Lymphocyte kinetics in HTLV-1 infection

Charles Bangham

Department of Immunology

Wright-Fleming Institute

Human T-lymphotropic virustype 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.

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?

The proviral load of HTLV-1 is high and correlates with the risk of HAM/TSP

Nagai et al 1998 J. Neurovirol. 4, 586

median proviral load

(copies/100 PBMCs)

HAM/TSP: 5.4

asymptomatic: 0.34

How is the high proviral load maintained?

Retroviruses replicate by two routes:

mitotic

infectious

host cell

polymerase

- provirus latent

- little sequencevariation in virus

reverse

transcriptase

- provirus

- sequence

expressed

variation

Evidence for latency of HTLV-1

amino acid position

variability (Kabat-Wu)

HIV-1

HTLV-1

Env

1. HTLV-1 varies little in sequence:

2. HTLV-1 mRNA and proteins are usually undetectable in PBMCs.

3. Virions are absent and plasma is non-infectious.

Daenke et al. 1990: J. Virol. 64, 1278

‘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

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?

• abundant

• chronically activated

Anti-HTLV-1 CTLs are -

• chiefly directed against the HTLV-1 Tax protein

Bangham 2000 Curr Opin Immunol 12, 397

but CTL frequency, specificity and activation do not differbetween HAM/TSP and asymptomatics

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?

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:

=SN DD / 1.15AC (N=4) HAM (N=4)

0.42

2. Naturally occurring Tax variants escape CTL recognition:

Niewiesk et al 1994: J. Virol. 68, 6778Niewiesk et al 1995: J. Virol. 69, 2649

% specific lysis

Positive selection at individual Tax residues

HLA-A2-restricted CTL epitopes

Positive selection at individual residues

AC1 CD8L cluster: 33 genes

HAM CD8Lcluster: 17 genes

A1 (AJ001687): NKG2D2 (M57888): Granzyme B3 (U20350): CX3CR14 (M30894): TCR-gamma5 (M12824): CD8 A6 (AF031824): Leukocystatin7 (M18737): Granzyme A8 (M85276): Granulysin (NKG5)9 (S69115): NKG7

D1 (M17016): Granzyme B precursor2 (M1121): RANTES (1)3 (M1121): RANTES (2)

B1 (X57352): IFITM3 (I-8U)

C1 (U26174): Granzyme K2 (M28393): Perforin

AC2 CD8L cluster: 16 genes

9 genes were overexpressed in CD8+ cells from individuals with a low HTLV-1 proviral load

Vine, Heaps et al., 2004: J Immunol 173, 5121

5

91 2

3 220

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

• halved the odds of HAM/TSP

Jeffery et al: PNAS (1999) 96, 3848; J Immunol (2000) 165, 7278Vine et al: J Infect Dis (2002) 186, 932

2. HLA class 1 heterozygosity was associated with a lower proviral load.

HLA-A2 and HLA-Cw8 prevent 36% of potential HAM/TSP cases.

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.

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.

yzcdtdy ε−=

y = freq. of infected cells

z = freq. of CTLs

ε = lysis rate constant - ‘CTL efficiency’

c = constant

Quantification of anti-viral CTL efficiency

HAM patient

carrier

Asquith, Mosley et al., 2005: J Gen Virol 86, 1515

0

1

2

3

4

5

6

0 10 20 30 40 50 60 70

% CD8+ cells

% T

ax+

CD

4+ c

ells

a

fter 1

8hrs

repeat 1: observed data

repeat 1: best theoretical fit

repeat 2: observed data

repeat 2: best theoretical fit

80% by block of perforin

29% by partial HLA mismatch

100% by full HLA mismatch

ε is reduced:

Quantification of anti-viral CTL efficiency

Asquith, Mosley et al., 2005 J Gen Virol 86, 1515

Prediction 1: CTL lytic efficiency determines HTLV-1 proviral load in vivo

prov

iral l

oad

(cop

ies/

100

PB

MC

)

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.

Asquith, Mosley et al.2005: J Gen Virol 86, 1515

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.

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

2H-glucose kinetics in vivo: CD45RO+ cells turn over faster than CD45RA+

0

0.05

0.1

0.15

0 2 4 6 8 10

L01-HS

0.00

0.05

0.10

0.15

0 2 4 6 8 10

L07-TBI

0.00

0.05

0.10

0.15

0 2 4 6 8 10

L10-HBO

time (d)

2H incorporation

AC 1

AC 2

HAM 1

CD45RO+

CD45RA+

Asquith et al. 2006: submitted

Prediction 2a: mean proliferation rate of CD4+ T cells in HAM/TSP > asymptomatic carriers

0

1

2

3

4

5C

D4+ C

D45

RO

+ T ly

mph

ocyt

epr

olife

ratio

n ra

te (d

ay-1

)

Asymptomatic carriers

HAM/TSP patients

Asquith et al. 2007 PNAS 104, 8035-8040

p = 0.01 (Mann-Whitney,

2-tailed)

In vivo turnover rate of CD4+CD45RO+ cells correlates with Tax expression ex vivo

0

1

2

3

4

5

0 2 4 6 8 10

Tax expression ex vivo(Tax+CD4+/CD4+ after 18h culture)

CD

4+ CD

45R

O+ p

rolif

erat

ion

rate

in

vivo

(% p

er d

ay)

P = 0.016 (Spearman rank correlation)

Asquith et al. 2007 PNAS 104, 8035-8040

Prediction 2b:Tax-expressing cells turn over faster

than Tax – cells in vivo

00.010.020.030.040.050.060.07

0 2 4 6 8 10

00.020.040.060.08

0.10.120.14

0 2 4 6 8 10

L02-HAY

L07-TBI

Tax+

Tax+

Tax –

Tax –

turnover rate:

7% per day

13% per day

Asquith et al. 2007 PNAS 104, 8035-8040

Prediction 3: proviral load correlates with Tax expression

y = 0.86x + 3.08R2 = 0.71

y = 0.61x + 0.56R2 = 0.82

0

5

10

15

20

25

0 5 10 15 20

Proviral load (% of PBMC infected)

% T

ax e

xpre

ssio

n (T

ax+C

D4+

/CD

4+)

HAM/TSPAC

Asquith et al. 2005: Retrovirology 2, 75

Prediction 4: advantage conferred by Tax expression falls as CTL lysis rate increases

model prediction

experimental results

Asquith et al. 2005: Retrovirology 2, 75

Incr

ease

in p

rovi

ral l

oad

caus

ed b

y in

crea

se in

% T

ax e

xpre

ssio

n

Rate of CTL lysis (Tax+ cells killed/day/CD8+ cell)

Quantification of HTLV-1 infection dynamics in vivo

resting HTLV-1+

cellHTLV-1-expressing cell

0.05 – 5% per day

1d

(30d) XCTL

1d

1d (70d)

1d

cytokines (Il-2, IL-15) antigen

? Asquith and Bangham 2008: Trends Immunol. 29, 4-11

Tax expression in HAM/TSP > carriers at a given proviral load

Asquith et al. 2005: Retrovirology 2, 75-83

Toulza et al. 2008: Blood 111, 5047-5053

- and a given FoxP3+

frequency

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?

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 NlaIIIsites in genome

Kiran Meekings, 2008: PLoS Pathogens 4(3): e1000027 .Statistical analysis carried out in collaboration with Rick Bushman, Jeremy Leipzig, Chuck Berry

HTLV-1 integration frequency in vivo correlates with gene density

p = 1.8 x 10-5 (logistic regression)Meekings et al. 2008: PLoS Pathogens 4(3): e1000027

0

2

4

6

8

10

1 2 3 4 5 6

Gene density (number genes within 25kb)

Obs

erve

d/E

xpec

ted

in vitro

in vivo

HTLV-1 proviral integration in vivo predominates:

01

23

4

1 2 5 10 25

Distance from integration site (kb)

Obs

erve

d / E

xpec

ted

In VitroIn Vivo

00.5

11.5

22.5

3

1 2 5 10 25

Distance from Integration site (kb)

Obs

erve

d / E

xpec

ted

00.050.1

0.150.2

0.250.3

0.350.4

1 2 5 10 25

Integration Site Window (kb)

Prop

ortio

n of

site

s

Tax PositiveTax Negative 0%

10%

20%

30%

40%

50%

AC

HA

M/T

SP

% in

gen

e

- near CpG islands - near transcription start sites

- Tax expression

- and is associated with

- and with HAM-TSP

Meekings et al. 2008: PLoS Pathogens 4(3): e1000027

Rate of proviral expression & rate of CTL lysis determine HTLV-1 load and

risk of HAM/TSP

fast

slow

fast

slow

HAM

moderate load

HAM

high load

AC

low load

AC

moderate load

CTL lysis

HTL

V-1

prov

iral e

xpre

ssio

n

Conclusion: persistence of HTLV-1 in vivo

proviralexpression

CTLs

T-cellproliferation

integration into transcriptionally active units

histone deacetylation

CD4+FoxP3+

cells

p30, Rex, HBZ

infectious spread (viral synapse)

p12I