An Antibody Fusion Protein for Cancer Immunotherapy Mimicking IL-15 trans-Presentation at the Tumor Site

The application of cytokines to induce or promote the generation of an effective antitumor immune response is an attractive approach in cancer immunotherapy. Among a continuous growing list of cytokines with antitumor potential (1), interleukin (IL)-15 appears as a candidate with especially promising features (2, 3).

IL-15 belongs to the 4 α-helix bundle cytokine family and was initially described in relation to IL-2, with whom it shares several biological functions, for example, the induction of proliferation and differentiation of natural killer (NK), T, and B cells (4, 5, 6). Currently, the major function of IL-15 appears to focus on the generation, proliferation, and activation of NK cells and the support of the survival of CD8⁺ memory T cells (7, 8). Unlike IL-2, IL-15 rather inhibits activation-induced cell death (9) and seems not to exert crucial influence on regulatory T cells (10), important properties in view of an immunotherapeutic application.

IL-15 and IL-2 have 2 receptor units (IL-2/15Rβ and γ_c) of their heterotrimeric receptors in common, while the third receptor unit (IL-2Rα and IL-15Rα) is specific for each cytokine. Functional differences observed between IL-2 and IL-15 presumably ascribe mainly to the interaction with their respective α receptor unit. Although IL-2 is secreted by activated T cells and binds in a soluble form to its high-affinity heterotrimeric receptor (IL-2Rαβγ) on activated cells, IL-15 acts mainly in a membrane-bound form, presented in trans in association with its specific IL-15Rα subunit by monocytes and dendritic cells (antigen-presenting cells, APC) to NK or CD8⁺ T cells, expressing the heterodimeric intermediate affinity receptor IL-2/15Rβγ (ref. 11; Fig. 1A). For the process of trans-presentation, intracellular IL-15 binds to the IL-15Rα chain that shuttles to the cell surface where it presents IL-15 to neighboring cells expressing the receptor IL-2/15Rβγ. Thus, a more directed and controlled action of the cytokine is achieved (12). The sushi domain of the IL-15Rα chain has been identified as the structural region involved in IL-15 binding (13). There is also the possibility that IL-15 binds to soluble versions of IL-15Rα forming soluble complexes with stimulatory or inhibitory activity. In this case, soluble IL-15Rα might arise from proteolytic cleavage of the membrane form by metalloproteinases (14, 15).

The antitumor potential of IL-15 was observed in mouse models employing recombinant IL-15 (16, 17), antibody–IL-15 fusion protein (18), IL-15 gene-modified tumor cells (19, 20), and IL-15 transgenic mice (21). The relevance of IL-15 trans-presentation for an efficient antitumor response was shown in a mouse model with IL-15Rα–transfected MC38 tumor cells. Here, the expression of IL-15Rα dose dependently delayed the formation of fatal pulmonary metastases through the activation of NK cells (22). Subsequently, it was shown in vitro and in vivo that soluble complexes of recombinant IL-15 and IL-15Rα were much more stimulatory than soluble IL-15 alone (23). Treatment of mice with IL-15xIL15RαFc complexes enhanced the proliferation of CD8⁺ memory T cells and NK cells in vivo. Furthermore, tumor burden could be reduced and survival extended in a systemic B16 melanoma mouse model (24, 25). Similar results were obtained by an IL-15/IL-15Rα domain fusion protein (26, 27).

We hypothesize that targeting a fusion protein consisting of IL-15 and an extended IL-15Rαsushi domain by an antibody moiety to the tumor site will mimic locally the trans-presentation of IL-15 by the IL-15Rα, enhancing its stimulatory potential, thus further improving the antitumor effect of IL-15 (Fig. 1A). In this study, we report the generation of the antibody–cytokine fusion protein scFv_RD_IL-15, composed of the fibroblast activation protein (FAP)-specific scFv (scFv36), an IL-15Rα fragment (RD) and IL-15. We show antibody-mediated binding and cytokine activity for this construct in solution or targeted form and describe the influence of the RD on the receptor activation properties of the fusion protein. Proliferation of an IL-15Rβγ receptor-expressing cell line (Mo7e) was enhanced by the RD-containing fusion protein in solution, whereas proliferation of a cell line expressing the IL-15Rαβγ receptor (CTLL-2) was reduced. Nevertheless, in target-bound form, the proliferation effect on CTLL-2 was mainly RD independent. Also, peripheral blood mononuclear cells (PBMC) could be effectively stimulated by scFv_RD_IL-15 in targeted and soluble form. Here, scFv_RD_IL-15 was more efficient than scFv_IL-15 enhancing the proliferation and cytotoxicity response of unstimulated and activated T cells, respectively. Furthermore, scFv_RD_IL-15 was more effective than scFv_IL-15 or RD_IL-15 in the treatment of a lung metastasis mouse model. Thus, the combination of antibody-mediated tumor targeting and IL-15Rα domain associated trans-presentation of IL-15 in form of a fusion protein appears as a promising approach for cancer immunotherapy.

Antibodies were purchased from Biolegend (αCD3-PerCP and αCD8-PE), Immunotools (αCD4-PE and αCD56-PE), Miltenyi Biotec (αhexahistidyl-tag-PE), and Santa Cruz (αIL-15, αhexahistidyl-tag-HRP, and αCD107a-FITC). Recombinant human IL-2 and granulocyte-macrophage colony-stimulating factor (GM-CSF) were obtained from Immunotools and recombinant FAP from R&D Systems. B16-FAP and B16 cells (K. Pfizenmaier, IZI) were cultured in RPMI-1640, 5% FBS. Mo7e (H. van der Kruip, IKP) and CTLL-2 (P. Scheurich, IZI) were cultured in RPMI-1640, supplemented with 20% FBS, 10 mmol/L HEPES, 2 mmol/L l-glutamine, 0,13 mmol/L l-asparagine, 0,05 mmol/L β-mercaptoethanol, 1 mmol/L natriumpyruvate, nonessential amino acids, and 10 ng/mL rhGM-CSF (Mo7e) or 400 IU/mL rhIL-2 (CTLL-2), respectively. Cells were tested for mycoplasms and their morphologic appearance monitored by microscopic means. Antigen (FAP) expression and cytokine growth dependence of the respective cell lines was verified by flow cytometry and proliferation assays, respectively. PBMCs were isolated from buffy coat of healthy donors (blood bank, University Medical Center Ulm) by Ficoll gradient and were cultivated in RPMI-1640, 10% FBS. C57BL/6JRj mice were purchased from Elevage Janvier. Animal care and all experiments carried out were in accordance with federal guidelines and have been approved by university and state authorities.

The procedure is described in the Supplementary Data S1.

A total of 2.5 × 10⁵ cells (B16-FAP/B16) were incubated with fusion protein (200 nmol/L or titrated, accordingly) for 2 hours at 4°C. After washing, cells were incubated with a hexahistidyl-tag directed phycoerythrin (PE)-conjugated mouse monoclonal antibody, for 1 hour at 4°C. Washing and incubation steps were carried out in PBS, 2% FBS, and 0.02% sodium azide. Finally, cells were analyzed in a Cytomics FC500 (Beckman Coulter).

Cytokine growth-dependent cells (CTLL-2/Mo7e) were washed and suspended in culture medium with neither IL-2 nor GM-CSF. A total of 20,000 cells per well were seeded in 96-well plates and starved for 4 hours. Then the fusion proteins were added to the cells. After 3 (CTLL-2) or 4 (Mo7e) days, proliferation of the cells was measured by MTT assay (28). The cytokine activity of the fusion proteins presented in an antibody-mediated membrane-bound form was analyzed in the following setting: 10,000 cells per well (B16-FAP/B16) were seeded in a 96-well tissue culture plate. Next day, cell growth was arrested by treatment with 10 μg/mL mitomycin C (Sigma), for 2 hours at 37°C. Cells were washed twice with medium before addition of the fusion proteins. After 1 hour incubation at room temperature, the supernatant was discarded and adherent cells washed twice with medium. Then previously starved CTLL-2 cells (20,000 cells per well) were added and incubated for 3 days. Finally, the cells in suspension (CTLL-2) were carefully transferred to a new 96-well plate, followed by the MTT assay (28).

PBMCs were stained at a concentration of 1 × 10⁶ cells/mL with 625 nmol/L carboxyfluorescein diacetate succinimidyl ester (CFSE) using the CellTrace CFSE Cell Proliferation Kit (Life Technologies), following the instructions of the manufacturer. Proliferation induced by the fusion proteins in solution was assessed by incubating 2 × 10⁵ PBMCs per well with the corresponding fusion protein for 5 days, followed by flow cytometry analysis. Additional antibody-mediated staining was carried out to identify T cells (CD3-PerCP), CD4⁺T cells (CD3-PerCP/CD4-PE), CD8⁺T cells (CD3-PerCP/CD8-PE), and NK cells (CD3-PerCP/CD56-PE).

To determine PBMC proliferation induced by targeted fusion proteins, recombinant FAP (3 μg/mL) was coated overnight at 4°C on an ELISA plate. Fusion proteins (1 μmol/L) were added and incubated for 1 hour at room temperature. After washing with PBS, 2 × 10⁵ PBMCs per well were added. Alternatively, the fusion proteins (50, 100, 200 nmol/L) were incubated for 1 hour with 2 × 10⁵ B16-FAP cells per well. Then, B16-FAP cells were washed and 2 × 10⁵ preactivated PBMCs per well added. After 5 days, proliferation of PBMCs or T cells (additional CD3-PerCP staining) was determined by flow cytometry. Preactivated T cells were used in the targeted assay, as coculture with B16-FAP cells inhibited the responsiveness of unstimulated T cells. Preactivation of PBMCs was achieved by coculture with adherent B16-FAP cells in presence of 2.5 pmol/L bispecific antibody scDbFAPCD3 for 24 hours. Subsequently, PBMCs were harvested and washed before their addition to the assay as indicated above.

Cytotoxic potential of T cells was assessed in terms of degranulation and determined by measuring CD107a exposure on the cell surface by flow cytometry analysis. In the first step, 2 × 10⁵ unstimulated PBMCs per well were incubated with the fusion proteins in solution or alternatively 2 × 10⁵ preactivated PBMCs (as indicated above) were incubated with the fusion proteins targeted to B16-FAP cells (as indicated above). After 5 days, cells were washed and cytotoxic T-cell degranulation triggered. A total of 2 × 10⁴ B16-FAP cells per well were seeded the day before and incubated with 30 pmol/L bispecific antibody scDbFAPCD3 for 1 hour, before the addition of pretreated PBMCs for additional 6 hours in presence of 1.4 μL per well BD GolgiStop (monensin; BD Biosciences). Degranulation of T cells was monitored by CD107a-FITC/CD3-PerCP staining and flow cytometry analysis.

Groups of 5 C57BL/6JRj mice (4–6 months) were injected i.v. with 1 nmol fusion protein per mouse. Blood samples were taken at different time intervals (3, 30, 60 minutes, 2, 6, and 24 hours) from the tail and serum concentration of the fusion proteins analyzed by sandwich ELISA. In brief, anti–IL-15 monoclonal antibody (3 μg/mL) was coated. After blocking, serum and purified fusion proteins were titrated and bound fusion protein detected via horseradish peroxide (HRP)-conjugated mouse hexahistidyl-tag–specific antibody as described elsewhere (29). Serum concentration of the fusion proteins was determined by interpolating the corresponding standard curve. For comparison, the first value (3 minutes) was set to 100%. Pharmacokinetic parameters [(t_1/2α, t_1/2β, area under the curve (AUC)] were calculated via Excel using the first 3 to 4 time points to calculate t_1/2α and the last 3 measurable time points to calculate t_1/2β. AUC was determined for the time interval of the first 6 hours.

C57BL/6JRj mice (4 months) were injected i.v. with 8.5 × 10⁵ B16-FAP cells per animal. Groups of 5 to 6 animals were treated with PBS, scFv_IL-15, scFv_RD_IL-15, or RD_IL-15, respectively. Treatment of 0.2 nmol fusion protein was applied i.p. on days 1, 2, and 10 after cell injection. Mice were sacrificed on day 20. Lungs were removed and metastasis counted.

For comparison of multiple groups in the in vitro experiments, the one-way ANOVA followed by the Tukey post test was applied. Statistical analysis of the in vivo experiments was conducted by Student t test (GraphPad Prism software).

The fusion protein scFv_RD_IL15 is composed of an antibody moiety, a receptor domain, and a cytokine component. The antibody component corresponds to an scFv that consists of the variable regions of the heavy and light chain of the antibody mo36 (ref. 30; scFv36) directed against the tumor stromal FAP. This antibody is cross-reactive in mice and humans and had been previously generated at the institute (30). It was used in our study to target a stably FAP-transfected B16 melanoma cell line, suitable for in vitro and in vivo studies. Thus, FAP is used here as a model antigen to evaluate the feasibility of our approach, which might be extended later on to other tumor targets. The receptor domain (RD) comprises the amino acids 31 to 107 of the human IL-15Rα (sushi domain plus the subsequent 12 amino acids). The cytokine component corresponds to human IL-15. In the scFv_RD_IL-15 construct, the scFv36 is located at the N-terminus, followed by the RD and the IL-15 domain. Furthermore, scFv_IL-15, an antibody fusion protein without RD was generated as well as the control fusion proteins scFv_RD and RD_IL-15, missing IL-15 and scFv, respectively. All constructs are provided with a hexahistidyl-tag at the C-terminus (Fig. 1B).

The fusion proteins were produced in stably transfected HEK293 cells and purified by immobilized metal ion affinity chromatography with yields of 2.6 mg/L (scFv_RD_IL-15), 0.26 mg/L (scFv_IL-15), 0.66 mg/L (scFv_RD) and 0.88 mg/L (RD_IL-15), respectively. Analysis of the purified proteins by SDS-PAGE (Coomassie staining) under reducing conditions (Fig. 1C) revealed high degree of purity and a single band for the scFv_RD construct (36 kDa), pronounced double bands for the fusion proteins scFv_RD_IL-15 (48 and 53 kDa) and RD_IL-15 (24 and 28 kDa), and 3 bands for scFv_IL-15 (42, 46, and 50 kDa). The IL-15 sequence comprises 3 potential N-linked glycosylation sites. The N-glycosylated nature of the fusion proteins with IL-15 component was shown by PNGase F digestion of the proteins, resulting in a reduction of the slower migrating bands in SDS-PAGE (data not shown). The bands correlated with the predicted molecular mass of the fusion proteins: scFv_RD_IL15 (51.4 kDa), scFv_IL-15 (41.5 kDa), scFv_RD (37.2 kDa), and RD_IL-15 (25 kDa). The identity of the fusion proteins was further confirmed by immunoblot analysis with an anti–hexahistidyl-tag antibody (Fig. 1D). High-performance liquid chromatography analysis of scFv_RD_IL-15 and scFv_IL-15 revealed a mainly homogeneous population, taking into account the glycosylation status of these proteins. A second minor peak probably corresponds to a dimeric form, as scFvs might dimerize to a certain degree. No higher aggregate formation was observed (Supplementary Data S2).

Specific binding of the antibody fusion proteins to FAP was shown by flow cytometry (Fig. 1E). Stably FAP-transfected B16 cells (B16-FAP) were used as FAP-positive cell line, and B16 wild-type cells (B16) were used as FAP-negative control cell line. ScFv_RD_IL-15, scFv_RD, and scFv_IL-15 showed binding to FAP-positive B16-FAP cells but not to FAP-negative B16 cells. No binding was detected for RD_IL-15 neither on B16-FAP nor on B16 cells. Furthermore, titration of the antibody fusion proteins revealed similar binding affinity of scFv_IL-15 and scFv_RD_IL15 that was approximately 7- and 5-fold higher than the binding affinity shown by the scFv and scFv_RD, respectively (Fig. 1F).

The cytokine activity of the fusion proteins was analyzed in proliferation assays with the cytokine growth-dependent cell lines CTLL-2 and Mo7e. Human IL-15 induced the proliferation of both cell lines, allowing the analysis of stimulatory properties in the context of different IL-15 receptor chain compositions. CTLL-2, a murine cytotoxic T lymphocytic cell line, expresses IL-15Rα and IL-15Rβγ, while Mo7e, a human megakaryocytic leukemic cell line, expresses IL-15Rβγ chains only. Proliferation assays with the fusion proteins in solution revealed cytokine activity on both cell lines for all fusion proteins containing IL-15 (Fig. 2A and B). On CTLL-2 cells, scFv_IL-15 showed high activity at picomolar concentrations, while fusion proteins comprising the RD fragment (scFv_RD_IL-15 and RD_IL-15) were clearly less active (approximately 50-fold, comparing constructs with and without scFv, respectively; Fig. 2A). In contrast, on Mo7e cells, proliferation in response to scFv_RD_IL-15 and RD_IL-15 was also comparable but significantly more active (20- to 30-fold) than scFv_IL-15 (Fig. 2B). The scFv and the scFv_RD did not show any stimulatory effect neither on CTLL-2 nor on Mo7e cells.

Next, the cytokine activity of the constructs was investigated in their membrane-bound form, targeting the FAP antigen on the cell surface, thereby mimicking a trans-presentation of IL-15 either in presence (scFv_RD_IL-15) or in absence (scFv_IL-15) of the receptor α-chain domain (RD). The fusion proteins were incubated on B16-FAP or B16 cells and the IL-15 activity of bound constructs was determined by the proliferation of cocultured CTLL-2 cells. Here, proliferation was observed in response to the antigen-dependent membrane-bound presentation of both, scFv_RD_IL-15 and scFv_IL-15 (Fig. 2C). Unlike in solution, both constructs displayed a similar proliferation effect. As expected, no stimulation of CTLL-2 was observed for the experimental setting with FAP-negative B16 cells (Fig. 2D).

Then, we analyzed the proliferation effect of the fusion proteins on PBMCs. Similar to the proliferation pattern observed for Mo7e cells, fusion proteins in solution presenting IL-15 and RD (RD_IL-15 and scFv_RD_IL-15) achieved a stronger proliferation response (11- to 15-fold) than the fusion protein with IL-15 only (scFv_IL-15). Also in this case, scFv_RD_IL-15 and RD_IL-15 were equally effective, while scFv_RD did not induce proliferation at all (Fig. 3A). In addition, scFv_RD_IL-15 and scFv_IL-15 were also shown to be active in their target-bound form (Fig. 3B). Furthermore, we determined the proliferation effect of scFv_RD_IL-15 and scFv_IL-15 in solution in regard to different immune cell populations (Fig. 3C). Here, NK and CD8⁺ T cells were more responsive to IL-15 stimulation than CD4⁺ T cells. Irrespective of the cell type, scFv_RD_IL-15 was significantly more effective than scFv_IL-15 at low concentration (1 nmol/L). Improved proliferation of T cells in response to IL-15 presented in combination with RD was shown for the fusion proteins in solution (Fig. 4A) as well as for the fusion proteins in target-bound state (Fig. 4B). For the latter, scFv_IL-15 and scFv_RD_IL-15 were bound via the antibody moiety to FAP-expressing cells (B16-FAP), presenting IL-15 or RD_IL-15, respectively on the cell surface. Thus, by mimicking IL-15 trans-presentation, the proliferation effect on preactivated T cells was determined, indicating an advantage for the membrane presentation of IL-15 in an additional RD-bound form. Furthermore, the influence of the fusion proteins on the cytotoxic potential of T cells was analyzed (Fig. 4C and D). In solution, scFv_RD_IL-15 as well as scFv_IL-15 was able to enhance the percentage of degranulating T cells, whereby at low concentrations (0.1 and 1 nmol/L) scFv_RD_IL-15 showed a stronger effect than scFv_IL-15 (Fig. 4C). Also in the target-bound form, scFv_RD_IL-15 was at lower concentration more effective than scFv_IL-15, increasing the cytotoxic potential of preactivated T cells (Fig. 4D). Therefore, the fusion of RD to IL-15, presented either in soluble or targeted form, improves significantly the response of T cells in terms of proliferation and cytotoxicity.

Pharmacokinetic analysis in mice revealed some differences in the half-life of the fusion proteins. ScFv_RD_IL-15 showed an approximately 2-fold longer t_1/2α than RD_IL-15 and scFv_IL-15 (P < 0.05). t_1/2β was similar for scFv_RD_IL-15 and RD_IL-15, but 6-fold (P < 0.01) and 3.5-fold (P < 0.001) increased in comparison with scFv_IL-15. The AUC calculated for scFv_RD_IL-15 was approximately 1.7- and 2.5-fold higher than the AUC determined for RD_IL-15 and scFv_IL-15, respectively (P < 0.01; Table 1; Fig. 5A).

Finally, the therapeutic potential of the fusion proteins was investigated in a B16-FAP lung metastasis mouse model, whereat dosing and treatment schedule was chosen based on data of the IL-15/IL-15Rα domain fusion protein (RLI) in the literature (26). B16-FAP cells were injected i.v. into C57BL/6JRj mice on day 0 and treated on days 1, 2, and 10 with a single dose of 0.2 nmol fusion protein (scFv_RD_IL-15, scFv_IL-15, and RD_IL-15), respectively. After 20 days, lungs were removed and the number of tumor cell foci counted (Fig. 5B). In comparison with the PBS control group, all groups treated with IL-15 fusion proteins showed significantly less metastasis. Strongest effect was achieved by treatment with scFv_RD_IL-15 (approximately 86% metastasis reduction), followed by application of RD_IL-15 and scFv_IL-15 with similar outcome (approximately 46% and 53% metastasis reduction, respectively). Thus, comparing the targeted versions of the IL-15 fusion proteins (scFv_IL-15 and scFv_RD_IL-15), the presence of RD significantly enhanced the therapeutic potential. On the other hand, comparing the targeted and nontargeted versions of RD bearing IL-15 fusion proteins (scFv_RD_IL-15 and RD_IL-15) showed that the targeted form was more effective. Thus, scFv_RD_IL-15, combining tumor targeting and IL-15 presentation in the RD context, revealed the best therapeutic potential. No signs of toxicity (e.g., weight loss) were observed during the therapeutic experiment.

Initially, soluble IL-15Rα, generated by shedding from human cell lines was observed to exert an antagonistic effect. Binding of soluble IL-15Rα to soluble IL-15 led to competition with cell surface receptors inhibiting IL-15–mediated function (14). Later, complex formation of recombinant soluble IL-15 and IL-15Rα-Fc molecules was reported to enhance the bioactivity of IL-15, inducing proliferation of NK cells and memory T cells in vitro and in vivo (23, 25). Structural studies identified the “sushi” region in the IL-15Rα as the main responsible element for IL-15 binding (31). In vitro, soluble complexes of recombinant IL-15 and the IL-15Rα sushi domain were described to induce agonistic effects on the intermediate affinity receptor (IL-15Rβγ), inducing stronger proliferation than IL-15 alone (32). This effect could be even enhanced when the sushi domain of IL-15Rα was extended by 13 amino acids of the exon 3 (IL-15Rαsushi+), thereby increasing the binding affinity for IL-15 (33). Although this complex showed reduced bioactivity on high-affinity receptor IL-15Rαβγ–expressing cell lines, the combination of IL-15 and IL-15Rαsushi+ into a fusion protein (RLI) retrieved a molecule that presented maximal enhancement of IL-15Rβγ stimulation without losing stimulatory capacity for IL-15Rαβγ (32, 33). Furthermore, the fusion protein (RLI) showed efficiency in vivo by inducing antitumor effects in diverse mouse models (26). Therefore, for the generation of the targeted scFv36_RD_IL-15 fusion protein, we decided to relay on the RLI model with the IL-15Rαsushi+ variant building the IL-15Rα domain (RD). We could show that the fusion to the scFv component did not interfere with the cytokine activity of IL-15 in combination with the RD fragment, as our untargeted control construct RD_IL-15 displayed similar activity than scFv_RD_IL-15 in solution. On the other hand, fusion of the scFv to the cytokine component and or the receptor unit did not impede specific binding of the construct to the tumor target. Thus, scFv_RD_IL-15 and scFv_IL-15 bound with similar affinity to FAP-expressing cells. Furthermore, we analyzed the cytokine activity of the antibody fusion proteins in the context of intermediate (IL-15Rβγ) and high-affinity receptor IL-15Rαβγ–expressing cell lines. In solution, constructs presenting IL-15 in combination with the IL-15Rα domain (RD) stimulated more efficiently the proliferation of IL-15Rβγ–expressing Mo7e cells. This observation was in accordance with the data reported for the (RLI) fusion protein and the IL-15/IL-15Rαsushi+ complex, showing enhanced cytokine effect on Mo7e cells in comparison with IL-15 alone (32, 33). In regard to the stimulation of the high-affinity receptor IL-15Rαβγ, the presence of the RD fragment in the fusion proteins decreased the cytokine activity of the constructs in solution, independently of the presence of the scFv. Thus, proliferation of CTLL-2 (IL-15Rαβγ) cells was induced with picomolar concentrations of scFv_IL-15, but rather nanomolar concentrations of scFv_RD_IL-15 and RD_IL-15. This resembles the results described for soluble complexes of IL-15 and IL-15Rαsushi+ on high-affinity receptor–expressing cells, in which the extension of the sushi domain interferes with IL-15 receptor stimulation, but differs from the stimulatory effect described for the RLI fusion protein that showed to be equivalent or even better than IL-15 in terms of IL-15Rαβγ activation (33, 26). The latter might be due to the particular cell system used (Kit225 expressing intermediate and high-affinity receptors, in which intermediate receptor activation is specifically blocked by an antibody), although differences in the sterical presentation inherent to the configuration of the particular recombinant fusion protein cannot be excluded. Nevertheless, the inhibitory effect in cytokine activity observed for scFv_RD_IL-15 in solution was abolished in the targeted form, that is, bound to B16-FAP cells, presenting RD_IL-15 in a membrane-bound form. Here, scFv_RD_IL-15 as well as scFv_IL-15 showed equivalent stimulatory potential on CTLL-2 cells. Thus, for targeted constructs, effective trans-presentation of IL-15 to IL-15Rαβγ–expressing cells appeared independent of the association to RD.

Resting NK and naive T cells express the intermediate affinity IL-15Rβγ phenotype (10, 22, 34). Thus, proliferation assays with unstimulated PBMCs retrieved similar results than those obtained with the Mo7e (IL-15Rβγ) cell line in response to the fusion proteins in solution, achieving stronger effects with scFv_RD_IL-15 than with scFv_IL-15. NK as well as T cells (CD8⁺ T cells and CD4⁺ T cells) were identified to be responsive, following this activation pattern. Overall higher proliferation response of CD8⁺ T cells and NK cells in comparison with CD4⁺ T cells was observed, correlating with the report of very low IL-15Rβ expression levels on naïve CD4⁺ T cells (35). In addition, better performance of scFv_RD_IL-15 in comparison with scFv_IL-15 was also determined for the constructs in targeted form, that is, bound to FAP-expressing tumor cells, enhancing proliferation and cytotoxicity of activated T cells. Considering that for effector T cells high expression levels of IL-15Rα and therefore potential expression of high and intermediate affinity IL-15 receptors have been described (10), our results likely reflect the target-directed trans-presentation of RD_IL-15 to both receptor types. Even in this case, better performance of scFv_RD_IL-15 in comparison with scFv_IL-15 appears reasonable, considering that in regard to IL-15Rαβγ activation, both constructs showed in targeted form comparable cytokine activity on CTLL-2 cells (IL-15Rαβγ), while stimulation of IL-15Rβγ is clearly favored by the presence of RD in the construct.

In vivo analysis of scFv_RD_IL-15, scFv_IL-15, and RD_IL-15 in the B16-FAP lung metastasis mouse model resulted in clear antitumor effects for all 3 constructs, corroborating the antitumor potential of this cytokine. The best result was achieved with scFv_RD_IL-15, followed by scFv_IL-15 and RD_IL-15 to similar extent. Pharmacokinetic studies in mice indicated small differences between the constructs in half-life and AUC, favoring scFv_RD_IL-15. According to the in vitro data obtained with these constructs, cytokine activity could be remarkably enhanced by presenting IL-15 in association with the receptor α domain to IL-15Rβγ–expressing cells. This effect was also reported for the RLI fusion protein in comparison with recombinant IL-15 (approximately factor 500; ref. 26). Also here, in vivo analysis in a B16F10/lung metastasis model revealed a stronger antitumor effect for RLI than for IL-15. Nevertheless, pharmacokinetic analysis of RLI and IL-15 indicated a 6-fold and 34-fold increase in half-life and AUC for RLI, respectively, suggesting that improved pharmacokinetic properties as well as enhanced cytokine activity might contribute to the stronger antitumor effect observed for RLI. In our study, scFv_IL-15 and RD_IL-15 achieved similar antitumor effects; an outcome probably influenced by the tumor targeting properties of scFv_IL-15. Enhanced antitumor activity by directing IL-15 to the tumor in form of an antibody–cytokine fusion protein has been described previously (18). Here, IL-15 was fused to a fibronectin extra domain-B (EDB)-specific antibody in a diabody format, resulting in a homodimeric molecule. Accumulation at the tumor site as well as target-mediated tumor growth inhibition was shown in an s.c. F9 tumor mouse model, ascribing a crucial role for CD8⁺ T cells in the therapeutic effect. Tumor directed targeting of IL-15/IL-15Rα as a complex or fusion protein has not been reported so far. Here, by means of scFv_RD_IL-15, we could show that the strategy of combining tumor targeting by a recombinant antibody and trans-presentation of IL-15 in the IL-15Rα domain context was able to further enhance the antitumor potential of this powerful cytokine. Thus, scFv_RD_IL-15 appears as a promising reagent that sets itself apart from other recombinant molecules developed for the application of IL-15 in cancer immunotherapy.

R.E. Kontermann is a consultant to BioNTech. No potential conflicts of interest were disclosed by the other authors.

Conception and design: D. Muller

Development of methodology: V. Kermer, V. Baum, N. Hornig

Acquisition of data: V. Kermer, V. Baum

Analysis and interpretation of data: V. Kermer, V. Baum, D. Muller

Writing, review, and/or revision of the manuscript: V. Kermer, V. Baum, R.E. Kontermann, D. Muller

Administrative, technical, or material support: N. Hornig, V. Kermer

Study supervision: D. Muller

This work was supported by Deutsche Forschungsgemeinschaft (MU 2956/2-1).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.