Transcript No Slide Title

Computer simulation of the immune system: study of cross-reactive Tc
memory against heterologous viruses
Yiming Cheng, Dario Ghersi, Claudia Calcagno, Roberto Puzone, Franco Celada
Division of Rheumatology, Department of Medicine, New York University School of Medicine, New York, NY, USA
Abstract
Model Setup
Cross-reactivity
Split cross-reactivity
IMMSIM[1] was constructed to tackle the staggering complexity of the Immune
System by comparing it with the responses of an agent-based computer model,
where the agents are minimalistic portraits of lymphoid cell types, body’s target
cells and typical invaders. Once launched, the response develops in virtual 3D
(2 dimensions in space and 1 dimension in time) computer spaces, and all
similarities with and differences from biological consensus serve to shape,
confirm or discard ideas and hypotheses. In the present study we simulate the
formation and the recall of anti-virus memory, and try to predict what will
happen both to the responses and to the memory cells when the second infecting
virus is partly different from the first one, and when the cross-reactivity of the
two branches, humoral and cellular of the immune system is surprisingly
asymmetrical, or split. The in machina simulations explore all epitope distances,
and measure all changes in affinity, cellularity and efficiency in clearing the
infection. Besides the obvious cooperation, they reveal powerful competitions
between the two branches and between the cross-reacting memory and new
responses when the latter suffer the competition by preformed cell-rich but
inefficient clones. Memory, usually an asset, has become a liability.
Each virtual mouse (VM) in the present setup is endowed with 2500 cells of
each type: B, Th (Th1+Th2), Tc, APC. The random generation of 16-bit
strings assures a total repertoire of 65536 different specificities for B and T
receptors. Each of the specific cell types (B and T) is endowed with the
clonal repertoires shown in the table.
The experiments have been conducted in a MATRIX scheme to ensure at
least 25 repetitions of each experiment for every one bit at a time increased
distance between the primary infectant and the challenging virus. Since
only 0, 1, 2 and 3 mismatches between paratope and epitope are effective, 2
viruses are potentially related when they carry up to 6 bit differences
between their respective peptides, as shown in Cartoon (A).
Selin et al. [2] discovered that many serologically heterologous viruses are
cross-reactive at the level of cellular responses. This is defined as Split
cross-reactivity[3].
Theoretical Total Repertoire of the species
65536
Number of lymphocytes/individual mouse
2500
Expected Individual total repertoire
2453
Total Species Repertoire specific for one epitope
697
Expected specificity overlap between two mice
92
Expected Individual specific repertoire for one epitope
26
Expected number of clones specific for one epitope in common
between two individuals
(A)
<1
The overlap of specificity in two randomly generated VM is illustrated in
the figure below and is compatible with the consensus on private specificity.
IMMSIM model
(B) maps all distances between 5 viruses; (C) shows the combinations
between primary and secondary challenge used in the matrix.
(B)
(C)
(a)
Homologous Responses
0.5
Repeated infections by the same viruse generate primary, secondary and
tertiary responses by the humoral and cellular branches of the immune
system. At each successive infection, the number of virus particles decreases
H
0.45
Recall efficiency, in terms of
clones and of cell numbers,
declines in parallel in the
humoral response, but the
decline is biphasic in the
cellular, with no decrease up
to bit distance 3.
Bc cellularity
0.4
Bc clone
0.35
Recall ratio
The agents of IMMSIM behave like cellular automata (a), but they represent all
types of cells of the immune system with their specific receptors and paratopes,
the antigens and the invading viruses. T cell helps B cell upon antigen binding
of the B receptor, leading to the initiation of the humoral response (b). The
maturation of the antibody response is governed by the scarcity of T help in the
primary response and the antigen competition in the secondary response (c). (d)
illustrates all simulated steps of the viral infection and the combined cellular and
humoral response leading to the cure of infection.
(b)
(a)
Running side-by-side H+C and C (Split) responses allowed us to dissect
two zones of contrast affecting the cellular response. (a) The Tc strength
recall at bit distance 0-3 suffers a competitive thwarting caused by the
presence of the humoral response. The probable effector is antibody, as
indicated by what-if experiments introducing virtual Ab half lifetime.
Tc cellularity
0.3
Tc clone
0.25
0.2
0.15
(b) The Tc strength at bit distance 2-4 is thwarted and this is caused by the
presence of cellular cross-reactive memory.
0.1
0.05
0
0
1
2
3
4
5
6
7
8
Distance
C
The strength (weighted affinity) of the secondary response relative to the
primary response toward the primary infectant and against the secondary
infectant. In the humoral response, the latter curve descends harmoniously
from 1.3 to 1 as expected. In the cellular response, there is an initial dip and
the level of the primary response is reached only at distance 5 and higher.
(c)
V
(b)
Responses are decreasing in clone repertoire.
H
C
(d)
H
How can memory be a burden instead of an asset? It can, if the affinity of
memory is low on account of the distance of the cross-reactants: the sheer
member of the memory cells may outcompete the naïve cells and offer a
lower quality defense.
C
Responses are increasing in clone average affinity (strength).
Conclusions
Comparing the clonal engagement in the secondary response relative
to the primary for humoral and cellular responses show trimming at bit
distance 0, 1, 2, 3, 4 and then expansion at distance 5-8. The
significance of this observation is under study.
Like all model results, these conclusions are provisional. They await the
conformation by wet-lab researches (of the UMASS team) before being
used in the discussion about – e.g. – the influenza vaccine.
This work is supported by NIH Grant R01-AI054455
References
H
C
H
C
1. Celada F, Seiden PE. A computer model of cellular interactions in the
immune system. Immunol Today 1992;13(2):56-62
2. Selin LK, Lin MY, Kraemer KA, Pardoll DM, Schneck JP, Varga SM, et
al. Attrition of T cell memory: selective loss of LCMV epitope-specific
memory CD8 T cells following infections with heterologous viruses.
Immunity 1999;11(6):733-42
3. Celada F, 2005, Personal Communication.