Chronic Lymphocytic Leukemia B Cells Can Express CD40 Ligand and Demonstrate T-Cell Type Costimulatory Capacity

Approval for the research protocol was obtained from the internal review boards of The New York Hospital and The Hospital for Special Surgery, and verbal, informed consent was obtained before phlebotomy. In each case, the diagnosis of CLL was established by the coexpression of CD5 and CD19 in an expanded, clonal population of peripheral blood lymphocytes. Mononuclear cells were isolated from 10 to 30 mL of fresh, heparinized peripheral blood by Ficoll-Hypaque centrifugation. For most of the experiments, T and B lymphocytes were separated by rosetting with sheep red blood cells according to standard techniques, and the degree of separation was checked by immunofluorescence flow cytometry.

Unless otherwise indicated, cells were analyzed immediately subsequent to their purification. For some experiments, as indicated, cells were cultured in C50 media (RPMI supplemented with penicillin-streptomycin with L-glutamine and 10% fetal calf serum [FCS]) only or with IL-2 at 40 U/mL (Hemagen Diagnostics, Waltham, MA) and a 1:60,000 dilution of formalinized Staphylococcus Aureus Cowan 1 (SAC; Pansorbin; Calbiochem, San Diego, CA), recombinant IL-4 at 10 ng/mL (Genzyme Diagnostics, Cambridge, MA), monoclonal antibody (MoAb) to CD40 at 1 μg/mL (IgG_κ1; Genzyme), an F(ab′)₂preparation of MoAb to human IgM at 0.5 μg/mL (Jackson Immunoresearch Laboratories Inc, West Grove, PA), phorbol myristate acetate (PMA) at 10 ng/mL (Sigma) and ionomycin at 1.25 μg/mL (Calbiochem), or pokeweed mitogen (PWM; Life Technologies, Gaithersburg, MD). For cDNA preparation, Jurkat mutants that constitutively express CD40L (clone D1.1) were used as a source of CD40L.41 These mutant Jurkat cell lines were the kind gift of Dr Seth Lederman (Columbia University, New York, NY). The Ramos Burkitt's lymphoma cell line was obtained from American Type Culture Collection (ATCC; Rockville, MD). For immunoprecipitation experiments, in addition to the CLL samples, we used purified CD4⁺ polyclonal T-cell lines derived from patients with B-CLL that respond to the toxic shock syndrome toxin-1 (TSST-1; Toxin Technology, Sarasota, FL) superantigen. These cells were maintained in culture with IL-2 at 40 U/mL and were periodically triggered with TSST-1 (used at 10 ng/mL) in the presence of allogeneic antigen-presenting cells, as described previously.27 42 

Cells were washed in cold Hank's Buffered Saline Solution (HBSS) and incubated with antibodies according to standard techniques. Cell fluorescence was measured using a Becton Dickinson FACScan (Becton Dickinson, San Jose, CA). Cells were assessed for forward and side scatter, and events in the viable cell gate were analyzed for fluorescence intensity using the CellQuest program (Becton Dickinson). For acid washing of cells before analysis, the cells were exposed to acidic PBS (pH 4.1) for 3 minutes at room temperature and then washed twice with normal pH before exposure to MoAbs, as indicated, similar to the technique reported by Grammer et al.39 

Sterile antibodies used in cell culture experiments included anti-CD40 (murine IgG₁; Genzyme), anti-CD40L (murine IgG_1,, clone M-90; Genzyme), and anti-CD23 (EBVCS2; ATCC). In the blocking coculture experiments, the dose of each antibody used was 5 μg/mL. For single-color fluorescence flow cytometry, the antibodies used were anti-CD3 (OKT3; ATCC), anti-Vβ13 (murine IgGκ1), anti-CD19 (Immunotech, Westbrook, ME), anti-CD40 (Genzyme), and anti-CD40L (clone M-90, murine IgGκ1; Genzyme), in each case followed by goat antimouse Ig-fluorescein isothiocyanate (FITC). Antibodies used for two-color analyses included anti-CD3-FITC (Immunotech), anti-CD40-FITC (Biosource, Camarillo, CA), anti-CD19-FITC (Immunotech), anti-CD3-PE (Biosource), anti-CD5PE (Biosource), anti-CD40L-PE (TRAP clone; PharMingen, San Diego, CA), and anti-CD40L-PE (clone 89-76; Becton Dickinson).

RNAs were isolated from approximately 1 × 10⁷ cells using the TRIzol reagent (Life Technologies/GIBCO BRL, Grand Island, NY), according to the manufacturer's instructions, and the yield in each case was measured by spectroscopy. For each sample, cDNAs were prepared using approximately 1 μg of RNA, using a reverse transcriptase kit (GIBCO). For CD40L, the primers used were 5′-GGCCATTATGCACAGGTTGAAT-3′ (1100-1122) and 5′-GGGGAGGGAAGAGACTGACAAA-3′ (1469-1489),43which yielded a single amplified CD40L fragment of 396 bp. The quality of each cDNA preparation was checked by amplification of GAPDH sequences using primers 5′-GAAGATGGTATCACCTGGAC-3′ and 5′-GAAGAAGCTGCTGGTGTAGT-3′ (Stratagene, La Jolla, CA), which yield a single, amplified fragment size of 600 bp.

Freshly isolated purified CLL B cells were placed in culture under a variety of circumstances. After 36 or 60 hours as indicated, the cells were washed and cell surface proteins biotinylated using Sulfo-NHS-LC-Biotin (Pierce, Rockford, IL).44 Lysates were then subjected to immunoprecipitation overnight at 4°C using 3 μg of antibody to CD40L (TRAP clone; PharMingen) or with 3 μg of a control, isotype-matched antibody to IL-4 (Genzyme), in each case conjugated to protein A Sepharose 4 Fast Flow beads (Pharmacia Biotech, Piscataway, NJ). Precipitated materials were boiled for 4 minutes before sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to a polyvinylidene difluoride (PVDF) membrane, exposed to horseradish peroxidase (HRP)-conjugated streptavidin (Pierce) for 90 minutes, and then imaged using a chemiluminescence substrate (Amersham, Buckinghamshire, UK).

IgG and IgA secreted into the supernatants was measured by a standard enzyme-linked immunoabsorbance assay (ELISA) protocol. In each experiment, standard curves for the particular isotype were determined, and the samples were diluted such that the absorbance readings fell in the linear range of detection. For each circumstance and in each experiment, antibody measurements were determined in triplicate. Results were discounted if the standard deviation among the three triplicate readings represented greater than 15% of the mean or if the result for the diluted sample did not fall within the linear range of the standard curve. IgG production, as shown in Fig 6, reflects the results of single-case analyses, with error bars indicating the standard error for measurements of each experimental circumstance.

CLL B cells from patients with untreated CLL were purified from heparinized peripheral blood samples by Ficoll hypaque centrifugation and separated from the T lymphocytes by rosetting with sheep red blood cells. Of the non-T cells obtained, greater than 96% of the cells in each case expressed CD19. In the majority of cases, greater than 99% of the purified cells were B cells. These cells were analyzed directly, without stimulation, by immunofluorescence flow cytometry for B- and T-cell antigens, including CD40L. In the first series (Fig 1A), B cells from consecutive, unselected cases were purified and examined using an MoAb to CD40L (clone M-90) or, as a negative control, an isotype-matched control antibody to a T-cell receptor (TCR) β chain (anti-Vβ13). As shown, the level of CD40L expression varied considerably among the CLL samples but significant antibody binding was evident in 6 (cases no. 1, 4, 10, 11, 13, and 18) of these first 13 cases analyzed.

To demonstrate definitively that the CD40L signal was not due to a small proportion of T cells in the samples, two-color immunofluorescence flow cytometry was performed using a directly conjugated fluorescent antibody (TRAP clone) on additional unselected samples, also immediately after purification of the cells. As shown in Fig 2, in some cases, the CD19⁺ (B) cells demonstrated CD40L expression above background, as determined with a PE-conjugated antibody to CD3. Using this technique, the molecule was detected in approximately one fourth of the cases analyzed. In 1 case (no. 10) for which we had demonstrated fairly strong expression in initial analyses (Fig 1), we were able again to obtain fresh cells 6 months later. This two-color immunofluorescence pattern in the upper panel of Fig 2 represents the best example of the two-color stains we have performed. More typical were intermediate results, as shown in the middle panel of Fig 2 (6 cases), or low-level expression, as in the bottom panel (15 cases). Of note, in two-color FACS analyses of greater than 20 fresh specimens from patients with CLL, we have never observed significant CD40L expression in the unstimulated T cells.

Because there is some evidence that detection of CD40L by immunofluorescence flow cytometry may be dependent on the particular antibody used and that recognition of CD40L at the cell surface may be hindered by the presence of soluble CD40,45 we elected to evaluate CD40L in CLL B cells using an additional MoAb (clone 89-76). In addition, we examined the effect of acid-washing CLL B cells with PBS (pH 4.1) before exposure to MoAb to CD40L for FACS analysis, according to the technique of Grammar et al,39 to remove soluble CD40 that may be adherent to the B cell, particularly as may occur in CD40-expressing B-cell tumors such as CLL.45 Using the 89-76 clone, 3 of 6 cases analyzed were positive, including 2 that were also positive with the TRAP clone. The effect of acid-washing was similar to that reported by Grammer et al39 using lymphoblastoid B cells and is shown in Fig 1B.

In cases that were positive, CD40L expression diminished rapidly over time in culture with media only, such that the signal was less intense 48 hours after isolating the cells, became further diminished after 72 hours, and was undetectable thereafter (data not shown). However, some stimuli did appear to augment CD40L in cases that had initially undetectable levels. Figure 3 shows CD40L expression in CD19⁺ cells from a single case that were treated for 48 hours after purification under a variety of experimental circumstances. As shown in Fig 3B, the combination of IL-2 and SAC in culture was associated with increased CD40L expression in the CLL B cells. This effect was observed in each of 5 cases for which we analyzed this effect. In this particular sample, IL-4 did not result in augmented CD40L expression (Fig 3C), but the effect of this cytokine has varied among the 6 cases analyzed so far. Other B-cell stimulants also appear to augment CD40L expression in the CLL cells. For example, signaling induced by MoAb to CD40 (Fig 3D) and through ligation of the B-cell surface antigen receptor (Fig 3E) both resulted in increased CD40L expression at the cell surface.

B cells were purified from 2 clinical CLL samples and placed in culture for 16 hours before RNA extraction for analysis by RT-PCR. As shown in Fig 4, CD40L message was readily detected in both of the CLL samples exposed in vitro to media alone (lanes 1 and 5). In addition, there was an increase in message in cells from the second case after exposure to the combination of IL-2 and SAC (lane 6), to PMA and ionomycin (lane 7), and to IL-4 (lanes 8), as compared with the amplified signal for the housekeeping gene GAPDH. These results correlated well with those obtained by immunofluorescence flow cytometry.

MoAb (TRAP clone) was used to precipitate CD40L from several samples. For these experiments, lysates were prepared after biotinylation of cell surface proteins. Material from 6 × 10⁶ cells was precipitated with either antibody to CD40L or with an isotype-matched control antibody to IL-4 (Fig 5A). As shown, we observed major bands at 39 and 32 kD corresponding to the transmembrane forms of CD40L,46,47 which in this case were most evident after exposure of the cells to IL-4 (lane 3). In addition, a doublet was evident at 18 kD and another, single band at 15kD, as have been reported48 and attributed to soluble CD40L.49 The CD40L bands were not seen after immunoprecipitation with a control antibody to human IL-4 (lanes 2 and 4). In a separate experiment, to control for the amount of protein in each sample rather than cell number, CLL cells were stimulated for 60 hours with IL-2 and SAC and then examined for CD40L using 210 μg of biotinylated protein in each sample (Fig 5B). As shown, there is a prominent 39-kD band apparent in the stimulated cell lysate precipitated with antibody to CD40L, but not with the control antibody.

Finally, to determine that the CD40L signal was not due to a small fraction of T cells in each sample, we used activated polyclonal CD4⁺ T cells, derived from a CLL patient and triggered periodically with TSST-1 and IL-2, for immunoprecipitation after biotinylation according to the same technique. These CLL patient-derived T cells bear an activated phenotype and express CD40L.49a As shown in Fig 5C, CD40L was not detected by immunoprecipitation in our system when 5 or 2.5 × 10⁵ biotinylated T cells were used (lanes 2 and 3), but was readily observed after 1 × 10⁶ T cells were used (lane 1). Thus, the absolute minimum number of T cells required to provide the signal we observed in the CLL samples would be greater than 8% of the 6 × 10⁶ cells used in each immunoprecipitation, which was not the case. In the cells used for the experiment in Fig 5A, the proportion of B cells (CD19⁺, CD3⁻) was 97% after culture with media alone and 98% after exposure to IL-4. In Fig 5B, the proportion of B cells was greater than 99.5% at the time of lysate preparation.

Other investigators have determined that ligation of CD40 results in CLL B-cell activation31,49 and proliferation.50In our laboratory, we performed thymidine incorporation assays to examine CLL B-cell proliferation in culture under a variety of circumstances. Given that the CD40/CD40L molecular pair is an important modulator of CLL B cell growth, one might anticipate that the CD40, CD40L-expressing tumor cells would proliferate spontaneously in culture. However, as has been observed by other investigators, we did not observe significant thymidine incorporation in any of the 6 cases we studied without the addition of irradiated, CD40L⁺ T cells (data not shown). These negative results suggest that other factors besides CD40L, such as additional signals provided by accessory molecules or T-cell–derived cytokines, are necessary for B-cell proliferation in the context of CD40 ligation.

Because CD40L has a central role in T-cell–induced B-cell differentiation and antibody formation, we investigated the capacity of CLL B cells to trigger antibody production by normal, bystander B cells in coculture. For these experiments, we used as targets B cells purified from the peripheral blood of healthy volunteers. Effector cells, used to stimulate antibody production by the target B cells, included CLL B cells, autologous T cells obtained from the normal donor in each experiment, and also T cells purified from the CLL patients. These effector cells were irradiated and then placed in culture with the target B cells as indicated in Fig 6. Supernatants were removed and analyzed for IgG and IgA secretion by ELISA.

The CLL B cells were variable in their capacity to stimulate antibody production, but in general the effects were most clear in the presence of B-cell mitogens such as PWM. As shown in Fig 6A, the unirradiated target B cells alone did not generate significant levels of IgG, even with the addition of PWM. This indicates that the IgG produced was not simply due to a few contaminating T cells in the target B-cell preparation. In addition, the irradiated, effector CLL B cells alone did not produce IgG in the presence of PWM. However, when irradiated effector cells were added to the target B cells, the degree of help provided by the CLL B cells was of the same order of magnitude as that provided by purified, autologous T cells. Although it is possible that some contaminating tumor-derived T cells were present among the irradiated, effector CLL B cells, those would be unlikely to account for a signal so strong as that obtained with 100% T cells derived from the patient's blood. Therefore, it is highly unlikely that this result is due to a small percentage of CLL patient-derived T cells that may have been present in the effector B-cell population.

Figure 6B shows the effects of a blocking antibody to CD40L (clone M-90) in this coculture system. Here, the irradiated CLL B cells were exposed either to antibody to CD40L, to CD40, or to CD23 before coculture with the target B cells. As shown, each of the antibodies was associated with reduced IgG production, but the result was most marked with antibody to CD40L. Similar assays for IgA demonstrated only low level secretion of IgA (on the order of 100 to 150 ng/mL), even in the presence of PWM (data not shown). This result is consistent with a complex, redundant system in which multiple factors, including CD40L, instigate antibody production and isotype switching in B cells. Our data demonstrate that under some circumstances, costimulatory function and isotype switching, normally induced by T cells, can be mediated by malignant CLL B lymphocytes.

CD40L is primarily expressed in activated T cells and has not been considered a regular feature of any B-cell tumor. We have observed that freshly purified B cells from a significant proportion of CLL cases express CD40L and can induce antibody production by target, nonmalignant B cells in a process that is partially blocked by antibody to CD40L. Although previous investigators have established that the CD40/CD40L molecular pair is an important modulator of CLL B-cell growth,28,49,50 we and other investigators have observed that CLL B cells do not proliferate spontaneously in vitro.49-52 Therefore, the extent to which CLL B-cell–derived CD40L impacts tumor growth may depend on costimulation via accessory molecules and the presence of T-lymphocyte–derived cytokines, such as IL-4.50 Taken together, the data and the literature suggest a model for CLL tumor growth and associated immune aberrations due to tumor cells that, at least in some cases, express both CD40 and CD40L (Fig 7).

In work quite similar to our findings, Nusslein et al53demonstrated that irradiated or even formaldehyde-fixed CLL B cells have the capacity to stimulate IgE production by nonmalignant B cells in the presence of IL-4. However, our data contrast with those of Brugnoni et al,54 who stained non-T cells from the peripheral blood of 11 CLL patients and did not detect CD40L expression in those cells. It is possible that the negative result found by Brugnoni et al54 was due to the heterogeneity of CD40L expression in CLL and its subtle expression in most cases. In addition, those investigators used only the TRAP clone for CD40L, which in our laboratory offered a positive signal in only one fourth of the cases. Closer to our work, Cerutti et al55 analyzed human B-1 (CD5⁺) B cells and noted above-background CD40L expression, at levels similar to what we have observed in the human CLL cells. Most recently, while this report was under review, Trentin et al56 described CD40L expression in some CLL cases by FACS analysis and by RT-PCR. In this report, we include data obtained using three distinct MoAbs and, in addition, demonstrate that the signal for CD40L in the B cells can be modulated by acid-washing the cells. Thus, the results may depend on the particular reagent used and also on the presence of soluble CD40 in the sample.

Ranheim et al57 reported that CLL B cells express both CD27 and its ligand, CD70, and that expression of these molecules in the CLL cells is affected by ligation of CD40. Thus, although CLL cells are considered as resting B cells, unable to proliferate in vitro without T-cell–derived signals, there may be a dynamic aspect of the tumor in vivo due to infiltrating CD4⁺ T cells that provide cytokines and antigen-dependent costimulatory signals. Based on the observations that CD40 ligation results in NF-κB activation58 and that constitutive, high NF-κB activity is observed in 293 cells expressing both CD40 and CD40L,59even low-level, transient expression of CD40L by the CD40-expressing tumor cells may have profound effects on tumor cell growth, differentiation, and apoptosis.

We did not have sufficient clinical information to correlate CD40L expression in the CLL B cells with clinical features of autoimmune disease in these patients. Given that the Fas antigen and its ligand have been identified as critical mediators of apoptotic deletion of peripheral, autoreactive B cells,60 we speculate that the failure to delete autoreactive B cells in patients with CLL might be due to failed Fas-mediated apoptosis in those nonmalignant cells. Specifically, CD40L-stimulated B cells induced to express the Fas antigen would be protected from Fas-mediated deletion by cross-linking of the surface antigen receptor.26 Thus, most B cells that receive inappropriate, CLL B-cell–derived help but bear Ig that is not engaged by antigen would undergo apoptosis, which might account for the hypogammaglobulinemia characteristic of this disease. However, B cells responsive to a particular viral or other infectious agent present in the host, or those reactive with autologous structures, might survive Fas ligation subsequent to CD40 engagement and differentiate into antibody-producing cells.

The authors thank the CLL patients and their physicians for providing clinical samples. We thank Dr Radha K. Vakkalanka for expert guidance in RT-PCR.