Transcript Rutgers

Specificity, Diversity, and Self-tolerance
of T-Cell Receptors by Thymic Selection
Andrej Kosmrlj
Abhishek Jha
Mehran Kardar
Eric Huseby
Arup K. Chakraborty
T Cells orchestrate adaptive immunity, via
Receptors that recognize short peptides from pathogen proteins
T Cell
TCR
pMHC
Antigen Presenting Cell
T Cells Receptors (TCRs) must be:
Self-tolerant, i.e. weakly binding to
endogenous pMHC, to avoid autoimmune
diseases, e.g.
Multiple sclerosis,
diabetes,
…
pMHC
Diverse, to recognize an evolving landscape
of pathogens.
Specific/degenerate, to lock on specific
pathogen, and tolerate its mutants.
What is the shortest peptide length that ensures self-tolerance?
n*= 5.3
Length of peptide, n
In principle all human peptides can be recognized from sequences of length 5-6.
What is the physical mechanism that enables recognition of self-peptides?
T Cells are designed in the Thymus
where a diverse repertoire of thymocytes is
culled by encounters with self pMHC.
Positive selection TCR must bind sufficiently
strongly to at least one self pMHC (implicated
in MHC restriction, and sensitivity).
Negative selection TCR must not bind any self
pMHC too strongly (deleting autoimmune
TCR).
Model for TCR selection in the Thymus
conserved
variable
TCR {L}
E
N 10
 U L , J 
i 1
pMHC {J}
MHC
i
i
MiyazawaJernigan
peptide
Surviving T cells: E > EP for at least one peptide,
E<EN for all peptides
+
✓
--
E
Diversity
The selection process biases the composition of amino-acids in mature TCRs:
Negative selection leads to a slight preference for weak amino-acids.
Frequency of TCR amino-acids
from known TCR-pMHC crystal structures
(Thermophiles used for ordering,
Shakhnovich et al., PLOS, 2007)
Frequency in TCR/Frequency in proteome
(MJ used for ordering)
STRONG
WEAK
Specificity
Eric S. Huseby et al, Cell (2005); Nature Immunol. (2007),
compared the T cells of normal mice, with mice genetically engineered to
present only one type of peptide in their thymus.
T cells selected in the thymus are challenged with an antigenic peptide, and
reactive T cells identified.
Does a reactive T cell remain reactive upon mutating the peptide’s amino acids?
If mutations to an amino acid destroy reactivity with at least half the T cells,
the amino-acid is labeled a “hot spot”.
• Main results:
– Single peptide selection: few hot spots – cross-reactive T cells
– Many peptide selection: many hot spots – specific T cells
• Specificity to antigen peptide:
– Single peptide: mutations don’t matter – cross-reactive T-cells
– Many peptides: mutations destroy reactivity – specific T-cells
frequency
Numerical results for hot-spots† mirror the experimental situation
† Hot-spots are defined as locations along the sequence, where mutations of a peptide amino
acid destroy reactivity with more than half the reactive T cells
Frustration during negative selection constrains TCR
sequences
One peptide
T C R
Ep < E < EN
selected
Frustration during negative selection constrains TCR
sequences
One peptide
T C R
EN  i U (li , ji )  E p
selected
Frustration during negative selection constrains TCR
sequences
One peptide
T C R
EN  i U (li , ji )  E p
selected
Many peptides
T C R
E > EN
negatively selected
Optimizing interactions with one peptide can lead to “bad” interactions with another –
FRUSTRATION.
Positive selection does not involve frustration.
Specificity, Diversity, and Self-tolerance
of T-Cell Receptors by Thymic Selection
Andrej Kosmrlj
Abhishek Jha
Mehran Kardar
Eric Huseby
Arup K. Chakraborty
Frustration during negative selection constrains TCR
sequences
M




P e (l )   (1  ( E (l , j )  E N ) p ( j )
j 1
 
M

~ exp  ln( 1  ( E (l , j )  E N )) 
 j 1

 
 exp M ln( 1  ( E (l , j )  E N ))


1  (E ) ~ exp  exp( bE )
  I 10
E (l , j )   U (l , j )
 1
Frustration during negative selection constrains TCR
sequences
Solution:
Prob. that a TCR
is selected
10 20



P
(l )  exp  M   hij 
i 1
j 1


 (h11  h12  h13  ..)( h21  h22  ..)...( h20,1  h20, 2  ..)
escape
hij  exp bU (li , j ) p( j )
negative selection lead to choice of TCR
amino acids which minimize interactions
with other amino acids in En peptides.
Frequency in TCR/Frequency in proteome
COMPUTATIONAL RESULTS
strong
weak
Robust to variations in potential
AMINO ACID FREQUENCIES FROM TCR-PMHC CRYSTAL STRUCTURES
Frequency in TCR/Frequency in proteome
(Thermophiles used for ordering, Shakhnovich et al., PLOS, 2007)
STRONG
WEAK
Distribution of single site contact energies for
selected T cell-antigenic peptide interactions
Increased number of moderate
interactions
Decreased
number of
strong
interactions
strong
weak
Selection with many peptides: TCR sequences contain amino
acids that interact with Ag peptide amino acids moderately
Modest interactions lead to specificity:
Weak multivalent interactions stabilize the interface,
making each interaction important for recognition
Selection against one peptide – only few important sites
T C R
How much Free Energy of Binding is due to each amino
acid for specific versus degenerate TCRs?
P-1E
P-1E
B3K 506 TCR
C57BL/6 derived
MHC + peptide specific
P2Q P3K
P8K
P5K
YAe62.8 TCR
IAb-SP derived
MHC + peptide degenerate
IAb + 3K
KD
G
IAb + 3K
KD
G
Mutation
M
kcal/mol
Mutation
M
kcal/mol
WT
P-1 A
P2 A
P3 A
P5 A
P8 A
7
26
278
> 550
> 550
92
0.0
0.7
2.2
> 2.6
> 2.6
1.5
WT
P-1 A
P2 A
P3 A
P5 A
P8 A
8
9
56
62
> 550
7
0.0
0.0
1.1
1.2
> 2.5
-0.1
TCR sequences are specific, but diverse
One type of sequences are selected – ones with a
predominance of weak amino acids
+
+
+
+
-
+
+
+
+
+
-
+
+
+
TCR sequences are cross-reactive (degerate)
Several antigenic peptides composed of
sufficient number of strong amino acids can
interact productively with a given TCR
ADAPTIVE IMMUNITY IN HEALTH AND DISEASE
Flexible system to combat
diverse pathogens
Mis-regulation leads to
autoimmune diseases
Multiple Sclerosis
Diabetes
The challenge: develop principles that
govern the emergence of an immune
response or autoimmunity and design
rules for therapies
The problem: underlying mechanisms
characterized by cooperative dynamic
processes involving many components
and a spectrum of length/time scales
T CELLS RECOGNIZE SHORT PEPTIDES DERIVED FROM
PATHOGEN’S PROTEINS
Extraordinary Sensitivity of T cells for Antigen
mixture of self (En) and antigenic (Ag/agonist)
As few as 3 Ag molecules in a sea of 30,000
En can activate a T cell (Nature, 2002)
How does this exquisite sensory apparatus work
without frequent “noise”-induced autoimmune
responses?
T cell sensitivity to Ag pMHC is predicated upon
degenerate weak interactions with En pMHC.
With Mark Davis’ lab. (Nature Imm., 2004; Nature, 2005; PNAS, 2007;
unpublished)
INFLUENCE OF PARAMETERS IN MODEL
Parameters: Ec, EN, Ep
Ec-EN is large (conserved TCR-MHC interactions are very weak)
Ec
Ep Ec
Ep EN
EN
Small
gap
Large
gap
Positive selection limiting; These are the TCRs that are not
positively selected (MHC restriction); one or many types of
peptides lead to similar consequences
Ec-EN is small (conserved TCR-MHC interactions are very strong)
Ec EpEN
Ep
Ec
EN
Small
gap
Large
gap
Negative selection very easy; For many types of
peptides almost all T cells are negatively selected.
An approach at the intersection of disciplines
Life
Sciences
Physical
Sciences
Engineering
sciences
Theory/computation
statistical physics
chemical kinetics
Experiments
genetics
biochemistry
imaging
THYMIC SELECTION THRESHOLDS
Palmer lab, Nature (2006)
strongest
weakest
Sharp boundary separates positive and negative selectors
MOLECULAR MECHANISM (w/A.Weiss’ lab.)
A MEMBRANE-PROXIMAL SIGNALING MODULE IMPLICATED
(with Jayajit Das, Ashok Prasad; Jeroen Roose, Art Weiss@UCSF)
A positive feedback loop results in digital signaling
and a sharp threshold Cell, PNAS, in review (2008)
SHORT PEPTIDES ARE SUFFICIENT TO RENDER T CELLS
SELF-TOLERANT AND REACTIVE TO FOREIGN
SHORT PEPTIDES ARE SUFFICIENT TO RENDER T CELLS
SELF-TOLERANT AND REACTIVE TO FOREIGN
Longer peptides (length = n) will enable
sampling of more distinct peptides ~ 20n
# Unique peptides of length
n
SHORT PEPTIDES ARE SUFFICIENT TO RENDER T CELLS
SELF-TOLERANT AND REACTIVE TO FOREIGN
8
10
7
10
6
10
5
10
4
10
3
10
2
10
n=5.3
1
10
5
10
15
20
25
Length of peptide, n
Condition for plateau:
20  N ;
n
Total # of unique peptides of size, n
ln N
n
ln 20
Total size of human proteins
ACKNOWLEDGMENTS
Group members:
Collaborators:
J. Das, A. Prasad, M. Artomov,
C. Govern, H. Zheng, A. Jha,
J. Locasale, K. Fowler,
M. Wolfson, A. Prabhakar,
M. Yang, F. Liang, A. Kosmrlj.
M. Davis, A. Shaw, P. Allen, A.
Weiss, J. Roose, H. Ploegh
M. Dustin, M. Kardar, A
Perelson, J. Chen, U. von
Andrian, H. Eisen, V. Kuchroo,
E. Palmer, E. Huseby.
Funding:
NIH
Immune Response Consortium
Is there an optimal peptide length?
Fraction of Reactive T Cells
0.05
Constant Negative Selection Threshold
E=E
ENegENeg=0.4
0.04
Most TCR
negatively
selected
0.03
0.02
0.01
0.00
0
5
10
15
Length of Peptide n
20
25
IMPORTANCE OF PEPTIDE IN ALLO REACTIVITY
DEPENDS UPON THE ALLO MHC
Bigger change
Modest change
Theoretical models and experimental tests
(Nature Imm. (2004); Nature (2005); PNAS (2007); unpublished)
T cell sensitivity to Ag pMHC is predicated
upon degenerate weak interactions with En
pMHC that are tuned in the thymus.