Basic Fibroblast Growth Factor Is Expressed by CD19/CD11c-Positive Cells in Hairy Cell Leukemia

Seven patients suffering from HCL and seven healthy donors were investigated. HCL was diagnosed on the basis of cell morphology, double immunofluorescence staining using monoclonal antibodies (MoAb) against CD11c and CD19 and BM and splenic histology.

NIH-3T3 mouse fibroblasts were obtained from American Type Culture Collection (ATCC; Rockville, MD). D15-7 (established in our own laboratory) are human skin fibroblasts from an 8-year-old healthy donor. The erythroleukemic cell line, K562, was kindly provided by Dr G. Klein (Karolinska Institute, Stockholm, Sweden).

PB was taken after informed consent from patients with HCL and healthy volunteers. PBMC were separated by density gradient centrifugation using Ficoll-Hypaque (Pharmacia, Uppsala, Sweden). Cells banding at the interphase were washed twice in phosphate-buffered saline (PBS), resuspended in culture medium, and seeded in 6-well plates at 2 × 10⁶ cells/mL. In certain experiments, T lymphocytes were depleted by a rosetting technique using neuraminidase-treated sheep red blood cells.

PBMC were grown in RPMI 1640 supplemented with 2 mmol L-glutamine, penicillin (100 U/mL), streptomycin (100 μg/mL), and 10% heat inactivated fetal calf serum (FCS) at 37°C in a humidified atmosphere containing 5% CO₂. For certain experiments, cells were activated by pokeweed mitogen (PWM, 10 μg/mL) or 12-o-tetradecanoyl-phorbol-13-acetate (TPA, 10 ng/mL) plus calcium-ionophore A23187 (Ca-Ip, 10 ng/mL).

If not stated otherwise, culture supernatant was harvested from unstimulated cultures after 48 hours, centrifuged at 200g for 10 minutes, sterile filtered, and stored at −20°C until assayed.

Cell culture media (RPMI 1640, MCDB 104), penicillin-streptomycin, L-glutamine, and FCS were obtained through GIBCO, Life Technologies (Paisley, UK). PWM, TPA, and Ca-Ip were purchased by Sigma Chemicals (St Louis, MO). 2-CdA was provided by Ortho Biotech (Raritan, NJ).

Monoclonal antibodies against CD3, CD5, CD11c, CD14, CD19, and the respective isotype controls, as well as ABComplex for Western blot analysis, were obtained through Dako (Copenhagen, Denmark), a Cy3-labeled anti-CD11c MoAb was purchased from Sigma Chemicals. A neutralizing mouse anti-bFGF MoAb and recombinant human bFGF (rhbFGF) were purchased from R&D Systems (Minneapolis, MN) and a biotin-labeled goat anti-mouse antibody was obtained through Accurate Chemicals (Westbury, NY). For Western blotting experiments, a mouse anti-bFGF MoAb (clone Ab-3, Oncogene Science, Uniondale, NY) was used.

Immunofluorescence analysis using a Cy3-labeled anti-CD11c and an anti-bFGF MoAb (R&D) was performed on aceton-fixed cells. Incubation steps with a biotinylated goat antimouse antibody and conjugates supplied in the TS amplification kit (Du-Pont-NEN, Boston, MA) were performed following instructions of the manufacturer.

MoAbs directed against CD3, CD5, CD11c, CD14, CD19 were used to analyze the cellular composition of the PBMC. For FACS analysis, 2 × 10⁵ cells in PBS/bovine serum albumin (BSA) (1%) containing 0.1% sodium azide were stained with the fluorescein isothiocyanate (FITC)-labeled MoAb or isotype control for 30 minutes at 4°C. Samples were analyzed on a FACScan (Becton Dickinson, San Jose, CA). Gates for data analysis were set using the respective isotype staining as negative control. Quantitative analysis of apoptotic cells was performed using an Annexin V-FITC detection kit obtained from Pharmingen (San Diego, CA).

Sera, BM aspirates, cell lysates, and CM from cultures were tested for presence of bFGF by a commercially available ELISA-system (Bio-Trak; Amersham, Backinghamshire, UK) following the manufacturer’s protocol. Samples were tested in duplicates, optical density was measured, and bFGF values were calculated with a computer assisted plate reader (Dynatech, Guernsey Channel Islands, UK). The detection limit of the test is 1 pg/mL.

Freshly isolated PBMC (10⁷ cells) from a patient with HCL (66% CD19⁺/CD11c⁺ cells) and a healthy donor were cultured as described above. Supernatants from controls and PWM-stimulated cultures were passed through a 0.2-μm filter and incubated with 300 μL heparin-coated beads (AF-Hep 650M; Toso Haas, Tokyo, Japan) for 1 hour at room temperature (RT). rhbFGF was treated identically serving as positive control. Beads were washed twice in 1 mmol/L Tris/HCl, 0.2 mol NaCl, and bound proteins were eluted by boiling for 3 minutes in 300 μL electrophoresis buffer (0.125 mmol/L Tris/HCl; pH 6.8; 1% sodium dodecyl sulfate [SDS]). A total of 2 × 10⁶ cells was resuspended in 1 mL electrophoresis buffer and lysed by boiling for 3 minutes. Unsoluble material was removed by centrifugation and aliquoted samples were stored at −20°C. Thirty-microliter samples were separated on homogenous gels under reducing conditions and transferred onto nitrocellulose by semidry blotting. Western blot analysis was performed using a mouse anti-bFGF MoAb followed by a biotin-streptavidin-horseradish-peroxidase incubation step. Positive reactions were visualized by enhanced chemiluminescence (Super Signal; Pierce, Rockford, IL) on ECL-Hyperfilm (Amersham, Backinghamshire, UK).

NIH-3T3 fibroblasts and primary human dermal fibroblasts were cultured under serum-free conditions for 24 hours before addition of CM from PBMC of HCL patients or healthy donors. In a separate set of experiments, CM was preincubated for 1 hour at 37°C with a neutralizing anti-bFGF antibody (10 μg/mL) before addition to cultures. After 2 or 4 days, fibroblasts were pulsed with 1 μCi of³H-thymidine (³H-TdR; NEN) for 8 hours, harvested, and ³H-TdR uptake was determined using a Beta-Plate Scintillation Counter (Wallac-Pharmacia, Uppsala, Sweden). Cell cultures were performed in triplicates using 96-well plates.

To measure proliferation of PBMC, 2 × 10⁵ cells/well were cultured in 96-well plates for 72 hours with and without 2-CdA and recombinant human bFGF, pulsed with ³H-TdR for 8 hours, and thymidine uptake was determined in a liquid scintillation counter.

Cells were harvested 48 hours after onset of cultures, washed twice in ice-cold PBS, and counted automatically (Coulter Counter, Coulter Electronics Ltd, Luton, UK). A total of 5 × 10⁶ cells were subjected to RNA isolation by the guanidium-isothiocyanate-phenol-chloroform extraction procedure. cDNA was synthesized using 1 μg RNA as a template and a commercially available kit (Pharmacia). Polymerase chain reaction (PCR) was performed in a final volume of 50 μL in the presence of 0.25 mmol/L of each dNTP, 1 U Taq polymerase (Finnzyme, Helsinki, Finland), and 100 pmol/L of either bFGF- or β-actin–specific primers. For β-actin PCR (upstream primer: 5′-AGG CCG GCT TCG CGG GCG AC-3′; downstream primer: 5′-CTC GGG AGC CAC ACG CAG CTC-3′),23 30 cycles were performed at 94°C for 1 minute (denaturing), 60°C for 90 seconds (annealing), and 72°C for 90 seconds (extension). The final extension was performed at 72°C for 3 minutes. Each experiment was repeated at least two times and results showed no difference in β-actin expression, indicating equal amounts of mRNA. For bFGF-specific PCR (upstream primer: 5′-ACC ACG CTG CCC GCC TTG CCC-3′; downstream primer: 5′-CTT ATA GCC AGG TAA CGG TTA GCA-3′),12 40 cycles of amplification were performed (denaturation at 92°C/1 minute, annealing at 62°C/90 seconds, extension at 72°C/1 minute, final extension step at 72°C/5 minutes). Amplified DNA fragments were visualized by agarose gel electrophoresis. Southern blotting, hybridization to a digoxigenin (DIG)-labeled bFGF-specific cDNA probe (by courtesy of Dr N. Frazer, University of Oxford, Oxford, UK) and exposure to x-ray films was performed following standard procedures.

Serum derived from seven patients with HCL and seven healthy donors (HD) was screened by a bFGF-specific ELISA (detection limit, 1 pg/mL) (Table 1). Figure 1A indicates that in serum from all patients with HCL, bFGF was detectable, ranging from 1 to 90 pg/mL with a mean of 29 pg/mL. With the exception of one individual (4.6 pg/mL), no bFGF could be found in the serum from HD. Because BM is a primary site of HC appearance, BM aspirates were also tested. As indicated in Fig 1B, the mean concentration of bFGF in HCL-derived BM (641 pg/mL) was more than 16 times higher than in samples from HD (mean, 39 pg/mL).

Because serum- or BM-derived bFGF might originate from nonhematopoietic cells or from cells that are disrupted during BM aspiration, we investigated whether PBMC from HD and HCL patients are capable of producing bFGF. Cells were cultured for 48 hours and culture supernatants were analyzed for bFGF by ELISA (Fig 2). While in cultures of unstimulated cells from HD no bFGF was detected, significant concentrations were found in HCL (mean value, 55 pg/mL). Stimulation with PWM led to a minute release of bFGF in only two of seven HD (2.4 pg/mL and 1.9 pg/mL). In HCL, however, an increase of up to 100% was seen (mean value, 110 pg/mL).

To examine if HCL-derived bFGF induces mesenchymal cell growth, human dermal fibroblasts were incubated with CM obtained from PBMC of an HCL patient and an HD. After 48 and 96 hours, cell proliferation was assessed by ³H-TdR uptake. As shown in Fig 3, CM from HCL induced a much higher increase in fibroblast proliferation than CM derived from HD. The addition of a neutralizing bFGF-specific MoAb markedly reduced the growth stimulatory effect of CM from HCL, but not that from HD.

Because binding to heparin is a feature of bFGF, CM from HCL patients was incubated with heparin beads for 30 minutes at RT. Heparin treatment of CM clearly reduced fibroblast proliferation triggered by HCL-derived CM or by 10 ng rhbFGF (data not shown).

To test whether bFGF production results from constitutive expression or from in vitro activation, PBMC from HDs and patients with HCL were subjected to mRNA isolation immediately after blood donation and gradient centrifugation. Results of a RT-PCR are shown in Fig 4A. bFGF mRNA was highly expressed in HCL, while no or only minute mRNA expression could be detected in PBMC from HDs.

To examine if the expression of bFGF mRNA is paralleled by the synthesis of bFGF protein, lysates from freshly isolated cells from both HD and patients with HCL were subjected to bFGF-ELISA. As shown in Fig 4B, high amounts of intracellular bFGF were detectable in HCL-derived cells only.

Stimulation of PBMC for 48 hours by a mitogen (PWM) or activators of protein kinase-C (TPA+Ca-Ip) (Fig 4C) enhanced bFGF-specific mRNA expression. In HCL, the basic expression of bFGF mRNA was enhanced by both PWM and TPA+Ca-Ip. In HDs, treatment with either substances led to weak mRNA expression.

The concentration of extracellular bFGF protein in unstimulated HCL cultures did not change significantly over a period of 6 hours, but then gradually declined. In contrast, stimulation by TPA+Ca-Ip resulted in upregulation of bFGF after 24 hours, possibly indicating a correlation between cellular activation via protein kinase C and bFGF production (data not shown).

bFGF is expressed in several isoforms.24 To investigate if a particular isoform is predominantly produced in HCL, immunoblotting was performed. Samples were separated on SDS gels under reducing conditions (see Materials and Methods), blotted, and analyzed using an anti-bFGF MoAb. The pattern of bFGF protein expression detected in lysates derived from PBMC of HCL patients showed three distinct isoforms of approximately 18, 23, and 25 kD, while a lysate from an HD was negative (Fig 5A). In a second set of experiments, PBMC-derived cell-free CM from an HD and an HCL-patient were subjected to heparin chromatography. Bound proteins were eluted and analyzed as above. As demonstrated in Fig 5B, CM obtained from an HD again was essentially devoid of bFGF. In sharp contrast to lysates, the supernatant derived from a patient with HCL contained only the 23-kD high-molecular-weight isoform.

Double immunofluorescence analysis was performed to confirm whether HC are the producers of bFGF. PBMC from a patient with 66% CD19⁺/CD11c⁺ cells were examined. A typical reaction pattern of a bFGF- and a CD11c-specific MoAb is shown in Fig 6, showing that most of the CD11c-positive cells also express bFGF. In contrast, PBMC from an HD were negative for bFGF (data not shown). Because these results strongly suggested that CD19⁺/CD11c⁺ cells might be the producers of bFGF, we investigated the capacity of this particular cell population to produce bFGF. PBMC from the same patient and from an HD were investigated for coexpression of CD11c and CD19 by FACS analysis. While 66% of the PBMC from the patient expressed CD19, as well as CD11c, only 1.9% of the PBMC from the HD were double-positive for these antigens. When cells were cultured for 48 hours, only PBMC from the patient with HCL released bFGF (272 pg/mL; Fig 7A). In a parallel set of experiments, PBMC were separated from E-rosetting cells by gradient centrifugation. While in the case of HCL, this procedure yielded 91% CD19⁺/CD11c⁺ cells (<2% CD3 or CD5 and <1% CD14-positive cells), the non–T-cell population from the healthy donor contained 5% CD19⁺/CD11c⁺ cells only (<5% CD3, CD5, and CD14-positive cells, Fig 7B). Again, non-T cells from HD did not produce any detectable bFGF. In contrast, CM from PBMC enriched for CD19⁺/CD11c⁺ cells of the HCL patient contained high levels of bFGF (539 pg/mL). Therefore, enrichment for CD19⁺/CD11c⁺ cells leads to an increase of almost 100% (272 pg/mL to 539 pg/mL).

Endogenous bFGF production has a beneficial effect on cell proliferation and rescue from apoptotic cell death in various cell types, as has been reported earlier.25-28 To test whether bFGF exerts a protective effect on HC, cultures were set up in the presence of 2-CdA, a drug that induces apoptosis in HC29and is successfully used for treatment of HCL.30 As can be seen from Fig 8A, bFGF-producing cells from HCL could only be partially inhibited by the addition of 2-CdA. In contrast, proliferation of PBMC derived from an HD was completely abrogated. A possible explanation for the weak inhibition of HC might be that these cells express bFGF spontaneously and therefore are protected from the cytopathic effects of 2-CdA. This assumption is supported by results of experiments in which cultures treated with 2-CdA were supplemented with rhbFGF. Despite the presence of 2-CdA, exogenous rhbFGF restored the proliferative capacity of PBMC from a patient with HCL and an HD, suggesting that bFGF protects PBMC from the cytopathic action of 2-CdA. In vivo, such protection would require a constitutive expression of this cytokine even in the presence of 2-CdA. Thus, we examined if 2-CdA has an influence on bFGF release by HC. PBMC derived from two patients suffering from HCL were cultured for 48 hours and supernatants were screened for bFGF by ELISA. Indeed, PBMC from patient no. 1 produced high levels of bFGF, regardless of the presence of 2-CdA, while PBMC from patient no. 2 were inhibited by 22% only (Fig 8B). In this regard, it is interesting to note that patient no. 1 is resistant to therapy with 2-CdA.

In the next set of experiments, we tested whether bFGF expression in HC is paralleled by a reduced sensitivity for 2-CdA–mediated apoptosis. Cells were cultured for 48 hours and analyzed for Annexin-V binding, a feature of cells undergoing apoptosis. Figure 8C shows that 2-CdA induced Annexin-V binding in 52% of PBMC from a healthy donor, while only 26% of bFGF-producing HC bound Annexin-V. These results might indicate resistance to 2-CdA via bFGF.

The present study reports on the detection of elevated levels of bFGF in serum and BM from patients with HCL. This study also indicates that in HCL, bFGF is released in a biologically active form by the malignant cells, as CM derived from HC was shown to induce a proliferative response on fibroblasts, which could be diminished by treatment with a neutralizing anti-bFGF antibody. Stimulation of PBMC with PWM or TPA plus Ca-Ip led to an increase in bFGF mRNA expression in HCL, substantiating that bFGF expression can be upregulated by cellular activation. In cell lysates, three different isoforms can be detected by immunoblotting, but only a single bFGF isoform of approximately 23 kD is released by HC. Double-immunofluorescence and cell separation studies showed that the cytokine is produced and exported by CD19⁺/CD11c⁺ cells. Finally, bFGF expression is not blocked by 2-CdA and seems to mediate resistance to 2-CdA–mediated apoptosis.

Expression of bFGF and its receptors has been shown in few other hematologic malignancies, and the cytokine was implicated to play a potential pathological role in these diseases. Thus, in patients with myelofibrosis, an increased production of bFGF has been correlated with a possible involvement with the fibrotic process.14 In chronic lymphocytic leukemia, elevated bFGF levels were found to correlate with the stage of disease activity and resistance to fludarabine.15 

HCL, a rare form of chronic leukemia, is characterized by several unique and unusual features (reviewed in Hess31). Some of them might be explained by abnormalities in cytokine production32 leading to BM insufficiency or deregulation of adhesion molecule expression.33 

It is important to note that release of bFGF seems to be a feature of tumor cells, as transformed cells derived from breast carcinoma or fibrosarcomas export bFGF, while in nontransformed fibroblasts, bFGF remains cell-associated.34-36 In addition, acquisition of a signal peptide by molecular cloning converts bFGF to induce transformation of mouse fibroblasts, which are tumorigenic and metastatic in nude mice.37 

The bFGF detected in culture supernatants from HCL patients was capable of inducing proliferation of primary human fibroblasts. This might implicate that the cytokine exerts important effects also on cells in the close proximity to HC, ie, on mesenchymal cells, or molecules deposited within the extracellular matrix, eg, heparan sulfate proteoglycans (HSPG). Complex formation of HSPG with bFGF facilitates binding to FGF receptors and thereby triggers cellular activation.38,39 

In HCL, bFGF might also be operative through other growth factors. Among these transforming growth factor-β1 (TGF-β1) is a possible candidate. Interactions between bFGF and TGF-β1 have been described in several biological systems40-42 and may also be relevant for BM fibrosis seen in HCL. In this context, bFGF might contribute to BM fibrosis via activation of latent TGF-β1 in the BM microenvironment and consequently induction of extracellular matrix (ECM) synthesis and accumulation.43Indeed, both active and latent forms of TGF-β are highly elevated in HCL, particularly in BM (M. Shehata, in preparation).

One possible mechanism by which bFGF might exert a direct role in leukemogenesis could be through the prevention of apoptosis. There is evidence presented by Menzel et al15 who showed that bFGF accumulates intracellularly in malignant B cells from chronic lymphocytic leukemia (B-CLL) and delays fludarabine-induced cell death. Similarily, experiments described in this report showed that recombinant bFGF was capable of protecting HD-derived, as well as HCL-derived PBMCs, from the cytopathic effects of 2-CdA. Moreover, production and release of endogenous bFGF was not affected by 2-CdA. Therefore, overexpression and export of bFGF may mediate resistance to therapy and contribute to tumor progression.

Besides a possible direct effect on the survival and growth of CD19⁺/CD11c⁺ B cells, bFGF may also influence immune surveillance. In vitro and in vivo studies have shown that bFGF inhibits adhesion of natural killer (NK) cells to endothelial cells.33 Provided that bFGF functions in a similar way on HC, it may exert a protective mechanism for the malignant cells in the presence of NK cells. Evidence for a reduced NK cell activity in HCL has been provided earlier.44,45 

In primary mesenchymal cells, bFGF is produced as an 18-kD glycoprotein without a signal sequence.46,47 In addition to the 18-kD form, at least three N-terminal extended bFGF isoforms of approximately 22, 23, and 25 kD molecular weight have been described. They originate from unusual translational initiation at CUG codons upstream of the original AUG initiation codon.48 While the 18-kD form of bFGF is present in the cytosol, these high molecular weight forms are predominantly associated with the nucleus due to a nuclear localization signal in the alternatively expressed N-terminal region.49Interestingly, bFGF derived from HCL displayed a molecular weight (MW) of 23 kD, an isoform that is characterized by its nuclear localization. It is feasible that high MW forms of bFGF also control gene expression because the presence of a high MW form within the nucleus may facilitate interaction with transcription factors. It has been shown by Kevil et al50 that the translation initiation factor eIF-4 preferentially increases translation of CUG-1 isoforms, resulting in release of large quantities of the 23-kD isoform. Whether or not transcription factor-mediated mechanisms are also responsible for the elevated bFGF expression seen in HCL remains to be elucidated.

Both production and distribution of a particular isoform may be instrumental for cell proliferation, as well as development and progression of tumors. Therefore, bFGF may play a role in the development of pathological conditions in HCL and hence be involved in myelofibrosis, prevention of apoptosis, and proliferation of HC.