Synthesis of aromatic 13 C / 2 H-α-ketoacid precursors to be used in selective phenylalanine and tyrosine protein labelling †

Recent progress in protein NMR spectroscopy revealed aromatic residues to be valuable information sources for performing structure and motion analysis of high molecular weight proteins. However, the applied NMR experiments require tailored isotope labelling patterns in order to regulate spin-relaxation pathways and optimize magnetization transfer. We introduced a methodology to use α-ketoacids as metabolic amino acid precursors in cell-based overexpression of phenylalanine and/or tyrosine labelled proteins in a recent publication, which we have now developed further by providing synthetic routes to access the corresponding side-chain labelled precursors. The target compounds allow for selective introduction of C–H spin systems in a highly deuterated chemical environment and feature alternating C–C–C ring-patterns. The resulting isotope distribution is especially suited to render straightforward C spin relaxation experiments possible, which provide insight into the dynamic properties of the corresponding labelled proteins.


Introduction
Aromatic amino-acids represent a sensitive source of structural and dynamic parameters in high-molecular weight protein NMR spectroscopy. 1 Phenylalanines and tyrosines are substantially overrepresented at protein binding interfaces due to their ability to contribute to hydrophobic, as well as to electrostatic interactions. 2Examples from the literature have proven the importance of aromatic residue based NOE data to complement the set of methyl-group derived distance restraints for structure calculation. 3Moreover, aromatic side chains display a remarkable flexibility in dynamic motion, which can be sensitively probed by 13 C-1 H spin pair relaxation. 4Insufficient chemical shift dispersion, extensive 13 C-13 C spin coupling and retarded side-chain motion strongly affect the signal assignment and analysis in the aromatic spectral region.
Selective stable-isotope patterns are required to enable effective magnetisation transfer and well defined spin relaxation, which is both necessary to decrypt the structural information buried in these residues.Alternating 12 C-13 C-12 C and/ or 2 H-1 H-2 H arrangements in the aromatic ring systems have been shown to result in well resolved NMR signals due to sig-nificant reduction of scalar and dipolar couplings. 5Isolated 13 C-1 H spin systems in an otherwise 2 H-containing aromatic ring have additionally been used as valuable tools to elucidate the aromatic side chain motion by erasing unwanted relaxation pathways. 6Reports on labelling phenylalanine and tyrosine residues with stable isotopes include cell-free (CF) protein synthesis, 7 as well as cell-based expression systems. 6,8CFapproaches require the sophisticated synthesis of 15 N-labelled amino acids, but display highly selective isotope composition in the target proteins.Cell-based overexpression, on the other hand, makes use of amino acid precursor compounds, which are introduced to the metabolism of a protein expressing organism. 9lthough economically preferred, cell-based methods often suffer from low incorporation rates and selectivity due to the loss of heavy isotopes at metabolic crossroads.In order to expand the methodology of introducing stable isotopes at distinct positions of a target protein, we recently presented highly selective phenylalanine-and tyrosine-residue labelling based on the corresponding metabolic α-ketoacid precursors sodium phenylpyruvate and sodium 4-hydroxyphenylpyruvate (Scheme 1). 10 Protein synthesis using an E. coli overexpression host in the presence of the labelled aromatic α-ketoacids thus resulted in the incorporation of 13 C without any cross-labelling to other residues.This new methodology combines the robustness and versatility of in-cell overexpression with high incorporation selectivity, which is usually the domain of cell-free protein synthesis.In order to further develop our α-ketoacid precursor based approaches towards selective side-chain labelling, 11 we developed a synthetic route to sodium phenylpyruvate 1 containing 13 C-1 H at meta-positions in an otherwise perdeuterated chemical environment.We could already demonstrate that this side-chain labelled precursor is selectively converted to Pheresidues in an E. coli expression medium. 10This article describes the synthetic details to obtain the 13 C/ 2 H aromatic α-ketoacids illustrated in Scheme 2. In addition to the already mentioned precursor 1, synthetic approaches to access para 13 C-1 H labelled phenylalanine precursor 2, as well as the meta 13 C-1 H tyrosine precursor 3 are presented.The routes feature acetone and heavy water as 13 C and 2 H sources, respectively.Labelling of backbone positions is feasible by application of 13 C-glycine as shown previously. 10

Results and discussion
The approach to access the target compounds 1-3 (Scheme 2) is based on the synthesis of the aromatic ring by the reaction of labelled acetone with nitromalonaldehyde in basic aqueous solution. 12Selective deuteration at activated ring-positions was planned in acidic D 2 O using aniline or 4-aminophenol as electron rich substrates at elevated temperatures. 13On the one hand, this synthetic concept was designed as an economically practicable way of synthesizing enough material to be used in cell-based protein overexpression (quantitative isotope incorporation at 100-200 mg L −1 minimal medium) due to the relatively cheap sources of stable isotopes and robust reaction steps.On the other hand, the routes should be flexible enough to access alternative isotope patterns by simply switching to commercially available starting compounds with different stable isotope composition (e.g.various patterns of labelled acetone for side-chain-, or glycine as a 13 C-source for backbone labelling).
The synthesis of sodium 3,3-dideuterio([3,5-13 C 2 ]2,4,6-trideuteriophenyl)pyruvate 1 was performed as outlined in Scheme 3. Initially, a straightforward way to access the aromatic ring system in one step was applied by the reaction of commercially available [1,3-13 C 2 ]acetone 4 with sodium nitromalonaldehyde 5. Compound 5 can be prepared from mucobromic acid as a stable solid. 14Subsequent deoxygenation of [2,[6][7][8][9][10][11][12][13] C 2 ]4-nitrophenol 6 was performed in a two-step reaction sequence via the 1-phenyl-1H-tetrazolylether 7. 15 Compound 7 was prepared by reaction of the phenolic hydroxy group with 5-chloro-1-phenyl-1H-tetrazole in the presence of KOtBu.Hydrogenation using palladium on charcoal at room temperature and a pressure of 4 bar removed the oxygen from the aromatic ring, while at the same time the nitro-group was reduced yielding [3,5-13 C 2 ]aniline 8. 16 At this stage, the deuterium pattern at the aromatic ring was installed, as compound 8 shows highly selective 1 H/ 2 H exchange at the electron-rich ortho/para positions in the presence of D 2 O and HCl under microwave irradiation. 13Subsequent formation of [3,5-13 C 2 ] 2,4,6-trideuteriobenzonitrile 10 was achieved by using potassium tetracyanonickelate in ammonium chloride buffer. 17eduction of compound 10 using diisopropylaluminium hydride yielded [3,5-13 C 2 ]2,4,6-trideuteriobenzaldehyde 11 which was then used in the subsequent condensation step with hydantoin. 18The preparation of labelled benzalhydantoin 12 was done in the presence of ammonium acetate, which provided higher and more reproducible yields than the use of sodium acetate reported in the literature.18a Finally, the hydantoin ring of compound 12 was hydrolysed using 20% NaOD solution, which simultaneously introduced 2 H at the C 3 -position.Labelled sodium phenylpyruvate 1 was obtained by lyophilisation from aqueous solution as a stable white powder in an overall yield of ∼16% in 8 steps from [1,3-13 C]acetone 4.
In this case, the Pd/C-mediated hydrogenation was again conducted in the continuous-flow hydrogenation reactor, leading to an isolated hydrogen atom in the para position of the resulting labelled aniline 18.The following reaction steps were performed analogously to the reaction sequence reported for the preparation of the labelled sodium phenylpyruvate 1 leading to the target compound sodium 3,3-dideuterio([4-13 C]-2,3,5,6-tetradeuterio-phenyl)pyruvate 2 in 9 steps and an overall yield of ∼11%.
A more selective deuteration pattern can be achieved, if required, as shown in Scheme 7. Methylation of 4-nitrophenol 20 and subsequent reduction of the nitro group yielded p-anisidine 28, which showed no reactivity in the deuteration step meta to the amino group (28 → 29). 21Demethylation using HBr in the presence of a phase transfer catalyst (Aliquat-336®) gave selectively deuterated aminophenol 30. 22This sequence, which was verified using unlabelled 4-nitrophenol as a starting material, increases the number of reactions in the route to prepare sodium 3,3-dideuterio([3,5-13 C 2 ]2,6-dideuterio-4-hydroxyphenyl)pyruvate 3 by two steps, but represents an effective approach to avoid partial deuteration at the 13 C labelled aromatic positions in the target compound 3.The aromatic α-ketoacids 1-3 display high stability in their lyophilized forms as sodium salts, but undergo oxidative degradation in basic solution in the presence of atmospheric oxygen.18d NMR spectra of compounds 1-3 in D 2 O show mainly the keto forms, whereas in DMSO-d 6 the enol forms predominate, which is in accordance with literature data. 23

Conclusions
An efficient synthetic concept is presented to access labelled metabolic precursor compounds of phenylalanine and tyrosine allostery and enzymatic catalysis.The straightforward and economic synthetic protocols shown below will further promote the efforts to turn aromatic residue labelling into a routinely used concept and complement the techniques of NMR-based analysis of protein dynamics, which traditionally rely on the interpretation of spin relaxation residing at the backbone or 13 C and 2 H methyl bearing side-chains. 24
Sodium nitromalonaldehyde monohydrate 5. Sodium nitrite (30 g) was dissolved in water (30 mL) using a three necked round bottomed flask, equipped with a thermometer, a dropping funnel and a tube to drain the evolved gases.The mixture was slightly warmed to dissolve all of the NaNO 2 .A solution of mucobromic acid (30 g) in ethanol (30 mL) was slowly added for a period of 1 h.After additional stirring for 15 minutes, the reaction mixture was cooled to 0 °C and the precipitate was filtered off.The resulting solid was transferred into a round bottomed flask and stirred under reflux with ethanol (50 mL) and water (10 mL).The hot solution was filtered and the filtrate was subsequently cooled to 0 °C, which led to product precipitation.The solid was filtered off and washed with small portions of cold ethanol.Drying the product under vacuum gave 7.67 g (42%) of sodium nitromalonaldehyde monohydrate 5 as a white solid, which was stored over CaCl 2 . 1  [2,6-13 C 2 ]4-nitrophenol 6.An aqueous NaOH solution (4.4 g in 20 mL) was slowly added to a mixture of sodium nitromalonaldehyde monohydrate 5 (3.25 g) and [1,3-13 C 2 ]acetone 4 (1 g) in H 2 O (200 mL) at 0 °C using a dropping funnel.After the addition was complete, the flask was tightly closed and stirred for 6 days at 4 °C.The resulting brown solution was cooled to 0 °C and 6 N HCl (26 mL) was slowly added.Filtration of the solution resulted in a dark solid, which was taken up in 6 N HCl (26 mL) and boiled gently for 10 minutes.The warm mixture was filtered and the two combined filtrates were extracted with diethyl ether (6 × 100 mL).Subsequent drying of the combined organic phases over MgSO 4 and evaporation of the diethyl ether under reduced pressure yielded a yellow solid.The crude product was purified over a silica gel chromatography column by elution with hexane-ethyl acetate (6 : 4 v/v).The reaction yielded 1.47 g (63%) of [2,6-13 C 2 ]4-nitrophenol 6. 1  ]4-nitrophenol 6 (1.4 g) in dry dimethylformamide (18.4 mL) was stirred at room temperature, while potassium tert-butoxide (1.31 g) was added within 5 minutes in small aliquots under a constant stream of argon.After 1 h of vigorous stirring under an argon atmosphere, a solution of 5-chloro-1-phenyl-1H-tetrazole (1.9 g) in dry dimethylformamide (8 mL) was added and the reaction mixture was stirred for further 3 h.The solution was warmed to 65 °C and stirring continued overnight.Precipitation of the crude product was induced by pouring the mixture in ice water (100 mL) and completed at 4 °C in 12 h.The resulting precipitate was separated by filtration and washed with small portions of ice water.The reaction yielded 2.3 g of a crude product, which was further purified by column chromatography.Elution with hexane-ethyl acetate (8 : 2) gave 5-([2,6-13 C 2 ]4-nitrophenoxy)-1-phenyl-1H-tetrazole 7 (2.07 g, 74%) as a white solid. 1  [3,5-13 C 2 ]aniline 8. Palladium on charcoal (10%, 1.04 g) was added to a solution of 5-([2,6-13 C 2 ]4-nitrophenoxy)-1-phenyl-1H-tetrazole 7 (1.04 g) in dry toluene (150 mL) in a thick walled hydrogenation flask.The flask was mounted on a hydrogenation Parr-apparatus and a pressure of 4 bar of hydrogen was applied.After 12 h of agitation, the pressure was released, the flask flushed with argon and the black solid palladium catalyst separated from the solution by filtration.The catalyst was washed with toluene (30 mL) and the combined filtrates poured on a 0.5 N NaOH solution (150 mL).After separation of the two layers, the aqueous phase was extracted with toluene (3 × 100 mL).The combined organic phases were then extracted using 0.5 N HCl (3 × 100 mL).Addition of concentrated HCl (0.5 mL) to the combined aqueous phases was followed by reducing the volume of the resulting solution by half under reduced pressure at 50 °C.NaOH (1 N) was added until the solution showed a pH of ∼10.The product was extracted from the solution with dichloromethane (3 × 100 mL).Drying of the combined organic phases over MgSO 4 and subsequent careful evaporation of the solvents under reduced pressure (>100 mbar) gave 394 mg of a product/ dichloromethane mixture which was used for further conversion.The reaction yield was determined by integrating the corresponding NMR signals to be 338 mg (97%). 1  in HCl (0.4%, 160 mL) at 0 °C using a dropping funnel.After 2 h of stirring at 0 °C, the reaction mixture was brought to pH 7 by addition of saturated aqueous Na 2 CO 3 .The resulting solution was slowly added to potassium tetracyanonickelate hydrate (1.06 g) in NH 3 -NH 4 Cl buffer (60 mL, pH = 10).Stirring was continued for 15 min at 60 °C.The solution was then filtered and the solid residue was washed with small aliquots of water.The combined filtrates were extracted with diethyl ether (4 × 100 mL) and the combined organic phases were dried over MgSO 4 .Evaporation of the solvents under reduced pressure gave a crude product, which was further purified by bulb-to-bulb distillation (30 mbar; up to 120 °C) to yield [3,5-13 C 2 ]2,4,6-trideuteriobenzonitrile 10 (387 mg, 84%) as a slightly yellow liquid. 1  108.07;found 108.0.

Scheme 1 EScheme 2
Scheme 1 E. coli based overexpression of a model protein in the presence of labelled aromatic precursor compounds of phenylalanine and tyrosine results in selective protein isotope labelling as shown in previous studies.