Synthesis and hyperpolarisation of eNOS substrates for quantification of NO production by 1H NMR spectroscopy

Graphical abstract


Introduction
Nuclear magnetic resonance (NMR) is the most powerful technique used for identifying and characterizing organic molecules. 1 Unfortunately, NMR is inherently affected by a lack of sensitivity. 2 Magnetic resonance imaging (MRI) is a widely used clinical imaging technique that shares the same principles of NMR, and presents the same inherent problem of low sensitivity, which limits its application in many fields. 3 Recently, several hyperpolarization techniques have been developed to overcome the sensitivity issue of NMR and MRI, allowing detection of trace amounts of a certain compound in a complex mixture, 4,5 detection of a metabolite in cells 6 or real-time imaging in small rodents. 7 The most used hyperpolarisation techniques are dynamic nuclear polarization (DNP), 8,9 spin-exchange optical pumping (SEOP), metastability exchange optical pumping (MEOP), 10 parahydrogen induced polarization (PHIP) 11 and spontaneous amplification by reversible exchange (SABRE), [11][12][13] which recently emerged as a promising modality for in vivo pre-clinical and clinical MRI. 14,15 SABRE exploits the singlet spin state of parahydrogen (pH 2 )one of the spin states of the H 2 molecule -in order to increase the NMR signal. Compared to other hyperpolarization techniques, SABRE has the remarkable advantages that the substrate to hyperpolarize remains structurally unchanged and it is relatively non-expensive. Polarization transfer occurs through an iridium(I) complex that acts as a catalyst. Generally the catalytic system contains a carbene group that aids the process (Fig. 1). 16,17 SABRE has been shown to be effective in the hyperpolarisation of various spin ± ½ nuclei such as 1 H, 18 15 N 19-22 or 31 P. 23 Biologically relevant molecules have been hyperpolarised, 24,25 the catalyst can be deactivated 26 and hyperpolarisation in aqueous and biologically compatible media has also been achieved. [27][28][29] Nitric oxide synthases (NOS) are a family of enzymes that catalyse the synthesis of nitric oxide (NO) from L-arginine through an http 30 Three NOS isoforms are known: neuronal (nNOS, NOS-1), inducible (iNOS, NOS-2) and endothelial (eNOS, NOS-3). 31 nNOS and eNOS are constitutive isoforms and Ca 2+ dependent, while iNOS is Ca 2+insensitive and activated by pro-inflammatory cytokines in certain situations of stress or disease as a response of the immune system. 32 NO is a free radical molecule involved in many functions, such as vascular tone and blood pressure regulation, 33 blood flow in the kidney, 34 penile and clitoral erection, 35 immune response 36 and neuronal transmission. 37 It has been reported that abnormal NO formation or NO dysregulation play an important role in certain conditions such as type 2 diabetes, 38 heart failure, 39 haemolytic disorders 40 or critical illness myopathy. 41 Therefore, detection of NO production in vivo would be highly beneficial. Direct in vivo detection of NO is challenging because of its very short biological half-life (a few seconds). Moreover, current methods for detecting and quantifying NO in vivo are affected by significant drawbacks. 42 Conversion of L-arginine or L-NOHA to L-citrulline derivatives could be indirectly but effectively tracked by the shifting of the hyperpolarized NMR signal from the enzymatic substrate (L-arginine or L-NOHA) to the product L-citrulline. However, preliminary experiments showed that native L-arginine cannot be efficiently hyperpolarised via SABRE because -unlike pyridine and its derivatives -it is not a sufficiently good ligand for the iridium(I) catalyst.
Herein we report on the effective hyperpolarization of L-arginine-type pyridyl-amide substrates of endothelial nitric oxide synthase (eNOS). These molecular probes can be used to detect in vitro the activity of eNOS by 1 H NMR spectroscopy and might find future use as MRI probes in in vivo studies.

Design of L-arginine derivatives as eNOS substrates
It has been previously demonstrated that the guanidine moiety of eNOS substrates cannot be modified without causing a dramatic loss of efficiency in the enzymatic conversion to urea and consequent release of NO. In fact, small structural modifications of the L-arginine side chain, such as guanidine methylation or replacement of the d-methylene with oxygen or carbonyl function afforded substrates with significantly lower affinity or even converted them into potent inhibitors of the NOS enzyme. [43][44][45][46][47][48] On the other hand, Grant et al. 47 reported that the a-amino moiety is involved in the binding through the formation of a hydrogen bond in the active pocket of NOS. We therefore decided to focus on the L-arginine carboxylic function as a plausible site for the introduction of the iridium(I)-binding pyridine group, by installing a pyridyl-amide on L-arginine analogues. Diverse pyridyl amines differently substituted and equipped with a spacer were selected (structures 2a-i, Scheme 2), since the length of the spacer and the pyridine substitution was expected to affect the level of polarization transfer by SABRE.
In addition, considering that L-NOHA activity is similar or higher than that of L-arginine, 46,48 we decided to explore also L-NOHA analogues incorporating a pyridine ring, such as 11a (Scheme 4). Since it has been shown that neither the carboxylic nor the a-amino functions of L-NOHA are required for enzymatic recognition -in fact, some N-substituted alkyl hydroxyguanidines displayed comparable activity to that of native L-arginine (i.e. N-butylhydroxyguanidine) 49,50 -we included in our study also L-NOHA derivatives lacking the a-amino function (such as 11b, Scheme 4) or having a protected a-amino group (11c, Scheme 4).

Synthesis of L-arginine derivatives
Compounds 4a-i were synthesised starting from commercially available Boc-L-Arg(Pbf)-OH 1 (Scheme 2, Eq. 1), that was coupled with the corresponding amino pyridines 2a-i with HATU as the coupling agent to afford the protected amides 3a-i in good yields. Pbf and Boc protecting groups were cleaved from 3a-i in a TFA/ CH 2 Cl 2 95:5 mixture to yield the final free compounds 4a-i. Purification of 4a-i was performed on a Sep-Pak Ò C-18 cartridge using water as the eluent. The TFA counter-ion was exchanged to hydrochloride by dissolving 4a-i in diluted HCl and freeze-drying as many times as necessary. In order to obtain 4d and 4i, commercially available 3-(4-pyridyl)alanine 5 and 3-(3-pyridyl)alanine 6 respectively (Scheme 3) were dissolved in methanol and treated with SOCl 2 to afford the methyl ester derivatives 2d and 2i, which were coupled to 1 as previously described. Compound 4j required deprotection of the methyl ester from 4i (Scheme 2, Eq. 2), using LiOH overnight to yield 3j, which was converted into 4j as described above.
Synthesis of the second generation of compounds, e.g. L-NOHA analogues 11a-c, is shown in Scheme 4. Compound 11a was prepared by starting from the commercially available amino acid Boc-Cit-OH 7 which was coupled to 4-picolylamine to afford 8 (Scheme 4, Eq. 1). Compound 11b (Scheme 4, Eq. 2) was prepared from 5-aminovaleric acid 12, whose acid function was protected as methyl ester 13, followed by N-Boc function formation 14. The methyl ester was then hydrolysed to yield 15, which was coupled to 4-picolylamine to give 16. After acidic N-Boc removal, the Nsubstituted urea 17 was prepared using potassium cyanate in aqueous HCl. Finally, 11c (Scheme 4, Eq. 3) was synthesised from 8 by deprotecting the N-Boc group in acidic conditions and treating it with acetic anhydride in basic aqueous solution to yield urea 19.
The three urea derivatives 8, 17 and 19 were treated with CH 3 -SO 2 Cl in pyridine at 40°C to afford the N-cyanamides 9, 18 and 20 (Scheme 4), which were purified by flash chromatography, when possible. These N-cyanamides were poorly stable, to the extent that 10 was used as a crude, without purification or characterisation. L-NOHA derivatives 10 and 11b-c were prepared by nucleophilic addition of hydroxylamine to 9, 18 and 20 in good yields and purified by reverse phase HPLC. The Boc protecting group in 10 was cleaved in dioxane with 4 M HCl (quantitative yield) to afford 11a. 11b-c were obtained as TFA salts after HPLC purification, then TFA was exchanged to hydrochloride following the procedure described above.

Synthesis of L-citrulline derivatives
The three L-citrulline derivatives 21a-c (Scheme 5) were synthesised by coupling commercially available Boc-L-Cit-OH 7 and the corresponding pyridine function, namely 4-picolylamine, 3picolylamine or 4-aminopyridine, using HATU to yield 8 (described above), 22 and 23, respectively, in good yields. The Boc protecting group was cleaved by reaction with 4 M HCl in dioxane to afford L-citrulline derivatives 21a-c.

Enzymatic activity assays
NO production experiments were performed using (bovine recombinant) eNOS enzyme. NO Production was measured by the well-established nitrate/nitrite -also called lactate dehydrogenase (LDH) -colorimetric assay, which exploits the Griess reaction to quantify the nitrites generated from NO. 51 The concentrations of NO-derived nitrites produced from 4a-j and 11a-c were higher or comparable to that of L-arginine when incubated in the presence of eNOS (Table 1). L-Arginine produced a 25 lM nitrites concentration, while for 4a, 4g, 4j and 11a-c higher concentrations of nitrites were measured. The remaining compounds displayed NO production similar to L-arginine. Dipeptide derivatives 4d and 4i were not tested for enzymatic activity because -as explained belowthey did not polarise successfully.

SABRE hyperpolarisation
Hyperpolarization experiments were performed on 4a-j and 11a-c. All experiments were done in 0.6 mL of methanol-d 4 in Young tubes previously degassed and filled with pH 2 with a pressure of 3 bar. 1,15 The catalyst precursor used was [Ir(COD)(IMes) Cl] [COD = cyclooctadiene; IMes = 1,3-bis(2,4,6-trimethylphenyl) imidazole-2-ylidene], since it was previously demonstrated that this catalyst is one of the most versatile precursors for polarization transfer. 16 We used the so called ''shake-and-drop" method, according to which the sample was shaken for 10 s in a given magnetic field prior to dropping it into the NMR magnet and rapidly acquiring the NMR spectrum. 15 For these experiments a single 90°radiofrequency (RF) pulse was applied. All of the samples contained a concentration of 40 mM of the ligand 4a-j or 11a-c and 5 mM of the catalyst precursor. Polarization transfer was performed at both 65 G and Earth's magnetic field (MF) in order to study the effect of MF on the polarization transfer onto the ligands. These MF were chosen based on literature reports. 12,18,24 Scheme 2. Synthesis of first generation compounds 4a-j. Polarization transfer is maximised when the externally applied MF allows the couplings J (between the hydrides and the accepting protons) and the chemical shift difference between them to become optimally aligned. The effect of a co-ligand was also studied. In this context, a co-ligand is a small molecule that has lower binding affinity for the iridium catalyst centre than that of the ligand to be hyperpolarized. When the residence time on iridium is too long, the co-ligand can help driving dissociation of the already hyperpolarized ligand, thereby enhancing the build-up rate of this hyperpolarised agent in solution. In a second approach, the co-ligand can enable the hyperpolarization of more bulky targets which might not initially bind due to steric hindrance. The coligand, being a small molecule, helps freeing up space on the iridium centre to enable the bulky target to form an active catalyst. Acetonitrile generally behaves as a good co-ligand for SABRE, and in these experiments 2 lL of acetonitrile-d 3 was used (final concentration of 60 mM) to avoid polarization being lost into acetonitrile protons during catalysis.
The signal enhancements obtained during this study can be seen in Table 2. Generally, the 4-substituted pyridines hyperpolarized better in the Earth's MF, while the 3-substituted pyridines showed better performance at 65 G. As expected, when hyperpolarizing with acetonitrile-d 3 the results followed the trend that bulkier molecules polarize slightly better in the presence of acetonitrile-d 3 . Conversely, 4d and 4i-j, the dipeptides derivatives, did not show any significant enhancement of signal, possibly due to steric hindrance even when using a co-ligand. The second generation of compounds also failed to show good performance under these SABRE conditions as only 11b was hyperpolarized, while 11a and 11c failed to hyperpolarize significantly. We believe that the hydroxyl-guanidyl group, despite being less basic than the guanidyl, may bind the metal in an even stronger manner, partially inactivating the catalyst. Compound 4e was therefore the best performing compound in the SABRE experiment, giving the highest enhancements (Fig. 2). The hyperpolarized 1 H NMR spectrum (in blue) can be compared to its thermal (in red), where thermal spectrum is vertically enlarged 32 times.
Two enhanced signals were expected, but four were actually observed in the spectrum. Signals labelled as A and C on Fig. 2 correspond to the free compound 4e in solution, while signals B and D correspond to the fraction bound to the catalyst. The bound fraction may readily be dissociated by using a catalyst deactivator, such as bipyridine. 26 Deactivation of the catalyst may be achieved by adding a chelator to the solution, which is a molecule with a high affinity for the iridium centre. The total enhancement for all four observed enhanced signals was 870-fold, which would correspond to only two peaks if the catalyst were deactivated. This corresponds to a 2.8% achieved polarization at 9.4 T. Hydride signals can be observed in the d À21 to À24 region, indicating that as proposed pH 2 effectively enters the catalytic complex and is coupled to the pyridine of the ligand to hyperpolarise.

In vitro spectroscopy
The first in vitro spectroscopy attempt was performed using 4e, i.e. the compound that showed the highest level of hyperpolarisation. The experiment was carried out under conditions used for an optimal performance of the eNOS enzyme. The reaction was monitored by NMR. The L-citrulline derivative 21c was expected to give similar, yet distinguishable, peaks to those of 4e with the maximum difference being 0.05 ppm in the aromatic region. Fig. 3 shows spectroscopically the fate of analogue 4e after 3 h of reaction. Signals A and C correspond to 4e and after 25 min new signals B and D started to form, with a 41% conversion after 3 h. However, HPLC analysis of the crude enzymatic mixture revealed that L-arginine and 4-aminopyridine were the main components of the mixture. Further analyses showed that peaks B and D belong to 4-aminopyridine, as confirmed independently by recording the spectrum of 4-aminopyridine alone (not shown), and thatto our surprise -compound 4e was simply being hydrolysed by the PBS buffer solution, even in the absence of eNOS.
Analogues 4a and 4f also gave good levels of hyperpolarisation (268 and 270-fold enhancement respectively), therefore investigation of their suitability for NMR spectroscopy was carried out next. Both 4a and 4f proved to be stable under the reaction conditions overnight. However, when PBS buffered D 2 O solutions of a 1:1 mixture of 4a or 4f and their corresponding citrulline derivatives 21a and 21b respectively were analysed by 1 H NMR spectroscopy at 400 MHz we could not discriminate between enzymatic product and starting material via peaks in the aromatic peaks region which would receive SABRE hyperpolarisation.
Gratifyingly, when a 2D COSY or NOESY spectrum of a 1:1 mixture of 4f and the enzymatic product 21b was recorded, the amino acid back-bone signals can be distinguished as they exhibit chemical shift differences of 48 Hz and 46 Hz at a field of 9.4 T, while the corresponding benzylic protons appear as a singlet at d 4.60 and an AB doublet at d 4.57 and 4.68 respectively (Fig. 4). Hence, selective excitation of the overlapping signals at d 8.70 for the H-5 ring protons of the pyridyl unit of 4f in conjunction with 1D COSY or NOE methods enabled the observation of a resolved connection to the distinguished benzylic protons. It is worth noting that similar 1D COSY methods have been used by Tessari et al. to establish the magnitude of hydride substrate couplings in the catalyst, 52 whilst use of the NOE approach has been described by some of us. 5 The ratio of these peaks is proportional to the amount of material in solution. Hence, this enhanced hyperpolarized response can be used to probe NO production in vitro by 1 H NMR spectroscopy.
Finally, relaxation times T 1 ranging from 2.0 to 3.6 s were experimentally determined by 1 H NMR in physiological conditions for selected pyridyl protons of compounds 4a,e,f (Fig. 5). Although these T 1 values are currently too low for in vivo MRI applications of these probes (T 1 of at least 8-10 s would be required to guarantee sufficient polarization half-life), it is known that deuteration of adjacent protons could significantly increase T 1 values, 6 therefore deuterated versions of molecules 4 might be viable probes for future in vivo studies.

Conclusions
Novel eNOS substrates, L-arginine or L-NOHA analogues incorporating a pyridyl group, have been successfully synthesised for SABRE hyperpolarization. Compounds 4a-j and 11a-c showed a comparable or higher NO production by eNOS than L-arginine. Polarization transfer by SABRE was optimized to yield a maximum hyperpolarization of 870-fold enhancement on 4e. In vitro spectroscopy was successfully carried out in the presence of eNOS, and a future application of MRI spectroscopy for imaging NO production using SABRE-hyperpolarised tracers in vivo might be possible, provided further increase of both hyperpolarisation -for achieving sufficient signal-to-noise ratio -and T 1 -for achieving sufficiently long hyperpolarization half-life -can be achieved. The potential to resolve such a reaction in vivo depends on how the relaxation times of these resonances respond to the biochemi-  cal environment. It is well known, however, that 13 C detection in vivo is also possible with hyperpolarised agents and hence should further tests reveal this to be a problem we expect to be able to harness a 13 C labelling strategy to ensure success with these agents. To our knowledge, the library of compounds reported herein is the first attempt to develop SABRE-hyperpolarised tracers for quantification of NO production.   lamp and/or staining with a ceric ammonium molybdate or potassium permanganate solution. Flash chromatography was performed on silica gel (60 Å, particle size 0.040-0.062 mm). Yields refer to chromatographically and spectroscopically pure compounds, unless stated otherwise. Abbreviations used: DCM for dichloromethane, EtOAc for ethyl acetate, Et 2 O for diethyl ether, MeOH for methanol, THF for tetrahydrofuran, MeOD for deuterated methanol, HATU for 1-[Bis(dimethylamino)methylene]-1H-1,2,3triazolo[4,5-b]pyridinium 3-oxid hexafluorophosphate, TEA for triethylamine, TFA for trifluoroacetic acid, Boc for tert-butyloxycarbonyl, Pbf for 2,2,4,6,7-pentamethyldihydrobenzofuran-5-sulfonyl.

General procedure for the synthesis of compounds 2d and 2i
A solution of the corresponding pyridyl-L-alanine (150 mg; 0.90 mmol) in 2.5 mL of methanol was cooled down to À10°C. SOCl 2 (328 mL; 4.51 mmol) was added dropwise and the mixture stirred at r.t. for 36 h. Solvent was evaporated to yield the dihydrochloride products in 90%-quantitative yields.

General procedure for the synthesis of compounds 4a-i
Precursor was dissolved in 1 mL of TFA and stirred at r.t. for 2 h, unless stated otherwise. Solvent was evaporated under a current of air, triturated with Et 2 O, dissolved in water and passed through a Sep-Pak Plus Ò C18 cartridge previously conditioned by passing methanol (10 mL) followed by water (10 mL). Solvent was finally freeze dried and TFA salt exchanged to HCl salt by adding some drops of HCl 2 M prior to freeze-drying. Products were obtained in 38%-quant yield.

Hyperpolarisation and NMR experiments
para-hydrogen was produced by cooling H 2 gas over a spinexchange catalyst (Fe 2 O 3 ) at 30 K temperature. This method was able to provide para-hydrogen with more than 93% purity. Samples involved in the analysis were prepared with $ 6 mg of ligand ($60 lM) and 2 mg of IMes precursor catalyst (3 lM) dissolved in 0.6 ml deuterated methanol solvent in a 5 mm NMR tube fitted with a J. Young's tap. The resulting solutions were then degassed by 3 cycles of freeze-pumpthaw method before filling the tube with pH 2 at a pressure of 3 bar. Once filled with pH 2 , the tubes were shaken vigorously for $10 s in a fringe field of $65 Gauss around a 9.4 T Bruker spectrometer or Earth's magnetic field. Immediately after that, tubes were rapidly transported inside the spectrometer for subsequent NMR detections. Enhancement factor was calculated by taking the ratio of the integrals of peaks in the hyperpolarised spectra and thermal equilibrium spectra. Catalyst precursor was synthesised in our laboratory according to a literature procedure 1 [IMes = 1,3-bis(2,4,6-trimethylphenyl) imidazole-2-ylidene and COD = cis,cis-1,5-cyclooctadiene]. T 1 measurement experiments were performed using 8 mg of compound in the form of HCl salt (same amount used in the SABRE hyperpolarisation experiments) dissolved in 0.6 mL of PBS buffer pH = 7.4 and 37°C.

Enzymatic tests
eNOS (bovine recombinant) was purchased from Cayman chemicals and used as received. 1 U of enzyme produced 1 nM/ min of NO at 37°C in 50 mM HEPES buffer (pH 7.4) with 1 mM CaCl 2 , 20 mg/ml CaM, 0.1 mM NADPH, 50 mM L-arginine and 12 mM tetrahydrobiopterin. The amount of enzyme in 1 U was calculated from the batch specific activity as shown on the data sheet.
L-Arginine analogues 4a-i and 12a-c were tested using 1 U of enzyme in the stated conditions against an L-arginine positive control and a negative control solution in a total reaction volume of 100 lL. The reaction was optimised to 40 min. 3 lL of an ice-cold solution containing 20 mM HEPES (pH 5.5), 2 mM EDTA and 2 mM EGTA were added to stop the reaction. The concentration of NO was determined using the colorimetric Griess reaction by following the nitrate/nitrite colorimetric assay kit LDH method provided by Cayman chemicals.