Toward Bifunctional Chelators for Thallium-201 for Use in Nuclear Medicine

Abstract

Auger electron therapy exploits the cytotoxicity of low-energy electrons emitted during radioactive decay that travel very short distances (typically <1 μm). ²⁰¹Tl, with a half-life of 73 h, emits ∼37 Auger and other secondary electrons per decay and can be tracked in vivo as its gamma emissions enable SPECT imaging. Despite the useful nuclear properties of ²⁰¹Tl, satisfactory bifunctional chelators to incorporate it into bioconjugates for molecular targeting have not been developed. H₄pypa, H₅decapa, H₄neunpa-NH₂, and H₄noneunpa are multidentate N- and O-donor chelators that have previously been shown to have high affinity for ¹¹¹In, ¹⁷⁷Lu, and ⁸⁹Zr. Herein, we report the synthesis and serum stability of [^nat/201Tl]Tl³⁺ complexes with H₄pypa, H₅decapa, H₄neunpa-NH₂, and H₄noneunpa. All ligands quickly and efficiently formed complexes with [²⁰¹Tl]Tl³⁺ that gave simple single-peak radiochromatograms and showed greatly improved serum stability compared to DOTA and DTPA. [^natTl]Tl-pypa was further characterized using nuclear magnetic resonance spectroscopy (NMR), mass spectroscopy (MS), and X-ray crystallography, showing evidence of the proton-dependent presence of a nine-coordinate complex and an eight-coordinate complex with a pendant carboxylic acid group. A prostate-specific membrane antigen (PSMA)-targeting bioconjugate of H₄pypa was synthesized and radiolabeled. The uptake of [²⁰¹Tl]Tl-pypa-PSMA in DU145 PSMA-positive and PSMA-negative prostate cancer cells was evaluated in vitro and showed evidence of bioreductive release of ²⁰¹Tl and cellular uptake characteristic of unchelated [²⁰¹Tl]TlCl. SPECT/CT imaging was used to probe the in vivo biodistribution and stability of [²⁰¹Tl]Tl-pypa-PSMA. In healthy animals, [²⁰¹Tl]Tl-pypa-PSMA did not show the myocardial uptake that is characteristic of unchelated ²⁰¹Tl. In mice bearing DU145 PSMA-positive and PSMA-negative prostate cancer xenografts, the uptake of [²⁰¹Tl]Tl-pypa-PSMA in DU145 PSMA-positive tumors was higher than that in DU145 PSMA-negative tumors but insufficient for useful tumor targeting. We conclude that H₄pypa and related ligands represent an advance compared to conventional radiometal chelators such as DOTA and DTPA for Tl³⁺ chelation but do not resist dissociation for long periods in the biological environment due to vulnerability to reduction of Tl³⁺ and subsequent release of Tl⁺. However, this is the first report describing the incorporation of [²⁰¹Tl]Tl³⁺ into a chelator–peptide bioconjugate and represents a significant advance in the field of ²⁰¹Tl-based radiopharmaceuticals. The design of the next generation of chelators must include features to mitigate this susceptibility to bioreduction, which does not arise for other trivalent heavy radiometals.

Introduction

Molecular radionuclide therapy (MRT) involves the delivery of a lethal dose of ionizing radiation emitted by a radionuclide specifically to diseased tissues or tumors. For example, α (such as ²²⁵Ac) and β⁻ (e.g., ¹⁷⁷Lu, ⁹⁰Y) emitting radionuclides, attached to antibodies and peptides targeting the prostate-specific membrane antigen (PSMA), have recently shown clinical promise for treating prostate cancer.¹⁻⁴ PSMA is expressed on normal prostate cells, but its expression is greatly increased in malignant prostate tissues while remaining low in most other healthy tissues, making it a useful target for MRT.⁵ Following treatment with [¹⁷⁷Lu]Lu-PSMA-617, 70% of patients experienced a decline in prostate-specific antigen (PSA) levels in the blood.² A similar response was observed using [²²⁵Ac]Ac-PSMA-617, where patients saw a decline of ≥50% in PSA levels, which is closely associated with better overall survival.³

Because their typical range in tissues greatly exceeds cellular dimension, β⁻ particles are highly effective at damaging large tumors through the crossfire effect but are much less effective against single tumor cells and small cell clusters.^6,7 In comparison, α particles and radionuclides emitting Auger electrons (AEs) have a high linear energy transfer (LET) (80–100 and 4–26 keV/μm, respectively), potentially enabling them to target and kill micrometastases and circulating tumor cells.^8,9 α and β⁻ particles travel 40–80 μm and 0.1–10 mm, respectively, which can lead to off-target tissue toxicity to healthy tissues. This can be partially mitigated by choosing radionuclides with emissions that match the tumor size.⁹ AEs, on the other hand, travel typically <1 μm, making the likelihood of off-target effects much lower. AE-emitters thus make an exciting group of radionuclides for potentially effective MRT of micrometastases, with few side effects. This is exemplified by a recent report detailing in vitro and preclinical cytotoxic and antitumor effects of AE-emitting [¹²⁵I]I-DCIBzL as a prostate cancer therapy in preclinical mouse models.⁸¹⁶¹Tb has also shown therapeutic efficacy through the emission of both beta particles and AEs. In vivo studies using [¹⁶¹Tb]Tb-PSMA-617 showed an improved antitumor effect compared to [¹⁷⁷Lu]Lu-PSMA-617 despite the two agents having comparable pharmacokinetics.¹⁰ Furthermore, Vallis et al. have used [¹¹¹In]In-DTPA-hEGF in Phase 1 clinical trials with 16 patients with metastatic EGFR-positive breast cancer.¹¹ Radiation doses to the kidney and liver were within radiation toxicity limits, and high tumor accumulation was observed; however, for a therapeutic effect, dose escalation will be required.¹¹ Michel and co-workers have highlighted the therapeutic potential of antibodies labeled with ⁶⁷Ga, as more potency was observed when compared to ¹¹¹In and ¹²⁵I.¹² Pirovano et al. have developed an ¹²³I-labeled PARP1 inhibitor ([¹²³I]I-MAPi) utilizing the Auger electron emissions as the basis of a potent radiotherapeutic for use in glioblastoma tumors.^13,14

Thallium-201 (²⁰¹Tl, t_1/2 = 73 h) has the potential to be a highly effective therapeutic radionuclide in future MRT applications, as it emits 37 Auger and other high LET secondary electrons per decay (c.f. 25 and 12 AEs emitted by ¹²⁵I and ¹⁶¹Tb, respectively).^9,15 Like other AE-emitters, ²⁰¹Tl could also facilitate a theranostics and personalized approach with accurate dosimetry as it releases gamma and X-rays, enabling single photon emission computed tomography (SPECT) imaging. Historically, ²⁰¹Tl has been used as a SPECT myocardial perfusion imaging agent but has been largely phased out since the introduction of ^99mTc agents like tetrofosmin and sestamibi.

We have recently shown that nontargeted delivery of ²⁰¹Tl (in the form of [²⁰¹Tl]TlCl) shows short- and long-term toxicity in prostate cancer cells.¹⁶ A dramatic decrease in clonogenic survival was achieved at only 0.29 Bq/cell, significantly lower than for other AE-emitting radionuclides such as ⁶⁷Ga and ¹¹¹In.^17,18 However, [²⁰¹Tl]Tl⁺ has little intrinsic selectivity for tumors: it accumulates in the myocardium via the Na⁺/K⁺ ATPase pump. Thus, although it has been a very useful imaging agent for heart function, a targeted approach is required for other in vivo applications.¹⁹

To date, targeted delivery of ²⁰¹Tl to cancer cells has been hindered due to the lack of suitable bifunctional chelator chemistry. Despite the high importance of ²⁰¹Tl during the early years of nuclear medicine, thallium chelation has been poorly investigated. Previous attempts using proteins conjugated to the most common and broadly useful chelators such as DTPA or DOTA have shown complex instability.²⁰⁻²² More recent studies carried out by our group have confirmed that Tl³⁺ complexes of EDTA, DTPA, and DOTA, despite forming Tl³⁺ complexes with very high association constants, do not possess adequate kinetic stability for MRT, highlighting the continuing need for new thallium chelators that will form kinetically stable complexes.²³

Recently, Orvig and co-workers introduced a range of branched polydentate picolinic acid based chelators for evaluation as chelators for large, high-valent metal ions such as In³⁺, Lu³⁺, Sc³⁺, and Ac³⁺.²⁴⁻²⁶ H₄pypa, H₅decapa, H₄neunpa-NH₂, and H₄noneunpa (Figure 1A) are chelators with a cavity size ideal for these radiometal ions, which have ionic radii (1.01–1.26 Å) similar to that of Tl³⁺ (1.12 Å).²⁷ Recent studies demonstrated that H₄pypa can also be labeled with [⁴⁴Sc]Sc³⁺ and [⁸⁶Y]Y³⁺ and bioconjugated to PSMA-targeting radiopharmaceuticals.^28,29 Herein, we describe the preliminary evaluation of these ligands as chelators for [²⁰¹Tl]Tl³⁺. As all of these chelators could be efficiently radiolabeled with [²⁰¹Tl]Tl³⁺, we selected H₄pypa and its previously described isothiocyanate bifunctional derivative H₄pypa-NCS for further study. This include synthesis,in vitro and in vivo characterization of the [²⁰¹Tl]Tl-pypa-PSMA conjugate in healthy mice and PSMA-positive and -negative tumor models in mice.²⁹

Figure 1.

(A) Structures of H₄pypa, H₅decapa, H₄neunpa-NH₂, and H₄noneunpa. (B) Analytical HPLC traces of [²⁰¹Tl]Tl-pypa, [²⁰¹Tl]Tl-decapa, [²⁰¹Tl]Tl-neunpa-NH₂, and [²⁰¹Tl]Tl-noneunpa (black = counts per second) (HPLC method A). (C) Stability studies in human serum for [²⁰¹Tl]Tl-pypa, [²⁰¹Tl]Tl-decapa, [²⁰¹Tl]Tl-neunpa-NH₂, and [²⁰¹Tl]Tl-noneunpa (n = 3).

Results

Radiolabeling Chelators

In a preliminary radiochemical screening study, we oxidized [²⁰¹Tl]Tl⁺ to [²⁰¹Tl]Tl³⁺ using Iodobeads and assessed radiolabeling reactions of [²⁰¹Tl]Tl³⁺ with each of the chelators H₄pypa, H₅decapa, H₄noneunpa, and H₄neunpa-NH₂.²³ Each chelator (0.02 mg) was incubated with [²⁰¹Tl]Tl³⁺ (5–10 MBq, 20–30 μL) in an aqueous solution at pH 5 at ambient temperature for 10 min followed by HPLC and RP-TLC analysis.²³ Under the HPLC conditions employed here, [²⁰¹Tl]Tl-pypa eluted at t_R = 10.09 min, [²⁰¹Tl]Tl-decapa at 8.15 min, [²⁰¹Tl]Tl-noneunpa at 8.44 min, and [²⁰¹Tl]Tl-neunpa-NH₂ at 8.17 min, whereas unchelated [²⁰¹Tl]TlCl₃ eluted earlier at 2.03 min. Radiolabeling was rapid in all cases; radiochromatograms (Figure 1B) show radiochemical yields of >97% after only 10 min of incubation at room temperature (RT). Each chelator was also evaluated with [²⁰¹Tl]Tl⁺ (i.e., without prior treatment with Iodobeads); no complexation reaction was observed by HPLC in these experiments (Figure S3).

In Vitro Stability

Each [²⁰¹Tl]Tl-labeled complex was left standing in an ammonium acetate solution (1 M, pH 5) for 48 h (Figure S4), and each showed no degradation. However, all complexes showed modest stability when incubated in human serum at 37 °C (Figure 1C). After 24 h in human serum, 68.7 ± 6.5% of the [²⁰¹Tl]Tl-pypa complex was intact, decreasing to 57.7 ± 15.1% after 48 h. [²⁰¹Tl]Tl-decapa, [²⁰¹Tl]Tl-neunpa, and [²⁰¹Tl]Tl-noneunpa showed very similar complex stability to [²⁰¹Tl]Tl-pypa in human serum after 48 h. In addition to serum, the stability of all four [²⁰¹Tl]Tl-labeled complexes was evaluated in the presence of excess apotransferrin. All showed resistance to transmetalation over 24 h (Figures S5–S7).

All four evaluated chelators could be efficiently labeled with [²⁰¹Tl]Tl³⁺, but all [²⁰¹Tl]Tl-labeled complexes were susceptible to demetallation in serum (55–80% intact after 48 h incubation). We elected to further assess bioconjugates of H₄pypa for [²⁰¹Tl]Tl³⁺ receptor-targeted delivery to prostate cancer cells. A bifunctional H₄pypa derivative, alongside bioconjugates of H₄pypa with a monoclonal antibody and an alternative PSMA-targeted peptidic motif, has been previously described.²⁸⁻³⁰ We therefore based the selection of the H₄pypa ligand for further evaluation on the ease of H₄pypa incorporation into bioconjugates.

Synthesis and Characterization of [Tl(Hpypa)]

The chelator H₄pypa³⁰ was reacted with thallium trichloride hydrate ([^natTl]TlCl₃(H₂O)₄) in an ammonium acetate solution (pH = 5) at RT for 15 min to yield [Tl(Hpypa)] as a white solid. Following purification, the complex was characterized using nuclear magnetic resonance (NMR) (Figure 2 and Figures S10–S27) and high-resolution electrospray ionization mass spectrometry (HR-ESI-MS) (Figures S28–S33). HRMS data show the formation of a 1:1 complex of H₄pypa with Tl³⁺. Due to the poor solubility of [Tl(Hpypa)] in D₂O, a small amount of Na₂CO₃ (in D₂O) was added, adjusting the pH to 8–9. This greatly increased solubility, presumably by the formation of [Tl(pypa)]⁻, enabling NMR (¹H, ¹³C, and 2D NMR) spectroscopic studies to be carried out. Under more acidic conditions, the [Tl(Hpypa)] complex was insufficiently soluble to obtain NMR spectra. Previous reports in the literature show that when complexed with In³⁺, Lu³⁺, and La³⁺ ions, H₄pypa forms rigid complexes giving rise to sharp ¹H NMR peaks suggesting little fluxionality. However, the ¹H and COSY NMR data for [Tl(pypa)]⁻ suggest that there are at least two species in the solution (Figure 2). In complexes of pypa, methylene protons are diastereotopic, with coupling between geminal, diastereotopic methylene protons. In the ¹H COSY spectrum (Figure S25) of the pypa complex of Tl³⁺, 12 cross peaks between methylene protons are observed, indicating that at least two chemically distinct [^natTl]Tl-pypa complexes are present in the solution that do not interconvert rapidly within the NMR time scale.

Figure 2.

¹H NMR spectra (D₂O) of H₄pypa (top) and [^natTl]Tl-pypa (bottom).

X-ray quality single crystals of [Tl(Hpypa)] were obtained by the slow evaporation of equimolar mixtures of TlCl₃ and H₄pypa solutions in water with the pH adjusted to 2 by the addition of HCl (0.1 M).³⁰ The crystal structure of [Tl(Hpypa)] is shown in Figure 3, and selected bond lengths can be found in Table 1. Full crystallographic information can be found in Figure S34. The complex has an octacoordinated Tl³⁺ in a distorted square antiprismatic geometry, and when grown from a solution at pH 2, one of the carboxylic acid groups is protonated and does not coordinate to Tl³⁺. The Tl³⁺ ion is coordinated by eight (N₅O₃) of the nine potential donor atoms of the ligand. The Tl–O bond lengths are between 2.258(6) and 2.496(5) Å, and Tl–N bond lengths are between 2.311(7) and 2.525(7) Å. These are comparable to bond lengths previously reported for Tl³⁺ complexes.³¹⁻³³

Figure 3.

Crystal structure of [Tl(Hpypa)] (50% probability ellipsoids).

Table 1. Selected Bond Lengths and Angles in [Tl(Hpypa)].

bond lengths			bond angles
atom	atom	length (Å)	atom	atom	atom	angle (°)
Tl	O006	2.496(5)	O006	Tl	N00H	136.2(2)
Tl	O007	2.258(6)	O006	Tl	N00J	82.2(2)
Tl	O00A	2.370(6)	O007	Tl	O00A	156.0(2)
Tl	N00E	2.358(6)	O007	Tl	N00E	96.5(2)
Tl	N00F	2.348(7)	N00E	Tl	O006	67.6(2)
Tl	N00G	2.311(7)	N00F	Tl	N00J	141.6(2)
Tl	N00H	2.530(6)	N00G	Tl	N00F	120.5(2)
Tl	N00J	2.525(7)	N00J	Tl	N00H	129.8(2)

A low symmetry is observed due to the uncoordinated carboxyl group. Numerous attempts were made to grow an X-ray quality crystal at neutral pH or with an alternative counter ion, for example, tetrabutylammonium, but were not fruitful. Under more basic conditions, it is possible that both carboxylate groups coordinate the metal ion, allowing for a higher degree of symmetry.

Synthesis of H₄pypa-PSMA

As a basis for bioconjugate synthesis, an isothiocyanate derivative of H₄pypa, H₄pypa-NCS, was synthesized using the method previously described by Li et al.^28,30 To deliver ²⁰¹Tl to PSMA-expressing cells, H₄pypa-NCS must be coupled to the PSMA targeting vector via a linker molecule. Structure–activity relationships (SARs) of several PSMA targeting variants have demonstrated the significant role that linker design can have on the pharmacokinetic profile of a tracer.³⁴ The linker used here, incorporating a naphthyl group, was chosen due to the desirable characteristics of PSMA-617 in vivo,³⁴ including the high affinity for PSMA (assisted by the lipophilic linker binding to the hydrophobic PSMA pocket) and fast renal clearance shown by derivative PSMA-617.³⁴

To prepare the PSMA peptide analogue for coupling to H₄pypa-NCS, we adapted a previously reported method, as shown in Scheme 1.³⁵ In brief, l-glutamic acid di-tert-butyl ester was reacted with carbonyldiimidazole (CDI), forming the activated glutamic acid 1. This was then reacted with the cbz-protected l-lysine tert-butyl ester to yield the urea 2. The cbz group was then removed via catalytic hydrogenation, generating the urea derivative 3. Cbz-3-(2-naphthyl)-d-alanine was added via HATU mediated amide coupling in DMF to furnish compound 4 followed by a hydrogenation reaction to remove the cbz group (5). The coupling and cbz deprotection procedures were repeated with cbz-trans-4-(aminomethyl)cyclohexanecarboxylic acid to generate 6 and 7, respectively.

Scheme 1. Reagents and Conditions for the Synthesis of Compounds 1–7.

(i) CDI, MeCN/DMF (4:1), RT, 24 h, 51%. (ii) H-Lys(cbz)-OtBu, DIPEA, DMF, RT, overnight, 83%. (iii) Pd/C, MeOH, RT, overnight, 92%. (iv) Cbz-3-(2-naphthyl)-d-alanine, HATU, DIPEA, DMF, RT, overnight, 55%. (v) Pd/C, MeOH, RT, overnight, 89%. (vi) cbz-trans-4-(aminomethyl)cyclohexanecarboxylic acid, HATU, DIPEA, DMF, RT, overnight, 63%. (vii) Pd/C, MeOH, RT, overnight, 91%.

The reaction of a basic solution of H₄pypa-NCS in chloroform with 7 at ambient temperature led to the formation of conjugate 8 (Scheme 2). The tert-butyl groups of 8 were cleaved using trifluoracetic acid in DCM (1:1) to generate H₄pypa-PSMA (9), which was purified using reversed-phase HPLC. HR-MS confirmed the formation of the final product 9.

Scheme 2. Reagents and Conditions for the Synthesis of Compounds 8 and 9.

(i) CHCl₃, NEt₃, RT, overnight, 56%. (ii) TFA/DCM, overnight, 75%.

The method previously described for the radiolabeling of H₄pypa with ²⁰¹Tl, incorporating prior oxidation of [²⁰¹Tl]Tl⁺ to [²⁰¹Tl]Tl³⁺, was used to radiolabel 9 in good radiochemical yields (95 ± 3%). HPLC analysis indicated that [²⁰¹Tl]Tl-pypa-PSMA eluted at 15.9 min (10.7–24.5 MBq, 20 mmol) (Figure 4). A HPLC UV trace of the unlabeled H₄pypa-PSMA is included in Figure S8.

Figure 4.

Analytical radio-HPLC trace of [²⁰¹Tl]Tl-pypa-PSMA using the HPLC method A (orange = counts per second) (HPLC method A).

[²⁰¹Tl]Tl-pypa-PSMA uptake was then evaluated in DU145 PSMA-positive and PSMA-negative cells after 15 and 60 min of incubation (Figure 5). The amount of cell-associated [²⁰¹Tl]Tl was similar for PSMA-positive and PSMA-negative cells lines, indicating that [²⁰¹Tl]Tl accumulation is not specific to PSMA expression. Additionally, and consistent with this, co-incubation with an excess of the PSMA inhibitor PMPA (2-phosphonomethyl pentanedioic acid) did not meaningfully reduce [²⁰¹Tl]Tl accumulation in either cell line.

Figure 5.

Cell uptake experiments in DU145 PSMA-positive and -negative cell lines with [²⁰¹Tl]-pypa-PSMA or [²⁰¹Tl]TlCl after 15 or 60 min of incubation at 37 °C.

The uptake of [²⁰¹Tl]TlCl was also measured under the same conditions: the amount of ²⁰¹Tl associated with cells was in fact higher for cells incubated with [²⁰¹Tl]TlCl compared to cells incubated with [²⁰¹Tl]Tl-pypa-PSMA. Lastly, co-incubation with an excess of KCl reduced the uptake of [²⁰¹Tl]Tl-pypa-PSMA in both PSMA-positive and PSMA-negative cells lines.

Cumulatively, the data suggest that in the presence of cells, [²⁰¹Tl]Tl-pypa-PSMA releases ²⁰¹Tl and that this dissociation is potentially mediated by the reduction of [²⁰¹Tl]Tl³⁺ to [²⁰¹Tl]Tl⁺ by endogenous reductants. Released [²⁰¹Tl]Tl⁺ then behaves as a K⁺ mimic and is taken up by both PSMA-positive and PSMA-negative cells, with accumulation (via potassium channels, including the Na⁺/K⁺-ATPase pump) inhibited by co-incubation with excess K⁺.

In Vivo Biodistribution in Healthy Animals

To compare the biodistribution of [²⁰¹Tl]Tl-pypa-PSMA, [²⁰¹Tl]TlCl, and [²⁰¹Tl]TlCl₃, all three tracers were administered intravenously via the tail vein to healthy male SCID/beige mice. SPECT/CT images were acquired at 15 min intervals up to 1 h after injection (Figure 6A). Mice were then culled, and organs were collected for ex vivo biodistribution (Figure 6C).

Figure 6.

(A) In vivo images of [²⁰¹Tl]TlCl, [²⁰¹Tl]TlCl₃, and [²⁰¹Tl]Tl-pypa-PSMA at 15, 30, 45, and 60 min in healthy animals (n = 3 per group). (B) Regions of interest (ROIs) drawn from the SPECT images around organs of interest (bladder, heart, and kidneys) for [²⁰¹Tl]TlCl, [²⁰¹Tl]TlCl₃, and [²⁰¹Tl]Tl-pypa-PSMA at 15, 30, 45, and 60 min (n = 1 per radiotracer). (C) The ex vivo biodistribution of [²⁰¹Tl]TlCl, [²⁰¹Tl]TlCl₃, and [²⁰¹Tl]Tl-pypa-PSMA in healthy SCID beige mice culled at 1 h post injection (n = 3 per radiotracer).

SPECT/CT images showed that compared to [²⁰¹Tl]Tl-pypa-PSMA, ²⁰¹Tl administered as either Tl⁺ or Tl³⁺ has an initially high heart uptake at 15 min (4.5% and 3.6% IA (percentage injected activity), respectively) followed by washout, a high degree of retention in the kidneys (10.0–12.9% IA), and relatively low excretion via the urine/bladder (<1.7% IA at all time points) (Figure 6A). In contrast, [²⁰¹Tl]Tl-pypa-PSMA showed a lower myocardial accumulation at 15 min (2.1% IA) and significant [²⁰¹Tl]Tl activity associated with the urine/bladder (8.4% at 60 min).

Ex vivo biodistribution data showed that blood values were low for [²⁰¹Tl]TlCl, [²⁰¹Tl]TlCl₃, and [²⁰¹Tl]Tl-pypa-PSMA with only 0.24, 0.18, and 0.19% activity, respectively, present in blood at 1 h post injection (p.i.) (Figure 6C). [²⁰¹Tl]TlCl and [²⁰¹Tl]TlCl₃ have a high heart uptake of 10.3 ± 0.1% injected activity per gram (IA/g) and 15.4 ± 2.6% IA/g at 1 h p.i., respectively, while [²⁰¹Tl]Tl-pypa-PSMA showed a lower uptake (8.0 ± 0.4% IA/g) (Figure 6C), consistent with SPECT imaging analysis. All three ²⁰¹Tl compounds were predominantly cleared via the kidneys, with [²⁰¹Tl]TlCl having 74.4 ± 6.3% IA/g, [²⁰¹Tl]TlCl₃ having 104.5 ± 6.9% IA/g, and [²⁰¹Tl]Tl-pypa-PSMA having 61.0 ± 3.0% IA/g accumulating in kidneys at 1 h p.i. Clearance through the liver was much lower for all three groups, with [²⁰¹Tl]TlCl having 12.3 ± 0.6% IA/g, [²⁰¹Tl]TlCl₃ having 17.5 ± 2.0% IA/g, and [²⁰¹Tl]Tl-pypa-PSMA having 15.3 ± 4.2% IA/g accumulating in the liver by 1 h p.i.

[²⁰¹Tl]Tl-pypa-PSMA in a Prostate Cancer Animal Model

The biodistribution of [²⁰¹Tl]Tl-pypa-PSMA was studied in SCID/beige mice bearing either (i) DU145 PSMA-expressing tumors (PSMA-positive) or (ii) DU145 tumors that do not express the PSMA receptor (PSMA-negative) to determine if [²⁰¹Tl]Tl-pypa-PSMA accumulated in prostate cancer tissues via PSMA receptor binding. This model has previously been used to show the PSMA-specific uptake of tracers.³⁶ Each group of mice was administered [²⁰¹Tl]Tl-pypa-PSMA (10.7–24.5 MBq, 20 nmol) prior to SPECT/CT scanning for 2 h. At the conclusion of the SPECT/CT scan, each mouse was culled, and organs were dissected, weighed, and counted for radioactivity to obtain quantitative data on radiotracer biodistribution.

SPECT imaging analysis indicated that radioactivity concentration in DU145 PSMA-positive tumors was consistently higher than in DU145 PSMA-negative tumors and, at early time points only, this difference was statistically significant. At 30 min, the ²⁰¹Tl radioactivity concentration in PSMA-positive DU145 tumors measured 3.5 ± 1.4% IA/g (p = 0.0219) and decreased to 2.9 ± 0.9% IA/g at 2 h p.i. (Figure 7C). For PSMA-negative DU145 tumors, ²⁰¹Tl radioactivity concentration at 30 min was 2.1 ± 0.2% IA/g and remained steady until 2 h p.i. Biodistribution data 2 h p.i. corroborated SPECT imaging analysis: ²⁰¹Tl concentration at 2 h p.i. in DU145 PSMA-positive tumors measured 3.7 ± 2.8% IA/g, and in the PSMA-negative tumors, this ²⁰¹Tl radioactivity concentration measured 2.9 ± 1.5% IA/g (Figure 7B). Imaging and ex vivo biodistribution data further evidenced that [²⁰¹Tl]Tl-pypa-PSMA is cleared from the blood mainly via a renal pathway, with high levels of radioactivity observed in the kidneys and bladder/urine evident in both imaging and ex vivo biodistribution data.

Figure 7.

(A) In vivo SPECT image (0–30 min) of [²⁰¹Tl]Tl-pypa-PSMA in mice bearing DU145 positive and negative tumors at 0–30 min. SG = salivary glands, T = tumor, L = liver, K = kidneys, and B = bladder. (B) Ex vivo biodistribution of [²⁰¹Tl]Tl-pypa-PSMA in mice bearing DU145 positive and negative tumors 2 h p.i. (n = 3 per group). (C) Uptake in DU145 PSMA-positive and PSMA-negative tumors using regions of interest drawn from the SPECT images at 30, 60, 90, and 120 min. Tumor to blood (D) and muscle (E) ratios were calculated using biodistribution data (2 h p.i.). Tumor to blood ratios were taken from ROIs drawn on the SPECT images at various time points (F).

Ex vivo biodistribution data also indicated that the tumor/blood ratio for PSMA-positive tumors (11.1 ± 1.4) was significantly higher than that for PSMA-negative tumors (3.9 ± 3.0) at 2 h p.i. (p = 0.0385). The tumor/muscle ratio was similarly higher in mice bearing PSMA-positive tumors (ratio of 1.5 ± 0.4) than in mice bearing PSMA-negative tumors (ratio of 0.7 ± 0.2) (Figure 7E). SPECT image analysis was also used to determine tumor/muscle ratios for [²⁰¹Tl]Tl-pypa-PSMA. The tumor/muscle ratio for animals bearing PSMA-negative tumors was approximately 1 from 30 min to 2 h p.i. However, the tumor/muscle ratio for animals bearing PSMA-positive tumors measured 2.1 ± 0.7 at 30 min and decreased to 1.2 ± 0.4 at 2 h p.i.

Discussion

The premise of this work is that to explore the potential of ²⁰¹Tl as a therapeutic radionuclide, we need better chelators for thallium, capable of both convenient radiolabeling under mild conditions and resistance to dissociation or transchelation in the biological environment. Chelation of Tl⁺ is likely to be challenging due to the similarity of its aqueous chemical properties to those of group 1 alkali metals.³⁷ Therefore, in this study, we chose to focus on Tl³⁺.

Established general-purpose chelators widely used for a range of radiometals in nuclear medicine, such as DOTA and DTPA, are excellent chelators for In³⁺ (the closest periodic analogue of Tl³⁺) and indeed form well-defined complexes with Tl³⁺ with high affinity. Nevertheless, the DOTA and DTPA complexes of Tl³⁺ quickly decompose in serum and cannot be used in Tl³⁺ radiopharmaceuticals.²¹ No binding constants of either Tl⁺ or Tl³⁺ to endogenous serum proteins have been reported in the literature.³⁸ However, Li et al. have estimated the binding of Tl³⁺ to transferrin to have an association constant of 10²² based on the linear relationship that they have observed between the first hydrolysis constant of the other trivalent group 13 metal ions and their transferrin binding constant.³⁹ An alternative route to dissociation of Tl³⁺ complexes, not available to their In³⁺ analogues, is reduction of Tl³⁺ to Tl⁺ by reducing agents present in biological media. Because of this unique vulnerability to reduction of Tl³⁺, the analogy to In³⁺ and other trivalent heavy metals such as bismuth and lanthanides offers only limited guidance in the design of thallium chelators.

Nevertheless, as a starting point for the evaluation of chelators for Tl³⁺, we chose to evaluate a range of polydentate acyclic chelators containing amine, pyridine, and carboxylate donors (Figure 1) that have shown great promise with In³⁺ and other trivalent metal ions. Initial evaluation of the radiolabeling of these ligands with [²⁰¹Tl]Tl³⁺, after oxidation of [²⁰¹Tl]Tl⁺ to [²⁰¹Tl]Tl³⁺ with Iodobeads, showed that all of them were able to chelate [²⁰¹Tl]Tl³⁺ quickly and efficiently under mild conditions and in this respect represent an improvement on DOTA, which required longer reaction times (60 min at room temperature).^23,40 The radiochromatograms of the labeling mixtures each showed single peaks, suggesting the absence of major isomerism or that isomers were rapidly interconvertible (although, at least in the case of the pypa complex, this interpretation is not consistent with ¹H NMR discussed below). On this basis, all four complexes warranted the evaluation of stability in biological media.

The complexes showed no measurable dissociation when incubated in an ammonium acetate buffer or in the presence of transferrin but showed slow decomposition over hours to days in human serum. Although this rate of dissociation is suboptimal, it does not necessarily preclude the use of these chelators in ²⁰¹Tl radiopharmaceuticals, and it is significantly better than that reported for EDTA, DTPA, and DOTA:^23,40 [²⁰¹Tl]Tl-DTPA decomposed quickly in human serum (<10% intact after 1 h), and only 42.7 ± 20.8% of [²⁰¹Tl]Tl-DOTA remained intact after 24 h.²³

The stabilities of [²⁰¹Tl]Tl-pypa, [²⁰¹Tl]Tl-decapa, [²⁰¹Tl]Tl-noneunpa, and [²⁰¹Tl]Tl-neunpa-NH₂ in human serum were comparable after 1 h, with varying degrees of stability after 24 and 48 h. As none of the candidates were ideal with respect to stability, we based the selection of ligands for further evaluation on the ease of incorporation into bioconjugates.^28,30 Additionally, small peptide imaging agents, such as PSMA-617, have very rapid blood clearance (<1 h), so prolonged complex stability (up to 24 or 48 h) is not essential but would be desirable. Thus, for a more detailed evaluation, we selected H₄pypa, for which a PSMA-targeted peptide bioconjugate has recently been reported.²⁸

The ¹H NMR spectrum of the [^natTl]Tl-pypa complex under mildly basic conditions could be interpreted as consistent with the presence of at least two non-interconverting (on the NMR time scale) species. This is not consistent with the HPLC data reported above for the [²⁰¹Tl]Tl³⁺ complex, which may indicate that the HPLC method used was not capable of resolving multiple isomers/species. An alternative explanation is that in the acidic mobile phase used in HPLC analysis, interconversion between multiple species was rapid because of the dissociation of one or more carboxylate donor groups, which is suppressed under the basic conditions of ¹H NMR but would have allowed the substitution of a carboxylate donor by water or an accessible dissociative mechanism of isomerization.

Crystals of the [Tl(Hpypa)] complex, enabling single crystal XRD analysis, were obtained from an acidic solution. The solid phase structure consists of a complex where one carboxylate group is pendant and protonated, with a Tl³⁺ coordination number of eight instead of the potential nine. This suggests that carboxylate group coordination is labile, and while this does not lead to the immediate dissociation or transchelation of Tl³⁺ in biological media, it might be expected to increase vulnerability to reduction by decreasing the coordination number and hence reducing electron density on the metal center. This is pertinent to the biological behavior of the complex bioconjugate, discussed below.

The PSMA-pypa conjugate was efficiently radiolabeled with ²⁰¹Tl under conditions similar to those used for unconjugated H₄pypa. The radiolabeled conjugate was biologically evaluated in vitro and in vivo using the prostate cancer cell line DU145 with and without PSMA expression. The in vitro data (Figure 5) indicate that in the presence of cells, ²⁰¹Tl is released from the labeled bioconjugate complex, likely in the form of Tl⁺: the uptake of radioactivity in cells was initially low but increased with time, and over a period of an hour, the uptake pattern shifted to one that became similar to that of [²⁰¹Tl]TlCl—that is, it reached levels similar to those typically observed for [²⁰¹Tl]TlCl. ²⁰¹Tl radioactivity uptake was similarly inhibited by potassium ions, was not selective for PSMA-positive cells, and was unaffected by a PSMA-specific blocking agent. This behavior can be interpreted on the basis that during the first few minutes of incubation, before the PSMA-specific binding of the radioconjugate has time to occur to a measurable extent, reducing agents secreted by cells into the medium prior to and after addition of the radioconjugate cause the reduction of [²⁰¹Tl]Tl³⁺ to [²⁰¹Tl]Tl⁺ and consequent release from the chelator. As this process develops over a period of minutes, the ²⁰¹Tl radioactivity behaves biologically as Tl⁺ and is taken up efficiently by cells through the activity of the Na⁺/K⁺-ATPase pump, irrespective of PSMA expression.

This interpretation also accounts for the in vivo behavior as observed by SPECT imaging and ex vivo biodistribution. [²⁰¹Tl]Tl⁺ shows the characteristic early myocardium uptake expected of a Na⁺/K⁺-ATPase substrate and myocardial imaging agent. This behavior is not greatly changed when the [²⁰¹Tl]Tl⁺ is oxidized to [²⁰¹Tl]Tl³⁺ before administration, consistent with the very rapid reduction upon initial exposure to the biological environment when unprotected by a Tl³⁺ chelator. The radiolabeled bioconjugate, on the other hand, shows a greatly reduced early uptake in the myocardium, indicating that the chelator survives and protects the Tl³⁺ from reduction and dissociation long enough to allow blood clearance (mainly via the kidney), potentially allowing the opportunity for modest selective uptake in PSMA-expressing tumors, as observed in the in vivo experiments on tumor-bearing mice. Although both suppression of myocardial uptake and a degree of PSMA-specific tumor uptake are observed, the tumor uptake is far below that required for effective imaging or treatment and is much less than is commonly observed with other PSMA-based tracers in this tumor model.³⁶ The results are consistent with the hypothesis that dissociation is promoted by the reduction of the radiometal. This may well be facilitated by the acid-promoted release of a carboxylate donor, as observed in the X-ray crystal structure. The metal is left with reduced electron density and hence greater susceptibility to reduction.

Conclusions

Seeking effective chelators for Tl³⁺, we have evaluated a series of polydentate N, O-ligands that have previously been shown to be effective chelators of other trivalent heavy metal ions often used in nuclear medicine. The findings indicate that the ligands form Tl³⁺ complexes more rapidly and efficiently than conventional chelators (DOTA, DTPA) and resist dissociation or transchelation in buffers free of biomolecules or reducing agents. In serum, however, dissociation occurs over several hours, albeit more slowly than is the case for DOTA and DTPA complexes.²³ With H₄pypa as an example studied in more detail, it became clear that bioreductive dissociation occurred much more quickly in the presence of living cells than in serum, leading to cellular uptake in vitro that was characteristic of [²⁰¹Tl]TlCl. In vivo, a [²⁰¹Tl]Tl-labeled pypa-PSMA conjugate possessed sufficient kinetic stability to show suppression of myocardial uptake and observable but practically inadequate selective delivery to PSMA-positive tumors. We conclude that the class of ligands studied here represents an advance on DOTA and DTPA but is not satisfactory as a basis for thallium-chelating bifunctional chelators. Further design improvement is needed, and this needs to take into account not only simple association/dissociation constants but also protection against reduction—by maximizing electron density donated to metal by maximizing the coordination number (by building in rigidity and preorganization) and incorporating more strongly electron donating donor groups.

Materials and Methods

Unless stated otherwise, chemicals and solvents were purchased from commercial suppliers (Merck, Fisher Scientific, Fluorochem). H₄noneunpa, H₅decapa, and H₄neunpa-NH₂ were synthesized as reported.^24,41,42 [²⁰¹Tl]TlCl in saline was purchased from Curium Pharma, UK. Oxidation was performed using Pierce Iodination beads (Thermo Scientific). ¹H NMR, ¹³C NMR, HSQC, and COSY data were acquired on a Bruker 400 MHz and analyzed using the MestReNova software. Flash chromatography purification was performed on a Biotage Isolera 4 flash chromatography system using Sfar chromatography columns (silica and C18). HPLC was performed on an Agilent 1260 Infinity instrument with UV spectroscopic detection at 254 nm and a Lablogic Flow-Count detector with a Bioscan Inc. B-FC-3200 photomultiplier tube detector and analyzed using the Lablogic Laura software. The mobile phase used for analytical and semipreparative reversed-phase HPLC was composed of (A) water with 0.1% TFA and (B) MeCN with 0.1% TFA. LC/MS data were acquired on an Agilent 1200 Series Liquid Chromatograph with UV spectroscopic detection at 254 nm and the same column details as in reversed-phase HPLC, interfaced with an Advion Expression LC/MS mass spectrometer with an electrospray ionization source. The mobile phase used for LC/MS was composed of (A) water with 0.1% formic acid and (B) MeCN with 0.1% formic acid using an Eclipse XDB-C18 column (4.6 × 150 mm, 5 μm). High-resolution electrospray mass spectrometry was carried out by Dr. Lisa Haigh (mass spectrometry service at Imperial College). Crystallographic data were collected using an Agilent Xcalibur PX Ultra A diffractometer, and the structures were refined using the SHELXTL⁴³ and SHELX-2013⁴⁴ program systems. Radioactive samples were measured using a Capintec CRC25R dose calibrator or an LKB Wallac 1282 Compugamma CS gamma counter for which data were collected using the EdenTerm software. SPECT/CT images were acquired using a NanoSPECT/CT scanner (Mediso Ltd., Budapest, Hungary) with 1.3 mm pinhole collimators and two energy windows at 72.3 keV ± 10% and 140.51 keV ± 10%. Images were reconstructed using the software package HiSPECT (ScivisGmbH) and analyzed using the VivoQuant software (version 3.5, InviCRO Inc.). Lyophilization was performed using an Edwards Freeze-Dryer Modulyo.

Oxidation of [²⁰¹Tl]Tl⁺ to [²⁰¹Tl]Tl³⁺

This procedure is adapted from Rigby et al.²³ [²⁰¹Tl]TlCl (40 MBq, 100 μL) was added to one Pierce iodination bead in a 1.5 mL Eppendorf tube followed by the addition of HCl (0.5 M, 10 μL). The tube was vortexed, a small aliquot (2 μL) was removed for ITLC analysis, and the supernatant was pipetted into a clean tube. To measure the radiochemical conversion, acetone was used as the mobile phase and silica gel ITLC strips (iTLC-SG) were used as the stationary phase, giving good separation between [²⁰¹Tl]Tl⁺ (R_f = 0) and [²⁰¹Tl]Tl³⁺ (R_f = 1).

Radiolabeling H₄pypa, H₅decapa, H₄neunpa-NH₂, and H₄noneunpa with [²⁰¹Tl]Tl³⁺

Chelators studied were H₄pypa, H₅decapa, H₄neunpa-NH₂, and H₄noneunpa. The chelator solution (1 mg/mL in water, 20 μL) was added to [²⁰¹Tl]TlCl₃ (39.5 MBq, 108 μL) and 1 M ammonium acetate (pH 5, 20 μL). This was vortexed and agitated in a Thermomixer (500 rpm) at RT for 10 min. Radiochemical yield and purity were evaluated using RP-ITLC (unbound [²⁰¹Tl]Tl⁺, [²⁰¹Tl]Tl³⁺R_f = 0, [²⁰¹Tl]Tl³⁺ complex R_f = 1) and HPLC (method 1). To measure radiochemical conversion, reversed-phase TLC plates (TLC Silica Gel 60 RP-18 F254s MS-grade) were used as the stationary phase, and acetonitrile (30%) with water was used as the mobile phase. All TLC plates were imaged using a Cyclone Plus Phosphor Imager (PerkinElmer, Inc. USA).

Stability of [²⁰¹Tl]Tl-pypa, [²⁰¹Tl]Tl-decapa, [²⁰¹Tl]Tl-neunpa-NH₂, and [²⁰¹Tl]Tl-noneunpa

Human serum (300 μL, Merck) was added to an Eppendorf tube followed by the addition of [²⁰¹Tl]Tl-pypa, [²⁰¹Tl]Tl-decapa, [²⁰¹Tl]Tl-neunpa-NH₂, or [²⁰¹Tl]Tl-noneunpa (200 kBq, 12–15 μL). The tubes were then incubated at 37 °C for up to 48 h. Aliquots (2 μL) were removed at intervals and analyzed using RP-TLC to assess the stability. In addition to human serum, this process was repeated using an ammonium acetate solution (1 M, pH 5).

Radiolabeling H₄pypa-PSMA

A 1 mg/mL solution of H₄pypa-PSMA was prepared in an ammonium acetate solution (1 M, pH 5). An aliquot of the H₄pypa-PSMA solution (20 μL, 0.1 μM) was added to [²⁰¹Tl]TlCl₃ (110 MBq, 200 μL) followed by ammonium acetate (1 M, pH 5, 50 μL). This solution was vortexed and agitated in a Thermomixer (500 rpm) at RT for 10 min. Radiochemical yield and purity were evaluated using HPLC (method A, [²⁰¹Tl]TlCl₃t_R = 2.03 min; [²⁰¹Tl]Tl-pypa-PSMA t_R = 15.02 min).

Tissue Culture

DU145 (PSMA-negative) and DU145-PSMA (PSMA-positive) human prostate cancer cells were cultured in an RPMI-1640 medium supplemented with 10% fetal bovine serum, 2 mM l-glutamine, and penicillin/streptomycin (Sigma-Aldrich, UK) and maintained at 37 °C in a humidified atmosphere with 5% CO₂.⁴⁵ PSMA expression was evaluated using FACS, and the results can be found in Figure S1.

SPECT Scanning and Biodistribution in Healthy and DU145-PSMA Tumor-Bearing Animals

Animal studies were carried out in accordance with the UK Home Office Animals (Scientific Procedures) Act 1986. Experiments complied with UK Research Councils’ and Medical Research Charities’ guidelines on responsibility in the use of animals in bioscience research under UK Home Office project and personal licenses. The reporting of this study complied with the Animal Research: Reporting In Vivo Experiments (ARRIVE) guidelines (https://www.nc3rs.org.uk/arrive-guidelines).

Healthy SCID/beige animals (male 5–7 weeks old, n = 3 per radiotracer) were injected via tail vein injection under isoflurane anesthesia (1.5–2.5% in oxygen at 1 L/min) with [²⁰¹Tl]TlCl (17–22.9 MBq), [²⁰¹Tl]TlCl₃ (11.2–23.8 MBq), or [²⁰¹Tl]Tl-pypa-PSMA (14.1–16.9 MBq). Mice were then kept under continuous anesthesia on a heated pad for the duration of the experiment (1 h), and one mouse per group was imaged by SPECT/CT until 1 h post injection when animals were euthanized by cervical dislocation.

To study tracer uptake in tumors, SCID/beige mice (male 5–7 weeks old, n = 3 per group) were injected subcutaneously with DU145-PSMA or DU145 cells (4 × 10⁶ cells in 100 mL PBS) in the left shoulder. Once tumors had reached 5–10 mm in diameter (4–5 weeks after inoculation), [²⁰¹Tl]Tl-pypa-PSMA (10.7–24.5 MBq, 20 mmol) was administered via tail vein injection under isoflurane anesthesia. Mice were maintained under continuous anesthesia and imaged by SPECT/CT for up to 2 h post injection. Animals were then euthanized by cervical dislocation. SPECT images were reconstructed using the HiSPECT (Scivis GmbH) reconstruction software package at 0.3 mm isotropic voxel size using standard reconstruction with 35% smoothing and nine iterations. After euthanasia, organs were harvested from the mice, weighed, and gamma counted.

Image Analysis

Images were analyzed using VivoQuant 2.5 (InviCRO LLC., Boston, USA), enabling the delineation of regions of interest (ROIs) for quantification of radioactivity. ROIs for the tumor and organs (heart, muscle, etc.) were drawn using CT images, and volumes were determined. The total activity in the whole animal (excluding the majority of tail, out of SPECT field of view) at the time of [²⁰¹Tl] agents’ administration was defined as the injected activity (IA), and the percentage of injected activity per cm³ (% IA/cm³) and amount of radioactivity in tissues (MBq) were determined. A 5 mL syringe with 3 mL of [²⁰¹Tl]TlCl (40 MBq) was used to calibrate the SPECT/CT and ensure correct co-registration between the SPECT and CT.

Statistical Analysis

Data are reported as average ± standard deviation. Statistical analysis was performed using Graphpad Prism Version 7.0c with unpaired t tests used in uptake and a two-way ANOVA with Sidak’s multiple comparisons test used for in vivo studies; *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001.

Synthesis

Di-tert-butyl (1H-Imidazole-1-carbonyl)glutamate (1)

1 was synthesized using a previously reported method by Duspara et al.³⁵l-Glutamic acid di-tert-butyl hydrochloride (3.56 g, 12.04 mmol) and carbonyldiimidazole (2.15 g, 13.24 mmol) were dissolved in a 1:5 mixture of DMF/MeCN (50 mL) and stirred at RT overnight. MeCN was then removed in vacuo, and the remaining DMF was diluted with EtOAc (100 mL) and washed with water (3 × 50 mL) and brine (3 × 50 mL). The organic layer was then dried over magnesium sulfate, and the solvent was removed in vacuo. The crude product was then purified using a Biotage Isolera flash chromatography system (20–80% EtOAc/petroleum ether) to yield the desired product as a colorless oil that solidified upon standing (2.1 g, 51%). ¹H NMR (400 MHz, chloroform-d) δ 8.16 (t, J = 1.1 Hz, 1H), 7.57 (d, J = 6.8 Hz, 1H), 7.41 (t, J = 1.5 Hz, 1H), 7.07 (dd, J = 1.6, 0.9 Hz, 1H), 2.48–2.38 (m, 2H), 2.27–2.05 (m, 2H), 1.47 (s, 9H), 1.43 (s, 9H). ¹³C NMR (101 MHz, chloroform-d) δ 174.00, 173.50, 173.34, 171.69, 162.78, 157.36, 149.11, 136.24, 135.17, 129.96, 121.71, 116.40, 77.43, 77.11, 76.79, 53.49, 52.72, 52.53, 52.38, 52.35, 52.12, 51.83, 51.78, 36.61, 31.51, 30.33, 30.07, 27.98, 27.94, 27.78, 26.15. ESI-MS: calc. for [C₁₇H₂₇N₃O₅ + H]⁺ 354.42; found 354.35.

Tri-tert-butyl 3,11-Dioxo-1-phenyl-2-oxa-4,10,12-triazapentadecane-9,13,15-tricarboxylate (2)

2 was synthesized by adapting a previously reported method by Duspara et al.³⁵ H-Lys(Z)-OtBu·HCl (3.47 g, 9.34 mmol) was dissolved in DMF (20 mL). DIPEA (1.63 mL, 9.34 mmol) was added to the solution followed by 1 (3 g, 8.49 mmol, dissolved in 10 mL DMF) dropwise and allowed to stir overnight at RT. The reaction was diluted with EtOAc (100 mL) and washed with water (3 × 100 mL) and brine (3 × 100 mL). The organic layer was then dried over magnesium sulfate, and the solvent was removed in vacuo. The crude product was purified using a Biotage Isolera flash chromatography system (20–80% EtOAc/petroleum ether) to yield the desired product as a colorless oil (4.5 g, 83%). ¹H NMR (400 MHz, chloroform-d) δ 7.38–7.30 (m, 5H), 5.22–5.02 (m, 5H), 4.33 (dd, J = 8.1, 4.9 Hz, 2H), 3.17 (dd, J = 6.4, 3.7 Hz, 2H), 2.28 (td, J = 9.6, 6.4 Hz, 2H), 1.44 (d, J = 1.1 Hz, 18H), 1.43 (s, 10H). ¹³C NMR (101 MHz, chloroform-d) δ 172.41, 156.85, 156.59, 136.71, 128.46, 128.05, 128.00, 82.10, 81.75, 80.51, 77.33, 77.02, 76.70, 66.55, 53.29, 53.02, 40.65, 32.65, 31.60, 29.36, 28.36, 28.08, 28.03, 28.00, 22.24. ESI-MS: calc. for [C₃₂H₅₁N₃O₉ + H]⁺ 622.36; found 622.3.

Di-tert-butyl ((6-Amino-1-(tert-butoxy)-1-oxohexan-2-yl)carbamoyl)glutamate (3)

The cbz protected urea 2 (3.6 g, 5.79 mmol) was dissolved in methanol (20 mL) and added to Pd/C (10%, 0.125 g, 1.16 mmol). The reaction flask was evacuated before being flushed with two balloons of hydrogen gas and a third balloon left connected to the vessel for the duration of the experiment. TLC analysis of the reaction showed completion after 90 min. The Pd/C was removed via filtration through Celite, and the solvent was removed in vacuo to yield a colorless oil. This was then purified using the Biotage Isolera flash chromatography system (reversed-phase SFar C18 column, 0–60% MeCN/0.1% FA:H₂O/0.1% FA) to yield the desired product as a colorless oil that solidified under a vacuum (2.62 g, 92%). ¹H NMR (400 MHz, chloroform-d) δ 6.37 (d, J = 8.1 Hz, 1H), 6.23 (d, J = 8.0 Hz, 1H), 4.31 (s, 2H), 2.98 (s, 2H), 2.32 (dd, J = 6.5, 3.2 Hz, 2H), 1.71 (s, 4H), 1.44 (d, J = 1.8 Hz, 18H), 1.43 (s, 10H). ¹³C NMR (101 MHz, chloroform-d) δ 173.62, 172.80, 172.36, 157.65, 82.11, 81.54, 80.53, 77.33, 77.01, 76.70, 53.12, 52.88, 39.20, 31.78, 31.28, 28.10, 28.04, 27.20, 21.68. ESI-MS: calc. for [C₂₄H₄₅N₃O₇ + H]⁺ 488.64; found 488.45.

Di-tert-butyl ((6-(2-(((Benzyloxy)carbonyl)amino)-3-(naphthalen-2-yl)propanamido)-1-(tert-butoxy)-1-oxohexan-2-yl)carbamoyl)glutamate (4)

Z-3-(2-naphthyl)-d-alanine (0.395 g, 1.13 mmol) and HATU (0.858 g, 2.26 mmol) were dissolved in dry DMF (10 mL) followed by the addition of DIPEA (0.54 mL, 3.08 mmol), with the solution turning from colorless to yellow. This was left to stir for 15 min at RT, after which 3 (0.5 g, 1.03 mmol), dissolved in dry DMF (5 mL), was added to the stirring solution and left at RT to stir overnight. During this time, the reaction had turned dark brown in color. The reaction was diluted with EtOAc (100 mL) and washed with water (3 × 50 mL) and brine (3 × 50 mL). The organic layer was then dried over magnesium sulfate, and the solvent was removed in vacuo. The crude product was then purified using a Biotage Isolera flash chromatography system (20–70% EtOAc/petroleum ether) to yield the desired product as a yellow oil (0.46 g, 55%). ¹H NMR (400 MHz, chloroform-d) δ 7.83 (s, 3H), 7.77 (s, 1H), 7.71 (d, J = 8.2 Hz, 3H), 7.66 (s, 2H), 7.59 (d, J = 10.9 Hz, 3H), 7.51–7.33 (m, 4H), 7.27 (d, J = 11.8 Hz, 4H), 7.22–7.11 (m, 1H), 5.09 (s, 5H), 4.96 (d, J = 13.0 Hz, 1H), 4.35 (s, 1H), 4.12 (q, J = 7.1 Hz, 1H), 3.43 (s, 1H), 3.23 (s, 1H), 3.19–3.03 (m, 1H), 3.03–2.73 (m, 1H), 1.93–1.64 (m, 2H), 1.44 (d, J = 2.5 Hz, 20H), 1.41 (s, 9H). ¹³C NMR (101 MHz, chloroform-d) δ 133.47, 132.36, 128.52, 128.26, 128.14, 127.67, 126.23, 80.57, 77.34, 77.02, 76.70, 60.39, 53.41, 53.11, 31.83, 28.11, 28.02, 21.04, 14.20. HR-ESI-MS: calc. for [C₄₅H₆₂N₄O₁₀ + H]⁺ 819.4544; found 819.4550.

Di-tert-butyl ((6-(2-Amino-3-(naphthalen-2-yl)propanamido)-1-(tert-butoxy)-1-oxohexan-2-yl)carbamoyl)glutamate (5)

4 (0.185 g, 0.24 mmol) was dissolved in methanol (10 mL) and added to Pd/C (10%, 0.007 g, 0.05 mmol). The reaction flask was evacuated before being flushed with two balloons of hydrogen gas and a third balloon left connected to the vessel for the duration of the experiment. TLC analysis of the reaction showed completion after stirring overnight. The Pd/C was removed via filtration through Celite, and the solvent was removed in vacuo to yield a pale-yellow oil (0.149 g, 89%). ¹H NMR (400 MHz, chloroform-d) δ 7.84–7.73 (m, 3H), 7.67 (s, 1H), 7.57 (s, 1H), 7.47–7.41 (m, 2H), 7.35 (d, J = 8.2 Hz, 1H), 4.29 (s, 1H), 4.19 (s, 2H), 3.32 (s, 1H), 3.20 (s, 1H), 3.08 (s, 2H), 2.38–2.19 (m, 2H), 2.13–1.95 (m, 1H), 1.91–1.78 (m, 1H), 1.74 (t, J = 3.3 Hz, 2H), 1.41 (s, 10H), 1.40 (d, J = 1.6 Hz, 18H). ¹³C NMR (101 MHz, chloroform-d) δ 172.93, 172.83, 172.50, 170.65, 157.44, 133.44, 133.20, 132.53, 128.53, 128.40, 127.68, 127.64, 127.29, 126.28, 125.91, 82.13, 81.60, 80.62, 77.35, 77.03, 76.72, 55.40, 53.43, 52.96, 39.02, 38.65, 31.71, 28.59, 28.28, 28.06, 27.99, 27.97, 22.00. HR-ESI-MS: calc. for [C₃₇H₅₆N₄O₈ + H]⁺ 685.4176; found 685.4188.

Di-tert-butyl ((6-(2-(4-((((Benzyloxy)carbonyl)amino)methyl)cyclohexane-1-carboxamido)-3-(naphthalen-2-yl)propanamido)-1-(tert-butoxy)-1-oxohexan-2-yl)carbamoyl)glutamate (6)

Trans-4-(cbz-amino)cyclohexanecarboxylic acid (0.153 g, 0.53 mmol) and HATU (0.852 g, 1.1 mmol) were dissolved in dry DMF followed by the addition of DIPEA (0.16 mL, 1.56 mmol), with the solution turning from colorless to yellow. This was left to stir for 15 min at RT, after which 5 (0.3 g, 0.44 mmol), dissolved in dry DMF (5 mL), was added dropwise to the stirring solution and left to stir overnight. During this time, the reaction turned dark brown in color. The reaction was diluted with EtOAc (100 mL) and washed with water (3 × 50 mL) and brine (3 × 50 mL). The organic layer was then dried over magnesium sulfate, and the solvent was removed in vacuo. The crude product was then purified using a Biotage Isolera flash chromatography system (20–70% EtOAc/petroleum ether) to yield the desired product as a yellow oil (0.24 g, 63%). ¹H NMR (400 MHz, chloroform-d) δ 7.89–7.78 (m, 1H), 7.73 (t, J = 6.9 Hz, 2H), 7.63 (d, J = 1.6 Hz, 1H), 7.52–7.42 (m, 2H), 7.34 (d, J = 4.6 Hz, 6H), 5.26 (d, J = 7.7 Hz, 1H), 5.20 (d, J = 8.3 Hz, 1H), 4.90 (s, 1H), 4.71 (d, J = 7.4 Hz, 1H), 4.37–4.26 (m, 1H), 4.14 (q, J = 4.1 Hz, 1H), 3.39–3.26 (m, 1H), 3.20 (d, J = 6.4 Hz, 1H), 3.17–3.07 (m, 4H), 3.02 (t, J = 6.4 Hz, 2H), 2.32 (td, J = 9.5, 6.4 Hz, 2H), 2.06 (tdd, J = 11.3, 5.2, 3.1 Hz, 2H), 1.89–1.72 (m, 8H), 1.70–1.61 (m, 2H), 1.44 (d, J = 1.5 Hz, 18H), 1.42 (s, 10H). ¹³C NMR (101 MHz, chloroform-d) δ 175.84, 172.80, 172.51, 172.46, 171.54, 157.03, 136.59, 134.54, 133.41, 132.32, 128.51, 128.12, 127.62, 126.23, 125.79, 82.11, 81.46, 80.63, 77.34, 77.03, 76.71, 66.68, 54.79, 53.23, 52.94, 45.02, 39.16, 38.39, 37.55, 31.78, 31.66, 29.67, 28.92, 28.74, 28.59, 28.51, 28.11, 28.03, 21.87. HR-ESI-MS: calc. for [C₅₃H₇₅N₅O₁₁ + H]⁺ 958.5541; found 958.5559.

Di-tert-butyl ((6-(2-(4-(Aminomethyl)cyclohexane-1-carboxamido)-3-(naphthalen-2-yl)propanamido)-1-(tert-butoxy)-1-oxohexan-2-yl)carbamoyl)glutamate (7)

6 (0.081 g, 0.83 mmol) was dissolved in dry MeOH (10 mL) and Pd/C (10%, 0.003 g, 0.016 mmol) added. The reaction flask was evacuated before being flushed with two balloons of hydrogen gas and a third balloon left in the vessel for the duration of the experiment. TLC analysis of the reaction showed completion after stirring overnight. The Pd/C was removed via filtration through Celite, and the solvent was removed in vacuo to yield 7 as a yellow oil (0.062 g, 91%). ¹H NMR (400 MHz, chloroform-d) δ 7.77–7.67 (m, 3H), 7.65 (s, 1H), 7.42–7.31 (m, 3H), 5.75 (d, J = 18.6 Hz, 2H), 4.74 (s, 1H), 4.28 (d, J = 6.6 Hz, 1H), 4.08 (d, J = 6.4 Hz, 1H), 3.17 (d, J = 10.3 Hz, 2H), 3.06 (s, 2H), 2.72 (s, 3H), 2.31 (q, J = 7.0, 6.3 Hz, 3H), 2.16–1.98 (m, 4H), 1.92–1.81 (m, 2H), 1.69 (s, 4H), 1.43 (d, J = 1.7 Hz, 18H), 1.41 (s, 9H). HR-ESI-MS: calc. for [C₄₅H₆₉N₅O₉ + H]⁺ 824.5168; found 824.5174.

6,6′-((((4-(4-(3-((4-((7,11-Bis(tert-butoxycarbonyl)-2,2-dimethyl-19-(naphthalen-2-yl)-4,9,17-trioxo-3-oxa-8,10,16-triazanonadecan-18-yl)carbamoyl)cyclohexyl)methyl)thioureido)phenethoxy)pyridine-2,6-diyl)bis(methylene))bis((carboxymethyl)azanediyl))bis(methylene))dipicolinic Acid (26)

20 (0.01 g, 14.3 μmol) was dissolved in CHCl₃ (1 mL). 7 (0.012 g, 14.3 μmol) was separately dissolved in CHCl₃ (1 mL). Triethylamine (2 × 4 μL, 57 μmol) was added to each solution, and then both solutions were mixed. This was allowed to stir at RT overnight, after which the CHCl₃ was removed in vacuo. The product was then purified using reversed-phase semipreparative HPLC (A: MeCN/0.1% TFA, B: H₂O/0.1% TFA, 5–80% A over 60 min, 4 mL/min). UV-active fractions were analyzed using LC–MS (HPLC method B); pure fractions were combined and freeze-dried to yield the product as a white solid (0.012 g, 56%) ESI-MS: calc. for [C₇₉H₁₀₁N₁₁O₁₈S + H]⁺ 1525.80; found 1525.04.

((5-(2-(4-((3-(4-(2-((2,6-Bis(((carboxymethyl)((6-carboxypyridin-2-yl)methyl)amino)methyl)pyridin-4-yl)oxy)ethyl)phenyl)thioureido)methyl)cyclohexane-1-carboxamido)-3-(naphthalen-2-yl)propanamido)-1-carboxypentyl)carbamoyl)glutamic Acid (27)

21 (0.01 g, 6.6 μmol) was dissolved in DCM/TFA (1:1, 4 mL) and allowed to stir at RT overnight. The solution was then concentrated in vacuo, and the residue was redissolved in deionized water and purified using reversed-phase semipreparative HPLC (A: MeCN/0.1% TFA, B: H₂O/0.1% TFA, 5–80% A over 60 min, 4 mL/min). UV-active fractions were analyzed using LC–MS (HPLC method B); pure fractions were combined and freeze dried to yield the product as a white solid (0.006 g, 75%). HR-ESI-MS: calc. for [C₆₇H₇₇N₁₁O₁₈S + H]⁺ 1356.5247; found 1356.5298.

[^natTl]Tl-pypa

TlCl₃ hydrate (0.05 g, 96 μmol) was dissolved in ammonium acetate solution (1 M, pH 5, 0.5 mL) and added to a solution of H₄pypa (15) (0.037 g, 96 μmol) also dissolved in ammonium acetate solution (1 M, pH 5, 0.5 mL). The mixture was agitated for 5 min at RT, and an aliquot was removed for analysis using LC–MS. The complex was purified using reversed-phase preparative HPLC (A: MeCN/0.1% TFA, B: H₂O/0.1% TFA, 5–60% A over 40 min, 10 mL/min). UV-active fractions were analyzed using LC–MS (HPLC method B); pure fractions were combined and freeze-dried to yield the product as a white solid (0.055 g, 80%) HR-ESI-MS: calc. for [C₂₅H₂₃N₅O₈²⁰⁵Tl + H]⁺ 726.1291; found 726.1306.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.bioconjchem.2c00284.

• NMR (¹H, ¹³C, COSY, HSQC) spectra for compounds 1–7 and [^natTl]Tl-pypa; HR-MS spectra for compounds 1–7 and [^natTl]Tl-pypa; X-ray crystallographic data for [^natTl] [Tl(Hpypa)]; validation of PSMA expression in the cells used; description of the HPLC methods used, as well as the [²⁰¹Tl]TlCl and [²⁰¹Tl]TlCl₃ HPLC traces and the reactions between [²⁰¹Tl]TlCl with each chelator; further stability data against transmetalation in excess transferrin and in the buffer for all four [²⁰¹Tl]Tl-complexes; analytical HPLC trace for H₄pypa-PSMA; and additional information on [²⁰¹Tl]Tl-pypa-PSMA cell uptake experiments (PDF)

Author Contributions

^# S.Y.A.T and V.A. contributed equally as last authors.

The authors declare no competing financial interest.