NIRF imaging is emerging as a powerful tool for noninvasive imaging diseases in pre-clinical models. It also has a great potential for clinical use in providing both real time surgical information and functional/molecular information of the disease. NIRF probes with capability of imaging specific molecular targets and events are therefore under active investigation. Our research was intended to develop NIR fluorescent glucose analogs for imaging tumor metabolism in living subjects. Although several NIR 2-deoxyglucose analogs which are able to accumulate in the tumors were successfully synthesized recently (24
), whether these compounds are delivered and trapped in tumor cells via the GLUTs/hexokinase pathway remains to be answered. Moreover, the ability of these NIR 2-deoxyglucose analogs to imaging tumor metabolism in small animals has not been demonstrated yet.
It was reported that 2-NBDG could be incorporated by glucose transporting systems in living cells. More recently, it was found that 2-NBDG can accumulate into malignant tumor cells such as the MCF-7 breast cancer cells and the HepG2 liver cancer cells (27
). In this work, the cell uptake of 2-NBDG by U87MG was also examined using fluorescence microscopy imaging. It was found 2-NBDG did accumulate in the cytoplasm of U87MG cells (). We also found this fluorescent glucose analog showed good uptake in several other tumor cell lines such as C6, A375M, B16F0, and MDA-MB-435 (data not shown). More importantly, uptake of 2-NBDG by U87MG was significantly inhibited by 10 mM and 50 mM d
< 0.05) but not by 50 mM l
-glucose (), strongly indicating that the uptake of 2-NBDG is likely mediated by glucose transporters. However, the short excitation and emission wavelength (Ex: 475 nm; Em: 550 nm) of 2-NBDG limits its applications for imaging small living subjects.
Figure 2 (A) Light (top panel) and fluorescence (bottom panel) images of U87MG cells. The cells were incubated with 10 µM 2-NBDG for 10 min at 37 °C in the absence of d or l-glucose (a and b), and presence of 50 mM l-glucose (b and e) and 50 mM (more ...)
Encouraged by these results, we hypothesized that conjugation of d-glucosamine with a carefully designed NIRF dye may generate a fluorescence deoxyglucose analog which can be still recognized by GLUTs and hexokinase. Therefore, Cy5.5-2DG was synthesized and evaluated in both cell culture and mice, because of the longer excitation/emission wavelengths of Cy5.5 and its easy availability. Synthesis of Cy5.5-2DG was achieved through conjugation of Cy5.5-NHS ester with the 2-amino group of the d-glucosamine. The desired products were purified by semi-preparative HPLC. The retention time on analytical HPLC for Cy5.5-2DG was found to be 15.2 minutes. The yield of Cy5.5-2DG conjugate was typically over 70%. The purity of Cy5.5-2DG was over 95% from analytical HPLC analysis. The purified Cy5.5-2DG was characterized by MALDI-TOF-MS. Cy5.5-2DG: m/z = 1078.26 for [M+H]+ (C47H56N3O18S4, calculated MW = 1078.24). The absorption and fluorescence emission characteristics of Cy5.5-2DG (λab = 675 nm, λem = 695 nm) were identical to those of free Cy5.5, as apparent from the spectra measured in H2O (). The stability of Cy5.5-2DG was further evaluated by incubation the compound with mouse serum for 1 hour. No dissociation of Cy5.5 from Cy5.5-2DG conjugate was found by using HPLC analysis, suggesting the good serum stability of Cy5.5-2DG.
Absorption (675 nm) and emission fluorescence (695 nm) spectra of Cy5.5-2DG in water.
Fluorescence microscopy imaging studies were performed to investigate the cell uptake of Cy5.5-2DG in different tumor cell lines. It was found that Cy5.5-2DG does accumulate in several tumor cell lines such as A375M, B16F10, C6, U87MG, and MDA-MB-435 at 37 °C, while the uptake at 4 °C was minimal. This result indicates that the cellular uptake of Cy5.5-2DG is due to an active and temperature dependent process, instead of simple passive diffusion. In order to further elucidate whether its cell uptake is related to GLUTs, Cy5.5-2DG was incubated with either 50 mM d
-glucose or l
-glucose. However, Cy5.5-2DG localizes within U87MG cells with similar fluorescence intensity in the presence and absence of 50 mM d
-glucose and l
-Glucose failed to block the cell uptake of Cy5.5-2DG, suggesting GLUTs are not likely the transporter system responsible for Cy5.5-2DG uptake. In order to determine whether Cy5.5-2DG is a substrate for hexokinase, Cy5.5-2DG was incubated with glucose reagent (Infinity Glucose Reagent; ThermoDMA, Louisville, CO) at 37 °C for 3 minutes (12
). From MALDI-TOF-MS analysis, we were not able to observe the formation of phosphorylated Cy5.5-2DG (data not shown). Cy5.5-2DG does not seem to mimic glucose/deoxyglucose phosphorylation. More studies are needed to pinpoint the uptake and retention mechanism of Cy5.5-2DG in cells.
For comparison, Cy5.5-NHS was used as a control for staining the tumor cell lines. Fluorescence microscopy imaging studies demonstrate that Cy5.5-NHS can label these tumor cell lines as well as Cy5.5-2DG at 37 °C, and also shows negligible fluorescence signal when incubated at 4 °C. In a previous study (28
), it was reported that very little Cy5.5-NHS accumulates in two breast cancer cell lines, MDA-MB-648 and MDA-MB-435. This is likely due to different filter settings and detection sensitivity of the fluorescence microscopes used in our study compared to the published study. In that report, a 775 nm/845 nm (excitation/emission) filter was equipped to obtain Cy5.5 fluorescence images. However a Cy5.5 filter set (Exciter, 650/20 nm; Emitter, 675/35 nm) was used in our study. Considering the maximum excitation and emission wavelength of Cy5.5 dye is 675 nm and 694 nm respectively, the 775 nm/845 nm filter set is not the proper setting, and one is unlikely to observe enough Cy5.5 fluorescence signal using such a filter.
Whole-body optical imaging of subcutaneous tumor xenografted mice was then performed by using an IVIS200 system to monitor in vivo biodistribution of Cy5.5-2DG and Cy5.5-NHS. As shown in , the subcutaneous U87MG tumor could be clearly distinguished from the surrounding background tissue from 30 min up to 24 h pi of either 0.5 nmol of Cy5.5-2DG or Cy5.5-NHS. Quantitative analysis of these images was performed, and the fluorescence intensities in the tumor and the normal tissues as a function of time for Cy5.5-2DG and Cy5.5-NHS are depicted in , respectively. It was found that both probes exhibited fast tumor targeting characteristics in vivo. The tumor uptake of Cy5.5-2DG and Cy5.5-NHS reached a maximum 30 minutes pi and slowly washed out over time. A similar pattern was found for the uptake of both probes in normal tissue. At all time points pi (0.5 h to 24 h), there is no statistical significant difference in tumor and normal tissue uptake between Cy5.5-2DG and Cy5.5-NHS (P > 0.05) (). The fluorescence intensity ratio between the tumor and normal tissue (T/N) for Cy5.5-2DG (represented by solid line) and Cy5.5-NHS (represented by dotted line) was calculated and shown in . Cy5.5-NHS shows significantly higher T/N ratio than Cy5.5-2DG at later time points (4 h and 24 h) (P < 0.05). For example, T/N ratio for Cy5.5-2DG, and Cy5.5-NHS are found to be 2.81 ± 0.10 and 3.34 ± 0.23, respectively, at 24 h pi. U87MG tumor tissues from the mice 24 h after administration of fluorescent probes were then taken out and sliced for NIRF imaging. Fluorescence and brightfield images for the tumor slices of mice injected with Cy5.5-2DG and Cy5.5-NHS were obtained. It was found that both probes accumulate in the tumor cells, and the fluorescence intensity is similar in both samples.
Figure 4 (A) In vivo fluorescence imaging of subcutaneous U87MG glioblastoma tumor bearing nude mice after intravenous injection of 0.5 nmol Cy5.5-2DG (top) or Cy5.5-NHS (bottom). The position of the tumor is indicated by an arrow. Fluorescence signal from probes (more ...)
The tumor targeting abilities in living subjects of Cy5.5-2DG and Cy5.5-NHS were also evaluated in A375M xenografted mouse model. shows typical NIR fluorescence images of scid mice bearing subcutaneous A375M melanoma tumor after intravenous (iv) injection of 0.5 nmol of probe. Similar as the distribution in U87MG mouse, both Cy5.5-2DG and Cy5.5-NHS show higher uptake in the tumor compared to normal tissue uptake. Moreover, significant difference in T/N is observed between Cy5.5-2DG and Cy5.5-NHS at 48 hours pi (P = 0.01). The T/N ratio for Cy5.5-2DG, Cy5.5-NHS at 48 hours are 1.63 ± 0.06 and 2.58 ± 0.36, respectively ().
Figure 5 (A) In vivo NIRF images of subcutaneous A375M melanoma bearing scid mice after intravenous injection of 0.5 nmol Cy5.5-2DG (top) and Cy5.5-NHS (bottom). The position of the A375M tumor is indicated by an arrow. Fluorescence signal from NIRF probes was (more ...)
The radioactive deoxyglucose analog, [18F]FDG was then used to image the nude mice bearing human U87MG using microPET. [18F]FDG preferentially localizes and is retained in U87MG tumors, as clearly shown in . The tumor/muscle ratios are 3.89 ± 0.99, 3.96 ± 0.39, and 4.08 ± 0.65, at 30 min, 1 h and 2 h pi of the probe. Obviously, the tumor contrast for [18F]FDG at these time points is significantly higher than that of Cy5.5-2DG (T/N ratios at 30 min to 2 h pi are all below 2, as shown in ). This is probably caused by: 1). conjugation d-glucose with a big fluorophor Cy5.5 resulting in a compound with reduced cell uptake and/or retention; 2) auto-fluorescence of normal tissue increasing the background signal. Furthermore, compared to optical imaging, [18F]FDG-microPET is able to quantify the probe accumulation in most organs of interest such as brain, heart, kidney, etc. However, except for the disadvantages as described in the introduction, PET imaging can only monitor the [18F]FDG distribution up to several half lives of F-18. While optical imaging can observe the Cy5.5-2DG NIR signal in living mouse even several days pi.
Figure 6 (A) Decay-corrected whole-body coronal microPET images of nude mouse bearing human U87MG tumor at 30 min, 1 h, and 2 h (5-min static image) after injection of 5.55 MBq (150 µCi) of [18F]FDG. The position of the U87MG tumor is indicated by an arrow. (more ...)
The small animal optical imaging data shows that both Cy5.5-2DG and Cy5.5-NHS can accumulate very well in U87MG and A375M tumor models. Even 4 days pi of the probes, the tumor showed the high fluorescent signal and could be clearly visualized (data not shown). Another preliminary study also revealed that Cy5.5-2DG and Cy5.5-NHS can localize in the MDA-MB-435 tumor in an orthotopic tumor model (data not shown). Although it is not clear how Cy5.5-2DG and Cy5.5-NHS localize and are retained in the tumor, their high specificity, high accumulation and retention in tumors will likely make them very useful in 1) labeling cancer cells for cell trafficking; 2) in vivo staining of tumor such as breast cancer, melanoma, and glioma for guidance of tumor surgery and radiation therapy in conjunction with proper optical imaging systems, etc. Moreover, conjugation of Cy5.5-NHS with cancer chemotherapeutical drugs or radionuclides such as F-18 may generate novel tumor targeting agents.
The in vivo tumor seeking ability of Cy5.5-NHS documented here is very different from the results reported previously (28
). In that report, the authors did not observe any significant difference between the uptake in the tumor region and normal tissue region after injection of Cy5.5-NHS. This inconsistency may be due to several factors such as the depth and size of tumor, different settings of the optical imaging system, and the amount of dye injected (2.9 nmol/mouse in their report vs 0.5 nmol/mouse in our study), etc. Our finding also highlight the importance of a complete separation of free Cy5.5-NHS and Cy5.5 labeled biomolecules (peptides or antibodies, etc) in order to avoid any false-positive results.
Based on our data reported here and previous reports (19
), 2-NBDG has been proven to be a substrate of GLUTs, while Cy5.5-2DG is not. Considering Cy5.5 is a much larger molecule than 2-[N
-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino] (NBD) moiety (), it is likely that size of the fluorophor in the 2 position is the reason that Cy5.5-2DG cannot be delivered and trapped in tumor cells via the GLUT/hexokinase pathway. It looks like the resulting conjugate no longer behaves like d
-glucose if the dye gets too large. These results clearly suggest the importance of selection of NIR dyes for preparation of NIRF d
In conclusion, the NIRF glucose analog, Cy5.5-DG and Cy5.5-NHS both demonstrate tumor targeting abilities in cell culture and in living mice but not through the GLUT/hexokinase pathway. More studies are warranted to further explore these two imaging probes for optical tumor imaging in cell culture and living subjects.