Development of the designed ankyrin repeat protein (DARPin) G3 for HER2 molecular imaging

Purpose Human epidermal growth factor receptor-2 (HER2) overexpression is a predictor of response to anti-HER2 therapy in breast and gastric cancer. Currently, HER2 status is assessed by tumour biopsy, but this may not be representative of the larger tumour mass or other metastatic sites, risking misclassification and selection of suboptimal therapy. The designed ankyrin repeat protein (DARPin) G3 binds HER2 with high affinity at an epitope that does not overlap with trastuzumab and is biologically inert. We hypothesized that radiolabelled DARPin G3 would be capable of selectively imaging HER2-positive tumours, and aimed to identify a suitable format for clinical application. Methods G3 DARPins tagged with hexahistidine (His6) or with histidine glutamate (HE)3 and untagged G3 DARPins were manufactured using a GMP-compatible Pichia pastoris protocol and radiolabelled with 125I, or with 111In via DOTA linked to a C-terminal cysteine. BALB/c mice were injected with radiolabelled G3 and tissue biodistribution was evaluated by gamma counting. The lead construct ((HE)3-G3) was assessed in mice bearing HER2-positive human breast tumour (BT474) xenografts. Results For both isotopes, (HE)3-G3 had significantly lower liver uptake than His6-G3 and untagged G3 counterparts in non-tumour-bearing mice, and there was no significantly different liver uptake between His6-G3 and untagged G3. (HE)3-G3 was taken forward for evaluation in mice bearing HER2-positive tumour xenografts. The results demonstrated that radioactivity from 111In-(HE)3-G3 was better maintained in tumours and cleared faster from serum than radioactivity from 125I-(HE)3-G3, achieving superior tumour-to-blood ratios (343.7 ± 161.3 vs. 22.0 ± 11.3 at 24 h, respectively). On microSPECT/CT, 111In-labelled and 125I-labelled (HE)3-G3 could image HER2-positive tumours at 4 h after administration, but there was less normal tissue uptake of radioactivity with 111In-(HE)3-G3. Preadministration of trastuzumab did not affect the uptake of (HE)3-G3 by HER2-positive tumours. Conclusion Radiolabelled DARPin (HE)3-G3 is a versatile radioligand with potential to allow the acquisition of whole-body HER2 scans on the day of administration. Electronic supplementary material The online version of this article (doi:10.1007/s00259-014-2940-2) contains supplementary material, which is available to authorized users.


Introduction
Overexpression of human epidermal growth factor receptor-2 (HER2) enhances signal transduction through the PI3K/Akt and the Ras/Raf/MEK/MAPK pathways, enabling cancer cell proliferation and survival [1]. HER2 is overexpressed in 20 -25 % of patients with breast cancer and gastrooesophageal cancer, and in these cancers HER2 is an established therapeutic target [2,3]. There are a range of HER2-targeted therapies, including the monoclonal antibody trastuzumab, licensed for the treatment of breast and gastric cancers, and the trastuzumab-drug conjugate, T-DM1, licensed for the treatment of breast cancer [3,4]. Novel therapies targeting the HER2 receptor, downstream effectors and compensatory signalling pathways are in clinical development [1].
Currently, patients are selected for anti-HER2 therapy based on histological analysis, using immunohistochemistry or fluorescence in situ hybridization (FISH) of biopsied or surgically resected tissues [5]. These analyses are limited by the use of single-site and single time-point sampling, and furthermore they fail to provide information about heterogeneity of expression or changes in expression that occur over time [6]. Dependence on histological analyses risks misclassifying patients' HER2 status and selecting a suboptimal therapy. Discordance in HER2 status between the primary breast tumour and synchronous as well as metachronous metastases has consistently been reported [7]. HER2 status has also been reported to change, from negative to positive and vice versa, and also between metastatic disease relapses [8,9]. Although the impact of HER2 misclassification on clinical outcomes is unclear, 20 -30 % of patients with HER2-positive advanced breast cancer do not respond to first-line treatment combinations, despite advances in anti-HER2 therapies [10]. In the adjuvant setting, some patients with breast cancer histologically classified as HER2-negative benefit from adjuvant anti-HER2 therapy, casting doubt on their classification [11].
HER2 molecular imaging could potentially overcome the limitations of histological analysis by providing new information on HER2 expression at all metastatic sites. Radiolabelled trastuzumab has been used for this purpose, but although the pharmacokinetics of trastuzumab are well suited to therapy, its long half-life compromises tumour-to-blood ratios needed for imaging. Consequently, HER2 scans must be performed at least 2 days after radiolabelled trastuzumab administration, and even then tumours in highly vascular regions may be difficult to visualize [12]. In clinical studies, SPECT imaging with 111 In-trastuzumab and PET imaging with both 64 Cutrastuzumab and 89 Zr-trastuzumab have shown lower rates of tumour detection in patients with advanced HER2positive breast cancer than conventional imaging, including CT [12][13][14]. Therefore, a novel approach is needed to overcome these limitations, potentially utilizing low molecular weight high-affinity proteins [15].
Designed ankyrin repeat proteins (DARPins) are recombinant binding proteins composed of ankyrin repeats, which stack together to make a contiguous binding surface [16,17]. Each ankyrin repeat consists of 33 amino acids, which form a β-turn followed by two anti-parallel α-helices and a loop binding to the β-turn of the next ankyrin repeat [18,19]. Synthetic DARPin libraries have been designed [18], from which specific binders can be selected and further evolved by methods such as ribosome display [20].
The DARPin G3 is a low molecular weight protein (14 -15 kDa) with picomolar affinity (91 pmol/L) for HER2 [21][22][23]. It has a short half-life in mice (<3 min), lacks biological activity and binds to HER2 in the presence of trastuzumab and pertuzumab in vitro [21,23]. The DARPin G3 tagged with hexahistidine (His 6 ) and labelled with 99m Tctricarbonyl ([ 99m Tc(CO) 3 ] + ) can be used to visualize HER2positive tumours [21]. Our goal was to use the DARPin G3 for routine clinical HER2 SPECT and PET imaging, and we set out to generate and evaluate different radiolabelled G3 formats to select a lead for clinical development.
The His 6 tag can be employed for purification of recombinant proteins by immobilized metal affinity chromatography (IMAC). We evaluated the effect of the His 6 tag on DARPin G3 biodistribution and compared it with the DARPin G3 tagged with histidine glutamate (HE) 3 , which has a negative excess charge at physiological pH. The (HE) 3 tag has been reported to reduce background liver uptake in some cases, while still allowing tag-mediated IMAC [24,25]. We developed a GMP-compatible Pichia pastoris production platform that allows cleavage of histidine-based tags after IMAC purification, enabling comparisons among variants of G3.
We hypothesized that the DARPin G3 would be capable of selectively imaging HER2-positive tumours and aimed to identify a suitable format for clinical application. Thus, we systematically investigated the effect of tag and label on the quality of imaging. First, we assessed the sensitivity and specificity of DARPin G3 radiolabelled with [ 99m Tc(CO) 3 ] + via a His 6 tag in HER2-positive and HER2-negative tumourbearing mice. Subsequently, we assessed the biodistribution of His 6 -G3, (HE) 3 -G3 and untagged G3 DARPins radiolabelled with 111 In and 125 I in non-tumour-bearing mice. Thus, both residualizing and non-residualizing radioisotopes were tested, as they have different cellular fates which can affect tumourto-normal tissue ratios. Finally, the construct with the lowest normal tissue uptake was taken forward for evaluation as an imaging agent.
DARPin G3 radiolabelling DOTA-conjugated DARPins (5 -60 μg) in 0.2 M ammonium acetate, pH 6.5, were mixed with a solution of 111 InCl 3 (Covidien, The Netherlands; 10 -30 MBq) and incubated for 2 h at 37°C (reaction volumes 40 -60 μl). The reactions were stopped by adding 0.1 M disodium edetate (EDTA) and the radiolabelled DARPins were purified by elution into PBS using a NAP-5 column (GE Healthcare, Little Chalfont, UK) pre-equilibrated with PBS. Radiochemical purity was determined using instant thin-layer chromatography (iTLC), using iTLC silica gel (SG) strips (Varian, Palo Alto, CA). To test for 111 In-EDTA, iTLC strips were eluted with 0.1 M ammonium acetate containing 25 mM EDTA (final pH 5.5) in which system 111 In-EDTA eluted to the solvent front, while 111 In-G3 DARPin and insoluble material remained at the origin. Formation of radioactive insoluble material was evaluated using iTLC-SG eluted with 35 % ammonia (v/v)/ethanol/ water (1:2:5), in which system 111 In-DOTA-G3 DARPin and 111 In-EDTA both had Rf values >0.5, while any insoluble material present in the reaction mixture remained at the origin. If insoluble material was detected, reaction mixtures were filtered through a 0.2-μm sterile syringe filter with a Supor membrane (Pall Life Science, Portsmouth, UK). The radiochemical purity of 111 In-G3 DARPins was 70 -80 % before purification and >95 % after purification (see below for specific activities, SA).
Iodine radiolabelling was performed in precoated Pierce iodination tubes (Thermo Scientific, Runcorn, UK) with unconjugated G3 DARPins (5 -60 μg) in PBS for 10 min at room temperature, using either 10 MBq 125 I (PerkinElmer, Llantrisant, UK) or 15 -20 MBq 123 I (GE Healthcare). The iodination reactions were stopped by adding sodium metabisulphite to a final concentration of 1 μM. Radioiodinated DARPins were purified by buffer exchange into PBS with a NAP-5 column. Radiochemical purity was assessed with iTLC-SG strips (Varian) using 0.1 M ammonium acetate containing 25 mM disodium EDTA (final pH 5.5) as 123/125 I-G3 DARPin remained at the origin and free 123/125 I eluted at the solvent front. Radiochemical purity increased from 60 -70 % before purification to >95 % after purification (see below for SA). An equivalent molar amount of 'cold' potassium iodide (3.4×10 −10 mol) was coadministered with 123 I-G3 DARPin to minimize normal tissue uptake of 'free' 123 I (formed by deiodination in vivo). iTLC analysis was supplemented with size-exclusion highperformance liquid chromatography (HPLC). A Beckman System Gold 128 solvent module and a 168 UV detector module (monitoring at 280 and 220 nm; Beckman Coulter, High Wycombe, UK), combined with a Raytest GABi radiochemical detector (Raytest, Straubenhardt, Germany) were used. The radiolabelled DARPin G3 was eluted from a YMC-Pack Diol-60 (YMC Europe, Dinslaken, Germany) column, dimensions 300 mm length × 8.0 mm inner diameter, spherical shape, 5 μm particle size and 6 nm pore size, using a mobile phase of 0.2 M phosphate buffer (pH 6.8) at a flow rate of 0.5 ml/min.

In vitro assessment
The unconjugated and DOTA-conjugated G3 counterparts had comparable subnanomolar binding affinities for the HER2 extracellular domain (ECD) assessed by surface plasmon resonance, indicating that DOTA conjugation to the Cterminus does not compromise binding to HER2 (data not shown). Saturation binding assays of radiolabelled DARPin G3 were performed with BT474 HER2-expressing human breast cancer cells (ATCC, Manassas, VA) as previously described [26]. On BT474 cell-binding assays, 111 In-(HE) 3 -G3 and 125 I-(HE) 3 -G3 had saturable binding and similar binding affinities to each other.
To assess stability of the label, aliquots of radiolabelled DARPin G3 were incubated in PBS or in human serum/PBS (1:1). Serum stability was assessed at 37°C for 24 h, while PBS stability was assessed at 4, 20 and 37°C for 24 h after radiolabelling by iTLC, as previously outlined. After incubation, the stability of radiolabelled G3 was assessed by SDS-PAGE using non-reducing conditions, followed by autoradiography using a Cyclone storage phosphor system (PerkinElmer).

Biodistribution studies
All animal studies were ethically reviewed and performed in accordance with the UK Animals (Scientific Procedures) Act 1986, UK Home Office regulations and local regulations.

MicroSPECT/CT imaging
Female SCID-beige mice bearing BT474 tumours were injected intravenously with either 10.5 MBq of 123 I-G3 (2.1 MBq/μg, SA 30.1 MBq/nmol) or 8.4 MBq of 111 In-G3 (4 MBq/μg, SA 58 MBq/nmol). Preliminary work had demonstrated that HER2-positive tumour signals on SPECT scans were compromised at lower SAs of radiolabelled G3 DARPin (data not shown). Thus, the SA for imaging was increased compared to the biodistribution studies to enhance the quality of images, but mice assessed for biodistribution and imaging studies received a similar molar dose of G3 DARPin. The radiolabelling reactions for imaging studies were performed with both a lower amount of G3 DARPin and higher radiation activity within the parameters outlined in the radiolabelling section.
Imaging was performed under 2 % isoflurane anaesthesia 4 h after administration using a microSPECT/CT animal scanner (Bioscan, Poway, CA). Helical SPECT images were acquired in 20 projections over 30 -40 min using a four-headed camera with 4×9 (1.4 mm) pinhole collimators. The CT images were acquired in 180 projections with an exposure time of 1,000 ms using a peak kilovoltage of 45 kVp over 6 min. Radionuclide images were reconstructed using HiSPECT (Scivis, Göttingen, Germany) iterative reconstruction software and fused with CT images using proprietary InVivoScope (Bioscan) software.

DARPin G3 specificity for HER2 in vitro
The DOTA-conjugated G3 DARPins were labelled using naturally abundant indium chloride ( nat InCl 3 ; Sigma-Aldrich, St. Louis, MO) in 0.05 M HCl solution, and unconjugated G3 DARPins were labelled using Na nat I (Sigma-Aldrich) in 0.1 M sodium hydroxide, using the same conditions required for 111 In and 125 I radiolabelling reactions, respectively.
The assays were stopped by removing the medium and washing the cells with 2 ml ice-cold PBS. The cells were lysed with 1 ml of 1 M sodium hydroxide and the lysate was collected. The wells were washed twice with 1 ml of PBS. The washes were pooled with the cell lysate for analysis. The radioactivity was measured in a gamma counter. The mean percentage of radioactivity was calculated for each condition in relation to the controls treated with [ 99m Tc(CO) 3 ] + -G3 DARPin-His 6 alone. EDTA for 10 min. The cells were then transferred to tubes and centrifuged (4°C, 4 min, 1,000 rpm). The EDTA solution was removed and 5 ml of fresh medium was added. Cells were counted and diluted to 1 million/ml, and 1 ml was used for each test condition. After washing with cold PBS, 200 μl of DARPin at 10 μg/ml was added followed by incubation for 1 h at 4°C. Cells were washed with cold PBS and incubated with 200 μl of mouse anti-DARPin (K. Chester, unpublished) for 1 h at 4°C. Subsequently the cells were washed with cold PBS and incubated with 200 μl of Alexa Fluor® 488 goat antimouse IgG (Life Technologies, Paisley, UK) for 30 min at 4°C. After further washing with cold PBS, the cells were suspended in 500 μl of cold PBS. Samples were analysed on a CyAn ADP high-performance flow cytometer (Becton Dickinson, Oxford, UK); cells were gated according to size scattering, forward scattering and pulse width so only single cells were analysed. A total of 10,000 cell events were recorded per sample and data were analysed using FlowJo software (Tree Star, Ashland, OR).

Statistical analysis
An independent samples t test was performed with SPSS Statistics 21 software (IBM, Armonk, NY) to compare normal liver uptake between the different radiolabelled G3 DARPins (p values <0.05 were considered statistically significant).

In vitro stability of label
Over 24 h, 111 In-(HE) 3 -G3 DARPin was stable in PBS at 4, 20 and 37°C and in serum at 37°C, as >95 % of radiation activity remained bound to (HE) 3 -G3 and each sample contained a single-sized radiolabelled protein that had the appropriate molecular weight (Supplementary Table 1 and Supplementary Fig. 2).
Flow cytometry demonstrated that G3 DARPins labelled with cold indium and cold iodine bound to HER2-positive human breast cancer cells (BT474) and HER2-positive human gastrooesophageal junction (OE-19) adenocarcinoma cells, but did not bind to HER2-negative human breast adenocarcinoma cells (MDA-MB-468). The non-labelled G3 DARPin-His 6 and the assessed cold-labelled DARPins demonstrated HER2 specificity in vitro (Fig. 3). Thus, HER2 specificity of His 6 -G3, (HE) 3 -G3 and untagged G3 DARPins in vitro was confirmed.   Table 2). At 4 h, 111 In-(HE) 3 -G3 had significantly lower liver uptake than 111 In-His 6 -G3 (p = 0.001) and 111 In-untagged-G3 (p=0.001). Also at 24 h after administration, 111 In-(HE) 3 -G3 liver uptake was significantly lower than that of 111 In-untagged-G3 (p=0.002) and 111 In-His 6 -G3 (p=0.001). Interestingly, there was no significant difference in liver uptake between the 111 In-untagged-G3 and 111 In-His 6 -G3 at 4 and 24 h after administration (Fig. 4). This suggests a specific favourable influence of the (HE) 3 -tag, rather than some level of liver targeting of the His 6 tag. Bone uptake was largely attributed to the marrow, as marrow uptake was similar to intact bone uptake. Kidney uptake was greater than 200 % ID/g for all 111 In-G3 DARPins at 4 h ( Fig. 4 and Supplementary Table 2).
In the study of the radioiodinated DARPins (Fig. 5), where we also have to take enzymatic dehalogenation into account,  Table 3).
Normal tissue uptake of 125 I was <1.3 % ID/g for the 125 Ilabelled G3 DARPins at 24 h, while for 111 In-G3 DARPins normal tissue uptake was better maintained between 4 and 24 h (Figs. 4 and 5).

Biodistribution in mice bearing HER2-positive tumours
The biodistribution of (HE) 3 -G3 was assessed in HER2 tumour-bearing mice, and this construct was chosen on the basis of its lower normal tissue uptake (Figs. 4 and 5), which could facilitate imaging of HER2-positive tumours by achieving greater contrast between tumours and normal tissues. The normal tissue uptake of both 111 In-(HE) 3 -G3 and 125 I-(HE) 3 -G3 in tumour-bearing mice was similar to that in non-tumourbearing mice (Supplementary Figs. 3 and 4). Similarly, normal tissue uptake of 111 In-(HE) 3 -G3 in tumour-bearing mice was lower than that of 125 I-(HE) 3 -G3 at 4 h, except in the kidneys. At 24 h, the differences in normal tissue uptake between 111 In-(HE) 3 -G3 and 1 2 5 I-(HE) 3 -G3 were smaller (Supplementary Table 4). Kidney uptake of 111 In-(HE) 3 -G3 was higher than that in other tissues tested; it peaked at 4 h after administration at 232.0±24.1 % ID/g and decreased to 196.5±31.0 % ID/g at 24 h after administration (Fig. 6a, b and Supplementary Table 4).
MicroSPECT/CT imaging HER2 tumour uptake could be detected on the 4-h 123 I-(HE) 3 -G3 microSPECT/CT scan but there was non-specific uptake in the stomach, kidneys, bladder and thyroid. In contrast, on the 4-h 111 In-(HE) 3 -G3 microSPECT/CT scan, HER2 tumour uptake was detected and there was minimal non-specific uptake in non-renal organs (Fig. 7).
The G3 DARPins were radioiodinated directly at their tyrosine residues. It is likely that only one of the two tyrosine residues within each G3 DARPin was radioiodinated, owing to the molar ratios of radioiodide and G3 DARPin used for radiolabelling. Although direct radioiodination is convenient and the most widely used methodology, it is recognized that the radioligand is likely to undergo in vivo dehalogenation. In future studies, we plan to compare methods using bifunctional coupling agents, to determine whether this approach can improve in vivo stability of radioiodinated G3 DARPins [33]. Nanobodies (12 -15 kDa) are isolated from the heavy chain of Camelidae antibodies. The anti-HER2 5F7GGC Nanobody radioiodinated via the residualizing agent N-succinimidyl 4guanidinomethyl 3-125/131 I-iodobenzoate (*I-SGMIB-Nanobody) has superior targeting of HER2-positive human breast carcinoma in vitro and in vivo than its directly radioiodinated counterpart, as *I-SGMIB-Nanobody has enhanced tumour retention of radioactivity [34].
Importantly, we observed that 111 In-(HE) 3 -G3 was able to bind to HER2 in the presence of trastuzumab. This confirms the results of structural modelling which demonstrate that DARPin G3 and trastuzumab bind to non-overlapping epitopes of HER2 domain IV [35]. Thus, 111 In-(HE) 3 -G3 DARPin has the potential to image both treatment-naive patients and patients receiving concomitant trastuzumab without requiring a delay to treatment. 111 In-(HE) 3 -G3 imaging would provide completely new information with the potential to improve understanding of HER2 breast cancer evolution and heterogeneity. Current literature based on biopsy sampling is inconsistent with a wide-range of reported rates of HER2 expression discordance between primary and metastatic sites, from 0 to 34 % [7]. Whole-body assessment of HER2 expression at tumour sites over time and during therapy with HER2 imaging is both more feasible and acceptable to patients than serial and multiple biopsies. Furthermore, considerably less is known about gastric cancer HER2 biology, and 111 In-(HE) 3 -G3 imaging could help address this knowledge gap [36].
HER2 is an important therapeutic target in cancer. DARPin HER2 imaging could be used to improve the selection of patients for anti-HER2 therapy, by identifying increased HER2 expression and the need to introduce anti-HER2 therapy, as well as HER2 loss, necessitating discontinuation of anti-HER2 therapy [1]. This could not only save patients from unnecessary treatments, but also improve healthcare economics by ensuring appropriate use of expensive anti-HER2 therapies.
The pharmacokinetics of 111 In-(HE) 3 -G3, including short half-life in serum, high tumour-to-blood ratios and low nonrenal tissue uptake, are well suited to HER2 imaging. The kidneys received the highest radiation dose. However, based on medical internal radiation dose (MIRD) estimates, the maximum absorbed kidney dose in patients with the proposed clinical dose of 111 In-(HE) 3 -G3 (300 MBq) is 1,000-fold lower than the dose associated with a 5 % rate of radiation nephritis within 5 years of administration (TD 5/5 ; 9.75 mGy vs. 10 Gy) [37].
Imaging trials of most HER2 radioligands have had disappointing results due to limitations related to their pharmacokinetics for imaging which were already apparent in preclinical assessment [28][29][30][31]. It should be noted that the majority of clinical HER2 imaging studies have not assessed HER2 tumour status histologically for correlation with HER2 scans, thus limiting the evaluation of these imaging agents. Trastuzumab radiolabelled with 64 Cu, 89 Zr or 111 In has demonstrated the potential for HER2 imaging by identifying known and/or previously unknown tumour lesions in patients, but lacks sensitivity compared with conventional imaging owing to a long half-life and high blood pool activity [12][13][14]. Fab fragments of trastuzumab were developed for HER2 imaging because their lower molecular weight is associated with a shorter half-life in serum and faster clearance of background radiation from the blood compared with trastuzumab. However, 68 Ga-F(ab') 2 -trastuzumab PET imaging failed to identify any tumours among four of the eight patients with metastatic HER2-positive breast cancer assessed, due to relatively high liver uptake and blood pool activity as well as potential competition with therapeutically administered trastuzumab [38].
A radiolabelled HER2 binding Affibody molecule, Z HER2:342 (about 7 kDa), was used to assess three patients with HER2positive metastatic breast cancer by PET and/or SPECT imaging. Although not all known tumour lesions were identified in two patients, there was limited tumour biopsy sampling to systematically evaluate the accuracy of HER2 Affibody molecule imaging. Unfortunately, there was high background liver uptake in this initial study [39]. Recently, a phase I trial demonstrated that the second generation Affibody 111 In-ABY-025 has improved distribution, dosimetry and accuracy in assessing patients with HER2-positive and HER2-negative metastatic breast cancer. There was a good correlation between HER2 status assessed by SPECT/CT 111 In-ABY-025 imaging and immunohistochemistry with the HercepTest, including confirmation that a patient who had had HER2-positive primary disease developed HER2-negative metastases [15].
In our evaluation of the DARPin G3, the (HE) 3 tag is a beneficial component as it appeared to lower normal liver uptake of the DARPin G3 without compromising uptake in other assessed normal tissues. This is advantageous for clinical application as HER2 molecular imaging requires high tumour-to-liver tissue ratios and the liver is a common site for breast cancer metastases, yet is also involved in drug metabolism and excretion. Histidine-based tags enable IMAC purification and can also be used for chelation of [ 99m Tc(CO) 3 ] + [21,27]. The His 6 tag is well established and has been used safely in patients [40], but (HE) 3 -G3 DARPin was superior to His 6 -G3. Affibody molecules with an (HE) 3 tag also have lower normal liver uptake than counterparts with alternative histidine-based tags [24,25]. The G3 DARPins assessed in this study only differed by the presence or composition of their tags and thus the study was able to confirm that (HE) 3 -G3 DARPin had the lowest normal liver uptake.
The mechanism of (HE) 3 tag-mediated reduction in liver uptake of DARPin G3 has not been established. For Affibody molecules it has been proposed that positive charge and hydrophobicity in the tag play a crucial role in liver uptake [25]. However, the untagged and His 6 -tagged Affibody molecules have basic isoelectric points (pI), which are brought to a more acidic region by switching to an (HE) 3 tag, such that the overall pI of the different Affibody constructs tested differed greatly. By contrast the (HE) 3 -tagged, His 6 -tagged and untagged G3 DARPins have similar pI values of 4.79, 5.41 and 4.71, respectively [25]. Furthermore, the (HE) 3 -tagged, His 6tagged and untagged G3 DARPins have similar grand average Although in preclinical studies histidine-tagged proteins have been shown to exert greater immunogenicity than their untagged counterparts, several histidine-tagged proteins have been well tolerated in human trials [40]. For example, Endostar a novel recombinant human endostatin which inhibits angiogenesis has a histidine tag, and was safe and well tolerated in large clinical trials [41]. Interestingly, a Blast search (NCBI) for a short input sequence with HEHEHE as the query, yielded a variety of hits in the database of almost identical sequences, e.g. HEHEH in the human zinc transporter (gene ID SLC39A9) and HEHEQE in kinase 3 (gene ID TAOK3). The HHHHHH sequence cannot be found using a Blast search in the Homo sapiens database. Instead clusters of histidine are found that are coordinated with metal ions, e.g. Ca 2+ , Mg 2+ and Zn 2+ . The occurrence of the HEHEH sequence in human proteins could mean that the HE 3 tag is a potentially safer alternative to the His 6 tag.

Conclusion
A clinically valuable radioligand for HER2 molecular imaging in breast cancer and gastric cancer would require minimal normal liver and stomach uptake, as well as the ability to bind to HER2 in the presence of concomitant anti-HER2 therapy. We have demonstrated that 111 In-(HE) 3 -G3 DARPin has specificity for HER2, binds in the presence of trastuzumab, and achieves high tumour-to-blood ratios and reasonable tumour to non-renal tissue ratios, including tumour-to-liver and tumour-to-stomach ratios. Based on the presented preclinical data, 111 In-(HE) 3 -G3 could realize the clinical potential of HER2 imaging and may be suitable for assessment in a first-in-human study. Table 2 Tumour-to-normal tissue ratios of 111 In-(HE) 3 -G3 and 125 I-(HE) 3 -G3 in female SCID-beige mice bearing HER2-positive human breast tumours (BT474)