Advantages of paramagnetic CEST complexes having slow-to-intermediate water exchange properties as responsive MRI agents

Abstract

Paramagnetic saturation transfer chemical exchange (PARACEST) complexes are exogenous contrast agents that have great potential to further extend the functional and molecular imaging capabilities of magnetic resonance. Due to the presence of a central paramagnetic lanthanide ion (Ln³⁺ ≠ La³⁺, Gd³⁺, Lu³⁺) within the chelate, the resonance frequencies of protons and water molecules bound to the PARACEST agent are shifted far away from the bulk water frequency. This large chemical shift combined with an extreme sensitivity to the chemical exchange rate make PARACEST agents ideally suited for reporting significant biological metrics such as temperature, pH, and the presence of metabolites. Also, the ability to turn PARACEST agents “off” and “on” using a frequency selective saturation pulse gives them a distinct advantage over Gd³⁺-based contrast agents. A current challenge for PARACEST research is translating the promising in vitro results into in vivo systems. This short review article first describes the basic theory behind PARACEST contrast agents, their benefits over other contrast agents, and their applications to magnetic resonance imaging. It then describes some of the recent PARACEST research results. Specifically, pH measurements using water molecule exchange rate modulation, T₂-exchange contrast due to water molecule exchange, the use of ultra-short echo times (TE<10 μs) to overcome T₂-exchange line-broadening, and the potential application of T₂-exchange as a new contrast mechanism for magnetic resonance imaging.

1. Introduction

MR imaging agents

Magnetic resonance (MR) imaging uses a strong magnetic field and radiofrequency (RF) radiation to create an image representing the spatial distribution of water molecule protons within the subject (1,2). Since the human body is composed of mostly water (about 70 %) almost every tissue type generates an MR signal. This is one reason why MR gives such excellent anatomic detail, where the various tissue types (differentiated by their endogenous T₁ and T₂ relaxation times and proton densities) appear slightly brighter or darker than surrounding tissue (3). Yet, if two different tissue types have similar T₁ and T₂ values (and proton densities) they will appear similar in standard MR images. This can create a problem, for example, when trying to differentiate between cancerous and healthy tissue, especially at the early stages of disease. In order to highlight specific anatomic features or measure certain dynamic processes, the endogenous T₁ and T₂ tissue contrast can be further enhanced by modifying the echo and repetition times (TE and TR respectively) in the RF pulse sequence. If the endogenous contrast is insufficient then exogenous contrast agents can be introduced into the subject to further extend the level of contrast. Currently, the most common exogenous MR contrast agents are chelates that consist of an organic ligand and a Gd³⁺ ion (4). Among the lanthanide ions, Gd³⁺ has the largest paramagnetic relaxation enhancement of water protons and exhibits no pseudocontact (i.e., hyperfine) shift in nearby protons (5). Rapid exchange of water molecules between a single inner-sphere binding site on Gd³⁺ and bulk water reduces the effective T₁ of bulk water protons causing regions of agent uptake to appear brighter than the surrounding tissue, allowing those regions to be easily identified. Conversely, T₂* agents such as Dy³⁺-based chelates and superparamegnetic iron oxide nanoparticles (SPIO) reduce the effective T₂ of bulk water protons by creating B_o inhomogeneities within the subject, causing regions of agent uptake to appear darker than the surrounding tissue.

A relatively new method for both endogenous and exogenous MR contrast uses chemical exchange saturation transfer (CEST). The CEST method creates negative contrast (i.e., darkening) in magnetic resonance images using a combination of spin saturation and proton exchange with bulk water (6). Protons bound to a CEST agent are frequency shifted away from the bulk water frequency with the difference defined as Δω. RF spin saturation at the exchanging proton frequency (Δω) can produce indirect partial saturation of the bulk water proton pool through chemical exchange, provided that the rate of exchange (k_ex) between the bound and bulk proton pools is slower than their frequency difference (i.e., k_ex ≤ Δω) (7). The magnitude of the bulk water signal reduction is a function of concentration of exchangeable protons associated with each CEST agent, the endogenous T₁ relaxation time of tissue water protons, the RF saturation power and duration, and the proton exchange rate. For further details about CEST theory, please refer to the article by McMahon, et al., which appears in this same special issue.

Similar to the initial diamagnetic chemical exchange saturation transfer (DIACEST) agents (7), paramagnetic chemical exchange saturation transfer (PARACEST) agents also create negative contrast. The main difference between DIACEST and PARACEST agents is the latter contain a paramagnetic ion chelated by an organic multidentate ligand. Most often, the paramagnetic ion is a lanthanide ion (Ln³⁺ ≠ La³⁺, Gd³⁺, or Lu³⁺) (8), but complexes consisting of transition metal ions like Fe²⁺ have also been reported (9). The presence of the paramagnetic ion gives PARACEST agents two main advantages over DIACEST agents. First, the presence of a paramagnetic center typically increases the chemical shift of any ligand associated –NH and –OH proton resonance by approximately an order of magnitude or more. For example, the –NH and –OH proton chemical shifts of a typical DIACEST agent range from 1 to 5 ppm whereas the chemical shift of a similarly exchanging proton in a PARACEST agent can be 50 ppm or even larger (10). Second, the paramagnetic center often provides an additional exchange mechanism via molecular exchange of a water molecule residing in the inner coordination sphere of a lanthanide ion with other water molecules in the bulk solvent. The ability to have molecular exchange is significant because the bound water molecule chemical shift (Δω) can range from 50 ppm (for Eu³⁺) to as much as −720 ppm (for Dy³⁺) (8). Large Δω values such as these allow PARACEST agents to have faster chemical exchange rates than DIACEST agents while remaining below the intermediate exchange limit (i.e., k_ex ≤ Δω). A large Δω value also allows RF saturation at the bound proton frequency without direct saturation at the bulk water frequency, as well as the possibility of creating new agents having exchange frequencies (Δω) outside the tissue magnetization transfer (MT) window.

The tissue MT signal arises from dipolar exchange of protons with endogenous tissue macromolecules and typically spans from approximately +100 to −100 ppm (11). The MT signal can be much larger than the signal from a PARACEST agent at low concentrations and can severely mask the contrast produced by the agent. In most cases, water molecule exchange in a PARACEST agent is much faster than typical –NH or –OH proton exchange, therefore PARACEST agents require more saturation power (higher B₁) than DIACEST agents to achieve the same level of bulk water saturation. Unfortunately, the use of higher power B₁ pulses of not only increases the MT signal dramatically but also increases the chance of reaching specific absorption rate (SAR) safety limits.

As mentioned previously, the magnitude of a CEST signal depends upon the number of exchanging protons per molecule, whether it be a DIACEST molecule or a PARACEST agent. Typically there are more –NH or –OH proton exchange sites per molecule than water exchange sites in a PARACEST agent, giving proton exchange an apparent advantage over water exchange mechanisms. However, the higher water molecule exchange rates characteristic of most PARACEST agents can compensate for this concentration difference to some extent simply because one can saturate more protons per unit time during the application of a saturating RF pulse. Also, it has been demonstrated that the number of lanthanide ions per molecule of agent (and correspondingly the number of water molecule exchange sites) can be increased by a factor of 20 or more using simple polymerization procedures (12).

All together, DIACEST and PARACEST agents have shown great potential to further extend the imaging capabilities of diagnostic magnetic resonance imaging. Several of these applications are described for PARACEST agents in the next section.

PARACEST Applications

Since they were first described around a decade ago (6), numerous in vitro demonstrations of PARACEST agents in MR imaging have been reported ranging from general anatomic contrast to functional (13), cellular (14), and molecular imaging as well (15,16). This wide range of potential applications originates with the exquisite sensitivity of the CEST signal to proton and/or water molecule exchange rates. Here, the magnitude of CEST produced by a PARACEST (or DIACEST) agent is defined as the percent drop in the bulk water signal intensity when comparing the “Off” (saturation at −Δω) to the “On” (saturation at Δω) MR signal (i.e., (1−On/Off)×100) (6). If the proton or water molecule exchange rate is too fast or too slow, CEST is not observed and the agent is considered “inactive”. However, at rates intermediate between these two extremes, a reduction in the bulk water signal is observed upon RF activation at frequencies near Δω. This is best illustrated by collecting a complete CEST spectrum (a plot of bulk water intensity versus RF saturation frequency) of any potential new agent. The general shape and bandwidth of the exchange peaks in a CEST spectrum provides qualitative insights into the exchange mechanism and can provide quantitative data about the exchange rates when the spectra are fit to a proper model (17).

The sensitivity to exchange rate can be used to measure certain microenvironment biometrics surrounding the PARACEST chelate such as temperature and pH. For example, PARACEST agents have been shown to measure temperature using both a ratiometric method and a CEST exchange frequency method, where both methods were independent of agent concentration (18,19). Likewise, the sensitivity of –NH proton exchange to pH was demonstrated in the initial CEST publication on DIACEST agents (7). This was soon followed by several publications demonstrating pH measurements using the –NH proton exchange of PARACEST agents and a concentration-independent ratiometric method (10,18,20). It was recently shown that pH can also be accurately measured using the water molecule exchange of a Eu³⁺DOTA-based PARACEST agent and a ratiometric method (21), of which further details are described later in this article.

The proton and water exchange rates in a PARACEST agent are often found to be sensitive to nearby binding events, therefore this feature has been utilized to create metabolite-responsive agents. For example, it was shown that the PARACEST signal is not observed for Eu³⁺DOTA-2M-2PB in the absence of glucose because the water molecule exchange rate is too fast. Yet when bound to a single glucose molecule, the rate of water molecule exchange slows enough to turn on the PARACEST effect (22). This agent has been used to detect glucose being released from a perfused mouse liver after the addition of glucagon (23). In a second example, the PARACEST signal from a different Eu³⁺DOTA-based agent was deactivated upon binding to a single Zn²⁺ ion because, in this case, the presence of a Zn²⁺ ion catalyzed proton exchange from a single Eu³⁺-bound water molecule so effectively that the exchange rate no longer fit into the intermediate regime (24). This responsive agent could potentially be used to image the release of insulin from pancreatic beta-cells since insulin release also involves release of free Zn²⁺ ions into the surrounding tissue regions (13). Similar techniques have also been used to measure nitric oxide (25), enzyme activity (26), and phosphate esters (27).

Review Objectives

This review summarizes the evolution of responsive PARACEST agents, from those that show a change in CEST peak intensity in response to some physiological event, to new class of agent that switches CEST peak frequency with changes in pH. The advantage of the new class of agent is quite evident as one considers in vivo applications where the concentration of agent may not be well known. A second deterrent for in vivo applications of such sensors is the excessive line broadening contribution to the tissue water proton brought about by the simple presence of a paramagnetic complex having intermediate-to-slow water exchange kinetics. We demonstrate that this T₂ exchange (T_2ex) contribution to the tissue water line-width can be overcome by the use of ultra-short echo pulse sequences (e.g., UTE, SWIFT) for CEST imaging. Finally, the potential application of T_2ex as a new T₂ contrast mechanism, chemical approaches to optimize the sensitivity of such agents, and the challenges of translating PARACEST agents for in vivo measurements on a clinical scanner are also discussed.

2. Responsive PARACEST agents

A major focus in the development of MRI contrast agents for molecular imaging has been in the design of agents capable of responding to biologically important indices such as specific enzyme activity, pH, temperature, or tissue redox. CEST has been recognized as one of the best imaging methods available for measuring such physiological parameters for several reasons: 1) image contrast can be turned “on” and “off” at will by applying a frequency-selective radio frequency pulse; 2) CEST contrast is exquisitely sensitive to proton exchange rates (k_ex) and these are easily modulated by chemistry and/or physiology so responsive CEST agents are relatively easy to design; 3) it is possible to create responsive agents that provide for concentration-independent quantitative measures using ratiometric imaging methods. These features are not easily available in Gd³⁺-based agents.

It is now well established that water exchange in Eu³⁺-DOTA-tetraamide complexes occurs via a dissociative mechanism (28) and this translates into an important fundamental property - water exchange can be modulated in such paramagnetic complexes by altering the electronic properties of the ligand. A general mechanism, which appears to hold for all agents reported so far, is that electron-donating substituents on a DOTA-amide ligand donate more electron density through the coordinating amide onto the central lanthanide ion and this weakens the lanthanide ion-water molecule interaction and promotes faster water exchange. Conversely, electron-withdrawing groups on the amides would tend to slow the rate of water exchange. This was demonstrated experimentally in a series of complexes in which one of the appended amide side-arms was modified to vary the electron-donating ability of the ligand (Figure 1) (29). The experimental results were consistent with the expectation that k_ex was slower for the complexes with an electron withdrawing group (-CO₂tBu, -CN) and faster in those with an electron-donating group (-OMe). Even though the frequency of the bound water exchange CEST peak in these complexes varied somewhat, the magnitude of the CEST signal showed the largest variation with alterations in ligand electronics (Figure 1). These experimental results catalyzed the very important idea that responsive PARACEST agents could be built using the concept of variable electron donating ability. Such a platform design would be even more powerful if one could create systems where the ligand donor atoms or appended groups produce changes in CEST resonance frequency (Δω) rather than CEST intensity.

Figure 1.

Top: chemical structures of Eu(III) complexes; Bottom: CEST profiles arising from the coordinated water molecules of those complexes. The figure is reproduced from reference (29).

Bleaney’s theory of magnetic anisotropy (30) predicts that the second-order crystal field coefficient B_o² determines the dipolar NMR shift of paramagnetic lanthanide complexes. This was confirmed experimentally by correlating optical and NMR spectral data for a series of macrocyclic lanthanide complexes (30). Based on that study, the relationship between ligand electronic properties, k_ex and Δω could be summarized as follows: ligands that produce stronger ligand fields result in complexes with larger frequency shifts (larger Δω) and faster water exchange. This relationship was nicely demonstrated in a comparison of CEST properties for various DOTA-tetraamide ligands (31) prepared from proline, leucine or methionine (Figure 2). Density functional theory (DFT) calculations indicate that the negative charge density on the amide oxygen in this ligand series increased in the order, DOTA-(met)₄ < DOTA-(leu)₄ < DOTA-(pro)₄. More importantly and quite unexpectedly at that time, the three individual complexes showed dramatically different water exchange frequencies (Δω) and lifetimes. In this series, the water exchange peak was found at Δω = 64, 54, and 45 ppm for EuDOTA-(pro)₄, EuDOTA-(leu)₄, and EuDOTA-(met)₄, respectively, while the bound water lifetimes (τ_M) varied from 55, 81 and 200 μs for this same series (Figure 2). This verified experimentally for the first time that complexes with larger water exchange frequencies (an advantage for in vivo imaging) have the shortest bound water lifetimes (a disadvantage for PARACEST activation). Nevertheless, the approximate 5 ppm differences in Δω observed for these three EuDOTA-tetraamide complexes was sufficient to allow selective imaging of each individual agent by applying a long presaturation pulse at the appropriate offset frequency. This result led to a new design principle for responsive agents whereby the triggering event (change in pH, enzyme activity, etc.) results in a change in Δω with, preferably, little to no change in τ_M.

Figure 2.

Top: chemical structures of Eu(III) complexes; Bottom: CEST profiles arising from the coordinated water molecules for Eu(III)-DOTA-(met)₄ (red), Eu(III)-DOTA-(leu)₄ (green), and Eu(III)-DOTA-(pro)₄ (blue). The figure is reproduced from reference (31).

The first example of this new type of responsive PARACEST agent based on the above design principle was built by replacing one of the amide side chains in DOTA-(gly)₄ with ketone-phenol unit (Figure 3) (21). This particular chemical entity was chosen because deprotonation of the phenolic proton would result in a delocalization of the excess negative charge on the phenolate oxygen through the aromatic ring and onto acetyl oxygen atom yielding a quinone-like resonance structure with negative charge on the donor oxygen atom bonded to the Eu³⁺. This transformation increases the overall ligand-field strength produced by the ligand that in turn results in larger hyperfine shifts in all NMR active nuclei in the complex. This is also reflected in a larger hyperfine shift of the exchanging water molecule that results in a downfield shift from 50 to 54 ppm as the phenolic proton is removed between pH 6.0 and 7.6 (Figure 3). This unique CEST property proved to be key for direct imaging of pH using ratiometric principles. For example, the ratio of bulk water intensities after presaturation of the agent at 55 ppm (high pH position) versus 49 ppm (low pH position) was found to be linearly proportional to pH and independent of agent concentration. In addition, the τ_M of deprotonated and protonated species were estimated at 120 and 239 μs, respectively, consistent with the theory described above.

Figure 3.

Left: and the deprotonation of the phenolic proton in the pH responsive PARACEST agent results in conjugation of the resulting quinone-like structure with the acetyl oxygen atom coordinated to the Eu(III) ion. Right: pH dependence of CEST spectra for the corresponding agent. Reproduced with permission from (21) © American Chemical Society.

3. T₂ contrast from slow-to-intermediate water exchanging systems

One of the current goals in PARACEST agent detection by MR imaging is to translate these promising in vitro results to in vivo applications, specifically mouse models of human disease. Upon doping a PARACEST agent containing a small amount of free ligand with ⁶⁴Cu (a β⁺ emitter) to follow the tissue biodistribution by positron emission tomography (PET), it was easy to demonstrate that the untargeted Eu³⁺DOTA-based agents clear via renal filtration and are quickly deposited in the bladder (32). For a typical clinical dose (0.1 mmol/kg), both the kidneys and bladder showed a strong PET signal with signal remaining in the kidneys for over an hour. Imaging the agent within the bladder using standard spin-echo or gradient-echo pulse sequences and simple “Off” (saturation at −Δω) minus “On” (saturation at +Δω) PARACEST techniques was straightforward and similar to in vitro imaging (6,32). Yet in the kidneys, where the tissue MT signal provides a significant background and the agent concentration is much higher, the imaging results were markedly different. In this case, the MR signal intensity in kidney images decreased by more than 60% simply due to the presence of the PARACEST agent. Notably, this decrease in signal intensity occurred even before a presaturation pulse was applied to activate the agent. This indicated that the PARACEST agent was behaving like a susceptibility or T₂^* contrast agent (33), which was quite surprising for such a weakly paramagnetic complex. A similar T₂ effect in kidney images was recently observed using a Tm³⁺-based PARACEST agent (34). This effect is problematic for imaging PARACEST agents in vivo because having the regions of agent uptake appear dark in both the “Off” and “On” images means that the desired effect of the PARACEST agent itself (i.e., “Off” minus “On”) is more difficult to detect.

Since Eu³⁺ has little paramagnetic relaxation effect on either T₁ or T₂ in comparison to several other lanthanide ions (5), it was clear that the bulk water T₂ line broadening originated with some chemical feature of this slow-to-intermediate exchanging system other than magnetism. It has since been shown that the same slow-to-intermediate water molecule exchange feature that enables CEST can also cause a local decrease in the bulk water T₂ through the T₂-exchange (T_2ex) mechanism (35). Diamagnetic molecules containing exchangeable –NH and –OH protons have long been used to suppress the strong bulk water peak in high-resolution proton NMR experiments by reducing the bulk water T₂ through chemical exchange (36–38). Added line broadening due toT_2ex has also been reported for diamagnetic CEST agents (39) including the clinically approved x-ray contrast agent Iopamidol^™ (40). Water peak suppression in high-resolution NMR experiments using water molecule exchange with paramagnetic complexes has also been previously demonstrated (41), where it was shown that an increase in Δω reduced the amount of agent required to achieve the same level of water suppression compared to diamagnetic –NH and –OH exchange by a factor of ~200 (i.e., 2.5 mM versus 500 mM).

The basic theory of T_2ex is further described in detail in a recent publication (35). This publication used Eu³⁺ instead of Dy³⁺-based chelates and showed both in vitro and in vivo data. The fact that Eu³⁺ has one of the lowest paramagnetic relaxation enhancement values of the lanthanides (5) guaranteed that the reduction in the bulk water T₂ was mainly due to water molecule exchange and not due to paramagnetism. It was shown that the transverse relaxivity due to chemical exchange (r_2ex) is a function of the chemical shift (Δω) and the bound water lifetime (τ_B) as given below:

$$r_{2ex}=\left(1.8\times {10}^{-5}\right)·\frac{{\tau }_B\mathrm{\Delta }{\omega }^2}{1+{\tau }_B^2\mathrm{\Delta }{\omega }^2}$$ (1)

Equation (1), derived using a limiting case of Swift-Connick theory (37,42), was in exact agreement with results from Bloch-McConnell simulations of a two-pool system with chemical exchange (43). Equation (1) was used to create a “Swift-Connick” plot of r_2ex (s⁻¹ mM⁻¹) versus τ_B (s) for Eu³⁺ complexes at 9.4 T where the chemical shift of the exchanging water molecule (Δω) was fixed at 50 ppm (see Figure 4). It can been seen from Figure 4 that r_2ex is minimized at fast and slow water exchange rates, but reaches a well-defined maximum (0.96 s⁻¹ mM⁻¹) at a single intermediate exchange rate (τ_B = Δω⁻¹, or 9.4 μs). The value of r_2ex is highly dependent on the water molecule exchange rate as demonstrated by the experimental data points for EuDOTA (fast exchange) and EuDOTA-(gly)₄ (slow to intermediate exchange) measured at three different temperatures (22, 37, and 52 °C) (35). It can be seen that the bound water lifetime for each compound varies by an order of magnitude over the 30°C change in temperature. Furthermore, the bound water lifetimes predicted by these fits are in agreement with recent exchange rate data for similar EuDOTA-based compounds (44). These T_2ex data nicely explain why images of mouse kidneys appear dark after injection of EuDOTA-(gly)₄ (intermediate exchange) but remain bright after injection of EuDOTA (fast exchange) or EuTETA (no exchange) in images collected with conventional spin-echo or gradient-echo MR sequences at 9.4 T (35).

Figure 4.

A Swift-Connick plot showing the relation between exchange transverse relaxivity (r_2ex) and bound water lifetime (τ_B) for water molecule exchange with Eu³⁺ at 9.4 T. The markers show the total r₂ values of each Eu³⁺ compound (measured at 9.4 T) fitted to the Swift-Connick plot.

The sensitivity of T_2ex to the rate of water molecule exchange can also be demonstrated in vivo by simply changing the internal temperature of the animal. Figure 5 shows images of healthy mouse kidneys collected at two different temperatures, 37 °C (top row) and 25 °C (bottom row). These images were collected on the same mouse but on different days after an intravenous injection of 0.5 mmol/kg EuDOTA-(gly)₄. From the difference images, it can be seen that the level of T_2ex contrast in the kidneys is much lower at 25 °C than 37 °C, in agreement with the Swift-Connick plot (Figure 4) for EuDOTA-(gly)₄.

Figure 5.

(a–c) Fast spin-echo images of healthy mouse kidneys showing T_2ex contrast due to a 0.5 mmol/kg intravenous dose of EuDOTA-(gly)₄ at 37 °C; (a) pre injection; (b) 15 minutes post injection; (c) pre minus post. (d–f) Fast spin-echo images of healthy mouse kidneys showing T_2ex contrast due to a 0.5 mmol/kg intravenous dose of EuDOTA-(gly)₄ at 25 °C; (d) pre injection; (e) 15 minutes post injection; (f) pre minus post. The same mouse was used for each experiment, but on different days. These results agree with the EuDOTA-(gly)₄ data from Figure 4.

4. Overcoming T_2ex with SWIFT-CEST

As discussed previously, both the PARACEST and T_2ex effects are highly dependent upon the water molecule exchange rate. Fortunately, these two contrast mechanisms appear to reach a maximum at different exchange rates for a given lanthanide ion chelate (see Discussion section). This means that one could select an exchange rate where the PARACEST signal is maximized while the T_2ex is minimized. Also, since T_2ex contrast is highly dependent on Δω (and therefore B_o), one might choose to image PARACEST agents having intermediate-to-slow water exchange rates at lower magnetic fields where the T_2ex effects are less prominent (see Figure 6). Another option to overcome T_2ex is to use imaging sequences designed to detect short T₂ components by using very short echo times (TE). The standard spin-echo and gradient-echo pulse sequences used previously had minimum TE times of about 10 ms and 1 ms, respectively, which were not short enough to capture the MR signal from regions of reduced T₂ (e.g., the mouse kidneys) caused by the presence of the PARACEST agent. By using the sweep imaging with Fourier transform (SWIFT) pulse sequence, one can acquire MR images with essentially zero TE times (<10 μs) (45). It was recently shown that the ultra-short TE times used in SWIFT can indeed overcome bulk water T₂ shortening caused by T_2ex with PARACEST agents (46). SWIFT-CEST enables fast and straightforward imaging of PARACEST agents using simple “Off” minus “On” image techniques. Along with fast and silent acquisition of 3D data, SWIFT also has the advantage of being insensitive to both motion and flow, as well as B_o inhomogeneities due to changes in susceptibility.

Figure 6.

A Swift-Connick plot of exchange transverse relaxivity (r_2ex) versus bound water lifetime (τ_B) for water molecule exchange with Eu³⁺ at several different B_o strengths. As B_o increases, the maximum r_2ex value increases proportionally and moves towards faster exchange. The vertical line represents the water molecule exchange rate of EuDOTA-(gly)₄ at 37 °C (52 μsec).

Although SWIFT can be used to overcome T_2ex, the MT effect is still present in tissues like the kidneys. In order to overcome the MT signal, the dose of PARACEST agent used in the SWIFT-CEST experiment was increased to 1.0 mmol/kg (~10-fold higher than a “standard” clinical dose of Gd³⁺). Unfortunately, higher doses not only increase the contribution from T_2ex but also move us away from the goal of improving MR sensitivity in functional and molecular imaging at lower doses. An alternative way to improve the MR sensitivity of PARACEST agents is to use a lanthanide ion with a bound water frequency (Δω) far outside the MT frequency window (+100 to −100 ppm). An increase in Δω will also increase T_2ex proportionally. Also, unlike Eu³⁺, other lanthanide ions will have paramagnetic contributions that will further reduce the T₂ of the bulk water. Therefore, SWIFT-CEST or other ultra-short echo time sequences will likely be needed to image PARACEST agents having large Δω values. Detection of a Tb³⁺DOTA-based agent (Δω = −600 ppm) without interference from MT was recently demonstrated in vivo using SWIFT-CEST at 9.4 T (47).

5. Embracing T_2ex for T₂ Contrast

Given that typical water exchange-based PARACEST agents also serve as T_2ex agents, there may be useful in vivo applications where this imaging feature could be used to advantage. Since T_2ex is proportional to the bound water chemical shift, choosing a lanthanide ion complex with a large Δω value and working at high magnetic fields (B_o) would maximize the effect. Figure 6 illustrates Swift-Connick plots for Eu³⁺ (Δω = +50 ppm) at several different B_o strengths. It can be seen that as B_o increases, the maximum r_2ex increases proportionally with the curve maximum shifting toward faster exchange at higher fields. Note that for EuDOTA-(gly)₄ at 37 °C, which is represented by the vertical line at τ_B = 52 μs, there is very little change in r_2ex once B_o is greater than 4.7 T. Yet there is significant reduction in r_2ex at fields lower than 4.7 T. This suggests that for certain PARACEST complexes, imaging at lower B_o field strengths might be advantageous because r_2ex will be minimized. Figure 7 illustrates Swift-Connick plots for several different lanthanide ions at 9.4 T. It can be seen that Dy³⁺ (Δω = −720 ppm) gives the highest value for exchange-based transverse relaxivity (r_2ex) simply because Δω is largest for this lanthanide ion. A Dy³⁺-based PARACEST chelate with a bound water lifetime of around 500 ns would have a transverse exchange relaxivity of about 16 s⁻¹ mM⁻¹. It is important to note that data shown in Figure 7 is for r₂ due to water molecule exchange only (r_2ex) and does not include the paramagnetic contribution to the total r₂. Unlike Eu³⁺, lanthanide ions such as Dy³⁺ will also have a significant paramagnetic component added to the total r₂, which would simply shift the Swift-Connick peaks in the positive vertical direction and increase the overall r₂ for a given bound water lifetime. Although the total r₂ for a Dy³⁺ PARACEST chelate is lower than that of typical iron oxide (T₂^*) contrast agents such as superparamagnetic ion oxide (SPIO) nanoparticles (48), it is important to note that the origin of T₂ contrast for the two species are completely different. By creating relatively low molecular weight polymers of PARACEST chelates (12), the total r₂ (per molecule) could easily be enhanced by a factor of 100 or more. Using this approach, a highly sensitive T₂ contrast agent with a r₂ comparable to (or even greater than) superparamagnetic ion oxide nanoparticles might be possible while retaining the advantages of a low molecular weight molecule. This principle was recently demonstrated by synthesizing a polymerized version of the Eu³⁺DOTA-(gly)₄ chelate (12) and performing in vivo imaging in mice (49). Through a modest 20-fold degree of polymerization the effective r_2ex was increased from approximately 0.4 s⁻¹ mM⁻¹ to 8 s⁻¹ mM⁻¹ at 9.4 T. Figure 8 shows the effect of the Eu³⁺DOTA-(gly)₄ polymer in the mouse kidneys and surrounding tissue after an intramuscular injection of 70 μmol/kg of polymer (1.4 mmol/kg of Eu³⁺). Note, a 50% decrease in total signal intensity was observed in images of the kidneys. Similarly, since T_2ex is proportional to Δω, by using Tb³⁺ (Δω = −600 ppm) instead of Eu³⁺ (Δω = 50 ppm) the same level of contrast in the kidneys (50%) can be achieved at a much lower dose of lanthanide (0.18 mmol/kg versus 1.40 mmol/kg) as shown in Figure 9 (47). Subsequent polymerization of the Tb³⁺-based T_2ex agent could then reduce the required molecular dose into the low μmol/kg or even nmol/kg range. Such a platform may prove useful for imaging the movement of stem cells or for detecting agents targeted to specific biological structures in vivo.

Figure 7.

A Swift-Connick plot of exchange transverse relaxivity (r_2ex) versus bound water lifetime (τ_B) for water molecule exchange with several different Ln³⁺ ions at 9.4 T. As Δω increases, the maximum r_2ex increases proportionally and moves towards faster exchange. Reproduced with permission from John Wiley & Sons Ltd. (35).

Figure 8.

In vivo T_2ex contrast demonstrated by 2 mm thick fast spin-echo coronal slices of healthy mouse kidneys (a) before and (b) 2 hours after a 70 μmol/kg intramuscular dose of Eu³⁺DOTA-(gly)₄ polymer (1.4 mmol/kg of Eu³⁺). A 50% reduction in kidney signal can be seen in the post-injection image. (c) The pre-injection image minus the post-injection image revealing the agent uptake in the kidneys and surrounding tissue. Note that there is no uptake within the stomach.

Figure 9.

In vivo T_2ex contrast demonstrated by 2 mm thick fast spin-echo coronal slices of healthy mouse kidneys (a) before and (b) 10 minutes after a 0.18 mmol/kg intravenous dose of Tb³⁺DOTA-4AmP-4Bu monomer. A 50% reduction in kidney signal can be seen in the post-injection image. (c) The pre-injection image minus the post-injection image revealing the agent uptake in the kidneys.

6. Discussion

The wide range of PARACEST applications stem mainly from their exquisite sensitivity to proton and water molecule exchange rates. If exchange is too slow, a PARACEST signal is not observed but as the exchange rate increases into the intermediate-exchange region, a decrease in tissue water signal intensity at Δω can be detected in a CEST spectrum. As the exchange rate increases further, the CEST intensity at Δω reaches a maximum then begins to broaden and decrease while the exchange peak shifts towards the bulk water frequency (0 ppm) (17). CEST then disappears completely when the exchange rate becomes too fast (k_ex > Δω) and no differentiation between the bound and bulk protons can be made. These resonance-like characteristics of the “CEST peak” in the CEST spectrum can be observed by simply modulating the exchange rate using temperature (44).

In order to improve the MR sensitivity of these agents, it is important to identify which slow-to-intermediate exchange rate gives the maximum PARACEST effect. NMR Bloch theory predicts (43) that CEST will be maximal for the experimental condition:

$${\text{k}}_{\text{ex}}=2\pi {\text{B}}_1$$ (2)

so targeting an appropriate B₁ for a particular in vivo application allows one to predict the optimal exchange rate that one should target for the agent. However, as shown in Figure 7, T_2ex also reaches a maximum value at a predictable exchange rate. Fortunately, it appears that the optimal exchange rate for CEST (assuming some appropriate B₁ of 100–200 Hz) and T_2ex differ to some extent. For water molecule exchange with Eu³⁺ at 9.4 T, the PARACEST signal would be maximal at an exchange rate of ~10 kHz (τ_b = 1×10⁻⁴ s) assuming a B₁ of 10 μT (17), while the maximal contribution from T_2ex (independent of B₁) would occur at an exchange rate at ~100 kHz (τ_b = 1×10⁻⁵ s), an order of magnitude faster (35). Therefore it may become important, especially at higher B_o fields, to pick an appropriate chemical system that has exchange rate that maximizes the PARACEST signal and minimizes T_2ex. This consideration along with the use of ultra-short echo time sequences like SWIFT or UTE (46) will help insure detection of PARACEST signals at as low a concentration as possible in vivo.

Even at the ideal proton or water molecule exchange rate, the sensitivity of PARACEST contrast agents is quite low compared to nuclear imaging modalities such as positron emission tomography (PET) and single-photon emission computed tomography (SPECT). Nuclear medical imaging methods like PET and SPECT detect and image the radiation emitted from the nucleus of a single unstable isotope (e.g., the 140 keV gamma ray from ^99mTc). This sensitivity, combined with the low background radioactivity in the human body, means that doses as low as nM to pM can be used for imaging. Conversely, MR depends on an ensemble of thousands of hydrogen nuclei to generate signal adequate for imaging so the typical dose of an MR contrast agent is in the mM range. In order to truly extend the functional and molecular imaging capabilities of MR, the sensitivity of PARACEST contrast agents must be increased substantially. One can of course improve the sensitivity of PARACEST agents by simply increasing the number of water molecule exchange sites per molecular unit by forming polymers of the agent (10² increase) (12), coating the surface of perfluorocarbon-filled nanoparticles with agent (10⁵ increase) (50), and even liposomal encapsulation of the agent (10¹⁰ increase) (51,52). While a polymerized form of a PARACEST agent yielded an enhancement of only about 10² so far, this might be improved upon substantially while retaining the advantages of relatively low molecular weight species for improved biological uptake.

An alternative way to improve the sensitivity of PARACEST agents for molecular imaging is to use a paramagnetic chelate that has a bound water molecule frequency (Δω) outside of the tissue MT window (100 to −100 ppm). This would allow activation of the PARACEST agent without co-activation of the MT signal and potentially provide more CEST contrast at lower B₁ power, thereby also reducing SAR. It was recently shown that a Tb³⁺DOTA-based PARACEST agent (Δω = −600 ppm) can be imaged in vivo using the ultra-short echo times of SWIFT to overcome any T_2ex and paramagnetic reduction of the bulk water T₂ (47).

7. Conclusions

PARACEST contrast agents show great promise for extending the functional and molecular imaging capabilities of MR. The sensitivity of the PARACEST signal to proton or water molecule exchange rates make these agents well-suited to image and measure relevant biomarkers such as temperature, pH, and specific metabolites. Translating the encouraging early in vitro results to in vivo applications and eventually the clinic promises to be challenging, but will have high rewards for the study and diagnoses of human disease such as cancer and diabetes.

Abbreviations used

CEST

chemical exchange saturation transfer

DIACEST

diamagnetic chemical exchange saturation transfer

MR

magnetic resonance

MRI

magnetic resonance imaging

MT

magnetization transfer

NMR

nuclear magnetic resonance

PARACEST

paramagnetic chemical exchange saturation transfer

PET

positron emission tomography

RF

radiofrequency

SAR

specific absorption rate

SPECT

single photon emission computed tomography

SPIO

superparamagnetic iron oxide nanoparticles

SWIFT

sweep imaging with Fourier transform

TE

echo time

TR

repetition time

UTE

ultra-short echo time