Functionalized ¹²⁹Xe contrast agents for magnetic resonance imaging

Abstract

The concept of ‘xenon biosensor’ for magnetic resonance imaging (MRI) was first proposed by a Berkeley team in 2001, with evidence that hyperpolarized ¹²⁹Xe bound to a biotin-labeled cryptophane can detect streptavidin at much lower concentrations (nM-μM) than is typical for contrast-enhanced MRI experiments. ¹²⁹Xe biosensors have undergone many recent developments to address challenges in molecular imaging. For example, cryptophanes have been synthesized that exhibit 10-fold higher xenon affinity with distinct ¹²⁹Xe magnetic resonance spectra. Also relevant are dendrimeric cryptophane assemblies and inorganic zeolites that localize many ¹²⁹Xe atoms to rare targets. Finally, this article considers biosensors that produce measurable changes in ¹²⁹Xe chemical shift based upon the activity of oligonucleotides, proteins, or enzymes, and includes the first cell studies.

Introduction

Hyperpolarized (hp) ¹²⁹Xe biosensors were introduced in 2001 as novel contrast agents for magnetic resonance imaging (MRI) and spectroscopy (MRS) [1]. Xenon biosensors put a new (nuclear) spin on an old story: In 1973, Lauterbur [2] and Mansfield [3] obtained the first MR images from free induction decay (FID) data acquired in the presence of magnetic field gradients. Today, noninvasive proton (¹H) MRI is one of the most widely used and versatile techniques for scanning deep tissue in the diagnosis of human disease. However, intrinsic ¹H MRI signals typically provide poor detection sensitivity. Consequently, contrast agents must be introduced into patients for image enhancement, as occurs in more than half of the roughly 30 million MRI procedures performed annually in the U.S.

Contrast media based on gadolinium or iron-oxide particles have been reviewed extensively [4–5]. Efforts to enhance contrast further have produced ‘smart’ Gd³⁺ [6], iron-oxide [7], and also ¹⁹F-based agents [8] that provide signal in response to a biological trigger. However, fundamental limitations in sensitivity have motivated the investigation of alternate nuclei such as ¹²⁹Xe, ¹³C, ⁸³Kr, and ³He [9–12], which can be hyperpolarized to produce unpaired spins that result in much larger NMR signal.

Here, we focus on molecular imaging with ¹²⁹Xe gas, which is spin-½ and can achieve a hyperpolarized state with spin-lattice relaxation time (T₁) of roughly 70 min. 129Xe hyperpolarization yields ~10⁵ signal enhancement relative to the normal Boltzmann distribution of nuclear spins. The process of generating hp ¹²⁹Xe through spin-exchange optical pumping (Figure 1A) [13] involves the transfer of angular momentum from circularly polarized light to an alkali metal (here we consider Rb). Polarized Rb subsequently interacts with the nuclear spin of xenon through dipolar coupling. Recently, large improvements in ¹²⁹Xe hyperpolarization technology have achieved > 50% hp ¹²⁹Xe at flow rates of 1–2 L per min, which enables human clinical applications [14].

Figure 1.

A) Process of optical pumping to produce hyperpolarized ¹²⁹Xe. B) Schematic representation of hp ¹²⁹Xe NMR spectrum showing (from left) resonances of free xenon gas in aqueous solution, Xe-encapsulated biosensor bound to corresponding bioreceptor, and Xe encapsulated in free biosensor. Legend shows components of biosensor: molecular cage, linker, and recognition moiety.

Hyperpolarized ¹²⁹Xe offers potential advantages over conventional MRI contrast agents, namely: roughly 10⁴–fold more NMR signal per nucleus than a corresponding proton, which takes into account the greater gyromagnetic ratio and isotopic abundance of ¹H relative to ¹²⁹Xe; lower toxicity than Gd³⁺ and most other paramagnetic metal ions, as xenon has low affinity for proteins and is relatively ‘inert’ in the body when introduced below the anesthetic threshold, which is 70% by volume of inhaled gas mixtures; exquisite NMR chemical shift sensitivity, with a greater than 200 ppm spectral window in aqueous solution, which facilitates the detection of multiple species simultaneously (i.e., multiplexing); and, significant polarizability, which contributes to xenon’s millimolar solubility in water [15] and promotes binding to molecular hosts via London dispersion forces. ¹²⁹Xe biosensors (Figure 1B) that exploit xenon’s unique physical properties create exciting opportunities for molecular imaging. This article focuses on very recent advances in xenon biosensor design and application.

Xenon Biosensing: The Host Molecule

We define ‘xenon biosensor’ simply as a xenon-encapsulating agent attached to one or more bioactive molecules. Due to its filled valence electron shell, xenon resists forming covalent bonds; nonetheless, it shows modest affinity for hydrophobic cavities found in many organic molecules [16–19]. For example, the association constant, K_A, is 200 M⁻¹ for hemicarcerands [20] and 20 M⁻¹ for α-cyclodextrin [17]. A calix[4]arene derivative exhibited a K_A of only ~14 M⁻¹ at 298 K in water [21] whereas for water-soluble cucurbit[6]uril, a xenon binding constant of 3000 M⁻¹ was measured [18]. In fact, the non-covalent, dynamic association of ¹²⁹Xe with these host molecules creates opportunities for xenon biosensing applications, as xenon can be hyperpolarized separately as the gas and subsequently introduced to the host molecule, at multiple time points for longitudinal studies.

Cryptophane-A, first reported in 1981 [22], is the most studied Xe-binding cage and is comprised of two cyclotriveratrylene caps joined by three ethyl linkers, which give an internal van der Waals volume of approximately 95 Ǻ³ (Figure 2). As a result of its connectivity, cryptophane-A has D3 symmetry and is chiral. Cryptophane-A is also commonly referred to as cryptophane-2,2,2, based on the three 2-carbon spacers. Cryptophane-A binds Xe reversibly with K_A ~ 3,900 M⁻¹ in C₂D₂Cl₄ at 278 K [23]. In a series of related cryptophanes incorporating 3-carbon spacers, xenon affinity decreases with increasing size of the cavity [24].

Figure 2.

Alternate mono- and tri-functionalization approaches for attaching targeting moieties, dye molecules, and water-solubilizing groups to cryptophane-A.

Cryptophane-1,1,1 was recently synthesized [25], and exhibits the highest Xe association constant measured in organic solvent, K_A = 10,000 M⁻¹ at 293 K. This host has a calculated internal volume of 81 ų, making the xenon volume (42 ų) to host ratio equal to 0.52, which is very close to Mecozzi and Rebek’s prescribed 0.55 ratio for nonpolar host-guest complexes mediated purely by London forces [26]. Cryptophane-1,1,1 shows very slow decomplexation kinetics, which produces a narrow ¹²⁹Xe NMR resonance useful for direct detection, but also hinders the Xe exchange needed for indirect detection [27]. As this example shows, both thermodynamics and kinetics impact the design of host molecules for xenon biosensing applications.

Several water-soluble cryptophanes have been synthesized to date, many with exceptional xenon-binding characteristics [28–30]. Huber’s water-soluble cryptophanes [28] replaced all six methoxy groups with OCH₂COOH. The Xe affinity of hexa-acid cryptophane-A was measured to be 6800 M⁻¹ at 293 K. Conformational changes were observed in the hexa-acid cryptophanes, including a non-xenon-binding crown-saddle isomer. Our lab subsequently developed synthetic protocols for attaching three carboxylates on just one of the cyclotriveratrylenes [29]. This creates a dipole moment along the axis of the cage that may interact favorably with the xenon electron cloud.

In order to perform more accurate and rapid measurements of xenon binding, our lab developed fluorescence quenching and isothermal titration calorimetry (ITC) assays [30]. Reacting tri-propargyl cryptophane with three azido-propionates gave a water-soluble cryptophane with association constant of 1.7 × 10⁴ M⁻¹ at 293 K [29]. ITC indicates that entropy is an important factor in aqueous xenon binding, as −TΔS is of similar magnitude to ΔH. As a result, xenon affinity increases with temperature: K_A = 3.0 × 10⁴ M⁻¹ at 310 K. Importantly, xenon binding studies with the same compound in human plasma showed similar affinity (K_A = 2.2 × 10⁴ M⁻¹ at 310 K), which supports use of cryptophanes for cellular imaging. In order to improve xenon binding further, a cryptophane was synthesized with the carboxylates placed closer to the cavity [30]. Presently, a tri-acetate cryptophane-A shows the highest affinity for xenon of any host molecule in water, K_A = 3.3 × 10⁴ M⁻¹ at 293 K.

Importantly for multiplexing studies, the known cryptophanes produce very different ¹²⁹Xe NMR chemical shifts [25,28,31]. Among Huber’s hexa-acid cryptophanes in water, the encapsulated ¹²⁹Xe shows a single resonance that shifts ~30 ppm upfield with increasing size of the cavity: 222-cage, 64 ppm; 223-cage, 52 ppm; 233-cage, 42 ppm; 333-cage, 35 ppm [28]. A similar trend is observed for the corresponding organic-soluble cryptophanes in C₂D₂Cl₄ [13]. This range of chemical shifts will enable multiplexing experiments that employ different cryptophanes to target various biomarkers. Furthermore, these data suggest that as the cavity becomes smaller, the electron cloud of the xenon feels a greater perturbation from the cage, leading to a larger chemical shift. Interestingly, the Xe@cryptophane-1,1,1 peak is at 31.1 ppm, fully 36 ppm upfield from Xe@cryptophane-2,2,2 in C₂D₂Cl₄. Although this appears to violate the volume-chemical shift trend, cryptophane-1,1,1 lacks the six methoxy substituents on the phenyl rings. This suggests that incorporating electron-donating and withdrawing substituents on the cryptophane should allow efficient stereoelectronic modulation of ¹²⁹Xe NMR chemical shift for molecular imaging applications.

Xenon Biosensing: The Targeting Molecule

Attached to the host is the targeting molecule, which may be a recognition moiety with high affinity and specificity for a single analyte, e.g., biotin-streptavidin [1,32], peptide-antigen [33], DNA-DNA hybrid [34], or affinity tag for a specific cancer biomarker [35–36]. Attachment of a substrate molecule has the potential to allow even more sensitive detection of an enzyme via the conversion of multiple substrates to products [37]. However, activity-based enzyme sensors must ensure that ¹²⁹Xe is sufficiently near to the substrate to detect product formation. By the same token, the host molecule should not interfere with enzyme activity.

A ¹²⁹Xe-biosensor complex freely diffusing in solution typically results in a single NMR resonance (Figure 1B) that does not distinguish between cryptophane enantiomers. Binding to a biological receptor typically produces one or more additional ¹²⁹Xe peaks. To date, many xenon biosensors have relied on a monofunctionalized cryptophane approach [32–33,35,37]. Newer approaches for tri-substituting cryptophane-A facilitate the introduction of additional targeting moieties, water solubilizing groups or dye labels for fluorescence microscopy (Figure 2) [30,36]. Such flexibility is useful for tuning the biological and spectroscopic properties of the xenon biosensor.

Hyperpolarized ¹²⁹Xe Biodetection: Overview

It is now possible to collect MR images of inhaled hp ¹²⁹Xe in the lungs, brain, and other organs of the body [38–41]. However, ¹²⁹Xe MRI biosensors will require even greater sensitivity to achieve in vivo molecular imaging. Two different biodetection strategies are possible: 1) Target localization, with maximal hp ¹²⁹Xe localized to the target and minimal background signal elsewhere in the specimen; 2) Multiplexing, whereby multiple xenon biosensors identify several biomarkers. Colocalization of different ¹²⁹Xe MR spectroscopic signals should help to identify unusual cellular biochemistry, as might be associated with a disease state.

Xenon delivery strategies have sought to maximize the signal-to-noise ratio of the hyperpolarized ¹²⁹Xe MR signal, by minimizing the decay of polarization during transport or by dilution [38,42]. The T₁ of ¹²⁹Xe varies widely in solution, depending on its molecular environment [1,43–46]: T₁ = 4 s in deoxygenated blood, T₁ = 13 s in oxygenated blood, T₁ = 20–40 s in cryptophanes, and T₁ ≈ 1000 s in deuterated saline solution [43,47]. Direct inhalation is the most clinically feasible method of delivering hyperpolarized ¹²⁹Xe in humans, and some organs, such as the lungs and brain are accessible in this way [48]. In other cases, direct intravenous [42] or arterial [38] injection of hyperpolarized ¹²⁹Xe is feasible, particularly in rodents. Thus, it is practical to deliver hp ¹²⁹Xe for sensitive in vivo MRI detection following introduction of the biosensor. Experiments with radioisotope ¹³³Xe in pigs confirmed that xenon is cleared from the lungs and other organs, with retention only in fatty tissues after two hours [49]. Thus, introduction of fresh hp ¹²⁹Xe will allow longitudinal studies, provided the biosensor is retained.

Hyperpolarized ¹²⁹Xe Biodetection: Synthetic Strategies

Water-soluble dendrimers have been shown to encapsulate multiple cryptophanes, while presenting recognition moieties that deliver the cargo to a specific target [50]. The published approach exploited electrostatic and hydrophobic interactions to incorporate the cryptophanes, which avoided synthetic steps and the creation of cryptophane diastereomers that might produce additional ¹²⁹Xe NMR resonances. This invites exploration of polymeric systems that carry a larger payload of cryptophane and improve bioavailability. Relevant examples are micelles and liposomes that can enhance MRI contrast by encapsulating large numbers of paramagnetic species [51].

Another interesting approach for localizing xenon is the use of nanoporous materials. Lerouge prepared silica-based, 65-nm zeolites that contained many pores capable of binding hp ¹²⁹Xe with a fairly long T₁ [52]. Biocompatibility and colloidal stability of these nanoparticles were achieved by functionalization with PEG chains and a peptide targeting biological receptors; localization in a mouse was studied by ¹¹¹In scintigraphy. It may be possible to create zeolite structures with enhanced xenon-binding and spectroscopic characteristics, as needed for hp ¹²⁹Xe MRI studies.

Hyperpolarized ¹²⁹Xe Biodetection: NMR Strategies

Sensitive NMR methods for detecting hp ¹²⁹Xe have been developed, as elaborated in a recent review [53]. Most promisingly, a hp ¹²⁹Xe chemical exchange saturation transfer (hyper-CEST) strategy allows the selective depolarization of biosensor-bound hp ¹²⁹Xe using pulsed excitation [27]. This leads to the accumulated depolarization of bulk hp ¹²⁹Xe through rapid exchange of xenon atoms between “cryptophane-bound” and “free” states. Monitoring depletion of the “free” hp ¹²⁹Xe NMR signal provides roughly 10³ signal enhancement, based on the large pool of aqueous xenon (~ 3 mM at 1 atm and 310 K) [15] compared to the typically micromolar concentration of ¹²⁹Xe@biosensor that can be delivered to cells and other biological targets. However, some challenges remain. For example, it will be difficult to achieve a constant polarization level in vivo. Moreover, a decrease of spatial resolution is likely, due to xenon diffusion during the long saturation time.

Biological Applications: Achieving Large ¹²⁹Xe NMR Chemical Shifts

The initial Berkeley biotin biosensor gave a hp ¹²⁹Xe NMR peak at 70 ppm. Addition of 40 nanomoles of streptavidin produced a new hp ¹²⁹Xe NMR peak ~2.5 ppm downfield [1]. Shorter tethers that enforce the streptavidin-cryptophane interaction shifted the ¹²⁹Xe NMR spectrum by as much as 4 ppm [54–56]; however, the ¹²⁹Xe NMR resonances also became considerably broadened (> 1 ppm), which limited biodetection. The origin of the ¹²⁹Xe NMR chemical shift, as well as the appearance of multiple “bound” peaks in some systems, have been investigated computationally by Jameson [55,57]. It is now evident that cryptophane-biotin diastereomers can produce overlapping ¹²⁹Xe NMR spectra. An additional complication is that streptavidin (a dimer of dimers) binds multiple biotins, which may promote Xe-cryptophane:Xe-cryptophane interactions.

We subsequently designed an enzyme-responsive ¹²⁹Xe biosensor for the detection of a cancer biomarker, matrix metalloproteinase-7 [37]. The probe was designed such that proteolysis of a peptide containing three positively charged residues caused a change in the electrostatic environment of ¹²⁹Xe, which could possibly affect ¹²⁹Xe NMR chemical shift. ¹²⁹Xe NMR peaks of 61.9 and 62.5 ppm for the intact peptide diastereomers gave upfield shifts of 0.5 and 0.3 ppm upon proteolysis. Although these Δδ values are too small for in vivo detection, this work illustrated the potential for studying enzyme activity.

Roy et al. subsequently designed a cryptophane biosensor for in vitro DNA detection [34]. In order to avoid formation of diastereomers, a chiral (−)-cryptophane-A was synthesized and covalently attached to a 20-mer DNA probe. The NMR spectrum of hp ¹²⁹Xe with this biosensor alone or in the presence of noncomplementary DNA gave a peak at 68.5 ppm. Addition of the complementary DNA strand produced a new resonance at 67.0 ppm that clearly identified the DNA target in solution. Although unique DNA sequences are rare targets in the cell nucleus, this illustrates the utility of enantiopure cryptophanes for biosensing applications. More recently, Schlundt introduced a peptide-labeled ¹²⁹Xe biosensor that binds to a major histocompatibility complex (MHC) class II protein [33]. In this case, a 1-ppm downfield shift was observed upon protein binding, and larger chemical shift changes may be possible using a shorter linker. This work demonstrated sensitive biodetection using hyper-CEST NMR techniques.

Finally, our lab has developed ¹²⁹Xe biosensors for a second cancer biomarker, carbonic anhydrase (CA), which serves as a useful model system for elucidating the origin of ¹²⁹Xe NMR chemical shifts [36]. Tri-propargyl cryptophane was substituted with linkers varying between 6 and 8 bonds to CA-specific p-benzenesulfonamide ligand to yield non-diastereomeric biosensors with a single ¹²⁹Xe NMR resonance. These Xe biosensors were designed to bind CA isozymes I and II by considering the depth of the active-site channel, the size of the cryptophane, and the propensity of related compounds to bind CA. X-ray crystallographic studies confirmed binding of benzenesulfonamide to Zn²⁺ at the CAII active site (Figure 3A), with the 8-bond-linked cryptophane positioned at the mouth of the protein channel [58].

Figure 3.

A) X-ray crystal structure of CAII bound to benzenesulfonamide-8-bond-linker cryptophane, only one enantiomer shown. Data reveal sulfonamidate coordination to Zn²⁺ (gray atom), and Xe (green atom) bound inside cryptophane. B) Laser-polarized ¹²⁹Xe NMR spectra: (i) 7-bond-linker biosensor free in solution; (ii) biosensor bound to CAI and (iii) CAII.

The biosensors complexed with CA I and II yielded “bound” hyperpolarized ¹²⁹Xe NMR resonances of narrow linewidth at approximately 64 ppm that were shifted between 3.0 and 7.5 ppm downfield upon CA binding. These chemical shift differences clearly distinguished between CA I and II in solution. One of the most intriguing findings from this work came with the 7-bond-linker biosensor, which showed one “bound” resonance for CAI and distinctively two “bound” resonances for CAII (Figure 3B).

Biological Applications: Xenon Biosensors in Cells

Our lab has synthesized Xe biosensors targeting cancer biomarkers and performed cell studies [35]. Mono-propargyl cryptophane reacted with azido-peptides in a copper(I)-catalyzed [3+2] cycloaddition. Delivery of cryptophane into human cancer and normal cell lines was achieved with cationic cell penetrating peptides. A tetraRGD peptide targeting α_vβ₃ integrin increased specificity of cryptophane cell uptake (Figure 4). The peptido-cryptophanes were determined to be relatively nontoxic at the micromolar concentrations needed for hp ¹²⁹Xe NMR experiments.

Figure 4.

Cryptophane-tetraRGD biosensor targeting α_vβ₃ integrin. Inset: Uptake of Cy3-labeled biosensor in (A) NCI-H1975 and (B) HFL-1 cells.

Conclusion: A Bright and Undetermined Future for ¹²⁹Xe MRI Biosensing

¹²⁹Xe MRI biosensors have been designed to bind xenon with high affinity and target oligonucleotides or proteins [33–34,36–37]. Such compounds have the potential to detect various diseases, by localizing hp ¹²⁹Xe to a tissue and/or by providing multiplexed detection of different biomarkers. Improved capabilities for hp ¹²⁹Xe detection and chemical shift resolution are also coming from new ¹²⁹Xe hyperpolarization technology [13], organic and inorganic assemblies for localizing large numbers of ¹²⁹Xe atoms [50,52], and hyper-CEST imaging strategies [27]. Uptake of xenon biosensors in cells and mice has been verified [35,52]. In addition, proof-of-concept MRI studies have demonstrated the ability to resolve spatially cryptophanes that are targeted to labeled beads, different compartments or solvents [27,59–61]. Intravenously injected hp ¹²⁹Xe-saturated saline was shown to be a useful probe for pulmonary perfusion or gas exchange evaluation in the lung [62], further validating ¹²⁹Xe MRI as a sensitive method for imaging lung diseases [63]. With these rapid technological advances, it should now be possible to detect specific bio-receptors in cells and animals via contrast-enhanced hp ¹²⁹Xe MRI and MRS. Experimental validation appears to be close at hand, which will guide the continued development of xenon biosensors for molecular imaging and help to identify the best in vivo applications for this technology.