Coordination Chemistry of Emerging Meitner-Auger Electron-Emitting Radiometals for Targeted Radionuclide Therapy

Abstract

Targeted radionuclide therapy (TRT) selectively delivers radionuclides to cancer cells, leveraging their radioactive emissions to treat primary and metastatic malignancies. Substantial fundamental research in this area has enabled the clinical development of a growing library of radiopharmaceuticals for this application, several of which have become blockbuster cancer drugs approved by the U.S. Food and Drug Administration. To date, these therapeutics have focused on radiometals emitting long-range β⁻ particles and shorter-ranged α particles. More recently, efforts to use a different type of cytotoxic radiation, Meitner-Auger electrons (MAEs), have been undertaken. MAEs are of interest for TRT because they travel very short distances—on the order of nanometers—that enable the possibility of single-cell and subcellular delivery of cytotoxic radiation. This high spatial precision is valuable for the treatment of diffuse malignancies such as lymphoma and leukemia, as well as for inducing unique cell death pathways with efficacy against conventional radiation resistance. As a number of emerging MAE-emitting radiometals are identified and produced, there is a need to develop suitable chelators that can form stable complexes with them so that they can be applied in targeted constructs. This review summarizes the coordination chemistry of emerging MAE-emitting radiometals, which span different regions of the Periodic Table, including the transition metals, main group elements, and lanthanides. In addition, approaches for their production and purification are also described. The major emphasis of this manuscript provides a summary of different chelators and chelation approaches for these novel MAE-emitting radiometals. This comprehensive summary of the coordination chemistry of these radiometals provides a guide for researchers in this area and emphasizes the chelator-design challenges that need to be overcome for their eventual clinical use.

Graphical Abstract

1. Introduction

Targeted radionuclide therapy (TRT) is a therapeutic approach that delivers a cytotoxic radioactive payload selectively to cancer cells [1–3]. This selectivity for cancer cells is afforded by either leveraging intrinsic physiological differences within the tumor microenvironment or by using specific targeting vectors, like antibodies and peptides, that bind to receptors overexpressed on cancer cells [2,4,5]. Early forms of TRT began in the 1940s when β⁻-emitting iodine-130 and iodine-131 were used for the treatment of thyroid disorders [6]. These treatments used simple iodide salts, leveraging the intrinsic uptake of this anion by the thyroid gland [7,8]. In modern iterations of TRT, the use of more selective, receptor-based targeting vectors, like small molecules, peptides, and antibodies, is preferred because it minimizes off-target effects and presents a more tractable approach for the delivery to different types of cancer. These constructs require that the radiometals of interest be complexed by a chelator, which serves as the attachment point to the targeting vector [9–11].

To date, several TRT drugs have been approved by the U.S Food and Drug Administration (FDA) for the treatment of metastatic castration-resistant prostate cancer (mCRPC) and gastroenteropancreatic neuroendocrine tumors (GEP-NETs) [4]. These drugs leverage α-emitting radium-223 (²²³Ra) as [²²³Ra]RaCl₂ (Xofigo^®, Bayer, treatment of mCRPC), β⁻-emitting yttrium-90 (⁹⁰Y) as [⁹⁰Y]Y-ibritumomab tiuxetan (Zevalin^®, Biogen Idec Inc.), and β⁻-emitting lutetium-177 (¹⁷⁷Lu) as [¹⁷⁷Lu]Lu-DOTATATE (Lutathera^®, Novartis, treatment of GEP-NETs) and [¹⁷⁷Lu]Lu-PSMA-617 (Pluvicto^™, Novartis, treatment of mCRPC). In addition to these three radiometals, a wide variety of others have been investigated for TRT applications in both clinical and preclinical studies [2,12]. In this review, we focus our discussion on an emerging class of radiometals for TRT that deviate from conventional therapeutic α and β⁻ emitters. Specifically, this review summarizes the coordination chemistry of radiometals that emit Meitner-Auger electrons (MAEs, Table 1). The following introductory sections highlight and introduce their therapeutic potential, and the majority of this manuscript presents a more comprehensive summary of their coordination chemistry and the strategies that have been investigated to chelate them for radiopharmaceutical applications.

Table 1.

Meitner-Auger electron-emitting radionuclides discussed in this review.

Radionuclide	Half-life	MAEs per decay	MAE energy (keV/nt)a	CEs per decay	CE energy (keV/nt)a	Gamma energiesb (keV)	Applications
58mCo	9.04 h	4.17	3.98	0.99	18.89	<60	Therapy
103mRh	0.94 h	5.88	2.72	0.99	34.97	<60	Therapy
119Sb	38.19 h	23.68	8.86	0.84	16.97	<60	Therapy
135La	19.50 h	12.30	6.57	0	-	-	Therapy
155Tb	5.32 d	13.95	12.59	0.77	30.82	87 (32.0%) 105 (25.2%) 180 (7.45 %)	Therapy, SPECT
161Tb	6.91 d	10.96	8.94	1.42	39.29	-	Therapy
165Er	10.36 h	7.28	7.99	0	-	-	Therapy, SPECTc
197mHg	23.80 h	19.37	13.50	1.60	203.50	134 (33.5%) 279 (6.1%)	Therapy, SPECT
197gHg	2.71 d	23.18	16.10	0.81	54.15	77 (18.7%)	Therapy, SPECT
191Pt	2.80 d	13.27	17.77	0.99	57.09	82 (4.9%) 172 (3.5%) 160 (6.0%)	Therapy, SPECT
193mpt	4.33 d	27.39	10.93	2.99	126.80	-	Therapy
195mpt	4.02 d	36.55	23.12	2.78	161.40	99 (11.4%)	Therapy, SPECT
201Tl	3.04 d	20.91	14.79	0.91	29.90	167 (10.0%)	Therapy, SPECT

^aTotal energy of emitted electrons.

^bOnly including gamma energies 60–400 keV useful for SPECT imaging.

^cSPECT possible from X-ray photons [136].

1.1. Radionuclides for TRT

In principle, TRT can use radionuclides that emit either α-particles, β⁻-particles, or low-energy MAEs or conversion electrons (CEs). These three types of particles have different energies and penetration ranges in tissue. Consequently, their linear energy transfer (LET) values, the amount of energy deposited per unit distance traveled, are widely different [9,13,14]. For therapeutically relevant α-emitting radionuclides, the average LET of the ejected α particles is 80 keV/μm, which results from emission energies ranging from 5–9 MeV and pathlengths of 40–100 μm (5–10 cell diameters). By contrast, β⁻-particle energies span a wider range of values (0.05–2.3 MeV), albeit with a lower magnitude, and can travel much larger distances (0.05–12 mm, 50–1000 cell diameters), resulting in a low LET of 0.2 keV/μm. The large penetration depths of β⁻ particles have found particular value for treating tumors with heterogeneous vasculature and intratumor variability [13,15]. Compared to both α and β⁻ particles, MAEs and CEs provide the most localized energy deposition. These low-energy electrons (1–10 keV) have a moderate LET (4–26 keV/μm), but their short path lengths (2–500 nm) relegate their energy deposition to the cellular and subcellular level [16–20].

In addition to emitting one or more of these three particles, radionuclides for TRT must also have appropriate half-lives (t_1/2). Generally, half-lives between six hours and seven days are ideal because they allow for transport and preparation of the radiopharmaceutical while maintaining sufficient activity [21,22]. The other important requirement for the application of radionuclides for TRT is a reliable and high-yielding production and purification route. These factors in the context of MAE-emitting radionuclides have been recently reviewed [23,24]. The final important consideration in the use of radionuclides for TRT is the availability of suitable chelators that can afford a kinetically stable radiometal complex under mild radiolabeling conditions [11,25,26]. This review focuses on chelator-design efforts for MAE-emitting radiometals.

1.2. Meitner-Auger Electrons

The unique physics and properties of low-energy electrons emitted during nuclear decay have received growing interest as therapeutic alternatives to α and β⁻ particles [19,27,28]. MAEs, first reported by Lise Meitner in 1922 and Pierre Auger in 1923 [29,30], are low-energy electrons (<25 keV) emitted from atoms in high-energy core-hole excited states, which can be generated via internal conversion (IC) or electron capture (EC or ε) radioactive decay [20,31,32]. Through IC, when a nucleus decays and transfers energy to an inner-shell electron (K-, L-, or M-shell), that electron, termed a conversion electron (CE), is ejected and leaves behind a core-hole vacancy. In the EC process, a proton-rich nucleus absorbs an electron from one of the core orbitals, resulting in the core-hole vacancy. In both cases, upon relaxation of this high-energy state, its energy can be transferred directly to another core electron, leading to its ejection (Figure 1).

Figure 1.

Schematic of Meitner-Auger electron emission process as mediated by both EC and IC decay. Created in BioRender.

The ejected electron, called a MAE, has an energy equal to the difference between the initial vacancy-filling transition and the ionization energy of the shell from which it originated [32]. When the ejected electron originates from a subshell higher than that of the electron involved in the relaxation process it is referred to as a Coster-Kronig electron. A special case of this transition occurs when the vacancy (e.g., in the M shell) is filled by an electron from a higher subshell of the same shell (e.g., also in the M shell). In this case, the transition and ejected electron are termed super Coster-Kronig [33]. However, Coster-Kronig and super Coster-Kronig electrons are subtypes of MAEs and are commonly referred to collectively as MAEs. Because the energies of these ejected electrons are related to specific core-shell excited states, MAE spectroscopy has become a powerful tool for characterizing the elemental composition of materials [34–36]. In the context of nuclear decay, the energies of emitted MAEs are characteristic of the radionuclide, and their low energy can be used for therapeutic applications [20].

Although this review primarily focuses on radionuclides which emit MAEs, CEs, which are ejected during the IC process (Figure 1), can also be leveraged for therapy. These electrons possess higher energies than MAEs (typically >100 keV) and also travel short distances in tissue. In many cases, MAEs and CEs are treated similarly and collectively referred to as low-energy electrons due to their similar radiobiological properties [13,20].

1.3. MAE Radiobiology

Despite carrying substantially less energy than β⁻ particles, the short penetration length of MAEs confers them with a substantially larger LET. Consequently, MAEs can induce DNA damage localized within single cells [19,20,28]. MAEs have also received attention for targeting the cell membrane and other subcellular components [18,37–39]. The very short range and potential to target individual cells makes MAEs particularly well suited for treating circulating tumor cells, which can be used to treat diffuse malignancies like lymphoma and leukemia, and to prevent metastasis [40–46].

DNA damage induced by MAEs can arise from both direct and indirect effects. Indirect DNA damage is mediated by the radiolysis of water, which generates a significant concentration of reactive oxygen species. Although most studies show that DNA damage is more substantial when MAEs are within close proximity to DNA [13,47,17,48–51], some reports suggest that reactive oxygen species, generated from MAE-induced radiolysis in the cytoplasm, can diffuse sufficient distances to damage critical targets, including DNA, leading to cell death [16,52]. Additionally, it is known that radiation damage to the mitochondria can cause mitochondrial apoptosis [53–55], and MAE emitters localized to the cell membrane can effectively kill cells [37–39].

The first investigations of the therapeutic effects of MAEs used radioisotopes of iodine. Both iodine-125 (¹²⁵I, t_1/2, 59.4 days, EC, 23.0 MAEs/decay, 11.96 keV) and iodine-123 (¹²³I, t_1/2, 13.3 h, EC, 1.37 MAEs/decay, 7.23 keV) were demonstrated to cause DNA damage, corresponding approximately to one double-strand break per decay, in bacteria and bacteriophages [56,57]. Accordingly, both radioisotopes could also induce cytotoxic effects in mammalian cells [58–62]. Subsequent studies showed that ¹²⁵I has promising anticancer properties in mice models [63–65], and ¹²⁵I-labeled 5-iodo-2′-deoxyuridine was used to treat neoplastic meningitis in a patient [66]. Although radioiodine has been invaluable in establishing MAE radiobiology, its radiochemistry typically relies on the formation of covalent C–I bonds, rather than chelation, and further discussion of radioiodine is outside the scope of this review [67,68].

1.4. Bifunctional Chelators

The majority of radioisotopes that possess promising properties for TRT belong to metal ions. Consequently, the formation of radiolabeled constructs with these radiometals requires the use of bifunctional chelators (BFCs). BFCs should ideally form highly stable radiometal complexes under mild conditions, while also providing an additional functional group for covalent attachment to different targeting vectors (Figure 2).

Figure 2.

Bifunctional chelators possess a reactive functional group to which a targeting vector can be conjugated prior to complexation with a radiometal. Created in BioRender.

The efficacy of a chelator or BFC with different radiometal ions can be characterized by both the resulting thermodynamic and kinetic stability of the radiolabeled complexes. The thermodynamic stability of a complex can be quantified via the stability constant, K_ML = [ML]/[M][L], where the concentrations are those of the complex ML, free ligand L, and free metal ion M under equilibrium conditions. This constant, often reported on a logarithmic scale (log K_ML), is a quantitative measure of the thermodynamic stability of a metal-ligand complex, and it can be measured with several different techniques, including potentiometric and spectrophotometric titrations [11]. These values provide a reliable and convenient metric for comparing the efficacies of different chelators with different metal ions. Although thermodynamic stability measurements are valuable, another important criterion for the use of a complex in vivo is kinetic stability or inertness. The inertness of a complex is more difficult to quantify, especially for comparing systems across different studies and labs, but it is often measured by competing ion and ligand challenge experiments or directly in biologically relevant solutions like saline or serum. Given the many different endogenous chelators, metal ions, and reducing agents in vivo, kinetic stability under relevant conditions must be considered when evaluating the suitability of a metal–ligand complex for therapeutic use.

Bifunctionalization of a chelator requires that it possesses a reactive functional group that can form covalent bonds with targeting vectors (Figure 2). Common functional groups for this application include activated carboxylic acids and isothiocyanates, which can couple to the N-terminus or lysine residues within a protein, furnishing amide and thiourea linkages, respectively [69,70]. Similarly, squaric acid diesters have become popular moieties for conjugating targeting vectors to BFCs because of their enhanced stability compared to isothiocyanates [71,72]. Copper-catalyzed click reactions (Huisgen 1,3-cycloaddition) between alkynes and azides, as well as strain-promoted Diels-Alder reactions, have also been used [73].

For detailed discussion of bioconjugation methods, the reader is referred to other reviews on that topic [10,69,70,74,75]. The goal of this review is to highlight the coordination chemistry and chelators for emergent, therapeutically relevant MAE-emitting radiometals. This review will focus primarily on radiometals that have not been extensively explored. In this context, the chelation chemistry of radioisotopes, like gallium-67 (t_1/2 3.3 d, EC, 4.96 MAEs/decay, 6.64 keV), has been thoroughly discussed previously, primarily because of the relevance of gallium-68 as a β⁺-emitter for PET imaging [76–81]. Likewise, another potentially valuable therapeutic MAE emitter, technetium-99m (^99mTc, t_1/2 6.0 h, isomeric transition (IT)/β⁻, 4.41 MAEs/decay, 0.94 keV) [53,55,82–87], has been thoroughly studied for its primary role as a single photon emission computed tomography (SPECT) imaging agent [76,88–100]. Lastly, the common SPECT radionuclide indium-111 (¹¹¹In, t_1/2 2.8 days, EC, 7.43 MAEs/decay, 6.88 keV) also has significant potential as a therapeutic MAE emitter, but its coordination chemistry and radiochemistry have been thoroughly covered in earlier literature [11,76,77,90,101]. Because of the numerous review articles on radioisotopes of gallium, technetium, and indium, these radiometals will not be discussed in this manuscript.

1.5. Nuclear Decay Data Sources

Nuclear decay data (half-lives, γ-photon energies and yields, and MAE and CE energies and yields) for all radionuclides discussed in this review (Table 1) were obtained from the International Commission on Radiological Protection DECDATA database compiled by Eckerman and Endo and the most recent corresponding Nuclear Data Sheets [102–135]. Reported MAE and CE values are given as the total energy per decay.

2. Cobalt

Cobalt-59 (⁵⁹Co) is the only naturally occurring stable isotope of cobalt. Several synthetic radioisotopes of cobalt are accessible, and among these, cobalt-58m (^58mCo, t_1/2 9.0 h, EC/IT) is of interest for MAE therapy. ^58mCo has a promising emission profile of both MAEs (4.17/decay, 3.98 keV) and CEs (0.99/decay, 18.89 keV), as well as a half-life that is suitable for biological applications. Two other cobalt radioisotopes, cobalt-55 (⁵⁵Co, t_1/2 17.5 h, EC/β⁺) and cobalt-57 (⁵⁷Co, t_1/2 271.7 days, EC) are also of interest for their use in PET and SPECT, respectively, providing the possibility of theranostic Co-based radiopharmaceuticals. The ^58mCo isotope decays to ground state cobalt-58g (^58gCo), which has a long but not negligible half-life (t_1/2 70.9 days) that must be considered for dosimetry [137]. Although the intermediate half-life of ^58gCo presents dosimetric complications, its β⁺ emissions (average energy 201 keV) are potentially useful for PET imaging, conferring ^58mCo with theranostic properties [138]. The non-radioactive, naturally occurring ⁵⁹Co is also a useful NMR nucleus ($I=\raisebox{1ex}{$7$}\!\left/ \!\raisebox{-1ex}{$2$}\right.$, γ = 10.1 MHz·T⁻¹). Its chemical shifts, which span a large range of 18,000 ppm, are highly sensitive to its electronic environment, as well as external factors like temperature [139–141]. Because of these properties, ⁵⁹Co-based magnetic imaging agents have been proposed and are under development [142–144]. ⁵⁹Co NMR spectroscopy will be discussed in Section 2.4 in the context of BFC complexation [145].

2.1. Production of Cobalt-58m

^58mCo can be produced by irradiating nickel targets with fast neutrons through the ⁵⁸Ni(n,p)^58mCo reaction, as well as by the proton irradiation of iron via the ⁵⁸Fe(p,n)^58mCo reaction. These reactions are limited by the natural abundance of nickel-58 (⁵⁸Ni, 68.1% abundance) and iron-58 (⁵⁸Fe, 0.28% abundance) [146]. An early study irradiated solid [^natNi]NiSO₄ with >0.5 MeV neutrons to produce ^58m/gCo [147]. This radioisotope was isolated after dissolving the target in acetic acid in the presence of carrier Co²⁺ ions, allowing its subsequent precipitation. This low-specific activity material was redissolved in HCl or HNO₃ and then used for radiolabeling. The relative amounts of ^58mCo and ^58gCo afforded from this reaction depend on the energy of the neutrons used for bombardment [148]. Because fast-neutron beams are rare, the ⁵⁸Fe(p,n)^58mCo production method, which uses more widely available proton cyclotrons, is more desirable.

This reaction was optimized by using enriched ⁵⁸Fe electroplated onto silver disks as the target, which was subjected to 1h irradiation with 10.1 MeV proton [137]. The target was then etched with dilute H₂O₂ in concentrated HCl and passed through an anion exchange column with HCl, affording ^58m/gCo with 97% radiochemical yield. With this method, only 0.85% of this activity of the radiocobalt arose from ^58gCo with the remainder comprising the desired metastable isomer ^58mCo. However, cadmium-107 (¹⁰⁷Cd) and cadmium-109 (¹⁰⁹Cd) were also produced as byproducts via the action of protons on the silver disks of the target materials.

Deuteron bombardment of enriched iron-57 (⁵⁷Fe) targets also produces ^58mCo through the ⁵⁷Fe(d,n)^58mCo reaction [138]. For this process, targets, prepared by electroplating enriched ⁵⁷Fe onto silver disks, were irradiated with 8.2 MeV deuterons for up to 2 h, after which nearly pure ^58mCo was isolated with less than 0.009% ^58gCo. After bombardment, the Fe layer was dissolved in HCl and treated with H₂O₂ to drive formation of [⁵⁷Fe]Fe³⁺. This solution was then diluted in ethanol, eluted sequentially through an anion exchange column and extraction resin, and dried. This process could also recover the costly, enriched ⁵⁷Fe from the anion exchange column with 94% efficiency after four production runs. A more streamlined approach for this radiochemical separation uses cation exchange instead of anion exchange resin, as this modification retains [^58m/gCo]Co²⁺ while [⁵⁷Fe]Fe³⁺ is eluted [149]. This approach is advantageous because it requires substantially less resin to capture the trace amounts of [^58m/gCo]Co²⁺ compared to the macroscopic quantities of the Fe target material that are retained on the anion exchange resin.

Most recently, optimization of the target material for the ⁵⁸Fe(p,n)^58mCo route has led to improved radiochemical isolation of ^58mCo [150]. In this study, a siphon-style liquid target comprising an aqueous solution of [^natFe]Fe(NO₃)₃ was irradiated with a 13 MeV proton beam. By using the liquid target, the radioisotopes of Co generated by this reaction could be isolated rapidly by circumventing the slow solid target dissolution step. This faster radiochemical separation is important because it allows for higher ratios of ^58mCo to ^58gCo to be obtained. Although this study used ^natFe (91.8% iron-56, ⁵⁶Fe), which predominately led to cobalt-56 (⁵⁶Co) via the ⁵⁶Fe(p,n)⁵⁶Co reaction, the targetry and radiochemical aspects are applicable for the production of ^58mCo.

2.2. Chemistry of Cobalt

Cobalt is a first-row transition metal with ground state electron configuration [Ar]3d⁷4s²; its coordination chemistry is dominated by the Co²⁺ ([Ar]3d⁷) and Co³⁺ ([Ar]3d⁶) oxidation states. Co²⁺ has a d⁷ configuration and forms complexes with coordination numbers (CNs) ranging from 3–8; four-coordinate tetrahedral and six-coordinate octahedral complexes are the most common [151]. Co²⁺ has an ionic radius of 0.65 Å (CN 4) and is a borderline Lewis acid (η = 8.22 eV, Table 2) according to hard-soft acid base theory (HSAB) [152,153]. Co³⁺ is a d⁶ ion, forms complexes with CNs from 4–6 with six-coordinate octahedral complexes being the most common [151], has an ionic radius of 0.625 Å (CN 6) [152], and is a hard Lewis acid (η = 8.9 eV) [153]. Compared to Co²⁺, Co³⁺ forms inert complexes, a feature that is conferred by the ligand field stabilization effects of a low-spin d⁶ ion. As previously mentioned, ⁵⁹Co is an NMR-active nucleus [140], and consequently ⁵⁹Co NMR spectroscopy has arisen as a valuable tool to probe the speciation and properties of diamagnetic Co³⁺ coordination complexes [154–157].

Table 2.

Chemical properties of metal ions discussed in this review.

Ion	Ionic radius (Å)a	Absolute hardness, η (eV)b
Co2+	0.65 (LS)	8.22
	0.745 (HS)
Co3+	0.545 (LS)	8.9
	0.61 (HS)
Rh3+	0.665	11.2
Sb3+	0.76	-
Sb5+	0.60	-
La3+	1.032	15.39
Tb3+	0.923	-
Er3+	0.89	-
Hg2+	1.02	7.7
Pt2+	0.80c	8.0
Tl+	1.50	7.16
Tl3+	0.885	10.4

High spin (HS). Low spin (HS).

^aSix-coordinate ionic radii from Ref. [152].

^bRef. [153].

^cFour-coordinate ionic radius from Ref. [152].

Coordination complexes of Co²⁺ and Co³⁺ have found use in various pharmaceutical applications, including antivirals and cancer treatments [158–165]. Significantly, a Co³⁺ Schiff-base complex Doxovir^™ is active against herpes simplex virus and has passed phase II clinical trials as a treatment for herpes simplex virus type 1 [166,167]. Although cobalt-60 has been used for external-beam radiotherapy since the 1950s, interest in cobalt radioisotopes for TRT has recently been renewed [168].

2.3. Acyclic Chelators for Cobalt

A recent study investigated the acyclic, amino-polypyridyl TZTPEN chelator (N¹-((triazol-4-yl)methyl)-N¹,N²,N²-tris(pyridin-2-ylmethyl)ethane-1,2-diamine, Figure 3) for radiolabeling with ⁵⁷Co as a surrogate for ^55/58mCo [169]. This chelator is an analogue of TPEN (N¹,N¹,N²,N²-tetrakis[(pyridin-2-yl)methyl]ethane-1,2-diamine, Figure 3), which forms stable complexes with Co²⁺ (log K_ML = 16.6) [170]. TZTPEN uses a click chemistry-derived triazole in place of one of the pyridyls of TPEN for Co²⁺ coordination [169]. The triazole was afforded by the conjugation of this chelator with GE11 peptide, which targets epidermal growth factor receptor (EGFR) [171]. This peptide-conjugate of TZTPEN was radiolabeled with [⁵⁷Co]CoCl₂ at pH 5.5 (using 10⁻⁵ M final chelator concentration) for 10 min at 60 °C with microwave irradiation in 5% DMSO. Under these conditions, the radiochemical conversion (RCC) was 99%.

Figure 3.

Acyclic TPEN and TZTPEN chelators radiolabeled with [⁵⁷Co]Co²⁺. The triazole moiety of TZTPEN is used for targeting vector conjugation.

Once formed, the stability of the [⁵⁷Co]Co-TZTPEN-GE11 complex in phosphate-buffered saline (pH 5) at 37 °C was investigated. The radio-high-performance liquid chromatography (radio-HPLC) chromatogram of this material revealed two [⁵⁷Co]Co-labeled complexes with different retention times (corresponding to 83 ± 11% and 17 ± 11% of the ⁵⁷Co activity). These two signals were attributed to complexes of cobalt in its +2 and +3 oxidation states, suggesting that oxidation occurred during radiolabeling. Based on the redox potential of the related complex [Co(TPEN)]^3+/2+ (0.497 V vs. NHE at pH 7) [172], the authors suggested that the majority of the radiolabeled complex contained [⁵⁷Co]Co²⁺ with a smaller amount attributed to the [⁵⁷Co]Co³⁺ species. This oxidation to Co³⁺ may be driven by radiolysis of water as was demonstrated earlier in [⁵⁵Co]Co²⁺ complexes and will be discussed in Section 2.4 [173]. Although the radiolabeled [⁵⁷Co]Co-TZTPEN-GE11 complex showed EGFR-specific uptake in vitro, in vivo mouse studies showed significant off-target accumulation, suggesting that the complex is not stable under these conditions [169].

2.4. Cyclic and Bifunctional Chelators for Cobalt

The coordination chemistry of Co²⁺ and Co³⁺ is extensive. However, there have only been few studies investigating radiometal chelation with radioisotopes of this metal like ⁵⁷Co, ⁵⁵Co, and ^58mCo. DOTA, a ubiquitous chelator for many different types of metal ions, has also proven to be effective for Co radioisotopes. For example, the BFC DOTATOC (Figure 4) could be radiolabeled with [⁵⁷Co]CoCl₂. A concentration of 350 μM DOTATOC incubated with 1.6 MBq [⁵⁷Co]CoCl₂ at pH 5.0 (NaOAc buffer) for 30 min at 95 °C afforded a RCC of 99.5% and an apparent molar activity (AMA) of 185 MBq/nmol [145]. To further study the coordination chemistry and stability of [⁵⁷Co]Co-DOTATOC, unfunctionalized DOTA and a simplified bioconjugate of DOTA containing a single amino acid residue, DOTA-D-PheNH₂ (Figure 4), were investigated. Starting from non-radioactive CoCl₂, [Co(H₂DOTA)] and [Co(DOTA-D-PheNH₂)]⁻ were synthesized. The absence of ⁵⁹Co NMR signals and the purple color of the complexes were attributed to the fact that these complexes contained paramagnetic Co²⁺ (t_2g⁵e_g², S = 3/2), which did not spontaneously oxidize under aerobic conditions.

Figure 4.

DOTA and its bifunctional derivatives DOTA-D-PheNH₂, DOTATOC, and DOTATE. The differences in DOTATOC and DOTATE targeting vectors are highlighted in blue.

This conclusion was further supported by a crystal structure of [Co(H₂DOTA)] (Figure 5), which revealed the pseudo-octahedral geometry. This structure shows that only two of the four carboxylate pendent arms bind directly to the Co²⁺ center. The equatorial Co–N and axial Co–O interatomic distances were very similar at 2.171 and 2.203 Å, respectively, corroborating the lack of a significant Jahn-Teller distortion and confirming the high-spin Co²⁺ state. In addition, the thermodynamic stability of [Co(DOTA-D-PheNH₂)]⁻ was measured via a ⁵⁷Co radiotracer method and found to be log K_ML = 19.6 (Table 3). In addition, the kinetic stability of [⁵⁷Co]Co-DOTATOC was assessed by challenging this complex with a 10⁴-fold excess of DTPA at pH 5.0, revealing the complex to remain >98% intact after 1 week. In human serum at 37 °C, this radiocomplex did show some degradation but with a long half-life of 800 h. Collectively, these results show that DOTA is an effective chelator for radiocobalt ions, even though it does not use all four of its pendent carboxylates for binding.

Figure 5.

Crystal structure of the [Co(H₂DOTA)] complex (CCDC Number 616730) [145].

Table 3.

Selected stability constants of Co–L complexes.

Ion	Chelator	logKML	Ref.
Co2+	EDTA	16.49	[178]
	DTPA	19.3	[179]
	TPEN	16.6	[170]
	DOTA	19.9	[145]
	DOTA-D-PheNH2	19.6	[145]

In a similar fashion, DOTA conjugated to L19K-FDNB, a peptide which binds covalently to vascular endothelial growth factor [174], is an effective chelator for [⁵⁷Co]Co²⁺ even though only three of its four pendants bind to the metal center [175]. Building upon these results, DOTATOC was radiolabeled with ^58mCo [176]. Incubating DOTATOC (76–127 μM final concentration) with [^58mCo]CoCl₂ at pH 4.5 at 80–85 °C for 30 min resulted in a RCC >99% with a moderate AMA of 4.1 MBq/nmol. The complex remained stable after radiolabeling for at least 45 h after radiolabeling, showing that it survives the decay of ^58mCo to ^58gCo, presumably due to the small recoil of this event. Another peptide conjugate of DOTA, DOTATATE (Figure 4), was radiolabeled with ^58mCo and ⁵⁷Co, as well as ¹¹¹In and ¹⁷⁷Lu, and these complexes were used to study the differential biological effects of these radionuclides [177]. Radiolabeling and stability of all complexes was confirmed, and these results were used to verify that [^58mCo]Co-DOTATATE is substantially more therapeutically effective than [¹¹¹In]In-DOTATATE and [¹⁷⁷Lu]Lu-DOTATATE in vitro, highlighting the promise of ^58mCo.

In addition to DOTA, the smaller macrocycle NOTA and its bifunctional derivative NOTA-TRC105 (Figure 6) have also been studied with ^58mCo [138]. In this study, NOTA was conjugated to an anti-CD105 antibody known as TRC105 [180]. CD105 is a surface glycoprotein commonly overexpressed in tumor vasculature and has been targeted by vectors such as TRC105 in various pharmaceuticals [181–185]. Treating 0–300 ng NOTA with [^58mCo]CoCl₂ (pH 7.5, 25 °C, 30 min) leads to successful radiolabeling, affording complexes with an AMA of 26 ± 7 MBq/nmol [138]. Building upon these results, a conjugate of NOTA with an anti-CD105 antibody known as TRC105 [180] was investigated. Using similar conditions (pH 7.5, 25 °C, 2 h), NOTA-TRC105 could be radiolabeled to furnish [^58mCo]Co-NOTA-TRC105 in 90 ± 2% radiochemical yield and an AMA of 1.2 MBq/μg after purification by HPLC. The [^58mCo]Co-NOTA-TRC105 complex was then used in an in vivo biodistribution study in mice bearing mammary carcinoma xenografts. Successful targeting of the tumor vasculature was observed via PET imaging of the ^58gCo daughter, providing further confirmation of the in vivo stability of the ^58mCo complex.

Figure 6.

NOTA and its bifunctional derivative conjugated to the TRC105 antibody through a thiourea linker.

BFCs constructed from NO2A, DO3A, and their analogues NODAGA and DOTAGA (Figure 7), were radiolabeled with [⁵⁵Co]Co²⁺ and [⁵⁷Co]Co²⁺ [186]. These chelators were conjugated to the RM26 peptide, an antagonist to gastrin-releasing peptide receptor overexpressed in prostate and breast cancers (Figure 7) and then treated with [⁵⁵Co]Co²⁺ or [⁵⁷Co]Co²⁺ (pH 5.5, 90 °C by microwave heating, 3 min) to afford complexes with RCCs of >98% for all four of the BFCs. Their kinetic stabilities were assessed by challenging them with a 1,000-fold excess of ethylenediaminetetraacetic acid (EDTA) at room temperature, revealing them to remain >99% intact after 1 h.

Figure 7.

The RM26 peptide is conjugated to NO2A and DO3A through a PEG₂ linker and to NODAGA and DOTATGA through the terminal amide of the pendant.

As noted in the TZTPEN-GE1 discussion in Section 2.3, radiocobalt complexes can exist in both the +2 and +3 oxidation states under the conditions used for radiolabeling [169]. In fact, the ability of cobalt to cycle between these two oxidation states under biological conditions has been leveraged to create oxygen-sensitive magnetic resonance imaging probes [187]. Unsurprisingly, the Co^2+/3+ redox chemistry should have important implications on radiopharmaceutical agents based on this metal ion.

This hypothesis was investigated in more detail via a series of three NOTA- and DOTA-based chelators that were studied with [⁵⁵Co]Co^2+/3+ [173]. Prior investigations involving these chelators demonstrated that their Co²⁺ complexes could be oxidized with oxygen or excess H₂O₂. This process could readily be interrogated via UV-vis spectroscopy that shows the diagnostic 520 nm d-d transition of Co³⁺ [145,187,188]. Accordingly, radiolabeling of NODAGA, NO2A, and DOTA with [⁵⁵Co]CoCl₂ in the presence of excess H₂O₂ afforded their oxidized [⁵⁵Co]Co³⁺ complexes, which was verified via HPLC retention time comparisons to their non-radioactive analogues. Notably, [⁵⁵Co][CoDOTA)]⁻ was found to be spontaneously reduced to the [⁵⁵Co][Co²⁺(DOTA)]²⁻ species, a property which could be attributed to its more positive redox potential [189,190].

Model BFCs of the three chelators were then prepared by linking them to a tryptophan (W) residue, furnishing NODAGA-W, NO2A-W, and DO3A-W (Figure 8). Tryptophan was chosen for the increased lipophilicity it confers to the complexes, its convenient spectroscopic absorbance, and its 1.07 V (vs. NHE) redox potential that is far from the electrochemically relevant window of Co^2+/3+ redox examined in this study. Although the cobalt complex of DO3A-W complex remained in the +2 oxidation state, the complexes of NO2A-W and NODAGA-W were spontaneously oxidized by air at a rate that correlated with their redox peak potentials. By comparison, the radioactive [⁵⁵Co]Co²⁺ complexes were oxidized to a much larger extent under similar conditions compared to their non-radioactive analogues. The authors posited that this effect was due to the radiolytic generation of ROS in solution.

Figure 8.

NODAGA, NO2A, and DO3A conjugated to tryptophan (W) by an amide bond to the −NH₂ group.

Lastly, the authors investigated these ligands in PSMA-617-conjugated derivatives. Specifically, the prostate-specific membrane antigen (PSMA) conjugates of DO3A and NOTA were investigated because these ligands stabilize Co²⁺ and Co³⁺, respectively, allowing the effect of metal oxidation state on biodistribution to be assessed. Both complexes showed good biodistribution with high tumor uptake. Furthermore, radio-HPLC metabolite analysis of urine showed that within the [⁵⁵Co][Co²⁺(DOTA-PSMA-617)]⁻ and [⁵⁵Co][Co³⁺(NODAGA-PSMA-617)] complexes, the Co²⁺ and Co³⁺ oxidation states were maintained in vivo. Thus, with appropriate ligands, either oxidation state of radiocobalt ions can be used for nuclear medicine.

In addition to the macrocyclic chelators discussed above, the bicyclic chelator diaminosarcophagine (DiAmSar, Figure 9) has also been radiolabeled with [⁵⁵Co]Co²⁺ [191]. DiAmSar was radiolabeled with [⁵⁵Co]Co²⁺ at pH 8 and elevated temperature (80 °C) for 4 h to give an AMA value of 45 ± 9 MBq/nmol [191]. Higher pH solutions afforded higher AMA values as a consequence of the lower extent of protonation of the ligand under these conditions. Mass spectrometry of the non-radioactive [^natCo]Co-DiAmSar complex revealed the Co ion to attain the +3 oxidation state, which results from exposure to oxygen. Stability studies revealed this complex to remain >95% intact in both human serum or in the presence of greater than 5 × 10⁴ excess of EDTA at 37 °C over 96 h. After demonstrating the suitability of DiAmSar, it was conjugated to the SR142948 neurotensin receptor antagonist for targeted applications (Figure 9) [192]. This conjugate was radiolabeled with [⁵⁵Co]Co²⁺ in a sodium gentisate solution, a known radioprotectant to help prevent radiolysis of the complex [193–195], giving 87% radiochemical yield after purification. In this case, the [⁵⁵Co][Co³⁺(DiAmSar-SR142948)]³⁺ complex was obtained, in part due to the radiolysis products of gentisate that can accelerate the oxidation of Co²⁺. Overall, this study established precedent for the DiAmSar chelator for use with radioisotopes of cobalt.

Figure 9.

Diaminosarcophagine (DiAmSar, R = H) and the SR142948 neurotensin receptor antagonist to which it is conjugated via an amide bond.

3. Rhodium

The rhodium-103m (^103mRh) radionuclide is the decay product of both ruthenium-103 (¹⁰³Ru, t_1/2 39.3 d, β⁻-decay) and palladium-103 (¹⁰³Pd, t_1/2 16.99 d, EC). As such, ^103mRh can be obtained from generators using either of these longer lived parent radioisotopes, as has been proposed since the 1970s [196,197]. ¹⁰³Ru can be obtained as a fission product of uranium-235 (²³⁵U) and can also be made in a reactor from enriched ruthenium-102 (¹⁰²Ru) through the ¹⁰²Ru(n,γ)¹⁰³Ru reaction. ¹⁰³Pd can be made either with reactor or cyclotrons, via the ¹⁰²Pd(n,γ)¹⁰³Pd, ¹⁰³Rh(p,n)¹⁰³Pd, and ¹⁰³Rh(d,2n)¹⁰³Pd reactions.

As of yet, no chelators have been explored specifically with ^103mRh. There have been a handful of studies, however, on chelation approaches for the therapeutic β⁻-emitter rhodium-105 (¹⁰⁵Rh, t_1/2 35.4 h, 1 β⁻/decay, 306 and 319 keV γ-photons), and these efforts are likely to be amenable for use with ^103mRh MAE therapy.

3.1. Production of Rhodium-103m

The neutron bombardment of ^natRu has been investigated using both Ru metal and RuCl₃ as targets. Their dissolutions in H₅IO₆ afforded [¹⁰³Ru]RuO₄, which was subsequently extracted into carbon tetrachloride and stored as a biphasic mixture in contact with aqueous H₅IO₆ and HCl to stabilize it. As the [¹⁰³Ru]RuO₄ decays, the [^103mRh]Rh³⁺ partitions from the organic to the aqueous layer [198,199]. For samples generated by the fission of ²³⁵U, a similar radiochemical strategy can be applied, but an initial precipitation step is first required to remove other fission products. The handling of ¹⁰³Ru-based generators is complicated by the intense and high-energy γ emission of ¹⁰³Ru—the most intense γ photon (0.91/decay) having an energy of 497 keV—which requires significant shielding that is often only afforded by hot-cells [198]. By contrast, ¹⁰³Pd-based generators circumvent this problem because the γ emissions of ¹⁰³Pd are less intense (most intense contributes 2.2×10⁻⁴/decay) and of lower energy (357 keV). To make ¹⁰³Pd, palladium targets enriched to 80% ¹⁰²Pd in the form of [¹⁰²Pd]PdCl₂ or [¹⁰²Pd]Pd(NO₃)₂ can be irradiated with thermal neutrons. The resulting [¹⁰³Pd]Pd²⁺ can be dissolved in water or low molar HNO₃ and then extracted into toluene with 1 mM acetylacetone. The [^103mRh]Rh³⁺ produced from the decay of [¹⁰³Pd][Pd(acac)₂] in toluene can then be extracted into aqueous HCl.

Alternatively, more clinically suitable, ¹⁰³Pd-based ^103mRh generators have more recently been developed that would enable ^103mRh elution at the point of care. One approach involves the use of reactor-produced [¹⁰³Pd]PdCl₂, which is chelated by a lipophilic 16aneS4 derivative (16aneS₄-C8, Figure 10) and then adsorbed onto a C18 column [200]. As ¹⁰³Pd decays on this column, the resulting [^103mRh]Rh³⁺ can be eluted with aqueous HCl. A similar generator can also be assembled by loading [¹⁰³Pd]Pd²⁺ onto an anion exchange column [201].

Figure 10.

Lipophilic 16aneS4 derivative (16aneS4-C8) used to immobilize [¹⁰³Pd]PdCl₂ on a C18 column for use as a [^103mRh]Rh³⁺ generator.

3.2. Chemistry of Rhodium

The coordination chemistry of rhodium is dominated by its stable +1 and +3 oxidation states. Mononuclear complexes containing this metal in the −1, 0, +2, +4, and +5 states have been reported but are rare [151,202]. Rh³⁺ ([Kr]4d⁶) is a d⁶ ion with a radius of 0.665 Å (CN 6) and is a borderline Lewis acid (η = 11.2 eV) [152,153], although arguments have been made that it is a soft acid [203]. Its hydrolysis constant (defined as log K but reported here as −log K for comparison to the pH scale) of 3.6 may require acidic media for radiolabeling [204]. It most commonly attains an octahedral geometry, affording low-spin (t_2g)⁶ complexes that are kinetically inert. By contrast, Rh⁺ is a softer d⁸ ion that forms square planar complexes. In studying the chemistry of Rh, the use of ¹⁰³Rh (I = ½, γ = −13.5 MHz·T⁻¹) for NMR spectroscopy is useful, especially when applied with two-dimensional heteronuclear multiple quantum coherence (HMQC) techniques [139,205].

3.3. Acyclic Chelators for Rhodium

Although chelators have not yet been investigated with ^103mRh, studies involving the β⁻ emitter ¹⁰⁵Rh serve as a logical starting point for developing radioconjugates of this MAE emitter. Some of the earliest examples of chelators radiolabeled with [¹⁰⁵Rh]Rh³⁺ include multidentate amines [206–208], oximes [209], a porphyrin [210], and cysteine [211]. These initial studies, however, required addition of macroscopic amounts of non-radioactive RhCl₃ [212]. No-carrier-added radiochemical syntheses were later performed using acyclic diaminedithioether chelators, which gave moderate RCCs of 46–61% [213]. Higher RCCs of >90% were obtained for tetrathioether chelators with terminal carboxylate groups. However, both these classes of chelators required high temperatures of 80–85 °C (refluxing 10% ethanol solutions) to achieve reasonable RCC, which is not amenable for use with many temperature-sensitive biological targeting vectors. The high-temperature radiolabeling in ethanol likely reduces [¹⁰⁵Rh]Rh³⁺ to the more labile [¹⁰⁵Rh]Rh⁺, thereby facilitating ligand exchange [214] before the metal is re-oxidized to the +3 state. This ligand class has been explored extensively [212], and specific key examples are discussed below.

An initial investigation probed the importance of the carbon-chain length, and the resulting differences in chelator size and flexibility, between thioether donors within this ligand class, comparing ethylene and propylene linkers (222-S₄-(CH₂CO₂H)₂ and 333-S₄-(CH₂CO₂H₂)₂, Figure 11). Their coordination chemistry with Rh³⁺ revealed that the shorter linker ligand gave the cis isomer [RhCl₂(222-S₄-(CH₂CO₂H)₂]⁺, whereas the longer linker ligand formed the trans isomer [RhCl₂(333-S₄-(CH₂CO₂H)₂]⁺ [215]. In both cases, the terminal carboxylates remained protonated and were not coordinated to the Rh³⁺ center. Subsequent radiolabeling of these chelators (6–10 × 10⁻⁶ M) with [¹⁰⁵Rh]RhCl₃ at pH 4 at high temperatures (90 °C) led to high RCCs >90% but was less successful at lower temperatures [216]. Their kinetic stabilities in phosphate-buffered saline were also sufficient. [¹⁰⁵Rh][RhCl₂(222-S₄-(CH₂CO₂H)₂]⁺ remained >95% intact after 5 days, whereas [¹⁰⁵Rh][RhCl₂(333-S₄-(CH₂CO₂H)₂]⁺ was 80–90% intact after 5 days.

Figure 11.

Acyclic thioethers radiolabeled with [¹⁰⁵Rh]RhCl₃.

Given the success of 333-S₄-(CH₂CO₂H)₂, it was conjugated to the BBN(7–14)NH₂ peptide, which binds to the gastrin-releasing peptide receptor expressed in multiple types of cancer [212,217]. When radiolabeled with [¹⁰⁵Rh]RhCl₃ at pH 4–5 in refluxing aqueous ethanol, only low RCCs <10% were observed. The authors posited that alternative methods to reduce [¹⁰⁵Rh]Rh³⁺ to [¹⁰⁵Rh]⁺, instead of refluxing ethanol, could improve the radiolabeling efficiency.

In an effort to achieve more facile reduction to [¹⁰⁵Rh]Rh⁺, chelators bearing reducing phosphine moieties were developed [218,219]. Two bis(phosphino-amine) chelators (P₂N₂Ph₆ and P₂N₂Ph₂, Figure 12) and two bis(phosphino-thioether) chelators (P₂S₂Me₂ and P₂S₂Ph₂, Figure 12) were prepared and radiolabeled with [¹⁰⁵Rh]RhCl₃ [218]. The lipophilicity of the P₂N₂Ph₆ chelator complicated its radiolabeling, which could only be performed in 50% ethanol/chloroform, whereas the less lipophilic P₂N₂Ph₂ chelator could be radiolabeled at similar efficiency but within 10% ethanol. The bis(phosphino-thioether) chelators could also be radiolabeled under these conditions with comparable efficiencies. Like the lipophilic P₂N₂Ph₆, P₂S₂Ph₂ also required an organic cosolvent (acetonitrile) for radiolabeling. The [¹⁰⁵Rh]Rh-P₂N₂Ph₂ and [¹⁰⁵Rh]Rh-P₂S₂Me₂ complexes were incubated in phosphate-buffered saline at 37 °C, showing negligible decomplexation over 48 h. Compared to the previous studies with the S₄-(CH₂CO₂H)₂ chelators, the incorporation of phosphine moieties enabled >90% RCC at milder conditions (pH 5–7 vs. 4–5, 60 °C vs. ≥ 80 °C, 5–10% ethanol vs. ≥40%), but required higher chelator concentrations (10⁻³ M vs. 10⁻⁵ M). These milder radiolabeling conditions should facilitate the use of pH- and temperature-sensitive targeting vectors, but the requirement for higher chelator concentrations will lead to constructs with diminished specific molar activity [212].

Figure 12.

Bis(phosphino-amine) chelators (P₂N₂Ph₆ and P₂N₂Ph₂) and bis(phosphino-thioether) chelators (P₂S₂Me₂ and P₂S₂Ph₂) radiolabeled with [¹⁰⁵Rh]RhCl₃.

3.4. Cyclic Chelators for Rhodium

Building upon the linear thioether chelators described above, macrocyclic thiacrown ethers have also been investigated with [¹⁰⁵Rh]Rh³⁺ [220–222]. Among the thioether macrocycles investigated, 16S₄-diol (Figure 13) was studied with both stable and radioactive Rh³⁺ [220]. A crystal structure of [trans-RhCl₂(16S₄-diol)]⁺ revealed the Rh³⁺ center to attain a six-coordinate octahedral geometry comprising four equatorial sulfur donors and two axial chlorides.

Figure 13.

Macrocyclic thioether 16S₄-diol radiolabeled with [¹⁰⁵Rh]RhCl₃ and the crystal structure of [trans-RhCl₂(16S₄-diol)]⁺. Crystal structure reprinted with permission from Ref. [220]. Copyright © 1996 Elsevier, B. V.

Radiolabeling studies with [¹⁰⁵Rh]RhCl₃ showed complex formation that could be confirmed by matching its electrophoretic migration and chromatographic retention time to that of the non-radioactive [trans-RhCl₂(16S₄-diol)]⁺ complex. High RCCs (≥95%, pH 4, 10–15% ethanol) could be achieved with short incubation times, but with the requirement of elevated temperatures (60 min at 80 °C). Alternatively, longer incubation times at lower temperatures (90 min at 64 °C) could also be used to obtain good RCCs. The excellent kinetic stability of [¹⁰⁵Rh][trans-RhCl₂(16S₄-diol)]⁺ was evidenced by the lack of any decomposition in phosphate-buffered saline over 4 days and only 8% loss after 4 days in human serum at 37 °C. Compared to the acyclic chelators discussed above, 16S₄-diol could be labeled at much lower ligand concentration (10⁻⁶ M) and is therefore a promising chelator for ^103mRh. The 16S₄ chelator may also prove useful for other soft radiometals which has been proposed with a recent report of bifunctional 16S₄ derivatives [223].

4. Antimony

Naturally occurring antimony comprises two stable isotopes, antimony-121 (¹²¹Sb) and antimony-123 (¹²³Sb). The radioisotope antimony-119 (¹¹⁹Sb, t_1/2 38.2 h) was first proposed for use as a therapeutic MAE emitter in the early 2000s [224]. Its low-energy emissions (average 8.9 keV/decay) and large number of MAEs (average 23.68 MAEs/decay) make it highly favorable for this application. Antimony-117 (¹¹⁷Sb, t_1/2 2.8 h) serves as a promising diagnostic counterpart that emits β⁺ particles for PET imaging and 159 keV γ photon emission for SPECT [107,225]. The redox chemistry, speciation, biodistribution, and chelation of antimony with a focus on its use in radiopharmaceuticals has been recently reviewed [226].

4.1. Production of Antimony-119

Early production strategies for ¹¹⁹Sb employed the ¹¹⁹Sn(p,n)¹¹⁹Sb reaction, which used 10.1 MeV protons to bombard silver targets that were electroplated with enriched ¹¹⁹Sn [227]. This approach is also successful with clinical cyclotrons that operate in the energy range for producing common radionuclides like ¹⁸F, suggesting that access to ¹¹⁹Sb could be widespread [228]. Isolation of ¹¹⁹Sb from these targets is achieved by first dissolving them in concentrated HCl and then treating them with H₂O₂ to oxidize Sn²⁺ to Sn⁴⁺ and Sb³⁺ to Sb⁵⁺. From here, ion-exchange chromatography can effectively separate the ¹¹⁹Sb from the bulk Sn because of their different oxidation states. The ¹¹⁹Sb isotope can also be produced as a mixture of other Sb radionuclides (denoted ^1XXSb) by deuteron irradiation of natural tin (^natSn) targets through ^natSn(d,n)^1XXSb and ^natSn(d,2n)^1XXSb reactions [225].

Another approach to production uses a metastable isomer of tellurium-119 (^119mTe, t_1/2 4.7 days), which ultimately decays to yield ¹¹⁹Sb. The long half-life of ^119mTe enables its possible use as ¹¹⁹Sb generators, which can provide a multiweek supply of this radioisotope [229]. Although ^119mTe production requires high proton energies (34–68 MeV), it can be produced from naturally occurring antimony (^natSb), which comprises ¹²¹Sb and ¹²³Sb, through the ¹²¹Sb(p,3n)^119mTe and ¹²³Sb(p,5n)^119mTe reactions, respectively [230]. This generator production approach is limited by the difficult purifications needed to rigorously remove the excess ^natSb from the target.

4.2. Chemistry of Antimony

Antimony is a pnictogen (group 15) with a ground state electron configuration of [Kr]4d¹⁰5s²5p³. Although antimony usually behaves as a metal and exhibits oxidation states of +3 ([Kr]4d¹⁰5s²) and +5 ([Kr]4d¹⁰), it can also attain negative oxidation states down to −3 [231]. Sb³⁺ can exhibit weak Lewis base character but is generally considered to be a soft Lewis acid [232–235]. Antimony typically forms complexes with CNs ranging from 3–6 with a six-coordinate ionic radii of 0.76 Å and 0.6 Å for Sb³⁺ and Sb⁵⁺, respectively [152]. Coordination complexes of Sb³⁺ are often anisotropic, a feature that arises from the stereochemical activity of its 5s² lone pair. Under biological conditions, Sb⁵⁺ can be readily reduced to Sb³⁺, suggesting that Sb³⁺ may be more relevant in vivo [236,237]. Antimony in both the +3 and +5 oxidation states exhibits complex speciation in water and is very susceptible to hydrolysis, forming oligomeric and insoluble hydroxides [238]. An additional concern is the oxidation to Sb³⁺ to Sb⁵⁺ by H₂O₂ formed upon radiolysis of H₂O [239–241]. This problem is less significant if radiolabeling processes are rapid [226].

Although the coordination chemistry of antimony for TRT is largely underexplored, antimony compounds are currently used therapeutically for amphotericin-resistant Leishmaniasis, a tropical parasitic disease [242]. Further bioinorganic studies of antimony compounds have investigated their genotoxicity and potential use as chemotherapeutics for cancer and other biological targets which have been reviewed elsewhere [243,244]. Although chelators for applications of antimony in biology are sparse, a vast majority of these complexes rely on thiolates, thiocarbamates, thiocarbazates, and other sulfur-containing ligands that can stabilize the soft Sb³⁺ ion [245–247,243]. The soft Lewis acidity of Sb³⁺ gives it a high affinity for endogenous thiols such as glutathione (log K_ML = 25.2 ± 0.4), which is present at mM concentrations in vivo [248,249]. These properties must be carefully considered when designing chelators for in vivo use.

4.3. Acyclic Chelators for Antimony

Sb³⁺ forms stable complexes with carboxylate-containing chelators, such as EDTA, DTPA, lactic acid, tartaric acid, and citric acid with log K_ML values ranging from 7.84–19.5 (Table 4) [250]. Acyclic chelators with both hard and soft donors, such as carboxylates and thiolates, respectively, provide insight on the donor atom preferences of Sb³⁺. For example, Sb³⁺ complexes of glutathione and trypanothione show this ion to exclusively bind to the thiolate moieties [245,248,251,252].

Table 4.

Selected stability constants of Sb–L complexes.

Ion	Chelator	logKML	Ref.
Sb3+	glutathione	25.2	[248]
	EDTA	19.5	[250]
	PCTA	23.2	[253]

The preference of Sb³⁺ for soft thiolate donors prompted researchers to investigate the ligand trithiol (Figure 14) for ¹¹⁹Sb. The Sb complex of this ligand was prepared from SbCl₃, forming the neutral [Sb(κ³-trithiol)] complex (Figure 14) [254]. Its crystal structure shows the Sb³⁺ ion to attain a trigonal pyramidal coordination geometry consistent with the stereochemical activity of the 5s² lone pair [255]. The average S–Sb–S angle is 91.5°, and the average S–Sb bond length is 2.444(7) Å. Multiple intermolecular S⋯Sb contacts exist throughout the lattice with an average distance of 3.2960(8), which is shorter than the sum of S and Sb van der Waals radii [256–258]. Notably, the As³⁺ analogue of this complex shows a much longer average intermolecular S⋯As distance of 3.5369(12), which is comparable to the sum of their van der Waals radii [259]. Therefore, this shorter S⋯Sb interaction highlights the greater Lewis acidity of Sb³⁺ compared to As³⁺, as a result of the smaller of stereochemical activity of the Sb³⁺ 5s² lone pair [254].

Figure 14.

Trithiol, its bifunctional derivative trithiol-BFC, and the corresponding crystal structures of their Sb³⁺ complexes [Sb(κ³-trithiol)] (CCDC Number 1976847) [254] and [Sb(κ³-trithiol-BFC)] (CCDC Number 2071807) [225].

The bifunctional derivative of trithiol (trithiol-BFC, Figure 14) was used for radiolabeling with ^1XXSb, and the crystal structure of the [^natSb][Sb(κ³-trithiol-BFC)] complex shows similar coordination chemistry to the previous report (Figure 14) [225]. The Sb³⁺ center attains a trigonal pyramidal geometry with an average S–Sb–S bond angle of 91.6° and average S–Sb bond length of 2.440(8) Å. This angle is much more consistent with an octahedron missing three vertices, as opposed to a tetrahedron missing one vertex, which would give bond angles closer to 109.5°, and is consistent with a known decrease in orbital hybridization for heavy main group elements. Like the structure of Sb(κ³-trithiol), intermolecular interactions between S and Sb are apparent with average distances of 3.4962(2) Å. As noted above, these short distances, which are less than the sum of the van der Waals radii for these elements, indicate that this interaction has strong Lewis acid character.

The trithiol-BFC was successfully radiolabeled with a radioisotopic mixture of [^117/119Sb]Sb³⁺ (8.4 MBq/pmol ¹¹⁷Sb and 42 kBq/pmol ¹¹⁹Sb). Unlike more conventional radiolabeling processes, the procedure used in this study involved the direct incubation of the chelator (10⁻⁵ M) with the dissolved target for 30 min at room temperature. This process afforded an AMA of 30 nmol of trithiol-BFC incorporating 5.8 pmol ^1XXSb (49 MBq and 240 kBq of ¹¹⁷Sb and ¹¹⁹Sb, respectively). This approach of directly radiolabeling from the unseparated target material is a promising strategy that can streamline the clinical advance of novel radiopharmaceuticals. Although thermodynamic stability constants were not reported, the [^117/119Sb]Sb-trithiol-BFC complex shows high kinetic stability when incubated with 25 mM cysteine (91 ± 9% intact after 72 h) and fetal bovine serum (97.5 ± 1.6% after 72 h).

A recent study investigated the hydrotris(methimazolyl)borate (Tm⁻, Figure 15), which is known to chelate Sb³⁺ [260,261], for radiolabeling with Sb radioisotopes [262]. A mixture of Sb radioisotopes, comprising antimony-118m, 120m, and 122, was produced from the 10.8–13.1 MeV proton bombardment of ^natSn targets, which were then subjected to radiochemical processing and separation to yield [^1XXSb]Sb⁵⁺. Incubation of the radioisotope mixture with 2.8 mM NaTm at room temperature for 1 h at pH <4 afforded a complex with a RCC of 99.5 ± 1.0%. Using isotopically pure [¹¹⁷Sb]Sb⁵⁺, obtained from an enriched [¹¹⁷Sn]Sn target, radiolabeling with NaTm also proceeded effectively, giving a complex with AMA of 8.8–33.0 MBq/μmol after incubating at room temperature for 45 min at pH <4. Although [^1XXSb]Sb⁵⁺ was used for these radiolabeling procedures, it is reduced to the +3 oxidation state by excess Tm⁻ ligand, affording [^1XXSb]Sb(Tm)₂ as the most likely product. Although [^1XXSb][Sb(Tm)₂]⁺ was unstable at pH values of 4 or higher, this complex showed high kinetic stability in the presence of a high concentration of cysteine (25 mM) under acidic conditions (pH 0–1).

Figure 15.

Hydrotris(methimazolyl)borate (Tm⁻) chelator and its corresponding [Sb(Tm)₂]⁺ crystal structure (CCDC Number 274857) [261].

The crystal structure of [Sb(Tm)₂]⁺ reveals an Sb³⁺ center bound two Tm⁻ ligands (Figure 15) [261]. One of the Tm⁻ ligands interacts with Sb³⁺ in a tridentate κ³ manner, whereas the other binds with only two coordinating sulfur atoms in a κ² mode. The five-coordinate [Sb(κ³-Tm)(κ²-Tm)]⁺ complex exhibits a square pyramidal geometry, that is distorted by stereochemical effects from the 5s² lone pair.

The room-temperature ¹H NMR spectrum of the complex exhibits broad peaks, which the authors attributed to fluxionality of the Tm⁻ ligands. The ligand fluxionality, in addition to the exposed Sb³⁺ center, may explain the significant hydrolytic instability of the radiolabeled [^1XXSb][Sb(Tm)₂]⁺ complex [262]. Crystal structures of the halide-bound [Sb(κ³-Tm)(κ¹-Tm)X] complexes (X = Br⁻, I⁻) were obtained when SbBr₃ and SbI₃ were used as starting materials. These complexes attain similar square pyramidal geometries with a Sb-bound halide, one Tm⁻ ligand binding in a κ³ mode, and the second Tm⁻ ligand in a κ¹ mode. Furthermore, solution studies showed that the iodide complex is in equilibrium with a dimeric [{Sb(κ³-Tm)(μ₂-I)I}₂] structure, where the κ¹-Tm ligand of the mononuclear structure is replaced with two bridging μ₂-I ions. Because radiolabeling involved the use of HCl solutions of [^1XXSb]Sb^3+/5+, similar halide complexes of radioactive Sb may be formed, which should be considered for downstream biological applications. Considering the results of Tm⁻ radiolabeling with [^1XXSb]Sb³⁺ [262], the ligand fluxionality observed by NMR [261], and the anion-dependent crystal structures and equilibria [261], it is clear that Sb–Tm complexes exhibit complicated and dynamic coordination chemistry. Further studies into the composition and speciation of the radiolabeled complexes are necessary to assess the feasibility of Tm⁻ as a chelator for radio-antimony.

Efforts to develop chelators to stabilize the +5 oxidation state of Sb have also been pursued. Compared to this metal in the +3 oxidation state, Sb⁵⁺ lacks the stereochemically active lone pair, is a much harder ion, and has a substantially smaller ionic radius (0.6 Å compared to 0.76 Å, CN 6) [250,263,264]. Based on these properties, chelators containing hard anionic donors, instead of soft sulfur donors, were investigated. Solutions of 1 mM of either deferoxamine (DFO), DOTA, or TREN-CAM were radiolabeled with 21 kBq [^1XXSb]Sb⁵⁺ at 80 °C in pH 4 solution for 60 min. Of these three chelators, only TREN-CAM (Figure 16) could be radiolabeled, as determined via radio-HPLC. Optimization of these radiolabeling conditions showed that slightly lower temperatures (60 °C) and lower ligand concentration (10⁻⁴ M) were also effective. The high hydrolytic instability of Sb⁵⁺ was reflected by the fact that at slightly higher pH conditions (increasing from 2 to 4 to 6), a significant quantity of an intermediate hydroxide complex with only two coordinated catecholate donors [^1XXSb][Sb(H₂TREN-CAM)(OH)₂]⁻ was observed by radio-HPLC with a retention time shorter than the [^1XXSb][Sb(TREN-CAM)]⁻ complex. Consequently, it is important to use precise chromatographic characterization techniques, like radio-HPLC, to verify product formation.

Figure 16.

Tris-catecholate chelator TREN-CAM and the crystal structure of its [Sb(TREN-CAM)]⁻ complex (CCDC Number 2381819) [265].

Once formed, [^1XXSb][Sb(TREN-CAM)]⁻ was shown to be kinetically stable assessed in human serum at 37 °C, exhibiting <4% decomplexation after 72 h. The complex was also stable in the presence of 2.8 mM glutathione over 72 h at 37 °C, reflecting the ability of the TREN-CAM ligand to stabilize the +5 oxidation state. By contrast, under the same conditions, the intermediate [^1XXSb][Sb(H₂TREN-CAM)(OH)₂]⁻ species fully decomposed in the presence of glutathione after 1 day, revealing that hexadentate coordination by the TREN-CAM chelator is needed to prevent Sb⁵⁺ reduction. Both the efficient radiolabeling and radiochemical stability of the TREN-CAM complexes of [^1XXSb]Sb⁵⁺ indicate that further studies to test bifunctional analogues and biological conjugates are warranted.

To further understand the high efficacy of TREN-CAM for Sb⁵⁺, further solution and solid-state structural studies were carried out. Notably, the use of Sb K-edge X-ray absorption spectroscopy (XAS) and extended X-ray absorption fine structure (EXAFS) elucidated its properties in solution. Within the X-ray absorption near-edge structure (XANES) region, the edge energy was consistent with the +5 oxidation state. The EXAFS region could be modeled using a six-coordinate Sb center with average Sb–O interatomic distances of 1.992(5) Å. A single-crystal X-ray diffraction study of [^natSb][Sb(TREN-CAM)]⁻ verified these results (Figure 16), revealing a six-coordinate Sb center interacting with all six catecholate donors in a pseudo-octahedral geometry. The agreement between the CN and average interatomic distances measured by XAS indicate that this structure remains intact in solution. These structural studies provide a launching point for rational chelator design effort for radioisotopes of Sb in the +5 oxidation state.

4.4. Cyclic Chelators for Antimony

As of yet, cyclic chelators have not been investigated for ¹¹⁹Sb, but several reports have characterized complexes of this type with stable Sb³⁺ salts. Efforts toward the use of DOTA for this application were recently reported. Despite the widespread implementation of DOTA as a highly effective chelator, it was not able to form a complex with Sb³⁺ under aqueous conditions, presumably due to the extensive hydrolysis of this ion. Using ethanol as a solvent, however, enabled the isolation of [Sb(DOTA)]⁻ [253,266]. Challenging [Sb(DOTA)]⁻ with one equivalent of Bi³⁺, Y³⁺, and Sc³⁺ led to stochiometric displacement of the Sb³⁺ ion, and treating this complex with two molar equivalents of biologically relevant divalent ions, Ca²⁺, Mg²⁺, and Zn²⁺, resulted in 65%, 55%, and 38% release of Sb³⁺, respectively [266]. After incubating aqueous solutions of Na₄DOTA and SbCl₃ for 2 h at 50 °C, the vapor diffusion of ethanol afforded crystals of [Sb(DOTA)]⁻, apparently avoiding the formation of insoluble oxy-hydroxide species. Within the crystal structure of [Sb(DOTA)]⁻ (Figure 17), the Sb³⁺ center sits on a crystallographic 2-fold axis, conferring the complex with C₂ symmetry. All four macrocyclic N atoms and all four pendent carboxylates coordinate to the Sb³⁺ center, providing an eight-coordinate N₄O₄ coordination environment. The N–Sb distances range from 2.4277(11) to 2.4563(11) Å, and the O–Sb distances span 2.5011(10) to 2.6929(11) Å. The unique coordination chemistry of Sb³⁺ is apparent when comparing to that of the Sc³⁺ ion [267]. Both ions have similar eight-coordinate ionic radii of 0.87 Å [266], but Sb³⁺ is softer. Additionally, its electron configuration possesses a 5s² valence lone pair that can be stereochemically active. These differences are clear when comparing the crystal structures of [Sb(DOTA)]⁻ and [Sc(DOTA)]⁻. Despite the similar ionic radii of Sb³⁺ and Sc³⁺, the O–Sc distances (2.223(2) Å) are substantially shorter than the O–Sb distances of 2.501(1)–2.693(1) Å (Figure 17), while the N–M distances are comparable as is the distortion out of the N₄ plane (1.336 Å for both Sc³⁺ and Sb³⁺). Therefore, the lability of [Sb(DOTA)]⁻ cannot be attributed to geometric distortion of the DOTA chelator because it is actually less strained than the inert [Sc(DOTA)]⁻ complex. A possible explanation for its poor stability is most likely a mismatch between the HSAB properties of the DOTA carboxylate and soft Sb³⁺ ion. Hence, chelators bearing softer donors, such as thiolates or thioethers, may be more promising for Sb³⁺ chelation, as indicated by the relative success of trithiol-BFC.

Figure 17.

Crystal structures of [Sb(DOTA)]⁻ (left, CCDC Number 2128466) [266] and [Sc(DOTA)]⁻ (right, CCDC Number 984842) [267]. The unit cell of [Sb(DOTA)]⁻ contains both Λ(λλλλ) and Δ(δδδδ) enantiomers; only the Λ(λλλλ) is shown here for clarity. Coordinated K⁺ ion omitted from [Sc(DOTA)]⁻ structure for clarity.

Another cyclic aminopolycarboxylate PCTA (3,6,9,15-tetraazabicyclo[9.3.1]pentadeca-1(15),11,13-triene-3,6,9-triacetic acid, Figure 18) has also been studied for its ability to bind Sb³⁺ [253]. The [Sb(PCTA)] complex was formed by mixing equimolar amounts of Sb³⁺, prepared via the dissolution of Sb₂O₃ in 6 M HCl to avoid hydrolysis products, and H₃PCTA at 2.5 mM concentration in water, adjusting the pH to neutral with NaOH solution, then stirring at 70 °C for 5–7 days. Crystals were then grown from methanol solutions, furnishing two different crystal types. The first contained [Sb(PCTA)], with all three carboxylates deprotonated, cocrystallized with NaCl as [NaCl][Sb(PCTA)], and the second comprised the cationic complex [Sb(HPCTA)]⁺, with one carboxylate remaining protonated, as the chloride salt (Figure 18). In the [NaCl][Sb(PCTA)] structure, the Na⁺ ion is coordinated to one of the deprotonated carboxylate arms of the PCTA ligand, giving rise to a six-coordinate Sb³⁺ center. In the [Sb(HPCTA)]⁺ structure, the Sb³⁺ center is also six-coordinate because the protonated carboxylate does not bind to the central ions, but instead hydrogen bonds with a lattice water molecule. Consequently, both crystals show the Sb³⁺ center to attain a distorted pentagonal pyramidal geometry supported by an N₄O₃ environment. Only the [NaCl][Sb(PCTA)] crystal, which refined to better residuals, will be discussed. The Sb³⁺ center is six-coordinate, and the complex forms a distorted pentagonal pyramidal geometry similar to the crystal structure with EDTA [268]. In the absence of cocrystallized NaCl, density functional theory (DFT) calculations show a seven-coordinate structure which the authors state is the most likely coordination environment in the solution phase. The DFT structure shows a nonbonding lone pair of electrons on the metal center consistent with Sb³⁺ coordination complexes [255]. The pyridyl nitrogen of the macrocycle coordinates Sb³⁺ with a N–Sb interatomic distance of 2.512(3) Å, a value that is similar to those of the adjacent macrocyclic N donors (N1 and N3), which give rise to an average N–Sb distance of 2.546(3) Å. Notably, the macrocyclic N donor distal to the pyridyl group affords a substantially shorter N–Sb distance of 2.303(3). The O–Sb distances of the two coordinated carboxylates O atoms are 2.148(4) and 2.221(4) Å, whereas the Na-bound carboxylate O atom (O6) is 2.812(4) Å away from the Sb³⁺ center. The Sb³⁺ center resides 0.504 Å above the plane formed from by the macrocycle N atoms. The authors note that Sb³⁺ coordination complexes with EDTA⁴⁻ [268], HEDTA³⁻ [269], NTA³⁻ (nitrilotriacetate) [270], tartrate [271], and oxalate [272] adopt similar distorted pyramidal structures. These pyramidal structures—originating from the Sb³⁺ lone pair activity—prevent the donor atoms from completely encapsulating the Sb³⁺ center and therefore make it vulnerable towards transchelation [253].

Figure 18.

PCTA chelator and the crystal structures of its [NaCl][Sb(PCTA)] (CCDC Number 2060909) and [Sb(HPCTA)]Cl (CCDC Number 2060910) complexes [253].

To assess the stability of the [Sb(PCTA)] complex, its stability constant was measured via UV-vis titration to be log K_ML = 23.2(1). Notably, this value is larger than that of [Sb(EDTA)]⁻ (log K_ML = 19.5), marking it as one of the largest Sb³⁺ stability constants reported to date [250]. Given the susceptibility of Sb³⁺ towards hydrolysis, however, [Sb(PCTA)] underwent significant spontaneous and hydroxide-assisted degradation near physiological pH with dissociation half-lives of 82.0, 7.0, and 1.8 s at pH 6.8, 8.0, and 8.6, respectively. The authors attribute the lability of the complex to the situation of the Sb³⁺ ion above the plane of the macrocycle, as observed in the crystal structures, underscoring this geometric vulnerability for transmetalation and transchelation as a point for optimizing future chelators. Although the PCTA chelator does not appear to be suitable for Sb³⁺ chelation for radiopharmaceuticals, the coordination chemistry and thermodynamic and kinetic studies performed for this complex provide useful data for future ligand design efforts.

5. Lanthanum

Lanthanum has two naturally occurring isotopes. Stable lanthanum-139 (¹³⁹La) comprises the majority (99%) with the remaining trace amounts existing as lanthanum-138 (¹³⁸La, t_1/2 1.03 × 10¹¹ years). Among its synthetic radioisotopes, lanthanum-135 (¹³⁵La, t_1/2 19.5 h, EC), which decays to stable barium-135, has gained increasing attention as a MAE emitter for TRT because of its suitable half-life and high yield of MAEs with appropriate energies. Specifically, an average of 12.3 MAEs are emitted per decay with a total kinetic energy of 6.6 keV. Further, the use of stable La³⁺ as a non-radioactive surrogate for studies of the α-emitter actinium-225 (²²⁵Ac)—due to their similar ionic radii and Lewis acidities—has provided preliminary data for its coordination chemistry with BFCs prior to interest in the therapeutic isotope ¹³⁵La.

5.1. Production of Lanthanum-135

¹³⁵La can be prepared by proton irradiation of metallic natural barium (^natBa, 6.59% in ¹³⁵Ba) targets predominantly through the ¹³⁵Ba(p,n)¹³⁵La reaction using 16 MeV cyclotrons as discussed in a recent review [273]. Purification by cation-exchange resin furnishes >98% radionuclidic purity after letting the irradiated target stand for 12–24 h to allow for the decay of the coproduced, shorter-lived isotopes lanthanum-132, 133, 134, and 136 [274]. The activities of longer-lived isotopes lanthanum-132 (t_1/2 4.6 h), lanthanum-133 (t_1/2 3.9 h), and barium-135m (t_1/2 28.7 h) were significantly less than that of ¹³⁵La (92%) at end-of-saturation bombardment, and after 20 h, the radionuclidic purity of ¹³⁵La was 98% [274]. A different purification method was performed involving dissolution of the target in HNO₃, precipitation of bulk barium, and isolation of the lanthanum radionuclides on a column [275]. However, this method involves a much shorter decay of the target (2–3 h) prior to dissolution, thus leaving a mixture of lanthanum isotopes, namely 2.47% of the β⁺-emitting ¹³²La isotope. As this purification route was performed for preclinical studies, the ^132/135La mixture was then used to radiolabel DOTA for preliminary imaging and biodistribution studies without improving the radionuclidic purity. ¹³⁵La can also be prepared from barium-136 (¹³⁶Ba) through the ¹³⁶Ba(p,2n)¹³⁵La reaction. This process uses enriched metallic ¹³⁶Ba targets for irradiation with 14–30 MeV proton beams [276]. This reaction, however, affords lower yields compared to methods using ^natBa targets in the same proton energy regime.

Although most of the targets for lanthanum radionuclide production are barium metal, the use of natural and isotopically enriched barium carbonate ([^natBa]BaCO₃ and [¹³⁵Ba]BaCO₃) has also been investigated [277]. These targets were irradiated with both 11.9 and 23.3 MeV proton beams. The higher-energy protons yielded higher activities but gave a mixture of lanthanum-133 (¹³³La) and ¹³⁵La. When using a lower-energy 11.9 MeV proton beam, no ¹³³La was observed, giving good radionuclidic purity of ¹³⁵La, albeit with much lower activities. This trend was observed for both the [^natBa]BaCO₃ and [¹³⁵Ba]BaCO₃ targets, as well as for metallic ^natBa targets, with much higher activities afforded by the enriched [¹³⁵Ba]BaCO₃ target. This initial method with BaCO₃ targets was further improved via the use of an intermediate proton energy of 16.5 MeV [278]. After dissolution in HCl, Ba²⁺ was removed by precipitation as BaCl₂, and the lanthanide isotopes were chemically purified on a column. Using the [^natBa]BaCO₃ target, a gamma spectrum of the purified product after 21 h revealed a mixture of radioisotopes, with the majority of the activity arising from ¹³⁵La. Using this same method with enriched [¹³⁵Ba]BaCO₃ targets, ¹³⁵La was obtained in high radionuclidic purity with negligible amounts of other lanthanum isotopes after only 30 min from the end of bombardment. This method of Ba²⁺ precipitation during the purification also enables recycling of the ¹³⁵Ba to be reused for new targets.

Precipitation and chromatographic purification of lanthanum isotopes from barium target materials are the most commonly used radiochemical separation approaches, but a recent study has reported electrochemical separation by electroplating ^132/135La metal onto a platinum anode [279]. After the initial electroplating, the polarity of the electrochemical cell was then reversed to oxidatively remove ^132/135La, yielding [^132/135La]LaCl₃ in >99.99% radionuclidic purity.

5.2. Chemistry of Lanthanum

Lanthanum is the lightest lanthanide, and its ground state electron configuration is [Xe]5d¹6s². Like the other lanthanides, it exists almost exclusively in the +3 oxidation state ([Xe]), with only a few examples of air- and water-sensitive molecular complexes of this metal ion in the +2 oxidation state [280,281]. La³⁺ is the largest lanthanide ion; its six- and eight-coordinate ionic radii are 1.03 and 1.36 Å, respectively (Table 2) [152], and it typically forms complexes with high CNs ranging from 6–12. According to the HSAB principle [232], La³⁺, like the other lanthanide ions, is a hard Lewis acid (η = 15.39 eV, Table 2) [153]. This property is reflected by its high affinity for oxygen donors, like water and carboxylates. For example, it has a hydrolysis constant (−log K for La³⁺ + H₂O ⇌ La(OH)²⁺ + H⁺) of 8.89, suggesting necessary use of neutral or acidic media for radiolabeling, and a relatively high heat of hydration (ΔH° = 6.9 kcal/mol), which may require high temperatures for radiolabeling [282,283]. Based on these properties, La³⁺ forms stable complexes with chelators containing many oxygen donors [280,281,284].

5.3. Cyclic Chelators for Lanthanum

Unsurprisingly, the ubiquitous DOTA was among the first cyclic chelators to be radiolabeled with [^132/135La]La³⁺ and was investigated because of the high stability constant of the [La(DOTA)]⁻ complex (log K_ML = 20.7, Table 5) [275,285]. Aliquots of [^132/135La]LaCl₃ (3.7 MBq, ¹³⁵La basis) were incubated with increasing concentrations of DOTA (0–250 μM) in aqueous NaOAc buffer (pH 4.5) at 80 °C for 30 min, resulting in a complex with an AMA of 22 MBq/nmol [275]. Using similar radiolabeling conditions but at a slightly higher pH of 5.5, an AMA of 25 MBq/nmol was reported [286]. A separate study reported quantitative [^133/135La]La³⁺ labeling of DOTA (97.9% incorporation) using 2 μM DOTA (pH 4.5, 30 min, 80 °C) [287]. However, another study reported insufficient complexation of [¹³⁵La]La³⁺ by DOTA at higher pH (pH 5.5, 80 °C, 30 min) with an AMA value of 0.74 MBq/nmol, suggesting that lower pH is favorable for radiolabeling [288].

Table 5.

Selected stability constants of La–L complexes.

Ion	Chelator	logKML	Ref.
La3+	EDTA	15.5	[178]
	DTPA	19.48	[179]
	DOTA	20.7	[285]
	DO3Apic	21.17	[289]
	macropa	14.99	[290]
	macrodipa	12.13	[291]
	py-macrodipa	14.31	[288]
	py2-macrodipa	16.68	[292]

In developing the first TRT complexes with ^132/135La, a DOTA derivative bearing one picolinate pendant (DO3Apic, Figure 19) was radiolabeled with ^132/135La [286]. Using milder conditions (pH 5.5, 30 min, 25 °C) than those used for DOTA, an AMA of 14 MBq/nmol was obtained. Although this AMA is lower than that obtained for DOTA, DOTA required high temperature (80 °C), which is not desirable for downstream applications [289]. The kinetic stability of [^132/135La][La(DO3Apic)]⁻ was also found to be exceptional, as this radiocomplex remained intact in the presence of 1000-fold excess of DTPA for at least 21 days. Given its efficacy, this chelator was also conjugated to a targeting vector (Section 5.4).

Figure 19.

Macrocyclic chelators for La³⁺ discussed in Section 5.3.

The expanded macrocycle macropa (Figure 19), which has been used extensively for large α-emitting radionuclides like ²¹²Pb [293], ²¹³Bi [294], ²²³Ra [265], and ²²⁵Ac [295], has also proven to be effective for ^132/135La [286]. Although the [La(macropa)]⁺ stability constant log K_ML = 14.99 is much smaller than that of DOTA [290], macropa can be effectively radiolabeled with [^132/135La]LaCl₃, affording a large AMA of 161 MBq/nmol under mild room temperature conditions [286]. A later study reported 98.1% incorporation of [^133/135La]La³⁺ under mild conditions (pH 4.5, 10 min, 22 °C) with 5.7 μM macropa [287]. Like the DO3Apic complex, [^132/135La][La(macropa)]⁺ is also kinetically inert in the presence of 1000 equivalents of DTPA.

An isomer of macropa (macrodipa, Figure 19) and a derivative with a rigidified macrocyclic core (py-macrodipa, Figure 19) form La³⁺ complexes with moderate stability, as characterized by log K_ML values of 12.13 and 14.31, respectively. Furthermore, they could be radiolabeled with [¹³⁵La]La³⁺ under mild conditions (pH 5.5, 25 °C, 15 min), affording complexes with AMAs of 0.105 and 55 MBq/nmol for macrodipa and py-macrodipa, respectively [288]. Once formed, both [¹³⁵La]La-macrodipa and [¹³⁵La]La-py-macrodipa remained stable in human serum at 37 °C for 24 h. Differences were apparent, however, when the kinetic stabilities of the cold lanthanum (^natLa) complexes [^natLa]La–macrodipa and [^natLa]La–py-macrodipa were assessed by challenging them with a 10³-fold excess of DTPA. Under these conditions, the half-life of [^natLa]La–macrodipa was 1678 ± 33 s, whereas the half-life of [^natLa]La–py-macrodipa was substantially longer, 6.3 ± 0.3 days. Thus, py-macrodipa is favorable over macrodipa, both in terms of radiolabeling efficacy and kinetic stability. Insights into these differences can be gleaned from the crystal structures of the two complexes and DFT calculations of their La³⁺-binding energies. The crystal structures of [La(macrodipa)]⁺ and [La(py-macrodipa)]⁺ (Figure 20) reveal them to attain the same conformation and possess similar interatomic distances between donor atoms to the La³⁺ centers [288,291]. The pyridine group within py-macrodipa, however, provides stronger binding to the La³⁺ center and a more rigid chelator structure that suggests a higher degree of preorganization. DFT calculations on the binding energy and chelator strain energy support this assertion because they show py-macrodipa has a 21.3 kJ/mol stronger binding energy than macrodipa with La³⁺, and the chelator stain energy of py-macrodipa is 6.9 kJ/mol smaller than that of macrodipa [288]. Although the strain energy is a thermodynamic parameter, it reflects more restricted conformational flexibility of py-macrodipa, which may result in the improved kinetic stability of its La³⁺ complex.

Figure 20.

Crystal structures of [La(macrodipa)]⁺ (left, CCDC Number 2003601) [291] and [La(py-macrodipa)]⁺ (right, CCDC Number 2085495) [296].

5.4. Bifunctional Chelators for Lanthanum

Bifunctional versions of DO3Apic (DO2Apic) and macropa were conjugated to a PSMA-binding Glu-urea-Glu peptide derivative (DUPA, Figure 21) [297] and assessed for their ability to bind to radioisotopes of lanthanum [286,298]. The targeting capabilities of these constructs were validated using an in vitro radioactive displacement assay. In this experiment, La(DO2Apic)-DUPA and La(macropa)-DUPA were added to PSMA-expressing PC-3 PiP cells, and their PSMA affinity was verified by showing no displacement from the cells when they were challenged with a clinically validated ^99mTc-based MIP-1427 agent [286]. In addition, both BFC complexes were stable to a DTPA challenge over the course of 21 days. Radiolabeling the macropa-DUPA BFC with [¹³⁵La]La³⁺ was equally effective as the unconjugated macropa, as characterized by an AMA of 149 MBq/nmol. The DO2Apic-DUPA BFC, however, exhibited poorer radiolabeling properties than DO3Apic, presumably due to conversion of one of the anionic carboxylate donors to a neutral carboxamide. Radiolabeling of DO2Apic-DUPA required elevated temperature (80 °C) to achieve an AMA (13 MBq/mmol) that was comparable to DO3Apic obtained at room temperature (14 MBq/nmol). Macropa has also been conjugated to the well-established PSMA-617 vector for use with radiolanthanum. The resulting macropa-PSMA-617 BFC was quantitatively radiolabeled at 1 μM with [¹³³La]LaCl₃ (pH 6, 25 °C, 15 min) giving a large AMA value of 330 MBq/nmol [299], further verifying the suitability of this chelator for La radioisotopes.

Figure 21.

DO2Apic- and macropa-DUPA bifunctional chelators.

6. Terbium

Terbium has four synthetic isotopes with potential applications within nuclear medicine: terbium-149 (¹⁴⁹Tb, t_1/2 4.1 h, α-/β⁻-therapy), terbium-152 (¹⁵²Tb, t_1/2 17.5 h, PET/SPECT), terbium-155 (¹⁵⁵Tb, t_1/2 5.3 days, SPECT/MAE therapy), and terbium-161 (¹⁶¹Tb, t_1/2 6.9 days, β⁻-/MAE therapy) [300]. Hence, terbium is uniquely poised for use in matched-pair theranostic radiopharmaceutical agents that employ different radioisotopes of the same element to provide both diagnostic and therapeutic modalities. Of these four radioisotopes, ¹⁵⁵Tb and ¹⁶¹Tb are promising MAE emitters, which release 13.9 and 10.9 MAEs/decay, respectively. These two isotopes also emit CEs (0.77 and 1.42 per decay, respectively), which may also be useful for therapy over larger distances. Although radiopharmaceuticals based on ¹⁵⁵Tb have mostly focused on the use of this radioisotope for SPECT imaging, its high yield (13.9 MAEs/decay) with suitable energies (total energy 13.5 keV), also make it very promising for MAE-based TRT. Likewise, ¹⁶¹Tb has mostly received attention for its β⁻ emissions, which are similar to those of clinically approved lutetium-177 (¹⁷⁷Lu), and its low-energy γ photons (74.6 keV (9.8%) and 48.9 keV (14.8%)), which are suitable for SPECT imaging [300,301]. In addition to these emissions, ¹⁶¹Tb also releases short-range CEs (1.4 per decay, 39.3 keV) and MAEs (10.9 per decay, 8.9 keV), which are key contributors to its therapeutic potential [302–304]. In fact, Monte-Carlo simulations predict that the MAEs and CEs make this radioisotope substantially more effective for treating micrometastases compared to conventional ¹⁷⁷Lu [300,305–307].

6.1. Production Terbium-155 and 161

Terbium-155 (¹⁵⁵Tb, t_1/2 5.3 days) can be produced by high-energy (1.4 GeV) proton spallation of natural tantalum (^natTa, 99.99% abundant in tantalum-181) targets with advanced mass separation methods. Due to the high proton energies and need for on-line mass separation, production through this route is limited to advanced accelerators such as the ISOLDE and MEDICIS systems at CERN [308]. Alternatively, bombardment with lower, more accessible proton energies (30 MeV) on Gd₂O₃ targets highly enriched in gadolinium-155 ([¹⁵⁵Gd]Gd₂O₃) by the ¹⁵⁵Gd(p,n)¹⁵⁵Tb reaction can furnish the desired ¹⁵⁵Tb isotope. However, this reaction requires that the target be nearly 100% enriched in ¹⁵⁵Gd, which presents a significant hurdle for ¹⁵⁵Tb production. To circumvent the need for isotopically enriched targets, the 100% natural abundance of stable terbium-159 (¹⁵⁹Tb) as the only native isotope of Tb can be leveraged. In this strategy, natural Tb is irradiated with protons to produce dysprosium-155 (¹⁵⁵Dy) via the ¹⁵⁹Tb(p,5n)¹⁵⁵Dy reaction. However, the excitation function for this reaction reaches its maximum at 48 MeV and consequently requires high proton energies [309]. Subsequent electron capture decay of ¹⁵⁵Dy (t_1/2 9.9 h) affords ¹⁵⁵Tb [309]. Similarly, α-particle bombardment of natural gadolinium (^natGd) targets can produce the ¹⁵⁵Dy intermediate (¹⁵²Gd(α,n)¹⁵⁵Dy; ¹⁵⁴Gd(α,3n)¹⁵⁵Dy; ¹⁵⁵Gd(α,4n)¹⁵⁵Dy; ¹⁵⁶Gd(α,5n)¹⁵⁵Dy; and ¹⁵⁷Gd(α,6n)¹⁵⁵Dy), which decays to ¹⁵⁵Tb [310]. Although both of these latter strategies are advantageous because they use non-isotopically enriched targets, they are limited because of the requirements for high-energy cyclotrons (66 MeV protons or 60 MeV α-particles) and very efficient radiochemical processes to separate the adjacent Dy³⁺ and Tb³⁺ ions [309,310].

The production of terbium-161 (¹⁶¹Tb, t_1/2 6.9 days) can be achieved by neutron bombardment of enriched gadolinium-160 oxide ([¹⁶⁰Gd]Gd₂O₃) targets to make gadolinium-161 (¹⁶¹Gd) via ¹⁶⁰Gd(n,γ)¹⁶¹Gd, which undergoes β⁻-decay to ¹⁶¹Tb. Alternatively, natural terbium (^natTb, monoisotopic in ¹⁵⁹Tb) targets also provide a reactor-based route to ¹⁶¹Tb via ¹⁵⁹Tb(2n,γ)¹⁶¹Tb. Although neutron bombardment of ¹⁶⁰Gd₂O₃ can furnish TBq quantities of ¹⁶¹Tb, it requires development of more efficient Gd isotope enrichment strategies to become a viable approach. By comparison, ¹⁶¹Tb produced from ^natTb is not limited by the target, but it has low AMA because of the inability to isolate ¹⁶¹Tb from the bulk ^natTb target material and low isotopic purity because the long-lived, β⁻-emitting isotope terbium-160 (¹⁶⁰Tb, t_1/2 72.3 days) is also produced. Irradiation of ¹⁶⁰Gd with deuterons (¹⁶⁰Gd(d,n)¹⁶¹Tb and predominantly ¹⁶⁰Gd(d,p)¹⁶¹Gd(ε)→¹⁶¹Tb) has also been reported but is limited by both the need for enriched ¹⁶⁰Gd targets and the coproduction of ¹⁶⁰Tb. Thus, further efforts in optimizing target enrichment and radiochemical separations are needed to increase the availability of ¹⁶¹Tb.

6.2. Chemistry of Terbium

Terbium is in the middle of the lanthanide series and has a ground state electron configuration of [Xe]4f⁹6s². Under the aerobic and aqueous conditions typically used for radiochemistry, the only stable oxidation state for this metal is +3 ([Xe]4f⁸). The half-shell stability of the +4 oxidation state, however, ([Xe]4f⁷) has also afforded molecular complexes of Tb⁴⁺ [311,312], and extensive efforts with strong reductants have yielded isolable Tb²⁺ complexes under strictly air- and water-free conditions [313]. The position of Tb³⁺ near the middle of the lanthanide series places it in the region of the so-called “gadolinium break,” a point where the ionic radius contracts enough to afford substantial differences in coordination chemistry between the light and heavy lanthanides. The six-coordinate ionic radius of Tb³⁺ is 0.923 Å (Table 2), and this ion usually forms complexes with coordination numbers of 8 or 9 and is a hard Lewis acid [152]. The hydrolysis constant (−log K for Tb³⁺ + H₂O ⇌ Tb(OH)²⁺ + H⁺) of Tb³⁺ is 7.60, indicating that radiolabeling should be carried out near pH 7 or lower to avoid extensive formation of unreactive hydroxides [283]. Compared to other lanthanides, Tb³⁺ has a heat of hydration (ΔH° = 4.1 kcal/mol) intermediate for the lanthanide series—consistent with its position in the middle of the series—and should allow for radiolabeling at mild temperatures [282].

6.3. Acyclic Chelators for Terbium

Much of the acyclic chelator chemistry for terbium radiopharmaceuticals has been dominated by DTPA (Figure 22) due to its ability to rapidly chelate Tb³⁺ and form a complex with high thermodynamic stability (log K_ML = 22.71, Table 6) [314]. Although the flexibility of DTPA enables facile chelation of Tb³⁺ at room temperature, this same feature contributes to the poor kinetic stability of its Tb³⁺ complexes. Second generation analogues of DTPA have been investigated to improve upon the deficiencies of this ligand for application in radioactive Tb chelation.

Figure 22.

DTPA and its bifunctional p-SCN-Bn-CHX-A′′-DTPA derivative.

Table 6.

Selected stability constants of Tb–L complexes.

Ion	Chelator	logKML	Ref.
Tb3+	EDTA	17.83	[178]
	DTPA	22.71	[179]
	DOTA	24.2	[315]
	macropa	11.8	[290]
	py-macrodipa	12.0	[296]
	py2-macrodipa	14.8	[292]

DTPA conjugated to octreotide, a somatostatin analogue which binds to receptors commonly overexpressed in neuroendocrine tumors, was radiolabeled with [¹⁶¹Tb]Tb³⁺ using conditions previously employed for [¹¹¹In]In–DTPA-octreotide [316–318]. Using a 40–70-fold molar excess of DTPA-octreotide to [¹⁶¹Tb]TbCl₃, the [¹⁶¹Tb]Tb–DTPA-octreotide complex was formed at 25 °C after 10 min with a RCC value ≥95% [317]. The complex showed serum stability up to 24 h. Unfortunately, no other radiolabeling characterization was performed, but this study serves as one of the first reports of a DTPA-based BFC radiolabeled with ¹⁶¹Tb.

Derivatives of DTPA, including CHX-A′′-DTPA, which rigidifies part of the aliphatic backbone with a cyclohexyl moiety, have also been explored for radiolabeling. The bifunctional derivative p-SCN-Bn-CHX-A′′-DTPA (Figure 22) contains a reactive benzyl thiocyanate functional group tethered to CHX-A′′-DTPA for biological target attachment.

As a proof-of-principle of conjugating this chelator to a macromolecule, one study used this reactive handle to attach the BFC to human serum albumin (HSA) to serve as a model for targeting vectors; mass spectrometry analysis showed ~2 chelators per HSA. The construct CHX-A′′-DTPA-HSA (10⁻⁵ M) was then radiolabeled using [¹⁶¹Tb]TbCl₃ for 1 h at 25 °C [319]. A RCC value of >98% was reported. A control condition of radiolabeling unconjugated HSA showed only 6.0 ± 1.2%, thus indicating that the chelator is required. The [¹⁶¹Tb]Tb-CHX-A′′-DTPA-HSA complex showed good kinetic stability when incubated in phosphate-buffered saline (pH 7.4) at 37 °C with >95% of the complex still intact after 24 h. However, incubating in human serum at 37 °C showed ~10% degradation after 24 h, and an EDTA transchelation challenge (10³-fold excess EDTA) showed >35% transchelation of [¹⁶¹Tb]Tb³⁺. Although CHX-A′′-DTPA-HSA exhibited excellent radiolabeling performance, its limited kinetic stability has motivated investigation of other acyclic chelators.

Other acyclic chelators based on the well-studied HOPO class have also been explored [320]. Two HOPO derivatives, octadentate HOPO-O₈ and decadentate HOPO-O₁₀ (Figure 23), were successfully radiolabeled at 10⁻⁵ M final concentration with [¹⁶¹Tb]TbCl₃ at room temperature for 15 min at pH 6. HOPO-O₈ gave a higher RCC value of 99 ± 1% compared to 88 ± 10% for HOPO-O₁₀. The corresponding AMAs were 81 MBq/nmol (HOPO-O₈) and 85 MBq/nmol (HOPO-O₁₀). Further, the [¹⁵⁵Tb]Tb-HOPO-O₈ and [¹⁵⁵Tb]Tb-HOPO-O₁₀ complexes (using [¹⁵⁵Tb]TbCl₃ due to isotopic availability) showed remarkable kinetic stability with >98% remaining intact after incubation in human serum at 37 °C for 5 days [321]. As such, HOPO-O₈ and HOPO-O₁₀ have arisen as leading chelators for ¹⁵⁵Tb and ¹⁶¹Tb and warrant further investigations with bifunctional-targeted analogues [322].

Figure 23.

Acyclic HOPO-O₈ and HOPO-O₁₀ chelators.

6.4. Cyclic Chelators for Terbium

Bifunctional derivatives of DOTA have frequently been used for ¹⁵⁵Tb and ¹⁶¹Tb radiolabeling because they form thermodynamically stable complexes with this ion (log K_ML = 24.2) [315,323,319]. Only recently, however, was the ¹⁵⁵Tb-radiolabeling efficacy of the nontargeted DOTA chelator examined. Using the same conditions required to quantitatively form [¹⁵⁵Tb]Tb-HOPO-O₁₀ discussed above (pH 6, 80 °C, 15 min), the RCC for labeling DOTA was ~75% at 10⁻⁴ M chelator concentration and only ~25% at 10⁻⁵ M chelator concentration [321].

Several peptide-conjugated derivatives of DOTA have been radiolabeled with [¹⁶¹Tb]Tb³⁺, including DOTA conjugated to HSA [319], GA-HSA [319], folate [323,324], and a short somatostatin tetrapeptide (Phe–D-Trp–Lys–Thr–OMe) [325]. These chelators could be quantitatively radiolabeled at slightly acidic pH and at elevated temperatures (40–90 °C) and show relatively high kinetic stability at 37 °C in either human or fetal bovine serum. Radiolabeling under mildly basic conditions (pH 8) could be performed at high temperatures (90 °C) for long reaction times (90 min) [325].

A smaller macrocyclic chelator NETA, which comprises a 1,4,7-triazacyclonane (TACN) macrocyclic core appended with two carboxylic acids and one aminodicarboxylic acid, and its bifunctional 3p-C-NETA (Figure 24) derivative have been investigated with terbium radioisotopes [326,327]. Compared to other macrocycles, 3p-C-NETA is unique because the aminodicarboxylic acid pendant group is multidentate and serves as an acyclic chelator appended to the macrocyclic chelator core. The incorporation of this moiety allows 3p-C-NETA to benefit from the kinetic radiolabeling advantages of acyclic chelators, while maintaining the thermodynamic stability afforded by macrocyclic chelators [300]. The 3p-C-NETA chelator demonstrated large RCC (>95%) when radiolabeled with [¹⁶¹Tb]TbCl₃ under mild conditions (pH 4.1, 55 °C, 12 min) and showed no significant decomplexation after incubating at 37 °C in phosphate-buffered saline or human serum for 10 days [328]. It was also conjugated to HSA and radiolabeled using [¹⁶¹Tb]TbCl₃ (pH 4.7, 40 °C, 60 min) to give a RCC of >98%, showing compatibility of its mild radiolabeling conditions with conjugated macromolecules [319].

Figure 24.

Cyclic chelators 3p-C-NETA and DOTI-Me.

The cyclen core of the DOTA and NETA chelators have also been functionalized with different pendent arms and investigated with Tb radioisotopes. A cyclen chelator with methylimidazole pendent arms (DOTI-Me, Figure 24), for example, was radiolabeled with ¹⁶¹Tb [329]. The efficiency of this process was relatively poor, as RCC values of <20% and 90.3 ± 9.7% were obtained even when heating at 95 °C at pH values of <5 and 8.2, respectively. The [¹⁶¹Tb]Tb-DOTI-Me complex showed fair kinetic stability with 93% intact after incubating in human serum at 37 °C for 44 h. These results show that by comparison DOTA, which provided pendent carboxylates, is a better chelator for [¹⁶¹Tb]Tb³⁺.

In addition to the aza-crown macrocycles, macrocycles with ethereal oxygen donors have also been radiolabeled with ¹⁶¹Tb. The expanded macrocycle 2,2′,2′′,2′′′-(1,10-dioxa-4,7,13,16-tetraazacyclooctadecane-4,7,13,16-tetrayl)tetraacetic acid (crown, Figure 25), which was originally developed for ²²⁵Ac, has been demonstrated to be effective for ¹⁵⁵Tb and ¹⁶¹Tb [330,331]. Although thermodynamic stability constants for Tb³⁺ complexes of this chelator have not yet been measured, radiolabeling studies of a crown–octreotate conjugate (crown-TATE, Figure 25) with [¹⁵⁵Tb]Tb³⁺ and [¹⁶¹Tb]Tb³⁺ (10⁻⁴ M crown-TATE in NH₄OAc buffer at pH 6.0 for 10 min at room temperature) gave quantitative RCCs. The [^155/161Tb]Tb-crown-TATE complexes showed exceptional kinetic stability with >99% of the complex remaining intact after incubating in human serum at 37 °C for 7 days.

Figure 25.

Cyclic chelator crown and its bifunctional crown-TATE derivative.

Another large macrocycle TRICA (Figure 26), a nonadentate triaza-18-crown-6 with three carboxylate pendants, was recently reported and radiolabeled at room temperature with [¹⁶¹Tb]TbCl₃ to afford a complex in quantitative RCC with an AMA of 1 MBq/nmol [332]. The analogous [¹⁵⁵Tb]Tb-TRICA complex, which was prepared with the same procedure used for [¹⁶¹Tb]TbCl₃, remains 98% intact in human serum at 37 °C over 5 days, reflecting its good kinetic stability. Because Tb³⁺ is paramagnetic, diamagnetic Y³⁺ was used in its place for NMR studies. The ¹H NMR spectrum of a mixture of TRICA (1.1 equivalents) and Y³⁺ reveals that two pendent carboxylates bind to the metal ion from one face and the thirds binds from the opposite, analogous to the coordination mode observed in the crystal structure of [Gd(TRICA)] (Figure 26) [333]. By comparison, the ¹H NMR spectrum of [La(TRICA)] revealed a higher degree of symmetry, consistent with the three carboxylates all binding on the same face. Because Y³⁺ and Gd³⁺ have comparable ionic radii as Tb³⁺ (1.075 Å, 1.107 Å, and 1.095 Å, respectively, CN 9), the NMR and crystal structure studies from these former metal ions provide a good indication of the properties of [Tb(TRICA)] [152].

Figure 26.

Cyclic TRICA chelator and the crystal structure of its [Gd(TRICA)] complex (CCDC Number 1283556) [333].

7. Erbium

Erbium (^natEr) has six naturally occurring, stable isotopes, erbium-162, 164, 166, 167, 168, and 170, with erbium-166 (¹⁶⁶Er) being the most abundant at 33.5%. Synthetically produced erbium-165 (¹⁶⁵Er, t_1/2 10.4 h) emits 7.28 MAEs/decay with a total energy of 7.99 keV/decay as it decays to stable holmium-165. ¹⁶⁵Er is unique because it emits MAEs exclusively, except for a few low-energy X-rays, making its dosimetry easier to determine than for other isotopes. The average MAE energy from ¹⁶⁵Er (7.99 keV) is similar to the well-studied cytotoxic iodine-125 (11.96 keV), indicating that ¹⁶⁵Er likewise has high antiproliferative potential [20,334–338].

7.1. Production of Erbium-165

Direct and indirect production methods for ¹⁶⁵Er have been investigated. Proton or deuteron bombardment of natural holmium (^natHo, monoisotopic in holmium-165, ¹⁶⁵Ho) can produce ¹⁶⁵Er through the ¹⁶⁵Ho(p,n)¹⁶⁵Er and ¹⁶⁵Ho(d,2n)¹⁶⁵Er reactions, respectively [339]. Although numerous studies have experimentally measured the cross sections for ¹⁶⁵Er production, only a few studies have pursued methods to isolate and purify the [¹⁶⁵Er]Er³⁺ from the holmium targets [25,136,334,339–348].

A comparison of proton (16 MeV beam) and deuteron (17.5 MeV beam) bombardment of metallic ^natHo foil targets showed that deuteron bombardment (¹⁶⁵Ho(d,2n)¹⁶⁵Er) yielded larger amounts of activity compared to proton bombardment (¹⁶⁵Ho(p,n)¹⁶⁵Er). However, the deuteron bombardment also led to a higher level of holmium-166, as an undesired byproduct [339]. Separation of ¹⁶⁵Er from the large quantities of ^natHo is challenging because of the similar coordination chemistry of these two adjacent lanthanides. For these production methods, the dissolved target was eluted through a single LaNthanide 2 (LN2) column, giving a relatively poor separation factor of 2.75. Hence, subsequent studies have shifted towards different target materials and the development of more advanced purification methods.

In another study, [^natHo]Ho₂O₃ pellets were prepared and bombarded with a 13.4 MeV proton beam [334]. The targets were then dissolved in HNO₃, and the resulting solution was treated with α-hyodroxyisobutyric, passed through a cation exchange column, and then eluted through a LN3 column with HCl. Although this process afforded a high radionuclidic purity of [¹⁶⁵Er]Er³⁺ (99.9%), a 60-fold excess of ^natEr remained, which was most likely a consequence of it being a contaminant in the [^natHo]Ho₂O₃ target material

The 12.5 MeV proton irradiation of metallic ^natHo targets was investigated. After irradiation, the targets were dissolved in HCl and evaporated to dryness, and the resulting residue was dissolved in HCl and passed through a series of three columns [346]. Similar to earlier methods, the first two columns contained a cation exchange resin and LN2 resin. The final third column used was DGA resin. This resin substantially improved the isolation of [¹⁶⁵Er]Er³⁺, which was obtained in 64 ± 2% yield with a Ho/¹⁶⁵Er separation factor of (2.8 ± 1.1) × 10⁵.

A related procedure for obtaining [¹⁶⁵Er]Er³⁺ followed the same approach described above, but used a different third column [347]. In this case, the last column (Triskem TK221) could release the [¹⁶⁵Er]Er³⁺ with dilute HCl, affording a solution that is ready for radiolabeling without additional processing. This method provided [¹⁶⁵Er]Er³⁺ in 78.0 ± 5.6% radiochemical yield and achieved a very high separation factor of (1.14 ± 0.25) × 10⁶. This purification method was adopted to prepare [¹⁶⁵Er]ErCl₃ for radiolabeling of DOTA-PSMA-617 [136].

The production of ¹⁶⁵Er via the decay of thulium-165 (¹⁶⁵Tm) has also been of substantial interest. In particular, the longer half-life of ¹⁶⁵Tm (t_1/2 = 30.1 h) could be useful in the assembly of a ¹⁶⁵Tm→¹⁶⁵Er generator [344,345,348,349]. ¹⁶⁵Tm can be produced by both the ¹⁶⁶Er(p,2n)¹⁶⁵Tm and ^natEr(d,xn)¹⁶⁵Tm reactions. The use of ^natEr targets bypasses the need for expensive isotopically enriched ¹⁶⁶Er but is limited by the coproduction of long-lived thulium-167 (t_1/2 9.3 days) and thulium-168 (t_1/2 93.1 days) [25,339].

To establish a proof-of-concept for a ¹⁶⁵Tm→¹⁶⁵Er generator, targets prepared from [^natEr]Er₂O₃ deposited onto aluminum backings were irradiated with 60 MeV α-particles [348]. The target was then dissolved in HNO₃, and ytterbium and thulium radioisotopes were affixed to an LN column. After waiting 1 day for ingrowth of the ¹⁶⁵Er from the decay of ¹⁶⁵Tm, [¹⁶⁵Er]Er³⁺ was eluted in 89% radiochemical purity. Because the thulium within this sample comprised a mixture of thulium-166 (t_1/2 7.7 h), thulium-167 (t_1/2 9.3 days), and thulium-168 (t_1/2 93.1 days) in addition to the desired ¹⁶⁵Tm [350], their decay also led to the generation of stable erbium-166, erbium-167, and erbium-168, leading to a decreased specific activity of the therapeutically relevant [¹⁶⁵Er]Er³⁺. Although this study demonstrated a proof-of-principle for a ¹⁶⁵Tm→¹⁶⁵Er generator, the α bombardment on ^natEr reaction used to produce ¹⁶⁵Tm has a substantially lower cross-section than for proton or deuteron bombardment of ^natHo, ^natEr, and enriched ¹⁶⁶Er [349].

7.2. Chemistry of Erbium

Erbium is a heavy lanthanide with electron configuration of [Xe]4f¹²6s². Like the other lanthanides, the coordination chemistry of erbium coordination chemistry is dominated by the +3 oxidation state ([Xe]4f¹¹). Complexes of this ion in the +2 oxidation state can be formed, but only under unique circumstances with rigorously air- and water-free conditions [313]. Er³⁺ has a narrow f-f emission line at 1.54 μm in the near-infrared, which has made it valuable within applications like telecommunications where it can facilitate optical amplification in fiberoptics [351–353]. This near-infrared emission also lies in the tissue-transparent window, and this feature has been leveraged for Er³⁺-based photoluminescent biological probes in vitro and in vivo, as recently demonstrated in a Er³⁺-bacteriochlorin complex [354]. As a heavy lanthanide, Er³⁺ has a smaller ionic radius of 0.89 Å (CN 6) compared to the other lanthanides discussed in this review [152]. Consistent with its smaller radius, Er³⁺ has a lower hydrolysis constant (−log K = 7.46 for Er³⁺ + H₂O ⇌ Er(OH)²⁺ + H⁺) and heat of hydration (ΔH° = 3.1 kcal/mol) compared to La³⁺ (Section 5.2) and Tb³⁺ (Section 6.2) [282,283]. These properties suggest that more acidic conditions, but potentially lower temperatures, should be used for radiolabeling with [¹⁶⁵Er]Er³⁺. Like the other lanthanides, Er³⁺ is a hard Lewis acid.

7.3. Cyclic and Bifunctional Chelators for Erbium

To date, no acyclic chelators have been radiolabeled with ¹⁶⁵Er. The macrocycles DOTA (Figure 4) and crown (Figure 25), and their PSMA-conjugated BFCs, have been radiolabeled with [¹⁶⁵Er]ErCl₃. In an early study, DOTA-PSMA (4 × 10⁻⁵ – 8 × 10⁻⁶ M) was radiolabeled with [¹⁶⁵Er]ErCl₃ (pH 5.7, 80 °C, 30 min), affording RCCs that ranged from 100% at the highest chelator concentration to 49% at the lowest concentration [346]. The highest achievable AMA under these conditions was 82 ± 45 MBq/nmol, and the [¹⁶⁵Er]Er-DOTA-PSMA complex showed good kinetic stability with >95% intact after incubating in human serum at 37 °C for 12 h.

In a subsequent study, [¹⁶⁵Er]ErCl₃-labeling properties were verified, confirming that this radiocomplexation proceeds effectively at high temperature (80 °C), but poorly at lower temperature (37 °C). Collectively, these radiolabeling data are very similar to those for DOTA BFCs and ¹⁷⁷Lu (e.g., ¹⁷⁷Lu-DOTA-PSMA-617 and ¹⁷⁷Lu-DOTATATE with an AMA of 60 MBq/nmol) [355,356], underlying the similar chemistry of these heavy lanthanides [347]. Crown, however, could be radiolabeled at lower concentrations under very mild temperatures (10⁻⁶ M, pH 5.5, 37 °C) with this process being quantitative after 60 min. The crown ligand was conjugated to the octreotide peptide, and the conjugate crown-TATE was radiolabeled with [¹⁶⁵Er]ErCl₃. The radiolabeling proceeded efficiently at the lower-temperature condition (pH 5.5, 37 °C, 30 min), but required a higher chelator concentration (10⁻⁴ M) to give an AMA of 345.6 MBq/nmol. Thus, these radiolabeling studies show that crown is superior to DOTA for [¹⁶⁵Er]Er³⁺, mirroring a similar conclusion for its use with [^155/161Tb]Tb³⁺ (Section 6.4).

Despite the poorer radiolabeling properties of DOTA, it has been used to deliver ¹⁶⁵Er to prostate cancer. A PSMA-targeting bioconjugate DOTA-PSMA-617 BFC (10⁻⁶ M) was quantitatively radiolabeled with [¹⁶⁵Er]ErCl₃ (pH 5.5, 90 °C, 10 min), giving a large AMA value of 348.8 ± 23.3 MBq/nmol [136]. The [¹⁶⁵Er]Er-DOTA-PSMA-617 complex is stable, remaining intact in human serum at 37 °C for 12 h. To assess the PSMA-binding affinity of the radiolabeled construct, it was incubated with PSMA-receptor-expressing LNCaP cells for 4 h. Cellular displacement assays, afforded by challenging the complex with unlabeled PSMA-617, showed that it retained nanomolar affinity for the PSMA.

8. Mercury

Naturally occurring mercury comprises seven stable isotopes, mercury-196, 198, 199, 200, 201, 202, and 204, with mercury-202 (²⁰²Hg) being the most abundant at 29.7%. Also of importance for coordination chemistry, mercury-199 (¹⁹⁹Hg) is an NMR-active nucleus (I = ½, γ = 7.7 MHz·T⁻¹, ~16.9% abundance) [357–360,139].

Two isomers of the synthetic mercury-197 (¹⁹⁷Hg) isotope are amenable for MAE-based therapy. The ground state isomer (^197gHg, t_1/2 2.7 days) and metastable isomer (^197mHg, t_1/2 23.8 h) are separated by 299 keV and emit 23.2 and 19.4 MAEs per decay with total energies of 16.1 and 13.5 keV, respectively. ^197mHg decays to either ^197gHg (branching ratio 91%) or ¹⁹⁷Au (branching ratio 9%), releasing two γ photons at 134 keV (from ^197mHg) and 279 keV (from ^197gHg) that are suitable for SPECT imaging [361]. ¹⁹⁷Hg had been used for imaging since the 1950s and 1960s, in the form of the radiolabeled mercurial diuretic [¹⁹⁷Hg]chloromerodrin. The low specific activities for this compound, however, required chemotoxic doses that led to its removal from the clinic [362–364]. In recent years, more sophisticated productions methods have made high specific activity ¹⁹⁷Hg available, renewing interest in its use for imaging and therapy [365].

8.1. Production of Mercury-197m/g

Proton or deuteron irradiation of natural Au (^natAu), which exists as monoisotopic gold-197 (¹⁹⁷Au), is the most effective way to produce ^197m/gHg, in part because it circumvents the need for enriched targets. Although deuteron irradiation to afford the ¹⁹⁷Au(d,2n)^197m/gHg reaction has higher cross sections, the proton irradiation reaction ¹⁹⁷Au(p,n) ^197m/gHg is more often used due to the fact that most cyclotrons are currently operating with proton beams [366]. Spallation of lead targets and α-particle irradiation of platinum foils (¹⁹⁴Pt(α,n)^197m/gHg and ¹⁹⁵Pt(α,2n)^197m/gHg) have also been used to produce ^197m/gHg [367,368]. Highly pure [^197m/gHg]HgCl₂ can be obtained by chromatographic separation on a LN resin, which is the preferred method over liquid-liquid extractions and other chromatographic separations [369,370].

8.2. Chemistry of Mercury

The most stable oxidation state of mercury under aqueous and aerobic conditions is Hg²⁺ [236,371]. Hg²⁺ is classified as a soft Lewis acid (η = 7.7 eV, Table 2) [233], forms complexes with a wide range of CNs from 2–8, and has a six-coordinate ionic radius of 1.02 Å (Table 2) [152]. The large relativistic effects contribute to the large polarizability of this ion, which in turn lead to more covalent Hg²⁺ ligand bonds [372]. Hg²⁺ forms particularly strong bonds with soft donors such as phosphorous, tellurium, selenium, and sulfur. In fact, the name mercaptan, often used for thiols, originates from the Latin translation of “mercury capturer.” Despite the HSAB preferences of this ion, Hg²⁺ can also form stable complexes with harder donors like nitrogen based amines and heterocycles [360]. Owing to its closed shell electron configuration, Hg²⁺ does not benefit from any ligand-field stabilizing effects and consequently its complexes are typically quite labile, a feature that poses challenges when designing chelators for radiopharmaceutical applications.

In contrast to other divalent metal ions, Hg²⁺ often forms weaker complexes with macrocycles compared to acyclic chelators, as demonstrated by ¹⁹⁹Hg NMR studies [359,372]. The common four-coordinate tetrahedral geometry attained by Hg²⁺ is difficult for small macrocycles to accommodate, and the enthalpic penalty of chelator distortion to encapsulate the large ionic radius outweighs the entropic favorability of preorganization that is characteristic of the macrocyclic effect [372–376]. However, for macrocycles large enough to accommodate the ionic radius of Hg²⁺, stable Hg²⁺ complexes can be formed. Macrocycles can also form particularly stable Hg²⁺ complexes if they contain soft donor atoms (e.g., thiacrowns) to match the soft Lewis acidity of Hg²⁺ which will be discussed in Section 8.4 [377,378].

Several reports have demonstrated the success of sulfur-containing macrocyclic chelators for radiolabeling with ^197m/gHg, as is discussed below. Although ^197m/gHg coordination chemistry is in its infancy for radiopharmaceutical applications, these studies have years of prior work to build upon that was motivated by developing chelators and sensors to mitigate the chemotoxic effects of Hg²⁺.

8.3. Acyclic Chelators for Mercury

Several acyclic chelators for Hg²⁺ have been extensively studied and used as potential antidotes for mercury poisoning. One of the most prominent examples is the dithiol 2,3-dimercaptopropanol, also known as British anti-Lewisite (BAL) or dimercaprol (Figure 27), which was also used as an antidote against the organoarsenic chemical warfare agent Lewisite during World War II [379]. The soft sulfur donors on BAL make it a highly effective chelator for Hg²⁺ (log K_ML = 25.7, Table 7), and consequently it is currently recommended for the treatment of acute mercury poisoning [380,381]. Similarly, the chelators 2,3-dimercapto-1-propane sulfonic acid (DMPS, Figure 27) and 2,3-dimercaptosuccinic acid (DMSA, Figure 27) have been developed and used to treat acute mercury poisoning [382,383]. Like BAL, DMSA forms a remarkably stable complex with Hg²⁺ (log K_ML = 28.5 [384]. To date, acyclic chelators have not been investigated with ^197/197mHg, but their extensive use as mercury poisoning antidotes suggests that they are worth further investigation for radiopharmaceutical applications.

Figure 27.

Hg²⁺ chelators dimercaprol, DMPS (2,3-dimercapto-1-propane sulfonic acid), and DMSA (dimercaptosuccinic acid).

Table 7.

Selected stability constants of Hg–L complexes.

Ion	Chelator	logKML	Ref.
Hg2+	BAL	25.7	[381]
	EDTA	22.07	[381]
	DMSA	28.5	[384]
	DTPA	26.7	[179]
	NS4	20.0	[377]
	cyclen	23.8	[358]
	DOTA	29.0	[358]
	DO4S	32.5	[358]
	DO3S	32.3	[358]
	DO2A2S	31.7	[358]

8.4. Cyclic Chelators for Mercury

The affinity of soft sulfur donors for Hg²⁺ has prompted a number of studies on the coordination chemistry of thiacrown ethers with this ion (Figure 28) [385,386]. Compared to more conventional azacrown ethers used in radiopharmaceutical applications, thiacrown ethers exhibit distinct coordination chemistry with Hg²⁺. For example, six-coordinate complexes of thiacrowns typically form octahedral geometries, whereas the azacrowns typically furnish trigonal prismatic environments, a behavior unique to mercury; Cd²⁺ analogs with azacrowns typically form octahedral complexes [386]. Small thiacrowns, such as 1,4,7-trithiacyclonane (9S3, Figure 28), the thioether equivalent of TACN, consistently form 1:2 coordination complexes of the type [Hg(9S3)₂]²⁺ with a distorted octahedral geometry [386–392]. By contrast, the coordination chemistry of larger thiacrowns is more diverse, affording different products depending on reaction conditions and counteranions. For example, the expanded 14S4 chelator (Figure 28) affords distinct coordination complexes with different salts of Hg²⁺. HgCl₂ gives a dinuclear tetrahedral [(HgCl₂)₂(14S4)]²⁺ complex [393], HgI₂ forms a polymeric Hg-bridged [HgI₂(14S4)]²⁺ complex with tetrahedral symmetry around the Hg²⁺ bridge, and Hg(ClO₄)₂ furnishes a five-coordinate [Hg(14S4)(OH₂)]²⁺ complex of C_4v symmetry with an axial water ligand [393,394]. Hence, the counterions should be taken into consideration when using [^197m/gHg]Hg²⁺ salts for radiolabeling.

Figure 28.

Thiacrown ethers 9S3, 14S4, and NS4 for chelating Hg²⁺.

More recently, a larger cyclic thioether (NS₄, Figure 28) and its coordination chemistry with [^197m/gHg]HgCl₂ and non-radioactive Hg(NO₃)₂ were reported [377]. The secondary amine of this chelator was functionalized with a benzylamide to emulate the effects of conjugation with targeting vectors. Although crystal structures of its Hg²⁺ complex were not obtained, NMR spectroscopy and DFT indicated that all five NS₄ macrocyclic donor atoms, as well as the carbonyl O atom of the benzylamide, directly interacted with the Hg²⁺ ion. Because this chelator bound Hg²⁺ at low pH, direct potentiometric measurements of the thermodynamic stability of its Hg²⁺ complex were not possible, but a minimum stability constant of log K_ML = 20.0 ± 0.05 for Hg²⁺ complex of the NS₄ macrocycle without the benzylamide pendent arm could be determined. Radiolabeling with ^197m/gHg at pH 7 for 1 h demonstrated the successful formation of radiometal complexes. At 80 °C, quantitative incorporation of ^197m/gHg was achieved at NS₄ concentrations of 10⁻⁴ M, decreasing to 96.0 ± 0.3% at 10⁻⁵ M and 31.3 ± 2.0% at 10⁻⁶ M. When radiolabeling was carried out under the milder condition of 37 °C, good RCCs were only obtained when using a high ligand concentration of 10⁻⁴ M. At lower ligand concentrations under these conditions, the RCCs were substantially worse, 33.8 ± 2.1% and 0.0 ± 0.0% at 10⁻⁵ and 10⁻⁶ M ligand concentrations, respectively. To assess kinetic stability, [^197m/gHg]Hg–NS₄ was incubated with a physiologically relevant concentration of glutathione (2.12 mM) at 37 °C [249]. Under these conditions, 97.2 ± 0.3% and 92.8 ± 0.5% of the radiometal complex remained intact after 10 h and 3 days, respectively, signifying its good stability. Upon incubation in human serum at 37 °C, the complex was substantially less stable, as only 84.7 ± 3.4% and 74.1 ± 7.0% did not degrade after 2 and 24 h, respectively. The high stability of this complex under these various conditions supports the viability of NS₄ chelators for Hg-based radiopharmaceuticals, providing an initial ligand scaffold for further optimization.

Following this report, a series of cyclen derivatives, DO4S, DO3S, and DO2A2S, with different numbers of thioether pendent arms was explored for radiolabeling with ^197m/gHg (Figure 29) [358]. Titrations of these chelators in the presence of Hg²⁺ revealed the stability constants to be log K_ML= 32.5 ± 0.1 (DO4S), log K_ML = 32.3 ± 0.1 (DO3S), and log K_ML = 31.7 ± 0.1 (DO2A2S). Notably, these values are all substantially larger than those afforded by DOTA (log K_ML = 29.0 ± 0.1) and cyclen (log K_ML = 23.8 ± 0.1), indicating that the soft thioether donors provide significant stabilization to the soft Hg²⁺ center. In addition, within this series, the presence of more pendent thioethers gives rise to larger stability constants, confirming the importance of matching soft-soft acid-base donor interactions.

Figure 29.

Cyclen derivatives DO4S, DO3S, DO2A2S, and DO3SAm for chelating Hg²⁺.

Although crystal structures of Hg-complexes with this series of ligands were not obtained, NMR characterization (¹H–¹H COSY, ¹H–¹³C HMQC, ¹H–¹⁹⁹Hg HMQC, and NOESY) was used to understand the coordination environment of the Hg²⁺ ion. In the [Hg(DO4S)]²⁺ complex, coupling between ¹H nuclei on the thioether pendant and the central ¹⁹⁹Hg nucleus, measured via ¹H–¹⁹⁹Hg HMQC NMR spectroscopy, revealed that all four sulfurs of the thioether pendants and all four nitrogen atoms of the macrocyclic core participate in complexation, providing an eight-coordinate N₄S₄ environment around the Hg²⁺ center. Within the DO3S chelator, which only contains three pendent thioethers, NMR analysis of [Hg(DO3S)]²⁺ suggests that two thioether arms bind to Hg²⁺ on opposite sides of the macrocyclic core, while the third arm is either uncoordinated or in dynamic exchange with the other bound donors. All four N atoms of the macrocyclic core, however, are coordinated to Hg²⁺, giving a six-coordinate N₄S₂ structure. Finally, the NMR spectra of the mixed carboxylate/thioether complex [Hg(DO2A2S)] reveal all four macrocyclic N atoms and pendent arms to directly coordinate to the Hg²⁺ center to afford an eight-coordinate N₄S₂O₂ structure. Similarly, [Hg(DOTA)]²⁻ exhibits an eight-coordinate N₄O₄ structure as evidenced by NMR spectroscopy. In contrast to the eight-coordinate environment in solution, the solid-state [Hg(H₂DOTA)] crystal structure (Figure 30) reveals the Hg²⁺ center to attain a seven-coordinate geometry, where one of the pendent carboxylates is protonated and engaging in hydrogen bonding with a lattice water molecule. Two of the other carboxylates bind to the metal ion with Hg–O distances of 2.301(2) and 2.425(2) Å, and the fourth carboxylate is protonated, but still interacts with Hg²⁺ over a longer Hg–O distance of 2.929(2) Å. These results highlight differences between solution-state and solid-state chemistry, as well as effects of pH and ligand protonation states, that should be considered in whole when assessing metal complex geometries.

Figure 30.

Crystal structure of [Hg(H₂DOTA)] (CCDC Number 2347201) [358].

Radiolabeling of these chelators with [^197m/gHg]Hg²⁺, which was produced from a ¹⁹⁷Au target by ¹⁹⁷Au(p,n)^197m/gHg, was performed using a relatively high concentration (1.0 × 10⁻³ M) at pH 7 at either room temperature or 50 °C, and RCC values were obtained after incubating for 10 min and 1 h. In addition to the three thioether-appended chelators and DOTA, the N-methyl amido derivative of DO3S (DO3SAm, Figure 29) was studied to investigate the effects of bioconjugation on its radiolabeling efficacy. At room temperature, the RCC of DO3SAm was low at both 10-min and 1-h incubation times, <30% and ~80%, respectively. At 50 °C, however, the RCC increased to 60% after 10 min and >95% after 1 h. The RCC for DO3S remained below 20% and that of DOTA below 1% for all conditions examined. This result was attributed in part to the infancy of ^197m/gHg production and the resulting uncertainty in molar activities and potential for contamination by gold from the ¹⁹⁷Au target. Despite the low and uncertain radiolabeling performance, kinetic stability studies were examined in human serum. The radiolabeled complexes [^197m/gHg][Hg(DO4S)]²⁺, [^197m/gHg][Hg(DO3SAm)]²⁺, and [^197m/gHg(DO2A2S)] all remained ~90% intact after incubating in human serum at 37 °C for 2 h. At longer incubation times, however, significant degradation of all complexes was observed with stabilities of <75% observed after 24 h.

A derivative of DOTA with pendant amides instead of carboxylates (DOTAM, Figure 31) was radiolabeled with [^197m/gHg]Hg²⁺ [395]. The complex of this ligand with non-radioactive mercury has been previously reported, and its crystal structure reveals a six-coordinate environment in which all four macrocyclic nitrogen atoms and two oxygen atoms from adjacent amide pendants coordinate the Hg²⁺ center (Figure 31) [396]. The authors suggested that the softer nature of the amide pendants of DOTAM may provide a stronger binding affinity for Hg²⁺ compared to the harder carboxylate pendants of DOTA [395]. Despite this hypothesis, radiolabeling of DOTAM (10⁻⁴ M) with [^197m/gHg]Hg²⁺ proceeded with a very low RCC of 5.2 ± 1.7% at high temperatures (pH 5, 80 °C, 1 h), marking only a moderate improvement over DOTA. As discussed below, the bifunctional derivative of DOTAM is substantially more effective.

Figure 31.

DOTAM macrocyclic chelator and the crystal structure of its Hg complex (CCDC Number 241674) [396].

An 18-membered thiacrown ether was recently reported and used to examine the effects of pendent arms of varying softness: picolinate (N₂S₄-Pa), pyridyl (N₂S₄-Py), and thioether (N₂S₄-Thio) (Figure 32) [378]. A previously reported crystal structure of Hg²⁺ chelated by the thiacrown macrocycle without pendent arms ([Hg(N₂S₄)]²⁺) showed participation of all six macrocyclic donor atoms in coordination of the Hg²⁺ center [397], reflecting the thiophilicity of this ion [378]. NMR analysis of the complexes of the three chelators with stable mercury revealed substantial differences in their fluxional behavior. For both [Hg(N₂S₄-Py)]²⁺ and [Hg(N₂S₄-Thio)]²⁺, the resonances of the macrocyclic hydrogens appeared as broad peaks within their ¹H NMR spectra at 25 °C. Increasing the temperature to 45 °C led to coalescence of these signals to sharper peaks, revealing the temperature-dependent fluxionality of [Hg(N₂S₄-Py)]²⁺ and [Hg(N₂S₄-Thio)]²⁺ complexes. By contrast, the ¹H NMR spectrum of [Hg(N₂S₄-Pa)] revealed sharp well-resolved resonances for the macrocyclic hydrogens at 25 °C, indicating a greater degree of rigidity. DFT geometry optimizations of the Hg²⁺ complexes of these ligands were also investigated. Within these computations, the [Hg(N₂S₄-Thio)]²⁺ complex attained a five-coordinate, distorted trigonal pyramidal structure, whereas both [Hg(N₂S₄-Py)]²⁺ and [Hg(N₂S₄-Pa)] adopted six-coordinate, distorted octahedral geometries. In contrast to the [Hg(N₂S₄)]²⁺ crystal structure [397], these DFT calculations predicted Hg²⁺ cannot interact equally with all six macrocyclic donor atoms when the pendent arms are present [378]. Consequently, the DFT-optimized structures of these three ligands show that Hg²⁺ is situated asymmetrically in the macrocyclic core of [Hg(N₂S₄-Thio)]²⁺ and [Hg(N₂S₄-Py)]²⁺, a feature that is corroborated experimentally by NMR spectroscopy. The strong coordination of the picolinate arms in [Hg(N₂S₄-Pa)] is consistent with its more symmetric NMR spectra and was rationalized to be a result of the electrostatic interaction between the negatively charged picolinate pendants and the Hg²⁺ center.

Figure 32.

Thiacrown ether N₂S₄ with different pendent arms.

These structural differences were reflected by the radiolabeling performance of the three chelators. When radiolabeled with [^197m/gHg]Hg²⁺ (pH 7, 80 °C, 1 h), N₂S₄-Py and N₂S₄-Thio quantitatively formed complexes with this radiometal at chelator concentrations of 10⁻⁴ M. The picolinate ligand N₂S₄-Pa, however, gave a RCC value of 26 ± 1% at the same concentration under the same conditions. Notably, radiolabeling of N₂S₄-Thio occurred quantitively even at room temperature (25 °C) and 10⁻⁴ M concentration. From these results, the authors suggested that the softer Lewis base character of the thioether pendent arm contributes to its enhanced affinity for [^197m/gHg]Hg²⁺. The kinetic stabilities of [^197m/gHg][Hg(N₂S₄-Py)]²⁺ and [^197m/gHg][Hg(N₂S₄-Thio)]²⁺ in human serum at 37 °C were moderate with 71 ± 1% and 76 ± 2% of these complexes remaining intact, respectively, after 1 h. [249]. Another stability study was performed by challenging these complexes with a 10⁶-fold excess mixture of biologically relevant metal ions, ZnCl₂, FeCl₃, CuCl₂, MgCl₂, and CoCl₂. These studies confirmed the good selectivity of the N₂S₄ macrocycle for Hg²⁺, as less than 30% of either underwent transmetallation after 82 h under these conditions. [398]. Although the authors concluded that their previously reported, smaller NS₄ chelator is better suited for [^197m/gHg]Hg²⁺ radiolabeling [377], they suggest that the larger N₂S₄ chelator may be suitable for other soft radiometals such as [¹¹¹Ag]Ag⁺, [¹⁹⁹Au]Au⁺, and [²⁰¹Tl]Tl⁺ [378].

8.5. Bifunctional Chelators for Mercury

There have been several examples of ^197m/gHg bifunctional chelators reported in the literature. An early study in 1992 investigated a trithiamacrocycle with a pendent activated ester moiety (9S3-ester, Figure 33) [399]. Following radiolabeling with [^197m/gHg]Hg²⁺, the complex was conjugated to a rabbit immunoglobulin protein. This bioconjugate was stable in saline for over 24 h but was substantially less stable in serum.

Figure 33.

Trithiamacrocycle 9S3-ester radiolabeled with [^197m/gHg]Hg²⁺.

DOTAM (Figure 31) and the common bifunctional chelators p-SCN-Bn-DOTA and p-SCN-Bn-DOTAM (Figure 34) were also investigated with ^197m/gHg [395]. In comparison to the Hg²⁺ complex of the non-functionalized DOTAM discussed in the previous section, the ¹H NMR spectrum of [Hg(p-SCN-Bn-DOTAM)]²⁺ showed a sharpening, splitting, and downfield shift of the aliphatic peaks relative to the free ligand, features which the authors attributed to greater structural rigidity of the complex. This conclusion was further supported by variable temperature and 2D NMR studies.

Figure 34.

Bifunctional chelators p-SCN-Bn-DOTA and p-SCN-Bn-DOTAM.

The p-SCN-Bn-DOTA and p-SCN-Bn-DOTAM BFCs were then radiolabeled with [^197m/gHg]Hg²⁺. In comparison to DOTA, the p-SCN-Bn-DOTA BFC could successfully be radiolabeled with [^197m/gHg]Hg²⁺, achieving a RCC of 70.0 ± 1.1%, under identical conditions (10⁻⁴ M, pH 5, 80 °C, 1 h) that fail for DOTA. Likewise, p-SCN-Bn-DOTAM yielded the ^197m/gHg complex with quantitative RCC under the same conditions, whereas the non-functionalized DOTAM only achieved a RCC of 5.2 ± 1.7%. Furthermore, radiolabeling of p-SCN-Bn-DOTAM could be achieved at lower chelator concentrations (10⁻⁵ M, RCC = 91.6 ± 0.4%) and over shorter times (10⁻⁴ M, 10 min, RCC = 85.8 ± 4.7%). Given the unexpectedly improved radiolabeling properties of p-SCN-Bn-DOTA and p-SCN-Bn-DOTAM compared to the non-functionalized ligands, the authors hypothesized that the S atom in the –NCS could stabilize the thiophilic Hg²⁺ ion. This hypothesis was tested by using phenylisothiocyanate, which lacks a macrocyclic chelating agent. Surprisingly, this compound could be radiolabeled by ^197m/gHg with a RCC of 85.7 ± 3.7%. This result suggests that the –NCS groups within p-SCN-Bn-DOTA and p-SCN-Bn-DOTAM are not innocent and likely play a significant, but as of yet unknown, role in binding to [^197m/gHg]Hg²⁺.

To investigate p-SCN-Bn-DOTAM as a bifunctional chelator, it was conjugated to the PSMA-617 vector, and the resulting DOTAM-PSMA-617 was then radiolabeled. The RCC value for [^197m/gHg][Hg(DOTAM-PSMA-617)]²⁺ was reasonably high, 91.3 ± 3.8%, but significantly lower than the quantitative RCC attained by p-SCN-Bn-DOTAM. These results may suggest that the thiourea resulting from the conjugation of the isothiocyanate group is less effective at binding to the Hg²⁺ ion compared to the -NCS sulfur atom on p-SCN-Bn-DOTAM.

Both [^197m/gHg][Hg(p-SCN-Bn-DOTAM)]²⁺ and [^197m/gHg][Hg(DOTAM-PSMA-617)]²⁺ were incubated (37 °C) in glutathione (50 mM) and human serum to assess their kinetic stabilities. [^197m/gHg][Hg(p-SCN-Bn-DOTAM)]²⁺ was moderately stable in the presence of glutathione (85.9 ± 4.5% intact after 24 h), but underwent significant degradation in human serum with only 67.5 ± 2.9% remaining intact after 24 h. The bioconjugated complex [^197m/gHg][Hg(DOTAM-PSMA-617)]²⁺ was even less stable, with only 58.6 ± 9.6% and 59.6 ± 0.9% of it remaining intact in the presence of glutathione and human serum after 24 h. The poorer stability of the PSMA-617 bioconjugate compared to p-SCN-Bn-DOTAM correlated with their radiolabeling efficacies, further implicating a stabilizing effect afforded by the isothiocyanate group.

9. Platinum

Platinum has five naturally occurring stable isotopes, platinum-192, 194, 195, 196, and 198, with platinum-194 (¹⁹⁴Pt) and platinum-195 (¹⁹⁵Pt) being the most abundant at 32.9% and 33.8%, respectively. Three radioisotopes of platinum have received interest for Auger therapy: platinum-191 (¹⁹¹Pt, t_1/2 2.8 d), platinum-193m (^193mPt, t_1/2 4.3 d), and platinum-195m (^195mPt, t_1/2 4.0).

¹⁹¹Pt decays via electron capture to stable iridium-191, emitting 13.27 MAEs and 0.99 CEs per decay with respective average energies of 17.8 and 57.1 keV. This radioisotope has a small branching ratio for γ emission as well (82 keV (4.9%), 172 keV (3.5%), and 360 keV (6.0%)) that make it amenable to SPECT imaging [400]. Despite the potential of this radioisotope, ¹⁹¹Pt has only been studied in several studies, many of which use ¹⁹¹Pt as [¹⁹¹Pt]cisplatin for tracking the in vivo biodistribution of this drug [400–407].

^193mPt emits 27.39 MAEs and 2.99 CEs with respective average energies of 10.9 and 126.8 keV through its decay to radioactive ¹⁹³Pt, which subsequently transforms via electron capture to iridium-193 (¹⁹³Pt(ε)¹⁹³Ir, t_1/2 50 years). ^193mPt emits very few γ photons at an energy suitable for imaging (136 keV, 0.11%). The large number of MAEs per decay, however, has generated attention for its use as a therapeutic radioisotope for MAE-based TRT.

^195mPt emits 36.55 MAEs and 2.78 CEs per decay with respective energies of 23.1 and 161.4 keV, decaying to stable ¹⁹⁵Pt. Of the platinum isotopes discussed here, ^195mPt also emits the largest amount of γ photons (99 keV, 11.4%) suitable for SPECT imaging.

Radioplatinum chemistry has been dominated by radiolabeled cisplatin (cis-[Pt(NH₃)₂Cl₂]) and its derivatives. Since an initial study in 1967 using cis-[¹⁹¹Pt]Pt(NH₃)C1₄ to examine the effects of radiation delivered to bacteria and the first report of [^195mPt]cisplatin in a 1974 clinical study, radio-cisplatin has been the focus of numerous studies both in the clinic and in academic research [400–403,408–411]. Cisplatin forms covalent crosslinks with DNA [412,413], and thus radiolabeled derivatives have the potential to deliver short-range MAEs directly to genomic DNA. [400–402,409,411,414].

9.1. Production of Platinum-191, 193m, and 195m

¹⁹¹Pt has been produced since the 1960s by proton bombardment of natural iridium (^natIr, iridium-191 (37.3%) and iridium-193 (62.7%)) targets through the ^natIr(p,xn)¹⁹¹Pt reaction, and this approach is still the primary method for making it [406,408]. ¹⁹¹Pt is then isolated from the iridium target by dissolving in HCl/KCl solution, furnishing [¹⁹¹Pt]PtCl₄²⁻ in 90–95% radionuclidic purity. As of 2021, ¹⁹¹Pt has also been produced at the CERN MEDICIS facility and is separated from other radionuclides by mass separation [415]. Although this method has proven useful for fundamental research with ¹⁹¹Pt, the ability to generate it from ^natIr targets on 30 MeV cyclotrons makes it a promising isotope for therapy, based on its potential for large-scale production.

^193mPt was first produced by α-particle bombardment on natural osmium (40.8% osmium-192 (¹⁹²Os)) through the ¹⁹²Os(α,3n)^193mPt reaction [411]. Decades later, this method remains the most successful that has been optimized by using enriched ¹⁹²Os electrodeposited onto nickel targets [416,417]. The cross-sections for the ¹⁹²Os(α,3n)^193mPt are optimized at α-particle energies from 24–40 MeV. Lower energy α-particle energies of 18–24 MeV afford higher cross section for the ¹⁹²Os(α,n)^195mPt reaction, thus demonstrating that the same target material can be used for producing both of these radioisotopes [416].

9.2. Chemistry of Platinum

Platinum is a third-row transition metal with a ground state electron configuration of [Xe]4f¹⁴5d⁹6s¹. Its coordination chemistry is dominated by the Pt²⁺ ([Xe]4f¹⁴5d⁸) and Pt⁴⁺ ([Xe]4f¹⁴5d⁶) oxidation states. Pt²⁺ has an ionic radius of 0.6 Å (CN 4) and a d⁸ configuration that leads it to adopting four-coordinate, square planar geometries in almost all of its complexes [152]. As a soft Lewis acid (η = 8.0 eV, Table 2) [153], it forms stable bonds with sulfur-based donors like thiols. By contrast, the d⁶ ion Pt⁴⁺ has an ionic radius of 0.625 Å (CN 6) [152] and preferentially adopts octahedral geometries.

Platinum has a long history within medicinal chemistry, and an even longer history in establishing the basic principles of coordination chemistry [414,418]. The antiproliferative properties of the Werner coordination complex cisplatin were first reported in 1965, and approved for clinical use in 1978 [419,420]. Since then, several additional Pt²⁺-based drugs have been approved worldwide, Pt⁴⁺-based drug candidates have also been extensively studied [413]. Thus, many efforts in Pt radiochemistry have been focused on making radioactive analogues of these widely used drugs, as discussed above. In addition to these studies briefly discussed in Section 9, several acyclic chelators and BFCs have also been investigated with radioisotopes of Pt.

9.3. Acyclic and Bifunctional Chelators for Platinum

One of the first acyclic chelators radiolabeled with ¹⁹¹Pt was a bidentate 1-(pyridin-2-yl)methanamine group that was attached to an estrogen agonist [404]. Although radio-HPLC confirmed labeling, comparative structural studies using non-radioactive Pt were not carried out, and therefore the resulting coordination mode in this complex is not known. Another acyclic chelator, ethylenediamine, was attached to an antibody and radiolabeled with [¹⁹¹Pt]PtCl₄²⁻ to track its biodistribution [421].

Based on the intrinsic DNA-targeting properties of cisplatin, [^189/191Pt]cisplatin was synthesized and demonstrated to induce DNA double-strand breaks in cells via its MAE emissions, even though only ~2% of this compound localized to the nucleus [422]. To achieve better nucleus localization, the DNA-targeting dye Hoechst33258 was conjugated to either DTPA or cysteine to bind to [¹⁹¹Pt]PtCl₄²⁻ and consequently radiolabel the Hoechst33258 conjugate [406]. Although the high denticity of DTPA makes it an unlikely candidate for binding Pt²⁺, radiolabeling of this free chelator with [¹⁹¹Pt]PtCl₄²⁻ at 60 °C for 1 h gave RCCs of 70–80%, and the resulting complex was fairly stable in serum with 42% of the radioactive components shown to bind to proteins. In addition, the isolated amino acid cysteine also could be radiolabeled and afford stable complexes of ¹⁹¹Pt. Consequently, both the DTPA-Hoechst33258 and cysteine-Hoechst33258 could furnish ¹⁹¹Pt complexes albeit with low RCCs after high-temperature labeling. These complexes, along with the ¹¹¹In analogue [¹¹¹In]In-DTPA-Hoechst33258, were investigated in U2OS cells for their abilities to induce DNA double strand breaks. Overall, these results showed that the ¹⁹¹Pt-labeled complexes afforded more DNA double strand breaks, which may be a consequence of the greater MAE/decay of ¹⁹¹Pt compared to ¹¹¹In.

Building upon this strategy, another DNA-targeting motif pyrrole–imidazole polyamide (PIP), which binds specifically to the MYCN oncogene, was radiolabeled with ¹⁹¹Pt [407,423]. This compound was constructed by combining the PIP group, a cysteine residue for [¹⁹¹Pt]Pt²⁺ radiolabeling, a tri-arginine peptide for cell-penetration, and a coumarin dye to enable tracking via fluorescence microscopy. A BFC with a MYCN vector truncated to half of the MYCN chain (GCC-PIP, Figure 35) was also prepared as a control with lower affinity for the target.

Figure 35.

Bifunctional polypeptide chelator GCC-PIP. The Pt-binding motif is highlighted in blue.

Both compounds were radiolabeled with 4–96 MBq [¹⁹¹Pt]Pt²⁺ in DMSO containing Tween-80 and ethanol at 45 °C for 60 min to obtain the complexes with 50–70% RCCs. Following HPLC to obtain the complexes with 95% radiochemical purity, they were incubated with neuroblastoma cell lines to assess their biological effects. As expected, the targeted complex [¹⁹¹Pt]Pt-MYCN-PIP was taken up by these cells and accumulated in the nucleus to a much larger extent than the untargeted analogue [¹⁹¹Pt]Pt-GCC-PIP. Consequently, [¹⁹¹Pt]Pt-MYCN-PIP also exhibited greater cytotoxicity and induced more DNA double strand breaks than [¹⁹¹Pt]Pt-GCC-PIP, and its biological activity correlated with MYCN expression levels in a panel of several cell lines. Although these in vitro results were promising, in vivo biodistribution of [¹⁹¹Pt]Pt-MYCN-PIP in mice bearing Kelly tumor xenografts showed poor tumor uptake and accumulation in the lungs and spleen, indicating that further modifications are needed to optimize these complexes [407].

10. Thallium

Thallium has two naturally occurring stable isotopes, thallium-203 (²⁰³Tl, 29.5%) and thallium-205 (²⁰⁵Tl, 70.5%). The coordination chemistry of thallium can be investigated by NMR as ²⁰⁵Tl is an NMR-active nucleus (I = ½, γ = 24.9 MHz·T⁻¹) and has been used to characterize ligand binding as well as speciation given the large difference in Tl⁺ and Tl³⁺ chemical shifts [139,424,425]. The synthetic isotope thallium-201 (²⁰¹Tl, t_1/2 72.9 h, 20.91 MAEs/decay, 14.9 keV) decays by EC, emitting a γ photon at 167 keV that makes it suitable for SPECT imaging [426]. Based on its valuable MAE and γ emissions, it has been used clinically since the 1970s in the form of [²⁰¹Tl]TlCl for myocardial perfusion imaging [427,428]. Despite several advantages, [²⁰¹Tl]TlCl has been largely replaced by technetium-based imaging agents. Shortages in reactor-produced molybdenum-99 (⁹⁹Mo), needed for ^99mTc, over the last decade, led to a brief resurgence in its use [428]. More recently, the therapeutic potential of ²⁰¹Tl, based on its large MAE yield, has prompted further investigation of this isotope.

10.1. Production of Thallium-201

The earliest reports of ²⁰¹Tl production used high-energy (400 MeV) protons to bombard mercury-202 (²⁰²Hg) targets to initiate the ²⁰²Hg(p,2n)²⁰¹Tl reaction [429]. Lower proton energies of ~30 MeV can be used to produce ²⁰¹Pb, which decays to ²⁰¹Tl with a 9.4-h half-life, from enriched ²⁰³Tl via the ²⁰³Tl(p,3n)²⁰¹Pb reaction [430]. This production route is one of the most commonly used. The most typical production route involves 30 MeV proton irradiation of copper targets electroplated with enriched ²⁰³Tl followed by extraction and/or chromatographic purification [431,432]. In these cases, residual ²⁰³Tl from the target is separated from Pb isotopes on a column, and after ~32 h (~3.4 half-lives of ²⁰¹Pb), the ²⁰¹Tl formed from the decay of ²⁰¹Pb is eluted and purified by anion exchange chromatography to remove remaining Pb isotopes [431].

10.2. Chemistry of Thallium

Thallium exists in its Tl⁺ ([Xe]4f¹⁴5d¹⁰6s²) and Tl³⁺ ([Xe]4f¹⁴5d¹⁰) states. The standard reduction potential for the Tl³⁺/Tl⁺ couple is +1.25 V, indicating that Tl³⁺ is a powerful oxidant [433]. In addition, Tl³⁺ is also susceptible to hydrolysis, precipitating as solid Tl₂O₃ at pH 1 [425,434]. Therefore, highly effective chelators are needed to stabilize Tl³⁺ under biological conditions. The HSAB properties of thallium are dependent on its oxidation state with Tl³⁺ classified as a hard acid (η = 10.4 eV, Table 2) and Tl⁺ as a soft acid (η = 7.16 eV, Table 2) [153]. In addition, the ionic radii difference between these two ions is substantial; Tl⁺ has an ionic radius of 1.5 Å (CN 6), whereas Tl³⁺ has a six-coordinate ionic radius of 0.885 Å (Table 2). Therefore, it is apparent that very different chelators would be required for Tl³⁺ compared to Tl⁺. For example, these metal ions have very different log K_ML values with the ubiquitous chelators EDTA and DTPA (Table 8) [152,435]. The stability constants of Tl⁺ and Tl³⁺ complexes with EDTA and DTPA, both chelators containing hard nitrogen and oxygen donors, show substantially greater stability with the harder Tl³⁺ ion compared to Tl⁺ [178,179]. In general, the large ionic radius and low ionic charge of Tl⁺ gives rise to substantially less stable coordination complexes compared to those of Tl³⁺. Consequently, researchers have optimized procedures to selectively generate this radioisotope in the +3 oxidation state under mild conditions using Pierce iodination beads [43]. For example, [²⁰¹Tl]TlCl₃ for radiolabeling has been prepared by placing 100 μL [²⁰¹Tl]TlCl (40 MBq) in a plastic tube with one iodination bead and 10 μL 0.5 M HCl, then vortexing for 10 s [436].

Table 8.

Selected stability constants of Tl–L complexes.

Ion	Chelator	logKML	Ref.
T1+	EDTA	6.5	[178]
	DTPA	5.97	[179]
Tl3+	EDTA	37.8	[178]
	DTPA	10.9	[179]

10.3. Acyclic Chelators for Thallium

Both EDTA and DTPA have been successful radiolabeled with [²⁰¹Tl]TlCl₃ [437]. A recent investigation explored these radiolabeling reactions in more detail, showing that they can be carried out under mild conditions. The incubation of EDTA (2.4 mM) or DTPA (1.8 mM) with 3 MBq [²⁰¹Tl]TlCl₃ in 0.25 M NH₄OAc buffer (pH 5) at room temperature for 5 min is sufficient to form complexes with RCCs of >95% [43]. Both the [²⁰¹Tl]Tl–EDTA and [²⁰¹Tl]Tl–DTPA complexes, however, showed poor kinetic stability in human serum, undergoing complete decomposition after 24 h.

More recently, a series of acyclic chelators containing picolinates, H₄pypa, H₅decapa, H₄neunpa-NH₂, and H₄noneunpa (Figure 36), have been investigated with ²⁰¹Tl. Although they could not effectively complex the monovalent form [²⁰¹Tl]TlCl, they were all found to be good chelators for the trivalent [²⁰¹Tl]TlCl₃ [436,438–441]. The four chelators could be radiolabeled with 5–10 MBq [²⁰¹Tl]TlCl₃ in water (pH 5) at room temperature for 10 min, achieving RCCs of >95% [436]. Their kinetic stabilities were assessed in human serum at 37 °C, revealing them to remain between 55–80% intact after 48 h. Insight into the moderate stability of these complexes was gleaned from the crystal structure of the cold thallium (^natTl) complex [^natTl][Tl(Hpypa)]. Within this structure, eight of the nine donor atoms of pypa coordinate the Tl³⁺ center, conferring it with an N₅O₃ environment in a distorted square antiprismatic geometry (Figure 37). The final potential donor atom, a carboxylate, is protonated, preventing its interaction with the Tl³⁺ center. The authors suggested the lability of this fourth carboxylate increases the vulnerability of Tl³⁺ towards reduction to Tl⁺ by decreasing electron density on the Tl³⁺ center. The proposed reduction to Tl⁺ as a decomposition pathway was validated by in vitro and in vivo studies that implicated the monovalent ion exchanger Na⁺/K⁺-ATPase pump in the observed biological properties of the radiothallium. Consequently, these chelators still require additional optimization for use with ²⁰¹Tl and must meet the particular challenge of stabilizing the +3 oxidation state.

Figure 36.

Picolinate-containing acyclic chelators for thallium.

Figure 37.

Crystal structure of [Tl(Hpypa)] (CCDC Number 2115641) [436]. O4 remains protonated and prevents its interaction with the Tl³⁺ metal center. The TlCl₃ coordinated to O3 is omitted for clarity.

10.4. Cyclic Chelators for Thallium

DOTA has been radiolabeled with [²⁰¹Tl]TlCl₃ under several different conditions. In an early study, [²⁰¹Tl]TlCl₃ was incubated with H₄DOTA (~2 mM) at 18 °C for 90 min after flushing with ozone to form [²⁰¹Tl]Tl³⁺, giving an RCC value of ~70% [437]. The complex showed moderate stability with ~75% remaining intact after incubation in human serum for 24 h at 37 °C. For comparison, a recent study carried out radiolabeling in the absence of added ozone, by combining 2.5 mM H₄DOTA in NH₄OAc buffer (pH 5) and 3 MBq [²⁰¹Tl]TlCl₃ at room temperature for 60 min [43]. Although a higher RCC (>95%) was achieved, the complex formed under these conditions showed lower stability in human serum with only 78 ± 12% remaining after 1 h. Thus, it is likely that the ozone present in the reaction mixture plays an important role in stabilizing Tl³⁺ within the DOTA complex.

To gain a deeper insight on this relevant Tl³⁺ coordination chemistry, non-radioactive [^natTl][Tl(DOTA)]⁻ was characterized by NMR spectroscopy and single-crystal X-ray diffraction [442]. The crystal structure of [C(NH₂)₃][Tl(DOTA)]·H₂O shows an eight-coordinate Tl³⁺ supported by an N₄O₄ set of donor atoms (Figure 38). This Tl³⁺ DOTA complex attains the twisted square antiprismatic (TSAP′) geometry, comprising the Λ(λλλλ)/Δ(δδδδ) enantiomeric pair [443]. This structure is similar to the TSAP′ geometry of [Bi(DOTA)]⁻ and in contrast to the square antiprismatic (SAP) geometry of [Lu(DOTA)]⁻ [444,445].

Figure 38.

Crystal structure of [Tl(DOTA)]⁻ (CCDC Number 1025647) [442]. The unit cell contains both Λ(λλλλ) and Δ(δδδδ) enantiomers; only the Δ(δδδδ) enantiomer is shown here for clarity.

The conformation of this complex in the solution state was also verified by multinuclear and multidimensional NMR spectroscopy. Furthermore, the dynamics of the complex in interconverting between the enantiomeric forms (Λ(λλλλ) ⇌ Δ(δδδδ)) was examined by variable-temperature ¹H and ¹³C NMR spectroscopy, along with 2D ¹H–¹H COSY and ¹H–¹³C HSQC NMR spectroscopy [446]. The clearest example of this interconversion is seen by the ¹³C peaks for the two chemically nonequivalent carbon atoms [442]. At low temperature (272 K), there are two distinct ¹³C doublets at 49.2 and 53.6 ppm corresponding to the two macrocyclic carbons. The peaks are split into doublets due to ²⁰⁵Tl coupling with respective ²J_13C–205Tl coupling constants of 156 and 161 Hz. As the sample is heated, these ¹³C NMR resonances coalesce at ~315 K, resulting in a single ¹³C–²⁰⁵Tl doublet that signifies a rapid interconversion between the Λ(λλλλ) and Δ(δδδδ) enantiomers.

Line-shape analysis of the variable-temperature ¹³C NMR signals for these two macrocycle carbon atoms was used to generate an Eyring plot to afford the activation enthalpy (ΔH^‡) of interconversion. The ΔH^‡ for [Tl(DOTA)]⁻ (66 kJ/mol) is between those for [Bi(DOTA)]⁻ (40 kJ/mol) and [Lu(DOTA)]⁻ (101 kJ/mol). Although there are large differences in activation enthalpies, the small entropic component for [Tl(DOTA)]⁻ and negative entropic component for [Bi(DOTA)]⁻ give similar ΔG^‡ values for [Tl(DOTA)]⁻, [Bi(DOTA)]⁻, and [Lu(DOTA)]⁻ (59, 63, and 66 kJ/mol, respectively, at 298 K) [442,444,445]. DFT analysis of the Λ(λλλλ) ⇌ Δ(δδδδ) interconversion suggests a two-step process of ring inversion (ΔG^‡ = 60.8 kJ/mol) and a rate-limiting arm rotation (ΔG^‡ = 67.5 kJ/mol) [442]. Although this high activation energy does limit the interconversion to an extent, the ability of this interconversion to occur at moderate temperatures (NMR resonances coalescence at ~315 K, ~42 °C) may explain the limited kinetic stability at 37 °C [43]; during the ring inversion and arm rotation, the macrocycle N donor atoms and pendent arm O donor atoms elongate, as calculated by DFT, suggesting transiently weakened stability of the complex [442]. Although interpretation of these differences in enthalpic and entropic components of the interconversion across the series of three metal ions was not provided in the original study, it suggests that careful analysis of chelator complex stability is necessary, especially when using chelators for a broad scope of metal ions.

Another class of macrocyclic chelators recently explored for [²⁰¹Tl]Tl³⁺ are the texaphyrins [447]. Texaphyrins provide an N₅ coordination environment, but with a cavity that is ~20% larger than conventional porphyrins. Consequently, texaphyrins can successfully form stable 1:1 complexes with large di- and trivalent metal ions, such as Bi³⁺ and La³⁺ [448,449]. Given the good affinity of texaphyrins for large ions [450], as well as their intrinsic tumor-targeting properties [451], thallium complexes of a micelle-assembling texaphyrin-lipid conjugate were investigated [447]. Spectroscopic analysis verified that Tl⁺ did not form any complexes with the texaphyrin but that Tl³⁺ was readily incorporated. Following this initial investigation with cold thallium, texaphyrin-lipid conjugate was radiolabeled with [²⁰¹Tl]TlCl₃, achieving a RCC of 25.5 ± 4.5%. The stability of the resulting radiothallium complex was lacking; it degraded when heated to 37 °C, most likely due to acid-catalyzed decomposition of the imine macrocycle [447,452]. The [²⁰¹Tl]Tl³⁺-labeled texaphyrin-lipid complexes were then self-assembled into micellar nanoparticles, with the addition of other lipids (DPPC, cholesterol, and DSPE-PEG₂₀₀₀), for in vitro cell uptake studies. These constructs exhibited poor stability with only 29.1 ± 7.3% of them remaining intact after 72 h at 37 °C and 56.0 ± 1.4% remaining intact at 4 °C over the same time period [447]. It is unclear if degradation was from disassembly of the micelle or decomplexation of [²⁰¹Tl]Tl³⁺. Although this texaphyrin analogue was not suitable for [²⁰¹Tl]Tl³⁺, other expanded porphyrins or porphyrinoids may be more effective.

As noted above, reduction of Tl³⁺ is a common decomposition pathway of its coordination complexes. Tl⁺ is a weak Lewis acid that typically forms poorly stable complexes because of its low charge and large ionic radius. However, several studies have validated the potential of crown ethers for binding to Tl⁺ [453]. For example, benzo-crown ethers show a greater relative affinity for Tl⁺ over alkali metal ions compared to aliphatic crown ethers. Furthermore, the use of crown ethers with larger binding pockets is another approach to achieve stronger binding to the large Tl⁺ [435,453].

Building upon simple crown ethers, cryptands are multicyclic analogues that provide additional donor atoms to stabilize metal ions. Given the efficacy of cryptands for binding to large monovalent cations, Kryptofix 222 (K222, Figure 39) analogues were studied for their abilities to bind to [²⁰¹Tl]Tl⁺ [454]. Although radiolabeling standard crown ethers, like 18-crown-6, dibenzo-18-crown-6, and dicyclohexyl-18-crown-6, with [²⁰¹Tl]Tl⁺ were unsuccessful, the bicyclic, K222-based cryptand K222-BFC (Figure 39) afforded a [²⁰¹Tl]Tl⁺ complex with >95% RCC via a 15 min room temperature incubation of the ligand (125–500 μM final concentration) and [²⁰¹Tl]TlCl (~1 MBq) in pH 11 water. The highly basic conditions used for radiolabeling were required because the K222 ligand could be protonated even under neutral pH conditions. Although the crystal structure of the Tl⁺ complex could not be obtained, DFT calculations predicted that its structure possesses an eight-coordinate symmetrically situated Tl⁺ center.

Figure 39.

Cryptand K222 and its bifunctional derivatives used for radiolabeling with [²⁰¹Tl]Tl⁺.

A bifunctional K222-PSMA derivative (Figure 39) was also studied. Radiolabeling of this analogue using identical conditions as the free cryptand yielded the complex with >98% RCC. To evaluate its kinetic stability under physiological conditions, the [²⁰¹Tl][Tl(K222-PSMA)]⁺ complex was incubated in human serum at 37 °C. Initially, 10% of the ²⁰¹[Tl]Tl⁺ was lost, but no further decomplexation was observed after 2 h, boding well for its use in vivo. Despite these promising results, an in vitro study looking at the PSMA-mediated receptor uptake of [²⁰¹Tl][Tl(K222-PSMA)]⁺ in PSMA+ cells was less successful; similar uptake of the K222-PSMA-617 complex and free [²⁰¹Tl]TlCl were observed for both PSMA+ and PSMA− cell lines, suggesting that [²⁰¹Tl]Tl⁺ dissociates from the cryptand under these conditions.

11. Conclusion and Outlook

The development of new TRT pharmaceuticals is a rapidly expanding field that requires significant interdisciplinary contributions from oncologists, medical physicists, biologists, chemists, and other specialized roles [2]. In this context, coordination chemistry plays a critical role by guiding efforts to develop chelators that can enable the use of different therapeutic radiometal ions, including those with novel radionuclides and distinct types of radioactive emissions.

Currently, the only chelator that is applied for FDA-approved TRT therapeutics is DOTA. Although it is suitable for ¹⁷⁷Lu, its efficacy for a wider range of radiometal ions with disparate chemical properties is often not as promising. As explored in this review, metal ions with radioisotopes that are suitable for MAE therapy often have underexplored coordination chemistry. As such, there is a significant need to develop promising chelators for these less-studied metal ions by optimizing for the different HSAB properties, ionic radii, and oxidation states. This need is reflected by a paucity of in vitro and in vivo anticancer studies which cannot be obtained unless the MAE emitters are stably attached to suitable biological targeting vectors.

The studies highlighted in this review provide a reflection on recent efforts to address these challenges. In doing so, the coordination chemistry of a significant portion of the Periodic Table, including the transition metals, main group elements, and lanthanide metals, was discussed. Continued investigations, both fundamental and applied, will drive the development of new chelator types for many of these less-studied metal ions. Given the therapeutic potential of MAEs, this research is an exciting and necessary endeavor to leverage their full potential for the treatment of cancer while also embarking on rich coordination chemistry studies.

Highlights.

• The unique cytotoxic potential of Meitner-Auger electrons is discussed.

• Current production and purification strategies of each radiometal are summarized.

• State-of-the-art cyclic and acyclic chelators for each radiometal are summarized.

• Chelator radiolabeling conditions with each radiometal are summarized.

• Chelator design principles and their impact on chelator performance are emphasized.

List of Abbreviations and Acronyms

AMA

apparent molar activity

BFC

bifunctional chelator

Bq

Becquerel, unit of radioactivity

CE

conversion electron

Ci

Curie, unit of radioactivity

COSY

homonuclear correlation spectroscopy

CN

coordination number

DFT

density functional theory

EC (or ε)

electron capture

EXAFS

extended X-ray absorption fine structure

FDA

U.S. Food and Drug Administration

GEP-NET

gastroenteropancreatic neuroendocrine tumor

HMQC

heteronuclear multiple quantum coherence

HPLC

high-performance liquid chromatography

HSA

human serum albumin

HSAB

hard-soft acid-base

HSQC

heteronuclear single quantum coherence

IC

internal conversion

IT

isomeric transition

LET

linear energy transfer

LN

LaNthanide resin

MAE

Meitner-Auger electron

mCPRP

metastatic castration-resistant prostate cancer

NOESY

nuclear Overhauser effect spectroscopy

PSMA

prostate-specific membrane antigen

RCC

radiochemical conversion

SAP

square antiprismatic

SPECT

single photon emission computed tomography

t _1/2

half-life

TRT

targeted radionuclide therapy

TSAP

twisted square antiprismatic

XANES

X-ray absorption near-edge structure

XAS

X-ray absorption spectroscopy

List of Ligand Abbreviations and Acronyms

BAL

British anti-Lewisite (dimercaprol)

DFO

deferoxamine

DiAmSar

diaminosarcophagine

DMPS

2,3-dimercapto-1-propane sulfonic acid

DMSA

2,3-dimercaptosuccinic acid

DTPA

diethylenetriamine pentaacetate

DUPA

Glu-urea-Glu peptide

EDTA

ethylenediaminetetraacetic acid

K222

Kryptofix^® 222

NTA

nitrilotriacetate

PIP

pyrrole-imidazole polyamide

TACN

1,4,7-triazacyclonane

Tm

hydrotris(methimazolyl)borate

Data availability

No data was used for the research described in the article.