To define the distribution of folate receptor β (FRβ)–expressing cells in various tissues, including rheumatoid arthritis (RA) synovial tissues, and to verify the effects of an immunotoxin composed of an anti-FRβ monoclonal antibody (mAb) and truncated Pseudomonas exotoxin A (PEA) on apoptosis and tumor necrosis factor α (TNFα) production by adherent synovial mononuclear cells from RA patients.
Anti-FRβ mAb were produced by immunizing mice with FRβ-transfected murine pre–B cells. The distribution of the FRβ antigen was examined by immunohistochemical analysis using anti-FRβ mAb and macrophage-specific anti-CD163 mAb. Anti-FRβ mAb was chemically crosslinked with truncated PEA. FRβ-expressing macrophages were produced by the transfection of adenovirus vector containing the FRβ gene. Apoptotic cells were detected by staining with propidium iodide. TNFα was measured by enzyme-linked immunosorbent assay.
FRβ-expressing cells were not present in peripheral blood leukocytes and their activated cells. In all of the tissues examined, most FRβ-expressing cells were CD163+. The immunotoxin significantly induced the apoptosis of FRβ-transfected macrophages and adherent RA synovial mononuclear cells and inhibited TNFα production by adherent RA synovial mononuclear cells.
We demonstrated the limited distribution of FRβ-expressing cells in various tissues. The immunotoxin targeting FRβ-expressing cells will provide a therapeutic tool for rheumatoid synovitis.
Activated synovial macrophages are thought to play an important role in the pathogenesis of rheumatoid arthritis (RA) (1–3). These macrophages release proinflammatory cytokines, proteinases, and other chemical mediators that lead to the development of synovitis and joint destruction. The removal of macrophages decreases the severity of joint disease in animal models of RA (4, 5). Several antirheumatic drugs, including D-penicillamine and gold salts, affect synovial macrophages (6, 7). Furthermore, one of the actions of anti–tumor necrosis factor α (anti-TNFα) therapy may be the result of antibody-dependent cellular cytotoxicity of membranous TNFα-expressing macrophages (8, 9). Thus, reagents that target synovial macrophages may be very effective in the treatment of RA.
We have shown that folate receptor β (FRβ) messenger RNA (mRNA) and protein are highly expressed by synovial macrophages, but not by other synovial cells, in the inflamed joints of patients with RA (10). In a rat arthritis model, 99mTc-labeled folate was found to specifically accumulate in affected joints (11). Moreover, we have shown that treatment with a drug that blocks FRβ reduces disease activity in murine collagen-induced arthritis (12).
FRβ is expressed by acute myelogenous leukemia cells (13, 14). In vitro studies suggest that folate liposomes that contain cytotoxic drugs might be useful for the treatment of this disease (15). In addition, immunotoxins, which are composed of specific antibodies combined with toxins, such as Pseudomonas exotoxin A (PEA), diphtheria toxin, gelonin, saporin, and ricin A, effectively bind to malignant cells that bear the appropriate surface antigens (16, 17) and monocyte/macrophages (18–20). However, there are no reports of immunotoxins that target FRβ. Before we develop and test these immunotoxins, it is important to determine the distribution of FRβ-expressing cells in normal and inflamed tissues from humans.
In the present study, we produced monoclonal antibodies (mAb) against human FRβ. We used these mAb to detect FRβ-expressing cells in a variety of tissues and cells. We then prepared an immunotoxin composed of an anti-FRβ mAb and truncated PEA (16, 17) and verified that the immunotoxin selectively targets RA synovial macrophages in vitro.
MATERIALS AND METHODS
Synovial tissue was obtained from 10 RA patients who underwent total knee replacement surgery. These patients fulfilled the American College of Rheumatology (formerly, the American Rheumatism Association) 1987 criteria for RA (21) and had previously been treated with either methotrexate, D-penicillamine, gold salts, sulfasalazine, bucillamine, or prednisone (<7.5 mg/day) or combinations of these drugs. Adherent synovial mononuclear cells were prepared as previously described (22). Briefly, synovium was digested with type V collagenase (Sigma-Aldrich, St. Louis, MO) for 45 minutes at 37°C. Cell suspensions were filtered through a stainless steel mesh. Mononuclear cells were isolated on a Ficoll-Hypaque gradient, incubated in plastic plates for 1 hour at 37°C, and washed 10 times with phosphate buffered saline. These steps for obtaining adherent cells were repeated. Using the same procedures, adherent lung mononuclear cells were prepared from lung tissues obtained by surgical resection for lung carcinoma from 4 patients.
Samples of skin (n = 3), lung (n = 3), liver (n = 2), spleen (n = 2), kidney (n = 1), intestine (n = 3), and lymph node (n = 3) obtained by surgical resection for cancers were used for immunohistochemistry analysis to detect FRβ-expressing cells.
Peripheral blood was drawn from 4 healthy donors, and synovial fluid was drawn from the swollen knee joints of 3 RA patients. Monocytes were prepared from the peripheral blood of healthy donors and were stimulated with lipopolysaccharide (LPS; derived from Escherichia coli 0111:B4) (Sigma-Aldrich), macrophage colony-stimulating factor (M-CSF) (Pierce, Rockford, IL), or interferon-γ (IFNγ) (Pierce) as described previously (22).
Informed consent was obtained from all donors in accordance with the requirements of the Human Investigations Committee of Kagoshima University.
Production of mAb against the FRβ antigen.
Murine FRβ-expressing pre–B cells were prepared for immunogens as follows. FRβ complementary DNA (cDNA) containing the Kozak consensus and coding sequence was prepared from a reverse transcription–polymerase chain reaction (RT-PCR) product derived from RA synovial cells, as previously described (10). The primer sequences were 5′-AGAAAGACATGGTCTGGAAATGGATG-3′ (upstream) and 5′- GACTGAACTCAGCCAAGGAGCCAGAGTT-3′ (downstream). The cDNA fragment was ligated with pCR2.1-TOPO vector (Invitrogen, Carlsbad, CA), and its identity was confirmed by sequencing. The Eco RI–digested insert was then ligated into pEF-BOS (23). This construct was transfected into B300-19 cells (murine pre–B cell line) with the use of Lipofectamine (Invitrogen). After 48 hours, transfected cells were selected in 1,000 μg/ml of G-418 disulfate (Nakarai Tesque, Kyoto, Japan) in Dulbecco's minimum essential medium (DMEM) containing 10% fetal calf serum. Resistant colonies were examined by RT-PCR using the above primers.
BALB/c mice were immunized with the FRβ-transfected B300-19 cells. Their spleen cells were fused with NS-1 myeloma cells. Hybridomas were screened for reactivity with FRβ-transfected B300-19 cells by flow cytometric and Western blot analyses. Two anti-FRβ mAb (36b [IgG2a isotype] and 94b [IgG1 isotype]) were selected for further evaluation.
Preparation of truncated PEA.
Truncated PEA was purified as previously described (24). Briefly, the plasmid pMS8-38-402, which encodes the truncated LysPE38QQR protein, was transfected into BL21 (DE3) cells (Stratagene, La Jolla, CA). After induction with IPTG, periplasm was collected. Truncated PEA was purified through Poros HQ (Applied Biosystems, Foster City, CA) and TSK 3000 SW (Tosoh, Tokyo, Japan) columns using a Vision Workstation Liquid Chromatography system (Japan Perceptive, Tokyo, Japan). The activity of truncated PEA was evaluated by the Carroll method (25). Briefly, the ADP ribosylation of truncated PEA was measured using wheat germ extract containing ADP ribosyltransferase (Promega, Tokyo, Japan) and 14C-labeled nicotinamide adenine dinucleotide (Daiichikagaku, Tokyo, Japan).
Preparation of immunotoxin.
Anti-FRβ immunotoxin was prepared according to a slight modification of Hassan's procedure (24). Briefly, 1 ml of anti-FRβ mAb (3 mg/ml) was incubated with 100 μg of succinimidyl trans-4-(maleimidomethyl) cyclohexane-1-carboxylate (SMCC; Sigma-Aldrich) for 1 hour at room temperature. One milliliter of truncated PEA (10 mg/ml) was incubated with 400 μg of succinimidyl 3-(2-pyridyldithio)propionate (SPDP; Sigma-Aldrich) for 12 hours at 4°C. Excess SMCC and SPDP were removed using a PD-10 column (Amersham Pharmacia, Tokyo, Japan). Truncated PEA coupled with SPDP was activated with tris(2-carboxyethyl)phosphine (Molecular Probes, Eugene, OR) and mixed with anti-FRβ mAb–SMCC. This mixture was purified using PD-10, Poros HQ, and TSK 3000 SW columns.
Determination of the binding capacity of immunotoxin to the FRβ antigen.
A total of 2 × 105 FRβ-transfected B300-19 cells, FRβ-transfected B300-19 cells pretreated with anti-FRβ mAb (10 μg/ml for 15 minutes), or nontransfected B300-19 cells were sequentially incubated with the immunotoxin (1 μg/ml) or with the control mAb (1 μg/ml) and then with rabbit anti-PEA antibody, followed by fluorescein isothiocyanate–conjugated goat anti-rabbit antibody (Zymed, South San Francisco, CA). Cells were then analyzed by flow cytometry.
Adenovirus-mediated FRβ gene transfer into monocyte/macrophages.
FRβ cDNA was excised from the pCR2.1 vector by Xba I digestion, blunted using a DNA blunting kit (Takara Shuzo, Kyoto, Japan), and subcloned into the Swa I site of pAxCAwt, which had been generated using an adenovirus expression vector kit (Takara Shuzo). Following the manufacturer's instructions, human embryonic kidney 293 cells (RIKEN Cell Bank, Tsukuba, Japan) were transfected with this cosmid vector and adenovirus genomic DNA–terminal protein complex to produce replication-incompetent E1- and E3-deficient adenoviruses expressing FRβ. Titers of recombinant adenovirus were determined by plaque assays of 293 cells (26). To obtain high titers of adenovirus, infection of the virus into 293 cells was performed 4 times. The adenoviruses were suspended in culture medium, adjusted to 2 × 108 plaque-forming units/ml, and stored at −80°C until they were used.
Monocyte/macrophages were transfected with the adenovirus as previously described (27). Briefly, monocytes were incubated with 20 ng/ml of M-CSF for 24 hours, transfected with the adenovirus vector at a multiplicity of infection of 50, and centrifuged at 2,000g for 2 hours at 37°C.
Measurement of apoptotic cells.
FRβ-expressing B300-19 cells (2 × 105/ml), B300-19 cells (2 × 105/ml), macrophages (1 × 106/ml), adherent RA synovial mononuclear cells (1 × 106/ml), and adherent lung mononuclear cells (1 × 106/ml) in DMEM (Nikken, Kyoto, Japan) containing 10% human AB serum were placed into 24-well plates and incubated at 37°C in an atmosphere of 5% CO2 for various times (see below) in the presence of either immunotoxin or a mixture of equimolar amounts of anti-FRβ mAb and truncated PEA at the concentrations indicated below. Propidium iodide (Molecular Probes) staining was performed as described by Nicoletti et al (28). Briefly, cells were stained with propidium iodide (50 μg/ml) in 0.1% sodium citrate plus 0.1% Triton X-100 for 20 minutes at 4°C. Apoptotic cells were detected by flow cytometry.
Immunoprecipitation and Western blot analyses.
FRβ-expressing B300-19 cells, B300-19 cells, and adherent RA synovial mononuclear cells were biotinylated with Sulfo-NHS-LC-Biotin (Pierce) and lysed in buffer containing 1% Triton X and proteinase inhibitors. The lysates were immunoprecipitated with either anti-FRβ mAb or control mAb bound to protein G–agarose (Santa Cruz Biotechnology, Santa Cruz, CA). The immunoprecipitates were run on sodium dodecyl sulfate (SDS)–polyacrylamide 12% gels under reducing conditions, transferred to Hybond ECL nitrocellulose membranes (Amersham Pharmacia), and stained with streptavidin–peroxidase (Zymed). The membranes were incubated in enhanced chemiluminescent reagents (Amersham Pharmacia). Chemiluminescence was detected on Hyperfilm ECL (Amersham Pharmacia).
The immunotoxin, anti-FRβ mAb, and truncated PEA were run on SDS–polyacrylamide 10% gels under reducing conditions, transferred to Hybond ECL nitrocellulose membranes, and reacted with rabbit anti-PEA antibody (a gift of Dr. Ira Pastan, Laboratory of Molecular Biology, National Cancer Institute, National Institutes of Health, Bethesda, MD) and either horseradish peroxidase (HRP)–conjugated goat anti-rabbit antibody or HRP-conjugated goat anti-mouse IgG antibody (both from Bio-Rad, Hercules, CA). The blots were then developed as described above.
Serial frozen sections were stained using biotinylated anti-FRβ mAb or biotinylated anti-CD163 mAb (Maine Biotechnology Service, Portland, ME), streptavidin–peroxidase, and aminoethylcarbazole reagent (Nichirei, Tokyo, Japan). Cryosectioned RA synovial tissues were stained using biotinylated anti-FRβ mAb followed by rhodamine–avidin D (Vector, Burlingame, CA) and Alexa Fluor 488–conjugated anti-CD163 mAb (Molecular Probes). Double-stained sections were imaged by confocal laser microscopy (Leica Microsystems, Wetzlar, Germany) using a 488-nm argon ion laser and a band-pass 515–545-nm emission filter for Alexa Fluor 488 and a 568-nm argon ion laser and a band-pass 570–630-nm emission filter for rhodamine. Large field of view images were acquired using a 20× objective lens with detector pinholes completely opened to give pseudoconventional images. The resulting images were 12 bit, and 1,024 × 1,024 pixels, giving a pixel resolution of 0.45 μm pixel size. Contrast and brightness were set using bright samples to ensure that there was no saturation of the pixels.
Measurement of TNFα production by adherent RA synovial mononuclear cells.
RA synovial mononuclear cells (1 × 106/ml) were suspended in DMEM containing 10% human AB serum, placed in 24-well plates with either the immunotoxin or a mixture of equimolar amounts of anti- FRβ mAb and truncated PEA, and incubated for 12 hours at 37°C in an atmosphere of 5% CO2. After 2 washes with phosphate buffered saline, the cells were incubated for 12 hours with 10 μg/ml of LPS. TNFα in culture supernatants was measured in duplicate by an enzyme-linked immunosorbent assay using an anti-TNFα antibody and a biotinylated anti-TNFα antibody according to the manufacturer's protocol (Pierce). Serial dilutions of recombinant TNFα (Pierce) were used as the standard for the determination of TNFα concentrations.
The nonparametric Mann-Whitney U test was used to test for differences. P values less than 0.05 were considered significant.
Characterization of FRβ-specific mAb.
The 36b and 94b mAb reacted with FRβ-transfected B300-19 cells but not with B300-19 cells or with KB nasopharyngeal epidermoid carcinoma cells expressing the FRα antigen (Figure 1A). Moreover, these antibodies immunoprecipitated a 40-kd protein from FRβ-transfected B300-19 cells and adherent RA synovial mononuclear cells (Figure 1B), findings consistent with previously reported results using a polyclonal FRβ-specific mAb (10).
We used the mAb to evaluate the expression of FRβ on peripheral blood leukocytes and in various tissues. The antibodies did not react with freshly obtained peripheral blood leukocytes or with monocyte/macrophages that had been stimulated with LPS, M-CSF, or IFNγ (data not shown). In contrast, FRβ-expressing cells were observed in samples of skin, lung, liver, spleen, kidney, intestine, and lymph node (data not shown), as well as in RA synovium (Figure 2). FRβ was expressed at high levels in RA synovial tissues and at much lower levels in the other tissues.
To determine which cells expressed FRβ, we compared anti-FRβ mAb–stained sections with serial sections stained with macrophage-specific anti-CD163 mAb. In all tissues, most FRβ+ cells also expressed CD163. FRβ-expressing cells in RA synovial tissues were distributed mainly in the sublining layer, while CD163+ cells were observed in the lining and sublining layers. Interestingly, alveolar macrophages did not stain with the anti-FRβ mAb, and Kupffer cells stained weakly with the anti-FRβ mAb.
Effectiveness of the immunotoxin composed of anti-FRβ mAb and truncated PEA in targeting RA synovial macrophages.
This limited distribution of FRβ-expressing cells led us to hypothesize that toxin-conjugated anti-FRβ mAb (immunotoxins) would specifically inhibit the activity of RA synovial macrophages and be useful for the treatment of RA. To test this hypothesis, we conjugated an anti-FRβ mAb (36b) with truncated PEA. Western blot analysis under reducing conditions showed that most of the immunotoxin molecules were of higher molecular weight than the anti-FRβ mAb heavy chain. Several immunotoxin bands reacted with both anti-PEA antibody and anti-mouse IgG antibody, indicating that the mAb chains were conjugated with truncated PEA (Figure 3A). Furthermore, flow cytometry studies indicated that the immunotoxin bound to the FRβ antigen (Figure 3B). Specifically, the immunotoxin reacted with FRβ-expressing B300-19 cells but not with B300-19 cells. Moreover, the binding of immunotoxin to FRβ-expressing B300-19 cells was inhibited by an excess of anti-FRβ mAb.
Next, we examined whether the immunotoxin could induce apoptosis of FRβ-expressing cells. As expected, the immunotoxin induced the apoptosis of FRβ-transfected B300-19 cells, but not B300-19 cells. In addition, a mixture of anti-FRβ mAb and truncated PEA did not induce apoptosis of FRβ-transfected B300-19 cells (Figure 4). These findings indicate the specific effect of the immunotoxin on FRβ-expressing cells.
The ability of the immunotoxin to induce apoptosis of FRβ-expressing macrophages was then evaluated. As described above, peripheral blood monocytes and cultured or in vitro–activated macrophages do not express FRβ. Thus, we transfected monocyte/macrophages with a recombinant adenovirus vector encoding the FRβ gene. The percentages of FRβ-expressing cells after transfection ranged from 30% to 40% (Figure 5A). When the immunotoxin was added to cultures of FRβ-expressing macrophages or FRβ-nonexpressing macrophages, the immunotoxin induced significant apoptosis of the FRβ-expressing macrophages as compared with the FRβ-nonexpressing macrophages. However, at the highest concentration tested (5 μg/ml), the mixture of anti-FRβ antibody and truncated PEA induced apoptosis of the FRβ-expressing and FRβ-nonexpressing macrophages (Figure 5B).
We then studied the effects of the immunotoxin on RA synovial macrophages. Adherent RA synovial mononuclear cells contained 40–50% FRβ-expressing cells (or CD14+ cells). Many freshly obtained RA synovial macrophages are already apoptotic (29). Thus, the maximum percentage of adherent RA synovial mononuclear cells that could be induced to undergo apoptosis by the immunotoxin would be 40–50%. We found that the immunotoxin induced apoptosis in fewer than 25% of adherent RA synovial mononuclear cells (Figure 6A). In contrast, the immunotoxin did not induce apoptosis in RA synovial fibroblast-like cells from long-term cultures (data not shown).
Adherent lung mononuclear cells contain FRβ– alveolar macrophages and FRβ+ tissue macrophages. As with adherent RA synovial mononuclear cells, ∼40–50% of adherent lung mononuclear cells expressed CD14. We found that the concentration of immunotoxin needed to induce apoptosis of adherent lung mononuclear cells was higher than that needed to induce apoptosis of adherent RA synovial mononuclear cells. These results suggest that RA synovial macrophages are more sensitive to the immunotoxin than are lung macrophages. Next, we examined the effects of the immunotoxin on TNFα production by LPS-stimulated adherent RA synovial mononuclear cells. At low concentrations, the immunotoxin significantly inhibited TNFα production, as compared with the mixture of anti-FRβ mAb and truncated PEA (Figure 6B).
We produced 2 mAb that react with FRβ but not with FRα. Immunohistochemical analysis revealed FRβ-expressing cells in a wide variety of human tissues; most of the FRβ-expressing cells also expressed the macrophage marker CD163. However, FRβ was not detected on peripheral blood leukocytes, even after in vitro stimulation with LPS, M-CSF, or IFNγ. Previous studies using a polyclonal antibody and Northern blotting techniques indicated that myeloid cells and activated macrophages express FRβ antigen and mRNA; however, FRβ on the myeloid cells was not functional (14). It is intriguing that epitopes recognized by our mAb are absent on myeloid cells. This suggests that these epitopes might be associated with the function of FRβ.
It has been reported that mice with the Folbp-2 deletion, which is equivalent to FRβ, showed normal growth (30). Furthermore, folic acid has a higher affinity for FRα than for FRβ. At present, the physiologic role of FRβ on hematopoietic cells is not well defined. We recently showed that peroxynitrite generated from nitric oxide and reactive oxide induced the nitration of folic acid; one of the products, 10-nitro-folic acid, has a higher affinity for FRβ than for FRα (31). These findings, combined with those from the current study, suggest that FRβ on tissue macrophages might be required for the incorporation of folate derivatives for the synthesis of proinflammatory proteins.
We found FRβ-expressing macrophages in the sublining layer of RA synovial tissues. In a previous study by Cauli et al (32), many 27E10+ early macrophages were found in the sublining layer, whereas 25F9+ mature macrophages were more abundant in the lining layers. Thus, the majority of FRβ-expressing macrophages in RA synovial tissues seem to be acute inflammatory macrophages. Further examinations using various pathologic tissues should help to clarify the nature of FRβ-expressing macrophages.
There are several reports about folate-containing reagents that target FRβ-expressing leukemia cells and macrophages (33, 34). However, folic acid binds more avidly to FRα than to FRβ. In addition, cells have routes for folic acid uptake other than via folate receptors. Thus, reagents with higher affinity for FRβ than for FRα should be used for targeting FRβ-expressing cells. Anti-FRβ mAb has many advantages over other methods of drug delivery via FRβ in terms of specificity. In the present study, we produced an immunotoxin composed of an anti-FRβ mAb and a truncated PEA (LysPE38QQR), which lacks the cell-binding domain but retains the translocation and the adenosine diphosphate ribosylation domains (24). After internalization and proteolytic processing, domain II functions to translocate the toxin to the cytosol; the domain catalytically ADP-ribosylates elongation factor 2 in the cytosol, leading to the arrest of protein synthesis and the induction of apoptosis (16, 17).
Although most malignant cells are actively proliferating, RA synovial macrophages in the cultures we used are nondividing. We assumed that monocyte/macrophages might be resistant to PEA, as compared with dividing tumor cells. However, in the present culture system, our immunotoxin induced apoptosis and inhibited the production of TNFα by adherent RA synovial mononuclear cells. FRβ was found only on macrophages in the adherent RA synovial mononuclear cell cultures. Moreover, macrophages are the main producers of TNFα in adherent RA synovial mononuclear cell cultures (1, 2). The immunotoxin did not induce apoptosis of cultured RA synovial fibroblast-like cells. Taken together, these findings support the notion that the immunotoxin mainly damaged RA synovial macrophages. The action of the immunotoxin is not mediated through macrophages Fcγ receptors, since the immunotoxin induced more apoptosis of the FRβ gene–transfected macrophages than of the antisense gene–transfected macrophages.
The data from the present study suggest that the immunotoxin might ameliorate joint inflammation in patients with RA by inhibiting the activity of synovial macrophages. However, improvements in the immunotoxin, such as using recombinant monomeric or dimeric Fv immunotoxins, would cause less immunogenicity and more accessibility to lesions (35, 36).
The toxic side effects of immunotoxins are of at least 2 types. One type results from damage to normal cells that express the target antigen. Another type is nonspecific and is usually characterized by damage to liver cells (37). A recent study indicated that an anti-CD64 mAb conjugated with ricin A induced apoptosis of RA synovial fluid macrophages and inhibited the production of TNFα and interleukin-1β in synovial tissue explants (20). However, the CD64 antigen is expressed on monocytes and activated neutrophils in addition to tissue macrophages. Thus, immunotoxins using an anti-FRβ mAb have advantages over those using an anti-CD64 mAb because of the limited distribution of FRβ and the low levels of FRβ expression in normal tissues (38). Severe systemic toxicity might be observed with the in vivo use of immunotoxins. Thus, we would prefer to develop strategies to target activated macrophages in the joints by local injections of the immunotoxin during the first clinical trial.
We have recently observed that our immunotoxin inhibited the growth of FRβ-expressing HL-60 cells (promyelocytic leukemia cells) implanted in SCID mice (Nagayoshi R, et al: unpublished observations). In addition to reducing the activity of RA and FRβ-expressing leukemias, the immunotoxin should reduce the activity of other diseases in which macrophage activation is involved in the pathogenesis (39).
The authors would like to thank Dr. Ira Pastan (Laboratory of Molecular Biology, National Cancer Institute, National Institutes of Health, Bethesda, MD) and Dr. S. A. Michie (Department of Pathology, Stanford University School of Medicine, Palo Alto, CA) for critical review of the manuscript.