Tumor Microenvironment Biosensors for Hyperpolarized Carbon-13 Magnetic Resonance Spectroscopy

Abstract

Hyperpolarization (HP) of a carbon-13 molecule via dynamic nuclear polarization (DNP) involves polarization at low temperature, followed by a dissolution procedure producing a solution with highly polarized spins at room temperature. This dissolution DNP method significantly increases the signal-to-noise ratio (SNR) of nuclear magnetic resonance (NMR) over 10,000-fold and facilitates the use of magnetic resonance spectroscopy (MRS) to image not only metabolism, but also the extracellular microenvironment. The extracellular tumor microenvironment (TME) closely interacts with tumor cells and stimulates their growth and metastasis. Thus, the ability to detect pathological changes in the TME is pivotal for the detection and study of cancers. This Review highlights the potential use of MRS to study features of the tumoral TME – elevated export of lactate, reduced interstitial pH, imbalanced redox equilibrium, and altered metal homeostasis. The promising outcomes of both in vitro and in vivo assays and suggest that DNP-MRS may be a useful technique to study aspects of the TME. With continued improvements, this tool has the potential to study the TME and provide guidance for accurate patient stratification and precise personal therapy.

Graphical Abstract

Introduction

The tumor microenvironment (TME) refers to the altered and pathological surroundings of growing tumor cells. It is mainly composed of tumor cells, tumor stroma, immune cells, and the extracellular matrix (ECM) [1]. The TME closely interacts with tumor cells and promotes tumor progression and therapeutic resistance [2]. In comparison to the normal tissue microenvironment, the TME exhibits many differences, including abnormal tissue architecture, elevated oxidative stress, extracellular acidification, metabolic alterations, and abnormal ECM dynamics [3]. A variety of strategies have been employed to study the tumor microenvironment using non-invasive imaging methods, including NMR spectroscopy of protons and other nuclei, chemical exchange saturation transfer (CEST), and nuclear imaging techniques such as positron emission tomography (PET) and single photon emission computed tomography (SPECT) [4, 5]. The initial successful translation of the dynamic nuclear polarization – magnetic resonance spectroscopy (DNP-MRS) technique to hyperpolarize (HP) [1-¹³C]pyruvate for clinical imaging of patients with prostate cancer (PCa) in 2013 [6] led to a new era of the use of this method in patient studies, with promising outcomes summarized in prior review articles [7]. Expanding the scope of DNP-MRS to characterize the TME may provide complementary information that could enhance the ability to diagnose and treat cancer effectively. In this review, we focus on the development and implementation of ¹³C-labeled TME biosensors.

Dynamic nuclear polarization of ¹³C labeled cellular substrates is an emerging clinical technique, which can significantly increase ¹³C NMR sensitivity over 10,000-fold, allowing for MRS detection in vitro and in vivo [8]. However, DNP-MRS has strict criteria for successful implementation, including the requirement for a high concentration preparation, the ability to form a glass upon freezing, low probe toxicity, long spin-lattice relaxation times (T₁) for the interrogated ¹³C nucleus, and spectrally resolvable chemical shift differences upon reaction [9]. The study of the TME using DNP-MRS is a relatively new concept. Therefore, the purpose of this review is to (i) summarize the basics of the DNP-MRS imaging technique; (ii) elucidate the biochemical processes responsible for four important possible TME imaging biomarkers: elevated cellular export of lactate, reduced interstitial pH, imbalanced cellular redox equilibrium, and altered metal homeostasis; and (iii) highlight in vitro and in vivo assays using DNP-MRS to characterize the status of TME and their correlation with tumor behavior.

1. Nuclear Magnetic Resonance Spectroscopic Imaging using Hyperpolarized Carbon-13 Biosensors via Dynamic Nuclear Polarization

Non-proton MRS can be challenging for imaging owing to low sensitivity. Spin polarization can be employed to increase the signal-to-noise ratio (SNR). DNP and parahydrogen-induced polarization (PHIP) are commonly used methods for spin polarization. DNP is a powerful technique that can substantially increase the nuclear spin polarization of ¹³C in cellular substrate molecules and improve its ¹³C NMR signal sensitivity over 10,000 fold [8]. The DNP procedure involves cooling a mixture of ¹³C biosensor compound, a free radical reagent, and in some cases a solvent or glass former to ~1 K. The sample is then irradiated with microwave frequency in a strong magnetic field to transfer the polarization from electron spins on the radicals to the ¹³C nuclei of the biosensor molecules. Finally, it is warmed up rapidly (dissolution) for subsequent experimental steps (Figure 1a) [10, 11]. For example, we polarized [1-¹³C]1,2-glycerol carbonate (GLC) to 29.5% at 5 T (back-calculated to time of dissolution), corresponding to ~29,400 fold enhancement of SNR after dissolution (Figure 1b). After dissolution, the polarization rapidly returns to its thermodynamic equilibrium, with exponential decay of hyperpolarized signal with time constant T₁ (Figure 1c). PHIP is an alternative modality to DNP, involving a catalyzed hydrogenation reaction by adding para-enriched hydrogen (para-H₂) to an unsaturated substrate molecule. However, the application of PHIP can be limited due to the availability of unsaturated precursors for the target substrate [12, 13]. An alternative approach called signal amplification by reversible exchange (SABRE) relies on transferring polarization from para-H₂ through the scalar coupling framework of a metal complex to the substrate, which can keep the substrate chemically unchanged [14]. In order to be eligible for animal models and human clinical imaging studies, the HP biosensor should preserve the polarization long enough for dissolution, intravenous injection, and imaging data collection. The basic criteria for using DNP-MRS are as follows: the HP ¹³C biosensor molecules should (i) be nontoxic, with safety of injecting sufficiently high concentrations of 50–250 mM for detection; (ii) have rapid cell transport or enzyme-catalyzed conversion; (iii) have a sufficiently long spin-lattice relaxation (T₁); (iv) polarize well enough to generate sufficient signal enhancement (typically greater than 5% polarization); and (v) demonstrate an observable chemical shift separation (≥5 ppm) between the agent and resulting HP metabolites [9]. In this Review, we focus on the application of dissolution DNP-MRS for interrogating the status of the TME.

Figure 1.

Basics of dynamic nuclear polarization. (a) Mechanism of increasing the signal to noise ratio of ¹³C MNR via DNP: (i) preparing ¹³C enriched biosensor molecules with stable radical species, (ii) cooling the mixture of biosensor molecules with radicals to ~1 K in a magnetic field of 3–5 T in order to generate high radical polarization, (iii) transferring the electron spin from radicals to nuclear spin on ¹³C of biosensor molecules using microwave irradiation. (b) ¹³C NMR spectra of ¹³C GLC acquired at (up) HP state (29.5% polarization) with pulse angle of 15° and repetition time of 1 s based on a T₁ of 60 s with full ¹H decoupling, and (low) thermal equilibrium state. The signal is averaged of 600 transients. (c) HP ¹³C T₁ decay spectra of (left) DEMA T₁ = 96.2 s, (middle) bicarbonate T₁ = 39.2 s, (right) DHA T₁ = 20.3 s. All measurements were taken at 11.7 T field strength, temporal spacing = 3s, with 500 timepoints. The large differences in SNR at later time points highlight the critical need for longer T₁ probes.

2. Hyperpolarized ¹³C MRS Biosensors for Imaging the Tumor Microenvironment

In this Review, we highlight four TME conditions, including elevated export of lactate, reduced interstitial pH, increased oxidative stress, and altered metal homeostasis. We describe the potential mechanism of using corresponding HP ¹³C biosensors to target their biomarkers for the detection and of the tumor. The HP ¹³C biosensor molecules and their fundamental mechanisms of action are summarized in Figure 2. DNP-MRS also boasts the capability of simultaneously studying the distorted tumor vasculature present within the TME using HP [¹³C]urea or other probes, which have been used for measuring regional blood flow in preclinical cancer models [15].

Figure 2.

¹³C-biosensors and their proposed related reaction mechanism for imaging TME. The position of the ¹³C label is indicated for each molecule with a circle. Published T₁ measurements are indicated.

2.1. Imaging HP ¹³C-Lactate in the Tumor Microenvironment

Unconstrained tumor proliferation and invasion and therapeutic resistance usually correlate with alterations in cellular metabolism and extracellular conditions, indicative of the Warburg effect – an upregulation of aerobic glycolysis and lactic acid fermentation through the lactate dehydrogenase (LDH) catalyzed reaction [16]. Figure 3 shows the metabolic pathway of pyruvate and the function of the monocarboxylate transporters (MCT). Intravenously injected pyruvate rapidly metabolizes into lactate, alanine, and CO₂, with the metabolic rates and metabolite ratios depending on the function of the tissues. Owing to the increased uptake of pyruvate via MCT1 and MCT4 and increased activity of LDH in the tumor cells, the increased conversion of pyruvate to lactate (or a high ratio of [lactate]/[pyruvate]) is observed in tumors [17–19]. Thus, HP [1-¹³C]pyruvate provides a method of estimating the flux through this pathway and characterizing the metabolic status of a tumor. This method has been successfully translated to human clinical application to noninvasively characterize alterations in tumor metabolism for patients [6].

Figure 3.

Biochemical scheme illustrating the metabolic pathway of converting HP [1-¹³C] pyruvate to HP [1-¹³C] lactate and HP [1-¹³C] alanine, and increased HP [1-¹³C] lactate export into the TME via the upregulated MCT4.

An important link between heightened glycolytic flux and the TME is the increased cellular export of lactate, a process that can be investigated using DNP MRS. Increased concentrations of exported lactate in the TME conditions to support the tumor, for example, by suppressing immune effector cells [20] and influencing nearby stromal and vascular cells [21]. K. R. Keshari et al investigated the HP [1-¹³C]pyruvate-to-HP [1-¹³C]lactate flux in real-time using DNP-MRS combined with a bio-reactor platform of living renal cell carcinoma (RCC) cells. They found RCC cells have much a higher flux rate than the normal renal proximal tubule cells and the metastatic RCC cells have more rapid export of lactate to the extracellular space. These differences are likely mediated by the differential expression of MCT4 [22]. Subsequently, K. R. Keshari et al injected HP [1-¹³C]pyruvate in a living human prostate tissue slice culture platform. They observed a significantly increased [1-¹³C]pyruvate to [1-¹³C]lactate flux in malignant in comparison to the benign prostate tissues and provided mechanistic evidence for HP[1-¹³C]lactate as a prostate cancer biomarker [17]. R. Sriram et al developed a DNP-MRS compatible 3D cell/tissue culture bio-reactor to dynamically interrogate HP [1-¹³C]lactate production and efflux in human RCC cells, aiming at testing the correlation of this process with cancer aggressiveness, metastasis, and response to therapy [23]. Using this DNP-MRS bio-reactor platform, B. L. Koelsch et al were able to separate extra- and intracellular weighted HP metabolites of RCC cells and assess the ability of membrane transport in real-time [24]. These techniques have been extended to in vivo model systems to estimate lactate export in tumors and to provide a link to the TME. B. Feuerecker et al co-polarized [1-¹³C]pyruvate and [1,4-¹³C₂]fumarate, and measured the apparent diffusion coefficient (ADC) of the produced lactate using HP diffusion-weighted MRS. They linked the ADC ratio of lactate to pyruvate to the distribution of lactate in the intra- and extracellular compartments to report the transport, while measuring fumarate-to-malate conversion to detect necrosis and rule out its interference with lactate ADC [25]. More recently, R. Sriram et al used dynamic diffusion-weighted HP ¹³C pyruvate MRI to interrogate tumor lactate production and compartmentalization in a murine orthotopic model of human RCCs [26]. Furthermore, X. Zhu et al used this method based on a slice-selective double spin echo sequence to monitor the cellular lactate transport on the transgenic adenocarcinoma of the prostate (TRAMP) model [27]. Taken together, these prior studies suggest feasibility for imaging lactate in the TME using hyperpolarized ¹³C.

2.2. Imaging decreased interstitial pH (pH_e)

Solid tumors have heterogeneous perfusion, high metabolic activity, rapid cell proliferation, and typically develop an acidic interstitial microenvironment with pH 6.5 – 7.2, in contrast to pH 7.4 in normal and healthy tissues [28]. Extracellular acidification is commonly associated with local invasion and metastasis in a variety of cancers, including melanoma, breast, colon, renal, and PCa [29, 30]. Mechanisms of extracellular pH (pH_e) acidification include monocarboxylate transporters (MCTs) [31], Na⁺/H⁺ exchanger 1 [32], vacuolar H⁺-ATPase [33], and outward-facing carbonic anhydrase isoforms [34]. Measurement of pH_e exhibits value for detecting an invasive and potentially metastatic phenotype. Methods for imaging pH_e include (i) pH sensitive contrast agents for ¹H MRI [35–37], and paramagnetic chemical exchange saturation transfer mechanism (paraCEST) [38]; (ii) other non-proton MRS, for example the ³¹P biosensor, 3-aminopropyl phosphonate [39–43], and ¹⁹F probes such as 3-[N-(4-fluoro-2-trifluoromethylphenyl)-sulphamoyl]-propionic acid (known as ZK-150471) [41, 44, 45] and 6-fluoropyridoxol [46, 47]; and (iii) hyperpolarized MRS, including the SABRE [48, 49] techniques introduced above, (iv) nuclear imaging techniques such as PET and SPECT [50, 51], and (v) dissolution DNP, which is the focus of this Review on ¹³C-labeled biosensors.

A variety of methods have been employed for imaging of pH using DNP-MRS [10, 52–56]. The most widely reported biosensor for pH imaging using DNP-MRS is H¹³CO₃⁻. The core mechanism of the method is that mammalian tissues preserve a highly regulated acid-base balance via an equilibrated reaction between HCO₃⁻ and CO₂, catalyzed by the enzyme carbonic anhydrase in the extracellular environment [57]. Theoretically, if exogenous HP H¹³CO₃⁻ can rapidly equilibrate in tissues and establish a chemical equilibrium with HP ¹³CO₂, their detectable signals can be used to image pH_e using the Henderson-Hasselbalch equation, pH = pK_a + log₁₀([HCO₃⁻]/[CO₂]), assuming pK_a is known in vivo [58]. F. D. Gallagher et al intravenously injected HP H¹³CO₃⁻ in a mouse tumor model. First, they measured the T₁s of H¹³CO₃⁻ and ¹³CO₂ in vivo as 10.1 ± 2.9 s and 9.8 ± 2.5 s respectively at 9.4 T and found their apparent T₁s become equal due to the rapid interconversion, so that calculated pH would not vary in the course of HP decay. They then imaged pH_e and demonstrated that the average tumor pH_e was significantly lower than the surrounding tissue [10]. The same group quantified carbonic anhydrase activity through the rapid exchange rate of H¹³CO₃⁻ and ¹³CO₂ catalyzed by carbonic anhydrase by analyzing the change in signal decay before and after selective saturation of the HP ¹³CO₂ resonance. They have found that tumor cells overexpress carbonic anhydrase to compensate for decreased enzymatic activity due to the acidic extracellular TME [59].

Although HP H¹³CO₃⁻ has been demonstrated as a useful biosensor compound for in vivo pH_e imaging using MRS [10, 60], H¹³CO₃⁻ has several challenges, including low polarization, difficulties in preparing for hyperpolarization at high concentration, fast T₁ relaxation, and the use of toxic reagents such as Cs for the formulation [60]. These challenges have been mitigated by developing methods that rapidly produce HP H¹³CO₃⁻ and HP ¹³CO₂ indirectly from a more readily polarized precursor compound via a chemical reaction, while preserving the nuclear spin order (polarization). For example, R.K. Ghosh et al. exploited the rapid decarboxylation of HP [1,2-¹³C] pyruvate with hydrogen peroxide (H₂O₂) to produce hyperpolarized [1-¹³C] acetate, ¹³CO₂, and H¹³CO₃⁻ [61]. More recently, tissue pH was measured in vivo using HP ¹³C-ethyl acetyl carbonate. The compound was rapidly hydrolyzed by esterase in vivo to ¹³C-monoacetyl carbonate, which then decomposed to HP ¹³CO₂ [62]. Similarly, we polarized non-toxic ¹³C-carbonated small organic precursor molecules, and subsequently hydrolyzed them to HP H¹³CO₃⁻ and biocompatible components via a base-catalyzed dissolution step [52, 54, 63]. We verified HP [1-¹³C]1,2-glycerol carbonate (¹³C-GLC) as an excellent candidate for this HP H¹³CO₃⁻ producing method since ¹³C-GLC and its hydrolytic products including H¹³CO₃⁻, ¹³CO₂, and glycerol are nontoxic at high concentration [64] and ¹³C-GLC can rapidly (< 30 s) achieve >98 % breakdown at elevated temperature (~75 °C) in the presence of NaOH. In combination with hyperpolarized ¹³C MRS, we employed this method to image pH_e in the TRAMP model and correlated decreases in tumor pH to the increased grade of tumors (Figure 4). The subsequent in vitro assays and in vivo evaluation on PCa staging demonstrated this method can achieve a high polarization, produce high concentrations of HP H¹³CO₃⁻ in solution, and consequently obtain large signal gains for pH_e imaging [52, 54]. One important general consideration in the use of precursor molecules to generate H¹³CO₃⁻ is the potential for the generation of side products during chemical production. In order to facilitate future clinical translation of these methods, careful optimization of reaction conditions would be necessary.

Figure 4.

DNP-MRS imaging the interstitial pHe of TRAMP mice using HP H¹³CO₃⁻ demonstrates decreases in tumor pH associated with high grade disease. (A) HP ¹³C pH_e images of low-, high-, and separate distinct low- and high-grade tumors in the same TRAMP mouse. High grade tumor regions verified at pathology are circled in red, and low grade regions circled in blue. (B) and (C) scatter plots demonstrating the significant difference of pH_e between low- and high-grade regions over all mice on mean and regional-minimum (*p < 0.05).

In addition to the method of measuring the ratio of HP H¹³CO₃⁻ to HP ¹³CO₂ to calculate pH_e using the Henderson-Hasselbalch equation, other HP compounds exhibiting a pH-dependent chemical shift can be used to measure pH. In principle, a pH dependent chemical shift method has advantages including a potentially greater SNR because the signal is not split between two resonances. In contrast, bicarbonate has the advantages of being a non-toxic, endogenous molecule, routinely used in clinical medicine, suggesting feasibility for clinical translation. These chemical shift pH probes include N-(2-acetamido-)2-aminoethanesulfonic acid (ACES) [53], zymonic acid (ZA) [56], and [2-¹³C, D₁₀]diethylmalonic acid [55], among others [65, 66], all of which exhibit a large pH-dependent ¹³C chemical shift over the physiologic range. For example, co-polarization of [2-¹³C, D₁₀]diethylmalonic acid with [1–¹³C,D₉]tert-butanol, a reference compound with a pH-independent chemical shift, could accurately image pH via ¹³C NMR and MRS in phantom experiments [55]. Importantly, S. Duwel et al imaged pH_e within healthy rats’ bladders and kidneys, and xenograft tumors using co-polarized ZA and urea. They demonstrated the feasibility of measuring pH_e via monitoring the chemical shift variation of ZA signals relative to urea, or between ZA’s two ¹³C labeled sites upon pH changes. This method exhibits several advantages, in particular, the independence of concentration, temperature, ionic strength, and protein concentration [56]. Overall, these prior promising studies suggest a future potential for clinical translation of hyperpolarized ¹³C pH imaging for the detection of the acidic tumoral microenvironment in aggressive tumors.

2.3. Imaging imbalanced redox processes

Oxidative stress arises when the production of cellular reactive oxygen species (ROS) such as hydroxyl (OH^•), superoxide (O2^−•), hydrogen peroxide (H₂O₂), and nitric monoxide (NO^•), generated as metabolic by-products by biological systems overwhelm the intrinsic antioxidants. For example, mitochondria use oxygen to produce adenosine triphosphate, leading to the production of ROS [67]. Overproduction of free radicals can cause damage to biomolecules and lead to the development of chronic diseases including cancer [68, 69]. Moreover, some other non-radical molecules such as hydrogen peroxide (H₂O₂), nitric oxide, hypochlorous acid, and peroxynitrite that can generate free radicals under various conditions like the Fenton reaction may therefore cause oxidative stress [70]. The human body has built-in immune defense systems of counteracting oxidative stress by producing antioxidants as free radical scavengers to prevent and repair damages caused by radicals [71]. Consequently, rapid cancer cell proliferation and drug resistance are correlated with a highly reduced intracellular microenvironment as a result of a high-level accumulation of reducing agents such as glutathione [72, 73] and nicotinamide adenine dinucleotide phosphate (NADPH) [74] to maintain the reduction-oxidation (redox) balance [75]. This intracellular reducing environment is generally coupled with an oxidizing TME.

Ascorbic acid (AA) is an essential reducing agent (antioxidant) and cofactor of numerous enzymes to activate biosynthetic reactions [76–79]. Importantly, as shown in Figure 5a, AA and its oxidized form, dehydroascorbic acid (DHA), are closely coupled to two primary cellular antioxidants: glutathione (GSH) and NADPH [76]. Cells have two types of transporters to uptake AA: sodium-dependent vitamin C transporters (SVCT1, 2) directly transport AA into cells [80], while most tissues acquire AA through the transport of DHA into cells via glucose transporters (GLUT1, GLUT2, and GLUT4), which exhibit a similar affinity for both DHA and glucose and are overexpressed in most tumors [81, 82]. Inside the cells, DHA is rapidly reduced to AA through two reactions with either GSH or GSH-dependent enzymes including the thiol-disulfide oxidoreductases, glutaredoxin, and protein disulfide isomerase, and the predominant NADPH-dependent enzyme such as selenoprotein and thioredoxin reductase [76, 83, 84]. Thus, the rate of intracellular reduction of DHA to AA reflects oxidative stress since it is determined by the availability of NADPH [85, 86].

Figure 5.

The redox couple DHA and AA (VitC) is used as a biosensor for measuring oxidative stress. (a) Biochemical scheme illustrating the metabolic pathway between HP [1-¹³C] DHA and HP [1-¹³C] AA. HP [1-¹³C] DHA was used as an endogenous redox sensor for in vivo metabolic imaging: (b) distribution of HP [1-¹³C] AA and HP [1-¹³C] DHA in a TRAMP mouse, (c) representative ¹³C spectra from liver, kidney, and prostate tumor in a TRAMP mouse, and (d) summary of average metabolite ratios of HP [1-¹³C] AA / (HP [1-¹³C] AA)+ HP [1-¹³C] DHA) for normal liver, kidneys, and prostate (n = 5).

Importantly, the properties of AA and DHA make them the prime candidates for in vivo metabolic investigation using DNP-MRS [84, 87]. DHA shares the same membrane transporters with glucose and therefore exhibits rapid uptake [88], and the interconversion between DHA and AA is also rapid, enabling detection of intracellular DHA reduction on the hyperpolarized timescale. Therefore, researchers have evaluated the feasibility of using [1-¹³C]-DHA and [1-¹³C]-AA as ideal redox couple biosensor candidates for hyperpolarized redox imaging and considered their potential for clinical translation. For example, two significant initial research works published in 2011 elucidated the above-described redox mechanism, demonstrated a relatively high level of hyperpolarization of [1-¹³C]-DHA and [1-¹³C]-AA, and importantly proved the hypothesized mechanism. K. R. Keshari et al [87] interrogated the in vivo rapid transformation of HP [1-¹³C] DHA to HP [1-¹³C] AA in different organs and tumors, such as in liver, kidney, within tumor and in TRAMP model, as well as in normal rat brains, where DHA are well perfused (Figure 5b). TRAMP mice exhibited an increased conversion ratio of AA/[AA + DHA] within the tumor in comparison to the normal prostate gland or the surrounding benign tissue, but there was no significant difference in the liver and kidney between TRAMP and normal mice. Thus, the feature suggests the feasibility of using HP [1-¹³C] DHA as a biosensor for the detection of PCa. The cardiopulmonary effects of DHA on mice under anesthesia are well-known, but use of this redox sensor in humans may be possible in the future based on better preparative methods, and more sensitive hyperpolarized ¹³C imaging tools. Interestingly, a high and rapid reduction of HP [1-¹³C] DHA within the brain was observed, identifying DHA as an excellent biosensor to cross the blood-brain barrier to detect the oxidative stress inside the brain [89].

Another research work reported by S. E. Bohndiek et al [84] implicated AA and DHA could be used as extra- and intercellular biosensor candidates respectively. They observed that the extracellular pool of [1-¹³C]-AA was oxidized rapidly in suspensions of hypoxic EL4 lymphoma cells, suggesting the generation of oxidative stress with hydrogen peroxide from cancer cells that deplete oxygen and nutrients from medium [90, 91]. They also found that intracellular [1-¹³C]-DHA was reduced rapidly in both EL4 tumor cell suspension and in vivo EL4 tumor xenografts. Therefore, these results suggest HP [1-¹³C]-AA could be used for detecting ROS in the extracellular environment, where high concentrations of superoxide and hydrogen peroxide may be present [90]. Additionally, imaging AA/DHA metabolism provides an approach for determining in vivo metabolism of NADPH and the ratio of GSSG/GSH ratio, both of which indicate oxidative stress [92]. In a follow up study from our groups, we extended the hyperpolarized ¹³C AA approach to a similar ¹¹C PET imaging approach, where we found that extracellular [1-¹¹C]AA could be oxidized to [1-¹¹C]AA, and then internalized [93]. Thus, HP [1-¹³C]-DHA and AA (and potentially, their ¹¹C analogs) might be used to image redox status in vivo for the characterization of cancers including PCa [87, 94].

Hydrogen peroxide (H₂O₂) is an important metabolite involved in cellular redox metabolism reactions and processes. The concentrations of H₂O₂ in the intra- and extracellular spaces play a crucial role in determining the proliferation, survival, or death of the cells [95]. The intracellular homeostatic concentration ranges from 0.1 ~ 100 nM, depending on the cell type, while about 100-fold concentration gradient from extracellular to intracellular has been estimated. A higher concentration of H₂O₂ evokes inflammatory responses and leads to tumor growth, metastasis, growth arrest, and cell death [96]. Thus, the concentration of H₂O₂ acts as a potential biomarker for detecting the presence and progress of a variety of pathological changes. Accordingly, ¹³C biosensors have been developed for imaging H₂O₂. A. R. Lippert et al imaged oxidative H₂O₂ using HP ¹³C-benzoylformic acid as a ¹³C MRI contrast agent, which selectively and rapidly reacts with H₂O₂ to produce HP ¹³C-benzoic acid [97]. Wibowo et al developed ¹³C thiourea (TU) to image the dynamics of H₂O₂ in vivo with high sensitivity and spatiotemporal resolution. They demonstrated that ¹³C-TU is highly polarizable (10.4 ± 1.1% polarization) with a long T₁ (53.8 ± 3.7 s at 3T). HP ¹³C TU is oxidized by H₂O₂ to produce chemically distinguishable product HP ¹³C thiourea dioxide that is subsequently hydrolyzed to urea [98]. Overall, the biological relevance of these H₂O₂ measurements, and limits of in vivo detection remain an area for future study. Taken together, these prior studies demonstrate a potential to use hyperpolarized ¹³C MRI for the detection of alterations in redox homeostasis in the TME.

2.4. Imaging Alterations of Trace Divalent Metal Ions

Trace divalent ions, such as zinc, calcium, copper, iron, and selenium, play a vital role in many biochemical reactions for maintaining biological processes and function. Alterations in metal homeostasis are associated with pathological changes in the metabolic microenvironment of cancers [99–101]. Thus, the distribution of these metal ions could be potential biomarkers used for the detection of cancers and the prediction of therapy [102].

In principle, DNP-MRS could be well suited to the detection of metals. Two prior publications reported by A. Mishra et al [103] and S. Wang et al [104] established a solid foundation for the further development of this area. Mishra et al initially demonstrated the feasibility of using HP ¹³C-labeled metal chelators EDTA and EGTA to identify divalent metals including, Pb²⁺, As²⁺, Cd²⁺, Zn²⁺, Mg²⁺, and Ca²⁺ via the distinct carboxyl resonance variation by metal coordination (Figure 6a). They subsequently applied this technique to differentiate between biologically essential and toxic divalent metals and determined the concentration of Ca²⁺ in human serum [103]. However, one challenge potentially limiting the application of this method as an in vivo biosensor is the high concentration of Ca²⁺, leading to challenges in generating image contrast.

Figure 6.

(a) (left) Identification of divalent metals by metal-specific chemical shift of ¹³C-EDTA, (right) the binding curves for different metals as plots of the area under the curve (AUC), normalized to the sum of all AUCs to the chemical shift peak originating from metal-bound or unbound ¹³C-EDTA. (b) HP [1-¹³C] Cys phantom imaging experiment demonstrating accurate Zn²⁺ quantification in biological samples, including serum and prostate extracts.

S. Wang et al focused on the development of hyperpolarized amino acid-derived sensors for the detection of Zn²⁺ [104]. In normal prostate, zinc accumulates intracellularly to a remarkably high concentration, and then coordinates with citrate to be secreted into the prostatic fluid. In contrast, the zinc accumulation property becomes dysfunctional in de-differentiated PCa, leading to a 60–80% decrease in zinc concentration [105]. S. Wang et al identified [1-¹³C]L-cysteine ([1-¹³C]Cys) and [1-¹³C₂] iminodiacetic acid ([1-¹³C]IDA) as compounds with specificity and targetability of Zn²⁺ binding over other endogenous cations Na⁺, K⁺, Ca²⁺, and Mg²⁺ and negligible chemical shift changes over the physiological pH range from 6.5 to 7.4. Importantly, these two biosensors showed (i) a good level of hyperpolarization (13.4 ± 0.6% and 6.6 ± 0.9% back-calculated polarization for Cys and IDA, respectively), (ii) a large chemical change in the presence of Zn²⁺ (+4.6 and +7.3 ppm in the presence of equimolar Zn²⁺, respectively), (iii) the linear correlation of the chemical shift differences ([1-¹³C]Cys) or area% of bound peak ([1-¹³C]IDA) as the function of the concentrations of Zn²⁺ at low concentration range (0.25 and 0.5 eq.), and (iv) sufficient T₁ for MRI scans. However, due to substantial signal line broadening, [1-¹³C]IDA exhibited limited applicability. In subsequent assays, it was verified that HP [1-¹³C]Cys could accurately determine Zn²⁺ concentration in biological samples – rat serum and prostate extracts with low and high concentrations. The current results suggest feasibility of using HP ¹³C metal chelators as potential biosensors to image pathologies associated with alterations in metal ions homeostasis such as PCa, neurodegenerative diseases, and diabetes.

3. Summary and Prospect of HP TME biosensors in Medical Applications

The TME plays a critical role in the processes of cancerous growth, invasion, and metastasis. The DNP-MRS methods described in this review could be used to interrogate the abnormal TME. In turn, they could provide information for disease prognostication and precise personal therapy. The TME imaging methods described in this review have the potential for clinical utility. For example, decreases in tumor pH_e measured with HP pH imaging methods could be used to detect transformation from indolent to aggressive cancer phenotypes in diseases such as prostate or breast cancer. This is under active investigation in our laboratories. In order to drive these methods into broader use in a clinical setting, an impact on patient management would be needed, for example, to detect locally aggressive disease, to help guide appropriate surgical or oncologic management. In this review, we considered four cancer properties in TME, including elevated export of lactate, reduced interstitial pH, increased oxidization stress, and altered metal homeostasis. Imaging of these pathologies with HP MRS has strong potential to be included in future clinical studies.