Objective

In biological systems, R₁-dispersion curves (R1=1/T₁ versus ¹H Larmor frequency ν₀) acquired by Fast-Field-Cycling NMR (FFC-NMR) often exhibit local R₁ enhancements between 1.5 and 3.5 MHz, called quadrupolar peaks (QP) that result from quadrupolar interactions between nitrogen 14 of amino-acids in proteins and water protons (¹H-¹⁴N)¹. This study is focused on glioma, an aggressive cerebral tumor, with two objectives: (i) to describe if (¹⁴N-¹H) QPs signals can discriminate between invasion and proliferation tissues. To test this, we used two glioma mouse models, Glio6 and Glio96, validated as invasion/migration models^2,3, compared with the U87 characterized by a high glioma proliferation. (ii) To determine the intra- or extra-cellular origin of the proteins involved in the QP. To that purpose, glioma tissues characterized by their extra and intracellular compartments were compared to cell pellets with relaxation dominated by intracellular compartments. In parallel, we performed additional FFC and QP measurements of cells treated by trypsin, a digestive enzyme of the serine proteases family that breaks down proteins. Trypsin was specifically used because it is an extracellular biochemical substance that does not cross cell membranes and therefore can only degrade the extracellular compartment.

Subjects and Methods

Cell culture, cell pellet and Trypsin treatment: Glio6 and Glio96 cell lines were obtained from human surgical glioma resection prior to this work. Both were grown under hypoxia in 3% O₂ and 5% CO₂ in untreated flask with DMEM/F-12 (1:1) + GlutaMAX supplemented with BFGF and EBF growth factors (20ng/ml). U87 glioma cells (ATCC HTB-14) were grown in 5% CO₂ under 20% O₂ in DMEM + GlutaMAX (Gibco) and 10% fetal bovine serum and penicillin-streptomycin (100 U/ml). Cell pellets were obtained after centrifugations (5min, 1200rpm) and removing the supernatant. For Glio6 and Glio96, we added 1 ml cell dissociation solution and washed 5 min later, resuspended twice with 4 ml and 1ml PBS respectively, centrifuged and removed the supernatant. 20 to 40.10⁶ cells were used for FFC-NMR measurements, acquired 40 min after preparations. For Trypsin experiments, the cells were treated by adding 20-fold excess of trypsin by weight. Cells were prepared at 37˚C and cell viability was assessed using trypan blue exclusion method before and after FFC-NMR acquisitions.
Animal models: All animal procedures were conducted according to the license C3818510003). Mouse-glioma models were obtained using immune-deficient nude-mice (30 to 35 g). Glioma cells were injected into the right caudate nucleus under isoflurane (3%). In vivo MRI follow-up was performed at 4.7T with T2W- RARE sequence (TR/TE=3500/33 ms), every week during the first month and every 2 days during the second and the third months. At late stage, brains were removed (U87 n=11, Glio6 n=6, Glio96 n=7) and glioma tissues extracted (30-210mg) and stored at -80°C.
FFC-NMR: FFC-NMR as described in Figure 1 was performed using a Stelar SPINMASTER-2000 relaxometer with a 10 mm bore diameter. Samples were put in 5mm diameter NMR tube. To increase SNR we accumulated 128 echoes of a Carr–Purcell–Meiboom–Gill (CPMG) sequence with 10ms echo-time. We used 30 evolution fields with logarithmic sampling between 5KHz and 30MHz. For each field the magnetization was measured over 12 evolution times selected logarithmically between 0.01 and 4 times the estimated T1 value. QP were acquired between 1.5 and 3.5MHz with a high sampling (n=30). The total time of both acquisitions: 90min.
FFC analysis: The amplitude of the QP was estimated after subtracting the background contribution, fitted with a power-law model⁴, and analyzed according to reference⁵. Comparison between the different groups was assessed using the Student test at p<0.05 level while the normality was ensured with the Kruskal-Wallis (KS).

Result

Figure 2 displays the average R₁-dispersion curves of the three glioma tissues. As reported in our previous work⁶, this is lower for invasive glioma than for proliferative ones, as quantified by the vertical offset (A). Additionally, the exponent of the power law β also discriminates invasion from proliferation, as expected since it relates to water molecular dynamics. In all curves the QP signal was present and its amplitude was also discriminative of invasion and proliferation (p<0.05). Figure 3A compares the R₁-dispersion curves of glioma tissues with their corresponding cell pellets and details the QP signal after subtraction of the curve background. Figure 4 is a table which reports the statistical analyses testing for differences between glioma tissue and their corresponding cell pellets. This is particularly marked for the peaks amplitude (AQP) which is related to the protein concentrations⁷, meaning that proteins responsible for QP are intracellular. Indeed, under trypsin treatment, (¹H-¹⁴N) QP of cells remain unchanged (Figure 3B), confirming thereby the intracellular origin of these signals.

Conclusion

From this study, we can conclude that FFC-NMR is an appropriate tool to discriminate invasion from proliferation. This study especially demonstrates that the (¹H-¹⁴N) QP peak signals predominantly arise in glioma from proteins localized inside the cells. This study aimed to determine the origins of the QP signals which are modulated by motion of the protein backbones, and constitutes an important step when exploring the role of this parameter as a new glioma biomarker of large intracellular proteins.

Acknowledgements

This project is as part of the European H2020 PHC-11-2015 IDentIFY project 668119.