In addition to the reduced folate carrier, the folate receptor (FR) can facilitate the cellular uptake of folic acid, a water-soluble vitamin required in the biosynthesis of nucleotide bases and other essential natural products.1 Whereas normal cells rely almost exclusively on the reduced folate carrier (KM for folate ∼ 10−5 M) to supply their folate needs, solid tumors of epithelial origin2, 3, 4 and hematopoietic cancers of the myeloid lineage5 frequently overexpress FR (KD for folate ∼ 10−10 M) on their cell surfaces, perhaps to enhance their capacity to compete for the vitamin. This upregulation of the high-affinity FR on cancer cell surfaces has rendered the receptor an attractive target for cancer-specific interventions.6, 7, 8 Thus, not only have imaging and therapeutic agents been targeted to FR-expressing cells using anti-FR-specific monoclonal antibodies,9, 10, 11 but direct conjugation of drugs to folic acid has been documented to lead to selective accumulation of the attached pharmaceuticals in the tumor tissue.12, 13 Importantly, the high affinity of free folic acid for the FR is generally not compromised by linking a pharmaceutical to the vitamin, thereby allowing the synthesis of low-molecular-weight drugs that can localize selectively to cancer cells in vivo.
Interest in developing folate-targeted cancer therapeutics has recently grown as estimates of the incidence and magnitude of FR upregulation by malignant cells have increased. Thus, whereas early measurements of FR expression in primary tumor masses suggested that only ∼ 1/3 of human cancers might upregulate the receptor,2, 3 more recent comparisons of primary and metastatic tumor tissues demonstrate that even when FR is absent from the primary mass, it may be abundantly expressed in the metastatic disease (personal observations). Consistent with this observation, higher-grade and later-stage ovarian cancers have been reported to express more FR on their cell surfaces than their earlier-stage counterparts.3 Further, cancers that are resistant to standard hormone therapy14 or chemotherapy15 may also express higher levels of FR than more responsive tumors of the same type. Taken together, these results suggest that cancers that may be difficult to treat by classical methods may also be especially targetable with folate-linked drugs.
One class of emerging therapies for treatment of otherwise refractory cancers has involved the mobilization of the immune system first to recognize and then to attack the malignancy. Thus, in order for cancer cells to survive and proliferate in vivo, they must somehow escape normal immune surveillance.16 Whereas many tumors may express clearly aberrant proteins, other tumor-associated antigens are only weakly immunogenic, largely because they are predominantly self-proteins that are simply expressed more abundantly in cancer tissues. In order to render tumors more immunogenic, we have covalently linked foreign haptens to folic acid and delivered the resulting folate-hapten conjugates to tumor cell surfaces, exploiting the high affinity of the conjugate for FR as a means of concentrating the hapten in the tumor mass. Our hypothesis has been that folate-linked haptens would act as high-affinity bridges between tumor cells and antihapten antibodies, concentrating the autologous IgG onto the tumor cell surface and thereby marking it for destruction by antibody-dependent processes. Using syngeneic tumor models in immune competent mice, we have previously demonstrated that folate targeting of the hapten, fluorescein (FITC), does in fact lead to complete elimination of established FR-positive tumors in fluorescein-immunized mice if the immune system is simultaneously boosted with low to moderate levels of IL-2 and IFN-α.17 Although subsequent work has shown the same therapy to be effective in a second tumor model,18 and even though the treatment is characterized by little detectable toxicity, unfortunately little research has been conducted to determine the mechanism responsible for tumor regression. Therefore, the purpose of the present article was to investigate both the molecular and cellular mechanisms that underlie the folate-targeted immunotherapy. We demonstrate here that folate-hapten conjugates must bridge between antihapten antibodies and FRs on cancer cell surfaces to achieve therapeutic efficacy, and that this antibody binding together with the active participation of NK cells, macrophages, CD4+ T cells and CD8+ T cells can lead to not only complete tumor regression, but also development of a cellular immunity that leads to long-term protection against subsequent challenge with the same cancer cells.
Folate-(γ)-ethyleneamine-fluorescein (folate-FITC) conjugate was supplied by Endocyte (West Lafayette, IN). BSA, KLH, mouse IgG (Balb/c mice), carrageenan, silica and (5-)aminofluorescein (single isomer) were purchased from Sigma Chemical (St Louis, MO). FITC (isomer I) was purchased from Molecular Probes (Eugene, OR). TiterMax Gold adjuvant was obtained from CytRx (Norcross, GA) and GPI-0100 (a saponin-based adjuvant) was obtained from Galenica (Birmingham, AL). TriColor-macrophage (F4/80) monoclonal antibody (mAb) was purchased from Caltag Laboratories (Burlingame, CA), human recombinant IL-2 was from PeproTech (Rocky Hill, NJ) as a lyophilized powder with no additives and human recombinant IFN-α A/D was obtained from PBL Biomedical Laboratories (Piscataway, NJ). For in vivo use, IL-2 and IFN-α, alone or in combination, were prepared in sterile phosphate-buffered saline (PBS, pH 7.4) containing 1% syngeneic serum as described elsewhere.17 Polyclonal rabbit antiasialo GM1 antibody was purchased from Wako Pure Chemicals USA (Richmond, VA), and anti-CD4+ and anti-CD8+ T-cell mAbs were generated as ascites in pristane-treated nude mice by injecting 5–14 × 106 cells/animal of ATCC hybridomas GK1.5 and 2.43, respectively (QED Bioscience, San Diego, CA).
Animals, cell lines and tumor models
Female Balb/c mice, 6–8 weeks of age, were purchased from Harlan Sprague Dawley (Indianapolis, IN) and maintained under pathogen-free conditions until they were 6–8 weeks old. FR-positive M109 tumors were maintained in Balb/c mice and regenerated as described previously.17 M109 cells were cultured in folate-deficient RPMI-1640 medium (Gibco-BRL) supplemented with 10% v/v heat-inactivated fetal calf serum (HIFCS), 100 units/ml penicillin and 100 μg/ml streptomycin (FDRPMI). For intraperitoneal implantation, 5 × 105 viable early passage tumor cells were suspended in 400 μl of folate-deficient RPMI-1640 supplemented with 1% syngeneic mouse serum (antibiotic-free) and injected into the peritoneal cavity. For subcutaneous implantation, 1 × 106 tumor cells in 100 μl of the above medium were injected under the skin in the shoulder region. All animal experiments were carried out in accordance with procedures approved by the Purdue Animal Care and Use Committee.
To induce anti-FITC antibodies in mice, BSA- or KLH-conjugated FITC were prepared as described previously and used as immunogens.17 When TiterMax Gold was used as adjuvant, mice were immunized twice against a total of 90–100 μg BSA-FITC or KLH-FITC emulsified in TiterMax Gold. When GPI-0100 was used as adjuvant, mice were immunized 3 times with 35 μg KLH-FITC admixed with 100 μg GPI-0100. Pooled FITC-antiserum from KLH-FITC/GPI-0100-immunized mice was routinely collected 1 week after the final vaccination and used for in vitro assays. The effect of adjuvant composition on anti-FITC antibody production and isotype distribution in mice has been published in a recent review.18
Purification of total serum IgG from FITC-immunized mice
Total serum IgG from KLH-FITC/GPI-0100-immunized animals (hereafter referred to as anti-FITC immune IgG) was isolated using an ImmunePure IgG (Protein A) purification kit (Pierce, Rockford, IL) following the manufacturer's recommendations. Semiquantitative ELISA analysis was performed according to previously described procedures17 and used to measure the FITC-specific isotypes (IgG1, IgG2a, IgG2b, IgG3) in the IgG preparation at approximate ratios of 1:0.65:0.24:0.19, respectively.
Isolation of murine effector cells
NK cells were prepared from splenocytes using a nylon wool fiber column (Polyscience, Warrington, PA) and then cultured with 6000 units/ml of IL-2 in RPMI-1640 medium (Gibco-BRL) supplemented with 10% v/v HIFCS, 15 mM HEPES, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 50 μM 2-mercaptoethanol, 100 units/ml penicillin and 100 μg/ml streptomycin.19 The adherent NK cells harvested 6 days later are reported to be almost exclusively asialo-GM1+ with less than 2% containing T-cell markers.20 Flow cytometric analysis indicated that the adherent effector cells we routinely isolated for antibody-dependent cellular cytotoxicity (ADCC) assays were ∼ 95% asialo GM1+. No CD8+ T cells and F4/80+ macrophages were detected (data not shown). Murine macrophages were obtained from thioglycolate-elicited peritoneal exudate cells. Briefly, mice were injected intraperitoneally with 1 ml of 0.5% (w/v) thioglycolate (Difco Laboratories, Detroit, MI). Three days later, peritoneal macrophages were isolated and suspended at 5 × 105 cells/ml of RPMI-1640 medium containing 10% v/v HIFCS.
Complement-dependent cytotoxicity (CDC) assay
M109 tumor cells were incubated with 200 nM folate-FITC in FDRPMI for 30 min at 37°C and the unbound ligands were removed by washing 3 times with ice-cold fresh medium. Protein A-purified anti-FITC immune IgG (10 μg/ml) or control IgG (10 μg/ml) was added to the folate-FITC-labeled tumor cells and the incubation was continued for 1 hr. After removing unbound IgG, mouse complement (Rockland, Gilbertsville, PA), restored according to the manufacturer's instructions, was diluted 2-, 5- and 10-fold with RPMI-1640 medium supplemented with 1% BSA. After 1-hr incubation at 37°C, cell lysis was determined by propidium iodide (PI) staining of the cells (20 μg/ml) and analyzed by flow cytometry. Tumor cell lysis was calculated according to the following formula: 100 × (% viable cells with control IgG − % viable cells with immune IgG)/(% viable cells with control IgG).
Specific lysis of FR-positive M109 tumor cells was determined by 51Cr-release assay using isolated murine NK cells or macrophages as effector cells. Briefly, 51Cr-labeled M109 cells (1 × 107 cells) were incubated with 1–200 nM folate-FITC for 1 hr at 4°C and unbound ligands were removed with ice-cold fresh medium. The cells were then suspended at 1 × 105/ml, and 100 μl of this cell suspension was added to each well of a round-bottom 96-well plate with 50 μl of anti-FITC immune IgG or control IgG (Sigma) at 1–100 μg/ml. One hundred μl of various concentrations of murine effector cells were then added to each well at effector-to-target (E/T) ratios of 50:1, 25:1, 12.5:1 and 6.25:1, and cytotoxic interactions were allowed to proceed for 4 hr at 37°C. As controls, aliquots of 51Cr-labeled M109 cells were subjected to the above ADCC assay in the absence of folate-FITC or in the presence of folate-FITC supplemented with a vast excess of free folic acid (0.2 mM) to block folate-FITC binding to the cancer cell surfaces. To examine the bimodal effect of folate-FITC on ADCC activity, the murine immune IgG were precomplexed with folate-FITC before adding to 51Cr-labeled M109 cells pretreated with 200 nM folate-FITC, a concentration that guarantees all cell surface FRs will be occupied. All assays were performed in triplicate, and target cell lysis was determined by counting 100 μL of supernatant in a gamma scintillation counter after removal of intact cells by centrifugation. Spontaneous and total lyses were determined by incubating radiolabeled target cells in the absence of effector cells or by adding 1% Tween 100 (v/v) to the radiolabeled target cells, respectively. The percentage of cell lysis was calculated by the following formula: lysis (%) = 100 × (experimental release − spontaneous release)/(total lysis release − spontaneous release).
Thioglycolate-elicited macrophages, isolated as described above, were allowed to adhere by incubation of 1.5 × 106 cells in 2 ml RPMI-1640 medium containing 10% HIFCS for 4 hr at 37°C in each well of a 6-well tissue culture plate. Nonadherent cells were removed by 2 washes with fresh medium. The adherent macrophages were at least 95% pure based on staining with TriColor-F4/80 mAb (data not shown). Meanwhile, M109 tumor cells were incubated with 200 nM folate-FITC for 30 min at 37°C, washed 3 times and incubated with preimmune serum or FITC-antiserum at 1:4 dilutions for 1 hr at 37°C. After 3 washes, the cells were resuspended at a concentration of 5 × 106/ml in FDRPMI. One ml of the opsonized tumor cells was then added to each well containing the adherent macrophages and incubated for 30 min at 37°C, followed by 3 more washes with PBS containing 1% BSA. TriColor-F4/80 mAb was then added to each well (0.4 μg/ml) and cells were incubated for another 30 min at 37°C to allow labeling of the macrophages. The cells were then subjected to a final wash, recovered from the plate and analyzed by dual-color flow cytometry.
To study the dynamics of phagocytosis of folate-FITC-labeled tumor cells by macrophages, thioglycolate-elicited macrophages were added to glass coverslips and allowed to adhere for 3 hr. M109 tumor cells, pretreated with folate-FITC and then FITC-antiserum (as described above), were added to the adherent macrophages and incubated for 5, 15, 30 and 90 min at 37°C. After the incubation, TriColor F4/80 mAb (0.4 μg/ml) was added to the cell mixture to label the macrophages (30 min, 37°C). Cells were then washed, fixed with 1% formaldehyde solution and imaged for cell-associated FITC and TriColor fluorescence with a Bio-Rad MRC-1024 UV/Vis Confocal Laser Scanning Microscopy system equipped with a Nikon Diaphot 300 inverted microscope, 488 nm and 568 nm Argon lasers, a 60× Fluor Phase objective, a 522/35 PMT filter for FITC and a 585 long-pass filter for TriColor.
To reduce mouse serum folate levels to serum concentrations characteristic of humans (∼ 20 nM), mice were maintained on a folate-deficient diet for approximately 3 weeks prior to any folate-FITC treatment and then returned to a high-folate diet 2–3 days after the last treatment. This dietary regimen lowers the serum folate concentration from a level that is artificially elevated by the large doses of folic acid in commercial rodent chows to a level (∼ 25 nM) that approaches the concentrations in humans and wild mice. All drug treatments were administered intraperitoneally and the therapeutic efficacies were evaluated based on mouse survival or tumor volume.
Treatment regimen 1.
Mice immunized with BSA-FITC in TiterMax Gold adjuvant bearing intraperitoneal M109 tumors were injected with either PBS or 1,500 nmol/kg folate-FITC on days 4 and 7 after tumor cell implantation. This was followed by injection of 5,000 units/day IL-2 on days 8–12.
Treatment regimen 2.
Mice immunized (3 × 2-week intervals) with KLH-FITC in GPI-0100 adjuvant bearing intraperitoneal M109 tumors were treated with folate-FITC (1,500 nmol/kg) in combination with IL-2 (5,000 units/day) and IFN-α (25,000 units/day) on days 7, 8, 9, 11 and 14 after tumor cell implantation.
Treatment regimen 3.
Mice immunized with KLH-FITC in TiterMax Gold adjuvant bearing intraperitoneal M109 tumors were treated with folate-FITC (60–6,000 nmol/kg) in combination with IL-2 and IFN-α, as described in regimen 2.
Treatment regimen 4.
KLH-FITC/GPI-0100 immunized mice (3 × 1-week intervals) were implanted subcutaneously with M109 cells 3 days after the third immunization. After allowing the tumor to establish in vivo for 7 days (≤ 50 mm3), the mice were given a 3-week treatment of 1,800 nmol/kg folate-FITC (5 times/week), 40,000 units/day IL-2 (5 times/week) and 25,000 units/day IFN-α (3 times/week). Tumor dimensions were measured twice a week using a caliper, and tumor volumes were calculated by the following formula: 0.5 × a × b2, where a is the longest axis across the tumor and b is the shorter axis perpendicular to a.
Tumor rechallenge analyses.
Mice that had been recently cured of established M109 tumors using the standard folate-hapten immunotherapy17 were reimplanted with fresh M109 cells after first depleting their CD4+ T cells, CD8+ T cells, or both using the appropriate depleting mAbs described below.
Antiasialo GM1 antibody treatment
In treatment regimen 1, a group of mice was injected with antiasialo GM1 antibody to induce depletion of NK cells according to a published protocol.21 Briefly, the lyophilized product of polyclonal antiasialo GM1 antibody (Wako Pure Chemicals) was reconstituted in 1 ml double-deionized water and used as the stock solution; 20 μl of this stock solution was diluted 10 times with PBS and injected intraperitoneally into Balb/c mice to induce removal of NK cells.
Depletion of T cells and macrophages in vivo
In treatment regimen 2, depletion of CD4+ or CD8+ T-cell populations was accomplished by intraperitoneal injection of 20 μl anti-CD4 (clone GK1.5) or 5 μl anti-CD8 (clone 2.43) mAb ascite preparations diluted 10 times with PBS. The effectiveness of a particular depletion schedule was evaluated by FACS staining of splenocytes in a separate set of mice, and these analyses indicated that the samples were 98–99% free of CD4+ and CD8+ T cells (data not shown). For macrophage depletion, both carrageenan22 and silica23 were injected intraperitoneally in the same treatment regimen. Briefly, carrageenan (type IV λ from Sigma) was dissolved in sterile PBS, pH 7.4, at 5 mg/ml (heated to 56°C for complete solubilization); 200 μl of this solution was injected into each mouse 2 days prior to folate-FITC treatment (day 5). Silica (Sigma) was dissolved in 5% dextrose, sonicated prior to use and injected at a concentration of 1 g/kg in a 400 μl volume on days 5 and 12. Phagocytosis of either carrageenan or silica leads to macrophage inactivation.22, 23
Spleens were aseptically harvested from both naive mice and mice with protective memory that was developed against M109 tumor cells following a curative immunotherapy (similar to treatment regimen 4). Splenocytes were pooled to enrich CD8+ T cells by negative selection using the SpinSep separation system following the manufacturer's instructions (StemCell Technologies, Vancouver, Canada). The resulting cell populations containing ∼ 87% CD8+ T cells were used as effector cells. To measure tumor-specific cytotoxicity, short-term (6 hr) and long-term (18 hr) 51Cr-release assays were performed at 37°C in complete DMEM (Gibco-BRL) supplemented with 10% v/v HIFCS, 100 units/ml penicillin and 100 μg/ml streptomycin against 51Cr-labeled M109 cell targets using E/T ratios of 50:1, 25:1 and 12.5:1. The percent cytotoxicity was calculated as described for the ADCC assay.
The Prism (GraphPad Software, San Diego, CA) computer program was used for statistical analyses. Student's t-test was used for in vitro data analysis. Differences in animal survivals or tumor volumes between treatment groups were analyzed using a log-rank or ANOVA test, respectively. A difference was considered statistically significant when p ≤ 0.05.
Bimodal dependence of therapeutic efficacy on folate-hapten concentration
We have previously demonstrated that targeting of FR-positive cancer cells with folate-FITC conjugates leads to regression of established tumors in FITC-immunized mice.17 Although data were presented to suggest that folate-FITC could promote anti-FITC IgG binding to cancer cells in vitro, direct evidence that folate-FITC serves a bridging function to promote attachment of autologous anti-FITC IgG to FR on cancer cell surfaces in vivo was still lacking. To explore this hypothesis, the dependence of therapeutic efficacy on folate-FITC concentration was evaluated. Thus, theory would predict that little efficacy should be observed at low folate-FITC concentrations (due to an insufficient number of folate-FITC bridges), and also at high folate-FITC (due to saturation of both ligands for the bridge, i.e., IgG and FR), while at intermediate folate-FITC concentrations, where simultaneous occupancy of a cell surface FR and a soluble IgG by the same folate-FITC would allow bridging to occur, high therapeutic activity would be expected. As seen in Figure 1, therapeutic efficacy does in fact rise to a maximum and decline again to a minimum in the anticipated bimodal fashion. Such behavior would not have been expected if folate-FITC were directly cytotoxic, or if the bispecific ligand needed to bind only one of its bridging partners for cytotoxicity (e.g., by stimulating the FR). However, a bimodal response curve would be predicted if folate-FITC had to bridge between an antibody and the cancer cell surface for efficacy. It is important to point out that immunized mice subjected to the highest dose of folate-FITC (6,000 nmol/kg; 5 doses) displayed no signs of toxicity based on general appearance, body weight, or histopathology (data not shown). In fact, a follow-up GLP toxicology study was conducted in guinea pigs where folate-FITC was given 5 days/week for 3 consecutive weeks. There were no significant toxicologic findings at 12.7 mg/kg, a dose equivalent to 39,350 nmol/kg in mice based on body surface area conversion. It is also notable that any changes in the tumor burden at the time of therapy (Fig. 1) would not have affected the folate-FITC distribution in vivo, since tumor uptake only amounts to a small fraction of the total injected dose based on pharmacokinetic biodistribution studies with 3H-folate-FITC (data not shown). Due to the apparent lack of toxicity of folate-FITC, we conclude that the observed therapeutic efficacy depends on the bivalent bridging function of folate-FITC.
Folate-hapten conjugates mediate ADCC and antibody-dependent phagocytosis of FR-positive tumor cells in vitro
Hypothetically, antibody-coated tumor cells could be eliminated in vivo by CDC, ADCC, phagocytosis, or a combination of the above.24, 25, 26 Solid tumors, however, often display resistance to antibody-triggered CDC due to expression of complement regulatory proteins on their surfaces27 and/or secretion of soluble complement inhibitors.28 For a preliminary evaluation of the role of CDC in our hapten-targeted immunotherapy, we examined the cytotoxicity of mouse complement against folate-FITC-labeled cancer cells in the presence of anti-FITC immune IgG in vitro. Although intense IgG binding could be demonstrated upon addition of the immune IgG to the folate-FITC marked cells, no complement-mediated lysis of the tumor cells could be detected even at 10–50 vol % complement concentration (data not shown). This suggests that some resistance mechanism/incompatibility must exist to prevent complement-mediated lysis of the cancer cells in vitro.
To assess the potential role of ADCC in tumor cell killing, we evaluated the cytotoxicity of NK cells and macrophages toward the folate-FITC/antibody-marked cancer cells using a standard 51Cr-release assay. The cell lysis assay revealed that only NK cells (not macrophages; data not shown) caused direct lysis of the folate-FITC-labeled M109 cells in the presence of anti-FITC antibody (Fig. 2a; p < 0.05 for all antibody concentrations and E/T ratios compared to effector cells alone). This NK-mediated cytotoxicity was, importantly, FR-specific, since it was reduced to background levels by competition with excess folic acid, which blocks folate-FITC binding to FR on the cancer cell surfaces (Fig. 2b; p = 0.004). Since a bimodal therapeutic response of folate-FITC was seen in vivo, we looked for a similar effect on NK-mediated ADCC in vitro by varying the fractional saturation of cultured tumor cell FR with folate-FITC. As shown in Figure 3, a dose-dependent increase in ADCC with increasing folate-FITC was observed until both cancer cell FRs and anti-FITC immune IgG were presaturated with folate-FITC. The specific lysis of folate-FITC-treated tumor cells in the presence of anti-FITC IgG was significantly reduced compared to the results obtained after presaturating the antibody with folate-FITC (p = 0.001). This observation appears to be analogous to the situation in vivo, where sufficient folate-FITC is added to saturate both cancer cell surface FR and anti-FITC IgG simultaneously.
Because macrophages did not appear to be directly involved in the antibody-dependent cell lysis, we examined the ability of autologous macrophages to engulf folate-FITC-bound tumor cells opsonized with FITC-antiserum. As shown in the dual-color FACS analysis of Figure 4, after only 30 min of coincubation, ∼ 34% of the macrophages (identified by F4/80 mAb) had either bound or ingested opsonized tumor cells, whereas tumor cells that were not allowed to bind anti-FITC IgG were only marginally affected (6.4%). To clarify the type of association that mediates macrophage docking with opsonized cancer cells, confocal fluorescence images were taken at different time points following mixing of the cancer cells with the macrophages, and sites of cell-cell interaction were analyzed. As shown in Figure 5, initial docking of macrophages to tumor cells occurred almost immediately (∼ 5 min) upon mixing, with the area of interfacial contact covering about one-fifth of the average tumor cell surface. By 15–30 min postmixing, significant phagocytosis/pitting of the cancer cells was observed, as evidenced by the presence of green (FITC) fluorescent particles inside the red macrophages. In most cases, macrophages were not able to engulf an entire tumor cell completely, but tumor cell death still appeared to occur, as indicated by the fragmentation of the cancer cells and the accumulation of their fluorescent particles within the macrophages. By 90 min postmixing, few intact cancer cells remained in the cell culture dish, and it appeared that nearly all the folate-FITC/anti-FITC antibody-coated tumor cells had been internalized as the result of FcR-mediated phagocytosis. It might be argued that some of the tumor cells had been stripped of their folate-FITC/anti-FITC immune complexes via FcR-mediated internalization, but since the few remaining cancer cells were still fluorescent, we do not consider this interpretation likely. Taken together, these results suggest that both NK cells and macrophages can participate in the killing and/or clearance of folate-FITC/anti-FITC antibody-marked tumor cells and that the NK-mediated ADCC can be reduced if excess folate-FITC is allowed to block all binding sites on both the antibody and tumor cell surface.
Requirement of NK cells for antitumor activity in vivo
Since NK cells were found to contribute to antibody-dependent tumor cell lysis in vitro (ADCC; Figs. 2 and 3), we decided to explore the effect of NK cell depletion on the survival of tumor-bearing mice treated with the hapten-targeted immunotherapy in vivo. For this purpose, mice were implanted intraperitoneally with M109 tumor cells and treated on days 4 and 7 with a noncuring dose of the hapten-targeted immunotherapy. A suboptimal dosing frequency of folate-FITC supplemented with low doses of IL-2 (5,000 units/day) was specifically selected to ensure the eventual death of all mice from cancer. This would enable us to detect either an improvement or decline in mouse lifespan upon NK cell depletion. For NK cell removal, a regimen of 4 intraperitoneal injections of a polyclonal antiasialo-GM1 antibody (which recognizes an antigen found predominantly on mouse NK cells29) was performed during the course of the folate-FITC/IL-2 treatment. As shown in Figure 6, elimination of the NK cells completely abolished any tumor-suppressive effect of folate-FITC plus IL-2 immunotherapy (p < 0.0001). Whereas normal tumor-bearing mice treated with the suboptimal immunotherapy survived a median of 42 days, NK cell-depleted mice subjected to the same therapy survived only ∼ 19 days, a lifespan similar to untreated tumor-bearing mice. It is worth mentioning that IL-2 alone under this condition (5,000 units/day × 5 days) has no antitumor activity as demonstrated previously.17 These data support a prominent role for NK cells in the antitumor activity of the hapten-targeted immunotherapy. However, it should also be noted that other populations of cells may also express some asialo-GM1+, such as activated macrophages30 and a subpopulation of cytotoxic T cells,31 and elimination of these cells might also have contributed to the loss of therapeutic efficacy seen in Figure 6.
Requirement of CD4+ T cells, CD8+ T cells and macrophages for the hapten-targeted immunotherapy
Because of the prominent impact of NK cells on the folate-FITC/IL-2-mediated immunotherapy in vivo, it seemed prudent to evaluate the potential contributions of other immune cells such as macrophages, CD4+ and CD8+ T cells to the same hapten-targeted immunotherapy. Here we supplemented folate-FITC with both IL-2 (5,000 units/day) and IFN-α (25,000 units/dose), because these 2 cytokines have been demonstrated to synergize with our folate-hapten targeted immunotherapy,17 and because IFN-α has been shown to drive Th1 immune responses that often result in antigen-specific adaptive cellular immunity.32 Once again, the strategy for clarifying the roles of these immune cells in vivo was to deplete/inactivate each cell type in tumor-bearing mice and evaluate the impact of this immune cell depletion on the survival of the immunotherapy-treated mice. Anti-CD4+ (T helper) and anti-CD8+ (cytotoxic T cell) mAbs were administered to FITC-immunized mice during the folate-targeted immunotherapy to deplete CD4+ and CD8+ T cells, respectively, while macrophages were inactivated by administration of either carrageenan22 or particulate silica,23i.e., modulators that block the cytotoxic/phagocytic activity of the macrophages. As shown in Figures 7 and 8, depletion of any of these immune cells significantly reduced or even abolished the antitumor activity (anti-CD8+, p = 0.0006 for Fig. 7a and 0.028 for Fig. 7b; anti-CD4+, p = 0.006; carrageenan, p = 0.009; silica, p = 0.010). In fact, within experimental error, immunotherapy in the absence of CD4+ T cells, CD8+ T cells, or macrophages was no more effective than nontargeted therapy (treatment of immunized animals with nontargeted aminofluorescein), hapten-targeted therapy in nonimmunized mice, or no therapy at all (see controls in Fig. 7a).
To establish the role of CD4+ and CD8+ T cells in the immunotherapy, we investigated the role of effector T cells in mice that had already developed an antitumor memory response as a consequence of a previous curative immunotherapy protocol.17 By depleting T-cell subsets during tumor rechallenge with mAbs against CD4+ and CD8+ T cells (Fig. 9), we found that depletion of CD4+ or CD8+ T cells alone was largely ineffective in reversing the protective effect, but removal of both T-cell subsets simultaneously effectively abolished the antitumor immunity in almost all mice (double depletion vs. single depletion: anti-CD4+, p = 0.006; anti-CD8+, p = 0.002). Since the mere presence of hapten with cytokine (IL-2/IFN-α) was ineffective without folate targeting or without the presence of antihapten antibody (Fig. 7a), these data suggest that a combination of immune cells is probably required to achieve both complete tumor regression and the long-term memory leading to subsequent immunity against the same type of cancer.
Ex vivo evidence of tumor-specific T-cell memory
In order to identify tumor-reactive memory T cells, CD8+ T cells were enriched from splenocytes from naive mice or mice with immune memory against M109 tumors following a successful immunotherapy (see Material and Methods). These CD8+ T-cell populations were used as effector cells in 51Cr-release assays against fresh M109 cell targets at various effector-to-target ratios. As shown in Figure 10, significantly higher cytolytic activity was found in the spleens of mice that had been previously cured of the same cancer using folate-targeted immunotherapy than in the naive animals (p = 0.004 for E/T 50:1; p < 0.05 for E/T 25:1 and E/T 12.5:1). Interestingly, the tumor-reactive memory T cells displayed higher activity against M109 cells after a relatively long-term (18 hr) rather than short-term (6 hr) incubation. Such delayed cytotoxicity may be indicative of the specific effector function mediated by the memory T cells, in that rapid killing would have suggested cell cytoxicity mediated by release of perforin and granzymes, whereas the delayed killing might suggest killing primarily via activation of the death receptor (Fas).33 Notably, Fas expression has been detected on M109 cells with anti-Fas mAb (data not shown). Further, cytokines produced during the long-term incubation may have also sensitized the tumor cells to effector cell-mediated lysis; a similar phenomenon has been observed by others.33
Effectiveness of immunotherapy against subcutaneous M109 tumor
In FITC-immunized mice bearing 7-day-old peritoneal M109 tumors, we noted that by extending the duration of treatment and optimizing cytokine dose, 100% cures could be achieved with little or no toxicity if folate-FITC (1,800 nmol/kg, 5 times/week) was supplemented with IL-2 at 40,000 units/dose (5 times/week) and IFN-α at 25,000 units/dose (3 times/week; data not shown). To evaluate the efficacy of the immunotherapy in a subcutaneous tumor model, mice bearing ≤ 50 mm3 subcutaneous M109 tumors were subjected to the 3-week treatment regimen (see Material and Methods) and examined for tumor growth. As shown in Figure 11, mice treated with the folate-FITC/cytokine combination exhibited a complete response (p < 0.01), whereas animals treated with cytokines alone displayed only retarded tumor growth (p < 0.05). These data demonstrate that the immunotherapy is also effective in subcutaneous tumor models.
Folate-FITC treatment of FITC-immunized mice can lead to complete cures of established tumors in a number of murine tumor models upon costimulation with low to moderate levels of cytokines17, 18 (and data not shown). Because little or no toxicity has been detectable at any dose of the therapy (Endocyte, personal communication), an optimized protocol has recently entered human clinical trials for treatment of kidney and ovarian cancers. Unfortunately, despite this progress in development of the drug, little information has been available on the mechanism of action of the hapten-targeted immunotherapy. Therefore, in the present article, we have explored the roles of antibody binding, CDC, ADCC, phagocytosis and acquisition of cellular immunity in the tumor rejection process.
Previous evidence that folate-FITC might form a bridge between anti-FITC antibodies and FR-positive tumor cells came from 2 observations: folate-FITC promotes anti-FITC IgG binding to tumor cells in vitro, and folate-FITC induces IgG accumulation in the tumor mass in vivo.17 However, documentation that this antibody accumulation is in fact critically involved in the mechanism of the folate-FITC-targeted immunotherapy was not available until it was shown here that folate-FITC can mediate ADCC-related killing of tumor cells in vitro (Fig. 2), and that the dependence of therapeutic efficacy on folate-FITC concentration displays the predicted bimodal shape both in vivo and in vitro (Figs. 1 and 3). Add to this observation that folate-FITC is totally ineffective in nonimmunized animals (Fig. 7a), and the conclusion is compelling that folate-FITC has no intrinsic cytotoxicity, but rather kills tumor cells by bridging anti-FITC IgG to FR on their cell surfaces.
Complement-dependent cytotoxicity was not expected to be a major contributor to the immunotherapy of the M109 solid tumors, because complement proteins do not readily perfuse solid tumors,34 because solid tumors often express regulatory proteins that can inactivate complement27, 35 and because solid tumors may secrete soluble inhibitors of complement.28 Thus, our inability to obtain evidence for complement participation in the folate-FITC/anti-FITC IgG-mediated cytotoxicity was not surprising. Nevertheless, it would still be naive to generalize that CDC will not participate in hapten-targeted therapy of all tumors, since some FR-expressing cancers are readily exposed to complement components (e.g., chronic myelogenous leukemia, acute myelogenous leukemia5, 36) and others may express fewer complement regulatory proteins.27
Antibody-dependent cellular cytotoxicity is typically mediated by FcR-expressing immune cells, primarily NK cells, macrophages and neutrophils.37, 38 Involvement of NK cells and macrophages in the hapten-targeted immunotherapy was confirmed by studies both in vitro and in vivo. Thus, NK cells were found to promote tumor cell lysis in vitro (Figs. 2 and 3), and macrophages were observed to bind and phagocytose juxtaposed regions of attached tumor cells in culture (Figs. 4 and 5). Further, depletion of either cell type in vivo was shown to lead to a significant decrease in therapeutic potency (Figs. 6 and 8). Although the lack of an appropriate antibody for neutrophil depletion precluded direct evaluation of this cell's role in the immunotherapy, preliminary immunohistochemistry data nevertheless demonstrate that folate-FITC administration induces a large influx of neutrophils (as well as other immune cells) into the tumor mass within 48 hr of treatment (data not shown). Thus, a significant role for neutrophils in the immunotherapy cannot be excluded.
In recently published17, 18 and unpublished studies, it has been observed that whenever complete tumor eradication is achieved by the hapten-targeted immunotherapy, a long-term immunity develops, such that subsequent rechallenge with the same tumor cells leads to their rejection without additional treatment. In general, long-term immunity of this sort is attributed to induction of memory T cells.39, 40, 41 In this study (Fig. 9), mice that had previously been cured of established M109 tumor and were subsequently rechallenged with fresh tumor cells readily rejected the new tumor cells under the same conditions that led to the rapid death of naive mice inoculated with the same cells. More importantly, a parallel cohort of cured mice that were first depleted of both CD4+ and CD8+ T cells and then rechallenged with the same tumor cells showed little extension of lifespan. Curiously, depletion of either T-cell subset alone was found to be largely ineffective in reversing the same tumor protective effect, suggesting that the tumor-specific immunity involves both CD4+ and CD8+ T cells. Further, ex vivo tumor-reactive T-cell activity was elevated in the spleens of immunotherapy-treated mice that had developed tumor protective memory (Fig. 10). Since CD4+ and CD8+ T cells were also found to be essential in the initial tumor regression process (Fig. 7), these data would suggest that both T-cell subsets contribute to both the initial effector phase and subsequent memory phase of the hapten-targeted immunotherapy. Indeed, 2 characteristics of the initial stages of the immunotherapy would be expected to predispose the immune system to this type of cross-priming. First, endogenous tumor antigen presentation should be facilitated by the enhanced tumor cell phagocytosis associated with direct macrophage/dendritic cell recognition of the antibody-decorated tumor cells. Second, the fact that tumor cell phagocytosis is triggered by antibody binding rather than tumor cell apoptosis or necrosis should strongly bias the antigen presenting cells to induce (rather than suppress) activation of their cognate T cells.42, 43, 44, 45 Taken together, the heavy hapten-induced deposition of autologous antibodies on tumor cell surfaces probably induces the infiltrating antigen presenting cells to promote development of a cellular immunity that would not have arisen if folate had been used to target a cytotoxic drug to kill the cancer cells. Finally, the proinflammatory cytokines (IL-2, IFN-α) that strongly synergize with the hapten-targeted immunotherapy may also play an important role in assisting the development of cellular immunity. Thus, IFN-α is well known to enhance tumor immunogenicity, promote antigen presentation and play a significant role in the generation of adaptive tumor immunity.32, 46
Since immunotherapies are rapidly emerging as viable strategies for the treatment of cancer,47, 48 it is perhaps important to identify the potential strengths and weaknesses of any new immunotherapeutic methodology that is presented for consideration. From our perspective, hapten-targeted immunotherapy suffers primarily from the delay that is required between tumor diagnosis and initiation of therapy. Thus, before a patient can respond to a hapten-targeted immunotherapy, he or she must first be immunized against the hapten (unless an exogenous antihapten antibody is administered). In cases where the cancer is progressing rapidly, the time needed to generate a sufficiently high antihapten antibody can impose an unacceptable delay to the patient. Second, the hapten-targeted immunotherapy is largely dependent on the ability of the cancer patient to generate antibodies against the foreign hapten. While research from other laboratories18 suggests that even cancer patients who cannot develop immunity against an endogenous tumor antigen can still elicit an immune response against a foreign hapten, the strength of the induced immunity can be somewhat variable. Thus, it is not unlikely that antihapten titers will have to be monitored and the vaccination schedule adjusted to accommodate variations in a patient's response. Third, unlike administration of a humanized antibody (e.g., herceptin), hapten-targeted immunotherapy requires the introduction of a foreign drug substance into the patient (in our case, folate-FITC). While a vitamin linked to a dye may seem relatively innocuous, its metabolism by the body must still be monitored to ensure no toxicity ensues.
On the positive side, folate-hapten conjugates have subnanomolar affinities for the cell surface FR, are intrinsically small in size and can penetrate solid tumors easily. Recently, optical imaging studies have been conducted to visualize the localization of folate-FITC in mice bearing FR-positive M109 and L1210A tumors.49 Kennedy et al.49 conclude that because of its small size and water solubility, folic acid can efficiently deliver the attached FITC molecule to virtually all malignant cells in both primary and metastatic tumor sites while maintaining a sharp contrast between tumor cells and the surrounding normal tissues. With FR-positive tumors expressing between 2 and 10 million FR/cell, we estimate that an average FR-positive cancer cell will be coated with an average of ∼ 5 × 106 hapten molecules per cell. Folate-hapten conjugates are also inexpensive, nonperishable, nonimmunogenic (i.e., except following vaccination against the hapten) and easy to prepare in large quantities. In contrast to exogenous monoclonal antibodies, autologous antihapten antibodies (induced by antihapten vaccination) may also recruit host effector cells more efficiently into solid tumors.8 Naturally induced antibodies are also nonimmunogenic, continuously generated in vivo and longer-lived in circulation than exogenous monoclonal antibodies.50 Moreover, the therapeutic efficacies of autologous antibodies are not affected by the Fc receptor polymorphisms that can compromise the coupling between exogenous antibodies and endogenous Fc receptor-expressing cells, especially in mediating ADCC.51, 52 Finally, endogenous antibodies may also better stimulate cross-presentation of tumor antigens, promoting a stronger hapten-independent elimination of tumor cells by the immune system. Thus, in addition to its disadvantages, the ability to retarget an endogenous humoral immune response to tumor cells with foreign haptens may have merits that will enable it to find a niche in the cancer field.
The authors acknowledge Andrew Hilgenbrink, Kathy Lagheb and Jennifer Sturgis for their technical assistance.