Network Biology

Vol. 4, No. 2, 1 June 2014

International Academy of Ecology and Environmental Sciences

Network Biology
ISSN 2220-8879
Volume 4, Number 2, 1 June 2014

Editor-in-Chief
WenJun Zhang
Sun Yat-sen University, China
International Academy of Ecology and Environmental Sciences, Hong Kong
E-mail: zhwj@mail.sysu.edu.cn, wjzhang@iaees.org

Editorial Board
Ronaldo Angelini (The Federal University of Rio Grande do Norte, Brazil)
Sudin Bhattacharya (The Hamner Institutes for Health Sciences, USA)
Andre Bianconi (Sao Paulo State University (Unesp), Brazil)
Danail Bonchev (Virginia Commonwealth University, USA)
Graeme Boswell (University of Glamorgan, UK)
Jake Chen (Indiana University-Purdue University Indianapolis, USA)
Ming Chen (Zhejiang University, China)
Daniela Cianelli (University of Naples Parthenope, Italy)
Kurt Fellenberg (Technische Universitaet Muenchen, Germany)
Alessandro Ferrarini (University of Parma, Italy)
Vadim Fraifeld (Ben-Gurion University of the Negev, Israel)
Alberto de la Fuente (CRS4, Italy)
Pietro Hiram Guzzi (University Magna Graecia of Catanzaro, Italy)
Yongqun He (University of Michigan, USA)
Shruti Jain (Jaypee University of Information Technology, India)
Sarath Chandra Janga (University of Illinois at Urbana-Champaign, USA)
Istvan Karsai (East Tennessee State University, USA)
Caner Kazanci (University of Georgia, USA)
Vladimir Krivtsov (Heriot-Watt University, UK)
Miguel ángel Medina (Universidad de Málaga, Spain)
Lev V. Nedorezov (Russian Academy of Sciences, Russia)
Alexandre Ferreira Ramos (University of Sao Paulo, Brazil)
Santanu Ray (Visva Bharati University, India)
Dimitrios Roukos(Ioannina University School of Medicine, Greece)
Ronald Taylor (Pacific Northwest National Laboratory,U.S. Dept of Energy, USA)
Ezio Venturino (Universita’ di Torino, Italy)
Jason Jianhua Xuan (Virginia Polytechnic Institute and State University, USA)
Ming Zhan (National Institute on Aging, NIH, USA)
TianShou Zhou (Sun Yat-Sen University, China)

Editorial Office: networkbiology@iaees.org

Publisher: International Academy of Ecology and Environmental Sciences

Address: Flat C, 23/F, Lucky Plaza, 315-321 Lockhart Road, Wanchai, Hong Kong
Tel: 00852-6555 7188
Fax: 00852-3177 9906
Website: http://www.iaees.org/
E-mail: office@iaees.org
Network Biology, 2014, 4(2): 31-46

Article

Fibrillar organization in tendons: A pattern revealed by percolation

characteristics of the respective geometric network

Daniel Andrés Dos Santos1, María Laura Ponssa2, María José Tulli2, Virginia Abdala1,2,3
1
Instituto de Biodiversidad Neotropical, Facultad de Ciencias Naturales e Instituto Miguel Lillo, Universidad Nacional de
Tucumán – CONICET. Horco Molle S/N, Yerba Buena, Tucumán, Argentina
2
Instituto de Herpetología, Fundación Miguel Lillo-CONICET. Miguel Lillo 251, San Miguel de Tucumán, Tucumán, Argentina
3
Cátedra de Biología General, Facultad de Ciencias Naturales e IML, Universidad Nacional de Tucumán. Miguel Lillo 251, San
Miguel de Tucumán, Tucumán, Argentina
E-mail: dadossantos@csnat.unt.edu.ar

Received 7 January 2014; Accepted 10 February 2014; Published online 1 June 2014

Abstract
Since the tendon is composed by collagen fibrils of various sizes connected between them through molecular
cross-links, it sounds logical to model it via a heterogeneous network of fibrils. Using cross sectional images,
that network is operatively inferred from the respective Gabriel graph of the fibril mass centers. We focus on
network percolation characteristics under an ordered activation of fibrils (progressive recruitment going from
the smallest to the largest fibril). Analyses of percolation were carried out on a repository of images of digital
flexor tendons obtained from samples of lizards and frogs. Observed percolation thresholds were compared
against values derived from hypothetical scenarios of random activation of nodes. Strikingly, we found a
significant delay for the occurrence of percolation in actual data. We interpret this finding as the consequence
of some non-random packing of fibrillar units into a size-constrained geometric pattern. We erect an ideal
geometric model of balanced interspersion of polymorphic units that accounts for the delayed percolating
instance. We also address the circumstance of being percolation curves mirrored by the empirical curves of
stress-strain obtained from the same studied tendons. By virtue of this isomorphism, we hypothesize that the
inflection points of both curves are different quantitative manifestations of a common transitional process
during mechanical load transference.

Keywords percolation; collagen; fibril network; interspersion; pattern recognition.

Network Biology     
ISSN 2220­8879   
URL: http://www.iaees.org/publications/journals/nb/online­version.asp 
RSS: http://www.iaees.org/publications/journals/nb/rss.xml 
E­mail: networkbiology@iaees.org 
Editor­in­Chief: WenJun Zhang 
Publisher: International Academy of Ecology and Environmental Sciences 

1 Introduction
It has been often stressed that collagen fibrils in a connective tissue exhibit a network organization (Purslow et
al., 1998; Berthod et al., 2001; Chandran and Barocas, 2006; Rigozzi et al., 2011; Shirazi et al., 2011). It is

IAEES www.iaees.org
32 Network Biology, 2014, 4(2): 31-46

basically proposed that the microstructural properties of the collagen network contribute to continuum
mechanical tissue properties that are strongly anisotropic with tensile-compressive asymmetry (Shirazi et al.,
2011). Besides, extensible connective tissues (e.g. skin, blood vessels, fascia) contain networks of fibrillar
collagen embedded into an amorphous matrix. It is the reorientation of the collagen fibres within these
networks that allows large extensions of the tissues and is responsible for their non-linear stress–strain curves
(Wainwright et al., 1976; Purslow et al., 1998). This characterization of the collagen organization as a network
is used to explain mechanical properties of soft biological tissues. Despite distinct mechanical functions,
biological soft tissues have a common microstructure in which a ground matrix is reinforced by a collagen
fibril network (Purslow et al., 1998). Likewise, the existence of a collagen network in hard biological tissues
such as cartilage (Långsjö et al., 2009; Långsjö et al., 2010; Julkunen et al., 2010) is also widely accepted.
Collagen fibrils are not isolated functional entities, but they integrate a network system in which the proximate
fibrils could exhibit a functional connection. In spite of the currently assumption that the collagen fibril
network is responsible of the main mechanical properties of tendon, to our best knowledge the underlying
geometrical network has been never formalized in graph terms where the nodes and edges represent the fibrils
and their cross-links respectively.
The properties of connective tissues are known to depend on a wide variety of factors such as the type and
maturity of the tissue, the chemical nature of the covalent cross-links, the type and quantity of the
glycosaminoglycans throughout the extracellular matrix (ECM) and the content of elastic fibres, water and
minerals (Parry et al., 1978).Two main classes of extracellular macromolecules make up the tendon matrix:
proteoglycans (PGs), which play a complex role in force transmission and maintenance of tendon tissue
structure (Reed and Iozzo, 2002; Rigozzi et al., 2010) and collagen fibrils. Studies that have investigated the
relationship between structural and mechanical properties have generally focused on one major component,
either collagen or proteoglycans (PG), with studies focusing on collagen fibril morphology being more
common (Rigozzi et al., 2010).The amount of interactions between the collagen fibrils and the surrounding
matrix influents the stiffness of the tissue, and this may prevent changes in shape after the removal of stress
(Parry et al., 1978). The degree of interaction between the collagen fibrils and the amorphous matrix is
function of the collagen fibril diameter distribution. The relevant role of the ECM is also visible during tendon
development, because collagen fibrillogenesis generates a tendon-specific extracellular matrix that determines
the functional properties of the tissue (Zhang, 2005; Zhang et al., 2005). Other studies indicate that
biochemical deficiencies the amorphous ECM may be a primary causative factor in certain tendon pathologies
(Battaglia et al., 2003; Mikic et al., 2001).
Networks are a collection of elements (nodes or vertices) connected by some relationships of interest
(links or edges) (Zhang, 2012a, 2012b, 2012c). The internet, airline routes, and electric power grids are all
examples of networks whose function relies crucially on the pattern of interconnection between the
components of the system. Thinking of systems as networks and studying their patterns of connection can
often lead to new and useful insights (Newman, 2010; Ferrarini, 2013, 2014; Zhang, 2012a, 2012b, 2012c,
2013). An important property of such connection patterns is their robustness—or lack thereof— to removal of
network nodes, which can be modeled as a percolation process on a graph representing the network (Callaway
et al., 2000). Percolation theory is a branch of probability theory dealing with properties of random media
(Berkowitz and Ewing, 1998; Zhang, 2012a, 2013). It is one of the simplest models in probability theory
which exhibits what is known as critical phenomenon. This usually means that there is a natural parameter in
the model at which the behavior of the system drastically changes (Grimmett, 1999). Percolation statement is
simple: every site on a specified lattice is independently either occupied (recruited), with probability q, or not
with probability 1 − q. In particular, the system percolates when it exhibits a continuous phase transition at a

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 31-46 33

finite value of q which, on a regular lattice, is characterized by the formation of a cluster large enough to span
the entire system across its dimensions in the limit of infinite system size (Newman and Ziff, 2001). From a
mathematical perspective, the percolation theory describes the behavior of connected clusters in a random
graph (Cuestas et al., 2011). The percolation method allows us to evaluate the network resilience to deletion of
network nodes (Callaway et al., 2000).
In this work we delineate a network approach for studying the tendon organization (system composed of
interconnected fibrils) that could contribute to a better understanding of its biomechanical responses. Relative
neighborhood networks between collagen fibrils are here erected as proxies for the underlying collagen
network. This type of objects constitutes an appropriate candidate to make mathematically tractable the
collagen network. The links (edges) between the discrete units (nodes) of this network are derived from the
relationships of spatial proximity between fibrils. However, the spatial gap between fibrils connected by
proximity is occupied by the amorphous matrix and should be also considered a relevant component for the
functionality of the entire system. Amorphous matrix is assumed to be the physical substrate over which the
information can flow across the nodes of the network. In dealing with the term transference of information, we
adopt the meaning implicit to the information theory (Shannon, 1948) that involves the transmission of data or
any state change in a system. This notion liberates us from considering the ECM network as a theoretical
model uniquely associated to the context of force transmission. The model of ECM network is useful to
address the topic of functional integrity. One way to do that is to evaluate if information can propagate
throughout the structure of the network, or equivalently to study its characteristics of percolation. Percolation
theory may contribute to the understanding of the information flow such as force transmission from the
beginning of the tensile activity (tendon activated by a contractile force) until the resulting response
(movement of skeletal pieces).
This paper is interested on the morphology and spatial organization of the tendon collagen fibrils in
addition to their functional consequences. It is structured around the following assumptions: i) collagen fibrils
are units that mediate the transmission of information; ii) the tendon is an assembly of interconnected fibrils
that can be modeled with the approach of geometric networks; iii) molecular cross-links provides the material
evidence about the connection between fibrils; iv) percolation of information throughout the network is
directly associated to the notion of functional integrity; v) phase transition involved by the percolation
threshold can be traced to a point on the non-linear stress-strain curve. In tight correspondence with the above
premises, and considering a hypothetical scenario of fibril recruitment by increasing size, we ask the following
questions: 1) which is the critical threshold that allows the information to traverse the physical dimensions of
the network in which fibrils reside? Based on the observed percolation, 2) is it possible to infer some peculiar
geometrical feature of the network structure? To answer this, we need to compare the observed percolation
pattern against the random expectations under a stochastic shuffling of the fibril sizes but maintaining the
topology of the network. Taking into account that percolation of a system implies a phase transition, 3) which
is the influence of this putative phenomenon in the mechanical properties exhibited by tendons? Can the stress-
strain curves be used to deal with the last issue? We think that a right comprehension of all these inquiries will
bring us new insights to grasp the outstanding mechanical properties of the tendon.

2 Material and Methods

Electron microscopy analysis was conducted with samples of the flexor tendon of the Digit IV obtained from
adults of anurans (Scinax nasicus, Phyllomedusa sauvagii, Rhinella arenarum, Leptodactylus latinasus and
Leptodactylus chaquensis) and squamatan reptiles (Liolaemus elongates, L. coeruleus, L. bibroni and
Tupinambis rufescens). Details about examined specimens are provided in the Appendix. We selected the

IAEES www.iaees.org
34 Network Biology, 2014, 4(2): 31-46

flexor tendon of Digit IV because of the importance of this digit in locomotion as stressed by Teixeira-Filho et
al. (2001) and Tulli et al. (2011).
Samples were placed overnight in 0.1-M phosphate buffer with 2.5% glutaraldehyde and 4%
paraformaldehyde. The tissue was then fixed in 1% osmium tetroxide, dehydrated in graded acetones, and flat
embedded in Epon plastic 812 (Ernest F Fullam, Inc, Latham, NY) in a cross-sectional orientation. Sections
(85 nm) were obtained and stained with 0.25% lead citrate and 5% uranyl acetate in 50% acetone and then
observed and photographed in a JEOL100CX transmission electron microscope (LAMENOA, Universidad
Nacional de Tucumán, Argentina). Collagen fibril diameters of each species were measured on each micro-
graph using the Image J 1.44p (Wayne Rasband, National Institutes of Health, USA,http://rsbweb.nih.gov/ij/).
Diameter of each fibril present in the selected area was measured. Each fibril included in the selected area was
identified in a coordinate system using the particle analysis option of the Image J software.
2.1 The collagen fibril network
Each cross section of a tendon can be represented as a set of points distributed in a 2-dimensional space. Since
these points are assumed to interchange information they give rise to a spatial network. In fact,
glycosaminglycancross-links connect each fibril with the adjacent ones. To address the problem of the
geometrical organization of the network, we need to operationalize the concept of the fibrillar network.
Because spatial networks subsume into the category of geometric graphs, we choose this mathematical model
to work with. A geometric graph G = (V, E) is an embedding of the set V of nodes as points in the plane, and
the set E of edges as straight line segments joining pairs of points in V. Locations for the center mass of each
fibril are adopted as point occurrences of fibrils throughout the study system (here, the cross section of the
tendon). We also say that the graph G is planar if no two of its edges intersect except perhaps at their end
points. In the computational geometry literature, there are several classes of planar geometric graphs arising
from what are known as proximity graphs (see the survey of Jaromczyk and Toussaint 1992 for more details).
In this paper, we will model the topology of the collagen fibril network via a Gabriel graph. The rationale
for this choice relies on the capacity of the Gabriel graph to capture a fair representation of the proximity
structure portrayed by the data. A geometric graph G= (V, E) is called a Gabriel graph if the following
condition holds: for any u, vϵV, an edge (u, v) ϵE if and only if the circle with uvas diameter does not contain
any other point of V (Gabriel and Sokal, 1969) (Fig. 1). The ultimate meaning of observing two adjacent fibrils
A and B in the underlying Gabriel graph (connected by an edge) is that their area of reciprocal influence is
unique and not interfered by nearby fibrils. If the stress force does flow throughout the interfibrillar matrix, it
would freely flow across the space bridging fibrils directly linked in the respective Gabriel graph.

Fig. 1 Example of Gabriel graph (right) obtained from the Delaunay triangulation (left) applied on a set of 4 points, i.e. {a, b, c
and d}. The edge joining a toc is discarded because it corresponds to the diameter of a circle that includes another point of the set
(i.e. point d).

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 31-46 35

2.2 Percolation
The network of collagen fibrils is used to represent the propagation of information across the tendon. Here,
information can be understood as the propagation of the mechanical stress across the tendon, or the flow of the
cross-linking molecules through the amorphous matrix, or any other factor that can change the integral
response of the tendon. Vertices on the graph are considered occupied or not, depending on whether the
network nodes they represent (fibrils) are activated or recruited. We have examined the site percolation of the
network in which the occupation is a function of the fibril size. So, we have considered percolation under a
targeted activation of nodes. In this way, we have simulated a sequential recruitment of fibrils increasingly
ordered by their sizes and checked then the spanning area of the putative percolating cluster.
In site percolation of spatial networks one views the clusters formed by the occupied vertices. A cluster is
defined as set of neighboring or adjacent vertices that are occupied. Of particular interest is to establish if there
exists a cluster that connects the borders of the area spanned by the network (here, the cross section of the
tendon). Such a cluster is called a percolating cluster and its presence represents a qualitative change in the
structure of the system from a disconnected state to a connected one (functionality transition). We estimate the
percolation threshold (q) from the inflection point of the S-shaped curved relating the relative size of the
spanning cluster and the fraction of activated sites. The following fitting model was used (Pawłowska and
Sikorski, 2013):

1 1 exp

where a is a parameter that dictates the slope of the curve. Here, percolation probability P is function of the
fraction of activated sites. In operative terms, nonlinear least square regression was applied on the values
of relative size of the percolating cluster against the respective fraction of activated sites. The relative size of
the percolating cluster was taken as the ratio between the area of the bounding box enclosing the points of the
largest component and the overall rectangular area of the system under study. Percolation threshold was
calculated for the case of targeted sequential activation of nodes ordered by increasing size. In a next instance
of analysis, this observed score for q was compared against values obtained from random scenarios of node
activation. Certainly, if the observed value would deviates from random expectations then we could suspect
some pattern of spatial organization where the size of fibrils has a prominent role.
We conducted separate recruitment scenarios for each empirical case of study. Each scenario consisted of
21 instances in a gradual sequence of node activation, using for that purpose a vector of cutoff values that
dictate the activation (or not) of the nodes. Whenever a node has a size (= fibril diameter) lower or equal than a
specified cutoff value it is activated. The 21 elements of the referred vector are increasingly ordered and
correspond to the percentiles by steps of five (0, 5, 10, …, 100) found on the statistical distribution of fibril
sizes. Thus, the first element of this vector corresponds to the minimum size of fibril, its second element
corresponds to the fifth percentile, its third element is the tenth percentile, and so on. The last element
concerns to the largest fibril. Although exploring the entire set of unique values of size would have yielded
more accurate results, we have binned the data by percentile intervals of five because of computational
facilities. The full activation of nodes translates into the original Gabriel graph that accounts for the collagen
fibril network. As the activation of nodes proceeds from the minimum cutoff value to the maximum one, nodes
are connected by an edge if they are actually neighbors in the underlying Gabriel graph. The critical instance,
or scene, of the recruitment scenario is that where the percolating cluster firstly appears. The inflection point
on the percolation curve, obtained by adjustment of the above equation to the 21 instances of node activation,
was used to estimate the percolation threshold q. The q quantile for the set of fibril diameters was subsequently

IAEES www.iaees.org
36 Network Biology, 2014, 4(2): 31-46

employed for inferring the critical size for fibril activation around which percolation occurs.
To test the significance of the observed values of percolation, we compared them with scores obtained via
random simulations. We conserved the network topology and changed the original assignment of size values
among the nodes. We set the level of significance at 5% over a total of 100 random simulations. Rejection of
null hypothesis would suggest a geometrical pattern shaped by the size of the collagen fibrils.
2.3 Balanced interspersion of heteromorphic fibrils
We will consider a model of non-random interspersion to explain the particular arrangement of polymorphic
fibrils, i.e. a pattern where a given fibril is surrounded by others of dissimilar size. We will study an
algorithmic arrangement of fibrils on a square lattice able to mimic the percolating behavior observed across
the empirical cases of study.
2.4 Stress-strain relationship and inflection point
Tendon samples of Leptodactylus chaquensis, Phyllomedusa sauvagii, Rhinella arenarum and Tupinambis
rufescens were carefully aligned and gripped using Instron® at room temperature located at Facultad de
Odontología, Universidad Nacional de Tucumán. Tendons were tested in tension up to break at the same
elongation rate of 0.033 mm/s. Initial gauge length and width of specimens were measured before performing
the tensile test. Once the ultimate tensile strength was achieved, we decided to report any additional reading of
stress-strain above the 90% of this maximum caused by the gradual breakage of tendons by defibrilation.
Curves were generated through piecewise-cubic splines on the raw data. The inflection point of a curve is
the geometrical place where a change of concavity is produced, and they are linked with phase transitions in
the dynamics of a system. We estimated the inflection point directly from the raw data following the
methodology of Christopoulos (2012). In the context of this paper, the inflection point corresponds to the
transition from a convex strain-stress response to a concave one. All statistical and network analyses were
performed with the R platform (R Core Team, 2012). The scripts are available from heading author upon
request. This study was approved by the Ethics Committee of Universidad Nacional de Tucumán.

3 Results
The Fig. 2 shows the distributions of the fibrils according to their size. The box plots show a segregation of the
data in three pools clearly distinguishable: one composed by small fibrils (the majority less than 50 nm),
another conformed by medium fibrils (most of them between 50 and 100 nm), and other by large fibrils (the
majority of them larger 100 nm). Segregation of the taxa proceeds regardless of the phylogeny. The range of
variability of fibril size seems to increase in direct relationship with the median of fibrils.
3.1 Percolation
The Fig. 3 shows two scenes taken from the progressive recruitment process, namely (i) the scene concerning
to the stage near to the formation of the percolating cluster, (ii) the scene concerning to the stage immediately
posterior to the formation of the percolating cluster. The percolation corresponds to the inflection point in the
fitted S-shaped curves relating quantile probabilities and fraction of spanning area. Percolation thresholds
calculated for the empirical cases are in Table 1.

IAEES www.iaees.org
 Networkk Biology, 2014
4, 4(2): 31-46 37

Fig. 2 Sizze distribution of

o digital flexoor tendon fibrilss across some representativess of South Ameerican herpetoffauna. For eachh
species, daata coming fromm all samples were
w pooled intoo a single data set.
s Species cann be segregated into three size categories: onee
dominatedd by small fibriils (< 50 nm), another characcterized by inteermediate fibrils (50-100 nm) and the last on ne where largee
fibrils prevvail (> 100 nm)). Outliers havee been removed and whiskers extend
e to the most extreme datta point which is
i no more thann
t interquartilee range from the box.
1.5 times the

Tablee 1 Percolation characteristics of tendon cross sections.

Percentagee of Med dian of the
Numbber
Averaage percolatio
on threshold images that fibril size at the
of
resulted in a high obseerved
Speciess samplling
percolationn percolation
imagees Tragetted Ranndom
threshold (P< thresshold
analyzzed activaation actiivation
0.05) (nm))
Tupinammbis rufescenns 1 0.72 0.53
3 100% 241
Leptodaactylus chaquuensis 11 0.62 0.51
1 82% 187
Phyllom
medusa sauvaagii 7 0.65 0.48
8 100% 153
Rhinellaa arenarum 3 0.63 0.49
9 100% 92
Leptodaactylus latinaasus 5 0.59 0.49
9 80% 89
Scinax nasicus
n 1 0.65 0.48
8 100% 88
Liolaem
mus bibroni 1 0.59 0.47
7 100% 83
Liolaem
mus coeruleuss 1 0.64 0.52
2 100% 36
Liolaem
mus elongatuss 1 0.70 0.49
9 100% 23

IAEES www.iaees.org
w
38 Networkk Biology, 2014
4, 4(2): 31-46

Fig. 3 Fibrril proximity neetworks at two contrasting stagges of node acttivation. Each fibril
fi is proportiionally represen
nted to its crosss
section areea. The underlyying Gabriel graph (overall infferred network)) is also shownn and its nodes are located at the t mass centerr
of each fibbril. Node activvation proceedss orderly progreessing from the smallest node to the largest oone. Images at the t left columnn
represent snapshots
s of neetworks captureed from activattion instances close
c to the form
mation of the reespective perco olating clusters..
Images at the right colum mn illustrate sceenes of fibril reecruitment oncee the percolatioon threshold hadd been recently y surpassed. Ass
nodes are activated they shift
s their fillinng color from grrey to red. Therre exists a blue link between a pair of adjacen nt nodes if bothh
of them arre activated; othherwise they arre connected thrrough a grey lin nk. In each scenne, dotted bounnding boxes forr both the studyy
area and thhe largest compponent of activaated nodes are displayed. Notee that the area spanned by thee bounding box of the putativee
percolatingg cluster (right column) closely matches the entire study arrea despite som me nodes are still inactivated. Scale
S bars: 2500
nm.
In thhe bulk of data of random
m simulationn, the observeed value of percolation
p thhresholds wass higher thann

IAEES www.iaees.org
w
 Networkk Biology, 2014
4, 4(2): 31-46 39

random expectancy
e (P
P< 0.05) (Tabble 1).Random m simulation ns revealed thhat there is a ggeometrical pattern
p linkedd
to the siize of the fibbrils in mosst of the sam
mples, becausse the percollation occurss at earlier stagess of thee
progressiive recruitmeent when the attribute sizee for the fibriils is randomlly decoupledd from the truue location off
them (Fiig. 4). With a slight abuuse of wordss, the high valuesv of perrcolation threesholds deteccted imply a
geometriical pattern coonstrained byy size.

Fig. 4 Perccolation curvess. The percolatioon threshold is estimated as thhe inflection poiint of the adjustted S-shaped cu
urve that relatess
the fractioon of activated fibrils againstt the relative reectangular areaa occupied by the largest com mponent of theem. The dottedd
horizontal line passes throough the inflecttion points of fitting
fi curves. For the scenario of progressive size-dependentt recruitment off
fibrils (bluue curve), the respective percoolation thresholld falls always beyond the onee-sided 95% coonfidence interv val. Confidencee
intervals were
w created aftter considering the curve param meters associateed to 100 scenaarios of random recruitment off fibrils.

3.2 Balanced intersp persion of polymorphic fiibrils

We have been ablee to reproduuce conditionns of delayeed percolatioon through a non-random m layout off
heteromoorphic units (elements
( of different sizees) on a squarre lattice. The pattern undder consideration is calledd
by us as a Balanced Interspersionn of Polymoorphic Units (BIPU). BIIPU consists of a regulaar alternatedd
occurrennce of larger and smaller fibrils througghout the phy ysical dimenssions of the ssystem. Oncee the pool off
fibrils haas been partittioned into tw
wo sets of coontrasting sizees (large andd small), we aallocate themm following a
chess booard pattern (Fig. 5). Seequential activation of siites in this pre-ordered
p template maay achieve a
percolation thresholdd similar to the observeed ones in the analysis of cross-seection imagees of tendonn
ultrastruccture, i.e. q ~ 0.64 (Table 2).

IAEES www.iaees.org
w
40 Networkk Biology, 2014
4, 4(2): 31-46

Fig. 5 Moodel of balanced interspersioon of polymorpphic fibrils. On ne hundred fibrils have beenn sampled from m two differentt
distributioons of cross section diameter inn nanometers. Fifty
F elements were drawn froom a normal vaariable S1 ~ (1440, 20) whereass
the other fifty
f elements were
w sampled from
f S2 ~ (100, 20). Subsequeently, the totaliity of elements is arranged across a 10 X 100
square latttice where the larger
l elements (in any order) alternate with the
t smaller onees (in any order) in a chess boaard pattern. Forr
an easy baacktracking of iddeas, the generaating R script iss displayed nex
xt to the respectiive layout.

Table 2 Statistical syntheesis (mean pluss minus standarrd deviation) off percolation exxperiments perfoormed on severral hypotheticall
square latttices in which putative fibrilss of various sizzes were allocaated following the pattern of balanced intersspersion (chesss
board layoout where largeer fibrils alternaate side-to-side with smaller fiibrils). For eachh lattice configuuration, 100 rep
plications weree
run.

Targeted Random
Size of the square lattice
n
activation acctivation

10 X 10 0.65 (± 0..03) 0.52 (± 0.05)

10 X 11 0.66 (± 0..03) 0.52 (± 0.05)
10 X 12 0.65 (± 0..03) 0.51 (± 0.05)
11 X 11 0.66 (± 0..03) 0.51 (± 0.05)
11 X 12 0.66 (± 0..03) 0.52 (± 0.05)
11 X 13 0.67 (± 0..03) 0.52 (± 0.05)
12 X 12 0.66 (± 0..03) 0.52 (± 0.06)
12 X 13 0.65 (± 0..03) 0.52 (± 0.05)
12 X 14 0.66 (± 0..03) 0.53 (± 0.05)
13 X 13 0.66 (± 0..03) 0.52 (± 0.05)
13 X 14 0.66 (± 0..03) 0.52 (± 0.05)
13 X 15 0.66 (± 0..03) 0.52 (± 0.04)
14 X 14 0.66 (± 0..03) 0.53 (± 0.04)
14 X 15 0.67 (± 0..03) 0.52 (± 0.05)
14 X 16 0.66 (± 0..02) 0.52 (± 0.05)
15 X 15 0.66 (± 0..03) 0.53 (± 0.05)
15 X 16 0.66 (± 0..02) 0.52 (± 0.04)
15 X 17 0.66 (± 0..02) 0.53 (± 0.04)
16 X 16 0.66 (± 0..03) 0.53 (± 0.05)
16 X 17 0.66 (± 0..03) 0.53 (± 0.04)
16 X 18 0.66 (± 0..02) 0.53 (± 0.04)
17 X 17 0.67 (± 0..03) 0.53 (± 0.04)
17 X 18 0.66 (± 0..02) 0.53 (± 0.04)

IAEES www.iaees.org
w
 Networkk Biology, 2014
4, 4(2): 31-46 41

17 X 19 0.66 (± 0..02) 0.53 (± 0.03)

18 X 18 0.66 (± 0..02) 0.53 (± 0.04)
18 X 19 0.67 (± 0..02) 0.53 (± 0.04)
18 X 20 0.67 (± 0..02) 0.54 (± 0.04)

3.3 Stresss-strain relaationship and d inflection point

p
Typical stress-strain
s curves for annurans and sqquamatans digital flexor tendons are shown in Fig g.6. Ultimatee
tensile sttress (UTS) for
f each species and summ mary statistics about the stress and straain found at the
t inflectionn
point aree showed in Table
T 3.

Fig. 6 Streess-strain curvees. Direct readiings from the Instron®

I devicee were adjustedd through pieceewise-cubic splines. Inflectionn
points marrked as red dotss.

Table 3 Basic
B measuremments of stress-sstrain relationshhip. Reported values
v are the medians
m of the respective sammples. Scores off
critical strrain and stress are those recorrded at the infllection point off the polynomiaal cubic fit of tthe empirical data.
d Maximumm
stiffness corresponds to thhe slope of the tangent at suchh inflection poin
nt. UTS= ultimaate tensile strenngth.
Strain at Criticall
UTS Criticalsttrain mumstiffness
Maxim
Speciess UTS stress
(MPa) (mm/mm m) (MPa)
(mm/mm)) (MPa)
Rhinellaa arenarum(nn = 6) 0.36 11.22 0.25 6.64 81.27
Phyllom
medusa sauvaagii (n = 5) 0.23 65.50 0.16 50.14 569.65
Leptodaactylus chaquuensis (n = 6)) 0.62 51.49 0.45 29.82 176.56
Tupinam
mbis rufescenns(n = 5) 0.27 47.52 0.15 21.89 488.13

IAEES www.iaees.org
w
42 Network Biology, 2014, 4(2): 31-46

4 Discussion
The survival of living organisms is dependent on the functional integrity of all their organs and organ systems.
The tendon is a biomechanical system of force transference from muscles to bones in which its functional
integrity translates into a controlled movement of joints. Without loss of generality, the functional integrity of
any system relies on the ability of their components to propagate information and offer an adaptive response to
external influences. It is hard to conceive such a property in a system of poorly connected components or,
equivalently stated, in a system with a relaxed network structure underpinning it. In this paper, we have
assumed a network organization for the fibrils of the tendon and consequently explored the patterns of
connections between them under the approach of network percolation. At least to our knowledge, percolation
theory is the most adequate conceptual framework to answer inquiries about the functional integrity of the
tendon, because it deals explicitly with the subject of information propagation throughout the physical
dimensions where the system resides.
While Svensson et al. (2013) have recently pointed out that fibrils are not evenly loaded within the tendon
butare sequentially recruited throughout the initial stress-strain region, the analysis of our data revealed
moreover a size-dependent effect for that sequential recruitment of fibrils. The percolation thresholds are
consistently biased towards the upper tail of the statistical distribution of the size of fibrils. It is then
comprehensible that the following taxa: Leptodactylus chaquensis, Phyllomedusa sauvagii, and Tupinambis
rufescens, exhibit great values for the percolation threshold size. All they surpass easily the 100 nm of
diameter in the cross section of fibrils. Ultimately, this means that taxa with great fibrils do not achieve
percolation by activation of their small fibrils alone, larger fibrils are also necessary to be activated. A
marginal essay performed by us using images about tendon ultrastructure available from literature (horse:
Parry, 1988; mouse: Ameye et al., 2002; rabbit: Gill et al., 2004) showed the same pattern. In mammals it also
seems to be necessary the activation of fibrils around the 65th percentile to achieve percolation. When
compared with the random activation of fibrils, the observed values for percolation thresholds also resulted
significantly higher than random expectations.
The delay detected for the occurrence of percolation during the recruitment process indicates that collagen
fibrils are spatially arranged according to a geometrical non-random model, in which the location and size
attribute are both important. We propose the model BIPU of fibril arrangement characterized by the even
interspersion of fibrils of different size category throughout the ECM. This pattern would account for a
uniform fibrillar density, that makes restricted regions to be very similar with regards of the overall tendon,
and also would explain the delayed percolation threshold observed along our experiments of targeted sites
activation. In fact, the balanced interspersion of polymorphic unit simply that larger fibrils makes a shadow
effect on nearby smaller fibrils decreasing thus the chance of direct connections between the latter ones.
Additionally, the balanced interspersion would also imply that heterogeneous fibrilsare close each other
facilitating the access to information managed by the different fibrils. An open question is if the delay for
percolation represents a way to optimize the physiological range of tendons.
Traditionally, a typical stress-strain curve for a tendon has been characterized by three regions: the toe,
linear and non-linear regions (Wang, 2006). On the contrary, we interpret the resulting stress-strain curve as
composed by two phases: the convex and concave regions, being the point of inflection the transition between
them. Functionally, they would reflect two quite different behaviors: in the first region, stiffness increases by a
targeted activations of fibrils depending upon their sizes, whereas in the second region the tendon begins to
yield and fracture. This change in the approach to analyze the stress-strain curve abruptly breaks with the
conventional perspective of considering an intermediate linear region in which the respective slope reflect
some biomechanical property intrinsically linked to the structure of tendons. In other words, we think that the

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 31-46 43

task of calculating the Young's modulus as a unique and distinctive measure seems to be a sum-zero exercise
in analyzing functional tendon properties. Consequently, we propose to rely on the point of inflection to
calculate several metrics that account for the mechanical properties of tendons such as critical stress, critical
strain (physiological range), and critical stiffness. We propose an isomorphic linkage between stress-strain and
percolation curves, and we hypothesize that the inflection points of both curves reflect a phase transition
associated to the same underlying process. This process is probably the propagation of fibril disfunction
throughout the extracellular ground of connective tissue. When failure percolates the overall system begins to
elongate elastically, so the percolation threshold could indicate the upper bound for the physiological range of
the tendon.
In our approach, the physical properties of a tendon depend on the pattern of connections behind the
geometrical network of fibrils where location and difference in size play a key role. It is interesting to note that
a polimodal distribution of fibril sizes characterize those connective tissues able to resist tensional forces such
as tendons. On the contrary, those connective tissues that are commonly under no tensile force show a
typically unimodal distribution of fibril size (e.g. buccal gingival mucosa collagen: Ottani et al., 1998; skin
collagen: Danielson et al., 1997; Silver et al., 2001; corneal stroma lamella: Parry et al., 1978). Likewise, those
tendons that are not yet functional, such as embryonic tendons exhibit also anunimodal distribution of collagen
fibril size (Fleischmajer et al., 1988; Parry et al., 1978; Zhang et al., 2005). Unimodal distribution of collagen
fibril size is also present in regenerated tendons (Gill et al., 2004). We suggest that for acting efficiently, the
polymorphism must be accompanied of another geometrical feature of spatial organization. This spatial
organization seems to be subsumed into a pattern of balanced interspersion of different fibrils throughout the
ECM. An immediate application of our proposal concerns with the design of strategies for tissue engineering
provided of biomimetic-synthetic nanofibrous composites.

5 Concluding Remarks
The polymorphic nature of collagen fibrils in addition to the presence of molecular cross-links between them
lead us to think in an heterogeneous network of fibrils that influences the mechanical behavior of tendon as a
whole. The main contribution of our work is to combine geometrical and network considerations into a single
framework by using the percolation approach. This analysis allows a holistic study of the structural properties
of a tendon based on their architectural design. The geometrical pattern we have suggested (non-random
packing constrained by size) is amenable with the idea of a progressive and sequential recruitment of fibrils
dependent on their size. This pattern offers a delay to percolation, which could act a mechanism for expanding
the physiological range. As a surplus of our work, we also address the circumstance of being percolation
curves mirrored by the empirical curves of stress-strain obtained from the same studied tendons. By virtue of
this isomorphism, we hypothesize that the inflection points of both curves are different quantitative
manifestations of a common transitional process during mechanical load transference.

Appendix Specimens examined

L: personal collection of María Laura Ponssa; FBC: personal collection of Felix B. Cruz; FML: Fundación
Miguel Lillo; GS: personal collection of Gustavo Scrocchi; ST: personal collection of Sebastián Torres. An
asterisk is appended to the access number if the respective specimen was used for studying cross sectional
tendon images, otherwise specimens were used for stress-strain analysis.
Leptodactylus chaquensis: L103, L766, L850*, L344, L950, L950-951*, L964*, L971*, ST103; Leptodactylus
latinasus: L937a*, L939-942*, L944-945*;
Phyllomedusa sauvagii: L849*, L936*, L938*, L946-947, L947(1), L946-948*, L971;

IAEES www.iaees.org
44 Network Biology, 2014, 4(2): 31-46

Rhinella arenarum: L51, L851*, L909, L935, L935*, L937b*, L961, L965;
Scinax nasicus: L949*;
Liolaemus bibroni: FBC 1265*;
Liolaemus coeruleus: FBC 1265*;
Liolaemus elongatus: GS3227;
Tupinambis rufescens: FML 07256*, FML7554.

Acknowledgements
This work was supported by the ANPCyT and CONICET through the following research grants: PICT 2012-
1067, PICT 2012-1910, PIP 112-200801-00225, and BID-PICT 606. All authors are grateful to CONICET for
supporting our work via its program of post-graduate fellowships and research grants. We thank Gabriela
Pacios (Facultad de Odontología, Universidad Nacional de Tucumán)for assistance with Instron, and
NicolásNieva (Facultad de Ciencias Exactas, Universidad Nacional de Tucumán) for comments that greatly
improved the structure of the manuscript.

References
Ameye L, Aria D, Jepsen K, et al. 2002. Abnormal collagen fibrils in tendons of biglycan/fibromodulin-
deficient mice lead to gait impairment, ectopic ossification, and osteoarthritis. FASEB J, 16: 673-680
Battaglia TC, Clark RT, Chhabra A, et al. 2003. Ultrastructural determinants of murine achilles tendon
strength during healing. Connective Tissue Research, 44: 218-224
Berthod F, Germain L, Li H, et al. 2001. Collagen fibril network and elastic system remodeling in a
reconstructed skin transplanted on nude mice. Matrix Biology, 20: 463-473
Berkowitz B, Ewing RP. 1998. Percolation theory and network modeling applications in soil physics.
SurvGeophys, 19: 23-72
Callaway DS, Newman MEJ, Strogatz SH, Watts DJ. 2000. Network robustness and fragility: percolation on
random graphs. Physical Review Letters, 85: 5468-5471
Chandran PL, Barocas VH. 2006. Affine versus non-affine fibril kinematics in collagen networks: theoretical
studies of network behavior. Journal of Biomechanical Engineering, 128: 259-270
Christopoulos DT. 2012. Developing methods for identifying the inflection point of aconvex/ concave curve.
arXiv:1206.5478v1 [math.NA]
Cuestas E, Vilaró M, Serra P. 2011.Predictibilidad de la propagación espacial y temporal de la epidemia de
influeza A-H1N1 en Argentina por el método de percolación. Revista Argentina de Microbiología, 43:
186-190
Danielson KG, Baribault H, Holmes DF, et al. 1997. Targeted disruption of decorin leads to abnormal
collagen fibril morphology and skin fragility. Journal of Cell Biology, 136: 729-743
Ferrarini A. 2013. Exogenous control of biological and ecological systems through evolutionary modeling.
Proceedings of the International Academy of Ecology and Environmental Sciences, 3(3): 257-265
Ferrarini A. 2014. True-to-life friction values in connectivity ecology: Introducing reverse flow connectivity.
Environmental Skeptics and Critics, 3(1): 17-23
Fleischmajer R, Perlish JS, Timpl R, et al. 1988. Procollagen intermediates during tendon fibrillogenesis.
Journal Hist Cyt, 36: 1425-1432
Gabriel KR, Sokal RR. 1969. A new statistical approach to geographic variation analysis. Systematic Zoology,

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 31-46 45

18:259-278
Gill SS, Turner MA, Battaglia TC, et al. 2004. Semitendinosus regrowth. Biochemical, ultrastructural, and
physiological characterization of the regenerate tendon. American Journal of Sports Medicine, 32: 1173-
1181
Grimmett G. 1999. Percolation (2nd edition). Springer-Verlag, New York, USA
Jaromczyk JW, Toussaint GT. 1992. Relative neighborhood graphs and their relatives. Proceedings of the
IEEE, 80: 1502-1517
Julkunen P, Livarinen J, Brama PA, et al. 2010. Maturation of collagen fibril network structure in tibial and
femoral cartilage of rabbits. Osteoarthritis and Cartilage, 18: 406-415
Långsjö TK, Arita M, Helminen HJ. 2009. Cartilage collagen fibril network in newborn transgenic mice
analyzed by electron microscopic stereology. Cells, Tissues and Organs, 190: 209-218
Långsjö TK, Vasara AI, Hyttinen MM, et al. 2010. Quantitative analysis of collagen network structure and
fibril dimensions in cartilage repair with autologous chondrocyte transplantation. Cells, Tissues and
Organs, 192: 351-360
Mikic B, Schalet BJ, Clark RT, et al. 2001. GDF-5 deficiency in mice alters the ultrastructure, mechanical
properties and composition of the Achilles tendon. Journal of Orthopaedic Research, 19: 365-371
Newman MEJ. 2010. Networks: An Introduction. Oxford University Press, Oxford, UK
Newman MEJ, Ziff RM. 2001. A fast Monte Carlo algorithm for site or bond percolation. Physical Review E,
64: 016706
Ottani V, Franchi M, Depasquale V, et al. 1998. Collagen fibril arrangement and size distribution in monkey
oral mucosa. Journal of Anatomy, 192: 321-328
Parry DAD. 1988. The molecular and fibrillar structure of collagen and its relationship to the mechanical
properties of connective tissue. Biophysical Chemistry, 29: 195-209
Parry DAD, Barnes GRG, Craig AS. 1978. A comparison of the size distribution of collagen fibrils in
connective tissues as a function of age and a possible relation between fibril size distribution and
mechanical properties. Proceedings of the Royal Society of London B, 203: 305-321
Pawłowska M, Sikorski A. 2013.Monte Carlo study of the percolation in two-dimensional polymer systems.
Journal of Molecular Modeling (DOI 10.1007/s00894-013-1892-y).
Purslow PP, Wess TJ, Hukins DWL. 1998. Collagen orientation and molecular spacing during creep and
stress-relaxation in soft connective tissues. Journal of Experimental Biology, 201: 235-242
R Core Team. 2012. R: A language and enviroment for statistical computing. R Foundation for statistical
computing Vienna, Austria. Available at: http://www.R-project.org/.
Reed CC, Iozzo RV. 2002. The role of decorin in collagen fibrillogenesis and skin homeostasis.
Glycoconjugate Journal, 19: 249-255
Rigozzi S, Müller R, Snedeker JG. 2010. Collagen fibril morphology and mechanical properties of the Achilles
tendon in two inbred mouse strains. Journal of Anatomy, 216: 724-731
Rigozzi S, Stemmer A, Müller R, Snedeker JG. 2011. Mechanical response of individual collagen fibrils in
loaded tendon as measured by atomic force microscopy. Journal of Structural Biology, 176: 9-15
Shannon CE. 1948. A mathematical theory of communication. Bell System Technical Journal, 27: 379-423,
623–656.
Shirazi R, Vena P, Sah RL, Klisch SM. 2011. Modeling the collagen fibril network of biological tissues as a
nonlinearly elastic material using a continuous volume fraction distribution function. Mathematics and
Mechanics of Solids, 16: 707-716
Silver FH, Freeman JW, Devore D. 2001. Viscoelastic properties of human skin and processed dermis. Skin

IAEES www.iaees.org
46 Network Biology, 2014, 4(2): 31-46

Research and Technology, 7: 18-23

Svensson RB, Mulder H, Kovanen V, et al. 2013. Fracture mechanics of collagen fibrils: influence of natural
cross-links. Biophysical Journal, 104: 2476-2484
Teixeira-Filho PF, Rocha-Barbosa O, Paes V, et al. 2001. Ecomorphological relationships in six lizard species
of Restinga da Barra de Marica, Rio deJaneiro, Brazil. Rev Chil Anat, 19: 45-50
Tulli MJ, Abdala V, Cruz FB. 2011. Relationships among morphology, clinging performance and habitat use
in Liolaemini lizards. Journal of Evolutionary Biology, 24: 843-855
Wainwright SA, Biggs WD, Currey JD, Gosline JM. 1976. Mechanical Design in Organisms. Edward Arnolds,
London, UK
Wang JHC. 2006. Mechanobiology of tendon. Journal of Biomechanics, 39: 1563-1582
Zhang G. 2005. Evaluating the viscoelastic properties of biological tissues in a new way. Journal of
Musculoskeletal and Neuronal Interactions, 5: 85-90
Zhang G, Young BB, Ezura Y, et al. 2005. Development of tendon structure and function: regulation of
collagen fibrillogenesis. Journal of Musculoskeletal and Neuronal Interactions, 5: 5-21
Zhang WJ. 2012a. Computational Ecology: Graphs, Networks and Agent-based Modeling. World Scientific,
Singapore
Zhang WJ. 2012b. Modeling community succession and assembly: A novel method for network evolution.
Network Biology, 2(2): 69-78
Zhang WJ. 2012c. Several mathematical methods for identifying crucial nodes in networks. Network Biology,
2(4): 121-126
Zhang WJ. 2013. Network Biology: Theories, Methods and Applications. Nova Science Publishers, New York,
USA

IAEES www.iaees.org
Network Biology, 2014, 4(2): 47-57

Article

In silico prediction of three-dimensional structure and interactome

analysis of Tubulin α subfamily of Arabidopsis thaliana

Jasmin Šutković, Mohamed Ragab Abdel Gawwad

Genetics and Bioengineering Department, International University of Sarajevo, Ilidza, 71220 Bosnia and Herzegovina
E-mail: mragab@ius.edu.ba

Received 29 October 2013; Accepted 3 December 2013; Published online 1 June 2014

Abstract
Microtubules are essential components of cytoskeleton, rigid hollow rods approximately 25 nm in diameter.
Microtubules are dynamic structures being continuously assembled and disassembled within the cell. The
basic building blocks of microtubules are heterodimers of globular α- and β-tubulin subunits In Arabidopsis
thaliana tubulin subunits are encoded by small gene families, six for α-tubulin and nine for β-tubulin.Both α-
and β-tubulin bind GTP, which functions analogously to the ATP bound to actin to regulate polymerization. It
is shown that tubulin alpha forms hydrogen bonds with the GTPase domain of b-tubulin. Multiple sequence
alignment revealed highsimilarity between the family subunits. Due to the missing of three dimensional
structuresin A. thaliana, structural models were predicted and validated. Additionally, protein domains search
revealed that all tubulin α family subunits contain GTPase domain as the tubulin C terminal domain,
confirming previous research. Finally the interactome analysis revealed several interactomes.AtTUA6 shows
strong interaction with embryosac development arrest 10 protein (EDA10), involved in stimulating the
exchange of guanyl nucleotides, enabling the replacement of GDP by GTP in association with a GTPases.

Keywords 3D structure; interactome; microtubules; tubulin α; functional annotation.

Network Biology     
ISSN 2220­8879   
URL: http://www.iaees.org/publications/journals/nb/online­version.asp 
RSS: http://www.iaees.org/publications/journals/nb/rss.xml 
E­mail: networkbiology@iaees.org 
Editor­in­Chief: WenJun Zhang 
Publisher: International Academy of Ecology and Environmental Sciences 

1 Introduction
Microtubules are dynamic cytoskeletal polymers essential for various cell functions such as intracellular
organization, ordered vesicle transport, cell division and establishment of cell polarity. The basic building
blocks of microtubules are heterodimers of globular α- and β-tubulin subunits. They are arranged in a head-to-
tail fashion to form 13 protofilaments that constitute cylindrical microtubules with outer diameter around 25
nm. In Arabidopsis thaliana tubulin subunits are encoded by small gene families, six for α-tubulin (Schröder et
al., 2001) and nine for β-tubulin (Snustad et al., 1992). It has been proposed that the function of microtubules
is modulated by highly diverse posttranslational modifications of tubulin dimers (Blume et al., 2010). The
alpha- and beta-tubulins are the major components of microtubules, while gamma-tubulin plays a major role in

IAEES www.iaees.org
48 Network Biology, 2014, 4(2): 47-57

the nucleation of microtubule assembly. Microtubules are capable of performing various tasks during the life
cycle of eukaryotic cells in spite of ahighly conserved basic structure. The different functions appear to be, at
least in part, mediated by differentialaction of diverse associated proteins, including motor proteins (Schröder
et al., 2001). However, the in vivo threedimensional structure of Tubulin alpha proteins is not resolved. In this
study the in silicon three dimensional structure of all Tubulin α proteins in Arabidopsis thaliana will be
predicted, phylogenetic relationships and Interactome for new functional annotations as well.
Microtubules
Microtubules play central roles in several of the most basic processes of eukaryotic cells: cell division, cell
motility, intracellular transport, and the control of cell shape. In plant cells, rigid cell walls obviate the need for
direct cytoskeletal maintenance of cell shape (Steven et al., 1987). Microtubules of the eukaryotic cytoskeleton
perform essential and diverse functions and are composed of a heterodimer of alpha and beta tubulins. The
genes encoding these microtubule constituents belong to the tubulin superfamily, which is composed of six
distinct families. Genes from the alpha, beta and gamma tubulin families are found in all eukaryotes. The alpha
and beta tubulins represent the major components of microtubules, while gamma tubulin plays a critical role in
the nucleation of microtubule assembly. There are multiple alpha genes, which are highly conserved among
species. This gene encodes alpha tubulin and is highly similar to mouse and rat TubA1 gene (Levilliers et al.,
1998; Jolly et al., 2007).
About 13 of these protofilaments associate in parallel making the microtubule wall, and giving rise to a
polymer with a well-defined polarity. Essential to the function of microtubules is their ability to switch
stochastically between growing and shrinking phases (dynamic instability), a non-equilibrium behaviour of
tubulin that is based on nucleotide binding and hydrolysis. Each tubulin monomer binds one molecule of GTP.
The nucleotide bound to αtubulin, at the so called N-site, is non-exchangeable. The resulting metastable
microtubule structure is thought to be stabilized by a cap of remaining GTP-tubulin subunitsat the ends, the
loss of which results in rapid depolymerization (Lowe et al., 2001).As mentioned above,the alpha tubulin
heterodimer is the structural subunit of microtubules. Tubulin alpha shares minimum 40% amino-acid
sequence identity with tubulin beta , exist in several isotype forms, and undergo a variety of posttranslational
modifications. The structures of alpha and beta tubulins are basically identical: each monomer is formed by a
core of two beta-sheets surrounded by alpha-helices. The monomer structure is very compact, but can be
divided into three regions (domains) based on function: the amino-terminal nucleotide-binding region, binds
GTP , the carboxy-terminal region which probably constitutes the binding surface for motor proteins and an
intermediate taxol-binding region (Nogales et al., 1998).
It is important to note that various isotypes of the two major microtubule subunits, Alpha and beta tubulin,
can occur even within a single organism and that these may interact differentially with the proteins. Tubulin
isotypes comprise post-translational modifications of a common progenitor and/or members of multigene
families (MacRae and Carrie, 1989).In higher plants, as in higher animals, it is now accepted that tubulins are
generally encoded by multigene families. Even in Arabidopsis, with its ‘minimal’ genome, families with at
least six different α and nine β tubulin members have been found (Kopczak et al., 1992; Chu et al., 1993).
Tubulin α subfamily in Arabidopsis thalina
Tubulin α subfamily is the major constituent of microtubules. It binds two moles of GTP, one at an
exchangeable site on the beta chain and one at a non-exchangeable site on the alpha-chain.This
subfamilyincludessixtubulin α proteins, as show in Table 1 (Steven et al., 1987).
Alpha-tubulin 1 (TUA1) is encoded by the tua1 gene in A. thaliana.This protein is primarilyexpressed in
stamens and mature pollen. TUA1 is one of the main constituent of cytoskeleton,involved in microtubule-

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 47-57 49

based process. The location of TUA1 was measure in microtubules cytoskeleton, cytosol and plasma
membrane (Ludwig et al., 1987, 1988).
Alpha-tubulin 2 (TUA2) proteinis also located in cytoskeleton and represent the main structural constituent
of it.This protein is involved in cadmium transport (salt stress), microtubule-based processes and in the
response to salt stress.TUA2 is located in cytoskeleton, cytosol and cell wall and cell membrane and in
chloroplast. TUA2 is expressed 26 plant structure during 15 growth stage (Ludwig et al., 1987).
Alpha-tubulin 3 (TUA3)is involved is the function of cellular response to gravity.Like all other tubulin
alpha proteins it is an important structural part of cytoskeleton.TUA3 is mainly expressed in mature pollen
stage and during the germination stage of pollen. TUA3 shares the location with other tubulin alpha members
(Ludwig et al., 1987; Saito et al., 2003).
Alpha-tubulin 4 (TUA4) functions in structural constituent of cytoskeleton.TUA4 is involved in response
to cadmium ions, microtubule-based process, cellular response to gravity. This protein is located in cytosol,
cell wall, plasma membrane and chloroplast. TUA4 is expressed in 29 plant structures during 15 growth stages
(Ludwig et al., 1987, 1988; Saito et al., 2003).
Alpha-tubulin 5(TUA5), together with all tubulin alpha members, represent an important structural
constituent of cytoskeleton.TUA5 regulates cadmium ion transport and microtubule-based processes.As
atubulin complex it is located in cytosol, cell wall and cell membrane.It is expressed ion 6 plant structures
(Mattingly et al., 2001; Saito et al., 2003; Abe and Hashimoto, 2005).
Alpha-tubulin 6 (TUA6) protein, like other tubulin alpha members it is a structural constituent of
cytoskeleton, involved in response to salt stress, microtubule cytoskeleton organization and cellular response
to gravity. Furthermore, it is involved protein polymerisation (Ludwig et al., 1987; Mattingly et al., 2001;
Saito et al., 2003; Abe and Hashimoto, 2005; Ban et al., 2013).

2 Methodology
The fasta sequences of Tubulin α subfamily proteins (TUA1, TUA2, TUA3, TUA4, TUA5 and TUA6) were
obtained from NCBI databases. The Gene accession numbers from NCBI and TAIR database are shown in
Table 1.

Table 1 Tubulin α proteins and their NCBI accession IDs.

Tubulin α members NCBI accession IDs TAIR IDs
TUA1 NP_176654.1 AT1G64740
TUA2 NP_175423.1 AT1G50010
TUA3 NP_197478.1 AT5G19770
TUA4 NP_171974.1 AT1G04820
TUA5 NP_197479.1 AT5G19780
TUA6 NP_849388.1 AT4G14960

The multiple sequence alignment was performed with the ClustalW2 program (Larkin et al., 2007). Default
parameters were applied and aligned sequences were executed using this software. Phylogenetic tree was
constructed using Maximum likelihood method in Robust Phylogenetic Analysis software, where phylogenetic
relationships between amino acids of TUA subunits were reconstructed and analysed (Dereeper et al., 2008).
In order to predict the three dimensional structure of Tubulin alpha subfamily in Arabidopsis
thaliana,protein data bank (Pdb) files were obtained based on homology modelling using CPH models 3.2

IAEES www.iaees.org
50 Network Biology, 2014, 4(2): 47-57

servers (Nielsen et al., 2010). Additionally to validate the 3-D structures for all TUAs, by testing
stereochemistry quality using Ramachandran plots obtained from RAMPAGE program (Lovell et al., 2003).
The protein’s genetically mobile domain annotations were obtained using Pfam database analysis program
(Finn et al., 2010). For the protein-protein interaction analysis, Arabidopsis interaction viewer was used
(Popescu et al., 2009). Subcellular localization for each protein was predicted and confirmed by WoLF
PSORT program (Horton, et al., 2007).
Arabidopsis Interactions Viewer V.2.0 was used for the protein- protein interaction in order to find the
interactome for all AtRFC subunits (Popescu et al., 2009).

3 Results and Discussion

3.1 Modeling of 3D structure in Tubulin α subfamily
The three-dimensional (3-D) structure details of proteins are of major importance in providing insights into
their molecular functions. 3-D structure of all six subunits showed highly structural similarities. Additionally
outputs have been confirmed by Ramachandran plots (Fig. 1a and 1b). The modelled enzymes are monomers
folded into α/β domains. AtTUA family consists of minimum 15 central stranded β sheet flanked by 11α
helices. The similar structure is shared by other tubulin alpha enzymes from mammalian origins but the
number of helices and β sheets differ from one species to another. The Ramachandran plots in Fig. 1a and 1b
indicate the region of possible angle formations by w (phi) and c (psi) angles. The conventional terms
represent the torsion angles on either side of alpha carbon in peptides. The plot was divided into three regions:
favoured (91.5%), allowed (6%) and outlier (2.5%). This result is significant since the high percentage of
residues in favoured region (90%). This indicates that the built model is of good quality, as shown in Fig. 1a
and Fig. 1b.

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 47-57 51

Fig. 1a 3D modelsand Ramachadran plots of AtTUA1/2/3.

IAEES www.iaees.org
52 Network Biology, 2014, 4(2): 47-57

Fig. 1b 3D structure and Ramachadran plots of AtTUA4/5/6.

3.2 Multiple sequence alignment and phylogenetic tree analysis

Aligned sequences showed substantially significant homology, and constructed phylogenetic tree representing
the already proven relationship between proteins in this family which play their major role as GTPase binding
proteins.
The protein sequence of TUA1, TUA2, TUA3, TUA4, TUA5 and TUA6 were obtained from the NCBI
sequence database (Table 1). The amino acid sequences are highly similar to the sequences of the homologous
human TUA subunits, respectively, and also show amino acid sequence similarity to functionally homologous
proteins of other organisms.

Fig. 2 Tubulin α subfamily phylogenetic tree.

The phytogenic relationships among the five subunits revealed that the two proteins (AtTUA3 and
AtTUA5) are identical and that they diverged from AtTUA1. AtTUA2 and AtTUA4 evolved together with
AtTUA6 with reserved evolutionary relationships (Fig. 2).This confirm the similarity in function of AtTUA
subunits along the sequence alignment results. The sequence alignment scores in the TUA subfamily are
shown in Table 2.

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 47-57 53

Table 2 ClustaW2 alignment result.

SeqA Name Length SeqB Name Length Score
1 AtTUA1 450 2 AtTUA2 450 88.6667
1 AtTUA1 450 3 AtTUA3 450 91.7778
1 AtTUA1 450 4 AtTUA4 450 88.6667
1 AtTUA1 450 5 AtTUA5 450 91.7778
1 AtTUA1 450 6 AtTUA6 450 88.4444
2 AtTUA2 450 3 AtTUA3 450 93.5556
2 AtTUA2 450 4 AtTUA4 450 100.0
2 AtTUA2 450 5 AtTUA5 450 93.5556
2 AtTUA2 450 6 AtTUA6 450 99.5556
3 AtTUA3 450 4 AtTUA4 450 93.5556
3 AtTUA3 450 5 AtTUA5 450 100.00
3 AtTUA3 450 6 AtTUA6 450 93.3333
4 AtTUA4 450 5 AtTUA5 450 93.5556
4 AtTUA4 450 6 AtTUA6 450 99.5556
5 AtTUA5 450 6 AtTUA6 450 93.3333

3.3 Domain search in Pfam database

Pfam domain search shows that all members in the TUA subfamily contains the Tubulin GTPase domain and
the Tubulin C-terminal domain, confirming the sequence alignment results. Tubulin is homologous to FtsZ
protein, responsible for bacterial cell division (alpha-beta tubulin structure). FTsZ and tubulins are GTPases,
having the GTPase domain. FtsZ can polymerise into tubes, sheets, and rings in vitro and is ubiquitous in
bacteria and archaea. FtsZ is found at the C-terminal domain (Nogales et al., 1998).
In all AtTUA the ATPase domain extends from 49 to 246 amino acid residues, while C-terminal domain
extends from 248 to 393 residues, shown in Table 3.
All six subunits show conserved regions characteristic of Tubulin/FtsZ family GTPase binding domain and
the Tubulin C-terminal domain (Nogales et al., 1998).

Table 3 AtTUA family domains.

Tubulin/FTSZfamily, Tubulin c-terminal

Domains
GTPase domain domain
Proteins Start End Start End
TUA1 49 246 248 393
TUA2 49 246 248 393
TUA3 49 246 248 393
TUA4 49 246 248 393
TUA5 49 246 248 393
TUA6 49 246 248 393

IAEES www.iaees.org
54 Network Biology, 2014, 4(2): 47-57

3.4 Subcellular localization

The Subcellular localizations of AtTUAs are analysed with using WoLF PSORT program. Results showed
good and significant reliabilities that all Tubulin alpha members are located in cytoskeleton and/or cytosol.
TAIR software confirmed those localization results (Table 4).

Table 4 AtRFC family localization results.

Protein Subcellular localization
AtTUA1 Cytostol
Cytoskeleton
AtTUA2 Cytostol
AtTUA3 Cytoskeleton
Cytostol
AtTUA4 Cytostol
Cytoskeleton
AtTUA5 Cytostol
Cytoskeleton
AtTUA6 Cytostol
Cytoskeleton

3.5 Interactions
According to data obtained from the Arabidopsis Interactions Viewer (TAIR) we found several interactions
belonging to all six subunits of AtTUA family. The interactome network of TAIR analysis revealed that
AtTUA1, AtTUA4 and AtTUA6 subunits showed strongly interaction with Tubulin folding cofactor B (Fig. 3).
Additionally AtTUA4 and AtTUA6interact with Prefoldin 6 protein and EDA10 protein, whereas only
AtTUA6 interacts with Prefoldin 3 protein.

Fig. 3 AtTUAfamilyinteraction network (only siginifcant interactions are shown).

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 47-57 55

Tubulin cofactor B is involved in embryo development and cell division (Du et al., 2010). Prefoldin 3 and
6, members of the PFD complex, are thought to function along with the TCP ring complex to properly fold
microtubule proteins. Additionally, Prefoldin 3 is shown to play essential role in Arabidopsis tolerance is salt
stress (Rodríguez-Milla and Salinas, 2009). Embryo sac development arrest10 protein (EDA10) is involved in
guanyl-nucleotide exchange factor activity by stimulating of the exchange of guanyl nucleotides associated
with a GTPase (Pagnussat et al., 2005).
The common partner for AtTUA2 and AtTUA4 are tubulin folding factors A, E and B. Tubulin folding
factors are mostly involved in the control of the correct balance of α- and β-tubulin monomers (Victor et al.,
2002). AtTUA3 and AtTUA5 show no significant interactions (Table 5 and Fig. 3).

Table 5 Interactomes of AtRFC subunits.

AtRFCs Interactom ID Interactome name and function
TUA1 EMB2804 Tubulin folding cofactor B
TUA2 ATRBP47C RNA-binding protein 47C
TUA2 ATCHIP Carboxyl terminus of HSC70-interacting protein
TUA2 At3g46220 Unknown name and function
TUA2 TUB6 Tubulin beta-6 chain
TUA4 EDA10 Embryo sac development arrest 10
TUA4 TUB5 Tubulin beta-5 chain
TUA4 PFD6 Prefoldin 6 chain
TUA4 PFI_TFC E Tubulin folding cofactor E Pfifferling (PFI)
TUA4 TFCA Tubulin folding cofactor A (KIESEL)
TUA4 PFD3 Prefoldin 3 chain
TUA4 EMB2804 Tubulin folding cofactor B
TUA6 PFD6 Prefoldin 6
TUA6 PFI_TFC E Tubulin folding cofactor E / Pfifferling (PFI)
TUA6 TFCA KIS_tubulin folding cofactor A (KIESEL)
TUA6 EMB2804 Tubulin folding cofactor B
TUA6 EDA10 Embryo sac development arrest 10
TUA6 TUB6 Tubulin beta-6 chain
TUA6 PFD3 Prefoldin 3

4 Conclusions
Tubulin family is a multi protein complex consisting of several subunits. Tubulin α subfamily is the major
constituent of microtubules. It binds two moles of GTP, one at an exchangeable site on the beta chain and one
at a non-exchangeable site on the alpha-chain. This subfamily includes six tubilin α proteins. The 3D structure
and inteactome analysis confirmed their similarity and functions.Additinally, through the strong interaction
between AtTUA6 and Embryo sac development arrest10 (EDA10), AtTUA may additionaly functions in the
exchange of guanyl nucleotides associated with a GTPase. It was expected that AtTUA1, AtTUA3 and
AtTUA4 interacts with various tubulin coding factors, confirming previous research. Interestingly, AtTUA4
and AtTUA6 are shown to interact with Prefoldin2 and Profoldin 6 protein that supports the proper folding of
microtubules and have important role in Arabidopsis tolerance to salt stress.

IAEES www.iaees.org
56 Network Biology, 2014, 4(2): 47-57

References
Abe T, Hashimoto T. 2005. Altered microtubule dynamics by expression of modified alpha-tubulin protein
causes right-handed helical growth in transgenic Arabidopsis plants. Plant Journal, 43(2): 191-204
Ban Y, Kobayashi Y, Hara T, et al. 2013. Alpha tubulin is rapidly phosphorylated in response to hyperosmotic
stress in rice and Arabidopsis. Plant and Cell Physiology, 54(6): 848-858
Blume Y, Yemets A, Sheremet Y, et al. 2010. Exposure of beta-tubulin regions defined by antibodies on an
Arabidopsis thaliana microtubule protofilament model and in the cells. BMC Plant Biology, 10(29): 1471-
2229
Chu BY et al. 1993. Alteration of P-Tubulin gene expression during low-temperature exposure in leaves of
Arabidopsis thaliana. Plant Physiology, 103: 371-377
Dereeper A, Guignon V, Blanc G, et al. 2008. Robust phylogenetic analysis for the non-specialist. Nucleic
Acids Research, 36(l2): 465-469
Du Y, Cui M, Qian D, et al. 2010. AtTFC B is involved in control of cell division. Frontiers in Bioscience,
1(2): 752-63
Finn RD, Mistry J, Tate J, et al. 2010. The Pfam protein families database: Bateman. Nucleic Acids Research,
38: 211-222
Horton P, Park KJ, Obayashi T, et al, 2007. WoLF PSORT: protein localization predictor. Nucleic Acids
Research, 35: 585-587
Jolly C, Mitar I, Sattentau QJ. 2007. Requirement for an intact T-cell actin and tubulin cytoskeleton for
efficient assembly and spread of human immunodeficiency virus type 1. Journal of Virology, 81(11):
5547-5560
Kopczak SD, Haas NA, Hussey PJ, et al. 1992. The small genome of Arabidopsis contains at least six
expressed alpha-Tubulin genes. Plant Cell, 4: 539-547
Larkin MA, Blackshields G, Brown NP, et al. 2007. Clustal W and Clustal X version 2. Bioinformatics, 23(21):
2947-2948
Levilliers J, Crozet F, Chaib H, et al. 1998. Sequence characterization of a newly identified human alpha-
tubulin gene (TUBA2). Genomics, 47(1): 125-130
Lowe J, Amos LA.1998. Crystal structure of the bacterial cell-division protein FtsZ. Nature, 391: 203-206
Lowe J, Li H, Downing KH, et al. 2001. Downing and E. Nogales. Refined structure of αβ-Tubulin at 3.5 A
resolution. Journal of Molecular Biology, 313(5): 1045-1057
Lovell SC, Davis IW, Arendall WB, et al. 2003. Structure validation by Calpha geometry: phi, psi and Cbeta
deviation. Proteins, 50(3): 437-450
Ludwig SR, Oppenheimer DG, Silflow CD, et al. 1987. Characterization of the alpha-tubulin gene family of
Arabidopsis thaliana. Proceedings of the National Academy of Sciences of USA, 84(16): 5833-5837
Ludwig SR, Oppenheimer DG, Silflow CD, et al. 1988. The α1-tubulin gene of Arabidopsis thaliana: primary
structure and preferential expression in flowers. Plant Molecular Biology, 10(4): 311-321
MacRae TH, Carrie ML. 1989. Tubulin synthesis, structure, and function: what are the relationships?
Biochemistry and Cell Biology, 67(11-12): 770-790
Mattingly KS, Beaty BJ, Mackie RS, et al. 2001. Molecular cloning and characterization of a metal responsive
Chironomus tentans alpha-tubulin cDNA. Aquatic Toxicology, 54(3-4): 249-260
Nielsen M, Lundegaard C, Lund O, et al. 2010. CPHmodels-3.0 - Remote homology modeling using structure
guided sequence profiles. Nucleic Acids Research, 38
Nogales E, Downing KH, Amos LA, et al. 1998.Tubulin and FtsZ form a distinct family of GTPases. Nature
Structural and Molecular Biology, 5: 451-458

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 47-57 57

Nogales E, Sharon GW, Kenneth HD. 1998. Structure of the αβtubulin dimer by electron crystallography.
Nature, 391:199-203
Pagnussat GC,Yu HJ, Ngo QA, et al. 2005.Genetic and molecular identification of genes required for female
gametophyte development and function in Arabidopsis. Development, 132(3): 603-614
Popescu SC, Popescu GV, Bachan S, et al. 2009. MAPK target networks in Arabidopsis thaliana revealed
using functional protein microarrays. Genes and Development, 23: 80-92
Rodríguez-Milla MA, Salinas J.2009. Prefoldins 3 and 5 play an essential role in Arabidopsis tolerance to salt
stress. Molecular Plant, 2(3): 526-534
Saito Y, Soga K, Wakabayashi K, et al. 2003. Increase in expression level of alpha-tubulin gene in
Arabidopsis seedlings under hypergravity conditions. Biological Sciences in Space, 17(3): 177-180
Schröder J, Stenger H, Wernicke W.2001. Alpha-tubulin genes are differentially expressed during leaf cell
developmentin barley (Hordeum vulgare L). Plant Molecular Biology, 45: 723–730
Snustad DP, Haas NA, Kopczakand SD, et al. 1992. The small genome of Arabidopsis contains at least nine
expressed P-tubulin genes. Plant Cell, 4: 549-556
Steven RL, David GO, Carolyn DS, et al. 1987. Characterization of the α tubulin gene family of Arabidopsis
thaliana. Proceedings of the National Academy of Sciences of USA, 84: 5833-5837
Victor K, Paul EG, Jaideep M, et al. 2002. The Arabidopsis Tubulin-folding Cofactor A gene is involved in the
control of the α/β-Tubulin monomer balance. Plant Cell, 14(9): 2265-2276

IAEES www.iaees.org
Network Biology, 2014, 4(2): 58-66

Article

Invasive cancer as an empirical example of evolutionary suicide

Ahmed Ibrahim
44 El-Geish St. Mansoura , Dakahlia, Egypt
E-mail: wetawdt@gmail.com

Received 20 March 2014; Accepted 7 April 2014; Published online 1 June 2014

Abstract
In recent years, a large portion of the literature has focused on evolutionary suicide. Which is about the
extinction of population via highly fit invasive mutation possessing a strategy favored by selection. ''Darwinian
extinction'' or evolutionary suicide it is one of the most important findings in adaptive dynamics but
unfortunately remains in need of more empirical examples. On the other hand, much literature has been
published on somatic evolution and how carcinogenesis is an evolutionary process caused by
mutation/selection, and how it competes on resources and space, and evades predation. Therefore, its micro-
environment reflects most of ecological interactions in independent organisms. Today, there are many
mathematical models describe this unique case of somatic evolution and show invasion fitness of tumor Clone
cells as an unbeatable strategy leading to normal cell extinction and evolutionary stable strategy (ESS). When
we combine the studies in these two lines of research, we almost inevitably arrive at the conclusion that
evolutionary theory falls short of adequately explaining the phenomenon of life in its fullness and complexity.
This is due to the fact that when we look at cancer as an empirical example for evolutionary suicide, we find
that the latter is not a rare or special case and that it can occur in the most common ecological conditions.
Therefore, it can be argued that in the absence of a mechanism for preventing evolutionary suicide, there will
be no adequate explanation for life.

Keywords Darwinian extinction adaptive dynamics; tragedy of the common; population dynamics; somatic
evolution; invasion fitness; evolution theory; tumor progression.

Network Biology     
ISSN 2220­8879   
URL: http://www.iaees.org/publications/journals/nb/online­version.asp 
RSS: http://www.iaees.org/publications/journals/nb/rss.xml 
E­mail: networkbiology@iaees.org 
Editor­in­Chief: WenJun Zhang 
Publisher: International Academy of Ecology and Environmental Sciences 

1 Introduction
In simple terms, evolution as a theory aims to explain the phenomenon of life in its complexity and diversity
via basic mechanisms such as genetic mutation and natural selection. According to the theory, such
mechanisms are necessary and sufficient for life to occur as we know it. Logically speaking, a sufficient cause
necessarily entails the occurrence of its result. So if it happens that the mutation-selection binary does not
ensure the occurrence of its result, then it is not sufficient, and therefore is not a sound explanation of life.

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 58-66 59

According to EPP Susanna (2010), in other words, to say “r is a sufficient condition for s” means that the
occurrence of r is sufficient to guarantee the occurrence of s.”
No one can deny the existence and impact of mutation/selection; however whether they are necessary and
sufficient for explaining the existence of biological systems or not is a different question (Zhang, 2012).
Moreover, what are we to do if it turns out that such mechanisms at the individual-population level may drive
populations to extinction instead of survival? The eventuality of self-extinction, formally known as
“evolutionary suicide”, has been confirmed in adaptive dynamic (Metz, 1996; Geritz, 1998; Dieckmann and
Law, 1996; Geritz, 1997). According to Gyllenberg and Parvinen (2001), evolutionary suicide is an
evolutionary process where a viable population adapts in such a way that it can no longer persist. Evolutionary
suicide has been discovered in models of life-history evolution (Matsuda and Abrams, 1994; Ferrière, 2000;
Gyllenberg et al., 2002 ;Webb, 2003; Parvinen, 2010 ; Rousset and Ronce, 2004) and in various other models
(Zayed and Packer, 2005; Dercole et al., 2006; Hedrick, et al., 2006; Parvinen, 2007; Gandon and Day, 2009)
In fact, evolutionary biologist Haldane has observed that animals and plants are not quite such ruthlessly
efficient strugglers as they would be if Darwinism were the whole truth…it does not pay a species to be too
well adapted. A variation making for too great efficiency may cause a species to destroy its food and starve
itself to death. It should be noted therefore that any gains that may be benefited from mutation-selection in the
population will inevitably perish by one mutant drives population to evolutionary suicide.
Evolutionary suicide occurs because of evolution (mutation/selection) favors the genotype allele of
invasive mutant whose phenotypic trait has higher adaptive fitness landscape, its adaptation progress to the
point that a population no longer can exist. Fitness, furthermore, is defined by the ability to both survive and
reproduce at the individual and group-level, even if we had to consider kin selection, group selection, and
inclusive fitness (Maynard Smith, 1964; Hamilton, 1987).The model of evolutionary suicide follows the
modern interpretation associated with Fisher's fundamental theory of natural selection (Parvinen and
Dieckmann, 2013; Frank and Slatkin, 1992; Okasha, 2008).The occurrence of evolutionary suicide is a
potentially possible event owing to the fact that mutant invasion fitness drives the population to extinction,
even if we consider all of the above mentioned conceptions and principles. In fact, evolutionary suicide may
occur even under the ideal situation of "Optimizing Selection" (Parvinen and Dieckmann, 2013). Moreover,
evolutionary processes tend to promote suicide instead of preventing it, and when it occurs it will cause all of
the adaptive benefits that it had once been acquired to disappear. Hence, we cannot accept a theory for life
which not only fails to prevent evolutionary suicide but also promotes it. Our questions are
(1) Is there an empirical example of evolutionary suicide?
(2) Is evolutionary suicide a particular case which occurs in very rare ecological conditions for mutant invasion
fitness or is it possible through evolutionary processes in the most common ecological conditions and
interactions known thus far?
(3) Are there any solutions which the theory of evolution can provide for this problem?

2 Is there an empirical example of evolutionary suicide?

The answer is 'yes'; among them are:
(1) Single-cell organisms (Fiegna and Velicer, 2003)
(2) Multi-cellular organisms (Rankin and Kokko, 2006 ; Muir and Howard, 1999)
(3) Virulence parasite (De Roode et al., 2005)
(4) Plants (Gersani et al., 2001)
(5) Still, the present paper proposes a novel example represented in the case of invasive cancer, which in turn
possess three important advantages:

IAEES www.iaees.org
60 Network Biology, 2014, 4(2): 58-66

(a) Cancer arises directly from evolutionary mechanisms (i.e., mutation, selection, genetic drift) thus
supporting the results of the mathematical models which argue that evolutionary theory mechanisms cannot
prevent the potential of evolutionary suicide.
(b) Cancer is the single empirical example of how evolution scenario of life on earth can occur (Casás-Selves
and DeGregori, 2011; Merlo et al., 2006). Arguing that it is an empirical example of evolutionary suicide, one
may plausibly argue that evolutionary suicide occurs at the heart of this scenario and therefore must not be
ignored.
(c) Cancer has many and various ecological interactions (identical to the same ecosystem of independent
organisms). This supports the main aim of this article, namely that evolutionary theory happens to be an
insufficient explanation for the phenomenon of life because it fails to prevent evolutionary suicide. This
becomes the more evident in the light of earlier evidence that it is not a rare or particular case but can occur in
the most common ecological conditions.

3 Invasive malignant cancer is an evolutionary process and an empirical example of evolutionary suicide
3.1 Cancer is an evolutionary process
Since Peter Nowell has described tumor progress as an evolutionary fashion (Nowell, 1976) and many studies
have supported it, evolutionary biologists believe cancer reflects species evolution (Casás-Selves and
DeGregori, 2011).
The neoplasm is genetically and epigeneticall heterogeneous. Inside it, the population of individual cells
competes for space and resources. It evades predation by the immune system and can even cooperate, disperse
and colonize new organs (Merlo et al., 2006).
3.1.1 Mutations
Cancer arises from genetic mutation such as mutations activating ontogenesis and inactivating tumor
suppressor genes (Gatenby and Vincent, 2003a; Roth et al. 2001; Maley et al., 2006; Gonzalez-Garcia et al
2002; Frank and Nowak, 2004; Ibrahim et al., 2011) and epigenetic mutations such as hypermethylation
(Weisenberger et al., 2006; Horie-Inoue and Inoue, 2006).
3.1.2 Genetic drift
Genetic drift plays a role in the co-existence of various colonies in the neoplasm and in neutral mutations
fixation (Merlo et al., 2006).
3.1.3 Natural selection
Natural selection (NS) will favor neoplasm colonies because of the effective fitness strategies characterizing
cancer as one of its major hallmarks (Hanahan and Weinberg, 2010):
(1) Sustaining proliferative signaling
(2) Evading growth suppressors
(3) Resisting cell death
(4) Enabling replicative immortality
(5) Inducing angiogenesis
(6) Activating invasion and metastasis
(7) Reprogramming of energy metabolism
(8) Evading immune destruction
Moreover, natural selection will also favor neoplasm colonies due to their genetic instability. Hence,
cancer appears to invite all the conditions necessary for natural selection to act upon it. According to Ridley
(1996): (a) reproduction. Entities must reproduce to form a new generation; (b) heredity. The offspring must
tend to resemble their parents; (c) variation in the individual characters among the members of the population;

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 58-66 61

(d) variation in the fitness of organisms according to the state they have for a heritable character.
3.1.4 Ecological interactions
(1) Competition: to compete on resources of nutrition and space, every competitor will have a negative effect
on the others (Caignard et al., 1985; Miller et al 1980; Guba et al., 2001).
(2) Amensal: this has to do with partial competition having a partial effect as when a colony stimulates an
immune response against its competitors only (Heppner et al., 1983).
(3) Predation: a tumor may use mechanisms which help it escape immune system predation (Seliger, 2005).
(4) Parasitism: when a colony benefits at the expense of other ones (Rundhaug, 2005; Nagy, 2004).
(5) Commensalism: when a colony increases other colonies' fitness without payoff (Heppner and Miller, 1998;
Jouanneau et al., 1994).
(6) Mutualism: cooperation among all that benefits all (Axelrod et al., 2006; Ishiguro et al., 2006; Fukino et al.,
2004; Mueller and Fusenig, 2004; Tlsty, 2001).
Ultimately, one finds that most of the important and common ecological interactions among independent
organisms in the biological system are mirrored or present in cancer colonies with their micro-environments.
3.2 Invasive cancer is an empirical example of evolutionary suicide
A malignant cancer in the invasive phase (Vineis and Berwick, 2006) is a phenotype that has evolved by
mutation/selection and one which possesses invincible strategies, (local) maximum adaptive fitness landscape,
and evolutionary stable strategy (Gatenby and Vincent, 2003b). It therefore invades the population, leading to
extinction of normal cells via competition (Gatenby and Gawlinski, 1996; Vineis and Berwick, 2006; Gatenby
and Vincent, 2003b; Kareva, 2011a, b). Eventually, this leads to self-extinction following the destruction of
resources (death of cancer patient). One may therefore argue that invasive cancer is an empirical example of
evolutionary suicide. Contrary to Darwin's belief in natural selection “According to Darwin (1859) Natural
selection will never produce in a being any structure more injurious than beneficial to that being, for natural
selection acts solely by and for the good of each. ” Evolutionary suicide shows that adaptive traits can harm the
population and drive it towards whole extinction. The same problem persists in neo-Darwinism, even when
one takes into account 'inclusive fitness'; defined as (Bryden, 2007), the expected reproductive success of a
trait/organism due to its phenotype and the frequency of the trait/related organisms in the population”. We can
see cancer evolutionary dynamics as selfish behavior that prevents cooperation between cells in a tissue
(Franziska Michor, 2005). Therefore, the mathematical dynamic models, which describe the invasion that
follows from the selfish strategies, also describe the tumor progress and vice versa. As a result of competition
between cooperator and defector strategies, Parvinen refers to what he calls the ''Kamikaze-mutant" case in this
model (Parvinen, 2010). I think this case of evolutionary suicide fully applies to tumor cells in the invasive
phase since it appears to bear all the advantageous fitness traits unique to cancer. It is (ESS) and causes
extinction of normal cells and the others genotypes colonies surrounding it. Below are some quotations from
five important studies describing the ecological dynamics of the tumor progress (the competition among
normal and invasive cancer, adaptive fitness landscape of normal and invasive cancer):
3.2.1 It should be noted that Kareva (2011a) showed cancer as a case of evolutionary suicide in this model, it
“eventually entering the domain of attraction of the stable equilibrium of another, larger game, which can lead
to evolutionary suicide ? Now glycolytic cells that have become numerous enough are cooperating, jointly
increasing the toxicity of the surrounding microenvironment, and becoming more efficient competitors as a
group, eventually killing the host and consequently killing themselves. In the model, this is captured through
introduction of the additional toxicity term that captures increased mortality of aerobic cells proportional to the
amount of lactic acid secreted by glycolytic cells. Indeed, one can observe that the cell population initially
grows; peaks and then eventually collapses, going to extinction (see Figure 7). So, the either equilibrium

IAEES www.iaees.org
62 Network Biology, 2014, 4(2): 58-66

within the same game of prisoner's dilemma can become attracting not because of the changes in payoffs for
each cell but due to different initial composition of the population of players, which happens solely through
natural selection”.
3.2.2 According to Gatenby and Vincent (2003b), the tumor population can evolve to an ESS, driving the
normal cells to extinction, resulting in an invasive cancer.
3.2.3 According to Gatenby and Vincent (2003b), only after evolving strategies that allow for increased
substrate uptake will the system come to equilibrium with complete destruction of all normal cells, i.e., an
invasive cancer. Because its strategy is at a maximum on its adaptive landscape, the cancer population cannot
be dislodged through the introduction of other mutant strategies (Derived from same G-function) this
prediction of a decline in somatic evolution and, therefore, decreased Cellular heterogeneity in an established
invasive cancer has been observed.
3.2.4 According to Gatenby and Vincent (2003b), a genotype allowed to evolve to an ESS will invade and
destroy adjacent non-ESS populations. Furthermore cellular strategy at an ESS it will have no further pressure
to evolve at least within the current microenvironment. Thus, the transition from a premalignant state to an
invasive cancer should be accompanied by a transition from heterogeneous cellular population to one that is
comparatively homogeneous.
3.2.5 According to Vineis and Berwick (2006), the effect of normal cells over cancer cells becomes negligible.
3.2.6 According to Gatenby and Gawlinski (1996), for aggressively invasive cancers there is no region of
coexistence between tumor and normal tissues and therefore the Lotka-Volterra competition has no effect on
the structure and dynamics of the tumor-host interface.
3.2.7 According to Kareva (2011b), another consequence of tumor heterogeneity is the possibility of so-called
evolutionary suicide —in their quest for higher growth rates, lower death rates, and increased competitiveness
and with their ability to migrate out and colonize distant organs, cancer cells defy “cooperation” with somatic
tissue, eventually killing the host and thus killing themselves. This evolutionary experiment is run within each
cancer patient, sometimes leading to cancer cells committing evolutionary suicide at the expense of the host.
From an ecological perspective, one can look at this process as an attempt of new species (cancer cells), which
have different metabolic and reproductive strategies compared with the “resident” population (somatic cells) to
invade a new habitat (tissue). Successful invasion will result in the formation of a primary solid tumor.

4 Is evolutionary suicide a particular case that occurs in rare ecological conditions or is it possible
through evolutionary processes in the most common ecological conditions and interactions that we know?
In the light of the ecological interactions of our empirical example (cancer), (i.e. Competition, Amensal,
Predation, Parasitism, Commensalism, and Mutualism.), it can be argued that evolutionary suicide may indeed
occur in the most important and common ecological conditions and interactions of life biological system

5 Are there any solutions which the theory of evolution can provide for this problem?
This problem is similar to the other problem known as "the tragedy of the common" (Hardin, 1968) and which
is defined as (Rankin et al., 2007): a situation in which the selfish actions of individuals result in the complete
collapse of the resource over which they are competing. It thus appears that the solution heavily requires a
great deal of ''intelligence'' in order to solve the dilemma and protect the biological system. According to
Rankin et al. (2007), the lesson drawn from these studies is that solving the dilemma often requires negotiation
and sanctions on disobedient individuals. This Changes the payoffs, so that group-beneficial behavior also
becomes optimal for the individual” Kin selection, group selection, and so on may explain how natural
selection opts for certain useful strategies such as cooperative and altruistic behaviors but cannot prevent a

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 58-66 63

phenotype which is fit and which has actually been favored by selection.

6 Discussion
6.1 Are there differences between cancer evolution in single organism and independent organism's
evolution?
Yes, there are differences and asexual reproduction may be the major one. However, this should not negatively
impact the findings of this paper since cancer was cited as an empirical example for evolutionary suicide in the
various and most common ecological conditions and interactions. Moreover, there are other empirical
examples for evolutionary suicide in sexual reproduction (Rankin and Kokko, 2006 ; Muir and Howard, 1999).
In spite of such differences, cancer still remains the only empirical example of how the evolutionary scenario
of life on earth may occur.
6.2 Conclusion
Since the occurrence of evolutionary suicide in most populations and various ecological conditions is
theoretically (adaptive dynamics) and empirically (cancer) possible, and since such an occurrence leads to full
population- extinction, then every extant population at the cellular and organismal level is a phenomenon
which no theory may sufficiently explain without accounting for a mechanism which has the capacity to
prevent evolutionary suicide. Since the theory of evolution does not account for such a mechanism, then it can
be argued that the theory is inherently insufficient as an explanation of the phenomenon of life.

References
Axelrod R, Axelrod DE, Pienta KJ. 2006. Evolution of Cooperation among tumor cells. Proceedings of the
National Academy of Sciences of USA, 103: 13474-13479
Caignard A, Martin MS, Michel MF, et al. 1985. Interaction between two cellular subpopulations of a Rat
colonic carcinoma when inoculated to the Syngeneic host. International Journal of Cancer, 36(2): 273-279
Darwin C. 1859. The Origin of Species by means of Natural Selection.
Dercole F, Ferrière R, Gragnani A, et al. 2006. Coevolution of slow–fast populations: evolutionary sliding,
evolutionary pseudo-equilibria and complex Red Queen dynamics. Proceedings of the Royal Society
B:Biological Sciences, 273(1589): 983-990
De Roode et al. 2005. Virulence and competitive ability in genetically diverse malaria infections. Proceedings
of the National Academy of Sciences of USA, 102(21):7624-7628
Dieckmann U, Law R. 1996. The dynamical theory of coevolution: a derivation from stochastic ecological
processes. Journal of Mathematical Biology, 34(5-6): 579-612
EPP Susanna S. 2010. Discrete Mathematics with Applications (4th Edition). DePaul University, Chicago,
USA
Ferrière R. 2000. Adaptive Responses To Environmental Threats: Evolutionary Suicide, Insurance, And
Rescue. Options Spring 2000, IIASA, Laxenburg, Austria
Fiegna F, Velicer GJ. 2003. Competitive fates of bacterial Social parasites: persistence and self-induced
extinction of Myxococcus Xanthus cheaters. Proceedings of the Royal Society B: Biological Sciences, 270:
1527-1534
Frank SA,Nowak MA. 2004. Problems of somatic Mutation and cancer. Bioessays, 26(3): 291-299
Frank S, Slatkin M. 1992. Fisher's fundamental theorem of natural selection. Trends in Ecology and Evolution,
7(3): 92-95

IAEES www.iaees.org
64 Network Biology, 2014, 4(2): 58-66

Franziska Michor. 2005. Evolutionary dynamics of cancer. PhD thesis, Department of Organismic and
Evolutionary Biology, Harvard University, USA
Fukino K, et al. 2004. Combined total genome loss of Heterozygosity scan of breast cancer stroma and
Epithelium reveals multiplicity of stromal targets. Cancer Research, 64: 7231-7236
Gandon S, Day T. 2009. Evolutionary epidemiology and the dynamics of adaptation. Evolution, 63(4): 826-
838
Gatenby RA, Vincent TL. 2003a. Application of quantitative models from population biology and evolutionary
game theory to tumor therapeutic Strategies. Molecular Cancer Therapeutics, 2: 919-927
Gatenby RA,Vincent TL. 2003b. An evolutionary model of carcinogenesis. Cancer Research, 63: 6212-6220
Gatenby RA, Gawlinski ET. 1996. A reaction-diffusion model of cancer invasion. Cancer Research, 56: 5745-
5753
Geritz SAH, Kisdi E, Meszena G, et al. 1998. Evolutionary singular strategies and the adaptive growth and
branching of the evolutionary tree. Evolutionary Ecology, 12: 3537
Geritz SAH, Metz JAJ, Kisdi E, et al. 1997. the dynamics of adaptation and evolutionary branching. Physical
Review Letters, 78: 2024-2027
Gersani M, et al. 2001. Tragedy of the commons as a result of root competition. Journal of Ecology, 89: 660-
669
Gonzalez-Garcia I, Sole RV, Costa J. 2002. Metapopulation dynamics and spatial heterogeneity in cancer.
Proceedings of the National Academy of Sciences of USA, 99: 13085–13089
Guba M, et al. 2001. A primary tumor promotes dormancy Of solitary tumor cells before inhibiting
angiogenesis. Cancer Research, 61: 5575-5579
Gyllenberg M, Parvinen K, Dieckmann U. 2002. Evolutionary suicide and Evolution of dispersal in structured
metapopulations. Journal of Mathematical Biology, 45: 79-105
Gyllenberg M, Parvinen K. 2001. Necessary and sufficient conditions for evolutionary suicide. Bulletin of
Mathematical Biology, 63: 981-993
Haldane JBS. 1985. On Being the Right Size and Other Essays (Smith J, ed). Oxford University Press, UK
Hamilton WD. 1987. Discriminating nepotism: expectable, common and overlooked. In: Kin Recognition in
Animals (Fletcher DJC, Michener CD, eds). Wiley, New York, USA
Hanahan D, Weinberg RA. 2011. Hallmarks of Cancer: The Next Generation. Cell, 144: 646-674
Hardin G. 1968. The tragedy of the commons. Science, 162: 1243-1248
Hedrick PW, Gadau J, Page Jr RE. 2006. Genetic sex determination and Extinction. Trends in Ecology and
Evolution, 21: 55-57
Heppner GH, Miller BE, Miller FR. 1983. Tumor Subpopulation interactions in neoplasms. Biochimica et
Biophysica Acta, 695: 215-226
Heppner G, Miller F. 1998. The cellular basis of tumor progression. International Review of Cytology, 177: 1-
56
Horie-Inoue K, Inoue S. 2006. Epigenetic and proteolytic Inactivation of 14–3-3sigma in breast and prostate
cancers. Seminars in Cancer Biology, 16: 235-239
Ibrahim SS, Eldeeb MAR, Rady MAH. 2011. The role of protein interaction domains in the human cancer
network. Network Biology, 1(1): 59-71
Ishiguro K, Yoshida T, Yagishita H, et al. 2006. Epithelial and stromal genetic instability contributes to
genesis of colorectal adenomas. Gut, 55: 695-702
John Bryden. 2007. The evolution of social organisms: modelling reproduction strategy. PhD Thesis.
Univerisity of Leeds, UK

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 58-66 65

Jouanneau J, Moens G, Bourgeois Y, et al. 1994. A minority of carcinoma cells producing acidic fibroblast
growth factor induces a community: Effect for tumor progression. Proceedings of the National Academy
of Sciences of USA, 91: 286-290
Kareva I. 2011a. Prisoner's dilemma in cancer metabolism. PLoS ONE, 6(12): e28576.
doi:10.1371/journal.pone.0028576
Kareva I. 2011b. What can ecology teach us about cancer. Translational Oncology, 4(5): 266-270
Maley, C. C. et al. (2006) Genetic clonal diversity predicts Progression to esophageal adenocarcinoma. Nat
Genet. 38, 468–473
Mark Ridley. 1996. Evolution (2nd edition). 71-72, Blackwell, Oxford, UK
Matias Casás-Selves, James DeGregori. 2011. How cancer shapes evolution and how evolution shapes cancer.
Evo Edu Outreach, 4: 624-634
Matsuda H, Abrams PA. 1994. Timid consumers: self-extinction due to adaptive change in foraging and anti-
predator effort. Theoretical Population Biology, 45: 76-91
Maynard SJ. 1964. Group selection and kin selection. Nature, 201: 1145-1147
Merlo LMF, Pepper JW, Reid BJ, et al. 2006. Cancer as an evolutionary and ecological process. Nature
Reviews Cancer, 924-935
Metz JAJ, Geritz SAH, Meszéna G, et al. 1996. Adaptive dynamics, a geometrical study of the consequences
of nearly faithful reproduction. In: Stochastic and Spatial Structures of Dynamical Systems (VanStrien SJ,
VerduynLunel SM, eds). 183-231, North-Holland, Amsterdam, Netherlands
Muir WM, Howard RD. 1999. Possible ecological risks of transgenic organism release when transgenes affect
mating success: sexual selection and the trojan gene hypothesis. Proceedings of the National Academy of
Sciences of USA, 96: 13853-13856.
Mueller MM, Fusenig N E. 2004. Friends or foes bipolar Effects of the tumor stroma in cancer. Nature
Reviews Cancer, 4: 839-849
Nagy JD. 2004. Competition and natural selection in a Mathematical model of cancer. Bulletin of
Mathematical Biology, 66: 663-687
Nowell PC. 1976. The clonal evolution of tumor cell populations. Science, 194: 23-28
Okasha S. 2008. Fisher's fundamental theorem of natural selection-a philosophical analysis. Britain Journal for
the Philosophy of Science, 59(3): 319-351
Parvinen K. 2010. Adaptive dynamics of cooperation may prevent the coexistence of defectors and cooperators
and even cause extinction. Proceedings of the Royal Society B: Biological Sciences, 277: 2493-2501
Parvinen K. 2007. Evolutionary suicide in a discrete-time metapopulation model. Evolutionary Ecology
Research, 9: 619-633
Parvinen K, Dieckmann U. 2013. Self-extinction through optimizing selection. Journal of Theoretical Biology,
333: 1-9
Rankin DJ, Bargum K, Kokko H. 2007. The tragedy of the commons in evolutionary biology. Trends in
Ecology and Evolution, 22(12): 643-651
Rankin DJ, Kokko H.(2006. Sex, death and tragedy. Trends in Ecology and Evolution, 21: 225-226
Roth MJ, et al. 2001. Genetic progression and Heterogeneity associated with the development of Esophageal
squamous cell carcinoma. Cancer Research, 61: 4098-4104
Rousset F, Ronce O. 2004. Inclusive fitness for traits affecting metapopulation demography. Theoretical
Population Biology, 65: 127-141
Rundhaug JE. 2005. Matrix metalloproteinases and angiogenesis. Journal of Cellular and Molecular Medicine,
9: 267-285

IAEES www.iaees.org
66 Network Biology, 2014, 4(2): 58-66

Seliger B. 2005. Strategies of tumor immune evasion. BioDrugs, 19: 347-354

Tlsty T. 2001. Stromal cells can contribute oncogenic signals. Seminars in Cancer Biology, 11: 97-104
Vineis P, Berwick M. 2006. the population dynamics of cancer: a Darwinian perspective. International Journal
of Epidemiology, 35(5): 1151-1159
Webb C. 2003. A complete classification of Darwinian extinction in ecological interactions. American
Naturalist, 161: 181-205
Weisenberger DJ, et al. 2006. CpG island methylator Phenotype underlies sporadic microsatellite instability
and is tightly associated with BRAF mutation in Colorectal cancer. Nature Genetics, 38: 787-793
Zayed A, Packer L. 2005. Complementary sex determination substantially Increases extinction proneness of
haplodiploid populations. Proceedings of the National Academy of Sciences of USA, 102: 10742-10746
Zhang WJ. Computational Ecology: Graphs, Networks and Agent-based Modeling. World Scientific,
Singapore, 2012

IAEES www.iaees.org
Network Biology, 2014, 4(2): 67-73

Article

Comparative structural analysis of HAC1 in Arabidopsis thaliana

Amar Ćemanović, Jasmin Šutković, Mohamed Ragab Abdel Gawwad

Genetics and Bioengineering department, International University of Sarajevo, Ilidza, 71220 Bosnia and Herzegovina
E-mail: mragab@ius.edu.ba,mrgawad@hotmail.com

Received 30 July 2013; Accepted 4 September 2013; Published online 1 June 2014

Abstract
Histone acetylation is an important posttranslational modification correlated with gene activation. In
Arabidopsis thaliana, the histone acetyltransferase 1 (AtHAC1) is homologous to animal p300/CREB (cAMP-
responsive element-binding protein)-binding proteins, which are the main histone acetyltransferases
participating in many physiological processes, including proliferation, differentiation, and apoptosis. In this
study the 3-D structure of the HAC1 protein in Arabidopsis thaliana was predictedusing 4 homology-based
prediction servers: ESyPred3D, 3D-JIGSAW, SWISS-MODEL and PHYRE2. The homology modeled
structureswere evaluated and stereochemical analysis done by Ramachadran plot analysis. The amino acid
sequences of Arabidopsis thaliana HAC1protein are highly similar to the sequence of the homologous human
p300/CREB. SWISS MODEL and Phyre2 servers computed the identical 3D structures. Validation and
verification methods, using Z-score and 3D-1D score, showed that these 3D models are of good and acceptable
quality.

Keywords Arabidopsis thaliana; HAC1 protein; 3-D structure; structure prediction, homology.

Network Biology     
ISSN 2220­8879   
URL: http://www.iaees.org/publications/journals/nb/online­version.asp 
RSS: http://www.iaees.org/publications/journals/nb/rss.xml 
E­mail: networkbiology@iaees.org 
Editor­in­Chief: WenJun Zhang 
Publisher: International Academy of Ecology and Environmental Sciences 

1 Introduction
Arabidopsis thaliana, also known as thale-cress or mouse-ear cress, or simply Arabidopsis,is a small flowering
plant belonging to the Brassicaceae family, a plant family comprising species such as radish and cabbage.
Although not of agronomic importance, it is a widely known plant due to its use as a model organism in
various genetic and plant biology studies. In addition, it is the first plant that had its genome sequenced. The
prominent features of Arabidopsis are its small genome (135 Mb), rapid life cycle and ease of cultivation in
laboratory conditions (The Arabidopsis Genome Innovative, 2000).
In cells of eukaryotic organisms, including plants, the DNA is associated with histone proteins to form
structures called nucleosomes. In nucleosomes DNA is wound around a group of eight histones, which makes
the region inaccessible to transcription factors and therefore transcriptionally inactive. Histone
acetyltransferases (HATs) are a group of enzymes which function in transcriptional regulation by means of
histone modification. Specifically, they transfer acetyl groups from acetyl-CoA to specific residues in histone

IAEES www.iaees.org
68 Network Biology, 2014, 4(2): 67-73

tails, thereby lowering the affinity of histones for DNA and, accordingly, making the DNA more accessible to
the transcription machinery. The process of acetylation is reversible, with the deacetylation being performed
by deacetylase enzymes (Sterner and Berger, 2000).
Based on sequence similarity, histone acetyltransferases have been organized into four families: GNAT,
MYST, p300/CBP and TAFII250. The p300/CBP family is especially important since it acetylates not only
histones, but also various other proteins, including CREB, p53, Stat3 and HIV-1 Tat protein (Sterner and
Berger, 2000).This protein family in Arabidopsis thalianaincludes the following members: HAC1, HAC2,
HAC4, HAC5 and HAC12 (Pandey et al., 2002).
From the members of the p300/CBP family, HAC1 has the highest homology with the human and animal
p300/CREB protein (Deng et al., 2007). Although the human p300/CREB is well characterized and has an
experimentally determined crystal structure, little is known about its homolog in Arabidopsisthaliana. It has
been found that it is located in the nucleus and exhibits, besides its acetyltransferase activity, also transcription
cofactor activity. Furthermore, it is found to acetylate primarily histones H3 and H4 (Earley et al., 2007).
The aim of this paper is to predict the best 3D model for the HAC1 protein in Arabidopsis thaliana, using
homology modeling techniques by model comparison and evaluation of the predicted structures.

2 Materials and Methods

The protein sequence and information (Name and Origin, Protein attributes, General annotation and Entry
information) of HAC1 was retrieved from the National centre of Biotechnology information (NCBI), in
FASTA format, with the accession number NP_565197. Additional information were taken from the
Arabidopsis Information Resource (TAIR), with the identification number AT1G79000. Secondary Structure
Prediction of HAC1 protein was done by GOR IV server prediction server, based on the information theory
(Garnier et al., 1996). There is no defined decision constant. GOR IV uses all possible pair frequencies within
the window of 17 amino acid residues.
The protein sequence was subjected for comparative homology modelling via SWIS-MODEL Server
(Schwede et al., 2003) and PHYRE2 server (Kelley LA and Sternberg MJE, 2009). SWISMODEL software
assured a better control of the homology modelling process and promoted a deeper understanding of various
features of the protein structure meant to be modeled.Phyre2 is a major update to the original Phyre server with
a range of new features, accuracy is improved, using the alignment of hidden Markov models via HH search to
significantly improve accuracy of alignment and detection rate (Jefferys et al., 2010).For additional structural
and comparative analysis, ESyPred3D (Cristophe et al., 2002) and 3DJIGSAW (Bates et al., 2001) online
servers predicted the 3D structure of HAC1.
Finally, once the 3D structures were generated, structural evaluation and stereo-chemical analysis was
performed using different evaluation and validation tools. Backbone conformation of all models was evaluated
by analysis of Psi/Phi Ramachandran plot using RAMPAGE program (Lovel et al., 2002). Energy calculations
on a protein chain were performed by atomic empirical mean force potential ANOLEA (Melo et al., 1997).
The program performs energy calculations on a protein chain, evaluating the "Non- Local Environment" (NLE)
of each heavy atom in the molecule. A plot obtained from the Verify3D (Eisenberg, D., 1997) structure
Evaluation Server that represented the average3D-1D profile score. Verify3D analyzes the compatibility of an
atomic model (3D) with its own amino acid sequence (1D).

3 Results and Discussion

The most successful general approach for predicting the structure of proteins involves the detection of
homolog’s of known three-dimensional (3D) structure—the so-called template-based homology modeling or

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 67-73 69

fold-recognition. Protein 3D structure is very important in understanding the protein functions, interactions and
localizations (Denise Chen, 2009). In this study, for Arabidopsis thaliana histone acetyltransferase 1 protein ,
with 1697 amino acids, the best homolog found in all modeling programs, is the human p300 histone
acetyltransferase (HAT), with protein data bank ID ‘3BIY’. The best homolog found in both modelers,
Phyre2and SWISS modeler, with 45.8% sequence similarity, is a human p300 histone acetyltransferase (HAT).
HAT1 regulates gene expression by acetylating histones and other transcription factors (Xu et al., 2001).
The secondary structural analysis of the protein was done and random coil was found to be most frequent
(54.63%), followed by alpha helix (28.87%). Extended strand (Ee) was found to be least frequent (16.50%)
(Data not shown).The dominance of the coiled regions indicates the high level of conservation and stability of
the protein structure (Neelamathi et al., 2009).The predicted models were visualized by PyMol software, in the
mode of cartoon view, with the lines removed.

A B

Fig. 1 Predicted models of HAC1 protein by (A)ESyPred3D and (B)3D-JIGSAW servers.

A B

A B
Fig. 2 Predicted models of HAC1 protein by (A) Phyre2and (B) SWISS-MODEL servers.
A B
IAEES www.iaees.org
70 Network Biology, 2014, 4(2): 67-73

Benchmark of 4 different homology modeling programs: 3DJIGSAW, Swiss-Model, EasyPred3D and

Phyre2, has been performed used to transform the alignment to a 3D model. The difference between the
programs is how the information contained in the alignment is used to build a 3D model.
One of the most frequently used modelling programs; SWISS-MODEL generates the 3D models from a
small number of rigid bodies, obtained from the core of the aligned regions (Blundell et al., 1987). The final
three-dimensional (3D) structure was then confirmed by the Phyre2 server, an automatic fold recognition
server for predicting the structure and the function of proteins based on homology modeling (Lawrence, 2011).
Errors of the model are usually estimated either from the energy of the model or from theresemblance of a
given characteristic of the model to real structures (Sippl, 1995). Predicted models of HAC1 (Figs. 1-4) are
evaluated by ANOLEA and Verify3D tools (Table 1).

Table 1 Model evaluation of proteins by using ANOLEA and Verfy3D tools.

High
Modeling Evaluation 3D-1D
Protein energyzone Z-score
program program profile
(%)
ANOLEA 88.27% 18.04 -
ESyPred3D
Verify3D - - 0.68
ANOLEA 51.67% 6.15 -
3DJIGSAW
Verify3D - - 0.72

AtHAC1 SWISS ANOLEA 36.95% 3.51 -

MODEL Verify3D - - 0.71
ANOLEA 36.95% 3.51 -
PHYRE2
Verfy3D - - 0.71

Verify3D method assesses protein structures using three-dimensional profiles. This program analyzes the
compatibility of an atomic model (3D) with its own amino acid sequence (1D). Each residue is assigned a
structural class based on its location and environment (alpha, beta, loop, polar, apolar etc).The scores ranges
from -1 (bad score) to +1 (good score) (Eisenberg, 1997). ANOLEA program performs energy calculations on
a protein chain, evaluating the "Non- Local Environment" (NLE) of each heavy atom in the molecule (Melo et
al., 1997).ANOLEA results (Tab.1) are shown as High energy (%) and pseudo Z-score. High energy zone
(HEz) in the protein profile correlates with errors in the structure. Z-score is represents as pseudo energies of
target protein sequences, where lower Z-score means higher reliability and vice versa.
Energy calculations performed by ANOLEA, gave the Z-score of 3.51 in 3D models generated by Phyre 2
and SWISS-MODEL server, indicating same quality of the models. While the HAC1 models obtained from
EasyPred3D and 3DJIGSAW gave 18.04 and 6.15 Z-scores, two to six fold higher scores compared to SWISS
Models and Phyre2 servers. The high Z-score obtained from EasyPred3D server is caused by the specific
alignment method, obtained by combining, weighting and screening the results of several multiple alignment
programs. Therefore the final three-dimensional structures is built using the modeling package MODELLER,
which is not done in this study. 3D-1D profile scores, obtained from Verify3D, of HAC1 models is 0.71
obtained from Phyre 2 and SWISS-MODEL servers. EasyPred3D and 3DJIGSAW homology modeler resulted
in 0.72 and 0.68 respectively.
The 3D structures generated in SWISS-MODEL and Phyre2 servers are identical, having the same number

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 67-73 71

of α helixes and β sheets. Z-score of SWISS-MODEL and Phyre2 modeler is much smaller compared to
EasyPred3D and 3DJIGSAW server. Furthermore, the 3D-1D score of 0.72 (maximum is 1) indicates a good
quality 3D model, compared to the human homolog model. Ramachandran plots analysis, generated by the
RAMPAGE program (Lovel et al., 2002) for all four 3D models, shows all residues in the most favorable
regions and disallowed regions (Fig. 3).

A
BB

D
C

Fig. 3 A. Ramachandran plot of the ESyPred3D HAC1 model; B. Ramachandran plot obtained from the3D-JIGSAWserver;
C. Ramachandran plot from the SWISS-MODEL server; D. Ramachadran plot of the PHYRE2HAC1 model.

The number of residues in favored region in the ESyPred3D 3D model of HAC1 is 73.0% (Fig. 3A).
Ramachandran plot analysis for 3DJIGSAW HAC1 model (Fig. 3B), resulted in 83.9% of all residues placed
in favored regions. These two results confirm the model check analysis of Verify3D and ANOLEA, indicating
that 3DJIGSAW and ESyPred3D servers predicted the models with insufficient residues in the expected
favored region. However, SWISS-MODEL and PHYRE2 servers were capable to produce a good 3D structure
of Arabidopsis thaliana HAC1 protein, having 95% of all residues located in expected regions in
Ramachandran plot analysis (Fig. 3 C and D).

IAEES www.iaees.org
72 Network Biology, 2014, 4(2): 67-73

Although the high similarity of 45.8 with thep300/CBP HAT homolog from humans , with a well predicted
X-ray crystallography structure and known function, it is difficult to attribute some the functional
characteristics to our predicted AtHAC1(Arabidopsis histone acetyltransferase of the CBP family) just
according the structural similarities, that are stated in this paper.
The p300/CBP HAT protein from humans, with the sequence length of 2414 amino acids, is a bigger
protein than HAC1 in Arabidopsis thaliana, with 1697 amino acids.
For future research, structural prediction and conformations of other Arabidopsis thaliana members
p300/CBP histone acetyltransferase proteins, this HAC1 structural analysis could be used to find out the other
proteins that not yet have been identified in this pathway. Furthermore, interactome analysis of all HAC
member of the CBP family is recommended in order to assign new functions to this family.

4 Conclusion
An attempt was made to predict the three-dimensional structure of Arabidopsis histone acetyltransferase 1 of
the CBP family. The structural analysis approach was based on homology modeling, using SWISS-MODEL,
Phyre2, ESyPred3D and 3DJIGSAW model servers. For 3D model validations, Verify3D and ANOLEA tools
have been used. Additionally, Ramachandran plot analysis was performed in order to verify whether all
residues of our predicted models lye in the most favorable regions. SWISS MODEL and Phyre2 servers
computed the identical 3D structure. Validation and verification methods, using Z-score and 3D-1D score,
showed that these 3D models are of good and acceptable quality.

References
Arabidopsis Genome Initiative. 2000. Analysis of the genome sequence of the flowering plant Arabidopsis
thaliana. Nature, 408(6814): 796-815
Arnold K, Bordoli L, Kopp J, et al. 2006. The SWISS-MODEL Workspace: A web-based environment for
protein structure homology modelling. Bioinformatics, 22: 195-201
Bates PA, Kelley LA, MacCallum RM, et al. 2001. Enhancement of Protein Modelling by Human Intervention
in Applying the Automatic Programs 3D-JIGSAW and 3D-PSSM. Proteins: Structure, Function and
Genetics, Suppl(5): 39-46
Blundell TL, Sibanda BL, Sternberg MJ, et al. 1987. Knowledge-based prediction of protein structures and the
design of novel molecules. Nature, 326: 347-352
Chen D. 2009. Structural GENOMICS: Exploring the 3D Protein Landscape. Simbios, the NIH National
Center for Physics-Based Simulation of Biological Structures,11-18
Deng W, Liu C, Pei Y, et al. 2007. Involvement of the histone acetyltransferase AtHAC1 in the regulation of
flowering time via repression of FLOWERING LOCUS C in Arabidopsis. Plant Physiology, 143: 1660-
1668
Earley KW, Shook MS, Brower-Toland B, et al. 2007. In vitro specificities of Arabidopsis co-activator histone
acetyltransferases: implications for histone hyperacetylation in gene activation. The Plant Journal, 52: 615-
626
Eisenberg D, Luthy R, et al. 1997. VERIFY3D: assessment of protein models with three-dimensional profiles.
Methods in Enzymology, 277: 396-404
Garnier J, Gibrat JF, Robson B. 1996. GOR secondary structure prediction method version IV. Methods in
Enzymology, 266: 540-553
Greer J. 1990. Comparative Modeling Methods: Application to the Family of the Mammalian Serine Proteases.

IAEES www.iaees.org
 Network Biology, 2014, 4(2): 67-73 73

Proteins, 7: 317-334
Kelley LA, Sternberg MJE. 2009. Protein structure prediction on the web: a case study using the Phyre
server. Nature Protocols, 4: 363-371
Kelley LA, Bennett-Lovsey R, Herbert A, et al. 2011. Phyre: Protein Homology/analogY Recognition Engine.
Structural Bioinformatics Group, Imperial College, London, UK
Lovell SC, Davis IW, Arendall III WB, et al. 2002. Structure validation by Calpha geometry: phi,psi and Cbeta
deviation. Proteins: Structure, Function and Genetics, 50: 437-450
Melo F, Feytmans E. 1998. Assessing protein structures with a non-local atomic interaction energy. Journal
of Molecular Biology, 2775: 1141-1152
Neelamathi E, Vasumathi E, Bagyalakshmi S, et al. 2009. Insilico prediction of structure and functional
aspects of a hypothetical protein of Neurosporacrassa. Journal of Cell and Tissue Research, 93: 1889-1894
Pandey R, Müller A, Napoli CA. 2002. Analysis of histone acetyltransferase and histone deacetylase families
of Arabidopsis thaliana suggests functional diversification of chromatin modification among multicellular
eukaryotes. Nucleic Acids Research, 30(23): 5036-5055
Raghava GPS. 2002. APSSP2 : A combination method for protein secondary structure prediction based on
neural network and example based learning. CASP5. A-132
Sippl MJ. 1995. Knowledge based potentials for proteins. Current Opinion in Structural Biology, 5: 229-235
Söding J, Biegert, A, Lupas AN. 2005 TheHHpred interactive server for protein homology detection and
structure prediction. Nucleic Acids Research, 33(Web Server issue): W244-8
Sterner DE, Berger SL. 2000. Acetylation of Histones and Transcription-Related Factors. Microbiology and
Molecular Biology Reviews, 64(2): 435
The PyMOL Molecular Graphics System, Version 1.3. Schrödinger, LLC
Xu W, Chen H, Du K, et al. 2001. A transcriptional switch mediated by cofactor methylation. Science, 294:
2507-2511

IAEES www.iaees.org
 Network Biology
The Network Biology (ISSN 2220-8879) is an open access, peer-reviewed online journal that
considers scientific articles in all different areas of network biology. It is the transactions of the
International Society of Network Biology.It dedicates to the latest advances in network biology. The
goal of this journal is to keep a record of the state-of-the-art research and promote the research work
in these fast moving areas. The topics to be covered by Network Biology include, but are not limited
to:

 Theories, algorithms and programs of network analysis

 Innovations and applications of biological networks
 Ecological networks, food webs and natural equilibrium
 Co-evolution, co-extinction, biodiversity conservation
 Metabolic networks, protein-protein interaction networks, biochemical reaction networks,
gene networks, transcriptional regulatory networks, cell cycle networks, phylogenetic
networks, network motifs
 Physiological networks
 Network regulation of metabolic processes, human diseases and ecological systems
 Social networks, epidemiological networks
 System complexity, self-organized systems, emergence of biological systems, agent-based
modeling, individual-based modeling, neural network modeling, and other network-based
modeling, etc.

We are also interested in short communications that clearly address a specific issue or completely
present a new ecological network, food web, or metabolic or gene network, etc.

Authors can submit their works to the email box of this journal, networkbiology@iaees.org and(or)
wjzhang@iaees.org. All manuscripts submitted to Network Biology must be previously unpublished
and may not be considered for publication elsewhere at any time during review period of this journal.

In addition to free submissions from authors around the world, special issues are also accepted. The
organizer of a special issue can collect submissions (yielded from a research project, a research group,
etc.) on a specific topic, or submissions of a conference for publication of special issue.

Editorial Office: networkbiology@iaees.org

Publisher: International Academy of Ecology and Environmental Sciences

Address: Flat C, 23/F, Lucky Plaza, 315-321 Lockhart Road, Wanchai, Hong Kong
Tel: 00852-6555 7188
Fax: 00852-3177 9906
E-mail: office@iaees.org
Network Biology
ISSN 2220-8879
Volume 4, Number 2, 1 June 2014

Articles

Fibrillar organization in tendons: A pattern revealed by percolation

characteristics of the respective geometric network
Daniel Andrés Dos Santos, María Laura Ponssa, María José Tulli, et al. 31-46

In silico prediction of three-dimensional structure and interactome analysis of

Tubulin α subfamily of Arabidopsis thaliana
Jasmin Šutković, Mohamed Ragab Abdel Gawwad 47-57

Invasive cancer as an empirical example of evolutionary suicide

Ahmed Ibrahim 58-66

Comparative structural analysis of HAC1 in Arabidopsis thaliana

Amar Ćemanović, Jasmin Šutković, Mohamed Ragab Abdel Gawwad 67-73

The International Academy of Ecology and Environmental Sciences (IAEES) is a nonprofit and registered
international organization. It devotes to promote global ecology and environmental sciences and protect global
ecological environments, by publishing scientific publications, conducting research activities, launching
environmental programs, disseminating knowledge and technologies, sponsoring conferences, and providing
information and discussion spaces, etc. The responsibility for these publications rests with the International
Academy of Ecology and Environmental Sciences.

IAEES
http://www.iaees.org/