved. These residues were called centrally
conserved residues.
After obtaining
the Z scores, as shown above, the following values were calculated and
tabulated for HALCVD sequences of each antibody:
I.
Total Percentage of critical
central residues that lie in the CDR regions.
II.
Percentage of
significant central residues that lie in the regions flanking CDR L1, CDR L2, and
CDR L3.
III.
Percentage of peripheral
residues that lie in the CDR L1, CDR L2, and CDR L3 regions.
IV.
Total percentage of
peripheral residues that lie in the CDR regions.
3
RESULTS AND
DISCUSSIONS:
3.1 Multiple sequence alignment:
MSA tools are
used to determine regions of conserved residues in closely or distantly related
sequences. In our study, we used the
MSA technique to determine conserved regions in 28 HALCVD
sequences of the dataset. For this, we used Multiple Alignment using Fast
Fourier Transform (MAFFT) program(10).
Figure 1: MSA
output through MAFFT
As mentioned earlier, the variable domain of an
antibody comprises 3 CDR regions separated by 4 Fr regions. The MSA output is
shown in Figure1, clearly reflected that the maximum amount of variability was
present in the three CDR regions. The hyper-variability regions saw a high
ratio of different amino acids in a given position. These amino acids were
found to be related to the most common amino acid found in that position. This
result supports the fact that CDR regions contribute to the antibody’s
specificity for binding antigens(11,12).
Conserved residues
were seen localized to the Fr regions which form the β sheet to form a
scaffold that holds the CDR residues in place for effective interaction with
its corresponding epitope on the antigen. Thus most of the residues of the Fr
regions comprise of stable amino acids which are also conserved.
3.2
Statistically
significant central residues and peripheral residues
The only
functionally significant parts of the variable domains are the CDR regions(13,14). Thus the values of their closeness centrality values
were expected to be the highest. However, on progressing in our study by
calculating the Z scores of all the residues in the antibodies of the dataset,
the expected output was not seen in the result. The real observations are
tabulated below in Table 2.
3.2.1
Statistically
significant central residues
Figure 2:
Location of statistically significant central residues in the HALCVD sequences
All the residues
of the HALCVD sequences with a Z score greater than or equal to 1, were
considered as statistically significant central residues. According to the
structure of the antibody paratope, most of these residues were expected to
belong to the CDR regions. However, this most likely outcome was contradicted
as only 4% of the statistically significant central residues belonging to the
CDR L1, CDR L2, and CDR L3 regions. While almost 60% of the statistically
significant central residues fell in the Fr regions flanking the CDR regions.
When this data was further broken down, it was found that almost 9% of the
statistically significant residues were found flanking CDR L1, whereas nearly
50% of the statistically significant residues were involved in flanking CDR L3,
and only 0.38% of the statistically significant residues were flanking CDR L2.
The same is represented in the form of a graph in Figure 2.
3.2.2
Statistically significant peripheral residues
To study the
relevance of closeness centrality in determining the residues of the CDR regions, all the HALCVD sequences residues with a
Z score less than or equal to-1 were considered as statistically significant
peripheral residues. Theoretically, these would comprise the surface residues
of the antibody network.
Thus by
filtering the residues of HALCVD sequences with Z scores equal to or below -1
it was observed that 61% of the statistically significant peripheral residues
belonged to the CDR regions, of which almost 23% fell in CDR L1, 29% in CDR L3
and only 9% in CDR L2. The same is depicted in Figure 3.
Figure 3:
Location of statistically significant peripheral residues in the HALCVD
sequences
From this outcome,
we understand that even though CDR L1, CDR L2, CDR L3, CDR H1, CDR H2 and CDR
H3 form a large cleft for antigen binding, the residues in the CDR regions on
light chain do not interact with the residues in the CDR regions on the heavy
chain. As a result of this residue in the CDR regions do not appear as a part
of a large cleft in the antibody network. Instead, CDR L1, CDR L2, and CDR L3
appear as three consecutive bulges on the surface of L chain. As a result of
them forming variable loops on β sheets created by Fr region and the same
applies to the CDR regions on the heavy chain. From these findings, we can say
that even though the CDR regions are the major functional parts of HALCVD
sequences, the parameter of closeness centrality cannot be used in identifying
the CDR regions’ residues due to their shape.
However, the question, why some of the CDR
flanking residues were recognized as statistically significant central residues?
remained unanswered. To find a valid explanation for the same the results of
MSA and closeness centrality studies were merged to give centrally conserved
residues.
3.3 Centrally
conserved residues
Statically
significant central residues that are also conserved are called centrally
conserved residues. From the previous analysis of the centrally conserved
residues and their location in the network, it is known that most of them are
found clustered in surface cavities or clefts. However, the parameter of closeness
centrality makes sure that these cavities or clefts contain residues essential
for protein function (2).
From our study
on closeness centrality values of the residues in HALCVD sequences, we observed
that 60% of the statically critical central residues belong the Fr region,
flanking the CDR regions. When we collaborated this information with the result
obtained from MSA, we obtained a result depicted in Figure 6. In this figure
the residues of the CDR regions are specified as CDR L1 in red, CDR L2 in blue
and CDR L3 in green and the area shaded in grey consists of all the centrally
conserved residues.
Figure 4:
Multiple sequence alignment of HALCVD sequences showing CDR L1, CDR L2, CDR L3
and centrally conserved residues
Thus on
observing the Figure 4, we saw that the residues flanking CDR L1 and CDR L3 are
centrally conserved thus these residues are functionally important in the
antibody network.
These central
residues are capable of efficient transmission of information in the physical
or chemical form of the other amino acid residues in the antibody, even though
they are not a part of the CDR regions. Since these centrally conserved residues of the Fr
region are a part of the variable domain, it is safe to assume that their function
is also related to the formation of the antigen-antibody complex.
Since these
centrally conserved residues belong to the Fr regions that form the β
sheet scaffold to hold the CDR regions in place, they are embedded in the paratopic
cleft and exposed like CDR residues in the form of consecutive bulges(15). However, because they are the residues that flank
the CDR residues, they are not very deeply buried in the β sheet scaffold.
Therefore, they too are capable of interacting with the antigen epitope, when
the antibody encounters the antigen.
Thus
functionally, these CDR flanking centrally conserved residues (CDRFCCR), may
have the following possibilities:
· They trigger a conformational change when
the CDR regions (paratope) recognize the specific epitope on the antigen. This
conformational change enables proper docking of the antibody on the antigen.
(ball in glove)
· On the formation of the antigen-antibody
complex, they trigger signals through the antibody that enables the constant
domain to execute a suitable immune response.
4. CONCLUSION:
The goal of this study is to determine how useful
closeness centrality is in identifying the CDR regions’ residues. According to
our study, the efficiency of closeness centrality in identifying functionally
important residues is related to the shape of the site. In antibodies, the 6
CDRs, three each on the light and heavy chain form a large cleft for antigen
binding. Moreover, the only function of the variable domain is to identify the
specific epitope and bind to it to form a reversible antigen-antibody complex.
After noting the facts mentioned above, it was assumed that the only
functionally significant parts of the variable domain are the CDR regions. Thus
the values of their closeness centrality values were expected to be the
highest. However, on progressing in our study by calculating the Z scores of
all the residues in the antibodies of the dataset, the expected output was not
seen in the result. Only 4% of the statistically significant central residues
belonged to the CDR L1, CDR L2, and CDR L3 regions. While almost 60% of the
statistically significant central residues fell in the Fr regions flanking the
CDR regions. Moreover, a staggering 61% of the statistically significant
peripheral residues were found to belong to the CDR regions. From these findings,
we concluded that even though the CDR regions are the major functional parts of
HALCVD sequences, the parameter of closeness centrality cannot be used in
identifying the CDR regions’ residues. As residues in the CDR regions do not
appear as a part of a large cleft in the antibody network, instead CDR L1, CDR
L2, and CDR L3 appear as three consecutive bulges on the surface of L chain.
5. ACKNOWLEDGEMENT:
The author thanks the Vellore Institute of
Technology for providing the necessary computational equipment to carry out the
research work.
6. REFERENCES:
1. Boginski V, Commander CW. Identifying critical
nodes in protein-protein interaction networks. In: Clustering challenges in
biological networks. World Scientific; 2009. p. 153–67.
2. del Sol A, Fujihashi H, Amoros D, Nussinov
R. Residue centrality, functionally important residues, and active site shape:
Analysis of enzyme and non-enzyme families. Protein Sci. 2006 Sep;15(9):2120–8.
3. Adolf-Bryfogle J, Xu Q, North B, Lehmann A,
Dunbrack RL. PyigClassify: A database of antibody CDR structural
classifications. Nucleic Acids Res. 2015;
4. Kabat EA, Wu TT, Bilofsky H. Unusual
distributions of amino acids in complementarity determining (hypervariable)
segments of heavy and light chains of immunoglobulins and their possible roles
in specificity of antibody-combining sites. J Biol Chem. 1977;
5. Kabat Ea, Wu Tt. Attempts To Locate
Complementarity‐Determining Residues In The Variable Positions Of Light
And Heavy Chains. Ann N Y Acad Sci. 1971;
6. Estrada E. Virtual identification of
essential proteins within the protein interaction network of yeast. Proteomics.
2006;
7. Dunbar J, Krawczyk K, Leem J, Baker T, Fuchs
A, Georges G, et al. SAbDab: The structural antibody database. Nucleic Acids
Res. 2014;
8. Sheriff S, Silverton EW, Padlan EA, Cohen
GH, Smith-Gill SJ, Finzel BC, et al. Three-dimensional structure of an
antibody-antigen complex. Proc Natl Acad Sci. 1987;
9. Huang Y, Niu B, Gao Y, Fu L, Li W. CD-HIT
Suite: a web server for clustering and comparing biological sequences.
Bioinformatics. 2010;26(5):680–2.
10. Katoh K, Misawa K, Kuma K, Miyata T. MAFFT:
a novel method for rapid multiple sequence alignment based on fast Fourier
transform. Nucleic Acids Res. 2002;
11. Ofran Y, Schlessinger A, Rost B. Automated
Identification of Complementarity Determining Regions (CDRs) Reveals Peculiar
Characteristics of CDRs and B Cell Epitopes. J Immunol. 2008;
12. Wu TT, Kabat EA. An analysis of the
sequences of the variable regions of Bence Jones proteins and myeloma light
chains and their implications for antibody complementarity. J Exp Med. 1970;
13. Mirsky A, Kazandjian L, Anisimova M.
Antibody-specific model of amino acid substitution for immunological inferences
from alignments of antibody sequences. Mol Biol Evol. 2015;
14. Wang W, Singh S, Zeng DL, King K, Nema S.
Antibody structure, instability, and formulation. Journal of Pharmaceutical
Sciences. 2007.
15. Krawczyk K, Baker T, Shi J,
Deane CM. Antibody i-Patch prediction of the antibody binding site improves
rigid local antibody-antigen docking. Protein Eng Des Sel. 2013;