Introduction

Targeted alpha therapy (TAT), a radiopharmaceutical strategy that delivers high-energy alpha (\(\alpha\)) particles to cancerous cells in vivo through biological targeting vectors, is a nascent approach to treat certain types of cancers1,2. Compared to generalized cancer treatments, TAT has shown high efficacy with minimal toxicity by directly targeting diseased cells while causing limited damage to the surrounding tissue3,4. The \(\alpha\)-emitting radioisotope actinium-225 [225Ac; t1/2 = 9.920(3) days] is particularly attractive for this type of radiopharmaceutical therapy given its relatively short half-life and a decay chain with multiple \(\alpha\)-emissions to promote an in vivo generator of \(\alpha\) particles5,6,7. The potential of 225Ac-TAT has been established in clinical trials8, such as in the treatment of advanced-stage prostate cancer9. As 225Ac is not naturally occurring, the burgeoning demand for this radioisotope has been met through strategic developments in isotope production10,11. Despite these advancements, the clinical progression of 225Ac-TAT has been slow, with 225Ac-TAT studies suffering from poor retention at the targeted site in vivo, leading to nonspecific radiotoxic effects from 225Ac and its progeny12,13. Underlying these issues is the paucity of knowledge regarding the chemical behavior of this element.

Despite its discovery at the turn of the century, little research has been dedicated to understanding actinium (Ac) chemistry over the past several decades, with only a handful of studies reported with simple inorganic or organic ligands14. Interest in fundamental Ac coordination has experienced a revival due to the recent realization of the potential of 225Ac for TAT15,16,17,18,19, yet developments are scarce as the chemistry is technically challenging, requires specialized facilities, and is currently limited to scales of micrograms or less due to the current production capabilities for 225Ac and other Ac isotopes11,20. Nonetheless, important advancements have been made in the theoretical and experimental gas- and solution-state chelation of Ac with synthetic and naturally occurring systems, in addition to pioneering X-ray absorption spectroscopic measurements to probe its coordination behavior in relevant solutions21,22,23,24,25,26. On the other hand, with no single crystal structure reported to date, our understanding of Ac coordination behavior in the solid-state is still limited. Foundational powder X-ray diffraction measurements, mostly from 195014,27, have established Ac chemical bonding concepts, but recently, the accuracy of these measurements have been scrutinized due to the possibility of impurities in the diffraction samples28. As such, there is a clear need to assemble a platform capable of addressing these multidimensional challenges to uncover even the most rudimentary information about this element, with particular emphasis on limiting interference from contaminants that would otherwise skew Ac bonding metrics.

Our approach to probe Ac chemical behavior was multifaceted. First, we used a multidentate, synthetic siderophore, termed 3,4,3-LI(1,2-HOPO) and referred to as HOPO hereafter, that has demonstrated efficient binding for heavy metal decorporation in vivo29. As HOPO is particularly amenable for chelation of lanthanide (Ln) and actinide (An) ions in aqueous solutions, we hypothesized that HOPO should bind AcIII analogously to other trivalent MIII ions (MIII = Sc, Y, La–Lu, Am, Cm, Cf, Es)30,31,32,33,34,35, while also providing a spectroscopic handle to probe AcIII solution thermodynamics that would otherwise be inaccessible given the AcIII electronic ground state (5f 06d 0)30,31. Although HOPO can serve to examine Ac coordination in solution, an additional tactic is needed to access solid-state structural information. Macromolecular scaffolds offer an opportunity to crystallize nanogram to microgram quantities of a metal or metal-complex through high-molecular weight carriers to reveal insight for metal coordination via protein crystallographic measurements36. The mammalian protein, siderocalin (Scn), binds siderophores and their ferric complexes as part of the innate immune system, and has also been shown to selectively bind a diverse range of Ln- and An-siderophore complexes37,38. By adding Scn to a solution containing AcIII and HOPO, recognition of the metal complex in the binding site of Scn will occur, and this macromolecular AcIII–HOPO–Scn ternary construct can be crystallized from solution. Taken together, HOPO and Scn provide a unique lens to view the fundamental coordination chemistry of Ac on a microgram scale, outlined in Fig. 1. Herein, we isolate and characterize an AcIII–HOPO complex contained within a macromolecular scaffold.

Fig. 1: Generalized scheme of the work presented herein that aims to understand the coordination of an actinium hydroxypyridinone complex contained within a macromolecular scaffold.
figure 1

a Synthetic route to an actinium small molecule complex encapsulated in a macromolecule through chelation of 227AcIII with the 3,4,3-LI(1,2-HOPO) (HOPO) chelator and the siderocalin (Scn) protein. Under buffered conditions, HOPO binds AcIII to yield a monoanionic coordination complex, nominally termed [AcIII(HOPO)]1–. Through electrostatic interactions with ligand aryl groups and amino acid residues, [AcIII(HOPO)]1– is then contained in the binding pocket of Scn to yield a ternary AcIII–HOPO–Scn protein complex. The binding between HOPO, AcIII, and Scn was probed using spectrofluorimetric competition titrations and fluorescence quenching binding assays. b The ternary AcIII–HOPO–Scn protein complex was crystallized to structurally characterize actinium coordination within Scn. After AcIII–HOPO–Scn was formed in solution, further purification was provided through protein concentration filters wherein superfluous ions, such as the radioactive daughters that had grown in since purification of the AcIII stock solution, were filtered out. After one week, diffraction-quality crystals of the ternary AcIII–HOPO–Scn complex were observed and structurally characterized with protein crystallography.

Full size image

Results

Purification and quantification of an actinium stock solution

The half-life of 225Ac, the limited availability of this isotope, and the need to possess significantly higher mass quantities to perform non-trace chemical studies were circumvented by employing the longest-lived isotope of actinium, 227Ac [t½ = 21.772(3) years], in these studies as the closest surrogate to the radiopharmaceutically-relevant isotope. The synthesis of an 227Ac complex required a chemically and radiochemically purified stock solution attained through the reprocessing of legacy 227Ac samples according to established methods, which combine anion exchange chromatography and extraction chromatography to remove chemical impurities and 227Ac decay products (Supplementary Fig. 1)23,24,25,39. Purifications were performed immediately before the experimental studies in order to mitigate safety concerns associated with the highly radioactive, gamma-emitting daughters of 227Ac, namely 227Th and 223Ra, and ensure the overwhelming majority of the metal stock solution contained 227Ac instead of its daughter isotopes (Supplementary Fig. 2). Further nuances arose in the quantification of the final metal stock after purification, as the low energy beta (\(\beta\)) and gamma (γ) emissions of 227Ac make determining its concentration directly after purification particularly challenging. As such, a combination of liquid scintillation counting to directly probe 227Ac \(\beta\)-emission and gamma spectroscopy to indirectly probe 227Ac with \(\gamma\)-emissions from daughter ingrowth were used (Supplementary Fig. 3). A thorough discussion of the purification and quantification of 227Ac is provided in the Supplementary Methods section.

Determination of the [AcIII(HOPO)]1– stability constant

Upon achieving a chemically and isotopically pure stock solution of 227Ac, the formation constant (log \({\beta }_{{{{{{\rm{ML}}}}}}}\)) of AcIII with HOPO (Fig. 1A) was determined to establish the viability of this chelator for TAT applications and understand the complexation of actinium with HOPO in the context of other +3 metal ions. With an eye towards isotope recovery and limited mass quantities, spectrofluorimetric competition titrations were chosen over other methods, such as solvent extractions or potentiometric titrations, to assess complexation thermodynamics. The high sensitivity of fluorescence spectroscopy enables thermodynamic stability constants to be determined indirectly through metal competition experiments for non-luminescent metals, like AcIII, using europium (EuIII) as the luminescent reference31. As the experiments were performed at specific pH and ionic strength values, the thermodynamic stability constants are conditional and therefore termed herein as log \({\beta }_{{{{{{\rm{ML}}}}}}}^{{\prime} }\) values instead of log \({\beta }_{{{{{{\rm{ML}}}}}}}\) values. To determine the stability constant of AcIII with HOPO via spectrofluorimetric competition titrations with microgram quantities of AcIII, the concentration of EuIII was scaled down considerably. Rather, typical competition titrations that contained 3 \(\mu\)M of the EuIII complex (in this case, [EuIII(HOPO)]1–) were reduced down to 10 nM, which only required approximately 1.70 μg of competing AcIII instead of 1.02 mg for triplicate measurements. Numerous surrogate experiments were performed separately with LaIII to support the feasibility and reproducibility of spectrofluorimetric measurements at these reduced scales (Supplementary Discussion section, Supplementary Figs. 4–8, Supplementary Table 1). Consistent log \({\beta }_{{{{{{\rm{ML}}}}}}}^{{\prime} }\) values for [LaIII(HOPO)]1– were observed, regardless of the metal concentrations at 3 \(\mu\)M [EuIII(HOPO)]1–, 50 nM [EuIII(HOPO)]1–, or 10 nm [EuIII(HOPO)]1– underscoring the viability of this approach for AcIII thermodynamic studies (Fig. 2A–C, E).

Fig. 2: Spectrofluorometric competition titrations used to assess the stability constants of trivalent lanthanides and actinides with the HOPO chelator.
figure 2

a–c, e Surrogate experiments with LaIII were successively scaled down to repeat the experiment with limited mass quantities of AcIII. Competition against the [EuIII(HOPO)]1– complex at concentrations of (a) 3 \(\mu\)M [EuIII(HOPO)]1–, (b) 50 nM [EuIII(HOPO)]1–, and (c) 10 nM [EuIII(HOPO)]1– were performed with LaIII to demonstrate the merit of the methodology, resulting in similar conditional stability constants (log \({\beta }_{{{{{{\rm{ML}}}}}}}^{{\prime} }\)) tabulated in e. The error reported is the standard deviation corresponding to the last digit in triplicate measurements. d Spectrofluorimetric competition titrations with 227AcIII and [EuIII(HOPO)]1– resulting in a conditional stability constant (log \({\beta }_{{{{{{\rm{ML}}}}}}}^{{\prime} }\)) tabulated in e. ([EuIII(HOPO)]1– = 10 nM, I = 0.5 M KCl, pH = 7.36, T = 25 °C, λex = 325 nm. f Stability constants of trivalent lanthanide and actinide ions with HOPO to yield the respective [MIII(HOPO)]1– complex plotted as a function of ionic radii. The log \({\beta }_{{{{{{\rm{ML}}}}}}}^{{\prime} }\) values for Ac and La are reported herein (c, d), whereas the remaining values are from refs. 31,32,33. The error reported is the standard deviation corresponding to the last digit in triplicate measurements, except for americium, which is the standard deviation corresponding to the last digit in duplicate measurements. The ionic radii are from ref. 28. (Ac) and ref. 40. for CN = 6. Note, the stability constants increase across the 4f and 5f series.

Full size image

Spectrofluorimetric competition titrations between AcIII and [EuIII(HOPO)]1– were monitored by fluorescence spectroscopy. As seen in Fig. 2D, the intensity of [EuIII(HOPO)]1– decreased with the increasing concentrations of AcIII, which corresponds to the formation of the [AcIII(HOPO)]1– complex and the release of EuIII ions from HOPO chelators. These changes were modeled using a nonlinear least-squares refinement (Supplementary Fig. 9), and included refinement parameters such as the HOPO protonation constants measured under these solution conditions (Supplementary Figs. 10, 11, Supplementary Table 2), the hydrolysis constants of the metal ions (Supplementary Table 3), and the [EuIII(HOPO)]1– formation constant31. In aqueous solution at pH 7.36, the conditional stability constant of the [AcIII(HOPO)]1– complex was determined to be 17.0(1). This value can be compared to [LaIII(HOPO)]1– under the same conditions with a log \({\beta }_{{{{{{\rm{ML}}}}}}}^{{\prime} }\) value of 16.50(3), or at higher concentrations with a log \({\beta }_{{{{{{\rm{ML}}}}}}}^{{\prime} }\) value of 17.08(4) (Fig. 2E). In addition, these [LaIII(HOPO)]1– values are in agreement with a published [LaIII(HOPO)]1– conditional stability constant collected under different solution conditions through metal competition titrations, with a reported value of 16.4(3)31. Generally, the conditional stability constants of AcIII and its non-radioactive surrogate LaIII with HOPO are lower than the rest of the lanthanides (Fig. 2F). When compared to other +3 actinides, the conditional stability constant of AcIII with HOPO is also lower, with [AmIII(HOPO)]1– and [CmIII(HOPO)]1– values at 20.4(2) and 21.8(4), respectively32,33. More broadly, the log \({\beta }_{{{{{{\rm{ML}}}}}}}^{{\prime} }\) value determined herein for [AcIII(HOPO)]1– is consistent with established bonding trends of the f-elements; stability constants with HOPO increase as a function of decreasing ionic radii and increasing Lewis acidity across the 4f- and 5f-series (Fig. 2F)28,40.

Chelation of [AcIII(HOPO)]1– by a protein, siderocalin

Armed with solution thermodynamic data to support the complexation and formation of the [AcIII(HOPO)]1– complex, the next step was to establish the recognition of this complex in solution with the protein, siderocalin (Scn). To assess the affinity of Scn for [AcIII(HOPO)]1– to form the ternary complex, referred to herein as AcIII–HOPO–Scn, the dissociation constant (KD) was determined through a fluorescence quenching binding assay (Fig. 1A). The KD value can be quantified by monitoring the intrinsic fluorescence of tryptophan residues that reside in the binding cavity, or calyx, of Scn, which are thereby quenched upon metal complex recognition37,41. Surrogate experiments were performed with LaIII (Supplementary Discussion section, Supplementary Figs. 12–14).

A buffered solution that contained the [AcIII(HOPO)]1– complex was titrated into a solution of Scn and the tryptophan emission was monitored by fluorescence spectroscopy. As seen in Fig. 3, the normalized intensity of Scn decreases with the increasing concentrations of [AcIII(HOPO)]1–, which is based on the recognition of the actinium complex by the protein. Fluorescence data were analyzed by nonlinear regression analysis of fluorescence response versus [AcIII(HOPO)]1– concentration using a one-site binding model (Supplementary Table 4)42. In aqueous solution at pH 7.4, the KD of the [AcIII(HOPO)]1– complex with Scn was determined to be 6(1) nM. This value can be compared to the free ligand, [HOPO]4–, and [LaIII(HOPO)]1–, with KD values of 43(17) nM and 20(5) nM, respectively (Supplementary Tables 5, 6). Notably, the nanomolar affinity of Scn for [AcIII(HOPO)]1– is significantly higher compared to HOPO and the LaIII complex (Fig. 3). Moreover, Scn shows a greater affinity for the AcIII complex compared to other Ln- and An-HOPO complexes, such as EuIII, AmIII, and CmIII with KD values of 14(1) nM, 29(1) nM, and 22(5) nM, respectively38. The obtained KD value is closer to those of yttrium (YIII) and scandium (ScIII) HOPO complexes, [YIII(HOPO)]1– and [ScIII(HOPO)]1–, with KD values of 2(1) nM and 6(1) nM, respectively43. It should be noted that the recognition of [AcIII(HOPO)]1– by Scn does not quite match the avid affinity of the protein for one of its known endogenous targets, the ferric complex of enterobactin, [FeIII(Ent)]3– [0.4(1) nM]37. Nonetheless, the thermodynamically favorable formation of [AcIII(HOPO)]1– coupled with the complex’s binding affinity with Scn collectively generate an advantageous platform to probe actinium coordination in the solid-state.

Fig. 3: Fluorescence quenching binding assays as a function of titrant concentration and subsequent determination of dissociation constants and cumulative stability constants.
figure 3

a Luminescent behavior of tryptophan residues within the siderocalin (Scn) protein. Fluorescence (\(\lambda\)ex = 281 nm; blue trace) is quenched upon titrant recognition in the Scn binding pocket (\(\lambda\)ex = 281 nm; red trace), which can be used to quantify the interaction between the protein and the titrant to yield a dissociation constant (KD). Fluorescence quenching analyses of Scn with HOPO (b), [LaIII(HOPO)1– (c), and [AcIII(HOPO)1– (d) at pH 7.4. Blue squares, purple triangles, and green diamonds denote the normalized fluorescence intensities at 340 nm (\(\lambda\)ex = 281 nm) and the error bars represent the standard deviation of three separate trials. The dotted gray fit lines were calculated with a one-binding site model42. e Tabulated KD values from fluorescence quenching analyses in b–d. The AmIII value is from ref. 38. The log \({\beta }_{2}^{{\prime} }\) value, which represents the binding constant of the ligand or metal complex with Scn, was calculated by taking the inverse log of the KD. The error reported is the standard deviation corresponding to the last digit in triplicate measurements. f The cumulative stability constant (log \({\beta }_{{{{{\mathrm{1,2}}}}}}^{{\prime} }\)) of HOPO–Scn with AcIII compared to other chelators, including DOTA19, BATA19, and EDTA45. The asterisk denotes a conditional stability constant.

Full size image

Crystallization of actinium in a macromolecular scaffold

The ternary AcIII–HOPO–Scn complex was crystallized using methods previously demonstrated to routinely and effectively crystallize other +3 metal–HOPO–siderocalin macromolecules quickly (Fig. 1B)38. After one week, diffraction-quality crystals were isolated and analyzed at the Advanced Light Source (Beamline 5.0.2). Crystals of the ternary LaIII–HOPO–Scn complex were also characterized for direct comparison between the +3 La/Ac metal centers. As demonstrated previously, Scn is capable of specific, tight binding to a wide variety of metals complexes with natural and synthetic siderophores44. In general, the binding of ligands/complexes in the protein calyx is driven by electrostatic and cation–\(\pi\) interactions, particularly by interactions between side chains and ligand aryl groups. Although perfected to accommodate hexadentate ligands like enterobactin in three pockets within the calyx, siderocalin manages to bind octadentate structures like HOPO by distorting the coordination environment of the calyx only mildly. The fourth HOPO ring of the octadentate chelator cannot fit within the calyx and rather extends outwards (Fig. 4A), yet the protein shows remarkable structural rigidity and conservation as highlighted by the structural similarities between trivalent lanthanide and actinide structures (Supplementary Fig. 15)38,44. The AcIII–HOPO–Scn structure is no different in these generalities, emulating binding in the protein scaffold akin to other f-elements. The most noteworthy structural rearrangement in the AcIII complex involves the movement of two aryl groups of the HOPO chelator, the one occupying Pocket #3 and the other extending out of the calyx, above another HOPO group, in Pocket #1 (Fig. 4A, red arrows). These movements displace the two aryl groups outwards, likely to accommodate the larger AcIII ion, and are not observed in the corresponding LaIII structure. Otherwise, the side-chains of residues lining the Scn calyx remain largely unperturbed, with the largest movement (0.7 \(\mathring{\rm A}\)) occuring at the N\(\zeta\) position of a lysine residue (K125) that defines Pocket #1 (Fig. 4A). Further evidence of AcIII–HOPO–Scn crystallization were determined through qualitative radiolytic degradation of the crystals over time and quantitative radioactivity measurements of the crystals (Supplementary Discussion section, Supplementary Figs. 16–18).

Fig. 4: Protein crystallographic renderings of [AcIII(HOPO)]1– and [LaIII(HOPO)]1–complexes inside the calyx of siderocalin (Scn).
figure 4

a Overlay of [AcIII(HOPO)]1– and [LaIII(HOPO)]1–complexes inside the calyx of Scn. The protein is shown in a molecular surface representation (gray), the chelating groups are shown in a licorice-stick representation, colored by atom type (carbon = gray, nitrogen = blue, oxygen = red), and the metals are shown as colored spheres (Ac = magenta, La = gray). The protein binding pockets are noted in yellow text and one of the amino acid residues (K125) that defines Pocket #1 is labeled. Red arrows highlight the positions of the HOPO groups wherein the rings are geometrically different between the La and Ac structures. b Representation of the interactions about the AcIII metal center with HOPO rings, identifying the oxygen interactions with either N-oxide or ketone oxygen atoms.

Full size image

Discussion

Evaluation of actinium solution and solid-state metrics

Notwithstanding microgram quantities of 227AcIII, the combination of the HOPO chelator and siderocalin enabled the collection of meaningful solution and solid-state information. The conditional stability constant of the [AcIII(HOPO)]1– complex, with a log \({\beta }_{{{{{{\rm{ML}}}}}}}^{{\prime} }\) value of 17.0(1), expands the very limited library of actinium thermodynamic data14. Put in the context of other chelators, the [AcIII(HOPO)]1– formation constant is greater than that of ethylenediaminetetraacetic acid (EDTA), with a log \({\beta }_{{{{{{\rm{ML}}}}}}}\) value of 14.22 (I = 0.1 M, pH = 2.8)45, and is less than the formation constants of AcIII with 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (H4DOTA) and H4BATA with log \({\beta }_{{{{{{\rm{ML}}}}}}}\) values of 20.32(3) and 25.74(1), respectively19. The lanthanide-utilizing protein lanmodulin (LanM) has been shown to recognize actinium, with a polymetallic formation constant of 36.2(5) for 228Ac3LanM at pH 7, but should be noted as the product of three individual stability constants, one for each metal binding site22. In the same vein, by combining the affinity of siderocalin for [AcIII(HOPO)]1– (measured through KD), the cumulative stability constant (\({\beta }_{{{{{\mathrm{1,2}}}}}}^{{\prime} }\) = K1K2; Supplementary Discussion section, Supplementary Eqs. (5)–(7)) for AcIII–HOPO–Scn is 25.2, making it one of the most effective actinium chelators to date (Fig. 3F). Insight into this favorable interaction was supported through protein crystallographic data that illustrated a different geometry of the actinium complex in the protein binding cavity compared to other trivalent ions.

Herein, actinium structural characterization in the solid-state was enabled through its crystallization within a macromolecular scaffold. The structure crystallized in the P41212 space group with a resolution of 2.08 Å (Supplementary Table 7; Supplementary Fig. 19). Three unique side chains (A, B, and C) are observed; each side chain possesses one crystallographically unique AcIII–HOPO complex (Fig. 4A). In the complex, HOPO surrounds the AcIII ion in an approximate square pyramid geometry with an average AcIII–OHOPO distance of 3.2(7) Å (Supplementary Table 8). In all cases, the averaged Ac–O distances involving N-oxide oxygen atoms [Ac–ON-oxide = 2.9(5) Å] were considerably shorter than the averaged Ac–O distances involving ketone oxygen atoms [Ac–Oketone = 3.5(8) Å] (Fig. 4B). These distances can be considered in the larger context of previous solution-state measurements: Ac–OH2O bond lengths were found to be 2.63(1) Å for the actinium-aquo ion by X-ray absorption spectroscopy (XAS) at the AcIII L3-edge24. Perhaps more relevant are actinium XAS measurements with the DOTP8– macrocycle, wherein Ac–ODOTP and Ac–NDOTP bond lengths were found to be 2.49(1) Å and 2.87(3) Å, respectively25. When compared to the structure of LaIII–HOPO–Scn that also crystallized in the P41212 space group with a resolution of 2.0 Å (Supplementary Table 7; Supplementary Fig. 19), the average La–OHOPO distance was 2.5(2) Å, which is nearly identical to the average La–OH2O bond length [2.55(5) Å] in the [LaIII(H2O)9]CF3SO3H solid-state structure (Supplementary Table 9)46. Compared to other +3 actinide–HOPO–Scn structures, the average AmIII–OHOPO and CmIII–OHOPO, are 2.5(2) Å and 2.6(2) Å, generally consistent with typical bonding trends across the 5f series.(67)38 The CfIII–OHOPO distance is 3.3(9) Å, which is considerably longer than typical Cf–O bonds and possibly is owed to the low resolution of this structure (2.7 Å)34.

The caveat of protein crystallographic solid-state analyses is that the data do not typically provide the same level of atomic resolution that is obtained in small molecule crystallography. Nonetheless, protein crystallography can be a useful tool to reveal coordination chemistry, particularly with heavy elements like actinium wherein the high Z is clearly distinguishable against the low Z macromolecular scaffold. When supported by additional studies, the solid-state structure of AcIII–HOPO–Scn provided structural evidence for the differences in KD values compared to its surrogate LaIII, and other +3 actinides including AmIII, CmIII, and CfIII, with a geometrical change about the metal center adopted for AcIII that is not observed in the other MIII–HOPO–Scn structures. We can speculate that the larger size of actinium (AcIII ionic radius = 1.065 \(\mathring{\rm A}\), CN = 6)28 enables closer interactions between the metal complex and key amino acids in the protein calyx, which are observed at longer distances in the lanthanum system (LaIII ionic radius = 1.032 \(\mathring{\rm A}\), CN = 6)40. For example, the two lysine residues (K125, K134) that define Pocket #1 have an average Ac···N\(\zeta\)K125 distance of 4.0(4) Å vs. 4.5(1) Å for La···N\(\zeta\)K125 and an average Ac···N\(\zeta\)K134 distance of 3.9(1) Å vs. 4.5(1) Å for La···N\(\zeta\)K134 (Supplementary Table 10; Supplementary Fig. 20). Still, the binding cavity of Scn is very rigid, and thus the HOPO coordination to AcIII seems to adjust in order to allow for AcIII–HOPO recognition in the protein. To illustrate this alteration, the AcIII–HOPO structure in Scn can be overlayed with an optimized DFT structure previously calculated for the [AcIII(HOPO)]1– complex to emphasize the protein scaffold’s influence on actinium coordination, for which is not observed in the LaIII–HOPO–Scn structure (Supplementary Fig. 21)47. Future work will leverage computational analyses to tease out the impacts of siderocalin on actinium bonding. Meanwhile, these results emphasize differences in the coordination behavior of lanthanum and actinium and collectively illustrate an example wherein the use of a non-radiological surrogate as a proxy for the radioactive ion did not fully capture their chemical complexities. As underscored by others12,19,48, surrogate chemistry is not always sufficient to fully understand actinide behavior, and therefore efforts to continue to explore actinium chemistry, and also develop systems to tackle the unique challenges associated with working with this element, must continue to be pursued.

As is true for most of the actinide series, our fundamental understanding of actinium chemistry is severely lacking – a consequence of its rarity, difficulty of isolation, radiotoxicity, and until recently, a lack of commercial application. Even the position of this element on the Periodic Table is still debated49. The potential of 225Ac-TAT has since motivated the exploration of the coordination, bonding, stability, and reactivity of this element. In combining a synthetic siderophore, HOPO, to chelate AcIII and encapsulate the AcIII–HOPO complex within the binding site of the siderophore-seeking protein siderocalin, solid-state structural information of an actinium complex within a macromolecular scaffold was obtained. The crystallography illustrated geometrical changes about the AcIII metal center that deviated from other trivalent metals, such as actinium’s lanthanide surrogate, lanthanum, and heavier actinides like americium, curium and californium. The ability to compare the structural behavior of these +3 heavy metals in the solid-state captured the limitations of using surrogates in lieu of their radioactive counterparts, as the translation of their coordination chemistry may not always be identical. Taken together with the stability constant determined for the [AcIII(HOPO)]1– complex, these results represent a platform that not only reveal unique Ac coordination in the solid-state, but more broadly aid in understanding periodicity across the actinide series and this element’s position on the Periodic Table, while also directing renewed efforts for chelation strategies in TAT development. Future studies will focus on applying this platform to radiopharmaceutically-relevant 225Ac in vitro and in vivo studies, as we have previously demonstrated that siderocalin can be fused with antibodies and antibody fragments for the precise delivery of cargo isotopes within targeted cancer cells43.

Methods

All experimental details, which include general considerations, instrumentation, preparation and quantification of actinium stock solutions, potentiometric titrations, fluorescence competition semi-batch assays, fluorescence quenching assays, macromolecular crystallography details, and additional characterizations, are provided in the Supplementary Methods section in the Supplementary Information.