Intravital Imaging to Monitor Therapeutic Response in Moving Hypoxic Regions Resistant to PI3K Pathway Targeting in Pancreatic Cancer

Abstract

Application of advanced intravital imaging facilitates dynamic monitoring of pathway activity upon therapeutic inhibition. Here, we assess resistance to therapeutic inhibition of the PI3K pathway within the hypoxic microenvironment of pancreatic ductal adenocarcinoma (PDAC) and identify a phenomenon whereby pronounced hypoxia-induced resistance is observed for three clinically relevant inhibitors. To address this clinical problem, we have mapped tumor hypoxia by both immunofluorescence and phosphorescence lifetime imaging of oxygen-sensitive nanoparticles and demonstrate that these hypoxic regions move transiently around the tumor. To overlay this microenvironmental information with drug response, we applied a FRET biosensor for Akt activity, which is a key effector of the PI3K pathway. Performing dual intravital imaging of drug response in different tumor compartments, we demonstrate an improved drug response to a combination therapy using the dual mTORC1/2 inhibitor AZD2014 with the hypoxia-activated pro-drug TH-302.

Keywords: AZD2014; Akt; FRET; PI3K pathway; PLIM; hypoxia; intravital imaging; nanoparticles; pancreatic cancer; pro-drug.

Graphical abstract

Figure 1

The Presence of Hypoxia and the Associated Molecular, Phenotypic, and Resistance Effects in the KP^flC and KPC Mouse Models of PDAC (A) Immunofluorescence staining of the GEM KP^flC and KPC PDAC mouse models for Akt(Ser473) (red), DAPI (cyan), and pimonidazole (green). Scale bars, 50 μm. (B) Quantification of Akt(Ser473) grey value/cell in pimonidazole negative (normoxic) and positive (hypoxic) regions (n = 5 tumors/mouse model). Mean ± SEM. p values are from a one-sample t test. (C) Propidium iodide staining of cell cycle phase distribution in the KP^flC and KPC primary PDAC cell lines. Mean ± SEM. p values were calculated using a two-way ANOVA with a Tukey correction for multiple comparisons. (D) Representative western blots of the KP^flC and KPC primary PDAC cell lines, incubated for 48 hr in normoxia or hypoxia (5%, 1%, or 0.1% oxygen; n = 5). (E) A simplified schematic of the PI3K pathway, indicating the targets of the PI3K pathway inhibitors used in this study. (F and G) IC₅₀ curves demonstrating the response of KP^flC (F) and KPC (G) primary PDAC cell lines to the PI3K pathway inhibitors rapamycin, NVP-BEZ235, and AZD2014, in both normoxic (black lines) and hypoxic (0.1% oxygen, red lines) conditions (n = 3). An extra sum-of-squares F test was performed between the best-fit parameters of each curve. ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. See also Figure S1.

Figure 2

Organotypic Invasion Assay of KPC Cells Treated with PI3K Pathway Inhibitors and/or TH-302 (A) A schematic representation of the organotypic matrix assay and a representative area of invading KPC cells stained with pimonidazole (a marker of hypoxia). (B) Representative images of invading KPC cells stained with the epithelial cell marker pan-cytokeratin or the proliferative marker Ki67. Scale bars, 100 μm; insets, 25 μm. (C) Normalized Ki67 staining from the same experiments scored for cells on top of the organotypic matrix (normoxia) and cells invading into the matrix (hypoxia). Mean ± SEM. (D) Normalized invasive index of KPC cells treated with NVP-BEZ235, AZD2014, or rapamycin (n = 3). Mean ± SEM. (E and F) Representative images of invading KPC cells stained with the hypoxia marker pimonidazole (E) or the proliferative marker Ki67 (F). Scale bars, 50 μm; insets, 10 μm. Mean ± SEM. (G) Quantification of Ki67-stained KPC cells invading into organotypic matrices (n=4). (H and I) Assessment of Ki67-stained KPC cells either on top of (H, normoxia) or invading into (I, hypoxia) organotypic matrices (n = 4). A one-sample t test was performed on normalized data in all panels. ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. See also Figure S2.

Figure 3

Assessment of KPC Subcutaneous Xenografts Treated with TH-302 (50 mg/kg) and AZD2014 (2.5 mg/kg) (A and B) Tumor volume measurements (A) and average linear growth rate (B) of tumors over 7 days from vehicle/saline (n = 10), vehicle/TH-302 (n = 9), AZD2014/saline (n = 10), and AZD2014/TH-302 (n = 10). (C and D) Immunohistochemistry (IHC) staining of drug response in tumors assessed for the proliferative marker Ki67 (n = 5 tumors/treatment) (C) and DNA damage response (γH2AX, n = 5 tumors/treatment) (D). Scale bars, 100 μm; insets, 10 μm. Mean ± SEM. (E and F) Staining for the necrotic (using H&E) (E) and hypoxic (pimonidazole IHC) (F) tumor fractions from vehicle/saline (n = 10), vehicle/TH-302 (n = 9), AZD2014/saline (n = 10), and AZD2014/TH-302 (n = 10) treatments. Scale bars, 1 mm; insets, 100 μm. Mean ± SEM. (G) HIF1α (n = 5 tumors/treatment) IHC staining of tumors. Scale bars, 100 μm; insets, 10 μm. Mean ± SEM. (H) Carbonic anhydrase IX (CAIX) IHC staining of tumors from vehicle/saline (n = 10), vehicle/TH-302 (n = 9), AZD2014/saline (n = 10), and AZD2014/TH-302 (n = 10) treatments. Scale bars, 100 μm; insets, 10 μm. Mean ± SEM. p values are from a Student two-tailed parametric t test in all panels. ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. See also Figure S3.

Figure 4

Characterization of the Cell Cycle Response of KPC Subcutaneous Xenografts Treated with TH-302 (50 mg/kg) and AZD2014 (2.5 mg/kg) (A) IHC staining of drug response in tumors assessed for the mitosis marker HistoneH3(Ser10) (n = 5 tumors/treatment). Scale bars, 100 μm; insets, 10 μm. Mean ± SEM. p values are from a Student two-tailed parametric t test. (B–E) Phenotypic quantification of four distinguishable stages of mitosis, namely late G₂ (B), prophase (C), metaphase (D), and anaphase (E), where HistoneH3(Ser10) presents a unique phenotype. Scale bars, 10 μm. Mean ± SEM. p values are from a Student two-tailed parametric t test. (F) Propidium iodide staining of cell cycle phase distribution in KPC primary PDAC cells after treatment with AZD2014 (500 nM) and/or TH-302 (1 μM) at 0.1% oxygen. Mean ± SEM. p values were calculated using a two-way ANOVA with a Tukey correction for multiple comparisons. (G) HistoneH3(Ser10)/propidium iodide dual-parameter FACS analysis of KPC primary PDAC cells after treatment with AZD2014 (500 nM) and/or TH-302 (1 μM) at 0.1% oxygen. Mean ± SEM. p values are from a Student two-tailed parametric t test. (H) Schematic representation of the proposed combination effect of AZD2014 on cells prior to mitotic entry by TH-302. ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001.

Figure 5

Tracking of Tumor Hypoxia with EF5 and Pimonidazole (A and B) Immunofluorescence of (A) KPC GEM tumors and (B) KPC xenograft tumors for EF5 (red) and pimonidazole (green), chemical indicators of tumor hypoxia, after either (i) co-injection or (ii) 24-hr delayed treatments. Scale bars, 100 μm. (iii) Quantification of overlapping (yellow) regions of staining between EF5 and pimonidazole in KPC GEM tumors (n = 3 mice/group) and KPC xenograft tumors (n = 4 mice/group). Mean ± SEM. p values are from a Student two-tailed parametric t test in all panels. See also Figure S4.

Figure 6

Dual FLIM/PLIM Imaging of KPC Cells Stably Expressing the Eevee-Akt-mT2 Intramolecular FRET Biosensor, Treated with Oxygen-Sensitive Nanoparticles (A) A schematic illustration of the methodology applied to modify the multiphoton FLIM detection system to detect oxygen-sensitive PLIM through modulation of the EOM. (B) A schematic demonstrating the use of a glass coverslip to induce a hypoxic response in vitro. (C) qRT-PCR analysis of relative mRNA expression of hypoxia response genes upregulated in cells incubated for 2 hr in hypoxia (5%, 1%, and 0.1% oxygen) or under a coverslip, compared to normoxia, normalized to Rplp0 (n = 5). Mean ± SEM. (D) Dual FLIM/PLIM imaging of Akt activity in KPC cells treated with MitoImage NanO2 and incubated with a glass coverslip for 1 or 2 hr (n = 4). Scale bars, 50 μm; insets, 10 μm. Mean ± SEM. p values are from a Student two-tailed parametric t test in all panels. ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. See also Figures S5 and S6, and Tables S1 and S2.

Figure 7

Dual FLIM/PLIM Intravital Imaging of Drug Response and Tumor Oxygen Content (A) A schematic representation of the intravital imaging setup for the dual FLIM/PLIM imaging. Xenografts of KPC cells stably expressing the Eevee-Akt-mT2 intramolecular FRET biosensor were allowed to reach 350 mm³, before treatment with oxygen-sensitive nanoparticles (NanO2) and dual FLIM/PLIM imaging. (B) FLIM-FRET analysis of vehicle/saline (n = 5), vehicle/TH-302 (n = 4), AZD2014/saline (n = 7), or AZD2014/TH-302 (n = 4)-treated mice. Mean ± SEM. p values are from a Student two-tailed parametric t test. (C–F) Representative FLIM and PLIM maps are provided for cells with a short PLIM value (high oxygen content, normoxic) and a long PLIM value (low oxygen content, hypoxic) for vehicle/saline (C), vehicle/TH-302 (D), AZD2014/saline (E), and AZD2014/TH-302 (F) treatments. Scale bars, 25 μm. (G–J) These same cells are then highlighted on their representative Deming regression curves, where blue and red points highlight cells with short or long PLIM values, respectively. The 95% confidence intervals emphasize whether the slope of the Deming regression is significantly non-zero for vehicle/saline (G), vehicle/TH-302 (H), AZD2014/saline (I), and AZD2014/TH-302 (J) treatments. (K) Average Deming slopes of mice from each treatment group, assessed for departure from zero with a one-sample t test. Mean ± SEM. ^∗p < 0.05, ^∗∗p < 0.01 and ^∗∗∗p < 0.001.