Effect of Small-Molecule–Binding Affinity on Tumor Uptake In Vivo: A Systematic Study Using a Pretargeted Bispecific Antibody

This article is featured in Highlights of This Issue, p. 1219

Radiolabeled agents have been used as delivery vehicles of ionizing radiation to specific disease sites for more than 50 years (1–6). Their systemic administration allows the treatment of widely disseminated disease, as opposed to external beam radiotherapy, which is used for the treatment of known disease sites. Therapeutic radiopharmaceuticals are designed to exhibit high specificity to the targeted disease site with low accumulation in normal tissues producing minimal radiation damage to normal cells.

A large number of molecules have been considered for targeted delivery of radioisotopes, including radiolabeled antibodies, antibody fragments, alterative scaffolds, and small molecules (7–10). While antibodies exhibit excellent binding specificity, they also exhibit long half-lives in the blood resulting in low tumor-to-background ratios. Antibody fragments and other smaller binding scaffolds exhibit faster blood clearance but result in high kidney and/or liver uptake. Radiolabeled small-molecule ligands generally exhibit more rapid blood clearance and lower background than antibodies and antibody fragments but usually result in poor specificity due to relatively low affinities for the desired target. Thus, there is a strong interest in developing small molecules with higher affinities both through improved high-throughput screening techniques (11) and through affinity enhancement using avidity (12–14).

Another approach to generate high-affinity binding of small molecules to disease sites is to use a method called pretargeted radioimmunotherapy (15–17). This approach couples the high-binding specificity of antibodies with the rapid clearance of radiolabeled small molecules, resulting in high tumor uptake yet fast clearance from nontumor tissue.

We have engineered a high-affinity antibody fragment with specificity for DOTA (1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid) metal chelates for pretargeted radioimmunotherapy applications (18). A particularly unique feature of the engineered scaffold is its ability to bind to different DOTA chelates with a wide range of affinities. Here, we use it in a pretargeted approach to systematically analyze the effect of affinity on tumor uptake in vivo. We compare these results with a compartmental model that has been extended from previous work and use simple analytic relations to derive design criteria to guide engineering efforts in the development of small-molecule radiotherapeutics. We present here a unique analysis of affinity in tumor targeting and discuss its implications in pretargeted radioimmunotherapy and small-molecule targeting.

DOTA, S-2-(R-aminobenzyl)-1,4,7,10-tetraazacyclododecane tetraacetic acid (DOTA-Bn), and S-2-(4-isothiocyanatobenzyl)-1,4,7,10-tetraazacyclododecane tetraacetic acid (DOTA-SCN) were purchased from Macrocyclics. All other chemicals were purchased from Sigma-Aldrich or Thermo Fisher Scientific unless specified otherwise. Sm3e/C825 bispecific antibody (bsAb) was produced by transient HEK cell transfection and purified as described (19).

Five milligrams (10 nmol) of 500-kDa amino dextran purchased from Invitrogen with 136 moles of amine per mole dextran was reacted with 3.7 mg (5.3 μmol) DOTA-SCN in 1 mL dimethyl sulfoxide with 1.9 μL (13.6 μmol) triethylamine (TEA) overnight at room temperature with mild vortexing. The dextran reaction mixture was diluted with 14 mL of 0.4 mol/L sodium acetate (pH 5.2) and 53 μmol yttrium nitrate was added. The mixture was incubated overnight at 37°C, dialyzed against water, and then dried down by vacuum centrifugation. The dried dextran compound was resuspended in PBS and purified by size exclusion chromatography using a Superdex 75 10/300 GL column. Fractions containing the dextran compound were combined, dialyzed against water twice, dried by vacuum centrifugation, resuspended in saline, and 0.2-μm filtered. The final dextran-DOTA-Y contained approximately 130 DOTA molecules as assessed by a TNBSA assay (Thermo Fisher Scientific).

DOTA and DOTA-Bn were dissolved at 0.5 mmol/L in ammonium acetate (pH 5.6). About 1 to 2 mCi ¹⁷⁷LuCl₃ (PerkinElmer) or ¹¹¹InCl₃ (Cardinal Health) were added to the metal chelate and incubated for 1 to 2 hours at 85°C to 95°C. The radiolabeled compounds were purified by reverse-phase high-performance liquid chromatography (RP-HPLC; refs. 14, 20) with γ detection on a 4.6 × 75 mm² Symmetry C18 column using a linear gradient from 0% to 40% B over 15 minutes, at a flow rate of 1 mL/min, where A = 10 mmol/L TEAA and B = methanol. The purified compounds were dried under vacuum, resuspended in saline, and filter-sterilized.

¹¹¹In-DOTA-dextran was prepared by synthesizing dextran-DOTA as described above, without loading with cold yttrium. Dextran-DOTA was incubated with 1 to 2 mCi ¹¹¹InCl₃ for 1 hour at 37°C followed by concentration and dilution with saline as described above.

All animal handling was conducted in accordance with Beth Israel Deaconess Medical Center Institutional Animal Care and Use Committee guidelines. LS174T and C6 cells were obtained from American Type Culture Collection and maintained under standard conditions. The cell lines were confirmed to be negative for mycoplasma and mouse pathogens by the Yale Virology Lab (Yale School of Medicine, New Haven, CT). Xenograft tumors were established in left and right flanks, respectively, of 5- to 6-week-old make NCRU-nu/nu mice (Taconic Farms) as described previously (19).

An IgG-based bsAb, Sm3e/C825, that binds to carcinoembryonic antigen (CEA) and DOTA metal chelates has been previously described (19). The bsAb binds to CEA with an apparent affinity of about 100 pmol/L and to ¹⁷⁷Lu-DOTA-Bn, ¹⁷⁷Lu-DOTA, ¹¹¹In-DOTA-Bn, and ¹¹¹In-DOTA with affinities of approximately 10 pmol/L, 400 pmol/L, 1 nmol/L, and 20 nmol/L, respectively (18, 19). Five hundred micrograms (2.5 nmol) Sm3e/C825 bsAb (19) was intravenously injected into LS174T and C6 (CEA-negative tumor used as a control for nonspecific tumor uptake) tumor-bearing mice followed by intravenous injection of 250 μg (0.45 nmol) dextran clearing agent 24 hours later to clear residual bsAb in the blood before administration of the radiolabeled DOTA. One hundred to 150 μCi (2–8 pmol) ¹⁷⁷Lu-DOTA-Bn, ¹⁷⁷Lu-DOTA, ¹¹¹In-DOTA-Bn, or ¹¹¹In-DOTA was injected intravenously 1 hour following clearing agent administration. Blood was collected from the tail vein using microcapillary tubes and counted on a model 1470 Wallac Wizard (PerkinElmer) 10-detector γ-counter. At various times, mice were euthanized by intraperitoneal injection of pentobarbital followed by cervical dislocation, a method consistent with the recommendations of the Panel on Euthanasia of the American Veterinary Medical Association. Organs and tumors were resected, washed 3 times in PBS, weighed, and counted as described above.

SPECT/CT (single-photon emission computed tomography/computed tomography) scans and image analyses were conducted using a rodent scanner (NanoSPECT/CT, Bioscan) equipped with an 8W X-ray source running at 45 kV (177 μA) and a 48-μm pitch CMOS-CCD X-ray detector. Mice were anesthetized in an anesthetic chamber with isoflurane and transferred to a bed on a gantry for imaging where gas anesthesia was maintained for the duration of the scan. After acquisition of a CT topogram, helical micro-SPECT was conducted using a 4-headed γ camera outfitted with multi-pinhole collimators (1.4 mm) and a total scan time of 45 minutes. SPECT images were acquired over 360 degrees in 24 projections each using a 256 × 256 frame size (1.0-mm pixels). Images were reconstructed with Bioscan HiSPECT iterative reconstruction software and fused with CT images. Immediately after scanning, mice were sacrificed; tissues and tumors were weighed and counted as described above.

Tumor uptake of radiolabeled small molecules was simulated using a mechanistic compartmental model extended from previous work (21, 22), with the assumption that radioisotope that is internalized into cells remains trapped within the cell (Supplementary Materials and Methods). The tumor concentration of residualizing isotope following a subsaturating intravenous injection of radiolabeled small molecule can therefore be described as follows:

where C_tumor is the overall concentration of the isotope in the tumor, t is time, C_p0 is the initial plasma concentration of the radiolabeled small molecule, k_e is the rate of endocytosis (s⁻¹), BP is the binding potential B_max/K_D, B_max is the concentration of total antigen in the tumor, and K_D is the binding affinity, K^trans is the transcapillary transport rate (s⁻¹), k_cl is the plasma clearance rate of the small molecule (s⁻¹), and ϵ is the available volume fraction of the small molecule in the tumor. Parameter values were measured experimentally as described or obtained from published literature (Supplementary Table S1).

Using the above model, the concentration of residualized isotope in the tumor as time goes to infinity is as follows:

The bsAb Sm3e/C825, composed of the engineered high-affinity antibody fragment C825 with specificity for DOTA metal chelates (18), was used in a pretargeted protocol to target DOTA chelates to CEA-expressing xenograft tumors. A schematic depicting the pretargeted approach is shown in Fig. 1. C825 binds to different DOTA chelates with widely varying affinities dependent on the chelated metal and the presence or lack of an aminobenzene group attached to a carbon in the macrocycle backbone of DOTA. The organ/tissue biodistribution at 24 hours postinjection of the hapten (¹⁷⁷Lu-DOTA-Bn, ¹⁷⁷Lu-DOTA, ¹¹¹In-DOTA-Bn, or ¹¹¹In-DOTA) was determined in tumor-bearing mice (Fig. 2 and Supplementary Table S2). The activity in the LS174T tumor increased with increasing hapten affinity from 0.5%ID/g ± 0.1%ID/g for ¹¹¹In-DOTA (K_D = 20 nmol/L) to 1.6%ID/g ± 0.3%ID/g for ¹¹¹In-DOTA-Bn (K_D = 1 nmol/L) to 14.3%ID/g ± 1.8%ID/g for ¹⁷⁷Lu-DOTA (K_D = 400 pmol/L). The tumor activity for ¹⁷⁷Lu-DOTA-Bn (K_D = 10 pmol/L) was 19%ID/g ± 4%ID/g and not significantly different than that of the 400 pmol/L affinity ¹⁷⁷Lu-DOTA. The activity in the C6 antigen–negative tumor also increased with affinity, due to higher-affinity binding to bsAb retained nonspecifically through the enhanced permeability and retention (EPR) effect (23). Activities in nontumor tissues are also higher for the highest affinity compounds due to higher affinity binding to residual bsAb. The tumor-to-kidney ratio increased from 1.2 ± 0.4 for about 20 nmol/L to 17 ± 3 for about 400 pmol/L but then decreased to 10 ± 2 for ∼10 pmol/L affinity due to higher uptake in the kidney yet similar tumor uptake. Similarly, the tumor-to-blood ratio was highest for 400 pmol/L affinity ¹⁷⁷Lu-DOTA at 380 ± 90.

One mouse from each affinity group was imaged by SPECT/CT (Fig. 3). For the ¹¹¹In isotope, visible tumor signal is observed in the antigen-positive LS174T tumor at 24 hours postinjection for 1 nmol/L ¹¹¹In-DOTA-Bn; however, no significant signal is observed for 20 nmol/L ¹¹¹In-DOTA. For the ¹⁷⁷Lu isotope, excellent tumor targeting is observed for both 400 pmol/L ¹⁷⁷Lu-DOTA and 10 pmol/L ¹⁷⁷Lu-DOTA-Bn. Some signal is also observed in the antigen-negative tumors, as expected from the biodistribution data. It should be noted that while ¹¹¹In and ¹⁷⁷Lu have similar reconstructed resolutions, the average sensitivity of ¹¹¹In is about 5 times greater than ¹⁷⁷Lu in mice (24).

The use of the clearing agent 1 hour before hapten administration resulted in significantly better tumor-to-background ratios compared with a 2-step protocol (Supplementary Fig. S1). For the pretargeted protocol, the dextran-DOTA compound was loaded with nonradioactive yttrium as it is one of the metals, when chelated to DOTA and DOTA-Bn that exhibits the highest affinity to C8.2.5. The clearing agent clears rapidly from the blood through the liver and spleen with no observable tumor accumulation (Supplementary Fig. S2).

Because the ¹⁷⁷Lu-DOTA compound resulted in the highest tumor-to-background ratios of the 4 DOTA haptens, it was further characterized for pretargeted radioimmunotherapy applications. At 4 hours postinjection of ¹⁷⁷Lu-DOTA, tumor uptake was 7.44%ID/g ± 0.41%ID/g in the antigen-positive tumor (Table 1 and Supplementary Fig. S3), approximately 90-fold higher than the tumor uptake observed for ¹⁷⁷Lu-DOTA alone (Supplementary Table S3). Tumor uptake in the antigen-negative tumor was 9.82%ID/g ± 0.35%ID/g at 4 hours, similar to the antigen-positive tumor due to the EPR effect. Over time, the tumor activity in the antigen-negative tumor decreased to 4.23%ID/g ± 0.54%ID/g at 24 hours and 2.89%ID/g ± 2.28%ID/g at 48 hours whereas the tumor activity in the antigen-positive tumor increased to 14.3%ID/g ± 1.8%ID/g at 24 hours and remained essentially constant at 48 hours. The LS174T tumor-to-blood ratio increased from 18 ± 2 at 4 hours to 380 ± 90 at 24 hours and was greater than 450 at 48 hours (Table 2). At 48 hours, the blood activity was not measurable above background. The LS174T tumor-to-kidney ratio increased from approximately 8 at 4 hours to about 20 at 24 and 48 hours.

We developed a mathematical model on the basis of an extension to a previously published compartmental model of tumor uptake of targeted agents (21). The model uses only measurable parameters, with no fit variables (Supplementary Table S1). The transcapillary transport rate of the DOTA-based compounds was assumed to be similar to that for Gd-DTPA (25). This model applies only to residualizing isotopes and targeted molecules that are not cell permeable. For radiotherapy, the goal is to retain the isotope for an extended period of time to allow for radioactive decay at the site of the tumor. It is known that radioactive metals and some forms of iodine are residualizing and retained intracellularly after internalization (26–28). Derivation of the model is provided in the Supplementary Materials and Methods. The analytic solution allows straightforward analysis of the effect of changing parameters on tumor uptake.

From this equation, we derive a metric for affinity, BP > K^trans/(ϵ × k_e), for maximum tumor uptake of isotope for a given radiotherapeutic/antigen system. From this metric, we predict that faster internalization will lead to a lower affinity requirement and that higher affinity is required for ligands with faster transcapillary transport and for antigen targets with lower B_max.

For the particular system studied here, |\frac{{{\raise0.7ex\hbox{${K^{{\rm trans}}}$} \!\mathord{\left/ {\vphantom {{K^{{\rm trans}}} \varepsilon}}\right.\kern-\nulldelimiterspace}\!\lower0.7ex\hbox{$\varepsilon $}}}}{{k_{\rm e}}}|{\approx 400}⁠, therefore, it is predicted that saturating levels of signal should be obtained with |${\rm BP}\;{{\rm = }}\;{\frac{{B_{\max }}}{{k_{\rm d}}}} > 400$|⁠. For our measured value of B_max = 200 nmol/L, this corresponds to K_d < 0.5 nmol/L, consistent with the experimental results (Fig. 4). Equation (D) also predicts a maximal residualized tumor signal of |${\lim _{t \to \infty}} = {\frac{{K^{{\rm trans}}}}{{k_{{\rm cl}}}}}C_{{\rm p0}}$| for the highest affinity capture; thus for the parameters in this system, |$\lim _{t \to \infty} C_{{\rm resid}} =$||$0.3\;C_{{\rm p0}} \cong 15$|%ID/g is the predicted highest dose attainable.

The experimental results of tumor uptake versus affinity compared very well with model prediction (Fig. 4) with the 24-hour tumor uptake increasing significantly from single-digit nanomolar to picomolar affinity and then reaching a plateau.

Here, we present a systematic study of the effect of affinity on tumor uptake of DOTA metal haptens using a previously engineered bsAb that binds with varying affinities to different DOTA chelates (18). The effect of binding affinity on tumor uptake has been previously described for antibodies and antibody fragments (29, 30). However, this is the first time, to our knowledge, that the effect of binding affinity on tumor targeting of a small molecule has been studied in vivo with the same target antigen resulting in unaltered internalization kinetics and B_max. Four compounds spanning a range of affinities over 3 orders of magnitude were studied. We show here that an affinity of 400 pmol/L is required for maximum uptake in the studied system with an internalization half-life of about 13 hours and a B_max on the order of 200 nmol/L (10⁵–10⁶ binding sites per cell, assuming typical cell densities for a vascular xenograft tumor; ref. 31). Further improvement in affinity to 10 pmol/L affinity does not significantly improve tumor uptake.

Tumor uptake of radiolabeled small molecules was simulated using a mechanistic compartmental model extended from previous work. The experimental results were consistent with model simulations. We further derived analytic relations to provide design criteria to guide engineering efforts in the development of small-molecule radiotherapeutics. The design criteria allow for prediction of a target affinity for the development of new radiotherapeutic agents. These relationships can guide experimental efforts in drug development.

The experimental and mathematical model results presented here suggest that a plateau exists for any given ligand–receptor pair such that further improvements in affinity result in no additional improvement in tumor uptake. The affinity range at which this plateau exists depends on the B_max, k_e, K^trans, and ϵ of the particular ligand and antigen. For example, in the system studied here, if B_max was reduced from 200 to 20 nmol/L, saturating levels of signal would require a 10-fold improvement in affinity. Beyond a given affinity, additional affinity improvement may result in decreased therapeutic efficacy in some applications by resulting in higher background due to improved binding to residual bsAb present at low concentrations in PRIT applications or improved uptake in normal tissues with low levels of antigen expression in one-step approaches.

It should be noted that the clearing agent did not appear to completely clear circulating bsAb, as the amount of background signal increased with increasing hapten affinity (Fig. 2). In addition, the LS174T tumor activity increases from 4 to 24 hours for pretargeted ¹⁷⁷Lu-DOTA (Supplementary Fig. S3). While the simplified model presented here does not take antibody kinetics into account, the correlation of the experimental data with the model suggest that the relationships derived here may be useful in the design of tumor-targeting small molecules. Additional experiments with more efficient clearing of the bsAb would provide further data to support the model.

In addition to the affinity series, we present a method for pretargeted radioimmunotherapy that uses an IgG-scFv bsAb, a dextran-based clearing agent, and radiolabeled DOTA. ¹⁷⁷Lu-DOTA has previously been shown to exhibit very rapid whole-body clearance from mice (32). Here, we show high LS174T tumor uptake and retention of ¹⁷⁷Lu-DOTA with fast clearance from nontumor tissue resulting in the highest yet reported tumor-to-kidney ratios at 48 hours postinjection for CEA targeting.

A significant amount of ¹⁷⁷Lu-DOTA uptake is observed in the CEA-negative tumors at early times. The EPR effect results in nonspecific tumor accumulation of high-molecular-weight compounds. While approximately 4-fold higher bsAb uptake is observed in LS174T tumors versus C6 tumors (19), a significant fraction of the bsAb localized to the LS174T tumors will be inaccessible to binding due to the approximately 13-hour internalization half-life of CEA (33), whereas all bsAb in the C6 tumors will be accessible to hapten binding. This is consistent with the observation of similar hapten activities in the 2 tumors at early times. At later times, unbound antibody slowly intravasates out of the CEA-negative tumor, whereas CEA-bound antibody in the LS174T tumors internalizes ¹⁷⁷Lu-DOTA compounds where the radiolabel is trapped within the cell.

An engineered IgG-like bsAb was used here to harness the established therapeutic advantages of IgGs. The bsAb possesses slow blood clearance resulting in high tumor uptake, retains potentially beneficial secondary immune function, and can be produced and purified in a fashion identical to that of an IgG (19). The system uses DOTA-chelated metal as the hapten, with no additional synthesis or modification required. This eliminates any issues with linker cleavage and peptide stability that have been reported for other haptens in PRIT applications (34, 35). DOTA chelated to gadolinium has been administered to human subjects in millimolar concentrations and has an established safety profile. DOTA metal chelates exhibit rapid blood clearance and whole-body clearance observed in mice (32) and humans (36). A particularly useful advantage to the approach described here is that only radiolabeled DOTA will bind to the pretargeted anti-DOTA–binding sites, whereas unlabeled DOTA exhibits no observable binding. This results in very high effective-specific activity without requiring complex and time-consuming purification schemes.

While 3-step pretargeted radioimmunotherapy adds complexity over 2-step procedures, it allows higher doses of bsAb to be administered resulting in higher achieved tumor doses as well as more homogenous distribution within the tumor (37). In addition, it allows for possible secondary immune effects resulting from the retained Fc domain that may prove significant (38). Two-step approaches may be sufficient for molecular imaging leading to improved cancer screening and staging (39, 40). However, it is anticipated that the increased number of hapten-binding sites afforded by 3-step approaches will prove critical for therapy.

We present here a method for pretargeted radioimmunotherapy and use it in a systematic study of the effect of small-molecule affinity on tumor uptake in vivo with affinities spanning a range of 3 orders of magnitude. In addition, we develop a mathematical model of tumor targeting using only known, measured parameters that correlates well with experimental observations. We predict that this model will be useful for rational design of new agents and to guide experimental efforts in the development and optimization of targeted cancer therapeutics.

No potential conflicts of interests were disclosed.

Conception and design: K.D. Orcutt, K. Dane Wittrup

Development of methodology: K.D. Orcutt, J.V. Frangioni, K. Dane Wittrup

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.D. Orcutt, J.J. Rhoden, B. Ruiz-Yi, J.V. Frangioni

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.D. Orcutt, B. Ruiz-Yi, J.V. Frangioni, K. Dane Wittrup

Writing, review, and/or revision of the manuscript: K.D. Orcutt, J.J. Rhoden, K. Dane Wittrup

Study supervision: K. Dane Wittrup

The authors thank David G. Whitehead and Fangbing Liu, PhD, for help with cell culture; Elaine P. Lunsford for assistance with SPECT/CT imaging; Hak Soo Choi, PhD, for helpful discussions; Lorissa Moffitt for editing; and Eugenia Trabucchi and Donald McGaffigan for administrative support.

The study was supported by National Science Foundation Graduate Research Fellowships (K.D. Orcutt and J.J. Rhoden); Lewis Family Fund (J.V. Frangioni); and NIH grant R01-CA-101830 (K.D. Wittrup).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.