Oxidizing potential of endosomes and lysosomes limits intracellular cleavage of disulfide-based antibody-drug conjugates

Abstract

Antibody-drug conjugate therapy entails targeted killing of cancer cells with cytotoxic compounds covalently linked to tumor-specific antibodies and shows promise in the treatment of several human cancers. Current antibody-drug conjugate designs that incorporate a disulfide linker between the antibody and cytotoxic drug are inspired by indirect evidence suggesting that the redox potential within the endosomal system is reducing. It is presumed that antigen-dependent endocytosis leads to disulfide linker reduction and intracellular release of free drug, but direct demonstration of such a mechanism is lacking. To determine whether the disulfide N-succinimidyl 4-(2-pyridyldithio)pentanoate (SPP) linker would be reduced during endocytic recycling of the anti-HER2 antibody trastuzumab (Herceptin, Genentech), we synthesized a trastuzumab-SPP-Rhodamine red conjugate and developed a linker cleavage assay by using the self-quenching property of this fluorophore. In breast carcinoma SKBr3 cells, no SPP linker cleavage was observed, as detected by fluorescence dequenching upon internalization. By contrast, the conjugate did display fluorescence dequenching when diverted to the lysosomal pathway by geldanamycin, an effect partly due to proteolytic degradation rather than disulfide reduction. To understand why linker reduction was inefficient, we measured redox potentials of endocytic compartments by expressing a redox-sensitive variant of GFP fused to various endocytic proteins. Unexpectedly, we found that recycling endosomes, late endosomes, and lysosomes are not reducing, but oxidizing and comparable with conditions in the endoplasmic reticulum. These results suggest that intracellular reduction is unlikely to account for the potency of disulfide-linked antibody-drug conjugates.

Fig. 1.

Fluorescence dequenching of a trastuzumab-SPP-RR conjugate upon SPP linker cleavage in vitro.(A) Crowding of RR moieties, as in the oxidized disulfide dimer form of RR thiol (diamonds) or trastuzumab-SPP-RR conjugate with 4.5 RR/antibody (squares), attenuates fluorescence intensity by self quenching. Relief of the crowding by sodium 2-mercaptoethanesulfonate reduction in vitro increases fluorescence intensity ≈2-fold. Less-crowded trastuzumab-SPP-RR conjugates with only 1.5 (triangles) or 0.5 (circles) fluorophores per antibody show no such fluorescence increase. Fluorescence intensity of each RR-containing construct was measured by spectrofluorimetry after the indicated incubation time at 22°C with 10 mM sodium 2-mercaptoethanesulfonate and results normalized to 100% intensity at time = 0. (B) Endoproteolytic degradation by incubation with 0.1 mg/ml subtilisin for the indicated times also relieves fluorescence quenching of the 4.5 RR/ab trastuzumab-SPP-RR conjugate (squares). Only minimal fluorescence dequenching is observed upon endoproteolytic degradation of an Alexa-488-conjugated trastuzumab (no SPP linker present).

Fig. 2.

Fluorescence dequenching of trastuzumab-SPP-RR^4.5 in live SKBr3 cells requires endocytosis and lysosomal routing. SKBr3 cells with surface-bound trastuzumab-SPP-RR^4.5 were treated and incubated at 37°C for the indicated times and fluorescence intensity measured by flow cytometry as described in Methods. Despite appreciable exposure of the surface pool to the endocytic recycling pathway in control samples, very little dequenching is observed (circles). Maximal dequenching occurs when lysosomal routing is induced with GA (diamonds) in a manner that is fully sensitive to clathrin disruption (crosses) and partially sensitive to PIs (triangles) or deacidification by chloroquine (squares). MFI, mean fluorescence intensity.

Fig. 3.

Red fluorescence dequenching of a dual labeled ⁴⁸⁸trastuzumab-SPP-RR^4.5 conjugate upon GA-induced routing to the lysosomal pathway. The green fluorophore Alexa-488 was conjugated to trastuzumab-SPP-RR^4.5 to produce a dual-labeled ⁴⁸⁸trastuzumab-SPP-RR^4.5 conjugate as described in Methods. SKBr3 cells with surface-bound ⁴⁸⁸trastuzumab-SPP-RR^4.5 were incubated with (D–F) or without (A–C) 3 μM GA for 3 h and then imaged by fluorescence microscopy as described in Methods. Red fluorescence dequenching upon GA-induced lysosomal routing can be recognized as a red shift in the merged dual color image (compare C with F). (Scale bars: 20 μm.)

Fig. 4.

roGFP1-tagged compartment proteins localize correctly in stably transfected PC3 cells. PC3 cells stably transfected with various roGFP1-tagged compartment proteins (A, D, G, J, and M, green channel) were fixed with 3% paraformaldehyde, permeabilized with saponin, and immunolabeled with antibodies against markers of the intended target compartments, detected with Cy3-labeled secondary antibodies (B, E, H, K, and N, red channel), and imaged by deconvolution microscopy. Colocalization is shown by the yellow color in the merged images (C, F, I, L, and O). Mito-ss-roGFP1 (A) fluoresces in the mitochondrial matrix, clearly visible inside the rings demarcated by cytochrome c in the intermembrane space (B); the boxed region in C is magnified 3-fold in the bottom left corner to show this more clearly. roGFP1-Cnx (D) completely colocalized with the ER enzyme protein disulfide isomerase (E); transferrin receptor-roGFP1 (G) exhibited the same staining pattern as Alexa-555-transferrin (H) after 1 h uptake at 37°C; the internal membrane lysosomal roGFP1-CD63 (J) and the limiting membrane lysosomal roGFP1-Lamp2a (M) both overlapped well with the late endosomal and lysosomal marker Lamp1 (K and N). (Scale bars: 30 μm in A, 3.3 μm in C Inset.)

Fig. 5.

roGFP1 fluorescence intensity measurements reveal that the endocytic pathway is oxidizing. (A) PC3 cells stably transfected with mito-ss-roGFP1 (Mito), roGFP1-calnexin (Cnx), transferrin receptor-roGFP1 (TrfR), roGFP1-Lamp2a (Lamp), and roGFP1-CD63 (CD63) were imaged at 380- and 490-nm excitation, and the fluorescence emission intensities (in the 528-nm channel) of the selected compartments were measured and plotted as 380/490-nm ratios (black bars). This process was repeated after oxidizing (a different well of) cells for at least 5 min with a membrane-permeable oxidant (100 μM aldithriol or 1 mM H₂O₂; gray bars) and after fully reducing the cells with 10 mM DTT (white bars). At least 20 cells were measured for each condition in up to four independent experiments, and the error bars represent the standard deviation of the mean of all of the samples together for each condition. (B) roGFP-CD63-transfected PC3 cells were either imaged as in A (Control), preincubated for 16 h with PIs (5 μM pepstatin A + 10 μg/ml leupeptin; +PIs) or for 16 h with 10 μg/ml cycloheximide (+Chx), or incubated for 24 h in a 3% oxygen incubator then imaged in the presence of 1.5 units/ml of the oxygen-depleting enzyme Oxyrase (low O₂). Results were plotted as in A before (black bars) and after equilibrating with oxidant (gray bars) or reductant (white bars).