Optical metabolic imaging identifies glycolytic levels, subtypes, and early-treatment response in breast cancer

Abstract

Abnormal cellular metabolism is a hallmark of cancer, yet there is an absence of quantitative methods to dynamically image this powerful cellular function. Optical metabolic imaging (OMI) is a noninvasive, high-resolution, quantitative tool for monitoring cellular metabolism. OMI probes the fluorescence intensities and lifetimes of the autofluorescent metabolic coenzymes reduced NADH and flavin adenine dinucleotide. We confirm that OMI correlates with cellular glycolytic levels across a panel of human breast cell lines using standard assays of cellular rates of glucose uptake and lactate secretion (P < 0.05, r = 0.89). In addition, OMI resolves differences in the basal metabolic activity of untransformed from malignant breast cells (P < 0.05) and between breast cancer subtypes (P < 0.05), defined by estrogen receptor and/or HER2 expression or absence. In vivo OMI is sensitive to metabolic changes induced by inhibition of HER2 with the antibody trastuzumab (herceptin) in HER2-overexpressing human breast cancer xenografts in mice. This response was confirmed with tumor growth curves and stains for Ki67 and cleaved caspase-3. OMI resolved trastuzumab-induced changes in cellular metabolism in vivo as early as 48 hours posttreatment (P < 0.05), whereas fluorodeoxyglucose-positron emission tomography did not resolve any changes with trastuzumab up to 12 days posttreatment (P > 0.05). In addition, OMI resolved cellular subpopulations of differing response in vivo that are critical for investigating drug resistance mechanisms. Importantly, OMI endpoints remained unchanged with trastuzumab treatment in trastuzumab-resistant xenografts (P > 0.05). OMI has significant implications for rapid cellular-level assessment of metabolic response to molecular expression and drug action, which would greatly accelerate drug development studies.

Figure 1

Graphical representation (a) of the relationship between HER2, NADH and FAD. HER2 activation drives an increase in glycolysis, which produces NADH. The pyruvate generated in glycolysis can enter the mitochondria as a reactant in oxidative phosphorylation. Oxidative phosphorylation consumes NADH and produces FAD. A net gain of NADH relative to FAD is observed with HER2 activation due to a relative increase in glycolysis. Inhibition of HER2 by trastuzumab reduces cellular glycolysis rates resulting in a decrease of cellular NADH relative to FAD. NADH and FAD are highlighted as the fluorescent molecules in this diagram, and molecules in bold indicate the net direction of the reaction. Optical redox ratio (b), NADH τ_m (c), and FAD τ_m (d) of MCF10A (non-malignant) cells before (n=9) and after (n=6) treatment with 4 mM NaCN. τ_m is the mean lifetime (τ_m = τ₁*α₁ + τ₂*α₂). ** P <0.001, *P <0.05.

Figure 2

Representative images of the optical redox ratio (NADH/FAD; first row), NADH τ_m (second row), and FAD τ_m (third row) of MCF10A (non-malignant) and malignant breast cells. Scale bar represents 30 μm. The redox ratio is normalized to the mean daily MCF10A mean redox ratio as a daily fluorescence standard. τ_m is the mean lifetime (τ_m = τ₁*α₁ + τ₂*α₂). The ER and HER2 status of each cell line is provided under its name.

Figure 3

(a) The optical redox ratio (mean +/− SE) is increased in malignant cells with ER overexpression (MCF7, green), and is further increased in cells with HER2 overexpression (MDA-MB-361, BT474, and SKBr3; blue). (b) The NADH τ_m (mean +/− SE) is decreased in triple negative cells (MDA-MB-231, red) relative to non-malignant, ER+, and HER2+ cells. The NADH τ_m is increased in HER2+ cells relative to non-malignant cells. (c) FAD τ_m (mean +/− SE) is increased in malignant cells relative to non-malignant cells. (d) A scatterplot of NADH τ_m versus redox ratio/% mitotic cells provides visual separation of the different molecular sub-types of breast cancer [non-malignant (NM), ER+/HER2−, HER2+, and triple negative (TNBC)]. (h) Correlation between redox ratio/% mitotic cells and glycolytic index. Unless indicated with a line, stars (*) indicate statistically significant differences with the MCF10A (non-malignant) cells and bullets (*) indicate statistically significant differences with the HER2+ cells grouped together. ** P <0.001, ** P < 0.001. n = 18 for all malignant cell lines and n = 58 for MCF10A.

Figure 4

(a) The redox ratio (mean +/− SE) decreases with trastuzumab-treatment in responsive (BT474) and partially responsive (MDA-MB-361) cells but remains unchanged in cells with acquired resistance to trastuzumab (HR6). (b) NADH τ_m (mean +/− SE; τ_m = τ₁*α₁ + τ₂*α₂) is shorter in trastuzumab-treated responsive cells (BT474) and cells with acquired resistance (HR6), but unchanged in cells with poor response (MDA-MB-361). (c) FAD τ_m (mean +/− SE) is shorter in trastuzumab-treated BT474 but unchanged MDA-MB-361 and HR6 cells. * indicates significance between control and treated, unless otherwise indicated. (d) Distribution density modeling (black line represents fit of untreated cell histogram, gray line represents fit of trastuzumab treated cells) of the redox ratio on a cell-by-cell basis reveals two distinct populations of MDA-MB-361 cells for the redox ratio, with 70% and 30% of cells in the majority and minority populations, respectively, for both control and treated cells. There is a significant decrease (P<0.05) in the mean of the minority population of treated cells (but not in the majority population), suggesting trastuzumab response in a sub-population of cells. ** P <0.001, * P <0.05. n = 18 images for control and treated for all cell lines.

Figure 5

(a) BT474 tumors treated with trastuzumab (10 mg/kg, 2 × weekly) decrease in size (mean +/− SE) compared to control IgG treated tumors. Trastuzumab vs. control at each time point. (b) Ki67 staining demonstrates reduced proliferation in trastuzumab-treated tumors. (c) Cleaved caspase 3 staining demonstrates increased apoptosis in trastuzumab-treated tumors at day 5. (d) Representative FDG-PET image (T=tumor). (e) FDG uptake increases in control tumors at day 12 compared to control tumors at day 2. No significant difference in FDG uptake between trastuzumab-treated and control tumors is observed. n = 10. (f) Representative OMI images. Scale bar is 50 μm. (g) Decreased redox ratio, (h) NADH τ_m, and (i) FAD τ_m are observed in trastuzumab-treated tumors at 2 days post trastuzumab-treatment. * P<0.05, ** P<0.01; *** P < 0.001; n = 6 tumors.

Figure 6

(a) HR6 tumor size (mean +/− SE) treated with trastuzumab (10 mg/kg, 2 × weekly) compared to control IgG treated tumors. (b) Ki67 staining of HR6 control IgG and trastuzumab-treated tumors. (c) Cleaved Caspase 3 staining of HR6 control IgG and trastuzumab-treated tumors. (d) Representative FDG-PET image (T=tumor). (e) FDG uptake increases in control tumors at day 12 compared to control tumors at day 2, * P<0.05. No significant difference in FDG uptake between trastuzumab-treated and control tumors is observed at any time. n = 10. (f) Representative OMI images, scale bar is 50 μm.. (g) Redox ratio, (h) NADH τ_m, and (i) FAD τ_m of control and trastuzumab-treated tumors. * P<0.05; n = 6 tumors.

Figure 7

(a) Representative images of the optical redox ratio (first row), NADH τ_m (second row), and FAD τ_m (third row) of untreated control and trastuzumab-treated (10 μg/kg; 48 hours) MDA-MB-361 mouse xenograft tumors. Fibrillar structures are due to collagen second harmonic generation (SHG). Scale bar is 50 μm. (b-d) Mean +/− SE on a per-image basis, and (e-g) distribution density modeling on a cell-by-cell basis in control and trastuzumab-treated tumors. (b) The optical redox ratio is reduced in trastuzumab-treated tumors compared to the control tumors. (c) NADH τ_m is reduced in the trastuzumab-treated tumors compared to the control tumors. (d) FAD τ_m is reduced in trastuzumab-treated tumors compared to the control tumors. (e-g) Distribution density modeling reveals two distinct populations of trastuzumab-treated cells for the redox ratio (e, maroon line) and FAD τ_m (g, maroon line) measurements, indicating heterogeneous cellular metabolic response to treatment. All control populations (e-g; blue lines) and the trastuzumab-treated NADH τ_m population (f, maroon line) are unimodal. (h) The increased coefficient of variation in the redox ratio and FAD τ_m of the trastuzumab treated xenografts indicates increased intra-cellular variation in metabolism. * P <0.05