Glucosamine and N-acetyl glucosamine as new CEST MRI agents for molecular imaging of tumors

The efficacy of glucosamine (GlcN) and N-acetyl glucosamine (GlcNAc) as agents for chemical exchange saturation transfer (CEST) magnetic resonance molecular imaging of tumors is demonstrated. Both agents reflect the metabolic activity and malignancy of the tumors. The method was tested in two types of tumors implanted orthotopically in mice: 4T1 (mouse mammary cancer cells) and MCF7 (human mammary cancer cells). 4T1 is a more aggressive type of tumor than MCF7 and exhibited a larger CEST effect. Two methods of administration of the agents, intravenous (IV) and oral (PO), gave similar results. The CEST MRI observation of lung metastasis was confirmed by histology. The potential of the clinical application of CEST MRI with these agents for cancer diagnosis is strengthened by their lack of toxicity as can be indicated from their wide use as food supplements.

CEST agents, D-glucose showed inferior sensitivity and shorter duration of the effect 19,20 owing to its rapid metabolism in the tumors. Both FDG and 2-DG enter the tumor cells through GLUT family members where they and their phosphorylated products accumulate, but they may be toxic at high concentrations. The non-metabolizable 3OMG may be a better CEST MRI alternative 21 , but detailed studies of its toxicity are lacking. In the present work the non-toxic GlcN and GlcNAc are tested as potential candidates for tumor imaging by CEST MRI.

Results
In vitro. In vitro CEST NMR experiments were performed to test the ability of the method to detect millimolar concentrations of GlcN and GlcNAc under physiological conditions. Figure 1 presents the % CEST plots at different RF saturation fields (B 1 ) for solutions of 20 mM GlcN and GlcNAc at 37 °C and pH 7.4, corresponding to the intracellular conditions. Both GlcN and GlcNAc showed a significant CEST effect around a frequency offset of 1.2 ppm from water that originated from the hydroxyl groups connected to carbons numbers 3 and 4. At this offset, and for B 1 of 2.4 μ T, the CEST effects for the 20 mM solutions of GlcN and GlcNAc at pH 7.4 amounted to 7% and 14% respectively. Considering the proton concentration in water, these results correspond to an enhancement by factors of 350 and 700, respectively, over the direct detection of the metabolites. As can be seen from the pH dependence shown in Fig. 1c,d, the CEST effect is significantly larger at lower pH values, which may be pertinent to the extracellular microenvironment of tumors known to be slightly acidic.
GlcN and GlcNAc showed significant CEST effects through the amino protons at a frequency offset of ~3 ppm and the amido protons at frequency offset of ~3.5 ppm (Fig. 1a,b). However these CEST effects were lower than that observed at frequency offset of 1.2 ppm.
In vivo. The in vivo experiments were carried out on two widely-used breast cancer xenograft models, 4T 1 tumors (aggressive mouse breast cancer cells) and MCF7 tumors (human breast cancer cells), implanted in 8-week-old female BALB/C and SCID/ICR mice, respectively. Two methods of administration of the CEST agents were tested: intra-venous through the tail vein (IV), and per-os by oral gavage (PO). The tumors in both 4T 1 and MCF7 models showed significant CEST effects lasting for more than an hour following administration of GlcN or GlcNAc (Figs 2, 3, 4 and 5).
In the 4T 1 model, PO administration of 0.38-2.0 g/kg GlcN increased the CEST signals of the tumors by 4-12% relative to the baseline (Fig. 2d). A one-way ANOVA was used to test for differences in the obtained % CEST among GlcN doses, showing statistically significant differences (P < 0.0001); Tukey's post hoc test showed that the differences are between 5-6 groups of doses. The relatively large CEST signal at baseline, 9.9 ± 0.6% (mean ± standard deviation (STD); n = 26) is probably due to the presence of other metabolites, such as the glycosaminoglycans 22,23 and poly-sialic acid residues of mucoproteins known to be abundant in many tumors 24,25 . However, following GlcN administration, there was a substantial increase in the CEST effect; for example, GlcN at a dose of 1.1 g/kg raised the CEST effect from a baseline of 10.4% to 20.2%, a 9.8% increase (Fig. 2c). IV administration of GlcN to mice carrying 4T 1 tumors showed the same effect as those for PO administration.
CEST MRI experiments were performed in 6 SCID-ICR mice carrying the human MCF7 cell line treated with a dose of 1.0 g/kg GlcN, 3 administered by PO and 3 by IV (Examples are shown in Fig. 3). The CEST signals of MCF7 tumors increased by 5.5 ± 1.1% (n = 3, P = 0.014) for the PO experiments and by 5.4 ± 0.8% (n = 3, P = 0.007) for the IV experiments. Signal changes were also seen in the urinary bladder, as expected, and no significant changes were observed in other organs. The % CEST obtained in the tumors rose rapidly during the first 10 mins after treatment (Fig. 3d,h) by both modes of delivery, followed by steady state persistence of the CEST effect for more than an hour. It is interesting to note that the CEST effects for the MCF7 tumors were consistently lower than those for the 4T 1 tumors. This is compatible with the less aggressive nature of the MCF7 tumors as seen from the time of growth of about 30 days compared to 10 days for the 4T 1 tumors.
CEST MRI experiments were performed in 10 Balb/C mice carrying 4T 1 tumors treated with a dose of 1 g/kg GlcNAc, 5 by PO and 5 by IV (Examples are shown in Fig. 4). The CEST percent of 4T 1 tumors increased by 5.9 ± 0.7% (n = 5, p < 0.0001) for the PO experiments and by 6.1 ± 0.9% (n = 5, p < 0.0001) for the IV experiments. The tumor in Fig. 4a occupies two areas and that in Fig. 4e three areas. In both cases the animal was treated with GlcNAc (1.0 g/kg PO and 1.1 g/kg IV, respectively). The CEST effect was observed in the whole tumor ( Fig. 4c,g), thus most of the tumors appear to be active.
An example showing the detection of metastasis by CEST MRI following the administration of 1.0 g/kg GlcN is shown in Fig. 5. This experiment was done in Balb/C mice bearing 4T 1 tumor, in their primary advanced stage (~14 days after injection of cells). These tumors are known to metastasize from the primary tumor in the mammary gland to multiple distant sites including lymph nodes, blood, liver and lungs 26 . The GlcN CEST MRI experiment points to three areas suspected for metastasis in the liver and in the lower portion of the right lung (as indicated by arrows in Fig. 5c). The % CEST at these points was 12-13% above the baseline (Fig. 5b). The presence of metastasis was confirmed by histology (Fig. 5d,e).

Discussion
The results of this study demonstrate the ability of CEST MRI using the amino monosaccharides GlcN and GlcNAc to serve as contrast agents to detect tumors. The validity of the method was established by the findings that both PO and IV administration of GlcN or GlcNAc to mice bearing 4T 1 and MCF7 implanted breast tumors significantly enhanced the CEST MRI contrast of the tumors. The improved CEST MRI images started within a few minutes of GlcN or GlcNAc administration and persisted for over an hour (Figs 3d,h and 4d,h); the mice appeared to be unaffected by the treatment.
The larger CEST effect obtained for the more aggressive 4T 1 tumors points to the technique as a possible indicator of the aggressiveness of the tumor, something that cannot be obtained by standard MRI experiments. The in vitro CEST signal produced in solutions of GlcNAc was higher than that produced in GlcN solutions (Fig. 1). On the other hand, the CEST contrast observed for 4T 1 tumors treated with GlcN was slightly higher than that observed with GlcNAc (Fig. 2b,c vs. Fig. 4b,c). This may be due to greater accumulation of GlcN in the tumor. It should be noted that not only the accumulated GlcN and GlcNAc are responsible for the observed CEST effects but some of their metabolic products 4,15-16 may also contribute to the effect. This phenomenon is presently under investigation.
A major advantage of these agents is their lack of toxicity, which makes them suitable for human clinical diagnosis. Indeed, they are being used in the treatment of osteoarthritis [27][28][29][30] and inflammatory bowel disease 31 . The reason for their use for osteoarthritis is the fact that GlcN and GlcNAc are precursors for glycosaminoglycans (GAGs) molecules, that are major components of joint cartilage. In addition to the clinical studies leading to the conclusion that GlcN and GlcNAc have little if any potential for harm [31][32][33] , human breast milk contains up to 1,500 milligrams of GlcNAc per liter 34 , which means that a breast-feeding newborn may consume as much as 900 mg of GlcNAc a day.  Oral administration of GlcN at very high doses (5-15 g/kg body weight) was found to be well tolerated in humans 4 . As GlcN and GlcNAc have been widely prescribed as a nutritional supplement for therapeutic usage, it should be possible to perform oral administration in the clinic. Indeed, GlcN is usually taken orally and in humans 90% is absorbed 35,36 .
Significant CEST MRI contrast was achieved in our study when the animal was treated with a dose of 380 mg/kg GlcN, PO (Fig. 2d). The equivalent human oral dose would be about 2 g per 70 kg (calculation based on the ratio of mouse to human surface area). Human plasma pharmacokinetics showed that when 3 g of glucosamine sulphate were administered orally to healthy volunteers, the GlcN plasma concentration reached its maximum of about 14 mM at about 3.5 hours, and declined to half the maximum at about 8 hours from administration 37 . This should give ample time to perform the CEST MRI examination.
An advantage of both GlcN and GlcNAc over D-glucose is their suitability for diabetic patients; some studies suggest that they have no significant effects on insulin or blood glucose values 38-44 . Our results were performed using cw irradiation at 1.2 ppm from the water peak on a 7 T MRI scanner. Human 7 T MRI scanners are not used in clinical practice at the present time. The proximity of the irradiation frequency to the water peak may impede the application of our technique using MRI scanners with lower fields, such as 3 T. Further development of the CEST technique is required to cope with this problem. The use of the amine and amide proton peaks of GlcN and GlcNAc that occur at 3.0 and 3.5 ppm from the water peak should also be explored; while they give smaller CEST effect than the peak at 1.2 ppm, their effect at lower field strengths is not known.
In summary, both GlcN and GlcNAc have demonstrated good tumor targeting properties with in vivo CEST MRI. This new modality offers good resolution, and sensitivity sufficient to detect metabolic changes. The method, which requires no additional equipment, can replace PET/CT or PET/MRI as a standard imaging modality for the detection of tumors and their metastases as well as their response to therapy. The method is especially suitable for diabetics as GlcN and GlcNAc do not appear to affect blood glucose levels or insulin sensitivity. The advantage of CEST MRI of glucose and its analogs over other MRI modalities is that it reflects metabolic changes that often occur with no apparent morphological changes, enabling clinicians to stage the disease at diagnosis.
In vitro studies. Solutions of GlcN sulfate and GlcNAc were prepared in 10 mM phosphate-buffered saline (PBS) containing 10% D 2 O for NMR field/frequency locking.
NMR Spectroscopy. NMR spectra were acquired on a Bruker DRX spectrometer (Germany), equipped with a 11.7 T wide-bore vertical magnet, operating at an RF of 500 MHz for 1 H detection. The probe temperature was set to 37 °C for all experiments. The spectrometer was de-tuned to avoid radiation damping effects. The 45° pulse duration was ~16 μ s, and the acquisition time and relaxation delay were 2 s and 8 s, respectively. Spectral width was 7500 Hz, data size 16 K, number of scans = 8, and a saturation pulse of 3 s was employed at a series of frequencies (Ω) in the range of − 5 to + 5 ppm relative to the water signal (0 ppm). Several RF saturation fields (B 1 ) in the range of 1-6 μ T were used. The data were recorded and processed using TOPSPIN 2.1 software (Bruker).
The CEST was measured by magnetization transfer asymmetry:

SAT S AT 0
In Vivo Experiments. Tumor-bearing animals. Cell culture. 4T 1 (mouse mammary cancer cells) and MCF7 (human mammary cancer cells) were obtained from American Type Culture Collection (ATCC). Cells were cultured in high glucose DMEM supplemented with 10% PBS, 100 μ g/ml penicillin, 100 U/ml streptomycin, 12.5 U/ml nystatin and 2 mM L-glutamine (Sigma, Israel) at 37 °C in a humidified atmosphere containing 5% CO 2 .
Orthotopic mammary fat pad implantation. BALB CEST MRI protocol. A Bruker 7 Tesla (T) BioSpec scanner with 30 cm bore size was used to scan implanted xenograft mammary tumors of mice before and after administration (IV or PO) of GlcN hydrochloride or GlcNAc (dissolved in saline, pH 7.4). Mice bearing 4T 1 or MCF7 tumors, with an average tumor volume of 5 ± 2 mm 3 , were fasted overnight with water access, anesthetized with isoflurane (1-2%), and scanned with a volume coil that was used for RF transmission (without surface coil). Mouse body temperature was maintained at 37 °C. T 2 RARE anatomical images (RARE factor-8, TR-3000 ms, TE-11.7 ms) were acquired to identify the slice presenting the maximum tumor size. The CEST images were generated as follows: a series of T 2 RARE images (RARE factor-1, TR-3000 ms, TE-11.7 ms) was collected from a single 1 mm coronal slice of the abdominal area (acquisition matrix before zero filling 128 × 64, field of view of 40 × 40 mm 2 ) after a 2 s presaturation pulse of B 1 = 2.4 μ T at the hydroxyl proton frequency offset of ± 1.2 ppm from the water signal. CEST images were acquired before and for 1-2 hours after GlcN or GlcNAc administration. The mean intensities in the selected region of interest (ROI) of the tumor were used for the % CEST plot. The reported statistical analyses in all experiments were given as standard deviations (± SD).
No corrections for the B 0 inhomogeneity were used in the present work because no significant change was observed after such corrections in our previous works. The line width of the water peak was approximately 30 Hz.
Histology. Immediately after the CEST MRI experiment, the tumor, lungs, kidney, ovaries, uterus and liver were excised and fixed in 4% formalin solution. Pathology analysis was performed by Patho-Lab (Rehovot, Israel). Statistical Analysis. The results were evaluated using the statistical software SPSS 23.0. All data were expressed as mean ± standard deviation (STD). One-way analysis of variance for multiple comparisons with Tukey post-hoc test were used to determine statistical significance, which was defined by a P value < 0.05.