Enhancement of Polymorphonuclear Cell-mediated Tumor Cell Killing on Simultaneous Engagement of FcγRI (CD64) and FcαRI (CD89)¹

Immunotherapeutic approaches that harness the cytotoxic ability of immune cells to reject tumor cells receive increasing levels of attention. Both T cells and myeloid cells are considered suitable effector cells and have documented ability to kill tumor cells (1). PMN³ cells represent an attractive effector cell population, and their numbers can be easily increased in vivo by G-CSF (2). To elicit cytotoxic responses, these cells require activation via trigger molecules that can be linked to target cells via mAb, and several mAb targeting to tumor cells have recently been approved for cancer therapy (3). The prototypic antitumor Ab rituximab, a chimeric anti-CD20 mAb, was shown to elicit prominent antitumor effects in patients with non-Hodgkin’s B-cell lymphoma (4, 5), whereas Herceptin, an anti-HER-2/neu mAb, induced promising results in breast cancer patients (3, 6). Because interaction with Fc receptors was reported to be crucial for therapeutic responses induced by antitumor mAb (7), the development of BsAb targeting to select Fc receptors may represent a way to further improve therapeutic activity (1, 8, 9).

Both the class I receptors for IgG (FcγRI, CD64) and IgA (FcαRI, CD89) have been identified as candidate therapeutic targets (9, 10, 11). These receptors exhibit a myeloid-restricted cell distribution and potently trigger effector functions like phagocytosis and tumor cell lysis (12, 13). PMN cells constitutively express FcαRI and can be induced to express FcγRI upon treatment with IFN-γ or G-CSF (2, 14). We thus posed the question whether it would be feasible to use both receptors simultaneously as trigger molecules for immunotherapy.

Both FcR classes, however, associate with the promiscuous FcR γ chain signaling subunit and are dependent on this FcR γ chain for stable surface expression (15, 16, 17, 18, 19). Consequently, it is conceivable that FcγRI and FcαRI “cross-compete” for the FcR γ chain, which would hinder the possibility of using both receptors in immunotherapy. The aim of this study was, therefore, to determine whether both receptors can function independently, irrespective of (limiting amounts of) the FcR γ chain. Furthermore, we analyzed whether simultaneous engagement of FcαRI and FcγRI on PMN cells facilitates destruction of malignant cells.

Surface expression of FcαRI was detected with FITC-conjugated F(ab′)₂ fragments of CD89 mAb A77 (Medarex, Annandale, NJ) or PE-labeled CD89 mAb A59 (PharMingen, San Diego, CA). FITC-conjugated CD64 mAb 22 (Medarex) was used to determine FcγRI expression. Mouse PMN cells were defined with PE-conjugated Gr-1 (PharMingen).

Fully human IgA1 antibodies targeting Ep-CAM were obtained by using phage display and engineering as described previously (20). BsAb FcγRIxHER-2/neu (22x520C9; MDXH210), FcαRIxHER-2/neu (A77x520C9), FcγRIxHLA-II (22xF3.3), and FcαRIxCD20 (A77x1F-5) were prepared as described in Ref. 21. mAb 520C9 (Medarex) recognizes HER-2/neu, a proto-oncogene product overexpressed on human carcinoma cells. mAb 1F5 and mAb F3.3 are directed against CD20, and MHC class II antigens, respectively, and were a kind gift from Dr. M. Glennie (Tenovus Research Laboratory, Southampton, United Kingdom).

Whole blood of mice was incubated with mAb (10 μg/ml) for 15 min at room temperature and subjected to fluorescence-activated cell sorting lysing solution (Becton Dickinson, San Jose, CA). Human PMN cells (2 × 10⁵), either freshly isolated or cultured overnight, were incubated with 22-FITC and A59-PE for 30 min at 4°C. Cells were analyzed on a FACScan (Becton Dickinson).

Generation of FcγRI (22) and FcαRI (19) Tg mice was described earlier. Briefly, FcγRI Tg mice were generated by injection of an 18-kb human genomic DNA fragment into FVB/N oocytes. A 41-kb cosmid clone served as a Tg construct to create FcαRI Tg mice. Both Tg mice expressed the human receptor solely on myeloid cells, which parallels the human situation. FcγRI Tg males were crossed with FcαRI Tg females to generate (FcαRI × FcγRI) double Tg (dTg) mice. Expression of transgenes was determined by flow cytometry of peripheral blood cells using anti-FcγRI mAb 22-FITC and anti-FcαRI mAb A77-FITC.

To induce FcγRI expression on PMN cells and to increase PMN cell counts in blood, mice were injected s.c. with 1.6 μg/mouse/day murine G-CSF for 4 days (23).

Peripheral blood (heparin anticoagulated) from healthy volunteers was collected and PMN cells were isolated by Ficoll-Histopaque discontinuous gradient centrifugation. PMN cells were collected at the interface between Ficoll and Histopaque and the remaining erythrocytes were removed by hypotonic shock. Both purity of PMN cells and viability checked with trypan blue exceeded 95%.

The breast carcinoma cell line SK-BR-3, overexpressing HER-2/neu and the malignant B-cell line ARH-77 were obtained from the American Type Culture Collection (Manassas, VA). Cells were cultured in RPMI 1640 medium (Life Technologies, Inc., Grand Island, NY) supplemented with 10% FCS and antibiotics. SK-BR-3 cells were harvested using trypsin-EDTA (Life Technologies, Inc., Paisley, United Kingdom).

In culture experiments, human or mouse cells were cultured with human or mouse cytokines, respectively. Human PMN cells were cultured overnight at 37°C with IFN-γ (300 units/ml; Boehringer Mannheim, Mannheim, Germany) to induce FcγRI expression. Mouse bone marrow cells were cultured in DMEM medium, supplemented with 4.5 g/liter glucose, 10% FCS, and antibiotics, with or without granulocyte macrophage colony-stimulating factor (GM-CSF, 50 ng/ml) and tumor necrosis factor α (TNF-α, 50 ng/ml), or IFN-γ (250 units/ml; Amgen, Thousand Oaks, CA). After 24 h, nonadherent cells were harvested and stained with A77-FITC or 22-FITC. Gr-1-PE was used to define mouse PMN. Dr. J. Andresen (Amgen) generously provided GM-CSF and G-CSF. TNF-α was kindly donated by Dr. W. Buurman (University Maastricht, the Netherlands).

⁵¹Cr release assays were used to evaluate the capacity of effector cells to lyse tumor cells (24). Either 1 × 10⁶ SK-BR-3 or 1 × 10⁶ ARH-77 tumor cells were incubated with 150 μCi of ⁵¹Cr (Amersham, Little Chalfont, United Kingdom) for 2 h at 37°C and washed three times.

In “cold target inhibition experiments,” 2.5 × 10^{3 51}Cr-labeled SK-BR-3 cells were plated with 2.5 × 10³ unlabeled ARH-77 cells (or vice versa) in 96-well round-bottomed microtiter plates. Fifty microliters of whole blood of G-CSF-treated dTg mice were added (E:T ratio, 120:1). Alternatively, 50 μl of RPMI 1640 medium containing 3 × 10⁵ (E:T ratio, 60:1), 6 × 10⁵ (E:T ratio, 120:1), or 9 × 10⁵ (E:T ratio,180:1) IFN-γ-treated human PMN cells were added. BsAb-mediated tumor cell lysis of ⁵¹Cr-labeled targets was compared with lysis of the same cells in the presence of a BsAb directed against competing unlabeled cells.

In an additional set of experiments, ⁵¹Cr-labeled SK-BR-3 or ⁵¹Cr-labeled ARH-77 cells (5 × 10³/well) were incubated with effector cells and increasing amounts of anti-FcαRI BsAb, anti-FcγRI BsAb, or both types of BsAb. Cells were incubated at 37°C for 6 h, after which ⁵¹Cr release in supernatants was measured.

Polystyrene tubes were coated with 100 μg/ml human serum IgA (Cappel, Aurora, OH), 100 μg/ml human IgG (CLB, Amsterdam, the Netherlands), or PBS for 3 h at 37°C. After washing three times with PBS, all tubes were blocked with HEPES complete [20 mm HEPES (pH 7.4), 132 mm NaCl, 6 mm KCl, 1 mm MgSO₄, 1.2 mm NaH₂PO₄, 1 mm CaCl₂, 5.5 mm glucose, 0.5% BSA, and 1.5 mm MgCl₂] for 1 h at 37°C. The luminol-enhanced chemiluminescence method was used for determination of real-time respiratory burst activity (25). Human PMN cells (2 × 10⁵/0.2 ml HEPES) were gently centrifuged (400 rpm, 5 min, 4°C) and placed in a 953 LB Biolumat (Berthold, Wildbad, Germany). Luminol (150 mm) was injected in all tubes, and light emission was recorded continuously for 30 min at 37°C.

Results were analyzed by means of the unpaired two-tailed Student’s t test (comparison experiments described in Fig. 6) or ANOVA tests (combination experiments; Fig. 5). Differences in competition experiments were analyzed by Wilcoxon rank sum tests (Fig. 4). Results are expressed as mean ± SE, and significance was accepted at P < 0.05.

Since both FcγRI and FcαRI crucially depend on the FcR γ signaling chain for stable surface expression, we investigated whether these receptors “cross-compete” for the FcR γ chain. We first checked expression of FcγRI and FcαRI under various conditions. Freshly isolated human PMN expressed FcαRI, but not FcγRI (Fig. 1,A, left panel). To induce surface FcγRI, PMN cells were incubated with IFN-γ. FcγRI expression was up-regulated after 24 h (Fig. 1,A, right panel), whereas FcαRI levels were unaffected (Fig. 1 B). Because FcαRI expression on human PMN cells is difficult to manipulate experimentally, we created mice Tg for human FcγRI or human FcαRI. It was previously shown that human FcγRI and FcαRI physically interact with the murine FcR γ chain (17, 19) and these Tg models enabled us to modulate receptor expression separately and in vivo. As described before (19, 22), cell distribution patterns in Tg mice closely parallels the human situation; FcγRI is constitutively expressed on monocytes and macrophages and induced on PMN cells by G-CSF or IFN-γ treatment (22). FcαRI Tg mice constitutively express FcαRI on PMN, whereas expression can be induced on macrophages by culture with GM-CSF and TNF-α (19, 26).

Culturing bone marrow-derived PMN cells of (FcαRI × FcγRI) dTg mice with GM-CSF and TNF-α enhanced FcαRI expression levels (Fig. 2,A, top panels), whereas IFN-γ selectively induced FcγRI expression (Fig. 2,A, bottom panels). Up-regulation of either receptor class, however, did not influence expression levels of the other class. No difference in FcγRI expression was found between FcγRI sTg and dTg PMN cells upon culture with GM-CSF/TNF-α, although the latter PMN cells expressed increased levels of FcαRI (Fig. 2,B, top right panel). Similarly, increased levels of FcγRI after culture with IFN-γ did not affect FcαRI expression levels (Fig. 2 B, bottom left panel).

When mice were injected with G-CSF, FcγRI was up-regulated on dTg PMN, whereas expression of FcαRI was unaltered (Fig. 3,A). Again, enhanced FcγRI expression levels did not affect levels of FcαRI (Fig. 3,B, left panel). In addition, PMN cells from dTg and FcγRI sTg mice showed identical FcγRI expression levels, indicating that FcαRI does not affect expression of FcγRI either (Fig. 3 B, right panel). In summary, FcαRI and FcγRI expression levels on resting and cytokine-stimulated PMN cells are comparable between sTg and dTg mice.

Since both FcαRI and FcγRI require the FcR γ chain for proper function (15, 16, 17, 18, 19), we investigated whether the FcR γ chain preferentially interacts with either one of these receptors. For this purpose, we set up a cold target inhibition assay. Effector cells were incubated with a mixture of ⁵¹Cr-labeled target cells X and unlabeled target cells Y (Fig. 4,A). Lysis of ⁵¹Cr(X), either in the absence (Fig. 4, B–D, open columns) or presence (Fig. 4, B–D, filled columns) of a BsAb targeting the competing unlabeled cell line Y, was measured after incubation for 6 h. IFN-γ-treated human PMN cells (Fig. 4, B and C) or whole blood of G-CSF-treated dTg animals (Fig. 4,D) were used as effector cells. When ⁵¹Cr-labeled ARH-77 was used as target cell line X and the unlabeled SK-BR-3 as Y, neither lysis via FcγRI nor that via FcαRI was influenced when BsAb directed against SK-BR-3 and the competing receptor were present (Fig. 4,B). In the alternate case, with ⁵¹Cr-labeled SK-BR-3 as X and ARH-77 as Y, no differences in FcγRI-mediated tumor cell lysis were observed, either in the absence or presence of BsAb targeting FcαRI and ARH-77 (Fig. 4,C, left panel). In addition, at E:T ratios 120:1 or 180:1 FcαRI-mediated lysis of ⁵¹Cr-labeled SK-BR-3 cell was unaffected in the presence of a BsAb targeting FcγRI (Fig. 4,C, right panel). However, at E:T ratios of 60:1 tumor cell lysis was somewhat reduced. No differences in FcγRI- or FcαRI-mediated lysis of either cell lines were observed when whole blood of dTg mice was used as effector population (Fig. 4 D). In an additional set of experiments in which ⁵¹Cr release was measured after 2 or 4 h of incubation, similar results were found (data not shown, n = 2).

Next, tumor cell kill upon simultaneous engagement of FcαRI and FcγRI was assessed. Maximal lysis of ARH-77 cell was observed with either 1 μg/ml BsAb FcγRIxHLAII or FcαRIxCD20, which was not increased in the presence of higher BsAb concentrations. Tumor cell lysis was, however, enhanced upon incubation with two targeting BsAb, relative to either one of them separately (Fig. 5,A). Comparable data were obtained with whole blood of G-CSF-treated dTg mice (data not shown, n = 3). We, furthermore, investigated whether the observed reduction in FcαRI-mediated tumor cell lysis upon engagement of FcγRI at E:T ratios of 60:1 would abrogate this enhancement in tumor cell lysis. Since BsAb FcαRIxCD20 did not induce tumor cell lysis at the E:T ratio 60:1 (data not shown), SK-BR-3 cells expressing both HER-2/neu and Ep-CAM were used. Because an IgA Ab targeting Ep-CAM was available (20), BsAb FcγRIxHER-2/neu and IgA anti-Ep-CAM Ab were used in these experiments. Maximal lysis of SK-BR-3 tumor cells was observed in the presence of 0.4 μg/ml FcγRIxHER-2/neu or 2.0 μg/ml IgA anti-Ep-CAM, and again enhanced tumor cell lysis was observed upon addition of two (bispecific) Abs, relative to addition of only one (Fig. 5 B).

Notably, FcαRI proved more efficient in triggering tumor cell kill than FcγRI (Fig. 6, A and B). Therefore, the capacity of FcαRI and FcγRI to initiate PMN cell signaling was investigated. Cross-linking of FcαRI resulted in a more rapid induction of rises in intracellular free calcium levels than cross-linking of FcγRI (Fig. 6,C). Initiation of respiratory burst activity was tested as a more distal signaling event using a sensitive chemiluminescence method. No PMN oxygen radical production was evoked when PBS/HEPES-coated tubes were used. Tubes coated with either IgA or IgG activated the PMN NADPH-oxidase complex, but respiratory burst activity was consistently higher with IgA-coated tubes (Fig. 6 D). This was observed with a range of IgA and IgG concentrations (data not shown, n = 2).

In this study, we aimed to evaluate whether simultaneous targeting of tumor cells to two types of trigger molecules on PMN cells results in enhanced tumor cell destruction. As trigger molecules, we chose FcγRI and FcαRI, because both are selectively expressed on myeloid effector cells which can easily be mobilized in vivo and potently trigger tumor cell lysis in vitro(2, 9, 10, 27). Moreover, treatment with a combination of G-CSF and BsAb, targeting FcγRI and idiotype, led to effective antitumor responses in lymphoma-bearing FcγRI Tg mice, which were not observed in control mice (including treated nontransgenic litter mates) (28). Similarly, treatment of FcαRI Tg mice with an anti-FcαRI BsAb resulted in prolonged survival, compared with control mice (29).

A possible drawback of therapies engaging both FcαRI and FcγRI, however, may be the (limiting) amount of FcR γ chain in effector cells restricting simultaneous triggering via two FcR γ chain-dependent receptors (15, 16, 17, 18, 19). In the present work, however, no evidence was found that the FcR γ chain limits expression of FcαRI or FcγRI, neither in vitro nor in vivo. Up-regulation of FcγRI did not result in decreased expression of FcαRI and vice versa (Figs. 1,2,3). Furthermore, cold target inhibition experiments showed that in most circumstances both receptors function independently of each other without any cross-competition for the FcR γ chain. Since BsAb targeting either HLA class II or HER-2/neu induced efficient tumor cell lysis, it is feasible that these BsAb induce maximal occupancy of receptors and associating FcR γ chain. In one situation, where FcαRIxHER-2/neu and FcγRIxHLAII BsAb were used at an E:T ratio 60:1, FcαRI-mediated tumor cell lysis was somewhat reduced upon engagement of FcγRI, suggesting competition for the FcR γ chain. No FcR γ chain competition was observed when FcγRIxHER-2/neu and FcαRIxCD20 BsAb were added, which is likely attributable to less efficient tumor cell lysis via BsAb targeting CD20 than HLA class II. This indicates that FcR γ chain cross-competition may occur in situations where both classes of receptors are maximally engaged, while numbers of effector cells are limited. However, at E:T ratios of 60:1 simultaneous triggering of both receptor classes still enhanced tumor cell killing (Fig. 5 B), suggesting that even under these conditions only minimal FcR γ chain competition occurs.

Notably, earlier data in mast cells supported strong competition between FcεRI and FcRγIIIa for the FcR γ chain (30, 31). It may, thus, be possible that differences exist between FcR γ chain levels in PMN cells versus mast cells or that expression of FcεRI and FcγRIIIa is more strictly regulated than expression of FcαRI and FcγRI. Additionally, in mast cells, FcγRIIIa and FcεRI exist as multisubunit complexes consisting of FcR β chain and FcR γ signaling units (32). Rather than FcR γ chain, the FcR β chain might, therefore, be limiting.

Since FcαRI was capable of triggering early and late signaling events more potently than FcγRI (Fig. 6; Refs. 33 and 34), it is conceivable that different signaling pathways are initiated upon either FcαRI or FcγRI cross-linking. Simultaneous engagement of both pathways may, therefore, amplify effector functions. The higher capacity of FcαRI to initiate PMN cell activation might well be attributable to the presence of the positively charged amino acid (Arg²⁰⁹) in the transmembrane region of FcαRI (18). Because the FcR γ signaling chain bears a negatively charged amino acid in its transmembrane region, we hypothesize that this results in a stronger association of the FcR γ chain with FcαRI than with FcγRI (which lacks such a positively charged amino acid) (15). Alternatively, it is possible that the increased number of tumor antigens bound by effector cells resulted in enhanced tumor cell lysis.

Whereas BsAb targeting FcγRI and CD20 were reported unable to initiate Ab-dependent cellular cytotoxicity, tumor cells were lysed in the presence of BsAb targeting FcαRI and CD20 (35). Also, in our experiments, FcαRI proved more efficient in initiating tumor cell killing. An attractive feature of FcγRI as target molecule, however, is the ability of this receptor to induce a “vaccine” effect. FcγRI was shown to initiate efficient antigen presentation in vitro and in vivo(22, 36), and recently a unique motif for antigen presentation has been identified in its cytoplasmic tail (37). We, therefore, speculate that targeting to FcγRI might induce a memory response to recirculating tumor cells. Indeed, FcγRI Tg mice with lymphoproliferative disease injected with BsAb (targeting FcγRI) were not only cured, but also protected against tumor rechallenge (28).

In conclusion, this study documents the FcR γ chain not to be limiting for either expression or function of FcαRI or FcγRI on PMN cells. Because of this, FcαRI and FcγRI can be simultaneously engaged for induction of cell lysis, resulting in improved tumor cell killing. Importantly, it was shown that IgA1 and IgG1 anti-Ep-CAM Abs do not synergize (20). This has been attributed to binding of IgG1 anti-Ep-CAM Abs to inhibitory FcγRIIb (CD32) receptors on PMN cells, which would inhibit rather than enhance FcαRI-mediated killing. The usage of BsAb, selectively targeting to activatory PMN FcR, would overcome this problem and may thus be a prerequisite for combined treatment. Moreover, whereas FcαRI was shown to be more active in killing malignant cells, FcγRI might more potently induce a vaccine response. Immunotherapy, involving combined engagement of FcγRI and FcαRI, may, therefore, constitute an attractive option for the treatment of malignant disorders.

We thank Toon Hesp for excellent animal care and Dr. H. van Ojik for critical reading of this manuscript.