Efficient Efficient 1818 F-Labeling of Large 37-Amino-Acid pHLIP Peptide F-Labeling of Large 37-Amino-Acid pHLIP Peptide Analogues and Their Biological Evaluation Analogues and Their Biological Evaluation

,


INTRODUCTION
Positron Emission Tomography (PET) is a non-invasive functional in vivo imaging modality, commonly used in tumor diagnosis. Tremendous effort has been made towards the development of additional efficient and widely applicable PET tracers, given the limitations of [ 18 F]-FDG, for the targeting of a large variety of cancers. However, the development of a universal tumor targeting PET tracer is limited due to tumor heterogeneity and differences within the tumor environment. 1 Therefore, targeting a physiological anomaly present in most cancers seems very promising for the development of a widely applicable diagnostic agent in oncology. The acidity of the tumor microenvironment is one of such anomalies, which plays a significant role in progression and is often associated with increased invasion and metastasis. 2 It is present in 90% of tumor microenvironments and is therefore considered a promising environmental marker for targeting. 3,4 A 37-residue peptide discovered in 1997 by Hunt et al. 5 was recently shown to display unique pH-dependent properties: the pH Low Insertion Peptide (WT-pHLIP, NH 2 -ACEQNPIYWARYADWLFTTPLLLLDLALLVDADEGTG-CO 2 H) is soluble in aqueous solution at physiological pH and shows transient interaction with cell membranes in tissues with neutral extracellular pH (pH e ). In tissues with an acidic pH e , Asp/Glu residues of pHLIP are protonated, which increases the overall hydrophobicity of the peptide and affinity to the cell membrane. As a result, pHLIP inserts and translocates itself through the phospholipid bilayer that is the cellular membrane. 6 This insertion involves a change in the peptide conformation to a α-helix, the N-terminus remaining in the extracellular space, while the C-terminus reaches the intracellular lumen.
The pHLIP peptide was recognized as a potentially useful platform for tissue acidity imaging and therefore, its potential application as a universal tool in oncology was envisaged. 7 We successfully demonstrated that fluorescent pHLIP could be used to detect tissue acidity and target tumors in mice. 8 In an extension to the study, 64 Cu-labeled pHLIP was evaluated for its suitability as a PET tumor imaging agent. The novel tracer displayed good imaging characteristics with the highest level of radioactivity accumulation in LNCaP tumor being reached 1 h after administration, but some demetallation 9 and a significant retention of activity in the GI tract were also observed.
Fluorine-18 is the most common PET nuclide due to its favorable physical properties (low positron energy, pure positron decay and 110 min half-life), availability, and dosimetry. 10,11 Therefore, the 18 F-nuclide was considered the ideal non-metal based alternative for the PETlabeling of pHLIP and its analogues. However, the 18 F-labeling of large peptides such as WT-pHLIP (molecular weight (MW) > 4000 Da) still remains challenging because nucleophilic substitution with [ 18 F]-fluoride requires harsh basic conditions and cannot be performed directly on peptides. 12 as a result, numerous approaches to label peptides with 18 F were developed, including bioconjugation with prosthetic groups like Nsuccinimidyl-4-[ 18 13 or more recent methodologies involving Al 18 F chelation 14 or the "less conventional" use of organosilicon-based fluoride acceptor. 15 Among the recent chemoselective 18 F-labeling strategies involving indirect labeling by means of [ 18 F]prosthetic groups, the popular alkyne-azide copper(I) catalyzed cycloaddition (CuAAC) "click" reaction, where the coupling of an azide and an alkyne leads to the formation of a triazole ring, appeared well suited for our purpose. It has been shown to be widely applicable and efficient for the labeling of peptides, mostly requiring a two-step synthetic approach. The success of this strategy and its wide range of application led to the development of a large variety of 18 F-labeled prosthetic groups as exemplified on Figure 1 and recently reviewed. 16,17 Thus, the one-step radiolabeling of suitable precursors followed by cycloaddition of the resulting 18 F-bearing alkyne prosthetic groups with azido functionalized peptides of interest was shown to result in total RCYs of 20 -75% (d. c.). [18][19][20] The corresponding radiosyntheses utilizing an 18 F-labeled azido prosthetic group and an alkyne functionalized peptide were shown to be equally efficient. [21][22][23] With regards to the radiochemistry however, only a few of the common 18 F-labeled prosthetic groups like 1-(azidomethyl)-4-[ 18 F]-fluorobenzene 24 or 4-[ 18 F]fluoro-N-methyl-N-(prop-2-ynyl)-benzenesulfonamide, 20 are UV-detectable. A common disadvantage of [ 18 F]-fluoroalkynes or azides is also their high volatility, which can be technically challenging. 13,18,19,21,25 Finally, the preparation of [ 18 F]-glycosylazides or azidomethyl-[ 18 F]-benzenes for example involves more than one step and is therefore more laborious. 22 Besides these inherent difficulties of a prosthetic group radiosynthesis, to the best of our knowledge, only Hausner et al. have demonstrated an efficient 18 F-radiolabeling of peptides with a high molecular weight (MW > 2000 Da), commenting on the increasing challenge to label peptides with increasing complexity and size. 13 Ramenda et al. have reported the 18 Fradiolabeling of azide-functionalized Human Serum Albumin but noted that the introduction of azide residues into HSA and subsequent radiolabeling via click chemistry significantly altered the structural and functional integrity of HSA. 20 More recently, in an eloquent approach, Gill et al. have demonstrated the utility of maleimide functionalized alkyne in labeling antibodies with [ 18 F]-fluorine via a click approach. 26 However, the presence of cysteine and lysine residues in the pHLIP amino acid sequence makes this approach unrealistic in our case, as it is necessary to limit the number of derivatization sites to one in order to preserve the unique pHLIP properties. Therefore, it appears that further efficient 18 F-labeling procedures are necessary for the development of a successful and widely applicable [ 18 F]-pHLIP derivative.
Herein we report on the use of a novel 18 F-labeled prosthetic group for the efficient two-step "click" radiolabeling of azidohexanoic acid derivatized pHLIP involving a UV-detectable prosthetic group. The ultimate aim of this study was to evaluate the novel prosthetic group for its universal utilization in [ 18 F]-peptide chemistry. Therefore, the reaction conditions were investigated with two different peptides: D-WT-pHLIP (all amino acids have D configuration) and L-K-pHLIP (all amino acids have L configuration), which has the exact same sequence as WT-pHLIP but two aspartic acid residues have been replaced with lysines. These aspartic acid residues are present in the transmembrane part after insertion and are critical for pH dependent behavior. Therefore L-K-pHLIP lacks pH-dependent behavior and was used as a negative control in our study. 6,27,28 In vitro stability of the [ 18 F]-D-WT-pHLIP in both human and murine plasma was also examined. Finally, PET imaging and biodistribution studies were performed in order to determine the in vivo properties of the novel [ 18 F]-pHLIP conjugates in LNCaP and PC-3 tumor xenografts bearing mice.

General methods
All chemicals and solvents were purchased from Sigma-Aldrich, Anichem or Fluka and were used without further purification unless specifically stated otherwise. All peptide starting materials were purchased from C S Bio Co. (Menlo Park, CA, USA). Ultra-pure water was used in this work (>18 MΩcm −1 at 20 ºC, Milli-Q, Millipore, Billerica, MA, USA). All instruments were maintained and calibrated regularly according to quality control procedures as previously reported. Thin layer chromatography performed on pre-coated silica gel 60 F245 aluminium sheets suitable for UV detection of compounds was used for monitoring the reactions. Radioactivity measurements were performed with a Capintec CRC1243 Dose Calibrator (Capintec, Ramsay, NJ, USA). Precise quantification of low radioactivity samples was achieved with a Perkin Elmer (Waltham, MA, USA) Automatic Wizard2 Gamma Counter. Nuclear magnetic resonance spectra were recorded on a Bruker AVIII 500.13 MHz spectrometer with an internal standard from solvent signals. Chemical shifts are given in parts per million (ppm) relative to tetramethylsilane (0.00 ppm). Values of the coupling constant, J, are given in hertz (Hz). The following abbreviations are used for the description of 1 H NMR and 13 C NMR spectra: singlet (s), doublet (d), doublet of doublets (dd), triplet (t), quartet (q). The chemical shifts of complex multiplets are given as the range of their occurrence. Low resolution mass spectra (LRMS) were recorded with a Waters Acquity UPLC with electrospray ionization SQ detector (ESI). High resolution mass spectra (HRMS) were recorded with a Waters LCT Premier system (ESI) and MALDI TOF analysis was performed on a Bruker Ultraflex TOF/TOF MALDI tandem TOF mass spectrometer. Quality control and purification of the final products in order to confirm the purity ≥ 95% of the radioactive intermediates and final products were achieved by analytical or semipreparative high performance liquid chromatography (HPLC). All HPLC experiments were performed on a Shimadzu HPLC system equipped with a Flow Count PIN diode radiodetector from BioScan, a DGU-20A degasser, two LC-20AB pumps, and SPD-M20A photodiode array detector and a SPD-M20A autosampler using reversed phase columns: (1)

Chemistry
Syntheses of cold reference compounds 2 to 5 are presented on Scheme 1.

2-Ethynyl-6-fluoropyridine (3)
To a solution of 2 (30 mg, 0.12 mmol) in dry tetrahydrofuran (THF, 2.5 mL) was added tetrabutylammonium fluoride (TBAF) trihydrate (120 mg, 0.38 mmol). After stirring the reaction mixture for one hour at RT, water (10 mL) was added. This mixture in water/THF (1:4) was used without further purification for the synthesis of the fluorinated reference peptides. For analytical purposes, compound 3 was purified by semipreparative HPLC using the column 1 and the gradient solvent system A. The retention time of compound 3 was 11.2 min. Due to the volatility of the product and the small scale of the synthesis, concentration of the product was achieved using a C18 light cartridge (Waters) preconditioned with 5 mL of EtOH, followed by 5 mL of water. The combined fractions were diluted with water and passed over the cartridge. Pure product 3 was eluted with CDCl 3 . 1 H NMR (500 MHz, CDCl 3 ): δ 7.76 (q, J = 7.8 Hz, 1H), 7.38 (d, J = 7.6 Hz, 1H), 6.94 (dd, J = 2.7, 2.9 Hz, 1H), 3.19 (s, 1H). 13

Radiosynthesis of [ 18 F]-3
[ 18 F]-fluoride (1480 -2220 MBq) was obtained via the 18 O(p,n) 18 F reaction of 11-MeV protons in an EBCO TR-19/9 cyclotron using enriched 18 O-water. QMA light cartridges (Waters) preconditioned with 0.5 M K 2 CO 3 (5 mL) and water (5 mL) were used for trapping of 18 F − from the aqueous solution. In order to elute 18 F − from the cartridge into a sealed 5 mL reaction vial, 1 mL of Kryptofix K 222 solution (Kryptofix K 222 , 2.5 mg; K 2 CO 3 , 0.5 mg in MeCN/water (3:1)) was slowly passed through the cartridge. The solvents were evaporated at 110 ºC under vacuum in the presence of slight inflow of argon gas. After addition of MeCN (1 mL), azeotropic drying was achieved under vacuum and with a slight argon inflow. For the complete removal of water traces, the procedure was repeated twice. A solution of the corresponding precursor 6 (Anichem), 1.5 -2 mg in 200 μL of dry DMSO, was added to the Kryptofix complex and the reaction mixture was heated at 130 ºC for 10 min before 1 mL of water and 0.4 mL of MeCN were added (total injection volume 1.6 mL). Purification by semipreparative HPLC was carried out using the reversed phase column 3 with the gradient solvent system C: the product was eluted with the same retention time as the cold standard 3: t R = 9.2 min. Under the conditions used, the bromo precursor 6 has a retention time of 10.8 min. Identification of [ 18 F]-3 was achieved by co-injection with the reference 3 on the reversed phase column 1 and the solvent system A: t R = 10.2 min ( Figure  2A).

Radiosynthesis of [ 18 F]-4
Stock solutions of copper (II) acetate (18 mg/mL) and (+)-sodium L-ascorbate (20 mg/mL), respectively, were freshly prepared. A v-shaped 4 mL HPLC vial (MT-IT TM , Prominence, SHIMADZU) was equipped with a stir bar, sealed and set under argon before 30 μL of the Cu(II) acetate solution and 60 μL of the (+)-sodium L-ascorbate solution were mixed. The prosthetic group [ 18 F]-3 (150 -500 MBq) was directly collected from the HPLC into the prepared 4 mL HPLC vial (V = 1.2 -1.6 mL). A solution of azidohexanoyl-D-WT-pHLIP (N 3 (CH 2 ) 5 CONH-ACEQNPIYWARYADWLFTTPLLLLDLALLVDADEGTG-CO 2 H, 1 mg, 0.2 μmol) in EtOH (150 μL) was added and the mixture was heated at 70 °C for 5 -10 min. After addition of MeCN (100 μL), purification of the product was achieved using semipreparative HPLC on reversed phase column 1 with solvent system A (total injection volume 1.5 -2.0 mL). Two main radioactive peaks were eluted with the same retention time as the cold reference compounds 4a and 4b that are conformational isomers. HPLC analysis indicated that the overall radiochemical purity of the compound 4 (4a + 4b) is >98%. The radiolabeled peptide could not be isolated from the peptide starting material resulting in lowered apparent specific activity. Only After evaporation of the solvent under vacuum and a slight argon inflow, which was completed within 6-8 minutes, the product was formulated in saline with 3% ethanol for injection.

Radiosynthesis of [ 18 F]-5
The radiosynthesis and purification were achieved in an analogous way to [ 18 F]-4. For the formulation of an injectable solution of [ 18 F]-5, a cartridge purification step was performed as described above. The product was eluted with 1.5 mL of acidified EtOH containing 0.02% 2N HCl and the solvent was removed under vacuum and a slight argon inflow.

Determination of the lipophilicity of [ 18 F]-4 and [ 18 F]-5 (logD pH7.4 )
As described by Wilson et al. 29 for the shake flask method, phosphate buffer (pH 7.4) saturated octanol (500 μL) and octanol saturated phosphate buffer (500 μL) were added to a microcentrifuge tube (1.5 mL, Eppendorf). Following the addition of 10 μL of the radiotracer solution (approximately 37 kBq), the samples were first vortexed and then shaken for 15 min. For phase separation, the samples were centrifuged at 2348 rcf for 5 min. 50 μL of each layer were withdrawn carefully from each phase and pipetted into microcentrifuge tubes (Eppendorf) for measurement of the distribution of the activity with a gamma-counter (Wizard 2 , Perkin Elmer). The experiment was performed in quintuplicate.

In vitro stability test
For the determination of the in vitro stability of 18

Tumor cell culture
Human prostate cancer cell lines LNCaP and PC-3 from the American Tissue Culture Collection (ATCC, Manassas, VA, USA) were cultured in a 5% CO 2 atmosphere at 37 °C. LNCaP cells were cultured in RPMI 1640 medium containing 10% fetal calf serum (FCS), 2 mM L-glutamine, 1 mM sodium pyruvate, 4.5 g/L glucose, 1.5 g/L sodium bicarbonate and 100 U/mL of penicillin and streptomycin. PC-3 cell were cultured in F-12 Kaighn's medium containing 10% FCS, 2 mM L-glutamine, 1.5 g/L sodium bicarbonate and 100 U/mL of penicillin and streptomycin. Cells were trypsinized in the absence of magnesium or calcium ions using a cocktail of 0.25% trypsin and 0.53 mM EDTA in Hank's buffered salt solution.

LNCaP and PC-3 tumor xenografts
Male athymic mice (NCRNU-M, 20-25 g) were obtained from Taconic Farms, Inc (Hudson, NY, USA) and were kept in the MSKCC vivarium for one week before any experimental handling was performed. The animals were allowed free access to water and food and all animal care and experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC). LNCaP and PC-3 tumor xenografts were induced on the right and the left shoulder of each mouse, respectively, by subcutaneous injection of the corresponding cell suspension in Matrigel (BD, Collaborative Biomedical Products, Inc., Bedford, MA, USA) and media (1:1) containing 5.0 × 10 6 -10 7 cells (viability > 90%) per injection (200 μL). All procedures were carried out under anesthesia (1.5% isoflurane). Because of the slow rate of growth of the LNCaP tumors, these tumors were implanted 14 days prior to the PC-3 tumor implantation to have similarly sized tumors for the imaging and biodistribution studies. Palpable tumors of similar size developed after 4 to 5 weeks after the initial tumor xenograft induction. An energy window of 350-700 keV and a coincidence timing window of 6 ns were used. Data were sorted into two-dimensional histograms by Fourier re-binning, and transverse images were reconstructed by filtered back-projection into a 128×128×63 (0.72×0.72×1.3 mm) matrix. The images were normalized to correct for non-uniformity of response of the PET, dead-time count losses, positron branching ratio and physical decay to the time of injection, but no attenuation, scatter or partial-volume averaging correction was applied. The measured reconstructed spatial resolution for the Focus 120 is approximately 1.6 mm in full width at half maximum at the center of the field of view. The counting rates in the reconstructed images were converted to percent of injected dose per weight (%ID/g) by use of a system calibration factor derived from the imaging of a mouse-sized water-equivalent phantom containing 18 F. Images were evaluated by region-of-interest (ROI) analysis using ASIPro VM software (Concorde Microsystems).

Combined histological/autoradiographic analysis
After the completion of the last imaging session, vascular perfusion marker Hoechst 33342 (15 mg/kg in 0.2 mL of sterile saline) was administered via the tail vein catheter. Series of contiguous fresh-frozen tumor sections of 10 μm thickness were cut and exposed to a phosphor-imaging plate (Fujifilm BAS-MS2325, Fuji Photo Film, Japan) for an appropriate length of time at −20 °C. Digital images of radioactivity distribution at 50 μm resolution were obtained. The same sections were subsequently used for immunofluorescence imaging of Hoechst 33342, and finally stained with H&E. Autoradiographic and histological images were registered using Adobe Photoshop CS3 software.  i.), respectively. The blood was collected immediately followed by quick removal of the tumors. The lung, liver, heart, kidneys, spleen, bone, muscle, stomach, small intestines (SI), large intestines (LI) and testes were taken as well. The individual tumors and further organs were weighed and the radioactivity in each sample was measured in the gamma counter. Activity concentrations were calculated as percentage injected dose per tissue weight. (%ID/g wet tissue).

Postmortem biodistribution studies
As a negative control, a formulated solution of [ 18 F]-5 [55 μCi (2 MBq), 3.5 nmol)] was administered to a further group of dual tumor bearing mice (n = 4, 19 -22 g). These mice were euthanized at 2 h p. i. and biodistribution performed as above.

RESULTS AND DISCUSSION
Recently, the CuAAC "click reaction" has attracted increasing interest for the 18 F-labeling of peptides, mainly because of the recent development of novel [ 18 F]-prosthetic groups and efficient protocols for this type of chemistry. 16,17,30 However, the feasibility of such methods for the synthesis of large [ 18 F]-peptides (MW > 2000 Da) has yet to be proven. Indeed, the preparation of an 18 F-labeled α v β 6 specific 20 amino acids peptide by Hausner et al. in 2008 is the only successful report, 13 which demonstrates the challenge of labeling large peptides efficiently without affecting their integrity.
We experienced numerous difficulties during our first attempts to synthesize 18 F-labeled WT-pHLIP via CuAAC using a previously described [ 18 F]fluoro-PEG-alkyne. Indeed, even though such a prosthetic group has been used successfully for the radiolabeling of RGD constructs, 18 no "click coupling" with azido-derivatized pHLIP analogues was observed in the standard CuAAC conditions that we used (Cu(II) acetate/(+)-sodium L-ascorbate in H 2 O/CH 3 CN 1:1). We were therefore interested in the development of a novel fluorinated prosthetic group and a corresponding synthetic procedure (Scheme 1), suitable for the radiolabeling of large peptides. We chose Fluoropyridinealkyne 3 because of the commercial availability of the corresponding bromo-precursor 6, its UV-detectability, and the fact that 2-[ 18 F]fluoropyridines were shown to display high in vivo stability. 31 Moreover, the small size of 3 was expected to have a very limited impact on the pharmacokinetic of peptides as large as pHLIP.

Radiochemistry
[ 18 F]-3 was prepared in one step by nucleophilic fluorination of the corresponding bromo precursor 6 (Scheme 2) using Kryptofix and K 2 CO 3 . Optimized conditions (DMSO, 10 min, 130 °C) led to [ 18 F]-3 in RCYs of 27.5 ± 6.6% (n = 11) decay corrected (d. c.), with conversion rates between 70 and 90%. This high labeling efficacy was not translated into higher RCYs because a loss of radioactivity was systematically observed during the transfer to the HPLC, presumably due to the volatility of [ 18 F]-3. Although a distillation-based purification might allow for higher yields, like those obtained with volatile [ 18 F]fluoroalkynes prosthetic groups, 19 it would also introduce MeCN, which in our case was not favorable for the subsequent "click" reaction. Instead, technical improvements or pH adjustments could potentially minimize the effect of the pyridine alkyne volatility issue. Our HPLC purification has as principal advantage to afford [ 18 F]-3 in high radiochemical purity (RCP > 98%, Figure 2A) in a mixture of ethanol and water, which appeared to be a perfectly suitable solvent system for the CuAAC with pHLIP. Indeed, full incorporation of [ 18 F]-3 was achieved within 10 min at 70 °C as demonstrated by the HPLC chromatograms on Figure 2B and 2C, where no unreacted material [ 18 F]-3 was detected.
The two peaks observed on Figure 2B are due to acidic HPLC conditions, which promotes the formation of two pH-dependent conformational isomers (4a and 4b). When either of the peaks were collected separately and reinjected, we observed two peaks in HPLC indicating that these are same chemical species. This fact has also been confirmed by HPLC-MS. No dimerization of peptide (through cysteine-cysteine disulfide linkage) was observed. It is also interesting to note that conformational isomers were only observed in the case of WT-pHLIP ( Figure 2B), which reflects the non acidity-dependent behavior of K-pHLIP ( Figure 2C). Figure 2 also reveals that the radiolabeled peptides [ 18 F]-4 and [ 18 F]-5 could not be separated from the azido-derivatized starting materials. Consequently, the apparent specific activities were low (0.1 -1 GBq/μmol), which should not question the use of [ 18 F]-pHLIP as an efficient PET tracer anyway. Indeed, given the mechanism of pHLIP insertion across the cell membrane, the number of insertion sites is theoretically unlimited wherever an acidic microenvironment is present, 7 as opposed to receptor based targeting systems. 32 Considering the size and the complexity of the [ 18 F]-pHLIP analogues, the detailed radiochemical data obtained with [ 18 F]-4 and [ 18 F]-5 (Table 1) show the high efficacy of our novel strategy using [ 18 F]-3. Total RCYs of up to 20% could be achieved with respect to the starting amount of [ 18 F]-fluoride activity in a short synthesis time (≤ 85 min) and in one step, whereas several of the UV-detectable prosthetic groups that were used for peptide labeling were produced after multi-step radiosyntheses. 22,23,33 The total radiochemical yields also reflects that some radioactivity was lost during the transfers and on the HPLC, which leads to reduced overall yields. In our preparative HPLC conditions we never observed any unreacted [ 18 F]-3 or other byproducts in the reaction. Although better yields were reported with small-size peptides, 16 the CuAAC for was never achieved in such high yields with large peptides. By comparison, Hausner et al. radiolabeled a 20 amino acid peptide via "click" reaction in yields below 10%. 13 Finally, our novel 18 F-labeling protocol was performed with low amounts of copper, and a minimum of 1 mg peptide starting material was necessary to achieve complete incorporation of [ 18 F]-3, which is in line with previously reported 18 F-"click" reactions on larger peptides. 13 For the characterization of the novel 18 F-labeled pHLIP analogues, a logD pH7.4 determination was carried out (Table 1). [ 18 F]-4a and [ 18 F]-4b eluted from the HPLC system with different retention times and relative lipophilicity, exhibited the same logD pH7.4 of −0.7 presumably due to isomerization and formation of the same equilibrium state at pH 7.4. [ 18 F]-5 was eluted from the HPLC system with shorter retention time, and it accordingly displayed a logD pH7.4 of −1.30 ± 0.03. It is well accepted that most of the [ 18 F]prosthetic groups (Figure 1) lead to more lipophilic 18 F-labeled peptides, especially as the "click" approach involves the formation of a lipophilic triazole ring, often considered as a surrogate for the amide bond. 16,34 Although a similar effect is expected with the use of [ 18 F]-3, the small size of the prosthetic group and the triazole ring should only have a limited impact on the pharmacological properties of future investigated peptides.
The stability of the [ 18 F]-prosthetic group in the [ 18 F]-peptide construct is important in terms of clearance and metabolism. In particular, in vivo radio-defluorination is easily detectable because of the related bone uptake of radioactivity. 32 Therefore, [ 18 F]-4a, as a prominent pHLIP analogue was first subjected to an in vitro stability test in human and murine plasma at 37 °C. The tracer displayed a good stability: after 60 min, 80% and 100% of the parent tracer remained intact in human and murine plasma, respectively ; 65% and 85% were still present after 120 min. The degradation products were more hydrophilic and displayed retention times (t R ) suggesting potential peptide fragments. [ 18 F]-Fluoride is very hydrophilic and an elution from the HPLC column within the first 3 min is expected. However, no radioactive peaks were detected during the first 10 min. The fact that peptides derived from D-amino acids were shown to display higher in vivo stability, 35,36

PET imaging studies
In order to prove the ability of [ 18 F]-4a to target acidity, we chose two prostate carcinoma tumor models, LNCaP and PC-3. Indeed, as determined by magnetic resonance, Vavere et al. reported a significantly more acidic average pH e for LNCaP tumors when compared with the PC-3 tumors. 9 Therefore, we were hopeful that this difference could be observed with PET following [ 18 F]-4a administration.
As a matter of fact, the LNCaP tumor could be clearly visualized at both time points ( Figure  3) and exhibited a radioactivity accumulation of 8.3 ± 1.5 and 8.5 ± 1.3%ID/g at 2 and 4 h p. i., respectively. The PC-3 tumor showed lower radioactivity accumulation of 4.2 ± 0.4 to 5.4 ± 0.5%ID/g at 2 and 4 h p.i, respectively, and was poorly visualized. The lower uptake is a good indication for the successful application of the pHLIP principle for PET imaging of acidic tumors. Another indication for the specificity of the pHLIP was the lack of any tumor accumulation observed with the negative control peptide [ 18 F]-K-pHLIP ( Figure 3).
The MR studies reported by Vavere et al. determined the total volume average pH e in the tumors, but the heterogeneity of the tumor microenvironment suggests the formation of acidic and non-acidic areas in tumors. Accordingly, the PET images acquired after [ 18 F]-4a administration displayed a heterogeneous distribution of the tracer in the tumor since hot spots within the LNCaP tumor could be visualized, indicating the specific enrichment of pHLIP analogues in acidic regions. Therefore, following PET imaging at 4h p. i., tumors were excised, sectioned and evaluated by digital autoradiography for 18 F-distribution. This distribution was subsequently compared to the distribution of the vascular perfusion marker Hoechst 33342 using the same section. A clear pattern of higher 18 F uptake in tumor regions lacking vascular perfusion was observed (Figure 4). Previous studies have shown that tumor regions with low vascular perfusion are associated with low pO2 and anaerobic glycolysis, resulting in lactate accumulation and reduced pH. 37 The accumulation of [ 18 F]-4a in tumor regions lacking vascular perfusion is consistent with the proposed mechanism of pHLIP accumulation, and corroborates the specificity underlying the heterogeneous tumor distribution observed in the PET imaging.  The highest accumulation of radioactivity at both time points was detected in the liver (32.5 ± 4.3%ID/g and 34.9 ± 5.8%ID/g at 2 and 4 h p. i., respectively), followed by the kidneys. As expected with large hydrophobic peptides the blood clearance was slow. [ 64 Cu]-pHLIP shows highest accumulation in LNCaP tumors at 1 h p. i. (4.5 ± 1.7%ID/g) and highest tumor to blood ratio at 24 h p.i. 9 .
The tumor-to-tissue ratios for [ 18 F]-4a and [ 18 F]-5 are shown in Table 2. The tumor (LNCaP)-to-liver and the tumor (LNCaP)-to-kidney ratios are 0.2 and 0.3, respectively, both at 2 and 4 h. From our limited biodistribution studies it is difficult to differentiate between excretory and metabolic factors contributing to these high values. As shown in table 2, the negative control peptide [ 18 F]-5 exhibits similar ratios in liver and kidney. It can be hypothesized that the activity in these organs is due to excretion rather than metabolism. The tumor-to-muscle ratio of 4.5 measured at 2 h p. i. with [ 18 F]-4a in comparison to 0.4 with [ 18 F]-5, demonstrates we are specifically targeting the tumor with [ 18 F]-4a and is not a result of non-specific accumulation. The tumor-to-muscle ratio is even higher at 4 h p. i. with a value of 5.7. These studies clearly demonstrate that this click approach with prosthetic group 3 has no negative impact on the biological properties of pHLIP. Additionally, higher %ID/g values in LNCaP tumors as compared to the PC-3 tumor indicate that the accumulation of the tracer reflects acidity of the tissue.
[ 18 F]-pyridines labeled at ortho-position to the nitrogen have been reported to be stable against in vivo defluorination. 31 However, we observe some bone uptake with [ 18 F]-5 (1.31 ± 0.63 %ID/g) and [ 18 F]-4a (4.14 ± 1.12 %ID/g) 2h p. i.. In comparison, [ 18 F]-Fluoropentyne based prosthetic group shows very minimal bone uptake (0.39 %ID/g at 1h p.i.). In our case, it is unclear whether the uptake is result of defluorination or specific accumulation, because the two peptides radiolabeled wth same prosthetic group show different bone uptake values.

CONCLUSION
We have extended the use of the CuAAC "click chemistry" with the development of 2ethynyl-6-[ 18 F]fluoro-pyridine as a new prosthetic group suitable for the 18 F-labeling of large peptides. This approach should be widely applicable and was shown to be efficient for the radiolabeling of pHLIP analogues with a molecular weight over 4000 Da. Two 18 Flabeled analogues were injected in mice in order to acquire the first in vivo [ 18 F]-pHLIP data. The WT-pHLIP construct showed good in vitro stability and only mild in vivo defluorination was observed in both cases. A milestone in the development of a 18 F-labeled pHLIP tumor imaging agent was achieved, as the use of [ 18 F]-3 will allow for the fast production and evaluation of second generation pHLIP analogues, designed for higher accumulation in the tumors and faster clearance from non-target tissues. Examples of [ 18 F]-labeled prosthetic groups for the CuAAC "click" reaction with functionalized peptides and proteins.       Table 2 LNCaP tumor-to-tissue ratios determined upon biodistribution studies with the tracers administered to tumor bearing mice.