Tricyclic cell-penetrating peptides for efficient delivery of functional antibodies into cancer cells

Abstract

The intracellular environment hosts a large number of cancer- and other disease-relevant human proteins. Targeting these with internalized antibodies would allow therapeutic modulation of hitherto undruggable pathways, such as those mediated by protein-protein interactions. However, one of the major obstacles in intracellular targeting is the entrapment of biomacromolecules in the endosome. Here we report an approach to delivering antibodies and antibody fragments into the cytosol and nucleus of cells using trimeric cell-penetrating peptides (CPPs). Four trimers, based on linear and cyclic sequences of the archetypal CPP Tat, are significantly more potent than monomers and can be tuned to function by direct interaction with the plasma membrane or escape from vesicle-like bodies. These studies identify a tricyclic Tat construct that enables intracellular delivery of functional immunoglobulin-G antibodies and Fab fragments that bind intracellular targets in the cytosol and nuclei of live cells at effective concentrations as low as 1 μM.

Graphical abstract

Extended Data Fig. 1. Membrane porosity following treatment with Tat-trimer.

(a, b) Addition of 40 μM propidium iodide (PI) 20 min after addition of 1 μM trimer; image at 30 min after start of experiment. Cells treated with tri-Tat A (a) co-stain with PI; cells treated with tri-cTat B (b) are PI negative. (c) Average fluorescence intensity of PI per cell, 45 min after the start of the experiment (n=25). Cells treated with tri-Tat A show significantly higher PI uptake, indicative of pore formation. (d) Cells treated with tri-Tat A (solid line) or tri-cTat B (dotted line) for 60 min and metabolic activity as an indicator of cell viability assessed using MTT assay after 1 h, 2 h, 4 h, 3 days (n=3 biologically independent experiments). Data presented as mean ± standard deviation. Scale bar: 20 μm.

Fig. 1. Synthesis of Trimer Tat constructs.

Synthesis of tri-Tat A and tri-cTat A (a), tri-Tat B and tri-cTat B (f). In silico generated ball and stick models of linear Tat trimers suggest that they adopt coiled helices which arrange in a more compact conformation in tri-Tat A. Cyclic peptides adopt tighter helices, due to the C-terminus being forced towards the centre of the molecule. In all four trimers, peptides align in an off-parallel fashion, with arginine side chains pointing outwards. Hydrogens removed for clarity, Carbon – grey, Oxygen – red, Nitrogen – blue, Sulphur – orange; geometry of trimer conjugates optimized using a Dreiding-like forcefield; the peptide backbones highlighted from N-terminus (blue) to C-terminus (red); the central carbon of the tetrakis core highlighted in yellow. Fluorescence excitation and emission spectra of trimer (black line) compared to AF488 (dotted line); y axis - normalized fluorescence intensity (A.U.); x axis – wavelength (nm).

Fig. 2. Live cell confocal microscopy of linear and cyclic Tat trimers in HeLa and CHO cells.

(a, f) 1 μM mono-Tat (a) or mono-cTat (f) are not taken up into HeLa or CHO cells. (b, g, h) 1 μM linear tri-Tat A (b) cyclic tri-cTat A (g) and tri-cTat B (h) are taken up into the cytosol and nucleoli of HeLa and CHO cells. (c) 1 μM linear tri-Tat B is only taken up into endosomes of HeLa and CHO cells. (d, i) Quantification of average fluorescence intensity per cell in cells treated with linear (d) or cyclic (i) constructs. (e, j) Quantification of the percentage of transduced cells (scored as positive when showing homogenous cytoplasmic and nucleolar fluorescence) in cells treated with linear (e) or cyclic (i) constructs. Data presented as mean ± standard deviation; n-numbers identical for treatment groups analysed in graphs d/e and i/j. BF = Brightfield image. Scale bar: 20 μm.

Fig. 3. Continuous live cell confocal microscopy of trimers in HeLa cells.

(a) Time course uptake of 1 μM tri-Tat A; and (b) 1 μM tri-cTat B. (c, d) Cross-sectional profile plot through a representative cell treated with 1 μM tri-Tat A (c) and 1 μM tri-cTat B (d); surface plot (middle) composed of cross-sectional profiles from 1 min to 15 min (y-axis: fluorescence intensity (AU); x-axis: distance along the cross section (mm); z-axis: time (min); colour added for clarity to denote y-axis values – dark blue 0-199 AU, orange – 200-399 AU, grey – 400-599 AU, yellow – 600-799 AU, light blue >800 AU). Tri-Tat A fluorescence moves from the plasma membrane inwards (c), while tri-cTat B fluorescence moves from a focal point in the cell outwards (d). (e-h) Time course analysis of cross-sectional profiles (1 – 15 min) through n=5 cells examined over 3 independent experiments treated with tri-Tat A (solid lines) and tri-cTat B (dotted lines), where regions of interest (ROIs) corresponding to the membrane (e), cytosol (f), nucleus (g), and nucleoli (h) are defined and average fluorescence intensity reported. Tri-cTat B shows significantly different membrane association and uptake kinetics into cellular compartments compared to tri-Tat A. Data presented as mean ± standard deviation. Scale bar: 20 μm.

Fig. 4. Co-delivery of antibodies and antibody fragments in live HeLa cells using tri-cTat B.

(a, b) Post-wash live cell confocal microscopy images of HeLa cells treated with 500 nM mouse Fab fragment AF647 conjugate (Fab-AF647) and 1 μM tri-cTat B for 30 min (a) or 166 nM mouse IgG AF647 conjugate (IgG-AF647) and 1 μM tri-cTat B for 45 min (b) show homogenous distribution of Fab in cytosol and nucleus of cells with green nucleoli staining typical of tri-cTat B delivery; (c) Continuous live-cell confocal microscopy of trimers in HeLa cells treated with 500 nM Fab-AF647 (red) and 1 μM tri-cTat B (green); (d) quantification of the percentage of cells transduced with cargo (IgG-AF647 or Fab-AF647 - scored as positive when showing homogenous cytoplasmic and nucleolar fluorescence). Data presented as mean ± standard deviation. Scale bar: 20 μm.

Fig. 5. Co-delivery of functional antibodies and antibody fragments in live HeLa cells.

HeLa cells were transfected with actin-RFP (a, b) or histone-RFP (c). 90 min post-wash live cell confocal microscopy images of cells treated with (a) 500 nM anti-β-actin mouse Fab-AF647 conjugate (β-actin-Fab-AF647) and 1 μM tri-cTat B for 30 min show co-localization of Fab (red) with actin stress fibres (orange); (b) 166 nM anti-β-actin mouse IgG_2b AF647 conjugate (β-actin-IgG-AF647) and 1 μM tri-cTat B for 30 min show co-localization of antibody with actin stress fibres; (c) 166 nM anti-RFP mouse IgG₁ AF647 conjugate (RFP-IgG-AF647) and 1 μM tri-cTat B for 30 min show co-localization of antibody with RFP fused to histone in the nucleus. G = green channel (tri-cTat B); O = orange channel (RFP fusion protein); R = red channel (AF647); Co-localization panel: co-localized pixels are shown as a mask of yellow pixels of constant intensity and all results shown present a significant correlation and are co-localized; BF = brightfield; scale bar: 20 μm.

Fig. 6. Proximity ligation assays (PLA).

PLA were performed in HeLa cells using an H2B-AF488 antibody and a histone H2A.Z antibody. H2B-AF488 (166 nM) was co-delivered into cells using cTat₃-alkyne (2 μM, 1h incubation at 37 °C, 5% CO₂). Negative controls consisted in the co-delivery of a mouse non-specific (NS) antibody and an antibody against β-actin (166 nM IgG, 2 μM cTat₃-alkyne, 1h incubation at 37 °C, 5% CO₂) and in the incubation of non-specific, β-actin, H2B-AF488 and H2A.Z, antibodies after fixation and permeabilization of the cells and. a) Confocal microscopy images showing Hoechst 33342 stained nuclei (blue), PLA signals (red) and an overlay of the fluorescent channels. b) The number of PLA signals was quantified in the nuclei using CellProfiler and represented using Origin in at least 50 cells (IgG incubation controls: No abs, n=57; NS, n=55; β-actin, n= 76; H2B, n=55; H2A.Z, n=100. IgG codelivery controls: NS, n= 87; β-actin, n=90. PLA delivery, n=98) over three independent experiments. Data presented as mean ± SD. A. One way analysis of variance (ANOVA) with Tukey’s test correction were employed in statistical analysis. ****p<0.0001. Scale bars: 20 μm.