An in vivo imaging method to concurrently characterize three metabolic pathways in murine models of breast cancer

Introduction: The greatest clinical challenge posed by breast cancer is treating recurrent and metastatic disease. Emerging evidence points to cancers’ ability to reprogram metabolism to resist treatment and recur. Cancers characteristically take up glucose to fuel aerobic glycolysis as a major source of energy and material to support proliferative biosynthesis. However, various metabolic pathways have been implicated in supporting recurrent disease, including fatty acid beta oxidation and oxidation of glucose through mitochondrial metabolism. Notably, our group and others have shown that aggressive breast cancers can rely on both glycolysis and oxidative metabolism giving them tremendous metabolic flexibility. Thus, characterizing multiple metabolic endpoints and capturing metabolic heterogeneity across in vivo disease models is of upmost importance for identifying metabolic vulnerabilities and targeting treatment. Our group has validated three fluorescent molecules to track metabolic pathways in vivo. TMRE, a fluorescent cation attracted to the proton gradient in active mitochondria, and fluorescent glucose analog 2-NBDG have been used in combination to report on glycolysis and oxidative phosphorylation. Bodipy FL C16, a fluorescent palmitate analog, has been developed to report on the uptake of fats. All three have been imaged through window chambers in vivo. However, while TMRE+2-NBDG and TMRE+Bodipy FL C16 have been robustly validated for use in combination in single animals, the overlapping excitation and emission spectra of 2-NBDG and Bodipy FL C16 have required them to be measured with a temporal separation of two days. Here, we sought to develop methods to combine all three agents in a concurrent injection.

Materials and

Methods: Combined fluorescence from two fluorophores can be unmixed using linear least squares fit and individual fluorophore references due to the assumption that each pixel contains a linear combination of each fluorophore: Ftotal = FBodipy * c1 + F2-NBDG * c2. Coefficients c1 and c2 can be solved for using a system of equations. For validation, we prepared homogenous phantoms containing Bodipy FL C16 and 2-NBDG where one fluorophore was held constant and the other increased linearly, and a matched phantom where the linearly increasing phantom was prepared individually. Concentrations were biologically relevant. We imaged these phantoms using a portable, novel, low-cost, LED powered microscope, leveraging a combination of two LEDs (450nm and 470nm) and two bandpass filters (535nm +/- 20 nm and 562nm +/- 20 nm) to take four images per FOV to construct a crude spectrum for each pixel, which can be unmixed using a single-fluorophore reference. This methodology was validated against spectroscopic measurements of the phantoms (gold standard). In cohorts of n = 6 female BALB/c mice, we implanted mammary window chambers and imaged the fluorophores in the current standard multi-day injection scheme (TMRE and 2-NBDG on one day, Bodipy FL C16 alone two days later) and in a concurrent combination of all three fluorophores (TMRE, 2-NBDG, Bodipy FL C16 in a single imaging session) in both healthy and orthotopic 4T1-tumor-bearing mice.

Results, Conclusions, and Discussions: In phantoms, we observed 2-NBDG values, when mixed with Bodipy, were within 12% of individually prepared 2-NBDG and retained an R2 = 0.9322, matching the individual preparation of R2 0.9819. Likewise, Bodipy FL C16 values when mixed with 2-NBDG were within 10% error of individually prepared Bodipy FL C16 and retained an R2 = 0.9628, matching the individual preparation of R2 = 0.9603. In vivo imaging of both healthy and 4T1 tumor-bearing mammary fat pads showed that mean image intensities of Bodipy FL C16, 2-NBDG, and TMRE imaged in the multi-day injection scheme were not significantly different from mean image intensities of Bodipy FL C16, 2-NBDG, and TMRE measured in the concurrent injection scheme (p = n.s. for each comparison by Wilcoxan rank sum test).
Leveraging our novel LED-powered microscope and spectral unmixing algorithm, we have developed a system for imaging three fluorophores concurrently in vivo. We have validated that our concurrent injection is not significantly different from the gold standard multi-day injection scheme. Our system has broad capabilities to track three major axes of metabolism spatially and temporally across a full disease time course in animal models as well as patient derived organoids and xenograft models to better understand metabolic vulnerabilities of the most aggressive, treatment resistant, recurrent cancers.

Acknowledgements (Optional):