Cellular and Molecular Imaging using Chemical Exchange Saturation Transfer (CEST)

Abstract

Chemical Exchange Saturation Transfer (CEST) is a powerful new tool well suited for molecular imaging. This technology enables the detection of low concentration probes through selective labeling of rapidly exchanging protons or other spins on the probes. In this review, we will highlight the unique features of CEST imaging technology and describe the different types of CEST agents which are suited for molecular imaging studies including CEST theranostic agents, CEST reporter genes, and CEST environmental sensors.

1. Introduction

Chemical Exchange Saturation Transfer (CEST) is a powerful new tool well suited for molecular imaging (1-4). This technology enables the detection of low concentration probes through selective labeling of rapidly exchanging protons or other spins on the probes. The transfer of saturation from one molecule to another was first observed in 1963 by Sture Forsén and Ragnar Hoffman (5,6), however it was not so clear that this technique would be applicable to magnetic resonance imaging until Steve Wolff, Robert Balaban and colleagues demonstrated the detection of compounds of interest for medical imaging with exchangeable protons through saturation transfer imaging, urea and ammonia (7,8). As has now been clearly shown, for exchangeable spins at low concentration (nM-mM range) a cumulative effect (Molar range) can be built up and detected provided the spins have a sufficiently high exchange rate (k_ex), allowing production of high contrast images (4,9). Fig. 1 demonstrates the basic principles of signal amplification using CEST imaging. As is shown, after applying a saturation pulse, signal contrast can be detected through measuring the asymmetric Magnetization Transfer Ratio (MTR_asym) which is the normalized difference in water signal between the two experiments, one with the RF field on resonance with the exchangeable peak and the other with the RF field at the same frequency away from water but on the opposite side (10-14). Based on this mechanism for generating contrast, CEST probes have a few unique features. Because the labile spins resonate over a range of chemical shifts, imaging schemes have been developed to discriminate between agents possessing these spins simultaneously, which has been called multi-color (15-17) or multi-frequency MRI (2,18). This labeling feature makes CEST agents unlike T1 or T2* MRI contrast agents, and instead more similar to optical imaging agents. CEST contrast generally increases as the scanner magnetic field increases, which makes this a favorable approach compared to T1 relaxation agents whose contrast will be reduced at higher fields through T1 relaxation time convergence (19). Because of these features, a large number of groups have focused on developing CEST contrast agents for a variety of cellular and molecular imaging applications.

Figure 1.

Schematic of Chemical exchange saturation transfer (CEST): principles and measurement approach for pure exchange effects. a, b) Solute protons (blue) are saturated at their specific resonance frequency in the proton spectrum (here 8.25 ppm for amide protons). This saturation is transferred to water (4.75 ppm) at exchange rate k_sw and nonsaturated protons (black) return. After a period (t_sat), this effect becomes visible on the water signal (b, right). c) Measurement of normalized water saturation (S_sat/S₀) as a function of irradiation frequency, generating a so-called Z-spectrum (or CEST spectrum or MT spectrum). When irradiating the water protons at 4.75 ppm, the signal disappears due to direct (water) saturation (DS). This frequency is assigned to 0 ppm in Z-spectra. At short saturation times, only this direct saturation is apparent. At longer t_sat the CEST effect becomes visible at the frequency of the low- concentration exchangeable solute protons, now assigned to 8.25 - 4.75 = 3.5 ppm in the Z-spectrum. d) result of magnetization transfer ratio asymmetry analysis of the Z-spectrum with respect to the water frequency to remove the effect of direct saturation. Reproduced from (4) with permission.

2. Small molecule CEST probes

A major emphasis in CEST imaging has been placed on development of natural, diamagnetic CEST (diaCEST) agents, as there is an enormous variety, which possesses labile protons. In fact, the first compounds presented as CEST agents suitable for medical imaging were urea and ammonia (7,8). More recent studies have investigated detection of D-glucose (20-22), glutamate (23,24), creatine (24,25), and L-arginine (26) on 7 T and higher field scanners. As a result of the interest in developing agents for detection on 3 T and lower field scanners, recent studies have emphasized the design of compounds with chemical shifts from water (Δω) > 4 ppm (27). Based on this criteria, barbituric acid (28,29), thymidine analogs (30), iodinated compounds (31-33), imidazoles (34) salicylic acid and anthranilic acid analogs have come to the forefront. The salicylates and anthranillates are particularly appealing based on their chemical structure consisting of a phenol proton hydrogen bonded to a carboxylate, which has been known with large chemical shift (35-37) which is particularly favorable for 3 T scanners based on their Δω = 6 -12 ppm and the capability of tuning k_ex from 400 - 3000 s⁻¹, suitable rates for detection of CEST contrast. Fig. 2 shows a range of MTR_asym spectra for these probes. The salicylates seem well suited for detection of perfusion (38). Some of the first investigational CEST patient studies have now been performed utilizing the diaCEST agents glucose and iopamidol, with the idea that these agents might present advantages for tumor imaging, with the initial images appearing quite encouraging (39,40).

Figure 2.

Representative examples of IM-SHY CEST agents with tunable exchangeable protons. Conditions: CEST data were obtained at 17.6 T using 10 mM concentrations, pH 7.3 - 7.4, t_sat = 3 sec, ω₁ = 3.6 μT and T= 37 °C. Experimental data are shown as closed circles, while the lines represent Bloch simulations. Reproduced from (27) with permission.

A second focus for this field is on design CEST probes using paramagnetic metal complexes similar to those used as T1 agents. Sherry (41,42), Aime (43,44), and others (45-52) have pioneered the use of paramagnetic metal based CEST agents with appropriately shifted labile protons, which have been termed as paramagnetic CEST (paraCEST) probes. The labile proton Δω can be increased to between 30-700 ppm for these systems based on hyperfine shifts allowing detection of protons with larger k_ex while still adhering to the slow-intermediate exchange condition (1,53-57). The labile protons can be a whole water molecule coordinated to the metal and exchanging on/off the complex, or they can be amide, amine or alcohol protons on the complex with suitable configurations with respect to the paramagnetic center. To date, paraCEST agents have been only applied in pre-clinical studies, however these agents represent the most promising for detection on 1 - 1.5 T scanners (58). Indeed, as HPDO3A has already been approved for patient studies as a safe chelator of Gadolinium and is in use in clinical practice (Prohance, Bracco Imaging, Milan, Italy), one could reasonably expect that paraCEST agents based on HPDO3A could be translated for generating pH maps of tumors, which could be useful in monitoring tumor microenvironment (59).

3. CEST probes as theranostic agents

An exciting area, which appears quite promising, is to utilize CEST imaging to monitor the delivery of therapeutics. This is a very active area of research. Based on sensitivity considerations, a majority of the studies have emphasized use of delivery systems with both CEST probe and therapeutic integrated into nanocarriers. The approaches have included: conjugating CEST probes to the surface of nanocarriers (60-63), entrapping CEST probes into the interior of the carriers (15,28,29), or a third approach, entrapping shift agents in the interior of the carriers and using the interior solvent as the CEST probe itself, the so-called lipo-CEST approach (64). A number of studies have shown that delivery of therapeutics to tumors can be detected and quantified using this technology (29,60,65). This has been motivated in large part by the possibility of performing multi-color CEST imaging in vivo without penetration depth limitations (Fig. 3). As is shown, two different therapeutics can be labeled with a different color and monitored independently using this technology. Moreover, it was recently demonstrated that CEST MRI can be used to detect a clinically used DNA alkylating anti-cancer agent directly without the need to conjugate or modify the drug (66). This is extremely important because even the addition of the Lest tag (such as a fluorine group) can significantly change the affinity of the drug to the target or reduce its activity. The future appears quite promising for this new technology as contrast agent chemistry, image acquisition schemes and MRI hardware improve.

Figure 3.

Representative two-color CEST image demonstrating simultaneous detection and visualization after injection of L-arginine CEST liposomes and poly-l-ysine CEST liposomes into the footpads of a mouse. Left panel: T2-weighted anatomical image with arrows indicating the location of popliteal lymph nodes; Middle panel: MTR_asym images at the frequency of interest for L-arginine (1.8 ppm, upper) and poly-L-lysine (3.6 ppm lower); Right panel: MTR_asym/T2w image overlay with CEST contrast highlighted using a 64-bit scaled single color of PLL (green, left lymph node) and L-Arg (yellow, right lymph node). Reproduced from (26) with permission.

4. CEST reporter genes for cellular imaging

After the observation was made that biological macromolecules can be detected using CEST agents (67), an intense effort has been put forth to develop CEST agents which might be expressed by cells, so-called CEST reporter genes. This is a particularly interesting area, as this technology would allow the generation of functional information about cells not normally accessible using MRI scans. Developing of these types of agents poses different challenges from the other agents mentioned earlier to formulate an optimal expressable CEST probe, with a number of studies that are summarized in Table 1. Each study was necessary for developing the next step. While early on it was shown by van Zijl and colleagues that Poly-LLysine (PLL) is a good CEST probe (67), optimization was needed to allow expression of this peptide by mammalian cells. Nevertheless, PLL was the foundation for a decade of development of new peptide-based agents. Another important aspect of this work was the development of screening methods for peptide libraries for formulating the next generation of agents (68,69). Advances in gene synthesis technologies significantly reduced costs and shortened the synthesis time (70), which has now led to an increased ability to express many of these proteins in both live mammalian cells and bacteria (71-74). The ability to synthesize whole proteins (longer the 50 amino acids) allowed deeper investigation of their biophysical properties (75).

Table 1.

selected milestones in the developing of protein based CEST agents

	Authors	Year	Topic	Findings relevant to the next gen LRP	Ref.
1	McMahon et. al.	2008	Quantifying exchange rate	Determining the kex of exchangeable protons	(70)
2	Gilad et. al.	2007	LRP prototype	LRP can be expressed in the brain	(77)
3	McMahon et. al.	2008	Multicolor peptides	Changing the amino acid sequence affect the resonances frequency and the contrast intensity	(82)
4	Liu et. al.	2010	Peptide screening with MRI	Developing tools for high-throughput screening	(69)
5	Airan et. al.	2012	PKA sensor	(1) Synthetic gene that with heterogeneous amino acid sequence (2) effect of phosphorylation on the contrast	(72)
6	Bar-Shir et. al.	2014	Human protamine as a reporter gene	DNA optimization improves contrast	(75)
7	Oskolkov et. al.	2014	Characterization of Human Protamine	(1) Effect of phosphorylation and (2) effect of intermolecular bonds on the contrast	(76)
8	Bar-Shir et. al.	2015	Supercharged GFP as a CEST based reporter	The contrast can be improved by changing only limited number of amino acids	(74)

In order to ensure the quality of these reporter genes is sufficient for cellular imaging, it is important to demonstrate reproducibility beyond the first “proof-of-concept” paper. This has been demonstrated for the LRP CEST reporter gene; following the initial study where it was shown that the contrast generated from LRP can be differentiated from controls, its practicality has been repeatedly demonstrated in live rodents and enabled collection of high temporal and spatial resolution images (76,77). Farrar and colleagues have confirmed that LRP can be visualized in live rats after delivery of an oncolytic virus, thus serving as a marker for treatment efficacy (78) (Fig. 4). In a separate study, Pomper and colleagues have demonstrated tumor-specific expression of LRP using a tumor-specific promoter (79) (Fig. 5). Interestingly, the PEG3 promoter is weaker than CMV and yet still provided a detectable CEST contrast. These two studies have presented new evidence that LRP is effective for detecting cells through gene expression. Moreover, these studies established the reproducibility of LRP as a reporter gene for MRI.

Figure 4. LRP In vivo.

A & C) before and B & D) after injection of LRP-expressing oncolytic virus into a brain tumor in a live rat brain. There is a significant (p=0.05) increase in tumor CEST contrast for G47Δ-LRP (n=7) than for controls G47Δ-empty (n=6). Reproduced from (78) with permission.

Figure 5.

a, b) Representative CEST maps superimposed on T2-weighted images:Left hemisphere has 9L tumors and right hemisphere has (a) 9L^CMV-LRP and (b) 9L^PEG-LRP. c) Temporal changes in the ΔMTR_asym values (mean ± s.d.; 8 mice) of each tumor type. t-test showed statistical difference of CEST contrast at 3.4-3.6 ppm, (*p<0.05; n=8). Reproduced from (79) with permission.

5. CEST environmental sensors for cellular imaging

While reporter genes such as LRP can be used to monitor cell survival and are well established in a pre-clinical setting, there are challenges to using reporter genes in patients based on the requirement of genetic manipulation of the cells for transplantation. Because of this, we have developed an alternative strategy, which instead employs implantable environmental sensors to monitor rejection of these cell grafts. These sensors are integrated into cell composites. This is a flexible method, as there are a range of MRI probes which have been developed for detecting changes in environment including the pH (32,33,69,80,81) or the concentration of inorganic ions such as Zn²⁺ or Ca²⁺ (50,82-85). Our initial idea was that one effective way to use these sensors would be to embed these in cell composites and monitor the environment post-transplantation. There are three components of these composites: the CEST sensors, the liposomes which entrap the sensors but are sufficiently permeable to water and the hydrogels which provide support for the cell graft (86) (Figure 6). This design isolates the sensor from the cells and also from the immune system while allowing free access of water and ions. One type of environmental change that can be sensed readily using CEST imaging is pH, which influences k_ex through acid and base catalysis of proton exchange, with pH changes also linked to cell death (87). Previous hydrogel based cellular MRI studies were concerned with visualizing their location (88,89), so our cell composites with arginine based CEST pH sensors added a new dimension. Our first investigation involved transplanting human hepatocytes subcutaneously within alginate composites, and in order to test the capabilities of these sensors, groups of mice with and without immunosuppression treatment were included with the expectation that cell survival would be prolonged in the immunosuppressed group (90). The hepatocytes were transfected with luciferase to allow usage of bioluminescence to monitor hepatocyte viability for validation. As expected, as the encapsulated cells died, there was a corresponding 33% drop in CEST MRI contrast detected, with also a substantial difference in hepatocyte survival for the immunosuppressed mice compared to the non-immunosuppressed mice, which corresponds to lowering the pH from 7.4 to 6.9. This work was followed up by a separate investigation of the host immune response to these cell composites, which depended on whether the highly immunogenic HepG2 hepatocytes were included within these composites (91). A higher immune cell infiltration was detected when live hepatocytes were included in the composite (compared to dead hepatocytes), which was presumably associated with xenogeneic molecules passing through the capsule surface. The amount of immune cells surrounding the composites containing hepatocytes was comparable to composites without hepatocytes when FK506 immunosuppression was administered. Based on these studies, we concluded that CEST imaging of cell composites should represent a promising technology for monitoring the functionality of cells after transplantation.

Figure 6. Schematic showing the principles of in vivo detection of cell viability using LipoCEST microcapsules as pH nanosensors.

The CEST contrast is measured by the drop in the signal intensity (ΔS) of water after selective saturation (i.e. removal of capability to generate signal) of the NH protons in L-arginine at 2 ppm. The L-arginine protons (red) inside the LipoCEST capsules exchange (k_SW) with the surrounding water protons. The k_SW is reduced at lower pH causing a significant drop in CEST contrast. Reproduced from (9) with permission.

6. Hyperpolarized CEST (hyperCEST) imaging

An alternative strategy to increase the sensitivity of CEST imaging further, is to move away from detecting exchange with water to detecting exchange of hyperpolarized ¹²⁹Xe, which has been termed hyper-CEST imaging. This strategy was initially reported by Leif Schroeder, Alex Pines and co-workers, and involved use of specialized cage structures which transiently trap and impart a chemical shift on hyperpolarized ¹²⁹Xe compared to gas phase, with rapid exchange of Xe occurring between the interior and exterior of the cage (92). For molecular imaging studies, ¹²⁹Xe appears to be the most promising of the noble gases which can be hyperpolarized based on solubility in aqueous solutions (93) and biological tissue (94,95), relative abundance of xenon gas, and also responsiveness of the ¹²⁹Xe chemical shift to molecular environment. Furthermore, hyperpolarizers have been designed to produce large quantities of polarized Xe-129 gas allowing clinical usage (96-99). This strategy of hyperCEST imaging envisions targeted imaging of these cages after inhalation of xenon gas by patients. Cages have been developed which are sensitized to lead, zinc, mercury and cadmium ions (100-102), which assemble onto a multivalent M13 bacteriophage (103), and report on cellular internalization (104). One particularly nice study showed that bioengineered bacterial gas vesicles enabling detection of picomolar concentrations of the gas-binding protein nanostructures expressed (105). Another interesting study showed that bacterial spores represent another nanoporous structure which can be detected using hyperCEST imaging (106). This strategy, which involves hyperpolarized Xenon with the polarization dropping with T1, has different requirements from water based CEST imaging, and as a result has spurred new imaging methods to allow faster acquisition (107,108).

7. Outlook for the future of CEST imaging

One of the major criticisms pointed at CEST MRI is its low sensitivity. Indeed, CEST MRI would be more appealing for the broad imaging community if the sensitivity could be improved. Improving the robustness of the technology could be achieved through three independent levels: (1) better probe design, (2) better MRI imaging sequences and (3) increased reproducibility (Fig. 7). For example, CEST data are often contaminated by multiple components that interfere with detecting the probe of interest. We have developed an approach, termed Length and Offset VARied Saturation (LOVARS) (109), in which the length (t_sat) and Δω of the saturation pulse is varied to modulate the water signal loss and impart differential phases on the interfering components. This allows their separation from the desired probe signal using post-processing techniques similar to those used to analyze time-varying signal changes in event-related fMRI (110-116), such as Fast Fourier Transform (FFT) to separate different frequency components(109). Finally, it is key to repeatedly use the against with the highest standard of rigorous and reproducibility(117), in order to achieve highly sensitive, user-friendly imaging probe. Other approaches including systematic variation of saturation pulse flip angle, of multiple frequency pulses or variable delay elements and use of frequency labeling pulses instead of saturation or improvements in modeling the Z-spectra (118-123). CEST imaging sequence design is still relatively immature as a field, and the expectation is that many improvements in probe, sequence and post-processing will be forthcoming. As a result, molecular imaging using CEST should have a bright future.

Figure 7.

Multi-dimensional development of molecular imaging using CEST.