2.1.

Optical Glucose Probe Preparation

2-( $N$ -(7-nitrobenz-2-oxa-1, 3-diazol- 4-yl)amino)-2-deoxyglucose (2-NBDG) was commercially purchased (Invitrogen, California) and diluted to a stock solution of $1\mathrm{M}$ in phosphate-buffered saline (PBS). The stock solution was stored in a $-20°\mathrm{C}$ refrigerator.

2.2.

Cell Culture

To demonstrate the relationship between 2-NBDG concentration and cellular uptake, the human metastatic colorectal cancer cell line HT-29 (ATCC, Manassas, Virginia) was grown in McCoy’s media supplemented with 10% fetal bovine serum (FBS). The cells were grown in 24 well plates at $37°\mathrm{C}$ with 5% $\mathrm{C}{\mathrm{O}}_2$ . When the cells reached 90% confluence, they were washed twice with Hank’s buffered salt solution (HBSS) and incubated for one hour in $1\mathrm{mL}$ of growth media. 2-NBDG was then added to the media to achieve concentrations of 0, 0.1, 0.25, 0.5, or $1.0\mathrm{mM}$ in the wells. The experiment was performed in triplicate, with three wells used for each concentration. The cells were incubated for $30\min$ with the probe, then washed in HBSS for $10\min$ and imaged using a commercial fluorescence imaging system (Olympus Small Animal Imaging System OV100, Olympus Corporation, Tokyo, Japan). The images acquired by this imaging platform were analyzed using CellProfiler⁹ to automatically segment and define cells within the image, and the mean fluorescence per cell was calculated in this manner.

We then sought to demonstrate that cellular uptake of 2-NBDG could be inhibited by excess glucose in the growth media. For this experiment, the highly glycolytic human breast cancer cell line MCF-7 (ATCC, Manassas, Virginia) was grown in McCoy’s media supplemented with 10% FBS and 1% penicillin/streptomycin. The cells were grown in 24 well plates at $37°\mathrm{C}$ with 5% $\mathrm{C}{\mathrm{O}}_2$ . When the cells reached 90% confluence, they were washed twice with HBSS and incubated for one hour in $1\mathrm{mL}$ of media. A solution containing D-glucose dissolved in PBS was then added to the wells to reach final concentrations of excess media glucose of 0, 5, 10, or $20\mathrm{mM}$ in the wells. Along with the glucose solution was added $50\text{-}\mu \mathrm{M}$ 2-NBDG. The experiment was performed in triplicate, with three wells for each glucose concentration. The cells were incubated with the 2-NBDG and excess glucose for $30\min$ ; following this, they were washed in HBSS for $10\min$ and then imaged and analyzed using the OV100 imaging system and CellProfiler software, as described previously.

The glucose inhibition experiment was repeated using FDG to assess for a similar inhibitory pattern. The procedure for this experiment was as before, except that following the $10\text{-}\min$ HBSS wash, the cells were harvested with trypsin and centrifuged at 1000 revolutions per minute for $3\min$ . The supernatant was discarded, and the cells were resuspended in HBSS. They were then analyzed using a gamma counterplate reader (Wallac Wizard 1480 Automatic Gamma Counter; GMI, Ramsey, Minnesota). To correlate the FDG results with the 2-NBDG data, the fluorescence and gamma counter results were normalized to their respective values for the cells that were incubated with no excess media glucose, and the fold decrease with the addition of the three concentrations of glucose was calculated.

2.3.

Subcutaneous Tumor Imaging

The human colorectal cancer cell line HT-29 was grown as described. Before the cells reached confluence, they were harvested by trypsinization, resuspended in HBSS, and injected into the subcutaneous space in the left flank of $(n=4)$ athymic nude mice (nu/nu; Taconic, Germantown, New York). Once the tumors were greater than $100{\mathrm{mm}}^3$ , the change in 2-NBDG uptake by fasting was assessed in a crossover trial designed in the following manner. Half of the mice were kept nil per os (NPO) for $24\mathrm{h}$ , while the other half were allowed to feed ad libidum; all mice then received $500\mathrm{nmol}$ of 2-NBDG via tail vein injection and then underwent multispectral fluorescence imaging (Maestro In-Vivo Imaging System, Cambridge Research and Instrumentation, Woburn, Massachusetts) thirty minutes later. Prior to 2-NBDG imaging, a dark fluorescence background was acquired as well. The mice were then allowed to feed freely for two days. Following this, the two groups were reversed: the original group of mice that had been kept NPO were allowed to feed ad libidum, while the other half were kept NPO for $24\mathrm{h}$ . After $24\mathrm{h}$ , $500\mathrm{nmol}$ of 2-NBDG were then injected via tail vein in all animals, and multispectral fluorescence imaging was again performed on the mice.

The fluorescence spectrum corresponding to the 2-NBDG signal within the tumor was isolated by removing the spectral contributions of the mouse skin and imaging plate autofluorescence and reflection. Regions of interest (ROIs) were drawn within the tumor in the 2-NBDG spectral image to quantify the change in 2-NBDG signal between the animals’ fed and fasting states.

2.4.

Seizure Model Imaging

Generalized seizures were induced in a mouse using a previously described method.^{10, 11} Seizures were generated by the intraperitoneal (i.p.) injection of $240\text{-}\mathrm{mg}∕\mathrm{kg}$ pilocarpine in a nude mouse; a control arm was established with the i.p. injection of saline in another mouse. Ten minutes prior to pilocarpine injection, $30\mathrm{mg}∕\mathrm{kg}$ of scopolamine was injected subcutaneously to prevent peripheral cholinergic activity. Within minutes of the pilocarpine injection, the animal was noted to demonstrate irregular muscle contractions and rapid eye blinking behavior. Both the seizure model animal and the control animal were administered $300\mathrm{nmol}$ of 2-NBDG i.p. 30 min after the pilocarpine/saline. One hour following this, the animals were sacrificed. Their brains were harvested and homogenized, and epifluorescence imaging of 2-NBDG was performed using the OV100 system. ROIs were drawn within the images of the cuvettes containing the homogenized brains to quantify the difference in fluorescence signal intensity between the seizure model brain and the control animal brain.

2.5.

Window Chamber Tumor Imaging

The rodent gliosarcoma cell line 9L (ATCC, Manassas, Virginia) was stably transfected with an expression vector encoding red fluorescent protein (RFP) using the SuperFect transfection reagent according to the manufacturer’s protocol (Qiagen, Valencia, California). This cell line was maintained in RPMI media supplemented with 10% FBS and 1% penicillin/streptomycin. When the cells reached 90% confluence, they were harvested by trypsinization and injected into a dorsal window chamber in a nude mouse surgically implanted by a previously published method.¹² After $10\text{days}$ , intravital laser scanning fluorescent microscopy was performed using a miniaturized four-laser scanning microscope as previously described.¹³ Excitation and emission filter settings were optimized for 2-NBDG and RFP. Prior to 2-NBDG administration, image stacks were acquired in the RFP and 2-NBDG channels to define the tumor location and measure tumor autofluorescence in the 2-NBDG channel, respectively. Following this, $500\mathrm{nmol}$ of 2-NBDG were injected i.p. and serial image stacks were acquired over one hour. The same photomultiplier tube gain settings, window, and level were used to collect and display the images. ROIs were drawn in the 2-NBDG channel images to quantify the change in signal intensity over the time course of the experiment.

2.6.

Quantitative Colonoscopy

Colorectal adenomas were generated in a mouse (balb/c; Charles River Laboratories, Wilmington, Massachusetts) by the i.p. injection of the chemical carcinogen azoxymethane (AOM) twice weekly for three weeks at a dose of $10\mathrm{mg}∕\mathrm{kg}$ . The animal was then imaged using a quantitative fluorescence colonoscopy system.^{14, 15} A discrete adenomatous lesion was identified, and a baseline autofluorescence measurement was acquired. The animal then was administered $500\text{-}\mathrm{nmol}$ 2-NBDG by tail vein injection, and the 2-NBDG signal intensity from the tumor was collected for $60\min$ .

2.7.

Metastatic Lymph Node Imaging

The rodent metastatic prostate carcinoma cell line JHU-31 (ATCC, Manassas, Virginia) was maintained in RPMI 1640 growth media supplemented with $250\text{-}\mathrm{nM}$ dexamethasone and 10% FBS. Before the cells reached confluence, they were harvested by trypsinization and injected into the dorsum of the right hind paw of an athymic nude mouse (nu/nu; Taconic, Germantown, New York). Two weeks after implantation, $500\mathrm{nmol}$ of 2-NBDG was administered via tail vein. Thirty minutes following this, the animal was placed under anesthesia with 2% isoflurane; the skin covering the hind paws was retracted, and the popliteal lymph nodes were identified. The 2-NBDG signal from both lymph nodes was imaged using a commercial epifluorescence system (bonSAI; Siemens Medical Solutions, Malvern, Pennsylvania).

2.8.

Intraperitoneal Tumor Imaging

Peritoneal wall tumors were generated in $(n=5)$ athymic nude (nu/nu; Taconic, Germantown, New York) mice in the following manner. The animals were placed under anesthesia with 2% isoflurane, and a $1\text{-}\mathrm{cm}$ transverse incision was made in the abdomen at the level of the umbilicus. The fascial and muscle layers were retracted, and HT29 human colorectal cancer cells, grown as before and harvested by trypsinization, were injected directly into the peritoneal wall. Two weeks following this procedure, FDG-PET imaging was performed on the mice (Inveon PET-CT, Siemens); the animals were kept NPO for $4\mathrm{h}$ prior to imaging. Two days later, the animals received $500\mathrm{nmol}$ of 2-NBDG by tail vein injection, and after thirty minutes, the animals underwent intravital epifluorescence imaging with the OV100 system following surgical exposure of the tumors; the animals were kept NPO for $4\mathrm{h}$ prior to this imaging as well. The tumors were then excised and stored in 70% ethanol for subsequent histologic validation.

2.9.

Statistical Analysis and Interpretation

The data in this study are presented as $\text{means}\pm \text{standard}$ errors. Statistical analysis was performed using the Student’s $t$ test, and significance was assigned for $p<0.05$ . The evolution of the 2-NBDG concentration within the tumor imaged with quantitative colonoscopy imaging was determined by drawing ROIs within the tumor and measuring the mean tumor signal intensity over the course of the experiment; the same ROI was used for all images. Target-to-background ratios (TBRs) for the FDG-PET and 2-NBDG imaging were calculated by dividing the mean signal intensity within the tumors by the mean signal intensity within the erector spinae muscles. All ROIs were drawn using the ImageJ software platform.

3.1.

In Vitro Evaluation of Malignant Cell Uptake of 2-NBDG

Our evaluation of 2-NBDG as a practical optical reporter of glucose metabolism begins with an investigation of its characteristics on the cellular level. As we show in Figs. 1a and 1b, and as has been shown previously,⁶ 2-NBDG uptake is concentration-dependent, with increased intracellular transport occurring with increased extracellular concentrations of the molecule. Moreover, we were able to replicate prior results demonstrating that 2-NBDG uptake can be inhibited by excess D-glucose in the growth media.⁷ However, we present for the first time a comparison of 2-NBDG versus FDG uptake inhibition by glucose, performed under identical experimental conditions. As Fig. 1c illustrates, the decrements in the two molecules’ uptakes are very similar for the three concentrations of excess glucose we evaluated. These results suggest that not only are 2-NBDG and FDG transported into the cell by identical mechanisms, but their retention within the cell, presumably by 6-phosphorylation, is also conserved.

Fig. 1

In vitro evaluation of 2-NBDG cellular uptake. (a) Fluorescence microscopy of 2-NBDG uptake in HT-29 cells after incubation in $1.0\text{-}\mathrm{mM}$ 2-NBDG for $30\min$ . (b) Cellular uptake of 2-NBDG by HT-29 cells is concentration dependent, with uptake increasing monotonically as the concentration of 2-NBDG in the culture media increases. (c) Correlation of FDG and 2-NBDG uptake inhibition by excess glucose. Transport of both 2-NBDG and FDG is reduced into MCF-7 cells by excess glucose in the growth media, with both molecules showing similar decrements in intracellular uptake as the extracellular glucose concentration is increased.

3.2.

Macroscopic Imaging of 2-NBDG Uptake into Highly Metabolically Active Tissue

We then sought to determine whether the glucose-sensitive uptake of 2-NBDG was extendible and could be visualized using a whole animal imaging modality. For this experiment, we investigated the uptake of 2-NBDG in highly glycolytic tumors implanted subcutaneously in mice using multispectral fluorescence imaging, and we measured the change in uptake between a normal fed state and a fasting, “glucose-starved” state. Figures 2a and 2b show the marked, statistically significant $(p<0.05)$ increase in 2-NBDG signal in the animals’ fasting states versus their fed states, implying that, similar to clinical observations in FDG-PET imaging, 2-NBDG uptake accurately reflects tissue utilization of glucose.

Fig. 2

Assessment of 2-NBDG as a marker of glucose metabolism in animals. (a) Multispectral fluorescence imaging of HT-29 subcutaneous tumors in nude mice $40\min$ following administration of $500\mathrm{nmol}$ of 2-NBDG in fed versus fasting states. Red arrows indicate the location of the tumor. There is a noticeable increase in 2-NBDG signal after the animal was kept NPO for $24\mathrm{h}$ . (b) Analysis of $(n=4)$ animals with subcutaneous tumors imaged, while fed and fasting shows a statistically significant $(p<0.05)$ increase in 2-NBDG uptake in fasting animals. (c) Statistically significant $(p<0.05)$ increase in 2-NBDG signal in the brain of an animal experiencing generalized seizures induced by injection of pilocarpine i.p. versus control. These data suggest that 2-NBDG localization reflects true glucose utilization rather than nonspecific extravasation.

Moreover, we illustrate in the prior experiment the ability of 2-NBDG imaging to accurately map highly metabolic areas, as seen by the sharp definition of the subcutaneous tumor against a benign background. We expanded on this observation by asking whether the molecule exhibits greater accumulation within nonmalignant tissue during episodes of increased activity through the imaging of 2-NBDG signal within the brain of an animal experiencing generalized seizures. We present these data in Fig. 2c, where we demonstrate a significant $(p<0.05)$ increase in 2-NBDG accumulation in the brain of the convulsing animal compared to control.

The ability of 2-NBDG imaging to precisely demarcate highly metabolically active tissue was further explored by high-resolution intravital laser scanning fluorescent microscopy, performed on a 9-L gliosarcoma tumor grown within a window chamber and stably transfected to constitutively express RFP (Fig. 3 ). The 2-NBDG molecule accumulates in the tumor over time, with the probe still visible within the vasculature after $10\min$ . However, by one hour postinjection, the 2-NBDG signal is confined to only the tumor itself, as defined by the RFP signal in the left-most panel of Fig. 3a. These data corroborate the previously presented cellular and whole animal imaging data that support 2-NBDG as a specific marker of glucose uptake.

Fig. 3

Intravital laser scanning fluorescent microscopy of an RFP-labeled 9-L gliosarcoma tumor implanted in a dorsal window chamber of a nude mouse. (a) The left-most image illustrates the location of the tumor based on its constitutive expression of RFP. The second left-most image shows the tumor autofluorescence in the 2-NBDG channel, and the subsequent images in the panel show accumulation of 2-NBDG in the tumor over time. By $60\text{-}\min$ postinjection, the 2-NBDG signal correlates very well with the RFP signal. (b) A plot of the 2-NBDG signal intensity in the tumor over time shows that the molecule accumulates and reaches a steady plateau after approximately $20\min$ .

3.3.

Clinically Relevant Imaging Applications of 2-NBDG

Having validated 2-NBDG imaging’s utility for the visualization of highly metabolic processes from the micro- to the macroscale, we then evaluated several different clinically relevant applications for this imaging probe. We first applied this molecule to the minimally invasive metabolic imaging of colorectal cancer (Fig. 4 ). Using an inhouse-designed quantitative fluorescence colonoscopy system,^{14, 15} we imaged 2-NBDG uptake over time in a chemical carcinogen-induced murine model of colorectal cancer. As Fig. 4b illustrates, the 2-NBDG concentration in the tumor initially increased to a sharp peak, a characteristic we interpret as representing the fluorescent probe in the blood pool phase, and then settled to a level that was constant after one hour, a finding that we believe represents true accumulation and retention of the probe within the tumor itself.

Fig. 4

Minimally invasive and intraoperative applications of 2-NBDG imaging. (a) and (b) depict data from a quantitative fluorescence colonoscopy performed on a mouse with chemical carcinogen-induced orthotopic colorectal adenomas. (a) White light (WL) image of an adenoma that was imaged using the quantitative catheter-based imaging system for one hour continuously, following the administration of $500\mathrm{nmol}$ of 2-NBDG. (b) Plot of 2-NBDG signal over time in a single adenoma shows an initial peak followed by a steady plateau. The peak likely represents 2-NBDG initially within the tumor vasculature, and the subsequent plateau reflects true accumulation and trapping of 2-NBDG within the tumor cells. (c) Epifluorescence imaging of popliteal lymph nodes in a mouse injected with JHU-31 metastatic prostate cancer cells in its right hind paw, $30\min$ after the administration of $500\mathrm{nmol}$ of 2-NBDG. Red arrows indicate the location of the lymph nodes. The lymph node ipsilateral to the tumor is noticeably larger in the WL image than the contralateral node; fluorescence imaging shows increased 2-NBDG uptake within the lymph node, a finding suggestive of metastatic spread.

Next, we investigated the intraoperative applicability of 2-NBDG imaging through two animal models. In the first, we implanted the rodent metastatic prostate cancer cell line JHU-31 in the right hind paw of a nude mouse and assessed for metastatic spread to nearby lymph nodes [Fig. 4c]. We were able to readily discriminate between 2-NBDG uptake in a popliteal lymph node ipsilateral to the implanted tumor versus uptake in the contralateral node. In the second (Fig. 5 ), we surgically generated a model of a peritoneal wall tumor and demonstrated first with FDG-PET/CT imaging the location and highly metabolic nature of these lesions. We then surgically exposed the tumors and performed intravital epifluorescence imaging. This imaging revealed the effectiveness of 2-NBDG to serve as a molecular beacon of pathologically elevated metabolic activity, highlighting the tumors against the benign abdominal viscera background. The TBRs for the FDG-PET and 2-NBDG imaging [Fig. 5b] were statistically similar (FDG-PET TBR 4.1 [CI: 1.1–7.0]; 2-NBDG TBR 3.4 [CI: 2.8–4.0]), and the malignant nature of the lesions was confirmed by histology [Fig. 5d].

Fig. 5

Correlation of FDG-PET imaging with intraoperative 2-NBDG imaging. HT29 cells were surgically implanted into the peritoneal wall of $(n=5)$ nude mice. The mice subsequently underwent imaging with FDG-PET/CT, and then two days later, underwent surface reflectance imaging $30\min$ following the administration of $500\mathrm{nmol}$ of 2-NBDG and surgical exposure of their abdomen. (a) FDG-PET/CT reveals two large intraperitoneal masses, shown with red arrows, that are highly FDG-avid in this representative mouse. (b) The same two lesions were well visualized with 2-NBDG imaging. Red arrows demonstrate the location of the tumors, and the blue box in the left-most image shows the approximate area of focus for the center and right-most images. (c) TBRs for FDG-PET and 2-NBDG imaging were statistically similar. (d) The malignant nature of the lesions was confirmed by histology.