Development and preclinical evaluation of cixutumumab drug conjugates in a model of insulin growth factor receptor I (IGF-1R) positive cancer

Overexpression of insulin growth factor receptor type 1 (IGF-1R) is observed in many cancers. Antibody drug conjugates (ADCs) with PEGylated maytansine (PEG6-DM1) show promise in vitro. We developed PEG6-DM1 ADCs with low and high drug to antibody ratios (DAR) using an anti-IGF-1R antibody cixutumumab (IMC-A12). Conjugates with low (cixutumumab-PEG6-DM1-Low) and high (cixutumumab-PEG6-DM1-High) DAR as 3.4 and 7.2, respectively, were generated. QC was performed by UV spectrophotometry, HPLC, bioanalyzer, and biolayer-interferometry. We compared the in vitro binding and internalization rates of the ADCs in IGF-1R-positive MCF-7/Her18 cells. We radiolabeled the ADCs with 111In and used microSPECT/CT imaging and ex vivo biodistribution to understand their in vivo behavior in MCF-7/Her18 xenograft mice. The therapeutic potential of the ADC was studied in vitro and in mouse xenograft. Internalization rates of all ADCs was high and increased over 48 h and EC50 was in the low nanomolar range. MicroSPECT/CT imaging and ex vivo biodistribution showed significantly lower tumor uptake of 111In-cixutumumab-PEG6-DM1-High compared to 111In-cixutumumab-PEG6-DM1-Low and 111In-cixutumumab. Cixutumumab-PEG6-DM1-Low significantly prolonged the survival of mice bearing MCF-7/Her18 xenograft compared with cixutumumab, cixutumumab-PEG6-DM1-High, or the PBS control group. Cixutumumab-PEG6-DM1-Low ADC was more effective. The study highlights the potential utility of cixutumumab-ADCs as theranostics against IGF-1R positive cancers.


Results
Synthesis of drug linker and ADCs. The synthesis of drug linker DM1-MAL-PEG 6 -NHS and immunoconjugates cixutumumab-PEG 6 -DM1-Low, cixutumumab-PEG 6 -DM1-High, and controls are outlined in Supplementary Fig. 1A,B. The drug linker DM1-MAL-PEG 6 -NHS was characterized by NMR and mass spectroscopy. Data from the NMR spectra is in excellent agreement with their assigned structures ( Supplementary  Fig. 2). The mass spectra of these compounds showed molecular ion peaks corresponding to their molecular masses ( Supplementary Fig. 3). The optimal reaction condition for antibody drug conjugation was 100 mM HEPES, pH 8.0, for 2 h at 37 °C, followed by at 4 °C for 20 h which gave the desired DAR. UV spectrophotometer was used to determine DAR and found 3.4 ± 0.3 (n = 10) for cixutumumab-PEG 6 -DM1-Low and 7.2 ± 0.1 (n = 10) for cixutumumab-PEG 6 -DM1-High. The spectra of conjugated and unconjugated IgG were sufficiently different and indicated the absorption maxima of the drug and the antibody. Cixutumumab-PEG 6 -DM1-Low and cixutumumab-PEG 6 -DM1-High showed a distinct difference in the UV absorbance at 254 and 280 nm.

Internalization of immunoconjugates.
A rapid time-dependent increase in red fluorescence was observed with cixutumumab immunoconjugates, but not isotype control antibody or media control, from the first time point (2 h) to end point (48 h) ( Fig. 1C and Supplementary Fig. 7). The red signal was observed in the cytosolic compartment of the cells, indicative of the localization and processing of the antibody in lysosomes and endosomes which was quantified as µm 2 /well. Internalization increased from 2 to 48 h and was highest at 48 h post incubation with cixutumumab-PEG 6 -DM1-Low > cixutumumab-PEG 6 -DM1-High > cixutumumab > DOTAcixutumumab-PEG 6 -DM1-Low > DOTA-cixutumumab > cixutumumab-PEG 6 -DM1-High > DOTA-Ctrl-IgG. The DOTA-modified conjugates were not internalized as much by MCF-7/Her18 cells.

Immunoconjugates
All mice were euthanized after the last imaging time point (144 h p.i.) for biodistribution studies, to validate the in vivo SPECT data (Fig. 4). At 144 h p.i., 111 In-cixutumumab was almost completely cleared from all organs except for the liver (5.4 ± 0.2%IA/g) and kidney (6.9 ± 0.2%IA/g). There was a low residual bone uptake of the 111 Incixutumumab-PEG 6 -DM1-High, and 111 In-cixutumumab-PEG 6 -DM1-Low compared to 111 In-cixutumumab with the highest uptake observed at 144 h time point. The highest absolute tumor uptake (7. 8 ± 0.1%IA/g) was found for 111 In-cixutumumab. The absolute tumor uptake of 111 In-cixutumumab-PEG 6 -DM1-Low (2.9 ± 0.3%IA/g) was lower than that of 111 In-cixutumumab, and higher than 111 In-cixutumumab-PEG 6 -DM1-High (1.8 ± 0.1%IA/g). Tumor-to-muscle ratio by biodistribution for 111 In-cixutumumab was 17.3 ± 0.9 at 144 h p.i. (n = 3; Supplementary Fig. 11).  The one mouse showed some response to cixutumumab therapy evidenced by slowed xenograft growth. In the cixutumumab-PEG 6 -DM1-Low, and cixutumumab-PEG 6 -DM1-High groups, 2/5 mice showed response to therapy that lasted throughout the study (185 days). Tumor growth index of individual mice for the different groups is presented ( Supplementary Fig. 12). No unwanted treatment-related deaths occurred in any group. Mice maintained or gained weight in the course of treatment ( Supplementary Fig. 13). Kaplan Meier survival curves showed significant differences in median survival between groups (Fig. 5B). Median survival in the PBS group was significantly (p < 0.05) lower than for the cixutumumab treatment groups. The median survival of the PBS group was 60 days, while that of cixutumumab group was 71 days. On the other hand, median survival of the cixutumumab-PEG 6 -DM1-Low and cixutumumab-PEG 6 -DM1-High groups 147 and 65 days, respectively.

Discussion
The initial excitement on the use of anti-IGF-1R antibodies to treat cancers has lived up to the hype in patients 9,10 . Therefore, other approaches aimed at targeting this receptor could yield important benefits to patients. Our study is the first to describe a cixutumumab ADC. ADCs show improved efficacy compared with naked antibodies. However, cancer cells invariable develop resistance to ADCs. A major cause of resistance to ADCs is the expression of drug efflux multidrug resistant gene (MDR1) pumps [16][17][18] . Like most small molecule chemotherapeutic agents, DM1 is an MDR1 substrate, however, PEGylated DM1 (PEG-DM1) used in this study has the same potency as DM1 but is not a substrate for MDR1 18,25 . A careful balance between the DAR, therapeutic window and optimum in vivo distribution is needed for ADCs, and this may vary with different antibodies. Zhao et al. prepared ADCs with a DAR of 9 using PEG linkers and demonstrated that the high DAR yielded more effective ADCs in vitro that those with DAR of 3-4 18 . We developed cixutumumab-PEG 6 -DM1-Low and cixutumumab-PEG 6 -DM1-High ADCs with DARs of 3.4 and 7.2, respectively. Although physico-chemical characterization of these PEGylated-DM1 ADCs have been recently done, no studies have characterized these pegylated-DM1 ADCs in real time using imaging technologies. In the current study, we used in vitro assays, non-invasive real-time microSPECT/CT imaging and ex vivo biodistribution to characterize the behavior of PEGylated cixutumumab ADC immunoconjugates by radiolabeling the cixutumumab-PEG 6 -DM1-Low/High with 111 In using a DOTA chelator.
To obtain a high DAR, commonly used linkers such as N-succinimidyl-4-(2-pyridyldithio)butanoate (SPDB) or succinimidyl-4-(N-maleimidomethyl)cyclohexane-1-carboxylate (SMCC) lead to the formation of aggregates that tend to abrogate the binding of the antibody 26,27 . The use of NHS-PEG 6 -DM1 resulted in ADCs with no aggregation as confirmed by HPLC and bioanalyzer. These ADCs preserved their binding to IGF-1R as confirmed by flow cytometry, and biolayer interferometry. The DOTA conjugation on cixutumumab (21.9 ± 5.9 nM) led to increase fourfold increase in K D compared with the unconjugated antibody-cixutumumab (5.2 ± 1.9 nM) as observed by flow cytometry. A similar effect was observed when DOTA was conjugated to cixutumumab ADCs. www.nature.com/scientificreports/ (Conjugation of DOTA on ADC constructs decreased the internalization rate of all constructs (Fig. 1C). The decrease in internalization rate of the DOTA immunoconjugates may be the result of additional negative charges introduced by DOTA. The introduction of positive charges on antibodies has been shown previously to significantly improve binding and internalization rates 28 . However, this effect was expected to be less when the DOTA conjugated antibody is complexed with 111 In (due to overall net charge), as used in imaging and biodistribution studies. At that point, factors other than the charge of the antibody play a predominant role in tissue uptake. To be potent, ADCs developed using non-cleavable linkers/drugs must be cleaved in lysosomes to release the cytotoxin PEG 6 -DM1. Using an anti-EpCAM antibody, Kovtun et al. showed that there was no difference in the lysosomal processing of PEG 4 -DM1 compared to DM1 conjugated ADCs 17 . Some literature reports suggest ADCs internalize better than the naked antibodies while other reports show no differences or even the opposite 28 . This depends on the antibody-antigen complex as well as other adapter molecules. Smith et al. observed higher internalization rate using anti-melanotransferrin ADC L49 conjugated to auristatin than for the naked antibody 29 . Furthermore, Law et al. using anti-CD20 antibody rituximab showed that the internalization rate was dependent on drug payload with rituximab conjugated to doxorubicin (rituximab-vcDox) showing better internalization than rituximab conjugated to monomethyl auristatin E (rituximab-vcMMAE) in CD20 positive cells 30 . We showed by live cell imaging that in the case of IGF-1R internalization rate of cixutumumab-PEG 6 -DM1-Low was higher at all times compared with the unconjugated antibody (Fig. 1C).
Cixutumumab cross reacts with murine IGF-1R ( Supplementary Fig. 5C) with a K D of 3.5 nM. Biodistribution and imaging studies show that tumor uptake in mice bearing MCF-7/Her18 xenograft decreases with increasing number of PEG 6 -DM1 on the antibody. Tumor uptake for 111 In-cixutumumab was slightly higher than for 111 In-cixutumumab-PEG 6 -DM1-Low even though this was not statistically significant (p > 0.05). However, tumor uptake of 111 In-cixutumumab was significantly higher (p < 0.0001) than 111 In-cixutumumab-PEG 6 -DM1-High, indicating a lower binding to IGF-1R of the radioimmunoconjugate in vivo. The microSPECT imaging further confirmed the significantly lower tumor uptake of 111 In-cixutumumab-PEG 6 -DM1-High at 24, 48, 96 and 120 h post injection.
As expected, cixutumumab ADCs were more cytotoxic than cixutumumab to MCF-7/Her18 cells. Compared with cixutumumab and cixutumumab-PEG 6 -DM1-high, cixutumumab-PEG 6 -DM1-Low had the highest internalization rate and the lowest EC 50 (most potent in vitro). Despite having similar K D values in vitro, the lower potency of cixutumumab-PEG 6 -DM1-high may be due to its slightly lower internalization rate. We studied the efficacy of the antibody and ADCs in MCF-7/Her18 xenograft. Cixutumumab-PEG 6 -DM1-Low (median survival of 147 days) prolonged the survival of tumor bearing mice compared with cixutumumab (median survival of 71 days) and cixutumumab-PEG 6 -DM1-High (median survival 65 days). Perez et al. showed that trastuzumab emtansine (trastuzumab-DM1) was not superior to trastuzumab plus taxanes as a first line treatment in HER2 positive breast cancer patients 31 . It remains to be seen in future preclinical studies if cixutumumab-PEG 6 -DM1-Low is inferior/superior to cixutumumab plus chemotherapeutics such as taxanes. Cixutumumab-PEG 6 -DM1-High was less effective despite having a higher drug to antibody ratio, likely due to the lower tumor uptake of the construct. Since cixutumumab cross reacts with mouse IGF-1R ( Supplementary Fig. 5C) one would expect any potential toxicity issues in humans to be evident in our mouse studies. A more detailed toxicity study needs to be performed in IND enabling studies to fully determine the maximum tolerated dose of these new immunoconjugates.

Materials and methods
General. All experiments were performed in accordance with the University of Saskatchewan's guidelines and regulations. Chemicals used in the conjugation, radiolabeling, and purification steps were American Chemical Society reagent grade or better. Water and buffers were rendered metal-free by passing through a column of Chelex-100 resin, 200-400 mesh (Bio-Rad Laboratories, Inc.), and were sterile-filtered through a 0.22 µm filter device. Cixutumumab was purchased from Creative Biolabs (Shirley, NY). DM1 drug was obtained from Toronto Research chemicals (Toronto, ON) and NHS-PEG 6 -Maleimide was purchased from Biochempeg (Watertown, MA). Bifunctional chelating agent 2-(4-isothiocyanatobenzyl)-1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (p-SCN-Bn-DOTA) was obtained from Macrocyclics (Plano, TX). Recombinant human and mouse IGF-1R were purchased from R&D Systems (Minneapolis, MN). 111 In was received from Nordion (Kanata, ON) as 111 InCl 3 in 0.1 M hydrochloric acid (optima grade, Sigma-Aldrich, St. Louise, MO). The 1 H NMR spectra were recorded on a DPX-500 MHz Bruker FT-NMR spectrometer (St. Louis, Missouri) using CDCl 3 as solvent. The chemical shifts were reported as parts per million (δ ppm) tetramethylsilane (TMS) as an internal standard. Mass spectra were obtained on an Agilent 6550 iFunnel QTOF mass spectrometer (Palo Alto, CA). The progress of the drug linker reaction was monitored on ready-made silica-gel plates (Merck, Etobicoke, ON) using ethyl acetate and hexane (3:1) as solvent 32 . Iodine was used as a developing agent or by spraying with the potassium permanganate reagent. Chromatographic purification was performed on a silica gel (100-200 mesh). The residues were obtained recrystallized by the addition of 30:70 hexane-chloroform. All chemicals and reagents obtained from Sigma-Aldrich (St. Louise, MO) were used without further purification 32 .
Cell culture. The IGF-1R-positive human breast cancer cell line MCF-7/Her18 was kindly provided by Dr.
Bioanlzyer. The analysis of molecular weight and purity of the immunoconjugates was performed on an Agilent 2100 Bioanalyzer using Agilent High Sensitivity Protein 250 Kit (cat # 5067-1575) following the manufacturer's protocol. The size and relative peak area were calculated using Agilent 2100 Expert software 32 .
Biolayer interferometry. Binding kinetics between the antibodies and IGF-1R were measured using biolayer interferometry (BLI) with ForteBio Octet RED384 (PALL Corporation, CA). Antibodies were immobilized on Anti-human IgG Fc Capture sensors (18-5060, Forte Bio) according to manufactures instructions. The equilibrium dissociation constant (K D ) was obtained using a 1 to 1 binding model with local fitting. Data analysis and curve fitting was performed using data analysis software 7. Radiolabeling and immunoreactivity. Radiolabeling of DOTA conjugates with 111 InCl 3 was performed as reported earlier 37 . The reaction was monitored using iTLC using 100 mM sodium citrate buffer (pH 5.0) as mobile phase. The iTLC strips were analyzed using ScanRam (LabLogic, Brandon, FL). The 111 In-labeled conjugates were purified using Amicon Ultra-4 centrifugal filters (10 K, EMD Millipore, Burlington, MA) with PBS, and purity was determined using size exclusion radio-HPLC and iTLC. A radiochemical purity (RCP) > 95% was considered good for in vitro and in vivo studies. The immunoreactive fraction of the 111 In-labeled immunoconjugates was determined using MCF-7/Her18 cells as previously described 38 .
In vitro cytotoxicity of ADCs. The in vitro cytotoxicity (EC 50 values) of cixutumumab immunoconjugates was determined using IncuCyte S3 live cell imager 33 . Briefly, 3000-5000 (MCF-7/Her18) cells were seeded 24 h prior to treatment in a 96 well plates. Media was removed and cells were washed with PBS. Cells were then incubated with IncuCyte ® Cytotox Red reagent diluted in complete media (1× , Essen Bioscience Cat #4632) for 3 h before treatment. Then, cells were treated with different concentrations (0.2-500 nM) of the antibodies and ADCs at 37 °C for 30 min prior to imaging. Live cell imaging was performed as described above and the EC 50 values for individual compounds were calculated using GraphPad Prism 5.
Tumor xenograft, microSPECT/CT imaging and biodistribution. Animal studies were approved by the University of Saskatchewan Animal Care and Use Committee as per protocol # 20170084. Female athymic CD-1 nude mice were purchased from Charles River Laboratory (Sherbrooke, QC). The mice were housed under standard conditions in approved facilities with 12 h light/dark cycles and provided ad libitum throughout the duration of the studies. For inoculation, MCF-7/Her18 cells (5 × 10 7 cells/mL) were resuspended in a 1:1 mixture of PBS: Matrigel (BD Biosciences, ON) and 0.2 mL suspension was injected in the right flank. The radiolabeled antibodies were passed through 0.22 µm Ultra free MC filter and 10-12 MBq (specific activity, 0.5 MBq/µg) was injected via tail a vein in tumor bearing mice. MicroSPECT images were acquired using Vector4CT scanner (MILabs, Utrecht) at 24, 48, 96, and 120 h post injection using a similar lab protocols 39 . The mice were anesthetized using a mixture of isoflurane/oxygen (5% of isoflurane in oxygen) and whole-body SPECT/CT images were obtained while anesthesia was maintained at 2%. Body temperature, heart rate, and breathing frequency were monitored continuously using physiosuite (Kent Scientific). All the scans were acquired in a list-mode data format with an ultra-high sensitivity (XUHS-2.0 mm) mouse pinhole collimator with 75 pinholes 40,41 . Corresponding CT scans were acquired with a tube setting of 50 kV and 480 μA, single frames of full scan angle were acquired in 480 steps and shoot rotation mode.