Emergent Dynamics of Thymocyte

Development and Lineage Determination

Sol Efroni1*, David Harel2, Irun R. Cohen3
1 National Cancer Institute Center for Bioinformatics, National Institutes of Health, Rockville, Maryland, United States of America, 2 Department of Computer Science,
Weizmann Institute of Science, Rehovot, Israel, 3 Department of Immunology, Weizmann Institute of Science, Rehovot, Israel

Experiments have generated a plethora of data about the genes, molecules, and cells involved in thymocyte
development. Here, we use a computer-driven simulation that uses data about thymocyte development to generate an
integrated dynamic representation—a novel technology we have termed reactive animation (RA). RA reveals emergent
properties in complex dynamic biological systems. We apply RA to thymocyte development by reproducing and
extending the effects of known gene knockouts: CXCR4 and CCR9. RA simulation revealed a previously unidentified
role of thymocyte competition for major histocompatability complex presentation. We now report that such
competition is required for normal anatomical compartmentalization, can influence the rate of thymocyte velocities
within chemokine gradients, and can account for the disproportion between single-positive CD4 and CD8 lineages
developing from double-positive precursors.

Citation: Efroni S, Harel D, Cohen IR (2007) Emergent dynamics of thymocyte development and lineage determination. PLoS Comput Biol 3(1): e13. doi:10.1371/journal.pcbi.
0030013

real-time interactive manipulation of the action. RA allows

Introduction
the experimenter to extract statistical and local information
The mammalian thymus receives stem cells from the bone from the running simulation. Moreover, the experimenter
marrow. These cells—thymocytes—go through a series of can intervene in the simulation and observe in silico the
anatomical subcompartments in a process termed T cell effects of thought experiments. Our simulation was built
education [1,2]. About 97% of candidate T cells die, while the primarily using for the first tier the language of Statecharts
remaining 3% are essential to the continuing development of [8], with its enhanced legibility and organizational structure.
the adaptive immune system [3]. For a recent review of thymic We added to Statecharts a second tier of animation.
architecture and cell traffic, see [4]. For a brief schematic Between the two tiers, we built a set of tools to facilitate
animation of thymic maturation, see Video S1. data-mining options, such as tracking cells, manipulation of
Extensive research in disparate disciplines has uncovered a surface receptors, inducement of apoptosis, tracking cell
mass of data regarding thymocyte development. Subfields of ancestry, data displays, visualization of chemokine gradients,
thymus research include genes, gene expression and differ- zooming in and out across scales, streaming reports,
entiation; molecules (integrins, chemokines, cytokines, re- statistical data, and more (see Figure 1B). Demonstrations
ceptors, antigens, and other ligands); cells (stem cells, of these tools are in Videos S2–S4 and online. The
thymocytes, epithelial cells, dendritic cells, and macrophages); references to papers we used for the database are listed in
cell behavior (adhesion, migration, and anatomic localiza- Text S1.
tion); cell states (differentiation states, cell cycle, prolifer- RA differs fundamentally from other approaches directed
ation, and apoptosis); and physiology (antigen expression, to network modeling [9–11] or to simulating the micro-
positive and negative selection, lineage choice, and antigen- architecture of the immunological synapse or membrane [12–
receptor repertoires). The technologies used to study 14]. We did not simulate membrane data in the present work
thymopoeisis include genetics, transgenes and gene knock- due to the lack of quantitative data. Another spatially based
outs, protein chemistry, microscopy and immunohistochem- system has been developed [15], and thus far it has been
istry, in vitro cell cultures and interactions, in vivo applied to molecular signaling [16].
phenotypes, cell and organ transfers, immunizations, and
more. A systematic integration of these data into an accurate Editor: Howard Petrie, Scripps Gerontology Center, United States of America
and comprehensive representation is much needed. We
Received July 27, 2006; Accepted December 11, 2006; Published January 26, 2007
address this need using reactive animation (RA) to reveal
multiscalar emergent properties and to guide experimenta- A previous version of this article appeared as an Early Online Release on December
11, 2006 (doi:10.1371/journal.pcbi.0030013.eor).
tion in thymocyte development.
RA is a computational approach to simulating complex Copyright: Ó 2007 Efroni et al. This is an open-access article distributed under the
dynamic systems. The technology of RA has been described terms of the Creative Commons Attribution License, which permits unrestricted
use, distribution, and reproduction in any medium, provided the original author
elsewhere [5–7]. Briefly, the RA simulation is built in two and source are credited.
tiers. The first tier is built, bottom-up, from the actual Abbreviations: CMJ, cortico–medulary junction; DN, double-negative; DP, double-
cellular and molecular data, and incorporates the program, positive; MHC, major histocompatibility complex; RA, reactive animation; SCZ,
the logic, and the dynamics of the simulation. The second subcortical zone; SP, single-positive
tier is a front-end visualization of the simulation capable of * To whom correspondence should be addressed. E-mail: [email protected]

PLoS Computational Biology | www.ploscompbiol.org 0127 January 2007 | Volume 3 | Issue 1 | e13
 Emergent Dynamics in the Thymus

Author Summary strates that the simulated thymic lobule faithfully produces
the fine anatomical relationships of real thymic structure;
Biological systems are the embodiment of complexity that defies thousands of thymocytes, individually computed, localize, as
intuitive understanding. Biologists have accumulated masses of data seen in histological sections, to particular anatomical sites
about the molecules, cells, and discrete interactions that compose according to twelve distinct developmental stages—color-
living systems, but the list of facts alone cannot explain how such coded in the legend box.
systems work dynamically. We have developed a hybrid, computa- Figure 2 shows the migratory path of a single thymocyte.
tional approach to the simulation of complex systems called reactive
Both the emerging anatomy and the emerging path are
animation (RA). RA uses a bottom-up integration of diverse
experimental data to create an integrated and dynamic representa- faithful to experimental results [2]. Video S1 and Video S3
tion of the system’s interacting cells and molecules. RA is faithful to show dynamic versions of RA figures. These results demon-
experimental fact, while it plays out the action in animated formats strate that the molecular data in hand suffice to generate the
directly accessible to the eye and mind. Most importantly, RA is macroscale thymus, and that RA can reveal this cross-scalar
analytical, interactive, and allows experimentation in silico. Here, we emergence.
use RA to reveal unexpected emergent properties of thymocyte
development. In particular, we now report that competition Role of CXCR4 in Thymocyte Development
between thymocytes for sites of stimulation could be important in The effects on thymus fine anatomy of chemokine
generating the fine anatomy of the thymus, in selecting for receptors CXCR4 [19] and CCR9 [20] have been studied
thymocytes with a range of migration velocities, and in explaining experimentally, so we could compare an RA simulation of
the paradox of CD4 to CD8 T cell lineage ratios. This study highlights knocking out these receptors to the experimental results.
the explanatory power and the potential aid to experimentation
Targeted gene deletion of CXCR4 [19] resulted in failed
offered by an animated, interactive simulation of complex sets of
data. cortical localization and developmental arrest. Figure 3A
(left) shows the thymic lobule as it was captured under the
microscope [19] and as it is captured during simulation
(right). In both cases, the thymocytes do not respond to
Results CXCR4 stimulation; both Figure 3 panels show that thymo-
Thymus Anatomy Emerges from Molecular Data cyte development gets hung up close to the cortico–medulary
Data at the level of single cells and their microenvironment junction (CMJ) in the DN1 stage (labeled red). Figure 3B
culled from hundreds of papers were coded to the simulation. shows the same time frame and anatomical section in a wild-
Anatomic localization is critical to thymus development; type thymus. Note that double-positive (DP) cells (blue cells in
thymocytes at different developmental stages migrate to the simulation) have spread into the cortex. Unlike the static
specific thymic compartments [2]. Validation of a bottom-up histology of the experimental model, RA provides a dynamic
simulation such as RA (and of the database itself) requires representation.
that the microscale molecular data put into the model suffice
Role of CCR9 in Thymocyte Development
to generate realistic macroscale thymocyte migration and
In contrast to the CXCR4 knockout, deleting CCR9 had no
anatomical location. Cell migration depends on thymocyte
major effect experimentally on intrathymic T cell develop-
receptor profiles, chemokine gradients, epithelial cells, cell
ment [20]. However, competitive transplantation experiments
proliferation, cell survival, cell velocity, and other factors that revealed that bone marrow cells from CCR9/ mice were less
enter the simulation. For example, a thymocyte at the double- efficient in repopulating the thymus of lethally irradiated
negative (DN)1 stage expresses a profile of surface markers Rag-1/ mice than were bone marrow cells from littermate
CD4CD8CD25CD44þLselectinlowCD69 [17]. Experimen- CCR9þ/þ mice [18,20].
tally, we know that thymocytes at the DN1 stage migrate The RA simulation results, presented in Figure 4 and in
towards the chemokine CCL25 [18]. The thymus stroma, too, Videos S1–S4, show both the influence of the lack of
influences migratory decisions; a thymocyte’s path to a response to CCL25, the chemokine ligand of CCR9/, and
chemokine may be blocked by another cell. Furthermore, the outcome of competition between CCR9/ and wild-type
the chemokine itself is involved in two dynamic processes: cells. Figure 4A shows the normal thymus at the same time
first, specific regions of the thymus continuously secrete the point as the altered thymus that appears in Figure 4B; the
chemokine, and second, the chemokine diffuses over time abnormal cells are coded gray here. Figure 4 is animated in
and space. Thus, the migration of a thymocyte continuously Video S3, where the upper panel shows the wild-type
changes as a function of secretion and diffusion of chemo- phenotype and the lower panel shows the CCR9/
kines and the current locations of other migrating thymo- phenotype: the CCR9/ cells congregate around the
cytes and stationary stromal cells. This and much additional subcortical zone (SCZ). Thymocytes at the transition from
information is included in the simulation. Figure 1 demon- DN to DP would normally migrate towards the chemokine

Figure 1. A Snapshot of the Simulated Thymus at Run Time

(A) Different colors stand for different stages during thymocyte selection. The legend box shows the different developmental stages and their
corresponding colors.
(B) Different icons beneath the figure stand for different tools for the control and the visualization of the simulation. The tools include the ability to
zoom into parts of the visualization, show and hide the two tiers of the simulation, highlight interacting cells, display or hide chemokine gradients,
display statistical information through real-time communication with Matlab, visualize apoptotic levels in real-time, initiate and stop a backtrack of a
cell’s migration, and more.
(A,B) In the lower right hand side, the elapsed time is displayed, together with a utility that enables the control of time progression.
doi:10.1371/journal.pcbi.0030013.g001

PLoS Computational Biology | www.ploscompbiol.org 0128 January 2007 | Volume 3 | Issue 1 | e13
 Emergent Dynamics in the Thymus

PLoS Computational Biology | www.ploscompbiol.org 0129 January 2007 | Volume 3 | Issue 1 | e13
 Emergent Dynamics in the Thymus

Figure 2. A Trace of the Path of a Single Thymocyte from Its Birth via Proliferation at the SCZ to Its Position as a Mature Cell in the Medulla
We generate the visual trace by utilizing the built-in tracing tool. The trace is color-coded: the current time is highlighted in red and the beginning of
the trace is enriched with blue. Intermediate times are marked as mixtures of these two colors on the trace line.
doi:10.1371/journal.pcbi.0030013.g002

CCL25 and enter the cortex, but the CCR9/ cells cannot Thymocyte Competition Influences Cell Apoptosis and
respond to this chemokine and are blocked. The blocked Velocity
thymocytes, however, can still move randomly, and popula- Thymocytes need to traverse developmental niches; thus,
tion pressure pushes them away from the SCZ, so that some when the number of thymocytes exceeds the space available
of them reach their next developmental checkpoint—the for antigen presentation sites on epithelial cells, the
cortical epithelial cells—passively. These fortunate cells can thymocytes pile up and those waiting their turn for
then mature into their next developmental stage and stimulation may undergo apoptosis from the lack of
migrate towards the medulla (via a different chemokine), interaction [21,22]. RA makes it possible to study the function
where they can further mature (depending on further of competition by modifying interaction time constants, as
selection events) into fully functional single-positive (SP) shown in Video S3. The results indicate that competition is
cells. RA discloses these dynamics, surmised from static essential to normal thymic development. Figure 6A shows the
experimental histology alone. normal pattern of apoptosis that occurs in the cortex in the
competition-enabled thymus. Figure 6B, in contrast, shows
Interthymocyte Competition Influenced by CCR9 that an abnormal pattern of apoptosis develops in a thymus
RA also made it possible to observe the dynamics in silico free of cell competition; here, most thymocytes die of
of a competitive experiment, in which equal numbers of negative selection in the medulla, rather than in the cortex.
CCR9/ and wild-type cells are seeded into the thymus: The RA simulation suggests that the waiting times for
Video S2 shows that the wild-type cells survive and mature in interactions with cortical epithelial cells constitute a bottle-
much higher numbers. RA makes it possible to quantify the neck that is a factor in normal thymus development.
ratios between mature wild-type cells and mature CCR9/ RA in silico experimentation suggests that competition also
cells (Figure 5): we can see an initial peak of maturing wild- selects for differential speeds of trafficking in response to
type cells, followed by a decrease and an eventual asymptotic chemokine gradients. Figure 7 shows that faster thymocytes
ratio, as the buildup of random pressure of CCR9/ cells enjoy greater chances of survival, at least up to a point.
eventually generates homeostasis. This competition has not Nevertheless, some thymocytes that are relatively slower may
avoid the negative selection suffered by some of their more
yet been performed experimentally, but RA simulation
speedy brothers. Thus, competition selects for a range of cell
predicts the outcome shown in Figure 5. Figure 5 shows an
velocities, and not only for a uniformly high velocity. How
asymptotic value of four wild-type thymocytes to every
selection for a range of T cell velocities might enhance
CCR9/ thymocyte. The dynamics of the asymptote and
defense against invaders [23] as well as for body maintenance
the final value are our predictions if such an experiment was
[24] needs to be investigated.
to be performed. A critical point evident from Figure 5 is the
overwhelming advantage that wild-type cells have immedi- Competition for Niche Determines Lineage Commitment
ately after seeding the thymus. Such a marked effect should Another prediction emerging from cell competition relates
be easy to witness experimentally. to lineage commitment. A developing thymocyte must choose

PLoS Computational Biology | www.ploscompbiol.org 0130 January 2007 | Volume 3 | Issue 1 | e13
 Emergent Dynamics in the Thymus

Figure 3. Comparing a Knockout to a Wild-Type Thymus

(A) The left panel shows a figure borrowed from [19] of the effect of a CXCR4 knockout. The right panel was taken from the simulation. Red-colored cells
in this right panel correspond to DN cells. The thymocytes in the experimental system, marked red fluorescence, remain very close to the CMJ, where
they first entered the thymus, which is situated at the lower right corner of the plate.
(B) A wild-type thymus (labeled lck[Cre]) shows thymocytes scattered throughout the thymic lobule both in the experimental plate from [19] on the
right, and in its in silico equivalent on the left. The blue cells correspond to DP cells. DN cells and DP cells spread throughout the thymus in a normal
fashion.
doi:10.1371/journal.pcbi.0030013.g003

whether to become an SP CD4 T cell (helper) or an SP CD8 T T cell receptor–MHC interaction instructs the T cell. The
cell (cytotoxic). The decision-making process is obscure ‘‘stochastic’’ approach proposes that SP CD4 or CD8
because mature SP CD4 and CD8 T cells evolve from thymocytes are ‘‘randomly’’ generated, and are later selected
precursors that are DP for both CD4 and CD8, yet CD4 cells according to their functional performance with the MHC.
predominate at a 2:1 ratio. Current theories of lineage These theories attempt to found lineage choice on its
commitment deal with the molecular details of the choice. molecular components aimed at showing where exactly,
The two most significant themes in the theories distinguish during development, the cell chooses its lineage [25–27].
between an ‘‘instruction’’ approach and a ‘‘stochastic’’ However, the emergence of competition between thymo-
approach [25]. The ‘‘instruction’’ approach proposes that cytes for interaction space provides a novel solution to the
the productive interaction of a T cell receptor with a CD4:CD8 2:1 paradox. If the dissociation rates of CD8 cells
particular major histocompatability complex (MHC) mole- from epithelial cells are lower than those of CD4 cells, then
cule, class I (for CD8) or class II (for CD4) as the case may be, the CD8 cells will remain longer at their epithelial-cell
rewards the thymocyte and induces a genetic choice to interaction stations (peptide-MHC I sites). As long as a CD8
differentiate to the CD8 or CD4 phenotype. The more fitting thymocyte lingers at a peptide-MHC 1 niche, this niche is

PLoS Computational Biology | www.ploscompbiol.org 0131 January 2007 | Volume 3 | Issue 1 | e13
 Emergent Dynamics in the Thymus

Figure 4. The Effect of Knocking Out CCR9 in Thymocytes

(A) Shows the distribution of wild-type cells (color coded as in Figure 1).
(B) Shows the altered distribution of the knockout cells (colored gray). The normal cells and the altered cells are colored differently to be compatible
with Video S2, where we show the results of a competition experiment between the different cell types. The major differences are in the abilities of
thymocytes to migrate from the SCZ into the cortex, after maturation from pre-DP stages. This altered behavior is responsible for the diluted numbers
of thymocytes in the cortex of the knockout thymuses.
doi:10.1371/journal.pcbi.0030013.g004

unavailable for other, competing CD8 thymocytes. CD8 slower than the dissociation rate of CD4 thymocytes. A
thymocytes, we propose, do not compete with the CD4 relatively greater avidity of CD8 cells for epithelial cell niches
thymocytes, because CD4 thymocytes compete among them- (by 1.7–3.3) would generate the observed lineage predom-
selves for stimulation by interacting with peptide–MHC II inance of CD4 T cells.
stations on epithelial cells. We tested the outcome on lineage
frequency of simulating different dissociation rates for Discussion
interactions between epithelial cells and CD4 and CD8
thymocytes. The results are shown in Figure 8. It can be seen RA analysis of thymocyte development sheds new light on
that about two-thirds of thymocytes will mature into CD4 T the dynamic relationship between molecules and cells in
cells and one-third into CD8 T cells (the de facto ratio) when generating the structure and function of the thymus organ.
the dissociation rate of CD8 thymocytes is 1.7 to 3.3 times First, we can see that the existing body of data, however

PLoS Computational Biology | www.ploscompbiol.org 0132 January 2007 | Volume 3 | Issue 1 | e13
 Emergent Dynamics in the Thymus

Figure 5. Competition In Silico between CCR9/ Cells and Wild-Type Cells

The three panels at the bottom, with context lines leading to the time-graph, show the ratios between the two cell populations developing over time.
An initial peak of maturing wild-type cells is followed by a decrease and an eventual asymptotic ratio, as the buildup of random pressure of CCR9/
cells eventually generates homeostasis. An asymptotic value of four wild-type thymocytes to every CCR9/ thymocyte is reached. See text for further
discussion.
doi:10.1371/journal.pcbi.0030013.g005

discrete and piecemeal, can be integrated by RA simulation epithelial cells. RA simulation also suggests that the absence
into a representation of the functional anatomy of the of thymic output resulting from CXCR4 inhibition can be
thymus seen in histologic sections. What we know about cells attributed to the nonmigratory behavior of cells entering the
and molecules can indeed account for what we see; the thymus.
macroscale organ emerges from the microscale mass of data Third, the visualization of cell dynamics through RA
in hand. In this regard, RA can be said to validate the provides a view of emergent physiology. Although the
database. Note, however, that classical histologic sections are thymus is packed full of cells, the existence of competition
two-dimensional slices of a three-dimensional organ frozen in among thymocytes for space and stimulation has not been a
time; RA simulation adds the dimension of time—dynamics— subject for experimentation or even discussion; competition
and so shows us the formative power of the dynamic flux of is simply not seen in static histologic sections. Since
cells, molecules, and interactions that give rise to the higher- competition was not recorded in the database, we did not
scale organ. In another project involving a different organ, we explicitly program competition into our model. Never-
are currently extending RA simulations to accommodate the theless, cell competition emerged before our very eyes as
third dimension in space; hopefully, the added complexity of we witnessed, via RA, the animated struggle between
the representation will enhance our understanding of the individual thymocytes for productive interactions with
biology. thymic epithelial cells. In silico manipulation of various
Second, RA simulation offers novel explanations for the parameters suggested that thymocyte competition might
observed outcomes of experimental intervention. In our case, function as an important factor in three emergent proper-
for example, RA simulation suggested that the lack of ties of T cell maturation: the functional anatomy of the
phenotype observed in mice with CCR9 knocked out thymus, the selection of thymocytes with a range of
(CCR9/) might be explained by dynamic compensation migratory velocities, and the relative preponderance of SP
through population pressure. RA simulation also explains the CD4 T cells. Obviously, these suggestions require exper-
competitive growth advantage enjoyed by wild-type cells over imental validation. Irrespective of the outcome, however, the
CCR9/ cells. Indeed, overexpression of CCR9 on thymo- animation arm of RA, in providing a higher-scale view of
cytes leads to an in vivo phenotype that can be explained by complex emergent properties [28], can alert us to new
RA as an untimely attraction of the thymocytes by cortical questions for experimentation. Ultimately, we would like to

PLoS Computational Biology | www.ploscompbiol.org 0133 January 2007 | Volume 3 | Issue 1 | e13
 Emergent Dynamics in the Thymus

Figure 6. The Influence of Competition on the Pattern of Apoptosis in the Thymus

We visualize the different levels of apoptosis by using different colors to show the relative numbers of apoptosing cells across the thymic grid. Red
zones correspond to higher levels of apoptosis.
(A) Shows the normal distribution of apoptosis primarily to the SCZ; some apoptosis is seen in the outer cortex (OC) and some in the medulla (M); the
inner cortex (IC) and CMJ show relatively fewer apoptotic cells.
(B) Shows the influence of removing competition between thymocytes for developmental niches. The lack of competition moves the bulk of apoptosis
from the SCZ to the CMJ and the M zones. In the wild-type thymus, most of the cells die in their DP stages, in the cortex, and in the SCZ. This is in
agreement with experimental results, where only 10% of cells survive to the SP stage (see review; [30]). In contrast, in the altered thymus displayed in
(B), most of the cells survive to the SP stage and die in the medulla of negative selection.
doi:10.1371/journal.pcbi.0030013.g006

PLoS Computational Biology | www.ploscompbiol.org 0134 January 2007 | Volume 3 | Issue 1 | e13
 Emergent Dynamics in the Thymus

Figure 7. Histogram of the Different Migration Velocities of Thymocytes

The faster cells are more likely to survive thymic selection by contacting the process of an epithelial cell; but slower cells also can have an advantage
(see text).
doi:10.1371/journal.pcbi.0030013.g007

model a complete biological system—an entire cell, organ, or limit the notion of prediction, confirmation, and verification
organism—in a way that is sufficiently realistic so as to be of emergent properties; see [29].
able to test the role of any known fact about the system. This
goal has been formulated as the ability of a model to pass a Materials and Methods
sort of Turing test, and can be viewed as taking to the utmost
Simulation. The RA simulation was written in Cþþ using the
Rhapsody tool, and so RA code was generated by Rhapsody’s code-
generation engine, initiated by the language of Statecharts. To this
automatically generated code, manually encoded objects and func-
tion were added. RA is the bridge made between the running
simulation and the animation. Communication is made over a TCP/IP
connection between a server implementing the Statecharts simula-
tion and built-in animation functions in Flash. We used Matlab to
analyze populations and population-level behavior. See [6] for further
details.

Supporting Information
Text S1. Collection of References Used in the Thymus Simulation
Found at doi:10.1371/journal.pcbi.0030013.sd001 (74 KB PDF).
Video S1. Survival of Wild-Type Cells in Comparison with Mutated
Cells
Found at doi:10.1371/journal.pcbi.0030013.sv001 (57.8 MB AVI).
Video S2. Migration Patterns in Wild-Type and Mutated Cells
Found at doi:10.1371/journal.pcbi.0030013.sv002 (10.9 MB MOV).
Video S3. Homeostasis in the Wild-Type Thymus
Found at doi:10.1371/journal.pcbi.0030013.sv003 (6.3 MB MOV).
Video S4. Following the Migratory Trail of a Single Thymocyte
Figure 8. Influence of CD4 and CD8 Dissociation Rates on Lineage Found at doi:10.1371/journal.pcbi.0030013.sv004 (3.3 MB AVI).
Commitment Ratio
Measuring the ratio of CD4 mature cells to CD8 mature cells, in silico, as a
function of the dissociation rate, we find that to achieve the
Acknowledgments
experimentally measured 2:1 ratio, the dissociation rate of CD8 cells The authors thank Dr. Daniel C. Douek, Vaccine Research Center/
should be anywhere between 0.38 to 0.45 that of CD4 thymocytes. The NIH for his helpful discussions and intellectual input.
expanded insert zooms in on this zone, which produces the
experimentally observed ratio of two CD4 cells for every CD8 cell. Figure Author contributions. SE, DH, and IRC conceived and designed the
8 shows that an increase in the dissociation rate of CD8 over CD4 could experiments, contributed reagents/materials/analysis tools, and wrote
result in a CD4:CD8 population equilibrium, or even to overpopulation of the paper. SE performed the experiments and analyzed the data.
the thymus with CD8 cells, depending on the relative dissociation rate. Funding. This research was partially supported by the John von
doi:10.1371/journal.pcbi.0030013.g008 Neumann Minerva Center for the Development of Reactive Systems

PLoS Computational Biology | www.ploscompbiol.org 0135 January 2007 | Volume 3 | Issue 1 | e13
 Emergent Dynamics in the Thymus

at the Weizmann Institute of Science and by the Center for the Study Competing interests. The authors have declared that no competing
of Emerging Diseases, Jerusalem. interests exist.

References Key role of local regulation in chemosensing revealed by a new molecular

1. Anderson G, Jenkinson EJ (2001) Lymphostromal interactions in thymic interaction–based modeling method. PLoS Comput Biol 2 (7): e82.
development and function. Nat Rev Immunol 1: 31–40. 17. Rothenberg EV, Taghon T (2005) Molecular genetics of T cell development.
2. Petrie HT (2003) Cell migration and the control of post-natal T-cell Annu Rev Immunol 23: 601–649.
lymphopoiesis in the thymus. Nat Rev Immunol 3: 859–866. 18. Wurbel MA, Philippe JM, Nguyen C, Victorero G, Freeman T, et al. (2000)
3. Janeway C (2001) Immunobiology: The immune system in health and The chemokine TECK is expressed by thymic and intestinal epithelial cells
disease. New York: Garland Science. 600 p. and attracts double- and single-positive thymocytes expressing the TECK
4. Takahama Y (2006) Journey through the thymus: Stromal guides for T-cell receptor CCR9. Eur J Immunol 30: 262–271.
development and selection. Nat Rev Immunol 6: 127–135. 19. Plotkin J, Prockop SE, Lepique A, Petrie HT (2003) Critical role for CXCR4
5. Efroni S, Harel D, Cohen IR (2003) Toward rigorous comprehension of signaling in progenitor localization and T cell differentiation in the
biological complexity: Modeling, execution, and visualization of thymic T- postnatal thymus. J Immunol 171: 4521–4527.
cell maturation. Genome Res 13: 2485–2497. 20. Uehara S, Grinberg A, Farber JM, Love PE (2002) A role for CCR9 in T
6. Efroni S, Harel D, Cohen IR (2005) Reactive animation: Realistic modeling lymphocyte development and migration. J Immunol 168: 2811–2819.
21. Ashwell JD, Lu FW, Vacchio MS (2000) Glucocorticoids in T cell develop-
of complex dynamic systems. Computer 38: 38–47.
ment and function*. Annu Rev Immunol 18: 309–345.
7. Harel D, Efroni S, Cohen IR (2003) Reactive animation. Lect Notes Comput
22. Lu FW, Yasutomo K, Goodman GB, McHeyzer-Williams LJ, McHeyzer-
Sci 2852: 136–153.
Williams MG, et al. (2000) Thymocyte resistance to glucocorticoids leads to
8. Harel D (1987) Statecharts: A visual formalism for complex systems. Sci
antigen-specific unresponsiveness due to ‘‘holes’’ in the T cell repertoire.
Comput Programming 8: 231–274.
Immunity 12: 183–192.
9. Sulzer B, Perelson AS (1996) Equilibrium binding of multivalent ligands to 23. Singer AL, Koretzky GA (2002) Control of T cell function by positive and
cells: Effects of cell and receptor density. Math Biosci 135: 147–185. negative regulators. Science 296: 1639–1640.
10. De Boer RJ, Perelson AS (1994) T cell repertoires and competitive 24. Cohen IR (2000) Discrimination and dialogue in the immune system. Semin
exclusion. J Theor Biol 169: 375–390. Immunol 12: 215–219. Discussion, 257–344.
11. Kohler B, Puzone R, Seiden PE, Celada F (2000) A systematic approach to 25. Germain RN (2002) T-cell development and the CD4–CD8 lineage decision.
vaccine complexity using an automaton model of the cellular and humoral Nat Rev Immunol 2: 309–322.
immune system. I. Viral characteristics and polarized responses. Vaccine 26. He X, He X, Dave VP, Zhang Y, Hua X, et al. (2005) The zinc finger
19: 862–876. transcription factor Th-POK regulates CD4 versus CD8 T-cell lineage
12. Li QJ, Dinner AR, Qi S, Irvine DJ, Huppa JB, et al. (2004) CD4 enhances T commitment. Nature 433: 826–833.
cell sensitivity to antigen by coordinating Lck accumulation at the 27. Bosselut R, Guinter TI, Sharrow SO, Singer A (2003) Unraveling a revealing
immunological synapse. Nat Immunol 5: 791–799. paradox: Why major histocompatibility complex I-signaled thymocytes
13. Chakraborty AK (2002) How and why does the immunological synapse ‘‘paradoxically’’ appear as CD4þ8lo transitional cells during positive
form? Physical chemistry meets cell biology. Sci STKE 2002: PE10. selection of CD8þ T cells. J Exp Med 197: 1709–1719.
14. Schatz DG (2004) Antigen receptor genes and the evolution of a 28. Cohen IR (2000) Tending Adam’s garden: Evolving the cognitive immune
recombinase. Semin Immunol 16: 245–256. self. San Diego (California): Academic Press. 266 p.
15. Meier-Schellersheim M (1999) SIMMUNE, a tool for simulating and 29. Harel D (2005) A Turing-like test for biological modeling. Nat Biotechnol
analyzing immune system behavior. Hamburg: Deutsches Elektronen- 23: 495–496.
Synchrotron (DESY). 30. Chen W (2004) The late stage of T cell development within mouse thymus.
16. Meier-Schellersheim M, Xu X, Angermann B, Kunkel EJ, Jin T, et al. (2006) Cell Mol Immunol 1: 3–11.

PLoS Computational Biology | www.ploscompbiol.org 0136 January 2007 | Volume 3 | Issue 1 | e13