Synthesis, ¹⁸F-labelling and radiopharmacological characterisation of the C-terminal 30mer of Clostridium perfringens enterotoxin as a potential claudin-targeting peptide

Abstract

The cell surface receptor claudin-4 (Cld-4) is upregulated in various tumours and represents an important emerging target for both diagnosis and treatment of solid tumours of epithelial origin. The C-terminal fragment of the Clostridium perfringens enterotoxin cCPE_290–319 appears as a suitable ligand for targeting Cld-4. The synthesis of this 30mer peptide was attempted via several approaches, which has revealed sequential SPPS using three pseudoproline dipeptide building blocks to be the most efficient one. Labelling with fluorine-18 was achieved on solid phase using N-succinimidyl 4-[¹⁸F]fluorobenzoate ([¹⁸F]SFB) and 4-[¹⁸F]fluorobenzoyl chloride as ¹⁸F-acylating agents, which was the most advantageous when [¹⁸F]SFB was reacted with the resin-bound 30mer containing an N-terminal 6-aminohexanoic spacer. Binding to Cld-4 was demonstrated via surface plasmon resonance using a protein construct containing both extracellular loops of Cld-4. In addition, cell binding experiments were performed for ¹⁸F-labelled cCPE_290–319 with the Cld-4 expressing tumour cell lines HT-29 and A431 that were complemented by fluorescence microscopy studies using the corresponding fluorescein isothiocyanate-conjugated peptide. The 30mer peptide proved to be sufficiently stable in blood plasma. Studying the in vivo behaviour of ¹⁸F-labelled cCPE_290–319 in healthy mice and rats by dynamic PET imaging and radiometabolite analyses has revealed that the peptide is subject to substantial liver uptake and rapid metabolic degradation in vivo, which limits its suitability as imaging probe for tumour-associated Cld-4.

Introduction

Claudins are a family of single-chain membrane proteins containing four membrane-spanning helices, which comprise at least 27 distinct members. Together with occludins, junctional adhesion molecules and the zonula occludens proteins, claudins constitute the proteinaceous component of specialised differentiations of epithelial and endothelial cell membranes called tight junctions, which are sites in which close contacts between adjacent cells are formed based on interactions of these membrane proteins (Takahashi et al. 2012; Salvador et al. 2016). Claudins are functioning by engaging in either homophilic interactions with claudins of identical subtypes at the surface of adjacent cells or heterophilic interactions with claudins of different subtypes (Piontek et al. 2017a). Most tumours originate from epithelial and endothelial cells, which implicate the involvement of tight junctions in oncogenic progression. With regards to the role of claudins in cancer, it has been found that their expression can increase or decrease depending on the kind and stage of the tumour (Singh and Dhawan 2015; Osanai et al. 2017). However, Cld-4, and to a lesser extent, Cld-3, have been shown to be upregulated in various malignancies such as gastrointestinal, breast, pancreatic, prostate, uterine and ovarian cancer (Ding et al. 2013; Kwon 2013). Their aberrant expression seems to impair the structure of tight junctions, which in turn supports tumour progression by enhancing cell migration, invasion and metastasis (Boireau et al. 2007). This renders Cld-4 an interesting target for the diagnosis and differentiation of tumours and potentially also as a target for therapeutics (Kominsky et al. 2007; Neesse et al. 2012).

A vector for targeting Cld-4 has been identified in the C-terminal domain of the enterotoxin of the pathogenic bacterium Clostridium perfringens. This 35-kDa single-chain proteotoxin exerts its cytotoxic action via the pore-forming N-terminal domain, while its C-terminal domain is responsible for targeting the surface of intestinal epithelial cells via interaction with Cld-4 and -3 (Mitchell and Koval 2010). The C-terminal domain of CPE comprises the amino acids 171–319 (Hanna et al. 1995). Its X-ray crystal structure has been solved, which has revealed that it exhibits a β-sandwich-type fold consisting of nine largely antiparallel β sheets (Van Itallie et al. 2008) (Fig. 1). Visualisation of tumour-associated Cld-4-has been shown for fluorophore (Neesse et al. 2013), indium-111 (Mosley et al. 2015) and Xe-labelled cCPE (Piontek et al. 2017b) using optical, SPECT and MR imaging, respectively. Due to the large molar mass of 14 kDa, drawbacks such as slow pharmacokinetics and potential immunogenicity (Suzuki et al. 2011) are associated to the use of cCPE as imaging agent. The site within cCPE that is responsible for the interaction with Cld-4 has been demonstrated to be confined to the fragment comprising the C-terminal residues 290–319 (cCPE_290–319; see Fig. 3 for sequence), both on the basis of radioligand binding experiments using ¹²⁵I-labelled cCPE fragments (Hanna et al. 1991) as well as using surface plasmon resonance (Ling et al. 2008). The 30mer peptide’s capability of tumour-targeting in vivo has been claimed for fluorescein-conjugated cCPE_290–319 in a mouse xenograft model of ovarian cancer (Cocco et al. 2010). Residues of cCPE that are critical for its interaction with Cld-4 have been identified by subjecting the 16 C-terminal amino acids of the protein to an alanine scan (Takahashi et al. 2008). This and other studies have revealed that the tyrosines 306, 310 and 312 and Leu 315 are important for receptor binding (Veshnyakova et al. 2010), which are located in a flexible loop between β sheets 8 and 9, i.e. residues Lys 304–Tyr 312, or on β sheet 9, respectively, according to the crystal structure shown in Fig. 1a. Very recently, these findings have been confirmed by solving the X-ray crystal structure of cCPE complexed with lipid bilayer-embedded Cld-4 (Shinoda et al. 2016b).

Fig. 1

a X-ray crystal structure of cCPE_194–319. The fragment covering the last 30 C-terminal amino acids are highlighted in dark blue, the other parts of the protein are coloured according to secondary structure elements (red: helices, green: β sheets, cyan: loops). Side-chain atoms that are critical for the interaction with Cld-4 are shown in stick representation. PDB entry: 2QUO (Van Itallie et al. 2008). b Topology diagram for the same protein, which shows that all β sheets except 1 and 3 are in antiparallel orientation to each other. For clarity, the short helix between β sheets 1 and 2 is not shown (colour figure online)

To the best of our knowledge, neither the chemical synthesis of cCPE_290–319 nor its labelling with radionuclides has been reported so far. Considering the overexpression of Cld-4 in various kinds of solid tumours, radiolabelled derivatives of cCPE_290–319 would be beneficial for diagnosis, and perspectively, also for therapy of such tumours. This can be expected from the successful application of radiolabelled peptides targeting peptide-activated G protein-coupled receptors in nuclear medicine. However, compared to Cld-4, the expression of such receptors is restricted to a minority of tumours such as neuroendocrine tumours, which overexpress somatostatin receptors (Fani and Maecke 2012; Ullrich et al. 2016). Herein, we envisage the preparation of cCPE_290–319 by solid-phase peptide synthesis and its labelling with fluorine-18, which represents the most advantageous radionuclide for positron emission tomography (PET) in terms of positron energy (635 keV) and β⁺/EC branching ratio (97%) (Hess et al. 2001; van der Born et al. 2017). Labelling of peptides with fluorine-18 can be straightforward achieved via ¹⁸F-fluorobenzoylation using N-succinimidyl [¹⁸F]fluorobenzoate ([¹⁸F]SFB) (Kim et al. 2014; Richter and Wuest 2014). Labelling of cCPE_290–319 by ¹⁸F-acylation reactions in solution is potentially challenging due to the presence of two internal lysine residues in the sequence. Therefore, the reaction of the side chain-protected 30mer with [¹⁸F]SFB on solid support was envisaged, as this approach has been revealed to be efficient for the site-selective labelling of lysine-containing peptides (Kuchar et al. 2012, 2018). Apart from the study presented herein, labelling of claudin-targeting peptides with fluorine-18 has been recently reported for the 17mer cCPE_303–319 C-terminally extended by three lysine residues and other non-CPE-related peptides using oxime-forming conjugation with 5-[¹⁸F]fluoro-5-deoxyribose. However, no results on their radiopharmacological properties were published (Feni et al. 2017).

Materials and methods

Solid-phase peptide synthesis

General remarks

All chemical reagents and solvents were obtained from commercial suppliers and used without further purification. Fmoc-protected amino acids and coupling reagents were purchased from MultiSynTech and Iris Biotech. Fmoc-protected pseudoproline dipeptides were obtained from Aapptec. Linker-functionalised polystyrene resins (Fmoc-Rink-Amide MBHA and Fmoc-Phe-preloaded Wang resin) were purchased from Multisyntech and Merck. HPLC systems used for analysis and purification of the synthesised peptides are documented in detail in ESM. Mass spectra were obtained on a Micromass Quattro/LC (Waters) or a Waters Xevo TQ-S mass spectrometer equipped with an electrospray ionisation source (operated in positive mode), each driven by the Mass Lynx software. ESI-MS spectra for compounds 1-3, 4a–g, 5a, b and 6 are shown in ESM.

Manual manipulation of the polystyrene-based resins was done in polypropylene syringes equipped with a 25-µm polyethylene filter (Multisyntech). The reaction mixtures were shaken on an orbital shaker (600 rpm) at room temperature, if not stated otherwise. Washing of resin was done on a filtration bottle supported by a vacuum pump.

If not described otherwise, peptides were synthesised by microwave-assisted, fully automated solid-phase peptide synthesis using Fmoc-protected l-amino acid with appropriately protected side chain functionalities as building blocks by employing a CEM Liberty Microwave Peptide Synthesizer combined with a CEM Discover microwave reactor as previously reported (Kuchar et al. 2012).

Loading of 2-ClTrtCl resin and manual coupling of the first three amino acids

A solution of Fmoc-Phe-OH (116 mg, 0.3 mmol) and DIPEA (0.8 mL, 4.5 mmol, 2.9 eq) in DMF (3.5 mL) were added to 1 g (1.55 mmol) of pre-swollen (30 min in 5 mL DMF) 2-ClTrtCl resin. After shaking the resulting mixture for 5 h, the resin was washed with DMF and CH₂Cl₂ (3 × 5 mL each). Subsequently, the resin was treated (4 × 5 min) with a mixture of CH₂Cl₂/MeOH/DIPEA (5 mL; v/v/v = 17:1:2), washed (CH₂Cl₂, EtOH, CH₂Cl₂; 3 × 5 mL each) and dried in vacuo. The loading degree was determined as previously described (Wodtke et al. 2015).

The pre-swollen resin (30 min in 5 mL DMF) loaded with the initial amino acid (0.325 mmol) was treated with 20% (v/v) piperidine/DMF (4 × 5 mL, 10 min each). Washing with DMF (4 × 5 mL each) was followed by treatment with 5% (v/v) DIPEA in DMF (4 × 5 mL, 1 min each) and repeated washing with DMF (2 × 5 mL, 1 min each). The next amino acid (1.3 mmol) was dissolved in DMF (5 mL) and added to the resin. After 10 min, HBTU (493 mg, 1.3 mmol), HOBt (176 mg, 1.3 mmol) and DIPEA (454 µL, 2.6 mmol) were added as neat compounds. After 4 h the resin was rinsed with DMF (4 × 5 mL each) and CH₂Cl₂ (4 × 5 mL each) and dried in vacuo.

Fragment condensation

The fully protected 15mer peptide cCPE_290–304 containing a free C terminus was suspended in 4 mL of solvent and added to the pre-swollen peptidyl resin (5 mL DMF, 30 min) containing N-terminally deprotected cCPE_305–319 followed by HATU and DIPEA as neat compounds. The synthesis was carried under the conditions outlined in the table below.

Solvent	N-terminal fragment (mmol)	C-terminal fragment (mmol)	HATU (mmol)	DIPEA (mmol)	Conditions
DMF	0.055	0.05	0.075	0.1	27.5 h, room temperature
NMP	0.055	0.05	0.1	0.1	30 min, 55 °C, 70 W (Heinlein et al. 2011)

N-terminal modifications

N-terminal 4-fluorobenzoylation of peptides was carried out as previously published (Kuchar et al. 2012), except that CH₂Cl₂ was used as solvent instead of DMF and the reaction time was extended to 5 h.

N-terminal attachment of the 6-aminohexanoic spacer was performed by adding a solution of Fmoc-6-aminohexanoic acid [85 mg, 0.24 mmol; synthesised according to Wojczewski et al. (2000)] in DMF (2 mL) to the pre-swollen peptidyl resin (0.06 mmol) containing N-terminally deprotected cCPE_290–319 followed by HBTU (91 mg, 0.24 mmol) and DIPEA (85 µL, 0.48 mmol) as neat compounds. After 4 h, the resin was washed with DMF (4 × 5 mL each). The Fmoc group was removed as described above. Finally, the resin was washed with DMF (3 × 5 mL each) and CH₂Cl₂ (3 × 5 mL each) and dried in vacuo.

For N-terminal conjugation with fluorescein, a solution of FITC (70.1 mg, 0.18 mmol) and triethylamine (25 µL, 0.18 mmol), was added to the pre-swollen peptidyl resin (0.06 mmol) containing Ahx-cCPE_290–319. After 5 h, the resin was washed with DMF, EtOH and CH₂Cl₂ (3 × 5 mL each) and dried in vacuo. Cleavage from the resin and purification of the FITC-conjugated peptide by semi-preparative HPLC (Method 3) were carried out as described below.

Cleavage from 2-ClTrtCl resin under conservation of side-chain protecting groups

The pre-swollen (30 min in 5 mL of CH₂Cl₂) peptidyl resin containing the cCPE_290–304 15mers was treated with HFIP/CH₂Cl₂ (5 mL; v/v = 1:4) for 15 min, which was repeated four times. The combined filtrates were concentrated to 1–2 mL in a N₂ stream, treated with ice-cold diethyl ether (20 mL) and kept on ice for 30 min. The white precipitate was filtered off, washed with ice-cold diethyl ether and dried in vacuo.

Resin cleavage under concomitant removal of the side-chain protecting groups

The pre-swollen (30 min in 5 mL of CH₂Cl₂) peptidyl resin was treated with a mixture of TFA/H₂O/triethylsilane 95:2.5:2.5 (v/v/v; 7 mL per 0.1 mmol of peptidyl resin) for 5 h. Cleavage in analytical scale (mini-cleavage) was performed with 10 mg of peptidyl resin in 1 mL of the cleavage cocktail for 1 h. The filtrate was collected and the remaining resin washed with TFA. The combined filtrates were concentrated to 1–2 mL in a N₂ stream, treated with ice-cold diethyl ether (20 mL) and kept on ice for 30 min. The precipitate was filtered off, washed with ice-cold diethyl ether and dried in vacuo.

Characterisation of distinct products

cCPE _290–319 (1)

Sequence: H-SLDAGQYVLVMKANSSYSGNYPYSILFQKF-OH

Peptide 1 served as starting material for all N-terminal modifications. It was synthesised by MW-assisted SPPS under standard conditions. Only a small amount was cleaved from the resin and purified to provide material for SPR measurements and to establish an HPLC system for radiolabeling.

Resin	Wang
Loading	0.37 mmol/g
Yield (for peptide-loaded resin, n = 5)	90 ± 7%
Yield (for isolated purified peptide)	24 mg, 29%
M (calculated; C154H226N36O46S)	3389.66 g/mol
M (found; ESI-MS+)	m/z = 1138.6 ([M+Na+2H]3+)
	m/z = 1131.2 ([M+3H]3+)
	m/z = 848.7 ([M+4H]4+)
RP-HPLC
System 1	tR = 22.3 min
System 2	tR = 22.4 min
System 3	tR = 25.5 min
System 4	tR = 24.0 min
Purity	≥ 98%

Fmoc-cCPE _305–319 (2)

Sequence: H-SYSGNYPYSILFQKF-OH

Compound 2 was solely characterised by mini-cleavage as resin-bound peptide was used for fragment condensation.

Resin	Wang
Loading	0.37 mmol/g
Yield (crude product, n = 2)	87 ± 15%
M (calculated; C88H120N18O24)	1813.9 g/mol
M (found; ESI-MS+)	m/z = 907,6 ([M+2H]2+)
RP-HPLC (system 1)	tR = 17.8 min

cCPE _290–304 (3)

Sequence: Fmoc-SLDAGQYVLVMKANS-OH

Synthesis was carried out at 2-ClTrtCl resin to ensure conservation of protecting groups after cleavage. Full characterisation was only done for the synthesis on the Ser^tBu-Asn^Trt-Ala-preloaded resin under standard conditions of solid-phase peptide synthesis. Yield refers to the amount of the crude, fully protected peptide. MS analysis was done on the deprotected peptide obtained after mini-cleavage.

Resin	2-ClTrtCl; preloaded with SertBu-AsnTrt-Ala
Loading degree	0.41 mmol/g
Yield (crude product, n = 3)	54 ± 10%
M (calculated; C84H124N18O25S)	1816.9 g/mol
M (found; ESI-MS+)	m/z = 1840.0 ([M+Na]+)
	m/z = 1818.0 ([M+H]+)
	m/z = 931.5 ([M+2Na]2+)
RP-HPLC
System 1	tR = 19.8 min

FBz-cCPE _290–319  × 2 TFA (4a)

Sequence: FBz-SLDAGQYVLVMKANSSYSGNYPYSILFQKF-OH × 2 TFA

The complete amount of resin-bound peptide was subjected to cleavage, purified and characterised after lyophilisation. Peptide 4a was used for SPR measurements, ECD spectroscopy and served as non-radioactive reference compound for [¹⁸F]FBz-cCPE_290–319. The yields stated below refer to the employed amount of peptide-loaded resin.

Yield (n = 2)	8.3–20.9 mg (9–12%)
M (calculated; C164H235FN36O47S)	3511.68 g/mol
M (found; ESI-MS+)	m/z = 1179.0 ([M+Na+2H]3+)
	m/z = 1171.8 ([M+3H]3+)
	m/z = 879.1 ([M+4H]4+)
RP-HPLC
System 1	tR = 25.8 min
System 2	tR = 26.8 min
System 3	tR = 28.2 min
System 4	tR = 31.0 min
Purity	≥ 97%

Ahx-cCPE _290–319

Sequence: H-Ahx-SLDAGQYVLVMKANSSYSGNYPYSILFQKF-OH

Free peptide was isolated in small amounts obtained by mini-cleavage for characterisation. The remaining resin was used for radiolabelling with [¹⁸F]SFB and to synthesise the corresponding non-radioactive reference compound and for conjugation with FITC.

Resin	Wang
Loading degree	0.37 mmol/g
Yield (gravimetrical for resin-bound peptide, n = 3)	93 ± 6%
M (calculated; C163H243N37O47S)	3502.75 g/mol
M (found; ESI-MS+)	m/z = 1183.4 ([M+2Na+H]3+)
	m/z = 1176.1 ([M+Na+2H]3+)
	m/z = 893.6 ([M+3Na+H]4+)
RP-HPLC
System 2	tR = 22.3 min
System 4b	tR = 26.6 min

FBz-Ahx-cCPE _290–319  × 2 TFA (5a)

Sequence: FBz-Ahx-SLDAGQYVLVMKANSSYSGNYPYSILFQKF-OH × 2 TFA

Purified 5a was used for inhibition in cell binding studies and SPR measurements.

Yield	19.8 mg, 20%
M (calculated; C170H246FN37O48S)	3624.77 g/mol
M (found; ESI-MS+)	m/z = 1228.3 ([M+Na+K+H]3+)
	m/z = 1209.8 ([M+3H]3+)
	m/z = 907.5 ([M+4H]4+)
RP-HPLC
System 2	tR = 26.2 min
System 3	tR = 28.1 min
System 4b	tR = 32.0 min
Purity	≥ 94%

FITC-Ahx-cCPE _290–319  × 2 TFA (6)

Sequence: FITC-Ahx-SLDAGQYVLVMKANSSYSGNYPYSILFQKF-OH × 2 TFA

Yield	55.7 mg, 23%
M (calculated; C184H254N38O52S2)	3891.78 g/mol
M (found; ESI-MS+)	m/z = 1298.3 ([M+3H]3+)
	m/z = 974.0 ([M+4H]4+)
RP-HPLC
System 1	tR = 25.9 min
System 2	tR = 26.8 min
System 3	tR = 28.2 min
Purity	≥ 95%

Ahx-cCPE _290–319 -amide

Sequence: H-Ahx-SLDAGQYVLVMKANSSYSGNYPYSILFQKF-NH₂

Synthesised under standard conditions of MW-assisted SPPS using pseudoproline dipeptides. The resin-bound peptide served as reactant for fluorobenzoylation. Only a small amount was cleaved from the resin and purified.

Resin	Rink-Amide-MBHA
Loading degree	0.55 mmol/g
Yield (gravimetrical for resin-bound peptide)	90%
M (calculated; C163H244N38O46S)	3501.76 g/mol
M (found; ESI-MS+)	m/z = 1169.2 ([M+3H]3+)
	m/z = 877.2 ([M+4H]4+)
RP-HPLC
System 1	tR = 23.2 min
System 2	tR = 22.1 min
System 4b	tR = 26.6 min

FBz-Ahx-cCPE _290–319 -amide × 2 TFA (5b)

Sequence: FBz-Ahx-SLDAGQYVLVMKANSSYSGNYPYSILFQKF-NH₂ × 2 TFA

Obtained by fluorobenzoylation of resin-bound Ahx-cCPE_290–319-amide.

Yield	16.6 mg, 14%
M (calculated; C163H244N38O46S)	3623.78 g/mol
M (found; ESI-MS+)	m/z = 1209.6 ([M+3H]3+)
	m/z = 917.1 ([M+2Na+2H]4+)
	m/z = 907.8 ([M+4H]4+)
RP-HPLC
System 1	tR = 25.6 min
System 2	tR = 26.2 min
System 3	tR = 28.1 min
System 4b	tR = 30.5 min
Purity	≥ 73%

FBz-cCPE _290–319 -Tyr306Ala × 2 TFA (4b)

Sequence: FBz-SLDAGQYVLVMKANSS ASGNYPYSILFQKF-OH × 2 TFA

This peptide was synthesised by N-terminal fluorobenzoylation of the corresponding resin-bound peptide as described for the wild-type peptide 4a. The yield refers to the used amount of resin-bound peptide.

Yield	3.3 mg, 3%
M (calculated; C160H237FN36O46S)	3419.65 g/mol
M (found; ESI-MS+)	m/z = 1155.9 ([M+2Na+H]3+)
	m/z = 1148.6 ([M+Na+2H]3+)
RP-HPLC
System 1	tR = 25.1 min
System 2	tR = 25.5 min
System 3	tR = 28.6 min
purity:	≥ 81%

FBz-cCPE _290–319 -Tyr306Phe × 2 TFA (4c)

Sequence: FBz-SLDAGQYVLVMKANSSFSGNYPYSILFQKF-OH × 2 TFA

This peptide was synthesised by N-terminal fluorobenzoylation of the corresponding resin-bound peptide as described for the wild-type peptide 4a. The yield refers to the used amount of resin-bound peptide.

Yield	29.1 mg, 22%
M (calculated; C161H229FN36O47S)	3495.69 g/mol
M (found; ESI-MS+)	m/z = 1174.6 ([M+Na+2H]3+)
	m/z = 1167.1 ([M+3H]3+)
	m/z = 875.7 ([M+4H]4+)
RP-HPLC
System 1	tR = 28.5 min
System 2	tR = 27.6 min
System 3	tR = 28.7 min
Purity	≥ 98%

FBz-cCPE _290–319 -Tyr306Trp × 2 TFA (4d)

Sequence: FBz-SLDAGQYVLVMKANSSWSGNYPYSILFQKF-OH × 2 TFA

This peptide was synthesised by N-terminal fluorobenzoylation of the corresponding resin-bound peptide as described for the wild-type peptide 4a. The yield refers to the used amount of resin-bound peptide.

Yield	11.4 mg, 9%
M (calculated; C166H236FN37O46S)	3534.70 g/mol
M (found; ESI-MS+)	m/z = 1187.0 ([M+Na+2H]3+)
	m/z = 894.0 ([M+2Na+2H]4+)
RP-HPLC
System 1	tR = 25.3 min
System 3	tR = 28.9 min
purity	≥ 99%

FBz-cCPE _290–319 -Tyr306pFPhe × 2 TFA (4e)

Sequence: FBz-SLDAGQYVLVMKANSS(pFF)SGNYPYSILFQKF-OH × 2 TFA

This peptide was synthesised by N-terminal fluorobenzoylation of the corresponding resin-bound peptide as described for the wild-type peptide 4a. The yield refers to the used amount of resin-bound peptide.

Yield	34.7 mg, 24%
M (calculated; C164H234F2N36O46S)	3513.68 g/mol
M (found; ESI-MS+)	m/z = 1194.7 ([M + 3Na]3+)
	m/z = 1187.1 ([M+2Na+H]3+)
	m/z = 1180.0 ([M+Na+2H]3+)
	m/z = 1173.0 ([M+3H]3+)
	m/z = 889.2 ([M+2Na+2H]4+)
RP-HPLC
System 2	tR = 28.0 min
System 3	tR = 29.2 min
Purity	99%

FBz-cCPE _290–319 -Leu315Ala × 2 TFA (4f)

Sequence: FBz-SLDAGQYVLVMKANSSYSGNYPYSIAFQKF-OH × 2 TFA

This peptide was synthesised by N-terminal fluorobenzoylation of the corresponding resin-bound peptide as described for the wild-type peptide 4a. The yield refers to the used amount of resin-bound peptide.

Yield	16.2 mg, 15%
M (calculated; C161H229FN36O47S):	3469.63 g/mol
M (found; ESI-MS+)	m/z = 1180.2 ([M + 3Na]3+)
	m/z = 1172.5 ([M+2Na+H]3+)
	m/z = 1165.3 ([M+Na+2H]3+)
	m/z = 879.5 ([M+2Na+2H]4+)
RP-HPLC
System 1	tR = 23.5 min
System 2	tR = 23.6 min
System 3	tR = 24.3 min
Purity	98%

FBz-cCPE _290–319 -Tyr306Ala-Leu315Ala × 2 TFA (4g)

Sequence: FBz-SLDAGQYVLVMKANSSASGNYPYSIAFQKF-OH × 2 TFA

This peptide was synthesised by N-terminal fluorobenzoylation of the corresponding resin-bound peptide as described for the wild-type peptide 4a. The yield refers to the used amount of resin-bound peptide.

Yield	15.8 mg, 12%
M (calculated; C155H225FN36O46S)	3377.61 g/mol
M (found; ESI-MS+)	m/z = 1141.9 ([M+2Na+H]3+)
	m/z = 1134.5 ([M+Na+2H]3+)
	m/z = 856.9 ([M+4H]4+)
RP-HPLC
System 1	tR = 23.1 min
System 2	tR = 23.6 min
System 3	tR = 26.5 min
Purity	≥ 99%

ECD spectroscopy

Stock solutions of peptide 4a were prepared at a concentration of 0.5 mg/mL in a mixture of 10 mM sodium phosphate pH 7.4 and CH₃CN (1:1, v/v). The high absorbance at this concentration did not allow measurements below 200 nm. To obtain ECD spectra of 4a at a concentration of 0.125 mg/mL, the stock solution was diluted with the buffer/CH₃CN mixture. The high absorbance for this solvent mixture did not allow measurements below 200 nm for both concentrations. To measure spectra of peptide 4a in the presence of 50% TFE, a stock solution was prepared in buffer/CH₃CN/TFE 1:1:2 (v/v/v) at a concentration of 0.5 mg/mL. 1% of TFA was added and the mixture was heated (50 °C) for complete dissolution. Dilution to 0.125 mg/mL was done with the identical solvent mixture. In this case, the ECD scanning was possible below 190 nm. The ECD spectra were measured in a quartz cuvette (Starna) with an optical path length of 1 mm using a J-810 spectropolarimeter (Jasco, Japan). The conditions of the measurements were as follows: a spectral region of 200 (180)–400 nm, a scanning speed of 20 nm/min, a response time of 8 s, a resolution of 1 nm, a bandwidth of 1 nm and a sensitivity of 100 mdeg. The final spectrum was obtained as an average of five accumulations. The spectra were corrected for a baseline by subtracting the spectra of the corresponding polypeptide-free solution. The ECD measurements were conducted at room temperature. Molar ellipticities were calculated according to Urbanova and Maloň (2012).

Labelling with fluorine-18

Synthesis of [¹⁸F]SFB

[¹⁸F]Fluoride was produced at a cyclotron (Cyclone 18/9, IBA, Belgium) via the ¹⁸O(p,n)¹⁸F nuclear reaction. The automated synthesis of [¹⁸F]SFB was performed according to Mäding et al. (2005) with implemented modifications as published in Kapty et al. (2011).

Radiolabelling of cCPE_290–319 with [¹⁸F]SFB

¹⁸F-labelling of the resin-bound 30mer, which contained a free N-terminal amino group but was otherwise fully protected, was carried out in orientation to Kuchar et al. (2012). 3–10 mg (0.9–3 µmol) of peptidyl resin were subjected to reaction with [¹⁸F]SFB. The major difference lies in the isolation of the cleaved radiolabelled peptide. The resin containing the ¹⁸F-labelled peptide was treated with 300 µL of TFA/H₂O/triethylsilane 95:2.5:2.5 (v/v/v) at 50 °C for 15 min. The filtrate was collected and the remaining resin washed with 100 µL of cleavage cocktail. The combined filtrates were diluted with 300 µL of CH₃CN/H₂O (v/v = 1:1) and the resulting solution subjected to purification by semi-preparative radio-HPLC. Alternatively, the combined filtrates obtained after cleavage were treated with ice-cold diethyl ether (1.5 mL) in a 2-mL test tube and centrifuged. The etheric layer was discarded and the pellet dissolved in 700 µL of CH₃CN/H₂O/TFA (v/v/v = 8:6:1) and the resulting solution subjected to purification by semi-preparative radio-HPLC.

Radiolabelling of cCPE_290–319 with 4-[¹⁸F]fluorobenzoyl chloride

To prepare 4-[¹⁸F]fluorobenzoyl chloride [in orientation to the procedures published in Lang et al. (1999), Kiesewetter and Eckelman (2001) and Seimbille et al. (2005)], the [¹⁸F]SFB radiosynthesis was interrupted at the stage of 4-[¹⁸F]fluorobenzoic acid, which was reacted with neat α,α-dichloromethyl methyl ether (1 mL) for 2 min at 90 °C. After cooling to 40 °C, the mixture was again heated to 90 °C for 2 min. The reaction of the resin-bound 30mer with 4-[¹⁸F]fluorobenzoyl chloride was performed equally to that with [¹⁸F]SFB, with the exceptions of reducing the temperature to 40 °C, applying non-aqueous solvent (300 µL of CH₂Cl₂/DMF 1:1 (v/v)) and the addition of DIPEA equimolar to resin-bound peptide. After the completed reaction, the resin was washed with CH₂Cl₂ (2 × 2 mL). Cleavage from resin was followed by the precipitation step as described above.

Radiolabelling of Ahx-cCPE_290–319 with [¹⁸F]SFB

The pre-swollen peptidyl resin was treated with a solution of [¹⁸F]SFB and DIPEA (equimolar amounts to peptide-bound resin) in 300 µL of DMF at 50 °C for 30 min under gentle magnetic stirring. After washing with CH₂Cl₂ (2 × 2 mL), the resin was treated with 300 µL of TFA/H₂O/triethylsilane 95:2.5:2.5 (v/v/v) at 30 °C for 15 min. The filtrate was collected in a 2-mL test tube, the remaining resin was washed with 100 µL of cleavage cocktail and the combined filtrates were treated with ice-cold diethyl ether (1.5 mL) and centrifuged. The etheric layer was discarded and the pellet dissolved in 700 µL of CH₃CN/H₂O/TFA (v/v/v = 8:6:1) at 60 °C. The resulting solution containing the ¹⁸F-labelled peptide was subjected to semi-preparative HPLC (Method 4_b). The product-containing eluate was collected according to the monitored γ trace of the chromatogram and diluted with H₂O (30 mL). The resulting solution was passed over a pre-conditioned LiChrolut^® RP-18 cartridge. The cartridge was washed with H₂O (3 mL) and eluted with ethanol/H₂O (v/v = 1:1; 4 × 500 µL).

Synthesis of [ ¹⁸ F]FBz-cCPE _290–319 ([ ¹⁸ F]4a) with [ ¹⁸ F]SFB

Sequence: [¹⁸F]FBz-SLDAGQYVLVMKANSSYSGNYPYSILFQKF-OH × 2 TFA

Resin	Wang
Loading degree	0.37 mmol/g
Resin-bound activity (n = 7)	7 ± 3%
Radiochemical yield (d.c.; n = 3)	1 ± 0.2%
RP-HPLC
System 2	tR = 26,8 min
System 4	tR = 37,0 min
Radiochemical purity	≥ 98%
Time of synthesis	ca. 140 min

Synthesis of [ ¹⁸ F]FBz-cCPE _290–319 ([ ¹⁸ F]4a) with [ ¹⁸ F]FBzCl

Sequence: [¹⁸F]FBz-SLDAGQYVLVMKANSSYSGNYPYSILFQKF-OH × 2 TFA

Resin	Wang
Loading degree	0.37 mmol/g
Resin-bound activity (n = 4)	24 ± 12%
Radiochemical yield (d.c.; n = 2)	2 ± 0.2%
RP-HPLC
System 2	tR = 26.8 min
System 4	tR = 37.0 min
Radiochemical purity	≥ 98%
Time of synthesis	ca. 150 min

Synthesis of [ ¹⁸ F]FBz-Ahx-cCPE _290–319 ([ ¹⁸ F]5a) with [ ¹⁸ F]SFB

Sequence: [¹⁸F]FBz-Ahx-SLDAGQYVLVMKANSSYSGNYPYSILFQKF-OH × 2 TFA

Resin	Wang
Loading degree	0.37 mmol/g
Resin-bound activity (n = 11)	65 ± 4%
Radiochemical yield (d.c.; n = 10)	7 ± 3%
Molar activity (n = 10)	1.5 ± 0.3 GBq/µmol
RP-HPLC
System 2	tR = 27.0 min
System 4b	tR = 31.0 min
Radiochemical purity	≥ 99%
Time of synthesis	ca. 130 min

Surface plasmon resonance

The SPR analyses were performed using a Biacore T100 (GE Healthcare) device. The ligand MKH₁₀AS-Cld-4_29–81-LEVLFQGP-Cld-4_139–160 (M = 10.6 kDa; purchased from Abcam, #ab124320) comprising both extracellular loops of human Cld-4 was immobilised on a CM5 sensor chip using the amine coupling kit (GE Healthcare) and PBS (GE Healthcare) following the protocol published elsewhere (Tondera et al. 2017). After the final coupling step using a solution of Cld-4 at a concentration of 20 µg/mL in 10 mM acetate buffer (pH 4) and the blocking of the surface using ethanolamine, 5500–6500 arbitrary response (resonance) units (RU) of Cld-4 were immobilised on the sensor surface. A reference cell was prepared by blank immobilisation. Kinetic data were obtained in the single-cycle mode at 25 °C using the respective cCPE_290–319 derivatives 1, 4a–g and 5a and b at concentrations of 2–10 µM. The flow buffer used was PBS containing 1% DMSO. Association and dissociation of all analytes were followed in real-time and measured at 25 °C at a flow rate of 30 µL/min. The surfaces were regenerated by a sequence of glycine (10 mM, pH 3.0, 5 s) and NaOH (0.05 mM, 5 s) followed by a stabilisation period of 300 s. The rate constants according to a two-state reaction model (see scheme below) were obtained by fitting the data including reference subtraction and blank buffer correction derived from both blank runs using the Biacore T200 Evaluation software 3.0 as described in Tondera et al. (2017):

$$R + L\underset{{k_{d1} }}{\overset{{k_{a1} }}{\rightleftharpoons}}RL\underset{{k_{d2} }}{\overset{{k_{a2} }}{\rightleftharpoons}}RL^{*} .$$

The parameters K_D, k_off and k_on were calculated from the individual rate constants by the following equations (Tummino and Copeland 2008):

$$K_{\text{D}} = \frac{{k_{d1} }}{{k_{a1} \left( {1 + \frac{{k_{a2} }}{{k_{d2} }}} \right)}},$$

(I)

$$k_{\text{off}} = \frac{{k_{d1} k_{d2} }}{{k_{a2} + k_{d1} + k_{d2} }},$$

(II)

$$k_{\text{on}} = \frac{{k_{\text{off}} }}{{K_{\text{D}} }}.$$

(III)

Measurements towards thermodynamic analysis for compound 1 were performed in multiple-cycle mode at 15, 20, 25, and 30 °C. The obtained K_D values were analysed by non-linear regression using the van’t Hoff equation for changing heat capacity (de Mol et al. 2005) as implemented in the Biacore evaluation software. The reference temperature was defined as 298 K.

Western blot analysis

Preparation of protein extracts from subconfluent cell cultures using RIPA buffer as well as western blot analysis was performed as described elsewhere (Mamat et al. 2012). In brief, 50 µg protein lysate were separated in a 10% v/v sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a PVDF membrane (Fisher Scientific) using a semi-dry transfer system (Bio-Rad). After blocking with 5% w/v skimmed milk for at least 1 h, PVDF membrane was incubated with polyclonal rabbit anti-human Cld-4 IgG (abcam, ab15104; 1:500) for 2 h at room temperature and afterwards overnight at 4 °C. Membranes were washed three times with TBS-T (TBS with 0.05% v/v Tween20) for 20 min and incubated with the appropriate horseradish peroxidase coupled secondary antibodies (goat anti-rabbit IgG, Sigma-Aldrich, A0545, 1:5000) for 1 h at room temperature. Protein bands were detected with Super Signal West Dura Chemiluminescent Substrate (Fisher Scientific) and the MF-ChemiBis 3.2 imaging system (Biostep). To verify equal protein loading, membranes were stripped (62.5 mm Tris pH 6.8, 2% SDS, and 0.7% β-mercaptoethanol) and reprobed with rabbit anti-actin IgG (Sigma-Aldrich, A5060) and horseradish peroxidase coupled goat anti-rabbit IgG (Sigma-Aldrich, A0545).

Immunocytochemistry

HT-29 and A431 cells were seeded in chamber slides (Corning) and cultivated over night. Cells were washed twice with PBS and subsequently fixed with PFA solution (4% (m/V) PFA, 2.5% (m/V) sucrose in PBS) for 30 min. Afterwards, the cells were blocked with blocking solution (10% FCS in PBS with 0.5% Tween 20) for 30 min before being incubated with the rabbit polyclonal anti-Cld-4 antibody (abcam ab15104, 1:50) for 1 h. Then, cells were washed three times with PBS-T (PBS with 0.5% Tween 20) and incubated with the secondary antibody (donkey anti-rabbit Alexa Fluor 546; Molecular Probes A10039, 1:200) for 1 h. Afterwards, they were washed three times for 15 min with PBS-T and nuclei were counterstained as described below. Cells were mounted with antifade mounting medium (Dako) and imaged with confocal microscopy (Olympus Fluoview 1200).

Microscopy and flow cytometry with FITC-Ahx-cCPE_290–319 (6)

HT-29 (12.6 × 10⁴) and A431 (6 × 10⁴) cells were seeded in 8-well chamber slides (ibidi) and cultured overnight. Then, cells were washed with PBS an incubated with 10 µM of 6 (1 mM stock in DMSO) or fluorescein (1.5 µL of 1 mM stock in water) in DMEM media (150 µL) without supplements. After 0, 1, 3 and 4 h of incubation, cells were washed three times with PBS and fixed with PFA solution (4% (m/V) PFA, 2.5% (m/V) sucrose in PBS) for 15 min. Then, they were washed again three times with PBS and nuclei were counterstained with Hoechst 33258 (10 µg/mL Sigma-Aldrich) for 5 min and analysed by confocal microscopy (Olympus Fluoview 1200). The experiment was repeated twice. For flow cytometry cells were seeded in 6-well plates (Greiner Bio-One) and the experiment was done analogously to the description above. After incubation with 6 and washing with PBS cells were detached with trypsin/EDTA (0.05%/0.02%) and resuspended in flow buffer (1% (m/V) BSA in PBS) before being analysed with flow cytometry (AttuneNxT, Thermo Fisher). Median fluorescence intensity (MFI) of three independent experiments was determined.

Stability of FITC-Ahx-cCPE_290–319 (6) in human serum

A 10 µM solution of FITC-Ahx-cCPE_290–319 (6) in human serum was prepared in Protein LoBind tubes from a 10 mM stock solution in DMSO. The solution was incubated at 37 °C and 700 rpm. Samples were taken before (T₀) and at various time points during incubation (T_x). Immediately after taking samples, an equal volume of EtOH/CH₃CN 1:1 (v/v) was added ($c_{{T_{0} }}$ = 50 µM) to inactivate serum proteases and precipitate all serum proteins. Precipitation was performed for at least 1 h at − 20 °C before centrifugation (3750g, 4 °C, 5 min). The supernatant was transferred into a 0.22-µm Corning^® Costar^® Spin-X^® centrifuge tube filter and centrifuged (3750g, 4 °C, 5 min). The discharge was precipitated again for at least 1 h at − 20 °C before further centrifugation (3750g, 4 °C, 5 min). For analysis by UPLC, the sample solutions were diluted 1:5 in the appropriate starting gradient ($c_{{T_{0} }}$ = 10 µM). Separation and analysis was performed by a UPLC–ESI-MS (Waters I-Class UPLC with PDA eλ Detector and coupled to XEVO TQ-S ESI-ToF-MS; Waters Acquity BEH C18 column (2.1 × 100 mm, 1.7 µm particle size), gradient 25–75% 1:1 CH₃CN/MeOH (0.1% AcOH)/H₂O (0.1% AcOH), 5 min). Detection was performed by absorbance measurement at a suitable wavelength (FITC, λ = 485 nm) and ESI-MS coupling. For quantitative comparison of peptide stability, absorbance measurements were analysed and the ratio of the peptide peak area at T₀ to total peak area was defined as 100% stability. All further samples were calculated as percentages to this ratio. All measurements were performed in triplicates.

Radiopharmacological characterisation of [¹⁸F]FBz-Ahx-cCPE_290–319 ([¹⁸F]5a) in vitro

Cell binding studies of [¹⁸F]FBz-Ahx-cCPE_290–319 ([¹⁸F]5a)

Human colorectal adenocarcinoma (HT-29; ATCC HTB-38) and squamous cell/epidermoid carcinoma (A431, ATCC CRL-1555) cell lines were routinely cultivated in DMEM supplemented with 10% (v/v) heat-inactivated foetal calf serum (FCS), penicillin (100 U/mL), streptomycin (100 µg/mL), glutamine (4 mM), 1% HEPES (1 M; A431 cells only) at 37 °C and 5% CO₂ in a humidified incubator. Radiotracer uptake studies were performed in monolayer cultures. Therefore, cells were seeded in 24-well plates at a density of 1.0 × 10⁵ cells/mL and grown to confluence.

Cell binding of [¹⁸F]5a was studied at temperatures of 4 °C and 37 °C and incubation times of 5, 10, 30 and 60 min in quadruplicate

15–30 MBq of [¹⁸F]5a were dissolved in ethanol/H₂O 1:1 (v/v, 2 mL) and diluted into PBS according to the number of wells. After removing the medium, 250 µL of PBS and 250 µL of the radiotracer-containing solution were added. Optionally, instead of PBS 250 µL of 10 µM FBz-Ahx-cCPE_290–319 (1% DMSO/PBS) were added 10 min prior to addition of the radiotracer. After the respective incubation time, the supernatants were discarded and the cell washed with ice-cold PBS (3 × 500 µL) and lysed with 500 µL of 1% SDS/0.1 M NaOH. Half of the lysate was used for activity measurement (decay corrected; relative to a mixture of 25 µL of the radiotracer-containing PBS solution and 225 µL of 1% SDS/0.1 M NaOH) with Cobra II gamma counter (Canberra-Packard). Activity measurements were corrected for unspecific tracer binding determined in empty (cell-free) plates using the same experimental conditions. The other half was used for determination of the total protein concentration in duplicate (2 × 25 µL) using the bicinchoninic acid assay (BCA; Pierce) and bovine serum albumin as protein standard. Uptake data for all experiments are expressed as percentage of injected dose per mg protein (%ID/mg protein).

To prove the stability during cellular incubation, the radiotracer-containing supernatant of HT-29 cells was analysed after 60 min at 37 °C by radio-HPLC.

Investigation of stability of [¹⁸F]FBz-Ahx-cCPE_290–319 ([¹⁸F]5a) in blood plasma

Plasma stability was investigated by incubating [¹⁸F]FBz-Ahx-cCPE_290–319 in blood plasma of a Wistar rat at 37 °C for 1 h. Subsequently, the sample was treated with ice-cold methanol for precipitation of plasma proteins, centrifuged, and the supernatant subjected to HPLC analysis using an (Agilent Technologies 1200 LC) using a Zorbax C-18 300SB column (250 mm × 9.4 mm) as stationary phase and a mixture of water (A) and acetonitrile (B) containing 0.1% TFA each as mobile phase. A binary gradient was run at a flow rate of 3 mL/min starting at 25% B to 60% B within 20 min.

Radiopharmacological characterisation of [¹⁸F]FBz-Ahx-cCPE_290–319 ([¹⁸F]5a) in vivo

Animal experiments were carried out according to the guidelines of the German Regulations for Animal Welfare. The protocol was approved by the local Ethical Committee for Animal Experiments (reference numbers 24D-9168.11-4/2007-2 and 24-9168.21-4/2004-1).

Investigation of in vivo stability of [¹⁸F]FBz-Ahx-cCPE_290–319 ([¹⁸F]5a)

For the investigation of metabolic stability, 20–30 MBq of [¹⁸F]5a dissolved in physiological saline (0.5 mL) were injected into the tail vein of male Wistar rats under anaesthesia with a gas mixture of 10% desflurane and 30% oxygen/air. At defined time points, blood samples were taken from right femoral artery via catheter. The exact volume and activity (decay-corrected according to time of injection) of the blood samples was determined for %ID/mL calculation. The samples were centrifuged at 3 min at 11,000g, the obtained supernatant was removed and treated with ice-cold methanol for precipitation of plasma proteins. The resulting suspension was centrifuged again (11,000g, 3 min) and subjected to HPLC analysis on an Agilent 1100 system equipped with UV/Vis DAD and a radiation detector (Canberra-Packard, Radiomatic Flo-one Beta 150TR—with PET flow cell). A Zorbax 300SB-C18 column (250 × 9.4 mm; 5 µm) was used as stationary phase, elution was done in gradient mode at a flow rate of 3 mL/min using the following programme: 0–5 min 95% water/5% acetonitrile/0.1% trifluoroacetic acid, 5–15 min gradient 100% acetonitrile/0.1% trifluoroacetic acid, 15–20 min plateau at 100% acetonitrile/0.1% trifluoroacetic acid, 20–23 min back to 0–5 min 95% water/5% acetonitrile/0.1% trifluoroacetic acid.

The ratio of area under the curve of original [¹⁸F]5a to the summed peak areas of all observed ¹⁸F-labelled species was calculated for every time point. The obtained fractions were multiplied with the activity concentration (%ID/mL) of the corresponding blood samples for the calculation of total clearance activity concentrations.

PET experiments

Pilot dynamic small animal PET experiments were then performed in both healthy male Wistar rats and NMRI Foxn1^nu/nu mice using 24 MBq and 7 MBq, respectively, of [¹⁸F]5a following the protocols published elsewhere with some minor modifications (Wolf et al. 2011; Kuchar et al. 2018). In brief, one rat and two mice were positioned under anaesthesia as described above and immobilised prone with their medial axis parallel to axis of the scanner with thorax and abdominal region (organs of interest: heart, liver, spleen, kidneys, large vessels) in the centre of field of view of a dedicated PET/CT scanner for small animals (NanoScan PET/CT scanner, Mediso). PET acquisition was started 20 s before intravenous (i.v.) infusion of the radiotracer through a needle catheter into a lateral tail vein and emission data were acquired continuously for a tracer-dependent duration of 0–60 min p.i. Acquired data were sorted into 28–32 time frames and reconstructed as described elsewhere (Pietzsch et al. 2005). Images were analysed by assigning three-dimensional regions-of-interest (ROI) over the heart region, the liver, and the kidneys using ROVER software (ABX GmbH). From these ROIs time–activity curves (TACs) representing the total (decay-corrected) fluorine-18 activity in a defined volume and expressed as radioactivity concentration, percent of maximum were obtained in each rat and mouse.

Results and discussion

Synthesis of cCPE_290–319 and derivatives

Initial attempts to prepare cCPE_290–319 (1) by sequential solid-phase peptide synthesis using Wang linker-functionalised polystyrene resin led to crude material that contained the desired 30mer peptide in insufficient amounts and that has revealed difficult to purify (Fig. 2a). Hence, the C-terminal fragment of cCPE comprising the amino acid residues 290–319 can be considered as “difficult peptide/sequence” (Tickler and Wade 2007; Paradis-Bas et al. 2016). For this reason, the 30mer sequence was deconstructed into two fragments of equal size. As the serine residue that is present at the C terminus of the resulting N-terminal fragment can be introduced as C-terminal pseudoproline dipeptide, epimerisation at this position during fragment condensation of both 15mers at solid support should be largely suppressed. This approach of peptide fragment condensation has been successfully applied in the synthesis of the C-terminally thioester-functionalised, N-glycosylated 1–39 fragment of RNase C (Heinlein et al. 2011). The C-terminal fragment 305–319 (2) of cCPE_290–319 was assembled by MW-assisted SPPS under standard conditions. The gravimetrically determined yield of the resin-bound peptide was in the range of 76–95%. For the synthesis of the N-terminal fragment Fmoc-cCPE_290–304 (3), the 2-ClTrtCl linker was chosen as anchor group for resin attachment, as this should enable the efficient release of the 15mer under conservation of all side-chain protecting groups under mildly acidic conditions (Bollhagen et al. 1994). To avoid cleavage of the resin-attached pseudoproline dipeptide consisting of Asn and Ser by diketopiperazine formation, removal of the Fmoc group was carried out under the optimised conditions identified by Heinlein et al., i.e. DBU/HOBt/DMF 2.1:1.7:69.2 (v/w/v) (Heinlein et al. 2011). The following amino acid was attached by non-automated manual coupling to enable determination of resin-loading at the tripeptide stage. All subsequent steps were carried out under standard conditions of MW-assisted automated SPPS. After cleavage from the resin with HFIP/CH₂Cl₂ 1:3 (v/v) the desired protected 15mer bearing Asn^Trt-Ser(ψ^Me,Mepro)-OH at the C terminus was isolated in a yield of only 6%. It was hypothesised that this might be due to diketopiperazine formation at later stages of peptide chain assembly by nucleophilic attack of the amidic N_α nitrogen atom at the C-terminal carbonyl carbon (Goodman and Stueben 1962). Therefore, the N-terminal 15mer was synthesised with the tBu-protected C-terminal serine residue under otherwise identical conditions, which resulted in yields ranging from 43 to 59% (n = 3) for the isolated peptide. Under the chosen conditions, Fmoc deprotection is carried out with 20% piperidine/DMF containing 0.1 M HOBt, which might lead to partial cleavage of the peptide from the resin due to repeated exposure to the slightly acidic medium. For this reason, the synthesis of the pseudoproline-free N-terminal 15mer was attempted under modified conditions suggested by Friligou et al. (2011), which, however, did not result in higher yields. Fragment condensation was carried out with HATU as coupling reagent to minimise epimerisation and DIPEA as base in DMF at room temperature. The N-terminal fragment has revealed to be incompletely soluble in that solvent. Analysis by mass spectrometry and HPLC after 22 h indicated the presence of a mixture of the desired 30mer peptide and unreacted C-terminal 15mer, which were difficult to separate from each other. Performing the coupling of the two fragments under MW irradiation [55 °C, 70 W, 30 min according to Heinlein et al. (2011)] and changing the solvent from DMF to NMP did not result in a more complete reaction progress. In contrast, the N-terminal fragment was still detectable in the crude mixture despite its improved solubility in NMP compared to DMF, which might be due to aggregate formation with the peptidyl resin under these conditions. Parameters and results for all synthetic trials are summarised in Table 1.

Fig. 2

HPLC chromatograms of crude products of cCPE_290–319 (1) obtained by sequential SPPS at Wang linker-functionalised resin. a Without pseudoproline dipeptides, b synthesis using three pseudoproline dipeptide building blocks for introducing Ser 304, 307 and 313. Chromatograms for all synthetic strategies are shown in Fig. S1 in ESM

Table 1 Summary of parameters and results for synthesis of cCPE_290–319 and its 15mer fragments

As the fragment condensation was not superior over the sequential SPPS of the 30mer, the latter approach was reconsidered. As there are three additional internal serine residues in the sequence, the introduction of serines 304, 307 and 313 using the pseudoproline dipeptides Fmoc-Asn^Trt-Ser(ψ^Me,Mepro)-OH (Ser 304) and Fmoc-Tyr^tBu-Ser(ψ^Me,Mepro)-OH (Ser 307 and 313) was envisaged. This resulted in a gravimetrically determined yield of 90 ± 7% (n = 5) for the resin-bound peptide, compared to 60% for conventional serine introduction. The composition of the crude product was reproducible and significantly more favourable than for the sequential synthesis without pseudoproline dipeptides (Fig. 2b). The use of pseudoproline dipeptides as building blocks for introducing β-hydroxyamino acids was found to be beneficial for the synthesis of several other long peptides (White et al. 2004; Harris et al. 2012; Šebestik et al. 2012; Vernieri et al. 2014; Winkler and Tian 2015) and has been mainly attributed to the disruption of β-structures because of the unique conformational properties imparted by the five-membered oxazolidine ring and masking of the amide proton, which in consequence prevents peptide aggregation and self-association (Tuchscherer and Mutter 2003; Mutter 2013). To investigate the influence of the resin anchor group, the synthesis was also performed at the 2-ClTrtCl resin, which resulted in a considerably lower gravimetric yield of resin-bound peptide of 43%. This result can be explained by partial loss of the resin-bound peptide during MW irradiation. This finding is in agreement with another study, which has revealed that peptide chlorotrityl esters are not stable during MW-assisted SPPS. This has been attributed to displacement of the peptidyl moiety by other nucleophiles that are present in the coupling mixture via a S_N1-type mechanism (Echalier et al. 2013).

The successfully established sequential SPPS of cCPE_290–319 enabled the synthesis of analogues, which are summarised in Fig. 3. Structural variations involved modifications at the N terminus such as 4-fluorobenzoylation and conjugation of FITC via a 6-aminohexanoic spacer (Jullian et al. 2009), which will serve as reference compounds for the ¹⁸F-labelled peptides or probes for fluorescence-based experiments, respectively. The 4-fluorobenzoyl group was introduced either by direct attachment to the amino group of the N-terminal Ser residue or via the 6-aminohexanoic spacer. For the directly 4-fluorobenzoylated cCPE_290–319, analogues in which amino acid residues that were suggested to be critical for the interaction with Cld-4 were exchanged by other amino acids, such as alanine or related non-proteinogenic amino acids.

Fig. 3

Structural variations for peptides derived from cCPE_290–319. pFF 4-fluorophenylalanine

Radiolabelling via ¹⁸F-fluorobenzoylation

Initially, labelling of cCPE_290–319 with fluorine-18 was attempted by reacting resin-bound, fully side-chain protected 1 with [¹⁸F]SFB (Scheme 1a) in orientation to the previously published procedure (Kuchar et al. 2012). This method has been demonstrated to be highly suitable for the site-selective labelling of lysine-including peptides and was therefore selected for radiolabelling of this 30mer peptide, which contains a lysine residue at the second-last C-terminal position (Fig. 3). Progress of the reaction was evaluated by determining the decay-corrected resin-associated radioactivity after filtration and washing in relation to initial activity of [¹⁸F]SFB as well as the radiochemical yields for the crude and purified product. In contrast to the conditions reported in Kuchar et al. (2012) that were optimised for peptides of less than 15 amino acids in length attached to Rink amide resin, two major modifications proved to be crucial for the successful preparation of 30mer [¹⁸F]4a. First, prior to acidolytic cleavage of the ¹⁸F-labelled peptide, the resin had to be carefully washed with dichloromethane. This step was necessary to remove DMF present from the labelling mixture, which acts as competing proton acceptor, because the Wang linker is considerably less acid-sensitive than the Rink anchor group. Second, the crude ¹⁸F-labelled peptide had to be precipitated with ice-cold ether and redissolved in CH₃CN/H₂O/TFA = 8:6:1 prior to purification by semi-preparative HPLC as immediate subjection of the [¹⁸F]4a-containing cleavage cocktail to HPLC impeded the separation process. Despite these methodological adjustments, the radiochemical yields for the isolated radiolabelled peptide did not exceed 1%, which was attributed to the insufficient reaction between [¹⁸F]SFB and the N terminus of protected, resin-bound 1 as obvious from the low retention of the [¹⁸F]SFB activity on the solid support.

Scheme 1

Radiolabelling of cCPE_290–319 by ¹⁸F-fluorobenzoylation directly at the N terminus (a) and via a 6-aminohexanoic spacer (b). The frame around the 30mer sequence refers to the resin-bound, fully side-chain protected peptide including Ser residues 304, 307 and 313 as acetone-derived pseudoprolines

Therefore, [¹⁸F]SFB was substituted by the more reactive 4-[¹⁸F]fluorobenzoyl chloride. This reagent was prepared by modifying the automated synthesis of [¹⁸F]SFB in the way that intermediate [¹⁸F]fluorobenzoic acid was reacted with α,α-dichloromethyl methyl ether in orientation to procedures reported in the literature (Lang et al. 1999; Kiesewetter and Eckelman 2001; Seimbille et al. 2005). In contrast to labelling with [¹⁸F]SFB, the addition of buffered aqueous solution was avoided. Even though the retention of ¹⁸F activity was approximately as twice as high compared to [¹⁸F]SFB, only 4.5 MBq of purified [¹⁸F]4a were obtained when using 1.9 GBq of 4-[¹⁸F]fluorobenzoyl chloride. As higher activities of this reagent would have been difficult to handle due to its volatility, further attempts to obtain [¹⁸F]4a in higher amounts by this approach were not undertaken.

As the reaction of both ¹⁸F-fluorobenzoylating agents with protected, resin-bound 1 resulted in insufficient radiochemical yields, an N-terminal 6-aminohexanoic spacer was introduced, which should improve the nucleophilicity of the resin-bound amino group for both steric as well as electronic reasons. [¹⁸F]SFB was reacted with the corresponding resin-bound precursor under non-aqueous conditions in the presence of DIPEA as base in equimolar amounts to the peptidic precursor, which resulted in considerably increased activity retention on solid phase compared to the radiolabelling in the absence of the spacer. When 2500 MBq of [¹⁸F]SFB were used, about 700–1000 MBq of the precipitated crude product were obtained. Purification by semi-preparative HPLC (Fig. 4) was accompanied by significant activity loss. The radiochemical yield of the isolated product was further diminished when the pooled fractions were concentrated under reduced pressure. Therefore, [¹⁸F]5a was alternatively isolated by solid-phase extraction of the HPLC fractions with a LiChrolut^®RP-18 cartridge. Elution with ethanol as biocompatible solvent provided [¹⁸F]5a in amounts ranging from 83 to 102 MBq, corresponding to decay-corrected radiochemical yields of 7–12%. Using non-radioactive 5a as calibration standard, a molar activity of 1.45±0.3 GBq/µmol (n = 10) was determined for [¹⁸F]5a.

Fig. 4

(Radio-)HPLC chromatogram for purification of [¹⁸F]FBz-Ahx-cCPE_290–319 ([¹⁸F]5a)

With the optimised approach for radiolabelling of cCPE_290–319 in hand, sufficient activities of [¹⁸F]5a were available for radiopharmacological experiments in cellulo and in vivo, which will be discussed in the sections below. The radiotracer [¹⁸F]5a was obtained in excellent radiochemical and chemical purities and sufficient molar activity in average synthesis times of ca. 130 min.

Interaction with (His)₁₀-Cld-4_29–81-LEVLFQGP-Cld-4_139–160 as Cld-4 mimicking construct

The kinetics and affinity of the interaction of cCPE_290–319 and its derivatives shown in Fig. 3 with Cld-4 were studied by surface plasmon resonance (SPR). For this purpose, a commercially available protein construct that is composed of both extracellular loops of Cld-4 which are connected by an artificial octapeptide linker (construct III in Fig. 5a) was immobilised on the sensor chip. Another SPR study investigating cCPE-Cld-4 interactions has used Cld-4 mimicking constructs that only contained the small extracellular loop of Cld-4 (constructs I and II in Fig. 5a) (Ling et al. 2008). A very recently published work has employed an immobilised complex of a detergent-stabilised phospholipid bilayer and C-terminally truncated Cld-4 containing all four transmembrane helices for that purpose (Shinoda et al. 2016b).

Fig. 5

Investigation of the interaction between cCPE_290–319 and derivatives and Cld-4 by SPR. a Schematic representation of Cld-4 and derived artificial soluble constructs described in the literature (I and II, Ling et al. 2008) and used in this work (III). b Representative sensorgram of FBz-Ahx-cCPE_290–319 (5a). Shown are the sequential measurements (single-cycle mode) for immobilised construct III and solutions of 5a in the concentrations of 2 µM, 4 µM, 6 µM and 10 µM. Sensorgrams for all compounds are included in ESM. c, d Thermodynamic analysis for the interaction of compound 1 with immobilised construct (III). van’t Hoff plot (c) and thermodynamic parameters derived thereof (d)

A representative sensorgram recorded in multiple-cycle mode is shown for peptide 4 in Fig. 5b. The obtained binding curves were analysed according to a two-step model as implemented in the machine software. The derived rate constants and dissociation constants are shown in Table 2.

Table 2 Dissociation and global rate constants for interaction of cCPE_290–319 (1) and its derivatives with the artificial Cld-4 mimicking protein construct III (see Fig. 5a)

Very recently, a dissociation constant of 3.4 nM has been determined for immobilised full-length cCPE and the detergent-solubilised full-length Cld-4-phospholipid bilayer complex (Shinoda et al. 2016a). Compared to this value, the affinities determined for cCPE (1) are lower by approximately three orders of magnitude. This difference can be mainly attributed to the dramatically increased conformational flexibility of the binding partners in the present system compared to the full-length proteins. Even though the Cld-4 interaction site of cCPE is mainly localised in the 30mer region 290–319, the remaining parts of the protein molecule restrain its spatial orientation. Furthermore, both extracellular loops of Cld-4 are highly flexible in the artificial construct, whereas in the full-length protein their attachment to the transmembrane helices results in conformational restriction that will partially preorganise the amino acid residues for interaction with cCPE. Experimental evidence for increased flexibility of the truncated binding partners was obtained from analysing the temperature dependence of the binding of 1 to the claudin-mimicking construct. The dissociation constants were determined in the range of 15–30 °C and analysed according to the van’t Hoff equation, which has revealed a value of −51 kJ/mol for ΔH and 22 kJ/mol for − TΔS (Fig. 5d). This indicates that the interaction of 30mer 1 with the Cld-4-derived construct is associated with an entropic penalty because complex formation requires the conformational restriction of both highly flexible binding partners. Opposing enthalpic and entropic contributions are commonly observed during the interaction of linear peptides with protein-binding partners (London et al. 2010). The negative-binding entropy suggests that the affinity of 1 and the derived 30mer peptides would be stronger towards membrane-embedded full-length Cld-4 compared to construct III because of the conformational restriction of the small extracellular loop imparted by the adjacent transmembrane helices in the native protein.

Obviously, removing the N-terminal positive charge 1 by acylation increases the binding affinity, as FBz-cCPE_290–319 (4a) and FBz-Ahx-cCPE_290–319 (5a) show lower K_D values. The improved affinity seems to be mainly due to increased values for the second-order rate constants k_on, which suggest reduced electrostatic repulsion for the N-terminally acylated derivatives during complex formation. This repulsive force might arise from a local positive charge at the small extracellular loop that is imparted by the residues Lys 157 and Arg 158 of Cld-4 (Piontek et al. 2017a). Apart from removing the positive charge from the N terminus, N-terminal acylation changes the overall charge state of the 30mer peptide from + 1 to 0 at pH 7.4.

As Leu 315 and Tyr 306 were identified as most crucial residues of cCPE with regards to the interaction with Cld-4, we decided to exchange these amino acids against alanine for validation of Cld-4 targeting of 4a (Harada et al. 2007; Takahashi et al. 2008; Kimura et al. 2010). Furthermore, analogues of diminished affinity to Cld-4 could potentially serve as negative control agents in imaging experiments. The K_D values determined for 4a and 4f indicate that exchange of either Tyr 306 or Leu 315 against alanine results in only slight affinity reduction. Notably, exchange of both residues against alanine (4g) leads to a drop in the K_D value by approximately one order of magnitude. Based on molecular modelling studies, Tyr 306, Tyr 310, and 312 were proposed to constitute a hydrophobic pit that can adopt Leu 151 of Cld-4, whereas Leu 315 can form contacts with Ala 153 (Protze et al. 2015). This model was recently confirmed by the co-crystal structure of the Cld-4–cCPE complex (Shinoda et al. 2016b). Substitution of each single residue of this triple-tyrosine motif to alanine in full-length cCPE was shown to attenuate binding to Cld-4 (Harada et al. 2007). In line with these results, replacement of Tyr 306 in the fluorobenzoylated 30mer by Ala resulted in slightly diminished binding affinity, even though the loss of affinity is smaller than that observed for the full-length protein. To investigate the influence of the hydroxy group at Tyr 306, this residue was substituted by other aromatic amino acids such as phenylalanine (4c), tryptophan (4d), and 4-fluorophenylalanine (4e), which resulted in only slightly reduced binding affinities. This result is in agreement with the co-crystal structure, which has revealed that the hydroxy groups of all three clustered tyrosines do not directly participate in polar interactions (Shinoda et al. 2016b). Compared to the replacement of Tyr 306, mutation of Leu 315 by Ala was more effective in attenuating the binding affinity. The similar K_D ratios of 4g/4f and 4b/4a indicate a similar change in the free-binding enthalpies upon exchange of Tyr 306 to Ala in both 30mer peptides 4f and 4a.

In vitro stability and ECD spectra

Considering the peptidic structure of [¹⁸F]5a, the radiotracer is potentially vulnerable towards degradation by proteolytic enzymes that are present in biological media or associated to cell surfaces. Therefore, the stability of [¹⁸F]5a in the presence of HT-29 cells and in blood plasma of Wistar rats was investigated by radio-HPLC. Furthermore, the stability of the FITC-conjugated analogue 6 in human serum was investigated by UPLC-DAD-MS. The results are shown in Fig. 6. [¹⁸F]5a proved to be stable in the presence of HT-29 cells as well as in blood plasma over 60 min. HT-29 is one of the Cld-4 positive cell lines that were selected for radiotracer binding studies reported below. In agreement with these findings, peptide 6 revealed to be virtually stable over 4 h in human serum. After 6 h a degradation product was detectable, with 84.9% of original peptide remaining (Fig. 6c, d). UPLC–MS data indicate that metabolite formation occurs by proteolytic cleavage after Asn 303 (see ESM).

Fig. 6

Radio-HPLC and UPLC investigation the stability of [¹⁸F]5a and 6, respectively, in biological media in vitro. a Radio-HPLC chromatograms for [¹⁸F]5a in HT-29 cell suspension, b radio-HPLC chromatograms for [¹⁸F]5a in rat blood, c UPLC chromatograms for 6 in human blood plasma, each at 37 °C, d time course of integrity of 6 in human blood plasma

The obtained results proved the radiotracer [¹⁸F]5a to be sufficiently stable for radiopharmacological investigations in cellulo and in vivo.

Considering the presence of proteases of diverse substrate specificity in blood plasma, the stability of the 30mer peptides [¹⁸F]5a and 6 in this body fluid is remarkable (Seitz et al. 2006). This might indicate a stabilised secondary structure of the cCPE-derived 30mer peptide (Tyndall et al. 2005). To support this assumption, ECD spectra of the fluorobenzoylated 30mer 4a were recorded in a mixture of water and acetonitrile in the absence and presence of TFE (Fig. 7). Interestingly, the shape of the ECD spectrum in the absence of TFE resembles that of a β-sheet structure, as it is characterised by a negative Cotton effect (single minimum) at 219 nm. Therefore, it can be concluded that 4a might adopt a β-hairpin conformation (Maynard et al. 1998; Woody 2002), which would explain its proteolytic stability (Madala et al. 2010). In the presence of 50% TFE, the appearance of the ECD spectrum is changing as two negative Cotton effects at 220 and 208 nm of reduced absolute molar elipticity can be observed, which is indicative of an α-helical conformation. Considering the linear structure of 4a, conformational changes induced by high concentration of TFE are not surprising. The induction of the β-sheet-α-helix transition upon addition of TFE has been observed for other peptides (Reiersen and Rees 2000) and even for conformationally highly constrained polypeptides of high β-sheet content (Jayaraman et al. 1996). Notably, the shape of the ECD spectra and the molar ellipticities is largely independent of the concentration both in the absence and the presence of TFE. This finding indicates that the conformational behaviour of the fluorobenzoylated 30mer 4a is not influenced by aggregation under the chosen conditions.

Fig. 7

Electronic circular dichroism (ECD) spectra of FBz-cCPE_290–319 (4a) dissolved in CH₃CN/H₂O 1:1 (v/v) at different concentrations (solid lines: 0.125 mg/mL, dashed lines: 0.5 mg/mL in the absence (black lines) and presence (red lines) of TFE (50% v/v) (colour figure online)

Cell binding of [¹⁸F]FBz-Ahx-cCPE_290–319-OH ([¹⁸F]5a) and FITC-conjugated analogue 6

To prove targeting of the radiolabelled cCPE_290–319 [¹⁸F]5a of Cld-4, radiotracer cell binding experiments have been performed with target-expressing cell lines. Expression of Cld-4 in the epithelial carcinoma cell lines A431 and HT-29 was confirmed by Western blot analysis and microscopic immune fluorescence staining of the viable cells (Fig. 8a, d).

Fig. 8

Cell binding of [¹⁸F]5a and its FITC-conjugated analogue 6. a Western blot detection of Cld-4 and actin (HT-29, A431). b Comparison of cell binding of [¹⁸F]FBz-Ahx-cCPE_290–319 ([¹⁸F]5a) for different tumour cell lines. c Quantification of cell-associated median fluorescence intensity (MFI) of 6 by flow cytometry. MFI expressed as means of three experiments with standard deviation (SD). Histograms are shown in ESM. d Fluorescence microscopic images of FITC-Ahx-cCPE_290–319 (6) and Cld 4-specific monoclonal antibody (rightmost images) to different tumour cell lines; the scale bars represent 20 µm in reality

Cell-binding studies were performed by incubating [¹⁸F]5a in an activity concentration range that corresponded to a molar concentration of 0.55 µM for the no-carrier added radiotracer. For both Cld-4 positive cell lines binding of [¹⁸F]5a steadily increased over the considered time range, while the amount of cell-bound radiotracer after 60 min was higher for the A431 cells (Fig. 8b). Preincubation for 10 min with non-radioactive 5a at a concentration of 10 µM resulted only in insignificant attenuation of radiotracer binding.

To obtain information towards Cld-4 targeting of cCPE_290–319 independent of radiotracer binding experiments, microscopic imaging with fluorescently labelled 6 was performed, which differs from 5a only in the replacement of the fluorobenzoyl group by the FITC-derived moiety. The microscopic images show that cell-associated fluorescence was observable for both cell lines and was stronger after 4 h compared to 1 h of incubation. Cell-associated fluorescence was significantly lower when fluorescein was used instead of 6 (Fig. 8c, d). Quantification of the fluorescence signal by flow cytometry has revealed that binding of 6 is more intense towards the A431 cells compared to the HT-29 cells, which is in accordance with the results obtained in the radiotracer binding studies with [¹⁸F]5a.

PET studies and in vivo stability

The favourable in vitro stability of [¹⁸F]5a encouraged its first orienting in vivo evaluation by dynamic PET imaging. Dynamic PET experiments were performed with healthy NMRI nude mice and rats. As shown in Fig. 9a, after 5 min p.i., the majority of the ¹⁸F activity in the mouse is located in the liver, while the kidneys are also visible and some activity is concentrated in the urinary bladder. At 30 min p.i., the ¹⁸F activity has mainly accumulated in the gall bladder, the intestines and the urinary bladder. The time–activity curves for the mouse (Fig. 9c) indicate that the gall bladder activity significantly increased at later time points, as reflected by the PET image for 60 min p.i (Fig. 9a).

Fig. 9

PET investigation of [¹⁸F]5a in mouse and rat. Representative coronal images of a mouse (a) and a rat (b) obtained from a small animal PET study, showing ¹⁸F-activity distribution in vivo as maximum intensity projection (0–60 min) after intravenous injection of [¹⁸F]5a. GB gall bladder, H heart, Li liver, Ki kidney, UB urinary bladder, V blood vessel (abdominal aorta and vena cava), S spleen. Yellow–red colours show highest activity concentration (Bq/mL). For anatomical orientation, the PET-CT image at 60 min p.i. is shown in a. The corresponding time–activity curves showing kinetics of the ¹⁸F-activity concentration during the entire study period of 60 min after injection of [¹⁸F]5a in mouse (c) and rat (d) as calculated from ROI analysis of dynamic small animal PET scans over the large abdominal vessel region (aorta, vena cava) representing the blood pool as well as liver and kidneys (colour figure online)

When Fig. 9a, b are compared, it becomes obvious that the PET images of [¹⁸F]5a in healthy Wistar rats largely resemble those obtained in mice, as extensive liver uptake can be discerned at 5 min p.i. Differently to mice, spleen uptake can be observed in rats at 5 min p.i., the kidney uptake is less obvious for the early phase and no gall bladder uptake can be observed because of the lack of that organ in this species. After 60 min p.i., the kidneys appear with more contrast and accumulation of activity is visible in intestine, while [¹⁸F]5a and potential metabolites derived thereof have been largely cleared from the liver. Of note, the rat urine bladder is out of the field of view in the PET images.

The PET images in Fig. 9 indicate that the elimination of [¹⁸F]5a occurs via both the hepatobiliary and renal pathways and that the clearance from the blood is very rapid in both rodent species. This statement is supported by the time–activity curve for the large abdominal blood vessels (vena cava, aorta) as region of interest (Fig. 9c, d). The blood activity decreases to approximately 20% of maximum within the first 2 min after injection. At later time points, the blood activity slightly increases in both species to reach a local maximum at 25 min p.i. in the mouse, which is probably due to re-extraction of hydrophilic metabolites into the blood. In accordance with the noted rapid blood clearance, the time–activity curve for mouse liver shows that maximum activity is reached within 2 min. The temporal complementarity between the blood and liver time–activity curves suggests that the majority of [¹⁸F]5a is directly extracted from the blood into the liver.

Such a pronounced liver uptake of a peptidic radiotracer is rather unusual, as elimination of peptides occurs predominantly by renal filtration even though the suppression of hepatic uptake is considered as a common challenge in the development of radiolabelled peptides for molecular imaging (Hosseinimehr et al. 2012). However, the branched peptide insulin, which is composed of 51 amino acids and with regards to its molar mass still comparable to 5a (3.6 vs 5.7 kDa), also undergoes rapid hepatic uptake (Kim et al. 2014), which was proven to be mediated by the insulin receptor (Sodoyez et al. 1983, 1985). Moreover, experiments with rat hepatocytes in vitro have been shown that insulin undergoes proteolytic degradation within endosomal compartments after the internalisation of the peptide–receptor complex (Juul et al. 1986). Reflecting the situation for the insulin receptor, claudin-4 has been shown to undergo clathrin-mediated endocytosis (Ivanov et al. 2004). Therefore, involvement of a receptor-mediated uptake could be assumed. It has been found that Cld-1, -2, -3, -5, -7, -8, -12, and -14 are expressed in normal human liver tissue while Cld-4 is absent in hepatocytes (Tsujiwaki et al. 2015). Besides Cld-4, full-length cCPE can bind to Cld-3 with similar affinity and lower affinity to Cld-6, -7, -8, and -14 (Fujita et al. 2000). Whether the observed liver uptake is claudin-mediated or involves other receptors such as scavenger receptors (expressed on liver- and spleen-residing macrophages) remains to be clarified within further studies.

Due to the extremely rapid elimination from the bloodstream, particularly in NMRI nude mice, the animal experiments were not extended to tumour-bearing xenograft mice.

Wistar rats were also employed to evaluate the metabolic stability of [¹⁸F]5a ex vivo. As shown in Fig. 10, the compound undergoes rapid degradation in vivo. As this finding is in contrast to the stability observed for [¹⁸F]5a in rat blood in vitro, the degradation is interpreted by liver uptake-induced metabolisation. PET images in mice indicate renal filtration as substantial route for excretion, and radio-HPLC analysis of rat urine 60 min p.i. suggests the elimination of one hydrophilic radiolabelled main metabolite (Fig. 10c). Noteworthy, no original radiotracer is detectable in the urine. Therefore, proteolytic degradation seems to be mainly taking place in the liver and is probably followed by re-transfer of hydrophilic metabolites into the blood that undergo there further processing to a final radiometabolite that is excreted into the urine. This assumption is supported by the PET-derived time–activity curves discussed above. Excretion of hydrophilic catabolites from hepatocytes into the blood or bile canaliculi is a fundamental process in drug metabolism and is mainly achieved by secondary and primary active transports mediated by carriers located in the basolateral and apical membranes, respectively (Hosseinimehr et al. 2012).

Fig. 10

Investigation of metabolic stability and arterial blood clearance of [¹⁸F]5a in healthy male Wistar rats. a Time course of metabolic transformation as derived from radio-HPLC analysis of withdrawn blood samples. b Time course of arterial blood clearance as calculated from activity measurements of blood samples and fractions of original peptide. c Radio-HPLC chromatograms of blood and urine samples taken at different time points

Conclusions

The efficient preparation of cCPE_290–310 via step-wise SPPS by incorporating three pseudoproline dipeptides has been established within this study. Its labelling with fluorine-18 was achieved via ¹⁸F-fluorobenzoylation on solid phase, which was dependent on the introduction of an N-terminal 6-aminohexanoyl linker for efficient reaction with [¹⁸F]SFB. The elaborated method for ¹⁸F-labelling on the Wang linker-functionalised polymeric support should be applicable for other peptides of longer sequence and accounted for the initial pharmacokinetic characterisation of cCPE_290–310 in vitro and in vivo. Binding experiments for [¹⁸F]5a and its FITC-conjugated analogue 6 with Cld-4-expressing tumour cell lines indicate the potential of cCPE_290–310-derived peptides for tumour targeting. However, the high liver uptake of [¹⁸F]5a and its rapid degradation in vivo limits its suitability for imaging of tumour-associated Cld-4.

Therefore, the binding affinity and pharmacokinetic properties of cCPE_290–319 should be improved for the future development of PET tracers derived from this peptide. This could be achieved by constraining the 30mer in a structure that resembles the one in the full-length protein. This in turn could be realised by designing bicyclic peptides (Rhodes and Pei 2017) based on the crystal structure of the full-length cCPE, for example.