ORIGINAL ARTICLE

Immunomodulatory effects of Toll-like receptor-7 activation on chronic lymphocyticleukemia cells

DE Spaner1,2,3,4, Y Shi1, D White1, J Mena1, C Hammond1,4, J Tomic1,4, L He5, MA Tomai6, RL Miller6, J Booth1 and L Radvanyi5,7

1Division of Molecular and Cellular Biology, Research Institute, Sunnybrook and Women’s College Health Sciences Center,Toronto, Canada; 2Toronto-Sunnybrook Regional Cancer Center, Toronto, Canada; 3Department of Medicine, University ofToronto, Toronto, Canada; 4Department of Medical Biophysics, University of Toronto, Toronto, Canada; 5Immunology platform,Sanofi-Pasteur, Toronto, Canada and 6Department of Pharmacology, 3M Pharmaceuticals, 3M Center, St Paul, MN, USA

Weak immunogenicity of chronic lymphocytic leukemia (CLL)cells may contribute to disease progression and inhibiteffective immunotherapy. Accordingly, agents that enhancethe immunogenicity of CLL cells may be useful in immunother-apeutic approaches to this disease. Since Toll-like receptors(TLRs) are major regulators of innate immunity and initiation ofadaptive immunity, we studied the effects of viral pathogenassociated molecular pattern agonists (that are recognized byTLRs) on the costimulatory phenotype and function of CLLcells. CLL cells (especially those with high endogenousexpression of CD38) responded to TLR7-activating imidazoqui-nolines and guanosine analogs by increasing costimulatorymolecule expression, producing inflammatory cytokines, andbecoming more sensitive to killing by cytotoxic effectors.Additional activation of protein kinase C pathways increasedthe ability to stimulate T-cell proliferation, blocked phosphor-ylation of the transcription factor, signal transducer andactivator of transcription (STAT)3, and resulted in the acquisi-tion of a dendritic cell surface phenotype by TLR7-activatedCLL cells. Normal B cells also responded to TLR7 activation byincreasing costimulatory molecule expression and cytokineproduction. These findings suggest a potential role for TLR7agonists in CLL immunotherapy.Leukemia (2006) 20, 286–295. doi:10.1038/sj.leu.2404061;published online 8 December 2005Keywords: B lymphocytes; tumor immunity; chronic lymphocyticleukemia; costimulatory molecules; cytokines; toll-like receptors

Introduction

Chronic lymphocytic leukemia (CLL), while incurable withconventional chemotherapy, may be susceptible to T cell-mediated approaches.1 Consistent with this, we found thatautologous tumor vaccines increased CLL-reactive T cellactivity, and caused partial responses, in some patients.1

However, the absence of complete remissions suggested a needfor additional measures.

Cytotoxic T cells (CTLs), which are required to clear cancercells, evolved likely to deal with viral infections.2 Along withantigen-reactive T cells, successful antiviral and antitumor

responses require that the target cells be immunogenic.Immunogenicity is a complex phenotype involving the produc-tion of costimulatory molecules, cytokines, and chemokines thatpromote CTL proliferation and function. Unfortunately, tumorcells have inherently weak immunogenicity and often makeimmunosuppressive factors.3 Methods to enhance tumor cellimmunogenicity might then improve the efficacy of cancervaccines, in general, and CLL vaccines, in particular.

Signaling events in virus-infected target cells during protectiveCTL responses provide a guide to develop strategies to enhanceCLL immunogenicity.2 Viruses provide antigenic targets for CTLsand also activate the adaptive immune system through toll-likereceptors (TLRs).4 TLRs recognize pathogen-associated molecu-lar patterns (PAMPs) and activate transcription factors thatcontrol the expression of costimulatory molecules (such as theB7-family members, CD80 and CD86, the adhesion molecule,CD54, and CD83), and type 1 immunity-promoting cytokinessuch as Interferon (IFN)a and IL12. TLR1, 2, 4, 5, and 9recognize mainly bacterial PAMPs4 and CpG oligonucleotides(which activate TLR9) are known to affect immunologicalparameters of CLL cells.5 TLR3 recognizes double-stranded viralRNA4 while TLR7 (expressed mainly by B cells and monocytes)recognizes single-stranded viral RNA in endosomes andactivates NF-kB and p38 mitogen activated protein kinase(MAPK) signaling pathways through a phospho-relay systeminvolving MyD88, interleukin-1 receptor associated kinase(IRAK)1, IRAK4, and TNF receptor associated factor (TRAF)6.4

TLR7 is of clinical interest because it can be activatedpharmacologically by imidazoquinolines6 and guanosine ana-logs.7 The experiments in this paper were designed to study theeffect of TLR7 agonists on immunogenic properties of CLL cells.

Materials and methods

Cell samplesBlood was taken from consenting CLL patients (with a persistentclonal expansion of CD19þCD5þ IgMlo lymphocytes8) whowere untreated for more than 3 months. Normal B cells werefrom healthy volunteers and phlebotomized hemochromatosispatients. Protocols were approved by the Local Review Board.Patient characteristics are described in Table 1.

Antibodies and reagentsFluorescent CD38, CD80, CD86, CD54, CD83, CD5, CD19,CD40, CD95, MHC, and tumor necrosis factor (TNF) antibodieswere from Pharmingen (San Francisco, CA). Lipopolysaccharide(LPS), Loxoribine, phorbol dibutyrate (PDB), phytohemaggluti-

Received 26 July 2005; revised 24 October 2005; accepted3 November 2005; published online 8 December 2005

Correspondence: Dr DE Spaner, Division of Molecular and CellularBiology, Research Institute, S-116A, Research Building, Sunnybrookand Women’s College Health Sciences Center, 2075 Bayview Avenue,Toronto, Ontario M4N 3M5, Canada.E-mail: spanerd@srcl.sunnybrook.utoronto.ca7Current address: Department of Melanoma Medical Oncology,University of Texas, MD Anderson Cancer Center, Houston, Texas,77030, USA

Leukemia (2006) 20, 286–295& 2006 Nature Publishing Group All rights reserved 0887-6924/06 $30.00

www.nature.com/leu

nin (PHA), chloroquine, and 2-mercaptoethanol (2-ME) werefrom Sigma Chemical Co. (St Louis, MO). Dexamethasone(Pharmascience Inc., Montreal, Quebec), IL2 (Chiron Corp.,San Francisco, CA), IFNa2b (Schering, Pointe-Claire, Quebec)and Remicade (Centocor, Horsham, PA) were from the hospitalpharmacy. Poly (I:C) was from Amersham Pharmacia Biotech Inc.(Piscataway, NJ). AG490, AG126, SB203580, and Bafilomycin A1were from Calbiochem (San Diego, CA). S286909 was from 3MPharmaceuticals (St Paul, MN). IL6 and IL10 antibodies were fromR&D Systems, Inc. (Minneapolis, MN). 5,6-carboxyfluoresceindiacetate succinimidyl ester (CFSE) was from Molecular Probes(Eugene, OR). Staphylococcal enterotoxin A (SEA) was from ToxinTechnology Inc. (Madison, WI). The TACE inhibitor, TAPI,10,11

was from Peptides International (Louisville, KY).

Cell purificationCLL, normal B, and T cells were isolated as describedpreviously.12

Activation of CLL cellsCLL cells (1.5� 106/ml) were cultured in serum-free AIM-Vmedium (GibcoBRL, Grand Island, NY) plus 2-ME (5� 10�5

M)in six-well plates (Becton-Dickinson Labware, Franklin Lake, NJ)for 3–4 days. S28690 and PDB were used at 1mg/ml and100 ng/ml, respectively. IL10 and IL6 antibodies were usedat 15mg/ml.

T-cell proliferation assaysMixed lymphocyte responses (MLRs) were performed asdescribed.12 PHA (10mg/ml) was added to some cultures as apositive control for proliferation.

SEA-directed cytotoxicity assayEffectors. Lymphokine activated killer (LAK) cells were madefrom normal peripheral blood mononuclear cells (PBMCs)cultured in AIM-V with IL2 (500 U/ml) for 7–10 days and fedevery 3 days. SEA-reactive T cells were made from PBMCsstimulated with SEA (1 ng/ml) in AIM-V plus 10% FCS for 10–14days. Effectors were collected by density gradient centrifugationand suspended in AIM-V.

Targets. CLL cells were suspended at 1� 107/ml in phos-phate-buffered saline (PBS) plus 5% FCS and then pelleted andincubated with 100ml of CFSE (50 mM in PBS) for 5 min at roomtemperature in the dark before being washed and resuspendedin AIM-V.

Assay. Effectors (at 1 and 3� 106/ml) and targets (at 1� 106/ml)were mixed (1:1) in a final volume of 200ml in 5 ml polystyrenetubes (BD Biosciences, Bedford, MA) and cultured for 4 h at371C. SEA (1 ng/ml) was added to some tubes to activate effectorcells and bind them to targets.13 The samples were then washedtwice and 7-amino actinomycin D (7-AAD) and Annexin V-PEwere added, according to the manufacturer’s instructions (BD-Pharmingen).

ImmunophenotypingFlow cytometric analysis was performed as described pre-viously.14 For cellular cytotoxicity, gates were set on CFSE-stained CLL cells using an FL1-histogram. The percentages oftarget cells that bound Annexin V-PE and/or 7AAD weredetermined by further gating in an FL2/FL3 dot-plot (seeFigure 4d for an example).

ImmunoblottingImmunoblotting was performed as described14 on resting CLLcells (with IRAK1, IRAK4, MyD88 (eBioscience, San Diego, CA)and TRAF6 (Santa Cruz Biotechnology, Santa Cruz, CA)antibodies) and activated CLL cells (with IkB, phospho-IkB (Ser32) (New England BioLabs Inc., Beverly, MA), signal transducerand activator of transcription (STAT)3, phospho-STAT3 (Y705),and b-actin antibodies (Cell Signaling, Beverly, MA).

Table 1 Clinical properties of CLL patients

Pt.No.a

Sex Raistageb

Age(years)

WBC Duration Treatmentc CD38(%)

1 F I 49 32 1.5 None 102 F I 52 27 7 None 83 M 0 59 15 1 None 34 M 0 59 27 1 None 97 F 0 40 23 2 None 49 F 0 32 22 1 None ND

12 M 0 45 13 2 None 613 F I 39 25 7 None 2514 M 0 49 12 1 None ND16 F 0 73 19 11 None 817 M I 42 34 7 None ND18 M II 57 146 6 Rad 3219 M II 69 121 2 None 123 M I/II 64 125 7 None 224 M 0 46 46 3 None 425 F I 55 87 7 None 127 F I/II 52 47 1 None 928 M I 59 77 4 None ND29 F I/II 77 37 4 None 4630 F III 70 142 4 Ch 331 M IV 59 59 10 F 1432 M IV 54 125 3 Ch,F ND35 F III 70 173 2 F 8036 F III 31 1 Ch 638 M IV 50 130 2 F,Ch,P,CHOP 739 F III 56 123 7 F,R 640 M IV 53 25 1 F, CHOP ND41 M IV 66 364 10 Ch,F,S 9343 M IV 47 512 7 Ch,F 4844 F IV 48 98 11 None 249 F III 71 25 1 None 1050 M IV 54 122 5 ‘‘ ND54 M 0 76 34 2 None 155 M 0 71 42 1 None 4060 M 0 41 18 1 None 164 M IV 82 49 10 Ch, Rads 2965 M II 65 25 2 None ND66 F IV 68 156 2 F,Ch 9467 F 0 56 13 2 None ND68 M 0 74 12 1 None 2069 M 0 64 12 2 None 270 M 0 82 12 1 None 196 M II 49 71 1 None 1597 F 0 60 14 2 None 1

aPatient numbers are maintained throughout this article.bRai stage 0¼ lymphocytosis; I¼with adenopathy; II¼with hepatos-plenomegaly; III¼with anemia; IV¼with thrombocytopenia.8cCh¼Chlorambucil, Cy¼Cyclophosphamide, H¼Adriamycin,O¼Vincristine, P¼Prednisone, F¼Fludarabine, S¼Splenectomy,Rads¼ local radiation, R¼Rituxan, ND¼not done.

Immunomodulation of CLL cells by TLR7 agonistsDE Spaner et al

287

Leukemia

Real-time PCR amplificationTotal RNA and cDNA were made as described before.12 TLR7and HPRT transcripts were amplified with the primers: TLR7

forward (50-CTAAAGACCCAGCTGTGACCAG-30), TLR7 reverse(50-CCAGTCCCTTTCCTCGAGACAT-30), HPRT forward (50-GAG-GATTTGGAAAGGGTGTT-30), and HPRT reverse (50-ACAATA-

None

S28690

30

10

50

70

90

CD80 CD86 CD83 CD95

% C

ells

.001 .001

.005.005

CD80 CD86 CD83 CD95 CD54 CD40 Class Class

1400

1200

1000

800

600

400

200

0

II MHC I MHC(n=31) (n=31) (n=31) (n=31) (n=31) (n=21) (n=16) (n=20)

MF

I

.05 .001 .02 .05

.01

.05

.05 .05

CD80 (%)

CD80 (MFI)

CD86 (%)

CD86 (MFI)

Loxoribine LPS Poly I:C IFNα (n=10) (n=15) (n=3) (n=19)

10

8

6

4

2

0

Fol

d In

crea

se

a b

c

CD38+ cells (%)

CD

80+ c

ells

(%

)

y=50.2+0.3x(p<0.05)

(n=25)

10 30 50 70 90

100

80

60

40

20

0

d

60 65 28 66 55 29 18 3 70 23 44 64 43 67 1 2 3 4 5 6

Patient Numbers Control Numbers

CLL Cells Normal PBMCs

TLR

7 m

RN

A t

rans

crip

ts

450

350

250

150

500

eIRAK1

IRAK4

TRAF6

MyD88

β-actin

Patient Numbers

32 97 23 1 96 64 25 27

∆

Immunomodulation of CLL cells by TLR7 agonistsDE Spaner et al

288

Leukemia

GCTCTTCAGTCTGA-30). PCR was performed on a DNA engineOpticont System (MJ Research Inc., Waltham, MA) usingSYBR Green I as a double-strand DNA-specific binding dye.Reactions were cycled 40 times after initial denaturation (951C,15 min) using the parameters: denaturation at 951C for 15 s,annealing at 541C (TLR7) and 521C (HPRT) for 20 s, andextension at 721C for 20 s. Fluorescent data were acquiredduring each extension phase. Melting curve analysesof amplification products were performed by cooling thesamples to 41C and then increasing the temperature to 951Cat 0.21C/s. Fast loss of fluorescence is observed uniquelyat the denaturing/melting temperature of the amplified DNAfragment. Standard curves were generated with serial 10-folddilutions of cDNAs obtained with the same primers used forreal-time PCR.

Cytokine measurementCytokines in supernatants were measured with the respectiveELISA kits (Pierce Biotechnology Inc., Rockford, IL), accordingto the manufacturer’s instructions.

Statistical analysisThe Student’s t-test and paired t-tests were used to determineP-values. Best-fit lines were determined by least-squares regression.

Results

Costimulatory molecule expression on TLR7-activatedCLL cellsA number of immunogenic molecules were measured on CLLcells treated with different TLR agonists. CD54, CD80, CD86,and CD83 levels increased on all CLL patient samples(regardless of clinical stage) in response to the imid-azoquinoline, S28690. CD40, Class I and Class II MHCmolecules – already expressed highly on resting cells –increased further upon stimulation with S28690 (Figure 1a andSupplementary Figure 1). Loxoribine (also a TLR7 agonist7) hadsimilar effects as S28690, while TLR3 and -4 agonists (poly (I:C)and LPS, respectively) and the antiviral cytokine, IFNa2b, didnot change costimulatory molecule expression by CLL cellssignificantly (Figure 1b).

While all patient samples responded to S28690, heterogeneityin responsiveness was apparent. For example, increases inCD80 were found to be proportional to base-line expression ofCD38, a surface molecule associated with aggressivity of CLL8

(Figure 1c). To address the issue of variable responsiveness toS28690, components of the TLR7-signaling pathway weremeasured.4 TLR7 protein could not be detected by immunoblot-

ting with commercially available antibodies. However, allpatient samples expressed TLR7 mRNA. Absolute levels ofTLR7 message varied somewhat but no obvious correlation withresponsiveness to S28690 (determined by changes in costimu-latory molecule expression) was noted. Interestingly, TLR7mRNA expression was several-fold lower in CLL cells comparedto normal PBMCs (Figure 1d). Other components of the TLR7-signaling pathway (MyD88, IRAK1, IRAK4, and TRAF6) wereexpressed by all CLL samples with some variation and withoutobvious relationship to S28690-mediated changes in costimu-latory molecule expression (Figure 1e).

Mechanism of costimulatory molecule changesTreatment with S28690 led to the rapid activation of NFkB(determined by phosphorylation of the inhibitor protein, IkB,which releases NFkB to engage in gene transcription4)(Figure 2a) and costimulatory molecule expression was inhibitedin an additive fashion by dexamethasone (an inhibitor of NFkB-mediated gene transcription15) and SB203580 (a specific p38kinase inhibitor16) (exemplified in Figure 2b and summarized inFigure 2c). The effects of S28690 on NFkB-mediated genetranscription were also inhibited by AG126 (an IRAK1inhibitor17) and the endosomal inhibitors, chloroquine, andbafilomycin7 (Supplementary Figure 2), consistent with signalingthrough TLR7.

Indirect STAT3 activation by S28690The immunogenic phenotype required for successful antitumorT-cell responses is also affected by other signaling pathways. Inparticular, the STAT family member, STAT3, is considered anegative regulator of immunogenicity.18 STAT3 was activated inS28690-treated CLL cells (indicated by the appearance oftyrosine-phosphorylated forms (Figure 2d)). However, phos-pho-STAT3 did not appear for several hours, and could beblocked by IL6 and IL10 antibodies (Figure 2d, upper panels),suggesting it resulted from autocrine production of thesecytokines, rather than direct signaling through TLR7.

Phorbol esters also block IL10 and IL6-mediated STAT3activation.19 Consistent with this, phospho-STAT3 levels weredecreased in S28690-activated CLL cells treated simultaneouslywith PDB (Figure 2d, middle panels). Similar results wereobserved with another inhibitor of STAT3 activation, AG49018

(Figure 2d, lower panels).

Cytokine and chemokine production by S28690-treatedCLL cellsThe prevention of STAT3 phosphorylation by anti-IL6 and -IL10suggested that cytokines stimulated by S28960 might affect the

Figure 1 Costimulatory molecule expression by TLR7-activated CLL cells. (a) Purified CLL cells were cultured alone, or with S28690, and surfaceexpression of the indicated molecules was determined after 2 days by flow cytometry. The average and s.e. of the percentages of cells expressingthe molecules (top graph) and the MFIs of expression (bottom graph) were determined from the indicated number of patient samples. Numbers overthe arrows are the P-values for the differences between sample means. (b) The ‘fold-increases’ in percentage of cells expressing CD80 and CD83(and MFI of expression) were calculated from the ratio of the values obtained by flow cytometric evaluation of cells activated with Loxoribine, LPS(100mg/ml), poly (I:C) (100mg/ml) or IFNa2B (500 U/ml) for 48 h compared to cells cultured alone. The average and s.e. of these relative increases,for the number of patients indicated on the x-axis, are shown. (c) For each patient sample, the difference in CD80 expression caused by S28690was calculated by subtracting the percentage of CD80þ cells, after culture alone, from the percentage after culture with S28690. The differencewas then plotted against the initial percentage of CD38þ CLL cells. A linear regression line (determined by the method of least squares) indicates astatistically significant (Po0.05) direct correlation between baseline CD38 expression on CLL cells and change in CD80 expression caused byS28690 (r2¼ 0.69). (d) CLL cells were purified directly from the blood of 14 different CLL patients and the absolute numbers of TLR7 mRNAtranscripts determined by real-time PCR, as described in the Materials and methods. For comparison, TLR7 mRNA expression in PBMCs of sixnormal controls is shown. (e) IRAK1, IRAK4, TRAF6, and MyD88 expression by eight different CLL patient samples were determined byimmunoblotting as described in the Materials and methods.

Immunomodulation of CLL cells by TLR7 agonistsDE Spaner et al

289

Leukemia

resulting immunogenic phenotype of CLL cells. Unlike normalmonocytes,20 only 1/8 CLL patient samples produced even lowlevels of IFNa or IL12 in response to S28690 (not shown).Production of TNFa, Lymphotoxin, IL6, IL10, and the chemo-

kines, CXCL8 (IL8) and CXCL10 (IP10), were most apparent,with low amounts of IL1b and IFNg made sometimes (Figure 3).As with costimulatory molecules (Figure 1), cytokine productionwas heterogeneous.

a

0

20

40

60

80

100

120

SB203580Dexamethasone

Dex/SB203580

Inhi

bitio

n (%

)

*

*

*

*

*

*

*

*

**

b

CD86

CD

54

CD83

CD

80 35% 90% 61% 69% 36%

2% 38% 13% 11% 7%

c d

1 2 3 4

None S28690 S28690+ αIL6/10

None S28690 S28/AG S28/AG

STAT3

pSTAT3

pSTAT3

pSTAT3

STAT3

STAT3

S28690 0’ 3’ 10’ 30’

p-IκB

IκB

β-actin

Pt. 19

50µM 17µMCD80 CD86 CD83 CD54

+DexNone S28690 +SB203580 +Dex+SB203580

100

101

102

103

104

100

101

102

103

104

100 101 102 103 104100 101 102 103 104 100 101 102 103 104100 101 102 103 104 100 101 102 103 104

None S28690 PDB S28/PDB

Figure 2 Role of NFkB, p38, and STAT3 in S28690-signaling. (a) IkB phosphorylation was measured by immunoblotting before, 3, 10, and 30 minafter stimulation with S28690. Similar results were obtained with three other samples. (b) Example of the effect of previously optimizedconcentrations of dexamethasone (15mg/ml) and/or SB203580 (10mM) on S28690-mediated changes in CD80, CD86, CD83, and CD54expression, determined by flow cytometry after 48 h. The numbers in the dot-plots are the percentages of CD80þ and CD86þ cells (upper andlower rows, respectively). (c) The percent inhibition (defined as 1-(the ratio of the MFI of expression of CD80, CD86, CD54, or CD83 on CLL cells,in the presence of S28690 and the inhibitors, to the MFI in S28690 alone) �100) was calculated for nine different patient samples, treated in thesame way. The averages and s.e. are shown. * and ** represent P-values of o0.02 and 0.05, respectively, and indicate the changes in CD80,CD83, and CD54 expression were significant. (d) CLL cells were cultured alone (lane 1), with S28690 (lane 2), with S28690 and IL10 and IL6antibodies for 24 h (lane 3) (top panels); or with PDB (lane 3) or both S28690 and PDB (lane 4) (middle panels); or with S28690 and differentconcentrations of AG490 (added 1 h before S28690) (bottom panels) for 12 h before lysis and immunoblotting. The blots were probed first with aphospho-specific STAT3 antibody (top) and then with a pan-specific STAT3 antibody (bottom) as a loading control. The results were similar forthree other samples.

Immunomodulation of CLL cells by TLR7 agonistsDE Spaner et al

290

Leukemia

Given their ability to inhibit STAT3 activation, we hypothe-sized that concomitant treatment with phorbol esters wouldenhance the immunogenic properties of S28690-treated CLLcells. Consistent with this, TNFa levels were about 10-foldhigher (and a trend towards increased IL6, IL8, and IL1production was noted) in the presence of both S28690 andPDB (Figure 3).

Effect of phorbol esters on costimulatory moleculeexpression by S28690-activated CLL cellsCostimulatory molecules were also measured on CLL cellstreated concomitantly with PDB. PDB, alone, caused CD83expression by B90% of CLL cells and increased percentages ofCD80þ and CD86þ cells (Figure 4b). Remarkably, thecombination of S28690 and PDB caused essentially all

None S28690 PDB S28/PDB

None S28690 PDB S28/PDB

LT-α

(pg/

ml)

104

103

102

101

1

IL8

(pg/

ml)

None S28690 PDB S28/PDB

101

102

103

1

IL10

(pg

/ml)

None S28690 PDB S28/PDB

103

104

102

101

1

IL1β

(pg/

ml)

None S28690

103

104

102

101

1

CLL IL8 (n=14)

CLL IL10 (n=14)

CLL IL1β (n=14)

CLL Lymphotoxin-α (n=14)

None S28690 PDB S28/PDB

None S28690 PDB S28/PDB

None S28690 PDB S28/PDB

104

103

102

101

1

105

104

103

102

101

1

104

103

102

101

1

104

103

102

101

1

IL6

(pg/

ml)

TN

F (

pg/m

l)IP

10 (

pg/m

l)IF

N-γ

(pg/

ml)

CLL IFN- γ (n=4)

CLL IP10 (n=10)

CLL TNF-α (n=14)

CLL IL6 (n=11)

Figure 3 Cytokine and chemokine production by CLL cells. CLL cells, from the number of patients indicated in each graph, were cultured alone,or with S28690, PDB, or both S28690 and PDB, for 48 h. Cytokines and chemokines in the supernatants were then measured by ELISAs. Theaverages of triplicate wells are shown for each patient sample. (For comparison with normal B cells, the averages and s.e. are summarized in thebottom panel of Figure 5).

Immunomodulation of CLL cells by TLR7 agonistsDE Spaner et al

291

Leukemia

b

a

CD83

0.8% 98.5%

CD86

CD

54

CD

80

16.8% 99 %

None S28690/PDB None S28690/PDB

c

d

0

10

20

30

40

50

60

70

80

90

None S28690

% ly

sis

p<.01

Annexin-V

7AA

D +SEA

Alone

None S28690

36% 54%

10% 6%

S28690 PDB S28690/PDB PHA

P

rolif

erat

ion

(A54

0-59

5nm

x103

)

100

0

200

300

400

(n=27) (n=9) (n=12) (n=9)

(p<.05)

CD80 CD86 CD83 CD80 CD86 CD83 CD54/10

100

80

60

40

20

0 0

50

100

150

200

MF

I

% C

ells

None (n=31) PDB (n=12) S28690/PDB (n=14)

.005

.005 .001 .001.00 1

.00 1

.00 1 .02

Figure 4 Effect of S28690 and phorbol esters on immunogenic properties of CLL cells. (a) Representative CLL cells were cultured alone, or withS28690 and PDB, for 2 days. CD80, CD83, CD54, and CD86 expression were then determined by flow cytometry. The percentages refer to CD80(left panels) and CD86 (right panels). (b) Summary of flow cytometric evaluation of percentage of CLL cells expressing CD80, CD83, CD54, andCD86 (left graph) (and the MFI of expression (right graph)) after culture alone, with PDB, or with PDB and S28690. The average and s.e. of theresults from the number of patients indicated in the graph legend are shown. For CD54, only MFI (divided by 10) is shown. (c) Purified CLL cellswere cultured alone or with S28690, PDB, or PDB and S28690 for 3–4 days, and then washed extensively, irradiated (2500 cGy), and used tostimulate T cells from normal donors. For comparison, PHA was added to cultures of T cells alone. After 5–6 days, Alamar Blue was added for 12 hand the difference between the readings at wavelengths of 540 (reduced state) and 595 (oxidized state) nm was used as a measure of T-cellproliferation (after subtracting the background of T cells cultured with untreated CLL cells). The average proliferation and s.e. from the number ofpatients indicated on the x-axis are shown. (d) SEA-reactive T cells and CFSE-labeled CLL cells (that had been cultured for 2 days with or withoutS28690) were cultured for 4 h with or without additional SEA. Target cell death was then determined by flow cytometry, as described in theMaterials and Methods. A representative example at an E:T ratio of 3:1 is shown on the left. The numbers in the dot-plots are the sums of thepercentages of cells in the upper and lower right quadrants, and represent the total percentages of dying CLL cells in the presence (upper panels)and absence (lower panels) of SEA. The results with seven different patient samples are shown in the right graph. The net percentage of dying cells(subtracting background from SEA-mediated killing) in each case is shown. The P-value (obtained with a paired t test) shows significant differencesin the death of S28690-treated CLL cells.

Immunomodulation of CLL cells by TLR7 agonistsDE Spaner et al

292

Leukemia

0

50

100

150

200

250

300

B c

ell n

umbe

r (x

104 )

Pro

lifer

atio

n (A

540-

595n

mx1

03)

0

50

100

150

200

p<.05

p<.05

None S28690 PDB S28/PDB None S28690 PDB S28/PDB

% C

ells

MF

I

CD80 CD86 CD83 CD80 CD86 CD83 CD54/10

a

b c

.02 .02

.05

.02None (n=10) S28690 (n=8) PDB (n=8) S28690/PDB (n=10)

.05

.02 .02

.05

120

100

80

60

40

20

00

20

40

60

80

100

d

Normal B cells

CLL cells

(10) (12) (12) (12 (8) (8) (4) (4)

(11) (14) (14) (14) (10) (8) (20) (4)

S28690None PDB S28690/PDB25000

20000

15000

10000

5000

0

25000

20000

15000

10000

5000

0

Cyt

okin

es (

pg/m

l)

IL6 IL8 TNF IL10 IP10 IFNa IL1b IFNg

IL6 IL8 TNF IL10 IP10 IFNa IL1b IFNg

Figure 5 Effect of S28690 on normal B cells. (a) Costimulatory molecule expression: Purified B cells from the indicated donors were activatedwith S268690 and/or PDB and the percentages of cells expressing CD80, CD86, and CD83, and the MFI of expression (and also CD54), weredetermined by flow cytometry. The averages and s.e. are shown. (b) T-cell stimulation: Proliferation of T cells stimulated by irradiated, activatednormal allogeneic B cells were determined in a colorimetric assay. This experiment was repeated four times with similar results. (c) B cellclearance in the presence of T cells: Autologous B and T cells from five different donors (each at 106 cells/ml) were mixed 1:1 and cultured alone,or with S28690 and/or PDB. The numbers of remaining B cells after 5–6 days (calculated by multiplying the percentage of CD19þ cells by thenumber of viable cells (determined by manual counting with a hemocytometer)) are shown. (d) Cytokine production: Cytokines were measured asdescribed for CLL cells in the legend to Figure 3. The averages and s.e. of the results for each different cytokine from the numbers of donorsindicated in parentheses below the x-axis are shown (top panel). For comparison, the results of the individual CLL patient samples shown inFigure 3 are displayed in a similar manner (bottom panel).

Immunomodulation of CLL cells by TLR7 agonistsDE Spaner et al

293

Leukemia

CLL cells (regardless of CD38 status) to acquire a CD83hi

CD80hiCD86hiCD54hi cell surface phenotype (Figure 4a, b),which is characteristic of potent antigen presenting cells (APCs),such as dendritic cells (DCs).12 The effects on CD80 and CD54were particularly strong.

Effect of S28690 on costimulatory function andsensitivity to cytotoxic effectorsThe desired outcome of modulating immunogenic properties oftumor cells is to increase their ability to support T-cell activity andsensitivity to killing by CTLs.3 Despite increased costimulatorymolecule and cytokine expression, S28690-treated CLL cells didnot cause strong proliferation of T cells in MLRs. Concomitanttreatment with phorbol esters caused S28690-treaed CLL cells tostimulate T cell proliferation more strongly, but not to levelsreached with the T-cell mitogen, PHA (Figure 4c).

A flow cytometric assay was developed to evaluate thesensitivity of CLL cells (treated with S28690 for 2 days) to CTLs.CLL targets were labeled with CFSE and then incubated witheffector cells. Annexin-V (to label apoptotic cells) and 7-AAD (toidentify necrotic cells) were then added and the percentages ofdead and dying CLL target cells determined by gating on FITCþ

cells.21 CLL cells were relatively resistant to killing by LAK cells,despite S28690 (not shown). Using SEA-reactive T-cell lines(shown previously to kill CLL cells13) as effectors, S28690-activated CLL cells were found to be significantly moresusceptible to killing (Figure 4d).

Effect of S28690 on normal B cellsNormal B cells were studied to determine if immunomodulatoryeffects of S28690 were specific to CLL cells (Figure 5). S28690caused qualitatively similar changes in CD80, CD86, CD54, andCD83 expression on normal B cells (Figure 5a). The majordifference was that the percentage of cells that expressed CD80,and the mean fluorescence intensity (MFI) of expression,(63.774.6% and 4074.7 (n¼ 10), respectively) after treatmentwith PDB and S28690 (Figure 5a) were less than CLL cells treatedsimilarly (97.570.6% of cells with MFI of 117.3742.8 (n¼ 15))(Po0.01 for both parameters). As with CLL cells (Figure 4c), theability of normal B cells to stimulate T-cell proliferation inallogeneic MLRs was increased only marginally by S28690 butincreased further with concomitant PDB treatment (Figure 5b).

Potential T-cell-mediated destruction of normal B cells thathave been made more immunogenic by S28690 in vivo is animportant clinical consideration. To model this situation, purifiedautologous T cells (which do not express TLR722) were culturedwith B cells in the presence of S28690 and/or PDB. The absolutenumber of B cells remaining at the end of the culture period wasthen calculated from the percentage of CD19þ cells and the totalviable cell number (Figure 5c). S26890 generally increased B cellnumbers at the end of the culture period. The effect of PDB wasmore variable, but resulted generally in more B cell clearance.

The spectrum of cytokines made by normal B cells in responseto S28690 was similar to CLL cells. However, the magnitude ofIL6, IL8 IL1b, and IFNg production was higher and stimulationwith both S28690 and PDB resulted in much greater productionof IL8 and IP10 (compare Figure 5d, top and bottom panels).

Discussion

The results in this paper show that TLR7 regulates a number ofimmunogenic properties of CLL cells. S28690 increased

costimulatory molecule expression by CLL cells (Figure 1)through TLR7-signaling pathways (Figure 2, SupplementaryFigure 2). Activation with both S28690 and PDB, caused CLLcells to acquire uniformly a cell surface phenotype character-ized by high CD80, CD86, CD54, and CD83 expression (Figure4a, b) and to increase proinflammatory cytokine production(especially TNFa) (Figure 3), while decreasing phosphorylatedSTAT3 levels (Figure 2d). These features are shared by matureDCs, which are considered the most potent APCs.12

Despite near uniformity in response to S28690 and phorbolesters (Figure 4b), heterogeneity in costimulatory moleculeexpression and cytokine production was observed when CLLcells were treated with S28690 as a single agent (Figure 1). Thisheterogeneity was related to CD38 expression (Figure 1) andperhaps to variations in expression of TLR7 (Figure 1d) and itsassociated signaling molecules (such as MyD88 and IRAK1(Figure 1e)). Further studies are required to explain the variationin TLR7-responsiveness, which may reflect the underlyingcytogenetic aberrations that drive CLL progression8 or possiblyprevious activation of CLL cells in vivo by endogenous TLRligands or cytokines whose receptors share components of theTLR7-signaling pathway.23

Previously, we found that autologous vaccines (made fromoxidized tumor cells) caused partial clinical responses in someCLL patients.1 The observation that CLL cells invariablyacquired properties of DCs when treated with both S28690and a protein kinase C (PKC) agonist could be exploited to makemore potent patient-specific vaccines in vitro. CLL cells treatedwith S28690 alone did not stimulate strongly T-cell proliferation(Figure 4c) but were killed more easily by superantigen-activated T cells in a redirected cytotoxicity assay (Figure 4e).While this assay does not mimic a putative in vivo situation withfidelity, it may suggest that S28690-activated CLL cells may bemore susceptible to killing by tumor-reactive T cells. Themechanism of enhanced killing by CTLs is not clear. It does notappear to be mediated through TNFa receptors (as TNFaantibodies did not block killing (not shown)) but may be relatedto enhanced expression of FAS by S28690-activated CLL cells(Figure 1a). Another TLR7-agonist (Loxoribine) increased thesensitivity of CLL cells to chemotherapeutic agents.24 Wesuggest that the ability of TLR7-agonists to promote killing ofCLL cells may be exploited in treatments for CLL.

Consistent with this, we found that topical administration of animidazoquinoline (Imiquimod) resulted in clearance of leukemicskin infiltrates in a CLL patient.9 Although little impact onsystemic disease was observed with this route of administration,the strong local effects (and the results of this paper) justify aclinical trial to investigate the effects of intravenous administra-tion of imidazoquinolines to achieve higher concentrations. Sucha trial would establish the toxicity in CLL patients (including thepotential effects on normal B cells (Figure 5)) and may reveal thatTLR7-agonists have activity as single agents (perhaps by sensitiz-ing CLL cells to tumor-reactive T cells found in some patients14).Alternatively, infusions of imidazoquinolines (to sensitize CLLcells to CTLs) following cancer vaccines or adoptive T-celltransfer (to increase CLL-reactive T cell numbers and activity) maybe an effective immunotherapeutic strategy for CLL.

Acknowledgements

Supported by grants from the Leukemia Research Fund of Canadaand the Ontario Cancer Research Network (OCRN) (to DS).RM and MT are employed by a company whose (potential)product was studied in this work.

Immunomodulation of CLL cells by TLR7 agonistsDE Spaner et al

294

Leukemia

References

1 Spaner DE, Hammond C, Mena J, Foden C, Deabreu A. A phase I/IItrial of oxidized autologous tumor vaccines during the ‘watch andwait’ phase of chronic lymphocytic leukemia. Cancer ImmunolImmunother 2005; 54: 635–646.

2 Zinkernagel RM. Immunity against solid tumors? Int J Cancer 2001;93: 1–5.

3 Spaner D. Amplifying cancer vaccine responses by modifyingpathogenic gene programs in tumor cells. J Leukoc Biol 2004; 76:338–351.

4 Kawai T, Akira S. Pathogen recognition with Toll-like receptors.Curr Opin Immunol 2005; 17: 338–344.

5 Decker T, Schneller F, Sparwasser T, Tretter T, Lipford GB, WagnerH et al. Immunostimulatory CpG-oligonucleotides cause prolifera-tion, cytokine production, and an immunogenic phenotype in CLLB cells. Blood 2000; 95: 999–1006.

6 Hemmi H, Kaisho T, Takeuchi O, Sato S, Tomizawa H, Takeda Ket al. Small anti-viral compounds activate immune cellsvia the TLR7-signaling pathway. Nat Immunol 2002; 3:196–200.

7 Lee J, Chuang TH, Redecke V, She L, Pitha PM, Carson DA et al.Molecular basis for the immunostimulatory activity of guaninenucleoside analogs: activation of TLR7. Proc Natl Acad Sci USA2003; 100: 6646–6651.

8 Shanafelt TD, Geyer SM, Kay NE. Prognosis at diagnosis:integrating molecular biologic insights into clinical practice forCLL patients. Blood 2004; 103: 1202–1210.

9 Spaner D, Miller R, Mena J, Grossman L, Sorrenti V, Shi Y.Regression of skin deposits in a CLL patient treated with the TLR7/8agonist, Imiquimod. Leuk Lymphoma 2005; 46: 935–939.

10 Darlak K, Miller RB, Stack MS, Spatola AF, Gray RD. Thiol-basedinhibitors of mammalian collagenase. J Biol Chem 1990; 265:5199–5205.

11 Bueno C, Rodriguez-Caballero A, Garcia-Montero A, Orfao A.A new method for detecting TNFa-secreting cells usingdirect-immunofluorescence. J Immunol Methods 2002; 264:77–87.

12 Hammond C, Shi Y, Mena J, Tomic J, Cervi D, He L et al. Effect ofserum and antioxidants on the immunogenicity of PKC-activatedCLL cells. J Immunother 2005; 28: 28–39.

13 Wallgren A, Festin R, Gidlof C, Dohlsten M, Kalland T, TottermanTH. Efficient killing of B-CLL cells by superantigen-directed T cells.Blood 1993; 82: 1230–1238.

14 Gitelson E, Hammond C, Mena J, Lorenzo M, Berinstein N,Spaner D. CLL-reactive T cells during tumor progression and afterautologous tumor vaccines. Clin Cancer Res 2003; 9: 1656–1665.

15 Auphan N, DiDonato JA, Helmberg A, Karin M. Immunosuppres-sion by glucocorticoids: inhibition of NFkB activity throughinduction of IkB synthesis. Science 1995; 270: 286–290.

16 Lee JC, Kassis S, Kumar S, Badger A, Adams JL. p38 mitogen-activated protein kinase inhibitors – mechanisms and therapeuticpotentials. Pharmacol Ther 1999; 82: 389–397.

17 Li L, Cousart S, Hu J, McCall CE. Characterization of interleukin-1receptor-associated kinase in normal and endotoxin-tolerant cells.J Biol Chem 2000; 275: 23340–23345.

18 Gamero AM, Young HA, Wiltrout RH. Inactivation of Stat3 intumor cells: releasing a brake on immune responses againstcancer? Cancer Cell 2004; 5: 111–112.

19 Sengupta TK, Talbot ES, Scherle PA, Ivashkiv LB. Rapid inhibitionof IL6 signaling and Stat3 activation mediated by MAP kinases.Proc Natl Acad Sci USA 1998; 95: 11107–11112.

20 Gorden KB, Gorski KS, Gibson SJ, Kedl RM, Tomai MA, Alkan SSet al. Synthetic agonists reveal differences between human TLR7and TLR8. J Immunol 2005; 174: 1259–1268.

21 Lecoeur H, Fevrier M, Garcia S, Riviere Y, Gougeon ML. A novelflow cytometric assay for characterization of cell-mediatedcytotoxicity. J Immunol Methods 2001; 253: 177–187.

22 Bekeredjian-Ding IB, Wagner M, Schnurr M, Endres S, HartmannG. Plasmacytoid DCs control TLR7 sensitivity of naive B cells viatype I IFN. J Immunol 2005; 174: 4043–4050.

23 Fan H, Cook JA. Mechanisms of endotoxin tolerance. J EndotoxinRes 2004; 10: 71–84.

24 Tosi P, Zinzani PL, Pellacani A, Ottaviani E, Magagnoli M, Tura S.Loxoribine affects fludarabine activity on freshly isolated B-CLLcells. Leuk Lymphoma 1997; 26: 343–348.

Supplementary Information is available on the Leukemia website (http://www.nature.com/leu/).

Immunomodulation of CLL cells by TLR7 agonistsDE Spaner et al

295

Leukemia