Chemical Exchange Saturation Transfer (CEST) MRI measures the effect of chemically exchanging labile protons associated with biomolecules. Application of the CEST MRI technique to measurements of the amide protons associated with mobile proteins in vivo is amongst the most popular of these methods, and the CEST effects at 3.5ppm downfield of water have been used to make pH measurements in acute stroke¹, or to differentiate tumour recurrence from radiation necrosis². One assumption about the CEST MRI signal is that it largely reflects the intracellular properties of tissue, since the majority of mobile protein is intracellular³. However, to date, no study has elucidated the exact contribution of the extracellular protein to the CEST MRI signal in vivo. If the contribution of the extracellular protein to the CEST signal could be determined, it may be possible to specifically quantify pH in the extracellular space, which would be of particular interest in imaging of tumours⁴. This study employed a novel pulse sequence that encodes both diffusion and CEST signals to selectively measure the Z-spectrum from extracellular or intracellular environments.

Mouse mammary carcinoma cells (4T1) in PBS were placed in a 14T vertical bore spectrometer. Water-suppressed (WATERGATE) ¹H spectra were acquired at various intervals to ensure cell survival during the experiments. Stimulated echo acquisition mode (STEAM) diffusion measurements (11 b-values altered by varying only the magnitude of diffusion gradients) and CEST spectra (CW saturation, 2s duration, B₁=1µT, ±5.1ppm in 0.3ppm steps, normalised by unsaturated acquisition) were acquired. The DiffusionCEST pulse sequence (Fig 1), comprising a STEAM diffusion sequence with CEST pulses between the second and third 90° RF pulses, was used to acquire Z-spectra with 11 diffusion gradient strengths and CEST saturation of 500ms CW, B₁=1µT.

Post-mortem naïve BALB/c mouse brain was imaged with similar sequences at 9.4T. CEST imaging was performed using single-shot spin-echo EPI following 300 gaussian pulses of duration 13ms, flip angle 180° (50% d.c, equivalent CW B₁=0.8µT). STEAM diffusion imaging was performed using 16 b-values (b-value altered by varying only the magnitude of diffusion gradients). DiffusionCEST was used to acquire Z-spectra with 8 b-values and CEST saturation of 35 CEST pulses, each pulse comprising a 14ms gaussian (flip angle 180°) followed by 1ms spoiler gradient (n=35 in Fig 1).

The presence of restricted and free water was confirmed in cells and post-mortem brain by fitting a bi-exponential decay function to the water signal as the diffusion gradient strength was varied, and assigned to intra- and extracellular, respectively. The Z-spectra from intra- and extracellular spaces were isolated, by fitting a bi-exponential function to the measured signal at each offset frequency in the acquired Z-spectrum as diffusion b-value was varied.

Bi-exponential fitting of the water signal as a function of diffusion-weighting revealed two water populations, both in the cells (Fig 2A) and in the post-mortem brain (Fig 2B). Z-spectra from the cells (Fig 3A) and a region of interest covering the whole brain (Fig 3B) revealed CEST effects associated with intracellular protein. DiffusionCEST results from the cells (Fig 4A) and post-mortem mouse brain (Fig 4B) revealed a bi-exponential decay in signal at all frequency offsets in the Z-spectrum, enabling the separation of intra- and extracellular Z-spectra (Fig 5A and B).

In the cells (Fig 5A), the extracellular Z-spectrum displays only direct saturation of water, whereas the intracellular spectrum displays features associated with intracellular protein. Similarly, in the post-mortem mouse brain (Fig 5B), the intracellular Z-spectrum displays prominent CEST effects upfield and downfield of water. However, despite being noisy, the extracellular Z-spectrum shows similar effects upfield, and less saturation downfield. The molar fractions of intra- and extracellular water calculated from their contributions to the overall Z-spectrum are 0.83 and 0.17, respectively, which are in remarkable agreement with literature values of intracellular water content of the brain.

By combining diffusion and CEST imaging, the Z-spectra from intra- and extracellular spaces have been separated in a simple cell system and post-mortem mouse brain. The extracellular Z-spectrum from the post-mortem mouse brain is noisy, making interpretation and quantification of the Z-spectrum difficult. However, this may be ameliorated with increasing acquisitions with low diffusion-weighting. The combination of STEAM diffusion and CEST imaging produces images with low signal-to-noise, and the acquisition time currently precludes in vivo application. Further study and optimisation of the pulse sequence are on-going.A novel combination of CEST with STEAM diffusion imaging enables the separation of Z-spectra from intra- and extracellular water, respectively. Following further study and optimisation, this sequence may enable quantification of pH from the intra- and extracellular environments separately.