1. Antigenic variation in malaria involves a highly structured switching pattern
Mario Recker Department of Zoology, University of Oxford

2. Mathematical approach to (understand) malaria
“the mathematical method of treatment is really nothing but the application of careful reasoning to the problems at issue.” Ross R, The Prevention of Malaria . London: John Murray.
Sir Ronald Ross
20. Aug. 1897
??? ? ?
Macdonald G, The Epidemiology and Control of Malaria
from McKenzie & Samba, AJTMH 2004

3. Most targets of protective immunity polymorphic surface proteins
Development of immunity / effective vaccines hindered by extensive antigenic diversity:
- mutation / recombination (genotypic change)
- antigenic variation (no genotypic change)
circumsporozoite protein (CSP)
merozoite surface proteins (MSP)
variant surface antigens (VSA)
diversity
Major multigene families:
rif > 150 copies per genome
stevor copies per genome
Pfmc-2TM 13 copies per genome
var 60 copies per genome
Scherf et al., Annu Rev Microbiol 2008

4. Sequence diversity of var genes is immense!
adapted from Gardner, M. et al., 2002, Kyes, S. et al., 2002
cumulative diversity of DBLa seqnuences
pairwise sharing among DBLa seqnuences
from Barry et al, PLoS Pathog. 2007

5. Antigenic variation in P.falciparum
PfEMP1 (P. falciparum Erythrocyte Membrane Protein 1)
embedded on surface of red cell
causes severe disease through adherence to host cell receptors
important immune target
IE binding to endothelium
IE binding to erythrocytes
IE binding to dendritic cell
EM by D. Ferguson, Oxford Univ.

6. Infected blood cells sequester in tissue capillaries
EM by D. Ferguson, Oxford Univ.

7. (Molecular) Requirement for antigenic variation
- every var gene recognised as part of a family
- mechanism to limit expression to a single copy
- activation coinciding with silencing of previously active gene
- cellular memory to avoid ‘early’ repertoire exhaustion
Result:
succeeding waves of parasitaemia dominated by a single variant of PfEMP1
Scherf et al., Annu Rev Microbiol 2008
PfSir2: P.falciparum silent information regulator
TPE: telomere position effect
what orchestrates expression at population level?

8. - use mathematical models to create and test hypotheses -
What orchestrates sequential dominance?
- use mathematical models to create and test hypotheses -
For example:
differences in growth rates or probabilities in switch rates (e.g. Kosinski, 1980)
differences in growth rates or probabilities in switch rates (e.g. Kosinski, 1980)
differences in growth rates or probabilities in switch rates (e.g. Kosinski, 1980)
differences in growth rates or probabilities in switch rates (e.g. Kosinski, 1980)
differential susceptibilities assigned to variants expressing two surface proteins (e.g. Agur et al., 1989)
differential susceptibilities assigned to variants expressing two surface proteins (e.g. Agur et al., 1989)
differential susceptibilities assigned to variants expressing two surface proteins (e.g. Agur et al., 1989)
differential susceptibilities assigned to variants expressing two surface proteins (e.g. Agur et al., 1989)
modifications of switch rates by ‘natural selection’ (Frank, 1999)
modifications of switch rates by ‘natural selection’ (Frank, 1999)
modifications of switch rates by ‘natural selection’ (Frank, 1999)
immunological interaction, e.g. cross-immunity (e.g. Recker et al, 2004)
immunological interaction, e.g. cross-immunity (e.g. Recker et al, 2004)

9. Agglutinating antibody titer
Increases in levels of antibodies to VSA expressed by heterologous isolates are transient and limited. 58 10 20 30 40 50 60 18 55
Percentage of infected red cells positive
Agglutinating antibody titer
time after infection

10. Model assumption: each variant comprises
a unique major epitope which elicits variant specific, long-lived immune response
a number of minor epitopes which elicits transient, cross-reactive immune response

11. The model m’>>m dynamics of variant i:
intrinsic growth rate
clearance by specific response
clearance by cross-reactive response
dynamics of specific response zi:
immune response proportional to antigen
decay rate
m’>>m
dynamics of transient, cross-reactive response, wi:
transient immune response proportional to antigen variants with shared epitopes

12. Mathematical model without switching
Recker et al, Nature 2004
Model suggested that parasite-host relationship has evolved to favour some short-lived immune responses that allow the parasite to persist and the host to survive

13. In vitro switching dynamics
Horrocks et al. (PNAS, 2004) showed
on and off rates for a given variant are dissimilar
on and off rates vary dramatically among different variants
var 1
var 2
var 3
time
Abundance
- rates appear to be intrinsic property of a particular gene -
→ could introduce a hierarchy of expression whereby stable variants are more prominently expressed, at least during the early phases of infection?

14. stable dominance of initial variant initial variant replaced
To investigate the nature of var gene switching, generate transcription profiles for the entire repertoire in clonal parasite populations and measure the change in that profile over time
stable dominance
of initial variant
1st generation
2nd generation
initial variant
replaced

15. change over generations determined by:

16. - use mathematical model to determine most likely switching pathway -
off rate
switch bias
on rate
use iterative approach to find ‘best-fit’ switch matrix and off-rate vector

17. Switch matrix:
Switch sequence: 1→2 → 4 → 3 →

18. Clone 3D7_AS2

19. Clone It_B2B

20. even for a stable clone…

21. Clone B10
Clone B12
Data provided by Dzwikowski, Frank & Deitsch

22. To test the validity of this prediction, examine the var transcript distribution in Clone 2 every few generations

23. protection of repertoire protection against immune attack
Evolutionary conflict:
protection of repertoire
vs.
protection against immune attack
repertoire protection:
immune evasion:

24. protection of repertoire protection against immune attack
Evolutionary conflict:
protection of repertoire
vs.
protection against immune attack
Assume var gene repertoire as a network where
- nodes = variants
- edges = switch / transition probabilities
Task: optimise network over two traits
- pathlength (= repertoire protection)
- robustness (= adaptability to selection pressure)

26. Clone 3D7_AS2

27. Clone It_B2B

28. Investigate effects of hierarchical switching for in vivo dynamics

29. naïve host
highly structured switching results in (significantly?) increased length of infection.

31. naïve host
highly structured switching results in (significantly?) increased length of infection.
semi-immune host
sms and lattice-type pathways far more flexible in overcoming pressure from pre-existing immune responses to help set up chronic infections.

32. Antigenic relationship between variants
minor epitope 1 a b c d e u v
minor epitope 2 x y z
Switch sequence: (au) → (bu,av) → (cx) → (dx,cy) →…

33. Summary
for pathogens with a limited antigenic pool, such as P. falciparum, tight control over variant expression is essential
tightly ordered gene activation requires every subsequent variant to be able to evade current immune responses and therefore may be compromised by previous infections
highly structured switching in P. falciparum has evolved as an evolutionary compromise between the protection of its limited antigenic repertoire and the flexibility to fully utilise this repertoire when needed

34. Acknowledgements Sunetra Gupta Sam Kinyanjui Caroline Buckee Pete Bull
University of Oxford
Department of Zoology
Sunetra Gupta
Caroline Buckee
Robert Noble
Sam Kinyanjui
Pete Bull
Kevin Marsh
Chris Newbold
Andrew Serazin
Sue Kyes
Zóe Christodoulou
Robert Pinches