A versatile pretargeting approach for tumour-selective delivery and activation of TNF superfamily members

TNFR superfamily (TNFRSF) members have important immunoregulatory functions and are of clear interest for cancer immunotherapy. Various TNFRSF agonists have been clinically evaluated, but have met with limited efficacy and/or toxicity. Recent insights indicate that ‘first-generation’ TNFRSF agonists lack efficacy as they do not effectively cross-link their corresponding receptor. Reversely, ubiquitous TNFRSF receptor(s) cross-linking by CD40 and Fas agonistic antibodies resulted in dose-limiting liver toxicity. To overcome these issues, we developed a novel pretargeting strategy exploiting recombinant fusion proteins in which a soluble form of TRAIL, FasL or CD40L is genetically fused to a high-affinity anti-fluorescein scFv antibody fragment (scFvFITC). Fusion proteins scFvFITC:sTRAIL and scFvFITC:sFasL induced potent target antigen-restricted apoptosis in a panel of cancer lines and in primary patient-derived cancer cells, but only when pretargeted with a relevant FITC-labelled antitumour antibody. In a similar pretargeting setting, fusion protein scFvFITC:sCD40L promoted tumour-directed maturation of immature monocyte-derived dendritic cells (iDCs). This novel tumour-selective pretargeting approach may be used to improve efficacy and/or reduce possible off-target toxicity of TNFSF ligands for cancer immunotherapy.

To address these issues, we here report on a two-step approach which involves pretargeting of cancer cells with fluorescein-labelled anticancer antibodies, followed by treatment with a recombinant scFv:TNFSF fusion protein with high-affinity binding capacity for fluorescein derivatives. These scFv:FITC:sTNFSF fusion proteins only gain full agonistic activity upon binding to cancer cells pretargeted with a FITC-labelled antibody. Using this two-step approach, tumour-selective pro-apoptotic activity of fusion proteins scFvFITC:sTRAIL and scFvFITC:sFasL was achieved towards various cell lines and primary patient-derived cancer cell types. In a similar pretargeting setting, fusion protein scFvFITC:sCD40L promoted tumour-directed maturation of immature monocyte-derived dendritic cells (iDCs).
Analogously, when CD7 pos /CD20 neg Jurkat or CEM T cell leukemic cells were pretargeted with anti-CD7-FITC, treatment with scFvFITC:sTRAIL dose-dependently induced apoptosis, whereas pretargeting with RTX-FITC did not (Fig. 1D). Similarly, scFvFITC:sTRAIL potently induced apoptosis in CD7 pos /CD47 pos /CD20 neg /CD14 neg Jurkat cells when pretargeted with anti-CD7-FITC or anti-CD47-FITC, whereas no apoptosis was induced in the absence of these antibodies or when pretargeted with anti-CD14-FITC ( Fig. 1F and G). Importantly, binding as well as pro-apoptotic activity of scFvFITC:sTRAIL was inhibited when treatment was performed in the presence of excess amount of fluorescein analogue Fluo-A (Fig. 1H, and I). Further, apoptotic activity by scFvFITC:sTRAIL in anti-CD47-FITC pretargeted Jurkat cells was blocked in the presence by TRAIL-neutralizing antibody 2E5 or total caspase-inhibitor zVADfmK (Fig. 1I).
Two step pretargeting with scFvFITC:sTRAIL selectively induces apoptosis in various solid cancer cell types. To further establish utility of the pre-targeting approach, ovarian carcinoma cell line OvCAR-3 (EpCAM pos /CD44 pos ) was pretargeted with anti-EpCAM-FITC or anti-CD44-FITC. Subsequent treatment with scFvFITC:sTRAIL dose-dependently induced apoptosis and reduced cell viability ( Fig. 2A-C). In contrast, no apoptosis or loss in cell viability was detected when cells were pretargeted with irrelevant anti-CD33-FITC antibody or upon treatment with anti-EpCAM-FITC or anti-CD44-FITC alone ( Fig. 2A-C). Similarly, elimination of MCSP pos cancer cells by scFvFITC:sTRAIL was only observed when pretargeted with anti-MCSP-FITC, whereas pretargeting with an irrelevant anti-CD33-FITC antibody failed to do so (Fig. 2D). Thus, target antigen-selective pro-apoptotic activity of scFvFITC:sTRAIL was detected when cancer cells were pretargeted with a relevant first step FITC-labelled anticancer antibody and is applicable to diverse cancer types.
Two step pretargeting with scFvFITC:sTRAIL selectively induces apoptosis in patient-derived cancer cells. In a more clinically-relevant setting, short-term cultures of primary patient-derived ovarian cancer (OC) cells were pretargeted with anti-CD44-FITC, upon which treatment with scFvFITC:sTRAIL induced apoptosis in 6 out of 7 treated samples (Fig. 2E). In contrast, pretargeting with control antibody anti-CD33-FITC did not induce apoptosis in 4 out of 5 samples (Fig. 2E). Similar results were obtained for treatment of patient-derived primary melanoma cells with scFvFITC:sTRAIL after pretargeting with anti-MCSP-FITC, with significant induction of apoptosis in 4 out of 4 samples tested (Fig. 2G). Of note, combination of the FITC-pretargeting scFvFITC:sTRAIL strategy with proteasome inhibitor bortezomib significantly enhanced apoptosis in 2 out of 3 primary OC samples when pretargeted with anti-CD44-FITC (Fig. 2F), whereas cotreatment with vemurafenib enhanced scFvFITC:sTRAIL-induced apoptosis in 4 out of 4 patient-derived primary melanoma cell cultures (Fig. 2H). Together, these results demonstrate that scFvFITC:sTRAIL selectively binds to FITC-tagged cells and can subsequently induce apoptosis in multiple types of cancer cells only when pretargeted with appropriate FITC-labelled anticancer antibody.

Two step pretargeting with scFvFITC:sFasL selectively induces apoptosis in leukaemia cells.
Similar to scFvFITC:sTRAIL, scFvFITC:sFasL selectively bound to Jurkat.CD20 only when pretargeted with RTX-FITC (Fig. 3A), whereas limited binding was detected on parental Jurkat cells (Supplementary data 1C). The binding of scFvFITC:sFasL was abrogated when treatment was performed in the presence of excess molar amounts of Fluo-A (Supplementary data Fig. 1D). In line with this, scFvFITC:sFasL dose-dependently induced apoptosis in Jurkat.CD20 cells when pretargeted with RTX-FITC, whereas Jurkat cells remained largely unaffected (Fig. 3B). This pro-apoptotic activity of scFvFITC:sFasL was abrogated in the presence of excess amounts of Fluo-A or FasL-neutralizing mAb NOK.2 ( Fig. 3C and D). Further, treatment of either Jurkat.CD20 cells or Jurkat cells with scFvFITC:sFasL alone did not induce apoptosis ( Fig. 3C and D). Treatment with scFvFITC:s-FasL dose-dependently induced apoptosis in CD20 pos Z138 B cell non-Hodgkin's lymphoma cells pretargeted with RTX-FITC ( Fig. 3D) with an EC50 value of 14.9 ng/ml. However, treatment with maximum concentration of scFvFITC:sFasL (200 ng/ml) did not induce apoptosis in non-pretargeted Z138 cells. No apoptotic activity was detected towards Z138 when cells were treated with scFvFITC:sFasL after pretargeting with mock antibody anti-CD7-FITC (Fig. 3E). This pro-apoptotic activity of scFvFITC:sFasL was blocked in the presence of excess amounts of Fluo-A or Fas-neutralizing antibody NOK.2 in Z138 and BJAB cells, although cell death in PR1 cells was not inhibited by the latter (Fig. 3F), which may be due to CD20 cross-linking induced apoptosis, to which PR1 cells are sensitive 16 .
Two step pretargeting with scFvFITC:sFasL selectively induces apoptosis in solid tumour cells. Treatment of OvCAR3 ovarian cancer cells pretargeted with anti-EGFR-FITC with scFvFITC:sFasL resulted in dose-dependent induction of apoptosis with a corresponding loss in cancer cell viability ( Fig. 4A and B). In this experiment, the EC50 value of scFvFITC:sFasL in OVCAR3 cells was calculated to be as low as 100 ng/ ml, whereas no apoptosis was induced in OVCAR3 cells upon even when treatment was at the maximum dose of scFvFITC:sFasL. Tumour-selective pro-apoptotic activity of scFvFITC:sFasL was confirmed in a panel of 4 carcinoma cell lines pretargeted with either anti-EpCAM-FITC or anti-CD44-FITC (Fig. 4C). Importantly, analogous treatment with scFvFITC:sFasL also markedly triggered apoptosis in 6 out of 7 primary patient-derived ovarian cancer samples when pretargeted with anti-CD44-FITC, but not with anti-CD33-FITC (4 out of 5) (Fig. 4D). Similarly, in a panel of MCSP-expressing cell lines expressing both MCSP and CD44, treatment with ) and CD20-negative cells (SEM) were pretargeted with RTX-FITC or anti-CD7-FITC followed by scFvFITC:FasL treatment (100 ng/ml) in the presence or absence of Fluo-A or anti-FasL MAb. Apoptosis in all experiments was determined by Annexin V/PI staining. All graphs represent mean ± SD. Statistical analysis was performed using two-way ANOVA (*p < 0.05, **p < 0.01, ***p < 0.001, n.s. not significant). scFvFITC:sFasL induced apoptosis when cells were pretargeted with anti-MCSP-FITC and anti-CD44-FITC, respectively (Fig. 4F). Of note, treatment with scFvFITC:sFasL induced significant apoptosis in 3 out of 3 primary patient-derived melanoma cells pretargeted with anti-MCSP-FITC, whereas no significant increase in apoptosis was observed in cells pretargeted with anti-CD14-FITC (Fig. 4G). Combining scFvFITC:sFasL with bortezomib significantly enhanced apoptosis in 3 out 3 primary ovarian cancer samples when pretargeted with anti-CD44-FITC (Fig. 4E). Similarly, co-treatment with B-RAF inhibitor vemurafenib synergistically triggered cell death in 2 out of 2 primary patient-derived melanoma cell samples (Fig. 4H).
Two step pretargeting with scFvFITC:sCD40L allows for tumour-localized iDC maturation. Of note, it was previously reported that only when applied at high concentrations sCD40L has capacity to induce iDC maturation (28). Indeed, treatment of iDC with 0.5 µg/ml scFvFITC:sCD40L markedly up-regulated expression levels of CD83, CD86 and HLA-DR (Fig. 5C) which was comparable to up-regulation of these markers upon treatment with LPS (5 µg/ml) (Supplementary data Fig. 2D). First, two step pretargeting of scFvFITC:sCD40L was confirmed using Jurkat.CD20 cells treated with RTX-FITC which resulted in strongly increased fluorescence intensity detected upon secondary anti-CD40L-Dye647 staining (Fig. 5B).  . Apoptosis in all experiments was determined by Annexin V/PI staining and cell viability was determined by MTS assay. All graphs represent mean ± SD. Statistical analysis was performed using two-way ANOVA (*p < 0.05, **p < 0.01, ***p < 0.001, n.s. not significant). Next, the two step pretargeting approach was evaluated for its capacity to promote tumour-localized iDC maturation using fusion protein scFvFITC:sCD40L (for a schematic representation see Fig. 5A). To this end, FaDu cells (EpCAM pos /CD33 neg ), pretargeted with anti-EpCAM-FITC were treated with increasing dose of scFvFITC:sCD40L in the presence of iDCs. This treatment scheme indeed led to a dose-dependent induction of IL-12 production by iDCs, whereas no significant increase IL-12 expression levels were detected when pretargeting with anti-EpCAM-FITC was omitted or when pretargeting was with an irrelevant anti-CD33-FITC antibody (Fig. 5D). Of note, upon treatment with scFvFITC:sCD40L prominent clustering of DCs was observed in co-cultures iDCs and FaDu cells, but only when the latter were pretargeted with anti-EpCAM-FITC. No clustering of DCs was observed when in this treatment scheme FaDu cells were not pretargeted with an irrelevant anti-EpCAM-FITC antibody (Fig. 5E). Concordantly, upon treatment with scFvFITC:sCD40L, CD83 levels on DCs were increased in co-cultures iDCs and FaDu cells, but only when the latter were pretargeted with anti-EpCAM-FITC. No increase in CD83 and IL-12 level was observed when in this treatment scheme FaDu cells were pretargeted with an irrelevant anti-CD33-FITC antibody or when treatment was in the presence of excess amounts of Fluo-A ( Fig. 5F and G). Furthermore, analogous scFvFITC:sCD40L treatment schemes in which iDCs were co-cultured with a panel of various cancer cells types significant levels of IL-12 were induced, but only when these cancer types were pretargeted with a FITC-labelled antibody of appropriate specificity (Fig. 5H and Supplementary Table 1). Collectively, these data indicated that scFvFITC:sCD40L promoted antigen-restricted DC maturation in the presence of cancer cells that were pretargeted with a relevant FITC-labelled anticancer antibody preparation.
No off-target toxicity of scFvFITC:sTRAIL, scFvFITC:sFasL and scFvFITC:sCD40L towards human hepatocytes. Hepatocytes are known to be highly sensitive to toxicity induced by cross-linked and/ or aggregated forms of FasL and CD40 agonists. To access for off-target toxicity, IHH hepatocytes (CD44 pos ) were treated with scFvFITC:FasL or scFvFITC:sCD40L. This treatment showed no signs toxicity towards IHH cells. In contrast, apoptosis induction in IHH cells was observed when these were pretargeted with anti-CD44-FITC (Supplementary Fig. 2A).
Next, we wondered whether leukaemia cells displaying scFvFITC:sFasL after appropriate two step pretargeting would induce off-target bystander toxicity in hepatocytes while passing through the liver. To mimic this in vitro, IHH cells were cultured in the presence of Ramos B cells pretargeted (or not) with anti-CD19-FITC or the irrelevant antibody anti-CD7-FITC. Subsequent treatment with scFvFITC:sFasL did not induce apoptosis in IHH cells in any of the treatment conditions assessed (Supplementary Fig. 2B and 2C). These data suggest that even in the continued presence of leukaemia cells displaying a surplus scFvFITC:sFasL molecules, only minimal innocent bystander toxicity is induced in hepatocytes.

Discussion
Here we described a novel two step approach that pretargets cancer cells with a FITC-labelled antibody directed at a tumour-associated cell surface antigen followed by treatment with an essentially inactive scFvFITC:TNFSF fusion protein. As a result of its selective immobilization at the tumour cell surface the inactive scFvFITC:TNFSF fusion protein locally regains its full agonistic activity towards its cognate TNFRSF receptor.
Our data confirm that fusion proteins scFvFITC:sTRAIL and scFvFITC:sFasL, indeed regain potent tumour-selective pro-apoptotic activity towards various types of cancer, both cancer lines and freshly isolated patient-derived primary cancer cells, but only when pretargeted with a relevant FITC-labelled anti-tumour antibody. Similarly, the capacity of fusion protein scFvFITC:sCD40L to promote tumour-directed maturation of immature monocyte-derived dendritic cells (iDCs) was significantly enhanced when tumour cells were pretreated with a relevant FITC-labelled anti-tumour antibody.
Toxicity of TNFSF ligands like FasL and CD40L towards hepatocytes has been a major hurdle for their clinical application. Our results indicated that hepatocytes (IHH cells) were resistant to prolonged treatment with relative high concentrations of scFvFITC:sTRAIL (up to 1000 ng/ml), scFvFITC:sFasL (up to 500 ng/ml) or scFvFITC:sCD40L (up to 500 ng/ml) (Supplementary data Fig. 2A). However, significant apoptosis was induced by scFvFITC:sFasL in CD44 pos IHH cells when pretargeted with an anti-CD44-FITC mAb. This is in line with high sensitivity of hepatocytes to Fas cross-linking agonists. This indicated that also for this pretargeting approach suitability of potential target antigens must be carefully evaluated both in vitro and in vivo to assess possible toxicity towards normal cell types expressing the same target antigen. In this respect, it is of particular interest that the activity of our TNFSF pretargeting approach can be rapidly attenuated by the application of fluorescein in the event of adverse effects. Sodium fluorescein is an FDA-approved fluorescent imaging agent (e.g. Fluorescite) for diagnostic retina and iris angiography 25 . Intravenous administration of 500 mg sodium fluorescein is well tolerated, has fast pharmacokinetics and shows excellent distribution into the interstitial space after. 25 . In a previous study, the activity of anti-FITC chimeric antigen receptor (CAR) T cells was shown to be effectively controlled and attenuated by fluorescein 26,27 .
Obviously, the efficacy of our pretargeting approach is dependent on the expression level of the target antigen of choice. For example, on primary melanoma cells expression of CD44 is higher than MCSP, which translated into a higher efficacy of the scFvFITC:sTRAIL when pretargeted with the anti-CD44 FITC-labelled antibody (Supplementary Fig. 1B and E). Thus, highly over-expressed tumour-associated antigens appear most suited for FITC-based pre-targeting.
It is well established that intratumoural heterogeneity may severely hamper the efficacy of antibody-based therapies directed at one specific target antigen 24 . The FITC-based pretargeting strategy may be used to circumvent this problem by combining two or more different FITC-labelled anticancer antibodies. Studies evaluating the various possible benefits of our pretargeting approach to simultaneously target multiple antigens or different TNFSF ligands to tumour cells, endothelial cells and/or immune cells in the tumour microenvironment are currently on-going.
This pre-targeting approach may be applied to various other TNFSF family members, particularly those that are known to require oligomerization for effective induction of cognate receptor signalling. In this respect, the 4-1BB/4-1BBL signalling axis may be of particular interest as it is crucial in providing co-stimulation to tumour-reactive TILs. Previously, 4-1BBL was shown to require higher order oligomerization to activate 4-1BB signalling 28 . However, ubiquitous 4-1BB activation by agonistic agents is associated with severe liver toxicity in humans 29 . Therefore, 4-1BBL appears of particular interest to be used as TNSF ligand in our pretargeting approach and as such it is currently being explored.
Taken together, in this study we provide proof-of-concept for pretargeting of both pro-apoptotic and co-stimulatory TNFSF family members using readily available FITC-labelled anticancer antibodies. This two-step pretargeting approach may be of use to improve efficacy and/or attenuate possible off-target toxicity of TNFSF ligands for cancer immunotherapy.

Material and Methods
Antibodies and reagents. Cell lines and primary patient-derived cancer cells. Jurkat, SEM, CEM, BJAB, Z138, HT1080, FaDu, A375m, DLD-1 and OVCAR-3 cell lines were purchased from the ATCC (Rockville, MD). Jurkat.CD20 cells were generated by transfection with pCMV-CD20. Cells lines were authenticated by Short Tandem Repeat analysis and routinely screened for mycoplasma. Immortalized human hepatocytes were from Department of Pediatrics, UMCG. Tumour samples from cancer patients were collected at surgery after informed consent. Cells were cultured in RPMI 1640 or DMEM (10% FCS, 37 °C, 5% CO2). Primary melanoma cells were analysed for CD44 and MCSP expression (Supplementary data Fig. 1B and C). The study protocols were approved by the Institutional Ethics Committees of University Medical Centre Groningen (METC 2011.206 and METC 2012/330 respectively) and informed consents were obtained from all donors. All methods were performed in accordance with the relevant guidelines and regulations.
Isolation, preparation and phenotyping of DCs. Peripheral blood mononuclear cells (PBMCs) were obtained from healthy volunteers after informed consent using density gradient centrifugation. Monocytes were isolated from PBMCs using anti-CD14-coated magnetic beads (Miltenyi Biotec, USA). For generation of immature DCs (iDCs), monocytes (3 × 10 6 /ml) were cultured in the presence of 500 U/ml GM-CSF and 1000 U/ml IL-4 for 7 d. As positive control, iDCs were treated with 1 μg/ml LPS for 48 h. DC phenotype was flow cytometrically analysed for CD83, CD14, HLA-DR and CD86 expression.

Assessment apoptosis and cell viability. Phosphatidylserine (PS) exposure was analysed using
Annexin-V-APC. Cell viability was assessed by MTS assay (Promega Benelux, Netherlands). Each experimental and control group was performed in triplicate.
Binding scFvFITC:TNFSF to cancer cells pretargeted with RTX-FITC. Jurkat and Jurkat.CD20 cells were incubated with RTX-FITC at 4 °C for 1 h, followed by two washes with PBS, and then incubated with scFvFITC:TNFSF, Binding was evaluated by flow cytometry using an APC-labelled antibody directed against the respective TNF-ligand at 4 °C for 45 min.
Apoptosis induction. Leukaemia cells (5 × 10 4 /well) and solid cancer cells (3 × 10 4 /well) were seeded in a 48-well plate and pretargeted with the indicated FITC-labelled anti-tumour antibody. Unbound antibody was removed by washing and followed by treatment with medium only, scFvFITC:sTRAIL or scFvFITC:sFasL. Synergistic effects of scFvFITC:sTRAIL with velcade (5 μM) or vemurafenib (10 µM) were evaluated using primary patient-derived cancer cells after pretargeting with the indicated FITC-labelled antibody and treatment with scFvFITC:sTRAIL. After 24 h, apoptosis induction was evaluated as indicated.
Tumour-localized DC maturation by scFvFITC:sCD40L. Tumour-localized maturation of iDC's by scFvFITC:sCD40L was evaluated by pretargeting (or not) cancer cells with the indicated FITC-labelled anticancer antibodies. In short, iDCs (Effector cells) were mixed with cancer (Target) cells at an E:T cell ratio of 5:1 and then treated with scFvFITC:sCD40L. In a control experiment, competing agent Fluo-A was added. After 48 h, DC activation was evaluated by CD83 expression using flow cytometry and IL-12 secretion by ELISA.
SCIENTIfIC REPORTS | 7: 13301 | DOI:10.1038/s41598-017-13530-w Assessment off-target toxicity. Off-target toxicity of the scFvFITC:TNFSF pretargeting approach towards hepatocytes was assessed using CD44 pos immortal human hepatocytes (IHH) 31 subjected in vitro to scFvFITC:T-NFSF alone or in the presence of Ramos cells pretargeted with anti-CD19-FITC. After 24 h apoptosis induction in IHH cells was separately evaluated. Statistical analysis. Statistical analysis was performed by one-way ANOVA followed by Tukey-Kramer post-test or, where indicated, by two-sided Student's t-test using Prism software. P < 0.05 was defined as a statistically significant difference. Where indicated *P < 0.05; **P < 0.01; ***P < 0.001.