Molecular Engineering of Nonmetallic Biosensors for CEST MRI

Abstract

Recent advancements in molecular and synthetic biology, combined with synthetic chemistry and biotechnology, have opened up new opportunities to engineer novel platforms that can monitor complex biological processes with various noninvasive imaging modalities. After decades of using gadolinium- or iron-based metallic sensors for MRI, the recently developed chemical exchange saturation transfer (CEST) contrast mechanism has created an opportunity for rational design, in silico, of nonmetallic biosensors for MRI. These biomolecules are either naturally occurring compounds (amino acids, sugars, nucleosides, native proteins) or can be artificially engineered (synthetic probes or recombinant proteins). They can be administered either as exogenous agents or can be genetically (over)expressed. Moreover, they can be precisely engineered to achieve the desired biochemical properties for fine tuning optimized imaging schemes. The availability of these agents marks the dawn of a new scientific era for molecular and cellular MRI.

Graphical Abstract

Understanding complex biological processes at the cellular level in live subjects is a key element in both basic science and clinical practice. Molecular and cellular imaging is a relatively new discipline that seeks to develop imaging toolboxes for the visualization, characterization, and quantification of (sub)cellular biologic processes. This can be accomplished using advanced imaging technologies, which include optical imaging, nuclear imaging, X-ray computed tomography (CT), ultrasound, and magnetic resonance imaging (MRI).^1–3 Synthetic imaging probes or genetically engineered imaging reporters have been developed for all of the aforementioned imaging modalities. These sensors are either biologically inert (to study location, a particular cell or organ, etc.) or biologically responsive (to study a process or function of interest). Among the imaging modalities, MRI is unique because longitudinal studies on a live subject can be performed, and the information obtained using biosensors or reporter genes can be coregistered with high-resolution anatomical images.

1. CEST MRI

In MRI, information is most commonly obtained by enhancing the contrast between the target and its surrounding tissue. This contrast can be enhanced by either manipulating the imaging acquisition parameters or by introducing chemical or biological entities also known as contrast agents (see Keywords). For many years, gadolinium (Gd³⁺)-based probes⁴ and super-paramagnetic iron oxide (SPIO)-based particles⁵ have been used to affect water relaxation properties (T₁ and T₂ or T₂*) and, consequently, the localized MR image contrast. Although Gd³⁺ agents have been used for the design of responsive MRI sensors, their metal origin might have adverse cellular effects that have not been elucidated yet in patients. The contrast generated by SPIO agents wipes out the ¹H MRI signal that is obtained from the imaged tissue, thus masking the anatomical information in the region of interest.

In 2000, a new contrast mechanism for MRI was proposed, termed chemical exchange saturation transfer (CEST).^6,7 In CEST MRI, a dynamic exchange process between an exchangeable proton and water protons is detected (Figure 1a; for a detailed review, see refs 8 and 9). By applying a radiofrequency (saturation) pulse at the resonance frequency (shifted by Δω from the water protons) of the exchangeable pool of protons, the magnetization of this proton’s pool is nullified (often referred to as saturated) and transferred to the bulk water protons through an exchange process. If the saturation pulse is applied for a sufficient time and the exchange process is fast enough (but still in the slow–intermediate exchange regime on the NMR time scale), then an effect can be detected through the reduction of the water signal. This mechanism allows the detection of low-concentration CEST biosensors (from hundreds of micromolar to several millimolar) through the high ¹H MRI signal of the water. Although CEST contrast can be detected from endogenous components (e.g., glutamate,¹⁰ creatine,¹¹ glycogen,¹² and others), it can be generated also by properly designed sensors. When combined with certain paramagnetic synthetic complexes, which induce highly shifted exchangeable protons (or water molecules), CEST can been termed PARACEST. PARACEST sensors were extensively studied, and their applications have been reviewed elsewhere;^8,13,14 however, since they are not based on natural compounds that are biodegradable, they are potentially toxic when not rapidly cleared from the body (as exemplified in the case of gadolinium). Most of the CEST MRI biosensors are based on proton exchange and detected through the changes in the water ¹H MRI signal. However, the CEST contrast that is based on two pools of exchangeable components, which have different chemical shift offsets (Δω) in NMR spectrum, was also recently extended to ¹⁹F- and ¹²⁹Xe-based MRI (Figure 1b,c).^15,16 As both ¹⁹F and ¹²⁹Xe nuclei do not exist in biological tissues, the information from heteronuclear-CEST (Hetero-CEST) can be overlaid on 1H MRI anatomical information and be presented as a hot-spot map.

Figure 1.

Principles of CEST MRI. A frequency-selective saturation pulse (saturation) is applied to label exchangeable protons (a) or molecules (b, c) to achieve a reduction in the bulk signal (ΔSI, change in signal intensity) from the chemical exchange process (detection). Shown are current examples for ¹H₂O (a), a ¹⁹F chelate (b), and ¹²⁹Xe gas (c).

2. RATIONALE FOR SELECTION AND DESIGN OF BIOSENSORS

The workflow in Figure 2 demonstrates the steps for CEST biosensor design toward a desired biological target. As a first step, a biological process or objective needs to be carefully specified. Such an objective may be, for example, a certain physiological condition (pH, temperature), an enzyme, a signaling or metabolic pathway, receptors, or a specific metal ion. In the next step, potential sensors (e.g., substrates, antigens, responsive agents) should be identified based on their putative CEST properties. It is important to determine the CEST contrast of the potential biosensor both before and after targeting the biological object (i.e., before and after enzymatic conversion, pH changes, etc.). It is essential to take into account that CEST contrast can be either turned on or off. A turn-on sensor may reflect a buildup of the contrast by changing proton exchange rates or by accumulation of the CEST probe. A turn-off sensor may reflect substrate conversion and elimination of the exchangeable proton.

Figure 2.

Rational design of developing a CEST biosensor.

Next, if a naturally occurring molecule that provides high contrast, high sensitivity, and high specificity can be identified, then in vivo imaging can be performed directly. Nevertheless, the sensor administration concentration and delivery method should be considered carefully (i.e., intravenously, (i.v.), intraperitoneal (i.p.), bolus or infusion, bare sensor, or embedded within a nanocarrier vehicle). Otherwise, if a naturally occurring probe does not exist or if CEST contrast is not satisfactory (e.g., poor contrast, small Δω, low sensitivity, and low specificity), then a molecular (synthetic or genetically) engineered CEST biosensor should be designed. By following a rational design for CEST biosensors, as reflected in the workflow depicted in Figure 2, both chemists and biologists will be able to generate new probes adapted specifically to their own research.

3. SMALL MOLECULES AS CEST BIOSENSORS

Small imaging probes are frequently used in a wide range of imaging modalities, including fluorescent or luminescent molecules for optical imaging, iodine-containing compounds for CT imaging, radioactive probes for positron emission tomography (PET) or single photon emission computed tomography (SPECT) applications, or contrast agents for MRI. The following section describes the use of small CEST biosensors for a range of applications from glucose uptake and metabolism imaging to real-time monitoring of gene expression. The depicted biosensors range from naturally occurring molecules (i.e., sugars, amino acids) to rationally designed synthetic molecules.

3.1. Imaging Glucose Uptake and Metabolism.

Primary and metastatic cancers are associated with upregulation of glycolysis, which results in increased levels of glucose uptake and consumption.¹⁷ Moreover, overexpression of glucose transporters and glycolytic enzymes has been found in malignant tumors.¹⁸ Therefore, monitoring glucose intake and metabolism in tumors has been a desired goal in cancer research for decades.¹⁹ The glucose analogue 2-deoxy-d-glucose (2DG) cannot be metabolized further during glycolysis.²⁰ Hence, the radioactive tracer 2-[¹⁸F]fluoro-2-deoxy-d-glucose (FDG),²¹ which accumulates in tumor cells, has been the flagship agent of clinical PET applications for several decades.

Although glucose uptake and metabolism can be detected by MR spectroscopy (MRS)-based methodologies using ¹³C-labeled glucose, which obviates the need for radioactive tracers, these methods suffer from low sensitivity and have poor-to-nonexistent spatial resolution.^19,20 Therefore, both the research groups of Golay and van Zijl capitalized on the five OH exchangeable protons of naturally occurring D-glucose (Figure 3a) to obtain spatial information from glucose uptake by tumors using CEST and have termed this methodology glucoCEST.^22,23 The glucoCEST contrast originates from saturating the glucose-exchangeable protons that resonate at a range of frequency offsets from the water resonance (Figure 3a; Δω = 0.8–2.1 ppm²² or Δω = 1.2–2.9 ppm²³). The systemic administration of D-glucose resulted in sufficient sensitivity for the detection of glucose accumulation in tumor-bearing mice.^22,23 These studies demonstrated the potential of using a naturally occurring MRI biosensor (d-glucose) for in vivo applications. However, since it is difficult to distinguish between the intra- and extracellular glucose and to account for downstream metabolism via the glycolitic pathway of the intracellular glucose, the contribution of each of these components to the CEST contrast is complex and not fully understood. The glucose analogue 2DG is phosphorylated to 2DG-6-phosphate (2DG6P) when taken up by cells, and it then accumulates in the cells since it cannot be furthered metabolized by the glycolytic pathway. Therefore, 2DG (Figure 3a) has been considered a turn-on CEST alternative for glucoCEST imaging^24,25 in the same manner as FDG is used for PET. Very recently, the using of 3-O-methyl-d-glucose (3OMG), a nontoxic and nonmetabolic derivative of glucose, was also used for CEST imaging of tumors.²⁶

Figure 3.

Small molecules as CEST biosensors. The chemical structure of small molecules (natural or synthetic) that are being used as CEST sensors for monitoring: glucose uptake (a) and metabolism, pH (b), and genetically encoded enzyme/substrate systems (c).

3.2. pH Sensing.

The acidity (pH) of intracellular compartments²⁷ in living cells plays a key role in many cellular processes, such as apoptosis, endocytosis, ion transport, and others. These events are studied mostly using fluorescent dyes²⁸ or genetically encoded fluorescent proteins engineered for pH sensing.²⁹ Since CEST contrast is affected dramatically by the exchange rate (k_ex) between the exchangeable protons and the water protons and the k_ex is dependent on pH,⁸ it is feasible to map pH using CEST MRI. Therefore, theoretically, each CEST agent can be used to measure pH, as demonstrated for several probes in the original papers of Balaban and co-workers.^6,7 Chan et al. relied on the guanidine exchangeable protons of the amino acid l-arginine (Figure 3b; Δω = 2.0 ppm) as a pH-sensitive CEST contrast agent for the noninvasive imaging of the viability of transplanted cells.³⁰ In this case, CEST contrast is turned off upon pH lowering as a consequence of cell death. To avoid dependency on the biosensor concentrations, compounds with more than one exchangeable proton pool, which resonates at a different Δω and has a different exchange rate with water, were suggested as pH CEST sensors. One example is the amide protons of iopamidol (Figure 3b; Δω₁ = 4.2 ppm, Δω₂ = 5.5 ppm), which can be used in a ratiometric method for pH assessment.³¹ Later, iobitridol, an analogue of iopamidol that contains a single set of amide protons, was used for in vivo pH mapping with CEST.³²

One of the main obstacles of the aforementioned CEST biosensors is their relatively low Δω, which may result in (i) direct saturation of the water protons (during the saturation of the magnetization of the exchangeable proton) and (ii) contributions from endogenous CEST contrast (mostly from sugars, amino acids, peptides, and proteins). Moreover, CEST sensors with a relatively low Δω may not be applicable for MRI scanners operating at low magnetic fields (especially 1.5–3.0 T clinical scanners, compared to the common 9.4–11.7 T preclinical scanners).⁸ The hydroxyl proton of salicylic acid (Figure 3b) and its derivatives was suggested as a CEST contrast generator with highly shifted exchangeable protons (Δω = 8.7–10.8 ppm), probably due to the intramolecular hydrogen bonding found in these molecules.³³ These probes, which demonstrate (so far) the largest Δω between water protons and exchangeable protons for CEST agents, are pH-sensitive and can be monitored in vivo. Since these compounds can be easily chemically modified, they appear to be ideally suited for further development of CEST sensors.^34,35

3.3. Monitoring Genetically Encoded Enzyme/Substrate Systems.

Monitoring enzyme activity longitudinally and noninvasively is crucial for many disciplines of the biomedical sciences. PET,³⁶ SPECT,³⁷ and bioluminescent imaging (BLI)^38,39 are well-established and highly sensitive modalities for imaging genetically encoded reporter systems. These reporter systems couple an efficient enzyme to a specific substrate (sensor) that can be sensitively imaged upon its conversion. However, some subjective and intrinsic limitations of these modalities restrict their widespread use or clinical translatability. The complexity of synthesis (requiring large investments in infrastructure including a cyclotron) and relatively short half-life times of radiolabeled compounds, in addition to a limited number of clinical scanners (~1:8 ratio of PET/MRI scanners in the US), restricts the application of nuclear imaging-based technologies. Optical-based imaging modalities are limited by tissue penetration of light and, consequently, are not applicable for imaging deep tissues, large animals, or humans. Therefore, there is an increasing need for MRI-based genetically engineered reporter systems as an alternative. Since ¹H CEST contrast is generated by exchangeable protons of the molecule of interest, there is no need to attach a metallic imaging moiety to alter the contrast. Such an addition dramatically changes the structure of the desired substrate and may significantly affect the conversion rate by the target enzyme. The use of CEST-based reporter probes (substrates) for imaging reporter genes (enzymes) has opened a new avenue for the design and use of small (natural or synthetic) molecules as imaging probes for monitoring enzyme expression and activity.^40–43

Cytosine deaminase (CDase) is an enzyme that catalyzes the conversion of cytosine to uracil (Figure 3c) and is expressed exclusively in bacteria and yeast. Given that CDase is absent in mammalian cells and since it can also catalyze the conversion of the prodrug 5-fluorocytosine (5FC) into the chemotherapeutic agent 5-fluorouracil (5FU), it was suggested for use as a therapeutic gene for cancer treatment.⁴⁴ Although CDase expression and activity (Figure 3c) were monitored in vivo using ¹⁹F MRS,⁴⁵ no spatial information could be observed. Liu et al. used CEST MRI to monitor CDase activity with its native substrate cytosine or the pro-drug 5FC.⁴² In that study, the CEST contrast derived from the NH₂ protons (Figure 3c, Δω = 2.0–2.4 ppm) of cytosine or 5FC was eliminated upon its conversion to uracil or 5FU. This system is depicted in the right-hand section of the workflow (Figure 2) and demonstrates the utilization of both natural (cytosine) and synthetic (5FC) substrates as imaging probes. The bacterial enzyme carboxypeptidase G2 (CPG2) mediates the release of the C-terminal glutamate residue from a wide range of N-acylating moieties. Accordingly, CPG2 was suggested as a therapeutic gene that converts a nitrogen mustard pro-drug (CMDA, 4[(2-chloroethyl)(2-mesyloxyethyl)amino] benzoyl l-glutamic acid)) to an active drug (4[(2-chloroethyl)(2-mesyloxyethyl)-amino] benzoic acid), which is a potent DNA cross-linking agent.⁴⁶ The synthetic molecule, 3,5-difluorobenzoyl-l-glutamate (3,5-DFBGlu, Figure 3c), was suggested to be a CPG2 substrate whose conversion could be imaged with CEST. Due to the fact that 3,5-DFBGlu conversion releases glutamate, through which its exchangeable protons (Figure 3c, Δω = 3.0 ppm) generate CEST contrast,¹⁰ a turn-on CEST contrast is obtained upon CPG2 activity. In this case, the different k_ex (and, therefore, the CEST contrast) of amide (in 3,5-DFBGlu) and amine (in glutamate) functional groups was exploited to monitor CPG2 enzyme activity with CEST MRI.⁴³ Note that the resulted effect on CEST contrast upon changing k_ex can be predicted in silico using mathematically simulations,^47,48 which may be beneficial for screening larger arrays of potential CEST probes.

In order to maximize the CEST imaging capabilities of genetically encoded enzyme/substrate systems, the following requirements should be fulfilled: (i) sufficient CEST contrast, (ii) efficient conversion of the substrate by the target enzyme, and (iii) selectivity, meaning the substrate should be minimally converted by other endogenous enzymes. The herpes simplex virus type-1 thymidine kinase (HSV1-tk) is an enzyme that phosphorylates a wide range of nucleoside analogues, including radioactive modified nucleosides that are accumulated in HSV1-tk-expressing cells upon their phosphorylation. This mechanism of action has been used to monitor many biological targets, using PET or SPECT in preclinical studies and in patients.^49–54 The large chemical shift offsets of imino protons of nucleosides from water protons make them potential CEST biosensors. However, the relatively fast k_ex of the imino protons of uridine and thymidine (k_ex = 3000–5000 s⁻¹ Δω = 6.2 ppm)^55,56 reduces their potential to be used as CEST sensors.

On the basis of the right-hand section of the rational design workflow (Figure 2), it is possible to transform thymidine into a CEST MRI reporter probe, which is a specific substrate for HSV1-tk but not for endogenous cellular kinases.^40,41 Hydro-genation of the thymidine double bond led to a reduction in the aromaticity of the product, dihydrothymidine (DHT), and to a dramatic increase in the pK_a of the imino proton as compared to thymidine. This resulted in a reduced k_ex and improved CEST characteristics (k_ex = 800–1700 s⁻¹, Δω = 5.0 ppm). Importantly, the pK_a value of the exchangeable protons of the synthetic CEST probe can be predicted in silico prior its preparation, using computational software,^57,58 enabling the screening of large numbers of potent probes.⁴¹ Although DHT generates high CEST contrast, the similarity of its structure to the native nucleoside thymidine leads to its phosphorylation by endogeneous kinases and therefore additional chemical modification is required to obtain further specificity to HSV1-tk. Therefore, the addition of a methyl group at the 5-position of DHT, which results in 5-methyl-dihydrothymidine (5-MDHT, Figure 3c), makes it a preferable CEST probe with high specificity to HSV1-tk and a negligible conversion by other cellular kinases. Accordingly, following i.v. administration, 5-MDHT accumulates only in HSV1-tk-expressing cells, thus enabling the differentiation of these cells from wild-type cells using CEST MRI.^40,41

4. GENETICALLY ENCODED PROTEINS AS CEST BIOSENSORS

The advantage of using fluorescent proteins as reporter genes is that no substrate administration is required.⁵⁹ This eliminates challenges such as probe delivery, accessibility, and toxicity. In MRI, reporter genes have been developed based on overexpressing iron storage and transporting proteins, which affect the T₂ and T₂* water relaxation times.^60–62 Although ferritin has been used in vivo in a variety of biological applications,^{60,61,63–65} it was shown that iron preloading of transfected, transplanted cells is important for a reliable imaging of cells in vivo.⁶⁶

4.1. CEST Reporter Genes.

Positively charged polymers were suggested as potential CEST contrast agents in the early 2000s.⁶⁷ One such polymer, poly-l-lysine, generates exceptionally high CEST contrast when the saturation pulse is applied at the resonance frequency of its amide protons (Δω = 3.6 ppm). An artificial protein consisting of 200 lysine residues (termed lysine-rich protein, LRP, Figure 4a), which was genetically engineered and expressed in cells, was therefore proposed as a new MRI reporter gene.⁶⁸ The applicability of LRP was recently demonstrated to be a valuable tool for direct monitoring of oncolytic virotherapy.⁶⁹ Subsequently, libraries of a variety of peptides were screened to identify optimal amino acids and peptide sequences for obtaining CEST contrast.^70,71 Among these sequences, arginine-rich peptides and proteins showed favorable CEST contrast originating from both their guanidine protons (Δω = 1.8 ppm) and amide protons (Δω = 3.6 ppm). The arginine-rich protein protamine from salmon, which consist of more than 60% arginine residues, showed a superior high CEST contrast.⁷² It was demonstrated that human protamine 1 (hPRM1, ~50% arginine, Figure 4b) can be used as a humanized CEST MRI reporter gene that allows differentiation between wild-type cells and cells that overexpress hPRM1.⁷³ One advantage of using genetically encoded systems is that libraries of biosensors can be generated using molecular biology tools (e.g., random mutagenesis and directed evolution), which can be used to select improved variants of the sensor.⁷⁴ However, efficient high-throughput screening, which is key for such an approach, should be carefully considered in order to allow the performance of this procedure within a manageable time scale. For example, for screening that relies on optical methods, an automated, confocal, real-time, and single-cell kinetic imaging system has been integrated into a single, compact unit.⁷⁵ Such a system will facilitate the screening and narrow it from millions of clones to a number of clones that can be scanned with MRI. Recently, lysine- and arginine-rich mutants of green fluorescent (GFP) proteins, aka supercharged GFP, showed increase in CEST contrast without compromising the fluorescent properties of the protein.⁷⁶ Thus, such reporters can be used as a template for high-throughput screening.

Figure 4.

CEST reporter genes. CEST reporter gene genes based on artificial (LRP) (a) and human (hPRM1) (b) proteins. (c) Genetically engineered responsive CEST reporter gene designed to image PKA activity. The full sequence of the protein is given by the single-letter amino acid code (i.e., K for lysine, R for arginine, and S for serine). The chemical structure of part of the sequence is presented.

4.2. Responsive CEST Reporter Genes.

Protein kinases are enzymes that that play a key role in cellular signaling, with protein kinase A (PKA) being one of the most investigated ones. It is not surprising that many imaging sensors were therefore developed to monitor its activity, including the genetically encoded sensor for PKA regulation based on fluorescence resonance energy transfer (FRET).^77,78 Metal-loprotein-based sensors have also been developed to image PKA activity with MRI.⁷⁹

As it is known that PKA phosphorylates the hydroxyl group (–OH) of serine (or threonine) with high specificity toward the peptide sequence LRRASLG,⁸⁰ the opportunity exists to image PKA activity with CEST MRI using a synthetic peptide (Figure 4c).⁸¹ Interestingly, the CEST contrast obtained from the guanidine protons (Δω = 1.8 ppm) and the amide protons (Δω = 3.6 ppm) of the LRRASLG peptide was quenched upon its phosphorylation to LRRASpLG. Therefore, an artificial gene that encodes eight tandem repeats of the LRRASLG sequence was developed. This PKA biosensor, expressed in Escherichia coli, showed a higher CEST contrast compared to that of nontransfected controls. In this case, a turn-off CEST sensor was created to monitor cellular signaling. This concept can be expanded to other protein kinases as long as the phosphorylated sequence contains arginine and lysine residues that generate the CEST contrast and a phosphorylation site that allows conditional quenching of the signal.

5. HETERONUCLEAR (¹⁹F AND ¹²⁹XE) CEST BIOSENSORS

One main intrinsic limitation of MRI contrast agents in general and of CEST agents in particular is the large concentration of water protons of the target tissue. Although the high proton concentration makes MRI a robust imaging methodology with a high signal-to-noise ratio (SNR), low concentrations of biosensors are less effective in altering the overall water ¹H MRI signal. Particularly for CEST contrast, submillimolar to several millimolar concentrations of the contrast agents are required for a detectable CEST contrast.⁸ Therefore, if the aim is to image a low-concentration target (e.g., metal ion levels or receptor of interest), then an alternative to ¹H CEST should be considered. Molecules with heteronuclear (i.e., other than ¹H) atoms, such as ³He, ¹³C, ¹⁵N, ¹⁹F, ³¹P, and ¹²⁹Xe, have been suggested as sensors for MR-based applications due to the fact that their MR-active isotopes do not exist (or occur in negligible concentrations) in normal biological tissues.

Heteronuclear NMR experiments using saturation transfer modules were used to measure the exchange between two pools of (nonproton) nuclear spins^82–84 much earlier than reported in the original CEST MRI papers.^6,7 The combination of heteronuclear NMR and CEST opened a new avenue for the design of MRI sensors since it exploits the benefits of both methodologies, i.e., (i) the amplification effect of the CEST mechanism, and from the use of heteronuclear spins; (ii) the large chemical shifts (several hundreds of ppm for nonproton spins); (iii) the high sensitivity of the obtained Δω to the local environment; and (iv) the lack of background signal. These advantages form the basis for the design and use of ¹⁹F and ¹²⁹Xe CEST biosensors.

5.1. ¹⁹F-CEST.

Metal ions play a crucial role in myriad biological processes, and the ability to monitor real-time changes in metal ion levels is essential for understanding a variety of physiological events. The design and use of fluorescent sensors to study variable changes in cellular levels of metal ions allows us to understand their role in cell biology.⁸⁵ Calcium (Ca²⁺) signaling is used by all cell types to carry information and regulate various biochemical processes. In 1980, a new technology was developed that enabled the detection of intracellular levels of Ca²⁺ by using the synthetic fluorescent dye, 1,2-bis(o-aminophenoxy) ethane-N,N,N,N-tetraaceticacid (BAPTA).⁸⁶ This technology revolutionized our understanding of Ca²⁺ physiology and kindled the development of a variety of synthetic-⁸⁵ and genetic⁸⁷-based calcium indicators. Although exciting advances in molecular biology now provide the opportunity to obtain long-term functional imaging of Ca²⁺ signaling in vivo via genetically encoded Ca²⁺ indicators based on fluorescent proteins,^88,89 the light source on which these methodologies are based is an obstacle with regard to deep tissue penetration and applicability for large animals and humans.

Interestingly, synthetic fluorinated derivatives of BAPTA showed a unique Δω on the ¹⁹F NMR spectrum and a different dissociation constant (K_d) between ion-bound and free F-BAPTA when fluorine was substituted at a different positions of the aminophenoxy ring (3-, 4-, 5-, or 6-position).^90,91 The 5F-BAPTA derivative was identified as a potential sensor for monitoring Ca²⁺ ions by combining ¹⁹F MRI and CEST MRI, since the k_ex between Ca²⁺-bound and free 5F-BAPTA was reported to be 150–550 s⁻¹,^84,91 an optimal k_ex for a Δω of ~6 ppm. In addition, the possibility of transferring magnetization between Ca²⁺-bound and Ca²⁺-free 5F-BAPTA during NMR experiments was demonstrated.⁸⁴ By performing a ¹⁹F CEST MRI experiment, low concentrations (500 nM) of Ca²⁺ ions could be imaged with high specificity (compared to competing ions). This resulted from the unique Δω (in the ¹⁹F-NMR spectrum, Figure 5a) between free 5F-BAPTA and ion-bound 5F-BAPTA resonances for each ion (e.g., Ca²⁺, Zn²⁺, or Mg²⁺), and the different k_ex between free and bound 5F-BAPTA for each examined ion. While Zn²⁺-5F-BAPTA or Mg²⁺-5F-BAPTA exchanged with free 5F-BAPTA too slow or too fast, respectively, to be detected by 19F CEST, the k_ex between Ca²⁺-5F-BAPTA and free bulk 5F-BAPTA allows the transfer of saturation and thus Ca²⁺ detection.¹⁵ This approach, termed ion CEST (iCEST), can be further extended and optimized to monitor different metal ions at variable levels by the rational design and synthesis of fluorinated ion chelates. One main advantage of the iCEST methodology is that the high specificity toward an ion of interest is reflected by two parameters, Δω and k_ex, making this approach more ion specific. In an additional report, it was shown that further modification of 5F-BAPTA to obtained 5,5,6,6-tetrafluoro-BAPTA (TF-BAPTA) as the ¹⁹F iCEST probe allows the detection of both Zn²⁺ and Fe²⁺ ions specifically and simultaneously and thus is termed multi-ion CEST (miCEST).⁹² For novel probe design, such an approach eliminates the need to attach an imaging-responsive moiety to the binding chelate, as required for metal-based MRI sensors,^93,94 allowing a straightforward, rational synthetic design of ion sensors. Finally, the rising interest in ¹⁹F MRI applications^95,96 as an alternative to (super)paramagnetic sensors^4,97 may lead to further use of ¹⁹F-based MRI

Figure 5.

Heteronuclear CEST (HeteroCEST) sensors. (a) Schematic depiction of the dynamic exchange process between free 5F-BAPTA and M²⁺-bound [M²⁺-5F-BAPTA] and the obtained ¹⁹F NMR spectra of 5F-BAPTA in the presence of different M²⁺ ions. (b) Two suggested systems for HyperCEST. (i, left and middle panels) Cryptophane cage-based system: The chemical structure of the Xe biosensor illustrating the cryptophane-A cage (green), the linker (black), and the targeting moiety (biotin in this case, orange) and the ¹²⁹Xe-NMR spectrum of free and caged Xe gas. The middle panel shows a schematic illustration of the HyperCEST concept. (ii, right panel) Schematic illustration of Xe atoms that exchange between the interior of the nanoemulsion droplets and the bulk aqueous pool, which leads to two unique peaks in the ¹²⁹Xe spectrum. Adapted from refs 15 and 100. Copyright 2013 American Chemical Society. Adapted with permission from ref 16. Copyright 2006 American Association for the Advancement of Science.

5.2. Hyperpolarized ¹²⁹Xe-CEST.

Since hyperpolarized nuclear spins show a nearly 10⁴-fold signal enhancement over nonpolarized elements, hyperpolarized-based approaches have the potential to significantly impact the field of molecular and cellular MR imaging. Hyperpolarized ¹³C has been extensively studied and has already been translated to the clinic.^{98 129}Xe can also be hyperpolarized⁹⁹ and is an interesting nucleus for molecular and cellular MRI applications. In an elegant approach, the combination of hyperpolarized ¹²⁹Xe gas and CEST (Hyper-CEST) was demonstrated.^16,100 By exploiting the chemical shift difference between entrapped and free hyperpolarized ¹²⁹Xe gas, the potential of using Hyper-CEST for molecular and cellular imaging applications was shown (Figure 5b).

Cryptophanes are a group of molecular cages that can reversibly encapsulate a variety of guests, including halomethanes, ammonium salts, and noble gases, in particular Xe.¹⁰¹ The ¹²⁹Xe NMR spectrum of the Xe–cryptophane complex in the presence of an excess of xenon shows two well-separated (Δω) signals at 229.5 and 62.3 ppm, which are assigned to free xenon in solution and to xenon trapped in the cryptohane cavity, respectively. Exploiting this Δω and by binding a biotin-labeled cryptophane cage to avidin-functionalized agarose beads in an aqueous environment, an extremely low concentration of the biosensor (5 μM cryptophane) could be observed using HyperCEST.¹⁶ The proof of concept of performing HyperCEST was recently established¹⁰² and extended to monitor live cells that were loaded with fluorescently labeled cryptophanes.¹⁰³ Cryptophane-A was also used to tag antibodies for targeted detection of cellular antigen CD14.¹⁰⁴ In a different route, surfactant-stabilized perfluorocarbon-in-water nanoemulsions (Figure 5b) were suggested as molecular sensors for HyperCEST applications.¹⁰⁰ Taking advantage of the Δω between the hyperpolarized ¹²⁹Xe at the aqueous solution and the interior of nanoemulsions and of the fact that ¹²⁹Xe freely exchanges between the two compartments, it was shown that the nanoemulsion can be detected at concentrations as low as 100 fM, corresponding to <1 μL of perfluorocarbon per liter of solution using HyperCEST. Due to the difference in the chemical shift of these two cages, it was possible to detect simultaneously two cell populations.¹⁰⁵ Finally, genetically encoded gas nanovesicles from buoyant microbes were used as cages that entrap hyperpolarized Xe. These nanovesicles are based on gas-binding protein nanostructures expressed by certain buoyant microorganisms, which can be cloned and expressed in bacteria. However, since it is a cluster of genes, a proper expression in mammalian cells maybe challenging. These nanovesicles are different in size and therefore displayed maximal hyperCEST contrast at different saturation frequencies.¹⁰⁶ These new strategies for molecular-engineered nanosize HyperCEST sensors have the potential to be further explored and to be added to the HyperCEST toolbox for imaging molecular and cellular events.

6. ADDITIONAL TECHNICAL CONSIDERATIONS

It is important to note that when using high-power RF pulses in a CEST experiment a direct saturation of the MR frequency of the surrounding water molecules occurs. This is an impediment to using CEST probes with a relative small Δω (0–2 ppm) and calls for the design and of CEST sensors having a larger Δω of their exchangeable protons.¹⁰⁷ Furthermore, the use of strong RF pulses may lead to exceeding approved specific absorbance rates (SAR), which is a measure of the potential for RF-induced heating. This has to be taken into account when CEST experiments are being performed in vivo, especially in a clinical setup. To overcome these SAR limitations, the commonly used continuous wave (CW) RF irradiation cannot be used. Alternative schemes using short repeated RF irradiation pulses (pulsed-CEST imaging) may be considered.^108,109

Although the absolute concentration of CEST sensors may be assessed quantitatively, many variables may limit an accurate quantification in vivo. This includes the unknown number and properties of other exchangeable pools present in the studied region and the T₂ exchange (T_2ex) effect.¹¹⁰ Endogenous magnetization transfer may also complicate the determination of CEST sensor concentrations.¹¹¹ In addition, pH changes that affect the CEST signal magnitude also interfere with quantification. Finally, B₀ and B₁ inhomogeneities restrict the ability to determine the concentration of the CEST sensor solely from the obtained CEST contrast. While B₀ and B₁ inhomogeneities can be mapped^112,113 and corrected post data acquisition, the determination of the effect of the other abovementioned parameters on CEST amplitude (number of exchangeable pools, T_2ex, endogenous MT, and pH) is not straightforward. Therefore, if applicable, the acquisition of baseline CEST data prior to the CEST sensor administration is recommended, allowing the normalization of the CEST contrast relative to endogenous contributions.

7. CONCLUSIONS

In summary, in light of the recent demand for new tools that will allow better investigation of complex processes, a new field has evolved at the interfaces of synthetic chemistry, molecular engineering, and cellular imaging. It is now clear that it would be very difficult to find a single perfect agent for imaging all biological targets. Therefore, a better approach is to tailor the imaging probe or the sensor to perform a specific task. The rising field of CEST MRI offers chemists and biologists the opportunity to design large arrays of metal-free MRI sensors that can tackle particular questions. To that end, following the workflow described in Figure 2 may be beneficial for the development of a desired sensor. After identifying the biological target, a potential sensor should be considered (natural or unnatural). If required, chemical modification or genetic manipulation of the sensor of interest should be performed based on a prediction (in silico, if applicable) of its properties. Screening of several optional probes may be performed prior to the selection of an optimal molecular probe that generates either CEST-off or CEST-on contrast for in vitro and in vivo applications. Following these steps may lead to the development of future nonmetallic biosensors for investigating complex biological questions.

KEYWORDS

(Bio)sensor

an imaging agent used to monitor a biological phenomenon or target

Contrast agent

An imaging probe that changes the target signal in the image relative to the background signal, creating contrast

Small molecule biosensor

Nonpolymeric chemical substance with low molecular weight, e.g., sugar, amino acid, nucleoside, or other naturally occurring or synthetic compound

Genetically engineered enzyme/substrate reporter system

A system that consists of two components, (i) a recombinant enzyme and (ii) its substrate, which allows imaging of enzyme expression and/or activity

Reporter gene

A gene whose product acts as a beacon for a certain imaging modality

Responsive reporter gene

A reporter gene that changes its imaging properties (enhances signal, changes color, etc.) upon a change in biological events (i.e., enzymatic activity, ion levels, regulation of cellular function)

CEST

Chemical exchange saturation transfer

HeteroCEST

CEST experiments that do not involve protons (e.g., ¹⁹F or ¹²⁹Xe)

iCEST

Ion CEST. A methodology that enables the detection of metal ions using their fluorinated chelate when performing ¹⁹F CEST MRI

HyperCEST

Hyperpolarized CEST. A methodology developed for molecular imaging applications using hyperpolarized ¹²⁹Xe gas as the imaging agent