IgG-single-chain TRAIL fusion proteins for tumour therapy

Single-chain formats of TNF-related apoptosis inducing ligand (scTRAIL) can serve as effector components of tumour-associated antigen-targeted as well as non-targeted fusion proteins, being characterized by high tumour cell-specific induction of apoptosis through death receptor activation. We studied the suitability of immunoglobulin G as a scaffold for oligovalent and bispecific TRAIL fusion proteins. Thus, we developed novel targeted hexa- and dodecavalent IgG-scTRAIL molecules by fusing scTRAIL to the C-terminus of either light (LC-scTRAIL) or heavy immunoglobulin chain (HC-scTRAIL), or to both ends (LC/HC-scTRAIL) of the anti-EGFR IgG antibody hu225. The binding specificity to EGFR and death receptors was retained in all IgG-scTRAIL formats and translated into high antigen-specific bioactivity on EGFR-positive Colo205, HCT116 and WM1366 tumour cell lines, with or without sensitization to apoptosis by bortezomib. In vivo, therapeutic potential was assessed for one of the targeted variants, HC-scTRAIL, compared to the non-targeted Fc-scTRAIL. Both molecules showed a significant reduction of tumour volume and synergism with a Smac mimetic in a Colo205 xenograft tumour model. The IgG-scTRAIL format allows directing a defined, highly bioactive form of TRAIL to a wide variety of tumour antigens, enabling customized solutions for a patient-specific targeted cancer therapy with a reduced risk of side effects.


Results
Design and production of anti-EGFR IgG-single-chain TRAIL fusion proteins. In this study, we used a humanized version (hu225) of EGFR-blocking antibody cetuximab as a basis to generate IgG-singlechain TRAIL (IgG-scTRAIL) fusion proteins. A scTRAIL derivative combining high bioactivity with enhanced thermostability (scTRAIL-FAVSGAA) 10 was fused either to the C-terminal end of the light chain or the heavy chain comprising an effector-deficient Fc region. Moieties were joined by a peptide linker of 15 amino acid residues comprising two N-glycosylation sites (Fig. 1a). Three IgG-scTRAIL variants, characterized by hexavalent (LC-scTRAIL, HC-scTRAIL) or dodecavalent arrangement (LC/HC-scTRAIL) of scTRAIL, were generated (Fig. 1b). Upon transient expression and subsequent anti-FLAG affinity chromatography purification, all variants yielded soluble protein, composed of full-length polypeptide chains with the expected molecular masses and only low amounts of impurities or degradation products, as proven by reducing SDS-PAGE (Fig. 1c). Under non-reducing conditions, Fc-mediated dimerization of the molecules could be confirmed, although fusion of scTRAIL to the light chain resulted in a more diffuse pattern of bands. The large, dodecavalent variant LC/HC-scTRAIL (~375 kDa) produced lower yields than the hexavalent variants LC-scTRAIL and HC-scTRAIL (~262 kDa). Thus, yields of up to ∼9 mg purified protein per litre cell culture supernatant were obtained for HC-scTRAIL (Table 1). Importantly, all proteins eluted essentially as single peaks in size-exclusion chromatography (SEC), similar to the humanized IgG (hu225 IgG) produced under the same conditions (Fig. 1d).
EGFR-specific binding of IgG-scTRAIL fusion proteins. As shown by flow cytometry analysis on EGFR-positive human colon carcinoma cell lines Colo205 and HCT116, hu225 IgG and cetuximab bound with similar EC 50 values to the target cells, demonstrating that the humanized IgG retained the EGFR binding activity (Supplemental Fig. S1). Next, we analysed whether fusion with scTRAIL has an influence on the binding properties of hu225 IgG to Colo205 and HCT116 (Fig. 2a, Table 2). Compared to hu225 IgG, all scTRAIL fusion protein variants bound with slightly higher EC 50 values, but still within the sub-nanomolar range, suggesting that the affinity of scTRAIL to its cognate receptors has no major impact on the overall avidity of the proteins to EGFR + cells at 4 °C. Interestingly, amongst the two hexavalent configurations, HC-scTRAIL exhibited EC 50 values close to those of the hu225 IgG. Upon competition of binding of HC-scTRAIL to Colo205 cells using a 200-fold molar excess of cetuximab or TRAILR2-Fc, respectively, a contribution of TRAILR binding and EGFR binding could clearly be confirmed, with the latter having the dominant activity (Fig. 2b). In an EGFR ELISA, sub-nanomolar EC 50 values were determined for binding of all IgG-scTRAIL fusion proteins to immobilized EGFR-Fc, although EC 50 values were approximately 2-to 3-fold increased, compared to hu225 IgG (Fig. 2c, Table 2). Regarding TRAIL receptor binding, HC-scTRAIL bound with significantly lower EC 50 values to TRAILR1 compared to LC-scTRAIL (P < 0.05), LC/HC-scTRAIL (P < 0.001) and Fc-scTRAIL-FAVSGAA (P < 0.05). Compared to TRAILR1, a stronger binding of the scTRAIL fusion proteins was observed for TRAILR2 with approximately 4to 8-fold lower EC 50 values (Fig. 2d, Table 2, Supplemental Fig. S2). Taken together, all three IgG-scTRAIL fusion proteins bound specifically and with high affinity to EGFR and both proapoptotic TRAIL receptors.

Cell death induction by IgG-scTRAIL fusion proteins in tumour cells in vitro.
Cell death-inducing activity of the EGFR-targeted IgG-scTRAIL fusion proteins in comparison to a non-targeted, hexavalent variant of the Fc-scTRAIL format (Fc-scTRAIL-FAVSGAA) was analysed in vitro using the EGFR + tumour cell lines Colo205, HCT116 (both colon carcinoma) and WM1366 (malignant melanoma) 20 (Fig. 3, Table 3). IgG-scTRAIL variants with a hexavalent TRAIL configuration (LC-scTRAIL and HC-scTRAIL) showed ~5-to ~22-fold lower EC 50 values and therefore an increased bioactivity compared to Fc-scTRAIL-FAVSGAA, suggesting a clear benefit of EGFR targeting in terms of cell death induction by hexavalent TRAIL formats. Importantly, competition of EGFR binding by cetuximab abrogated the targeting effect completely, resulting in bioactivities at the level of Fc-scTRAIL-FAVSGAA. As expected from the increase of TRAIL valence, dodecavalent LC/HC-scTRAIL showed, when normalized to scTRAIL units, approximately 2-to 5-fold higher bioactivity compared to the two hexavalent formats. Interestingly, cetuximab only partially blocked the bioactivity of the LC/HC-scTRAIL on the tested tumour cell lines. Moreover, the co-incubation of the scTRAIL fusion proteins with the clinically established proteasome inhibitor bortezomib resulted in an up to 5-fold increase of bioactivity (Table 3 In vitro stability and pharmacokinetics of HC-scTRAIL. Due to its favourable characteristics in terms of expression titres, receptor binding, bioactivity and molecule size, we selected HC-scTRAIL for further studies and focused first on protein stability in vitro. Thermal stability of HC-scTRAIL protein moieties was analysed by differential scanning calorimetry revealing a first melting point at an onset temperature of 54 °C, a second at 71.5 °C, and a third at 77 °C (Fig. 4a). The first melting point can be attributed to the denaturation of the scTRAIL moiety as the most thermosensitive component of the molecule. This was confirmed by the thermogram of Fc-scTRAIL-FAVSGAA, which showed the same first melting point. This fusion protein showed a second melting point of 77 °C, which can be attributed to the Fc portion, corresponding to the third melting point of HC-scTRAIL. The second melting point of HC-scTRAIL is most likely caused by denaturation of the Fab domains.
The plasma stability of HC-scTRAIL and Fc-scTRAIL-FAVSGAA was assayed via cell death assay and ELISA (Fig. 4b, Supplemental Fig. S4). After one week of incubation at 37 °C in human blood plasma, the bioactivity of both proteins was at least 50%, which corresponded to ~70-90% intact protein as measured by ELISA.
Pharmacokinetic properties of IgG-scTRAIL (HC-scTRAIL) were determined in immunocompetent CD-1 mice receiving a single dose intravenous (i.v.) injection (Fig. 4c). The protein showed a terminal half-life t 1/2 β of 16.1 ± 2.6 h and an area under the curve (AUC) of 76 ± 11 (µg/ml)h. Anti-tumour activity of HC-scTRAIL in Colo205 mouse xenografts. Finally, we investigated the anti-tumour activity of HC-scTRAIL in the established nu/nu mouse xenograft model using subcutaneously implanted Colo205 cells (Fig. 5). When the tumours reached an average size of 100 mm 3 , six doses of HC-scTRAIL or Fc-scTRAIL-FAVSGAA as reference (0.3 nmol each) were administered i.v. twice a week (first cycle). In contrast to the progressively growing tumours of the PBS control group, both proteins inhibited the growth of the tumours significantly. In a second cycle, starting from day 39, animals were treated four times with a combination of scTRAIL fusion protein and intraperitoneally injected Smac mimetic SM83, which is known to synergistically enhance TRAIL-induced cell death 21 . Upon this combination treatment, we observed an additional ~50% reduction of the average tumour sizes until volumes of ~80 mm 3 were reached at day 53. However, no difference regarding the monitored tumour volumes could be detected between the groups treated with EGFR-targeted HC-scTRAIL and non-targeted Fc-scTRAIL-FAVSGAA. Furthermore, no loss of body weight was observed, indicating that the administered doses of scTRAIL fusion proteins were well tolerated (Supplemental Fig. S5).

Discussion
In this study we developed a novel, modular platform utilizing human IgG1 for generation of multivalent single-chain derivatives of the apoptosis-inducing TNF homology domain of TRAIL in combination with targeting of tumour-associated antigens, here demonstrated for EGFR. Fusing scTRAIL either to the C-terminus of the HC or LC, or to both chains of an IgG results in hexavalent or dodecavalent configurations of TRAIL. Our data demonstrate that EGFR binding in hexavalent LC-scTRAIL and HC-scTRAIL is functional and, compared to non-targeted hexavalent Fc-scTRAIL molecules, resulted in an enhanced bioactivity in three EGFR-expressing tested tumour cell lines in vitro. LC-scTRAIL and HC-scTRAIL fusion proteins exhibited similar EC 50 values of cell death induction on these cell lines, indicating that fusion of the scTRAIL moieties to the C-terminus of both, the light and the heavy chain is suitable for generation of hexavalent IgG-scTRAIL fusion proteins. Furthermore, EGFR-dependent increase of apoptosis of tumour cells was independent of the sterical configuration of the molecule, i.e., the position of scTRAIL in the IgG molecule, as shown by competition with cetuximab. In addition to the hexavalent IgG-scTRAIL formats, we also showed that dodecavalent IgG-scTRAIL fusion proteins comprising 4 scTRAIL units can be produced as soluble proteins. Of note, this dodecavalent IgG-scTRAIL fusion protein, LC/HC-scTRAIL, exceeded the bioactivity of the hexavalent IgG-scTRAIL molecules. Most likely, this can be attributed to a fostered clustering of death receptors and consequently a more efficient DISC formation, especially of TRAILR2 [22][23][24] .
Competition of LC/HC-scTRAIL with cetuximab only revealed a partial reduction of the bioactivity (e.g., approximately 2-fold on Colo205) compared to the hexavalent IgG-scTRAIL fusion proteins (~4-fold for LC-scTRAIL, ~7-fold for HC-scTRAIL on Colo205). In principle, cetuximab-competed LC/HC-scTRAIL was not expected to reach the EC 50 values of the non-targeted Fc-scTRAIL fusion protein, because both molecules have different TRAIL valences. However, binding of the dodecavalent LC/HC-scTRAIL to EGFR and TRAIL receptors on EGFR + tumour cells was shown to be similar compared to hexavalent formats, suggesting a rather stronger reduction of bioactivity than observed upon competition with cetuximab. In fact, this finding led us hypothesize that TRAIL valences higher than six may dominate over bivalent EGFR targeting. In this regard, an anti-proliferative effect of the cetuximab-derived hu225 IgG can be excluded for two reasons: (i) the setting and the short duration (16 h) of the cytotoxicity assay performed here were chosen to assess TRAIL-induced apoptosis and are rather not suitable to monitor tumour cell growth, and (ii) all analysed tumour cell lines carry activating mutations in MAPK pathway proteins downstream of EGFR (WM1366: NRAS mut , Colo205: BRAF mut , HCT116: KRAS mut ), thereby abolishing anti-proliferative/pro-apoptotic effects of cetuximab due to constitutively active MAPK signalling which is independent of growth factor stimulation. Mechanistically, we therefore propose that the in vitro targeting effect of IgG-scTRAIL in the cell lines investigated here is predominantly mediated by a faster and more intense caspase activation, putatively in consequence of a rapid formation and/or stabilisation of TRAIL-TRAILR complexes, which is in line with observations from EGFR-targeted scFv-Fc-scTRAIL 17 . In principle, the hu225 V H and V L fragments used in IgG-scTRAIL have been shown to inhibit, as demonstrated for the Db-scTRAIL format, EGFR tyrosine autophosphorylation in a similar manner like cetuximab 13,21 , which can potentially result in additional growth inhibitory effects in EGFR/MAPK pathway wild-type tumour cells. However, the relevance of such effects for tumour therapy with IgG-scTRAIL is, in view of the much faster and robust apoptosis induction triggered by TRAIL, presumably insignificant.
For HC-scTRAIL we determined a terminal half-life in mice of approximately 16 h, which is slightly longer than the half-life of Fc-scTRAIL molecules but similar to a previously generated scFv-Fc-scTRAIL fusion protein also targeting the EGFR 17 . This supports recent findings demonstrating the importance of the Fc moiety for prolonged serum half-life, presumably by implementing FcRn-mediated recycling 17 . Interestingly, the HC-scTRAIL fusion protein, exhibiting a molecular mass and a Stokes radius exceeding that of the corresponding IgG, did not reach the long half-life of the IgG of approximately 9 days in mice 17 . Comparison of Fc-scTRAIL molecules comprising either wild-type or receptor-binding diminished scTRAIL units indicated a contribution of TRAIL-receptor-independent clearance effects to the half-life of scTRAIL fusion proteins in mice 17 . Irrespective of the mechanisms affecting in vivo half-life, the IgG-scTRAIL appears as a promising format for a highly active TRAIL therapeutic.  In a Colo205 xenograft tumour model, HC-scTRAIL demonstrated strong anti-tumour activity which was similar to that of a non-targeted Fc-scTRAIL. This finding is in accordance with recent results comparing an EGFR-targeting scFv-Fc-scTRAIL with Fc-scTRAIL in the same Colo205 model 17 . The apparent discrepancy to the in vitro data, demonstrating superior activity of tumour targeted versus non targeted hexavalent TRAIL molecules, but similar therapeutic potency in this particular in vivo tumour model is currently not fully understood, but important parameters begin to emerge. Thus both, antigen density and affinity of the targeting antibody in relation to the affinity of TRAIL to its cognate death receptors appear to be important 19 . Further studies using other in vivo tumour models are necessary to conclude on the benefit of targeting TRAIL to tumour cells for a maximum therapeutic output and safety. Especially for the dodecavalent IgG-scTRAIL format the role of tumour targeting for in vivo anti-tumour activity as well as safety appears of particular relevance, because excessive death receptor clustering may cause apoptosis in non-malignant, death receptor-positive tissues.
We also provide further data emphasizing the importance of combined therapy of IgG-scTRAIL with apoptosis sensitizers such as bortezomib and Smac mimetic SM83 to enhance cell death and sustained tumour eradication    in vivo, respectively. Particularly regarding the recently reported pro-tumourigenic and pro-metastatic roles of the TRAIL/TRAILR system in TRAIL resistant tumour cells 25,26 , future work will show whether appropriate combinations of TRAIL formats like IgG-scTRAIL with sensitizers will result in complete and lasting tumour remissions.  Table 3. EC 50 values (pM scTRAIL units) of protein bioactivity on EGFR-positive tumour cells (n = 3, mean ± S. D.). In conclusion, we provide evidence for the therapeutic potential of a novel IgG-based format for the generation of multivalent scTRAIL fusion proteins. By realization of two bioactivity-enhancing principles in one molecule, i.e. hexa-or dodecavalent configuration of TRAIL and binding to EGFR-positive tumour cells, IgG-scTRAIL combines efficient cell death induction in tumour cells with targeted delivery. Though demonstrated here for EGFR targeting, the antigen versatility of the IgG-scTRAIL format facilitates the generation of tailor-made molecules to fight different cancer entities by addressing respective tumour-associated antigens on the cell surface. Moreover, considering the ambivalent role of TRAIL in certain scenarios, antigen targeting by TRAIL-based protein therapeutics could be a suitable strategy to ensure systemic safety and tumour cell-restricted bioactivity. Due to the heteromeric composition of the IgG building block, a future perspective is to combine scTRAIL moieties with a second, e.g. immune-stimulating, effector moiety within one covalently linked heterodimer, thus creating a multifunctional molecule for cancer treatment.

Methods
Cell lines and materials. HEK293, Colo205 and HCT116 cell lines were purchased from ATCC. The WM1366 cell line was generated at the WISTAR Institute. HEK293 were cultured in RPMI 1640 medium (Thermo Fisher) supplemented with 5% foetal bovine serum (FBS Premium or FBS Brazil One, PAN Biotech). Colo205, HCT116 and WM1366 were cultured in RPMI 1640 medium with 10% FBS and Pen-Strep (Thermo Fisher). During production of recombinant proteins, HEK293 cells were cultivated in Opti-MEM ™ I (Thermo Fisher). HEK293-S cells were kindly provided by Dr. Verena Berger (Baliopharm GmbH, Jülich, Germany) and cultivated in Erlenmeyer shaker flasks in FreeStyle ™ F17 Expression Medium (Thermo Fisher) supplemented with GlutaMAX ™ (Thermo Fisher, 20 ml/L medium) and Kolliphor ® P 188 (Sigma-Aldrich, 0.1% w/v). All cell lines were cultivated at 37 °C with 5% CO 2 . Bortezomib was from UBPBio. Polyethylenimine (PEI) was purchased from Polysciences GmbH and a 1 mg/ml stock solution was prepared as described 27 . Tryptone N1 was from Organotechnie. Human plasma was obtained from the blood bank of the Klinikum Stuttgart. Cetuximab was provided by Dr. Thomas Mürdter (IKP, Stuttgart) and SM83 was provided by BioNTech (Mainz, Germany).
Production and purification of recombinant proteins. In preparation for transient production of scTRAIL proteins, HEK293 cells were grown to 50-75% confluence in T175 flasks and on day of transfection, the culture medium was replaced by fresh medium. 50 µg plasmid DNA and 150 µg PEI were each mixed in 2.5 ml Opti-MEM ™ I, combined and vortexed for 3 × 1 s, incubated for 20 min at room temperature and given to the flask, followed by careful swinging. 16 h after transfection, the medium was replaced by Opti-MEM ™ I medium + 50 µM ZnCl 2 and cells were cultivated for 3 days. For extended protein productions, genetically engineered HEK293-S suspension cells (M. Siegemund, unpublished data) were transiently transfected. The cells were grown to 1.5-2.0 × 10 6 cells/ml and transfection was done by diluting 1 µg plasmid DNA/ ml culture and 3 µg PEI/ml culture each in fresh culture medium (1/20 of culture volume). Both solutions were combined, treated as described above and given to the culture flask, followed by immediate shaking. 24 hours after, 0.5% tryptone N1 and 10 µM ZnCl 2 were added and cells were grown for another 5 days. All cultures were harvested by centrifugation at 2,000 × g for 30 min at 4 °C and proteins were purified from supernatants by anti-FLAG affinity chromatography as described 30 . To this, supernatants were incubated with anti-FLAG ® M2 Affinity Gel (Sigma-Aldrich, 3 ml bead volume/500 ml supernatant) for at least 2 h at 4 °C on a roller mixer. Beads were collected in an empty column and washed with 5 CV 1 × TBS. Bound proteins were eluted with 7 CV 100 µg/ml DYKDDDDK peptide/1 × TBS (peptides&elephants). After dialysis in 1 × PBS, eluates were concentrated with Vivaspin ® 20 devices, 50 kDa MWCO (Sartorius). In combination with production in genetically engineered HEK293-S cells, HC-scTRAIL and Fc-scTRAIL-FAVSGAA were further purified by preparative gel filtration on a Superdex ® 200 10/300 GL column (GE Healthcare). Protein concentrations were determined spectrophotometrically using the calculated extinction coefficients. Aliquots were stored at −80 °C. TRAILR1-Fc, TRAILR2-Fc and EGFR-Fc were produced and purified as previously described 15 . Biochemical and biophysical protein analysis. Purified proteins were analysed by SDS-PAGE under non-reducing and reducing conditions, followed by staining with InstantBlue ™ (Expedeon). For SEC, a TSKgel ® SuperSW mAb HR, 7.8 × 300 mm column (Tosoh) was equilibrated in 0.1 M Na 2 HPO 4 /NaH 2 PO 4 , 0.1 M Na 2 SO 4 , pH 6.7 and proteins were eluted at a flow rate of 0.5 ml/min. The globular proteins thyroglobulin (669 kDa, R s 8.5 nm), apoferritin (443 kDa, R s 6.1 nm), β-amylase (200 kDa, R s 5.4 nm), BSA (66 kDa, R s 3.55 nm) and carbonic anhydrase (29 kDa, R s 2.35 nm) were used as references. Denaturation temperatures of proteins were measured by differential scanning calorimetry using a MicroCal DSC instrument (Malvern).
ELISA. 96-well plates were coated with TRAILR1-Fc, TRAILR2-Fc (200 ng/well in 0.1 M sodium carbonate buffer, pH 9.5) or EGFR-Fc (300 ng/well in PBS) overnight at 4 °C and then blocked with 3% (w/v) skim milk powder in PBS (MPBS). The protein samples were diluted in MPBS, titrated 1:3 in duplicates starting from 150 nM (TRAILR ELISA) or 50 nM (EGFR ELISA) and incubated on the receptor-Fc-coated plates for 2 h at room temperature on a shaker. Relevant for TRAILR binding, molar concentrations of the samples were normalized to one scTRAIL unit. Bound protein was for TRAILR ELISA detected by 1:15,000 in MPBS diluted monoclonal anti-FLAG ® M2-peroxidase (HRP) antibody (Sigma-Aldrich) or for EGFR ELISA by 1:10,000 in MPBS diluted anti-human IgG (Fab specific)-peroxidase antibody (Sigma-Aldrich) in combination with 3,3′,5,5′-tetramethylbenzidine (TMB) as substrate (0.1 mg/ml TMB, 100 mM sodium acetate buffer, pH 6.0, 0.006% H 2 O 2 ). Reactions were stopped with 50 µl of 2 M sulfuric acid and absorbance was measured at 450 nm. Data were fitted with Prism (GraphPad Software) in order to calculate EC 50 values ± SD from three independent experiments. EC 50 values of binding were analysed by unpaired t-test or one-way ANOVA and Tukey post hoc tests for statistical evidence.
Flow cytometry. Cells were trypsinised and then washed once and re-suspended in FACS buffer (PBS, 2% FBS, 0.05% NaN 3 ). 1.5 × 10 5 cells were incubated for 2 h on ice in presence of serially diluted proteins in FACS buffer starting from 50 nM or without protein (control). For competition experiments, cells were incubated with a 200-fold molar excess of cetuximab or TRAILR2-Fc for 30 min on ice before the proteins were added. The cells were washed two times with ice-cold FACS buffer, followed by incubation of protein samples and controls with 1:100 in FACS buffer diluted anti-DYKDDDDK-PE labelled antibody (Miltenyi Biotec) on ice for 1 h. Cetuximab and hu225 IgG binding was detected with 1:500 in FACS buffer diluted anti-human IgG (γ chain specific)-R-phycoerythrin antibody (Sigma-Aldrich). After two washing steps, cells were analysed on a MACSQuant ® Analyzer 10 equipped with a 585/40 nm filter or a MACSQuant ® VYB with 586/15 nm filter (Miltenyi Biotec). Data were fitted with Prism from three independent binding curves.
Cell death assays. Colo205 (4 × 10 4 /well), HCT116 (1.5 × 10 4 /well) and WM1366 (2 × 10 4 /well) cells were seeded in 100 µl culture medium in 96-well plates and grown for 24 h, followed by treatment with serial dilutions (1:3) of proteins in triplicates. Molar concentrations were normalized to one scTRAIL unit. For positive control, cells were lysed with 0.5% Triton ™ X-100. Cell death assays were performed without sensitization of carcinoma cells or in the presence of the proteasome inhibitor bortezomib (250 ng/ml for Colo205, 5 ng/ml for HCT116, 250 SCientifiC REPORTS | (2018) 8:7808 | DOI:10.1038/s41598-018-24450-8 ng/ml for WM1366; UBPBio), which was added 30 min prior incubation with the proteins. After 16 h of incubation, cell viability was determined by crystal violet staining. Alternatively, to demonstrate EGFR-dependent enhancement of cell death, cells were pre-incubated for 30 min with 70 nM of cetuximab for competition. Data were analysed with Prism and EC 50 values ± SD were calculated, based on three independent experiments. Plasma stability assay. Proteins were incubated at a concentration of 0.5 µM in 50% human blood plasma.
The bioactivity was analysed by cell death assay on Colo205 cells, as described before, without incubation of the samples at 37 °C (control) or after 1, 3 and 7 days at 37 °C, respectively. EC 50 values ± SD were calculated with Prism from triplicates and normalized to the control.
Pharmacokinetic studies. Animal care and all experiments performed were in accordance with federal and European guidelines and have been approved by university (animal welfare officer) and state authorities (Regierungspräsidium Stuttgart). An i.v. injection of 25 µg of HC-scTRAIL was administered to female CD-1 mice (Charles River, 8 weeks old, 3 animals). Blood samples (50 µl) were taken at 0.05, 0.5, 1, 2, 6, 24, 72 and 168 hours and incubated on ice for 30 minutes. Clotted blood was centrifuged at 13,000 × g for 30 minutes, 4 °C and serum samples were stored at −20 °C. The serum concentration of HC-scTRAIL was analysed with BD OptEIA ™ Human TRAIL ELISA Set (BD) according to the manufacturer's instructions. The first value (0.05 h) was set to 100%. The initial (t ½ α) and terminal (t ½ β) half-lives and the bioavailability (area under the curve, AUC) were calculated with MS Excel. Xenograft mouse tumour model. Female NMRI nu/nu mice (Charles River, 12 weeks old) received s.c. injections of 3×10 6 Colo205 cells in 100 µl PBS at the left and right dorsal sides. Treatment started 14 days after tumour cell inoculation when tumours reached an average volume of 100 mm 3 . 0.3 nmol HC-scTRAIL or Fc-scTRAIL-FAVSGAA were administered twice a week in 100 µl PBS for a total of six doses. The control group received six doses of 100 µl PBS. In a second round of treatment, 100 µg Smac mimetic SM83 was co-administered i.p., together with a total of 4 i.v. doses of PBS or 0.3 nmol of the proteins twice a week. Tumour volume was monitored as described 9 . One-way ANOVA and Tukey post hoc tests were performed for statistical analysis.
Data availability statement. All data generated or analysed during this study are included in this published article (and its Supplementary Information files).