cRGD Peptide-Conjugated Icosahedral closo-B₁₂ ²⁻ Core Carrying Multiple Gd³⁺-DOTA Chelates for α_vβ₃ Integrin-Targeted Tumor Imaging (MRI)

Abstract

A vertex-differentiated icosahedral closo-B₁₂²⁻ core was utilized to construct a α_vβ₃ integrin receptor-targeted (via cRGD peptide) high payload MRI contrast agent (CA-12) carrying 11 copies of Gd³⁺-DOTA chelates attached to the closo-B₁₂²⁻ surface via suitable linkers. The resulting polyfunctional MRI contrast agent possessed a higher relaxivity value per-Gd compared to Omniscan, a small molecular contrast agent commonly used in clinical settings. The α_vβ₃ integrin receptor specificity of CA-12 was confirmed via in vitro cellular binding experiments and in vivo MRI of mice bearing human PC-3 prostate cancer xenografts. Integrin α_vβ₃-positive MDA-MB-231 cells exhibited 300% higher uptake of CA-12 than α_vβ₃-negative T47D cells. Serial T1-weighted MRI showed superior contrast enhancement of tumors by CA-12 compared to both a non-targeted 12-fold Gd³⁺-DOTA closomer control (CA-7) and Omniscan. Contrast enhancement by CA-12 persisted for 4 h post-injection, and subsequent enhancement of kidney tissue indicated a renal elimination route similar to Omniscan. No toxic effects of CA-12 were apparent in any mice for up to 24 h post-injection. Post-mortem ICP-OES analysis at 24 hours detected no residual Gd in any of the tissue samples analyzed.

INTRODUCTION

Magnetic resonance imaging (MRI) is one of the most powerful diagnostic modalities in clinical medicine and biomedical research. The resolution and sensitivity of MRI is further enhanced through the administration of a contrast agent (CA). Due to their positive contrast enhancement properties, paramagnetic gadolinium-based small molecular CAs have been used most widely in clinical settings.¹ A significant drawback of these CAs is their lack of specificity and failure to provide sufficient contrast at low concentrations. With regard to cancer detection, targeting of MRI CA to cellular markers specifically expressed by tumor cells would greatly facilitate diagnosis, monitoring of disease progression, and assessment of treatment success. Research toward the development of targeted CAs has been ongoing, with most efforts directed toward conjugation of a CA to specific antibodies or a peptide.^2,3

Efforts have also been directed at producing tissue specific MRI CAs.4 In particular, the α_vβ₃ integrin receptor has been extensively investigated as a molecular target due to its involvement in angiogenesis, proliferation, and metastasis of several tumor types.⁵ Integrin α_vβ₃ has high expression on proliferating endothelial cells during tumor angiogenesis and metastasis in contrast to resting cells.^6,7 Since the α_vβ₃ integrin receptors binds to certain extracellular matrix proteins via the peptide sequence Arg-Gly-Asp (RGD), diverse RGD peptide ligands have been investigated as a targeting agent for α_vβ₃ integrin.⁸ In particular, the cyclic RGD peptide (cRGD) has a relatively high and specific affinity for α_vβ₃ integrin receptors. The synthesis of several small molecular MRI CAs possessing a cRGD peptide conjugated to a Gd³⁺-chelate have been reported in the literature.⁵

However, small molecular CAs are in practice unsuitable for targeted imaging applications because of the high tissue concentrations of Gd³⁺ ions necessary to obtain sufficient contrast. Target receptors associated with particular tumor types or tissue pathologies are typically expressed at very low levels. In the case of α_vβ₃ integrin, the α_v and β₃ subunits number only 3×10³ to 1.4×10⁴ and 5.3×10² to 1.1×10⁴ copies per cell, respectively.⁹ In order to obtain sufficient contrast enhancement, the number of Gd³⁺ ions bound per target must be substantially increased.

The problem of insufficient Gd concentration in targeted imaging has been addressed by the use of macromolecules (dendrimers, liposomes, micelles, nanoparticles) to which large numbers of Gd³⁺-chelates are attached along with targeting moieties.¹⁰ Conjugation of Gd³⁺-chelates to macromolecules offers the additional advantage of increasing the rotational correlation time, resulting in improved relaxivity (r₁) and therefore increased contrast enhancement.¹¹ However, despite their satisfactory in vivo biodistribution profiles and high r₁ values, these macromolecular CAs suffer from the problem of polydispersity. Polydispersity has impact on the stability, solubility, appearance and texture of the macromolecule and thus can affect the overall efficacy of the drug or imaging agent.

Polyhedral boranes and carboranes have generated considerable interest in biomedical research because of their potential as monodispersed nanocarriers of diagnostic and therapeutic agents.¹² Hydroxylation of all 12 B-H vertices¹³ of icosahedral [closo-B₁₂H₁₂]²⁻ to [closo-B₁₂(OH)₁₂]²⁻ permits the attachment of up to 12 copies of a desired functionality through suitable linkers even at zero generation.¹⁴ The 12 linker arms are in close proximity to one another, a configuration that makes these compounds ideally suited for the construction of monodispersed nanomolecular MRI CAs. In a companion paper, we reported the synthesis of the first such contrast agent, an icosahedral closo-B₁₂² scaffold possessing 12 Gd⁺³-chelates (Figure 1).¹⁵

Figure 1.

Schematic representation of a vertex-differentiated [closo-B₁₂]²⁻ core supporting a targeting moiety (cRGD) and 11 copies of a MRI contrast agent.

Synthesis of a targeted nanomolecular CA requires the presence of heterobifunctionalized linker arms on the same scaffold to permit attachment of both the targeting moiety as well as multiple copies of Gd³⁺-chelates. We recently developed a synthetic methodology to differentiate a single B-OH vertex of [closo-B₁₂(OH)₁₂]²⁻ to provide [closo-B₁₂(OR)(OH)₁₁]²⁻. This methodology permitted the preparation of several vertex-differentiated closomers [closo-B₁₂(OR)(OR’)₁₁]²⁻ utilizing orthogonal chemical reactions.¹⁶

As proof of principle, we herein report the first application of this unique approach to synthesize a tumor-targeting nanomolecular MRI CA derived from a vertex-differentiated icosahedral closo-B₁₂²⁻ core. This closomer supports an α_vβ₃ integrin-targeting peptide (cRGD) covalently linked to one vertex, with the remaining 11 vertices carrying a high payload of Gd³⁺-chelates (Figure 1). Using a cell binding assay, we demonstrate the specificity of this targeted CA for α_vβ₃ integrin receptors. Also presented is a comparative analysis of the relaxivity and contrast enhancement kinetics of the targeted CA to a non-targeted 12-fold closomer CA and to the commercially available small molecular CA Omniscan.

RESULTS AND DISCUSSION

Abbreviations. N,N’-dimethylformamide (DMF), triethylamine (Et₃N), sodium sulfate (Na₂SO₄), sodium hydride (NaH), hexaethylene glycol (HEG), 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA), 1,4,7,10-tetraazacyclododecane-1,4,7-tris(t-butyl acetate) (DO3A-t-Buester), potassium hydrogen carbonate (KHCO₃) tetrahydrofuran (THF), dichloromethane (DCM), ethyl acetate (EtOAc ), methanesulfonyl chloride (MsCl) acetonitrile (ACN), ammonium chloride (NH₄Cl), methanol (MeOH), hydrochloric acid (HCl), N,N-diisopropylethylamine (DIPEA), room temperature (RT), diethylenetriaminepentaacetic acid (DTPA), diethylenetriaminetetraacetic acid (DTTA), trifluoroacetic acid (TFA), tetrabutyl ammonium (TBA), phosphate-buffered saline (PBS), molecular weight cut-off (MWCO), inductively coupled plasma optical emission spectroscopy (ICP-OES), and hours (h).

Synthesis of DOTA ligands

For this study, DOTA was selected as the Gd³⁺-chelating ligand because of the higher thermodynamic stability of Gd³⁺-DOTA (log K_GdL~23) relative to Gd³⁺-DTPA (log K_GdL~22) or Gd³⁺-DTTA (log K_GdL~17–19) chelates.^1c An amino-terminated DOTA ligand (4) was prepared via a six-step synthesis starting from hexaethylene glycol (HEG) (1) (Scheme 1). First, a phthalimide-functionalized HEG linker (1) was synthesized from commercially available HEG via a two-step process that converted one of the two hydroxyl groups to a phthalimide group in 44% yield. The remaining hydroxyl group in 1 was converted to a mesylate (−OMs) group, which upon reaction with sodium iodide, gave iodo-terminated HEG linker 2 in quantitative yield. The reaction of 2 with DO3A-t-Bu-ester produced DOTA derivative 3 in 96% yield. Deprotection of the phthalimide group on 3 using hydrazine gave the desired amino-terminated HEG-linked DOTA derivative 4 in 96% yield. The structures of compounds 1–4 were characterized by IR, NMR, and HRMS spectroscopy.

Scheme 1.

Synthesis of amino-terminated hexethylene glycol (HEG) linked DOTA ligand 4.

Reagents and conditions: (a) 1.0 eq. MsCl, Et₃N, DCM, 0 °C-RT, 3 h; (b) Potassium phthalimide, DMF, 110 °C, 15 h, 44% in two steps; (c) 1.5 eq. MsCl, Et₃N, DCM, 0 °C-RT, 3 h; (d) NaI, acetone, 65 °C, 15 h, 95% in two steps; (e) DO3A-t-Bu-ester, KHCO₃, DMF, 12 h, RT, 96%; (f) Hydrazine, ethanol, 65 °C, 4 h, 96%.

Synthesis of 12-fold Gd³⁺-DOTA-conjugated closomer CA-7

The preparation of a closomer conjugated to 12 Gd⁺³-DOTA chelates was accomplished in a stepwise procedure that relied upon our recently reported synthesis of 12-fold carbonate closomers capable of reacting with primary amines to yield 12-fold carbamate closomers.^14a The carbonate closomer 5 was reacted with the amino-terminated DOTA ligand 4 to produce the 12-fold DOTA conjugate 6 in 80% yield (Scheme 2). Closomer 6 was purified by size-exclusion column chromatography on Lipophilic Sephadex LH-20 using MeOH as the eluent and characterized by IR, NMR, and HRMS spectroscopy. To remove the tert-butyl ester groups present on the DOTA ligands, closomer 6 was treated with 80% TFA in DCM. Complete deprotection was confirmed by the absence of a large singlet at δ 1.47 ppm from the ¹H NMR spectrum of 6. Gd³⁺ chelation was achieved by reacting deprotected closomer 6 with GdCl₃·6H₂O in citrate buffer at pH ~7. Following extensive dialysis and lyophilization, the 12-fold Gd³⁺-DOTA-conjugated closomer CA-7 was obtained in 71% yield.

Scheme 2.

Synthesis of 12-fold Gd³⁺-DOTA closomer conjugate CA-7.

Reagents and conditions: (a) 4, ACN, 7 days, RT, 76%; (b) 80% TFA-DCM, 6 h, RT; (c) GdCl₃·6H₂O, citrate buffer, pH 7, 24 h, RT, sonication; (d) Dialysis, 1,000 MWCO, 48 h; (e) Lyophilization, 82%.

CA-7 was characterized by IR spectroscopy and ICP-OES analysis. The IR spectrum of CA-7 exhibited the characteristic shift of the carbonyl stretch from 1725 cm⁻¹ to 1605 cm⁻¹ indicative of complexation of Gd³⁺ with the DOTA ligands (Supporting Information). The gadolinium/boron content of CA-7 was determined using ICP-OES analysis. CA-7 contained an average of 11.4 Gd³⁺ ions per closomer, indicating essentially fully loaded DOTA ligands. Mass spectrometry analysis of CA-7 was unsuccessful; we could not obtain the desired mass peaks possibly due to the production of multiple charged species during the ionization step. The presence of free Gd³⁺ ions in the solution of closomer CA-7 was determined by a spectrophotometric method using xylenol orange.¹⁷ No free Gd³⁺ ions were detected. Dynamic light-scattering (DLS) analysis of CA-7 in PBS and 2% Tween-80/PBS solution gave an average particle size of 10 nm (Supporting Information).

Synthesis of targeted 11-fold Gd³⁺-DOTA conjugated closomer CA-12

The synthesis of a targeted MRI CA assembled on an icosahedral closo-B₁₂²⁻ core is based on our recently communicated work regarding preparation of vertex-differentiated closomers.¹⁶ Attachment of the targeting peptide cRGD was accomplished by an azide-alkyne click reaction of alkyne-terminated cRGD peptide derivative 8¹⁸ with the 11-fold carbonate closomer 9,¹⁶ which possesses an azido-termined linker at a single vertex (Scheme 3). The resulting cRGD-conjugated closomer 10 was purified by size-exclusion column chromatography on Lipophilic Sephadex LH-20 using MeOH as the eluent. Purified closomer 10 was characterized by NMR and HRMS spectroscopy. The ¹H NMR spectral analysis of 10 showed a singlet at δ 7.94 ppm attributed to the single proton of the 1,4-triazole group, confirming attachment of the cRGD peptide.

Scheme 3.

Synthesis of 11-fold Gd⁺³-DOTA-cRGD closomer conjugate CA-12.

Reagents and conditions: (a) CuI, DIPEA, THF-ACN, 15 h, RT, 87%; (b) 4, ACN, 7 days, RT, 83%; (c) 80% TFA-DCM, 6 h, RT; (d) GdCl₃·6H₂O, citrate buffer, pH 7, 24 h, RT, sonication; (e) Dialysis, 1,000 MWCO, 48 h; (f) Lyophilization, 82%.

Addition of DOTA ligands to the remaining 11 vertices of cRGD-conjugated closomer 10 was accomplished by reaction with amino-terminated DOTA derivative 4 (5 equiv per vertex) in ACN for seven days at RT to obtain the 11-fold carbamate closomer 11 in 83% yield (Scheme 3). Closomer 11 was purified by size-exclusion column chromatography on Lipophilic Sephadex LH-20 using MeOH as the eluent and its structure verified by IR, NMR, and HRMS spectroscopy. Removal of the tert-butyl ester groups from 11 was achieved by treatment with 80% TFA in DCM. Complete deprotection of tert-butyl ester groups from 11 was confirmed by the absence of the large singlet at δ 1.50 ppm from its ¹H NMR spectrum.

Deprotected closomer 11 was reacted with GdCl₃·6H₂O (10 equiv per vertex) in citrate buffer at pH ~7 to give closomer CA-12 in 82% yield. CA-12 was purified by exhaustive dialysis in ultrapure water and was characterized by IR spectroscopy and ICP-OES analysis. IR spectral analysis showed a characteristic shift of the carbonyl stretch from 1730 cm⁻¹ to 1595 cm⁻¹, confirming complexation of Gd³⁺ with the DOTA ligands. The mass spectrometry analysis of CA-12 was unsuccessful; however, the ICP-OES analysis of CA-12 showed an average of 10.6 Gd³⁺ ions per closomer, indicating the formation of essentially fully loaded complex. The presence of any free Gd³⁺ ions in solution of targeted closomer CA-12 was determined by a spectrophotometric method using xylenol orange.¹⁷ No free Gd³⁺ ions were detected in the solution. DLS analysis of CA-12 in PBS and 2% Tween-80/PBS solution gave an average particle size of 12–13 nm (Supporting Information).

Relaxivity Studies

The per-Gd relaxivity (r₁) of CA-7, CA-12 and Omniscan at 25 °C and 7 Tesla were compared in three different formulations: PBS, 2% Tween/PBS, and bovine calf serum (Figure 2). The relaxivity of each contrast agent was relatively unaffected by formulation type, but the r₁ values of CA-7 and CA-12 were slightly higher than Omniscan (4.2 mM⁻¹s⁻¹ in PBS) across all formulations. Nonetheless, the r₁ values for CA-7 and CA-12 (e.g., 4.8 mM⁻¹s⁻¹ per Gd and 5.9 mM⁻¹s⁻¹ per Gd in PBS) fall well within the range of other macromolecular MRI CAs carrying multiple Gd³⁺ chelates that have one inner sphere water-exchange site.¹¹ Macromolecular CAs are known to exhibit higher r₁ values due to their larger size and the confinement of the Gd³⁺ ions in a sterically constrained space. However, the slightly higher r₁ values of CA-12 compared to CA-7 may result from slower rotation (τ_R) of the Gd³⁺ chelates due to the attachment of the cRGD peptide to the icosahedral closo-B₁₂²⁻ core.

Figure 2.

Relaxivity (r₁) profile of CA-7, CA-12, and Omniscan in three different formulations at 25 °C and 7 T.

Molecular relaxivity values for CA-7 and CA-12 were calculated based on the number of Gd³⁺ ions per closomer. CA-7 and CA-12 have an average 11.4 and 10.6 Gd³⁺ ions per closomer, respectively, giving molecular relaxivity values of 54.7 and 62.5 mM⁻¹s⁻¹ per closomer. The T1-weighted MRI images of CA-12, CA-7 and Omniscan at 7T presented in figure 3 shows greater contrast enhancement for CA-12 and CA-7 compared to Omniscan due to their higher relaxivity values.

Figure 3.

T1-weighted MRI images of CA-12, CA-7, and Omniscan at 1mM Gd concentration in PBS (25 °C and 7 T).

Cell Binding Studies

To establish the specificity of cRGD-closomer conjugate CA-12 for α_vβ₃ integrin receptors, in vitro binding experiments were performed using MDA-MB-231 and T47D cells. MDA-MB-231 cells express high levels of α_vβ₃ integrins, (α_vβ₃ positive) whereas T47D cells lack α_vβ₃ integrin receptors (α_vβ₃ negative).¹⁹ Cells were incubated with CA-12 for 30 min at 37 °C (Gd concentration 5 mM, PBS), and the gadolinium concentration in lysed cell pellets was measured using ICP-OES analysis.The Gd concentration was 300% higher in MDA-MB-231 cells relative to T47D cells (Figure 4), demonstrating the ability of cRGD-containing closomer conjugate CA-12 to target cells expressing α_vβ₃ integrin receptors.

Figure 4.

Gadolinium concentration in lysed pellets of cells incubated with CA-12 (5 mM Gd, PBS) determined by ICP-OES.

In vivo MRI studies

The ability of CA-12 to specifically target α_vβ₃ integrin-expressing tumors in vivo was also investigated by MRI of severely compromised immunodeficient (SCID) mice bearing human PC-3 prostate cancer xenografts. Eighteen mice were subjected to MRI scanning following intravenous injection of either cRDG-containing CA-12, non-targeted closomer CA-7, or Omniscan at a gadolinium dose of 0.04 mmol/kg body weight. Six mice were included in each of the three experimental groups.

T1-weighted MRI scans showed that tumors of mice injected with targeted closomer CA-12 exhibited strong contrast enhancement for up to 4 h post-injection (Figure 5, panel A). Significant contrast enhancement by CA-7 persisted for only 1 hour post-injection (Figure 5, panel B). Contrast enhancement by Omniscan diminished within 30 min post-injection, and no enhancement was detectable by 4 h post-injection (Figure 5, panel C). MRI imaging also showed that contrast enhancement of the kidneys was higher and lasted longer in mice injected with CA-12 (Figure 5, panel A) compared to those injected with CA-7 or Omniscan (Figure 5, panels B and C), indicative of the relatively slower renal clearance of CA-12.

Figure 5.

Representative in vivo T1-weighted MRI scans of mice (n = 6 mice for each experimental group) pre- injection and at 1 min, 30 min, 1 h, 4 h and 24 h post-injection of (A) CA-12 (B) CA-7 and (C) Omniscan (gadolinium dose of 0.04 mmol/kg).

Calculation of the contrast enhancement ratios (CER) for tumor and kidney tissues permitted a quantitative assessment of the relative enhancement capacity of CA-12, the non-targeted closomer CA-7, and Omniscan. In tumor tissue, the CER did not decline significantly until 24 hours post-injection of CA-12, whereas it declined substantially by 4 h and 1 h post-injection of CA-7 and Omniscan, respectively (Figure 6). At 1 h and 4 h post injection of CA-12, tumor tissues had significantly higher CER compared to CA-7 (P < 0.01 at 1 h) and Omniscan (P < 0.01 at 1 h; P < 0.05 at 4 h). Similarly, the CER values for kidneys of mice injected with CA-12 and CA-7 initially increased then declined, though at a slower rate than for kidneys of mice injected with Omniscan (Figure 7). For example, the kidney signals are higher at 1 h post-injection of CA-12 compared to Omniscan (P < 0.01), and diminished to the baseline contrast at 24 h. The persistence of contrast in tumors and the slower renal clearance of CA-12 may be attributed to the specific binding to integrin receptors by the cRGD ligand.

Figure 6.

Contrast Enhancement Ratios (CER) for tumor tissues of mice as a function of time measured from the in vivo imaging experiments after intravenous injection of CA-12, CA-7, or Omniscan with a gadolinium dose of 0.04 mmol/kg. *P < 0.01, **P < 0.05.

Figure 7.

Contrast Enhancement Ratios or CER) for kidney tissues of mice as a function of time measured from the in vivo imaging experiments after intravenous injection of CA-12, CA-7, or Omniscan with a gadolinium dose of 0.04 mmol/kg. *P < 0.01.

After 24 h, all mice were euthanized, and tail, blood, heart, lungs, liver, spleen, stomach, large and small intestines, kidneys, brain, and muscle tissues were analyzed for the presence of gadolinium ions using ICP-OES. No gadolinium was detectable in any of the tissues, indicating complete elimination of the CAs.

CONCLUSIONS

We have shown for the first time the utility of a vertex-differentiated icosahedral closo-B₁₂²⁻ core in the synthesis of a novel α_vβ₃ integrin-targeted polyfunctional MRI CA. The closomer CA-12 carries a single cRGD peptide attached to the closo-B₁₂²⁻ core via a long PEG linker and holds 11 Gd³⁺-DOTA chelates attached through a HEG linker via carbamate linkages. This unique MRI CA exhibits a higher relaxivity per Gd than the small molecular CA Omniscan in PBS at 7T.

In vitro cell binding experiments and in vivo MRI analysis of tumor-bearing mice demonstrated the ability of CA-12 to selectively target integrin α_vβ₃-expressing cells. Binding studies showed that the Gd content was 3-fold higher in cells expressing α_vβ₃ receptors compared to cells in which receptor expression was absent. Serial T1-weighted MRI of mice possessing human PC-3 prostate cancer xenographs showed that higher and more persistent enhancement of tumors was achieved with CA-12 than with Omniscan or with CA-7, a closomer carrying 12 Gd³⁺-DOTA chelates but lacking the cRGD peptide. Mice exhibited no apparent toxic effects from any of the CA for 24 hours post-injection, and post-mortem ICP-OES analysis of numerous tissues found no detectable residual Gd.

The methodology presented here is versatile and should be highly useful in the development of a wide variety of high-payload, receptor-specific diagnostic and therapeutic agents. We are currently pursuing work in this direction.

EXPERIMENTAL SECTION

General information

Common reagents and chromatographic solvents were obtained from commercial suppliers (Sigma-Aldrich, Fisher Scientific, and Acros Organics) and used without any further purification. Lipophilic Sephadex LH-20 was obtained from GE Healthcare. DO3A-t- Bu-ester and cRGD peptide were purchased from Macrocyclics, Inc. and Peptides International, respectively. NMR spectra were recorded on Bruker Avance 400 and 500 MHz spectrometers. The high-resolution mass spectrometry analysis was performed using Applied Biosystems Mariner ESI-TOF. IR spectra were recorded on Thermo Nicolet FT-IR spectrometer. The dynamic light scattering analysis was performed on a Microtrac Zetatrac particle size analyzer. Gadolinium (Gd³⁺) concentrations of the samples were measured by inductively coupled plasma optical emission spectroscopy (ICP-OES) on a PerkinElmer OptimaTM 7000 DV instrument. All the Gd³⁺ and boron concentrations obtained via ICP analysis are the average of three measurements. The error bars provided in the figures are standard deviations.

Synthesis of 1

To a mixture of Et₃N (2.70 g, 26.6 mmol) and hexaethylene glycol (5.00 g, 17.7 mmol) in DCM (50 mL) stirring at 0 °C, a solution of MsCl (2.03 g, 17.7 mmol) in DCM (20 mL) was added over 1 h period by syringe pump. After addition of MsCl, the reaction was allowed to proceed at 0 °C for 1 h, and then at RT for 2 h. The reaction mixture was concentrated to dryness. The residue was dissolved in DCM (100 mL) and washed with 3% HCl-water (100 mL) and brine (100 mL). The organic layer was separated, dried over Na₂SO₄, then concentrated and dried under high vacuum. The dried sample was dissolved in DMF (30 mL) to which potassium phthalimide (4.26 g, 23.0 mmol) was added, and the mixture was stirred at 110 °C for 15 h. Following concentration under high vacuum to remove the DMF, the resulting solid was dissolved in Et₂O and filtered through a celite pad. The filtrate was concentrated, and the crude material was purified on a silica gel column to obtain pure product as a colorless oil. Yield: 3.2 g (44%). ¹H NMR (400 MHz, CDCl₃): δ 7.70 (dd, 2H, J = 2.8 & 5.2 Hz, −CH-Pth), 7.59 (dd, 2H, J = 2.8 & 5.2 Hz, −CH-Pth), 4.28 (brs, 1H, −OH), 3.77 (t, 2H, J = 5.6 Hz, −OCH₂HEG), 3.60 (m, 4H, −OCH₂HEG), 3.52-3.45 (m, 18H, −OCH₂HEG). ¹³C NMR (100.6 MHz, CDCl₃): δ 167.50, 161.98, 133.43, 133.03, 131.45, 122.53, 122.21, 72.03, 69.98-69.44 (multiple peaks), 67.21, 60.79, 36.68, 35.83, 30.73. HRMS (ESI): m/z calcd for C₂₀H₂₉NO₈ [M+Na]⁺ 434.1785; found 434.1575; for C₂₀H₂₉NO₈ [M+K]⁺ 450.1525; found 450.1038.

Synthesis of 2

To a mixture of Et₃N (1.10 g, 10.9 mmol) and 1 (3.00 g, 7.29 mmol) in DCM (30 mL) stirring at 0 °C, a solution of MsCl (1.08 g, 9.48 mmol) in DCM (10 mL) was added over 1 h by syringe pump. After addition, the reaction was allowed to proceed at 0 °C for 1 h and then at room temperature for 2 h. The reaction mixture was concentrated to dryness, and the residue was then dissolved in DCM (100 mL). After successive washing with 3% HCl-water (100 mL) and brine (100 mL), the organic layer was separated, dried over Na₂SO₄, then concentrated and dried under high vacuum. The dried sample was then dissolved in acetone, sodium iodide (4.37 g, 29.1 mmol) was added to it, and the resulting mixture was stirred at 65 °C for 15 h. The reaction mixture was then concentrated to dryness, DCM (100 mL) was added, and the solution filtered through a celite pad. The filtrate was concentrated and purified by silica gel chromatography to obtain pure product as a colorless oil. Yield: 3.6 g (95%). ¹H NMR (400 MHz, CDCl₃): δ 7.76 (dd, 2H, J = 3.2 & 5.6 Hz, −CH-Pth), 7.65 (dd, 2H, J = 2.8 & 5.2 Hz, −CHPth), 3.81 (t, 2H, J = 6.0 Hz, −OCH₂HEG), 3.67 (m, 4H, −OCH₂HEG), 3.59-3.50 (m, 16H, −OCH₂HEG), 3.18 (t, 2H, J = 7.2 Hz, −OCH₂HEG). ¹³C NMR (100.6 MHz, CDCl₃): δ 167.96, 133.77, 131.93, 123.02, 71.75, 70.45-69.92 (multiple peaks), 67.70, 37.10, 2.98. HRMS (ESI): m/z calcd for C₂₀H₂₈INO₇ [M+Na]⁺ 544.0803; found 544.0283.

Synthesis of 3

To a mixture of DO3A-t-Bu-ester (2.00 g, 3.88 mmol) and 2 (3.04 g, 5.83 mmol) in DMF (25 mL), KHCO₃ (0.78 g, 7.77 mmol) was added, and the mixture was stirred at 60 °C for 15 h. The reaction mixture was cooled and DCM (100 mL) added. The mixture was washed with brine (100 mL), and the organic layer was separated, dried over Na₂SO₄, and concentrated. After column chromatography over alumina (IV), the pure product was obtained as a viscous oil. Yield: 3.4 g (96%). ¹H NMR (400 MHz, CDCl₃): δ 7.63-7.60 (m, 4H, −CH-Pth), 3.68 (t, 2H, J = 5.6 Hz, −OCH₂HEG), 3.55 (t, 2H, J = 5.6 Hz, −OCH₂HEG), 3.45-3.36 (m, 18H, −OCH₂HEG), 3.10-2.00 (m, 24H, −OCH₂HEG & −CH₂DOTA), 1.28 (m, 27H, −^tBut-DOTA). ¹³C NMR (100.6 MHz, CDCl₃): δ 171.95, 167.61, 133.69, 131.41, 122.60, 81.63, 69.69-69.12 (multiple peaks), 67.24, 66.57, 55.81, 55.06, 53.37, 51.60, 49.49, 49.17, 36.65, 27.48. HRMS (ESI): m/z calcd for C₄₆H₇₇N₅O₁₃ [M+Na]⁺ 930.5410; found 930.5362.

Synthesis of 4

A mixture of 3 (4.4 g, 4.84 mmol) and hydrazine (1.55 g, 48.4 mmol) in EtOH (60 mL) was stirred at 65 °C for 3 h. Upon reaction completion, the reaction mixture was concentrated to dryness. The residue was dissolved in DCM (100 mL) and filtered using a sintered glass funnel. The filtrate was concentrated and dried under high vacuum. Following column chromatography of the crude product over alumina (IV), pure product was obtained as a viscous oil. Yield: 3.6 g (96%). ¹H NMR (400 MHz, CDCl₃): δ 3.24-3.16 (m, 20H), 2.88 (m, 2H), 2.78-2.59 (m, 4H), 2.56-1.83 (m, 20H), 1.07 (m, 27H, −^tBut-DOTA). ¹³C NMR (100.6 MHz, CDCl₃): δ 171.68, 171.40, 81.01, 80.98, 71.35, 69.35-69.04 (multiple peaks), 68.74, 68.55, 66.10, 55.28, 54.53, 53.11, 51.10, 49.47, 48.67, 40.35, 27.09, 26.83. HRMS (ESI): m/z calcd for C₃₈H₇₅N₅O₁₁ [M+Na]⁺ 800.5355; found 800.5723.

Synthesis of closomer 6

A mixture of 12-fold 4-nitrophenyl carbonate closomer 5 (0.20 g, 0.07 mmol) and DOTA ligand 4 (3.33 g, 4.28 mmol) in ACN (30 mL) was stirred for 7 days at RT under an argon atmosphere. The reaction mixture was then concentrated and purified via size-exclusion chromatography over Lipophilic Sephadex LH-20 using MeOH as the eluent. The product was obtained as a sticky, light-brown solid. M.P.: 65–68 °C. Yield: 0.60 g (80%). ¹H NMR (400 MHz, CD₃CN): δ 6.57 (brs, 10H), 3.79-2.13 (m, 554H), 1.47 (m, 324H). ¹³C NMR (100.6 MHz, CD₃CN): δ 173.8, 173.66, 171.50, 171.25, 82.74, 81.75, 71.02-70.35 (multiple peaks), 68.1, 57.19, 56.97, 56.31, 55.92, 55.28, 53.62, 53.17, 51.15, 50.66, 41.80, 28.34, 28.26, 28.14. ¹¹B NMR (160.4 MHz, CD₃CN): δ −18.19. HRMS (m/z): Average mass calcd. for C₄₆₈H₈₈₈B₁₂N₆₀O₁₅₆ [M+6H]⁴⁺+2K+Na+TBA 2838.7986. Found: 2838.5745, calcd. for C₄₆₈H₈₈₈B₁₂N₆₀O₁₅₆ [M+6H]⁴⁺+2K+2Na+TBA 2861.7986. Found 2861.6614.

Synthesis of 12-fold DOTA-Gd closomer CA-7

Closomer 6 (0.33 g, 0.03 mmol) was treated with 80% TFA/DCM (10 mL) for 6 h at RT under an argon atmosphere. The reaction mixture was then concentrated under reduced pressure. The resulting residue was dissolved in a minimum amount of methanol, and the product was precipitated by adding ether. The solid product was filtered, washed with ether, and dried under high vacuum overnight. The now deprotected closomer 6 was dissolved in 1 M citrate buffer (15 mL, pH 7) and added to a solution of GdCl₃.6H₂O (1.40 g, 3.78 mmol, in 15 mL of 1 M citrate buffer, pH 7) over 5 h at RT with vigorous stirring. The reaction mixture was maintained at approximately pH 7 using 0.3 N NaOH. When addition of 6 was complete, the reaction mixture was stirred for an additional 24 h at RT. To avoid possible aggregation and intermolecular chelation of Gd³⁺, the reaction mixture was sonicated a few times over the entire course of the reaction. To remove excess GdCl₃, the reaction mixture was dialyzed in deionized water for 48 h using 1000 MWCO dialysis tubing (Spectra/Por). Following lyophilization, the product CA-7 was obtained as an off-white solid. M.P.: 255–260 °C (decomposed). Yield: 0.22 g (71%). ICP-OES analysis: calcd. average Gd per closomer 12. Found: 11.4. The mass spectrometry analysis attempts were unsuccessful.

Synthesis of alkyne-terminated cRGD conjugate 8

Alkyne-terminated cRGD conjugate 8 was synthesized according to our recently published procedure.¹⁹

Synthesis of 11-fold 4-nitrophenyl carbonate closomer 9

Closomer 9 was synthesized according to our recently described method.¹⁶

Synthesis of closomer 10

Eleven-fold 4-nitrophenyl carbonate closomer 9 (0.03 g, 9.90 × 10⁻³ mmol), alkyne-terminated cRGD conjugate 8 (10.3 mg, 9.90 × 10⁻³ mmol), and copper (I) iodide (2.80 mg, 14.7 × 10⁻³ mmol) were dissolved in a 1:1 mixture of THF and ACN (6 mL). After addition of DIPEA (20 mg, excess), the reaction mixture was vigorously stirred at RT for 15 h under an argon atmosphere. Upon reaction completion, the reaction mixture was concentrated to remove THF and ACN. The residue was dissolved in ethyl acetate and filtered through a celite pad. The filtrate was concentrated, and the crude product was purified via size-exclusion column chromatography on Lipophilic Sephadex LH-20 using ACN as the eluent. The purified product was obtained as a sticky, light-yellow solid. Yield: 35.0 mg (87%). ¹H NMR (500 MHz, CD₃CN): δ 8.72 (m, 22H), 7.94 (s, 1H), 7.39 (m, 24H), 7.13 (m, 2H), 6.90 (m, 1H), 4.65-4.58 (m, 4H), 4.21 (s, 2H), 3.99-3.91 (m, 6H), 3.62 (m, 76H), 3.16 (m, 13H), 2.30 (m, 6H), 2.06 (m, 8H), 1.94 (m, 2H), 1.68 (m, 10H), 1.44-1.31 (m, 15H), 1.16 (m, 2H), 1.05 (m, 14H) (Fewer proton counts for TBA⁺ is due to exchange with H₃O⁺ and Na⁺). ¹³C NMR (125.7 MHz, CD₃CN): δ 156.63, 150.96, 150.69, 145.12, 127.63, 125.09, 122.31, 70.13-68.93 (multiple peaks), 63.78, 58.33, 46.93, 23.30, 19.33, 12.80, 8.23. ¹¹B NMR (160.4 MHz, CD₃CN): δ −14.34, −18.12. HRMS (m/z): Average mass calcd. for C₁₄₀H₁₅₀B₁₂N₂₂O₈₁ [M]²⁻ 1783.4705. Found: 1783.4463.

Synthesis of closomer 11

A mixture of closomer 10 (0.03 g, 7.40 × 10⁻³ mmol) and DOTA ligand 4 (0.32 g, 0.40 mmol) in ACN (10 mL) was stirred for 7 days at RT under an argon atmosphere. The reaction mixture was concentrated and purified via size-exclusion chromatography over Lipophilic Sephadex LH-20 using MeOH as the eluent. The pure product was obtained as a sticky, light-brown solid. M.P.: 72–75 °C. Yield: 65.0 mg (83%). ¹H NMR (400 MHz, CD₃OD): δ 8.09 (s, 1H), 7.30-7.16 (m, 5H), 7.05 (m, 2H), 6.80 (m, 2H), 6.70 (m, 1H), 4.66 (m, 7H), 4.52-2.35 (m, 590H), 1.54-1.51 (m, 300H), 1.16 (m, 6H), 0.95 (m, 6H). ¹³C NMR (100.6 MHz, CD₃CN): δ 173.78, 173.66, 171.25, 156.05, 130.28, 128.82, 124.9, 82.74, 82.72, 81.85, 71.02-69.87 (multiple peaks), 68.20, 57.19 (multiple peaks), 56.31, 55.71, 55.15, 53.55-53.23 (multiple peaks), 50.66 (multiple peaks), 48.43, 41.80, 28.26 (multiple peaks). ¹¹B NMR (160.4 MHz, CD₃CN): δ −14.54, −18.32. HRMS (m/z): Average mass calcd. for C₄₉₂H₉₂₀B₁₂N₆₆O₁₆₉ [M+6H]⁴⁺ + Na 2671.647030. Found: 2670.9688.

Synthesis of cRGD conjugated 11-fold DOTA-Gd closomer CA-12

Closomer 11 (0.09 g, 8.40 × 10⁻³ mmol) was treated with 80% TFA/DCM (10 mL) for 6 h at RT under an argon atmosphere. The reaction mixture was concentrated under reduced pressure, and the residue was dissolved in a minimum amount of methanol. Ether was added to precipitate the solid product. The product was filtered, washed with additional ether, and dried under high vacuum overnight. The now-deprotected closomer 11 was dissolved in 1 M citrate buffer (15 mL, pH 7) and added to a solution of GdCl₃.6H₂O (0.34 g, 0.92 mmol, in 15 mL of 1 M citrate buffer, pH 7) over 5 h at RT with vigorous stirring. The reaction mixture was maintained at approximately pH 7 using 0.3 N NaOH. When addition of 11 was complete, the reaction mixture was stirred at RT for an additional 24 h. To avoid possible aggregation and intermolecular chelation, the reaction mixture was sonicated a few times over the entire course of the reaction. The reaction mixture was then dialyzed in deionized water for 48 h using 1000 MWCO dialysis tubing (Spectra/Por). Following lyophilization, the product CA-12 was obtained as an off-white solid. M.P.: 207–210 °C (decomposed). Yield: 63.0 mg (69%). ICP-OES analysis: calcd. average Gd per closomer 11. Found: 10.6. Attempts at mass spectrometry analysis were unsuccessful.

Relaxivity Determinations

Solutions of CA-7 and CA-12, and Omniscan were prepared in PBS, 2% Tween-80/PBS and bovine calf serum. T1 relaxation rates were determined for each of the three formulations of CA-12 and CA-7 at Gd³⁺ ion concentrations of 0.25, 0.5 and 1 mM. The Gd³⁺ ion concentration of Omniscan was 1 mM. The measurements were repeated on two or more independently prepared samples to ensure consistency. A buffer matched blank sample (0 mM) was also used in the relaxivity measurements of each sample.

Measurements were performed using a 7 Tesla Varian Unity Inova MRI system (Varian Inc./Agilent Technologies) at RT. A T1-weighted MRI pulse sequence was applied with TE = 15 ms, and TR = 500 ms, slice thickness = 1 mm, matrix = 256 × 256, and FOV = 30 × 30 mm. A series of inversion recovery (IR) spin-echo images were acquired using TR = 3 s, TE = 15 ms, and the following inversion delays: 0.082, 0.1, 0.12, 0.16, 0.24, 0.32, 0.64, 1.28, 2.56, 3.6 s. Water signal intensities were measured using VnmrJ2.1D software (Varian Inc. /Agilent Technologies, 2005). Relaxation rates were calculated using a three-parameter experiential recovery fitting in Origin 8.5.0 (OriginLab Corporation, 2012).

Cell Binding Assays

MDA-MB-231 and T47D cells were obtained from the American Type Culture Collection (Manassas, VA). Cell culture reagents were purchased from Fisher Scientific (Chicago, IL) or Sigma-Aldrich (St. Louis, MO). Cells were cultured in complete growth medium [DMEM/F12 containing 10% heat-inactivated fetal bovine serum (FBS; Hyclone Laboratories, Logan, UT, USA), 1% antibiotics] in a humidified incubator at 37 °C and 5% CO₂. When cells had reached approximately 70% confluency (3 days), they were treated with trypsin-EDTA solution and washed twice with cold PBS. The cells were counted using a hemacytometer and diluted with PBS to a concentration of 1 × 10⁷ cell/mL. Cells were incubated with CA-12 (Gd concentration 5 mM) in a 37 °C water bath for 30 min. After incubation, cells were centrifuged at 800×g at 4 °C for 4 min, and the supernatant removed. The pellet was washed twice with PBS, and cells were resuspended in 200 µL PBS. ICP-OES was used to measure the gadolinium concentration of the cells.

In vivo MRI Analysis

In vivo MRI studies were performed in severely compromised immunodeficient (SCID) mice bearing human PC-3 prostate cancer xenografts. Animal studies were conducted in accordance with the highest standards of care as outlined in the NIH’s guide for care and use of laboratory animals and in accordance with policy and procedures for animal research at the Harry S. Truman Memorial Veterans Hospital. Four-to-five-week-old SCID mice were obtained from Taconic (Germantown, NY). Mice were housed four animals per cage in sterile micro-isolator cages in a temperature- and humidity-controlled room with a 12-hour light/dark schedule. The animals were fed autoclaved rodent chow (Ralston Purina Company, St. Louis, MO) and provided water ad libitum. Mice were inoculated with human prostate cancer PC-3 tumor cells on the right flank. Eighteen mice, six mice in each experimental group, were used in MRI experiments between 3–4 weeks post inoculation of tumor cells. Average body weight of mice was 25–30 grams at the time of MRI.

Imaging was performed using a 7T Varian Unity Inova MRI equipped with a Millipede quadrature RF coil (40 mm ID). A T1-weighted multi-slice spin echo sequence was performed to record images pre- and post-injection of CAs. The imaging parameters were TE = 15 ms, TR = 500 ms, slice thickness = 1 mm, 17 slices, average = 4, matrix = 128 × 256, and FOV = 35 × 70 mm (coronal). Mice were anesthetized using 2.5% isoflurane in oxygen via a nose cone. Once pre-injection imaging had been completed, CA-12, CA-7, or Omniscan dissolved in 2% Tween-80/PBS was injected through a tail vein (gadolinium dose 0.04 mmol/kg body weight). Mice were imaged immediately after introduction of the CAand again at 30 min, 1, 4, and 24 h post-injection. A water tube placed under the animal was treated as a reference signal. Vitals were continually monitored, and mice were released to their cage after each imaging session in order to fully recover.

Image analysis and processing were performed with VnmrJ2.1D software (Varian Inc. /Agilent Technologies, 2005). Regions of interest (ROIs) were manually drawn on the tumor tissue, muscle near the tumor, kidney cortex, and liver for each time point. Signal intensity (SI) was measured as the mean of the intensity over the segmented ROIs. SIs were normalized to the SI of muscle with the assumption that signal enhancement of muscle tissue is zero at 4 h and 24 h post-injection of CA. The contrast enhancement ratio (CER) for a tissue was calculated according to the equation

• CER = (SI_post – SI_pre) / SI_pre × 100%.

To assess persistence of the CAs in tissues, mice were euthanized 24 hours post-injection. Tumor, blood, heart, lungs, liver, spleen, stomach, large intestine, small intestine, kidney, brain, and muscle samples were collected, weighed and immediately frozen at −20 °C. Prior to the ICP-OES analysis, tissue samples were allowed to warm up to room temperature, acid digested and the gadolinium and boron content of the tissues was determined.

Statistical analysis

Quantitative data were expressed as the mean ± standard deviation (SD). Means were compared by analysis of a student’s t-test using t-test function in Microsoft Excel (2007). P values of less than 0.05 were considered to be statistically significant.