89Zr-Onartuzumab PET imaging of c-MET receptor dynamics

Purpose c-MET and its ligand hepatocyte growth factor are often dysregulated in human cancers. Dynamic changes in c-MET expression occur and might predict drug efficacy or emergence of resistance. Noninvasive visualization of c-MET dynamics could therefore potentially guide c-MET-directed therapies. We investigated the feasibility of 89Zr-labelled one-armed c-MET antibody onartuzumab PET for detecting relevant changes in c-MET levels induced by c-MET-mediated epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor erlotinib resistance or heat shock protein-90 (HSP90) inhibitor NVP-AUY-922 treatment in human non-small-cell lung cancer (NSCLC) xenografts. Methods In vitro membrane c-MET levels were determined by flow cytometry. HCC827ErlRes, an erlotinib-resistant clone with c-MET upregulation, was generated from the exon-19 EGFR-mutant human NSCLC cell line HCC827. Mice bearing HCC827 and HCC827ErlRes tumours in opposite flanks underwent 89Zr-onartuzumab PET scans. The HCC827-xenografted mice underwent 89Zr-onartuzumab PET scans before treatment and while receiving biweekly intraperitoneal injections of 100 mg/kg NVP-AUY-922 or vehicle. Ex vivo, tumour c-MET immunohistochemistry was correlated with the imaging results. Results In vitro, membrane c-MET was upregulated in HCC827ErlRes tumours by 213 ± 44% in relation to the level in HCC827 tumours, while c-MET was downregulated by 69 ± 9% in HCC827 tumours following treatment with NVP-AUY-922. In vivo, 89Zr-onartuzumab uptake was 26% higher (P < 0.05) in erlotinib-resistant HCC827ErlRes than in HCC827 xenografts, while HCC827 tumour uptake was 33% lower (P < 0.001) following NVP-AUY-922 treatment. Conclusion The results show that 89Zr-onartuzumab PET effectively discriminates relevant changes in c-MET levels and could potentially be used clinically to monitor c-MET status. Electronic supplementary material The online version of this article (doi:10.1007/s00259-017-3672-x) contains supplementary material, which is available to authorized users.

Introduction c-MET is a receptor tyrosine kinase with roles in embryogenesis, and tissue repair and regeneration in homeostasis. Dysregulation of c-MET is often found in solid cancers as a consequence of mutation, protein overexpression, gene amplification and auto/paracrine upregulation of its ligand hepatocyte growth factor (HGF). Aberrant activation of c-MET/HGF signalling leads to increased tumour proliferation, invasiveness and angiogenesis, resulting in poor prognosis in various human cancers, including non-small-cell lung cancer (NSCLC) and gastric cancer [1]. NSCLC patients harbouring mutations activating epidermal growth factor receptor (EGFR) can be treated with the EGFR tyrosine kinase Electronic supplementary material The online version of this article (doi:10.1007/s00259-017-3672-x) contains supplementary material, which is available to authorized users. inhibitors (TKIs) erlotinib and gefitinib. Resistance of NSCLC to EGFR TKIs invariably occurs in these patients due to secondary mutations or activation of bypass signalling pathways, such as c-MET/HGF [2,3]. Both c-MET and HGF have been implicated in resistance to EGFR-targeted therapies due to crosstalk via shared downstream signalling intermediates [2,4]. Concurrent transient exposure to HGF can result in lasting resistance to EGFR targeting TKIs by providing positive selection pressure for c-MET-upregulated resistant subclones [5,6]. c-MET-mediated resistance in EGFR TKIresistant NSCLC was detected in around 5-15% of patients using quantitative polymerase chain reaction and fluorescence in situ hybridization [2,7].
Several c-MET-directed treatment strategies have been reported. NVP-AUY-922, a non-geldanamycin heat shock protein 90 (HSP90) inhibitor, can potently downregulate EGFR and c-MET, and can overcome c-MET-mediated resistance in an EGFR TKI-resistant clone of NSCLC cell line HCC827 [8]. NVP-AUY-922 is currently in clinical trials in advanced EGFR mutant NSCLC refractory to EGFR TKIs (NCT01646125, NCT01124864), or with an EGFR exon 20 mutation (NCT01854034). The combination of NVP-AUY-922 and erlotinib in a phase I/II study of erlotinib-treated patients with progressive NSCLC has shown a 16% partial tumour response rate [9]. Another form of c-MET-directed therapy involves antibodies. Onartuzumab, is a 99-kDa onearmed antibody, reactive to human but not murine c-MET. Onartuzumab inhibits HGF binding, blocks c-MET receptor phosphorylation and downstream signalling and has shown antitumour efficacy in preclinical models [10].
In a randomized phase II trial in patients with advanced NSCLC, the addition of onartuzumab to erlotinib resulted in an improvement in progression-free and overall survival in patients positive for c-MET on immunohistochemistry (IHC) [11]. However, the METLung randomized phase III clinical trial in which erlotinib was combined with onartuzumab or placebo in patients with NSCLC and positive for c-MET on IHC was terminated early because of lack of effectiveness [12]. MET amplification and overexpression, on the other hand, are associated with increased responses to c-MET inhibitors [13]. Recently, splice alterations in exon 14 of the MET gene (METex14), that are found in about 3% of human lung cancers, have been found to be associated with in vitro and clinical sensitivity to the c-MET TKI capmatinib and the ALK/ROS1/MET TKI crizotinib [14]. METex14 leads to impaired c-MET downregulation and degradation, resulting in c-MET protein overexpression [15].
Selection of patients with c-MET amplification or overexpression could therefore potentially be beneficial for therapy decisions, due to their sensitivity to c-METdirected agents. IHC is generally performed on archived tissues and therefore reflects c-MET status of the tissues at the time of retrieval. c-MET status, however, can vary over time and among lesions, indicating a need for biomarkers able to capture this plasticity for the body as a whole [13,16]. PET may provide a noninvasive method for assessing whole-body c-MET status to capture c-MET dynamics after the emergence of c-MET-mediated EGFR TKI resistance, or as response to HSP90 inhibition.
For ease of clinical translation, we selected the therapeutic antibody-based PET tracer 89 Zr-onartuzumab, which has been reported to effectively discriminate c-MET expression in human gastric cancer xenografts [17]. Several other antibody and antibody fragment tracers have been reported to be of value for preclinical c-MET PET imaging [18]. These studies used an immunogenic full-length murine antibody DN30 [19] or antibody fragments and other protein scaffolds such as the 89 Zr-labelled H2 cys-diabody, H2 minibody [20], and anticalin 89 Zr-PRS110 [21]. In contrast to 89 Zr-onartuzumab, these tracers are not readily translatable to the clinic, as none of these tracer protein backbones has yet been administered to patients, and would therefore require extensive safety testing. Furthermore, administration of tracers based on the HGF ligand such as 64 Cu-HGF, as well as the bivalent antibody DN30 [19,22], might have the unwanted result of stimulating tumour growth by activating c-MET.
We therefore aimed to validate the ability of 89 Zronartuzumab PET to assess c-MET upregulation-mediated erlotinib resistance, as well as c-MET downregulation after HSP90 inhibitor NVP-AUY-922 treatment in human NSCLC xenografts.

Cell lines and chemicals
The human NSCLC cell line HCC827 was obtained from the American Type Culture Collection. Cells were authenticated in April 2015 by STR profiling using BaseClear and quarantined until screening showed that they were negative for microbial and mycoplasma contamination. Cells were subcultured twice weekly using Roswell Park Memorial Institute-1640 (RPMI-1640) medium supplemented with 10% fetal calf serum (Bodinco BV) and incubated at 37°C in a fully humidified atmosphere containing 5% CO 2 . HCC827ErlRes, a stable erlotinib-resistant subclone, was generated by culturing the parental cell line HCC827 in the presence of 1 μM erlotinib and 50 ng/mL recombinant human HGF (PeproTech) for 2 weeks, followed by 2 weeks in 1 μM erlotinib, a method analogous to that described by Turke et al. [6]. NVP-AUY-922 (Luminespib; LC Laboratories) and erlotinib (LC Laboratories) were dissolved in dimethyl sulfoxide, and aliquots stored at −80°C.
The immunoreactivity of Df-onartuzumab conjugates was assessed by a competition assay as described previously. In brief, NuncBreakApart 96-well plates were coated overnight at 4°C with 100 μL 100 ng/mL c-MET extracellular domain (Genentech) in 0.1 M Na 2 CO 3 buffer, pH 9.6. 89 Zr-Onartuzumab was diluted to 5,000 ng/mL in assay diluent which consisted of PBS, 0.5% bovine serum albumin fraction V (Bio-Connect) and 0.05% Tween 20 (Sigma-Aldrich). Plates were washed with washing buffer (PBS with 0.05% Tween 20) and blocked by incubation for 1 h at room temperature with 200 μL assay diluent, and washed again. Eight dilutions of unmodified onartuzumab at molar excesses between 0.004 and 156.25-fold were premixed 1:1 with diluted 89 Zr-onartuzumab solution, and 100 μL aliquots of the mixtures were incubated in wells for 1 h at room temperature. After washing, the wells were broken apart and counted using a calibrated well-type 1282 Compugamma system (LKB Wallac). Counts were plotted against concentrations of competing unmodified onartuzumab and the half-maximum inhibitory concentration (IC50) was calculated using GraphPad 5.0 (GraphPad Software, Inc.). The IC50 was divided by the final tracer concentration (2,500 ng/mL) to yield the immunoreactive fraction.

111
In-OA-NBC and 89 Zr-OA-NBC control tracer To determine nonspecific tracer distribution in organs, as well nonspecific tumour uptake, a one-armed isotype control antibody OA-NBC (Genentech) was coinjected and used as control tracer in all experiments. For 111 In labelling, OA-NBC was conjugated with p-SCN-Bn-DTPA (Macrocyclics) as described previously [27]. Radiolabelling was performed with 111 In-chloride (Mallinckrodt). The RCP of 111 In-OA-NBC labelling was checked by silica gel instant thin-layer chromatography (ITLC) using 0.1 M citrate buffer, pH 6.0, as eluent. Furthermore, OA-NBC was incubated with a 1:5 molar excess of OA-NBC to Df, in a procedure similar to that described for Df-onartuzumab, to yield Df-OA-NBC conjugate for 89 Zr-OA-NBC experiments.
c-MET expression on HCC827ErlRes cells in comparison with that on HCC827 cells was evaluated by inoculating a group of ten mice with 5 × 10 6 HCC827 cells into the right flank and 5 × 10 6 HCC827ErlRes cells into the opposite flank. The sizes of the HCC827 and HCC827ErlRes tumours were 457 ± 182 and 240 ± 74 mm 3 (P < 0.01), respectively, at the time of biodistribution analysis (day 6). Mice received 10 μg 89 Zr-onartuzumab (effective protein dose injected 9.7 ± 0.2 μg at 4.42 ± 0.09 MBq) coinjected with 10 μg 111 In-OA-NBC (1 MBq) via a penile vein. MicroPET scans were performed 6 days after injection, followed by ex vivo biodistribution analysis.
All microPET scans were performed using a Focus 220 rodent scanner (CTI Siemens). MicroPET scans were reconstructed and in vivo quantification was performed using AMIDE v. 1.0.4 [28]. Regions of interest (ROI) were drawn on the tumour based on the ex vivo weight assuming 1 g/cm 3 tissue density or measured tumour volume for longitudinal experiments. Scan data are presented as mean standardized uptake values (SUV mean ), which were calculated from the mean activity in the ROI divided by the injected dose (corrected for decay) per gram body weight, as described previously [29]. For biodistribution studies, organs and injected tracer standards were counted using the calibrated well-type LKB 1282 Compugamma system and weighed. 111 In and 89 Zr were counted in a single measurement at 311-500 keV and 758-1,144 keV, respectively, with crosstalk in the 111 In channel corrected using a reference 89 Zr dilution series. After decay correction, ex vivo tissue activities were expressed as the percentages of injected dose per gram tissue (%ID/g). Xenograft tumours were formalin-fixed and paraffinembedded for IHC analysis. All animal experiments were approved by the Institutional Animal Care and Use Committee of the University of Groningen.

Ex vivo analyses
Formalin fixed, paraffin-embedded tissue slices of thickness 4 μm were stained for c-MET using monoclonal rabbit antihuman c-MET antibody diluted 1:400 (ab51067; Abcam) [21]. Haematoxylin and eosin (H&E) staining was performed regularly to assess tissue viability and morphology. Digital scans of slides were acquired using a NanoZoomer 2.0-HT multislide scanner (Hamamatsu) and analysed with NanoZoomer Digital Pathology viewer software.

Statistical analysis
Data are presented as means ± SD. Statistical analyses were performed with GraphPad Prism 5.0 using the two-sided Mann-Whitney test for nonparametric data and the twosided paired Student's t test for paired data. P values ≤0.05 were considered significant.

Effects of erlotinib resistance and NVP-AUY-922 treatment on c-MET expression
An erlotinib-resistant clone, HCC827ErlRes, was generated from the parental cell line HCC827 by culturing cells for 2 weeks with 50 ng/mL HGF and 1 μM erlotinib, followed by 2 weeks culture in the presence of 1 μM erlotinib. Surface expression of c-MET on HCC827ErlRes cells as measured by flow cytometry was upregulated to 213 ± 44%, while EGFR surface levels were downregulated to 35 ± 17% of levels in the parental HCC827 cells (Fig. 1a). HCC827ErlRes cells were able to fully proliferate in the presence of up to 1,000 nM erlotinib as measured by the MTT assay, while parental HCC827 cells remained highly sensitive to erlotinib with an IC50 of 12 nM (Fig. 1b). NVP-AUY-922 treatment reduced surface expression of EGFR and c-MET (Fig. 1c). NVP-AUY-922 treatment was equally effective in reducing the viability of both HCC827 and HCC827ErlRes cells (Fig. 1d).

Zr-onartuzumab tracer development
Conjugation of Df to onartuzumab was approximately 60% efficient for all molar reaction ratios tested ( Supplementary  Fig. 1a). Df-onartuzumab conjugates were able to consistently bind ≥500 MBq 89 Zr per milligram of Df-onartuzumab with RCP ≥95% at ratios above 1:2 onartuzumab bound to Df (Supplementary Fig. 1b). The competition assay revealed a trend for lower immunoreactivity at higher conjugation ratios, signifying a need for balancing the required specific activity with the retained affinity of Df-onartuzumab conjugates ( Supplementary Fig. 1c). Aggregation and fragmentation of Df-onartuzumab conjugates were not observed. A conjugation reaction ratio of 1:5, yielding 3.11 ± 0.33 Df bound to onartuzumab, was chosen as the optimal ratio for animal studies. 89 Zr-Onartuzumab was stable in vitro, with a maximum observed decrease in RCP from 99.0 ± 0.2% to 91.0 ± 6.6% in human serum after 7 days at 37°C (Supplementary Fig. 1d). All 89 Zr-onartuzumab tracer batches had a RCP of ≥95% by trichloroacetic acid precipitation, while 111 In-OA-NBC batches had a RCP of ≥90% by ITLC. Zr-Onartuzumab tumour uptake increased over time for all tracer protein doses tested, with the highest tumour and least background organ uptake observed on day 6 after injection (Fig. 2a). 89 Zr-Onartuzumab tumour uptake was higher than that of the coinjected 111 In-OA-NBC control tracer in all tracer protein dose groups (Fig. 2b, Supplementary Fig. 2a, b). The 10 and 25 μg 89 Zr-onartuzumab dose groups showed similar tumour uptakes of 31.6 ± 8.7 and 29.8 ± 12.1%ID/g, while uptake in the 100 μg 89 Zr-onartuzumab dose group was lower but not significantly (P = 0.17 vs. the 10 μg group) at 23.5 ± 9.4%ID/g. SUV mean values in tumours of the 10, 25 and 100 μg 89 Zr-onartuzumab dose groups were 2.7 ± 0.9, 2.9 ± 0.7 and 2.2 ± 0.8, respectively (Fig. 2b, c, Supplementary   Fig. 2a, b). The 89 Zr-OA-NBC control showed no accumulation in HCC827 tumours over time (Fig. 2a, c, Supplementary  Fig. 2a, b), while ex vivo tumour uptake was similar to that of 111 In-OA-NBC (Fig. 2b, Supplementary Fig. 2a, b), showing the suitability of 111 In-OA-NBC as a proxy for 89 Zr-OA-NBC. Based on these results, a 89 Zr-onartuzumab protein dose of 10 μg was used in the next mouse cohorts.

In vivo effects HSP90 inhibition on 89 Zr-onartuzumab PET
The effects of NVP-AUY-922 treatment on c-MET expression were evaluated in HCC827 xenograft-bearing mice. In mice receiving NVP-AUY-922, the tumour SUV mean for 89 Zronartuzumab was 33 ± 10% lower after treatment than in the baseline scans (P < 0.001), while uptake values in vehicletreated mice before and during treatment were similar (Fig. 4a, b). Biodistribution studies confirmed the PET results revealing 27% lower 89 Zr-onartuzumab tumour uptake in NVP-AUY-922-treated mice than in vehicle-treated mice (Fig. 4c, Supplementary Figs. 4a, 4b, 6; P < 0.05). Similar comparisons of the tumour-to-muscle and tumour-to-blood ratios for the two groups showed similar trends, but the differences did not reach significance. IHC showed a marked decrease in c-MET expression in the tumours, with similar levels of necrosis, in NVP-AUY-922-treated and in vehicletreated animals (Fig. 4d, Supplementary Fig. 5b).

Discussion
This study illustrates the feasibility of in vivo imaging of c-MET dynamics by detecting upregulation after c-METmediated erlotinib resistance, as well as downregulation of c-MET following HSP90-directed therapy using 89 Zronartuzumab PET in human NSCLC xenograft-bearing mice.
Monitoring c-MET in vivo with 89 Zr-onartuzumab PET might be an attractive method for detecting c-MET-mediated emergence of resistance to EGFR TKIs, as well as for assessing downregulation of c-MET in response to HSP90 inhibition. Furthermore, it could be beneficial for patient selection for c-MET-directed therapy, as MET amplification and overexpression are associated with increased response to c-MET inhibitors [13].
Because of the different Df-suc-N-onartuzumab linker chemistry used for 89 Zr radiolabelling in the present study and for 89 Zr-Df-Bz-SCN-onartuzumab in the study by Jagoda et al. [17], we performed a tracer protein dose escalation study to optimize the 89 Zr-onartuzumab protein dose and a d b c  imaging time-point. HCC827 tumours in the present study showed high contrast and c-MET-specific 89 Zr-onartuzumab uptake at a tracer protein dose of 10 μg. Higher 89 Zronartuzumab protein doses showed a trend for lower tumour uptake. This blocking effect was also shown by Jagoda et al. using 1 mg cold onartuzumab, and is most likely caused by its nonreactivity with murine c-MET [10,17]. 89 Zr-Onartuzumab is remarkable in that respect, in that specific binding can be proven in vivo in a conventional blocking study. In contrast, cross-reactive antibodies might need a higher protein dose for optimal tumour contrast [30]. Similar protein doses of 89 Zr-OA-NBC and 111 In-OA-NBC one-armed isotype controls for 89 Zr-onartuzumab did not accumulate in tumour tissues, confirming the c-MET specificity of the 89 Zr-onartuzumab signal. 89 Zr-Df-p-SCN-onartuzumab uptake in MKN-45 gastric cancer xenografts 5 days after injection was lower at 22.5%ID/g than observed for 89 Zr-Df-suc-N-onartuzumab in HCC827 xenografts [17]. A direct comparison, however, is difficult due to differences in tumour models, desferal linker chemistries, tracer protein doses and day of biodistribution analysis.
Upregulation of c-MET in EGFR TKI gefitinib-resistant HCC827-GR6 xenografts was visualized with a 89 Zr-labelled H2 cys-diabody and H2 minibody, with twofold higher uptake observed in resistant tumours [20]. The reported biodistribution data for the resistant tumours, however, excluded cystic areas, which affected approximately half of the resistant c-MET-upregulated HCC827-GR6 tumours. Exclusion of these cystic areas might have artificially increased the apparent uptake of these tracers. Furthermore, a highly variable absolute tumour uptake was observed, differing by up to twofold between experiments, correlating with a similar variation in the remaining blood pool activity [20]. We observed robust tumour uptake of 89 Zr-onartuzumab. As the affinity of the H2 cys-diabody and H2 minibody for c-MET are comparable to that of onartuzumab, this is probably due to the high molecular weight of 89 Zr-onartuzumab that results in slower blood clearance and prolonged tumour exposure [10,20]. 89 Zr-Onartuzumab tumour-to-muscle and tumour-toblood ratios showed similar trends, but the differences between HCC827 and HCC827Erlres tumours did not reach significance. However, we included the 111 In-OA-NBC paired control molecule in each experimental animal as a valid control to determine specific tumour uptake [31], as well as ex vivo analysis for c-MET expression and necrosis. This strengthens the conclusion that the extra uptake in c-METupregulated HCC827ErlRes xenografts is indeed mediated by c-MET expression, and is not caused by possible variations in tumour size, blood pool tracer availability and permeability effects.
Pharmacokinetic readout of c-MET in response to c-MET TKI PHA665752 therapy has shown reduced total MET protein in human gastric cancer MKN-45 xenografts due to more necrosis which coincides with lower uptake of c-MET-specific peptide 99M Tc-AH-113018 [32]. The AH113804 peptide has also been labelled with 18 F for PET, and is able to detect recurrence of c-MET-expressing basal-like breast cancer xenografts after surgical resection [33]. A fluorescent labelled version of the same peptide has been used as a 'red flag' technique to detect polyps using a fluorescence endoscope technique in patients at risk of developing colon cancer [34].
We and others have shown potent downregulation of both EGFR and c-MET in human NSCLC cell lines by exposure to HSP90 inhibitors NVP-AUY-922 and 17-DMAG [8,35]. c-MET downregulation by treatment with NVP-AUY-922 has also been shown to overcome c-MET-mediated gefitinib and erlotinib acquired resistance in HCC827 sub clones [8]. We have visualized this downregulation of c-MET by treatment with NVP-AUY-922 in vivo using 89 Zr-onartuzumab PET, with sequential scans showing 33% lower uptake after treatment. This correlated with downregulation of c-MET protein on ex vivo IHC, while tumour necrosis levels were similar in NVP-AUY-922-treated and vehicle-treated animals.
A limiting factor in the present study may have been the relatively small effect sizes observed for upregulation and downregulation of tracer tumour accumulation. Preclinical studies using 89 Zr-labelled antibody tracers have shown that HSP90 inhibition and everolimus treatment downregulate other key oncogenic proteins, such as HER2, vascular endothelial growth factor A (VEGF-A), and insulin-like growth factor 1 receptor (IGF1R). The effect sizes of antibody tracer uptake reduction after target protein downregulation through HSP90 inhibition and everolimus treatment are comparable to the reduction in uptake of 89 Zr-onartuzumab after c-MET downregulation found in the present study [36][37][38][39], and the results of some of these preclinical studies have been translated to successful treatments in the clinic [40,41]. Furthermore, insight into whole-body c-MET target distribution via noninvasive 89 Zr-onartuzumab scans could potentially enlarge the patient population which might benefit from c-MET/HGFtargeted drugs [18].
Based on these promising preclinical results, we conclude that 89 Zr-onartuzumab c-MET PET provides a robust noninvasive and powerful tool that is easily translatable to the clinic for visualizing c-MET dynamics as a possible biomarker for c-MET-mediated resistance to erlotinib and for treatment response to HSP90 inhibition.
Ethical approval All applicable international, national, and/or institutional guidelines for the care and use of animals were followed. This article does not describe any studies with human participants performed by any of the authors.
Open Access This article is distributed under the terms of the Creative Comm ons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.