Actinium-225 for Targeted α Therapy: Coordination Chemistry and Current Chelation Approaches

Abstract

The α-emitting radionuclide actinium-225 possesses nuclear properties that are highly promising for use in targeted α therapy (TAT), a therapeutic strategy that employs α particle emissions to destroy tumors. A key factor, however, that may hinder the clinical use of actinium-225 is the poor understanding of its coordination chemistry, which creates challenges for the development of suitable chelation strategies for this ion. In this article, we provide an overview of the known chemistry of actinium and a summary of the chelating agents that have been explored for use in actinium-225-based TAT. This overview provides a starting point for researchers in the field of TAT to gain an understanding of this valuable therapeutic radionuclide.

Introduction

Targeted radionuclide therapy (TRT) is a powerful cancer treatment strategy that employs high-affinity tumor-targeting vehicles, such as antibodies, peptides, or small molecules, to deliver particle-emitting radionuclides to cancer cells, where they deposit a lethal payload of ionizing radiation (Fig. 1).^1,2 As a noninvasive, highly potent therapy in which off-target damage to noncancerous tissues is minimized, TRT is a valuable addition to the conventional treatment approaches of surgery, chemotherapy, and external beam radiation. The clinical utility of TRT has been validated by the FDA approvals of ⁹⁰Y-ibritumomab tiuxetan (Zevalin^®) and ¹³¹I-tositumomab (Bexxar^®) in 2002 and 2003, respectively. These radioimmunotherapeutic agents consist of a CD20 antigen-targeting monoclonal antibody labeled with a β particle-emitting nuclide for the treatment of B-cell non-Hodgkin's lymphoma.³

FIG. 1.

Schematic diagram depicting the concept of targeted radionuclide therapy employing either alpha (α) or beta (β⁻) particles in a tumor-targeting construct. The high LET and short range of α particles make them highly desirable for use in cancer therapy. LET, linear energy transfer.

Although β-emitting radiopharmaceutical agents have demonstrable efficacy for the ablation of tumors, their utility may be limited for some forms of metastatic cancer due to their physical characteristics.⁴ Depending on their energy, β particles can travel up to a centimeter from the site of nuclear decay, which can result in the nonspecific irradiation of healthy tissues neighboring the targeted malignancy. β particles also possess low linear energy transfer (LET), or energy deposition per unit pathlength, which reduces their cytotoxic efficacy and necessitates the administration of high doses of radioactivity to achieve therapeutic benefits. Using β emitters, the targeting of individual cells, smaller tumor burdens, and micrometastases is impossible.

In addition to β particles, other forms of radioactive emissions may be leveraged for the destruction of tumor cells. In particular, α particles are promising alternatives to β particles (Fig. 1). In comparison to β particles, α particles travel much shorter distances in biological media and exhibit higher LET, making α-emitting radionuclides attractive for use in TRT.^4–6 The path length of an emitted α particle ranges from 40 to 100 μm, or a few cell diameters. Therefore, the radiotoxicity of α emitters is confined to the targeted site, giving rise to high specificity for tumor cells. Furthermore, α particles deposit a mean energy of 80 keV per μm traveled in biological tissue, a value that is several orders of magnitude larger than that for β particles. This high LET of α particles causes more ionization events per track, contributing to irreparable, lethal DNA double-strand breaks.⁷ It is estimated that as little as one transversal of the cell nucleus by an α particle can sterilize a tumor cell,⁸ signifying the high potency of these emissions. By contrast, hundreds to thousands of decay events are required to annihilate tumor cells using β particles.

Despite the unparalleled cytotoxicity of α particles, the development of α-emitting radiopharmaceuticals has been hampered by the limited supply and unresolved chemistry of α emitters, problems that are not present for conventional β-emitting nuclides. A major milestone for the use of α emitters in cancer therapy was achieved in 2013 when the first-in-class α-emitting therapeutic agent radium-223 dichloride (²²³RaCl₂, Xofigo^®) was brought to market.⁹ As a bone-targeting calcium mimic, the ²²³Ra²⁺ ion is administered in an unchelated form as the simple chloride salt for the treatment of bone metastases stemming from metastatic castration-resistant prostate cancer.^10,11 To be more generally utilized in targeted α therapy (TAT) applications^12–14 to treat a wide range of cancers in addition to bone metastases, however, α emitters must be conjugated to tumor-targeting scaffolds using bifunctional chelating agents,¹⁵ following a similar design strategy as that employed in the β-emitting radiopharmaceuticals Zevalin and Bexxar.

Among the α particle-emitting ions that possess properties suitable for use in TAT, actinium-225 (²²⁵Ac) is highly promising.^16–18 The radiological half-life of ²²⁵Ac is 9.92 d,¹⁹ which enables the distribution of this radionuclide to clinics far from the site of production. Furthermore, this long half-life is compatible for use with macromolecular targeting vectors, such as antibodies or nanoparticles, which exhibit extended circulation times in vivo. As it decays to stable ²⁰⁹Bi, ²²⁵Ac generates eight short-lived progeny, emitting a total of four high-energy α particles that deliver a lethal radiation dose to cancer cells (Fig. 2).²⁰ Notably, ²²⁵Ac is significantly more potent than its daughter nuclide, ²¹³Bi, in both in vitro and in vivo models.^21–23 The superior efficacy of ²²⁵Ac compared to ²¹³Bi is most likely due to its 313-fold longer half-life and 3 additional α particle emissions.

FIG. 2.

Decay chain of ²²⁵Ac.

Despite the favorable nuclear properties and high cytotoxicity of ²²⁵Ac for use in TAT, two key challenges have prevented ²²⁵Ac from reaching its full clinical potential. First, the quantities of ²²⁵Ac that can be produced via current strategies cannot support its large-scale clinical use. The primary source of ²²⁵Ac is from the decay of ²²⁹Th (t_1/2 = 7340 years), which originates from reactor-bred ²³³U that was generated in the United States during the 1960s.^24,25 This supply only amounts to ∼1.7 Ci per year, which is not sufficient for a global medical application of this nuclide. Alternative production strategies employing high-energy accelerators at government-run labs are currently under investigation and will most likely provide a solution to this problem within the coming years.^26–30 The second challenge that has hindered the implementation of ²²⁵Ac-TAT is the need for the stable retention of ²²⁵Ac and its progeny to targeting vectors in vivo. The loss of ²²⁵Ac or its daughters from the targeting vectors will give rise to nonspecific radiotoxic effects.

The challenge of stable retention stems both from a lack of suitable chelating agents for ²²⁵Ac and from the α recoil effect, a phenomenon based on conservation of momentum laws that occurs upon release of an α particle. As described below, significant strides forward have been made in the development of chelators that form both thermodynamically stable and kinetically inert complexes with ²²⁵Ac. The recoil effect, however, remains an insurmountable challenge with respect to conventional chelation strategies. The total amount of energy released upon α decay due to the daughter nuclide recoil exceeds that of 10,000 chemical bonds. As such, the daughter nuclides will always dissociate from the targeting construct upon their formation. Several approaches to contain ²²⁵Ac and its daughters in nanoparticles have been explored with varying degrees of success.^31–38 The use of cell-internalizing targeting vectors may provide a strategy for retaining ²²⁵Ac within cancer cells, where the daughter nuclides have limited mobility.²²

Given the great interest in and value of ²²⁵Ac as a therapeutic radionuclide in TAT, we present an overview of the chemistry of this element and the status of suitable chelating agents for the biological delivery of ²²⁵Ac. Although there have been recent efforts to use nanoparticle-based delivery devices for this promising nuclide, our discussion will be limited to conventional bifunctional chelating agents.

Chemistry of Actinium

Background

Actinium, the lightest element in the actinide series, was first discovered by André Louis Debierne in 1899, and then several years later by Friedrich Oskar Giesel.³⁹ Since then, a total of 32 isotopes of actinium, ranging from ²⁰⁵Ac to ²³⁶Ac, have been identified or produced synthetically, and characterized. Of these isotopes, only ²²⁷Ac and ²²⁸Ac are present in nature as part of the naturally occurring decay chains of ²³⁵U and ²³²Th, respectively.^40,41 Within recent years, there has been a resurgence of interest in actinium, which has primarily been motivated by its potential use in TAT. This surge of interest, however, has outpaced the understanding of the chemistry of this element, which remains poorly developed owing to the limited supply and high radioactivity of all its isotopes. This incomplete understanding of the fundamental properties of actinium has slowed the development of novel chelation scaffolds, which are critical for the use of this nuclide in TAT. In this section, we summarize the known chemical and structural properties of the actinium ion, compiled in Table 1, and describe how these features can be leveraged for ligand design.

Table 1.

Chemical and Structural Properties of Ac³⁺ and La³⁺

Propertya	Ac3+	La3+
6-Coordinate ionic radius (Å)	1.12	1.03
pK1h	9.4	8.5
η (eV)	14.4	15.4
EA	2.84	3.90
CA	0.28	0.379
IA	10.14	10.29
Hydration number	10.9 ± 0.5	9.2 ± 0.37
M–OH2O (Å)	2.59(3)b, 2.63(1)c	2.54(3)

^apK_1h, first hydrolysis constant; η, absolute chemical hardness; E_A, electrostatic contribution to the formation constants of Lewis acid-base complexes; C_A, covalent contribution to the formation constants of Lewis acid-base complexes; I_A, ionicity in bonding, E_A/C_A.

^bRef.⁷¹.

^cRef.⁷².

Aqueous chemistry

In aqueous solution, actinium exists most stably in the +3 oxidation state. There is some evidence, however, that the +2 oxidation state may also be accessible.⁴² For example, in a radiopolarographic reduction study, a reduction half-wave potential was observed for aqueous solutions of ²²⁵Ac³⁺. This potential shifted in the negative direction by ∼150 mV upon the addition of 18-crown-6. Based on these data, this half-wave potential was attributed to the formation of the +2 actinium ion.⁴³ Conversely, on the basis of poor extraction yields of ²²⁷Ac with sodium amalgam in aqueous sodium acetate, an extraction procedure that is effective for lanthanides with stable +2 oxidation states, it was concluded that a +2 actinium ion is unlikely.⁴⁴ Additionally, theoretical studies predict highly negative standard reduction potentials of −4.9 V⁴⁵ and −3.3 V⁴⁶ relative to the normal hydrogen electrode for the Ac³⁺/Ac²⁺ couple, suggesting that this reduction is impossible in aerobic aqueous media.

The Ac³⁺ ion possesses chemical properties that are similar to the lanthanide, Ln³⁺, ions. Specifically, lanthanum, La³⁺, is a suitable nonradioactive surrogate for the Ac³⁺ ion. Quantitatively, La³⁺ differs from Ac³⁺ by its smaller ionic radius. The six-coordinate ionic radius of La³⁺ is 1.03 Å, compared with 1.12 Å for Ac³⁺.⁴⁷ Notably, Ac³⁺ is the largest +3 ion in the Periodic Table. As a result of its large size, it has correspondingly low charge density, a feature that makes Ac³⁺ the most basic +3 ion. Hydrolysis of the Ac³⁺ ion is observed between pH 8.6 and 10.4.^48–50 A precise value for the first hydrolysis constant (pK_1h) of Ac³⁺ was recently determined.⁵¹ This constant expresses the ability of the Ac³⁺ ion to polarize a coordinated water molecule sufficiently to effect the release of a proton and formation of AcOH²⁺. A pK_1h of 9.4 ± 0.1 was measured using an ion-exchange method. By contrast, the first hydrolysis constant of La³⁺ is 8.5.⁴⁹ The lower pK_1h of La³⁺ arises from the higher charge density of this ion, which more strongly favors the coordination of water and the subsequent release of a proton at a more acidic pH. An interesting aspect of the high pK_1h value for Ac³⁺ is that it indicates that hydrolysis does not occur until the pH is greater than 9. Thus, for radiolabeling considerations, basic buffers might prove to give effective radiochemical yields because little formation of the inert AcOH²⁺ is expected.

Coordination chemistry

Metal-ligand bonding in actinium complexes is driven primarily by electrostatic interactions and steric constraints. Because the stability of electrostatic interactions scales as the ratio of charge over distance, the large ionic radius of the Ac³⁺ ion gives rise to the formation of kinetically labile complexes. Furthermore, the lack of significant ligand-field stabilization energy effects for this actinide ion afford a high degree of structural diversity in Ac-ligand complexes that is limited only by the coordinative saturation of the Ac³⁺ ion and ligand–ligand steric interactions. Due to its lack of polarizability, Ac³⁺ is classified as a “hard” Lewis acid according to the Hard and Soft Acids and Bases (HSAB) theory,⁵² and is therefore likewise predicted to prefer “hard,” nonpolarizable, electronegative Lewis bases such as anionic oxygen donors. The hard/soft acid-base properties of a specific ion can be quantified using the concept of absolute chemical hardness. The absolute chemical hardness (η) of an ion is given by η = (I − A)/2, where I is the ionization energy and A is the electron affinity of the species of interest.^53,54 Based on reported values for the third and fourth ionization energies of Ac,⁵⁵ the Ac³⁺ ion is characterized by an absolute chemical hardness of 14.4 eV. For comparison, the absolute chemical hardness of La³⁺ is 15.4 eV.^53,54 It should be noted that both values are intermediate on the whole spectrum of hard and soft ions. Appreciably soft ions, like Au⁺, Ag⁺, and Cu⁺, exhibit absolute chemical hardness values that range from 5.7 to 6.3eV, whereas conventional hard ions, like Sc³⁺ and Al³⁺, are characterized by absolute chemical hardness values of greater than 24eV.^53,54 Therefore, the hardness values for Ac³⁺ and La³⁺ confirm that these ions are most certainly not soft, but are not among the hardest ions in the Periodic Table either.

Another useful concept to think about regarding HSAB properties of an ion is its corresponding E and C values. These E and C values, originally developed by Drago,⁵⁶ are parameterized based on known bond dissociation enthalpies. The assumption in this model is that bond dissociation enthalpies arise from a combination of electrostatic interactions, as characterized by the constant E, and covalent interactions, as characterized by the constant C. This model was further expanded by Hancock and Marsicano^57,58 for predicting stability constants in aqueous solutions [Eq. (1)].

In Equation (1), E_A and C_A are constants derived from experimental stability constant data that reflect the tendency of a cation or Lewis acid to engage in meaningful electrostatic or covalent interactions, respectively. E_B and C_B are the analogous parameters for a given ligand or Lewis base. The final terms, D_A and D_B, account for unfavorable steric interactions between the Lewis acid and base. Generally, for large ions, such as the third row transition metals, lanthanides, and actinides, this steric term is zero. In addition to having useful predictive power for stability constants, these E and C parameters can also be used to estimate the ionicity, I_A, of a given Lewis acid. This ionicity is taken as the ratio of the electrostatic and the covalent contributions to bonding, I_A = E_A/C_A. The ionicity can be interpreted as another quantitative measurement of chemical hardness, with a larger degree of ionicity corresponding to harder ions.

Based on known stability constants of Ac³⁺ with fluoride^42,59 and hydroxide,^48–51 the E and C values for this cation are calculated to be 2.84 and 0.28, respectively. The resulting ionicity is 10.14. The I_A value for Ac³⁺ is very close to that of La³⁺ (I_A = 10.30),⁵⁸ reinforcing the chemical similarity between these two ions. The individual values of E (3.90) and C (0.379) for La³⁺ are slightly greater than those of Ac³⁺. The smaller ionic radius and larger charge density of La³⁺, compared with Ac³⁺, provides a rationale for the larger E value of this ion. The C values for these ions are similar, indicating that neither has a strong covalent component to their bonding interactions. Using Equation (1), the E and C values should be valuable for the prediction of Ac³⁺ stability constants with representative ligands, which may aid in ongoing ligand design strategies.

Experimental conditional stability constants have been measured for several simple complexes of Ac³⁺.^42,59–64 These measurements are typically carried out using liquid–liquid extraction techniques at a fixed acidic pH using extractants, such as di-(2-ethylhexyl)phosphoric acid or dinonylnaphthylsulfonic acid, in the organic phase. From these studies, several trends are apparent. First, Ac³⁺ forms monohalido complexes of decreasing stability as heavier halides are employed. The stability constant β₁, where β_n = [ML_n]/[M][L]ⁿ, is ∼500–1000 times larger for the formation of AcF²⁺ compared to the formation of either AcCl²⁺ or AcBr²⁺. This large thermodynamic preference for F⁻ is consistent with the HSAB properties of this ion, as discussed above. The absolute chemical hardness of F⁻ (η = 7.0) is significantly greater than that of Cl⁻ (η = 4.7) or Br⁻ (η = 4.2).⁵³ As such, the fluoride ion is predicted to coordinate more strongly to hard acids like Ac³⁺. The Ac–F complex (β₁ ≈ 529) is also more stable than Ac³⁺ complexes of NO₃⁻ (β₁ = 1.31), SO₄²⁻ (β₁ = 15.9–22.9), and H₂PO₄⁻ (β₁ = 38.8). For these latter three inorganic anions, the stability constants increase with the basicity of the anion. Multidentate organic ligands predictably give rise to more stable Ac³⁺ complexes. For example, the stability constant for the formation of the citrate complex of Ac³⁺ is 9.55 × 10⁶.⁴² The complex of Ac³⁺ and EDTA is even more stable (β₁ = 1.66 × 10¹⁴).⁶⁵ The greater complex stability can be attributed to the higher denticity of EDTA, a hexadentate ligand, compared with citrate, a tridentate ligand. Ac³⁺ stability constants are similar to those of La³⁺, but usually slightly lower.^{60–63,66,67} The smaller Ac³⁺ stability constants are a consequence of the large radius of this ion, which gives rise to weaker electrostatic interactions with the ligand donor atoms.

Structural chemistry

An understanding of the structural chemistry of Ac³⁺ also provides useful information about the coordination preferences of this ion, insight that is of value in the design of new chelation platforms. The structural characterization of the Ac³⁺ ion, however, has been hindered by the lack of long-lived isotopes. The longest-lived isotope, ²²⁷Ac, has a half-life of only 22 years. The short half-life, limited availability, and radiological hazards of this isotope have challenged characterization techniques, such as X-ray crystallography, which require macroscopic quantities. Furthermore, the absence of appreciable spectroscopic features of this 5f⁰ 6d⁰ ion limits the use of optical spectroscopy techniques.

A series of actinium complexes have been characterized by X-ray powder diffraction, a technique that can be applied on microcrystalline powder samples. The solid-state compounds AcF₃, AcCl₃, AcBr₃, AcOF, AcOCl, AcOBr, Ac₂O₃, Ac₂S₃, and AcPO₄·0.5H₂O were all characterized in a single study published nearly 70 years ago.⁶⁸ Likewise, Ac₂(C₂O₄)₃·10H₂O and AcH₃ were prepared and characterized based on their powder patterns.^69,70 For all of these compounds, the powder pattern was comparable to that obtained for the La³⁺ analogs, indicating that the Ac³⁺ structures are isomorphous to those of La³⁺. This result is consistent with the chemical similarity of these ions, as described above. Furthermore, although direct Ac–L bond distances could not be measured, the unit cells determined from the Ac³⁺ diffraction patterns were consistently larger than those of La³⁺, reflecting the larger ionic radius of Ac³⁺. The data afforded by X-ray powder diffraction, however, is not significantly meaningful with respect to determining precise coordination geometries about Ac³⁺ centers.

In 2016, the first Ac–L bond distances were measured by extended X-ray absorption fine-structure (EXAFS) spectroscopy.⁷¹ These experiments employed ²²⁷Ac in aqueous solutions, taking advantage of the high sensitivity of the fluorescence detection mode of this spectroscopic technique. A solution of Ac³⁺ in aqueous HCl revealed photoelectron backscattering components that were assigned to inner-sphere Ac–O_H2O and Ac–Cl interactions. The interatomic distances were measured to be 2.59(3) and 2.95(3) Å for the Ac–O_H2O and Ac–Cl bonds, respectively. A subsequent follow-up study characterized the homoleptic Ac-aquo complex.⁷² The results from this EXAFS study reveal 10.9 ± 0.5 inner-sphere water molecules coordinated to the Ac³⁺ center. The average Ac–O_H2O bond length was measured to be 2.63(1) Å. By comparison, the La-aquo complex bears 9.2 ± 0.37 inner-sphere water molecules, with an average La–O_H2O bond length of 2.54(3) Å.⁷³ The higher hydration number and longer metal-oxygen bond length found in the Ac-aquo complex versus the La-aquo complex is a consequence of the larger size of the Ac³⁺ ion, which can accommodate more ligands in its inner coordination sphere. Furthermore, among homoleptic aquo complexes, the Ac-aquo complex is the only example of an 11-coordinate structure. These results highlight how Ac³⁺, as the largest +3 ion, may exhibit unexpected and divergent chemistry from the lanthanides and provide a promising first glimpse into the coordination chemistry of actinium. Ongoing EXAFS studies to characterize coordination complexes of Ac may reveal critical structural features that dictate complex stability.

Bifunctional Ligands for the Chelation of Actinium

Background

For the successful implementation of TAT with ²²⁵Ac, this radionuclide must be delivered with high specificity and retained within the vicinity of targeted cancer cells over the course of its nuclear decay. These conditions are best accomplished by covalently linking a tumor-targeting vector to a bifunctional chelator that forms a thermodynamically and kinetically stable complex with the ²²⁵Ac³⁺ ion. Unfortunately, the development of effective bifunctional chelating agents for ²²⁵Ac³⁺ has been hindered by the poorly understood coordination chemistry of this short-lived radioactive ion. This lack of knowledge regarding Ac³⁺ coordination chemistry makes it difficult to accurately foresee which ligands will form stable complexes in vitro and in vivo. Another challenge in the design of ligands for Ac³⁺ is the large ionic radius of this ion, which gives rise to a low charge-to-ionic radius ratio, a feature that leads to weaker electrostatic interactions with ligands.

In vivo, the instability of ²²⁵Ac-ligand complexes is reflected by the accumulation of free ²²⁵Ac in the liver and bone, where its radioactive emissions give rise to acute radiotoxic effects.^74,75 As such, the formation of kinetically inert complexes of ²²⁵Ac is a crucial prerequisite for the application of this radionuclide for TAT. In this section, we will describe research efforts to develop an effective bifunctional chelating agent for ²²⁵Ac. Specifically, the application of DOTA (Fig. 3) as the current state of the art for the chelation of the Ac³⁺ ion and the use of alternative chelating agents for this radionuclide are discussed.

FIG. 3.

Structures of DOTA and its bifunctional derivatives.

DOTA: the current gold standard

DOTA is a 12-membered macrocycle that provides octadentate coordination via 4 tertiary amine nitrogen donors and 4 carboxylic acid pendent arms (Fig. 3). This ligand is widely employed for the stable chelation of other tripositive radiometals such as ⁶⁸Ga³⁺, ¹¹¹In³⁺, ¹⁷⁷Lu³⁺, ^86/90Y³⁺, and ^44/47Sc³⁺,¹⁵ and is a critical component of FDA-approved peptide constructs for the diagnosis (⁶⁸Ga-DOTATATE) and treatment (¹⁷⁷Lu-DOTATATE) of somatostatin receptor-positive neuroendocrine tumors. With its established clinical efficacy that demonstrates its ability to stably coordinate chemically hard +3 ions, DOTA is expected to provide a suitable chelating scaffold for the Ac³⁺ ion.

In an initial study that examined the biodistribution of ²²⁵Ac-DOTA in normal BALB/c mice, the complex rapidly cleared from the blood; only a slight accumulation of activity was observed in the liver and bone of mice after 5 d (3.29%ID/g and 2.87%ID/g, respectively).⁷⁶ These results suggest that the ²²⁵Ac-DOTA complex is sufficiently stable in vivo, prompting further studies on the ²²⁵Ac-chelation efficacy of DOTA in tumor-targeting antibody constructs.²² Initial efforts to prepare ²²⁵Ac-DOTA-antibody constructs required a two-step procedure.⁷⁷ First, both p-SCN-Bn-DOTA and MeO-DOTA-NCS (Fig. 3) were radiolabeled with ²²⁵Ac at 55–60°C for 30 min in pH 5 acetate buffer. Under these conditions, no differences in radiolabeling efficiencies were observed between the two bifunctional DOTA ligands; both yielded the expected ²²⁵Ac-DOTA-NCS complexes, which were used interchangeably for the following bioconjugation step.

In the second step, ²²⁵Ac-DOTA-NCS was conjugated to a monoclonal antibody using the isothiocyanate (NCS) functional group, which reacts with lysine residues on the antibody to form stable thiourea linkages. This two-step procedure was necessary because temperature-sensitive antibodies cannot tolerate the 55°C–60°C temperature required to incorporate Ac³⁺ into the DOTA macrocycle. This process was successfully employed for the preparation of ²²⁵Ac-DOTA constructs with the antibodies HuM195 (anti-CD33), B4 (anti-CD19), trastuzumab (anti-HER2/neu), and J591 (anti-PSMA, prostate-specific membrane antigen), which target leukemia, lymphoma, breast/ovarian cancer, and prostate cancer, respectively.²² Greater than 95% of the ²²⁵Ac-DOTA-J591 construct remained intact after incubating in human serum for 15 d. Furthermore, all of the conjugates retained their biological activities, as they were readily internalized by cancer cells and gave rise to potent cytotoxic effects. The in vivo efficacies of ²²⁵Ac-DOTA-J591 and ²²⁵Ac-DOTA-B4 were evaluated in mice bearing either solid prostate carcinoma or disseminated human lymphoma, respectively. In both models, single doses of the ²²⁵Ac-labeled constructs led to tumor regression and prolonged survival of mice compared with untreated controls. From this pivotal study, ²²⁵Ac was established as a feasible radionuclide to deliver a series of lethal α particles to cancer cells, leading to its designation as an “atomic nanogenerator.”

The utility of DOTA as a bifunctional chelating agent for ²²⁵Ac-TAT constructs has subsequently been investigated in numerous preclinical studies that have shown promising results for the treatment of a range of different cancers.^78–86 Notably, ²²⁵Ac-DOTA constructs are currently being evaluated in several clinical trials for the treatment of leukemia, multiple myeloma, and prostate cancer.^87–90 These trials, which have demonstrated the incredible therapeutic potential of this radionuclide in humans (Fig. 4), have garnered great interest for ²²⁵Ac-TAT.

FIG. 4.

⁶⁸Ga-PSMA-11 PET/CT scans of a patient with metastatic castration-resistant prostate cancer before (A) and after (B, C) receiving several cycles of ²²⁵Ac-PSMA-617, a small-molecule construct. Complete imaging response and a remarkable reduction in prostate-specific antigen, a biomarker for prostate cancer, were achieved after the final treatment, demonstrating the extraordinary potential of ²²⁵Ac for TAT. This figure was reprinted with permission from ref. 89. Copyright © 2016 by the Society of Nuclear Medicine and Molecular Imaging, Inc. TAT, targeted α therapy.

Despite these advances in the field of ²²⁵Ac-TAT, the ²²⁵Ac-chelation properties of DOTA are not optimal for use in this application. Thermodynamically, the stability of metal ion complexes of DOTA is inversely related to the ionic radius of the metal ion, with larger metal centers giving rise to less stable complexes.^91,92 The thermodynamic preference of DOTA for smaller ions puts Ac³⁺ at a notable disadvantage, given its status as the largest +3 ion. In addition to thermodynamic considerations, the kinetic inertness of radiometal-ligand complexes is another important factor that dictates the suitability of chelating agents for TAT. Radiopharmaceuticals are administered in very low doses and are thus subject to highly dilute conditions in vivo, circumstances that enhance the off-rate kinetics of a radiometal-ligand complex. In this context, several studies have called into question the kinetic stability of ²²⁵Ac-DOTA constructs, as they report the loss of ²²⁵Ac from DOTA both in vitro⁷⁷ and in vivo.⁷⁶

In addition to the off-rate kinetics, the on-rate kinetics, which dictate the rate at which radiolabeling occurs, also present a major challenge to the widespread use of DOTA in the clinic for ²²⁵Ac-TAT. Complete radiolabeling of DOTA conjugates with ²²⁵Ac³⁺ in short periods of time requires the application of high temperatures.^{22,77,78,81,83,93} As such, the two-step labeling procedure,⁷⁷ as described above, has been applied for labeling antibodies with ²²⁵Ac to avoid directly subjecting them to temperatures above 37°C that lead to their denaturation.

Recently, a one-step method has been developed for the formation of ²²⁵Ac-DOTA-antibody constructs in which DOTA-antibody conjugates are directly radiolabeled with ²²⁵Ac at 37°C for 2 h.⁸⁴ This procedure offers several key advantages over the two-step method. For example, 10-fold higher radiochemical yields and 30-fold higher specific activities are observed for the one-step compared to the two-step radiolabeling methodology. Ideally, however, rapid radiolabeling (<20 min) at room temperature would greatly facilitate the use of ²²⁵Ac in the clinical setting and minimize radiolytic damage to sensitive antibody-based targeting vectors, ultimately improving specific activity while maintaining immunoreactivity. Aside from the 2 h one-step labeling of DOTA with ²²⁵Ac just described, no other more rapid or mild radiolabeling conditions have been discovered for this ligand. Collectively, these shortcomings indicate that DOTA is not ideal for use in ²²⁵Ac-TAT applications, highlighting the need for more suitable chelating scaffolds for ²²⁵Ac.

Aminocarboxylate and aminophosphonate ligands

Because of the limitations of DOTA, alternative ligands have been explored for the chelation of the ²²⁵Ac³⁺ ion for TAT. Early efforts to discover effective ²²⁵Ac-chelating agents were directed toward the evaluation of linear polyaminocarboxylate and polyaminophosphonate ligands, and other low molecular weight compounds, such as citrate.^75,94–96 Acyclic ligands are appealing for use in radiopharmaceuticals because they generally exhibit rapid radiolabeling kinetics, which permits incorporation of radiometals at room temperature within minutes.¹⁵ The investigation of this class of ligands with ²²⁵Ac, however, revealed that they did not form kinetically inert complexes. For example, the biodistribution of ²²⁵Ac complexes of EDTA, a hexadentate ligand, and CHX-A″-DTPA, an octadenate ligand (Fig. 5), in mice revealed significant accumulation of radioactivity in the liver and bone, consistent with the biodistribution pattern of unchelated Ac³⁺.^75,76 Notably, the Ac³⁺ complex of EDTA is ∼75% less stable than that of CHX-A″-DTPA,⁷⁵ a property that may be attributed to the fewer number of ligand donor atoms available for EDTA to coordinate the Ac³⁺ ion. Moreover, the preorganized diamino cyclohexyl group in the backbone of CHX-A″-DTPA may further increase the stability of the ²²⁵Ac complex, compared to the nonrigid ethylenediamine backbone of EDTA.

FIG. 5.

Structures of acyclic and macrocyclic aminocarboxylate and aminophosphonate ligands investigated for the chelation of ²²⁵Ac.

Additional preorganization in the form of macrocycles (the macrocyclic effect) significantly improves the stability of ²²⁵Ac-ligand complexes in mice.⁷⁶ For example, the ligand HEHA (Fig. 5), which is structurally related to DOTA by virtue of their polyaza macrocyclic cores that bear pendent carboxylate arms, was also found to be effective for Ac³⁺ chelation. HEHA differs from DOTA in the size of its macrocyclic cavity and in the number of nitrogen and oxygen donor atoms that it possesses; DOTA (N₄O₄) is a 12-membered macrocycle and HEHA (N₆O₆) is an 18-membered macrocycle. As noted above, for the DOTA complex, a slight accumulation of ²²⁵Ac was observed in the liver and bone of mice after 5 d (3.29%ID/g and 2.87%ID/g, respectively). This liver and bone accumulation, however, was dramatically less than that seen for the acyclic chelators EDTA (85.94%ID/g and 10.3%ID/g, respectively) and CHX-A″-DTPA (23.9%ID/g and 4.18%ID/g, respectively) over the same period of time.^75,76

Notably, the HEHA complex of Ac³⁺ is slightly more stable in vivo compared to that of DOTA; only a negligible quantity (<0.3%ID/g) of radioactivity was detected in any organ after 5 d.⁷⁶ These results suggest that the larger 18-membered macrocyclic cavity of HEHA may be better matched to the large size of the Ac³⁺ ion than the 12-membered macrocyclic cavity of DOTA, giving rise to a more stable complex with ²²⁵Ac³⁺. Another contributing factor to the slightly higher in vivo stability of ²²⁵Ac-HEHA may be the greater denticity of this ligand compared to DOTA. The EXAFS studies of the Ac-aquo complex, described above, indicate that the Ac³⁺ ion can accommodate very high coordination numbers.⁷² Hence, the 12 donor atoms of HEHA may more effectively saturate the coordination sphere of the Ac³⁺ ion compared to the 8 donor atoms of DOTA.

The bifunctional analog of HEHA, HEHA-NCS (Fig. 5), was subsequently prepared and conjugated to several different monoclonal antibodies. These HEHA-antibody conjugates were radiolabeled with ²²⁵Ac and evaluated for stability.⁹⁷ When tested in fetal bovine serum, all of the ²²⁵Ac-HEHA conjugates were unstable, as reflected by their >50% decomposition after 24 h. This instability was attributed to a combination of radiolytic decomposition of the ²²⁵Ac-HEHA conjugates and transchelation of ²²⁵Ac³⁺ by serum proteins. An ²²⁵Ac-HEHA conjugate of the thrombomodulin-binding monoclonal antibody 201B, which targets lung vasculature, was subsequently prepared and evaluated.⁹⁸ In normal mice treated with this construct, significant accumulation of ²²⁵Ac was observed in the liver and bone after 6 d, signifying the loss of this radionuclide from the HEHA conjugate. For mice bearing lung tumor colonies, the construct was effective in destroying the cancer cells, but also gave rise to severe radiotoxicity. This toxicity was attributed to the loss of ²²⁵Ac from HEHA due to both the instability of the chelate and the release of daughter nuclides arising from the α recoil effect. Although these results highlight the lack of long-term in vivo stability of Ac-HEHA complexes, it should be noted that the thrombomodulin-binding antibody does not enter cells.

Therefore, the use of cell-internalizing targeting vectors may minimize the redistribution of ²²⁵Ac and its daughters, avoiding the associated downstream toxic side-effects. Furthermore, the significant difference in stability between the unfunctionalized ²²⁵Ac-HEHA complex and the antibody-conjugated ²²⁵Ac-HEHA complex underscores the importance of evaluating long-term stability of chelating agents with targeting vectors that will prolong in vivo circulation time.

In addition to DOTA and HEHA, several other macrocycles have been investigated for their Ac-chelation properties (Fig. 5). PEPA is a 15-membered polyaza macrocycle that provides decadentate coordination via 5 tertiary amine nitrogen donors and 5 carboxylic acid pendent arms. Despite possessing a macrocyclic core that is intermediate in size between DOTA and HEHA, PEPA was not able to form complexes that stably retained ²²⁵Ac in vivo.⁷⁵ The striking difference between the stability of ²²⁵Ac complexes of DOTA and HEHA versus PEPA is not readily explained based on a comparison of the ligand structures, emphasizing the critical need to develop a better understanding of the coordination chemistry of actinium.

The macrocycles TETA, TETPA, and DOTPA (Fig. 5) did not strongly bind ²²⁵Ac in initial radiolabeling experiments and were therefore not investigated further.⁷⁷ TETA and TETPA are based on a cyclam macrocyclic core, which forms two 6-membered chelate rings upon metal coordination. The DOTPA ligand, which is based on a cyclen macrocycle like DOTA, bears four pendent carboxylate donors that are each linked to the macrocycle via a two-carbon chain. As such, these pendent donors form six-membered chelate rings upon metal coordination. By contrast, DOTA and HEHA, ligands that form complexes of higher stability with ²²⁵Ac³⁺, only generate five-membered chelate rings upon metal ion coordination. As previously documented, large ions form thermodynamically more stable complexes with ligands that give rise to smaller chelate rings.⁹⁹ Thus, for Ac³⁺, the largest +3 ion in the Periodic Table, there is a strong aversion to the larger six-membered chelate rings afforded by TETA, TETPA, and DOTPA. DOTMP (Fig. 5), an analog of DOTA in which the carboxylic acid pendent arms are replaced by phosphonic acid pendent arms, binds more strongly to ²²⁵Ac than does TETA, TETPA, and DOTPA, but rapidly releases this radionuclide in human serum.⁷⁷ The drop in kinetic stability in moving from DOTA to DOTMP can be ascribed to the lower basicity of the phosphonate donors in the latter, a property that can often be correlated to ligand-metal donor strength.

Calixarenes and texaphyrins

Another class of macrocycles, calixarenes, have also been investigated as chelators for ²²⁵Ac. A tert-butyl-calix[4]arene possessing four acetic acid arms appended to the phenolic groups of the core scaffold (Fig. 6) complexes ²²⁵Ac effectively, as demonstrated by its ability to nearly quantitatively extract this radionuclide from the aqueous phase in liquid–liquid extractions.¹⁰⁰ Additionally, the resulting ²²⁵Ac-calix[4]arene complex remained intact over the course of 5 h in the presence of high concentrations of competing alkali and alkaline earth metal ions. Based on these promising studies, a derivative of this calix[4]arene tetracarboxylic acid bearing an amine-reactive isothiocyanate was prepared.¹⁰¹ The further exploration of the utility of this ligand as a bifunctional chelator for ²²⁵Ac, however, was not investigated. The synthesis of two other bifunctional calix[4]arene tetracarboxylic acid-based ligands have also been reported.¹⁰² One of these ligands was successfully conjugated to human serum albumin and to the monoclonal antibody 506-A, which targets the pregnancy hormone hCG, but was ultimately not evaluated with ²²⁵Ac.¹⁰² Overall, these studies suggest that calix[4]arene-based scaffolds may be useful for ²²⁵Ac chelation, but further investigations are needed to fully explore the potential of these ligands.

FIG. 6.

Structures of t-Bu-calix[4]arene-tetracarboxylic acid, motexafin, and L^py.

Aromatic or pi-conjugated macrocycles may also possess suitable properties for chelating radiometals. Specifically, the class of expanded porphyrins, known as texaphyrins, bind strongly to large f-element ions and exhibit tumor-selective uptake properties.¹⁰³ With respect to Ac³⁺ chelation, motexafin (Fig. 6), a therapeutically relevant texaphyrin, was predicted by density functional theory calculations to form highly stable complexes with this large ion.¹⁰⁴ Experimental studies, however, were not able to confirm this prediction. Under high-temperature radiolabeling conditions (90°C), no ²²⁵Ac was incorporated into the texaphyrin, indicating that the N₅ donor system of this ligand is not effective for the chelation of the Ac³⁺ ion.¹⁰⁵ By contrast, its daughter nuclide ²¹³Bi was rapidly complexed by this macrocycle, consistent with results demonstrating the formation of cold Bi-texaphyrin complexes.¹⁰⁶

The preferential binding of motexafin to the ²¹³Bi daughter nuclide of ²²⁵Ac is a phenomenon that was previously observed for another nitrogen-rich macrocyle, L^py (Fig. 6).¹⁰⁷ This ligand carries the same 12-membered cyclen core as DOTA but bears pendent nitrogen-donor picolyl ligands, thereby providing an N₈ coordination sphere. Because L^py is structurally analogous to DOTA in that it only forms five-membered chelate rings, the large preference of this ligand for Bi³⁺ over Ac³⁺ was postulated to arise as a consequence of the nitrogen-rich coordination sphere provided by this ligand. Namely, the Ac³⁺ ion has a poor affinity for nitrogen donor atoms compared with the Bi³⁺ ion, which has a high affinity for these ligands. As such, these results highlight the need for the presence of chemically hard oxygen donor atoms in the development of ²²⁵Ac³⁺ chelators.

Recent advances in the development of ²²⁵Ac chelators

More recent studies have disclosed promising new candidates for ²²⁵Ac chelation. Bispa² is an octadentate ligand bearing two picolinic acid pendent arms attached to the tertiary amine nitrogen atoms of a bispidine scaffold (Fig. 7).¹⁰⁸ In radiolabeling experiments, bispa² (10⁻⁴ M) completely complexed ²²⁵Ac (40 kBq) at room temperature in under 30 min. Radiochemical yields decreased to 94%, 64%, and 2%, respectively, when lower ligand concentrations of 10⁻⁵, 10⁻⁶, and 10⁻⁸ M were used. Challenge experiments to assess the stability of the resulting ²²⁵Ac complex of bispa² were carried out in human serum or in an aqueous solution containing fivefold excess of La³⁺, with respect to ligand concentration, as a competing ion. Under these conditions, ∼90% and 72% of ²²⁵Ac-bispa² remained intact over the course of 7 d in human serum or in the presence of La³⁺, respectively. These preliminary studies point toward the potential utility of bispa² for ²²⁵Ac chelation; further in vivo studies employing biological targeting vectors are required, however, to establish the suitability of this ligand for ²²⁵Ac-TAT.

FIG. 7.

Ligands investigated recently for the chelation of ²²⁵Ac.

Ligands based on the diaza-18-crown-6 macrocycle have also been subjected to investigations for ²²⁵Ac chelation. For example, the ligand macropa (Fig. 7), which bears two pendent picolinate arms on the amine nitrogen atoms of the macrocyclic core, was probed for its ability to complex this large radiometal.¹⁰⁹ A unique feature of macropa, which prompted the studies with ²²⁵Ac, is that it forms more stable complexes with large over small lanthanide ions.¹¹⁰ Because Ac³⁺ has similar chemical properties as the lanthanides but is otherwise substantially larger, it was hypothesized that macropa would be a highly effective chelating agent for this ion. Radiolabeling of macropa with ²²⁵Ac (26 kBq) gave quantitative radiochemical yields after 5 min at room temperature with ligand concentrations as low as 0.59 μM. Furthermore, the stability of the ²²⁵Ac-macropa complex in an aqueous solution containing 50-fold excess of competing La³⁺ ions, with respect to ligand concentration, or in human serum was excellent; in both of these challenges, the ²²⁵Ac-macropa complex remained intact for over 7 d. Upon injection of the ²²⁵Ac-macropa complex into mice, no accumulation of activity in the liver or bone, signs of complex instability, was observed over the course of 5 h.

Based on these promising results, a bifunctional analog of macropa, macropa-NCS (Fig. 7), was synthesized, and it was conjugated to the antibody trastuzumab and to a small-molecule PSMA-targeting agent. The macropa-trastuzumab conjugate complexed ²²⁵Ac in quantitative radiochemical yield at room temperature, marking a distinct advantage over antibody conjugates of DOTA, which are unable to complex ²²⁵Ac under these conditions. The PSMA-targeting conjugate of macropa was similarly labeled with ²²⁵Ac under mild conditions. The ²²⁵Ac-labeled PSMA-targeting construct was administered to mice bearing prostate cancer xenografts. After 4 d, accumulation of ²²⁵Ac was only observed in the tumor, indicating that macropa can stably retain this nuclide in vivo over an extended period of time. Based on these results, macropa is a highly promising chelating agent for ²²⁵Ac that has key advantages over DOTA, which will favor its use for TAT.

Macropid (Fig. 7), another ligand based on the diaza-18-crown-6 macrocycle, was also evaluated for ²²⁵Ac chelation (Thiele NA, Brown V, Kelly JM, et al., Unpublished Results, 2017). This ligand, which bears two phenyliminodiacetate pendent arms, was shown to exhibit selectivity for large over small alkaline earth metal ions.¹¹¹ Macropid provides a total of 12 donor atoms, exceeding the 10 donor atoms provided by macropa. Attempts to radiolabel this ligand with ²²⁵Ac, however, were unsuccessful, even when high temperature conditions were applied. Likewise, EuK-106⁹³ (Fig. 7), a PSMA-targeting conjugate comprising a DOTA-like scaffold bearing additional carboxylic acid donor atoms,¹¹² failed to complex ²²⁵Ac under the high temperature conditions employed. Collectively, these results suggest that simply increasing the number of donor atoms in a ligand does not always increase the affinity for a large ion like Ac³⁺.

Conclusions

The use of ²²⁵Ac for TAT is an incredibly promising strategy for the development of new therapeutic radiopharmaceutical agents. Discussed above, a key challenge in the implementation of this nuclide in the clinic arises from its poorly understood coordination chemistry, which hinders the development of appropriate bifunctional chelating agents. In this review, we have highlighted the known coordination chemistry properties of this radioactive ion and summarized the chelation approaches that have been explored. A key feature of the Ac³⁺ ion that needs to be emphasized is its large ionic radius. The most successful ligands to date (DOTA, macropa) possess chelating scaffolds that can sufficiently accommodate the large size of this ion. The future development of ²²⁵Ac-based radiopharmaceutical agents will require the further exploration of these classes of ligands. Lastly, the rigorous EXAFS studies of this elusive ion that are ongoing will most certainly provide a greater insight into its coordination chemistry, bridging an important gap between our knowledge of the actinides and their practical utility.

Acknowledgments

The authors acknowledge funding support from a Pilot Award from the Weill Cornell Medical College Clinical and Translational Science Center, funded by NIH/NCATS UL1TR00457.

Disclosure Statement

N.A.T. and J.J.W. are inventors on a patent application 62/478,945, submitted by Cornell University, which covers a method for the use of the ligand macropa for actinium-225-based alpha therapy. This technology has been licensed to Noria Therapeutics, where J.J.W. is a member of the scientific advisory board.