Reengineering anthrax toxin protective antigen for improved receptor-specific protein delivery

Abstract

Background: To increase the size of the druggable proteome, it would be highly desirable to devise efficient methods to translocate designed binding proteins to the cytosol, as they could specifically target flat and hydrophobic protein-protein interfaces. If this could be done in a manner dependent on a cell surface receptor, two layers of specificity would be obtained: one for the cell type and the other for the cytosolic target. Bacterial protein toxins have naturally evolved such systems. Anthrax toxin consists of a pore-forming translocation unit (protective antigen (PA)) and a separate protein payload. When engineering PA to ablate binding to its own receptor and instead binding to a receptor of choice, by fusing a designed ankyrin repeat protein (DARPin), uptake in new cell types can be achieved.

Results: Prepore-to-pore conversion of redirected PA already occurs at the cell surface, limiting the amount of PA that can be administered and thus limiting the amount of delivered payload. We hypothesized that the reason is a lack of a stabilizing interaction with wild-type PA receptor. We have now reengineered PA to incorporate the binding domain of the anthrax receptor CMG2, followed by a DARPin, binding to the receptor of choice. This construct is indeed stabilized, undergoes prepore-to-pore conversion only in late endosomes, can be administered to much higher concentrations without showing toxicity, and consequently delivers much higher amounts of payload to the cytosol.

Conclusion: We believe that this reengineered system is an important step forward to addressing efficient cell-specific delivery of proteins to the cytosol.

Keywords: Anthrax toxin; Cytosolic protein delivery; DARPin; Protective antigen.

Fig. 1

Ribbon representation of the structures of PA constructs shown in their activated/furin-cleaved PA₆₃ version. a Previously published, retargeted PA_m-Ac2 [10]. b–d Schematic representation of the prepore-to-pore conversion at the respective pH of furin-cleaved PA_wt (b), PA_m fused to a retargeting DARPin, PA_m-Ac2 (c), and PA_wt fused to the wild-type receptor domain and the retargeting DARPin, PA_wt-sANTXR-Ac2 (d). e Newly designed stabilized PA_wt-sANTXR-Ac2 with PA_wt, the wild-type receptor CMG2 VWA domain, and the retargeting DARPin; PA shown in red, EpCAM-retargeting DARPin Ac2 shown in blue, CMG2 receptor VWA domain (sANTXR) shown in green, and prepore-stabilizing interaction region highlighted in black oval. Protein structures were adapted from PDB ID: 1TZN (PA prepore binding sANTXR), 1ACC (PA), and 4YDW (DARPin)

Fig. 2

Cytotoxicity of different PA variants in comparison with PA_wt-sANTXR-Ac2. a Viability assay of Flp-In 293-EpCAM-BirA with respective concentrations of PA_wt-sANTXR-Ac2, PA_m-sANTXR-Ac2, PA_m-Ac2, PA_wt-Ac2, and PA_wt (n = 3). Error bars indicate SEM. b Competition assay with Ac2 DARPin and 100 nM PA_m-Ac2 with increasing amounts of competitor Ac2-Flag (n = 3). Error bars indicate SEM. c Quantification of propidium iodide (PI)-positive Flp-In 293-EpCAM-BirA cells during time-lapse imaging, cells treated with either PA_wt-sANTXR-Ac2 (black squares) or PA_m-Ac2 (green circles) (n = 6). Error bars indicate SEM. d PI (red) staining for permeable cells and LF_N-eGFP staining (green) for binding to surface PA, comparing PA_wt-sANTXR-Ac2 and PA_m-Ac2 on Flp-In 293-EpCAM-BirA cells

Fig. 3

Effects of PA on different cell lines expressing EpCAM. a EpCAM surface expression data assessed via flow cytometry using an Alexa Fluor 488-labeled anti-EpCAM mouse mAb (n = 3). Error bars reflect SEM. b Confocal imaging of stained Flp-In 293-EpCAM-BirA cells with PA_wt-sANTXR-Ac2 and LF_N-eGFP to assess PA oligomerization. c Viability assays for a set of cell lines with PA_wt-sANTXR-Ac2, PA_m-sANTXR-Ac2, PA_m-Ac2, PA_wt-Ac2, and PA_wt (n = 3). Error bars reflect SEM

Fig. 4

Western blots of the BirA assay showing increased delivery of LF_N-cargo constructs with PA_wt-sANTXR-Ac2 on Flp-In 293-EpCAM-BirA cells. Cytosolically delivered cargo proteins are biotinylated by a cytoplasmically encoded BirA and stained with Streptavidin IRDye 680LT. Total cellular uptake measured via HA-tag on the LF_N-cargo. a, b Increasing concentrations of respective PA constructs incubated with a 5-fold excess of LF_N-NI₁C. Boxes indicate the bands of interest. c, d Quantification of western blot bands from a and b. Black bars, PA_wt-sANTXR-Ac2; red bars, PA_m-sANTXR-Ac2; green bars, PA_m-Ac2. The dotted line represents background signal (i.e., cells only), and the dashed line shows the signal of cargo at 20 nM PA_wt-sANTXR-Ac2. e, f Cytosolic localization (e) and total cellular uptake (f) of three different cargo DARPins delivered with 20 nM (lanes 1–3) and 100 nM (lanes 4–6) of PA_wt-sANTXR-Ac2 or 100 nM (lanes 7–9) for PA_m-Ac2; lanes 10 and 11 represent cells incubated with 100 nM LF_N-cargo without PA. "dest." refers to rationally destabilized versions of NI₂C and NI₃C DARPins [10]. Boxes indicate the bands of interest