The use of radiocobalt as a label improves imaging of EGFR using DOTA-conjugated Affibody molecule

Several anti-cancer therapies target the epidermal growth factor receptor (EGFR). Radionuclide imaging of EGFR expression in tumours may aid in selection of optimal cancer therapy. The 111In-labelled DOTA-conjugated ZEGFR:2377 Affibody molecule was successfully used for imaging of EGFR-expressing xenografts in mice. An optimal combination of radionuclide, chelator and targeting protein may further improve the contrast of radionuclide imaging. The aim of this study was to evaluate the targeting properties of radiocobalt-labelled DOTA-ZEGFR:2377. DOTA-ZEGFR:2377 was labelled with 57Co (T1/2 = 271.8 d), 55Co (T1/2 = 17.5 h), and, for comparison, with the positron-emitting radionuclide 68Ga (T1/2 = 67.6 min) with preserved specificity of binding to EGFR-expressing A431 cells. The long-lived cobalt radioisotope 57Co was used in animal studies. Both 57Co-DOTA-ZEGFR:2377 and 68Ga-DOTA-ZEGFR:2377 demonstrated EGFR-specific accumulation in A431 xenografts and EGFR-expressing tissues in mice. Tumour-to-organ ratios for the radiocobalt-labelled DOTA-ZEGFR:2377 were significantly higher than for the gallium-labelled counterpart already at 3 h after injection. Importantly, 57Co-DOTA-ZEGFR:2377 demonstrated a tumour-to-liver ratio of 3, which is 7-fold higher than the tumour-to-liver ratio for 68Ga-DOTA-ZEGFR:2377. The results of this study suggest that the positron-emitting cobalt isotope 55Co would be an optimal label for DOTA-ZEGFR:2377 and further development should concentrate on this radionuclide as a label.

proteins 10 . A general challenge for development of EGFR-imaging agents is expression of EGFR in normal tissues, particularly in liver. Sequestering of radiolabelled probes in liver might result in their insufficient uptake in tumours 11 . However, a clinical study has demonstrated that it is possible to find an injected anti-EGFR antibody dose capable of providing a partial saturation of EGFR in the liver without target saturation in tumours enabling visualization of EGFR-expressing malignancies using radiolabelled antibodies 11 .
We develop EGFR-imaging probes based on Affibody molecules. Affibody molecules are small (7 kDa) engineered scaffold proteins, which can be selected by phage display for binding with high affinity to different proteins 12 . In vivo molecular imaging using radiolabelled Affibody molecules has been demonstrated in preclinical studies for several cancer-related molecular targets, e.g. HER2, EGFR, HER3, and IGF-1R. Clinical studies have demonstrated specific and high-contrast imaging of HER2-expression in metastatic breast cancer using a 111 In-and 68 Ga-labelled ABY-025 Affibody molecule 13,14 . Earlier, we have shown that the anti-EGFR 111 In-DOTA-Z EGFR:2377 Affibody molecule can visualize EGFR expression in murine xenografts 15 . Importantly, this Affibody molecule has similar affinity to both human and murine EGFR. According to surface plasmon resonance measurements, equilibrium dissociation constants (K D ) were 0.8 and 28 nM for K D1 and K D2 , respectively, in the case of murine EGFR. Corresponding values for human EGFR were 0.9 and 28 nM 15 . Therefore, a murine model takes into account the target-specific interaction of the tracer both in tumours and in normal tissues and should thus ensure good translational potential. Our studies showed that injection of 30-50 µg 111 In-DOTA-Z EGFR:2377 provides partial saturation of EGFR in normal tissues without saturating the target in tumours 15 . Internalization of 111 In-DOTA-Z EGFR:2377 by EGFR-expressing cells was slow, less than 20% after 24 h. Therefore, normal EGFR-expressing organs acted as depot that release 111 In-DOTA-Z EGFR:2377 resulting in a relatively slow blood clearance. The tumour-to-blood ratios at optimal dosage were 7.0 ± 0.5 and 14 ± 4, at 4 and 24 h after injection, respectively.
A goal for the development of imaging agents is to obtain a probe with high uptake in tumour and low uptake in normal tissues. Earlier studies have demonstrated that different combinations of a chelator and a radionuclide can modify physico-chemical properties of a tumour-targeting peptide in different ways and influence its biodistribution 16,17 . Moreover, the use of the same chelator in combination with different nuclides has resulted in altered imaging properties of somatostatin, RGD-peptides and bombesin analogues [17][18][19][20][21][22] . For example, Heppeler and co-workers concluded after comparison of 57 Co-and 68 Ga-labelled somatostatin analogue DOTA-TOC that "the physical features (size, charge) of the radiometal strongly influence receptor affinity, biodistribution and the pharmacokinetics of radiometal-labelled peptides" 22 . A strong influence of a radiometal/chelator combination on imaging contrast has been observed also for Affibody molecules. For example, biodistribution and tumour-to-organ ratios of DOTA-conjugated anti-HER2 Affibody molecules were altered appreciably by substitution of 68 Ga by 111 In or 44 Sc and by substitution of 111 In by 57 Co [23][24][25] . Importantly, systematic studies concerning influence of radionuclides on biodistribution of tumour-targeting Affibody conjugates might help to select a probe with the optimal imaging characteristics.
Three cobalt radioisotopes, 55 Co (T 1/2 = 17.5 h), 57 Co (T 1/2 = 271.8 d) and 58m Co (T 1/2 = 9.04 h) are of interest for radionuclide tumour targeting. 55 Co is a long-lived positron emitter, which can be used for labelling of peptides for PET imaging 26,27 . 58m Co emits Auger electrons and has been proposed for radionuclide therapy 26,28 . The long-lived 57 Co is a convenient surrogate radionuclide, which can be used for preclinical development of radiocobalt-labelled probes 22,27 in the same way as 125 I is used as a substitute for 123 I, 124 I and 131 I.
Development of a radiocobalt-labelled Z EGFR:2377 derivative with favourable targeting properties would open the way for the use of 55 Co as a label. The increase of tumour-to-blood ratio of 111 In-DOTA-Z EGFR:2377 with time indicates that performing the imaging at the next day after injection would provide better contrast. For this purpose, a positron-emitting radionuclide with a half-life of 10-20 h would be optimal. Besides, our studies on the previous generation of anti-EGFR Affibody molecules demonstrated that residualizing radiometal labels provide appreciably better tumour retention of the radionuclide at 24 h after injection 29 . 55 Co can be produced by the 54 Fe(d, n) 55 Co reaction on enriched 54 Fe targets using low-energy cyclotrons available at PET-centres 26,30 . Importantly, the DOTA chelator provides stable coupling of radiocobalt to targeting proteins and peptides 22,25,26,28 . The use of a positron-emitting label for DOTA-ZEGFR:2377 would further improve detection of EGFR expression in vivo, since PET provides better imaging sensitivity compared with SPECT. A generator-produced positron emitter 68 Ga (T 1/2 = 67.7 min) is one obvious alternative to 55 Co, although the time window for imaging would be limited to three-four hour after injection.
The goal of this study was to confirm that the DOTA-conjugated Z EGFR:2377 can be labelled with 55 Co with preserved binding specificity and to compare tumour-targeting properties of radiogallium-and radiocobalt-labelled DOTA-Z EGFR:2377. The long-lived cobalt radioisotope 57 Co (T 1/2 = 271.8 d) was used in a number of experiments as a convenient surrogate of 55 Co.

Results
Tracer production and labelling. The C-terminal cysteine-containing anti-EGFR Affibody molecule Z EGFR:2377 was produced, conjugated with maleimido-derivative of DOTA and purified as described earlier 15 . Analytic reversed-phase high-performance liquid chromatography (RP-HPLC) demonstrated a purity for DOTA-Z EGFR:2377 of over 95% ( Supplementary Fig. S1). The molecular mass of DOTA-Z EGFR:2377 was determined by mass spectrometric analysis and showed a good agreement with the theoretical value (expected 7916.9 Da, found 7918 Da) ( Supplementary Fig. S2). Measuring of circular dichroism spectra ( Supplementary Fig. S3) demonstrated a high-fidelity refolding of DOTA-Z EGFR:2377 after heating up to 90 °C.
Labelling of DOTA-Z EGFR:2377 with cobalt isotopes provided yields of over 99%. The identity of conjugates was confirmed by radio-SDS-PAGE. Initial experiments concerning labelling with 55 Co were associated with precipitation of Affibody molecules. Exclusion of glassware from the radionuclide production solved this problem. Still, aggregation of 55 Co-DOTA-Z EGFR:2377 remains to be an issue. Similar effect for 57 Co-DOTA-Z EGFR:237 was Scientific RepoRts | 7: 5961 | DOI:10.1038/s41598-017-05700-7 observed only occasionally, e.g. in the case of prolonged (3 h) incubation at 85 °C. Remarkably, this phenomenon was observed only with Affibody molecules, but not during 55 Co-labelling of somatostatin or bombesin analogues [26][27][28] . Labelling of DOTA-Z EGFR:2377 with 68 Ga provided a yield of 98.9 ± 0.5%. A purification using NAP-5 column provided a purity of 100%. A summary of obtained yields and specific activities of imaging probes is provided in Supplementary Table S1. There was no measurable release of radionuclide from the conjugate after 4 h ( 57 Co) or 2 h ( 68 Ga) incubation with 500-fold excess of EDTA.
Data concerning internalization of 57 Co-DOTA-Z EGFR:2377 and 68 Ga-DOTA-Z EGFR:2377 by EGFR-expressing A431 cells are presented in Fig. 2. In agreement with previous data for 111 In-DOTA-Z EGFR:2377 15 , the internalization was slow, although the internalized fraction of the radioactivity increased over time. After 4-h incubation, the internalized fraction was less than 20% for both tracers. In the case of 57 Co-DOTA-Z EGFR:2377 , the cell-associated radioactivity reached maximum between 2 and 4 h with subsequent decline.  In vivo studies. Specificity of EGFR targeting in vivo was evaluated by pre-saturation of the receptors in normal murine tissues and A431 xenografts with high EGFR expression by pre-injection of a 10-fold excess of non-labelled Z EGFR:2377 . In addition, mice bearing SKOV-3 xenografts with very low EGFR expression were used as negative control. The results of the in vivo specificity experiments are presented in Fig. 3. Pre-saturation of EGFR in A431 xenografts caused a significant (p < 0.0005) decrease in tumour uptake of both tracers. Besides, there was a statistically significant (p < 0.05) decrease in radioactivity concentration in blood and lung, and in EGFR-expressing tissues, such as salivary gland, liver, and colon. The uptake of both 57 Co-DOTA-Z EGFR:2377 and 68 Ga-DOTA-Z EGFR:2377 in SKOV-3 xenografts was much lower (p < 10 −7 ) than in A431 xenografts.
Comparison of biodistribution of 68 Ga-DOTA-Z EGFR:2377 and 57 Co-DOTA-Z EGFR:2377 is presented in Table 1. The most prominent difference was between 68 Ga-DOTA-Z EGFR:2377 and 57 Co-DOTA-Z EGFR:2377 in hepatic, splenic and tumour uptake. The uptake of 68 Ga-DOTA-Z EGFR:2377 was 3.3-fold higher than the uptake of 57 Co-DOTA-Z EGFR:2377 in liver, and 4.4-fold higher in spleen. The tumour uptake of 57 Co-DOTA-Z EGFR:2377 was 2.1-fold higher than the uptake of 68 Ga-DOTA-Z EGFR:2377 . Besides, the uptake of 57 Co-DOTA-Z EGFR:2377 was 1.2-1.5 fold higher in salivary gland, lung, and colon. At 24 h after injection, the uptake of 57 Co-DOTA-Z EGFR:2377 was reduced 1.3-1.6 fold in tumour as well as in salivary gland, lung, and liver.
The difference in the distribution profile resulted in significantly (p < 0.05) higher tumour-to-organ ratios of 57 Co-DOTA-Z EGFR:2377 compared to 68 Ga-DOTA-Z EGFR:2377 for all measured organs and tissues at 3 h after injection ( Table 2). The most stunning was the difference in liver, where the use of 57 Co-DOTA-Z EGFR:2377 provided a positive contrast (tumour-to-liver ratio of 3.1 ± 0.5), while the tumour-to-liver ratio of 68 Ga-DOTA-Z EGFR:2377 was below one. At 24 h after injection, the tumour-to-blood ratio for 57 Co-DOTA-Z EGFR:2377 increased 2.6-fold, to 32 ± 7. There was minor but significant decrease of tumour-to-spleen and tumour-to-kidney ratios. injection. An injected protein dose was adjusted to 40 µg for all animals. For pre-blocking, a group of animals bearing EGFR-positive A431 xenografts was pre-injected with 400 µg non-labelled Z EGFR:2377 40 min before injection of the radiolabelled conjugate. A group with SKOV-3 xenografts was used as a negative control. Data are presented as average %ID/g ± SD (n = 4).  Table 1. Comparative biodistribution of 68 Ga-DOTA-Z EGFR:2377 and 57 Co-DOTA-Z EGFR:2377 after injection of 40 μg into BALB/C nu/nu mice bearing A431 xenografts. Data are presented as average %ID/g ± SD (n = 4), except for the gastrointestinal tract and carcass which are presented as %ID per sample. a Significant (p < 0.05) difference between uptake of 68 Ga-DOTA-Z EGFR:2377 and 57 Co-DOTA-Z EGFR:2377; b Significant (p < 0.05) difference between uptake of 57 Co-DOTA-Z EGFR:2377 at 3 and 24 h after injection.

h after injection 24 h after injection
Imaging using SPECT/CT and PET/CT devices for small animals, performed 3 h after injection (Fig. 4, full-scale image in Supplementary Fig. S5), confirmed the results of the biodistribution experiments. Kidneys were the organs with the highest uptake for both radionuclides. 68 Ga-DOTA-Z EGFR:2377 provided a high contrast of A431 xenografts, even though the hepatic uptake was higher than the tumour uptake. On the other hand, the tumour uptake of 57 Co-DOTA-Z EGFR:2377 was higher than the hepatic uptake.

Discussion
Anti-EGFR DOTA-Z EGFR:2377 Affibody molecule was successfully labelled with the nuclides 68 Ga and 55/57 Co. Both gallium-and cobalt-loaded DOTA-Z EGFR:2377 had subnanomolar IC 50 values (Fig. 1). 57 Co-DOTA-Z EGFR:2377 and 68 Ga-DOTA-Z EGFR:2377 showed specific binding to EGFR-expressing cells in vitro and in vivo (Fig. 3). Cellular processing of both 57 Co-DOTA-Z EGFR:2377 and 68 Ga-DOTA-Z EGFR:2377 was characterized by slow internalization (Fig. 2). Both conjugates enabled imaging of EGFR-expressing xenografts in vivo (Fig. 4). However, there was a striking difference in their biodistribution profile (Table 2). A common feature was a prominent hepatic uptake. The biodistribution profile of 68 Ga-DOTA-Z EGFR:2377 was quite similar to the profile reported for 111 In-DOTA-Z EGFR:2377 15 . The hepatic uptake was much higher than uptake in tumour or EGFR-expressing tissues, i.e. colon and salivary gland. In the case of 57 Co-DOTA-Z EGFR:2377 , the hepatic uptake was 3.4-fold lower and the  The images show threedimensional volume rendering of PET or SPECT data overlaid CT data. The images were rotated to provide view of liver. The relative colour scales were normalized to the highest activity and then adjusted to provide first red pixel in tumours (8,13 and 17% of the full scale, for Fig. 4a,b and c, respectively). Arrows point at tumours (Tu), kidneys (Ki) and livers (Li).
tumour uptake was 2-fold higher compared to 68 Ga-DOTA-Z EGFR:2377 at 3 h after injection of the same protein dose. This resulted in significantly higher tumour-to-organ ratios for all measured tissues. Most importantly, the tumour uptake of 57 Co-DOTA-Z EGFR:2377 was significantly higher than the hepatic uptake (tumour-to-liver ratio of 3.1 ± 0.5), which is a precondition for successful imaging of hepatic metastases. This is essential, as liver is the major metastatic site for many cancers. The tumour-to-bone ratio for 57 Co-DOTA-Z EGFR:2377 was also twice as high as for 68 Ga-DOTA-Z EGFR:2377 . At 24 h after injection, the tumour-to-blood ratio for 57 Co-DOTA-Z EGFR:2377 increased 2.7-fold compared with 3 h time point. The exact nature of this phenomenon is not quite clear. An important insight is provided by the fact that hepatic uptake of 68 Ga-DOTA-Z EGFR:2377 was three-fold higher than uptake of 57 Co-DOTA-Z EGFR:2377 (Fig. 3), when a blocking dose of non-labelled Z EGFR:2377 was pre-injected. Data for 111 In-DOTA-Z EGFR:2377 show that saturation of liver uptake by non-labelled Z EGFR:2377 was much less efficient than saturation of uptake in other EGFR-expressing tissues 15 . This suggests that there are at least two different mechanisms of DOTA-Z EGFR:2377 hepatic uptake: one is dependent on EGFR expression (and can be saturated) and one is independent. The use of 57 Co-DOTA-Z EGFR:2377 might reduce the hepatic uptake by the EGFR-independent mechanism. This is quite surprising, as the difference between 57 Co-DOTA-Z EGFR:2377 and 68 Ga-DOTA-Z EGFR:2377 is very subtle. Both Co 2+ and Ga 3+ have a six-coordination sphere and form DOTA complexes with distorted octahedral coordination geometry 22 . The major difference between Co-DOTA and Ga-DOTA complexes is that the cobalt complex carries a negative charge while the gallium complex is neutral. The literature data suggest that such small difference might have dramatic effect on affinity of radiolabelled peptides. For example, affinity of 68 Ga and 57 Co-DOTATOC to somatostatin receptors differs by one order of magnitude 22 . It is conceivable that the complex charge could also have an appreciable effect on DOTA-Z EGFR:2377 affinity to "scavenger receptors" in liver. The reduced liver binding of 57 Co-DOTA-Z EGFR:2377 makes it more available for binding to EGFR-expressing tumours and tissues, which is in agreement with experimental observations. The findings of this study suggest that the positron-emitting 55 Co (T 1/2 = 17.5 h) might be a suitable label for DOTA-Z EGFR:2377 and further development should concentrate on this radionuclide as a label. The half-life is long enough for imaging at 24 h after injection, but is not exceedingly long. The use of a positron-emitting label for DOTA-Z EGFR:2377 would further improve detection of EGFR expression in vivo, since PET provides better imaging sensitivity compared with SPECT.
Earlier, several approaches for preclinical development of PET probes specific to the extracellular domain of EGFR were evaluated. Labelling of the natural ligand, EGF, with 68 Ga 31 or 18 F 32 enabled specific imaging of EGFR-expressing xenografts with tumour-to-blood ratios in the range of 2-5. At low doses, both tracers had liver uptake exceeding tumour uptake several fold. In mice, the hepatic uptake has been reduced appreciably at higher injected protein doses. However, this approach is not feasible clinically because of adverse side effects associated with injection of high doses of EGF (nausea, vomiting, hypotension) 33 . Anti-EGFR therapeutic monoclonal antibodies cetuximab and panitumumab were labelled with long-lived positron emitters 64 Cu (T 1/2 = 12.7 h), 86 Y (T 1/2 = 14.7 h), and 89 Zr (T 1/2 = 78.4 h) [34][35][36][37][38][39] . The maximum tumour-to-blood ratio (range of 2-8) was obtained at 3-4 days after injection due to slow clearance of antibodies from blood. 57 Co-DOTA-Z EGFR:2377 provided a higher tumour-to-blood ratio already at 3 h after injection than any of these antibody-based agents at the best imaging time point. Labelling and preclinical in vivo evaluation of another clone of anti-EGFR Affibody molecule, Z EGFR:1907 , with 18 F and 64 Cu has been reported [40][41][42] . Z EGFR:1907 had an apparent cross-reactivity with murine EGFR, which made the murine model adequate. These probes provided tumour-to-blood ratios of 2-10, depending on the labelling approach. Still, tumour-to-liver ratios were around 1.4 at best. Recently, we performed pre-clinical evaluation of Z EGFR:2377 , which was site-specifically labelled with a long-lived positron emitter 89 Zr (T 1/2 = 78.4 h) using a maleimido derivative of a deferoxamine (DFO) chelator 43 . Direct comparison with 89 Zr-labelled anti-EGFR antibody cetuximab demonstrated that 89 Zr-DFO-Z EGFR:2377 provided appreciably higher tumour-to-blood, tumour-to-liver, and tumour-to-bone ratios. Yet, 57 Co-DOTA-Z EGFR:2377 provided 2-5-fold higher tumour-to-organ ratios compared to 89 Zr-DFO-EGFR:2377 . Taken together, these data indicate that the radiocobalt-labelled DOTA-Z EGFR:2377 demonstrated the best tumour-to-non-tumour ratios among all pre-clinically tested affibody-based agents for imaging of EGFR.
A clinical translation of the finding of this study would require considering the dosimetry difference between 68 Ga and 55 Co. Although the longer half-life of 55 Co permits a substantial improvement of tumour-to-organ ratios, it results in longer residence time and, therefore, higher dose burden on patients, compared to 68 Ga. Another issue is that the decay scheme of 55 Co includes a number of gamma emissions besides positrons, which increases whole-body dose (See Supplementary Table S2). These features are a common concern for all long-lived positron emitters. Still, the much more long-lived positron emitter 89 Zr (T 1/2 = 78.4 h) is used to label antibodies for clinical imaging 44 , and clinical studies show acceptable absorbed dose levels 45,46 . It has to be noted that 55 Co provides a better ratio between numbers of annihilation photons and co-emitted gamma photons compared to 89 Zr and the majority of other long-lived positron emitters (Supplementary Table S2). Compared to the long-lived positron emitter 64 Cu (T 1/2 = 12.7 h), 55 Co has higher positron abundance and does not emit beta particles. 55 Co also provides a better image quality compared to the long-lived positron-emitting radiometal 86 Y (T 1/2 = 14.7 h) 47 .
In conclusion, the chemical nature of a radionuclide can appreciable influence the biodistribution and targeting properties of Affibody molecules. In vivo studies using 57 Co as a model nuclide suggest that radiocobalt-labelled DOTA-Z EGFR:2377 is a very promising tracer for imaging of EGFR expression in vivo. Labelling of DOTA-Z EGFR:2377 with 55 Co providing a conjugate with preserved specificity of binding to EGFR-expressing cells is feasible. This justifies further preclinical development of 55 Co-labelled Affibody molecules for PET-imaging of EGFR expression in malignant tumours. Future studies should concentrate on development of large-scale production of 55 Co-DOTA-Z EGFR:2377 and quantitative PET imaging of xenografts with different EGFR expression levels.
The 68 Ge/ 68 Ga generator (50 mCi) was from Eckert and Ziegler (Vienna, Austria). Hydrochloric acid (0.1 M, prepared from 30% ultrapure HCl from Merck) was used for fractionated elution of the generator. The fraction containing the maximum radioactivity was used for labelling.
Labelling yield and radiochemical purity of the labelled DOTA-Z EGFR:2377 molecule were determined by radio-ITLC using 150-771 DARK GREEN strips (Biodex Medical Systems, New York, NY, USA) eluted with 0.2 M citric acid. In this system, free 68 Ga and 55/57 Co migrate with the solvent front and the radiolabelled DOTA-Z EGFR:2377 stays at the application point. The radio-ITLC was cross-validated by SDS-PAGE. The distribution of radioactivity along the thin layer chromatography strips and SDS-PAGE gels was measured using a Cyclone Storage Phosphor System (PerkinElmer, Wellesey, MA, USA).
The radioactivity uptake in the cellular processing and the biodistribution studies was measured using an automated gamma-counter with 3 inch NaI(TI) detector (1480 WIZARD, Wallac, Turku).
Statistics. Data on cellular uptake and biodistribution were analysed by unpaired, two-tailed t-test using GraphPad Prism (version 4.00 for Windows GraphPad Software) in order to determine significant differences (p < 0.05).
Tracer production and labelling. The C-terminal cysteine-containing anti-EGFR Affibody molecule Z egfr:2377 was produced, conjugated with maleimido-derivative of DOTA and purified as described earlier 15 . Before conjugation, a solution of Affibody molecules was treated with dithiothreitol (DTT; E. Merck, Darmstadt, Germany) in order to reduce spontaneously formed intermolecular disulfide bonds. For this purpose, a stock solution of Affibody molecules (2 ml, 2.3 mg/ml in PBS) was mixed with 100 µl 1 M Tris-HCl buffer, pH 8.0, and 63 µl DTT solution (0.5 M in water). The mixture was incubated at 40 °C for 30 min. The reduced Affibody molecules were then purified and the buffer was changed according to the manufacturer's instructions using a disposable PD-10 column (GE Healthcare) pre-equilibrated with 0.2 M sodium acetate, pH 5.5. Maleimido-mono-amide-DOTA (2 mg) was added and the mixture incubated for 1 h at 40 °C.
The Affibody molecule was purified and analysed by high-performance liquid chromatography and on-line mass spectrometry (HPLC-MS) using an Agilent 1100 LC/MSD system equipped with electrospray ionization and a single quadrupole (Agilent Technologies, Palo Alto, CA). For purification, a Zorbax 300SB C18 9.4 × 250 mm, 5 µm column was used. The analysis was performed using a Zorbax 300SB-C18 2.1 × 150 mm, 3.5 µm column. Solvent A comprised 0.1% trifluoroacetic acid (TFA) in water, and solvent B comprised 0.1% TFA in acetonitrile. The column oven temperature was set to 30 °C. For analysis, the column was eluted with a linear gradient of 35% to 60% solvent B over 18 min, with a flow rate of 0.5 ml/min. The purified conjugate (further designated as DOTA-Z EGFR:2377 ) was freeze-dried. Chemstation Rev. B.02.01 software (Agilent) was used for analysis and evaluation of HPLC data. The conjugate was aliquoted (50 µg/aliquot) and lyophilized.
The thermal stability and secondary structure of DOTA-Z EGFR:2377 were determined using a JASCO J-810 spectropolarimeter (JASCO, Tokyo, Japan). The samples were diluted in PBS to concentrations of 0.5 mg/mL and melting temperatures were assessed by measuring the ellipticity at 221 nm during a temperature gradient ranging from 25-90 °C. The secondary structural content was assessed by measuring the degree of ellipticity from 250 to 195 nm at 25 °C before and after the thermal stability measurement. 55 Co for labelling of Affibody molecules was produced via the 54 Fe(d,n) 55 Co nuclear reaction as described earlier 26 . Briefly, the electroplated 54 Fe target (99.84% enriched, Campro Scientific) was irradiated by 8.5 MeV deuterons using a GE PETtrace cyclotron. After the irradiation the produced 55 Co was separated from the target material on an anion exchange column (Dowex 1•8) using 4 M hydrochloric acid as eluent and further purified using a Chromafix 30-PS-HCO 3 cartridge 28 before evaporation to dryness. Labelling with radiocobalt has been performed according to previously described method 25 . Briefly, the lyophilized DOTA-Z EGFR:2377 (50 µg) was reconstituted in 40 µL of 0.2 M ammonium acetate, pH 5.5, mixed with radiocobalt stock solution (for 57 Co: up to 18 µL, 4-15 MBq; for 55 Co: 30-50 µL, 21-46 MBq) and incubated at 60°C for 30 min. The conjugate was purified using a NAP-5 size-exclusion column equilibrated with PBS.
For labelling with 68 Ga, the 68 Ge/ 68 Ga generator was eluted with 0.1 M hydrochloric acid as eluent, and fractions containing 500 µL were collected. The fraction containing the maximum radioactivity was used for labelling. The lyophilized DOTA-Z EGFR:2377 (50 µg) was reconstituted in 100 µL 1.25 M sodium acetate buffer, pH 3.6. 68 Ga-containing eluate, up to 3 MBq per microgram of DOTA-Z EGFR:2377 , was added to the mixture. The mixture was incubated at 95 °C for 15 min. Thereafter, 1000-fold excess of Na 4 EDTA was added, and the mixture was incubated at 95 °C for 10 min. After labelling, the conjugate was purified using a NAP-5 size-exclusion column equilibrated with PBS.
Labelling of DOTA-Z EGFR:2377 with 111 In for affinity measurements using LigandTracer and for determination of the half maximum inhibition concentration (IC 50 ) was performed as described earlier 15 . For IC 50 measurements, the conjugates were loaded with cobalt, indium and gallium of natural isotopic composition using the same conditions as for radiolabelling.
Scientific RepoRts | 7: 5961 | DOI:10.1038/s41598-017-05700-7 The yield and radiochemical purity of the conjugates were evaluated using ITLC eluted with 0.2 M citric acid, pH 2.0. To cross-validate the ITLC data, a radio-SDS page analysis was performed. The samples were treated with sample buffer (10 min at 70 °C) and analyzed using SDS-PAGE (at 200 V using NuPAGE 4-12% Bis-Tris Gels (Invitrogen AB, Sweden) in MES buffer (Invitrogen AB, Sweden). To provide a marker for low-molecular weight radioactivity, a sample of 68 Ga or 57 Co acetate was applied next to the analytes on the same gel.
To evaluate stability of labelling, the labelled conjugates were incubated with 500-fold excess of EDTA for 2 h at room temperature and then analysed using ITLC.
In vitro characterization of radiolabelled DOTA-Z EGFR:2377 . The EGFR-expressing A431 epidermoid carcinoma cell line (ATCC) was used. Ovarian carcinoma SKOV-3 cell line (ATCC) with very low EGFR expression was used as a negative control in in vivo studies. Cells were cultured using Ham's F10-medium supplemented with 10% foetal bovine serum (Flow, Stockholm, Sweden), L-glutamine (2 mM) and PEST (penicillin 100 IU/mL and streptomycin 100 mg/mL).
The in vitro binding specificity was verified by receptor pre-saturation with unlabelled Affibody molecules as described 15 . To test the binding specificity, 8 nM of radiolabelled DOTA-Z EGFR:2377 was added to 6 petri dishes containing 7 × 10 5 A431 cells/dish. A 50-fold molar excess of non-labelled DOTA-Z EGFR:2377 was added to groups of three petri dishes 15 min before the labelled conjugate to saturate the receptors. The dishes were incubated at 37 °C for 1 h in a humidified incubator. The media was collected, the cells were detached using trypsin-EDTA solution and radioactivity was measured. Percent of cell-bound radioactivity was calculated for both the pre-saturated and unsaturated cells.
The dissociation constant (K D ) of 111 In and 57 Co-labelled DOTA-Z EGFR:2377 binding to living A431 cells was measured by using LigandTracer Yellow (Ridgeview Instruments AB, Uppsala, Sweden) and analysed using InteractionMap software (Ridgeview Diagnostics AB, Uppsala, Sweden) as described earlier 48 . This device records kinetic binding and dissociation of radiolabelled tracers on living cells. In order to cover the concentration span needed for proper affinity estimation, three increasing concentrations of the radioconjugate (0.33, 1, and 3 nM) were added in each affinity assay. Analysis was performed in duplicates.
To evaluate the relative binding strength of gallium, cobalt and indium-labelled Z EGFR:2377 , the half maximum inhibition concentration (IC 50 ) were measured using 111 In-DOTA-Z EGFR:2377 . Monolayers of A431 cells were incubated for 4 h at 4 °C with nat Ga-, nat Co-or nat In-DOTA-Z EGFR:2377 (0-100 nM) in the presence of 10 nM 111 In-DOTA-Z EGFR:2377 . After incubation, the cells were washed with media and detached with trypsin. The detached cells were collected, and the cell-associated radioactivity was measured. The IC 50 values were determined using GraphPad Prism software.
The rate of internalization of 68 Ga-DOTA-Z EGFR:2377 and 57 Co-DOTA-Z EGFR:2377 by EGFR-expressing A431 squamous carcinoma cell during continuous incubation was studied using the acid wash method 49 . The labelled compound (protein concentration of 8 nM) was added to 15 petri dishes containing approximately 7 × 10 5 cells/ dish. The cells were incubated at 37 °C in a humidified incubator. At pre-determined time points (1, 2, 3, and 4 h after incubation start for 68 Ga-DOTA-Z EGFR:2377 ; 1, 2, 4, 8 and 24 h for 57 Co-DOTA-Z EGFR:2377 ), the medium from a set of three dishes was removed. The cells were washed with 1 mL of ice-cold serum free medium. To collect the membrane-bound radioactivity, the cells were treated with 0.5 mL of 0.2 M glycine buffer containing 4 M urea, pH 2.0, for 5 min on ice. Dishes were washed with 0.5 mL acidic buffer followed by 1 mL PBS, and the fractions were pooled. To collect radioactivity internalized by the cells, treatment with 0.5 mL of 1 M NaOH at 37 °C for 30 min was performed. Dishes were additionally washed with 0.5 mL NaOH solution followed by 1 mL PBS, and the alkaline fractions were pooled. The percentage of internalized radioactivity was calculated for each time point.
In vivo studies. The animal experiments were planned and performed in accordance with the national regulation on laboratory animals' protection and were approved by the Ethics Committee for Animal Research in Uppsala. Euthanasia was performed under Rompun/Ketalar anaesthesia. Female BALB/c nu/nu mice were purchased from Taconic M&B (Ry, Denmark). At the time of experiment, the animals were 10 weeks old and the average weight was 21.3 ± 0.8 g. EGFR-expressing xenografts were established by subcutaneous injection of 10 7 A431 cells in the right hind leg. The tumours were grown for 10-14 d. At the time of the study, the average tumour weight was 0.8 ± 0.3 g. The animals were randomized into groups of four. To test if the tumour uptake depends on EGFR expression, an additional group of mice was implanted with 10 7 SKOV-3 cells having very low EGFR expression. A biodistribution experiment was performed 14 days after implantation. At the time of the study, the average animal weight was 19.7 ± 0.8 g and the average tumour weight was 0.16 ± 0.03 g.
For biodistribution measurements, one group of mice was intravenously injected with 68 Ga-DOTA-Z EGFR:2377 (700 kBq in 100 µL PBS per mouse) and two groups with 57 Co-DOTA-Z EGFR:2377 (30 kBq in 100 µL PBS per mouse). The injected protein dose was adjusted to 40 µg/mouse by non-labelled DOTA-Z EGFR:2377 . The animals were sacrificed at 3 h after injection of 68 Ga-DOTA-Z EGFR:2377 and at 3 and 24 h after injection of 57 Co-DOTA-Z EGFR:2377 . Blood and organ samples were collected and weighed, and their radioactivity was measured. Tissue uptake was calculated as percent of injected dose per gram (% ID/g). To check the specificity of in vivo EGFR targeting, groups of four mice were subcutaneously pre-injected with 400 μg non-labelled recombinant Z HER2:342 affibody molecule before injecting 40 μg 68 Ga-DOTA-Z EGFR:2377 or 57 Co-DOTA-Z EGFR:2377 . The control groups were sacrificed at 3 h p.i. A group of mice bearing SKOV-3 xenografts with low EGFR expression was injected with a mixture of 57 Co-DOTA-Z EGFR:2377 (10 kBq per mouse) and 68 Ga-DOTA-Z EGFR:2377 (700 kBq per mouse). The injected protein dose was adjusted to 40 µg/mouse by non-labelled DOTA-Z EGFR:2377 . These mice were euthanized at 3 h after injection. Spectra of each sample and a standard were recorded. For each samples, radioactivity of 57 Co and 68 Ga were calculated by integration of counts in the energy range 10-160 keV and 300-160 keV, respectively. The data were corrected for background, gamma spectrometer dead time for each sample, 68 Ga decay during measurement and spillover of 68 Ga counts into 57 Co energy window.
Imaging was performed to obtain a visual confirmation of the biodistribution data. Mice bearing A431 xenografts were injected with 7 MBq 68 Ga-DOTA-Z EGFR:2377 or 9 MBq 57 Co-DOTA-Z EGFR:2377 . The injected protein dose was 40 µg/mouse. PET imaging was performed at 3 h after injection using Triumph ™ Trimodality system (Gamma Medica) PET/CT device for small rodents. The CT scan was performed at the following parameters: field of view (FOV), 8 cm; magnification, 1.48; one frame and 512 projections for 2.13 min. PET data were acquired in list mode during 30 min and reconstructed using OSEM-3D. CT raw files were reconstructed by filter back projection (FBP). SPECT/CT imaging was performed using nanoScan SC (Mediso Medical Imaging Systems, Hungary) at 3 and 24 h after injection at the following parameters: CT-energy peak of 50 keV, 670 μA, 480 projections, 2.29 min acquisition time. Helical SPECT acquisition was performed using 57 Co energy window (109.89 keV-134.31 keV), 110 projection and matrix of 256 × 256. The time per projection was 32 sec (totally 60 min) for imaging at 3 h and 65 sec (totally 12 min) for imaging at 24 h. CT raw files were reconstructed using Nucline 2.03 Software (Mediso Medical Imaging Systems, Hungary). SPECT raw data were reconstructed using Tera-Tomo ™ 3D SPECT reconstruction technology.