Pressure Dependent Signal Enhancement in Hyper-CEST

Introduction The so called hyper-CEST method promises tremendous potential on molecule-specific MR imaging using hyperpolarized Xe caged in functionalized cryptophane cages [1]. It was shown that these biosensors can be utilized as highly sensitive temperature probes [2] and that molecule-specific signals in the lower nM regime can be detected when experimental parameters are optimized to yield maximum saturation of the caged Xe. Here an alternative approach for enhancing the saturation efficiency and thus improving the overall sensitivity of hyper-CEST, by variation of the total xenon concentration in the solution is presented. Methods For the measurements an NMR tube of 13 mm diameter and fused with a glass valve on top was filled with the biosensor (cage-HA [3]) dissolved in 2 ml PBS buffer. After degassing the sample the MR tube was pressurized with hyperpolarized Xe gas (enriched to ~ 90 % Xe, PXe~15 %). For hyper-CEST experiments (performed on a BRUKER 30/100 whole body imager) we repeatedly (n=1 to 10) measured two FID signals (S2n-1 and S2n, Fig. 1a) of dissolved Xe (δbulk ≈ 196 ppm) within a 1 cm horizontal slice. Between these two excitations the caged Xe was saturated with a train of Nsat = 1000 successive selective pulses with a repetition time of Δtsat = 15 ms. Since the Xe nuclei transfer between cage and solution, the saturation of cage Xe causes a loss in the bulk signal. The signal reduction Sred=S2n/S2n-1 is governed by three major effects: T1-relaxation, signal loss from the first RF excitation of the bulk resonance, and damping by saturation transfer. To compensate for the first two effects, saturation is performed once on the cage resonance δcage ≈ 64 ppm and once at the mirror frequency referred to the bulk signal (i.e. at 330 ppm). The normalized hyper-CEST signal SCEST =Sred(on res.)/Sred(off res.) thus reflects the reduction solely due to saturation transfer. It is proportional to the difference of bulk xenon concentration Cbulk and Nsat times the cage concentration Ccage, ( ) bulk cage sat bulk CEST C C kN C S − = , where k=ksat × k129Xe × koc accounts for imperfect saturation ksat, Xe enrichment k129Xe and fractional xenon occupancy koc of the cages. For koc ≈ Ka × Cbuk / (1+Ka × Cbulk) with the xenon cage association constant Ka ≈ 6000 [2] one obtains SCEST = 1 − ksat × k129Xe × Ka × Nsat × Ccage / (1+Ka × Cbulk) (Eq.1). By varying the xenon gas pressure pXe in the NMR tube, the bulk concentration may be set freely, e.g. Cbulk = pXe × 3.4 mM/bar at T=35 °C [5]. Results and Discussion Hyper-CEST was performed with biosensor concentrations of Ccage = 5 μM and Ccage = 0.5 μM. In Fig. 1a) two consecutive FIDs are shown with caged xenon saturation in between. In Fig. 1b) the results of the signal reduction are shown for the 5 μM sample. The average signal reduction for off-resonance saturation is 0.85±0.01 which is slightly below the value of 0.87±0.01 as determined by T1-decay and flip-angle measurements [4]. Such deviations are characteristic to each sample and call for the normalization procedure described above (see also Fig. 1b). A similar experiment series was performed on the 0.5 μM sample but at even lower xenon gas pressures (Fig. 1c). Here, due to the lower biosensor concentration efficient saturation can only be achieved when the xenon bulk concentration in the solution is reduced appropriately. By fitting Eq. 1 to the data for both samples, a saturation efficiency of ksat ≈ 0.38 is obtained. For the experimental parameters used one expects ksat ≈ 1 − 2 × exp(-15 ms / 9 ms) ≈ 0.6 (’saturation’ was achieved by inversion of the caged xenon magnetization every 15 ms at 9 ms exchange time between bulk and caged xenon). Conclusion We have introduced a model for the hyper-CEST method. For a given xenon bulk concentration the efficiency of the experimentally applied saturation, and thus the lowest cage concentration to be detectable, can be determined. Assuming perfect saturation and a higher xenon cage association constant of Ka ≈ 30000, as is due to a modified cage construct [6], one would obtain with 50 nM biosensor concentration at bulk xenon concentration Cbulk = 0.14 mM (pXe = 40 mbar) a hyper-CEST signal SCEST as large as shown here for the 0.5 μM sample. Applying our model (Eq. 1) to the hyper-CEST measurements presented in [2] (Ccage = 10 nM, SCEST = 0.85 for 20 s CW satruration, Ka = 6000 M, k129Xe = 0.26, Cbulk = 190 μM) one obtains Nsat ≈ 2×10 assuming perfect saturation. Thus the exchange time between bulk and caged xenon would have to be less than 1 ms. The very smallness of that estimate, as well as the deviation of the model-derived and expected saturation efficiency ksat in our case, call for further investigations to improve the model which finally should allow for an optimization of the sensitivity of hyper-CEST in biosensor applications. References [1] Schröder, L et al. Science 314 (2006) 446-449 [2] Schröder, L. et al. Phys. Rev. Lett. 100 (2008) 257603 [3] Schlund, A. et al., Angew.Chem.Int.Ed. 48 (2009) 4142-4145 [4] Kilian, W. et al. MAGMA 22 Suppl 1. (2009) 70-71 [5] CRC Handbook of Chemisty and Physics, 89th Edition, 2008-2009, CRC Press, Chapter 8: SOLUBILITY OF SELECTED GASES IN WATER [6] Hill, P. A. et al., J. Am. Chem. Soc. 129 (2007) 9262-9263, J. Am. Chem. Soc. 131 (2009) 3069-3077