Radiolabeling of Human Serum Albumin With Terbium-161 Using Mild Conditions and Evaluation of in vivo Stability

Targeted radionuclide therapy (TRNT) is a promising approach for cancer therapy. Terbium has four medically interesting isotopes (149Tb, 152Tb, 155Tb and 161Tb) which span the entire radiopharmaceutical space (TRNT, PET and SPECT imaging). Since the same element is used, accessing the various diagnostic or therapeutic properties without changing radiochemical procedures and pharmacokinetic properties is advantageous. The use of (heat-sensitive) biomolecules as vector molecule with high affinity and selectivity for a certain molecular target is promising. However, mild radiolabeling conditions are required to prevent thermal degradation of the biomolecule. Herein, we report the evaluation of potential bifunctional chelators for Tb-labeling of heat-sensitive biomolecules using human serum albumin (HSA) to assess the in vivo stability of the constructs. p-SCN-Bn-CHX-A”-DTPA, p-SCN-Bn-DOTA, p-NCS-Bz-DOTA-GA and p-SCN-3p-C-NETA were conjugated to HSA via a lysine coupling method. All HSA-constructs were labeled with [161Tb]TbCl3 at 40°C with radiochemical yields higher than 98%. The radiolabeled constructs were stable in human serum up to 24 h at 37°C. 161Tb-HSA-constructs were injected in mice to evaluate their in vivo stability. Increasing bone accumulation as a function of time was observed for [161Tb]TbCl3 and [161Tb]Tb-DTPA-CHX-A”-Bn-HSA, while negligible bone uptake was observed with the DOTA, DOTA-GA and NETA variants over a 7-day period. The results indicate that the p-SCN-Bn-DOTA, p-NCS-Bz-DOTA-GA and p-SCN-3p-C-NETA are suitable bifunctional ligands for Tb-based radiopharmaceuticals, allowing for high yield radiolabeling in mild conditions.

Up until now, most terbium-labeling reactions were reported with ligands and peptides which are compatible with high radiolabeling temperatures (3,9). Biomolecule-based radionuclide therapies, e.g., using trastuzumab for cancers with overexpression of human epidermal growth factor receptor-2 (HER-2), have paved the way for a more targeted approach to theranostics (21). Radiolabeling with terbium, and most of the other f-block elements, were performed at temperatures (>90 • C), well-exceeding the temperatures compatible with heat-sensitive biomolecules such as monoclonal antibodies and antibody fragments (22). In this study, we developed a mild radiolabeling protocol (reaction temperature of 40 • C in an aqueous buffer), with a series of commonly used bifunctional ligands (Figure 1). We then used human serum albumin (HSA, 66.5 kDa) as a model protein to assess the in vitro and in vivo stability of the corresponding 161 Tb-labeled HSA conjugates. The high solubility, stability and plasma half-life of approximately 16-18 h make HSA the ideal vector to evaluate the stability of the 161 Tb-chelates in vivo.

Animals
Healthy albino Naval Medical Research Institute (NMRI) mice (age: 6-8 weeks, Envigo, Gannat, France) were housed in individually ventilated cages (IVC) in a regulated environment (22 • C, humidity, 12 h day/night cycle), with food and water. Animal experiments were conducted according to the Belgian code of practice and use of animal experiments were approved by the ethical committee for animal care from KU Leuven.

Instrumentation and Characterization
Mass spectra were recorded on an ultra-high-resolution timeof-flight mass spectrometer with electrospray ionization (ESI) (Bruker MaXis Impact, Bremen, Germany), coupled to a Dionex Ultimate 3,000 UPLC System (Thermo Fisher Scientific, USA). Quantification of protein concentration was determined using a microvolume UV-Vis spectrophotometer (NanoDrop One, Thermo Fisher Scientific). Quality assurance of the derivatized HSA constructs and 161 Tb-HSA constructs were carried out with size-exclusion chromatography (SEC) using a Superdex 200 10/300 GL column (GE Healthcare Bio-Science AB, Uppsala, Sweden), eluted with a sodium phosphate buffer (0.15 M sodium chloride, 0.01 M phosphate, pH 7.4, Thermo Fisher) at a flow rate of 0.75 mL/min. The column effluent was passed through a UV detector (2998 PDA detector, Waters) in series with a 3-inch NaI(Tl) radioactivity detector. Gamma counting was performed on a Wizard 2 3470 [crystal: NaI (Tl), 50 mm in height, 32 mm in diameter, dead time 2.5 µs; Perkin Elmer, Germany], with a detection profile referenced for 161 Tb decay (4). Counts were corrected for background radiation, physical decay and counter dead time.

Production of [ 161 Tb]TbCl 3
[ 161 Tb]TbCl 3 was produced using a method adapted from literature (4). In brief, enriched 160 Gd 2 O 3 (1.0 mg, 98.2 %, Isoflex USA) was loaded as a nitrate salt into a quartz ampoule and sealed. The ampoule was sealed inside an aluminum capsule and was irradiated for 10 days in the BR2 Reactor at the Belgian Nuclear Research Centre (SCK CEN) at a thermal neutron flux of 3.0 x 10 14 n/cm 2 /s. Following the irradiation and subsequent cooling for 5 days, the irradiated material was dissolved in tracemetal grade water. High-pressure ion chromatography (HPIC, Shimadzu), with a strong cation exchange column (ø: 6 mm, l: 50 mm, Shodex IC R-621), was used to separate the [ 161 Tb] from the [ 160 Gd] target matrix by elution with α-hydroxyisobutyric acid, with ammonium hydroxide (trace-metal grade) (added to adjust to pH 4.5). The collected fractions containing [ 161 Tb] were combined and concentrated by loading them onto a column packed with extraction resin (ø: 2.1 mm, l: 30 mm, LN3, TrisKem International) and eluted with 50 mM hydrochloric acid (trace-metal free). The isolated solutions of [ 161 Tb]TbCl 3 had a radionuclidical purity of 99.998% (determined by gamma spectroscopy), a concentration of ∼ 0.99 MBq/µL, and specific activity of ∼ 3.6 TBq/mg.

Human Serum Albumin (HSA) Ligand Constructs
A five-molar excess of bifunctional ligand (3 µmol) in 200 µl of a sodium bicarbonate solution (0.05 M, pH 8.5, 1.5 % DMSO) was added dropwise to a stirring solution of human serum albumin (400 µL, 0.6 µmol, CAF-DCF, Brussels, Belgium) in sodium bicarbonate (0.05 M, pH 8.5) in a LoBind vial (Eppendorf, Aarschot, Belgium). The mixture was then stirred for 2 h at room temperature and the conjugate was purified using a size exclusion chromatography cartridge (PD-10 column, GE Healthcare Bio-Science AB, Uppsala, Sweden) eluted with sodium acetate buffer (0.1 M, pH 4.7). The concentration of the HSA-ligand construct in the final reaction product was determined using spectrophotometry at 280 nm (NanoDrop R One, Thermo Fisher Scientific), with ε = 35,700 L/mol/cm and M = 66,477 g/mol. The purified product was analyzed using SEC using the method described above. UV detection of the eluate was performed at 280 nm. The number of chelators per protein was estimated by ESI-TOF-HRMS analysis considering the most abundant peak. The system was equipped with a Waters Acquity

In vitro Stability Studies
Stability of ligand complexes in phosphate buffered saline pH 7.4: The radiolabeled ligands were purified using a C18 Plus SEP-PAK cartridge (Waters, Antwerp, Belgium) by loading the reaction mixture, rinsing with water (5 mL) to remove unreacted [ 161 Tb]TbCl 3 , and eluting the purified complex with abs. ethanol (0.5 mL). 80 µL of the ethanolic solution was added to 720 µL of sodium phosphate buffer (0.15 M sodium chloride, 0.01 M phosphate, pH 7.4, Thermo Fisher) and incubated at 37 • C (n = 3). Samples were collected at different time points (10 min, 1 h, 4 h, and 24 h) and the percentage of intact 161 Tb-complex was determined using the same iTLC chromatography system as used above.
Stability of HSA-ligand in human serum: After radiolabeling and without purification, 50 µL of the 161 Tb-HSA radiolabeling solution was added to 720 µL human serum (Sigma Aldrich) and incubated at 37 • C (n = 3). Samples were collected at different time points (10 min, 1 h, 4 h, and 24 h) and the percentage of intact 161 Tb-HSA construct was determined using the same instant thin-layered liquid chromatography system as used in radiolabeling and referenced to the initial radiochemical yield. The in vitro stability was confirmed with the radio-SEC method described above at 1, 4 and 24 h.
Competition studies with EDTA: After radiolabeling and without purification, 50 µL of the 161 Tb-HSA radiolabeling solution was added to 50 µL EDTA solution (10 mM, 0.1M PBS, pH 7.4, Sigma Aldrich) and incubated at 37 • C (n = 3). Samples were collected at different time points after incubation (1 h, 4 h, and 24 h) and the percentage of intact 161 Tb-HSA construct was determined using the same iTLC method mentioned above.

Biodistribution Studies
Mice were anesthetized with 2.5% isoflurane in O 2 at a flow of 1 L/min and injected with ∼1 MBq of [ 161 Tb]TbCl 3 or 161 Tb-HSA construct (0.1-0.3 nmoles) via a tail vein. Animals were sacrificed by decapitation at 10 min, 1 h, 4 h, 24 h, or 7 days post injection (n = 3 animals per time point). Blood and organs were collected in tubes, weighed, and radioactivity was determined using an automated gamma counter as described above. Results are presented as standardized uptake values [SUV; determined using SUV = (MBq tissue /g tissue )/(MBq injected /g mouse )]. For calculation of percentage injected dose (%ID) in blood, bone and muscle, masses were assumed to be 7, 12, and 40% of mouse body weight, respectively (23,24). Blood data points (%ID calculated ) were fitted to a standard half-life equation (least-squares regression analysis), %ID calc = A · 0.5 k t , where A = constant, t = hours after injection (h), and k = 1/plasma half-life (h −1 ).

Statistical Analysis
Quantitative data are expressed as mean ± SD unless stated otherwise. Means were compared using a mixed model ANOVA analysis in GraphPad Prism 9.1.2. Values were determined to be statistically significant for p-values less than the threshold value of 0.05.

Optimization of Radiolabeling Conditions for Low Temperature Labeling
Radiolabeling efficiency of all ligands (DTPA, DOTA, DOTA-GA, NETA) was evaluated using four different ligand concentrations (0.1, 1, 5 and 10 µM) at 25 • C and 40 • C after 60 min reaction time. Results of the radiolabeling can be found in Figures 2A,B. Radiolabeling with DTPA resulted in >98% radiolabeling efficiency at all tested concentrations, even at 25 • C. All other ligands required a temperature of 40 • C to efficiently (>90%) chelate the terbium (III) ion. Radiolabeling using NETA at 40 • C resulted in quantitative yields in all investigated ligand concentrations (labeling efficiency >97%). Both DOTA and DOTA-GA required higher concentrations to reach suffucient radiolabeling efficiency.

In vitro Stability of Radiolabeled Ligands
Metal-ligand in vitro stability was determined in phosphate buffered saline (PBS, pH 7.4) at 37 • C and analyzed over 24 h (Figure 3). The amount of 161 Tb bound to the ligand was referenced to the initial radiochemical purity. For DTPA, DOTA and NETA ligand systems, >95% of the metal was still chelated to the ligand after 24 h. The DOTA-GA ligand was observed to have retained only 92.1 ± 6.8% of the initial radiochemical purity over the same period.

Synthesis, Radiolabeling and in vitro
Stability of Human Serum Albumin Conjugates DTPA, DOTA, DOTA-GA or NETA was reacted with HSA in a 5:1 molar excess. HSA-constructs were purified using a size exclusion cartridge and analyzed using HPLC-SEC. Unconjugated human serum albumin was found to be retained in the size exclusion column for 19 min and HSA-chelator constructs eluted at the same retention time (Supplementary Figures 1-6). Constructs were analyzed using mass spectrometry to determine the number of ligands attached to the HSA protein. An increase of 1,000-1,300 Da was observed for each conjugate, indicating an average number of two chelators per albumin molecule. The mass spectra data is summarized in Table 1.
HSA-conjugates and unconjugated HSA were radiolabeled at a concentration of 10 µM with [ 161 Tb]TbCl 3 at 40 • C to ensure maximum radiochemical yield. HSA (not conjugated to any chelator) only coordinated 6.0 ± 1.2% of the [ 161 Tb]TbCl 3 in the reaction mixture. The investigated conjugates were all labeled with quantitative yields (>98%) (Supplementary Table 1) as determined with iTLC and radio-SEC-HPLC (Supplementary Figures 3-6). The radiolabeled constructs (without further purification) were added to human serum and incubated at 37 • C for 24 h.  (Figure 4). Radio-SEC-HPLC chromatograms of 161 Tb-labeled HSA conjugates are provided in the supporting information (Supplementary Figures 7-10).
To assess the susceptibility of trans-chelation, the labeled HSA-conjugates were incubated with 1,000-fold excess of EDTA (Supplementary Figure 11). As observed in the previous study, 161 Tb leached from the DTPA-HSA ligand system, with only 64.4 ± 0.9% of the initial 161 Tb remained bound to HSA after 24 h at 37 • C. For the other conjugates, >90% of the initial fraction of the radiometal remained bound to HSA after 24 h.  (Figures 6A-D, Table 2). In contrast, in the mice injected with [ 161 Tb]Tb-DTPA-HSA, increasing bone uptake was observed (SUV: 0.8 ± 0.3 and 1.1 ± 0.3 at 24 h and 7 days p.i., respectively) in function of time ( Table 2), suggesting in vivo dissociation and absorption of free 161 Tb in bone. The blood half-life of the HSA constructs was significantly longer than for [ 161 Tb]TbCl 3 (8-15 h, Supplementary Figures 17-21). Standardized uptake value graphs are provided in Figure 6, with % injected activity diagrams provided in Supplementary Figures 12-16

DISCUSSION
In this study, we aimed to develop techniques that allow radiolabeling of heat sensitive biomolecules with 161 Tb. A series of ligands were preselected based on their lanthanide chelating capacity reported in literature (20,25,26). For each ligand, we evaluated the effect of ligand concentration and temperature on radiolabeling yields. Finally, stability of the different ligand complexes was evaluated in vitro and in vivo.
Coordination of terbium is pH sensitive, as too low pH blocks carboxyl coordination, which is the main coordinating moiety of the ligands of interest (Supplementary Table 3). In aqueous terbium solutions, hydrolysis (formation of Tb(H 2 O) x (OH) y species) will occur at increased pH (pH ∼ 6-7.6) (3,10,12,(27)(28)(29). Hydrolysis will dramatically reduce or prevent the overall formation of Tb-ligand complex, which might be mitigated by increasing the temperature during radiolabeling. We selected low pH conditions [sodium acetate buffer (0.1 M, pH 4.7)] to prevent hydrolysis and enable low-temperature chelation of Tb. When radiolabeling the DTPA ligand, quantitative radiochemical yields (>98%) were observed for all conditions investigated. Incubating  the reaction mixture at 25 • C was enough to obtain quantitative yields, even at low ligand concentrations (Figure 2A). This can be attributed to the flexible nature of the linear DTPA framework which makes chelating the terbium (III) ion easier (1). At 25 • C, radiochemical yields of DOTA and DOTA-GA were lower, with a maximum radiochemical yield of 91% (10 µM). This could be expected in view of the more rigid tetraaza ring of the latter two ligand structures. Increasing the temperature to 40 • C yielded no change in the maximum yields obtained for higher concentrations of DOTA and DOTA-GA but allowed for better radiochemical yields in the low concentrations tested ( Figure 2B). Finally, for NETA, a mean radiochemical yield of ∼60% was observed at 25 • C but quantitative yields (>95%), comparable to DTPA, were obtained at 40 • C. The hybrid nature of the NETA framework could explain the radiochemical yields similar to DOTA and DOTA-GA at 25 • C. Slightly increasing the temperature however seems to provide enough energy to allow terbium(III) to be incorporated more efficiently into the chelator binding pocket. The stability of the 161 Tb-ligand bond was evaluated in a phosphate buffered saline solution (pH 7.4) at 37 • C during a time period of 24 h using instant thin-layer chromatography (Figure 3). At the end of the study, >95% (relative to the initial radiochemical purity) of the metal remained intact for complexes with DTPA, DOTA and NETA. The complex with DOTA-GA was found to be the least stable, with 92.1 ± 6.8% of the initially chelated metal intact after 24 h.
After optimizing the radiolabeling conditions, we used these optimized conditions to radiolabel HSA conjugates, as a proof of concept. HSA is a heat sensitive molecule and is the most abundant protein in blood essential for the transport of many proteins throughout the body (30,31). It has a prolonged serum half-life (30), which also makes it advantageous for determining long-term in vivo stability of radiolabeled conjugates. Additionally, since HSA circulates in the blood and shows minimal physiological accumulation in tissue, it is the perfect tool to evaluate dissociation and potential accumulation of the free radiometal to other tissues. Bifunctional ligands were conjugated to HSA non-regioselectively, using lysine coupling. The ligands were reacted with HSA to afford conjugates L-HSA (where L = DTPA, DOTA, DOTA-GA or NETA), and analyzed by UV-HPLC and high-resolution mass spectrometry. Only a single peak (Rt = 19 min) was recorded in the UV channel (L-HSA), and their retention time is identical to that of underivatized HSA (Supplementary Figures 1, 3-6). No aggregation or degradation products were observed via SEC-HPLC. Furthermore, high resolution mass spectrometry was used to estimate the number of ligands conjugated to HSA for every conjugate. Unconjugated HSA was used as a reference for calculating the number of ligand molecules that are conjugated (66477-66485 Da) to HSA. The molecular mass of all the conjugates increased by 1,000-1,300 Da relative to HSA, which  suggests that the conjugates have an average of two ligands per HSA moiety ( Table 1).
Using the optimized labeling conditions (60 min, 40 • C), HSA constructs L-HSA were radiolabeled with 161 Tb and the labeling reaction mixture was analyzed by iTLC and radio-SEC. In addition, non-derivatized HSA was incubated with [ 161 Tb]TbCl 3 to determine if there is any non-chelator related binding of terbium to the protein. The 161 Tb-labeled conjugates  Figures 1-6).
Upon incubation in human serum, a radiochemical purity above 95% was maintained for the radiolabeled HSA constructs [ 161 Tb]Tb-DOTA-HSA, [ 161 Tb]Tb-DOTA-GA-HSA and [ 161 Tb]Tb-NETA-HSA over a 24-h study period. [ 161 Tb]Tb-DTPA-HSA had a noticeable decrease in radiochemical purity after 24 h from 98.6 ± 0.5% to 88.1 ± 1.3%. This is commonly observed with ligands bearing the DTPA chelating framework, as it is often labeled as an "easy-in-easy-out" ligand for metals (18). In a competition study with EDTA ( vitro results indicate that DTPA is a poor choice for chelation of terbium. As described before, HSA can be used as an effective model protein to evaluate the in vivo stability of radiolabeled complexes (30). First, a biodistribution was performed with [ 161 Tb]TbCl 3 to determine its in vivo fate. Free [ 161 Tb]TbCl 3 was observed to clear from the blood within the first 24 h (Figure 5; Supplementary Tables 4-6) and high uptake and retention were observed in liver and bone, with the highest values observed at 4 h p.i (SUV liver = 4.1 ± 0.4 and SUV bone 5.0 ± 0.5). At day 7, still high retention of radioactivity in bone and liver was observed (SUV liver = 1.4 ± 0.8, SUV bone = 3.5 ± 1.0), resulting in a strongly increasing bone-to-blood and bone-tomuscle ratio over the 7-day period (Supplementary Table 2). The high accumulation of radioactivity in bone, allowed us to identify this tissue as an indicator for leaching of the radionuclide from the radiopharmaceutical in vivo. The biodistribution of the 161 Tb-labeled HSA constructs showed the expected accumulation of activity in organs with high blood content (heart, lungs, spleen, etc.). As observed with the free [ 161 Tb]TbCl 3 , increased bone and liver uptake was observed over the  (Figures 6B-D). This result, together with the in vitro stability data in human serum and EDTA competition study, strongly suggests DTPA has fast radiolabeling kinetics but does not adequately retain the radiometal after chelation. In the in vitro test with human serum, 10% of the radioactivity of [ 161 Tb]Tb-DTPA-HSA was dissociated after 24 h. After 24 h in vivo studies showed 10% of the injected activity in the bone (Supplementary Figure 12), showing a good concordance between in vitro and in vivo results. No increase in retention of liver and bone activity was observed over 7 days after injection of radiolabeled constructs with DOTA-HSA, DOTA-GA-HSA and NETA-HSA (Figures 6B-D), suggesting high in vivo stability of 161 Tb complexes with ligands DOTA, DOTA-GA and NETA as compared to DTPA. This is an important result for further studies with radioactive terbium isotopes as the CHX-A"-DTPA framework (DTPA, Figure 1) is often seen and used as a generic chelator for different radiometals (32). DOTA, DOTA-GA and NETA have more rigid frameworks, which can explain the more stable chelation of metals in vivo.
As expected, radiolabeled HSA-constructs were retained longer in blood compared to  Figures 18-21). This minor variation in blood biological half-life of the different conjugates might be attributed to the non-regioselective coupling of ligands to HSA; potentially reacting with regions essential to biological circulating proteins (neonatal Fc receptor, etc.). Therefore, in future experiments, it is essential to make use of more site-specific targeting approaches (his-tag coupling, sortase A, etc.) (33)(34)(35) to avoid interfering with the binding affinity of the biomolecule.

CONCLUSION
This study is the first report on labeling with 161 Tb in mild conditions (25 • C and 40 • C in aqueous buffer). As a proof of concept, we successfully radiolabeled the heat-sensitive biomolecule HSA with 161 Tb, with high radiochemical yields. Several bifunctional ligands were evaluated for their radiolabeling properties, as well as their in vivo and in vitro stability. Of these ligands, radiolabeling with DTPA was highly efficient, even at room temperature. However, the DTPA-HSA construct showed the lowest stability, both in vitro and in vivo, leading to significant off-target bone uptake and retention. In contrast, complexes with a more rigid backbone (DOTA, DOTA-GA and NETA) required slightly higher radiolabeling temperatures but were found to be very stable in vitro and in vivo. These ligands have potential to be used with other vector molecules for diagnostic and therapeutic applications of the terbium radioisotope family. Research is currently ongoing to conjugate these ligands to other heat-sensitive vector molecules to allow delivery of 161 Tb or other terbium radioisotopes to the biological target of interest.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

ETHICS STATEMENT
The animal study was reviewed and approved by Ethical Committee for Animal Experimentation, KU Leuven.
Research for their contributions. Bernard Ponsard (BR2, SCK CEN) is thanked for the irradiations of the 160 Gd target and Frank Van der Linden for organization of the nuclear transports from SCK CEN to KU Leuven. SCK CEN Academy is gratefully acknowledged.