Near-simultaneous intravital microscopy of glucose uptake and mitochondrial membrane potential, key endpoints that reflect major metabolic axes in cancer

Abstract

While the demand for metabolic imaging has increased in recent years, simultaneous in vivo measurement of multiple metabolic endpoints remains challenging. Here we report on a novel technique that provides in vivo high-resolution simultaneous imaging of glucose uptake and mitochondrial metabolism within a dynamic tissue microenvironment. Two indicators were leveraged; 2-[N-(7-nitrobenz-2-oxa-1, 3-diazol-4-yl) amino]-2-deoxy-D-glucose (2-NBDG) reports on glucose uptake and Tetramethylrhodamine ethyl ester (TMRE) reports on mitochondrial membrane potential. Although we demonstrated that there was neither optical nor chemical crosstalk between 2-NBDG and TMRE, TMRE uptake was significantly inhibited by simultaneous injection with 2-NBDG in vivo. A staggered delivery scheme of the two agents (TMRE injection was followed by 2-NBDG injection after a 10-minute delay) permitted near-simultaneous in vivo microscopy of 2-NBDG and TMRE at the same tissue site by mitigating the interference of 2-NBDG with normal glucose usage. The staggered delivery strategy was evaluated under both normoxic and hypoxic conditions in normal tissues as well as in a murine breast cancer model. The results were consistent with those expected for independent imaging of 2-NBDG and TMRE. This optical imaging technique allows for monitoring of key metabolic endpoints with the unique benefit of repeated, non-destructive imaging within an intact microenvironment.

Full text

1
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
www.nature.com/scientificreports
Near-simultaneous intravital
microscopy of glucose uptake
and mitochondrial membrane
potential, key endpoints that
reect major metabolic axes in
cancer
Caigang Zhu1, Amy F. Martinez
1, Hannah L. Martin1, Martin Li1, Brian T. Crouch1, David A.
Carlson
2, Timothy A. J. Haystead2 & Nimmi Ramanujam1
While the demand for metabolic imaging has increased in recent years, simultaneous in vivo
measurement of multiple metabolic endpoints remains challenging. Here we report on a novel
technique that provides in vivo high-resolution simultaneous imaging of glucose uptake and
mitochondrial metabolism within a dynamic tissue microenvironment. Two indicators were leveraged;
2-[N-(7-nitrobenz-2-oxa-1, 3-diazol-4-yl) amino]-2-deoxy-D-glucose (2-NBDG) reports on glucose
uptake and Tetramethylrhodamine ethyl ester (TMRE) reports on mitochondrial membrane potential.
Although we demonstrated that there was neither optical nor chemical crosstalk between 2-NBDG
and TMRE, TMRE uptake was signicantly inhibited by simultaneous injection with 2-NBDG in vivo. A
staggered delivery scheme of the two agents (TMRE injection was followed by 2-NBDG injection after
a 10-minute delay) permitted near-simultaneous in vivo microscopy of 2-NBDG and TMRE at the same
tissue site by mitigating the interference of 2-NBDG with normal glucose usage. The staggered delivery
strategy was evaluated under both normoxic and hypoxic conditions in normal tissues as well as in a
murine breast cancer model. The results were consistent with those expected for independent imaging
of 2-NBDG and TMRE. This optical imaging technique allows for monitoring of key metabolic endpoints
with the unique benet of repeated, non-destructive imaging within an intact microenvironment.
Deregulation of cellular energetics is a hallmark of cancer1, and metabolic proling of tumors allows researchers
to investigate the mechanisms underlying cancer progression, metastasis, and resistance to therapies2–5. In spite
of variations in tissue site and signaling pathways, most cancers exhibit the common metabolic characteristic
of increased glucose metabolism relative to normal cells1. e ability to perform glycolysis regardless of oxygen
availability was coined the “Warburg eect”, aer Otto Warburg, who rst described aerobic glycolysis in cancer6.
More recently, the Warburg eect is challenged by a growing number of studies showing that many cancers rely
heavily on both mitochondrial metabolism and glycolysis to meet the increased energy demands required for
progression7–9. Even the most glycolytic tumor types may produce only 50–60% of their ATP by glycolysis, with
the balance from mitochondrial metabolism10,11.
Several important phenomena highlight the importance of measuring both glycolytic and mitochondrial
metabolism. Tumors with increased capacity for both glycolysis and oxidative phosphorylation tend to be aggres-
sive, with the ability to survive stressors such as cycling hypoxia or low nutrient availability. is adaptable met-
abolic phenotype promotes negative outcomes such as increased migration12, metastatic propensity13, and drug
resistance14. Further, recent evidence shows that some metastatic tumors rely almost primarily on mitochondrial
1Department of Biomedical Engineering, Duke University, Durham, NC, 27708, USA. 2Department of Pharmacology
and Cancer Biology, Duke University, Durham, NC, 27710, USA. Caigang Zhu and Amy F. Martinez contributed equally
to this work. Correspondence and requests for materials should be addressed to N.R. (email: nimmi@duke.edu)
Received: 18 May 2017
Accepted: 6 October 2017
Published: xx xx xxxx
OPEN
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
2
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
metabolism so that they can utilize “waste products” such as lactate from the surrounding microenvironment15,16;
this has been coined the “Reverse Warburg eect”15. Metabolic changes are also essential for tumor cells that
evade therapy and eventually recur. Some studies have found that dormant cells exhibit a relatively increased
dependence on mitochondrial metabolism17,18, conrming that the ability of tumor cells to shi their metabolism
between glycolysis and oxidative phosphorylation is essential for survival in changing environments19.
ere are a number of imaging methods that enable organ-level imaging of metabolic endpoints in vivo with
a resolution of 1–2 mm20. Positron Emission Tomography (PET) and Magnetic Resonance Spectral Imaging
(MR(S)I) are two such technologies20. PET imaging is a well-accepted technique for measuring glucose uptake
using uorodeoxyglucose ([18 F]FDG) as a tracer21. PET can be also used to image tissue hypoxia by incorporat-
ing additional probes (e.g. [18 F]FMISO)22. MR(S)I can report on both mitochondrial metabolism and glycolysis
endpoints23,24 using 31P or hyperpolarized 13C labeled compounds as tracers, and MRI can also quantify vascula-
ture based on blood ow eects25.
At the cellular level, measurements of glycolysis and mitochondrial metabolism are performed most com-
monly with in vitro cellular metabolism analyzers such as the Seahorse extracellular ux analyzer (Agilent,
USA)26–31. e Seahorse assay measures two metabolic endpoints: the extracellular acidication rate (ECAR),
which reports indirectly on glycolysis, and oxygen consumption rate (OCR), which reports on oxidative phos-
phorylation. ese assays are particularly useful in high-throughput experiments and can be used to compare the
ratio of glycolytic to oxidative metabolism across a spectrum of cell types32,33.
Metabolomics34,35 is a specialized technique based on mass spectrometry that reports on metabolic interme-
diates and end products in both glycolysis and the citric acid cycle, including glucose, pyruvate, lactate, citrate,
succinate, and ATP, among many others. Metabolomics operates on a complementary length scale to PET/MRI
and in vitro cellular metabolism analyzers by providing information at the tissue level. Unlike PET/MRI, metab-
olomics requires the destruction of tissue and therefore does not provide functional information. ere exists
an opportunity for new metabolic tools to bridge the resolution gap between in vitro analysis and whole body
imaging, while providing kinetic information to complement metabolomics.
We have developed a novel strategy to image the spatiotemporal relationship between glucose uptake and
mitochondrial metabolism in an intact tissue microenvironment using intravital microscopy. is approach ena-
bles imaging of the major metabolic axes, glycolysis and oxidative phosphorylation, that underpin important
tumor phenomena. Two indicators were leveraged to achieve this; 2-[N-(7-nitrobenz-2-oxa-1, 3-diazol-4-yl)
amino]-2-deoxy-D-glucose (2-NBDG) is an indicator of glucose uptake and Tetramethylrhodamine ethyl ester
(TMRE) reports on mitochondrial membrane potential. Our group36,37 and others38–40 have extensively validated
2-NBDG as a glucose analog in cells, window chambers, and ectopic and orthotopic tumor models. TMRE, a
rhodamine derivative, has been extensively used in vitro41–43. To complement previous in vitro eorts, we have
recently demonstrated through rigorous validation studies that TMRE reports on mitochondrial membrane
potential in vivo44,45.
In this study, we rst established using a combination of optical microscopy and mass spectrometry that
there is neither signicant optical crosstalk nor chemical crosstalk between 2-NBDG and TMRE in phantoms,
making them well suited for simultaneous imaging. However, TMRE uptake was signicantly inhibited by simul-
taneous injection with 2-NBDG in vivo. Further investigation demonstrated that the inhibitory eect was due
to 2-NBDG temporarily interfering with normal glucose usage which was veried using positive and negative
perturbations with 2-DG and glucose, respectively. A staggered delivery scheme, in which TMRE injection was
followed by a 2-NBDG injection aer a 10-minute delay, mitigated all cross-talk and permitted near- simultane-
ous in vivo microscopy of 2-NBDG and TMRE at the same tissue site. e staggered delivery strategy was evalu-
ated under both normoxic and hypoxic conditions in normal tissues as well as in a murine breast cancer model.
e results were consistent with those expected for independent imaging of 2-NBDG and TMRE. In summary,
near-simultaneous imaging of TMRE and 2-NBDG provides the unique capability to measure key metabolic
endpoints in high resolution with repeatable, in situ tumor imaging.
Results
There is neither chemical nor optical crosstalk between 2-NBDG and TMRE. Liquid chromatogra-
phy-mass spectrometry with electrospray ionization (ESI-LCMS) analysis of mixed 2-NBDG and TMRE solutions
conrmed that there was no inherent chemical reactivity or optical incompatibility between the two uorophores
in the absence of cells or tissue. Figure1 shows the ESI-LCMS data of four solutions: 1) 100 µM 2-NBDG, 2)
100 µM TMRE, 3) 100 µM 2-NBDG + 100 µM TMRE mixed for 1 hour, and 4) 100 µM 2-NBDG + 100 µM TMRE
mixed for 4 days. Each sample contained an internal standard with known spectral features (Hs10) to allow for
quantitative analysis. e chromatograms obtained from combined 2-NBDG + TMRE solutions show that all
features from the single-component solutions were maintained (Fig.1a). Further, integration of extracted ion
chromatograms revealed that the relative amounts of both 2-NBDG and TMRE, normalized to the internal stand-
ard, were not signicantly altered aer 1 hour or 4 days of mixing (Fig.1b).
Figure2 shows representative 2-NBDG (Fig.2a) and TMRE (Fig.2b) uorescence spectra obtained from
phantom sets with single-component 2-NBDG or TMRE only. e 2-NBDG and TMRE concentrations in each
phantom were determined based on our previous hyperspectral imaging studies36,37,45 in which we estimated a
range of relevant 2-NBDG and TMRE concentrations in tissues (described in the methods). e 2-NBDG con-
centrations in these phantoms were varied from 0 to 10 µM in 2 µM increments, while the TMRE concentrations
in the phantoms were varied from 0 to 15 nM in 3 nM increments. e 2-NBDG and TMRE emission peaks
occur at 545 nm and 585 nm, respectively. A linear correlation between uorescence intensity and uorophore
concentration was observed for each uorophore as expected (R2 = 0.993 and p < 0.003 for 2-NBDG, R2 = 0.999
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
3
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
and p < 0.0001 for TMRE). e phantom studies also demonstrate measurable changes in 2-NBDG and TMRE
intensity at concentration increments as low as 2 µM for 2-NBDG and 3 nM for TMRE.
In mixed-component phantoms, 2-NBDG uorescence intensity at a constant 2-NBDG concentration of 6 µM
was unaected by the addition of variable TMRE concentrations between 0 and 15 nM (Fig.2c). Similarly, TMRE
uorescence intensity at a constant TMRE concentration of 9 nM was minimally aected even when the highest
biologically relevant concentration of 2-NBDG (10 µM) was added (Fig.2d). Figure2e shows that TMRE emits
negligible uorescence compared to 2-NBDG upon excitation with the 488 nm laser typically used for 2-NBDG
excitation. Also, TMRE has negligible absorbance at the 2-NBDG emission band. Figure2f shows that 2-NBDG
emits negligible uorescence compared to TMRE upon excitation with the 555 nm laser typically used for TMRE
excitation. Moreover, 2-NBDG has negligible absorbance at the TMRE emission band. Taken together, the phan-
tom study and ESI-LCMS results indicate that there is neither signicant optical nor chemical crosstalk between
2-NBDG and TMRE, showing that they are suitable for combined imaging in vivo.
2-NBDG uptake is unaected by simultaneous injection with TMRE, while TMRE uptake is sig-
nicantly inhibited by simultaneous injection with 2-NBDG. Figure3 shows representative results of
2-NBDG or TMRE imaging in animals receiving a simultaneous injection (2-NBDG + TMRE) or an independent
injection (2-NBDG or TMRE alone). Figure3a shows that 2-NBDG uorescence is negligibly attenuated by the
presence of TMRE when both uorophores are injected simultaneously. Mean kinetic curves in Fig.3c further
demonstrate that the presence of TMRE has negligible eect on the uorescence of 2-NBDG even when the
two uorophores are injected simultaneously (p = NS for 2-NBDG vs. 2-NBDG + TMRE). e kinetic curves
can be used to create a delivery correction factor (RD) for 2-NBDG uptake, as demonstrated in subsequent g-
ures. Figure3b shows that TMRE uorescence is signicantly attenuated by the presence of 2-NBDG when both
uorophores are injected simultaneously. Figure3d demonstrates that TMRE uptake kinetics are signicantly
aected by the presence of 2-NBDG when the two probes are injected simultaneously (p < 0.01 for TMRE vs.
TMRE + 2-NBDG).
Attenuation eect of 2-NBDG on TMRE uorescence is attributed to 2-NBDG interference
with normal glucose usage during glycolysis. We hypothesized that the source of crosstalk was the
interference of 2-NBDG with glycolysis. To test this hypothesis, either glucose (normal glycolytic substrate)
or 2-DG (shown to inhibit glycolysis46) was simultaneously injected with TMRE. Figure4a shows time course
images from animals that were injected with TMRE alone, TMRE and glucose simultaneously, TMRE and 2-DG
Figure 1. 2-NBDG and TMRE are chemically compatible. Solutions containing single-component or combined
2-NBDG and TMRE solutions with an internal standard (Hs10) were analyzed by LC-MS for possible chemical
cross-reactivity. (a) Extracted ion chromatograms (EIC) showing 2-NBDG as a mixture of alpha- and beta-
anomers, TMRE as a mixture of methyl and ethyl esters, and Hs10 as a single peak. Chromatographic features
from combined solutions (2-NBDG + TMRE) and single-component solutions (2-NBDG or TMRE alone) were
maintained. (b) e areas under the curves (AUC) for 2-NBDG (AUC2-NBDG), TMRE (AUCTMRE), and Hs10
(AUCStd) were computed from summation of EI peak integrations related to each compound. Results are shown
as the ratio of AUCTMRE or AUC2-NBDG normalized to AUCStd.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
4
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
simultaneously, or TMRE and 2-NBDG simultaneously. Figure4b demonstrates that TMRE + 2-DG caused
altered kinetics compared to TMRE only (p < 0.01 for TMRE vs. TMRE + 2-DG). is was similar to the eect of
2-NBDG (p < 0.01 for TMRE vs. TMRE + 2-NBDG). However, the simultaneous injection of TMRE and glucose
had no eect on TMRE kinetics (p = NS for TMRE vs. TMRE + glucose). ese results indicate that simultaneous
injection with 2-NBDG attenuates the TMRE signal by temporarily interfering with normal glucose usage. is
data suggests that staggering the 2-NBDG injection following TMRE injection should enable combined use of
the probes.
A staggered injection strategy enables near-simultaneous microscopy of 2-NBDG and TMRE
uptake in vivo. Figure5 shows that a sequential injection strategy prevents attenuation of TMRE uptake.
TMRE was injected rst, followed by 2-NBDG injection aer a 10–15 minute delay. When sequential injection
with a 10–15 minute delay was used, TMRE uorescence closely recapitulated the results that were obtained when
TMRE was administered alone (Fig.5a). TMRE time course images from each group shown in Fig.5a were used
to create kinetic curves (Fig.5c). Figure5c clearly demonstrates that the sequential injection of TMRE followed
Figure 2. TMRE and 2-NBDG are optically compatible. A set of phantoms containing single-component or
combined 2-NBDG and TMRE was tested for optical crosstalk. (a) 2-NBDG spectra showed an emission peak
of 545 nm and peak intensity increased linearly with concentration. (b) TMRE spectra showed an emission peak
of 585 nm and peak intensity increased linearly with concentration. (c) 2-NBDG intensity was not aected by
the presence of TMRE. (d) TMRE intensity was not aected by the presence of 2-NBDG. e reduced scattering
coecient in all phantoms was 10 cm−1. (e) TMRE emits negligible uorescence compared to 2-NBDG
upon excitation with a 488 nm laser, which was typically used for 2-NBDG excitation. TMRE has negligible
absorbance at the 2-NBDG emission band. (f) 2-NBDG emits negligible uorescence compared to TMRE
upon excitation with a 555 nm laser, which was typically used for TMRE excitation. 2-NBDG has negligible
absorbance at the TMRE emission band. Excitation wavelength (488 or 555 nm) is shown on each panel.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
5
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
by 2-NBDG restored the expected kinetics (p = NS for TMRE vs. Delay: 10–15 min). Figure5b shows that the u-
orescence of 2-NBDG was negligibly aected by the presence of TMRE when TMRE and 2-NBDG were injected
sequentially with a 10–15 minute delay. Mean kinetic curves in Fig.5d further conrm that sequential injection
does not aect 2-NBDG kinetics (p = NS for 2-NBDG vs. Delay: 10–15 min).
Imaging of TMRE and 2-NBDG with a staggered injection strategy captures the expected met-
abolic responses to hypoxia. Figure6 shows the results of TMRE and 2-NBDG imaging in animals under
hypoxic conditions (10% oxygen) that received aTMRE injection only or a sequential injection of both agents
with a 10–15 minute delay between the administration of TMRE and 2-NBDG. As shown in Fig.6a, TMRE inten-
sity at 45 minutes (TMRE45) decreased signicantly under hypoxia compared to normoxia (21% oxygen) when
the animals received either an independent or sequential injection strategy. In contrast, 2-NBDG intensity at
60 minutes (2-NBDG60) increased under hypoxia compared to normoxia when the sequential injection strategy
was used. Figure6b and c show the mean uptake kinetics of TMRE and 2-NBDG respectively. Pixel distribution
curves were created to illustrate the fraction of pixels in each experimental group that exceeds a given uores-
cence intensity value (see Methods for details). Figure6d shows the pixel distribution curves generated from
Figure 3. 2-NBDG uptake is unchanged by simultaneous injection with TMRE, while TMRE uptake is
signicantly inhibited by simultaneous injection with 2-NBDG. 2-NBDG and TMRE kinetic imaging was
performed on non-tumor murine dorsal window chambers aer they received a simultaneous injection (2-
NBDG + TMRE) or an independent injection (2-NBDG or TMRE alone). (a) Representative 2-NBDG uptake
time course images for simultaneous injection and independent 2-NBDG injection. (b) Representative TMRE
uptake time course images for simultaneous injection and independent TMRE injection. (c) Mean uptake
kinetics of 2-NBDG. RD refers to the rate of delivery of 2-NBDG. 2-NBDGmax = the peak intensity of 2-NBDG.
Tmax = time (in seconds) at which 2-NBDGmax occurs. (d) Mean uptake kinetics of TMRE. NS = not signicant.
N = 5 mice/group. Comparison of mean kinetic curves across animal groups was performed with a two-way
Analysis of Variance (ANOVA) test using the MATLAB (Mathworks, USA) statistics toolbox.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
6
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
TMRE images at 45 minutes (TMRE45) for both injection strategies and both oxygenation conditions. Figure6e
shows the pixel distribution curves generated from the 2-NBDG images at 60 minutes (2-NBDG60) divided by
the rate of delivery (RD), as shown in Fig.3c. e individual pixel distribution curves from the animals in each
test group were averaged to create the curves shown (mean ± SE). Figure6f and g show the mean intensity from
TMRE45 images and mean intensity from 2-NBDG60/RD images, respectively. TMRE45 decreased signicantly
during hypoxia (p < 0.05), and 2-NBDG60/RD increased signicantly during hypoxia (p < 0.001).
Imaging of TMRE and 2-NBDG with a staggered injection strategy captures a distinct meta-
bolic phenotype in 4T1 tumors relative to non-tumor tissue. Figure7 shows the results of TMRE
and 2-NBDG imaging in animals with 4T1 tumors that received a sequential injection of TMRE (rst) and
2-NBDG (second) with a 10–15 minute delay. Figure7a shows TMRE imaging at 45 minutes post TMRE injection
and 2-NBDG imaging at 60 minutes post 2-NBDG injection in representative normal and 4T1 tumor-bearing
window chambers. Figure7b and c show the pixel distribution curves generated from images of TMRE45 and
2-NBDG60/RD respectively (see Methods for details). e individual pixel distribution curves from the animals in
each test group were averaged to create the curves shown (mean ± SE). Figure7b shows that TMRE45 increased
signicantly in 4T1 tumors compared to non-tumor tissues (p < 0.05). Similarly, Fig.7c shows that 2-NBDG60/
RD increased signicantly in 4T1 tumors compared to normal tissues (p < 0.05). ese results are consistent with
previous studies that evaluated TMRE and 2-NBDG uptake in 4T1 independently45.
Discussion
Near-simultaneous high-resolution imaging of mitochondrial membrane potential and glucose uptake in living
animals is well poised to enable unprecedented studies of metabolism in a variety of important disease models
and, in particular, cancer. ough several important metabolic imaging techniques are already being used20, our
complementary uorescence-based technique can be coupled with a variety of optical technologies to provide a
Figure 4. Simultaneous injection with glucose does not aect TMRE uptake, while simultaneous injection
with 2-DG decreases TMRE uptake by half. TMRE kinetic imaging was performed on murine dorsal
window chambers aer they received simultaneous injection of TMRE and glucose, 2-DG, or 2-NBDG. (a)
Representative TMRE uptake time course images for each group. (b) Mean uptake kinetics of TMRE for each
group. (c) Statistical comparison of the mean kinetic curves for the simultaneous injection groups vs. TMRE
alone. NS = not signicant. N = 4–5 mice/group. Comparison of mean kinetic curves across animal groups was
performed with a two-way Analysis of Variance (ANOVA) test using the MATLAB (Mathworks, USA) statistics
toolbox.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
7
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
resolution that enables investigation of the spatiotemporal relationship between glycolysis and oxidative phos-
phorylation at the tissue microenvironment level47. e technique can be utilized to assess novel therapies tar-
geted at tumor metabolism or to identify the metabolic changes that mark response or resistance in specic cell
populations. Further, studies of metabolic symbiosis between tumors and their microenvironments15, as well
as metabolic responses to environmental stress48,49, will benet from high-resolution, metabolic imaging of the
intact tissue microenvironment.
e ability of our microscope to image supercial tissues at capillary-level resolution with a millimeter-scale
single frame eld of view makes it particularly useful for imaging the tissue microenvironment in window cham-
ber models. Dorsal window chambers are by design optically thin and therefore provide an excellent model sys-
tem to image via microscopy50, where there is an inherent tradeo between sensing depth and lateral resolution.
e existing window chamber imaging techniques either provide a large eld of view (wide-eld imaging sys-
tems36,51) or high resolution (multiphoton52 or confocal microscopes53), but not necessarily both. Our microscope
has both high resolution (~2.2 µm) and a millimeter-scale single frame eld of view (2.1 mm × 1.6 mm), which
makes it well-suited to image normal tissue and small tumors in a window chamber model. Further, it has a sens-
ing depth of approximately 500 µm at the wavelength band for uorescence imaging (i.e. 545 nm for 2-NBDG u-
orescence and 585 nm for TMRE uorescence)54. e phantom studies showed measurable changes in 2-NBDG
and TMRE intensity when the concentration was varied as little as 2 µM for 2-NBDG and 3 nM for TMRE.
Figure 5. A sequential injection strategy rescues TMRE intensity from the decrease caused by simultaneous
injection with 2-NBDG. TMRE and 2-NBDG kinetic imaging was performed on murine dorsal window
chambers aer they received one of three distinct injection strategies: (1) TMRE alone, (2) 2-NBDG alone,
(3) TMRE followed by 2-NBDG with a 10–15 min delay. (a) Representative TMRE uptake time course images
for the sequential injection strategy and independent TMRE injection. (b) Representative 2-NBDG uptake
time course images for the sequential injection strategy and independent 2-NBDG injection. (c) Mean uptake
kinetics of TMRE for each injection strategy. (d) Mean uptake kinetics of 2-NBDG for each injection strategy.
NS = not signicant. N = 5–6 mice/group. Comparison of mean kinetic curves across animal groups was
performed with a two-way Analysis of Variance (ANOVA) test using the MATLAB (Mathworks, USA) statistics
toolbox.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
8
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
e stable binding and rapid equilibration of TMRE enabled us to perform simultaneous imaging of mito-
chondrial membrane potential and glucose uptake by injecting TMRE rst followed with 2-NBDG injection aer
a delay. It should be noted that there are several uorescent mitochondrial dyes that can be used for MMP meas-
urements43. Dierent probes are recommended for each usage paradigm, depending on the probe’s uptake kinet-
ics, concentration, and mitochondrial binding anity. Rhodamine 123 is recommended for applications that seek
to measure rapid changes in membrane potential43. JC-1 dye55 is commonly used for measurement of transient,
non-stable changes in MMP. Both TMRE and TMRM (Tetramethylrhodamine, methyl ester)43 are recommended
for measurement of pre-existing dierences in MMP, such as the stable dierences between tumor groups and
normal tissue that we desire to observe. TMRM can be used when only a short binding period is needed to mini-
mize disturbance to electron transport43,56. However, our primary goal in this study was to determine appropriate
time points for combined imaging with 2-NBDG, which has its own unique delivery and uptake kinetics. We thus
chose to use TMRE because of its fast equilibration and stable binding, which maximized the likelihood of nding
Figure 6. TMRE uptake decreases and 2-NBDG uptake increases in dorsal window chambers under hypoxic
gas breathing (10% oxygen). Normal dorsal window chambers were imaged with TMRE (rst) and 2-NBDG
(second) with a 10–15 min delay between injections. TMRE uptake and 2-NBDG uptake were captured under
either normoxia or hypoxia. (a) Representative images for each test group. (b) Mean uptake kinetics of TMRE.
(c) Mean uptake kinetics of 2-NBDG. (d) Pixel distribution curves show the mean distribution of pixels from
TMRE images taken at 45 minutes (TMRE45) for each group. (e) Pixel distribution curves show the mean
distribution of pixels of delivery-corrected 2-NBDG images taken at 60 minutes (2-NBDG60/RD) for each group.
(f) Mean intensity from TMRE45 images. (g) Mean intensity from 2-NBDG60/RD images. NS = not signicant.
N = 4–5 mice/group. Comparison of mean kinetic curves across animal groups was performed with a two-way
Analysis of Variance (ANOVA) test using the MATLAB (Mathworks, USA) statistics toolbox. Comparison of
mean pixel distribution curves across animal groups was performed with a Kolmogorov-Smirnov (KS) test
using the MATLAB (Mathworks, USA). Comparison of the mean intensity of TMRE45 or 2-NBDG60/RD across
animal groups was performed with two sample t-tests using the MATLAB (Mathworks, USA) statistics toolbox.
Figure 7. TMRE uptake and 2-NBDG uptake are increased in 4T1 tumors relative to normal tissues. Normal
and 4T1 dorsal window chambers were imaged with TMRE (rst) and 2-NBDG (second) with a 10–15 min
delay between injections. (a) Representative images for each test group. (b) Pixel distribution curves show
the mean distribution of pixels from TMRE images taken at 45 minutes (TMRE45) for each group. (c) Pixel
distribution curves show the mean distribution of pixels of delivery-corrected 2-NBDG images taken at
60 minutes (2-NBDG60/RD) for each group. NS = not signicant. N = 3–5 mice/group. Comparison of mean
pixel distribution curves across animal groups was performed with a Kolmogorov-Smirnov (KS) test using the
MATLAB (Mathworks, USA).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
9
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
a time point that was appropriate for imaging both probes. To characterize TMRE’s basic in vivo properties, our
study expands upon previous work by including a recommended TMRE dose, providing TMRE uptake kinetics
in both normal and tumor models, and validating TMRE imaging through multiple perturbations.
Previous studies have seen a range of interactions that can occur when multiple compounds are given simul-
taneously57–59, thus changing their kinetics or preventing full accumulation of the compounds. We established in
our tissue mimicking phantom study that no detectable chemical nor optical interaction was seen between the
uorophores. Consistent with phantom studies, the tissue studies demonstrated that TMRE negligibly aected
the fluorescence of 2-NBDG. Since the presence of TMRE did not affect 2-NBDG uptake, this experiment
also importantly conrmed that low concentrations of TMRE reach the tissue, and TMRE thus operates in the
non-quenching range. As a result, TMRE uorescence can be interpreted as dye accumulation corresponding to
more polarized mitochondria60.
We observed that simultaneous injection of 2-NBDG and TMRE changed TMRE uptake. To understand the
inhibitory mechanism that 2-NBDG exerts on TMRE, we imaged TMRE during co-injection with either glucose
or 2-DG. Co-injection with glucose had no eect on the TMRE signal; however, co-injection with 2-DG caused
a decrease in TMRE uptake similar to that caused by 2-NBDG. Glucose, 2-DG, and 2-NBDG are all taken up by
GLUT transporters and phosphorylated by hexokinase33,61–63. However, only glucose continues fully through
glycolysis to pyruvate, which is converted to acetyl-CoA and fed into the TCA cycle64. 2-NBDG and 2-DG remain
trapped in the cytoplasm aer phosphorylation62,63, which has been shown to have an inhibitory eect on glyc-
olysis. e resulting decrease in pyruvate to the TCA cycle may therefore be responsible for a drop in mitochon-
drial membrane potential and TMRE uptake. We know that, at the concentration used, any metabolic eects of
2-NBDG occur on a short time-scale, since we previously showed that multiple days of 2-NBDG imaging did not
cause an order eect65,66. It is yet unclear why 2-NBDG caused greater inhibition of TMRE uptake than 2-DG.
Compared to other uorescent glucose probes, 2-NBDG has a low molecular weight (MW = 342) and it directly
competes with both glucose and 2-DG for cellular uptake62,67. However, 2-DG has an even lower molecular weight
(MW = 164) and we hypothesize that this allows it to clear from tissue rapidly. is 2-DG clearance may be
responsible for partially restoring TMRE uptake to a level between the 2-NBDG group and the control group.
Toward our ultimate goal of metabolic imaging in diverse cancer applications, our current work in normal
tissues served to optimize and validate the sequential injection protocol that enabled near-simultaneous imaging
of 2-NBDG and TMRE. Sequential injection of TMRE followed by 2-NBDG with a 10–15 minute delay restored
the expected uptake and kinetics of both uorophores by allowing TMRE to equilibrate in the tissue and bind sta-
bly to mitochondria43 prior to 2-NBDG injection. Near-simultaneous imaging with the delayed injection strategy
consistently yielded results in line with the known metabolic response to hypoxia: increased glucose uptake and
decreased mitochondrial metabolism in normal tissue68. Imaging in 4T1 window chambers also indicated that
the sequential injection strategy developed here was appropriate for small tumors (~6 mm diameter). We saw that
4T1 tumors maintained both increased 2-NBDG uptake and increased TMRE uptake relative to normal tissue,
consistent with the ndings from our former study45 in which 2-NBDG and TMRE were injected in separate
cohorts of animals. e average intensity of TMRE45 increased ~1.5 fold and the average value of 2-NBDG60/
RD increased ~3.5 fold in 4T1 tumors compared to normal tissue, which is comparable to the changes observed
as a result of hypoxic stress in non-tumor window chambers. While the hypoxia and tumor studies illustrate the
dynamic range of TMRE and 2-NBDG imaging, the phantom studies speak to the sensitivity of the technique.
It is interesting to observe that uorescence signal was not uniform throughout the eld of view in the dorsal
window chamber studies. ere may be multiple biological phenomena that underlie the variable uorescence
signal. Most importantly, our previous studies have shown that vascular oxygenation is spatially heterogeneous,
even in non-tumor tissue36,37,45. e relationship between oxygenation and metabolism, as demonstrated here
by our hypoxic perturbation study, likely inuences the regional uptake of both probes. Oxygenation can have
profound eects on metabolism; specically, hypoxia is strongly associated with a shi toward a glycolytic pheno-
type in normal tissue and particularly in tumors69. As tumors grow, they develop natural regions of hypoxia due
to the combination of increased oxygen consumption during mitochondrial metabolism69, cell growth beyond
the oxygen diusion limit, and impeded delivery due to the immature and tortuous vessels created by angiogen-
esis65,70,71. Our previous work45 in which 2-NBDG and TMRE were injected in separate cohorts of animals has
demonstrated that decreasing the inspired oxygen concentration to 10% caused profound metabolic eects in a
panel of tumor lines (4T1, 4T07 and 67NR). Glucose uptake typically increased when inspired oxygen concen-
tration was decreased to 10%, while TMRE uptake typically decreased under the same forced hypoxic conditions.
However, both glucose uptake and TMRE increased during hypoxia in the metastatic 4T1 line. As tumors pro-
gress and develop regions of hypoxia, they will be likely characterized by a shi toward increased glucose uptake
and decreased MMP, though highly aggressive tumors may reveal special adaptations to hypoxic stress.
High-resolution imaging of glycolytic and mitochondrial endpoints will prove useful to study not only can-
cer, but also diabetes and other diseases characterized by metabolic aberrations, which until now have suered
from the lack of repeatable, high resolution metabolic imaging technologies. e extensive use of in vitro cellular
metabolism assays in the elds of immunology, neurobiology, nutrition, and cardiovascular research, among
many others, highlights the widespread usefulness of metabolic measurement technologies. By enabling in vivo
studies of glucose uptake and mitochondrial membrane potential at a length scale and resolution that comple-
ment existing methods, our imaging technique has the potential to ll an important need and facilitate novel
transdisciplinary studies of metabolism.
Methods
Liquid chromatography-mass spectrometry of uorophore samples. Quantitative LCMS was
performed on samples of 2-NBDG and TMRE with an internal standard, 2-(((1 R,4 R)-4-Hydroxycyclohexyl)
amino)-4-(3,6,6-trimethyl-4-oxo-4,5,6,7-tetrahydro-1H-indazol-1-yl)benzamide (Hs10), prepared as previously
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
10
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
described72, to analyze for uorophore stability. Four methanolic solutions were prepared: (i) 100 µM 2-NBDG,
100 µM Hs10; (ii) 100 µM TMRE, 100 µM Hs10; (iii) 100 µM 2-NBDG, 100 µM TMRE, 100 µM Hs10 (incubated
1 hour at 25 °C), and (iv) 100 µM 2-NBDG, 100 µM TMRE, 100 µM Hs10 (incubated 4 days at 25 °C). Electrospray
Ionization (ESI) LCMS analysis was performed using an Agilent 1200 Series liquid chromatography unit with
Agilent Ion Trap 6310 mass spectrometer detection (Agilent Technologies, Santa Clara, USA). Chromatography
was performed on an Agilent Eclipse Plus C18 column, 5 µm, 4.6 × 150 mm, 10 µL injection volume, using sol-
vents A: 0.2% formic acid in water; B: 0.2% formic acid in acetonitrile; gradient separation method: 0–100% B
over 9 minutes, ow rate 1 mL/min. Extracted Ion Chromatograms (EIC) were created by extraction of m/z sig-
nals for all ions related to analytes (2-NBDG, TMRE) and Hs10 from total ion chromatograms. EI for 2-NBDG:
m/z [M + H]+ = 343.0; [M + Na ]+ = 365; [2 M + Na]+ = 707.0. EI for TMRE: m/z [M]+ = 415.0 (ethyl ester);
[M]+ = 401.0 (methyl ester). EI for Hs10: m/z [M + H]+ = 411.0. All EI peaks related to analytes and Hs10 were
manually integrated. e summation of area under the curve for each analyte (AUCanalyte) was compared to the
AUC for Hs10 from each sample (AUCStd). Ratios of AUCanalyte/AUCStd were used to determine changes in analyte
concentration within samples (iii) and (iv) relative to samples (i) and (ii).
Spectral uorescence microscopy system. To further determine whether 2-NBDG and TMRE were
suitable for combined imaging, we performed a tissue-mimicking phantom study and animal imaging using a
custom designed microscope. In this study, our previously reported microscope73 has been modied as shown in
Fig.8 for optical imaging of both phantoms and in vivo animal tissue. In the illumination channel, a 488 nm crys-
tal laser (DL488–100-O, Crystal laser, Reno, NV, USA) and a 555 nm crystal laser (CL555-100-O, Crystal laser,
Reno, NV, USA) were utilized to excite 2-NBDG and TMRE, respectively. A 505 nm longpass dichroic mirror
(DMLP505R, orlab, USA) and a 573 nm longpass dichroic mirror (FF573-Di01-25 × 36, Semrock, Rochester,
New York, USA) were placed in the beam splitter wheel for 2-NBDG and TMRE imaging, respectively. e key
advantage of the uorescence system is its spectral capability, which is achieved by using a liquid crystal tuna-
ble lter (LCTF) (VariSpec VIS-7-35, PerkinElmer, Inc. Waltham, MA, USA) and a high resolution dual-modal
charge-coupled device (CCD) (ORCA-Flash4.0, Hamamatsu, Japan). e spectral microscope system was cali-
brated wavelength by wavelength using a standard lamp source (OL 220 M, S/N: M-1048, Optronic Laboratories,
USA).
A Nikon CFI E Plan Achromat 4x objective (NA = 0.1, Nikon Instruments Inc., USA) was used for all imaging
studies. e single frame eld of view (FOV) and lateral resolution were measured using a 1951 USAF reso-
lution target. e smallest element on the target (group 7, element 6), corresponding to a lateral resolution of
2.2 µm, was clearly resolved as shown in Fig.8. e single frame FOV was measured to be 2.1 mm × 1.6 mm,
and was limited by the illumination beam size rather than the CCD itself. e entire system was controlled via
custom-designed LabVIEW soware allowing the spectral imaging to be performed automatically and rapidly.
Note that while our microscope is capable of optical sectioning due to its structured illumination modality, we did
not utilize this feature in the current study.
Figure 8. Schematic of the uorescence spectral imaging system with millimeter-scale eld of view and
micron-level resolution. e 488 nm laser was used for 2-NBDG imaging while the 555 nm laser was used for
TMRE imaging. BX: Beam expander; BS: Beam splitter; CCD: Charge-coupled device; DBS: Dichroic beam
splitter; LCTF: Liquid crystal tunable lter; OBJ: Objective lens; P: Polarizer; RL: Relay lens; SF: Spatial lter;
SLM: Spatial light modulator; TL: Tube lens.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
11
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
Tissue-mimicking phantoms. A series of tissue-mimicking phantoms containing 2-NBDG or TMRE
at various biologically-relevant concentrations was prepared to validate the system’s spectral capability. e
2-NBDG and TMRE concentrations in each phantom were determined based on our previous hyperspectral
imaging study from which we estimated a range of relevant 2-NBDG and TMRE concentrations in animal tis-
sue (Fig.9). Specically, the 2-NBDG concentrations in these phantoms were varied from 0 to 10 µM in 2 µM
increments, while the TMRE concentrations in the phantoms were varied from 0 to 15 nM in 3 nM increments.
Two sets of mixed-component uorescence phantoms containing both 2-NBDG and TMRE were prepared to
investigate potential optical cross-talk between the two uorophores. In one set of mixed-component phantoms,
2-NBDG concentration was xed to be 6 µM while TMRE concentrations were varied from 0 and 15 nM. In the
second set of mixed-component phantoms, the TMRE concentration was xed to be 9 nM while 2-NBDG con-
centrations were varied from 0 to 10 µM. Polystyrene spheres (07310, Polysciences, Warrington, Pennsylvania)
were used as the scatterer in all phantoms. e reduced scattering level for all uorescence phantoms was 10 cm−1,
which closely mimics the scattering level of window chamber tissue described in literature74,75. No absorbers
were added to the uorescence phantoms since the absorption of window chamber tissue is negligible based
on previously published reports74,75. Deionized water was used to suspend the scattering beads and the uoro-
phores in each liquid uorescence phantom. e 2-NBDG (emission peak around 545 nm) uorescence images
were captured automatically from 500 nm to 700 nm in 5 nm increments with the help of the LCTF. In contrast,
the TMRE (emission peak around 585 nm) uorescence images were acquired from 565 nm to 700 nm in 5 nm
increments. e integration time for both 2-NBDG and TMRE imaging was set to 1 s for all phantom studies.
e absorbance spectra of pure TMRE solution (9 nM) and 2-NBDG solution (6 µM) were measured by a UV-Vis
spectrophotometer (Agilent Cary). In all of the uorescence measurements, background images of phantoms
without uorophores were subtracted from the uorescence images during data processing.
Murine dorsal skin ap window chamber model and imaging protocol. All experiments described
here were performed in accordance with approved guidelines and regulations. e protocol A114-15-04 was
approved by the Duke University Institutional Animal Care and Use Committee (IACUC). We surgically
implanted titanium window chambers on the backs of female athymic nude mice (nu/nu, NCI, Frederick,
Maryland) under anesthesia (i.p. administration of ketamine (100 mg/kg) and xylazine (10 mg/kg)) using an
established procedure50. All animals were housed in an on-site housing facility with ad libitum access to food and
water and standard 12-hour light/dark cycles. Mice were fasted for 6 hours before imaging to minimize variance
in metabolic demand76. e animals were randomly assigned to one of the imaging groups listed in Table1. e
uorescence probes were injected into mice via tail vein. e injection volume was held constant at 100 µL for all
experiments. Around 3 to 5 animals were used in each group as specied in the corresponding gures. Imaging
groups of normal animals under normoxia were designed to identify the potential biological cross-talk and opti-
mize the protocol for simultaneous imaging of TMRE and 2-NBDG in animals. Imaging groups of normal ani-
mals under hypoxia were designed to further validate that the optimized imaging protocol could enable optical
Figure 9. Estimated in vivo concentrations of 2-NBDG and TMRE in normal and tumor window chambers
using a hyperspectral imaging system44,45. Fluorescence images were captured in non-tumor (N.T.) window
chambers and in 67NR, 4T07, and 4T1 murine tumors aer single injection with 2-NBDG (0.1 mL of 6 mM)
or TMRE (0.1 mL of 25 µM). e estimated 2-NBDG and TMRE concentrations were then calculated by
comparing in vivo uorescence intensities to uorescence intensities of tissue-mimicking phantoms imaged
with the same instrument settings. Numbers in the tables correspond to uorescence intensities which were
then converted to estimated tissue-level concentrations.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
12
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
measurement of expected responses to known biological perturbations. Imaging groups of 4T1 tumors were
designed to test the feasibility of the optimized protocol for cancer metabolic imaging.
Background uorescence images of the window chamber were taken prior to the injection of any uoro-
phores. All of the injections were performed following the protocols listed in Table1. TMRE uorescence imaging
was performed for 45 minutes with a frequency of every 5 minutes. 2-NBDG imaging was performed for 60 total
minutes with a frequency as follows: every 2 minutes for the rst 10 minutes and then every 5 minutes for next
50 minutes of imaging. Only TMRE imaging was performed for imaging groups which involved injection of
glucose or 2-DG, with the same image capture frequency used in standard TMRE imaging. TMRE imaging was
performed at its peak emission wavelength, i.e. 585 nm, while 2-NBDG imaging was performed at 545 nm. e
integration time for all in vivo uorescence imaging was set to 5 s. All animals were anesthetized under inhaled
isourane (1–1.5% v/v) in room air or hypoxic gas during imaging. Each animal was euthanized aer the comple-
tion of all imaging based on the IACUC protocol.
Data processing and statistical analysis. Prior to any quantitative image processing, all images from
both the phantom study and the animal study underwent background subtraction rst and then calibration by
a uorescence slide (DeltaVision, Ex/Em: 488 nm/519 nm), to account for autouorescence and day-to-day sys-
tem variation, respectively. Since all of the phantoms were liquid solutions with no identiable features, it was
reasonable to average the spectral images into one spectrum for data analysis for the purpose of demonstrating
the spectral capability of the microscopy system. e average intensities of the previously processed uorescence
phantom images at all wavelengths were calculated to form a TMRE uorescence spectrum or 2-NBDG uores-
cence spectrum.
Image processing for the animal data was dierent compared to the phantom data due to the presence of blood
vessels. Previous studies36 have revealed that 2-NBDG extravasates into the parenchymal tissue, is taken up by
cells, and trapped in the cytosol within a few minutes post tail vein injection. Additional studies45 showed that
TMRE extravasates into the parenchymal tissue, enters cells, and is localized to mitochondria within 15 minutes
post tail vein injection. Minimal uorescence is observed in large vessels at our imaging time points of 45 minutes
(TMRE) and 60 minutes (2-NBDG) because the majority of the dye has already been localized to the cytosol
(2-NBDG) or mitochondria (TMRE). us, we have excluded these low-signal blood vessel regions during the
quantitative analysis in our study to reect only the 2-NBDG and TMRE uptake in the tissue space. To remove
the blood vessels from the quantitative analysis, a manually-traced blood vessel mask was applied to each set of
uorescence images. Only the tissue regions without blood vessels were considered for uorescence intensity cal-
culations of either TMRE or 2-NBDG. e average intensity values of the non-blood vessel tissue regions at every
time point were calculated to generate a time course kinetic curve. Comparison of mean kinetic curves across
animal groups was performed using a two-way analysis of variance (ANOVA) test followed by Tukey-Kramer
post-hoc tests.
Previous work by our group determined appropriate endpoints for measurement of TMRE and 2-NBDG in
vivo. We demonstrated that TMRE uptake 45 minutes aer injection responded as expected to perturbations of
mitochondrial membrane potential in both normal tissue and tumors45. TMRE uptake was also robust to minor
inter-animal variation in delivery kinetics. On the other hand, we found that although 2-NBDG uptake had
reached a stable plateau by 60 minutes aer injection (2-NBDG60), its nal intensity was profoundly inuenced
by the delivery kinetics of 2-NBDG delivery36,65. Accounting for inter-animal dierences in 2-NBDG delivery
with a correction factor (RD = 2-NBDGmax/Tmax, as shown in Fig.3c) resulted in more accurate measurement of
glucose uptake following known metabolic perturbations in normal tissue and tumors36,65. In the present study,
we therefore use the endpoints TMRE45 and 2-NBDG60/RD to represent TMRE uptake and delivery-corrected
2-NBDG uptake, respectively.
Pixels in the non-vessel space of TMRE45 and 2-NBDG60/RD images were used to create a pixel distribution
curve (1-cumulative distribution) for each animal. e proles illustrate the fraction of imaged pixels that meet or
exceed specic TMRE intensity values or delivery-corrected 2-NBDG intensity values at the time of measurement
Injection protocol Time point and dosage of tail vein injection
Normoxia in normal animals
TMRE only t = 0 min: 100 µL of 75 µM TMRE
2-NBDG only t = 0 min: 100 µL of 6 mM 2-NBDG
TMRE + 2-NBD G t = 0 min: 100 µL of 75 µM TMRE + 6 mM 2-NBDG
TMRE + Glucose t = 0 min: 100 µL of 75 µM TMRE + 6 mM glucose
TMRE + 2-D G t = 0 min: 100 µL of 75 µM TMRE + 6 mM 2-DG
TMRE → 2-NBDG (Delay: 10–15 mins) t = 0 min: 100 µL of 75 µM TMRE; t = 10–15 min: 100 µL of 6 mM 2-NBDG
Hypoxia in normal animals
TMRE only t = 0 min: 100 µL of 75 µM TMRE
2-NBDG only t = 0 min: 100 µL of 6 mM 2-NBDG
Optimal strategy Optimal strategy determined from normoxic imaging groups
4T1 tumors
Optimal strategy Optimal strategy determined from normoxic imaging groups
Table 1. Animal imaging protocols.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
13
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
(t = 45 min, or 60 min, respectively). e individual pixel distribution curves were averaged across multiple ani-
mals to create the nal curves (mean ± SE). Comparison of TMRE45 and 2-NBDG60/RD distributions among
dierent imaging groups was performed with a repeated measures Kolmogorov-Smirnov test.
e datasets generated and analyzed during the current study are available from the corresponding author on
reasonable request.
References
1. Hanahan, D. & Weinberg, . A. Hallmars of Cancer: The Next Generation. Cell 144, 646–674, https://doi.org/10.1016/j.
cell.2011.02.013 (2011).
2. roemer, G. & Pouyssegur, J. Tumor cell metabolism: Cancer’s Achilles’ heel. Cancer Cell 13, 472–482, https://doi.org/10.1016/j.
ccr.2008.05.005 (2008).
3. Cairns, . A., Harris, I. S. & Ma, T. W. egulation of cancer cell metabolism. Nat Rev Cancer 11, 85–95, https://doi.org/10.1038/
nrc2981 (2011).
4. Waler-Samuel, S. et al. In v ivo imaging of glucose uptae and metabolism in tumors. Nat Med 19, 1067–+, https://doi.org/10.1038/
nm.3252 (2013).
5. Walsh, A. J. et al. Optical metabolic imaging identies glycolytic levels, subtypes, and early-treatment response in breast cancer.
Cancer Res 73, 6164–6174, https://doi.org/10.1158/0008-5472.CAN-13-0527 (2013).
6. oppenol, W. H., Bounds, P. L. & Dang, C. V. Otto Warburg’s contributions to current concepts of cancer metabolism. Nat Rev
Cancer 11, 325–337, https://doi.org/10.1038/nrc3038 (2011).
7. Epstein, T., Xu, L., Gillies, . J. & Gatenby, . A. Separation of metabolic supply and demand: aerobic glycolysis as a normal
physiological response to uctuating energetic demands in the membrane. Cancer Metab 2, 7, https://doi.org/10.1186/2049-3002-
2-7 (2014).
8. Simoes, . V. et al. Metabolic Plasticity of Metastatic Breast Cancer Cells: Adaptation to Changes in the Microenvironment.
Neoplasia 17, 671–684, https://doi.org/10.1016/j.neo.2015.08.005 (2015).
9. Viale, A., Corti, D. & Draetta, G. F. Tumors and Mitochondrial espiration: A Neglected Connection. Cancer Res. https://doi.
org/10.1158/0008-5472.CAN-15-0491 (2015).
10. Tsytsarev, V. et al. In v ivo imaging of epileptic activity using 2-NBDG, a uorescent deoxyglucose analog. J Neurosci Meth (2011).
11. Soolo, L. Localization of functional activity in the central nervous system by measurement of glucose utilization with radioactive
deoxyglucose. Journal of cerebral blood ow and metabolism: ocial journal of the International Society of Cerebral Blood Flow and
Metabolism 1, 7 (1981).
12. Porporato, P. E. et al. A mitochondrial switch promotes tumor metastasis. Cell reports 8, 754–766, https://doi.org/10.1016/j.
celrep.2014.06.043 (2014).
13. Li, P. Y. et al. edox homeostasis protects mitochondria through accelerating OS conversion to enhance hypoxia resistance in
cancer cells. Scientic reports 6, doi:Artn 2283110.1038/Srep22831 (2016).
14. Singh, B. et al. Highly Adaptable Triple-Negative Breast Cancer Cells as a Functional Model for Testing Anticancer Agents. PloS one
9, doi:ATN e10948710.1371/journal.pone.0109487 (2014).
15. Pavlides, S. et al. e reverse Warburg eect: aerobic glycolysis in cancer associated broblasts and the tumor stroma. Cell cycle 8,
3984–4001 (2009).
16. Sotgia, F. et al. Mitochondrial metabolism in cancer metastasis Visualizing tumor cell mitochondria and the “reverse Warburg eect”
in positive lymph node tissue. Cell Cycle 11, 1445–1454, https://doi.org/10.4161/cc.19841 (2012).
17. Haq, . et al. Oncogenic BAF regulates oxidative metabolism via PGC1alpha and MITF. Cancer Cell 23, 302–315, https://doi.
org/10.1016/j.ccr.2013.02.003 (2013).
18. Yoshida, G. J. Metabolic reprogramming: the emerging concept and associated therapeutic strategies. J Exp Clin Cancer Res 34, 111,
https://doi.org/10.1186/s13046-015-0221-y (2015).
19. Lehuede, C., Dupuy, F., abinovitch, ., Jones, . G. & Siegel, P. M. Metabolic Plasticity as a Determinant of Tumor Growth and
Metastasis. Cancer research 76, 5201–5208, https://doi.org/10.1158/0008-5472.CAN-16-0266 (2016).
20. amamonjisoa, N. & Acersta, E. Characterization of the Tumor Microenvironment and Tumor-Stroma Interaction by Non-
invasive Preclinical Imaging. Frontiers in oncology 7, 3, https://doi.org/10.3389/fonc.2017.00003 (2017).
21. James, M. L. & Gambhir, S. S. A molecular imaging primer: modalities, imaging agents, and applications. Physiological reviews 92,
897–965, https://doi.org/10.1152/physrev.00049.2010 (2012).
22. Cho, H. J. et al. Noninvasive Multimodality Imaging of the Tumor Microenvironment: egistered Dynamic Magnetic esonance
Imaging and Positron Emission Tomography Studies of a Preclinical Tumor Model of Tumor Hypoxia. Neoplasia 11, 247–U245,
https://doi.org/10.1593/neo.81360 (2009).
23. Glunde, . & Bhujwalla, Z. M. Metabolic Tumor Imaging Using Magnetic esonance Spectroscopy. Seminars in oncology 38, 26–41,
https://doi.org/10.1053/j.seminoncol.2010.11.001 (2011).
24. Gillies, . J. & Morse, D. L. In vivo magnetic resonance spectroscopy in cancer. Annu Rev Biomed Eng 7, 287–326, https://doi.
org/10.1146/annurev.bioeng.7.060804.100411 (2005).
25. Gimi, B. et al. Molecular imaging of cancer: Applications of magnetic resonance methods. P Ieee 93, 784–799, https://doi.
org/10.1109/Jproc.2005.844266 (2005).
26. Crouch, S. P. M., ozlowsi, ., Slater, . J. & Fletcher, J. e Use of Atp Bioluminescence as a Measure of Cell-Proliferation and
Cytotoxicity. J Immunol Methods 160, 81–88, https://doi.org/10.1016/0022-1759(93)90011-U (1993).
27. Wu, M. et al. Multiparameter metabolic analysis reveals a close lin between attenuated mitochondrial bioenergetic function and
enhanced glycolysis dependency in human tumor cells. Am J Physiol-Cell Ph 292, C125–C136, https://doi.org/10.1152/
ajpcell.00247.2006 (2007).
28. uznetsov, A. V. et al. Analysis of mitochondrial function in situ in permeabilized muscle bers, tissues and cells. Nature protocols 3,
965–976, https://doi.org/10.1038/nprot.2008.61 (2008).
29. Chen, V. et al. Bezielle Selectively Targets Mitochondria of Cancer Cells to Inhibit Glycolysis and OXPHOS. Plos One 7, doi:ATN
e3030010.1371/journal.pone.0030300 (2012).
30. Lynam-Lennon, N. et al. Altered Mitochondrial Function and Energy Metabolism Is Associated with a adioresistant Phenotype in
Oesophageal Adenocarcinoma. Plos One 9, doi:ATN e10073810.1371/journal.pone.0100738 (2014).
31. Wang, S. Y. et al. 2-Deoxy-D-Glucose Can Complement Doxorubicin and Sorafenib to Suppress the Growth of Papillary yroid
Carcinoma Cells. Plos One 10, doi:ATN e013095910.1371/journal.pone.0130959 (2015).
32. Zh ang , J. et al. Measuring energy metabolism in cultured cells, including human pluripotent stem cells and dierentiated cells.
Nature protocols 7, 1068–1085, https://doi.org/10.1038/nprot.2012.048 (2012).
33. TeSlaa, T. & Teitell, M. A. Techniques to Monitor Glycolysis. Method Enzymol 542, 91–114, https://doi.org/10.1016/B978-0-12-
416618-9.00005-4 (2014).
34. Wu, H. F., Southam, A. D., Hines, A. & Viant, M. . High-throughput tissue extraction protocol for NM- and MS-based
metabolomics. Anal Biochem 372, 204–212, https://doi.org/10.1016/j.ab.2007.10.002 (2008).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
14
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
35. Want, E. J. et al. Global metabolic proling of animal and human tissues via UPLC-MS. Nature protocols 8, 17–32, https://doi.
org/10.1038/nprot.2012.135 (2013).
36. ajaram, N. et al. Delivery ate Aects Uptae of a Fluorescent Glucose Analog in Murine Metastatic Breast Cancer. Plos One 8,
doi:ATN e7652410.1371/journal.pone.0076524 (2013).
37. Frees, A. E. et al. Delivery-Corrected Imaging of Fluorescently-Labeled Glucose eveals Distinct Metabolic Phenotypes in Murine
Breast Cancer. Plos One 9, doi:ATN e11552910.1371/journal.pone.0115529 (2014).
38. Yamada, ., Saito, M., Matsuoa, H. & Inagai, N. A real-time method of imaging glucose uptae in single, living mammalian cells.
Nature protocols 2, 753–762, https://doi.org/10.1038/nprot.2007.76 (2007).
39. Tsytsarev, V. et al. In vivo imaging of epileptic activity using 2-NBDG, a uorescent deoxyglucose analog. J Neurosci Meth 203,
136–140, https://doi.org/10.1016/j.jneumeth.2011.09.005 (2012).
40. Cai, H. W. & Peng, F. Y. 2-NBDG Fluorescence Imaging of Hypermetabolic Circulating Tumor Cells in Mouse Xenogra model of
Breast Cancer. J Fluoresc 23, 213–220, https://doi.org/10.1007/s10895-012-1136-z (2013).
41. Gottlieb, E., Vander Heiden, M. G. & ompson, C. B. Bcl-x(L) prevents the initial decrease in mitochondrial membrane potential
and subsequent reactive oxygen species production during tumor necrosis factor alpha-induced apoptosis. Molecular and cellular
biology 20, 5680–5689 (2000).
42. Frezza, C. et al. Metabolic proling of hypoxic cells revealed a catabolic signature required for cell survival. Plos One 6, e24411,
https://doi.org/10.1371/journal.pone.0024411 (2011).
43. Perry, S. W., Norman, J. P., Barbieri, J., Brown, E. B. & Gelbard, H. A. Mitochondrial membrane potential probes and the proton
gradient: a practical usage guide. Biotechniques 50, 98–+, https://doi.org/10.2144/000113610 (2011).
44. Frees, A. E. et al. Hyperspectral Imaging of Glucose Uptae, Mitochondrial Membrane Potential, and Vascular Oxygenation
Dierentiates Breast Cancers with Distinct Metastatic Potential In Vivo. in Biomedical Optics 2016, OSA Technical Digest (online),
paper C4A.6 (Optical Society of America, 2016).
45. Martinez, A. et al. Metaboloptics: Visualization of the tumor functional landscape via metabolic and vascular imaging. Oncotarget
In press.
46. Pelicano, H., Martin, D. S., Xu, . H. & Huang, P. Glycolysis inhibition for anticancer treatment. Oncogene 25, 4633–4646, https://
doi.org/10.1038/sj.onc.1209597 (2006).
47. Sengupta, D. & Pratx, G. Imaging metabolic heterogeneity in cancer. Mol Cancer 15, doi:Artn 410.1186/S12943-015-0481-3 (2016).
48. eid, M. A. & ong, M. Dealing with hunger: Metabolic stress responses in tumors. J Carcinog 12, 17, https://doi.org/10.4103/1477-
3163.119111 (2013).
49. Casado, P., Bilanges, B., ajeeve, V., Vanhaesebroec, B. & Cutillas, P. . Environmental stress aects the activity of metabolic and
growth factor signaling networs and induces autophagy marers in MCF7 breast cancer cells. Mol Cell Proteomics 13, 836–848,
https://doi.org/10.1074/mcp.M113.034751 (2014).
50. Palmer, G. M. et al. In vivo optical molecular imaging and analysis in mice using dorsal window chamber models applied to hypoxia,
vasculature and uorescent reporters. Nature protocols 6, 1355–1366, https://doi.org/10.1038/nprot.2011.349 (2011).
51. Moy, A. J. et al. Wide-eld functional imaging of blood ow and hemoglobin oxygen saturation in the rodent dorsal window
chamber. Microvasc Res 82, 199–209, https://doi.org/10.1016/j.mvr.2011.07.004 (2011).
52. Alexander, S., oehl, G. E., Hirschberg, M., Geissler, E. . & Friedl, P. Dynamic imaging of cancer growth and invasion: a modied
sin-fold chamber model. Histochem Cell Biol 130, 1147–1154, https://doi.org/10.1007/s00418-008-0529-1 (2008).
53. Estrella, V. et al. Acidity Generated by the Tumor Microenvironment Drives Local Invasion. Cancer Res 73, 1524–1535, https://doi.
org/10.1158/0008-5472.Can-12-2796 (2013).
54. Upputuri, P. ., Sivasubramanian, ., Mar, C. S. & Pramani, M. ecent developments in vascular imaging techniques in tissue
engineering and regenerative medicine. Biomed Res Int 2015, 783983, https://doi.org/10.1155/2015/783983 (2015).
55. Perelman, A. et al. JC-1: alternative excitation wavelengths facilitate mitochondrial membrane potential cytometry. Cell Death Dis
3, e430, https://doi.org/10.1038/cddis.2012.171 (2012).
56. Scaduto, . C. & Grotyohann, L. W. Measurement of mitochondrial membrane potential using uorescent rhodamine derivatives.
Biophys J 76, 469–477 (1999).
57. Pang, . S., Peter, . M. & odrigues, A. D. Drug-Drug Interactions: What Have We Learned and Where Are We Going? Enzyme-
and Transporter-Based Drug-Drug Interactions: Progress and Future Challenges, 701–722, doi:https://doi.org/10.1007/978-1-4419-
0840-7_28 (2010).
58. Zhang, H. J., Sinz, M. W. & odrigues, A. D. Metabolism-Mediated Drug-Drug Interactions. Drug Metabolism in Drug Design and
Development: Basic Concepts and Practice, 113–136 (2008).
59. odrigues, A. D. & Lin, J. H. Screening of drug candidates for their drug-drug interaction potential. Curr Opin Chem Biol 5,
396–401, https://doi.org/10.1016/S1367-5931(00)00220-9 (2001).
60. Perry, S. W., Norman, J. P., Barbieri, J., Brown, E. B. & Gelbard, H. A. Mitochondrial membrane potential probes and the proton
gradient: a practical usage guide. BioTechniques 50, 98–115, https://doi.org/10.2144/000113610 (2011).
61. Silva, A. S. et al. Evolutionary approaches to prolong progression-free survival in breast cancer. Cancer research 72, 6362–6370,
https://doi.org/10.1158/0008-5472.CAN-12-2235 (2012).
62. O’Neil, ., Wu, L. & Mullani, N. Uptae of a Fluorescent Deoxyglucose Analog (2-NBDG) in Tumor Cells. Molecular Imaging and
Biology 7, 388–392, https://doi.org/10.1007/s11307-005-0011-6 (2005).
63. Yoshioa, . et al. A novel uorescent derivative of glucose applicable to the assessment of glucose uptae activity of Escherichia coli.
Biochimica et Biophysica Acta (BBA) - General Subjects 1289, 5–9, https://doi.org/10.1016/0304-4165(95)00153-0 (1996).
64. im, J. W., Tchernyshyov, I., Semenza, G. L. & Dang, C. V. HIF-1-mediated expression of pyruvate dehydrogenase inase: A
metabolic switch required for cellular adaptation to hypoxia. Cell Metab 3, 177–185, https://doi.org/10.1016/j.cmet.2006.02.002
(2006).
65. Frees, A. E. et al. Delivery-corrected imaging of uorescently-labeled glucose reveals distinct metabolic phenotypes in murine breast
cancer. PloS one 9, e115529, https://doi.org/10.1371/journal.pone.0115529 (2014).
66. oppenol, W. H., Bounds, P. L. & Dang, C. V. Otto Warburg’s contributions to current concepts of cancer metabolism (vol 11, pg 325,
2011). Nat Rev Cancer 11, 618–618, https://doi.org/10.1038/nrc3108 (2011).
67. Tseng, J. C., Wang, Y., Banerjee, P. & ung, A. L. Incongruity of imaging using uorescent 2-DG conjugates compared to 18F-FDG
in preclinical cancer models. Molecular imaging and biology: MIB: the ocial publication of the Academy of Molecular Imaging 14,
553–560, https://doi.org/10.1007/s11307-012-0545-3 (2012).
68. Semenza, G. L. HIF-1: upstream and downstream of cancer metabolism. Curr Opin Genet Dev 20, 51–56, https://doi.org/10.1016/j.
gde.2009.10.009 (2010).
69. Jiang, B. H., Agani, F., Passaniti, A. & Semenza, G. L. V-SC induces expression of hypoxia-inducible factor 1 (HIF-1) and
transcription of genes encoding vascular endothelial growth factor and enolase 1: involvement of HIF-1 in tumor progression.
Cancer Res 57, 5328–5335 (1997).
70. G o e l, S. et al. Normalization of the vasculature for treatment of cancer and other diseases. Physiological reviews 91, 1071–1121,
https://doi.org/10.1152/physrev.00038.2010 (2011).
71. Ostergaard, L. et al. e relationship between tumor blood ow, angiogenesis, tumor hypoxia, and aerobic glycolysis. C ancer research
73, 5618–5624, https://doi.org/10.1158/0008-5472.CAN-13-0964 (2013).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
15
Scientific RepoRTs | 7: 13772 | DOI:10.1038/s41598-017-14226-x
72. Hughes, P. F. et al. A highly selective Hsp90 anity chromatography resin with a cleavable liner. Bioorg. Med. Chem. 20, 3298–3305,
https://doi.org/10.1016/j.bmc.2012.03.043 (2012).
73. Fu, H. L. et al. Optimization of a Wideeld Structured Illumination Microscope for Non-Destructive Assessment and Quantication
of Nuclear Features in Tumor Margins of a Primar y Mouse Model of Sarcoma. Plos One 8, doi:ATN e68868 https://doi.org/10.1371/
journal.pone.0068868 (2013).
74. Bui, A. . et al. evisiting Optical Clearing With Dimethyl Sulfoxide (DMSO). Laser Surg Med 41, 142–148, https://doi.org/10.1002/
lsm.20742 (2009).
75. Vishwanath, ., Yuan, H., Barry, W. T., Dewhirst, M. W. & amanujam, N. Using Optical Spectroscopy to Longitudinally Monitor
Physiological Changes within Solid Tumors. Neoplasia 11, 889–900, https://doi.org/10.1593/neo.09580 (2009).
76. ajaram, N., eesor, A. F., Mulvey, C. S., Frees, A. E. & amanujam, N. Non-invasive, simultaneous quantication of vascular
oxygenation and glucose uptae in tissue. PloS one 10, e0117132, https://doi.org/10.1371/journal.pone.0117132 (2015).
Acknowledgements
is work was supported by generous funding from the Department of Defense Era of Hope Scholar Award
(http://cdmrp.army.mil/bcrp/era; W81XWH-09-1-0410). e funders had no role in study design, data collection
and analysis, decision to publish, or preparation of the manuscript. We would like to thank Dr. Fan Yuan for
his helpful discussion on our 2-NBDG and TMRE imaging data. We would like to thank Dr. Alaattin Erkanli
for his assistance in selecting the statistical tests for our data analysis. We also would like to thank Megan C.
Madonna, Marianne Lee and Helen A. Murphy for their assistance during the animal imaging. Many thanks
to Dr. Jenna H. Mueller, Dr. Fangyao Hu, and Christopher T. Lam and for their generous help with the imaging
system development.
Author Contributions
Conception and design: C.Z., A.F.M., T.A.J.H., N.R. Development of methodology: C.Z., A.F.M., N.R. Acquisition
of data: C.Z., A.F.M., H.L.M., M.L., B.T.C., D.C. Analysis and interpretation of data: C.Z., A.F.M., H.L.M. Writing,
review and/or revision of the manuscript: C.Z., A.F.M., N.R. Administrative, technical, or material support: D.C.,
T.A.J.H. Study supervision: T.A.J.H., N.R.
Additional Information
Competing Interests: e authors declare that they have no competing interests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre-
ative Commons license, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons license and your intended use is not per-
mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
© e Author(s) 2017
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com