Introduction

One of the most interesting questions about the prebiotic world has to do with homochirality—how did the L amino acids and D sugars form on Earth so polypeptides and nucleic acids could have well-defined structures, not as racemic or diastereomeric mixtures? We have described how meteoritic α-methyl amino acids, arriving on Earth with modest enantioexcesses of the S configuration, can generate normal L amino acids by a decarboxylative transamination under prebiotic conditions (Breslow and Levine 2006; Levine et al. 2008; Breslow et al. 2010; Breslow 2011). We and others have shown how the resulting small excesses can be amplified under either equilibrium (Morowitz 1969; Breslow and Levine 2006; Klussman et al. 2006, 2007; Hein and Blackmond 2012) or kinetic (Breslow et al. 2010; Breslow 2011) conditions to high enantioexcesses of the L configuration in water solution. We assume that the prebiotic processes occurred in water solution at moderate temperatures on the surface of the Earth where meteorites had landed.

In a preliminary publication we also described the formation of D-glyceraldehyde by aldol addition of formaldehyde to glycolaldehyde catalyzed by amino acids (Breslow and Cheng 2010). We found that all the L amino acids we examined preferentially formed an excess of D-glyceraldehyde with one exception; L-proline catalyzed the preferential formation of an excess of L-glyceraldehyde. We suggested that L-proline may have been present in small amounts prebiotically; it results from further reactions of L-glutamic acid. D-Glyceraldehyde is the likely precursor of other D sugars, building on additions to its aldehyde group, so our conclusion was that in spite of the result with L-proline it seemed likely that the D sugars were derived by catalysis from the L amino acids, which were originally derived from the meteoritic α-methyl amino acids. Glycolaldehyde is formed from formaldehyde and simple bases such as calcium hydroxide by the formose reaction (Breslow 1959) so we assume that there were places where formaldehyde was present on Earth. It is difficult to imagine how sugars, polymers of formaldehyde, could have been formed without it.

We also studied the amplification of our observed modest excesses of D-glyceraldehyde by the water evaporation method that was originally pioneered by Morowitz and used by him, Blackmond, and us to amplify small enantioexcesses of amino acids. We also used this technique to amplify small enantioexcesses of some D nucleosides to high levels of selectivity. We were able to amplify small excesses of D-glyceraldehyde to high levels with the same water evaporation technique. (Breslow et al. 2010). It depends on the fact that racemates are frequently less soluble than homochiral compounds, so evaporation of solutions with low enantioexcesses (ee’s) can concentrate them to very high ee’s as the racemates precipitate. This does not produce more of the homochiral compound, it simply concentrates the enantiomer in excess so the solutions can contain as much as 90 % or more of the L amino acid or the D nucleoside or D-glyceraldehyde along with minor amounts of the other enantiomer. Such high ee’s would presumably be enough for biology to start using the major enantiomer preferentially.

Blackmond had been studying L-proline catalysis of other reactions under mildly acidic conditions, and saw that under basic conditions the product configuration was reversed (Blackmond et al. 2010). She more recently reported that with our conditions L-proline afforded D-glyceraldehyde from glycolaldehyde and formaldehyde with added tetrabutylammonium acetate, not the L-glyceraldehyde formed without it (Hein and Blackmond 2012). Thus we have now examined the synthesis of glyceraldehyde at three defined pH regions, and with both primary and secondary chiral amino acids, to more fully understand the conditions that form D glyceraldehyde and thus the higher sugars derived from it. We find that L-proline at a measured pH of 7.6 reverses its preference to form D-glyceraldehyde 2.0 ± 0.5 % ee, as described below. However, if the products from the other L amino acids also reversed at higher pH, raising the pH would not bring all the amino acids, including L-proline, into line to form D-glyceraldehyde.

We have now seen that the preference for the formation of D-glyceraldehyde with catalysis by L amino acids with primary amino groups at low pH reverses to a small preference for L-glyceraldehyde at higher pHs (Table 1). We also see this reversal of stereochemistry at higher pHs when secondary L amino acids such as N-methyl-L-leucine were used as catalysts, just as with L-proline. Herein we report the detailed study of the glyceraldehyde synthesis catalyzed by amino acids in three different pH regions.

Table 1 Amino acid catalyzed glyceraldehyde synthesis under different pH conditions
Full size table

Methods and Materials

Formaldehyde (37 % solution in water), glycolaldehyde dimer, N-methyl amino acids, amino acids, 2,4–dinitrophenylhydrazine (stabilized with 33 % water), and HPLC grade hexane were purchased from Sigma Aldrich. Sodium bicarbonate, sodium carbonate and HPLC grade ethanol were purchased from Fisher. All chemicals and solvents were used without further purification.

In a typical reaction at pH 2.9–4.2, 1 mmol of glycolaldehyde, 40 mmol of formaldehyde (37 % in water) and 1 mmol of the amino acid were dissolved in 16 mL of water and stirred at room temperature in air for 3 days. At the end the pHs had dropped by a few tenths in all cases, possibly from further air oxidation of the formaldehyde. To the reaction mixture (1/6 of the total volume), were added 2,4-dinitrophenylhydrazine (10 mmol) and water (7 mL), and the reaction mixture was stirred at 50 °C for 8 h. The solid material was filtered away, and the filtrate was concentrated. The crude product was subjected to successive preparative thin layer chromatographies (First solvent system: 100 % ethyl acetate, Rf = 0.7; Second solvent system: 5 % methanol in dichloromethane, Rf = 0.4) to obtain pure glyceraldehyde 2,4-dinitrophenylhydrazone. The enantiomeric excess of glyceraldehyde hydrazone was analyzed by HPLC.

A Waters 600 HPLC system with a Waters 996 photodiode array spectrophotometer was used for HPLC analysis. TLC silica gel 60F254 (EMD 5715-7) was used for preparative TLC. HPLC conditions: A Chiralpak AD column was used to measure the enantiomeric excess. A mobile phase solvent hexane to ethanol gradient with 0.5 mL flow rate for 120 min was used and product dinitrophenylhydrazones (DNPs) were monitored at 350 nm. Under these conditions typical retention times for L-glyceraldehyde-DNP and D-glyceraldehyde-DNP are 93 and 122 min respectively. Honda et al. observed E & Z isomers of glyceraldehyde hydrazones in their HPLC conditions (Honda and Kakehi 1978). However, we did not observe mixtures of geometrical isomers in our HPLC conditions.

We did not use buffers since they were probably absent prebiotically, but the added amino acids have some buffering capacity. The pH of the aldol condensation of commercial formaldehyde and glycolaldehyde was found to be 2.9 to 4.2 depending upon the amino acids. We had a low pH without addition of acid since commercial formaldehyde (Sigma Aldrich 252549) has a measured pH of 3.4, from the presence of formic acid produced by air oxidation. Thus we carried out the glyceraldehyde synthesis with its initial acidic pH, as previously (Breslow and Cheng 2010) and then at higher pH’s produced by addition of base. In the low pH cases, where the interesting results were obtained, nothing extra was added to the solution, and the pH was generated by the small amount of formic acid in the formaldehyde. The reactions were performed in triplicate with small standard deviations, except for three examples done only once and only at an acidic pH. Of course the formaldehyde on prebiotic earth may not have contained formic acid as an impurity, but we do not know what the pH was of a solution on prebiotic earth so we explored all three pH regions.

Results and Discussion

In agreement with our earlier results we found that all the primary L amino acids under acidic conditions (pH 2.9–4.2) produced an excess of D-glyceraldehyde (Table 1, entry 1, 3–5). In particular, L-glutamic acid produced D-glyceraldehyde with 34.8 % ee (entry 1). As a check on the procedure we also used D-glutamic acid and saw a reverse of essentially the same ratio, L-glyceraldehyde in 33.8 % ee (entry 2).

When the secondary amino acid L-proline was used at low pH, L-glyceraldehyde was formed preferentially rather than D-glyceraldehyde, with 20.4 % ee (entry 6). To establish whether the secondary amino group explains the different behavior of L-proline in our system we examined a number of N-methylated L amino acids that lacked the ring structure of L-proline but also had secondary amino groups. We found that N-methyl-L-leucine, N-methyl-L-valine and N-methyl-L-glutamic acid all preferentially catalyzed the formation of L-glyceraldehyde (entries 7–9) at low pH, not the D-glyceraldehyde catalyzed without the N-methyl groups. However, proline afforded a larger ee than these other secondary amines, indicating that the cyclic character of proline also plays a role.

Then we studied the same reactions in the pH region 6.6–7.6, adjusted by the addition of sodium bicarbonate. In this pH region the primary L amino acids produced an excess of L-glyceraldehyde, not D-glyceraldehyde, with the modest enantiomeric excesses of 1.5 to 4.4 % (Table 1, entry 1,3–5). Although the enantioselectivity is quite a bit lower than that which resulted under acidic conditions, the reversal of chiral preference was clearly observed. With L-proline under non-acidic conditions the major enantiomer was now D-glyceraldehyde in 2 % ee (entry 6), consistent with the magnitudes of the other ee’s at measured neutrality. When N-methyl-L-leucine (secondary amino group) was used as a catalyst at higher pHs we also saw the formation of excess D-glyceraldehyde (entry 7) rather than the L-glyceraldehyde under acidic conditions. We did not do the higher pH studies with N-methyl-L-glutamic acid and N-methyl-L-valine.

When we performed the reactions at pH 8.7–9.8, adjusted by the addition of sodium carbonate, the results we obtained were very similar to the results in reactions performed at pH 6.6–7.6, but with lesser enantioselectivity.

We examined the recovered amino acid L-valine from reactions performed at both acidic and basic conditions, and saw no detectable racemization of the amino acid. We also ran the aldol reactions catalyzed by L-glutamic acid for 2 days and 4 days under acidic conditions and saw the formation of glyceraldehyde with 35.1 % ee and 33.6 % ee respectively, indicating that there was no appreciable racemization of the product glyceraldehyde under those conditions.

The chiral inductions and the reversal of chiral preferences between primary and secondary amines and with acidic versus non-acidic pHs could be explained by the following proposed mechanisms and depicted transition states (Scheme 1) modeled on those proposed by Barbas (Sakthivel et al. 2001; Ramasastry et al. 2007) for general aldol-type condensations with amino acid catalysts. Under acidic conditions (pH 2.9–4.2) primary L amino acids condense with glycolaldehyde 1 to form the Z-enamine favored by N-H–O hydrogen bonding. Formaldehyde is brought into the Re-face of the enamine through another hydrogen bond by the carboxylic acid group in the transition state TS-1 that leads to the formation of D-glyceraldehyde (+)-2. With the same reaction under basic conditions, formaldehyde approaches the Si-face of the enamine due to the absence of hydrogen bonding between carboxylate group and formaldehyde in the preferred transition state TS-3, and electrostatic repulsion by the carboxylate ion of the developing alkoxide anion, to give L-glyceraldehyde (−)-2.

Scheme 1
scheme 1

Proposed preferable transition states for the formation of glyceraldehyde

Full size image

When a secondary amino acid catalyzes the reaction of glycolaldehyde 1 and formaldehyde under basic conditions, in the absence of hydrogen bonding in the preferred transition state TS-2, formation of the E-enamine and approach of formaldehyde through the Re-face of the enamine are directed through steric bulk for the formation of D-glyceraldehyde (+)-2. Under acidic conditions, formaldehyde approaches the Si-face of the E-enamine because of hydrogen bonding between the carboxylic acid group and formaldehyde in the preferred transition state TS-4 that leads to the formation of L-glyceraldehyde (−)-2.

We performed our reactions under mild conditions, and to very low levels of completion to avoid further reactions of the glyceraldehyde with formaldehyde to form higher sugars. Such higher sugars are also formed in the simple formose reaction at high pH, where no amino acids are present. (Breslow 1959; Delidovich et al. 2009). In our reaction with L-glutamic acid catalyst, only 2.4 % of the glycolaldehyde was consumed after 4 days. The glyceraldehyde (20 % of the glycolaldehyde consumed) and dihydroxyacetone (13 % of the glycolaldehyde consumed), detected and measured by HPLC as DNP derivatives, were the only significant carbohydrate products. We found that there was no detectable loss of glyceraldehyde when it was incubated with formaldehyde and an amino acid at low pH, so the small yield of glyceraldehyde we found in our synthesis procedure does not reflect any loss of the product. These two products plus the recovered glycolaldehyde account for 98.4 % of the original glycolaldehyde. We have not been able to identify any other products; in particular, we do not see any product from Amadori reaction of glycolaldehyde with the amino acids under our conditions, nor ethanolamine from transamination.

Weber has studied the reaction of formaldehyde with glycolaldehyde under more vigorous conditions with amino acid catalysis (pH 5.5, temperature 50 ° C), and saw a variety of products derived from glyceraldehyde (Weber 2001). He observed the consumption of 89.8 % of the glycolaldehyde after 5 days. His study did not examine any chiral induction, the focus of our work.

Conclusion

As we have shown, small excesses of D-glyceraldehyde in water can be easily amplified to high concentration by water evaporation, since D-glyceraldehyde is a syrup while the racemate is a less-soluble solid (Breslow and Cheng 2010; Breslow 2011). D-Glyceraldehyde can then go on to form higher D sugars by aldol or ketol additions to the aldehyde group. Such studies under prebiotic conditions are underway in our laboratory, where both a ketolase and a transketolase addition reaction have been produced under prebiotic conditions (Breslow and Appayee 2013). The reactions by which meteoritic amino acids let us form L amino acids, which then lead to D sugars, are the simplest versions of the origin of homochirality on Earth for which actual experiments under prebiotic conditions, and findings in meteorites, produce supporting evidence.

As we have pointed out (Breslow 2011), an alternative scenario is that in a first step an α-methyl amino acid from a meteorite, with an enantioexcess of the S-isomer, could catalyze the formation of an excess of D sugars under acidic conditions, parallel to our findings, and in a second step the sugars might catalyze the formation of the L amino acids. Pizzarello and Weber found that L isovaline (α-methyl-homoalanine) catalyzed the self-condensation of glycolaldehyde to threose with an excess of the D enantiomer (Pizzarello and Weber 2004). Thus there is evidence that the first step in the alternative scenario is reasonable. However, there is yet no good example of the second step.

Blackmond has explored alternative processes to partially resolve two racemates. In one process she performed the Sutherland reaction (Powner et al. 2009) of 2-aminooxazole with racemic glyceraldehyde and a number of L amino acids to achieve a partial excess of D-glyceraldehyde (Hein et al. 2011). In the other experiment (Hein and Blackmond 2012) she performed the same Sutherland reaction with D-glyceraldehyde and racemic proline to achieve a partial excess of L-proline. Both the Blackmond and Sutherland processes depend on the spontaneous use of a special chemical, 2-aminooxazole, that is not known to have been present and available on prebiotic earth. This contrasts with our processes for producing L amino acids and D sugars, where no extra reactant is needed except for formaldehyde and the α-methyl amino acids that come to earth on meteorites, along with α-keto acids that can be derived from meteoritic molecules.

Our previous conclusion that the formation of D-glyceraldehyde under prebiotic conditions reflects catalysis by likely prebiotic L amino acids, except L-proline, is still correct provided the reactions occurred under acidic conditions. We have no evidence of the pH on Earth wherein this prebiotic reaction could have occurred, but conclude that if our mechanism is correct, but not necessarily otherwise, it must have been acidic in order to form D-glyceraldehyde guided by the predominant L amino acids.