Synthesis and Characterization of Folic Acid-Conjugated Terbium Complexes as Luminescent Probes for Targeting Folate Receptor-Expressing Cells

Several conjugates between folic acid and a series of kinetically stable lanthanide complexes have been synthesized, using amide coupling and azide–alkyne cycloaddition methodologies to link the metal-binding domain to folate through a variety of spacer groups. While all these complexes exhibit affinity for the folate receptor, it is clear that the point of attachment to folate is essential, with linkage through the γ-carboxylic acid giving rise to significantly enhanced receptor affinity. All the conjugates studied show affinities consistent with displacing biological circulating folate derivatives, 5-methyltetrahydrofolate, from folate receptors. All the complexes exhibit luminescence with a short-lived component arising from ligand fluorescence overlaid on a much longer lived terbium-centered component. These can be separated using time-gating methods. From the results obtained, the most promising approach to achieve sensitized luminescence in these systems requires incorporating a sensitizing chromophore close to the lanthanide.


INTRODUCTION
Chronic inflammation is prevalent in society with autoimmune diseases such as rheumatoid arthritis (RA), asthma, and Crohn's disease becoming increasingly common.Still, it can also appear as part of other diseases such as COVID-19 infection, and host against transplant rejection.Because of this, there is a great need to develop imaging agents that help selectively visualize inflammatory lesions in vivo to support early diagnosis and development of a personalized treatment approach.Macrophages play a crucial orchestrating role in the development and maintenance of chronic inflammation; therefore, these cells are promising targets for imaging inflammation. 1,2−6 While macrophages utilize FRβ during activation, rapidly dividing cancer cells require more folate for DNA synthesis.−15 To this end, using FRα and FRβ in imaging improves the management and treatment of a spectrum of conditions characterized by chronic inflammation and uncontrolled cellular proliferation.
Human folate receptors have a high binding affinity to folic acid and antifolates, 4 and as such there has been great interest in the development of probes based on their structure. 16,17In fact, the antifolate drug methotrexate is an anchor drug in several chronic rheumatic diseases including RA. 18 Folic acid consists of three distinguishable fragments: pterin, 4-aminobenzoic acid, and glutamic acid, the latter of which has two potential sites for conjugation (α-and γcarboxylates).The crystal structures of FRα 19 and FRβ 20 complexed with folic acid have been reported, helping to elucidate the important structural features of the folate-derivates within the FR binding pocket. 19,20The pteroic acid binds furthest within the FR binding cavity, and the terminal glutamic acid resides closest to the entrance of the receptor cavity with the glutamic acid carboxylates forming hydrogen bonds and charge−charge interactions. 20These structures support the determine the influence on probe emission and cellular FR binding properties.

Synthesis of Complexes.
For bioconjugation, to obtain the separate regioisomers of the folate-PEG 6 ligand, S2 α/γ the molecule was prepared by stepwise solid-phase synthesis adapted from literature and purified by RP-HPLC in low yields (γ = 19%, α = 5%). 22,44Folate-PEG 6 S2 α/γ was then coupled to DOTA-NHS ester, followed by metal complexation with terbium trifluoromethane sulfonate in water at pH 4 (Scheme S1).TbL 1γ and TbL 1α were obtained separately as yellow powders after purification by RP-HPLC and lypophilization. 22−43 The previously unreported 4′-ethynyl-2-acetophenone-DO3A, 4 was prepared as outlined in Scheme 1.Briefly, the 4′-ethynyl-2-bromoacetophenone, 1 was prepared by reacting 1,4-diethynylbenzene and 1,3-dibromo-5,5-dimethylhydantoin in the presence of N,N′-ethylenethiourea as catalyst in aqueous acetone at 45 °C and then purified by column chromatography to afford the product, 1 as a yellow oil.4′-Ethynyl-2-bromoacetophenone, 1 was then reacted with tertbutyl protected DO3A, 2 to yield 3. Subsequent deprotection with 1:1 trifluoroacetic acid and dichloromethane yielded 4. Metal complexation was carried out with terbium trifluoromethane sulfonate to produce Tb4.Azido folate, 6, was prepared using a modified literature procedure by Ke et al. as outlined in Scheme 2. 45 Folic acid (FA) was dissolved in dimethylsulfoxide and reacted with the azide-PEG n -amine, 5, in the presence of N,N′-dicyclohexylcarbodiimide (DCC) and pyridine to give 5, as a mixture of the αand γisomers.The mixture was found to be difficult to separate by RP-HPLC, so was used without further separation and clicked with the alkyne-bearing terbium conjugates, Tb7 and Tb4.
The final step in the synthesis of TbL 2 -TbL 4 was carried out by reacting the respective terminal alkynes, Tb7 and Tb4, with the azido folate, 6 in a copper-catalyzed azide−alkyne cycloaddition using copper(II) sulfate and sodium ascorbate as a reducing agent to produce the copper(I) catalyst in situ.All complexes were purified by RP-HPLC in acidic conditions to yield the conjugates as yellow powders and as the 1,4disubstituted triazoles.Complexes TbL 2 and TbL 4 were obtained as mixtures of the αand γ-regioisomers as these were difficult to separate by HPLC.TbL 3 , however, underwent successive rounds of HPLC purification to initially separate a purified mixture of TbL 3 and then separate the αand γregioisomers to yield TbL 3 major and TbL 3 minor .Due to the paramagnetic nature of the terbium center, regioisomers could not be assigned as the αor γ-regioisomer.
Synthesis of TbL 5 was carried out in the absence of a copper catalyst and the reaction was promoted by additional ring strain imposed by the two conjugated benzene rings on the Tb-DOTA-DBCO.In addition to the αand γ-regioisomers of 6, the SPAAC affords both the 1,3-and 1,4-triazole products resulting in a mixture of 4 isomers.Due to the high similarity of the complexes, these could not be separated by RP-HPLC and remained as a mixture of the four isomers.
2.2.Photophysical Characterization.2.2.1.Electronic Spectroscopy.The absorption spectra of the terbium folate complexes are shown in Figure 1.The absorption spectrum of folic acid is characterized by two absorption bands in the region of 250−300 and 300−400 nm, which correspond to the π → π* transition for the pterin and the n → π* transition for the p-amino benzoyl acid moieties respectively. 46All the terbium complexes exhibit these two distinct absorption characteristics of the FA moiety.Differences are observed for TbL , 4 which is red-shifted and broadened due to the overlap with the acetophenone moiety; similar absorbance spectra of acetophenone-folate complexes have been reported by Quici et al. 16 All complexes, except for TbL 4 , were excited through the pterin absorption band at ca. 280 nm using weak indirect excitation via Forster energy transfer to the terbium center.For TbL 4 , indirect Dexter-mediated energy transfer through the acetophenone chromophore was achieved following excitation at 301 nm (Figure 2C).The complexes all display strong fluorescence from the folate moiety (Figure 2), with the dominant pterin emission peak at ∼440 nm, while the characteristic terbium emission peaks are just visible at 545, 588, and 621 nm corresponding to the 5 D 4 → 7 F J (J = 5−3) respectively.This suggests that the variations in the structure with the αand γfolate regioisomers makes minimal difference in the absorption of the terbium emission spectra as shown for TbL 1 and TbL 3 .Similarly, the length of PEG chain appears to make little difference to the fluorescence emission for TbL 2 (n = 2) and TbL 3 (n = 5), as shown in Figure 2b.Furthermore, in the case of TbL 4 , it is clear that absorption is occurring through both the folate and the carboxyphenacyl chromophore; thus, while the lanthanide emission is stronger in this case, there is still a substantial background of folate fluorescence.
Therefore, time-gating techniques can be employed to remove the short-lived fluorescence from the folate and the cell.This is analogous to phosphorescence emission, where the phosphorescence emission is recorded over a longer period of time to capture the longer-lived terbium emission with an initial delay to remove the biological autofluorescence and organic folate-centered emission.Translating this over would utilize techniques such as Phosphorescence Lifetime Imaging Microscopy (PLIM).The phosphorescence emission spectra for TbL 1 -TbL 5 are shown in Figure 3, where it can be seen that the folate moiety is able to act as a sensitizer to indirectly excite the terbium center.In such a system, a variety of processes are possible: long-lived luminescence from the lanthanide and short-lived fluorescence from the ligand offer competitive routes to radiative emission.Interestingly, the length of the PEG chain spacer is important for the efficiency of this energy transfer, with the shortest PEG chain for TbL 2 (n = 2) resulting in the greatest emission when recorded under the same conditions.Subtle differences are also observed between the TbL 1 (PEG n = 6) with the weakest emission and the intermediate emission from TbL 3 , TbL 4 and TbL 5 (PEG n = 5).For TbL 4 , when excited at 300 nm through the acetophenone moiety, weaker emission is observed, therefore there may be other pathways that quench the excited state.The folate sensitization occurs through space via Forster energy transfer and is affected by the PEG spacer length.

Lifetimes and q
Values.The emission lifetimes of the complexes TbL 1 -TbL 5 were measured to determine the number of water molecules, q, coordinated to the metal center and are summarized in Table 1.This is calculated from the modified Horrocks equation that accounts for the difference in emission lifetimes when quenched by O−H or O−D oscillators. 47This is useful to inform future time-gating measurements for cell imaging.
The complexes all display weak molar extinction coefficients, which primarily arises from the folate moiety.Low quantum yields (Φ) were recorded for all complexes, with TbL 2 displaying the highest Φ.For these conjugates to function as fluorescent probes for the standard microscopy techniques,

Journal of Medicinal Chemistry
these are too weakly emissive to compete with the cellular autofluorescence.This is highlighted when comparing to the FDA-approved folate-heptamethine cyanine dye, OTL38 which has a Φ = 15.1% and ε = 272 000 M −1 cm −1 with λ ex = 776 nm, λ em = 793 nm. 10 This is able to excite at a lower energy wavelength, which is less damaging for biological imaging as well as emitting in the NIR-window where there is a reduced fluorescence from the cell.

Folate Receptor Fluorescent Folic Acid Binding Competition Assay.
To assess the binding affinity of the terbium-folate conjugates, a flow cytometry binding competition assay was carried out with folate-FITC and increasing    It is important to note that FR relative binding affinities for Tb-folate conjugates should be considered in comparison of 5methyltetrahydrofolate as the major circulating form of folate in plasma.Given that FR affinities for 5-methyltetrahydrofolate are approximately 10-fold lower than folic acid, 4,49 this implies that Tb-folate conjugates can displace natural folates from the receptor with proficient binding affinity.
Beyond binding affinities, we also examined whether the Tbfolate conjugates harbored any toxic effects.Following incubations of KB cells for 24 h with 500 nM of the Tb conjugates, no effects on cell viability were observed (results not shown).

In Silico Assessment of Membrane Permeability.
To predict the behavior of the Tb-folate compounds regarding the permeability through the blood-brain barrier and the cell membrane, in silico methods were utilized.The main parameters of focus were the total free energy penalty for the ligand to change state and enter the membrane (DG in ) and the predicted permeability through model cell lines such as Ralch Russ canine kidney cell (RRCK), Madin-Darby canine kidney cells (MDCK) as a model of the blood-brain barrier, and Caco-2 cells as a model of the gut-blood barrier. 50,51lycerol was used as a positive control for permeabilization.All Tb-folate compounds showed poor permeability profiles, both the blood-brain and the gut-blood barriers, indicating that these compounds cannot be orally administered or used for imaging brain inflammation or cancer (Table 2).
Finally, we focused on the Tb-folate compounds concerning Lipinski's and Jorgensen's rules to model their bioavailability.Lipinski's rule of five assesses this from the number of  n/c -No calculation bDG in is the total free energy penalty for the ligand to change state and enter the membrane.c Log P m : Logarithm of the RRCK permeability in cm/s d Predicted apparent Caco-2 cell permeability in nm/sec.Caco-2 cells are a model for the gut-blood barrier.QikProp predictions are for nonactive transport (<25 poor, > 500 great).e Predicted apparent MDCK cell permeability in nm/sec.MDCK cells are considered to be a good mimic for the blood-brain barrier.QikProp predictions are for nonactive transport (<25 poor, > 500 great).
hydrogen bond donors and acceptors, molecular weight, and partition coefficient, 52 and Jorgensen's rule of three is to model the permeability and solubility of the compounds.All compounds are characterized by low bioavailability (SI section 6).
2.3.Summary.This work has involved the synthesis of a range of folate-terbium bioconjugates, with the γ-isomer of the folic acid showing favorable FR binding over its α-analogue.The length of the PEG spacer was shown to alter the through space Forster energy transfer with shorter chains showing greater phosphorescence.These novel fluorescent complexes are a good initial set of probes which could be used for PLIM.The probes can be further improved by including a sensitizing chromophore to enhance fluorescent emission for use in more readily available microscopic techniques.Alternatively using different lanthanides in the same constructs, these could be extended for use as an MRI contrast agent (by using Gd) or a theragnostic for PET imaging and treatment (using radioactive 152 Tb and 149 Tb or 177 Lu).

CONCLUSIONS
This work has shown that a wide range of conjugation methods can be used to access folate bioconjugates with lanthanide complexes, and that these complexes retain appropriate affinities for folate receptors (particularly when conjugated via the folate γ-carboxylate).Due to the distance of the UVactive folate compound from the lanthanide center and the inefficient spectral overlap between the folate emission spectrum and the terbium absorption spectrum, the compounds suffered from the characteristic weak molar extinction coefficients associated with direct Ln 3+ excitation and therefore weak emission.While this can be resolved using time-gated techniques such as PLIM, it is clear that the next generation of luminescent probes must incorporate a sensitizing chromophore which absorbs light outside the range of folate absorption.Until that can be achieved, it will be difficult to achieve the kind of brightness associated with organic fluorophores in conventional confocal microscopy.
The ease with which lanthanide complexes can be incorporated into these systems also suggests another possibility.Use of 161 Tb or 177 Lu could be an effective route to PET contrast imaging of folate receptors for use in cancer and inflammation imaging.

EXPERIMENTAL PROCEDURES
4.1.Chemistry.4.1.1.General Procedures.Commercially available solvents and reagents were used without further purification unless otherwise stated.Reagents were laboratory grade and obtained from standard commercial sources including Sigma-Aldrich, Fisher Scientific, Alfa Aesar, Fluorochem, CheMatech, ChemicalPoint, and Insight Biotechnology.Anhydrous solvents used were obtained by passing through a MBraun MPSP-800 column (degassed with N 2 ) and then used immediately.Deionized water was obtained using an Elix Essential water purification system.Propargyl bromide was used as 80 wt % solution in toluene (Sigma-Aldrich).Neutral alumina for column chromatography was purchased from Alfa Aesar (aluminum oxide activated, neutral Brockmann grade I, pore size 58 Å, 60 mesh).Silica gel for column chromatography was purchased from Merck (Geduran, pore size 60 Å, 230−400 mesh, 40−63 μm).Aluminum oxide neutral TLC plates and silica TLC plates were TLC gel 60 purchased from Merck.Brine is a saturated sodium chloride solution.All terbium complexes are confirmed to be over >95% pure by HPLC analysis, except for TbL 2 (85%).
4.1.1.1.NMR Spectroscopy.NMR spectra were recorded at 298 K unless otherwise stated.600 MHz 1 H NMR and 151 MHz 13 C NMR were obtained using a Bruker NEO 600 with broadband helium cryoprobe equipped with a 14 T magnet.500 MHz 1 H NMR, 126 MHz 13 C NMR were recorded on a Bruker Avance III NMR equipped with a 11.75 T magnet.400 MHz 1 H NMR and 101 MHz 13 C NMR were obtained using a Bruker Avance III nanobay 400 NMR equipped with a 9.4 T magnet.
4.1.1.2.Mass Spectrometry.Low-resolution mass spectra were obtained using either an Agilent Technology 1260 Infinity or a Waters LCT Premier XE spectrometer.LC-MS were measured by a Waters LCT Premier benchtop orthogonal acceleration time-of-flight LC-MS system.High-resolution accurate mass spectra were performed by the staff of the Chemistry Research Laboratory, University of Oxford using Bruker μTOF or Waters Micromass GCT spectrometers and recorded to 4 decimal places.
4.1.1.4.Luminescence Spectroscopy.HORIBA Jobin Yvon FluoroLog3 fluorimeter (Hamamatsu R928 detector and a doublegrating emission monochromator) was used to acquire the luminescence spectra operated under FluorEssence software.The standard conditions for acquiring emission and excitation spectra are room temperature and steady-stated mode unless otherwise stated.HORIBA Jobin Yvon FluoroLog3 fluorimeter system equipped with a Xenon flash lamp was used to acquire emission lifetimes.Luminescence lifetimes were obtained by tail fit for Tb(III) complexes using Origin software.
4.1.3.1.Tb4.Following general procedure B in MeOH, the title compound was prepared from 4 (0.15 g, 0.31 mmol, 1.00 equiv) and Tb(OTf) 3 (0.20 g, 0.33 mmol, 1.05 equiv) and purified by column chromatography on neutral alumina (MeCN/H 2 O, 8:2 to 7:3) to yield the product as a pale beige-pink powder (0.108 g, 0.17 mmol, 79%).using pteroic acid (0.5 g, 1.60 mmol, 1.00 equiv) to yield S1 as a brown solid (0.48g, 1.17 mmol, 73%) and used directly without further purification.Prepared with minor adaptation from the procedure described by Chen et al. 22 and Kularatne et al. 44 Refer to Scheme S1 for an overview of the synthetic route.The procedure was carried out in parallel batches of four.1,2diaminoethane trityl resin (1.2−1.7 mmol/g, 70 mg, 0.1 mmol) was swollen with DCM (3 mL) for 2 h and then drained, followed by DMF (3 mL) for 2 h and then drained through a Biotage phase separator column.After swelling the resin, a solution of Fmoc-PEG 6 − OH (86 mg, 0.15 mmol, 1.50 equiv), HATU (57 mg, 0.15 mmol, 1.50 equiv), and DIPEA (35 μL, 0.2 mmol, 2.00 equiv) in DMF (4 mL) was stirred for 15 min before addition to the resin.N 2 was bubbled through the suspension overnight, and the resin was washed with DMF (3 × 15 mL) followed by i-PrOH (3 × 15 mL).The resin was then swollen in DMF (5 mL) for at least 30 min before the next step.The above resin was incubated with 20% piperidine in DMF (5 mL) for 30 min to remove the Fmoc-protection, then followed by washings with DMF (3 × 15 mL) and i-PrOH (3 × 15 mL).The resin was then swollen in DMF (5 mL) for at least 30 min before the next step.The above sequence was repeated for two more coupling steps for conjugation of γ-isomer: Fmoc-Glu-OtBu (CAS: 84793-07-7) or α-isomer: Fmoc-Glu(OtBu)−OH (CAS: 71989-18-9) (64 mg, 0.15 mmol, 1.50 equiv) and N 10 -TFA-Ptc−OH (61 mg, 0.15 mmol, 1.50 equiv).The product was cleaved from the resin using a TFA: H 2 O: TIPS mix (95:2.5:2.5)(5 mL) for 2−3 h at room temperature unstirred.The filtrate was collected, and the resin was washed with DCM (3 × 15 mL) and i-PrOH (3 × 10 mL) and the filtrate and washes were concentrated under vacuum.The product was washed with copious amounts of DCM that was removed under reduced pressure, then washed with MeOH and removed.The concentrated product was dissolved in minimal MeOH (1−3 mL) and precipitated in cold diethyl ether (∼400 mL) and centrifuged (4000 rpm, 4 °C, 10 min) and a yellow solid isolated that was characterized by MS (ES-MS (MeOH) m/z 915.3 [M + H] + ).The yellow precipitate was stirred with saturated Na 2 CO 3 solution at room temperature to cleave the trifluoroacetyl protecting group for ∼1 h followed by MS.The mixture was neutralized to pH 7 with 2 M HCl (aq.) and the solvent was removed under reduced pressure.The residue was dialyzed for at least 24 h.The precipitate was purified by HPLC (S3.2 HPLC conditions method 1).). Prepared with minor adaptation from the procedure described by Chen et al. 22 and Kularatne et al. 44 FA-PEG 6 -EDA-NH 2 , S2 (0.150 g, 0.183 mmol 1.00 equiv) was added to a flame-dried flask and dissolved in DMSO (1.8 mL, 0.1 M final concentration).DIPEA (112 μL, 0.641 mmol, 3.50−3.80equiv) and DOTA-NHS ester (0.153 g, 0.201 mmol, 1.10 equiv) were added and the reaction was stirred at 20 °C overnight under an Ar atmosphere.The reaction was monitored by LC-MS and upon completion (∼24 h) the unreacted DOTA-NHS ester was cleaved by the addition of water (1 mL) and stirred for 30 min.The reaction mixture was concentrated under reduced pressure and the crude product precipitated by dropwise addition into vigorously stirred cold (0 °C) acetone (∼600 mL).The suspension was centrifuged (4000 rpm, 4 °C, 10 min), and the yellow solid isolated.The solid was dissolved in water and freeze-dried to remove trace DMSO and yield the crude product as a yellow powder.The precipitate was purified by HPLC (S3.1 HPLC conditions method 1).
4. 1.3.16.17-Azido-3,6,9,12,15-pentaoxaheptadecan-1-amine/ Azide-PEG 5 -amine (5b).Prepared by the procedure described by Goswami et al. 56 and Kohata et al. 57 Refer to Scheme S2 for overview of the synthesis.To PEG 5 -diazide, S5b (2.30 g, 6.92 mmol, 1.00 equiv), 5% aqueous HCl (25 mL) was added and vigorously stirred at room temperature.To this mixture, a solution of triphenylphosphine (1.63 g, 6.23 mmol, 0.90 equiv) in Et 2 O (18 mL) was added dropwise over ∼3 h.The mixture was stirred for a further 24 h at room temperature.The reaction mixture was washed with ethyl acetate (3 × 100 mL) to remove unreacted starting materials and triphenylphosphine oxide that was formed during the reaction.The aqueous layer was collected, and 20 M potassium hydroxide was added slowly until the pH of the solution was basic (∼pH 12).The product was extracted by washing the aqueous layer with DCM (3 × 100 mL).Refer to Scheme S2 for overview of the synthesis.Folic acid (1.00 g, 2.27 mmol, 1.00 equiv) and dicyclohexylcarbodiimide (DCC) (0.935 g, 4.53 mmol, 2.00 equiv) were dissolved in dry DMSO (50 mL) under an inert atmosphere.Pyridine (10 mL) was added to the reaction mixture and stirred for 1 h before addition of the azide-PEG 2amine (0.40 g, 2.27 mmol, 1.00 equiv).The reaction mixture was stirred in the dark for a further 60 h.The resulting mixture was then filtered, under an inert atmosphere using a filter cannula, into a flask of vigorously stirring diethyl ether (200 mL) to produce a yellow precipitate.The suspension was continuously agitated with slow stirring and most of the ether was decanted.The yellow solid was washed twice, to remove remaining DMSO, with acetone (60 mL) and diethyl ether (100 mL); stirring was maintained and solvent was decanted each time.Any remaining solvent was then removed in vacuo to obtain the product as a fine yellow powder.
1 H NMR of a mixture of regioisomers (400 MHz, D 2 O with NH 4 OH, pH 9) δ 8.61 (s, 1H, H1), 7.67 (dd, 3 J H−H = 8.9, 7.6 Hz, 2H, H4), 6.81 (overlapping d/dd, 3  Prepared with minor adaptation from the procedure described by Jauregui et al. 41 Refer to Scheme S3 for overview.Triester, 2 (4.00 g, 6.72 mmol, 1.00 equiv) was added to dry MeCN (60 mL) with potassium carbonate (2.32g, 16.8 mmol, 2.50 equiv) and stirred for 20 min.The flask was cooled on ice, propargyl bromide (80 wt % in toluene) (0.75 mL, 5.83 mmol, 1.00 equiv) was added dropwise and the reaction was left stirring overnight at 15−25 °C under N 2 .Inorganic solids were removed by filtration and the solvent was removed under reduced pressure.The resulting oil was dissolved in toluene (75 mL) and washed with water (60 mL × 2).The organic layer was dried over MgSO 4 , filtered through Celite pad and the solvent removed to give S5 as a clear golden oil (2.61 g, 4.72 mmol, 70%).(7).Following general procedure A, compound 7 was prepared from S5 (0.5 g).The brown solid was further purified and was dissolved in minimal MeOH (∼8 mL) and precipitated upon addition of DCM (∼2 mL), centrifuged and the supernatant discarded to afford the product, 7 as the colorless trifluoroacetate salt (0.178 g, 0.462 mmol, 64%).4.1.4.5.Tb7.Following general procedure B in MeOH, the Tb7 was prepared from 7 (0.2 g, 0.4 mmol, 1.00 equiv) and Tb(OTf) 3 (0.25 g, 0.42 mmol, 1.05 equiv).The solvent was removed under reduced pressure and the solid dissolved in 2 mL water.The pH was adjusted to pH 14 to precipitate out excess lanthanide as lanthanide hydroxide.The solution was centrifuged.The supernatant was kept, and the pH was adjusted to pH 7, before removing the solvent.The complexes were dialyzed with Float-a-lyzer with MWCO 500 Da.The solution was lyophilized to yield Tb7 as a colorless powder (0.24 g, 0.45 mmol, 86%).).Tb.pDO3A, Tb7 (0.150 g, 0.28 mmol, 1.00 equiv), FA-PEG 2 -N 3 (6a) (0.185 g, 0.31 mmol, 1.10 equiv) and (CF 3 SO 3 Cu) 2 •C 6 H 6 (14.2 mg, 0.028 mmol, 0.10 equiv) were combined with MeOH (15 mL) and DMSO (5 mL).The solution was heated at reflux at 80 °C while stirring under Ar for 24 h.Methanol was removed in vacuo and the residue was dissolved in water before lyophilization to aid removal of the DMSO.The solid was dissolved in H 2 O (50 mL) and gently stirred with 2−3 g CupriSorb resin for 24 h to remove copper.The insoluble CupriSorb was removed by filtration and the filtrate was dried in vacuo.The compound was purified by HPLC (S3.5 HPLC conditions 5) and then dried by lyophilization to afford a yellowy-green powder.
Tb-pDO3A, Tb7 (0.031 g, 0.057 mmol, 1.00 equiv) and FA-PEG 5 -N 3 (6b) (0.050 g, 0.069 mmol, 1.20 equiv) were dissolved in H 2 O (5 mL) and degassed with three freeze−pump−thaw cycles.To this, (CuOTf) 2 •C 6 H 6 (5.0 mg, 0.01 mmol, 0.20 equiv) was added and degassed with a further freeze−pump−thaw cycle and heated at 40 °C under Ar for 3.5 days monitored by LC-MS.The reaction mixture was diluted in 50 mL H 2 O and gently stirred with 2−3 g CupriSorb resin to remove copper for 24 h.The insoluble CupriSorb was removed by filtration, and the filtrate was dried in vacuo to yield the crude as a yellow residue (106 mg, 0.083 mmol, crude 146%).
The crude products (258 mg) were combined and purified by HPLC twice.The first HPLC conditions (S3.6 HPLC conditions method 6-purification step 1) isolated the 54.6 mg compound as a mixture of regioisomers from the impurities, t R (prep) = 12.34−12.41min and t R (ana) = 2.98 min.The second HPLC conditions separated the major and minor isomers (S3.6 HPLC conditions method 6− purification step 2).
The compound was purified HPLC (S3.7 HPLC conditions method 7) isolated the compound as a mixture of regioisomers from the impurities.

Journal of Medicinal Chemistry
solvents were removed in vacuo to yield a pale yellow transparent oil.The residue was dissolved in 3.5 mL (8 H2O: 2 MeCN) and purified by preparative HPLC (S3.8 HPLC conditions method 8).UVdirected fraction collection λ = 220 nm.t R (prep) = 4.1 min.Solvent was removed in vacuo to yield the S6 as an off-white solid (66 mg, 69%).4.1.4.10.TbS6.To a solution of the ligand S6 (0.040 g, 0.06 mmol, 1.00 equiv) in 3 mL (H 2 O/MeCN, 7:3), the Tb(OTf) 3 (0.038 g. 0.063 mmol, 1.05 equiv) was added and the reaction mixture was stirred at 50 °C for 30 min.The pH was adjusted to 4/5 by dropwise addition of an aqueous 1 M NaOH solution.The reaction was left to stir at 50 °C for 48 h.The solvent was removed under reduced pressure and the residue was suspended in water.The pH of this solution was adjusted to 7 by addition of aqueous 1 M NaOH and the solvent was removed under reduced pressure giving the crude as a pale pink powder.The pale pink powder was purified by column chromatography (alumina MeCN/H 2 O, 8:2 to 7:3) to yield TbS6 as an off-white powder (75.7 mg, 0.092 mmol, 154%).R f = 0.32 (MeCN/H 2 O, 8:2).Excess mass attributed to salts; product used with salts.).Crude Tb-DOTA-DBCO, TbS6 (59 mg, 0.05 mmol, 1.06 equiv) was dissolved in H 2 O/MeCN (2 mL, 8:2).To this, FA-PEG 5 -N 3 , 6b (35 mg, 0.047 mmol, 1.00 equiv) was added and the reaction was stirred at 15−25 °C for 24 h.The reaction mixture was concentrated under reduced pressure and the residue dissolved in H 2 O (8 mL) and removed by lyophilization.The crude was purified by HPLC (S3.9 HPLC conditions method 9).(33.7 mg, 0.022 mmol, 44%) Analytical purity >97% t R (prep) = 14.1 min, t R (ana) = 1.5 min.4.2.Folate Receptor Affinities/Folate-FITC Competition.Human KB nasopharyngeal epidermoid carcinoma cells (expressing FRα) and FRβ-transfected Chinese Hamster Ovary cells were cultured in Folate-free RPMI medium (Thermo Fisher Scientific) supplemented with 20% Newborn Calf Serum, 1% antibiotics (penicillin and streptomycin) and 25 mM HEPES, in a humidified 37 °C incubator with 5% CO 2 as described by Van der Heijden et al. 4 The culture medium was refreshed every 3 days.Stock solutions (1 mM) of Tb-folate conjugates were made by dissolving in 50 mM NaHCO 3 and stored at −20 °C until use.Assessment of relative binding affinities of FR for Tb-folate compounds was performed by flow cytometry assays in competition with Folate-FITC as described previously. 59,60For KB cell experiments (25,000 cells/100 μL), Tbfolate compounds were diluted in PBS to reach the final concentration: 10, 25, 50, 75, 100, 250, 500, 1000 nM in the presence of 10 nM Folate-FITC.For CHO/FRβ cell experiments (25 000 cells/100 μL), Tb-folate compounds were diluted in PBS to reach the final concentration: 1, 2.5, 5, 7.5, 10, 25, 50, 100 nM in the presence of 1 nM Folate-FITC.As control competitor, folic acid was diluted in PBS, and added, in separate experiments, to reach the same concentration range as Tb-folate compounds.Cells treated with compounds were incubated on ice, in the dark for 10 min.Treated cells were then washed twice with ice-cold PBS by centrifugation (600 RCF for 5 min).Treated cells were kept on ice until the acquisition.Attune NxT flow cytometer (Thermo Fisher Scientific) was used for data acquisition.Flow cytometry data was analyzed in FlowJo v10.8 Software (BD Life Sciences).Statistical analysis was done in GraphPad Prism version 6.0.0 for Windows, GraphPad Software, San Diego, California USA.Data from 6−9 replicative experiments were analyzed.Example output of the competition assays is presented in SI section 7.
Considering FRα and FRβ share the same affinity for folic acid and Folate-FITC, relative affinities for Tb-folate compounds were expressed relative to folic acid. 59,60The relative affinity was calculated using Accurate prediction of membrane permeability is crucial for understanding absorption and distribution within the body.In this study, we utilized an improved computational method for estimating membrane permeability based on a physical model that incorporates multiple factors contributing to the free energy of desolvation and state changes, including neutralization and tautomerization upon membrane entry.The model accounts for the differences in dielectric properties between the low-dielectric continuum of the membrane and the high-dielectric continuum of water. 50,51irst, we looked at the properties of the DOTA fragment and then replaced terbium with calcium, as both calcium and terbium have a coordination number of 8 and are hard metal ions.Additionally, both have two valence electrons in their outermost shell.This similarity in the number of valence electrons can affect their chemical behavior.In silico prediction of the properties with Tb requires more computational power and time, therefore, we simplify the model to increase the speed of the in silico prediction (Table 2).
In the first step, QikProp, an absorption, distribution, metabolism, and excretion (ADME), prediction tool developed by Professor William L. Jorgensen was used. 61As an input, it takes the 2D structure of the compounds.It prepares the structure by assigning appropriate bond orders, atom types, and protonation states.QikProp computes a range of molecular descriptors that capture various physicochemical properties of the compound.These descriptors include molecular weight, hydrogen bond donors and acceptors, lipophilicity (LogP), polar surface area (PSA), and many others utilizing specific rules.These rules are based on desirable ranges and thresholds for specific properties, such as molecular weight Lipinski's rule of five criteria, 52 and Jorgensen's rule of three. 62(SI section 6).
To assess the permeability of the compounds and their ability to cross the blood-brain barrier we looked at values QPPMDCK1 and QPPCaco2.As a comparison, ChEBI_27470 (folic acid) and Glycerol were used (Table 2).We excluded the DOTA part from the prediction model as the algorithm used in the program is not parametrized for such large molecules.There is a limit on the number of atoms.In addition, the DOTA molecule itself is characterized by low permeability.As the "tails" are the main differences between the compounds, we focused on the behavior of the "tails".