Ab-IL2 fusion proteins mediate NK cell immune synapse formation by polarizing CD25 to the target cell-effector cell interface
- First Online:
- Cite this article as:
- Gubbels, J.A.A., Gadbaw, B., Buhtoiarov, I.N. et al. Cancer Immunol Immunother (2011) 60: 1789. doi:10.1007/s00262-011-1072-9
- 225 Views
The huKS-IL2 immunocytokine (IC) consists of IL2 fused to a mAb against EpCAM, while the hu14.18-IL2 IC recognizes the GD2 disialoganglioside. They are under evaluation for treatment of EpCAM+ (ovarian) and GD2+ (neuroblastoma and melanoma) malignancies because of their proven ability to enhance tumor cell killing by antibody-dependent cell-mediated cytotoxicity (ADCC) and by antitumor cytotoxic T cells. Here, we demonstrate that huKS-IL2 and hu14.18-IL2 bind to tumor cells via their antibody components and increase adhesion and activating immune synapse (AIS) formation with NK cells by engaging the immune cells’ IL-2 receptors (IL2R). The NK leukemia cell line, NKL (which expresses high affinity IL2Rs), shows fivefold increase in binding to tumor targets when treated with IC compared to matching controls. This increase in binding is effectively inhibited by blocking antibodies against CD25, the α-chain of the IL2R. NK cells isolated from the peritoneal environment of ovarian cancer patients, known to be impaired in mediating ADCC, bind to huKS-IL2 via CD25. The increased binding between tumor and effector cells via ICs is due to the formation of AIS that are characterized by the simultaneous polarization of LFA-1, CD2 and F-actin at the cellular interface. AIS formation of peritoneal NK and NKL cells is inhibited by anti-CD25 blocking antibody and is 50–200% higher with IC versus the parent antibody. These findings demonstrate that the IL-2 component of the IC allows IL2Rs to function not only as receptors for this cytokine but also as facilitators of peritoneal NK cell binding to IC-coated tumor cells.
KeywordsImmunocytokinesNatural killerCancerImmunotherapyImmune synapse
Interleukin-2 (IL-2) is a lymphocytotrophic cytokine that augments helper and cytotoxic T cell as well as natural killer (NK) cell-mediated antitumor immune responses . High levels of IL-2 in the tumor mass are indicative of increased immune cell infiltration which in turn is associated with positive clinical outcomes in cancer patients [2, 3]. IL-2 was therefore proposed as an immunotherapeutic agent for the treatment of various types of tumors [4, 5]. However, systemic administration of high amounts of IL-2, to induce immune-mediated activity against the tumor causes significant toxicity [6, 7]. The immunocytokine approach was developed to enable tumor-reactive mAbs to initiate antitumor immune responses via ADCC while also delivering IL-2 directly into the tumor microenvironment [8, 9].
Immunocytokines (ICs) are recombinant fusion proteins consisting of cytokine molecules, like IL-2, linked in most but not all cases, to the heavy chains of tumor-reactive mAbs. ICs that can bind to antigens expressed on the surface of cancer cells are being investigated for their antitumor effects. Prominent antitumor ICs that have been developed and studied in vitro and in vivo (mouse models and humans) include huKS-IL2 (anti-EpCAM), hu14.18-IL2, L19-IL2, T84.66-IL2 and scFv23-TNF [10–20]. huKS-IL2 and hu14.18-IL2 are constructed using humanized anti-EpCAM (huKS) and humanized anti-disialoganglioside GD2 (14.18) antibodies, respectively [19, 20]. For huKS-IL2 and hu14.18-IL2, a human IL-2 molecule is genetically linked to the carboxy terminus of each IgG heavy chain making the intact molecule bivalent with respect to IL-2. huKS-IL2 and hu14.18-IL2 are being tested in clinical trials in EpCAM positive cancers and in neuroblastoma and melanoma patients, respectively [12, 21]. Antitumor activity has been seen in children with neuroblastoma receiving hu14.18-IL2 in phase II clinical testing , and clinical efficacy has been demonstrated for an anti-GD2 antibody combined with IL2 and GM-CSF, when used for neuroblastoma patients in remission .
Treatment with ICs results in activation of T cell-mediated antitumor cytotoxicity  and in NK cell-mediated control of tumor growth [25, 26]. The antibody component of the IC, upon binding to antigens on the tumor cells, activates NK cells via CD16-Fc interaction . Consequently, tumor cell lysis occurs via ADCC . However, deglycosylated or Fc-epitope lacking ADCC-incompetent ICs are also able to mediate increased tumor cell lysis. Previous studies have demonstrated that ICs may mediate tumor cell lysis, presumably via ADCC-independent mechanisms [18, 29, 30]. In a recently published study , we have demonstrated that NK cell lines lacking CD16 but expressing high levels of the high affinity IL2R receptor are able to form an increased number of conjugates with GD2pos target cells in the presence of the hu14.18-IL2 IC, whereas only a basal level of such heterotypic cell–cell interactions were observed upon treatment with the hu14.18 antibody or with simultaneous addition of hu14.18 antibody and IL2. Furthermore, we have also demonstrated that the increased IC-mediated NK-tumor cell conjugations were highly productive as they resulted in a higher level of lysis of the target cells .
The ability of the IC to engage the IL2R in conjugating cytotoxic immune cells with tumor cells and thereby increase cytolysis of the targets is likely to be physiologically relevant. NK cells isolated from ovarian cancer patients are deficient in the Fc receptor, CD16, but exhibit increased tumor cell killing in the presence of huKS-IL2 in ex vivo experiments [10, 32]. The huKS-IL2-IL2R interactions with IL2R on these patients’ NK cells, similar to those observed in experiments with the CD16neg NK cell lines , may allow the ovarian tumor patients’ CD16-defecient NK cells to more effectively lyse the tumor targets when coated with IC.
Recognition of target cells by NK cells via natural cytotoxicity mechanisms requires the formation of an immune synapse at the site of contact between the effector and tumor cells . NK targeting of tumor cells via ADCC also occurs through the formation of an immune synapse . We therefore hypothesized that after the IC binds to tumor cells via the antibody, it next binds to effector cells via an IL-2/IL-2 receptor (IL2R) interaction. The IL2R then induces activating immune synapse (AIS) formation leading to lysis of the cancer cells via an IL-2 receptor-dependent mechanism.
Here, we demonstrate that tumor cells and NK cells can conjugate through the IL2 receptor when treated with IC. In addition, treatment with IC increases activating immune synapses (AIS) between tumor cells and NK cells. These AIS are formed through the interaction of the IL-2 component of the IC and the IL2R on the NK cell. IC and IL2R are then polarized at the interface between immune cells and targets, indicating that the IL2R can play a role in cell adhesion and polarized cell activation when it encounters the IL2 component of the IC bound to the surface of the target cell. These novel findings elucidating the adhesive and polarizing properties of the IL2R and its ability to contribute to cell–cell conjugate formation may have important implications in how ICs can be used in cancer treatment.
Materials and methods
Cell culture and immunocytokines
OVCAR-3 cells were obtained from ATCC and M21 and NKL cells were kind gifts from Drs. Ralph Reisfeld and Jenny Gumperz, respectively. huKS mAb, huKS-IL2 IC, hu14.18 mAb and hu14.18-IL2 IC were provided by The EMD-Lexigen Research Center (now EMD-Serono) in Billerica MA and were used at 0.25 μg/ml in all experiments. IL-2 (Proleukin, Chiron) was used at 750 U/ml and anti-CD25 blocking antibody (clone GL439) used in all experiments was 10 μg/ml. Unless noted otherwise, statistical analysis of all data was conducted using Mann–Whitney tests and graphs were plotted using the GraphPad Prism software.
Slide preparation and analysis for confocal microscopy
Target cells (7.5 × 104 cells) were mixed with effector cells at a 1:1 ratio in 400 μl phosphate buffered saline containing 1% bovine serum albumin (PBS-BSA) along with the IC, antibodies, IL-2, or CD25 blocking antibody. The cells were centrifuged for 5 min at 100×g and incubated at 37°C at 5% CO2 for 25 min. After incubation, the cells were fixed with 3% paraformaldehyde on poly-l-lysine coated coverslips, washed twice with 150 mM glycine, permeabilized with 0.1% Triton-X for 4 min and blocked overnight or for 30 min with PBS containing 5% goat serum. Some coverslips were stained (on separate coverslips) with murine antibodies, LFA-1 (1:200), CD2 (1:200) and antihuman CD25 non-blocking antibody (clone HB8784, 1 μg) in 5% goat serum or with a cocktail of goat anti-mouse FITC (1:300, Jackson ImmunoResearch), and rhodamine conjugated phalloidin (1:1,000, Invitrogen) for 30 min. Coverslips were washed, stained with DAPI, mounted in mounting media (Invitrogen), dried overnight and visualized using the Bio-Rad Radiance 2,100 MP Rainbow confocal microscope. Fifty conjugates between effector and target cells were counted on each coverslip. Each conjugate was scored for polarization at the contact interface for molecules LFA-1 and F-actin, CD2 and F-actin, CD25, or 14.18-IL2-FITC. Percent AIS formation was determined as: [(number of conjugates showing polarization of molecules/total conjugates) X100].
Plate adhesion assays
NKL (50,000 cells/well) cells labeled with Calcein AM were added to confluent cultures of OVCAR-3 or M21 tumor targets under the different control and test conditions. After 30 min incubation, the wells were washed twice with PBS-BSA and fluorescence of target cell-bound NKL were quantified using the Victor V-3 plate reader (Perkin Elmer).
Flow cytometry conjugate formation assays
OVCAR-3 and NKL cells were labeled with 0.5 M CellTracker Red (Invitrogen) and 1.25 pM CellTracker Green (Invitrogen), respectively. In other experiments involving M21 and NKL, the cells were labeled with BODIPY 630/650, and CFSE, respectively. The tumor targets and NKL cells were mixed at 1:1 ratio in the presence of IC or other specified test and control conditions. After 30 min incubation, tumor cell–NKL cell conjugation was measured by flow cytometry as described in our previous study .
NK cells from patients and healthy donors
Peripheral blood mononuclear cells from health donors and from ovarian cancer patients, and tumor associated NK cells from the peritoneal fluid of patients with ovarian cancer were obtained and purified for in vitro analyses as reported previously [32, 36]. All peripheral blood and peritoneal fluid samples were obtained with approval of the UW Health Sciences Human Subjects committee.
IC binding to patient NK cells
Previously frozen peritoneal fluid mononuclear cells were incubated with Goat IgG for 15 min and then with or without huKS-IL2 in PBS-BSA for 10 min. FITC-conjugated anti-CD25 antibody (clone 2A3, BD Biosciences, 20 μl/test) was added to some tubes and incubated for 10 min. After washing, the cells were labeled with fluorophore conjugated anti-NKp46, CD16, CD56 and CD3 to distinguish NK cells and samples were analyzed for anti-CD25 binding. Four patient samples were used in this analysis. A matched paired t test was conducted for statistical analysis using the statistical program JMP.
Immunocytokines increase NK AIS formation
IC binds to peritoneal NK cells of ovarian cancer patients via CD25
IC-CD25 interaction facilitates AIS NK-tumor cell conjugation
CD25 is polarized at the IC-mediated AIS
The data presented in this study conclusively demonstrate a role for CD25 in mediating attachment and formation of AIS by NK cells when these NK cells encounter tumor cells that display cell surface-bound IL2 that was delivered by the tumor-reactive IC. As such, the IC functions as a cross-linking agent. ICs have already been shown to mediate stable conjugate formation between T cells and tumor targets via IL2R . We have also studied the kinetics of IC-IL2R binding and demonstrated that the IC-IL2R complex is rapidly internalized by the NKL cells . Our studies further confirm that ICs presented on the target cells serve as ligands of IL2R and also indicate that IC-facilitated conjugation of NK cells with cancer cells may serve as an important mechanism in the therapeutic control of tumor growth.
NK cells play an important role in the elimination of tumor cells by ADCC as well as natural cytotoxicity (via NKG2D, DNAM-1, and other activating receptors) mechanisms [39–41]. NK cell-mediated killing of tumor targets via both ADCC and natural cytotoxicity mechanisms is enhanced when the effector cells are activated by IL-2 treatment. IL-2 treated NK cells have classically been referred to as lymphokine activated killer (LAK) cells due to their high potential in lysing tumor targets . In the extensive literature available on IL-2 stimulation of NK cells, the mechanism for activation involves the binding of IL-2 to its receptor on the NK cells thereby triggering an activating signaling cascade.
CD25 is the alpha chain of the IL-2 receptor and by itself does not have any signaling capacity . However, on the immune cell surface, CD25 is associated with the beta (CD122) and gamma chains of the IL-2 receptor that have the signaling motifs [44, 45]. Furthermore, CD25 has a very low affinity for IL-2 [46–48]. However, the trimeric IL-2 receptor (CD25 along with the beta and gamma chains) is a high affinity IL-2 receptor. Treatment with IC leads to stimulation of the NK cells via IL-2 thereby enhancing their antitumor responses . Our experiments with NK cells indicate that IC-facilitated binding of effector cells to tumor targets is significantly inhibited by anti-CD25 antibodies that block the recognition of IL-2 by this receptor. Comparable results were also obtained with NKL cells, which are virtually lacking Fc receptors. Therefore, in the presence of tumor cells that have been coated with IC, these NK and NKL cells are coming in contact with a cell surface that is now expressing a lattice of surface-bound IL2. In this setting, the IL2Rs of the NK and NKL cells bind to the tumor cell-bound IL2 and act initially as adhesion molecules, facilitating contact between NK and tumor cells via the IC.
The divalent presentation of IL2 on each IC molecule, and the presentation of cell-bound IL2 on the IC-labeled tumor cells may allow the CD25 molecules on the NK and NKL cells to bind to the IL2 fixed to the tumor cell with significantly greater affinity than that for soluble IL-2 . This theoretical higher affinity of cell-bound IC (versus soluble IL2) to CD25 is consistent with our cell adhesion blocking experiments, where an equimolar amount of excess soluble IL-2 was unable to inhibit IC-mediated binding between NKL cells and M21 (hu14.18-IL2 + IL2 in lower panel of Fig. 5). Similarly, a 3–5-fold excess of soluble IL2 does not inhibit conjugation of NKL cells to huKS-IL2 labeled OVCAR-3 cells (Fig. 4c). In contrast, almost complete inhibition of IC-facilitated NKL-tumor cell binding was observed when the anti-CD25 antibody was used as the blocking agent (Figs. 4b, c and 5), indicating that soluble IL-2 is not a strong competitor of the CD25 interaction between the IL2Rs on NK and NKL cells, and the cell-bound IL2 that IC has delivered to the tumor cell surface. This observation may help explain why in vivo administration of IC is so much more effective in eradicating tumor than is comparable amounts of monoclonal antibody + soluble IL2 . The IL2 component of the tumor-bound IC may attract effector cells via conjugate formation using CD25 as an adhesion molecule (as seen in vitro in Figs. 4 and 5). Furthermore, recruitment of immune cells to the tumor bed following IC therapy may lead to an increase in IL-2 synthesis and release at the tumor site. However, the secreted soluble IL-2, because of its lower affinity for the IL-2 receptor, is not likely to effectively compete with the tumor cell-bound IC, thereby enabling the IC to continue to recruit effectors to the tumor site .
Formation of the AIS is essential for NK cells to mediate lysis via their activating receptors [33, 51]. Lysis of target cells by ADCC is also associated with immune synapse formation [34, 52]. The AIS is associated with polarization of LFA-1 and CD2 along with the activating receptors . Our studies with fresh human NK cells show that the huKS-IL2 IC induces significantly greater AIS than does the corresponding huKS mAb, even when additional soluble IL2 is added (Fig. 1b, c). These conjugates formed by the IC show the typical phenotype of AIS, with LFA-1 and CD2 polarized to the synapse. This result suggests that the IL2 component of the IC, when fixed to the tumor surface, functions as an additional adhesive target that also induces effector polarization. Using NKL cells as a model, we demonstrate that CD25 not only facilitates binding of the effector cells to tumor targets but is also polarized to the site of the immune synapse (Fig. 7). A schematic diagram is included as Supplemental File 1, to demonstrate the relationships of GD2 distribution on M21 tumor cells, IL2R on NKL cells and the localization of FITC-conjugated hu14.18-IL2 to these M21 and NKL cells, when cultured alone, or when synapses are formed via the IC, based on the actual photomicrographs shown in Figs. 6 and 7. Separate analyses have recently demonstrated that these same NKL cells mediate greater GD2-specific killing of M21 cells when co-cultured with hu14.18-IL2 than when cultured with hu14.18 mAb + IL2 , indicating that the AIS facilitated between NKL and M21 cells by hu14.18-IL2 also leads to lysis. Future studies will evaluate the potential association between CD25 polarization (by cell-bound IC) and the actin cytoskeleton and WIP and WASp proteins that are required for the transport of the signaling molecules and perforin granules to the NK cell AIS [53, 54].
Studies in ovarian cancer patients have shown that NK cells isolated from the tumor microenvironment express reduced levels of the Fc receptor CD16 [32, 55]. This down-regulation of CD16 is likely brought about by the interaction of cancer antigens with tumor infiltrating NK cells . NK cells from the peritoneal environment of ovarian cancer patients therefore exhibit reduced ability to mediate ADCC as compared to circulating NK cells of the same patient; however, these cells demonstrate enhanced cytotoxicity in the presence of IC (10). The data we have generated with the FcR-deficient NKL cells suggests that NK-mediated antitumor activity with IC molecules may proceed even without activation of FcRs, as the IL2 that the IC brings to the tumor cell surface can induce adhesion and activation of AISs on NK cells. Our results are also consistent with in vivo antitumor effects seen with other ICs in settings where FcR-mediated ADCC was impossible [20, 25, 29].
Our in vitro data presented here are also consistent with a recent clinical observation from our phase II study of hu14.18-IL2 in children with relapsed/refractory neuroblastoma . An analysis of KIR/KIR-ligand genotype showed that antitumor-response (or clinical improvement) was associated with KIR/KIR-ligand mismatch . This suggests that the clinical effect of IC in these patients is mediated by NK cells, as shown in vitro in this report. That same study  also evaluated for possible associations between FcγR genotypes and response. In contrast to other studies showing an association between high affinity genotypes for the FcγRIIIa receptor and clinical responses to other therapeutic mAbs [58, 59], our study of hu14.18-IL2 showed no trend toward any association of high affinity FcγRIIIa alleles and response . While this lack of any trend toward association may reflect the small cohort of patients in this study, the data presented in this in vitro report provide another hypothesis. Namely, it may be that the activation of NK-tumor cell synapses via the IC may utilize either IL2Rs or FcRs (or both), thereby circumventing the potential need for (or benefit of) high affinity FcγRIIIa alleles. Further clinical confirmatory studies and in vitro analyses are needed to test this hypothesis.
Funding for this research was provided by grants from the Department of Defense (#W81XWH-04-1-0102), Ovarian Cancer Research Fund (UW/UWM.05), start-up funds from the Department of Obstetrics and Gynecology to MSP, a charitable donation from Jean McKenzie to JC and MSP, a T32 training grant to JAAG, and by U10CA98543 COG Group Chair Grant, and R01-CA-32685-25, P30-CA14520, CA87025, CA81403, RR03186, and grants from the Midwest Athletes for Childhood Cancer Fund, The Crawdaddy Foundation, The St. Baldrick’s Foundation, The Evan Dunbar Foundation and Abbie’s Fund to PMS. We are deeply grateful to Kathy Schell for her advice and help and acknowledge the support provided by the University of Wisconsin Comprehensive Cancer Centers Flow Cytometry facility which is supported by a core grant (CA14520) from the National Institutes of Health, to Lance Rodenkirk for help with confocal microscopy and image analysis, and to Lori Scardino for help with the schematic figure.