Toll-like receptor 9 signaling by CpG-B oligodeoxynucleotides induces an apoptotic pathway in human chronic lymphocytic leukemia B cells

Chronic lymphocytic leukemia (CLL) accounts for 25% to 30% of all leukemias and is the most prevalent type of leukemia in the Western world. Despite major advances in our understanding of the genetics, biology, diagnosis, clinical staging, and management of CLL,^1-5  the overall median survival of CLL patients remains approximately 6 years and is shorter for patients in the high-risk group. The disease is characterized by the progressive accumulation of long-lived malignant B cells with a characteristic CD19⁺CD5⁺CD23⁺BCR^low phenotype. These B-CLL cells accumulate in the G₀ phase of the cell cycle with a low proliferative index and have acquired resistance to apoptosis via aberrant up-regulation of prosurvival mechanisms.^6-10  Because B-CLL cells do not divide at a high rate, they are less susceptible to radiation therapy and chemotherapy. Alternative therapeutic approaches are needed to achieve longer survival and ultimately a cure in CLL patients. Immunotherapy is a promising approach for the treatment of CLL.^11-15  B-cell surface differentiation antigens and receptors can serve as tissue-specific targets for immunotherapy. Antibodies (Abs), such as the anti-CD20 (rituximab),^16-18  the anti-CD52 (alemtuzumab, Campath-1H),^19-21  and the anti-CD40 antagonist Ab (HCD122),²²  are being used clinically to achieve a complete and prolonged remission in CLL patients. However, most CLL patients who undergo antibody-based treatments eventually relapse. New immunotherapeutic strategies are needed to effectively prevent and treat leukemia recurrences.

Toll-like receptors (TLRs) are pattern recognition receptors that trigger innate and adaptive immune responses. The human TLR family includes 10 different TLRs that compose an elegant pathogen recognition system for host defense in innate immunity.²³  Triggering TLRs activates a set of common proinflammatory genes and leads to the expression of antimicrobial effector molecules as well as inflammatory cytokine production. Agonists for human TLR1 to TLR9 have been identified and are being developed as new drugs and vaccine adjuvants to treat cancer, allergies, and infectious diseases.^24,25  We are interested in exploring TLR-targeted immunotherapy of CLL with TLR agonists and in elucidating the molecular mechanism(s) underlying TLR agonist-mediated antitumor effects. Although the TLR expression profile in human B-CLL cells is unknown, studies have shown that TLR9 agonists, CpG oligodeoxynucleotides (CpG ODN), can activate B-CLL cells to up-regulate cell surface expression of costimulatory molecules and induce cytokine production.^26,27  In this study, we show that human B-CLL cells express high levels of TLR9. Triggering TLR9 with CpG-B ODN leads to nuclear factor-κB (NF-κB)–dependent production of autocrine interleukin-10 (IL-10), which activates JAK/STAT pathway-dependent tyrosine phosphorylation of signal transducer and activator of transcription 1 (STAT1) and thereby provokes an apoptotic death pathway in B-CLL cells.

Blood samples from 23 untreated CLL patients were drawn after obtaining written informed consent approved by the University of Minnesota Institutional Review Board in accordance with the Declaration of Helsinki. The sex, age, Rai stage, percentage of blood CD5⁺CD19⁺CD23⁺ B-CLL cells, and cytogenetic and mutational status of these CLL patients are summarized in Table 1. Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll-Paque density gradient centrifugation. B-CLL cells in PBMCs were purified using untouched B-cell isolation kit (Miltenyi Biotec) to more than 95% of CD5⁺CD19⁺ cells. Normal B and T cells were purified from PBMCs of healthy blood donors using B-cell or T-cell isolation kit (Miltenyi Biotec) to more than 98% of CD19⁺ cells or CD3⁺ cells, respectively.

Phosphorothioated unmethylated type A CpG ODNs (2216, 5′-ggGGGACGATCGTCgggggG-3′; PB4, 5′-tcgGACGATCGTCgggggG-3′; 1002, 5′-ggGGTCGTTCGTCGTTgggggG-3′) and type B CpG ODNs (2006, 5′-tcgtcgttttgtcgttttgtcgtt-3′; 684, 5′-tcgacgttcgtcgttcgtcgttc-3′; 685, 5′-tcgtcgacgtcgttcgttctc-3′) were purchased from Intergrated DNA Technologies. Lowercase letters represent phosphorothioate linkage, and upper letters represent phosphodiester linkage. CpG ODNs were resuspended in Tris-ethylenediaminetetraacetic acid buffer, diluted in phosphate-buffered saline (PBS), and tested at a final concentration of 5 μg/mL or as indicated.

RNA was extracted from 2 × 10⁶ purified B-CLL, normal B, or T cells, reversely transcribed to cDNA, and analyzed for TLR1-10 expression and real-time reverse-transcribed polymerase chain reaction (RT-PCR) for TLR9 expression. All PCR parameters and primer sequences for TLR1 to TLR-10 and β-actin were as published.²⁸  All RT-PCR reagents were from Invitrogen.

To determine the immune-stimulatory effects of CpG ODNs, purified B-CLL cells (1 × 10⁶/well) were cultured in 10% human AB serum RPMI 1640 media with or without CpG ODN in 48-well plates. After 3 days, cells were harvested and assessed for morphologic changes by Giemsa staining²⁹  and for phenotypic changes by flow cytometry. To determine CpG ODN treatment time or doses on apoptosis induction, B-CLL cells (2 × 10⁵/well) were cultured in media with or without CpG ODNs at 5 μg/mL for up to 9 days or at different doses (0.1-10 μg/mL) for 5 days in 96-well plates. Several wells of B-CLL cells from each culture condition were harvested at the indicated time points, counted, and stained with tetramethyl-rhodamine ethylester (TMRE, Invitrogen) or annexin V-FITC Apoptosis Detection Kit I (BD Biosciences). To determine the minimum treatment dose and exposure time required for CpG ODNs to achieve a 50% reduction of leukemia cells, B-CLL cells (4 × 10⁵/well) were cultured in 96-well plates in media with different concentrations of CpG ODNs for the indicated exposure time lengths. At the end of each designated treatment time point, cells were washed twice in the wells, resuspended in fresh media, and continued in culture for a total of 7 days. B-CLL cells of all treatment groups were then harvested, counted, stained with TMRE, and determined by flow cytometry. Viable B-CLL cell number was calculated by multiplying total cell counts with the percentage of TMRE-positive cells of each culture condition.

Cell surface markers were analyzed by staining with fluorescent Abs against CD3, CD5, CD19, CD23, CD40, CD45, CD54, CD80, CD86, HLA-ABC, HLA-DR, or isotype control Ab (BD Biosciences). Apoptotic B-CLL cells were assessed by TMRE or annexin V/propidium iodide (PI) staining. Mean fluorescence intensity (MFI) and positive cell percentages were determined.

B-CLL cells cultured with or without CpG ODNs for 3 days were irradiated at 30 Gy and used as stimulators to purified allogeneic T cells (1 × 10⁵/well) in mixed leukocyte reaction (MLR) assays. Plates were incubated at 37°C for 5 days and pulsed with 1 μCi/well of ³H-thymidine for the last 18 hours before harvesting. All determinations were conducted in triplicate and ³H-thymidine incorporation (cpm) was determined.

Eight-week-old female NOD-scid IL2Rγ^null mice, a NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ mouse strain that lacks of T, B, and natural killer (NK) cells (The Jackson Laboratory), were used. To determine whether CpG 2006 pretreated B-CLL cells were programmed to undergo apoptosis in vivo, groups of mice were injected intraperitoneally with 4 × 10⁷ purified B-CLL cells with or without 48 hours of preincubation with CpG 2006. To determine whether administration of CpG 2006 can inhibit the engraftment and survival of B-CLL cells in vivo, mice were intraperitoneally injected with 4 × 10⁷ B-CLL cells followed by 5 daily intraperitoneal injection with CpG 2006 (5 mg/kg body weight) or PBS. At day 5, cells from peritoneal cavity and spleen of each mouse were harvested, stained with anti-hCD45/CD19 Abs, and analyzed by flow cytometry. The number of B-CLL cells present in peritoneal cavity and spleen were calculated by multiplying total cell counts with the percentage of hCD45⁺/CD19⁺ cells.

Protein lysates were prepared from 5 × 10⁶ purified B-CLL, normal B, and T cells. Western blot was performed using primary Abs specific for TLR9, total, tyrosine- or serine-phophorylated STAT1 (Santa Cruz Biotechnology), cleaved caspase-3, caspase-9, or poly(ADP-ribose) polymerase (PARP; Cell Signaling Technology), respectively, and secondary anti–rabbit IgG-horseradish peroxidase (Santa Cruz Biotechnology), and visualized by enhanced chemiluminescence (Pierce Chemical). Blots were reprobed with an anti–β-actin Ab (Santa Cruz Biotechnology) as a loading control. Signals intensity was quantified using a Fluorochem 8000 imaging system (Alpha Innotech).

To block apoptosis with caspase inhibitors, B-CLL cells (2 × 10⁵/well) were pretreated with 20μM pan-caspase, caspase-3, or caspase-9 inhibitor (R&D Systems) for 2 hours in 96-well plates and cultured in media with CpG 2006. To block apoptosis with inhibitors specific for NF-κB and JAK/STAT pathways, B-CLL cells were pretreated with 5 μg/mL caffeic acid phenethyl ester (CAPE) and/or 50μM AG490 (Calbiochem) for 1 hour and cultured in media with or without CpG 2006. Kinetic expressions of tyrosine- or serine-phosphorylated STAT1 were determined by Western blot. Culture supernatants at 72 hours were assessed for cytokine production. AG490 pretreatment on CpG 2006-induced cleavage of caspase-9, caspase-3, and PARP in B-CLL cells were determined at day 5 by Western blots. To block apoptosis with STAT1 antisense ODN, B-CLL cells were pretreated with 12.5μM STAT1 phosphorothioated antisense ODN (5′-CCACTGAGACATCCTGCCACC-3′)³⁰  or sense ODN (5′-GGTGGCAGGATGTCTCAGTGG-3′) from Operon Biotechnologies for 24 hours and cultured in media with or without CpG 2006. Total, tyrosine- or serine-phosphorylated STAT1 expression was assessed by Western blot at 0, 24, and 48 hours. B-CLL cells in cultures were assessed for phenotypic changes at day 3 and apoptosis at day 5.

Serum-free culture supernatants from B-CLL cells incubated with or without CpG 2006 were collected at 72 hours or the indicated time points and assessed for cytokines. IL-1β, IL-2, IL-4, IL-6, IL-8, IL-10, interferon-γ (IFN-γ), tumor necrosis factor-α (TNF-α), granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage colony-stimulating factor (GM-CSF), and vascular endothelial growth factor (VEGF) were assayed with the Fluorokine MAP immunoarray (R&D Systems) using the Luminex 100 analyzer. IFN-α, IFN-β, and transforming growth factor-β1 (TGF-β1) were determined by enzyme-linked immunosorbent assay (R&D Systems). The lower limits of each cytokine detection were as follows (pg/mL): IL-1β, 0.15; IL-2, 0.2; IL-4, 0.7; IL-6, 0.5; IL-8, 0.2; IL-10, 0.3; IFN-γ, 4.0; TNF-α, 0.4; G-CSF, 1.0; GM-CSF, 2.0; VEGF, 0.4; IFN-α, 3.0; IFN-β, 50; TGF-β1, 4.6. The background level of TGF-β1 in serum-free medium alone is at 200 pg/mL.

Anti–IL-10 (10 μg/mL) or anti–TNF-α (10 μg/mL) Ab (R&D Systems) was added to B-CLL cell cultures with or without CpG 2006. The expression of tyrosine- or serine-phosphorylated STAT1 in B-CLL cells was assessed by Western blot at the indicated time points. In some experiments, graded doses (0.02-50.0 ng/mL) of exogenous rh-IL-10 (R&D Systems) were added to B-CLL cell cultures with or without the presence of anti–IL-10 or anti–TNF-α Ab. B-CLL cells in cultures with or without CpG 2006 and/or rh-IL-10 (5 ng/mL) were assessed for phenotypic changes at day 3 and apoptosis at day 5. Tyrosine- or serine-phosphorylated STAT1 expression in B-CLL cells cultured with or without IL-10 (5 ng/mL) in the presence or absence of anti–IL-10 Ab (10 μg/mL) were determined by Western blot at 24 hours.

Data from experiments are expressed as the mean plus or minus SD. Statistical analysis of the results between groups was performed by Student t test. Values of P less than .05 were considered significant.

B-CLL cells from blood samples of CLL patients express a characteristic CD19⁺CD5⁺CD23⁺ malignant B-cell phenotype. Like normal B cells, B-CLL cells expressed TLR1, TLR6, TLR7, TLR9, and TLR10 (Figure 1A; and data not shown). Real-time RT-PCR results showed that TLR9 gene expression was significantly higher (3.2-fold) in B-CLL cells than in normal B cells (Figure 1B). Western blot analysis confirmed that TLR9 protein expression was 3.5-fold higher in B-CLL cells than in normal B cells (Figure 1C). To study the immune-stimulatory effects of TLR9 agonists on B-CLL cells, 2 distinctive classes of CpG-A and CpG-B ODNs³¹  were used. Freshly isolated B-CLL cells display a typical leukemic B-cell morphology (Figure 1D). B-CLL cells cultured in media alone or with CpG-A ODN (CpG 2216) displayed a typical CLL morphology with cytologic features similar to that of fresh B-CLL cells. In contrast, CpG-B ODN (CpG 2006)–stimulated B-CLL cells formed aggregated cell clusters and underwent dramatic cytologic changes with marked cellular enlargement and abundant basophilic cytoplasm within 3 days in culture (Figure 1D). Fresh B-CLL cells are positive for HLA-ABC and HLA-DR but express negligible levels of CD40, CD54, CD80, and CD86. B-CLL cells cultured in media or with CpG 2216 for 3 days only modestly increased B-CLL cell surface expression of CD54, CD86, and HLA-DR (2.9-, 2.5-, and 2.2-fold, respectively). In contrast, more substantial and significant increases of surface expression of CD40, CD54, CD86, HLA-ABC, HLA-DR (11.5-, 12.5-, 5.2-, 4.5-, and 4.6-fold, respectively) on B-CLL cells were detected within 3 days after CpG 2006 stimulation (Figure 1E-F). We have comparatively tested additional CpG-A and CpG-B ODNs and consistently found that CpG-B ODNs strongly activate, whereas CpG-A ODNs weakly activate, B-CLL cells to up-regulate cell surface CD40 expression (Figure 1G). Fresh B-CLL cells are poor stimulators to allogeneic T cells in MLR.^32,33  Whereas untreated and CpG 2216-treated B-CLL cells induced weak proliferative responses from allogeneic T cells, CpG 2006-treated B-CLL cells induced a strong proliferation of allogeneic T cells in MLR (Figure 1H). These results demonstrate that triggering TLR9 with CpG-B ODNs potently activates B-CLL cells and induces phenotypic changes, consistent with the enhanced immunogenicity.

B-CLL cells can be sustained in culture for weeks despite undergoing spontaneous apoptosis. Kinetic analysis of B-CLL cell apoptosis showed that 10% to 25% of cells became apoptotic during a 9-day culture in media alone and 70% to 80% of the starting number of B-CLL cells could be recovered from days 5 to 9. The viability and cell counts of the CpG 2216-treated B-CLL group on days 1 to 9 were similar to that of B-CLL cells cultured in media only. Interestingly, CpG 2006-stimulated B-CLL cells maintained good viability during the first 2 days and then underwent profound apoptosis and greatly decreased the annexin V/PI-negative, TMRE-positive viable B-CLL cell number from days 3 to 9 (Figure 2A). Kinetic analysis showed significant decreases of annexin V/PI-negative, TMRE-positive viable B-CLL cells in cultures with CpG 2006 or CpG 685 at days 3 to 9 (Figure 2B). Similarly, the addition of CpG-B ODNs into patients' PBMCs containing different percentages (96%, 79%, or 47%) of B-CLL cells invariably induced the apoptosis and decreased the number of B-CLL cells in culture, in contrast to the results of B-CLL cells cultured in media only or with CpG-A ODNs (Figure 2C). The results of CpG-B ODN-induced apoptosis of B-CLL cells are consistent in 18 of 23 CLL patients (Table 1; Figure 2D). We also found that apoptotic B-CLL cells (R1 gate) lost their characteristic expression of CD23 and decreased surface expression of CD5 and CD19, whereas viable B-CLL cells (R2 gate) remained positive for CD23 (data not shown). B-CLL cells from 5 patients (patients 6, 10, 15, 17, and 20) were insensitive to CpG-B ODN treatment. Further studies are needed to determine whether B-CLL cells insensitive to CpG-B ODN treatment are from specific cytogenetic CLL subgroups and potentially correlate with the presence of certain cytogenetic aberrations. Nevertheless, these findings demonstrate that CpG-B ODN triggering of TLR9 activates B-CLL cells to undergo apoptosis.

Purified B-CLL cells were incubated with graded doses of CpG ODNs in a 5-day culture. CpG-B ODNs (2006 or 685) but not CpG-A ODNs (2216 or PB4) potently induced apoptosis and decreased B-CLL cell number in a CpG ODN dose-dependent manner (Figure 3A). A 50% reduction of B-CLL cells was achieved with just 12- to 24-hour exposure to CpG 2006 or CpG-B 685 at 3 μg/mL but was unachievable with the same dose of CpG 2216 or CpG PB4 at all treatment time points up to 7 days (Figure 3B). A 50% reduction of B-CLL cells was also achieved with 48-hour exposure to CpG 2006 or CpG 685 at 1 μg/mL but was unachievable when the dose of CpG 2006 or CpG 685 dropped to 0.3 μg/mL or lower in the 7-day culture (Figure 3C). These findings demonstrate that CpG-B ODN-induced apoptosis of B-CLL cells is treatment time- and dose-dependent.

To determine whether in vitro pretreatment of B-CLL cells with CpG 2006 can influence the engraftment and survival of leukemia cells in vivo, B-CLL cells with or without preincubation with CpG 2006 were injected intraperitoneally into NOD-scid mice. No human CD45⁺/CD19⁺ cells were detected in normal NOD-scid control mice. In contrast, hCD45⁺/CD19⁺ B-CLL cells were detected in the peritoneal cavity and spleen of the NOD-scid mice of No-CpG group. Mice injected with CpG 2006 pretreated B-CLL cells had significantly decreased B-CLL cells, at 25% and 27.5% of the B-CLL cell numbers recovered from the mice of No-CpG group, respectively, confirming in vivo the in vitro finding that CpG 2006-treated B-CLL cells are programmed to undergo apoptosis (Figure 4A). In addition, in vivo administration of CpG 2006 significantly decreased the number of B-CLL cells engrafted in the peritoneal cavity and spleen of NOD-scid mice, at 0.64% and 1.4% of the B-CLL cell numbers recovered from the No-CpG group mice, respectively (Figure 4B). These findings demonstrate that treatment of B-CLL cells with CpG 2006 in vitro and in vivo significantly inhibits the engraftment and survival of B-CLL cells in NOD-scid mice.

The molecular mechanisms involved in CpG-B ODN-mediated apoptosis of B-CLL cells are unknown. Treatment of B-CLL cells with CpG 2006 led to cleavage of caspase-9, caspase-3, and PARP in B-CLL cells at days 3 to 7 cultures (Figure 5A). Pretreatment of B-CLL cells with pan-caspase, caspase-3, or caspase-9 inhibitor significantly reduced CpG 2006-induced apoptosis of B-CLL cells (Figure 5B). These findings demonstrate that TLR9 signaling by CpG ODN activates the apoptotic pathway elements in B-CLL cells committing them to death, and this effect can be inhibited by specific caspase inhibitors.

TLR9 uses a myeloid differentiation factor 88-dependent pathway to induce expression of NF-κB-dependent target genes that promote inflammatory responses and cytokine production. Thus, a potential mechanism for CpG-B ODN to induce apoptosis of B-CLL cells is via autocrine cytokine stimulation by CpG ODN-triggered B-CLL cells. Testing this hypothesis, we examined whether the NF-κB signaling pathway inhibitor CAPE or the JAK/STAT pathway inhibitor AG490 could block CpG 2006-induced apoptosis of B-CLL cells. Pretreatment of B-CLL cells with CAPE and/or AG490 at the indicated treatment time and dose did not directly affect the viability of B-CLL cells but significantly reduced CpG 2006-induced apoptosis of B-CLL cells (Figure 5C). Moreover, pretreatment of B-CLL cells with AG490 effectively blocked CpG 2006-induced cleavage of caspase-9, caspase-3, and PARP in B-CLL cells (Figure 5C). These findings suggest that apoptosis induced by TLR9-CpG ODN ligation is dependent on the NF-κB and JAK/STAT signaling pathways. Preventing apoptosis by AG490 suggests that CpG ODN-induced apoptosis probably occurs via an indirect mechanism involving autocrine cytokines that provoke apoptosis of B-CLL cells.

Cytokines implicated in the pathogenesis of B-CLL have used the JAKs and STATs to transmit growth and survival signals. Constitutive activation of STAT1 can play a key role in leukemogenesis of acute and chronic leukemias.¹⁰  STAT1 is constitutively phosphorylated on a carboxy terminal serine residue (Ser727) in B-CLL cells but not in normal B cells.⁷  In contrast, tyrosine phosphorylation was not detected on any STATs in B-CLL cells.⁷  Our results confirmed that freshly isolated B-CLL cells expressed STAT1 constitutively phosphorylated on serine (pSer727-STAT1) but not tyrosine (pTyr701-STAT1; Figure 5D[b]). TLR9-CpG ODN ligation induced a significant increase of pTyr701-STAT1 expression in CpG 2006-treated B-CLL cells, which was detected at 24 hours, and increased steadily over the 72-hour culture (Figure 5D). The expression of pSer727-STAT1 in B-CLL cells was detected before and after CpG 2006 stimulation, transiently increased at 24 to 48 hours, and was maintained at high levels over the 72-hour culture. Pretreatment of B-CLL cells with CAPE or AG490 alone did not activate or alter STAT1 phosphorylation; however, CAPE and AG490 each effectively abolished CpG 2006-induced pTyr701-STAT1 expression without affecting pSer727-STAT1 expression in B-CLL cells (Figure 5E). These data provide additional evidence that CpG-B ODN-induced apoptosis of B-CLL cells occurs via the JAK/STAT pathway. Together with the finding that CAPE and AG490 each block CpG 2006-induced apoptosis in B-CLL cells, CpG ODN-induced tyrosine phosphorylation of STAT1 correlated with the induction of apoptosis.

To definitively determine whether STAT1 tyrosine phosphorylation impacts CpG ODN-induced apoptosis of B-CLL cells, antisense ODN against STAT1 was introduced to B-CLL cells before CpG 2006 stimulation. Pretreatment of B-CLL cells with STAT1 antisense ODN completely abolished pTyr701-STAT1, pSer727-STAT1, and total STAT1 expression in B-CLL cells (Figure 5F). STAT1 antisense ODN did not affect the survival of untreated B-CLL cells but significantly prevented CpG 2006-mediated apoptosis of B-CLL cells. The control STAT1 sense ODN did not affect the survival of untreated B-CLL cells and the apoptosis of CpG 2006-treated B-CLL cells (Figure 5G). Interestingly, pretreatment of B-CLL cells with these non-CpG STAT1 antisense or sense ODN did not interfere with CpG 2006-induced activation of B-CLL cells (Figure 5H). These results demonstrate that TLR9 signaling by CpG 2006 leads to tyrosine phosphorylation of STAT1, thereby provoking apoptosis of B-CLL cells.

One important biologic feature of CLL is the capacity of B-CLL cells to produce autocrine cytokines that impact their survival and apoptosis. To determine whether cytokine(s) are responsible for STAT1 tyrosine phosphorylation and apoptosis in CpG-activated B-CLL cells, supernatants from B-CLL cell cultures with or without CpG 2006 were assessed at time points up to 72 hours. Triggering TLR9 with CpG 2006 led to the production of IL-6, IL-8, IL-10, TNF-α, and VEGF by B-CLL cells (Figure 6A). A profound increase of IL-10 production was detected at 6 hours, peaked at 24 hours, and remained at high levels over the 72-hour culture. TNF-α production was significantly increased, peaked within 6 hours, decreased gradually but maintained at elevated levels over the 72-hour culture. No detectable levels of IL-1β, IL-2, IL-4, G-CSF, GM-CSF, IFN-α, IFN-β, and IFN-γ were produced by B-CLL cells in cultures with or without CpG 2006 (Figure 6A). The amounts (pg/mL) of IL-6 (98.7 vs 1.7), IL-8 (560.1 vs 178.8), IL-10 (287.6 vs undetectable), TNF-α (331.8 vs 2.9), and VEGF (417.5 vs 52.5) produced by B-CLL cells with or without CpG 2006 were significantly different (Figure 6B). TGF-β1 was slightly increased, but the difference was insignificant (313.3 ± 77.9 vs 246.3 ± 58.0). Similar cytokine profiles were detected in supernatants from unpurified PBMCs containing high percentages (> 70%) of B-CLL cells in cultures with or without CpG 2006 (Figure 6B).

The profound increases of IL-10 and TNF-α production by CpG 2006-activated B-CLL cells suggest that these cytokines may have a role in the apoptotic induction of B-CLL cells. Intriguingly, treating B-CLL cells with CAPE before CpG 2006 stimulation significantly blocked IL-10 production, reduced the elevated IL-6, IL-8, and VEGF production, but did not affect TNF-α production by CpG 2006-activated B-CLL cells (Figure 7A; data not shown). Treating B-CLL cells with AG490 slightly lowered IL-10 production but did not influence TNF-α production by CpG 2006-activated B-CLL cells. These results suggest that IL-10 might be the cytokine responsible for STAT1 modulation and apoptotic induction in CpG 2006-activated B-CLL cells. To test this hypothesis, anti–IL-10 or anti–TNF-α Abs were added into CpG 2006-stimulated B-CLL cell cultures. The addition of anti–IL-10 Abs significantly protected the survival of B-CLL cells, whereas anti–TNF-α Abs did not influence the survival of B-CLL cells in CpG 2006-stimulated cultures (Figure 7B). Moreover, anti–IL-10 Abs, but not anti–TNF-α Abs, significantly inhibited tyrosine phosphorylation of STAT1 (Figure 7C). Furthermore, adding exogenous rh-IL-10 into B-CLL cell cultures induced apoptotic B-CLL cells in an IL-10 dose-dependent manner, which could be specifically blocked by anti–IL-10 Abs (Figure 7D). These results confirmed that IL-10 can induce apoptosis of B-CLL cells, and demonstrated that TLR9 signaling by CpG-B ODN actually uses autocrine IL-10 to mediate the induction of apoptosis of B-CLL cells. In addition, treatment of B-CLL cells with IL-10 leads to tyrosine-phosphorylated STAT1 expression in B-CLL cells, which can be effectively blocked by anti–IL-10 Abs (Figure 7E). Notably, the addition of IL-10 into B-CLL cell cultures did not directly induce the activation of B-CLL cells or affect CpG 2006-induced activation of B-CLL cells (Figure 7F). Together, these findings demonstrate the molecular mechanism of TLR9-CpG ODN ligation induces apoptosis of human B-CLL cells is via the altered production of autocrine IL-10, which activates tyrosine-phosphorylation of STAT1 and thereby provokes apoptosis of B-CLL cells (Figure 7G).

This study provides the first demonstration that TLR9 signaling by CpG-B ODN leads to an NF-κB-dependent production of autocrine IL-10, which activates JAK/STAT pathway-dependent tyrosine phosphorylation of STAT1 proteins and thereby provokes an apoptotic death pathway in B-CLL cells. These findings provide new understanding of CpG ODN-mediated direct antitumor effects and new insights into the biology and therapy of human B-cell leukemias. Although the molecular mechanism(s) responsible for the survival and accumulation of B-CLL cells remains unknown, studies have suggested that B-CLL cells probably acquire resistance to apoptosis via aberrant up-regulated activities of kinase proteins in signaling pathways.^6-10  Constitutive activation of STATs can play an important role in leukemogenesis of B-CLL.¹⁰  STAT1 has a proapoptotic role by transmitting apoptotic signals and inducing proapoptotic regulatory genes, such as caspases, fas ligand, and TNF-related apoptosis-inducing ligand.^34-36  Constitutive serine phosphorylation without tyrosine phosphorylation of STATs may enhance prosurvival signals in B-CLL cells.^7,36  Our results confirm that B-CLL cells expressed STAT1 constitutively phosphorylated on serine, but not tyrosine, and demonstrate that TLR9 signaling by CpG-B ODN induces tyrosine phosphorylation of STAT1 in B-CLL cells. We think that this TLR9-CpG ODN ligation triggered NF-κB, IL-10, JAK/STAT pathway represents the main pathways for CpG-B ODN-mediated apoptosis in B-CLL cells. In a CD154 gene therapy study, tyrosine phosphorylation of STAT1 was detected in B-CLL cells in the majority of patients 24 hours after treatment and correlated with the decreased circulating leukemia cell counts.³⁷  These clinical observations agree with our findings that tyrosine phosphorylation of STAT1 promotes apoptosis in B-CLL cells.

There are several advantages supporting the development of TLR9-targeted immunotherapy for CLL. First, triggering TLR9 with CpG-B ODN induces apoptosis of B-CLL cells. This finding indicates that CpG-B ODN have a direct antileukemia effect as immunotherapeutic drugs for CLL. Although CLL is a heterogeneous B-cell malignancy with frequent and various cytogenetic abnormalities and defects in individual patients, the results from our study and another group³⁸  show that CpG ODN can induce apoptosis in most B-CLL samples. Second, TLR9-targeting of B-CLL cells with CpG-B ODN activates leukemia B cells and induces dramatic changes in morphology and phenotype. The resulting leukemia blasts might be more sensitive to chemotherapy drugs and more susceptible to killing by NK and T cells. Third, B-CLL cells are known to express tumor-associated antigens (eg, survivin) that are recognized by T cells^39,40  but are poor antigen-presenting cells (APCs) because of low expression of costimulatory molecules essential for T-cell activation.^32,33  CLL cells with poor APC function can also result in a contact-dependent impairment of immunologic synapse formation with T cells in CLL patients.^41,42  These defects can probably be overcome by treating B-CLL cells with TLR9 agonists. Fourth, B-CLL cells produce immunosuppressive cytokines that contribute to the defects in T-cell responses.¹³  Identifying TLR agonists that could preferentially alter B-CLL cell cytokine production to polarize a Th1 response may reverse T-cell abnormalities and enhance antileukemia responses in CLL patients. Fifth, dendritic cells (DCs) in CLL patients are dysfunctional and unable to stimulate an effective T-cell response,⁴³  which has prompted the use of in vitro generated DC vaccines against CLL.^40,44  CpG ODNs and CpG-induced cytokines can potently enhance DC maturation and their APC function, which may reverse the dysfunction of endogenous DCs. In addition, CpG-induced apoptotic B-CLL cells will be readily taken up by DCs to present tumor antigens to elicit T-cell antitumor responses. The further development of CpG-based immunotherapeutic approaches for CLL requires better understanding in the interactions of CpG-treated CLL cells with various immune effecter cells. Our studies in murine leukemia and brain glioma models have shown that administration of CpG ODNs induced in vivo activation of DCs to activate NK and T cell-mediated antitumor responses with potent therapeutic efficacy.^45,46  Adding CpG ODN to PBMCs from lymphoma patients induced activation of blood DCs to enhance T-cell and NK-cell function.⁴⁷  Besides the immune-enhancing effects, CpG ODNs are stable, chemically well defined, inexpensive, and safe agents with clinical drugs available for studies. These points support the development of TLR9-targeted therapy with CpG-B ODN as a therapeutic agent for CLL. Whether CpG-induced IL-10 production by B-CLL cells may alter serum IL-10 levels in CLL patients and affect the microenvironment with potential immunosuppresive effect on host immune responses is subject to further investigation in clinical trials.

Our results show that human B-CLL cells express TLR1, TLR6, TLR7, TLR9, and TLR10, with significant higher TLR9 expression in B-CLL cells than in normal B cells. Given that human naive B cells express undetectable levels of TLR9, triggering the B-cell receptor up-regulates the expression of TLR9 in naive B cells, and that memory B cells express high levels of TLR9,⁴⁸  our finding of high TLR9 expression in fresh B-CLL cells from patients indicates that B-CLL cells are probably derived from an “experienced” mature B cell or a memory B cell. Among the leukocyte populations examined to date,^28,49,50  TLR9 is expressed on human B cells and plasmacytoid DCs only. Therefore, TLR9-targeted immunotherapy of CLL is relatively specific for B-CLL cells. At least 3 distinct classes of CpG ODN with structural and functional differences have been identified.³¹  CpG-B ODNs are more potent TLR9 agonists than CpG-A ODNs for stimulating normal human B cells. Our results further demonstrated that CpG-B ODNs, but not CpG-A ODNs, induced significant activation and apoptosis of leukemic B cells. Given that B-CLL progresses slowly and can be treated to achieve a period of clinical remission, it is an ideal disease model for allowing the application of TLR agonists to manipulate host immune responses against leukemia cells in a minimum residual disease setting.

In conclusion, our results demonstrate that TLR9 targeting of B-CLL cells with CpG-B ODN induced NF-κB-dependent production of autocrine IL-10, which induces tyrosine phosphorylation of STAT1, thereby leading to apoptosis of B-CLL cells. This new understanding supports the development of TLR9-targeted therapy with CpG ODN as a therapeutic agent for treating human CLL. Both CpG 2006 (also known as CpG 7909 and PF-3512676) and CpG 685 (also known as GNKG168) are being tested in phase 1 clinical trials as monotherapy in patients with relapsed or refractory B-CLL.

This work was supported by the Leukemia & Lymphoma Society (research grants 6188-05 and R6241-08), Randy Shaver Cancer Research and Community Fund, and the Leukemia Research Fund (W.C.); National Institutes of Health (AI050737) and Leukemia & Lymphoma Society (Scholar award; M.A.F.); and National Institutes of Health (R01 CA72669; B.R.B.).

National Institutes of Health

Contribution: X.L. and E.A.M. performed research, analyzed data, and wrote the paper; V.B. and D.J.W. provided blood samples and clinical information of the CLL patients; M.A.F. and B.R.B. designed research and revised the paper; and W.C. designed and directed the overall study, analyzed data, and revised the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Wei Chen, University of Minnesota Cancer Center, MMC 806, 420 Delaware St SE, Minneapolis, MN 55455; e-mail: chenw@umn.edu.