A Systematic Evaluation of Antibody Modification and 89Zr-Radiolabeling for Optimized Immuno-PET

Abstract

Immuno-PET using desferrioxamine (DFO)-conjugated zirconium-89 ([⁸⁹Zr]Zr⁴⁺)-labeled antibodies is a powerful tool used for preclinical and clinical molecular imaging. However, a comprehensive study evaluating the variables involved in DFO-conjugation and ⁸⁹Zr-radiolabeling of antibodies and their impact on the in vitro and in vivo behavior of the resulting radioimmunoconjugates has not been adequately performed. Here, we synthesized different DFO-conjugates of the HER2-targeting antibody (Ab)-trastuzumab, dubbed T5, T10, T20, T60, and T200-to indicate the molar equivalents of DFO used for bioconjugation. Next we radiolabeled the immunoconjugates with ([⁸⁹Zr]Zr⁴⁺) under a comprehensive set of reaction conditions including different buffers (PBS, chelexed-PBS, TRIS/HCl, HEPES; ± radioprotectants), different reaction volumes (0.1-1 mL), variable amounts of DFO-conjugated Ab (5, 25, 50 μg), and radioactivity (0.2-1.0 mCi; 7.4-37 MBq). We evaluated the effects of these variables on radiochemical yield (RCY), molar activity (A_m)/specific activity (A_s), immunoreactive fraction, and ultimately the in vivo biodistribution profile and tumor targeting ability of the trastuzumab radioimmunoconjugates. We show that increasing the degree of DFO conjugation to trastuzumab increased the RCY (∼90%) and A_m/A_s (∼194 MBq/nmol; 35 mCi/mg) but decreased the HER2-binding affinity (3.5×-4.6×) and the immunoreactive fraction of trastuzumab down to 50-64%, which translated to dramatically inferior in vivo performance of the radioimmunoconjugate. Cell-based immunoreactivity assays and standard binding affinity analyses using surface plasmon resonance (SPR) did not predict the poor in vivo performance of the most extreme T200 conjugate. However, SPR-based concentration free calibration analysis yielded active antibody concentration and was predictive of the in vivo trends. Positron emission tomography (PET) imaging and biodistribution studies in a HER2-positive xenograft model revealed activity concentrations of 38.7 ± 3.8 %ID/g in the tumor and 6.3 ± 4.1 %ID/g in the liver for ([⁸⁹Zr]Zr⁴⁺)-T5 (∼1.4 ± 0.5 DFOs/Ab) at 120 h after injection of the radioimmunoconjugates. On the other hand, ([⁸⁹Zr]Zr⁴⁺)-T200 (10.9 ± 0.7 DFOs/Ab) yielded 16.2 ± 3.2 %ID/g in the tumor versus 27.5 ± 4.1 %ID/g in the liver. Collectively, our findings suggest that synthesizing trastuzumab immunoconjugates bearing 1-3 DFOs per Ab (T5 and T10) combined with radiolabeling performed in low reaction volumes using Chelex treated PBS or HEPEs without a radioprotectant provided radioimmunoconjugates having high A_m/A_s (97 MBq/nmol; 17.5 ± 2.2 mCi/mg), highly preserved immunoreactive fractions (86-93%), and favorable in vivo biodistribution profile with excellent tumor uptake.

Figure 1.. Bioconjugation and radiolabeling of antibodies for ⁸⁹Zr-immuno-PET.

A) Schematic showing the conjugation of a bifunctional chelator (p-SCN-Ph-DFO) to trastuzumab, and synthesis of radiolabeled antibody [⁸⁹Zr]Zr-DFO-trastuzumab for immuno-PET. B) Cartoon illustrating the syntheses of different trastuzumab-DFO-conjugates (T5, T10, T20, T60 and T200) by varying the molar equivalents (5, 10, 20, 60 and 200) of p-SCN-Ph-DFO. C) Analysis of the various trastuzumab-DFO conjugates using polyacrylamide gel electrophoresis (SDS-PAGE) showing changes in the migration of the antibody’s heavy chains (HC) and light chains (LC) attributed to changes (gain) in the molecular weight of these domains upon modification with the various molar equivalents of DFO.

Figure 2.. Comparative SPR kinetic analysis of trastuzumab-DFO-conjugates binding to HER2.

Sensorgrams showing dose-response curves and kinetic profiles for the binding of DFO-conjugated variants including A) unmodified trastuzumab-T0; B) T5; C) T10; D) T20; E) T60; and F) T200. The binding affinity (K_D), and kinetic rate constants – on-rate (k_a) and off-rate (k_d) for each immunoconjugate are mentioned alongside the corresponding representative sensorgram plots; G) Bar graph showing the inverse relationship between an increase in the degree of chelator/Ab ratio and progressively diminishing binding affinity from T0-T200; H) Bar graph showing the maximum impact on binding affinity caused by depreciating on-rates for the trastuzumab-DFO variants with higher degree of chelator conjugation; I) Bar graph representing the comparable dissociation kinetics for all the trastuzumab-DFO variants examined.

Figure 3.. CFCA analysis to determine active antibody concentration in the various trastuzumab-DFO-conjugates.

CFCA plots showing a direct comparison of the binding curves obtained from flowing 20 nM and 2 nM of unmodified trastuzumab (T0) shown in light blue at flow rates of 100 μL/min and 5μL/min versus T5 (orange) in A); T10 (green) in B); T20 (purple) in C); T60 (red) in D); T200 (gray) in E). The active concentration of trastuzumab was calculated from the region of the slope on the CFCA plots demarcated by the dashed vertical lines. F) CFCA trend plot revealing diminishing concentrations of HER2-reactive antibody in trastuzumab-DFO-conjugates with progressively increasing chelator/Ab ratios.

Figure 4.. Summary of radiolabeling studies including

A) the effect of the number of chelates per Ab on radiochemical yields at various masses (5, 25, 50 μg, n = 3 each; ~300 μL chelex treated PBS buffer, reaction performed at 37°C) showing quantitative radiolabeling yields for 50 and 25 μg of immunoconjugate used; B) the effect of buffer volume on A_s at constant mass of Ab (5 μg, n = 3 each), with more concentrated radiolabeling conditions providing higher A_s; C) the effect of number of chelators/Ab on the polarity (partition coefficient between octanol and PBS pH 7.4, n = 5, logD) of ⁸⁹Zr-labeled immunoconjugates, showing decreasing polarity with increasing number of DFOs/Ab (* indicates p-value = ≤ 0.05, ** = ≤ 0.01, *** = ≤ 0.001.); and D) the effect of buffer on A_s at constant volume (~300 μL) and Ab mass (5 μg), showing no clear trend but chelexed-PBS generally conferring optimal results.

Figure 5.. PET images delineating the in vivo profile of ⁸⁹Zr-labeled DFO-trastuzumab variants in a SKOV3 xenograft model.

A) Coronal 2D slices from PET images of [⁸⁹Zr]Zr-DFO-trastuzumab variants (~190 μCi, ~7 MBq, ~40 μg per mouse) in mice bearing subcutaneous SKOV3 xenografts; B) Corresponding maximum intensity projection (MIP) of PET images shown in A. Progressive washout from systemic circulation and a concomitant increase in activity concentration of targeted antibody-based tracer uptake in the tumor (yellow arrows) on left shoulder of mice injected with [⁸⁹Zr]Zr-T5, [⁸⁹Zr]Zr-T10 and [⁸⁹Zr]Zr-T20. PET images of [⁸⁹Zr]Zr-T200 showed dramatically low activity concentration in the tumor (yellow arrow) but heightened liver uptake (white arrow) as early as 24 h p.i. Increasing the DFO chelator conjugation ratio from 5–200 molar equivalents yielded 1–11 chelates/Ab, which appears to impact the in vivo tumor-targeting ability of trastuzumab.

Figure 6.

Comparative analysis showing ex vivo biodistribution data from mice used for PET imaging. Data from all collected organs presented in the supporting information (Supplementary Figure S7, Table SI). Statistical significance was assessed for tumor and liver activity concentrations for [⁸⁹Zr]Zr-T5, [⁸⁹Zr]Zr-T10 and [⁸⁹Zr]Zr-T20 in comparison to [⁸⁹Zr]Zr-T200; p-values * = ≤ 0.05, ** = ≤ 0.01, *** = ≤ 0.001. The bottom panel illustrates misdirected in vivo distribution of trastuzumab radioimmunoconjugates highlighted by the trend for decreased on-target (tumor) and increased off-target (liver) activity concentrations for the radioimmunoconjugates bearing higher number of DFOs/Ab – [⁸⁹Zr]Zr-T20 and [⁸⁹Zr]Zr-T200. The heightened liver uptake may also be related to a trend for decreasing logD (octanol/PBS partition coefficients) for the various trastuzumab-DFO conjugates (blue dashed lines).