A fluorogenic probe for granzyme B enables in-biopsy evaluation and screening of response to anticancer immunotherapies

Abstract

Immunotherapy promotes the attack of cancer cells by the immune system; however, it is difficult to detect early responses before changes in tumor size occur. Here, we report the rational design of a fluorogenic peptide able to detect picomolar concentrations of active granzyme B as a biomarker of immune-mediated anticancer action. Through a series of chemical iterations and molecular dynamics simulations, we synthesize a library of FRET peptides and identify probe H5 with an optimal fit into granzyme B. We demonstrate that probe H5 enables the real-time detection of T cell-mediated anticancer activity in mouse tumors and in tumors from lung cancer patients. Furthermore, we show image-based phenotypic screens, which reveal that the AKT kinase inhibitor AZD5363 shows immune-mediated anticancer activity. The reactivity of probe H5 may enable the monitoring of early responses to anticancer treatments using tissue biopsies.

Fig. 1. The hexapeptide H5 achieves high reactivity and selectivity for GzmB by accessing a unique binding pocket.

a Fluorogenic hexapeptide H5 and fluorescence quantum yields of intact and cleaved probe. b Tetra- and hexapeptide sequences, fluorescence increases and t₅₀ values upon incubation (25 μM) with hGzmB (20 nM). Data as means ± SEM (n = 3). c Time-course fluorescence of H5 (25 μM, green, 510 nm) and Ac-IEPD-AMC (25 μM, blue, 450 nm) after incubation with hGzmB (20 nM) at 37 °C. Probe H5 alone (25 μM, black, 510 nm). Data as means ± SEM (n = 5). d Cleavage rate of probe H5 by hGzmB (20 nM) as a function of substrate concentration. Data as individual replicates and kinetic values determined using the Michaelis-Menten equation (n = 3). e Fluorescence changes of probe H5 (25 µM) after incubation with proteases (20 nM) at 37 °C for 60 min. Data as means ± SEM (n = 3). For active GzmB vs pro-GzmB, p < 0.0001. For active GzmB vs GzmB+ inhibitor, p = 0.0003. For active GzmB vs GzmA, p < 0.0001. f Representative binding mode of IEPD-Dabcyl in T1 (left) and IEPDAL-Dabcyl in H5 (right) from the MD simulations. g Detailed interactions at P1’ and P2’ sites for the probe H5 with overlaid structures from 3 independent runs. P-values from two-tailed t-tests. Source data (b–e) provided as a Source Data file.

Fig. 2. Probe H5 detects GzmB-mediated anticancer activity of CD8+ T cells.

a Schematic procedure for co-culture assays. b Representative microscopy images of CD8+ T cells before/after reinvigoration and staining by anti-GzmB (10 nM, red) and Hoechst 33342 (1 µM, blue) (n = 3). Scale bar: 10 μm. c Flow cytometry (gating: Fig. S7) of E0771 cells in co-culture with active (IL-2, 24 h) or inactive (IL-2, 2 h) CD8+ T cells, or alone with staurosporine (1 μM, 1 h). Legends: viable (gray), H5-stained (green), apoptotic (red). H5 (5 μM), Annexin V-AF647 (10 nM). Data as means ± SEM (n = 3). d Confocal microscopy of mKate-expressing E0771 cancer cells (red) stained with H5 (25 μM, green) in co-culture with active T cells (left), non-active T cells (centre), or active T cells plus Ac-IEPD-CHO (right). Black arrows highlight T cells, yellow arrows highlight H5-stained intracellular GzmB puncta. Quantification of fluorescence intensity by image analysis shown in Fig. S8. Scale bar: 10 μm. e Time-course fluorescence microscopy of OT-I CD8+ T cells stained with Cell Tracker Orange (red) killing OVA-EL4 cancer cells in the presence of H5 (10 μM, green) and Sytox Blue (1 μM, blue) (n = 3). Scale bar: 10 μm. f Flow cytometric analysis of OVA-EL4 cancer cells from experiments in e. Data as means ± SEM (n = 3). P-values from two-tailed t-tests. Source data (c, f) provided as a Source Data file.

Fig. 3. Probe H5 detects T-cell-mediated tumor regression in a mouse model of squamous cell carcinoma.

a Experimental timeline of the CD8+ T-cell-mediated tumor regression model. b Cell populations found in wild-type SCC and SCC FAK (−/−) tumors. c–f SCC and SCC FAK (−/−) cells were injected into FVB immunocompetent mice (1 × 10⁶ cells/mouse) and tumors were harvested on day 14. Flow cytometry (gating: Fig. S15) of wild-type SCC and SCC FAK (−/−) tumors for cell viability with live/dead stain (c); CD8+ T cell infiltrates with anti-CD8-PE (p < 0.0001) (d); percentage of GzmB-positive SCC cancer cells by staining with H5 (5 μM, 30 min) (p = 0.0042) (e); fluorescence intensity of H5 inside SCC cancer cells (525 nm) (p = 0.0108) (f). Data in c–f as means ± SEM (n = 4) (p = 0.0005). g, h SCC FAK (−/−) tumors (g) and wild-type SCC tumors (h) (ex vivo stained with 5 μM compound H5) were analyzed by flow cytometry and presented as pseudo-colored two-dimensional tSNE (t-distributed stochastic neighbor embedding) plots to determine the distribution of the probe in different cell populations (left) and the fluorescence intensity of H5 staining (right) (n = 4). P-values from two-tailed t-tests using Welch correction. Source data (c–f) provided as a Source Data file.

Fig. 4. Probe H5 detects immunomodulatory action in phenotypic screens and T cell cytotoxic activity in human tissue from lung cancer patients.

a Experimental protocol of the phenotypic screen. b Fluorescence intensity of probe H5 in co-cultures of E0771 cells and IL-2-activated CD8+ T cells after incubation with small molecules (C1-C44). Wells received 100 U mL⁻¹ IL-2 (low IL-2) for T cell viability. High IL-2 (250 U mL⁻¹) used as a positive control for invigorated CD8+ T cells and rapamycin (0.3 μM) used as a negative control. Probe H5 (20 μM) and Hoechst 33342 (1 µM) were incubated for 1 h and fluorescence images were acquired with an ImageXpress^TM XLS. Data as means ± SEM (n = 8). Chemical structures of drugs showing H5 fluorescence signals above those with 250 U mL⁻¹ IL-2. c Representative H&E microscope images of non-tumor (left) and lung adenocarcinoma (right) paired samples. Scale bar: 50 µm. d Cytometry analysis (gating: Fig. S17) showing the percentage of EpCAM+ H5+ epithelial cells found in paired tissues (non-tumor vs tumor) from lung cancer patients after incubation with probe H5 (5 μM). Data as means ± SD (n = 5). P-value from two-tailed t-tests. Source data (b, d) provided as a Source Data file.