Enhancing PET Signal at Target Tissue in Vivo: Dendritic and Multimeric Tris(hydroxypyridinone) Conjugates for Molecular Imaging of αvβ3 Integrin Expression with Gallium-68

Tris(hydroxypyridinone) chelators conjugated to peptides can rapidly complex the positron-emitting isotope gallium-68 (68Ga) under mild conditions, and the resulting radiotracers can delineate peptide receptor expression at sites of diseased tissue in vivo. We have synthesized a dendritic bifunctional chelator containing nine 1,6-dimethyl-3-hydroxypyridin-4-one groups (SCN-HP9) that can coordinate up to three Ga3+ ions. This derivative has been conjugated to a trimeric peptide (RGD3) containing three peptide groups that target the αvβ3 integrin receptor. The resulting dendritic compound, HP9-RGD3, can be radiolabeled in 97% radiochemical yield at a 3-fold higher specific activity than its homologues HP3-RGD and HP3-RGD3 that contain only a single metal binding site. PET scanning and biodistribution studies show that [68Ga(HP9-RGD3)] demonstrates higher receptor-mediated tumor uptake in animals bearing U87MG tumors that overexpress αvβ3 integrin than [68Ga(HP3-RGD)] and [68Ga(HP3-RGD3)]. However, concomitant nontarget organ retention of [68Ga(HP9-RGD3)] results in low tumor to nontarget organ contrast in PET images. On the other hand, the trimeric peptide homologue containing a single tris(hydroxypyridinone) chelator, [68Ga(HP3-RGD3)], clears nontarget organs and exhibits receptor-mediated uptake in mice bearing tumors and in mice with induced rheumatoid arthritis. PET imaging with [68Ga(HP3-RGD3)] enables clear delineation of αvβ3 integrin receptor expression in vivo.


■ INTRODUCTION
Peptide-based imaging agents in nuclear medicine have tremendous utility in diagnosis, prognosis, and selection of therapeutic regimes for patients. Radiometals can be incorporated into clinically relevant peptides via a bifunctional chelator, providing effective and sensitive radiotracers that can be prepared conveniently in a radiopharmacy. The metallic isotope 68 Ga possesses decay properties that are suitable for positron emission tomography (PET) (68 min half-life, 1899 keV β + emission with 88% abundance), and the availability of a pharmaceutical grade 68 Ge/ 68 Ga generator means that the number of molecular imaging agents based on 68 Ga is likely to increase in the coming years. The somatostatin receptor 2targeted imaging agent, 68 Ga-DOTATATE for neuroendocrine tumors, 1−3 and more recently, the prostate specific membrane antigen targeted conjugate, 68 Ga-HBED-PSMA, 4,5 have dem-onstrated clinical utility in patient prognosis and management. Other chelator systems have exhibited efficient 68 Ga 3+ radiolabeling properties, and the resulting radiolabeled conjugates are effective at delineating target tissue in vivo. 6−12 Among these are the tripodal tris(hydroxypyridinone) chelators containing three 1,6-dimethyl-3-hydroxypyridin-4-one groups that can coordinate radiometallic Ga 3+ and Zr 4+ ions, 13−15 as well as Fe 3+ and Al 3+ with high affinity. 16,17 Upon deprotonation of hydroxyl groups, the hexadentate O 6 ligand, THP-Ac, can coordinate 68 Ga 3+ at mild pH (pH 6.5−7.5) and low ligand concentrations (10 μM) in <5 min. 13 Bifunctional derivatives of THP-Ac can be attached to peptides, with the resulting conjugates able to rapidly and quantitatively coordinate aqueous 68 Ga 3+ at room temperature at pH 5− 7. 18,19 Multimeric peptide-based imaging agents that incorporate more than one peptide targeting group (and in particular small peptides that target the α v β 3 integrin receptor) have demonstrated increased accumulation of activity at tissue targets in vivo relative to their monomeric analogues, and are effective contrast agents. 8,12,20−27 This increased accumulation of radiotracer in the case of multimeric compounds can be attributed to either (i) divalent binding, where two targeting peptides bind to two receptors simultaneously, requiring sufficient distance between the two intramolecular targeting ligands to form this bridge; or (ii) an increased local effective concentration of targeting peptide at the receptor site, resulting in a shift in equilibrium toward peptide receptor binding. In the case where the multimeric derivative has an increased persistence in the blood pool in vivo relative to the monomeric derivative, it is also possible that increased uptake is a result of increased bioavailability.
Dendritic and multimeric chelator systems for coordination of Gd 3+ and lanthanides provide enhanced contrast as MRI agents compared to their single-chelator counterparts, in large part due to the increased number of metal complexes present. 28−36 Multimeric constructs containing multiple chelating groups (including hydroxypyridinones) have also been synthesized for the purposes of sequestering trace metal ions. 17,37 A small body of work has been described in which the number of radiometal coordination sites on a radiolabeled protein/antibody has been increased by incorporation of either a dendrimer possessing multiple chelating groups, 38−41 or a pendant function (such as polylysine) on which multiple chelators are appended. 42 −44 In these cases, the functionalized/ radiolabeled biomolecule retained affinity for receptor targets in vitro 40,41,44 and in vivo. 38,39,42,43 Additionally, when compared to bioconjugates containing fewer chelating groups, dendrimer or multichelator conjugated proteins demonstrated higher specific activitythat is, a higher number of radionuclides were bound per bioconjugate compared to homologues containing a single chelator. 38,41−43 In some cases, this resulted in higher concentrations of radioactivity at target tissue (tumors) in vivo. 42 We are interested in using tris(hydroxypyridinone) ligands to explore new ways of increasing accumulated radioactivity at diseased tissue where target receptors are localized in vivo.
Here we aim to incorporate multiple in vivo targeting peptide groups and multiple chelator groups into a single molecule, thus increasing the affinity of the radiolabeled conjugate for receptors in vivo as well as increasing the specific activity of the radiolabeled species itself. This approach mimics that of multifunctionalized nanoparticles, where multiple copies of a targeting biomolecule/modifying agent and multiple chelating groups are appended to the surface of a nanoparticle. 45−48 We have employed the cyclic pentapeptide, c(RGDfK), (Chart 1) that targets the endothelial α v β 3 integrin receptor expressed in angiogenesis, and is associated with metastatic disease in cancers. The biology of such conjugates is well understood, and derivatives of c(RGDfK) are useful model compounds for assessing novel approaches to molecular imaging using peptidebased agents.
We have synthesized a novel bifunctional chelator construct that incorporates three hexadentate tris(hydroxypyridinone) metal binding sites for Ga 3+ and an isothiocyanate group that can be utilized as an attachment point to lysine residues of the N-terminus of peptides. We have also employed a scaffold for facile attachment of three c(RGDfK) peptide targeting groups that contain available Lys or N-terminal primary amine groups, ultimately providing a trimeric targeting motif. Conjugating these two motifs together results in a dendritic compound where an aryl group links a trimeric peptide targeting group and three tris(hydroxypyridinone) (THP) groups. We have also synthesized homologues of this compound to investigate the effect of incorporating multiple copies of a peptide/chelating group on in vitro uptake and in vivo biodistribution and target tissue accumulation. Finally, we have chosen the radiotracer with the best in vivo tumor to nontarget organ contrast and assessed the radiotracer in a model of rheumatoid arthritis to further illustrate the utility of 68 Ga-labeled THP conjugates that target integrin receptors for molecular imaging of in vivo events.

Synthesis.
To synthesize a multidentate chelator containing nine 1,6-dimethyl-3-hydroxypyridin-4-one groups (as three tris(hydroxypyridinone) moieties), the benzyl-protected tris-(hydroxypyridinone) (2) (3 equiv) was reacted with tricarboxylate (1), HATU, and diisopropylethylamine under microwave conditions (Scheme 1). The resulting compound (3) was purified using silica chromatography. Subsequent removal of the phthalimide group furnished 4, which possesses a primary amine that can be further functionalized for attachment to biomolecules. This species was converted to a phenyl isothiocyanate by reacting it with an excess of p-phenylene diisothiocyanate, resulting in 5, which was isolated via semipreparative reverse phase HPLC. Compound SCN-HP 9 was obtained upon deprotection of the benzyl-protected hydroxypyridinone groups by reaction of 5 with boron trichloride in dichloromethane, followed by addition of methanol.
This study aimed to assess a trimeric scaffold in which all three peptide motifs are equidistant from the bifunctional chelator and chemically equivalent to each other, thus ensuring that all three motifs possess the same affinity for the target receptor. To achieve this, tricarboxylate 1 was converted to a Nhydroxysuccinimide-activated intermediate (6) which was reacted with 3 equiv of c(RGDfK), to provide the multimeric compound 7. Compound 7 was purified and isolated using RP-HPLC. The phthalimide group was removed by reaction with hydrazine in ethanol, followed by addition of HCl (Scheme 2). Subsequently, RGD 3 , which contains three c(RGDfK) peptide groups and an apical primary amine, was isolated using RP-HPLC.
Reaction of RGD 3 with SCN-HP 9 at 120°C at 300 W in a microwave reactor provided HP 9 -RGD 3 (Chart 1) which was purified and isolated using RP-HPLC in ≥95% purity. HP 9 -RGD 3 possesses three c(RGDfK) groups and three tris-(hydroxypyridinone) groups, each attached through a tripodal scaffold to a central phenyl group via thiourea linkages. Similarly, the compound SCN-HP 3 (Chart 1), synthesized as previously described, 18 was reacted with (i) RGD 3 and (ii) c(RGDfK) under the same microwave conditions. This provided the derivatives HP 3 -RGD 3 and HP 3 -RGD (Chart 1) which are trimeric and monomeric c(RGDfK) homologues of HP 9 -RGD 3 , the former two conjugates containing only a single tris(hydroxypyridinone) ligand for coordination of a single metal ion. These conjugates were also purified using RP-HPLC and isolated in ≥95% purity (as previously reported for HP 3 -RGD 18 ).
Reaction of Hydroxypyridinone Derivatives with 68 Ga 3+ / nat Ga 3+ . Solutions of HP 9 -RGD 3 , HP 3 -RGD 3 , and HP 3 -RGD were added to an excess of nat Ga(NO 3 ) 3 . For each of these solutions, the reverse phase LCMS total ion chromatogram revealed Ga 3+ complexation of the hydroxypyridinone conjugates ( Figure 1). As predicted, the conjugate HP 9 -RGD 3 that contains nine hydroxypyridinone groups (three hexadentate tris(hydroxypyridinone) groups) coordinated three Ga 3+ metal ions, whereas the conjugates HP 3 -RGD 3 and H P 3 -RGD, each possessing only a single tris-(hydroxypyridinone) chelator, were bound to only a single Ga 3+ ion. In each case, the most intense set of peaks in the mass  Table 1.
Solutions of 68 Ga 3+ produced from a 68 Ge/ 68 Ga generator (80−120 MBq mL −1 , 0.1 M HCl aqueous solution, 1 mL) were added to HP 9 -RGD 3 , HP 3 -RGD 3 , and HP 3 -RGD and immediately adjusted to pH 6 by addition of ammonium acetate solution (1 M, 300 μL). As previously described, 18,19 these radiolabeling reactions proceeded rapidly at room temperature in less than 5 min, providing radiolabeled conjugates in >98% radiochemical yield. These mixtures were analyzed using the same HPLC conditions as those employed for the above LCMS studies. Each reaction solution gave rise to only a single signal in the radiochromatogram, each with a similar retention time to corresponding LCMS chromatogram signals ( Figure 1, Table 1). Under these conditions, the maximum occupancy of bioconjugate molecules by 68 Ga 3+ was lowfor HP 9 -RGD 3 , maximum occupancy was <0.03%, and for HP 3 -RGD 3 or HP 3 -RGD, maximum occupancy was <0.01%.
To discern whether [Ga(HP 9 -RGD 3 )] could be chromatographically distinguished from [Ga 2 (HP 9 -RGD 3 )] and [Ga 3 (HP 9 -RGD3)], solutions of HP 9 -RGD 3 were added to aqueous solutions containing 1, 2, or 3 equiv of Ga 3+ , and the samples were analyzed by reverse phase LCMS coupled to a UV detector ( Figure 2) using a less steep solvent gradient than that employed above. [Note: To avoid confusion over the oxidation state of the metal and the pH being monitored, we adopt normal convention in showing peptides as neutral or uncharged ligands either alone or in metal complexes, even though Arg/R side chains are protonated in water at pH < 9, and Asp/D side chains are deprotonated at pH > 5. Additionally, we have elected not to indicate the protonation state of HP groups in our nomenclature, even though at pH < 4.5 in water, HP groups are protonated bearing a single positive charge, and >8 HP groups are deprotonated bearing a single negative charge. At pH 5−7.5 in water, HP groups are neutral. Upon coordination to a metal ion, each bidentate HP group is deprotonated, formally bearing a single negative charge. Thus, HP 3 and HP 9 groups and trivalent metal complexes thereof are uncharged at pH 5−7.5 in aqueous solutions.] When an analytical HPLC column was utilized for LCMS analysis, the UV chromatogram of non-Ga-treated HP 9 -RGD 3 revealed at least three species (Figure 2b). The mass spectrum indicated that these species corresponded to [Al(HP 9 -RGD 3 Figure 2. (a) Mass spectrum (right) of HP 9 -RGD 3 acquired immediately after preparative HPLC purification (without analytical HPLC separation). Analytical HPLC UV chromatograms (left) and mass spectra (right) of solutions of HP 9 -RGD 3 with (b) no added Ga 3+ , (c) 1 equiv of Ga 3+ , (d) 2 equiv of Ga 3+ , (e) 3 equiv of Ga 3+ , (f) 68  these complexes came from storage vials and/or parts of the LCMS instrumentation, including the analytical reverse phase C18 column. Notably, samples of HP 9 -RGD 3 isolated immediately after preparative HPLC that were not passed through an analytical HPLC column, nor lyophilized and stored for extended periods of time, contained fewer metal complex species, as measured by mass spectrometry (Figure 2a), although the [Al(HP 9 -RGD 3 )] complex was still prominent. In solutions containing a single equivalent of Ga 3+ , the major Ga-containing species corresponded to [Ga(HP 9 -RGD 3 )] and [AlGa(HP 9 -RGD 3 )], and the corresponding UV chromatograms indicated the presence of multiple species (Figure 2c). In solutions containing 2 equiv of Ga 3+ , several Ga-containing species could be detected including [AlGa(HP 9 -RGD 3  The same "slow" LCMS gradient coupled to UV detection was used to analyze a sample nominally described here as [ 68 Ga(HP 9 -RGD 3 )] (Figure 2f,g). The concentration of 68 Ga was too low to allow detection of 68 Ga-containing species using the mass spectrometer available in our laboratory. The species of lowest molecular weight most closely matched stoichiometries of [AlFe(HP 9 -RGD 3 )], [Fe 2 (HP 9 -RGD 3 )], and [FeGa-(HP 9 -RGD 3 )]. Unlike samples of HP 9 -RGD 3 , neither [Al(HP 9 -RGD 3 )] nor [Fe(HP 9 -RGD 3 )] was detected. Generator eluate is known to contain significant amounts of Fe 3+ , 49 and it is likely that HP 9 -RGD 3 complexes this during the radiolabeling reaction. Indeed, unpublished data from our laboratory indicate that concentrations of Fe, Al, and nat Ga in generator eluate (0.1 M HCl, from an Eckert and Ziegler generator) are approximately 0.5 μM, 3 μM, and 1 μM, respectively. At these concentrations, Fe 3+ , Al 3+ , and nat Ga 3+ will compete with 68 Ga 3+ (present at picomolar concentrations) for ligand coordination, accounting for low percentage occupancy by 68 Ga 3+ of all three conjugates. Most importantly, the radiolabeled species in solution, referred to as [ 68 Ga(HP 9 -RGD 3 )] for the purposes of simple nomenclature, most likely corresponds to [M 2 68 Ga(HP 9 -RGD 3 )] or [M 68 Ga(HP 9 -RGD 3 )] (M = Al, Fe, or nat Ga). Thus, in vivo studies, described herein, measure the biodistribution of 68 Ga-labeled HP 9 -RGD 3 in which one or two tris(hydroxypyridinone) sites are occupied by nonradioactive metal ions. The resolution of the scintillation detector was lower than the resolution of the UV detector, and using the same gradient, the radiochromatogram of [ 68 Ga(HP 9 -RGD 3 )] gave rise to a single, very broad peak, coincident with the UV chromatogram signals.
Significantly higher amounts of 68 Ga 3+ were utilized to assess the maximum specific activity that could be complexed quantitatively by the conjugates. In this procedure, 68 Ga 3+ eluted from an iThemba generator was processed to preconcentrate the generator eluate and remove 68 Ge breakthrough. 50 The final 68 Ga 3+ solution consisted of 700−1000 MBq 68 Ga 3+ in H 2 O/ethanol (82%/18%, with [HCl] = 0.18 M, total volume = 1 mL). The conjugates HP 9 -RGD 3 , HP 3 -RGD 3 , and HP 3 -RGD were added to this solution followed by ammonium acetate and saline to obtain solutions (2.5 mL total volume) at pH 5.5−6.5 that were suitable for in vivo administration. Radiochemical yields of >97% were achieved using a minimum of 4 nmol of HP 9 -RGD 3 and 12 nmol of either HP 3 -RGD 3 or HP 3 -RGD. Decreasing the amount of either HP 3 -RGD 3 or HP 3 -RGD to 4 nmol decreased the radiochemical yield (86% and 88%, respectively). Again, these reactions proceeded rapidly (<5 min) at room temperature, illustrating the utility of tris(hydroxypyridinone) chelators for applications requiring rapid, kit-based labeling. The maximum specific activity of 68 Ga-labeled HP 9 -RGD 3 (180−240 MBq nmol −1 ) compared with that of HP 3 -RGD 3 or HP 3 -RGD (60− 80 MBq nmol −1 ) was 3-fold higher. The maximum occupancy by 68 Ga 3+ per molecule of bioconjugate was low in all cases. For HP 9 -RGD 3 , maximum occupancy was <0.3%, and for HP 3 -RGD 3 or HP 3 -RGD, maximum occupancy was <0.1%.
Hydrophilicity of 68 Ga-Radiolabeled Tracers, and Affinity for α v β 3 Integrin Receptor. The log D OCT/PBS values of the 68 Ga-labeled conjugates indicated that all three radiotracers possess comparable hydrophilicity (Table 1) Figure S1). The IC 50 value of [Ga(HP 3 -RGD)] (46 ± 16 nM) was comparable to that of c(RGDfK) (28 ± 8 nM), demonstrating that conjugation did not markedly affect affinity for the α v β 3 integrin receptor. However, the IC 50 values of [Ga(HP 3 -RGD 3 )] (73 ± 22 nM) and [Ga 3 (HP 9 -RGD 3 )] (79 ± 28 nM) were higher than those of c(RGDfK) and [Ga(HP 3 -RGD)], suggesting that, in this series of compounds, the presence of three copies of c(RGDfK) peptide on a single molecule did not impart higher affinity toward α v β 3 integrin compared to the monomeric THP derivative. This was surprising given that it has been reported that other dimeric, 22,23 trimeric, 27 and multimeric 51 RGD peptides demonstrate higher affinities for α v β 3 integrin than their monomeric homologues. The addition of the dendritic HP 9 chelator did not affect affinity of RGD 3   higher than that of [ 68 Ga(HP 3 -RGD 3 )], but this was not statistically significant. By 2 h PI, there were no differences in normalized tumor uptake between the three tracers.
Animals administered [ 68 Ga(HP 9 -RGD 3 )] showed increased persistence of activity in blood, and increased liver, spleen, muscle, and lung accumulation compared to animals administered either [ 68 Ga(HP 3 -RGD 3 )] or [ 68 Ga(HP 3 -RGD)] at both 1 and 2 h PI ( Figure 3, Table S1). At both 1 and 2 h PI, the concentration of kidney activity for animals administered [ 68 Ga(HP 9 -RGD 3 )] (1 h PI, 28.32 ± 3.76%ID g −1 ) was considerably higher than that of animals administered [ 68 Ga-(HP 3 -RGD 3 )] (1 h PI, 10.73 ± 0.18%ID g −1 ), which in turn was higher than that of animals administered [ 68 Ga(HP 3 - Additionally, ex vivo biodistribution data (Figures 3 and S2, Table 3) for all three tracers demonstrated that tumor uptake was target-specific at 1 h PI, with significantly lower tumor uptake for animals coadministered c(RGDfK) ("blocked" groups). In the blocked groups for all three tracers, uptake in lungs, heart, spleen, and muscle was generally lower than that observed in "nonblocked" groups. This is consistent with previous reports of receptor-mediated uptake of α v β 3 integrin targeted radiotracers, 18,22,25,27 and with the expression of α v β 3 integrin at low levels in normal vasculature tissue. 52,53 For all three tracers, radioactivity concentration in the kidneys was higher for the blocked groups than for the nonblocked groups. Similar patterns have previously been described for RGD peptide conjugates, 11 and have been attributed to slower kidney filtration rates in the presence of larger amounts of peptide.
PET imaging data for all three tracers was consistent with ex vivo biodistribution data. In PET scans of all three radiotracers acquired 1 h PI, tumors could be delineated, whereas in PET scans of animals coadministered c(RGDfK), tumors could not be delineated (Figure 3). Concentration of radioactivity in kidneys and livers for animals administered [ 68 Ga(HP 9 -RGD 3 )] was significantly higher than that of animals administered either [ 68 Ga(HP 3 -RGD 3 Figure S3); therefore, further imaging studies assessing α v β 3 integrin expression in a mouse model of rheumatoid arthritis utilized this tracer. Imaging Arthritis with [ 68 Ga(HP 3 -RGD 3 )]. To demonstrate the broad utility of tris(hydroxypyridinone) conjugates for molecular imaging of α v β 3 integrin receptor expression, [ 68 Ga(HP 3 -RGD 3 )] was used to image synovial inflammation in mice with induced rheumatoid arthritis. The α v β 3 integrin receptor is associated with cell types that are involved in rheumatoid arthritis at synovial joints, including bone resorbing osteoclasts, activated macrophages, and angiogenic endothelial cells. 53,54 Previous studies have demonstrated that radiolabeled RGD peptides target arthritic tissue 55−57 as well as osteoclasts. 58 In these experiments, [ 68 Ga(HP 3 -RGD 3 )] (8−13 MBq) was administered to healthy mice (C57Bl/6 male 6−8 week mice), and mice with induced mild and severe rheumatoid arthritis. 59,60 A separate group of animals with severe rheumatoid arthritis was administered both [ 68 Ga(HP 3 -RGD 3 )] and an excess of c(RGDfK) to assess specificity of in vivo targeted accumulation. Prior to induction of arthritis and administration of radiotracer, the diameters of wrist and ankle joints of all animals were measured using calipers to score the severity of arthritis. At 1 h PI of radiotracer, animals were euthanized by CO 2 asphyxiation followed by PET scanning and organ collection for ex vivo biodistribution.
For all animals, radioactivity concentrations in nontarget tissues and organs, as determined by ex vivo biodistribution data ( Figure S4), were consistent with results obtained from animals bearing U87MG tumors administered the same radiotracer. Blockade studies with c(RGDfK) in rheumatoid arthritis animals also indicated that some localization of [ 68 Ga(HP 3 -RGD 3 )] in all harvested organs except the kidneys was likely to be receptor-mediated (Table S2).
The ex vivo biodistribution data demonstrated that the amount of radioactivity in severely arthritic joints (increase in wrist and ankle diameter >0.5 mm) was higher than that in joints of healthy animals, or in joints of animals with mild arthritis (increase in wrist and ankle diameter <0.5 mm). In this analysis, joints were grouped according to their type (wrist or ankle) and the increase in swelling (caliper joint measurements) from point of disease induction to the point of imaging (8 days). Total %ID is reported rather than %ID g −1 , because of the large experimental error associated with dissecting and measuring the mass of mouse joint tissue. In severely arthritic joints, there was significantly higher radioactivity accumulation than in joints that were not classified as severely arthritic (including healthy animals and animals induced with mild arthritis) (Figure 4, Table 5). On the basis of %ID, severely arthritic joints could be differentiated from healthy joints, or joints that did not exhibit severe inflammation. Furthermore, blockade studies in severely arthritic animals demonstrated that uptake of [ 68 Ga(HP 3 -RGD 3 )] in inflamed arthritic joints was receptor-mediated ( Figure 4, Table 5). Joint uptake in severely arthritic animals coadministered [ 68 Ga(HP 3 -RGD 3 )] and c(RGDfK) peptide was significantly lower than that of severely arthritic animals administered solely [ 68 Ga(HP 3 -RGD 3 )].
When animals' joints were grouped according to extent of disease (classified according to whether animals were induced with mild or severe arthritis rather than increase in joint swelling), there were no statistically significant differences in joint uptake between healthy mice and mice with mild arthritis. There were statistically significant differences between healthy mice and mice with severe arthritis, and between mice with mild arthritis and those with severe arthritis ( Figure S5, Table  S3).
PET scans were consistent with biodistribution data ( Figure  5). In PET images acquired 1 h PI, severely arthritic joints could be distinguished from joints that were not inflamed, both

Bioconjugate Chemistry
Article between different animals (Figure 5a vs b) and within the same animal ( Figure 5b). For example, within one animal with severe arthritis, the left wrist was severely inflamed (a 1 mm increase in swelling from the point of disease induction) but the right demonstrated no swelling throughout the disease induction period. In the PET/CT scan, higher radiotracer uptake was discernible in the left wrist compared to the right wrist ( Figure  5b), consistent with ex vivo biodistribution data.

■ DISCUSSION AND CONCLUDING REMARKS
Multimeric receptor-targeted radiolabeled agents, in which each molecule contains either multiple copies of each targeting peptide, or multiple copies of a multidentate chelator, 42 can deliver increased radioactive payloads to target tissue relative to monomeric homologues. Dimeric, 25,61−63 trimeric, 8,27 tetrameric, 64,65 and octameric 66 RGD conjugates radiolabeled with metallic PET isotopes have demonstrated efficacious α v β 3 integrin targeting in vivo. Herein, we have described the first radiolabeled molecular compound that integrates multiple copies of a chelator and multiple copies of peptide, in this case, α v β 3 integrin-targeted c(RGDfK). In contrast to previous work on multimeric RGD peptide conjugates, the new dendritic system possesses tris(hydroxypyridinone) chelators that coordinate 68 Ga 3+ quantitatively at ambient temperature, near neutral pH, and micromolar concentrations of conjugate, allowing for rapid, one-step labeling. Synthesis of HP 9 -RGD 3 , which consists of a bifunctional enneakis(hydroxypyridinone) ligand attached to a trimeric peptide targeting α v β 3 integrin receptors, enabled assessment of whether dendritic chelator systems could enhance specific activity in radiolabeled molecular imaging agents. Incorporation of multiple hexadentate tris(hydroxypyridinone) groups into a single construct, HP 9 -RGD 3 , increased specific activity relative to HP 3 derivatives. At concentrations of ∼1.6 μM, nearquantitative radiochemical yields (>97%) were achieved for HP 9 -RGD 3 , whereas for HP 3 -RGD and HP 3 -RGD 3 , concentrations of 5 μM were required to achieve the same radiochemical yields. The maximum specific activity of 68 Galabeled HP 9 -RGD 3 (180−240 MBq nmol −1 ) was 3-fold higher when compared with that of HP 3 -RGD 3 or HP 3 -RGD (60−80 MBq nmol −1 ).
Analytical HPLC using a UV detector (interfaced with a mass spectrometer and a gamma scintillation detector) could distinguish numerous signals arising from a sample of [ 68 Ga(HP 9 -RGD 3 )]. For the same sample, line widths of signals in the radiochromatogram and the TIC chromatograms were much broader, and could not be precisely correlated with individual signals in the UV chromatogram. Mass spectral analysis indicated that all tris(hydroxypyridinone) sites in [ 68 Ga(HP 9 -RGD 3 )] were likely occupied by trivalent metal ions. Hexadentate THP-Ac has a very high affinity for Fe 3+ , with log K 1 = 32.52 and pFe 3+ = 28.47 at pH 7.4 ([Fe 3+ ] = 10 −6 M, [Ligand] = 10 −5 M). 17 Dendritic derivatives have similar affinities. 17 While affinities of 3-hydroxy-4-pyridinones for Al 3+ are normally several orders of magnitude lower than for Fe 3+ , they are still highfor example, for a series of tripodal tris(hydroxypyridinones), pAl 3+ consistently measured 21−22 at pH 7.4. 16 It is unsurprising that in 68 Ga-labeled HP 9 -RGD 3 , tris(hydroxypyridinone) sites unoccupied by 68 Ga 3+ complex adventitious Fe 3+ , Al 3+ , and nat Ga 3+ .
It was hypothesized that the higher specific activity and the trimeric peptide nature of [ 68 Ga(HP 9 -RGD 3  Additionally, for HP 9 -RGD 3 , it is likely that there was comparatively lower occupancy of target α v β 3 integrin in vivo by unlabeled conjugate. Comparable amounts of radioactivity were administered to each mouse for the three different radiotracers, but in the case of HP 9 -RGD 3 , ∼0.15 nmol of conjugate was administered to each animal, compared to ∼0.5 nmol for HP 3 -RGD 3 and HP 3 -RGD. There was an increase in radioactivity concentration in tumors at 1 h PI for [ 68 Ga(HP 9 -RGD 3 )] compared to [ 68 Ga(HP 3 -RGD 3 )] (mean difference = 0.41%ID g −1 ), and for [ 68 Ga(HP 3 -RGD 3 )] compared to [ 68 Ga(HP 3 -RGD)] (mean difference = 1.42%ID g −1 ) (Tables 3, 4). There was a statistically significant increase (mean difference = 1.83%ID g −1 ) in tumor accumulation at 1 h PI for [ 68 Ga(HP 9 -RGD 3 )] compared to [ 68 Ga(HP 3 -RGD)]. Thus, the combination of increased specific activity and a multimeric effect could be responsible for this increase in radioactivity concentration in tumors of animals administered the dendritic radiotracer. Such an observation lends credence to the hypothesis that dendritic radiotracers that incorporate multiple chelating groups and copies of targeting peptides can increase radioactive payload at tissue, relative to monomeric homologues. The higher mean difference in tumor uptake between [ 68 Ga(HP 3 -RGD 3  The distance between c(RGDfK) groups in HP 3 -RGD 3 and HP 9 -RGD 3 is not sufficient to allow simultaneous binding to two or more α v β 3 integrin receptors. It is possible that

Bioconjugate Chemistry
Article increasing this distance to allow simultaneous binding could result in higher uptake and retention in tissue that overexpresses α v β 3 integrin receptors. 62,63 We have previously demonstrated that THP peptide conjugates can rapidly and quantitatively complex 68 Ga 3+ at near-neutral pH, ambient temperature, and low chelator concentrations, and that the resulting radiotracers can delineate tumor tissue in vivo. 18,19 [ 68 Ga(HP 3 -RGD 3 )], prepared using simple radiolabeling procedures, can image synovial inflammation in mice with induced rheumatoid arthritis, distinguishing severely arthritic joints from disease-free joints. Few data are available on α v β 3 integrin targeted radiotracers for rheumatoid arthritis, 55,56 and the mouse models employed in previously published studies differ from the model utilized here. However, we note that [ 68 Ga(HP 3 -RGD 3 )] uptake in severely arthritic synovial joints of mice in a serum transfer-induced arthritis model (0.26 ± 0.02%ID in wrists and 0.68 ± 0.18%ID in ankles 1 h PI) is comparable to 111 In-DOTA-E(RGDfK) 2 uptake in mice in a collagen-induced arthritis model (0.2 ± 0.1%ID in arthritic joints at 1 h PI). 55 Rheumatoid arthritis is currently diagnosed using information from physical examination (signs and symptoms), blood tests, X-rays, and MRI. Ideally, it is diagnosed early so that treatment that slows or halts disease progression can begin. Current early diagnostic methods detect anatomical changes in patients' joints and so arthritic sites in patients that are not symptomatic or visually identifiable are missed. 67,68 PET/CT imaging of α v β 3 integrin expression using kit-based radiopharmaceuticals could provide whole-body diagnosis at early stages of disease before the manifestation of debilitating symptoms, as well as monitor patient response throughout treatment, providing greater effectiveness of treatment at early stages of disease. Such radiotracers could also be used to predict whether an individual patient will respond to a particular molecular therapeutic, based on the presence of a therapeutic receptor target. The availability of 68 Ga from a benchtop generator, coupled with a chelator conjugate that enables one-step, kitbased syntheses, which are not readily achievable with other chelators, brings about the possibility of widely available whole body PET molecular imaging of rheumatoid arthritis without the need for a local cyclotron or automated radiosynthesis equipment.

■ EXPERIMENTAL SECTION
Materials and Instrumentation. Chemicals and reagents were obtained from Sigma-Aldrich (Dorset, UK) unless otherwise specified. High-performance liquid chromatography (HPLC) analysis was carried out using an Agilent 1200 LC system with in-line UV and gamma detection (Flow-Count, LabLogic). NMR spectra were acquired on either a Bruker Avance 400 or 700 spectrometer (Bruker UK Limited, Coventry, UK). Spectra were referenced to residual solvent signals. Mass spectra were recorded in the positive ion mode on an Agilent 6510 Q-TOF LC/MS mass spectrometer coupled to an Agilent 1200 LC system (Agilent, Palo Alto, CA). Data were acquired and reference mass-corrected via a dual-spray electrospray ionization source, using the factory-defined calibration procedure. Analytical reverse-phase LCMS and radio-LCMS traces were acquired using an Agilent Eclipse XDB-C18 column (4.6 × 150 mm, 5 μm). Instant thin layer chromatography strips (ITLC-SG) were obtained from Varian Medical Systems UK, Ltd. (Crawley, UK), and ITLC strips were visualized using a Raytest Rita-Star TLC scanner.
Semipreparative reverse-phase HPLC was conducted using an Agilent Eclipse XDB-C18 column (9.4 × 250 mm, 5 μm) coupled to an Agilent 1200 LC system, with a 3 mL min −1 flow rate and UV spectroscopic detection at 220 nm. Mobile phase A contained water with 0.2% TFA and mobile phase B contained acetonitrile with 0.2% TFA. All methods started with 100% A at 0 min. For method 1, the concentration of B increased at a rate of 2% min −1 ; for method 2, the concentration of B increased at a rate of 1% min −1 ; and for method 3, the concentration of B increased at a rate of 0.5% min −1 . Analytical reverse-phase HPLC and radio-HPLC traces were acquired using two different instruments: (1) an Agilent 1200 LC system and an Agilent Zorbax Eclipse XDB-C18 column (4.6 × 150 mm, 5 μm) with a 0.25−1 mL min −1 flow rate and UV spectroscopic detection at either 214 or 220 nm. The radio-HPLC was coupled to a LabLogic Flow-Count detector with a sodium iodide probe (B-FC-3200). Aliquots (10−200 μL) of each radiolabeled sample were injected onto the column. Mobile phase A contained water with 0.1% TFA and mobile phase B contained acetonitrile with 0.1% TFA. For method 4, flow rate was 1 mL min −1 , the concentration of B increased at a rate of 5% min −1 , with 100% A at 0 min, and 100% B at 20 min. For method 5, flow rate was 0.25 mL min −1 , the concentration of B increased at a rate of 2% min −1 , with 100% A at 0−5 min, and 100% B at 55 min. (2) An Agilent Zorbax Eclipse XDB-C18 column (4.6 × 150 mm, 5 μm) with a 1 mL min −1 flow rate and UV spectroscopic detection at either 214 or 220 nm coupled to a Shimadzu HPLC consisting of a SCL-10AVP system controller, a SIL-10ADVP autoinjector, a LC-10 ATVP solvent delivery unit, a FCV-10AL control valve, a DGU-14A degasser, and a SPD-10AVP UV detector. This was coupled to a radiation detector consisting of an Ortec model 276 Photomultiplier Base with Preamplifier, Amplifier, BIAS supply, and SCA and a Bicron 1 M 11.2 Photomultiplier Tube. For method 6, the concentration of B increased at a rate of 6.67% min −1 , with 100% A at 0 min, and 80% B at 12 min. For initial radiolabeling and characterization studies that utilized <400 MBq, an Eckert and Ziegler 68 Ge/ 68 Ga generator (Berlin, Germany) was used. For biodistribution studies, and experiments that utilized >600 MBq 68 Ga, an iThemba Laboratories 1.8 GBq 68 Ge/ 68 Ga generator (IDB Holland BV, Netherlands) was used.
For biodistribution with U87MG tumor bearing mice, 68 Ga eluate from an iThemba Lab generator was preconditioned as previously described. 50 Briefly, a cation exchange cartridge containing AG 50Wx4 resin (50 mg) was conditioned by passing through aqueous HCl solution (4 M, 1 mL) and deionized water (1 mL) sequentially. To elute the 68 Ge/ 68 Ga generator, aqueous HCl solution (0.4 M, 5 mL) was passed through the generator and transferred directly onto the cation exchange cartridge. The cartridge was dried with air (1 mL), washed with 0.15 M HCl in water/ethanol (20%/80%), and again dried with air (1 mL). A solution of 0.9 M HCl in water/ ethanol (200 μL, 10%/90%) was used to elute 68 Ga (800−1000 MBq), which was diluted to a volume of 1 mL with deionized water. Lyophilized peptide conjugate (trifluoroacetate salt) HP 3 -RGD (25 μg, 12 nmol), HP 3 -RGD 3 (50 μg, 13 nmol), or HP 9 -RGD 3 (25 μg, 4 nmol)dissolved in 20−40 μL of water/ ethanol (50%/50%) was added to the solution containing 68 Ga, immediately followed by a solution of ammonium acetate (2 M, 400 μL) and 0.9% saline (1100 μL). An aliquot for ITLC analysis was immediately applied to an ITLC-SG plate. The ITLC-SG plate was developed using an aqueous citrate buffer Log P OCT/PBS Determination. A solution containing 68 Gaconjugate (50 μL, synthesized using eluate from an Eckert and Ziegler generator as described above) was added to 550 μL of octanol and 500 μL of aqueous phosphate buffered saline solution. The mixture was agitated using a vortex, and the phases separated by centrifugation (1000 rpm, 2 min). Aliquots from each phase were counted for radioactivity in a gamma counter. The experiment was repeated four times.
PET Scanning and Biodistribution of U87MG Tumor Bearing Animals. All animal experiments with U87MG tumor bearing animals were performed with approval from the Peter MacCallum animal experimentation ethics committee. Six-to eight-week-old female Balb/c nude mice (Animal Resources Centre, Western Australia) were implanted subcutaneously on the right flank with 4 million U87MG cells. Once the tumors reached a volume >250 mm 3 (two to three weeks) the animals were injected intravenously with 13−30 MBq of [ 68 Ga(HP 3 -RGD)] (1 μg of conjugate), [ 68 Ga(HP 3 -RGD 3 )] (2 μg of conjugate), and [ 68 Ga(HP 9 -RGD 3 )] (1 μg of conjugate) (100 μL of solutions described above). At 1 and 2 h, the animals were anaesthetized and imaged on a Philips MOSAIC small animal PET scanner for 10 min each. For blocking studies, animals were coinjected with c(RGDfK) peptide (400 μg), anaesthetized at 1 h, and imaged for 10 min each. The images were reconstructed using a 3D RAMLA algorithm, and radiotracer uptake was determined by ex vivo tissue counting as described previously. 70 On completion of the scan animals were euthanized by cervical dislocation and tissues harvested, weighed, and radioactivity counted using a gamma counter (Biomedex). Confidence intervals and p values were calculated using a two-tailed t-test in GraphPad Prism 5.04.
PET/CT Scanning and Biodistribution of Animals with Induced Rheumatoid Arthritis. All animal experiments on mice with induced rheumatoid arthritis complied with the Animals (Scientific Procedures) Act (UK 1986) and Home Office (UK) guidelines. A K/BxN serum transfer model of arthritis was utilized in these experiments, and serum was produced as previously reported. 59,60 Wild type C57Bl/6 male 6−8-week-old mice (Charles River) were injected with 150 μL arthritogenic serum (intraperitoneally) on days 0 and 2. The arthritic serum was either diluted 2-fold with PBS (to induce severe arthritis) or 8-fold with PBS (to induce mild arthritis). PBS alone was also injected into a third group to assess uptake in nonarthritic mice. Disease severity was evaluated in mice on days 0, 2 and 8 post-injection, by measuring weight loss, thickness of swollen paws, using microcalipers (Kroeplin), and visual clinical scoring on a scale of 0−3 per paw added for a

Bioconjugate Chemistry
Article total score out of 12 per animal. On day 8, animals from the control, the mild arthritis and severe arthritis groups (n = 3) were injected intravenously with [ 68 Ga(HP 3 -RGD 3  [ 68 Ga(HP 3 -RGD 3 )] was prepared using protocols described above for radiolabeling with an Eckert and Ziegler 68 Ge/ 68 Ga generator. Animals were injected under isoflurane anesthesia, after which they were immediately allowed to recover. One hour post-injection, animals were culled by CO 2 asphyxiation and PET scanned using a nanoScanPET/CT (Mediso Medical Imaging Systems, Budapest, Hungary) and organs/tissues harvested, weighed, and radioactivity counted using a gamma counter. PET/CT images were reconstructed using VivoQuant (inviCRO, LLC, Boston, USA). Confidence intervals and p values were calculated using a two-tailed t-test in GraphPad Prism 5.04. conceived the study, coauthored the manuscript, undertook synthesis, radiolabeling, LCMS analysis, and competitive binding assays, and conceived of and assisted with biodistribution studies. All authors have given approval to the final version of the manuscript.

Notes
The authors declare no competing financial interest.