Bivalent ligandshave been definedas “molecules having two pharmacophores covalently connectedby a linker”.¹ This category ofcompounds has been used to design new ligands for various G-proteincoupled receptors (GPCR) including the opioid,^1,2 cannabinoid,³ chemokine,⁴ or melanocortinreceptor systems among others.⁵ One ofthe interests of heterobivalent ligand (HBL) resides is the fact thatthe combination of two ligands for two different receptors may leadto improved pharmacological properties.⁶⁻⁹ The improvement of the pharmacological propertiesof bivalent ligands can arise from an increased affinity of the ligandfor tissues or cells coexpressing the receptors that the HBLs areaimed at. This could be due to the fact that the binding of a firstligand to its corresponding receptor might bring the second ligandclose to its targeted receptor, thus increasing its local concentrationand its possibility of binding to its corresponding receptor simultaneouslyor even after dissociation from the first receptor.^8,10 Inaddition, since some biomarkers may have different expression profilesdepending on the individual or within an individual (heterogeneity)or the stage of a disease (change in receptor expression level), HBLsallow to target a larger number of clinical cases with one singlemolecule.

HBLs find a particularly adequate application in nuclearimagingand therapy of tumors. This is due to the fact that tumors overexpressdifferent receptors, thus making multireceptor targeting a way ofgaining specificity in tumor targeting.¹¹ HBL-based radiotracers are expected to display increased uptakein disease tissue and improved tumor/background ratios at later timepoints.^12,13 A number of receptors such as regulatorypeptide receptors and chemokine and integrin receptors,^14,15 among others, have been identified at the surfaces of tumor cellsand could be used for multireceptor targeting.^11,16 An example for such multireceptor expression by tumor cells canbe found in prostate cancer (PCa). Prostate-specific membrane antigen(PSMA) is overexpressed in the vast majority of malignant tumors withthe PSMA expression level increasing with higher tumor stage and grade.¹⁷ The gastrin-releasing peptide receptor (GRPR)is also overexpressed by PCa cells, and much effort has been dedicatedto the development of GRPR-seeking imaging agents.^18,19 However, recent studies have shown that both receptors are differentlyoverexpressed and that GRPR is strongly upregulated in the majorityof PCa cases particularly on lower grade, whereas higher grade expressesabundantly the PSMA.^20,21

In the context of tumortargeting, where antibodies or peptidescan be used for imaging and therapy, the generation of peptide-basedHBLs is by far easier than the generation of antibody fragment-basedHBL conjugates. Furthermore, peptide conjugates benefit from fastdiffusion into the tumor as well as fast clearance from the body,and their use as dimers or multimers tends to increase theirin vivo half-life.^22,23 A downside of thisclass of compounds is the importance of the distance separating thetwo ligands. As a matter of fact, this distance is required to beoptimized to maximize the benefits of ligand heterobivalency,^2,4,12,24,25 implying the synthesis of libraries of compoundswith varied spacer length and nature. With the increased interestof medicinal chemists and radiopharmacists in HBLs, it is of primeimportance to provide easy and straightforward procedures to synthesizethem. The efficient synthesis of unlabeled HBLs might be performedby linear, sequential, or convergent synthesis. However, the use ofHBLs as imaging probes requires the addition of an imaging probe andadds another layer of complexity to the chemical synthesis process.The use of a third polyfunctional molecular entity like a probe (chelatorfor nuclear imaging or therapy, or an organic-based fluorophore forcellular microscopy or fluorescence imaging) can make classical linearpeptide synthesis cumbersome due to the use of multiple orthogonalprotective groups, tedious deprotection and coupling steps that requireintermediate purifications, the need of manipulating fully protectedpeptides, which can lead to solubility problems, or the inherent instabilityof some reporter probes (such as organic fluorophores) in classicalSPPS conditions.

Developing modular strategies to synthesizeradiolabeled HBLs isparticularly important, since these molecules often need to be optimizedto maximize the interactions of the ligands with receptors.^4,12,25,26 The use of a multifunctional molecular scaffold or template is ahighly efficient strategy to put together elaborated building blocks.Moreover, when those building blocks are delicate or difficult tosynthesize, it is highly desirable that they are incorporated ontothe scaffold via orthogonal click reactions.²⁷

Some amino acids, like lysine, are natural trifunctional platformsthat can be advantageously functionalized with orthogonally reactivebioconjugation moieties.²⁸ Similarly, trifunctionalizedbenzene rings can also be used as a starting material to provide auseful scaffold.²⁹ However, both optionsrequire multiple modifications of commercial compounds to decoratethe chosen scaffold with orthogonally reactive moieties.³⁰ Alternatively, trichloro-1,3,5-triazines havebeen noticed as a way to obtain clickable platforms from a readilyavailable commercial substrate.³¹ However,the high reactivity of triazines makes them vulnerable to hydrolysisand a large range of nucleophiles (including amines) and thereforerequires careful optimization and chemical expertise in their handling.^29,32

We have recently demonstrated that the use of disubstitutedtetrazinesbearing various chelating agents and fluorescent organic dyes wasa promising way for the effective synthesis of antibody-based multimodalimaging probes.³³⁻³⁵ To demonstrate both utility and scope of this approach,we report herein the usefulness of commercial 3,6-dichloro-1,2,4,5-tetrazine(dichloro-s-tetrazine) as a trifunctional platformfor the modular and straightforward synthesis of heterobivalent ligandsfunctionalized with an imaging probe. To reach this objective, a seriesof tetrazines derivatized with different tumor targeting vectors ofinterest in nuclear medicine were synthesized and reacted with a fluorescentorganic dye or a chelator to generate radiolabeled heterobivalentimaging and therapy agents (or precursors thereof) (Figure1). In an effort to make thismethodology even more practical, the previously published stepwiseprocedure was further improved to be carried out in one pot.

Thanks to the reactivity of the dichloro-s-tetrazinetoward nucleophilic species compared to the monosubstituted chlorotetrazine,nonsymmetrical tetrazines can be readily obtained by sequential aromaticnucleophilic substitution (S_NAr) of the chlorine atomswith an amine first and then with a thiol. The first substitutionreaction of the chlorine by an amino group leads exclusively to amonochlorotetrazine derivative. This is due to the fact that thissubstituent dramatically increases the electron density in the aromaticring yielding to a loss of reactivity of the carbon bearing the secondchlorine atom. To perform the second substitution, a stronger nucleophile,such as a thiol, is required and quickly affords a disubstituted tetrazinein good yield.³⁴ The disubstituted tetrazineis then reacted with a strained alkyne or alkene through an IEDDAreaction. It is worth noticing that the last step, the cycloadditionreaction, is completely chemoselective and is particularly well suitedto introduce unprotected chelators or moderately stable organic-basedfluorophores (Figure1).

Three receptors of interest in tumor imaging, GRPR, PSMA,and CXCR4,were chosen to illustrate the approach and targeted with compounds1,2, and3, respectively (Figure2).1 ([dPhe⁶, Sta¹³, Leu¹⁴]BBN(6–14),also known as JMV594³⁶) is a GRPR antagonistdisplaying nanomolar affinity toward GRPR. It is the targeting moietyof multiple GRPR-seeking radiotracers in clinical trials for prostateand breast cancer.³⁷2 (KuE,lysine-urea-glutamate) is a low-molecular-weight inhibitor of PSMA.Urea-based PSMA inhibitors have been heavily investigated for theiruse in nuclear medicine since the mid-2000s,³⁸ and KuE-based [¹⁷⁷Lu]Lu-PSMA-617 (Pluvicto) was FDA approvedin March 2022 for treatment of progressive, PSMA-positive metastaticcastration-resistant PCa. CXCR4 is a chemokine receptor overexpressedin more than 23 cancer types, such as kidney, prostate, lung, or brain.¹⁴3 (cyclo(DTyr-N(Me)DOrn-Arg-2-Nal-Gly))is a CXCR4 antagonist exhibiting nanomolar affinity toward the receptorand the targeting moiety of Pentixafor, an imaging agent for CXCR4in clinical trials for imaging multiple myeloma among other diseases.⁴⁰

The vectors were coupled to the tetrazine scaffoldvia spacersof similar length based on β-alanine. Different building blockswere designed depending on the chemical function involved in the couplingreaction with the tetrazine derivative; in this example,N-Fmoc-(βAla)₂-OH4,N-Boc-(βAla)₂-OH5, or 3-(3-(tritylthio)propanamido)propanoicacid6 were used (Figure3). Compounds4,5, and6 were synthesized from the corresponding amine and carboxylicacid using TSTU as a coupling reagent (seeSupporting Information).

Prior to reaction with dichloro-s-tetrazine, thiol-substituted7 was obtained by standardFmoc/tBu solid-phase peptide synthesis(SPPS) (Scheme1, seeSI for synthesis details). Amino-functionalizedKuE was obtained by peptide coupling of KuE(tBu)₃8 with4 using HATU as a coupling reagentto provide9 after Fmoc deprotection (Scheme1).tris-tBuester9 had to be used since we noticed that dichloro-s-tetrazine was able to react with both amines and carboxylates.Amino and thiol derivatives of compound3 were synthesizedby coupling of10 with5 or6 to provide after deprotection the desired CXCR4-targeting moieties11 and12 (Scheme1).

The procedure to generate heterobivalent ligandswas as follows:amino-functionalized vectors9 or11 werereacted with at least 1 equiv of dichloro-s-tetrazinein DMF in the presence of DIPEA for 10 min. The reaction mixture wasthen purified by RP-HPLC to yield chlorotetrazines13 and14. It is worth noting that monochlorotetrazinescan be prone to hydrolysis; thus, they must be stored at −20°C or immediately used in the next step (Scheme2). Additionally, it is important to stressthe fact that the first substitution is not chemospecific, and dichloro-s-tetrazine will react with amines or thiols but also withmild nucleophiles such as carboxylates, leading to the formation ofmonocarboxytetrazines,⁴¹ which will affordhydroxytetrazines after spontaneous hydrolysis. Amines were chosenas the first reactive partner, since substitution with an amine deactivatessignificantly the tetrazine core toward the next substitution; a firstsubstitution with a thiol might result in disubstituted compoundsif stoichiometry is not carefully controlled.

The thiol-functionalized7 or12 wasthen reacted with the isolated monosubstituted chlorotetrazine inDMF using DIPEA as a base (Figure4b). As an illustration of the deactivation of the chlorotetrazine,the second substitution must be performed at high temperature (80°C) and yielded the disubstituted tetrazine in 4 h (Figure4c). After completionof the reaction, a BCN-derivatized probe (1.2–3 equiv) wasadded to the reaction and stirred overnight at 37 °C to affordan HBL functionalized with a far-red fluorescent dye (Cy5) or differentchelators for labeling with radiometals for diagnosis and therapy(DOTA, NODAGA) in yields ranging from 4 to 40% (Scheme2). In the case of the cycloaddition, a higherreaction concentration resulted in shorter reaction times (Table1, lines 2 and 6).Interestingly, the reaction could be performed at concentrations aslow as 4 mM affording the target compounds in moderate yields (Table1, lines 1 and 5).In these cases, the use of an excess of alkyne was preferred. Thelowest synthetic yield was achieved with compound20,which can be explained by the use of the acid-sensitive Cy5 fluorescentdye. The replacement of TFA with formic acid for the cleavage oftert-butyl esters led to partial formylation of the peptidemoiety. Subsequent treatment of the crude mixture with aqueous ammoniaallowed the obtention of the target compound (seeSupporting Information). All these additional treatments,an additional purification, along with the sensitivity of the Cy5seem to lead to a low yield.

To increase the potency of this method, we havedeveloped a one-potversion of this procedure, which allowed to avoid several intermediatepurifications and lyophilizations, saving time and increasing theoverall yield. In order to validate the one-pot strategy, HBL18 was resynthesized and obtained in 26% yield (vs. 13% witha stepwise procedure). The analytical traces of each synthetic stephave been provided inFigure4 to show the efficiency of each reaction (Figure4).

In addition to beingbeneficial in terms of time—the synthesisof a HBL can be performed in less than 24 h—the one-pot procedureis also efficient, as it resulted in a 3-fold increase in yield. Thismethodology was successfully exemplified with the synthesis of othermonovalent and heterobivalent conjugates (seeSupporting Information).

In order to illustrate thefact that the assembly methodology providesfully functional conjugates without hampering the interaction of theligands with their receptor, we set out to evaluate the tumor targetingproperties of conjugate18in vivo.

Compounds were radiolabeled with [⁶⁸Ga]GaCl₃ in sodium formate buffer (6.6 mg, pH 3.5) for 10 min at 40 °C(18 andPSMA 11) or at 95 °C (AMBA). The radioconjugates were purified on a C18 Sep-Pakcartridge to achieve radiochemical purities >98%, as determinedbyradio-HPLC. Radiochemical yields ranged from 79 to 95% (not optimized).Molar activities were 12.7 ± 6.1 MBq/nmol forin vitro studies and ranged from 2.44 to 10.94 MBq/nmol forin vivo studies. Forin vivo experiments, radiolabeling,quality control, and intravenous injection were performed within 1h.PSMA 11 andAMBA were used as standardsfor validation of the PSMA and GRPR double xenograftsin vivo.⁴²

The distribution coefficientat pH 7.4 (logD) of [⁶⁸Ga]Ga-18 was determinedby the shake-flask method obtainedfrom saturated octanol-PBS solution (0.1 M, pH 7.4). The radiolabeledtracer exhibited a logD of −2.47 ± 0.19, which is typicalof peptide-based tracers and can be attributed to the presence ofseveral hydrophilic groups such as carboxylate groups and amide bonds.

Saturation binding experiments of [⁶⁸Ga]Ga-18 toward GRPR and PSMA were determined on PC3 cells (PSMA^–/GRPR⁺) and 22Rv1 cells (GRPR^–/PSMA⁺). AK_D value of 1.9 ± 1.2nM toward the GRPR was obtained on PC3 cells, and a value of 17.7± 4.9 nM toward the PSMA was obtained on 22Rv1 cells (22Rv1 expressesa small amount of GRPR; in this case, the GRPR interaction was blockedby coincubation with 10 μM of bombesin). The affinity of [⁶⁸Ga]Ga-18 toward the GRPR and the PSMA was lowerthan GRPR- and PSMA-targeting standards such as [⁶⁸Ga]Ga-RM2(K_D = 0.3 ± 0.2 nM on PC3 cells)⁴³ and [⁶⁸Ga]Ga-PSMA-617 (K_i = 2.3 ± 2.9 nM on LnCaP cells)⁴⁴ but still in the nanomolar range.⁴⁴

To assess [⁶⁸Ga]Ga-18in vivo, a model of athymic nude mice doubly xenografted with PC3 cellson the left shoulder and 22Rv1 cells on the right shoulder was used.⁴² To validate the model, two well-establishedradioligands of PSMA and GRPR, respectively, [⁶⁸Ga]Ga-PSMA11 (1.18 nmol, 7.67 MBq) and [⁶⁸Ga]Ga-AMBA (763 pmol, 5.41 MBq), were injected through retro-orbitalinjection.

Images recorded 50–70 min post injection (p.i.)showed thatboth tracers were able to exclusively accumulate in the xenograftoverexpressing their complementary receptor, which demonstrated uptakespecificity (3.46% ID/g for [⁶⁸Ga]Ga-AMBA inPC3 tumor and 3.22% ID/g for [⁶⁸Ga]Ga-PSMA11 in 22Rv1 tumor,Figure5). Both radioconjugates show uptake in the kidneys, whichis typical of peptide-based radiotracers that, with few exceptions,benefit from renal excretion. [⁶⁸Ga]Ga-AMBA shows uptake in the gastrointestinal tract which is due to the presenceof GRPR in the pancreas, stomach and colon.⁴⁵

At only 30–50 min p.i., [⁶⁸Ga]Ga-18 (1.88 nmol, 4.1 MBq) already showed accumulation in targetedtumorxenografts with 1.76% ID/g for PC3; 1.66% ID/g for 22Rv1 (measuredby quantitative PET imaging, seeSupporting Information) and GRPR- or PSMA-positive tissues such as pancreas and kidneys(Figure5, right).Quantitative PET measurements 20 min later did not show additionalaccumulation in both tumors (1.75% ID/g for PC3; 1.69% ID/g for 22Rv1).A significant decrease in the signal from the blood pool (1.02% ID/gfor heart), showing circulation of the radiotracer and washout fromthe background, let us anticipate an improved contrast at later timepoints in comparison with the images at 30–50 min (Figure5). Most signal wasobserved in the kidneys and in the bladder, suggesting that [⁶⁸Ga]Ga-18 is eliminated mainly through renal/urinaryexcretion. A weak signal in the gallbladder and high background inthe abdominal region suggested a small amount of hepatobiliary excretion.The uptake of [⁶⁸Ga]Ga-18 in each xenograft(1.75 and 1.69% ID/g in PC3 and 22Rv1 xenografts) at 50–70min was lower than the reference compounds (3.46 and 3.22% ID/g respectively).We hypothesize that this difference might be due to the structureof18, which has not been optimized likeAMBA orPSMA11.

These results were confirmed via abiodistribution study, performed2 h p.i.. Uptake specificity was confirmed by blocking experiments.In short, [⁶⁸Ga]Ga-18 was injected alone (533± 72 pmol) or in combination with a large excess of bombesin(BBN(1–14)) or PMPA (1000 mol equiv). Mice were euthanizedafter 2 h, organs were collected and weighed, and the amount of radioactivitywas measured in a γ-counter. Data showed a fast clearance fromthe blood and radioactivity accumulated in organs positive to PSMAand GRPR (kidneys, pancreas) and the tumor xenografts, thus confirmingthe behavior observed by PET imaging. Blocking of each receptor resultedin a statistically significant decrease of radioactivity in both PC3and PSMA xenografts, confirming the specificity of the uptake of [⁶⁸Ga]Ga-18. The uptake in the tumors was 1.31± 0.11% ID/g (PC3) and 1.01 ± 0.15% ID/g (22Rv1) (Figure6).

In summary, a straightforward method based on theuse of dichloro-s-tetrazine to quickly access labeledheterobivalent ligandshas been successfully developed. We have demonstrated that the methodis suited to easily assemble thiol- and amino-functionalized peptide-basedligands and different labels useful for medical imaging. The strategy,based on the use of multiple chemoselective reactions, successfullyprovided complex molecular structures in a straightforward, time-saving,and modular manner. The efficiency of our procedure has been furtherincreased by developing a one-pot approach giving 3 times higher yieldsin a shorter amount of time than the stepwise approach. To our knowledge,this is the first example of peptide-based heterobivalent ligandsfor imaging assembled in one pot.In vitro andin vivo evaluation of the pharmacological properties ofone of the conjugates demonstrated that the platform-based assemblymethod does not hamper the interaction of the ligands with their receptors.

Based on the increasing number of studies on bivalent and heterobivalentligands, we expect that this methodology will be of interest in thefields of drug design, drug delivery, imaging, and peptide receptorradiotherapy.