Synthetic approach for unsaturated precursors for parahydrogen induced polarization of choline and its analogs

Abstract

Reported here are (i) a new synthetic approach for preparation of (ii) a new compound class, of –OH, for example, an –OH group is replaced with acetyl protecting group, protected 1,2-dehydrocholine analogs and (iii) a new synthetic route for betaine aldehyde. The C=C bond of 1,2-dehydrocholine moiety can be used for molecular addition of parahydrogen producing –OH protected hyperpolarized choline by parahydrogen-induced polarization (PHIP). The reported synthetic approach allows for incorporation of ¹⁵N and deuterium labels, which are necessary for preparation of highly polarized PHIP contrast agents. Isotope labeling with ¹⁵N and/or deuterium was conducted. Hyperpolarized ¹⁵N-choline enabled by the reported synthetic approach can be potentially used as an imaging biomarker of cancer similar to choline positron emission tomography tracers.

Introduction

Nuclear magnetic hyperpolarization significantly increases nuclear spin alignment by several orders of magnitude.¹ This allows for a major NMR sensitivity enhancement enabling detection of dilute metabolites in humans.² As a result, molecular MR imaging of diseases, such as cancer with impaired metabolism, can report on disease grading,³ progression, response to treatment,⁴ and so on. Hyperpolarized imaging is similar to positron emission tomography (PET) imaging, because both imaging modalities use exogenous contrast agents that enable molecular imaging. In contrast to PET, hyperpolarized imaging can provide more metabolic information as it can distinguish between injected contrast agent and its in vivo etabolites by using molecular signatures discerned by MR. The main shortcoming of hyperpolarized contrast agents is that their lifetime is limited by spin lattice relaxation time T₁. As a result, ¹³C-enriched carboxyl sites of key biomolecules are exploited, because of their relatively long T₁ with respect to that of protons or ¹³C sites with directly attached hydrogens. Hyperpolarized 1-¹³C-pyruvate is the leading ¹³C molecular contrast agent tested in phase I clinical trials for imaging of prostate cancer in men.^2,5 It reports on elevated rate of glycolysis⁶ and has in vivo T₁ on the order of tens of seconds, which is sufficient for tracer injection and in vivo conversion to 1-¹³C-lactate.

Hyperpolarized ¹⁵N-choline⁷ is an attractive contrast agent for molecular imaging of cancer proliferation primarily for the following reasons: (i) known rapid uptake by tumors,⁸ (ii) long >2 min in vivo ¹⁵N T₁,⁹ and most importantly, (iii) its known efficacy for imaging and staging of cancer, as used in the form of PET imaging agent C11-choline or F18-choline.^10,11 Furthermore, over-expression of choline kinase and increase of choline kinase activity leading to elevated choline uptake and elevated levels of phosphocholine in tumors have been linked to tumor progression, invasiveness, drug resistance and hypoxia.^12,13

Although 1-¹³C-pyruvate can be efficiently polarized by dynamic nuclear polarization (DNP) to ¹³C nuclear spin polarization P ~ 30% in approximately 1 h, DNP polarization of ¹⁵N-choline is significantly less successful with P ~ 5% and polarization time of ~3 h. A faster and more efficient approach would clearly be useful for advancement of hyperpolarized ¹⁵N-choline as an imaging molecular probe in preclinical models of human cancer and in potential clinical trials.

Parahydrogen-induced polarization (PHIP)¹⁴ is a chemical hyperpolarized method that takes advantage of a singlet state of parahydrogen. The spin order of the parahydrogen molecule can be preserved during its molecular addition to an unsaturated molecular precursor. Once the symmetry of the singlet state is broken, the spin order can be converted to a large net magnetization¹⁵ of ¹³C site^16,17 via the mechanism of spin–spin couplings with polarization P approaching ~100%. In practice, several ¹³C hyperpolarized compounds with P > 10% were reported using PHIP, and parahydrogen and synthesis allow dramatically enhanced nuclear alignment phenomenon¹⁸ including 1-¹³C-succinate¹⁹ shown to be useful for interrogating cancer metabolism.^20,21 PHIP is an ultrafast polarization method capable of polarizing contrast agents within seconds to polarization of 20% and more.²²

The PHIP requires an unsaturated carbon–carbon bond for molecular addition of parahydrogen. Therefore, a typical molecular frame is C=C–S, Figure 1A, where S can be ¹³C, ¹⁵N or any other spin ½ nucleus preferably without any directly attached protons^23,24 allowing for long in vivo T₁ of more than 40 s. Unfortunately, PHIP polarization of ¹⁵N compounds was limited only to ¹⁵N trimethylethyl ammonium. Although ultralong in vitro ¹⁵N T₁ = 400 s was reported,²⁵ this agent has no metabolic relevance, and it is in fact highly neurotoxic.

Figure 1.

The molecular diagrams of parahydrogen induced polarization (PHIP). (A) General diagram of spin ½S nucleus (S = ¹³C, ¹⁵N, etc.) polarization using spin order of parahydrogen (para-H₂) added to an unsaturated carbon–carbon bond via molecular mechanism. (B) Diagram showing collapse of C=C bond in unstable hypothetical precursor 1,2-dehydrocholine leading to formation of betaine aldehyde that is not suitable for PHIP hyperpolarization. (C) The diagram of molecular hydrogenation of OH protected 1,2-dehydrocholine that can be potentially used for PHIP hyperpolarized leading to hyperpolarized protected ¹⁵N-choline.

The requirement for an unsaturated carbon–carbon bond severely limits the number of biomolecules that can be directly polarized by PHIP. For, example, 1-¹³C-pyruvate and ¹⁵N-choline are not amenable to PHIP, because the molecular framework is either not feasible or unstable as in the case of unsaturated molecular precursor for ¹⁵N-choline. The corresponding enol would simply collapse to betaine aldehyde, Figure 1B. A potential solution to circumvent this challenge is the use of –OH protected, for example, an –OH group is replaced with acetyl protecting group, unsaturated precursors. This approach was recently demonstrated for PHIP of ethanol²⁶ and 1-¹³C-phospholactate²⁷ by using acetate and phosphate moieties as protecting groups, respectively. The latter showed nearly complete molecular hydrogenation using water soluble Rh-based²⁸ catalyst and sufficiently long ¹³C T₁ suitable for in vivo imaging. Unlike 1-¹³C-phosphoenolpyruvate, which is a PHIP precursor for hyperpolarized 1-¹³C-phospholactate²⁷ and is commercially available, unsaturated PHIP precursors for protected ¹⁵N-choline have not been reported to the best of our knowledge.

Here, a new synthetic approach for a new class of compounds, unsaturated protected cholines, is described and exemplified by acetyl 1,2-dehydrocholine. This synthetic scheme allows for incorporation of ¹⁵N and deuterium labels. Although the requirement for spin ¹⁵N labeling is straightforward as it allows for sensitivity increase by >200 fold compared with the natural abundance level of ¹⁵N, complete or partial deuteration of the precursor is necessary to simplify the PHIP spin system to three spins—¹⁵N and two nascent protons from parahydrogen—to maximize the quantum yield of the polarization transfer sequence.¹⁶ The latter should lead to high PHIP polarization in future studies, which are outside the scope of this work.

Experimental section

General

All solvents were purchased from common vendors and were used as received. All new compounds were characterized by ¹H and ¹³C NMR and HR-MS. High-resolution mass spectra were recorded using a Synapt hybrid quadrupole/oa-TOF Mass Spetrometer (Waters Corp., Milford, MA) equipped with a dual chemical ionization/electrospray (ESCI) source. A post-acquisition gain correction was applied using a solution of 3-[(3-Cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS, FW 614.88, Sigma-Aldrich 3050 Spruce St. St. Louis, MO 63103, USA) as the lock mass spray.

Synthesis of new compounds

2,2-Dihydroxy-N,N,N-trimethylethanaminium bromide or betaine aldehyde bromide (2a)

Allyltrimethylammonium bromide²⁹ 1a (3 g, 16.7 mmol) was dissolved in methanol (300 mL) and water (2 mL). A small amount of Rose bengal was added, and the solution was cooled to −78 °C. Ozone was passed through the solution until the pink color of Rose bengal almost completely disappeared, and solution was flushed with air for 20 min. Dimethyl sulfide (30 mL) was added, and the reaction mixture was left at room temperature overnight. The solvent was removed under reduced pressure, and the resulting residue was washed with anhydrous acetone (3 × 50 mL). The solid residue was dried in vacuo for an additional 20 min. Then it was precipitated from anhydrous MeOH by acetone with 5:95 volume ratio. Crystals were washed with anhydrous ether, yielding betaine aldehyde bromide 2a with 62% yield (2.07 g, 10.3 mmol).

2-Acetoxy-N,N,N-trimethylethenaminium bromide (3a)

Betaine aldehyde bromide 2a (0.20 g, 1.0 mmol) dried overnight in lyophilizer was suspended in 20 mL of acetic anhydride. After 12 h, reaction mixture was filtered, and the filtrate was evaporated to dryness. Mixture of E- and Z-acetyl 1,2-dehydrocholine bromide was obtained as a major product with 20% yield (0.045 g), E : Z ratio of 1:0.46. The structure of the major isomer was elucidated using ¹H, ¹³C, COSY, HMBC, select NOE high-resolution NMR spectra. E-acetyl 1,2-dehydrocholine bromide: ¹H (DMSO-d₆, 400 MHz) δ: 7.93 (d, 1H, J(HH) = 11.6 Hz), 6.87 (d, 1H, J(HH) = 11.6 Hz), 3.34 (s, 9H), 2.23 (s, 3H); ¹³C{¹H} (DMSO-d₆, 100 MHz) δ: 167.2, 136.3, 127.3, 55.2, 20.2. Z-acetyl 1,2-dehydrocholine bromide: ¹H (DMSO-d₆, 400 MHz) δ: 7.30 (d, 1H, J(HH)= 5.6Hz), 5.92 (d, 1H, J(HH) = 5.6 Hz), 3.40 (s, 9H), 2.30 (s, 3H); ¹H (D₂O, 400 MHz) δ: 7.26 (d, 1H, J(HH) = 5.2 Hz), 5.76 (d, 1H, J(HH) = 5.6 Hz), 3.37 (s, 9H), 2.23 (s, 3H).

Allyltrimethylammonium chloride (1b)

Trimethylamine hydrochloride (5 g, 52 mmol) was dissolved in dry ethanol (200 mL). The solution was cooled to 0 °C, and ice cold 1 M NaOH ethanol solution (52 mL, 52 mmol) was added slowly. After 10 min, allyl bromide (4.5 mL, 52 mmol) was added drop-wise. The reaction mixture was left overnight at room temperature and then concentrated at reduced pressure by using a rotary evaporator. The resulting solid was re-suspended in 300 mL of dry ethanol. The solution was filtered to remove sodium chloride/bromide salt. The resulting filtrate was passed through an ion exchange column (~400 mL of IRA-400 Cl beads, total volume capacity 1.6 eq/L). The column was washed with excess ethanol, and the resulting solution was concentrated in vacuo. The solid residue was precipitated from ethanol (100 mL) by addition of diethyl ether (700 mL) resulting in allyltrimethylammonium chloride 1b (5 g) with 71% yield.³⁰

Betaine aldehyde chloride (2b)

Allyltrimethylammonium chloride 1b (2 g, 15 mmol) was dissolved in methanol (300 mL) and water (2 mL) solution. A small amount of Rose bengal was added, and the solution was cooled to −78 °C. Ozone was passed through the solution until the pink color of Rose bengal completely disappeared. Dimethyl sulfide (30 mL) was added, and the reaction mixture was left at room temperature overnight. The solvent was removed at reduced pressure, and the resulting residue was washed with anhydrous acetone (3 × 50 ml). The solid residue was dried in vacuo for an additional 20 min. Then it was precipitated from 60 mL of anhydrous methanol followed by the addition of anhydrous diethyl ether (750 mL). Betaine aldehyde chloride³¹ 2b (1.77 g) was produced with 76% yield as a white, extremely hygroscopic solid. ¹H (DMSO-d₆, 400 MHz) δ: 6.73 (d, 2H, J(HH) = 6.1 Hz), 5.26 (p, 1H, J(HH) = 5.8 Hz), 3.26 (d, 2H, J(HH) = 5.4 Hz) 3.13 (s, 9H); ¹³C (DMSO-d₆, 100 MHz) δ: 84.9, 68.5, 53.4. Note: upon dehydration ¹H NMR peak at 6.73 ppm disappears and ¹H NMR peak at 5.25 ppm collapses into a triplet. ¹H (D₂O, 400 MHz) δ: 5.46 (t, 1H, J(HH) = 5.5 Hz), 3.33 (d, 2H, J(HH) = 5.5 Hz), 3.13 (s, 9H); ¹³C{¹H} (D₂O, 100 MHz) δ: 84.9 (d, J = 3.9 Hz), 68.4 (t, J = 3.0 Hz), 53.4 (dd, J₁ = 3.0 Hz, J₂ = 3.9 Hz). Note: upon prolonged storage in D₂O for more than 4 h, ¹H NMR peak at 3.33 ppm and ¹³C NMR peak at 68.4 ppm disappear because of H/D exchange.

2-Acetoxy-N,N,N-trimethylethenaminium chloride (3b)

Betaine aldehyde chloride 2b (1.57 g, 10.1 mmol) was suspended in 400-mL acetic anhydride. After 24 h, reacted solution was evaporated to dryness. The residue was submerged in additional 200-mL acetic anhydride and evaporated to dryness. After 4 h in vacuo, the residue was partially re-dissolved in 250-mL acetic anhydride and filtrated. Filtrate was precipitated by addition of dry diethyl ether (1500 mL) producing a mixture of E- and Z-acetyl 1,2-dehydrocholine chloride 3b and diacetylbetaine aldehyde 4b (1.2 g) molar ratio of Z- : E- : diac of 0.25:1.00:0.74. Note: E-/Z- ratio varies from batch to batch and depends on storage time. E-acetyl 1,2-dehydrocholine chloride ¹H (DMSO-d₆, 600 MHz) δ: 7.91 (d, 1H, J(HH) = 11.6 Hz), 6.97 (d, 1H, J(HH) = 11.6Hz), 3.37 (s, 9H), 2.22 (s, 3H); ¹³C{¹H} (DMSO-d₆, 125 MHz) δ: 167.2, 136.2, 127.4, 55.1, 20.3; ¹H (D₂O, 500 MHz) δ: 7.90 (d, 1H, J (HH) = 11.5Hz), 6.58 (d, 1H, J(HH) = 11.5Hz), 3.31 (s, 9H), 2.15 (s, 3H); ¹³C{¹H} (D₂O, 125 MHz) δ: 169.5, 136.8, 127.1, 55.8, 19.9. Z-acetyl 1,2-dehydrocholine chloride ¹H (DMSO-d₆, 600 MHz) δ: 7.28 (d, 1H, J(HH) = 5.6 Hz), 5.95 (d, 1H, J (HH) = 5.6 Hz), 3.42 (s, 9H), 2.30 (s, 3H); ¹H (D₂O, 500 MHz) δ: 7.27 (d, 1H, J (HH) = 5.5 Hz), 5.77 (d, 1H, J(HH) = 5.5Hz), 3.38 (s, 9H), 2.24 (s, 3H); ¹³C{¹H} (D₂O, 125 MHz) δ: 167.3, 132.2, 119.0, 56.3, 19.8. Diacetylbetaine aldehyde ¹H (DMSO-d₆, 600 MHz) δ: 6.99 (t, 1H, J(HH) = 4.8 Hz), 3.88 (d, 2H, J(HH)= 4.9 Hz), 3.22 (s, 9H), 2.10 (s, 6H); ¹³C{¹H} (DMSO-d₆, 125 MHz) δ: 168.1, 84.5, 64.2, 53.5, 20.5; ¹H (D₂O, 500 MHz) δ: 7.08 (t, 1H, J(HH)= 4.4 Hz), 3.78 (d, 2H, J (HH) = 4.4 Hz), 3.21 (s, 9H), 2.09 (s, 6H); ¹³C{¹H} (D₂O, 125 MHz) δ: 170.6, 84.5, 65.1, 54.6, 20.0.

¹⁵N-Allyltrimethylammonium bromide (1c)

It was prepared from Me₃¹⁵N•HCl, 98% ¹⁵N, Sigma-Aldrich-Isotec, Miamisburg, OH, #608831, (0.96 g, 10 mmol) in manner similar to that of 1b, but without the anion exchange. Yield = 1.34 g (74%). ¹H (CD₃OD, 400 MHz) δ: 6.10 (m, 1H), 5.74 (m, 1H), 5.71 (d of m, 1H), 4.00 (d, 2H, J (HH) = 7.5 Hz), 3.13 (d, 9H, J(H-¹⁵N) = 0.7 Hz); ¹³C{¹H} (CD₃OD, 100 MHz) δ: 129.5 (d, J(¹³C-¹⁵N) = 1.6 Hz), 126.5, 69.5 (d, J(¹³C-¹⁵N) = 4.3 Hz), 53.2 (d, J (¹³C-¹⁵N) = 5.8 Hz); ¹⁵N{¹H} (CD₃OD, 40 MHz, auto calibrated) δ: 48.87. HR-MS calculated for C₆H₁₄¹⁵N: 101.1091; found: 101.1092 (0.99 ppm).

¹⁵N-2-Acetoxy-N,N,N-trimethylethenaminium bromide (3c)

¹⁵N-betaine aldehyde bromide was prepared in a manner similar to that of 2a from compound 1c (1.30 g). No purification was performed, and it was used directly in the next step. Acetyl ¹⁵N-1,2-dehydrocholine bromide 3c was prepared and purified in a manner similar to that of 3b. The reaction produced acetyl ¹⁵N-1,2,-dehydrocholine bromide 3c and diacetyl-¹⁵N-betaine aldehyde bromide 4c (1.09 g) with molar ratio of Z- : E : diac of 0.3:1.0:0.9. E-acetyl ¹⁵N-1,2-dehydrocholine bromide ¹H (DMSO-d₆, 400 MHz) δ: 7.92 (dd 1H, J(H-¹⁵N) = 1.5 Hz and J(HH) = 11.6 Hz), 6.89 (dd, 1H, J(H-¹⁵N) = 3.2 Hz and J(HH) = 11.6 Hz), 3.35 (broad s, 9H), 2.23 (s, 3H); ¹³C{¹H} (DMSO-d₆, 100 MHz) δ: 167.2, 136.3 (d, J(¹³C-¹⁵N) = 2.4 Hz), 127.3 (d, J(¹³C-¹⁵N)= 9.6Hz), 55.1 (d, J(¹³C-¹⁵N)= 5.1Hz), 20.2; ¹⁵N (DMSO-d₆, 40 MHz, auto calibrated) δ: 51.3; ¹H (D₂O, 400 MHz) δ: 7.87 (dd, 1H, J(H-¹⁵N)= 1.5Hz and J(HH)= 11.6Hz), 6.54 (dd, 1H, J(H-¹⁵N)= 3.4 and J(HH) = 11.6Hz), 3.28 (d, 9H, J(H-¹⁵N) = 0.6Hz), 2.11 (s, 3H); ¹⁵N{¹H} (D₂O, 40 MHz, auto calibrated) δ: 50.1. Z-acetyl ¹⁵N-1,2-dehydrocholine bromide ¹H (DMSO-d₆, 400 MHz) δ: 7.29 (t, 1H, J(HH) = 5.5Hz), 5.92 (dd, 1H, J(H-¹⁵N)= 3.6 Hz and J (HH)= 5.5Hz), 3.41 (d, 9H, J(H-¹⁵N)= 0.6Hz), 2.30 (s, 3H); ¹³C{¹H} (DMSO-d₆, 100 MHz) δ: 165.8, 131.6, 119.9 (d, J(¹³C-¹⁵N)= 7.9Hz), 55.7 (d, J(¹³C-¹⁵N)= 5.0Hz), 20.5; ¹⁵N (DMSO-d₆, 40 MHz, auto calibrated) δ: 52.0; ¹H (D₂O, 400 MHz) δ: 7.24 (t, 1H, J(HH) = 5.6Hz), 5.73 (dd, 1H, J(H-¹⁵N)= 3.7Hz and J (HH)= 5.6Hz), 3.34 (d, 9H, J(H-¹⁵N)= 0.6 Hz), 2.21 (s, 3H); ¹⁵N{¹H} (D₂O, 40 MHz, auto calibrated) δ: 50.4. ¹⁵N-diacetylbetaine aldehyde bromide ¹H (DMSO-d₆, 400 MHz) δ: 6.99 (dt, 1H, J(H-¹⁵N) = 1.2Hz and J(HH) = 4.8Hz), 3.84 (d, 2H, J(HH) = 4.8Hz), 3.21 (s, 9H), 2.10 (s, 6H); ¹³C{¹H} (DMSO-d₆, 100 MHz) δ: 168.1, 84.5, 64.3 (d, J(¹³C-¹⁵N) = 5.2 Hz), 53.6 (d, J(¹³C-¹⁵N)= 5.1 Hz), 20.5; ¹⁵N {¹H} (DMSO-d₆, 40 MHz, auto calibrated) δ: 48.1; ¹H (D₂O, 400 MHz) δ: 7.05 (dt, 1H, J(H-¹⁵N)= 1.7 Hz and J(HH) = 4.5 Hz), 3.75 (d, 2H, J (HH)= 4.5Hz), 3.17 (s, 9H), 2.06 (s, 6H); ¹⁵N{¹H} (D₂O, 40 MHz, auto calibrated) δ: 47.1.

¹⁵N-Allyltrimethylammonium-d₁₂ bromide (1d)

It was prepared from (CD₃)₃¹⁵N•HCl, 98% ¹⁵N, 99% D, Sigma-Aldrich-Isotec, #591815, (2.00 g, 19 mmol) and allylbromide-d₅ (2.51 g, 20 mmol) in a manner similar to that of 1b except eliminating the anion-exchange step. Yield = 84% (3.12 g). ¹³C{¹H} (CD₃OD, 100 MHz) δ: 128.9 (m), 125.9 (t), 68.3 (m), 52.2 (m); ¹⁵N{¹H} (CD₃OD, 40 MHz, auto calibrated) δ: 47.4. HR-MS calculated for C₆¹⁵ND₁₄: 115.1970; found: 115.1971 (0.87 ppm).

¹⁵N-2-Acetoxy-N,N,N-trimethylethenaminium-d₁₁ bromide (3d)

¹⁵N-betaine aldehyde-d₁₄ bromide was prepared in a manner similar to that of 2c starting from compound 1d (0.95 g, 4.7 mmol). Methanol-d₁ and acetone-d₆ were used instead of regular methanol and acetone. No purification was performed and ¹⁵N-betaine aldehyde-d₁₄ bromide was used directly in the next step. Acetyl ¹⁵N-1,2-dehydrocholine-d₁₁ bromide 3d was prepared from crude ¹⁵N-betaine aldehyde-d₁₄ bromide in a manner similar to that of 3c. No recrystallization was attempted. The reaction produced acetyl ¹⁵N-1,2-dehydrocholine-d₁₁ bromide 3d and diacetyl-¹⁵N-betaine aldehyde-d₁₁ bromide 4d. Yield over two steps = 0.99 g with molar ratio of Z- : E-: diac was 0.9:1.0:1.2. E-acetyl ¹⁵N-1,2-dehydrocholine-d₁₁ bromide ¹H (D₂O, 400 MHz) δ: 6.55 (m, 0.62H), 2.11 (s, 3H). Z-acetyl ¹⁵N-1,2-dehydrocholine-d₁₁ bromide ¹H (D₂O, 400 MHz) δ: 5.75 (dd, 0.79H), 2.21 (s, 3H). Diacetyl-¹⁵N-betaine aldehyde-d₁₁ bromide ¹H (D₂O, 400 MHz) δ: 2.09 (s, 6H).

Results and discussion

Despite scarce reports^32,33 of ketone-derived ylides, neither unsaturated choline analog nor any closely related compounds have been reported. Therefore, the synthetic approach to protected ¹⁵N-choline PHIP precursor was divided into two parts: (i) development of synthetic methodology without enrichment by stable isotopes and (ii) synthesis of ¹⁵N and deuterium isotopically enriched molecular precursors. A number of retrosynthetic approaches can be envisioned, Figure 2. Advantages and disadvantages of these retrosynthetic strategies are briefly discussed in the succeeding paragraphs.

Figure 2.

Five potential retrosynthetic approaches for preparation of protected ¹⁵N-choline PHIP precursor. The desired precursor shown in upper left corner has ¹⁵N label and 11 deuterium labels. The outlined five strategies are depicted with dashed red lines highlighting potential molecular assembling routes. X refers to a halogen atom. R refers to a protecting group such as alkyl, silyl or carbonyl.

The direct coupling approach # 1 shown in Figure 2 is potentially viable because of the relative ease of preparation of acetoxy-,³⁴ silyloxy-,³⁵ benzoxy-haloalkenes.³⁶ Unfortunately, direct replacement of vinylic halogen is extremely difficult, because alkylic halides, which are more active, react rather slowly with trimethylamine.³⁷

Both ethoxy-³⁸ and benzoxy-acetylenes³⁶ shown in approach #2, Figure 2, have been prepared previously by relatively straightforward methodologies. Ethoxyacetylene has been shown to react with trimethylamine, albeit with reverse regioselectivity, yielding (α-ethoxyvinyl)trialkylammonium hydroxides³⁹ unsuitable for ¹⁵N-choline PHIP.

Herkes and Simons⁴⁰ have shown a possibility of alkoxide addition to 1-bromovinyltrimethylammonium bromide, approach #3, Figure 2. The reaction with methanol (R =–CH₃) produced a mixture of 1-methoxyvinyltrimethylammonium bromide as well as an unknown product that was assigned as 2-methoxyvinyltrimethylammonium bromide. Despite all isolation attempts of 2-methoxyvinyltrimethylammonium bromide by fractional precipitation employing ethanol-ether or by recrystallization (acetonitrile) resulted in 1-methoxyvinyltrimethylammonium bromide as the only product.⁴⁰

Brookhart and co-workers⁴¹ have recently developed a transformation capable of producing silyl enol ethers similar to that shown in approach #4, Figure 2. Although this is by far the most promising methodology among all other approaches discussed earlier, it has two major disadvantages. First, isotopic labeling would require rather complex multistep synthesis. Second, this methodology is limited to the ethyl dimethyl silyl protecting group.

Enolization following by subsequent protection has been implemented in preparation of unsaturated t-butyl substituted phosphocholine analogs.³³ This methodology relies on stability of ylides derived from 2-trialkylammonium ketones with bulky substituents on the other side of the carbonyl group, approach #5, Figure 2.³² Unfortunately, this methodology is not applicable to the 2-trimethylammonium ethanal, betaine aldehyde, because of the competing aldol condensation. Therefore, a methodology for a betaine aldehyde enolization-protection under the mild conditions was developed.

A previously reported procedure³¹ required 2-day-long reaction at 100 °C in a sealed tube and produced only 2.3% betaine aldehyde yield after two steps. Insertion of rather expensive deuterium and ¹⁵N spin labels via betaine aldehyde required significantly more efficient synthetic route, which was not previously available to the best of our knowledge. Moreover, although betaine aldehyde with natural abundance level of ¹⁵N and deuterium isotopes is commercially available, it is very expensive for synthetic studies and large-scale production of contrast agents. For example, the Sigma-Aldrich price was $594 per 100 mg. Significantly more efficient and arguably more cost-effective methodology for preparation of betaine aldehyde is presented here. Moreover, betaine aldehyde produced in this fashion is amenable to isotopic ¹⁵N enrichment using relatively inexpensive, commercially available trimethylamine hydrochloride with ¹⁵N and/or deuterium isotopically enriched nuclei.

Allyltrimethylammonium bromide (1a) synthesized using previously described procedure²⁹ was ozonized in MeOH/H₂O and reduced with dimethyl sulfide in a manner similar to that described by Pappas and Keaveney, Figure 3, yielding betaine aldehyde (2a).⁴²

Figure 3.

Synthetic scheme for preparation of betaine aldehyde bromide (2a) and acetyl protected 1,2-dehydrocholine bromide (3a). (A) New betaine aldehyde synthesis. (B) Unsuccessful attempts to convert betaine aldehyde into protected 1,2-dehydrocholine. (C) Successful preparation scheme for acetyl 1,2-dehydrocholine.

This procedure, Figure 3A, allows producing large quantities of betaine aldehyde in its bromide form. Betaine aldehyde is a stable 1,1-diol (stable hydrated form of betaine aldehyde) even in its crystalline form. Several procedures, i–vi, Figure 3B, were evaluated for converting compound 2a to a stable protected PHIP choline precursor with a double carbon–carbon bond. Either starting material or polymerized products were obtained in all cases. However, betaine aldehyde was found to react with excess acetic anhydride, Figure 3C, resulting in acetyl 1,2-dehydrocholine with ~20% yield and with variable ratios of E and Z isomer.

Although the primary goal of this study is the development of an efficient synthetic route to a water-stable, isotopically enriched unsaturated choline PHIP precursor, the chemistry described here is new as well as the reaction mechanism, which is of interest from a fundamental perspective. The compound 3a was exhaustively characterized by multinuclear high-resolution NMR spectroscopy: 1D proton spectroscopy, 1D ¹³C spectroscopy, COSY, ¹H-¹³C HMBC and selective NOE, Figs. S1–S7, supplementary material. These studies provided peak assignments and structural characterization, Fig. S8.

Acetyl 1,2-dehydrocholine bromide (3a) is soluble in water, which is one of the major prerequisites for biomedical application of PHIP polarized contrast agents. A crude reaction mixture containing product 3a and starting material 2a was dried at reduced pressure and re-dissolved in 99.8% deuterium oxide. Proton high-resolution NMR spectra recorded after 4 min and 6 h, Figure 4, demonstrate the stability of this new compound in aqueous medium. Although a modest ~15% decomposition of 3a was observed after 6 h as measured by NMR signal intensities, the slow decay rate is very encouraging, because the PHIP polarization process takes less than 2 min. Therefore, overall 3a stability in aqueous medium is satisfactory.

Figure 4.

Proton high-resolution NMR spectra of compound 3a: spectrum in blue color represents ‘alkene region’ of 3a after being dissolved for 4 min in D₂O, spectrum in red corresponds to the same sample after 6 h.

The starting material to introduce the ¹⁵N PHIP spin label is (CH₃)₃¹⁵N•HCl (¹⁵N-trimethylamine hydrochloride, Figure 5) rather than allyltrimethylammonium bromide shown in Figures 3A. As a result of required chloride as a starting material, the previously described procedure²⁹ was modified to accommodate bromide to chloride ion substitution. The synthetic scheme shown in Figure 6 allows potentially introducing the ¹⁵N spin label from commercially available (CH₃)₃¹⁵N•HCl and (CD₃)₃¹⁵N•HCl. The viability of this modified synthetic route was initially tested with non-isotopically enriched compounds, because the latter are significantly less expensive than isotopically enriched ones. Moreover, stable ¹⁵N and D isotopes exponentially complicate structure characterization and validation of final products and intermediates by both high-resolution NMR and mass spectrometry because of the presence of hetero-nuclear spin-spin couplings and the complex distribution of molecular masses in the case of partial enrichment.

Figure 5.

Synthetic scheme for preparation of acetyl ¹⁵N-1,2-dehydrocholine bromide (3c) using ¹⁵N-trimethylammonium hydrochloride as a starting material. Achieved percentage yields are provided.

Figure 6.

Synthetic scheme for preparation of betaine aldehyde chloride (2b) and acetyl protected 1,2-dehydrocholine chloride (3b) using trimethylammonium hydrochloride as a starting material enabling potential introduction of ¹⁵N spin label from commercially available (CH₃)₃¹⁵N•HCl​ and (CD₃)₃ ¹⁵N•HCl. Achieved percentage yields are provided.

In the first step, (CH₃)₃N•HCl, trimethylamine hydrochloride, was converted to its free base by reaction with ethanolic solution of sodium hydroxide and reacted with allyl bromide. Anion exchange on an IRA-400 Cl column afforded known allyltrimethylammonium chloride (1b),³⁰ which in turn was converted to betaine aldehyde (2b) by an ozonolysis-reduction procedure described for (2a). In the absence of extensive drying of betaine aldehyde chloride improved overall yield of the desired product (3b) but produced substantial quantities of a side product 1,1-diacetyl choline chloride (4b). The final product was evaluated by means of 1D and 2D high-resolution NMR spectroscopy in both DMSO-d₆ as well as aqueous medium, 99.8% D₂O: 1D proton spectroscopy, 1D ¹³C spectroscopy, ¹H-¹³C HSQC and NOESY, Figs. S9–S21, supplementary material. The proven stability and solubility in water, and improved yield of the final product are essential for biomedical application of this class of molecules as hyperpolarized contrast agents.

Although the ultimate goal is the development of a fully ¹⁵N and deuterium labeled acetyl ¹⁵N-1,2-dehydrocholine-d₁₁ chloride analog, the preparation of ¹⁵N-1,2-dehydrocholine chloride was developed first, because the former requires more expensive starting materials and significantly complicates structure characterization and validation by both high-resolution NMR and mass spectrometry because of the presence of 11 deuterons. No anion exchange was performed to minimize potential loss of expensive ¹⁵N isotope label. Previous experience has shown that ozonolysis-reduction produces relatively pure betaine aldehyde chloride. Moreover, betaine aldehyde is extremely hygroscopic, and it is prone to undergo partial polymerization during extensive exposure to room temperature. Therefore, its recrystallization was replaced by repeated acetone washes in this procedure. Crude ¹⁵N-labeled betaine aldehyde bromide was applied to the final step producing a mixture of desired acetyl ¹⁵N-1,2-dehydrocholine bromide (3c) as well as diacetyl ¹⁵N-choline bromide (4c) after recrystallization with relatively good yields. The presence of 3c and 4c was confirmed by means of 1D high-resolution NMR spectroscopy, Figs. S25–S29 (supplementary material): 1D proton, ¹³C{¹H}, ¹⁵N{¹H} NMR spectroscopy and high-resolution mass spectrometry. We also report 1D proton, ¹³C{¹H}, ¹⁵N{¹H} high-resolution NMR spectra of compound 1c, Figs. S22–S24, supplementary material.

Although compound 3c carries a ¹⁵N spin label for preparation of potential PHIP, long-lived hyperpolarized states, it has too many protons with the anticipated outcome of dramatically decreased PHIP polarizing efficiency. This problem is usually mitigated by simplifying the spin system of the PHIP contrast agent to three spins: a long-lived heteronucleus and two nascent protons after parahydrogen addition. The key prerequisite for the molecular structure of unsaturated ¹⁵N PHIP precursor is the absence of any J-coupled protons to the ¹⁵N heteronucleus. This has been demonstrated with many ¹³C hyperpolarized contrast agents, when protons adjacent to the double bond were replaced with deuterons.^19,21,43,44 To introduce deuterium in the molecular structure of PHIP precursor, ¹⁵N-trimethylamine-d₉ hydrochloride and allyl bromide-d₅ were used as starting materials, Figure 7. Note that deuteration of the protecting acetyl group is not necessary, because its methyl protons are not J-coupled to ¹⁵N. Preparation of 1d was similar to that of 1b, whereas synthesis of final product 3d required deuterated solvents D₂O and methanol-d₁ during the ozonolysis step. Because of relatively high cost of ¹⁵N-trimethylamine-d₉ hydrochloride, no recrystallization of final product 3d was attempted. The presence of 3d and 4d was confirmed by means of 1D high-resolution NMR spectroscopy, Figs. S25–S29 (supplementary material): 1D proton and ¹³C{¹H} NMR spectroscopy. We also report ¹³C{¹H}, ¹⁵N{¹H} high-resolution NMR spectra of compound 1d, Figs. S30–S31, supplementary material.

Figure 7.

Synthetic scheme for preparation of acetyl ¹⁵N-1,2-dehydrocholine-d₁₁ bromide (3d) using ¹⁵N-trimethylamine-d₉ hydrochloride and allyl bromide-d₅ as starting materials. Achieved percentage yields are provided. Note the deuterium position susceptible to isotopic exchange and achieved % deuteration.

¹⁵N enrichment is nearly 100% in starting materials as confirmed by expected proton splitting patterns. Deuterium enrichment is expected to be similar to ¹⁵N for non-exchangeable sites, but it is likely to be more challenging for exchangeable sites in the final product or intermediates. For example, deuterons of methyl quaternary amine and deuteron in position Hb in 3d are non-exchangeable during synthesis progress. As a result, the residual protons are almost non-detectable by proton NMR spectroscopy, Figure 8. However, deuterons in position Ha can potentially exchange during synthesis. A likely step is conversion of ¹⁵N-betaine aldehyde-d₁₁ into 3d. This step requires large quantities of acetic anhydride, and a small acetic acid impurity is a likely explanation for acid catalyzed deuterium-hydrogen exchange via keto-enol tautomerization. As a result, deuteration of position Ha is only 30%, Figure 7. This potentially can be improved by additional purification of the acetic anhydride, which will be addressed in the future work.

Figure 8.

Proton high-resolution NMR spectra of compounds 3d and 3c and molecular diagram of acetyl ¹⁵N-1,2-dehydrocholine showing proton positions. (A) Acetyl methyl spectral region used for ‘fingerprinting’ of compound 3d lacking most of its other protons. (B) Molecular diagram of acetyl ¹⁵N-1,2-dehydrocholine. (C) Vinyl spectral region used for % D.

The PHIP hydrogenation is not stereospecific²⁷ in the sense that both E- and Z-acetyl ¹⁵N-1,2-dehydrocholine-d₁₁ can serve as a potentially new precursor for PHIP hydrogenation with parahydrogen to produce ¹⁵N hyperpolarized protected choline in a manner similar to recently demonstrated phosphate-protected ¹³C hyperpolarized phospholactate-1-¹³C.²⁷ Although the mechanism of this newly reported synthetic route for preparation of –OH protected unsaturated choline analogs is outside of the scope of this work, we can speculate that Z-isomer is produced as the major and perhaps the only product of this transformation, Figures 6 and 7. This speculation is supported by observation that a significantly greater proportion of Z-isomer is detected by NMR in the first 5–8 h of the final reaction step with acetic anhydride. Subsequent NMR measurements consistently detected more E-isomer and diacetylbetaine aldehyde. The latter is expected because of formation of the side product acetic acid during reaction of acetic anhydride and betaine aldehyde. Produced in this manner, acetic acid is likely responsible for yielding diacetylbetaine aldehyde and catalyzing Z- to E-isomer conversion. Although the presence of both E- and Z- isomers is not likely to impact PHIP hyperpolarization efficiency, diacetylbetaine aldehyde is clearly an undesirable product that cannot contribute to production of hyperpolarized ¹⁵N choline analogs because it lacks a double C=C bond.

Use of a phosphate group for protection of C=C–O– has been shown as a promising strategy significantly expanding the range of potential biomolecules amenable to PHIP hyperpolarization.²⁷ The transformation of betaine aldehyde to 1,2-dehydrocholine analogs as demonstrated here can be potentially tailored to preparation of ¹⁵N-1,2-dehydrophosphocholine-d₁₁. This is highly desirable, because phosphate protecting groups are resilient to hydrogenation. Therefore, PHIP hydrogenation of ¹⁵N-1,2-dehydrophosphocholine-d₁₁ would yield hyperpolarized ¹⁵N-phosphocholine-d₁₁ suitable for biomedical applications as a metabolic contrast agent without any additional chemical modifications. PHIP hydrogenation of the acetyl ¹⁵N-1,2-dehydrocholine-d₁₁ synthesized here could yield acetyl ¹⁵N-choline- d₁₁. However, the latter is highly neurotoxic and therefore requires chemical modifications before it can be used in biomedicine. These hydrogenation or/and other chemical modifications are much needed for biomedical use of this class of contrast agents, but well beyond the scope of this study.

Conclusion

Presented here is a robust synthetic approach for preparation of protected 1,2-dehydrocholine compounds. The synthetic approach allows for convenient incorporation of ¹⁵N and deuterium spin labels in a manner suitable for PHIP hyperpolarization of protected ¹⁵N-choline based contrast agents. The method is exemplified with natural abundance and ¹⁵N and deuterium spin labeled acetyl 1,2-dehydrocholine derivatives as described in US Patent Application US2012/0225020. Ac–O–C=C– motifs were successfully PHIP hyperpolarized earlier²⁶ suggesting that acetyl protected cholines can be polarized as well and hyperpolarized ¹⁵N-choline can be produced using a recently demonstrated strategy.²⁶

Moreover, this synthetic approach could also be extended to 1,2-dehydrophosphocholine PHIP precursor. The latter may be more suitable for molecular hydrogenation with water soluble Rh⁺ based catalysts, because the negative charge of the phosphate protecting group could significantly accelerate the hydrogenation rate of otherwise positively charged choline moiety. Although PHIP of ¹⁵N-phosphocholine using 1,2-dehydrophosphocholine as a precursor remains to be demonstrated in future studies, complete hydrogenation in PHIP of ¹³C-phospolactate was successfully demonstrated²⁷ suggesting that phosphocholine should also be amenable by PHIP.

Hyperpolarized ¹⁵N-choline is a potential molecular contrast agent of cancer, reporting on elevated choline uptake and elevated rate of phosphocholine synthesis due to hypoxia, HIF-1α over-expression and so on. Hypoxia is a universal hallmark of cancer.^45,46 Therefore, hyperpolarized ¹⁵N-choline enabled by the reported synthetic approach can be potentially used as a very specific imaging biomarker of cancer similar to 1-¹³C-pyruvate reporting on a different hallmark of cancer, elevated aerobic glycolysis.^46,47