Phage Display Selection, In Vitro Characterization, and Correlative PET Imaging of a Novel HER3 Peptide

Abstract

Purpose:

HER3 (ERBB3) is a receptor tyrosine kinase that is implicated in treatment resistance across multiple cancers, including those of the breast, lung, and prostate. Overexpression of HER3 following targeted therapy can occur rapidly and heterogeneously both within a single lesion and across sites of metastasis, making protein quantification by biopsy highly challenging. A global, noninvasive methodology such as positron emission tomography (PET) imaging can permit serial quantification of HER3, providing a useful approach to monitor HER3 expression across the entire tumor burden both prior to and following treatment. PET imaging of HER3 expression may permit a more personalized approach to targeted therapy by allowing for detection of HER3-mediated resistance, in addition to informing clinical trial patient selection for novel therapies targeting HER3.

Procedures:

Phage display selection targeting the HER3 extracellular domain was performed in order to develop a peptide with optimal blood clearance and highly accurate HER3 quantification.

Results:

The selection converged to a consensus peptide sequence that was subsequently found to bind HER3 with an affinity of 270 ± 151 nM. The peptide, termed HER3P1, was bound with high selectivity to HER3 over other similar receptor tyrosine kinases such as EGFR and HER2. Furthermore, HER3P1 was able to distinguish between high and low HER3-expressing cells in vitro. The peptide was radiolabeled with Ga-68 and demonstrated to specifically bind HER3 by in vivo PET imaging. Uptake of [⁶⁸Ga]HER3P1 was highly specific for HER3-positive tumors, with tumor-to background ratios ranging from 1.59–3.32, compared to those of HER3-negative tumors, ranging from 0.84–0.93. The uptake of [⁶⁸Ga]HER3P1 also demonstrated high (P < 0.001) correlation with protein expression as quantified by Western blot and confirmed by biodistribution.

Conclusions:

HER3P1 accurately quantifies expression of HER3 by PET imaging and has potential utility as a clinical imaging agent.

Introduction

Molecularly targeted cancer therapy, while providing benefits for a number of indications, has failed to produce durable responses in a majority of patients [1, 2]. While no single mechanism is entirely responsible for targeted therapy resistance, a recurring theme is the upregulation of feedback loops that circumvent blockade of oncogenic signaling pathways [3, 4]. Feedback loop signaling can be accomplished through alternative activation of a homologous pathway or through overexpression of a secondary oncogenic protein [5]. One such protein, human epidermal growth factor receptor 3 (HER3), has been implicated in targeted therapy resistance in a number of malignancies including breast, lung, and prostate cancer [4].

HER3 is a member of the epidermal growth factor (EGF) family of receptors that includes EGFR, HER2/ErbB2/neu, HER3/ErbB3, and HER4/ErbB4 [6]. The EGFR family canonically functions through ligand binding followed by dimerization, phosphorylation, and downstream signaling through the MAPK and PI3K/AKT pathways [6–8]. These receptors are major drivers of tumorigenesis, and multiple targeted therapies that block EGFR and HER2 are approved by the Food and Drug Administration [1, 9, 10]. Often, however, therapies that target EGFR and HER2 are only transiently effective, which may be due, in part, to a subsequent increase in membrane HER3 expression [4]. Upregulation of HER3 expression and transphosphorylation may lead to escape from therapeutic inhibition through the PI3K/AKT pathway, rendering initial targeted therapy ineffective. Although HER3 has been less explored historically due to its mutated kinase domain, concurrent discoveries that its intracellular domain can in fact signal through transphosphorylation and that it is overexpressed in patients with resistance to a number of targeted therapies have brought it to the forefront of targeted therapy development [4, 11, 12]. In fact HER3-mediated therapeutic resistance may not be limited to therapies that target EGFR or HER2, as evidence has also emerged to link HER3 with castration-resistant prostate cancer [13, 14].

Given its central role in targeted therapy resistance, a method to detect HER3 expression could provide oncologists with critical data necessary to inform treatment. While biopsy has traditionally been used to characterize protein expression, the rapid and transient nature of HER3 expression would require serial biopsies, which are not routinely performed, in order to accurately detect changes in expression [15]. A more tractable option to detect and quantify HER3 would be through positron emission tomography (PET) imaging. PET imaging provides rapid, non-invasive, and quantitative measurement of protein expression and can be performed at multiple time points in order to assay pretreatment and post-treatment levels of HER3 [16]. Due to the importance of HER3 in cancer therapeutic resistance biology, a clinically translatable HER3 PET imaging agent could help to diagnose targeted therapy resistance prior to tumor progression. This information could also be used to better select patient populations for current HER3 therapy clinical trials, which have yet to be successful despite growing evidence of the role of HER3 in tumor progression.

In order to generate a suitable clinically translatable imaging agent, the molecule should not only have high affinity and specificity for its target, but also demonstrate rapid clearance from the bloodstream and excellent tumor penetration. Peptides, which have a molecular weight of approximately 1–5 kDa and are non-immunogenic, demonstrate low off-target accumulation and are cleared through the renal system [17]. Given these favorable properties, small peptides or peptidomimetics such as DOTATOC/TATE and PSMA are used clinically or in clinical research in both the USA and Europe [18, 19]. In order to develop a peptide that targets HER3, a combinatorial selection technique termed bacteriophage (phage) display was used [20]. Phage display utilizes libraries of up to one billion unique peptides encoded in the DNA of filamentous viruses to pan for high affinity and specific peptides and has been used to successfully identify peptides that target HER2, the α_Vβ₃ integrin, and numerous other targets [21–23].

It was hypothesized that phage display would permit selection of a HER3 peptide that would have favorable properties for in vivo PET imaging. Therefore, we performed a phage display selection against the extracellular domain of HER3, and a peptide sequence that targeted HER3 was discovered. The peptide was characterized in vitro prior to analysis in HER3-expressing cancer models in vivo. The HER3 peptide was able to discriminate with high accuracy not only between high and low HER3 tumors, but also individual levels of HER3 expression among high-HER3 tumors. Because it demonstrates exquisite sensitivity in HER3 quantification, the novel HER3 peptide represents an attractive clinically translatable imaging agent for noninvasive detection of HER3 levels.

Materials and Methods

Materials

All chemicals and solvents were obtained from Sigma-Aldrich (St. Louis, MO) unless otherwise stated. MDA-MB-453, HCC1954, and 22Rv1 cells were purchased from American Type Culture Collection (Manassas, VA) and cultured in RPMI-1640 media supplemented with 10 % fetal bovine serum and 1 % penicillin and streptomycin.

Phage Display Selection

A cysteine-constrained randomized 7-mer library (New England Biolabs, Ipswich, MA) was utilized for the phage display selection. HER3 extracellular domain (ECD) (R&D Systems, Minneapolis, MN) was conjugated with long chain biotin (Thermo Scientific, Waltham, MA) utilizing standard NHS ester chemistry and purified by size exclusion chromatography with a 10-kDa molecular weight cutoff column (Genesee Scientific, San Diego, CA). Purified, biotinylated HER3 was bound to Dynabeads M-280 streptavidin beads (Thermo Scientific) and blocked with 1 % non-fat dry milk in Tris-buffered saline (TBS) plus 0.1 % Tween 20 (TBST). An aliquot of 2 × 10¹¹ plaque-forming units of phage was added to the beads and incubated for 1 h at 37 °C. The beads were then washed 10 times with TBST and eluted with 0.2 M glycine, pH 2.2, and subsequently neutralized by addition of 1 M Tris pH 8 to pH 7.2 The output was then amplified in ER2738 Escherichia coli (New England Biolabs) for 4.5 h at 37 °C followed and purified by the standard polyethylene glycol (PEG)/NaCl method [24]. The selection was performed for three rounds and following the third round, individual phage plaques were picked for analysis by enzyme-linked immunosorbant assay (ELISA).

Individual Phage Characterization

Individually selected phages were amplified for 4.5 h at 37 °C and supernatant was collected following centrifugation. HER3 ECD and bovine serum albumin as a control were adsorbed to Nunc Maxisorp 96-well plates (Sigma, St. Louis, MO). Supernatants corresponding to individual phages were added to HER3 and bovine serum albumin (BSA) wells and allowed to bind for 1 h at 37 °C. After incubation with peptide, wells were washed six times with TBST and bound phages were detected by the addition of a horseradish peroxidase (HRP)-conjugated anti-M13 antibody (GE Health Sciences). After a subsequent washing, 2,2′-Azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) substrate was added to wells and absorbance at 405 nm quantified by a Promega Glomax spectrophotometer (Promega, Madison, WI). The relative binding was compared to a control phage bearing no peptide and represented by the heat map in Fig. 1a. The phages with the five best binding ratios of HER3 to BSA were chosen for amplification and purification to ensure no other supernatant factor was obscuring the signal. Following purification of the phages, ELISA was performed in the exact same manner and the absorbance of each phage to either HER3 or BSA was quantified by the spectrophotometer. Phagemid DNA from each of the five phages in addition to three phages with no binding per the initial ELISA was recovered by plasmid miniprep isolation and automated Sanger DNA sequencing was performed by the CCIB DNA Core Facility at the Massachusetts General Hospital (Cambridge, MA).

Fig. 1.

a Phage display selection for HER3 peptide heat map of standardized phage supernatant binding to HER3 and BSA as control. Green indicates high binding and red indicates low binding. b Normalized binding of the five highest phages from supernatant phage ELISA; bars represent the mean of six replicates ± SEM. c Sequences of phages with high binding (1, 5, 17, 19, 26) and low binding (7, 22, 27) to HER3. **P < 0.01.

In Vitro HER3 Affinity and Specificity Characterization

After determination of a consensus peptide-binding sequence, the peptide sequence was covalently linked to a biotin-conjugated N-terminal triglycine linker using standard Fmoc chemistry. The purity and molecular weight were determined by high-performance liquid chromatography and mass spectrometry. In order to confirm HER3 affinity, the peptide was analyzed for binding to HER3 ECD immobilized to Nunc Maxisorp 96-well microtiter plates. Following blocking with 1 % non-fat dry milk, increasing concentrations of peptide were incubated with target protein for 1 h at 37 °C. Wells were washed with 0.1 % TBST, and bound peptide detected by addition of HRP-conjugated streptavidin (Abcam, Cambridge, UK), washing and ABTS substrate incubation for 10 min. Absorbance at 405 nm was read by the spectrophotometer.

In order to compare the specificity for HER3 against similar family members EGFR and HER2, extracellular domains of both EGFR and HER2 were immobilized and subjected to the same peptide ELISA at a single peptide concentration of 250 nM. Plates were washed and bound peptide once again detected by addition of streptavidin-HRP and ABTS and absorbance read at 405 nm.

HER3 Peptide Cellular Binding and Specificity Analysis

Although the peptide bound pure HER3, its binding capability in the context of the extracellular microenvironment needed to be analyzed prior to in vivo analysis. Two cell lines, MDA-MB-453 breast cancer cells which express moderate levels of HER3 and HCC-1954 cells which express low levels of HER3, were chosen to assess the ability of the peptide to discriminate between relatively close expression levels in vitro. Cells were seeded at a density of 1 × 10⁵ cells/well in 96-well plates (Fisher) and grown for 24 h in fetal bovine serum supplemented medium. HER3 peptide or a control peptide was added at 250 nM in media and incubated for 1 h at 37 °C. Cells were washed 3× with PBS and bound peptide detected by addition of streptavidin-HRP. Cells were once again washed and bound peptide detected by addition of ABTS and absorbance measured at 405 nm.

Peptide binding was confirmed visually by fluorescent microscopy. MDA-MB-453 and HCC-1954 cells were fixed in 10 % formalin and dried onto microscope slides overnight. Following rehydration with TBS, slides were blocked with 2 % BSA in TBS for 1 h prior to addition of peptide at a concentration of 250 nM in 0.1 % TBST and incubation at room temperature for 1 h. Cells were washed 3× with 0.1 % TBST and neutravidin-AlexaFluor488 was added to cells and incubated at room temperature for 1 h. Cells were washed and mounted with 4′,6′-diamidino-2-phenylindole, dihydrochloride (DAPI)-containing Vectashield mounting media (Vector Labs, Burlingame, CA). Peptide binding was visualized by inverted microscope (Olympus, Tokyo, Japan).

In order to analyze whether the substitution of biotin for 1,4,7-triazacyclononane-N,N′,N″-triacetic acid (NOTA), a bifunctional chelator that would permit radiolabeling and in vivo assessment of the peptide by PET imaging, would affect HER3P1 binding, the NOTA-conjugated HER3 peptide was synthesized. Synthesis occurred in the same manner as the biotinylated peptide, with NHS-ester NOTA (Macrocyclics, Dallas, TX) being substituted for biotin. The peptide was radiolabeled with Ga-68 eluted from a Ge-68/Ga-68 generator (Radiomedix, Houston, TX) in 0.05 M HCl. The pH of the eluate was adjusted to approximately 4 by 2 M-[4-(2-hydroxyethyl)piperazin-1-yl]ethanesulfonic acid buffer pH 8 and 100 ng of peptide reacted with the elution for 10 min at room temperature. Peptide was purified from free Ga-68 by reverse-phase C18 cartridge (Waters, Milford, MA) and purity determined by ITLC. Ten microcuries of purified [⁶⁸Ga]HER3P1 was then added to 1 × 10⁵ MDA-MB-453 cells and incubated for 1 h. After incubation, cells were washed 3× with TBS and bound radioactivity quantified by a Wallac gamma counter (Perkin Elmer, Waltham, MA). As a control, excess (100 μM) peptide was added to cells and bound peptide measured in the same manner.

Positron Emission Tomography Imaging and Ex Vivo Correlation

The [⁶⁸Ga]HER3P1 was labeled and purified in the same manner as described for in vitro analysis, and the final preparation was diluted in normal saline to approximately 300 MBq for injection into mice. Nu/nu mice bearing either 22RV1 (high HER3-expressing prostate cancer) or HCC-1954 (low HER3-expressing breast cancer) tumors were implanted in the right upper flank of mice and grown to 5–7 mm. Radiolabeled HER3 peptide was injected intravenously and allowed to circulate for 1 h prior to PET imaging. PET images were acquired on a Triumph microPET/CT (Trifoil, Chatsworth, CA) for 15 min in list mode, followed by CT acquisition. Images were constructed using 3D-MLEM (20 subsets) and corrected for scatter and randoms. The tumor and blood uptake was calculated in a 3D region of interest drawn around the tumor and heart, respectively, using CT guidance. Images were post-processed using VivoQuant (InviCRO, Boston, MA).

Following PET acquisition, tumors and relevant organs were removed from mice and weighed, and the total activity for each was quantified by a Wallac gamma counter (Perkin Elmer, Waltham, MA). After radioactive decay, tumors were lysed and analyzed by Western blot for correlation of tumor uptake to HER3 expression normalized to β-actin. HER3 (sc-81455, Santa Cruz Biotech, Dallas, TX) and β-actin (13E5, Cell Signaling, Danvers, MA) antibodies were used to detect protein followed by an HRP-conjugated goat-antirabbit secondary antibody (Abcam) and detection by SignalFire chemiluminescent substrate (Cell Signaling).

Statistical Analysis

For comparison of phage clones against BSA and peptide specificity against other receptor tyrosine kinases, a one-way ANOVA with a Dunnett test to correct for multiple comparisons was performed. A sigmoidal dose-response curve was fit to the peptide affinity data using GraphPad Prism 6.0. For peptide cell binding, a two-way ANOVA with Sidak’s multiple comparisons test was performed. A linear regression was fit to HER3 TBR versus Western blot data, with correlation significance calculated by a Pearson correlation. An unpaired t test was used to compare HER3 expression between 22RV1 and HCC-1954 tumors.

Results

Phage Display Selection

Following three rounds of selection, 28 individual phage clones were selected for initial ELISA screening. Phage binding was measured as absorbance and normalized to control phage. Of the resulting 28 phages screened, 23 had greater binding to HER3 than the control phage, and the five phages with the largest HER3 versus BSA differentials were chosen for further analysis (Fig. 1a). Each of these five phages had significantly higher background subtracted binding to HER3 as quantified by absorbance (range = 0.177–0.253) than the control phage, which had an absorbance of 0.007 ± 0.002 (P < 0.001) (Fig. 1b). Given the significant specificity of each phage, all five were sequenced to ascertain the amino acid composition of their displayed peptide. Surprisingly, each of the sequenced phage displayed an identical peptide. To further confirm that the convergence of the selection was not resultant from a target-unrelated peptide, three phages which did not demonstrate HER3-specific binding were also sequenced, and each had a different peptide sequence than the convergent sequence (Fig. 1c).

In Vitro Peptide Analysis

Since each of the top five phage clones analyzed displayed an identical sequence, this sequence was chosen for in vitro analysis to confirm peptide binding outside of the phage scaffold. For initial characterization, the peptide (HER3P1) was conjugated to biotin through a triglycine spacer. The biotinylated-HER3P1 peptide was first analyzed with purified HER3 ECD to obtain an approximate binding affinity. HER3P1 bound in a sigmoidal manner, with an affinity of 270 ± 151 nM (Fig. 2a). In order to ensure that the peptide was specific for HER3 and not EGFR or HER2, similar receptor tyrosine kinases also expressed on the cellular surface of many tumors; the binding of HER3P1 was tested against each receptor. The absorbance of HER3P1 binding to HER3 was 0.82 ± 0.03, whereas binding to EGFR was 0.21 ± 0.09, HER2 was 0.22 ± 0.07, and BSA was 0.15 ± 0.01 (P < 0.001 for all), indicating that the peptide was highly specific for HER3 and suitable for cell binding analysis (Fig. 2b).

Fig. 2.

Biotin-HER3P1 characterization against purified target. a Peptide ELISA with increasing concentrations of HER3P1 demonstrates saturable binding and an affinity of 270 ± 0.151 nM. Triangles represent the mean of six replicates ± SEM. b Measured HER3–1 absorbance against two similar receptor tyrosine kinases and BSA, demonstrating approximately tenfold selectivity over closely related proteins. Bars represent the mean of six replicates ± SEM. **P < 0.01.

Although the peptide bound HER3 in isolation, it was critical to assess its ability to interact with the receptor in a cellular context. To assay cellular binding, biotinylated-HER3P1 or a non-specific peptide was incubated with either MDA-MB-453 human breast cancer cells, which express moderate levels of HER3, or HCC-1954 cells which have lower levels of HER3 expression. Cell binding for HER3P1 was quantified and absorbance for MDA-MB-453 was 0.29 ± 0.06, whereas it was significantly lower for HCC-1954 cells 0.13 ± 0.06 (P < 0.05). Additionally, HER3P1 binding was significantly higher to MDA-MB-453 cells than the control peptide (0.12 ± 0.06, P < 0.05), whereas there was no difference between the control peptide and the HER3P1 for binding to HCC-1954 cells (Fig. 3a). The strong selectivity of the peptide was also apparent using fluorescent microscopy, with high binding to MDA-MB-453 cells and almost no visualization of HCC-1954 cells (Fig. 3b). Having demonstrated the HER3-targeting capabilities of the peptide sequence, confirmation of peptide binding after substitution of NOTA and radiolabeling was sought by competitive cell binding. The [⁶⁸Ga]HER3P1 bound to MDA-MB-453 cells, binding was significantly (P < 0.05) blocked by the addition of excess unlabeled NOTA-HER3P1, indicating specific peptide binding (Fig. 3c).

Fig. 3.

Biotin-HER3–1 cell binding. a Quantitative measurement of HER3–1 binding to moderate (MDA-MB-453) and low (HCC-1954) HER3-expressing cells reveals significant differentiation by HER3–1 between cell types, in addition to significantly higher binding to a moderate HER3 cell line than a control peptide. b Fluorescent microscopy visualizes peptide binding to MDA-MB-453 cells, with little binding to HCC-1954 cells. b Competitive cell binding analysis of [⁶⁸Ga]HER3P1 demonstrates specific binding to MDA-MB-453 cells; bars represent the mean of four replicates with error bars representing SEM. *P < 0.05.

Murine PET Imaging and Ex Vivo Analysis

After characterizing the ability of the HER3P1 to accurately distinguish between HER3 expression in vitro, the peptide was synthesized and conjugated to NOTA for in vivo analysis. NOTA-HER3P1 peptide was radiolabeled with Ga-68 and purified to greater than 95 % purity by reverse phase chromatography, as determined by ITLC. The specific activity of the labeled peptide was 296 ± 25.9 MBq/mg. A human prostate castration-resistant cancer cell line 22RV1, was chosen because of its high levels of HER3 expression, and HCC-1954 cells remained as the negative control for PET imaging studies. PET imaging demonstrated high [⁶⁸Ga]NOTA-HER3P1 tumor uptake in 22RV1 tumor bearing mice, with significantly lower levels in HCC-1954 tumor-bearing mice (Fig. 4a). Off-target peptide accumulation was minimal, with uptake in the kidneys and bladder consistent with the normal route of peptide clearance. Standardized tumor-to-blood ratios (TBR) for high HER3 22RV1 ranged from 1.60 to 3.32 (n = 4), whereas low HER3 HCC-1954 tumors ranged from 0.69 to 0.94 (n = 4).

Fig. 4.

[⁶⁸Ga]NOTA-HER3–1 PET imaging and tumor HER3 expression correlation. a Sagittal and axial maximum intensity projections of the radiolabeled peptide in high HER3-expressing 22RV1 and low HER3-expressing HCC-1954 tumors. Peptide uptake is visualized in the tumor and kidneys, which are the main route of clearance. b Western blot analysis of HER3 and β-actin for tumors excised from mice following PET imaging. c Quantification of HER3:β-actin ratio for 22RV1 and HCC-1954 tumors; bars represent the mean of four replicates ± SEM. d Linear regression with 95 % confidence intervals for each tumor comparing PET TBR to ex vivo HER3:β-actin ratio. Linear significance was determined by Pearson’s correlation. **P < 0.01; ***P < 0.0001.

Following PET acquisition, tumors were excised and HER3 and β-actin were quantified by Western blot (Fig. 4b). The HER3:β-actin ratio for each tumor was plotted against its corresponding TBR and fit to a linear regression with a goodness of fit R² = 0.96 and a Pearson correlation P value of less than 0.0001. This data confirmed that the HER3P1 uptake was highly correlated with HER3 expression and accurately quantified total HER3 in vivo (Fig. 4c). Additionally, biodistribution analysis of HER3P1 was performed. Sites of accumulation included the HER3+ tumors (0.50 ± 0.18 % ID/g) and HER3+ organs such as the stomach (0.30 ± 0.07 % ID/g), intestines (0.48 ± 0.15 % ID/g), and lungs (0.70 ± 0.14 % ID/g), in addition to kidneys (10.1 ± 1.67 % ID/g) as a route of clearance (Table 1). Accumulation in off-target organs was not due to blood accumulation, as tumor-to-organ ratios were significantly lower than tumor-to-blood ratios (P < 0.05 stomach, P < 0.001 intestines and lungs, Fig. 5a). Comparison between PET TBR and biodistribution TBR was highly correlated, as determined using Pearson correlation (P < 0.001, Fig. 5b).

Table 1.

Biodistribution in all relevant organs depicted as the mean injected dose per gram of tissue (%ID/g) of four 22RV1 tumor-bearing mice ± SD

Organ	% ID/g tissue ± SD
Tumor	0.5 ± 0.18
Blood	0.18 ± 0.04
Heart	0.31 ± 0.07
Lung	0.70 ± 0.14
Stomach	0.30 ± 0.07
Intestines	0.48 ± 0.15
Liver	0.70 ± 0.23
Kidney	10.1 ± 1.67
Spleen	0.38 ± 0.09
Muscle	0.28 ± 0.14
Bone	0.67 ± 0.38

Fig. 5.

Biodistribution and correlation of [⁶⁸Ga]NOTA-HER3–1. a Tumor-to-organ ratios calculated by biodistribution values from results in a demonstrating specific accumulation in murine sites of non-tumor tissue HER3 expression. b Linear regression of TBR from both biodistribution versus PET imaging. Significance of linearity determined by Pearson’s correlation. *P < 0.05; ***P < 0.001.

Discussion

The role of HER3 in targeted therapy resistance has been well documented, and efforts to pharmacologically inhibit its activity are currently being explored. Clinical trials examining the efficacy of anti-HER3 antibodies including patritumab (NCT02134015), MM-121 (NCT00734305), U3–1402 (NCT02980341), and GSK2849330 (NCT01966445) are ongoing or have been completed recently, with limited success to date [25]. One limiting factor for the administration of such agents is identifying patient populations most likely to benefit from the therapy. HER3 is particularly challenging to quantify by biopsy, because of its highly heterogeneous temporal and spatial expression. As drug approval continues to trend toward including a companion diagnostic with novel targeted therapies, an accurate method to quantify expression of HER3 to guide therapy is of paramount importance. PET imaging, which provides a global and repeatable methodology to assess target expression, is highly compatible with HER3 expression diagnosis. In order to facilitate such an assay, a novel HER3 peptide was selected by phage display and characterized for quantitative PET imaging of HER3-expressing tumors in murine models of multiple cancers.

Phage display, which is routinely used to screen for novel peptide ligands of various targets, was undertaken to identify a suitable peptide for HER3 imaging. Traditionally, phage display involves multiple rounds of selection, followed by screening of individual ligands to determine the highest affinity and specificity peptide among a host of candidates [26]. This process of selection and screening was undertaken for HER3, but all suitable peptides had identical sequences. Although this is not unique, it is relatively rare that a selection would converge on a single peptide sequence. Given the prevalence of phage display selections that are rendered unusable by contamination with target-unrelated peptides, sequences of non-binding phages were also obtained in order to determine if a single phage with a potential selective growth advantage dominated the selection [27]. However, the non-HER3 binding phages sequenced all had sequences different from both the HER3-avid phage and each other, limiting the possibility of a growth-advantaged phage dominating the selection. Further in vitro characterization of the displayed peptide conjugated to biotin demonstrated high affinity binding of 270 nM to HER3, with greater than tenfold specificity compared to that of other proteins, including the other similar receptor tyrosine kinases EGFR and HER2. Combined with specific binding of the peptide to HER3-expressing cancer cells, the peptide warranted further in vivo analysis.

Two tumor cell lines, 22RV1, a castration-resistant prostate cancer cell line with high levels of HER3, and HCC-1954, a breast cancer cell line with low levels of HER3, were chosen as models for binding analysis. PET imaging revealed high tumor-to-blood ratios in the HER3-positive 22RV1 cell line, with background levels in HER3-negative HCC-1954 tumors. Furthermore, ex vivo analysis of the tumors used for PET imaging provided a highly significant correlation between protein expression and PET TBR, indicating an accurate and robust method of HER3 quantification suitable for exploration in both pre-clinical and clinical trials.

To date, a small number of HER3 targeted PET imaging agents have been explored, with two reaching clinical trial. One clinical study has been performed by Lockhart et al. utilizing ⁶⁴Cu-labeled DOTA-conjugated patritumab to determine dosimetry and receptor occupancy; however, no correlation between tumor uptake and ex vivo immunohistochemical analysis could be determined, and specific accumulation measured by tumor-to-blood ratio was approximately 1 [28]. One trial (NCT02345174) has utilized PET imaging of an ⁸⁹Zr-labeled therapeutic antibody, GSK2849330, prior to administration of the same unlabeled antibody, but no results have been posted yet. Thus, there remains a need for an accurate PET imaging agent to quantify HER3 expression, which may help expedite the clinical approval of current or novel HER3 therapies.

Pre-clinically, a number of antibodies, antibody fragments, and affibodies have been used to image HER3 with a variety of PET isotopes including F-18, Ga-68, Cu-64, and Zr-89. Although antibodies provide accurate quantification, slow clearance prevents serial imaging and may limit clinical application [29]. The HER3 affibody ZHER3:8698 is cleared rapidly and may represent a suitable HER3 clinical imaging agent, but high renal uptake limits repeat imaging [30]. One small molecule pan-RTK inhibitor has been radiolabeled for imaging; however, the cross-reactivity with other RTKs greatly reduces its specific HER3 utility [31]. Given the current status of HER3 imaging agents, a peptide based imaging agent with high specificity and minimal background uptake represents a strong candidate for clinical translation.

Radiolabeled peptides, and specifically Ga-68 labeled peptides, have a strong presence in clinical targeted imaging. [⁶⁸Ga]DOTATOC and DOTATATE are both in use clinically, with the latter gaining FDA approval in 2016. The rapid pharmacokinetics and high tumor-to-background ratios of peptides have facilitated highly accurate quantitation of target expression, permitting not only detection but also characterization for therapeutic optimization of drugs such as Lutathera and Octreotide. The fundamental imaging paradigm surrounding these peptides provides a framework for future clinical translation of peptides.

Conclusion

HER3P1 represents a highly accurate peptide-imaging agent for HER3 with low off-target accumulation favorable for PET quantification. HER3 represents a critical protein in targeted therapies for a number of cancers, and its highly dynamic expression requires an equally dynamic approach to quantification. HER3 imaging with HER3P1 may represent an ideal method to quantify HER3 and provide critical cellular feedback analysis. As such, HER3P1 represents a novel HER3 imaging peptide with considerable promise as a clinical imaging agent.