Chronic lymphocytic leukemia B cells are endowed with the capacity to attract CD4⁺, CD40L⁺ T cells by producing CCL22

2.1 Chemokine expression by CLL cells from PB, BM and LN

As several evidences indicate that CLL cells retain the capacity to respond to microenvironmental signals delivered by activated T lymphocytes and as T lymphocytes are present in CLL LN and BM, we hypothesized that CLL cells might produce T cell chemoattractants. Therefore, our first step was to analyze neoplastic B lymphocytes purified from PB of 24 cases (Pt. 1–24, Table 1), BM of 4 other patients (Pt. 25–28, Table 1) and LN of 4 further cases (Pt. 29–32, Table 1).

All cases from LN, BM and 7 cases from PB were purified by cell sorting. Malignant B lymphocytes from the remaining PB were not purified, due to the high percentage of leukemic cells present in the preparations (>95%).

As shown in Table 2, only few chemokines could be detected by RT-PCR in LN-, BM- and PB-derived primary leukemia cells and at variable intensity.

CCL5 was detected in all samples tested regardless of their origin (Table 2 and Fig. 1A). CCL3 was absent in 4 PB cases (Table 2), and expressed, at variable levels, in all other samples tested (Fig. 1A). CXCL8 was absent in 12 samples (2 LN, 1 BM and 9 PB) (Table 2) anddetectable in the remaining tested cases (Fig. 1A). Interestingly, CLL cells obtained from LN and BM expressed a readily detectable band for CCL22 (in 4/4 LN cases and 2/4 BM cases) and a band of variable intensity for CCL17 (in 2/4 LN cases and 0/4 BM cases) (Fig. 1B). In contrast, both chemokines were absent in PB-derived cells (0/24 cases) (Fig. 1B).

2.2 CD40 cross-linking of leukemia cells induces RNA expression of the chemokines CCL17 and CCL22

These results suggested that signals available in the LN microenvironment might account for the different behavior of tissue-derived as compared to PB-derived malignant B cells. We focussed our attention on the CD40/CD40L interaction, which is the hallmark of T/B cell interactions in secondary lymphoid tissues. Fluorescent-activated cell sorting (FACS)-purified CLL cells from 7 patients' PB were cultured either in medium alone or in the presence of recombinant sCD40L for up to 3 days. CD40 ligation induced mRNA expression of both CCL22 and CCL17 (Fig. 1B), without altering the level of expression of either CCL3, or CCL5 or CXCL8 (Fig. 1A). Both CCL22 and CCL17 started being expressed already after 24 h of culture (data not shown) and maintained the same level of expression throughout the 3 days of culture (Fig. 1B). These results were then confirmed in 14 further CLL cases, whose malignant cells were not purified, due to the high percentage of leukemic cells present in the preparations (>95%).

2.3 CD40 cross-linking induces the secretion of CCL22 chemokine

Starting from these evidences, we next sought to determine whether CCL22 and CCL17 chemokines were actually secreted by leukemic cells and whether their production might be modified by cell activation. We prepared leukemic cell supernatants by culturing CLL cells at standard cell concentration (2×10⁶ cells/ml), in the presence or the absence of sCD40L. We collected the culture supernantants after three days of culture and measured by ELISA the presence of chemokines. CCL3, CCL5, CCL17 and CXCL8 were measured in 12 cases and CCL22 in 21 cases. Leukemic cells produced low levels (up to 0.36 ng/ml) of CCL3 in 8/12 cases and low levels (up to 3.16 ng/ml) of CXCL8 in 6/12 cases. CCL5 protein was detectable in all 12 supernatants tested, though at very variable levels, ranging between 0.07 to 36 ng/ml. The production of all these chemokines was not dependent on the culture condition i.e. no significant modifications occurred after CD40 activation.

Supernatants from unstimulated CLL cells were negative for both CCL17 and CCL22 and upon CD40-stimulation no CCL17 could be detected in the cell supernatants (data not shown). In contrast, CD40-stimulated malignant cells produced levels of CCL22 as high as 28.8 ng/ml (Fig. 2).

We investigated whether cell density might account for the difference between CCL17 and CCL22. CLL cells were cultured at high-density concentration (10×10⁶ cells/ml) but still no CCL17 was observed. On the contrary, the supernatant levels of CCL22 paralleled cell density (data not shown). To confirm that CCL22 was actually secreted by tumor cells we tested supernatants obtained from purified (>99% purity) malignant CLL cells of seven patients. Sorted cells produced levels of CCL22 comparable to those of the unsorted samples.

Not all patients produced CCL22 after CD40-stimulation. Two categories of patients were observed: CD40L-responders and CD40L-nonresponders. In 14/21 (67%) patients studied (patients 1–14, Table 1) CD40L stimulation induced CCL22 release in the supernatants. These patients were considered CD40L-responders. The remaining 7/21 cases (33%; Table 1: patients 15–21) were considered CD40L-nonresponders as CCL22 secretion remained unchanged after stimulation. Twenty patients of this series have been previously analyzed for the capacity to express Survivin mRNA after CD40 stimulation 22. In 17 out of these 20 patients, there was correspondence in the outcome of the response.

The ability of responding to CD40 stimulation did not correlate with Rai stage of the disease nor with CD38 expression, though the limited number of patients studied prevented any statistically significant conclusion.

2.4 Secreted CCL22 is functional and induces the migration of activated CCR4⁺, CD4⁺ and CD40L⁺ T cells

To determine whether CCL22 produced by CD40-stimulated CLL cells was functional, supernatants were tested for their capacity to induce the migration of activated CD4⁺ T cells 5, 11. To this end T cells were obtained from PBMC sorted for the expression of CCR4 and activated in the presence of irradiated feeder and IL-2, as described in the Sect. 4. As a result, these cells were CD4⁺ and uniformly expressed CD40L within 24 h from activation (Fig. 3). We studied these cells for the expression of several known chemokine receptors and in particular for the expression of CCR1, CCR3, CCR5, receptors for CCL3 and CCL5; CXCR1 and 2, receptors for CXCL8. T cell lines were negative for all these receptors. In contrast they uniformly expressed high levels of CCR4, the receptor for CCL22 (and CCL17) (Fig. 3).

Accordingly, recombinant CCL22, used as control, efficiently promoted the migration of recently activated T cells (Fig. 4A).

As shown in Fig. 4B, supernatants from CD40-stimulated CLL cells, obtained from six different patients, induced migration of activated T cells in a proportion significantly higher than that induced by control culture supernatants. As expected, the supernatant-induced migration was lower than that induced by recombinant CCL22 since the total amount of CCL22 present in the supernatants was lower than the optimal dose of chemokine necessary to induce optimal migration.

To demonstrate that the T cell chemotactic activity of the supernatants was specifically mediated by the CCL22 present in our supernatants, migration experiments were repeated using neutralizing antibodies. The use of anti-CCL22 antibody resulted in complete neutralization of the T cell migration to background levels (Fig. 4B). These results demonstrate that CCL22 is the only mediator of the T cell chemotactic activity present in the supernatants from CD40-stimulated CLL cells.

2.5 CD4⁺ CD40L⁺ T cells present in CLL LN and BM are interspersed within pseudo-follicles

Finally, as in vitro data were ascribing a role to CD40 ligation in inducing chemokine regulation in CLL cells, we evaluated whether this signal (i.e. CD40L⁺ T cells) was present in tissue biopsies. In BM and LN samples studied by cytofluorography analysis CD40L⁺ cells were clearly detectable among the CD3⁺ CD4⁺ T cells, and their proportion ranged between 5 and 30% of them. No CD19⁺ CD5⁺ B cells expressed CD40L in all the samples analyzed.

We next investigated LN and BM biopsies heavily infiltrated with malignant CLL cells by immunohistochemical analysis. In double-stained sections, CD4⁺ T lymphocytes were admixed with CD23⁺ malignant B cells (Fig. 5A) 24. Within pseudo-follicles, CD4⁺ T cells were interspersed with Ki67⁺ proliferating CLL cells (Fig. 5B), and close contacts were evident. As previously shown 22, CD3⁺ T cells are present in high numbers and are mainly localized within pseudo-follicles. The analyses of serial sections revealed CD40L expression in a proportion of lymphocytes corresponding to CD3⁺ T cells, preferentially clustered within pseudo-follicles (Fig. 5C).

4.1 Cells

Malignant B cells from 32 CLL patients were studied. Twenty-four samples were from PB, 4 were from BM and 4 from biopsied LN. Tissue samples were obtained following institutional guidelines. Diagnosis was based on standard clinical and laboratory criteria 34, 35. All patients were staged according to Rai's 36 and studied either before or at least 3 months after chemotherapy. Table 1 summarizes the main clinical data of the patients.

LN specimens were minced with a scalpel blade, reduced to single-cell suspension and passed through a fine wire mesh. All samples were enriched by density centrifugation over Lymphoprep (Nycomed Pharma AS, Oslo, Norway). When indicated, malignant B cells were purified by FACS.

CD4⁺ T cells were obtained from PB mononuclear cells (PBMC) of two healthy individuals.

4.2 Phenotypic analysis and cell sorting

Expression of cell surface molecules was determined by flow cytometry using standard methodology. The mAb used were: tricolor (TC)-conjugated anti-CD3, fluorescein (FITC)-conjugated anti-CD4 and anti-CD8 (Caltag, Burlingame, CA), FITC-anti-CD5 (Becton Dickinson, San Jose, CA) and phycoerythrin (PE)-anti-CD19 (Caltag), PE-anti-CD40L (DAKO A/S, Glostrup, DK); unconjugated CCR4 (IgG1), produced by us 37; PE-anti-CCR1, CCR2, CCR6, CXCR1, and CXCR2 and biotinylated-anti-CXCR5 (R&D Systems, Minneapolis, MN), PE-anti-CCR5, CXCR3 and CXCR4 (Pharmingen-Becton Dickinson,San Jose, CA). PE-conjugated goat-anti-mouse IgG1 and PE-streptavidin (Southern Biotechnologies, Birmingham, AL) were used as secondary reagents. Irrelevant isotype-matched antibodies were used as negative controls (Becton Dickinson). Samples were analyzed in a FACSCalibur flow cytometer (Becton Dickinson). At least 10,000 events were measured for each sample. For cell sorting of leukemic cells, CD5/CD19-labeled cells were concentrated, filtered, and purified using a FACStar cell sorter (Becton Dickinson). PBMC were stained with anti-CD4 and anti-CCR4 and then purified using a FACSVantage cell sorter (Becton Dickinson).

4.3 In vitro culture and CD40 stimulation of leukemic cells

Leukemic cells were cultured in RPMI medium with glutamine (Life Technologies Ltd, Paisley, Scotland) plus 10% fetal bovine serum (FBS) (Life Technologies Ltd), and gentamycin (10 μg/ml, Life Technologies Ltd) (complete RPMI-cRPMI) in the presence or the absence of 100 ng/ml soluble recombinant human CD40L plus 1 μg of enhancer, according to manufacturer's instructions (Alexis Corporation, San Diego, CA), at 37°C, 5% CO₂. Malignant cells were cultured at 2×10⁶ cells/ml and/or at 10×10⁶ cells/ml, as indicated. Supernatants and cells were collected at day 3. Cells were then lysed in RNazol (Biotecx Laboratories, Inc., Houston, TX). Supernatants were filtered and kept frozen at –20°C.

4.4 Generation of activated CD4⁺ T cells

FACS-purified CCR4⁺ CD4⁺ T cells were cultured in cRPMI and activated by repetitive stimulation using irradiated allogeneic PBMC plus 1 μg/ml PHA (Murex Diagnostics) and recombinant IL-2 (20 U/ml) for 5 days.

4.5 RT-PCR

RNA was extracted from 1×10⁶–2×10⁶ cells using RNazol (Biotecx Laboratories, Inc., Houston, TX) following the manufacturer's instructions. cDNA was prepared at42°C using a reverse transcription mix containing Superscript II (Life Technologies Ltd). The enzyme was inactivated for 4 min at 94°C. PCR amplification of cDNA samples was carried out using the primers whose sequences are shown in Table 3.

All reactions were carried out in a Perkin-Elmer thermocycler (Perkin-Elmer, Foster City, CA), at the following conditions: denaturation at 94°C for 45 s, annealing at 60°C for 30 s and extension at 72°C for 45 s (28 cycles).

4.6 ELISA

The production of the chemokines CCL3, CCL5, CCL17, CCL22 and CXCL8 was quantified by ELISA using commercially available kits (R&D Systems, Minneapolis, MN for CCL22 and CCL17; BioSource International Inc., Camarillo, CA for CCL3 and CCL5 and Bender MedSystem, Vienna, AU for CXCL8). The lower limit of detection of the ELISA was 0.002 ng/ml for CCL3, 0.003 ng/ml for CCL5, 0.007 ng/ml for CCL17, 0.0625 ng/ml for CCL22, and 0.011 ng/ml for CXCL8, respectively.

4.7 Chemotaxis assay

Chemotaxis assays were performed using the Transwell system (5-μm pores, Costar, Cambridge, MA), as previously described 38, 39. Cell supernatants obtained from leukemia cells cultured with and without CD40L (500 μl) were thawed and used at 100% vol/vol. Recombinant CCL22 (250 and 500 ng/ml) (R&D Systems, Minneapolis, MN) was added to cRPMI anddistributed on cluster plates and warmed for 15 min at 37°C. Following incubation with culture medium, the Transwell inserts were transferred into the pre-warmed cluster plates. CCR4⁺ CD4⁺ recently activated T cells (5×10⁵ cells for 6.5 mm inserts) were placed into the Transwell inserts. Plates were incubated for 1 h at 37°C, 5% CO₂. In parallel experiments, neutralizing anti-CCL22 antibodies (10 μg/ml, R&D) were added to the supernatants, prior to the chemotaxis assay.

All migrated cells were collected from the lower compartment and counted by flow cytometry. The flow cytometer settings were determined prior to the acquisition of the migrated cells using samples of the input population. The migration percentage was calculated by dividing the number of migrated cells by the number of input cells.

4.8 Immunohistochemistry

Frozen sections were obtained from snap-frozen fragments of LN (four samples) and BM trephine biopsies (three samples) from patients with CLL. To evaluate the relationship between proliferation centers and T cells we investigated LN samples with double-marker immunostaining technique, 40. Ki67-reactive cells, clustering in pseudo-follicles, were revealed together with CD4-expressing lymphocytes. Cryostat serial sections of LN were transferred onto glass slides covered with polylysine adhesive, fixed in chloroform-acetone 1:1 mixture for 5 min and air dried. Ki67 antibody (Dako, Glostrup, Denmark) was incubated first at 1:100 dilution, and revealed using a sensitive avidin-streptavidin-peroxidase technique (Biogenex San Ramon, CA). After development of peroxidase with H₂O₂ and 3,3′-diaminobenzidine containing 1% nickel-chloride, the preparations were thoroughly washed in PBS, incubated with anti-CD4 mAb 1:20 dilution (Dako) and revealed using an alkaline-phosphatase anti-alkaline phosphatase immunostaining technique (APAAP, Dako) with new-fuchsin development providing red immunostaining of CD4⁺ cells. The same technique was used to demonstrate in the same section Ki67 and CD40L, but the signal of the latter marker was unreliable.

To reveal CD40L⁺ cells, sections were fixed in cold acetone/chloroform 1:1 mixture for 1 min, and washed with PBS before immunostaining. CD40L antibody (clone TRAP1, Pharmingen-BectonDickinson) was used at 1:50 dilution in PBS for 30 min at room temperature. An avidin-streptavidin peroxidase immunoperoxidase system (Stravigen multilink kit, Biogenex) was used as detection system, following manufacturer's instructions.