Exploring the potential of the novel imidazole-4,5-dicarboxyamide chemical exchange saturation transfer scaffold for pH and perfusion imaging

Abstract

Here we describe and assess the potential of 14 newly synthesized imidazole-4,5-dicarboxyamides (I45DCs) for pH and perfusion imaging. A number of these aromatic compounds possess large labile proton chemical shifts (up to 7.7 ppm from water) due to their intramolecular hydrogen bonds and a second labile proton to allow for CEST signal ratio based pH measurements. We have found that the contrast produced is strong for a wide range of substitutions and that the inflection points in the CEST signal ratios vs pH plots used to generate concentration independent pH maps can be adjusted based on these subsitutions to tune the pH range that can be measured. These I45DC CEST agents have advantages over the triiodobenzenes currently being employed for tumor and kidney pH mapping both preclinically and in intial human studies. Finally, as CEST MRI combined with exogenous contrast has the potential to detect functional changes in the kidneys, we evaluated our highest performing anionic compound (I45DC-diGlu) on a unilateral urinary obstruction mouse model and observed lower contrast uptake in the obstructed kidney compared to the unobstructed kidney and that the unobstructed kidney displayed a pH ~ 6.5 while the obstructed kidney had elevated pH and increased range in pH values. Based on this, we conclude that the I45DCs have excellent imaging properties and hold promise for a variety of medical imaging applications, particularly renal imaging.

Graphical Abstract

This manuscript describes the synthesis and characterization of a novel series of CEST MRI pH imaging agents based on imidazoles and testing the top agent on a Unilateral Urinary Obstruction mouse model.

1 ∣. INTRODUCTION

Chemical exchange saturation transfer (CEST) MRI is an emerging MRI technique that involves selective saturation of labile protons (including hydroxyl (-OH), amide (-CONH), amine (-NH₂) and thiol (-SH) proton) on the molecules, followed by rapid saturation transfer between various labile protons and the bulk water pool^1-6. This process allows signal amplifications of factors of 500 or more to allow detection of low concentrations of CEST agents⁵. Another feature of CEST contrast is that the exchange rate is related to environmental pH, and clever design of CEST probes can be used to produce pH maps through looking at the ratio of two signals at different frequencies^7-9. This contrast can be produced both by endogenous compounds, which has been tested for providing additional diagnostic information for tumor patients^10,11 and exogenously administered compounds such as glucose in the first patient evaluations for highlighting tumors^12,13. However, most organic CEST agents suffer from reduced sensitivity at 3 T and below because of low SNR, shorter endogenous T1, and insufficient exchangeable proton chemical shift from water for detection at these clinical field strengths. Therefore, there is an emerging need to identify new high performance and well tolerated CEST agents.

Obstructive uropathy is a type of kidney injury based on impaired urine transit most commonly occurring due to obstructing stones, strictures, malignancy or congenital anatomic causes. While the relationship between urinary obstruction and acute kidney injury (AKI) is well established, a recent multicenter investigation revealed an independent association with chronic kidney disease (CKD), resulting in increased patient morbidity and mortality¹⁴. Despite the risk of irreversible renal injury, current laboratory and diagnostic techniques remain limited in assessing renal injury in the setting of urinary obstruction (especially in the setting of confounding comorbidities like diabetes and hypertension), which may lead to delayed intervention and permanent kidney damage. Standard blood and urine tests are not very informative about the extent of renal injury and the reversibility of renal impairment for UTO patients¹⁵. Multi-phase computed tomography is often utilized to assess the etiology of a urinary obstruction, but is unable to fully quantify loss in kidney function. Renal scintigraphy allows measurement of differential function but presents challenges due to its low resolution, challenges with handling motion and limits for performing after hours or on offsite patients^16,17. In this context, kidney tissue pH could provide more information on kidney function, particularly for patients with UTOs, because the kidneys are one of the main controllers of acid-base equilibrium and aberrations in renal parenchymal pH may be an indicator of incipient renal injury. Moreover, in the setting of nephrolithiasis, urine pH tests are widely utilized, as deviations in urine pH can predispose to stone formation. There are several MRI strategies for measuring pH including use of spectroscopic pH imaging probes such as ISUCA¹⁸, use of pH dependent relaxation probes^19,20 and use of hyperpolarized bicarbonate²¹. Recently PET/MRI was used to measure in vivo tumor pHe with PET/MRI co-agents, which showed great promise in clinical translation ²². However, all of these methods have limitations for translating to patients, for spectroscopic imaging probes there are sensitivity limitations, for relaxation probe pH mapping of the kidneys there are challenges due to difficulties teasing out changes in concentration from changes in pH and for hyperpolarized probes there are limitations imposed by the length of time that the polarization remains based on relaxation, for PET/MRI there are limitations imposed by requiring to perform both PET and MR and co-register these images. A method which can more readily separate these changes out with good sensitivity and no time limitations could render MRI based pH imaging as a good diagnostic for kidney imaging, particularly for managing UTO patients.

Based on the potential of CEST imaging for pH maps, we have long been interested in preparing CEST probes that are well suited for pH measurements. In order to get around the sensitivity limitations of CEST imaging, we and others have sought out and found special aromatic compounds with large labile proton chemical shifts (> 5.0 ppm) and suitable proton exchange rates based on the tuning of these two properties through intramolecular hydrogen bond including the salicylates, anthranilates, porphyrins, hydrazones and acetanilides^23-28. Previously, we have identified two Imidazole- 4,5-dicarboxyamides (I45DCs) which were well tolerated, possessing -NH protons resonating 7.7 – 7.8 ppm upfield from water and also with two labile protons to enable producing ratiometric pH maps, I45DC-diGlu and I45DC-diAsp²⁹. In the present study, we have explored the potential of an extended series of I45DCs including synthesizing 14 new I45DCs and testing them in vitro for their CEST MRI properties in the physiological pH range. Our most promising agent (I45DC-diGlu) was then tested in vivo on a unilateral urinary obstruction (UUO) mouse model of UTO to determine if administration of this compound could detect differences between obstructed and unobstructed kidneys.

2 ∣. MATERIALS AND METHODS

2.1. General Chemistry

Unless otherwise specified, the syntheses described below were carried out at ambient atmosphere and temperature. Anhydrous solvents such as N,N-Dimethylformamide (DMF), diisopropylethylamine (DIPEA) and benzene were purchased from Sigma-Aldrich. The solvent H₂O was obtained from a purification system from EMD Millipore. Deuterated solvents were purchased from Cambridge Isotope Laboratories. All other reagents and solvents were purchased from commercial vendors and used without further purification. HPLC purifications were performed using ¹²C columns (20 mm × 250 mm, Phenomenex Inc.) at 6 mL/min flow rate of water-acetonitrile eluent, unless otherwise specified. ¹H NMR spectra were recorded on a 500 MHz spectrometer. The chemical shifts are reported as δ values (ppm) relative to the water signal in deuterated methanol (MeOD). Mass spectral analyses were performed using the electrospray ionization (ESI) method.

2.2. Synthesis of imidazole-4,5-dicarboxamides

To a dry round-bottom flask I45DCA (3.12 g, 20.0 mmol) and 30 mL of benzene were added. To this stirred suspension 13.0 mL of thionyl chloride (180.0 mmol) and 0.772 mL of DMF (10.0 mmol) were added and the resulting mixture was refluxed for 16 h. After the mixture was cooled to room temperature, the solid product was collected by vacuum filtration, washed with two 20 mL portions of benzene, and dried under vacuum to yield acyl chloride, which was used without characterization. Then to a suspension of acyl chloride (1, 1.56 g, 5.0 mmol) and N,N-diethylaniline (1.60 mL, 10 mmol) in THF (30 mL), H-Glu(OtBu)-OtBu·HCl (2.96 g, 10.0 mmol) were added at −78±C. The reaction was kept for 1 hr at this temperature and then warmed to room temperature. After stirring 3 h, the yellow solid was precipitated from the solvent and filtered to give Glu-Pyrazine in 90% yield. Then to a stirring solution of Glu-Pyrazine (379 mg, 0.5 mmol) and H-Thr-OtBu·HCl (423.0 mg, 2.0 mmol) in THF (30 mL) DIPEA (517 mg, 4.0 mmol) was added. After stirring for 6h, the solvent was evaporated and crude product was purified by flash column chromatography to get the intermediate. This intermediate was dissolved in 5 mL TFA/DCM (1/1) for 2 h at room temperature. After all of the solvent was removed under vacuum, I45DC-GluThr was purified by HPLC. Compounds GluTyr, GluTrp, GluArg, GluLys, GluLeu, GluEa, diGlu, and GluTrp were synthesized with a similar procedure (Scheme 1).

Scheme 1.

Synthesis for I45DC-GluThr.

GluTyr yield 55% white powder. LCMS ESI⁺: 471.14 [M+Na], expected mass for C₁₉H₂₀N₄NaO₉⁺ 471.11. ¹H NMR (500 MHz, MeOD): δ 7.74 (s, 1H), 7.07 (d, J=10.0 Hz, 2H), 6.67 (d, J=10.0 Hz, 2H), 4.77-4.80 (m, 1H), 4.66-4.69 (m, 1H), 3.19-3.23 (m, 1H), 3.01-3.06 (m, 1H),2.43-2.54 (m, 2H), 2.31-2.36 (m, 1H), 2.08-2.15 (m, 1H). ¹³C NMR (126 MHz, MeOD): δ 176.29, 174.56, 174.49, 162.39, 157.34, 137.24, 131.47, 128.69, 116.23, 55.52, 53.06, 37.96, 31.08, 28.26.

GluTrp yield 52% white powder. LCMS ESI⁺: 494.15 [M+Na], expected mass for C₂₁H₂₁N₅NaO₈⁺ 494.13. ¹H NMR (500 MHz, MeOD): δ 7.71 (s, 1H), 7.55 (d, J=10.0 Hz, 1H), 7.29 (d, J=5.0Hz, 1H) 7.16 (s, 1H), 7.04 (t, J1=10.0 Hz, J2=5.0 Hz), 6.94 (t, J1=10.0 Hz, J2=5.0 Hz), 4.65 (t, J=5.0 Hz, 1H), 3.45-3.49 (m, 1H), 3.33-3.36 (m, 1H), 2.45-2.49 (m, 2H), 2.22-2.34 (m, 1H), 2.07-2.11 (m, 1H). ¹³C NMR (126 MHz, MeOD): δ 176.31, 174.88, 174.55, 137.96, 137.16, 128.80, 124.90, 122.32, 119.72, 119.31, 112.18, 110.52, 54.87, 53.03, 31.09, 30.05, 28.72, 28.26.

GluArg yield 60% white powder. LCMS ESI⁺: 442.19 [M+H], expected mass for C₁₆H₂₄N₇O₈⁺ 442.17. ¹H NMR (500 MHz, MeOD): δ 7.79 (s, 1H), 4.63-4.69 (m, 2H), 3.23-3.28 (m, 2H), 2.45-2.50 (m, 2H), 2.30-2.33 (m, 1H), 2.03-2.15 (m, 2H), 1.90-1.94 (m, 1H) 1.71-1.78 (m, 2H). ¹³C NMR (126 MHz, MeOD): δ 176.3, 174.6, 174.5, 162.6, 158.6, 137.4, 128.7, 114.4, 55.9, 53.2, 53.1, 41.9, 31.1, 30.1, 28.2, 26.2.

GluLys yield 47% white powder. LCMS ESI⁺: 436.16 [M+Na], expected mass for C₁₆H₂₃N₅NaO₈⁺ 436.14. ¹H NMR (500 MHz, MeOD): δ 7.78 (s, 1H), 4.66-4.69 (m, 1H), 4.61-4.63 (m, 1H), 2.97 (t, J=10.0 Hz, 2H), 2.46-2.49 (m, 2H), 2.30-2.34 (m, 1H), 2.09-2.14 (m, 1H), 2.00-2.04 (m, 1H), 1.91-1.95 (m, 1H), 1.71-1.74 (m, 1H). ¹³C NMR (126 MHz, MeOD): δ. 176.3, 174.8, 161.5, 137.4, 53.4, 53.0, 40.5, 32.3, 31.1, 28.2, 28.1, 23.6.

GluLeu yield 64% white powder. LCMS ESI⁺: 421.16 [M+Na], expected mass for C₁₆H₂₂N₄NaO₈⁺ 421.13. ¹H NMR (500 MHz, MeOD): δ. 7.77 (s, 1H), 4.67 (t, J=5.0 Hz, 1H), 4.63 (t, J=5.0 Hz, 1H), 2.42-2.53 (m, 2H), 2.30-2.35 (m, 1H), 2.08-2.15 (m, 1H), 0.99 (d, J=5.0 Hz, 3H), 0.97 (d, J=5.0 Hz, 3H). ¹³C NMR (126 MHz, MeOD): δ. 176.2, 175.8, 174.6, 137.3, 137.2, 53.1, 52.3, 42.0, 31.1, 28.3, 26.2, 23.4, 22.0.

GluThr yield 58% white powder. LCMS ESI⁺: 409.12 [M+Na], expected mass for C₁₄H₁₈N₄NaO₉⁺ 409.10. ¹H NMR (500 MHz, MeOD): δ 7.85 (s, 1H), 4.67-4.69 (m, 1H), 4.62 (d, J=5.0 Hz, 1H), 4.42-4.45 (m, 1H), 2.45-2.50 (m, 2H), 2.29-2.33 (m, 1H), 2.09-2.13 (m, 1H), 1.26 (d, J=5.0 Hz, 3H). ¹³C NMR (126 MHz, MeOD): δ176.2, 174.6, 173.6, 163.4, 161.7, 137.3, 132.5, 131.1, 68.6, 59.3, 53.1, 31.0, 28.2, 20.5.

GluEa yield 46% white powder. LCMS ESI⁺: 329.14 [M+H], expected mass for C₁₂H₁₇N₄O₇⁺ 329.11. ¹H NMR (500 MHz, MeOD): δ 7.77 (s, 1H), 4.66 (s, 1H), 3.72 (s, 2H), 3.53 (s, 2H), 2.44-2.52 (m, 2H), 2.26-2.31 (m, 1H), 2.11-2.16 (m, 1H). ¹³C NMR (126 MHz, MeOD): δ 174.8, 173.3, 135.7, 131.5, 129.7, 121.7, 60.2, 51.7, 41.5, 29.7, 26.8

To a dry round-bottom flask acyl chloride 1 (0.623 g, 2.0 mmol) was added followed by 10 mL of DCM and phenol (0.395 g, 4.2 mmol) under N₂. The suspension was cooled to 0 °C, and 320 μL of pyridine (4.0 mmol) was added dropwise. After 1 h, the solid was collected by vacuum filtration and washed with two 10 mL portions of DCM. The crude was purified by refluxing in DCM (30 mL), cooling to room temperature, and collecting the solid by filtration to yielded Diphenyl ester. Diphenyl ester (214 mg, 0.5 mmol) and 5 mL THF were added to a dry flask. Tyrosine-OtBu (237.0 mg, 1.0 mmol) and DIPEA were added to this suspension at 0 °C. After stirring at room temperature for 2 h, the reaction was refluxed for 2-4 days, monitored by TLC. The solvent was removed under vacuum and the tert-butyl protected intermediate was obtained by flash column chromatography. Then, this intermediate was dissolved in 5 mL TFA/DCM (1/1) for 2 h at room temperature. After all of the solvent was removed under vacuum, diTyr was purified by HPLC. Compounds diTrp, diHis, diAp, diEa and diPEG were synthesized using similar procedures (Scheme 2).

Scheme 2.

Synthesis of I45DCdiTyr.

diTyr yield 41% as white powder. LCMS ESI⁺: 505.15 [M+Na], expected mass for C₂₃H₂₂N₄NaO₈+ 505.13. ¹H NMR (500 MHz, MeOD): δ 7.80 (s, 1H), 7.06 (d, J=10.0Hz, 4H), 6.67 (d, J=5.0Hz, 4H), 4.28 (t, J=5.0Hz, 2H), 3.21-3.24 (m, 2H), 3.03-3.07 (m, 2H). ¹³C NMR (126 MHz, MeOD): 174.4, 162.1, 157.3, 137.2, 131.5, 128.7, 116.2, 55.5, 37.9.

diTrp yield 42% as white powder. LCMS ESI⁺: 529.20[M+H], expected mass for C₂₇H₂₅N₆O₆⁺ 529.18. ¹H NMR (500 MHz, MeOD): δ 8.12 (s, 1H), 7.07 (d, J=8.0 Hz, 2H), 7.12 (d, J=8.0 Hz, 2H), 6.96 (s, 2H), 6.86 (t, J=7.5 Hz, 2H), 6.75 (t, J=7.5 Hz, 2H), 4.76 (t, J=6.0 Hz, 2H), 3.30 (q, J=6.0 Hz, 2H), 3.17 (q, J=6.0 Hz, 2H); ¹³C NMR (126 MHz, MeOD): δ 173.28, 158.72, 136.56, 135.24, 128.77, 127.39, 123.53, 121.07, 118.46, 117.90, 110.95, 109.08, 53.93, 27.16.

diHis yield 30% as white powder. LCMS ESI⁺: 453.16 [M+H], expected mass for C₁₇H₁₈N₈NaO₆⁺ 453.12. ¹H NMR (500 MHz, D₂O): δ 8.53 (s, 2H), 7.87 (s, 1H), 7.25 (s, 2H), 4.83 (t, J=6.0 Hz, 2H), 3.34 (dq, J1=6.0 Hz, J2=15.0 Hz, 4H); ¹³C NMR (126 MHz, D₂O): δ 173.87, 162.34, 138.14, 134.29, 130.91, 129.44, 118.27, 53.80, 27.88.

diAp yield as white powder. LCMS ESI⁺: 303.20 [M+H], expected mass for C₁₁H₁₉N₄O₆⁺ 303.13. ¹H NMR (500 MHz, MeOD): δ 7.72 (s, 1H), 3.82 (t, J=5.0 Hz, 2H), 3.56-3.62 (m, 6H), 3.39-3.43 (dd, J₁=5.0 Hz, J₂=10.0 Hz, 2H). ¹³C NMR (126 MHz, MeOD): δ 161.9, 135.7, 130.7,70.6, 63.7, 41.9.

diEa yield as white powder. LCMS ESI⁺: 241.91 [M-H], expected mass for C₉H₁₃N₄O₄⁻ 241.09. ¹H NMR (500 MHz, MeOD): δ 7.71 (s, 1H), 3.71 (t, J=5.0 Hz, 4H), 3.52 (t, J=5.0 Hz, 4H). ¹³C NMR (126 MHz, MeOD): δ 160.7, 137.0, 134.7, 61.6, 43.0, 42.7.

diPEG yield 35% as clear oil. LCMS ESI⁺: 445.20 [M+H], expected mass for C₁₉H₃₃N₄O₈⁻ 447.24. ¹H NMR (500 MHz, MeOD): δ 7.63 (s, 1H), 3.65-3.68 (m, 14H), 3.55 (s, 4H), 3.38 (s, 6H). ¹³C NMR (75 MHz, CDCl₃): δ 163.6, 159.3, 135.23, 133.5, 128.3, 71.9, 70.6, 70.4, 69.7, 69.6, 59.0, 59.0, 39.5, 39.2.

To a stirring solution of I45DCA (312 mg, 2.0 mmol) in MeOH (20 mL) was added concentrated H₂SO₄ (2.0 mL). After refluxing for 6 h, the mixture was neutralized and extracted with ethyl acetate. Combined organic phase was dried over anhydrous Na₂SO₄, concentrated to give I45DCE. I45DCE was dissolved in MeOH (20 mL) and ethylenediamine (5 mL). The mixture was refluxed for 48 h. Then the solvent was evaporated and the residue purified by HPLC to get diEda (Scheme 3).

Scheme 3.

Synthesis of I45DC-diEda.

diEda yield 79%. LCMS ESI⁺: 263.15 [M+Na], expected mass for C₉H₁₆N₆NaO₂⁺ 263.12. ¹H NMR (500 MHz, MeOD): δ 7.71 (s, 1H), 3.47 (t, J=5.0 Hz, 4H), 2.85 (t, J=5.0 Hz, 4H). ¹³C NMR (126 MHz, MeOD): δ 163.4, 137.3, 132.2, 42.9, 42.0.

2.3. Phantom studies

25 mM of I45DCs were dissolved in 0.01 M phosphate-buffered saline (PBS) and titrated using high concentrations of NaOH/HCl to produce 0.15 ml solutions with pH values of 4.5, 5.0, 5.6, 6.0, 6.2, 6.5, 6.9, and 7.4. The samples were kept at 37°C during imaging. CEST data were acquired on 9.4 T and 11.7 T vertical bore Bruker Avance systems using a RARE imaging sequence (TR/TE = 10000/5.58 ms, RARE factor = 32, FOV = 17 × 17 mm², slice 1.5 mm, matrix size = 64 × 64, spatial resolution = 0.27 × 0.27 mm²) including a continuous-wave saturation pulse of 4 sec and six different saturation powers (B₁) of 2.4, 3.6, 4.8, 6.0, 8.4 and 10.2 μT. The CEST-weighted (M_z) images were collected at ninety-one frequency offsets between −13.5 ppm (−5400 Hz) and 13.5 ppm (5400 Hz) plus 1 at 20 000 Hz (M₀) to generate Z-spectra.

2.4. Calculation of exchange rates

The multiple saturation power I45DC Z-spectra for each pH were fit to numerical Bloch-McConnell solutions individually, using an open-source Matlab-based software (https://github.com/cest-sources/multiB1_fit_Z-cw_2pool)⁸. The Bloch-McConnell system consisted of either three or four pools, including ring NH protons (pool E, Δω ~ 7.7 ppm), amide, amine and hydroxyl protons (pool D, Δω ~ 2-4.5 ppm; pool B, Δω = 1.5 ppm) and water protons (pool A, Δω = 0 ppm). For the I45DC-diEda, an amide proton pool (pool E, Δω ~ 5.0 ppm) instead of the ring NH proton pool was simulated. The RARE readout was emulated by assuming the initial magnetization before the subsequent saturation phase to be small: M_i = 0.1×M₀. The longitudinal (R₁) and transverse (R₂) relaxation rates were fixed to: R_1A = 0.25 s⁻¹, R_1B = R_1D = R_1E = 1 s⁻¹, and R_2B = R_2D = R_2E = 66.66 s⁻¹, whereas the exchange rates (k_BA, k_DA, k_EA), relative concentrations (f_BA, f_DA, f_EA), and transverse relaxation of water, R_2A, were fit.

2.5. In vitro MR data post-processing and pH calculation

All post-processing was performed using in-house developed Matlab (version 2020b, MathWorks, USA) scripts. Briefly, the Z-spectra were interpolated on a voxel-by-voxel basis using smoothing splines, corrected for B₀ inhomogeneity, and saturation transfer (ST) value, ST = 1- M_z/M_0,, was measured. To generate a pH calibration curve, a region-of-interest (ROI)-based mean ST ratio = (1-ST_δ1)×ST_δ2/((1-ST_δ2)×ST_δ1) ³⁰ was plotted as a function of titrated pH and fit to a 3^rd degree polynomial with δ₁ = 2.0, 3.0, 3.5 or 4.5 ppm and δ₂ = 5.0 or 7.7 ppm corresponding to the frequency offsets of two CEST signal regions. We used a 3^rd degree polynomial function to fit our data because the fit quality as measured by R-squared was higher than when a linear function or a 2^nd degree polynomial function was used. Using this pH correlation function, the pH values for all pixels showing a ST% effect higher than 15% of the maximum CEST contrast detected at frequency offsets δ₁ or δ₂ were calculated. To provide a quantitative measure that reflects the sensitivity of the investigated compounds to pH, the slope of a line connecting the two extreme points of the calibration curve (at highest and lowest pH) were obtained as: slope = arctan (ΔpH/ΔST) and listed in Table 1. The range in detected pH values was determined by plotting the MRI pH measurements compared to that of electrode pH and selecting the linear region. Error bars were obtained by calculating the SD in pH over a single ROI enclosing the entire tube in the phantom at the respective pH.

Table 1.

CEST properties of I45DC contrast agents synthesized

Compound	Chemical shifts [δ1, δ2, ppm]	Detectable pH range	STratio slope (°)	KBA (s−1) at 7.7 ppm pH 6.5
Anionic Aliphatic
I45DC-diGlu	4.5, 7.7	5.6 – 7.0	53	5,081
I45DC-GluLys	3.0, 7.7	4.5 – 7.0	85	3,413
I45DC-GluArg	2.0, 7.7	4.5 – 6.5	79	3,449
I45DC-GluThr	4.5, 7.7	5.6 – 6.5	33	2,140
I45DC-GluEa	2.0, 7.7	5.6 – 6.5	70	3,950
I45DC-GluLeu	2.0, 7.7	5.6 – 6.5	78	6,767
Anionic Aromatic
I45DC-diTrp	3.5, 7.7	5.0 – 6.2	26	3,416
I45DC-diTyr	3.0, 7.7	5.0 – 6.5	74	2,628
I45DC-diHis *	-	-	-	-
I45DC-GluTrp	3.5, 7.7	5.6 – 6.5	32	2,499
I45DC-GluTyr	3.0, 7.7	5.6 – 6.5	62	3,539
Cationic Aliphatic
I45DC-diEda **	3.0, 5.0	4.5 – 7.4	60	381
Nonionic Aliphatic
I45DC-diAp	4.5, 7.7	5.6 – 7.4	65	9,442
I45DC-diEa	3.5, 7.7	6.2 – 7.4	30	2,745
I45DC-diPEG	3.5, 7.7	6.2 – 7.4	24	2,038

^*There is no peak at 7.7 ppm or 5 ppm observed.

^**Peak is at 5 ppm instead of 7.7 ppm. Conditions: CEST data were obtained at 25 mM concentration, t_sat = 4 sec and 37 ±C.

2.6. In vitro cellular cytotoxicity assay

The release of lactate dehydrogenase (LDH) from cells serves as an indicator of cell death. A commercial LDH toxicity assay kit (Abcam Cat. # ab65393, USA) was used according to the manufacturer's specification to determine the toxicity of the compounds in a variant of Human Embryonic Kidney (HEK293T) cell line. Briefly, released LDH in culture supernatants of HEK 293T cells after 24 h co-culture with different concentration of Iopamidol, diGlu and diEda was measured as the indicator of lysed cells. The percentage of cytotoxicity was measured by subtracting LDH content in remaining viable cells from total LDH in untreated controls. The final absorbance was measured at 450 nm. All experiments were performed in triplicate.

2.7. Unilateral Urinary Obstruction mouse model

C57BL/6 mice were anesthetized with ~3% isoflurane, and the incision site infiltrated with 25 mL 0.25% bupivicaine. The anesthetized animal was placed in ventral recumbency with tail towards surgeon and a midline incision was made and kidney and bladder dissected and identified. The left ureter was suture ligated utilizing a 6-0 silk suture.

2.8. In vivo MRI measurements

All experiments conducted with mice were performed in accordance with protocols approved by the Johns Hopkins University Institutional Animal Care and Use Committee (IACUC). Mice were anesthetized prior to infusion of 100 μl of I45DC-DiGlu at a concentration of 300 mM. The animals were then positioned in an 11.7 T horizontal bore Bruker Biospec scanner (Bruker Biosciences, Billerica, MA) and were under isoflurane anesthesia for the entire image collection period. High resolution T_2W images were acquired using a RARE sequence. To produce the CEST images, two sets of saturation images were collected, a WASSR set for B₀ mapping and a CEST data set for characterizing contrast. For the WASSR images, the saturation parameters were adjusted to stronger than normal to ensure water signal nulling in the presence of motion, t_sat = 2100 msec, B₁ = 1.2 μT, TE/TR = 3.55/6670 msec with saturation offset incremented from −1.5 to +1.5 ppm with respect to water in 0.1 ppm steps. For the CEST images, these were acquired using a RARE sequence with centric encoding (saturation length = 2100 msec at B₁ = 6 μT) and RARE = 32 on a Bruker 11.7 T horizontal scanner. Other parameters were: TR/TE = 11000/3.55 ms, Matrix size = 64×64 and axial slice thickness = 1.5 mm, FOV = 29 × 29 mm². A two offset CEST protocol was used with offsets of 7.7 ppm and 4.5 ppm repeatedly up to 1 h 16 min after I45DCs administration. We collected 8-10 pre-injection and 100 post-injection images at the two frequency offsets. For in vivo pH mapping, post-injection ST maps at 4.5 ppm and 7.7 ppm were calculated from 26 images acquired at the peak enhancement time. Averaged pre-injection ST maps were then substracted from the post-injection ST maps to eliminate the endogenous CEST signals. The pH maps were calculated using the ST ratio and a pH calibartion curve obtained from the in vitro experiments.

3. RESULTS AND DISCUSSION

In this study, we synthesized a series of 14 new I45DC compounds. We previously identified I45DC-diGlu as a extracellular pH (pHe) imaging agent, and as shown in Table 1, the Z-spectra of I45DC-diGlu are suited for measuring pH values between 5.6 and 6.9 with a ring NH k_BA ~ 5,081 s⁻¹ at pH 6.5, a typical pH for kidney tissue from CEST imaging measurements³⁰. We were interested in producing a series of pHe probes with reduced overall formal charge than diGlu which will in turn reduce the osmolarity of the contrast media, that still possessed two labile proton signals suitable for ratiometric measurement of pH values from at least 5.6 – 7.0 with suitable labile protons exchange rates (k_BA, k_CA) falling between 300 – 8,000 s⁻¹ based on our previous simulations for optimizing 3 T detection²³. We first prepared a series using amino acids with aliphatic sidechains, selecting the amino acid sidechains based on their potential for adjusting the intramolecular hydrogen bonding or slowing proton exchange and also to increase the formal charge of the compound³¹. We retained R₄ as glutamate and tested asymmetric substitutions with Fig. 1 showing their structures and Table 1 listing the CEST properties of these agents. The frequency offset and pH dependence of the second CEST signal varied among agents tested, and as a result, ratiometric CEST analysis provided slightly different pH detection ranges. Addition of weak electron acceptor groups led to narrower ranges of pH detection. We display the Z-spectra, pH calibration curves and pH maps for the best two in terms of ring NH k_BA (GluLys, GluArg,) in Fig. 2b,c. While the k_BA at pH 6.5 is better suited for CEST contrast than diGlu and their formal charge is lower as well, they displayed a very small change in ST_ratio over the range of physiological pH values with slopes of 85 ± and 79± respectively (Table 1) resulting in reduced quality pH maps (Fig. 2b,c far right panel). The guanidyl group and amino group can neutralize the charge of GluLys and GluArg, which also weakens the intramolecular hydrogen bonding of the ring NH through intermolecular hydrogen bonding and influencing the contrast efficiency. GluLeu and GluEa showed similar properties to GluArg so these are not displayed in Fig 2.

Figure 1.

Structure of I45DC scaffold and imaging agents synthesized and tested in this study.

Figure 2.

in vitro CEST Z-spectra, ST_ratio vs pH and pH maps for anionic I45DC agents; A-C) I45DC-diGlu; D-F) I45DC-GluArg; G-I) I45DC-GluLys; J-L) I45DC-diTrp; Conditions: CEST data were obtained at 25 mM concentration, t_sat = 4 sec and 37 ±C. Please note that some pH maps show values that are far from the titrated pH values which is due to the limited pH sensitivity range of the compounds.

In order to test whether bulkier sidechains might increase the protection factor of the ring NH and thus reduce k_BA further than the aliphatic variants, we prepared a series of anionic I45DC pH sensors with aromatic sidechain amino acids to determine how steric hindrance would influence CEST contrast. Of these, diHis did not possess ring NH protons suitable for detection. The unshared electron pair of the ring NH in imidazole easily leaves in the form of hydrogen ions, which can interact with the carbonyl group on R₅ position, leading to the quench of CEST signal. However, the CEST signal of diTrp is not influenced by the indole ring NHs. And the best (diTrp) is shown in Fig. 2d. While the formal charge is reduced, the k_BA is suitable at pH 6.5 and slope of ST_ratio vs pH suitable as well, the detectable pH range for this sensor is limited to pH values between 5.0 and 6.2 which also renders GluTrp, diTyr and GluTyr as reduced performance pH sensors. GluThr also performed similarly and as a result is not shown in Fig. 2. In addition, diTrp will aggregate in water below pH 6, which may be caused by the π -π stacking effect.

Based on these observations, we were interested in attaching basic and neutral substituents to this scaffold to determine if cationic or nonionic I45DC pH sensors would alter the detectable pH range of the I45DC sensors. Specifically, to get nonionic I45DCs, triethylene glycol monomethyl ether (PEG), 3-amino-1,2-propanediol (Ap) and ethanolamine (Ea) were conjugated and for cationic I45DC ethylenediamine (Eda) was conjugated (Figs. 1 & 2, Table 1). All these materials are very inexpensive and usually used as fragments of drugs. While nonionic diEa and diPEG displayed contrast at 7.7 ppm with a k_BA closer to the optimal ~ 1,000 s⁻¹ which should enhance their 3 T contrast, the detectable pH range was significantly reduced. The nonionic diAp also displayed contrast at 7.7 ppm (Fig 3b), although k_BA was too fast (> 9,000 s⁻¹ ) to be suitable for detection on 3 T clinical scanners. Based on these observations, we hypothesize that the lack of titratable groups on the nonionic derivatives resulted in the smaller range in detectable pH values. The Z-spectrum of cationic I45DC-diEda showed two distinct and relatively large peaks at 3.0 and 5.0 ppm instead, which originate from free amine proton groups on the side chains and ring NH on imidazole, respectively. In addition, diEda had the best k_BA (381 s⁻¹) for detection at clinical field strengths and largest detectable pH range (4,5 – 7.4, Fig 3a) as well. We postulate this smaller chemical shift of 5.0 ppm is due to weakening of the intramolecular hydrogen bonding between the R₅ carbonyl group and ring NH on imidazole from the terminal cationic amino groups, leading to a smaller chemical shift from bulk water. Based on preferring agents with large detectable pH ranges while restricting kBA, kCA to be less than 8,000 s⁻¹, we performed HEK293 cell toxicity studies on diGlu and diEda (Fig. 4). As can be seen, I45DC-diEda displays a higher toxicity at 5 mM, 10 mM, 20 mM and 40 mM concentrations compared to I45DC-diGlu and iopamidol which are well tolerated by these cells using these concentrations. Based on this, we selected I45DC-diGlu for further live animal testing.

Figure 3.

in vitro CEST Z-spectra, ST_ratio vs pH and pH maps for cationic and neutral I45DC agents; A) I45DC-diEda; B) I45DC-diAp; C) I45DC-diEa; D) I45DC-diPEG; Conditions: CEST data were obtained at 25 mM concentration, t_sat = 4 sec and 37 ±C. Please note that some pH maps show values that are far from the titrated pH values which is due to the limited pH sensitivity range of the compounds.

Figure 4.

Cell Cytotoxicity of the three CEST pH imaging agents Iopamidol, I45DC-diGlu, I45DC-diEda. This was measured by incubating HEK293T cells with these agents.

We then proceeded to test the performance of our I45DC-diGlu analog for CEST imaging in a complete Unilateral Ureter Obstruction (UUO) mouse model of UTOs. In this model, markedly lower renal uptake of ^99mTc-MAG3 was observed 1 day after suture ligation³² and histology showed tubular atrophy with interstitial fibrosis after 3 days³³. We administered this compound and acquired CEST MRI data on a single slice containing the center of both kidneys using two offsets, 4.5 and 7.7 ppm to characterize how the contrast varied in time after injection. As shown in Fig. 5, the CEST contrast in the unobstructed kidney peaked at a time of around 180 – 300 sec whereas for the UUO kidney this contrast persisted out to 17 min, which is similar to what has been seen in a renal scintigraphy study on a mouse model of UUO by Tantawy et al³⁷. In addition, the contrast is more uniform with larger ST values in the unobstructed kidney as seen in Fig 5A. We also produced pHe maps based on ratiometric contrast (Fig. 5B,D) and display histograms of pHe values across all pixels in each kidney (Fig. 5C,E) for n= 2 mice. These showed differences between the kidneys with the unobstructed kidney having relatively homogeneous pH values around 6.86±0.01 (Fig. 5B,C), while UUO kidneys showed an elevated pH of 6.97±0.01 and increased range in pHe values and the data being consistent between these mice. Based on our data, we concluded that I45DC-diGlu can depict differences in perfusion and pHe between unobstructed and UUO kidneys.

Figure 5.

In vivo CEST MRI of UUO mice for I45DC-diGlu; (A) ST maps at 7.7 ppm of a representative mouse calculated by averaging images acquired in different time ranges after administration of I45DC-diGlu overlayed on high resolution T_2W anatomic images; pHe maps (B, D) and (C, E) pHe histograms of two UUO mice following I45DC-diGlu injection.

In the present study, we have synthesized 15 I45DC compounds with 14 new and 13 of these possessing far shifted labile protons resonating at 7.7 – 7.8 ppm from water and 1 (I45DC-diEda) with a labile proton 5 ppm from water. Inspiration for the compounds studied here was derived from the catalytic site of serine proteases. Such proteases have a well defined hydrogen bonding network, which slows the exchange of the imidazole N-H groups in histidine sufficiently to allow detection through CEST contrast, with the exchangeable proton demonstrating chemical shifts as far as 13 ppm downfield from water^34,35. We were interested in maximizing the performance of the I45DC scaffold through testing a variety of substituents at the R₄ and R₅ positions and also testing whether symmetric and asymmetric substitutions would optimize the CEST properties for concentration independent measurement of pHe. The 14 I45DCs displaying strong contrast included anionic, cationic and nonionic compounds which show the versatility of this CEST imaging scaffold. Although, there is intramolecular hydrogen bonding for compounds of the I45DC class as described previously, the CEST contrast is pH and concentration dependent similar to CEST agents that don’t display intramolecular bonding. Furthermore, these 14 compounds have at least two CEST active protons with different pH dependencies which enables generation of pHe maps using the ratio of these two signals, although the inflection points for signal ratio vs pHe is not ideal for some to measure pHe over the entire physiologically relevant range. Compared to the triiodobenzene CEST pHe imaging agents including iopamidol and iopromide which have a significant advantage of FDA approval, the main labile protons are an additional 2 ppm further shifted from water, which should allow better discrimination from background tissue. In addition, the exchange rates are faster in general and therefore larger saturation B₁’s should be applied to detect these I45DC CEST agents. In summary, our phantom CEST experiments showed that a wide range of pHe values could be measured using these I45DC probes.

Furthermore, we demonstrate successful detection of our frontline compound (I45DC-diGlu) in a mouse model of UTO with clear differences in contrast kinetics and pHe maps seen using this agent. Specifically, we observed an elevated pHe for the UUO kidney. Our findings can be contextualized in the observations of previous studies, which demonstrated increased pHe values after glycerol-induced injury³⁶, after unilateral ischemia-reperfusion injury³⁷ and also after UUO when using iopamidol as a pH sensor ³⁸. Impaired urinary acidification has also been documented in patients with various types of obstructive nephropathy³⁹. One issue with using iodinated contrast agents is that contrast induced nephropathy is a known risk factor. Our I45DCs are similar to flumizole, dacarbazine and cimetidine and a number of other drugs which have LD50s of 500 −2,500 mg/kg and are used as antiinflammatories, anticancer, H₂ antagonists and others^40,41. Based on this similarity and our cell toxicity results, we expect that I45DC-diGlu and other I45DCs should be well tolerated.

A limitation of using CEST agents for ratiometric based pH imaging is the need to have sufficient signal intensity for both labile protons across all pH values to allow for precise pH measurements. By analyzing the CEST efficiency of our new I45DCs CEST agents, we suspect that intermolecular hydrogen bonding, intramolecular hydrogen bonding, molecular charge and π-π stacking effects are the main factors that influence the signal intensity as a function of pH. While we have not identified an ideal agent across all pH values of physiological interest, I45DC-diGlu can measure a fairly wide range of pH values and the other I45DCs agents synthesized and tested may provide ideas on designing more efficient diaCEST contrast agents in the future.

4. CONCLUSIONS

We have synthesized 14 imidazoles which can be used as CEST imaging agents for depicting perfusion and pH changes. I45DC-diGlu, which to date displayed the best exchange properties for a compound with large labile proton chemical shift (7.7 ppm), was able to depict differences in contrast kinetics and pH values between obstructed and unobstructed kidneys in a mouse model of UTO. Overall, our results suggest that I45DC compounds are very promising CEST probes, particularly for measuring pH.

Abbreviations

MRI

magnetic resonance imaging

CEST

chemical exchange saturation transfer

I45DC

imidazole-4,5-dicarboxyamide

Eda

ethylenediamine

Ap

3-amino-1,2-propanediol

Ea

ethanolamine

UTO

urinary tract obstruction

UUO

unilateral ureter obstruction

SD

standard deviation

DMF

N,N-Dimethylformamide

TFA

trifluoroacetic acid

DCM

dichloromethane

THF

tetrahydrofuran

DIPEA

diisopropylethylamine

ESI

electrospray ionization

MeOD

deuterated methanol

Δω

chemical shift difference with water in ppm

k_BA

chemical exchange rate from pool B to pool A

ST

saturation transfer

CKD

chronic kidney disease

AKI

acute kidney injury