Why does Mycobacterium tuberculosis use multiple mechanisms to inhibit antigen presentation

1
Why does Mycobacterium tuberculosis use multiple
mechanisms to inhibit antigen presentation?

• Stewart Chang
• Bioinformatics Program _at_ The University of
  Michigan
• Advisors Denise Kirschner and Jennifer Linderman

2
The macrophage during M. tuberculosis infection

• Dual roles during TB infection
• Preferred host cell
• Effector cell
• Effector cell function is cell-mediated
• Requires cytokine signal from CD4 T helper cell
• Macrophages must first present antigen bound to
  MHC class II

From Janeway CA et al. Immunobiology. New York
Garland Publishing, 2001.
3
One M. tuberculosis survival strategyInhibit
antigen presentation

• Experimental method
• Infect monocytes with M. tuberculosis at MOI 50
• Add soluble model antigen (tetanus toxoid)
• Measure T cell response by thymidine uptake
• Infected monocytes do not stimulate T cells as
  well as uninfected monocytes

TT pulse ()
Mtb TT pulse (x)
Gercken J, Pryjma J, Ernst M, Flad HD. 1994.
Defective antigen presentation by Mycobacterium
tuberculosis-infected monocytes. Infect. Immun.
623472-3478.
4
A number of different mechanismshave been
hypothesized

• Inhibition of antigen processing (H1)
• Inhibition of MHC class II maturation (H2)
• Inhibition of MHC class II peptide loading (H3)
• Inhibition of MHC class II transcription (H4)

Moreno C, Mehlert A, Lamb J. 1988. The inhibitory
effects of mycobacterial lipoarabinomannan and
polysaccharides upon polyclonal and monoclonal
human T cell proliferation. Clin. Exp. Immunol.
74206-210. Hmama Z, Gabathuler R, Jefferies WA,
de Jong G, Reiner NE. 1998. Attenuation of HLA-DR
expression by mononuclear phagocytes infected
with Mycobacterium tuberculosis is related to
intracellular sequestration of immature class II
heterodimers. J. Immunol. 4882-4893.
5
Conflicting data in the literature
Noss et al. (2000, above) found that Mtb inhibits
IFN-g-stimulated transcription of MHC class II
mRNA. Hmama et al. (1998, right) did not.
6
Questions asked in this study

• Why have multiple mechanisms been observed?
• What purpose do multiple mechanisms serve?
• Do some experimental protocols favor the
  detection of particular mechanisms?
• Do alternative mechanisms exist?

7
A review of antigen presentation

• Two pathways one for endogenous antigens, the
  other for exogenous antigens
• MHC class II acts as a receptor for peptides
  derived from exogenous antigens

From Mims C. et al. Mims Pathogenesis of
Infection Disease, Fourth Edition. London
Academic Press, 1985.
8
The antigen presentation model

• Above a certain threshold, the number of surface
  MHC class II-peptide complexes is determinative
  of T cell response
• Therefore, we use surface MHC class II-peptide
  complexes as our output variable

9
Model testing

• Parameter values were derived from the
  literature (mouse)
• Model behavior was checked against experimental
  results
• At right, behavior when IFN-g was added

10
Additional model testing Antigen presentation
11
Simulations of hypothesized mechanisms, effects
on antigen presentation levels

• In simulations, IFN-g and antigen were added at
  time 0 h, and relevant processes were inhibited
  to same extent
• Effect of H1 or H3 immediate but attenuates at 1
  h and 10 h
• H2 or H4 effective at time points gt 10 h
• Mechanisms may be complementary and allow M.
  tuberculosis to continuously inhibit antigen
  presentation

12
Application of the model to previous experimental
protocols

• Goal To determine if some experimental protocols
  favored the detection of particular M.
  tuberculosis mechanisms
• Two previous protocols
• Model accounts for differences in timings and
  concentrations but not differences in macrophage
  cell lines or M. tuberculosis strains

13
Surprising results for protocol of Noss et al.
(2000)

• In agreement with Noss et al. (2000), inhibiting
  MHC class II transcription (H4) significantly
  decreased antigen presentation levels
• However, inhibiting antigen processing (H1) or
  MHC class II peptide loading (H3) had a
  negligible effect on antigen presentation levels

14
Overview of sensitivity analysis

• Allows you to determine importance of inputs
  (e.g. parameters) to output variable
• Rationale Incomplete knowledge of parameters and
  extent to which M. tuberculosis inhibits
  processes
• Methodology, in general
• Specify distribution for each input, sample using
  LHS. For each set of input values, generate an
  output value (above right).
• Calculate correlation coefficient (e.g. PRCC)
  between output and input values.
• Plot correlation coefficients versus time to
  identify important inputs (below right).
• We specify uniform distributions with boundaries
  10 and 190 of baseline values

From Saltelli A et al. Sensitivity Analysis.
Chichester John Wiley Sons, 2000. Helton JC
and Davis FJ. 2002. Illustration of
sampling-based methods for uncertainty and
sensitivity analysis. Risk Analysis 22591-622.
15
Sensitivity analysis reveals other possible
mechanisms
When lower levels of antigen are used, other
processes are significantly correlated, e.g. rate
of lysosomal degradation of antigen, rate of self
peptide production at 1 h, 10 h, negatively
16
Return to the questionWhy multiple mechanisms?

• May allow continuous inhibition of antigen
  presentation
• Otherwise, inhibition may either abate with time
  or be delayed
• Our simulations of previous experimental
  protocols produce results consistent with their
  respective studies
• However, these protocols may favor detection of
  mechanisms targeting MHC class II expression
• Other mechanisms may be possible
• Possible targets IFN-g receptor-ligand binding,
  lysosomal degradation of antigen

17
Another application of the modelAid design of
new experimental protocols
Can we design an experiment to determine if
mechanisms not targeting MHC class II expression
are important to M. tuberculosis infection?
These results suggest mechanisms targeting MHC
class II expression should be less effective when
the duration of IFN-g stimulation is short
18
Predicted results using the proposed protocol

• Let Q percent reduction in antigen presentation
  levels of infected macrophages compared to
  uninfected control
• Q stays constant to the extent that mechanisms
  targeting processes other than MHC class II
  expression are effective

19
Current directions Applying the ODE model to M.
tuberculosis antigens and MHC class II alleles

• Important M. tuberculosis antigens are known
• Antigen 85 complex Ag85A, Ag85B, Ag85C
• 6-kDa early secretory antigenic target (ESAT-6)
• But many parameters need to be determined, e.g.
  binding affinity to MHC
• Some MHC class II alleles increase susceptibility
  to TB
• e.g. HLA-DR2 (old nomenclature), HLA-DRB11501
• Some MHC class II alleles decrease susceptibility
  to TB
• e.g. HLA-DR3
• Generally believed that MHC class II alleles
  differ in ability to bind peptides, but what
  happens at the macrophage surface?
• Hypothesis MHC class II from different alleles
• Differ in ability to bind Mtb antigens
• Leads to different numbers of MHC-Mtb antigen
  complexes on macrophage surface
• Elicits different T helper cell responses

20
In lieu of experimental data for M. tuberculosis
antigens, statistical models to predict binding
affinity

• A published additive model to predict binding
  affinity
• Step 1 Measure IC50 of standard peptide
• Step 2 Measure IC50 of derivatives (differ by
  only 1 aa perfect data set)
• Step 3 Find ratios of derivatives IC50 to
  standards IC50
• Step 4 Multiply ratios for peptide of interest
• Step 5 Multiply by IC50 of parent peptide
  (here, A13)
• Authors claim Predicts IC50 to within one order
  magnitude (peptides may vary five)

Marshall KW, Wilson KJ, Liang J, Woods A, Zaller
D, Rothbard JB. 1995. Prediction of peptide
affinity to HLA DRB10401. J. Immunol.
1545927-5933.
21
Do DR2 and DR3 differ in their binding affinities
for Ag85B?

• Target set Mtb Ag85B epitopes (18mers)
  recognized by CD4 T cells
• Trained model on 18mers in JenPep database
  (www.jenner.ac.uk/jenpep)
• Predicted IC50 values (binding affinities) differ
  by as much as two-fold

Mustafa AS, Fatema FA, Abal AT, Al-Attiyah R,
Wiker HG, Lundin KEA, Oftung F, Huygen K. 2000.
Identification and HLA restriction of naturally
derived Th1-cell epitopes from the secreted
Mycobacterium tuberculosis Antigen 85B recognized
by antigen-specific human CD4 T-cell lines.
Infect. Immun. 683933-3940.
22
Another view of the preliminary data

• The difference between binding affinities is
  statistically significant
• Could this account for differences in immune
  response?
• Or, could this result in different numbers of
  MHC-Ag85B complexes on the macrophage surface and
  different T cell responses?
• These numbers could be used in the ODE model to
  generate experimentally testable predictions

23
Acknowledgments

• Kirschner lab members
• Linderman lab members
• Helpful discussions Cheong-Hee Chang, Joanne
  Flynn, Eugenio de Hostos