Characterizing the metabolic heterogeneity in human breast cancer xenografts by 3D high resolution fluorescence imaging
© Xu et al. 2013
Received: 14 February 2013
Accepted: 18 February 2013
Published: 28 February 2013
Skip to main content
© Xu et al. 2013
Received: 14 February 2013
Accepted: 18 February 2013
Published: 28 February 2013
We previously reported that tumor mitochondrial redox state and its heterogeneity distinguished between the aggressive and the indolent breast cancer xenografts, suggesting novel metabolic indices as biomarkers for predicting tumor metastatic potential. Additionally, we reported that the identified redox biomarkers successfully differentiated between the normal breast tissue and the cancerous breast tissue from breast cancer patients. The aim of the present study was to further characterize intratumor heterogeneity by its distribution of mitochondrial redox state and glucose uptake pattern in tumor xenografts and to further investigate the metabolic heterogeneity of the clinical biopsy samples. We employed the Chance redox scanner, a multi-section cryogenic fluorescence imager to simultaneously image the intratumor heterogeneity in the mitochondrial redox state and glucose uptake at a high spatial resolution (down to 50 × 50 × 20 μm3). The mitochondrial redox state was determined by the ratio of the intrinsic fluorescence signals from reduced nicotinamide adenine dinucleotide (NADH) and oxidized flavoproteins (Fp including FAD, i.e., flavin adenine dinucleotide), and the glucose uptake was measured using a near-infrared fluorescent glucose-analogue, pyropheophorbide 2-deoxyglucosamide (Pyro-2DG). Significant inter- and intratumor metabolic heterogeneity were observed from our imaging data on various types of breast cancer xenografts. The patterns and degrees of heterogeneity of mitochondrial redox state appeared to relate to tumor size and metastatic potential. The glucose uptake was also heterogeneous and generally higher in tumor peripheries. The oxidized and reduced regions mostly corresponded with the lower and the higher pyro-2DG uptake, respectively. However, there were some regions where the glucose uptake did not correlate with the redox indices. Pronounced glucose uptake and high NADH were observed in certain localized areas within the tumor necrotic regions, indicative of the existence of viable cells which was also supported by the H&E staining. Significant heterogeneity of the redox state indices was also observed in clinical specimens of breast cancer patients. As abnormal metabolism including the Warburg effect (high glycolysis) plays important roles in cancer transformation and progression, our observations that reveal the 3D intratumor metabolic heterogeneity as a characteristic feature of breast tumors are of great importance for understanding cancer biology and developing diagnostic and therapeutic methods.
Abnormal metabolism and high variability in disease presentations are among the hallmarks of cancer (Heppner & Miller 1998; Hanahan & Weinberg 2011). The intratumor heterogeneity or diversity has been associated with tumor progression and aggressiveness (Heppner 1984; Allred et al. 2008; Park et al. 2010). It is therefore of great importance to characterize tumor metabolic heterogeneity.
Cancer metabolism has been an active research area for understanding tumor biology and developing biomarkers for cancer diagnosis and monitoring treatment response (Cairns et al. 2011; Vander Heiden et al. 2009). Higher than normal glucose uptake/metabolism as first identified by Otto Warburg in 1920s (Koppenol et al. 2011), lays the foundation for cancer staging by fluorinated 2-deoxyglucose positron emission tomography (FDG-PET) in the clinic. Studies also investigated the glucose uptake and its spatial heterogeneity in animal models (Kallinowski et al. 1988; Dearling et al. 2004). Another area under active research investigation is mitochondrial metabolism including TCA cycle and oxidative phosphorylation. Several mitochondrial metabolic enzymes have been identified as oncogenes or tumor suppressors in some cancers (Thompson 2009).
Previously we reported that the mitochondrial redox state and its heterogeneity in tumors imaged by the Chance redox scanner provide sensitive and potentially diagnosis-useful indices for differentiating among five human melanoma lines and two breast cancer lines with different metastatic potential in mouse models (Li et al. 2009a; Xu et al. 2010) . In addition, we reported that the redox imaging indices differentiated between the normal and cancerous breast tissues from breast cancer patients (Xu et al. 2013a). We showed that the aggressive or metastatic tumors have localized more oxidized areas (higher Fp redox ratio, Fp/(Fp + NADH)), and that the Fp redox ratios of the oxidized areas positively correlated with tumor metastatic potential and thus could be further developed for grading tumor aggressiveness of human melanoma and breast cancer. The pattern of the oxidized areas with surrounding reduced areas appears to be a distinct feature of intratumor heterogeneity in aggressive tumors. However, the 3D heterogeneity of the mitochondrial redox indices (NADH, Fp, and the Fp redox ratio) has not been thoroughly investigated for whole tumors.
Because of cancer cells’ higher than normal glycolysis, simultaneously imaging the glucose uptake/metabolism and the mitochondrial redox state in tumors adds one more dimension to our understanding of tumor metabolism. As the size of an oxidized tumor area can be as small as 1–2 mm (Xu et al. 2010), FDG-PET is not suitable for such a purpose due to its low spatial resolution (1–2 mm for small animals and ~4 mm for human subjects). Previously, Zheng and co-workers developed the near-infrared fluorescent glucose-analog Pyro-2DG and also confirmed that Pyro-2DG was selectively accumulated in tumor cells via the glucose transporters (Zhang et al. 2003). Using the high-resolution Chance redox scanner, they achieved simultaneous imaging of Pyro-2DG uptake and the redox indices of the 9L glioma and c-MYC-induced mammary tumors in animal models (Zhang et al. 2004). The imaging data showed that Pyro-2DG uptake correlated well with the reduced redox state and did not affect the redox state of the tumor tissue.
In the present study, we first report the 3D distributions of the mitochondrial redox state in the three breast tumor xenografts in nude mice. We also report the significant intra-tumor heterogeneity of the mitochondrial redox state observed in tumor samples from breast cancer patients. We then show the glucose uptake distribution pattern and its correlations with the redox indices in the breast tumor xenografts. A portion of the preliminary data from mouse models were published in a conference proceeding (Xu et al. 2012; Xu et al.2011a). The results are useful for deepening our understanding on tumor metabolism and its heterogeneity, which may drive the development of novel diagnostic and therapeutic methods for cancer management.
The animal protocols were approved by the Institutional Animal Care and Use Committee at the University of Pennsylvania. Subcutaneous xenografts of three human breast cancer cell lines with increasing metastatic potential (MCF-7 < MDA-MB-468 < MDA-MB-231, N ≥ 5 for each line) were grown to 6–10 mm in diameter in athymic nude mice for the 3D redox scanning. Another two MDA-MB-231 tumor xenografts (>10 mm in diameter and with an apparent necrotic center, average tumor volume = 989 mm3) and two MCF-7 tumor xenografts (>10 mm in diameter, average tumor volume = 750 mm3) were used for simultaneous imaging of Pyro-2DG uptake and the redox state. Pyro-2DG was administered to the anesthetized mice through tail vein injection at a dosage of ~2.5 mg/kg after the tumor-bearing mice were starved for 24 hrs. In approximately two hours, the anesthetized mice were snap-frozen using liquid nitrogen. The snap-freezing protocol maintained the in vivo mitochondrial redox state for the ex vivo redox scanning. The preparation of tissue samples for the redox scanning were previously described (Xu et al. 2010; Xu et al. 2009a). Briefly, the excised frozen samples were embedded with chilled mounting buffer (ethanol:glycerol:water = 10:30:60). Frozen reference standards of NADH, FAD, and Pyro-2DG with known concentrations (in 10 mM Tris–HCl buffer, pH ~7) were quickly mounted adjacent to the tissue for the purpose of calibrating the concentrations of these fluorescent molecules in tissue.
Tissue collection from patients was performed according to a protocol approved by the Internal Review Board of the University of Pennsylvania. Both normal and cancerous tissue specimens were collected from the cancer-bearing breasts of three patients shortly after surgical resection. Core biopsies and/or thin tissue blocks were obtained from tumor and normal adjacent breast tissue, respectively and were snap-frozen in liquid nitrogen within 5–15 minutes after the tissue removal from the body. These tissue specimens were embedded in the same way as described for the xenografts aforementioned.
The low-temperature redox scanner developed by Chance et. al (Chance et al. 1979; Gu et al. 2002; Li et al. 2009b; Quistorff et al. 1985) (the Chance redox scanner) was employed to obtain the multi-slice fluorescence images of NADH, Fp, and Pyro-2DG. The optical filters were 430 nm ± 25 nm and 525 nm ± 32 nm for Fp excitation and emission channel respectively; 360 nm ± 13 nm and 430 nm ± 25 nm for NADH excitation and emission channel, respectively; and 430 ± 28 nm and 670± 11 nm for pyro-2DG excitation and emission, respectively. For the 3D imaging study, the tumor samples were scanned section by section at different tissue depths with 400 ~ 800 μm spacing. Most tumors were scanned at least to the center. For some tumors the entire tumor was scanned. For the study of simultaneous imaging of NADH, Fp, and pyro-2DG, the fluorescent signals were obtained from 2 ~ 3 tissue sections with depth ranging from ~1700-3000 μm under the mouse skin.
The nominal concentration of NADH, Fp, and pyro-2DG in tissue was interpreted using the fluorescence intensity of the corresponding reference standard, respectively and further used to calculate Fp/(NADH + Fp) or NADH/(Fp + NADH). To account for the cell or mitochondria density difference at different regions within a tumor, the normalized Pyro-2DG, i.e. Pyro-2DG/(NADH + Fp) was also calculated based on the nominal concentrations of NADH and Fp.
In addition to the section average values, for the aggressive tumors, we also obtained the mean values of the redox indices of the core (defined as high Fp redox ratio regions) and those of the rim (defined as low Fp redox ratio regions) by fitting of two Gaussian functions to their corresponding histograms. The plots in the middle column of Figure 2 are for the MDA-MB-468 tumor. We can see that the core and the rim are clearly separated for all three redox indices in the entire range of tissue depth. Fp is higher in the core area but drops sharply at 2200 μm and similar pattern is seen for Fp in the rim area. NADH distributions are roughly opposite to those of Fp with the maxima at 1900 μm. The Fp redox ratio is markedly higher in the core than in the rim. In comparison, the redox indices of the core and rim of the MDA-MB-231 tumor exhibited a different depth-variation pattern, but had a similar separation between the rim and core except at one depth point for NADH (Figure 2, right column). All these patterns of the aggressive tumors can be characterized by a bimodal distribution of the mitochondrial redox state with a more and a less oxidized region.
Due to complete cell death, we can identify certain central regions where there were no fluorescence signals in all channels. However, within these necrotic centers, there are islands of cells exhibiting active metabolism indicated by strong NADH, Fp, and Pyro-2DG signals, particularly in the large MCF-7 tumor in Figure 4. Two islands of viable cells in the tumor central region exhibit differential Pyro-2DG uptake as shown by the Pyro-2DG/(Fp + NADH) image although their corresponding redox ratios are similar.
Tumor functional and metabolic heterogeneity has long been recognized. Studies have shown that clinical tumors exhibited more heterogeneous tissue oxygenation distributions than normal tissue, and the intertumor heterogeneity of tissue oxygenation was larger than intratumor heterogeneity; furthermore, the pO2 variations within breast tumors measured by needle electrodes did not correlate with the measuring site (tumor center versus periphery) (Vaupel et al. 2007; Vaupel et al. 1991). In addition to distinguishing between normal and (pre)malignant tissues (Xu et al. 2013b; Xu et al.2013a; Xu et al. 2011b), imaging the intratumor heterogeneity of the mitochondrial redox state in tumors may provide useful information for understanding tumor aggressiveness (Li 2012). Our previous studies showed that more oxidized mitochondrial redox state in tumor tissue was positively associated with tumor metastatic potential in the melanoma and breast cancer xenograft models (Li et al.2009a; Xu et al. 2010). We also observed that the clinical cancerous breast tissue was much more oxidized and more heterogeneous than the normal tissue (Xu et al. 2013a). The present study shows with great details of the metabolic heterogeneity in breast tumor xenografts and breast tumor specimens, which may be further investigated for developing high resolution biomarkers for clinical applications.
The present study is the first to reveal the 3D distribution of the mitochondrial redox state and Fp and NADH distribution in entire breast tumor xenografts. The three tumor lines we used in the study have the following ascending aggressiveness or invasive/metastatic potential, MCF-7 < MDA-MB-468 < MDA-MB-231 (Freund et al. 2003). Both MDA-MB-468 and MDA-MB-231 cells are triple-negative (negative for estrogen receptor, Her2/neu, and progesterone receptor) (Köster et al. 2009; Bartholomeusz et al. 2010; Anders & Carey 2008) and belong to aggressive tumors. From the redox images, we have observed significantly higher degree of heterogeneity in the more metastatic MDA-MB-231 and MDA-MB-468 tumors and less heterogeneity in the less metastatic MCF-7 tumors (≤10 mm). More quantitative analysis of the redox indices and their correlation with the aggressiveness of these three xenograft models is in progress. We also observed that MCF-7 tumors larger than 10 mm in diameter became more heterogeneous with distinct core-rim imaging pattern and the oxidized regions having very high Fp redox ratio (Figure 4). This result suggested that the large MCF-7 tumors may have acquired more aggressive features, consistent with the clinical observations that larger tumors tend to be more aggressive/metastatic.
We also investigated the glucose uptake by simultaneously imaging the fluorescence tracer pyro-2DG and the mitochondrial redox state in four large tumors. The results indicated that there were a large variety of cellular metabolic states existing under the complex tumor microenvironment. Although there are intertumor variations, the common patterns are that these tumors all exhibited more oxidized states in tumor central regions and more reduced states in tumor peripherals, and the normalized glucose uptake pyro-2DG/(Fp + NADH) grossly positively correlated with NADH/(Fp + NADH) in the tumor periphery. Early studies using bioluminescent enzymatic assay demonstrated the heterogeneous spatial distribution of glucose, lactate, and ATP in human breast cancer xenografts in rats (Kallinowski et al. 1988). Heterogeneous uptake of glucose by cancer cells in tumor xenografts was also quantitatively evaluated by radioluminography showing the selective uptake of deoxyglucose for viable cells over necrotic regions (Dearling et al. 2004). Such selectivity was also seen in the present study where tumor peripheral areas have apparently higher Pyro-2DG uptake, corresponding to higher NADH and the NADH redox ratio. This can be explained by the presumably higher perfusion and nutrient delivery and more NADH being metabolically generated in the tumor peripheral areas than in the central regions.
We also observed pronounced Pyro-2DG uptake in the areas inside the necrotic center along with significant NADH and Fp signals. High glucose uptake and NADH concentration inside the necrotic center indicate that these cells had high metabolic activities. It is likely that these cells were in State 3 where cells had sufficient nutrient resulting in rapid oxidative metabolism (high NADH, Fp and the Fp redox ratio) rather than in State 2 where cells were nutrient-starved (low NADH, high Fp and the Fp redox ratio) (Chance & Williams 1955; Chance & Baltscheffsky 1958; Chance & Schoener 1966). This also provides a possible answer to our previous question regarding whether the cancer cells in the oxidized regions were in State 2 (starvation) or State 3 (sufficient substrates) (Li et al. 2009a; Xu et al. 2010). According to the present study, it seems that the answer is some were in State 2 and some in State 3.
We reported the preliminary results from 3D high resolution mapping of the mitochondrial redox state of three breast cancer lines xenografted in mice, i.e., MCF-7, MDA-MB-468, and MDA-MB-231 with increasing ranks of metastatic potential. We identified significant intertumor and intratumor heterogeneity of the redox state among all xenografts and for the first time revealed the mitochondrial redox state spatial distribution in an entire tumor. Additionally, we showed that high degree of heterogeneity existed in the clinical biopsy samples from the breast cancer patients. We also reported the preliminary results of simultaneous imaging of the mitochondrial redox state and glucose uptake in larger breast cancer xenografts with apparent necrotic centers. We found that the glucose uptake was also heterogeneous (usually higher in tumor periphery) and positively correlates with NADH and the NADH redox ratio in general but not in all regions, and that islands of viable cells existed in tumor necrotic/oxidized centers exhibiting both high glucose uptake and NADH. These results indicated that a variety of metabolic states exist in a breast tumor and the metabolic heterogeneity warrants further investigation for developing potential metabolic biomarkers for clinical applications.
This work was supported by research grants from Susan G. Komen Research Foundation (KG081069, L.Z. Li) and National Institute of Health (R01CA155348 (L.Z. Li), RR02305, 2U24-CA083105, NCI Cancer Center Support Grant 2-P30-CA-016520-35), and the Linda and Paul Richardson Breast Cancer Research Funds (J. Tchou). We thank Mr. Aron Roxin for synthesizing Pyro-2DG, Dr. Anatoliy Popov and Dr. Zhihong Zhang for their advice on Pyro-2DG solution preparation, Dr. Xiaohong Ma for her assistance in Pyro-2DG tail-vein administration, and Dr. Beamon Agarwal for his assistance in H&E staining.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.