2. Human T-lymphotropic virus
type 1 (HTLV-1)
• infects 10-20 million people worldwide.
• endemic (1-20% of adults) in South America,
Caribbean, Central Africa, southern Japan.
• 2-3% develop an aggressive T-cell
leukaemia/lymphoma
• 2-3% develop a chronic inflammatory disease either of
CNS, eyes, muscles, joints, lungs or skin.
• 95% remain healthy carriers of HTLV-1.
3. HTLV-1 persistence and inflammatory disease
Three main questions:
1. How does HTLV-1 persist?
2. How does it spread?
3. Why do some develop HAM/TSP, whereas
most remain healthy carriers?
4. The proviral load of HTLV-1 is high and
correlates with the risk of HAM/TSP
median proviral load
(copies/100 PBMCs)
HAM/TSP: 5.4
asymptomatic: 0.34
Nagai et al 1998
J. Neurovirol. 4, 586
5. How is the high proviral load maintained?
Retroviruses replicate by two routes:
host cell - provirus latent
mitotic
polymerase - little sequence
variation in virus
reverse - provirus
infectious expressed
transcriptase - sequence
variation
6. Evidence for latency of HTLV-1
variability (Kabat-Wu) Env
1. HTLV-1 varies little in HIV-1
sequence:
HTLV-1
amino acid position
Daenke et al. 1990: J. Virol. 64, 1278
2. HTLV-1 mRNA and proteins are usually undetectable
in PBMCs.
3. Virions are absent and plasma is non-infectious.
7. ‘Standard model’ of HTLV-1 persistence
HTLV-1 is maintained by passive proliferation of provirus-
containing lymphocytes.
A fraction of cells express HTLV-1, but too few to allow the
immune response to make an impact on proviral load.
Supported by observation of
large clones of HTLV-1+
lymphocytes in vivo:
Wattel et al. 1995:
J. Virol.69, 2863
8. What is wrong with the ‘standard model’?
There is a strong T-cell and antibody response to HTLV-1.
Anti-HTLV-1 cytotoxic T lymphocytes (CTLs) are
chronically activated; virus-specific IgM is produced.
Passively proliferating HTLV-1+ cells would be outgrown if
any start to express HTLV-1: 1%/day 40-fold drop in
load over 1 year.
- does the CTL response make any impact?
9. Anti-HTLV-1 CTLs are -
• abundant
• chronically activated
• chiefly directed against the HTLV-1 Tax protein
but CTL frequency, specificity and activation do not differ
between HAM/TSP and asymptomatics
Bangham 2000 Curr Opin Immunol 12, 397
10. HTLV-1-specific CTL frequency is positively
correlated with proviral load
Kubota et al. 2000
Wodarz et al. 2001
- so do CTLs determine load, or passively reflect the load?
11. CTLs exert selection on HTLV-1 in vivo
1. The tax gene is under positive selection in
asymptomatic carriers but not in HAM/TSP patients:
AC (N=4) HAM (N=4)
DN / DS = 1.15 0.42
% specific lysis
2. Naturally occurring Tax variants
escape CTL recognition:
Niewiesk et al 1994: J. Virol. 68, 6778
Niewiesk et al 1995: J. Virol. 69, 2649
12. Positive selection at individual Tax residues
HLA-A2-
restricted CTL
epitopes
Positive
selection at
individual
residues
13. 9 genes were overexpressed in CD8+ cells from
individuals with a low HTLV-1 proviral load
A
AC1 CD8L AC2 CD8L 1 (AJ001687): NKG2D
cluster: cluster: 2 (M57888): Granzyme B
3 (U20350): CX3CR1
33 genes 16 genes 4 (M30894): TCR-gamma
5 (M12824): CD8 A
20 3 2 6 (AF031824): Leukocystatin
7 (M18737): Granzyme A
8 (M85276): Granulysin (NKG5)
9 (S69115): NKG7
9 B
1 (X57352): IFITM3 (I-8U)
1 2
C
1 (U26174): Granzyme K
2 (M28393): Perforin
5
HAM CD8L D
cluster: 1 (M17016): Granzyme B precursor
2 (M1121): RANTES (1)
17 genes 3 (M1121): RANTES (2)
Vine, Heaps et al., 2004: J Immunol 173, 5121
14. Protective role of HLA class 1 indicates
that CTLs limit HTLV-1 expression in vivo
1. Possession of either HLA-A*02 or HLA-Cw*08 :
• reduced proviral load by 3-fold HLA-A2 and HLA-Cw8
prevent 36% of
potential HAM/TSP
• halved the odds of HAM/TSP cases.
2. HLA class 1 heterozygosity was associated with a
lower proviral load.
Jeffery et al: PNAS (1999) 96, 3848; J Immunol (2000) 165, 7278
Vine et al: J Infect Dis (2002) 186, 932
15. Alternative scheme of HTLV-1 persistence
CTL
Spontaneously expressed Tax protein drives proliferation of provirus+ cells.
CTLs kill virus-expressing cells.
A fraction of daughter cells survive by shutting down expression (how?)
Proviral load is determined by an equilibrium between virus and CTLs : the
chief determinant of load is the rate – the ‘efficiency’ – of CTL killing.
Nowak and Bangham 1996: Science 272, 74-79; Bangham and Osame 2005: Oncogene 24, 6035–6046.
16. Predictions of model 2
1. Proviral load is determined by a) CTL efficiency &
b) rate of Tax expression.
2. Mean lymphocyte turnover rate a) is abnormally high in
HTLV-1 infection, especially in HAM/TSP patients, and
b) correlates with [Tax].
3. Proviral load correlates with Tax protein expression.
4. The advantage to the virus conferred by Tax expression
diminishes as the CTL lysis rate increases.
17. Quantification of anti-viral CTL efficiency
dy
= c − εyz
HAM dt
patient
y = freq. of infected cells
z = freq. of CTLs
ε = lysis rate constant - ‘CTL efficiency’
carrier
c = constant
Asquith, Mosley et al., 2005:
J Gen Virol 86, 1515
18. Quantification of anti-viral CTL efficiency
6
% T a x + C D 4 + c e lls repeat 1: observed data
5
repeat 1: best theoretical fit
a fte r 1 8 h rs
4
3 repeat 2: observed data
2
repeat 2: best theoretical fit
1
0
0 10 20 30 40 50 60 70
% CD8+ cells
ε is reduced:
80% by block of perforin
29% by partial HLA mismatch Asquith, Mosley et al., 2005
100% by full HLA mismatch J Gen Virol 86, 1515
19. Prediction 1: CTL lytic efficiency determines
HTLV-1 proviral load in vivo
proviral load (copies/100 PBMC)
Asquith, Mosley et al.
2005: J Gen Virol 86,
1515
lytic efficiency ε
(%Tax+ cells killed/CD8+ cell/day)
Conclusion: 30% to 50% of observed variation in HTLV-1
proviral load is accounted for by variation in ε.
20. Impact of CTL activity in HTLV-1 infection
1. The rate of CD8+ cell-mediated lysis is an important
determinant - perhaps the largest single determinant - of
variation in HTLV-1 load between individuals.
2. In a typical infected individual, each CD8+ cell kills
~5 HTLV-1-infected cells/day.
turnover rate of Tax+ cells of ~7% per day.
i.e. total of ~2 x 109 infected CD4+ cells killed/day.
21. Measurement of lymphocyte turnover
rates in vivo
• infuse 2H-labelled glucose (5% ) i.v. overnight
• carbon ring is incorporated into newly synthesized
nucleosides genomic DNA of newly divided cells
• decay of 2H/1H in DNA direct estimate of t½ of
specific (sorted) lymphocyte subsets
Macallan et al (1998) PNAS 95, 708
Asquith et al (2002) Trends Immunol. 23, 596
22. 2H-glucose
kinetics in vivo:
CD45RO+ cells turn over faster than CD45RA+
CD45RO+
0.15 0.15
L01-HS L07-TBI
AC 1 HAM 1
0.1 0.10
0.05 0.05
0 0.00
0 2 4 6 8 10 0 2 4 6 8 10
CD45RA+
0.15 2H incorporation
L10-HBO
AC 2
0.10
0.05
0.00
0 2 4 6 8 10
Asquith et al. 2006: submitted time (d)
23. Prediction 2a: mean proliferation rate of CD4+ T
cells in HAM/TSP > asymptomatic carriers
C D 4 + C D 4 5 R O + T ly m p h o cy te
5
p ro life ra tio n ra te (d a y-1 )
4 p = 0.01
(Mann-Whitney,
3
2-tailed)
2
1
0
Asymptomatic HAM/TSP
carriers patients
Asquith et al. 2007
PNAS 104, 8035-8040
24. In vivo turnover rate of CD4+CD45RO+ cells
correlates with Tax expression ex vivo
CD4 CD45RO proliferation rate in
5
4
vivo (% per day)
3
2 P = 0.016
+
(Spearman rank
1 correlation)
+
0
0 2 4 6 8 10
Tax expression ex vivo
(Tax CD4+/CD4+ after 18h culture)
+
Asquith et al. 2007
PNAS 104, 8035-8040
25. Prediction 2b:
Tax-expressing cells turn over faster
than Tax – cells in vivo
0.07
L02-HAY
0.06 Tax+ turnover rate:
0.05
0.04
0.03 7% per day
0.02 Tax –
0.01
0
0 2 4 6 8 10
0.14
L07-TBI
0.12 Tax+
0.1
0.08 13% per day
0.06
0.04 Tax –
0.02
Asquith et al. 2007
0
0 2 4 6 8 10 PNAS 104, 8035-8040
26. Prediction 3: proviral load correlates
with Tax expression
25
20
(Tax+CD4+/CD4+)
% Tax expression
y = 0.86x + 3.08
2
R = 0.71
15
HAM/TSP
AC
10
5
y = 0.61x + 0.56
2
R = 0.82
0
0 5 10 15 20
Proviral load
(% of PBMC infected)
Asquith et al. 2005: Retrovirology 2, 75
27. Prediction 4: advantage conferred by Tax
expression falls as CTL lysis rate increases
Increase in proviral load caused by
model prediction
increase in % Tax expression
experimental
results
Rate of CTL lysis Asquith et al. 2005: Retrovirology 2, 75
(Tax+ cells killed/day/CD8+ cell)
28. Quantification of HTLV-1 infection dynamics
in vivo
cytokines (Il-2, IL-15)
1d (70d)
antigen
(30d)
0.05 – 5%
per day
1d 1d
X CTL
resting
HTLV-1+
HTLV-1- 1d
cell
expressing
? cell Asquith and
Bangham 2008:
Trends Immunol.
29, 4-11
29. Tax expression in HAM/TSP > carriers
at a given proviral load
Asquith et al.
2005:
Retrovirology
2, 75-83
- and a given
FoxP3+ Toulza et al.
frequency 2008: Blood
111, 5047-5053
30. What determines the rate of HTLV-1
proviral expression in vivo?
- strain (sequence) of virus? No.
- proportion of defective proviruses? Unknown.
- T cell activation (ag, cytokines)?
Unlikely to explain observed between-individual variation.
- HTLV-1 regulatory products (p30II, Rex, HBZ)
Limit existing proviral expression, but do not control onset.
- epigenetic changes?
- genomic integration site?
31. Is HTLV-1 integration random?
Linker-mediated PCR
PBMCs taken from 10 HAM patients + 10 ACs
311 genomic integration sites mapped
Observed integration sites compared with random NlaIII
sites in genome
Kiran Meekings, 2008: PLoS Pathogens 4(3): e1000027 .
Statistical analysis carried out in collaboration with Rick Bushman,
Jeremy Leipzig, Chuck Berry
32. HTLV-1 integration frequency in vivo
correlates with gene density
10
Observed/Expected
8
6 in vitro
4 in vivo
2
0
1 2 3 4 5 6
Gene density (number genes within 25kb)
Meekings et al. 2008:
p = 1.8 x 10-5 (logistic regression) PLoS Pathogens
4(3): e1000027
33. HTLV-1 proviral integration in vivo predominates:
Observed / Expected
3
Observed / Expected
2.5
4
2
3
In Vitro 1.5
2 1
In Vivo
1 0.5
0 0
1 2 5 10 25 1 2 5 10 25
Distance from integration site (kb) Distance from Integration site (kb)
- near CpG islands - near transcription start sites
- and is associated with 50%
40%
% in g e n e
0.4 30%
Proportion of sites
0.35
0.3
20%
0.25 10%
Tax Positive
0.2
0.15 Tax Negative 0%
0.1 Meekings et al.
AC
H A M /T S P
0.05
0 2008: PLoS
1 2 5 10 25
Integration Site Window (kb)
Pathogens
4(3): e1000027
- Tax expression - and with HAM-TSP
34. Rate of proviral expression & rate of
CTL lysis determine HTLV-1 load and
risk of HAM/TSP
proviral expression
HAM HAM
fast moderate load high load
HTLV-1
AC AC
slow
low load moderate load
fast slow
CTL lysis
35. Conclusion: persistence of HTLV-1 in vivo
p12I CD4+FoxP3+
cells
histone
deacetylation
CTLs
integration into proviral infectious spread
transcriptionally expression (viral synapse)
active units
T-cell
proliferation
p30, Rex, HBZ