Biocompatible Monophosphonated Trityl Spin Probe, HOPE71, for In Vivo Measurement of pO₂, pH, and [P_i] by Electron Paramagnetic Resonance Spectroscopy

Abstract

Hypoxia, acidosis, and elevated inorganic phosphate concentration are characteristics of the tumor microenvironment in solid tumors. There are a number of methods for measuring each parameter individually in vivo, but the only method to date for non-invasive measurement of all three variables simultaneously in vivo is electron paramagnetic spectroscopy paired with a monophosphonated trityl radical, pTAM/HOPE. While HOPE has been successfully used for in vivo studies upon intratissue injection, it cannot be delivered intravenously due to systemic toxicity and albumin binding, which causes significant signal loss. Therefore, we present HOPE71, a monophosphonated trityl radical derived from the very biocompatible trityl probe, Ox071. Here we describe a straightforward synthesis of HOPE71 starting with Ox071, and report its EPR sensitivities to pO₂, pH, and [P_i] with X-band and L-band EPR spectroscopy. We also confirm that HOPE71 lacks albumin binding, shows low cytotoxicity, and has systemic tolerance. Finally, we demonstrate its ability to profile the tumor microenvironment in vivo in a mouse model of breast cancer.

Graphical Abstract

Introduction

Low-field electron paramagnetic resonance with the use of paramagnetic spin probes allows for non-invasive measurement of critical biomarkers in pre-clinical^1–3 and, more recently, in clinical settings⁴. Two types of paramagnetic probes are used for in vivo EPR, namely particulate probes such as chars, coals, lithium phthalocyanine (LiPc), lithium naphthalocyanine (LiNc), and soluble probes such as nitroxide and triarylmethyl (TAM or trityl) radicals.⁵ To date, only particulate lithium octa-n-butoxy-naphtalocyanine (LiNc-BuO) embedded inside a biocompatible polymer (polymeric dimethylsiloxane, PDMS) named OxyChip has been used in clinical trials to assess tumor oxygenation⁴. Particulate spin probes are limited to oxygen measurement and have to be implanted at the site of measurement. The lack of particulate distribution throughout the tissue makes imaging applications very challenging. On the other hand, soluble probes are ideal for EPR imaging because of their high diffusion in tissues. Moreover, soluble probes are not limited to oxygen measurement⁶; TAMs and nitroxides can report other important physiological parameters such as pH³, microviscosity^7–8, redox status⁹, inorganic phosphate concentration [Pi]¹, thiols concentration¹⁰ in vivo. Among soluble probes, TAM radicals (Figure 1) show a higher in vivo stability and narrower lineshapes than nitroxides which results in higher functional sensitivity and signal-to-noise ratio, which is critical for in vivo applications. Almost all the TAMs reported are based on the Finland trityl (FT, also named CT-03) because of its well-optimized synthesis^11–12.

Figure 1.

Structure of TAM radicals used for in vivo EPR and their functional sensitivity.

We recently reported the synthesis and application of an extracellular monophosphonated Finland-based TAM probe named pTAM or HOPE whose EPR spectrum is sensitive to pH, pO_2, and [P_i].^{1, 13}. The EPR spectrum of HOPE is a doublet with a hyperfine splitting dependent on the ionization state, therefore providing pH sensitivity, while the linewidth is sensitive to oxygen. The signal is also sensitive to [P_i] because the concentration of phosphate affects the proton exchange rate, which causes the different ionic species peaks to coalesce^{6, 14–16}. This probe represents an important tool in cancer research as these chemical tumor characteristics play an important role in tumor progression as well as treatment resistance. While HOPE showed great promise for the multifunctional profiling of tissue microenvironment, it also has some limitations. Indeed, HOPE is a derivative of the Finland trityl (Figure 1). FT-based TAMs bind to albumin through hydrophobic interactions, which drastically reduces the signal intensity through line-broadening^17–18. More importantly, FT-based TAMs are toxic upon intravenous injection¹⁹, which limits in vivo work to small-dose intratissue delivery.

Alternatively, Ox063 and its deuterated analogue, Ox071, with twelve additional hydroxyethyl groups, are very hydrophilic and do not bind to albumin¹⁸. They also have remarkably low toxicity in mice (LD₅₀=8 mmol/kg)²⁰ and have been used broadly for in vivo Overhauser-enhanced MRI and EPR oximetry studies upon systemic delivery^21–24. However, the synthesis of Ox063/71 derivatives was held back for almost 20 years by the lack of reproducible synthetic protocols. Only very recently, their synthesis became available in the scientific literature²⁵, which made the synthesis of Ox063/71-based TAMs possible. Hereby, we report a monophosphonated derivative of Ox071 as a biocompatible multifunctional trityl probe for simultaneous EPR measurement of pO₂, pH_e, and [P_i].

Materials and Methods

General

X-Band and L-Band EPR spectra were recorded using a Bruker Elexsys E580 (9.5 GHz) and a Magnettech (1.2 GHz) spectrometers, respectively. For X-band spectra, 50 μL of the solution was filled into a gas-permeable Teflon tube (1.14 mm diameter and 60 μm wall thickness) from Zeus, Inc., and the temperature and the nitrogen/oxygen gas mixture were controlled inside the resonator using a Noxygen temperature and gas controller. Gas was flushed for at least 10 min before the measurement or until the spectrum became time-independent. For L-Band spectra, 800 μL of the solution was filled into a 1.5 mL conical tube. The temperature was controlled using a circulation thermostat. Nitrogen/oxygen gas mixture from the gas controller was bubbled into the solution for 25–30 min before the measurement. HRMS spectra were recorded on a Thermofisher Scientific Q Exactive Mass Spectrometer with an Electron Spray Ionization (ESI) source. HPLC-MS analyses were carried out using a Water Alliance e2695 separation module, a Water 2998 PDA detector, and a Water SQD2 mass detector. Purifications on C18 columns were carried out using a CombiFlash Rf+ purification system using water (containing 0.1% TFA) and acetonitrile (containing 0.1% TFA). A Thermo Scientific Orion Star A111 Benchtop pH meter with a Fisherbrand accumet Micro Glass Combination Electrode was used and was calibrated with 4.00 and 9.00 buffer solutions from Fisher Chemical before the experiments.

Synthesis

Step 1: Monophosphorylation of OX071.

Ox071 tri-sodium salt (959 mg, 0.661 mmol, 1 eq.) synthesized according to known protocols²⁵, was dissolved to a concentration of 1 mM in Na-phosphate buffer (10 mM, pH=7.4, 660 mL). Then, a solution of potassium hexachloroiridate(IV), K₂IrCl₆, (958 mg, 1.983 mmol, 3 eq.) in deionized water (70 mL) was added to generate the triarylmethyl cation; the green solution turned deep green-blue. The reaction was stirred for 10 seconds, and triethyl phosphite, P(OEt)₃, (2.196 g, 13.220 mmol, 2.27 mL, 20 eq.) was added. The reaction mixture was stirred for 10 minutes, and the conversion was checked by HPLC-MS. The mixture was acidified to pH ≤ 2 using trifluoroacetic acid and loaded into a C18 loading cartridge (25 g). The crude product was purified by reverse-phase chromatography using a C18 column (86 g) with a gradient of acidic water/acetonitrile (95/5 to 85/15) and freeze-dried. 138 mg of 2 was recovered as a green solid (14% yield). 431 mg (47%) of Ox071 was also recovered.

Step 2: Deprotection of the phosphonic acid.

Diethyl phosphonate HOPE71 2 (138 mg, 0.094 mmol, 1 eq.) was dissolved in anhydrous dimethylformamide (15 mL) under argon. The solution was cooled to 0°C, and bromotrimethylsilane, TMSBr, in excess (1 mL) was slowly added. After stirring for 10 minutes at 0°C, the reaction was heated at 50°C for 5 hours, and the deprotection was monitored by HPLC-MS. The TMSBr was removed under reduced pressure. Methanol (10 mL) was added, stirred for 1 min, and removed under reduced pressure. The DMF solution was diluted 20 times with deionized water and freeze-dried. The final product was purified using a C18 column (26 g) with a gradient from acidic water to acidic water/acetonitrile (80/20) and freeze-dried. The residue was dissolved in water and titrated to pH=7 with NaOH (0.1M) and freeze-dried again to afford 121 mg (88%) of HOPE71 as a green powder. HRMS (ESI−) calcd for C₅₁H₃₉D₂₄O₁₉PS₁₂⁻: 1418.1862, found: 1418,1743.

Spectral sensitivity to oxygen, pH, and inorganic phosphate.

X-band Acquisition parameters.

For each condition, a focus on the low-field part of the spectrum was recorded six times. For certain conditions (identified below), a full spectrum was recorded three times. Settings were as follows: power; 0.04743 mW, mod. freq.; 30 kHz, mod. Ampl.; 50 mG, conv. time; 30.01 ms, number of points; 2054 for the full spectrum and 1024 for the low-field spectrum, sweep width; 6.0 G for the full spectrum and 2.0 G for the low-field spectrum. All spectra were measured at 37°C.

L-band Acquisition parameters.

For each condition, a focus on the low-field part of the spectrum was recorded four times. Parameter settings for the spectrometer were as follows: mod. freq.; 100 kHz, mod. Ampl.; 0.05 G, scan time; 30s, number of points; 4096, sweep width; 0.960 G and non-saturating power. All spectra were measured at 37°C.

Spectral fitting.

A MATLAB-based application developed in-house with a graphical user interface for a non-linear fitting function (lsqcurvefit) was used to fit each EPR spectrum (Figure S1)¹³. Fitting was performed on the low-field focused spectra. The following values were determined as constants for the spectral modelling: Gaussian linewidth, Lorentzian linewidth of the acidic and basic peaks, and the distance between the acidic and basic peaks. The Gaussian linewidth and Lorentzian linewidths were determined by sextuplicate EPR spectra of HOPE71 200 μM measured with no oxygen, and with acidic (4 < pH < 5) or basic (pH > 10) conditions so that only the acid or basic peak was present. The acidic to basic peak distance was determined by sextuplicate EPR spectra measured with no oxygen and no phosphate buffer at pH≈7. The variable parameters that also contributed to or were dependent on the curve-fitting include oxygen-induced Lorentzian line broadening, proton exchange rate, and the acidic fraction of the population.

X-band Oxygen calibration.

To determine the linewidth sensitivity to oxygen for both ionic forms. 200 μM solutions of HOPE71 in 1 mM phosphate buffer and 137 mM NaCl at pH=5 and 10 were prepared, and spectra were recorded for pO₂ = 0, 19, 38, 76, 114, 159 mmHg. Full spectra were recorded for both pH=5 and pH=10 with pO₂ = 0 mmHg. The spectral fitting described above was used to determine the oxygen-induced line broadening for each pO₂. Linear fit allows for the determination of the oxygen sensitivity for both ionic forms combined.

X-band pH calibration.

To determine the pKa of the probe, a 200 μM solution of HOPE71, 1 mM phosphate buffer and 137 mM NaCl was titrated by the addition of small amounts of HCl and NaOH. For each titrated pH, spectra were recorded with pO₂ = 0 mmHg. Full spectra were recorded of pH=7.1. The spectral fitting described above was used to determine the acidic fraction at each measured pH. A plot of pH versus acidic fraction (P_a) was fitted to the equation ${\mathit{P}}_{\mathit{a}}=\frac{1}{1+{10}^{\mathit{p}\mathit{H}-\mathit{p}\mathit{K}\mathit{a}}}$ to determine the pKa.

X-band Inorganic phosphate calibration.

To determine the exchange rate dependence on phosphate concentration, 200 μM solutions of HOPE71 at pH=7.2 were recorded for phosphate concentration of 0, 1, 4, 7 and 10 mM. The spectral fitting described above was used to determine the exchange rate for each phosphate concentration. The proton exchange between phosphate and HOPE71 is expressed by the following equation:

$$\text{HOPE}{71}^{3-}+{\text{HPO}}_4^{2-}\underset{{\text{k}}_r}{\overset{{\text{k}}_f}{\rightleftharpoons }}\text{HOPE}{71}^{4-}+{\text{H}}_2{\text{PO}}_4^{-}$$

The rate of proton loss of HOPE71³⁻ to HPO₄²⁻ is k_a = k_f [HPO₄²⁻]. The rate constant k_f can be determine by linear approximation of the dependence of k_a on [HPO₄²⁻] or the total phosphate concentration [P_i] using [HPO₄²⁻] = [P_i] (H⁺ + K_a^B₎/(2πK_a^B). Where $K_a^B$ is the dissociation constant of phosphate buffer ($K_a^B$=10^−6.66)²⁶. Linear fit of the equation $\frac{k_a\left(H^{+}+K_a^B\right)}{2\pi K_a^B}=k_f\left[P_i\right]$, allows determination of the phosphate sensitivity.

Validation of the calibration with blind samples. (X-band).

Three solutions with 200 μM of HOPE71 and 137 mM NaCl were prepared with various concentrations of inorganic phosphate (1–10 mM), pH, and pO₂ that were blinded to the researcher acquiring the spectra. The spectral fitting described above was used to extract the oxygen-induced Lorentzian linewidth broadening, proton exchange rate, and acidic fraction. The calibrations above were used to determine the EPR measured pO₂, pH, and [P_i].

L-band calibrations and validation.

The pO₂, pH, and [P_i] calibrations and validation were repeated for L-band EPR with 500 μM solutions of HOPE71. Oxygen calibration spectra were recorded for pO₂ = 0, 17, 34, 68, 103, 159 mmHg. Phosphate calibration spectra were recorded on solutions with 0, 1, 4, 7, 10 mM of inorganic phosphate.

In vitro Biological Characterizations.

Albumin binding.

To study the interaction of HOPE71 with albumin, 200 μM solutions of HOPE71 at pH=7 were prepared, with 0 and 200 μM bovine serum albumin (BSA) (1 eq.). For comparison, 200 μM solutions of HOPE at pH=7 were also prepared with 0, 50, 100 and 200 μM BSA. Spectra were recorded for each solution with X-band EPR. Spectra of HOPE were acquired with mod. Ampl. = 25 mG.

MTT assay for cell toxicity.

MDA-MB-231 (triple-negative breast cancer) cells were plated on a 96-well flat-bottom plate and allowed to grow overnight to 60–70% confluency. The medium used was DMEM with 10% fetal bovine serum. The cells were then incubated with increasing concentrations of HOPE or HOPE71 in the medium for 24 hours. DMSO (2% and 5% in medium) was used as a cytotoxic control. Medium with corresponding concentrations of HOPE or HOPE71 was used as a background control. All conditions were performed in quadruplicate. The MTT assay was performed using ThermoFisher Vybrant MTT Cell Proliferation Assay Kit according to the manufacturer’s protocol. Absorbance was measured at 570 nm minus 630 nm to correct for cell debris. This process was repeated with HUVECs (Human umbilical vein endothelial cells) using vascular cell basal medium plus ATCC Endothelial Cell Growth Kit-VEGF.

In vivo Applications.

Mouse model.

Female MMTV-PyMT (PyMT+/−) mice with spontaneous mammary tumors and their wild-type littermates (PyMT−/−), 10–16 weeks in age, were used for HOPE71 in vivo EPR studies. Mice were anesthetized by inhalation of isoflurane in air prior to injection and during the acquisition of spectra.

In vivo L-band Acquisition parameters.

For each in vivo EPR measurement of HOPE71, the resonator coil was placed on mammary gland 4 (MG4) or 9 (MG9). Focus on the low-field part of the spectrum was recorded six times. Parameter settings for the spectrometer were as follows: mod. freq.; 100 kHz, mod. Ampl.; 0.05 G, scan time; 30s or 60s, number of points; 4096, sweep width; 0.960 G and non-saturating power.

Intravenous Tumor Profiling.

Six female MMTV-PyMT mice, 10–16 weeks in age, with MG9 tumors approximately 1 cm in diameter (370–580 mm³), were administered a bolus dose of HOPE71 in saline (90 μL, 75mM, 10mg, 0.17–0.31 mmol/kg, pH=7) by retroorbital injection. EPR spectra focused on MG9 tumors were recorded immediately 5 min after injection.

Systemic Toxicity.

Six mice from the intravenous tumor profiling were observed for apparent signs of toxicity for 7–10 days after injection. The mass of five mice was tracked during this time.

Longitudinal Tumor profiling.

Eight female MMTV-PyMT mice were watched starting at 9 weeks old for tumor growth on mammary gland 9 (MG9). Stage III and IV tumors on MG9 were profiled by HOPE71 EPR once to twice per week (once for early stage; twice for late stage) until the tumor burden significantly impeded quality of life, about 15–16 weeks old. Four female wild-type (PyMT−/−) littermates were also profiled on the same days. HOPE71 in saline (15–50 μL, 2mM, pH=7) was injected intratissually into the tumor or the fat pad mammary gland. EPR spectra were recorded immediately after injection.

Intratissue versus intravenous injection.

To detect if there are significant differences in the tumor profiling based on delivery method, the pO₂, pH, and [P_i] of tumors profiled with intravenous injection were compared to those of sized matched tumors profiled by intratissue injection. Two-tailed T-test for independent means was used to test for statistically significant differences.

Results and Discussion.

Synthesis of HOPE71.

In order to develop a multifunctional pO₂, pH, [P_i] probe that maintains all the favorable spectral features of HOPE but prevents hydrophobic interaction with blood biomacromolecules and increases biocompatibility, we thought to synthesize a mono-phosphonated TAM, based on the Ox071 scaffold instead of the FT scaffold (see Figure 1). Indeed, Ox071 is remarkably non-toxic and does not bind blood biomacromolecules. In order to selectively substitute one carboxylic acid on Ox071 with a phosphonic acid, we used an ipso aromatic nucleophilic substitution reaction for TAM radicals²⁷. The one-electron oxidation of Ox071 using potassium hexachloroiridate(IV) in phosphate buffer leads to the trityl cation Ox071⁺ (Figure 2A). Then, the addition of triethyl phosphite leads to a selective nucleophilic attack of the phosphite at the para-position of one aryl ring, which triggers an oxidative decarboxylation leading to the mono-phosphonated trityl radical 2¹³. The recovery of ~50% of Ox071 radical indicates that the oxidant of intermediate 1 is the Ox071 cation. Therefore, for this mechanism, the theoretical yield of the mono-substituted product is capped at 50%. Also, the trityl cation was found to be very short-lived in an aqueous solution at neutral pH, leading to a mixture of Ox071 and quinone methide if left without the phosphite^{18, 28}. The best yield came from adding the phosphite within 10 seconds of adding the K₂IrCl₆. Then, bromotrimethylsilane efficiently deprotected the phosphonate to yield HOPE71 in only two steps from Ox071.

Figure 2.

Synthesis of HOPE71 and X-Band Spectra. A. Synthesis of HOPE71 starting with Ox071. B. Full X-band EPR spectra of HOPE71 measured under N₂ at pH 5, 7, and 10. HOPE71³⁻ has a greater hyperfine coupling constant (a_P) than HOPE71⁴⁻, forming resolvable peaks, and the ionized species are in equilibrium around physiological pH.

EPR Characterization of HOPE71.

HOPE71 was first characterized at X-band. Figure 2B shows the full spectrum of HOPE71 at pH=5, 7, and 10. The phosphorus (l=1/2) causes hyperfine splitting that is dependent on the protonation state (Fig. 2B and S2). HOPE71³⁻ at pH=5 has a greater hyperfine splitting constant (a_p = 3.54 G) than HOPE71⁴⁻ at pH=10 (a_p = 3.26 G). The peak-to-peak linewidth was 63 mG for HOPE71³⁻ and 64 mG for HOPE71⁴⁻. At pH=7, both ionic forms are present, and their EPR spectra are resolved, but the peaks for the low-field part of the spectrum show a better resolution because of the difference in g-factor between both ionic species (Δg=0.00002). The ~50/50 ratio indicates a pKa for the phosphonic acid close to 7.

To determine the sensitivity of the linewidth to oxygen, low-field spectra were recorded with increasing oxygen partial pressure on solutions of HOPE71 at pH=5 and 10. Figure 3A demonstrates the effect of oxygen on the linewidth and signal intensity at pH=10. The extent of line-broadening was determined by fitting each spectrum using a custom MATLAB app (see Figure S1), allowing the Lorentzian linewidth to vary. The oxygen-induced Lorentzian line-broadening reached 0.51 mG/mmHg (Figure 3B). Next, In order to determine the pKa of HOPE71, a solution of HOPE71 was titrated with small amounts of acid or base, and spectra were recorded for six pH points between 6 and 8 under a nitrogen atmosphere. The ratio of the peaks is directly correlated to the mole fraction of HOPE^3– and HOPE^4– (Figure 3C). Spectral line fitting was used to determine the acidic mole fraction ꭕ_{HOPE71 3−} for each pH point. By plotting the acidic and basic mole fractions versus pH, the titration curves show the pKa to be 7.1±0.1 (Figure 3D). Finally, the effect of P_i was investigated. Solutions of HOPE71 with increasing concentrations of P_i were recorded under nitrogen. The presence of phosphate buffer increases the proton exchange rate on the phosphonate group of HOPE71, which causes the acidic (HOPE71³⁻) and basic (HOPE71⁴⁻) peaks to coalesce (Figure 3E). The proton exchange rate (k_a) can be derived from spectral fitting. The exchange rate adjusted for pH has a linear relationship against [P_i] and is equal to 2.36±0.05×10⁴s⁻¹mM⁻¹ (Figure 3F)²⁶.

Figure 3.

Effect of pO₂, pH, and [Pi] on the X-Band EPR spectrum of HOPE71. A. Linewidth sensitivity to oxygen for the low field component at pH 10. B. The linear relationship between oxygen partial pressure (mmHg) and oxygen-induced line-broadening (mG) for pH=5 and pH=10 is 0.51±0.05 mG/mmHg (R²=0.998). C. Effect of pH on the ratio of the peaks corresponding to HOPE71³⁻ and HOPE71⁴⁻ for the low field component of the spectrum measured under N₂. D. Titration curves for the mole fraction of HOPE71³⁻ and HOPE71⁴⁻ versus pH show a pKa=7.1±0.1 (R²=0.999). E. Effect of the concentration of Pi on the low field component of the spectrum measured under N₂ and at pH 7. F. The proton exchange rate k_a, adjusted for pH, has a linear relationship versus inorganic phosphate concentration (k_f=2.36±0.05×10⁴s⁻¹mM⁻¹) (R²=0.999).

To validate that the three parameters can be extracted simultaneously from one spectrum, EPR spectra were recorded for HOPE71 solutions with random pO₂, pH, and [P_i] within the range of the calibrations that were blinded from the experimenter. The spectra were fitted, allowing the oxygen-induced line broadening, acidic fraction, and exchange rate to vary. The values for random pO₂, pH, and [P_i] were extracted using the calibrations in Figure 3. When comparing the prepared values to the EPR-derived values, the accuracy was within 2 mmHg of oxygen, 0.1 units of pH, and 0.3 mM of [P_i] (Table. 1), which validated the fitting procedure.

Table 1.

Validation of the X-band EPR Calibrations.

	Gas Controller Set (mmHg)	EPR measured pO2 (mmHg)	pH meter	EPR measured pH	[Pi] Prepared (mM)	EPR measured [Pi] (mM)
Sample 1	47.9	47.0 (±4.7)	7.28	7.26 (±0.1)	1.88	2.14 (±0.10)
Sample 2	34.9	32.2 (±3.2)	7.09	7.08 (±0.1)	6.88	7.00 (±0.10)
Sample 3	19.0	17.3 (±1.7)	6.94	6.97 (±0.1)	8.25	8.22 (±0.10)

^a.The spectra were fitted, and the corresponding pO₂, pH, and [Pi] were derived using the calibrations in Figure 3.

For in vivo applications, the penetration depth at X-Band (≈9.5 GHz) is not sufficient. Therefore, L-Band (≈ 1 GHz) or lower frequencies are required. The major difference between the spectrum of HOPE71 at X- and L-Band is the significant decrease of the effect of Δg, which caused the asymmetry of resolution between the low-field and the high-field components of the spectrum at X-Band (Figure S3). The oxygen and P_i calibrations needed to be repeated to adjust for small differences in line shape and width due to instrument variances. The pH calibration was repeated for rigor (See Figure S4A–D). The L-band slope for oxygen sensitivity was determined to be 0.41±0.05 mG/mmHg. The pKa measured by L-band EPR agrees with the X-band measurement of 7.1±0.1. For the calibration of inorganic phosphate, the L-band slope, or k_f, was found to be 1.86±0.05×10⁴s⁻¹mM⁻¹. Blind sample testing was repeated for the L-band calibrations, and the accuracy was estimated to be within 4 mmHg of oxygen, 0.1 units of pH, and 0.3 mM of [P_i].

Biocompatability of HOPE71.

The biocompatibility improvements of HOPE71 over HOPE were evaluated in several ways. First, X-band EPR spectroscopy was used to investigate the interaction of the probe with albumin. For HOPE71, there was no significant loss in the spectrum signal intensity when BSA was included in equal concentration to the probe (Figure 4A), showing no interaction between BSA and HOPE71. In contrast, when equal concentration was included with HOPE, there was over 80% loss in signal intensity (Figure 4B), showing strong interaction between BSA and HOPE. Indeed, when the HOPE interacts with BSA, the probe’s rotational correlation time (τ_corr) increases, which results in line broadening and a decrease in signal intensity.¹⁷ HOPE71 is significantly more hydrophilic with the addition of twelve hydroxyethyl groups which prevent hydrophobic interaction with BSA. Because albumin concentration is so high in the blood, this lack of binding is imperative for intravenous dose efficiency. Next, the cytotoxicity of HOPE71 and HOPE was assessed using MTT cell viability assays on rapidly dividing breast cancer cells (MDA-MB-231) and endothelial cells (HUVEC) for a 24h incubation period. On the breast cancer cells, both probes showed a minor loss in cell viability at higher concentrations (1–5 mM), (Figure 4C). Similarly, HUVECs had a significant loss in cell viability for both at the highest concentration (5 mM), but up to 1mM, cells treated HOPE71 did not show any decrease in cell viability in contrast to the cell treated with HOPE (Figure 4D). Because the local target concentration for L-band EPR spectroscopy is several hundred micromolar, we conclude that HOPE71 is not toxic. In addition, the pharmacokinetic half-life of Ox071 in the blood is less than 30 min²², while our incubation time was 24h. However, in vivo systemic toxicity is much more complex than cytotoxicity. HOPE can only be used in vivo by intratissue injection of a small dose¹, but when larger doses of HOPE or FT are given for systemic delivery (intravenous or retro-orbital), the mice quickly die¹⁹. Therefore, we observed the health of mice injected intravenously with a bolus dose of HOPE71 (n=5) required for EPR spectroscopy. No mice died in the short time after injection or showed apparent signs of distress or weight loss for a week following the injection (Figure 4E), suggesting the high biocompatibility of HOPE71. This is a significant improvement over HOPE.

Figure 4.

Biocompatibility of HOPE71. A. X-band EPR spectra of 200 μM HOPE71 with 0 or 200 μM BSA, measured at 37°C, pH = 7, and with 1mM phosphate buffer and 137 mM NaCl. No significant loss in signal is detected. B. X-band EPR spectra of 200 μM HOPE with 0, 50, 100, or 200 μM BSA, measured at 37°C, pH = 7, and with 1mM phosphate buffer and 137 mM NaCl. The loss in signal intensity with increasing concentration of BSA demonstrates HOPE binding to BSA. C-D. MTT cell viability assays using MDA-MB-231 breast cancer (C) and HUVEC (D) cells incubated in increasing concentrations HOPE71 or HOPE for 24 hours. DMSO was used as a cytotoxic control. E. Mass over time (days) of female MMTV-PyMT after intravenous injection of HOPE71 in saline (10mg).

Profiling tumors with HOPE71.

The in vitro results suggest that HOPE71 could be used for systemic delivery. The lack of albumin binding is also advantageous for intra-tissue delivery as the binding of HOPE to albumin results in a decrease in the EPR signal intensity, leading to a higher dose being required to gain sufficient signal-to-noise for spectral analysis. To demonstrate the potential of HOPE71 for profiling the tumor microenvironment, we used the MMTV-PyMT mouse model that develops spontaneous mammary tumors. This is a well-characterized mouse model of breast cancer²⁹. We compared intra-tissue and systemic delivery. For systemic delivery, we found that 10 mg (0.19–0.31 mmol/kg) of HOPE71 in saline was a sufficient bolus dose by intravenous injection to have a sufficient signal-to-noise ratio for L-band EPR spectroscopy for at least 30 minutes after injection. For the intra-tissue delivery, the required dose was significantly lower (45–150 μg, 1.2–3.0 μmol/kg). Note that the dose should be kept minimal to avoid self-broadening that could lead to artificially high pO₂ readings.³⁰ We compared the TME parameters pO₂, pH, and Pi for stages III and IV and the two modes of delivery. Figure 5A–C shows that while pO₂ was not significantly different between intra-tissue and intra-venous for stages III and IV, [Pi] is the first parameter to show differences at stage III. For stage IV, both [Pi] and pH are significantly different. The increased differences between systemic and local delivery support that there is a loss of good blood perfusion during tumor progression. Indeed, only the perfused region of the tumor could be reached with systemic delivery. We then compared tumors for stages III and IV versus healthy mammary glands of the wild-type littermates (Figure 5D–F). Again, Pi was the first parameter to deviate at stage III, while both Pi and pH were found to be significantly different between healthy mammary glands and tumors for stage IV. No statistical difference was found for pO₂ during the study. The results demonstrate that HOPE71 can be used to profile TME upon local and systemic delivery. For systemic delivery, the probe reports parameters from the extracellular space and vascular phase because of the lack of albumin binding. This can be seen as a disadvantage because the tissue parameters are more relevant. The use of a blood-pool paramagnetic broadening agent (e.g. Gd³⁺ contrast agent) in combination with HOPE71 could be helpful to subtract the vascular contribution to the measured parameters.

Figure 5.

Profiling mouse breast tumors using HOPE71. Measurements for pO₂, pH and [P_i] were derived from averages of six L-band spectra recorded after intratissue or intravenous injection of HOPE71 in saline. Breast tumor staging was estimated by age of female PyMT(+/−) mice (Stage III=10–12.9 weeks old, Stage IV=13–15.9 weeks old).²⁹ Wild type measurements were made in the mammary fat pad of female PyMT(−/−) littermates. A-C. Comparison between measurements of pO₂, pH and [P_i] after intravenous (IV) or intratissue (IT) injection of HOPE71 in stage III and stage IV mouse breast tumors. A. No significant difference in pO₂ between IV and IT injection. B. Significant difference in pH between IV and IT injection for stage IV breast tumors (p<0.001). C. Significant difference in [P_i] between IV and IT injection for stage III (p<0.001) and stage IV (p=0.01) breast tumors. (*p<0.05, **p<0.01, ***p<0.001) D-F. All measurements in these figures were made after intratissue injection of HOPE71 E. No significant difference in pO₂ between stage III and IV breast tumors and WT mammary glands. E. Significant difference in pH between stage IV breast tumors and WT mammary glands (p<0.001). F. Significant difference in [P_i] between stage III (p=0.007) and stage IV (p<0.001) breast tumors and WT mammary glands.

Conclusions

We propose HOPE71, a monophosphonated hydroxyethyl trityl radical probe, as a biocompatible multifunctional EPR probe for in vivo measurement of pO₂, pH, and [P_i]. HOPE71 can be synthesized starting with Ox071 by a 2-step process. The EPR-sensitivities of HOPE71 with X-band and L-band spectroscopy were verified and are comparable to HOPE. Furthermore, HOPE71 has clear biocompatibility improvements over HOPE, including a lack of albumin binding and systemic tolerance. The ability of HOPE71 to profile tumors in mouse models of breast cancer by EPR spectroscopy was demonstrated with both intravenous and intratissue injections. Moreover, HOPE71 represents an ideal spin probe for future in vivo simultaneous EPR imaging of pO₂, pH, and [P_i].³¹