Glucose deprivation is known to gradually increase total cellular transporter proteins [19, 20] and is also associated with decreased protein turnover in mammalian fibroblasts . Starved mammalian cells under low glucose concentration conditions undergo a p53 dependent G1 phase arrest that is quickly reversible upon glucose restoration . Fibroblasts have been showed to have a 24-hour survival capacity starting from 0.1 mM of glucose , which is in agreement with our results. Highest 18FDG uptakes were always obtained under a 2 mM glucose concentration, but such a low glucose concentration affected HUVEC morphology (Figure 4C-D). The spheroid HUVEC appearance observed under 1.5 mM glucose concentration can be due to cell-cell detachment caused by F-actin conversion in G-actin rather than being a consequence of cell death induction . However, not every preconditioning treatment susceptible to influence cell morphology can be considered for imaging cell or tissue substitutes.
When working with HUVEC, the optimal 18FDG uptake, while still maintaining cell integrity, was achieved using 3 mM glucose containing DMEM. The 3 mM glucose concentration also corresponded to the uptake plateau for fibroblasts. Hence, this glucose concentration was used for the cell density gradient experiments shown in Figure 5e and 5f and analysed in Figure 6B. Nearly all cell culture media contain a glucose concentration of at least 5.5 mM, which can be considered as a tremendous source of carbohydrate able to last for days . Most cells do not need that much glucose and quickly become saturated, making them produce and excrete lactate . Lowering the glucose level to 3 mM for 2 hours increased 18FDG uptake to approximately twice that found when using 5.5 mM media for both HUVEC and fibroblasts. Endothelial cells were shown to present different glucose metabolism and insulin responsiveness according to their organ of origin, so caution should be exercised when applying these results to other cell types .
The FBS supplemented M199 medium used for the cell culture contained 4 mM glucose, but since it was used at a concentration of 1% in the cell starvation media, this amount of added glucose was considered as non significant. Hiraki et al.  reported that glucose transport is also regulated by calf serum growth factors in a concentration-dependent manner. Considering that serum induced a first rise of sugar uptake within 10 minutes and a second at approximately 1 hour due to the activation of glucose transporter gene expression , it might be possible to gain extra signal by using 15% FBS containing DMEM before adjusting the glucose level to 3 mM.
Over-expression of insulin-receptors in HUVEC showed the presence of a functional insulin pathway . A small increase in 18FDG signal had been noticed in our data when using 10-8 M insulin, but this effect turned out not to be statistically significant. Insulin action might not be a major actor in our system, but anything that could bring some improvement is always welcome, so a 10-8 M physiological concentration was preserved in our protocol, as suggested by the maximum insulin effect observed by Gerritsen et al. [13, 27]. It must also be kept in mind for further studies that tissues such as heart, skeletal muscles and adipose tissues do present the insulin responsive glucose transporter GLUT4 [28–30], so even if our results with fibroblasts (known to have GLUT1, 3 and 4 ) turned out not to be significant, we strongly suggest that the insulin parameter always be tested.
Now that a protocol to maximize 18FDG cell uptake has been established, further studies are being planned to investigate additional parameters that are known to influence glucose uptake in cells, such as the presence of nitric oxide [12, 24, 26], growth factors , hypoxia , and proliferation.
More imaging studies will be needed to fully understand the importance of these factors in high cell density cultures, and PET imaging offers considerable potential to achieve this goal. Numerous key parameters must be dynamically monitored in real time in tissue cultures to optimize their development, which include morphology, viability, proliferation, metabolism, angiogenesis, perfusion, nutrient and oxygen consumption, hypoxia, apoptosis, and sometimes secretion of specific proteins. So far, only 18FDG, a cell glycolytic activity marker, has been investigated, but several other PET tracers are available, such as 18F - fluorothymidine (18FLT) and 11 C-methionine to, respectively, monitor DNA and protein synthesis, 18 F-fluoromisonidazole (18 F-MISO) for imaging hypoxia , and 18F- or 64Cu-labeled annexin-V for measurement of apoptosis [35, 36]. It would also be possible to adapt the proposed protocol for non-adherent (floating) cells by using microtubes for imaging instead of the square culture chambers. A centrifugation step would then be required prior to media removal for rinsing. Since radiotracers can sometimes bind to the microtube's wall, it may be advisable to transfer the suspension in a fresh tube after every rinse to avoid contamination from the container in the PET images.
Extension of the protocol used in this study to other cell types would be straightforward, provided that they have a similar growth rate and 18FDG delivery times. Due to the 109.8 minutes half-life of fluorine-18, the used experimental set-up with 18FDG could hardly be feasible under longer working conditions than the 12-h protocol used here. Obviously, other biological parameters could be monitored over extended observation periods with possibly longer incubation times using molecular probes labelled with longer half-life radiotracers, such as 64Cu (12.8 h), 89Zr (78 h) or 124I (4.18 d).