Harnessing Fc-Directed Bioconjugation for the Synthesis of Site-Specifically Modified Radioimmunoconjugates

Abstract

A growing body of preclinical and clinical evidence has shown that site-specifically and site-selectively modified immunoconjugates exhibit improved in vivo performance compared to their stochastically modified cousins. However, extant approaches to site-specific bioconjugation suffer from a variety of issues that make clinical translation challenging, including instability, complexity, and expense. Herein, we describe a novel chemical approach to the synthesis of site-specifically modified radioimmunoconjugates that is simple and straightforward. To this end, we leveraged an Fc-directed peptide to append free sulfhydryl moieties to unique sites within the Fc domain of the CA19-9-targeting antibody 5B1. These thiols were then modified with phenyloxadiazolyl methylsulfone-bearing variants of the chelator CHX-A″-DTPA, and the immunoconjugate was radiolabeled with [¹⁷⁷Lu]­Lu³⁺ to produce [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1 in high yield, purity, and specific activity. Subsequent analyses confirmed the site-specificity of the modification and demonstrated the high stability and immunoreactivity of the radioimmunoconjugate. Biodistribution studies in athymic nude mice bearing subcutaneous BxPC3 pancreatic cancer xenografts revealed that [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1 produced high activity concentrations in tumor tissue as well as high tumor-to-background activity concentration ratios and displayed performance that compared favorably to ¹⁷⁷Lu-labeled analogues synthesized with traditional stochastic and thiol-selective bioconjugation strategies.

Over the past two decades, immunoconjugates have become indispensable clinical tools for the imaging and treatment of cancer. Indeed, both full length monoclonal antibodies (mAb) and antibody fragments have emerged as effective vectors for the delivery of a wide range of cargoes to tumor tissue. Historically, immunoconjugates have been synthesized via the stochastic modification of solvent-exposed amino acids  most often lysines  on the surface of the immunoglobulin. Although this approach to bioconjugation is facile and inexpensive, it relinquishes control over the molecular location of the modification as well as the number of cargoes per antibody (i.e., the degree-of-labeling, DOL). As a result, stochastic bioconjugation strategies inevitably produce heterogeneous and poorly defined mixtures of immunoconjugates that can exhibit suboptimal in vitro and in vivo behavior.

To circumvent these issues, the field has increasingly turned to ‘site-specific’ and ‘site-selective’ methods of bioconjugation designed to attach cargoes to unique sites within immunoglobulins. A variety of innovative and effective approaches to site-specific and site-selective bioconjugation have been developed, including strategies based on unnatural amino acids, click chemistry, enzymatic reactions, the manipulation of the heavy chain glycans, and the modification of the interchain disulfides. , Critically, a growing body of preclinical and clinical evidence strongly suggests that these immunoconjugates boast improved in vivo performance compared to their stochastically modified cousins. − In addition, a recent clinical trial comparing the efficacy of stochastically and site-specifically labeled variants of ⁸⁹Zr-DFO-pertuzumab for the immunoPET of patients with metastatic HER2-expressing malignancies suggests that the latter may produce images with enhanced tumor-to-background contrast. However, each extant approach to site-specific bioconjugation comes with its own set of scientific, practical, and logistical limitations. Thiol-mediated strategies, for example, are simple and inexpensive but require the reduction of the mAb, produce mixtures of regioisomers, and frequently rely on maleimides that form unstable thioether linkages. Chemoenzymatic methods, in contrast, eschew reduction and offer more stability but require lengthy incubations with (frequently expensive) enzymes that can pose challenges in the context of clinical production. Clearly, approaches to site-specific bioconjugation that balance ease, selectivity, and clinical translatability remain an urgent unmet need.

This investigation is predicated on the development of a straightforward and facile approach to the synthesis of site-specifically modified radioimmunoconjugates. To this end, we have harnessed the ‘AJICAP’ reagent: a 17-amino acid cyclic peptide capable of selectively binding the Fc regions of IgG₁, IgG₂, or IgG₄ antibodies and facilitating the site-specific, covalent modification of their K248 residues. More specifically, the reagent recognizes a protein A recognition motif within the Fc domain known as Z34C, thereby positioning a reactive thiophenyl ester in close proximity to the primary amine of K248. Subsequent treatment of the resultant immunoconjugate with hydroxylamine cleaves an alkylthioester bond within the reagent, liberating the peptide and exposing a pair of free sulfhydryl moieties for modification with thiol-reactive probes. , Fujii et al. previously illustrated the versatility of this virtually ‘trace-less’ modification technology by creating stable and homogeneous  i.e. DOL = 1.9 ± 0.1 cargoes/mAb  antibody-drug conjugates of trastuzumab and rituximab using a variety of maleimide-bearing toxins (i.e., deruxtecan, maytansinoid, MMAE, MMAF, and tesirine).

Herein, we report the first application of this technology to the synthesis of radioimmunoconjugates. To this end, we have employed a model system with four key components (Figure ). The immunoglobulin at the heart of the investigation is 5B1, a clinically validated fully human mAb that targets CA19–9, a carbohydrate antigen that is overexpressed in a variety of malignancies but is most widely associated with pancreatic ductal adenocarcinoma (PDAC). − Next, we have selected lutetium-177 (¹⁷⁷Lu; t _1/2 ∼ 6.7 d), a β^–-emitting isotope frequently employed in radioimmunotherapy, as the radionuclide for this study. Finally, the bifunctional chelator pairs CHX-A″-DTPA (from here on referred to only as ‘DTPA’)  an acyclic chelator that has been used to stably coordinate [¹⁷⁷Lu]­Lu³⁺ in a wide variety of radiopharmaceuticals  with a phenyloxadiazolyl methyl sulfone (PODS) group that has been shown to selectively, rapidly, and irreversibly react with free sulfhydryl groups.

1.

Schematic of the synthesis of [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1 using the Fc-binding AJICAP reagent, NH₂OH, PODS-CHX-A″-DTPA, and ¹⁷⁷Lu.

The site-specifically modified immunoconjugate - DTPA-^PODSAJICAP-5B1 - was synthesized in a facile and straightforward manner (Figure ). First, the AJICAP reagent (5 equiv, of a 20 mM solution in DMF) was added to a solution of 5B1 (5.9 mg/mL in 20 mM sodium acetate buffer, pH 5.5), and the mixture was incubated for 1 h at room temperature. Subsequently, an excess of NH₂OH•HCl was added, and the solution was allowed to incubate for an additional 1 h. The product of this reaction, ^HSAJICAP-5B1, was then purified using an NAP-25 desalting column and eluted with phosphate-buffered saline supplemented with 10 mM EDTA (pH 7.4). Purified ^HSAJICAP-5B1 was then incubated with PODS-CHX-A″-DTPA in phosphate-buffered saline (Chelex-PBS, pH 7.4) for 2 h at RT. The chelator-bearing immunoconjugate was subsequently purified using size-exclusion chromatography, ultimately yielding the final product - DTPA-^PODSAJICAP-5B1 - in ∼ 65% yield from the parental mAb (11).

Two other chelator-bearing variants of 5B1 were also synthesized to facilitate in vitro and in vivo comparisons (Figure A). The first, DTPA-PODS-5B1, is a site-selectively modified probe synthesized via the reduction of the disulfide linkages of the mAb with TCEP and the subsequent reaction of the reduced antibody with PODS-CHX-A″-DTPA. The second, DTPA-5B1, is a stochastically conjugated immunoconjugate created via the incubation of the antibody with p-SCN-Bn-CHX-A″-DTPA under basic conditions. The purity of all three immunoconjugates was verified via SDS-PAGE (Supporting Figure S1), and size exclusion-HPLC of each of the probes revealed low levels (i.e., < 5%) of aggregation (Figure A and Supporting Figure S2). MALDI-ToF mass spectrometry was employed to determine the degree of labeling (DOL) of each chelator-bearing immunoconjugate, and the three probes all exhibited similar numbers of chelators per antibody: 2.9 ± 0.6 (DTPA-5B1), 2.8 ± 0.1 (DTPA-PODS-5B1), and 1.8 ± 0.1 (DTPA-^PODSAJICAP-5B1) (Supporting Table S1).

2.

(A) Schematic of the bioconjugation and radiolabeling of [¹⁷⁷Lu]­Lu-DTPA-5B1 (top), [¹⁷⁷Lu]­Lu-DTPA-PODS-5B1 (center), and [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1 (bottom); (B) radio-instant thin layer chromatograms of purified [¹⁷⁷Lu]­Lu-DTPA-5B1 (top), [¹⁷⁷Lu]­Lu-DTPA-PODS-5B1 (center), and [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1 (bottom). CPM = counts per minute.

3.

(A) SDS-PAGE of native 5B1, DTPA-5B1, DTPA-PODS-5B1, and DTPA-^PODSAJICAP-5B1; (B) autoradiography gel of [¹⁷⁷Lu]­Lu-DTPA-5B1, [¹⁷⁷Lu]­Lu-DTPA-PODS-5B1, and [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1. (C) 5-day serum stability curves of [¹⁷⁷Lu]­Lu-DTPA-5B1 (left), [¹⁷⁷Lu]­Lu-DTPA-PODS-5B1 (center), and [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1 (right); (D) bead-based immunoreactivity assays of [¹⁷⁷Lu]­Lu-DTPA-5B1 (left), [¹⁷⁷Lu]­Lu-DTPA-PODS-5B1 (center), and [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1 (right).

The radiolabeling of each of the immunoconjugates with [¹⁷⁷Lu]­Lu³⁺ was performed according to standard protocols (Figure A). Briefly, the chelator-bearing immunoglobulins (0.6 mg) were incubated with 3.0 mCi [¹⁷⁷Lu]­LuCl₃ in 50 mM NH₄OAc (pH 5.5) for 1 h at 37 °C. After an hour, the progress of the reaction was verified via radio-instant thin layer chromatography; the reaction, if complete, was quenched via the addition of 50 mM EDTA; and the radiolabeled mAb was purified via gel filtration chromatography (Figure B). Ultimately, each radioimmunoconjugate  [¹⁷⁷Lu]­Lu-DTPA-5B1, [¹⁷⁷Lu]­Lu-DTPA-PODS-5B1, and [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1  was isolated in >95% radiochemical yield, >99% purity, and a specific activity of ∼ 5 mCi/mg (Supporting Table S2). The serum stability of the radioimmunoconjugates was subsequently determined by incubating each ¹⁷⁷Lu-mAb in human serum for 5 d at 37 °C and periodically analyzing aliquots via radio-ITLC. All three radioimmunoconjugates proved stable to demetalation over this time period, with 92.2 ± 1.5% ([¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1), 95 ± 1.0% ([¹⁷⁷Lu]­Lu-DTPA-PODS-5B1), and 96.4 ± 0.2% ([¹⁷⁷Lu]­Lu-DTPA-5B1) remaining intact over the incubation period (Figure C). Next, the immunoreactivities of the ¹⁷⁷Lu-mAb were assayed via bead-based assay with CA19–9-coated magnetic particles (Figure D). All three radioimmunoconjugates displayed immunoreactive fractions >0.75 as well as ‘blockable’ binding of their antigen. Finally, the molecular location of the bioconjugation sites in each conjugate was interrogated via SDS-PAGE of the radioimmunoconjugates followed by autoradiography (Figure B). As expected given the location of K248, [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1 was radiolabeled exclusively on the heavy chain. Interestingly, however, both [¹⁷⁷Lu]­Lu-DTPA-PODS-5B1 and [¹⁷⁷Lu]­Lu-DTPA-5B1 also exhibited radiolabeling predominantly on the heavy chain, though some radioactive signal was associated with the light chain for the latter.

With the in vitro characterization of the trio of radioimmunoconjugates complete, the next step was to explore their in vivo performance in murine models of cancer (Figure ). To this end, athymic nude mice were inoculated with subcutaneous, CA19–9-expressing BxPC3 pancreatic ductal adenocarcinoma xenografts. Once the xenografts reached ∼100–200 mm³, the mice were administered [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1, [¹⁷⁷Lu]­Lu-DTPA-PODS-5B1, or [¹⁷⁷Lu]­Lu-DTPA-5B1 (100 μCi; 20 μg; in 100 μL 0.9% sterile saline) via the lateral tail vein. At three time points after inoculation  48, 96, and 144 h (n = 4 mice per radioimmunoconjugate per time point)  the mice were euthanized, and their tumors as well as several other tissues were collected, washed, dried, and assayed for radioactivity on a ¹⁷⁷Lu-calibrated gamma counter. These biodistribution data clearly showed that all three radioimmunoconjugates exhibited excellent tumor tropism and produced high tumor-to-healthy organ activity concentration ratios. Even at the earliest time point (i.e., 48 h postinjection), [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1, [¹⁷⁷Lu]­Lu-DTPA-PODS-5B1, and [¹⁷⁷Lu]­Lu-DTPA-5B1 produced tumoral uptake values of 17.9 ± 14.0, 32.5 ± 14.1, and 28.2 ± 14.6 %ID/g, respectively. At the same time point, the healthy tissues with the highest levels of uptake were the blood, liver, and spleen, all with activity concentrations at or below 13 %ID/g. As the experiment progressed, the radioimmunoconjugates cleared from the blood and other healthy tissues while the uptake in the xenografts increased, reaching values of 52.0 ± 24.5, 42.9 ± 13.7, and 18.8 ± 5.0 %ID/g for [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1, [¹⁷⁷Lu]­Lu-DTPA-PODS-5B1, and [¹⁷⁷Lu]­Lu-DTPA-5B1, respectively, at 144 h postinjection. While some of these values appear to differ, the only differences that are statistically significant is that between [¹⁷⁷Lu]­Lu-DTPA-5B1 and the two site-selectively modified constructs. At 144 h, the trio of radioimmunoconjugates produced broadly similar tumor-to-blood activity concentration ratios: 24.5 ± 15.6 ([¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1), 12.9 ± 6.2 ([¹⁷⁷Lu]­Lu-DTPA-PODS-5B1), and 10.6 ± 16.3 ([¹⁷⁷Lu]­Lu-DTPA-5B1). However, more variability was apparent in the tumor-to-bone activity concentration ratios at this time point, with [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1 (41.2 ± 22.7) higher than [¹⁷⁷Lu]­Lu-DTPA-5B1 (31.9 ± 9.9) and significantly higher than [¹⁷⁷Lu]­Lu-DTPA-PODS-5B1 (8.1 ± 3.2) (Supporting Table S3).

4.

Biodistribution data collected 48, 96, and 144 h after the administration of (A) [¹⁷⁷Lu]­Lu-DTPA-5B1, (B) [¹⁷⁷Lu]­Lu-DTPA-PODS-5B1, and (C) [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1 to athymic nude mice bearing subcutaneous BxPC3 xenografts; (D) Comparison of the biodistribution data in selected organs.

The chemical and in vitro data clearly illustrate that site-specific bioconjugation using the AJICAP reagent facilitates the synthesis of highly homogeneous, well-defined, and stable radioimmunoconjugates. The strength of this approach lies in the fact that it is mild, straightforward  i.e. the initial modification is a simple one-pot-two-step procedure  and purely chemical. Along these lines, the AJICAP technology provides several significant advantages over other extant approaches to site-specific and site-selective bioconjugation. Unlike strategies that rely upon unnatural amino acids, it eschews genetic engineering and can be employed with native antibodies. Unlike chemoenzymatic approaches, it avoids the use of enzymes as well as lengthy incubations. And unlike disulfide-mediated methodologies, it does not require the reduction of the antibody and is site-specific rather than site-selective. There are also several avenues for the continued development of this technology. As described here, the methodology requires three steps (two of them in a single pot) from antibody to completed immunoconjugate: (i) AJICAP conjugation, (ii) NH₂OH deprotection, and (iii) thiol-directed ligation. This strategy provides maximum modularity, as the first two yield a sulfhydryl-bearing construct that can react with any bifunctional, thiol-reactive cargo. However, a single-step procedure may also be useful. Indeed, it is possible to envision a one pot procedure in which an AJICAP-type reagent facilitates the site-specific installation of a cargo (i.e., a chelator or toxin) and is then removed to yield the completed immunoconjugate.

The tumor-to-bone activity concentration ratio of [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1 at 144 h postinjection was significantly better than that of [¹⁷⁷Lu]­Lu-DTPA-PODS-5B1, and the former’s tumoral uptake at the same time point was significantly superior to that of [¹⁷⁷Lu]­Lu-DTPA-5B1. Beyond these values, however, [¹⁷⁷Lu]­Lu-DTPA-^PODSAJICAP-5B1 did not yield broadly superior in vivo performance compared to the site-selectively and stochastically modified analogues. This is, from our point of view, not necessarily a demerit for the conjugation strategy, as 5B1 is a highly robust antibody that  even with the most haphazard stochastic bioconjugation  already yields high tumoral uptake and tumor-to-healthy organ activity concentration ratios. It is possible (even likely) that the in vitro and in vivo benefits of this site-specific bioconjugation will be more apparent with more fragile mAb. To explore this hypothesis, we are currently working to develop AJICAP-modified radioimmunoconjugates with several less optimized immunoglobulins.

In sum, this investigation demonstrates that the AJICAP reagent facilitates the straightforward synthesis of homogeneous, well-defined, and stable radioimmunoconjugates with excellent in vivo performance. We are currently working to simplify this approach, demonstrate its modularity across antibodies and chelators, and adapt the strategy for automation.

Glossary

Abbreviations

DTPA

diethylenetriaminepentaacetic acid

AJICAP

Ajinomoto Fc-directed cyclic peptide

CA19–9

carbohydrate antigen 19–9

Mal

maleimide

TCEP

tris­(2-carboxyethyl) phosphine

DOL

degree-of-labeling

mAb

monoclonal antibody

ADC

antibody-drug conjugate

K248

lysine 248

PDAC

pancreatic ductal adenocarcinoma

SDS-PAGE

sodium dodecyl sulfate-polyacrylamide gel electrophoresis

MALDI-ToF

matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry

SE-HPLC

size exclusion high performance liquid chromatography

PODS

phenyloxadiazolyl methyl sulfone

%ID/g

% of injected dose per gram

PBS

phosphate-buffered saline

DMSO

dimethyl sulfoxide

DMF

dimethylformamide

RPM

revolutions per minute

EDTA

ethylenediaminetetraacetic acid

iTLC

instant thin-layer chromatography

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.bioconjchem.5c00306.

• Experimental procedures, SE-HPLC chromatograms, SDS-PAGE results, tables of MALDI-ToF data, and tables of radiolabeling and biodistribution data (PDF)

⊥.

(C.G., J.S.) These authors contributed equally to this work.

The authors declare the following competing financial interest(s): Tomohiro Watanabe, Tsubasa Aoki, and Tomohiro Fujii are employees of Ajinomoto Co., Inc., which has intellectual property on the AJICAP reagent.