We investigated the possibility that amino acids can be synthesized from hydroxylamine and pyruvic acid or glyoxylic acid, under supercritical CO₂ conditions (60 ºC). Additionally, in order to compare the conditions, we repeated the same reaction under subcritical CO₂ conditions (31 ºC) as a control. For the detection of amino acids, an amino acid analyzer was used (Figure 1). Since this analysis system is adjusted for detecting amino-acid monomers, the differences before and after the hydrolysis reaction indicates the amount of polymer hidden in the original products.

Using pyruvic acid as carbon sourcs, two kinds of amino acids, alanine and glycine, were detected at 60 ºC and 31 ºC (Table 1). The free alanine yield at 60 ºC (60.7 μmol) was higher than that at 31 ºC (4.1 μmol), whereas the free glycine yield at 60 ºC (0.99 μmol) was lower than that at 31 ºC (2.2 μmol). These results suggested that alanine synthesis was conducted more effectively at 60 ºC.

After hydrolysis treatment with 6N HCl, 256.2 μmol and 70.2 μmol of alanine were detected at 60 ºC and 31 ºC, respectively (Table 1). Additionally, arginine (0.32 μmol) was detected at 60 ºC. From the significant differences of concentrations between before and after hydrolysis reaction, alanine polymers were synthesized in larger quantity at 60 ºC. However, the ratios of free alanine: alanine polymers, were 1:3.2 at 60 ºC; 1:16.5 at 31 ºC. Considering these results, alanine polymers may be more stable in lower temperature or degraded at high temperature.

In order to estimate the mechanism of alanine synthesis, we compared the reactions using glyoxylic acids as another carboxylic acid source. Different from the reaction using pyruvic acid, free alanine was not detected at both temperatures, whereas free glycine was detected at 60 ºC (866.3 μmol) and 31 ºC (183.3 μmol); free aspartic acid (2.5 μmol) was detected at 60 ºC (Table 1). These results suggested that the methyl group of alanine synthesized from pyruvic acid were derived from the methyl group of pyruvic acid, and that the alanine was synthesized from hydrogenation of pyruvic acid oxime, since alanine was not detected in the glyoxylic acid reaction.

Some methods have been proposed for the synthesis of alanine from pyruvic acid oxime. Hamlin et al. reported a hydrogenation of the oxime with a palladium-charcoal catalyst [19], and Borszeky et al. reported that a hydrogenation of pyruvic acid oxime with palladium/alumina catalysts provided the high yields of racemic alanine [20]. Although these hydrogenation were conducted with catalysts, we speculated that oxime hydrogenation also occurred under our supercritical CO₂ conditions.

After the hydrolysis of the product from glyoxylic acid, glycine (819.2 μmol) and aspartic acid (1.8 μmol) were detected at 60 ºC, whereas glycine (209.2 μmol) was detected at 31 ºC reaction (Table 1). Additionally, 1/2-cystine (1.2 μmol) and leucine (0.37 μmol) was detected at 60 ºC and 31 ºC, respectively (data not shown). The sulfur atoms of leucine and 1/2-cystine may be derived from impurities of the product of hydroxylamine hydrochloride (see Experimental Section).

The main detected amino acids both before/after hydrolysis conditions, alanine, glycine, and aspartic acid, are GNC code encoding amino acids (N means either of four bases). Ikehara et al. proposed in their hypothesis a [glycine, alanine, aspartic acid, and valine (GADV)]-protein world hypothesis on the origin of life, based on the GNC code encoding amino acids as the most primitive genetic code [21]. These GADV amino acids are known to be detected in the Miller discharge experiment [2] and the Miller volcanic spark discharge experiment using his apparatus performed after after his death [22]. Although valine was not detected in our experiments using pyruvic acid or glyoxylic acid, as a hypothesis, if the carboxylic acid source is changed to α-ketoisovaleric acid (3-methyl-2-oxobutanoic acid), valine may be synthesized under supercritical CO₂ conditions. Our data in CO₂ were thus partially agreement with the GADV-protein world hypothesis.

Surprisingly, in the glyoxylic acid reaction, the concentrations of glycine before and after hydrolysis reaction did not change dramatically (after hydrolysis reaction: before hydrolysis reaction = 209.2 μmol:183.3 μmol at 31 ºC; 819.2:866.3 at 60 ºC). Comparing the alanine data in the pyruvic acid reaction, it was estimated that glycine polymers were more unstable under supercritical or subcritical CO₂ conditions, since the ratio of alanine polymers was higher than that of free alanine in the above experiments. In contrast with the results, our previous study showed that glycine polymers, including decaglycine, were obtained from glycine monomer under hydrothermal condition (270 ºC, 10 MPa, and 27-second reaction) with an adiabatic expansion cooling system [23]. The cooling system may be an essential condition to obtain unstable glycine polymers. We speculated that alanine polymers were stable due to the formation of helix forms [24,25], but glycine polymers were unstable due to the large flexibility of the residues under CO₂ conditions, like in the aqueous model [25].

In order to confirm alanine polymer synthesis under supercritical CO₂ conditions, we analyzed the sample from pyruvic acid at 60 ºC with a LC-MS. We obtained 56 peaks under the gradient LC conditions used (Figure 2). Among these peaks, the positive ions, [M+H]⁺, of 232.13 (alanine trimer) and 303.16 (alanine tetramer) were detected at 33.7 and 34.6 min retention time, respectively (Table 2). The alanine dimer and 5 – 7 mers were not detected. These results suggested that alanine trimer and tetramer were stable under CO₂ supercritical conditions. From the ESR spectroscopic data, several alanine-based forms can adopt the 3₁₀ helix formation in aqueous solution [26,27]. Also under these supercritical conditions, the tetramers could be stable by forming the 3₁₀ helix and were thus detected in the aqueous sample due to their stability. However, we could not found out the reason why the trimer was stable under CO₂ supercritical conditions. We speculate that the trimer was a metastable structure in the supercritical CO₂; therefore the trimer was synthesized directly from three alanine monomers, not through the dimer structure, and the tetramer was synthesized from a trimer + monomer reaction. The polymers higher then pentamer may be separated out as deposits from supercritical CO₂. This deposition polymerization in supercritical CO₂ is known in ethylene polymerizations [28,29]. At the present, it is outside the scope of our research to investigate the deposition, but it will be the subject in future studies.

The reactor was made of stainless steel and was connected to a CO₂ inlet and outlet that have valves and a pressure gauge (Figure 3). Twenty six mmol of hydroxylamine hydrochloride (>97%; Tokyo Chemical Industry Co. Ltd.) and 26 mmol of pyruvic acid (98%; Alfa Aesar) or glyoxylic acid monohydrate (98%; Alfa Aesar) were mixed with 3 mL of Milli-Q (Millipore) water gently in a glass tube. The hydroxylamine hydrochloride could contain the following impurities: sulfate: max. 0.005%; copper: max. 2 ppm; lead: max. 2 ppm; iron: max. 5 ppm. After the glass tube was inserted into the stainless tube, the stainless tube was capped and CO₂ liquid was injected to purge the air inside. Then 20 g of CO₂ liquid was injected through the inlet of the reactor contained in an ice box. In order to compare the reaction conditions, we conducted reactions under two sets of conditions: supercritical condition (60 °C, over 7.4 MPa) and subcritical condition (31 °C, less than 7.2 MPa). This subcritical condition was chosen on the basis of our hypothesis that subcritical CO₂ conditions near the supercritical conditions are most effective for the reactions. The reactor was placed in an oil bath for two h in each condition. After the reactions, the reactors were rapidly moved into an ice box and the CO₂ gas released. One mL of Milli-Q water was added into the glass tube, and all the samples were transferred to a new glass bottle.

The aqueous samples (10 μL) were diluted with purified water (200-fold dilution) and filtered with 0.22-μm filters. On the other hand, the samples for hydrolysis studies were evaporated to dryness under reduced pressure. The dried residues were treated with 100 μL of 6N HCl. After vacuum sealing, the samples were hydrolyzed for 22 hr at 110 °C. After the hydrolysis, the hydrolyzed residues were evaporated to dryness under reduced pressure. The dried residues were dissolved with 200 μL of purified water and filtered with 0.22-μm filters. The filtrate samples were diluted with purified water (50-fold dilution). Both the aqueous samples and hydrolyzed samples were analyzed with an aminoacid analyzer, L-8500 (Hitachi High Technologies Co.), based on the absorbencies of 570 nm and 440 nm after ninhydrin reaction.

LC-MS analysis of the 60 °C sample using pyruvic acid was performed using LCMS-IT-TOF (Shimadzu Co.). The LC-MS components included a LC-20AD pump, a CBM-20A system controller, SPD-20A UV detector. The column used for reversed-phase LC analysis was a TSKgel ODS-80Ts (Tosoh Co.) packed with 5 μm particles (2.0 × 250 mm) and it was incubated at 40 °C.

The LC analysis consisted of the following conditions: Mobile phase A: 100% H₂O (0.1% formic acid); Mobile phase B: 80% ACN + 20% H₂O (0.1% formic acid); Flow rate: 0.200 mL/min; Gradient: 0 min 1% B, 10 min 1% B, 40 min 65% B, 40.01 min 100% B, 50 min 100% B, 50.01 min 1% B, and 60 min 1% B. Stop time: 60 min. A 100 μL of injection volume was used. The LCMS-IT-TOF was operated under the following conditions: ESI in positive mode; Detection voltage: 1.53 kV; MS scan range: 80 – 800 m/z. Both amino-acid analyses and LC-MS analyses were conducted at Toray Research Center. Inc.