Hydroxyapatite: catalyst for a one-pot pentose formation

K. Usami a and A. Okamoto *ab
aDepartment of Chemistry and Biotechnology, Graduate School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan. E-mail: okamoto@chembio.t.u-tokyo.ac.jp
bResearch Center for Advanced Science and Technology, The University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8904, Japan

Received 17th August 2017 , Accepted 20th September 2017

First published on 27th September 2017

One of the possible synthetic routes to pentoses is the formose reaction pathway from C1 and C2 carbon sources, but preferential ribose generation in a one-pot reaction without any control of conditions has not been reported. We have tested a one-pot pentose formation and analyzed the products and mechanism in the reaction, using 1H-NMR and mass spectrometry. Hydroxyapatite (HAp), which consists of phosphate and calcium ions, worked continuously for cross-aldol reactions and Lobry de Bruyn–van Ekenstein transformations to yield ribose from formaldehyde and glycolaldehyde. The continuous reaction proceeds in one pot in hot water only in the presence of a HAp catalyst, without any fine pH control or any complicated condition control at each reaction step. Ribose production by HAp may be a reason why a pentose backbone was incorporated into nucleic acids in the prebiotic world.


Ribose, which builds nucleic acid backbones, is the most fundamental pentose. This monosaccharide might be created from C1 or C2 carbon sources in a “prebiotic environment.” For example, formaldehyde (C1) was found in the dust within the coma of a comet1 and glycolaldehyde (C2, a dimer of formaldehyde) was found in the gas around a forming star.2 One of the possible synthetic routes to ribose is the “formose reaction” pathway suggested by Butlerow in 1861,3 for which a reaction mechanism was proposed by Breslow in 1959.4 This is a sequence of cross-aldol reactions and Lobry de Bruyn–van Ekenstein transformations from formaldehyde and glycolaldehyde (Scheme 1). In several attempts at reactions from formaldehyde and glycolaldehyde, strongly alkaline aqueous solutions containing calcium ions have been used, and the production of tar material, including some tetroses, pentoses and hexsoses, has been reported.4,5 Such a high pH condition containing concentrated calcium ions rapidly gives the reaction products, but a number of side reactions are induced such as self-aldol condensation, excessive condensation, and branched aldol formation. Pentose formation using borate mineral has also been reported.6 In this pathway, stabilized glyceraldehyde directly reacts with glycolaldehyde to afford pentose without passing through dihydroxyacetone. Formation of tar material was avoided in the borate method, but undesirable tetroses and hexoses as well as other pentoses (including arabinose, xylose and lyxose) were formed. There have been several attempts to generate nucleosides or nucleotides so far,6–9 but preferential ribose generation in a one-pot reaction without any control of condition has not been reported. In addition, mass spectrometry after gas or liquid chromatography has been used mainly in the product analysis in these attempts, although a variety of isomers would be expected in the reaction mixture. Here, we report that an ore powder catalyzes the production of ribose from C1 and C2 carbon sources in one pot in hot water. This catalyst worked continuously for cross-aldol reactions and Lobry de Bruyn–van Ekenstein transformations and gave ribose preferentially rather than the other monosaccharides.
image file: c7ob02051a-s1.tif
Scheme 1 Formose reaction.

Results and discussion

We have noted that phosphate ions play a key role in the synthesis and degradation of sugar molecules in biology. For example, phosphate ions catalyze the degradation of sugars to glyceraldehyde through several reversible processes in the glycolytic pathway. Phosphate ions work as a pH control in the microenvironment, in the coordination and organization of metal ions, as a proton donor and acceptor, as well as being a good leaving group. Apatite is a material containing phosphate and calcium ions that occurs in igneous deposits across the world as well as sedimentary deposits,10,11 and hydroxyapatite (HAp) is a major form among apatites, which can also be observed in our teeth and bones. Calcium ions are ordered by phosphate ions at the apatite surface; thus, we expect that calcium ions on the apatite surface, particularly HAp, catalyze aldol reactions with pH control of the microenvironment around the reaction site by surface phosphates. If so, HAp could well support the progress of the formose reaction. Indeed, three examples for sugar formation using apatites have been reported, but they were carried out under alkaline conditions12,13 and the analysis of the reaction products was insufficient.12–14

We added a catalytic amount of HAp powder (ϕ 30 μm, Fig. 1a) to an aqueous solution containing formaldehyde and glycolaldehyde. The pH of the solution was about 7, which is not controlled with any buffers. The solution was heated at 80 °C for 128 h (Table 1). The mixture after heating was still clear and the tar material observed in the reaction using calcium hydroxide was negligible in this reaction. HAp was removed from the mixture by filtration and then water was also removed in vacuo. The crude products were not easily analyzed by NMR and/or mass spectrometry because a variety of forms of starting aldehydes and products, such as monomers, dimers, and hydrates, were contained in the reaction mixture. To make treatment and analysis of the products easier, we acetylated the products quantitatively with acetic anhydride and analyzed the acetylated products. Acetylated β-ribofuranose (0.28% yield) was included in the reaction mixture, which was identified using 1H-NMR (Fig. 1b) and electrospray ionization mass spectrometry (ESI-MS, [M + Na]+ calcd 341.0843, found 341.0848) (Fig. S1). As seen in the mass spectrometry data, the mass corresponding to pentoses appeared clearly as a major product. The weak mass signal derived from hexoses was also observed ([M + Na]+ calcd 413.1054, found 413.1051), but we could not find any hexoses such as glucose and fructose in the mixture in the 1H-NMR spectroscopy measurement. This ribose formation did not proceed in the absence of HAp. HAp worked as a key catalyst for ribose formation from formaldehyde and glycolaldehyde.

image file: c7ob02051a-f1.tif
Fig. 1 Production of ribose through successive HAp-catalyzed reactions. (a) Powdered HAp (ϕ 30 μm) used in this reaction and the crystal structure of HAp. (b) 1H-NMR chart (600 MHz, dimethylsulfoxide-d6) of the reaction products after acetylation treatment.
Table 1 Conditions of each HAp-catalyzed reaction step
(mmol)   (mmol)   Main products HAp (mg) Temp. (°C) Time (h)
a Ribose is one of the products in the reaction mixture (0.28%). Dihydroxyacetone was also included in the reaction mixture (the singlet signal at 4.85 ppm in Fig. 1b).
Formaldehyde 1 + Glycolaldehyde 2 Ribosea 60 80 128
Formaldehyde 10 + Glycolaldehyde 1 Dihydroxyacetone 60 80 42
Glyceraldehyde 1 Dihydroxyacetone 60 80 25
dihydroxyacetone 1 + Glycolaldehyde 1 Ribulose + ribose 60 80 129
Ribulose 1 Ribose 30 80 90

To elucidate how HAp can catalyze ribose production and which pathway the ribose is produced through, we investigated each reaction step (Table 1). As the first step, glycolaldehyde was mixed with an excess amount of formaldehyde in the presence of a catalytic amount of HAp at 80 °C for 42 h. The reaction proceeded almost quantitatively and the main product in this short-time reaction was identified as dihydroxyacetone (Fig. 2a). If the path of the formose reaction is adopted, then dihydroxyacetone should be produced through the generation of a triose glyceraldehyde by the cross-aldol reaction between formaldehyde and glycolaldehyde, and subsequent isomerization from aldehyde to ketone. In fact, heating a solution of glyceraldehyde in water in the presence of HAp also gave the isomerized product, dihydroxyacetone, almost quantitatively (Fig. 2b), suggesting that dihydroxyacetone was made via the temporary formation of glyceraldehyde from formaldehyde and glycolaldehyde. Consumption of the starting aldehydes was not observed in these reactions in the absence of HAp followed by acetylation (Fig. S2). In addition, the formation of threose or erythrose via the homoaldolization of glycolaldehyde or the formation of fructose or dendroaldose via the homoaldolization of glyceraldehyde was not detected in these reactions (Fig. S3), which are quite different from the complex product mixture detected in the reactions in calcium hydroxide solution.5 Therefore, the catalytic surface of HAp powder is essential for both the first aldol condensation and the subsequent isomerization. The quantitative production of dihydroxyacetone on heating in the presence of HAp also suggests that the lifetime of glyceraldehyde, one of the intermediates in the formose reaction, is short under such reaction conditions, and the product dihydroxyacetone works as a key player in the next reaction step, although this is quite different from the previous reports that glyceraldehyde is used as an important intermediate in ribose formation.6,15,16

image file: c7ob02051a-f2.tif
Fig. 2 Analysis of each reaction step from formaldehyde and glycolaldehyde to dihydroxyacetone in the formose reaction pathway. (a) 1H-NMR spectroscopy of the products after heating the mixture of glycolaldehyde and excess formaldehyde in the presence of HAp and subsequent acetylation treatment. (b) 1H-NMR spectroscopy of the products after heating glyceraldehyde in the presence of HAp and subsequent acetylation treatment. * in (b) denotes ethyl acetate.

Formaldehyde would be released as a vapor from the reaction mixture by continuous heating and glycolaldehyde and dihydroxyacetone would be left behind in the hot water with HAp. The dihydroxyacetone in a glycolaldehyde-containing solution gave ribulose in the presence of the HAp catalyst. The production of ribulose and ribose was confirmed when the 1[thin space (1/6-em)]:[thin space (1/6-em)]1 mixture of dihydroxyacetone and glycolaldehyde was heated at 80 °C for 129 h (Fig. 3). This aldol reaction required a longer reaction time because of the use of the larger molecule, dihydroxyacetone, than in the first aldol reaction; thus, this reaction becomes the rate-determining step for the whole formose reaction. The reaction products after subsequent acetylation were the acetylated pentoses (Fig. S4) including ribulose (8%), ribose (20%), and xylulose (11%) as well as the starting material dihydroxyacetone (20%) and the oxidative decomposition product glycolic acid (40%). The pH of the reaction mixture decreased to about 2–3 during the long reaction time. The cross-aldol reaction between dihydroxyacetone and glycolaldehyde provided the pentoses with the ratio 7[thin space (1/6-em)]:[thin space (1/6-em)]3 of ribulose (plus ribose) and xylulose. The stereoselectivity of pentose formation can be explained as follows (Fig. 4): the formation of ribulose takes place through the formation of a thermodynamically stable E-calcium enolate, which undergoes a cross-aldol reaction through a chelated six-membered chair-like transition state under the long-time heating condition in the presence of HAp (trans-decalin-like formation). However, the formation of xylulose may be the result of a cross-aldol reaction through a Z-enolate in the cis-decalin-like transition state. The distances between two chelated calcium ions were estimated as ca. 5.3 and 4.7 Å in the E- and Z-enolate transition states, respectively, when the calcium ionic radius was 0.99 Å.17 The distance between two columnar calcium ions in the HAp surface was calculated as 5.4 Å (ref. 17 and 18) (Fig. 1a), which is very close to the distance of calcium ions in the E-enolate transition state, suggesting that the catalytic function of HAp works preferentially for ribulose formation.

image file: c7ob02051a-f3.tif
Fig. 3 1H-NMR spectroscopy of the products after heating of the glycolaldehyde and dihydroxyacetone mixture in the presence of HAp and subsequent acetylation treatment.

image file: c7ob02051a-f4.tif
Fig. 4 Proposed chelating structures in the cross-aldol reaction and subsequent aldose–ketose isomerization in the presence of two chelated calcium ions.

Isomerization from ribulose to ribose is required for the final step. Transformation from aldose to ketose is thermodynamically easy. In fact, the isomerization from glyceraldehyde to dihydroxyacetone proceeded quantitatively in the presence of HAp in a short time, as described above. However, the isomerization from ribulose to ribose is a structural change from ketose to aldose, and it might be unfavorable in view of the thermodynamics. Indeed, ribulose was 21.3 kJ mol−1 more stable than ribose, as determined by a B3LYP/6-31G* calculation (dihydroxyacetone was 8.6 kJ mol−1 more stable than glyceraldehyde) (Fig. 5a). However, this is a discussion based on the molecular energies of the open-chain forms of ribulose and ribose. The relationship between the energies of the ring forms of ribulose and ribose was reversed. Cyclized ribose (β-ribofuranose) was 23.0 kJ mol−1 more stable than cyclized ribulose and 69.7 kJ mol−1 more stable than linear ribose. Actually, this isomerization from ketopentose to aldopentose did take place, although its yield was not high. Ribulose isomerized to ribose (7%) and arabinose (3%) in 90 h heating in the presence of HAp (Fig. 5b). Xylulose was also converted into lyxose (4%) and xylose (1%) (Fig. S5). These isomerization steps did not occur in the absence of HAp at all. Although the yield of the isomerization from a starting material ribulose to ribose was 7%, the isomerization yield from ribulose to ribose was calculated to be 72% when the reaction started from dihydroxyacetone and glycolaldehyde as described above. When ribulose was used as the starting material, the conformation of ribulose must be changed from the cyclic form to the open-chain form prior to the aldose–ketose isomerization step. However, in the case of the reaction starting from dihydroxyacetone and glycolaldehyde, the ribulose as the reaction intermediate is already in the open-chain form chelating two calcium ions (Fig. 4). Because the ring-opening step is not required, the open-chain form of ribulose may directly convert into cyclic ribose through isomerization and cyclization, maintaining the HAp calcium-chelating trans-decalin-like transition state. In addition, the isomerization of ribulose is easier than that of xylulose. In fact, conversion of xylulose into the corresponding aldoses lyxose or xylose was negligible when the reaction started from dihydroxyacetone and glycolaldehyde. Calcium ions on the surface of HAp play a critical role in effective and stereoselective ribose formation from dihydroxyacetone and glycolaldehyde. The cyclized ribose was stable even in the presence of HAp (80 °C, 10 days, Fig. S6).

image file: c7ob02051a-f5.tif
Fig. 5 Analysis of the isomerization step from ribulose to ribose in the formose reaction pathway. (a) Energy diagram of aldose–ketose isomerization (B3LYP/6-31G*). (b) 1H-NMR spectroscopy of the products after heating of ribulose in the presence of HAp and subsequent acetylation treatment. # in (b) denotes β-ribofuranose.


As mentioned above, ribose was produced from simple carbon sources formaldehyde and glycolaldehyde by catalysis of the HAp surface, which consists of phosphate and calcium ions. This is the first example where HAp promotes the formose reaction at a lower pH and ribose is formed more preferentially than the other pentoses, tetroses and hexoses. HAp worked continuously for cross-aldol reactions and Lobry de Bruyn–van Ekenstein transformations, utilizing the effective positioning of columnar calcium ions on the surface of HAp. Although the reaction still included low-yielding steps and a large amount of glycolic acid was formed under an aerobic long-term heating condition, it proceeds in one pot in hot water in the presence of the HAp catalyst, without any fine pH control or any complicated reaction control at each reaction step. The major products of this HAp-catalyzed reaction from formaldehyde and glycolaldehyde were pentoses represented by ribose. The HAp-catalyzed formose reaction pathway may give us a hint for understanding the inevitability of ribose in nucleic acids.



Hydroxyapatite (powder form) was purchased from Sigma-Aldrich (St Louis, USA, 289396) and used without further purification.

HAp-catalyzed reaction and acetylation

The compounds as shown in Table 1 were added to water (2 mL). The amounts of the starting materials in each reaction are shown in Table 1. HAp powder, 60 mg, was added to the mixture and heated at 80 °C. After the reaction time shown in Table S1, the mixture was filtered to remove HAp powder and then the filtrate was concentrated in vacuo. Pyridine (2 mL) and acetic anhydride (2 mL) were added to the residue and the mixture was stirred at 80 °C for 15 min. The reaction mixture was concentrated in vacuo and then extracted with ethyl acetate and water. The combined organic layer was washed with brine and dried over magnesium sulfate. After being filtered and concentrated in vacuo, the residue was analyzed with 1H-NMR (600 MHz, dimethylsulfoxide-d6) and ESI-MS spectroscopy. Standard 1H-NMR and ESI-MS data used for identification of the reaction products in each step are as follows: 1,3-Dihydroxyacetone diacetate.δ = 4.85 (s, 4H), 2.10 (s, 6H); ESI-MS m/z [2 M + Na]+ calcd 371.0949, found 371.0955; Ribulose 1,3,4,5-tetraacetate.δ = 5.46 (d, 1H, J = 4.4 Hz), 5.33 (dd, 1H, J = 10.6, 4.7 Hz), 4.97 (d, 2H, J = 6.7 Hz), 4.20 (d, 2H, J = 4.7 Hz), 2.15 (s, 3H), 2.11 (s, 3H), 2.04 (s, 3H), 1.99 (s, 3H); ESI-MS m/z [M + Na]+ calcd 341.0843, found 341.0835; β-Ribofuranose 1,2,3,5-tetraacetate.δ = 6.02 (s, 1H), 5.25 (s, 2H), 4.32–4.30 (m, 2H), 4.06 (dd, 1H, J = 12.2, 5.2 Hz), 2.10 (s, 3H), 2.07 (s, 3H), 2.05 (s, 3H), 2.04 (s, 3H); ESI-MS m/z [M + Na]+ calcd 341.0843, found 341.0834; Glycolaldehyde acetate. This is a mixture of dimeric isomers (isomer A 39%, B 27%, C 23%, D 11%). δ = 6.28 (d, 1H, J = 3.8 Hz, D), 6.35 (dd, 1H, J = 4.1, 1.2 Hz, C), 5.88 (s, 2H, A), 5.79 (dd, 2H, J = 5.3, 2.9 Hz, B), 5.36 (t, 1H, J = 3.6 Hz, C), 5.26 (t, 1H, J = 4.2 Hz, D), 4.22 (d, 2H, J = 4.7 Hz, D), 4.15 (dd, 1H, J = 9.4, 4.4 Hz, C), 4.12 (d, 1H, J = 2.1 Hz, D), 4.11 (d, 1H, J = 3.2 Hz, C), 4.10 (d, 1H, J = 3.8 Hz, C), 4.07 (dd, 2H, J = 12.9, 1.5 Hz, A), 3.96 (dd, 1H, J = 9.7, 3.8 Hz, D), 3.92 (d, 1H, J = 9.4 Hz, C), 3.84 (dd, 2H, J = 12.0, 2.9 Hz, B), 3.74 (dd, 2H, J = 12.1, 5.6 Hz, B), 3.57 (d, 2H, J = 12.6 Hz, A), 2.09 (s, 6H); ESI-MS m/z [2 M + Na]+ calcd 277.0526, found 277.0528; Glyceraldehyde diacetate.δ = 5.77 (d, 1H, J = 7.9 Hz), 4.27 (dd, 2H, J = 12.3, 4.4 Hz), 4.16 (dd, 2H, J = 12.4, 2.8 Hz), 3.94–3.91 (m, 2H), 2.15 (s, 6H), 2.09 (s, 6H); ESI-MS m/z [2 M + Na]+ calcd 371.0949, found 371.9040; Xylulose 1,3,4,5-tetraacetate.δ = 5.53 (s, 2H), 4.97 (d, 1H, J = 17.9 Hz), 4.88 (d, 1H, J = 17.9 Hz), 4.23 (dd, 1H, J = 11.7, 4.4 Hz), 4.11 (dd, 1H, J = 11.7, 6.7 Hz), 2.17 (s, 3H), 2.09 (s, 3H), 2.02 (s, 3H), 1.99 (s, 2H); ESI-MS m/z [M + Na]+ calcd 341.0843, found 341.0841; Arabinose 1,2,3,5-tetraacetate. This is a mixture of isomers (isomer A 29%, B 20%, C 19%, D 18%, E 14%). δ = 11.98 (s, 1H, C), 6.26 (d, 1H, J = 4.7 Hz, E), 6.18 (d, 1H, J = 3.5 Hz, A), 6.09 (s, 1H, D), 5.74 (d, 1H, J = 7.6 Hz, B), 5.32–3.77 (5H), 2.22–1.91 (12H); ESI-MS m/z [M + Na]+ calcd 341.0843, found 341.0843; Lyxose 1,2,3,5-tetraacetate. This is a mixture of isomers (isomer A 45%, B 29%, C 18%, D 8%). δ = 6.22 (d, 1H, J = 3.9 Hz, D), 6.07 (d, 1H, J = 2.6 Hz, C), 6.03 (d, 1H, J = 2.0 Hz, B), 5.89 (d, 1H, J = 3.0 Hz, A), 5.51–3.59 (5H), 2.22–1.91 (12H); ESI-MS m/z [M + Na]+ calcd 341.0843, found 341.0839; Xylose 1,2,3,5-tetraacetate. This is a mixture of isomers (isomer A 47%, B 27%, C 18%, D 4%, E 4%). δ = 11.98 (s, 1H, C), 6.32 (d, 1H, J = 4.4 Hz, E), 6.12 (d, 1H, J = 3.5 Hz, A), 6.02 (s, 1H, D), 5.82 (d, 1H, J = 7.3 Hz, B), 5.40–3.63 (5H), 2.22–1.91 (12H); ESI-MS m/z [M + Na]+ calcd 341.0843, found 341. 0841; Glycolic acid acetate.δ = 4.53 (s, 2H), 2.08 (s, 3H).

All standard reagents were purchased from Sigma-Aldrich (St Louis, USA), Tokyo Chemical Industry (Tokyo, Japan), or Wako Pure Chemical Industries (Osaka, Japan) and acetylated without further purification.

Reaction yields were calculated from the area of the characteristic peak of each product: dihydroxyacetone diacetate, δ 4.85 (s, 4H); ribulose 1,3,4,5-tetraacetate, δ 5.46 (d, 1H, J = 4.4 Hz); xylulose 1,3,4,5-tetraacetate, δ 5.53 (s, 2H); β-ribofuranose 1,2,3,5-tetraacetate, δ 6.02 (s, 1H); arabinose 1,2,3,5-tetraacetate, δ 6.18 (d, 1H, J = 3.5 Hz); xylose 1,2,3,5-tetraacetate, δ 6.12 (d, 1H, J = 3.5 Hz); lyxose 1,2,3,5-tetraacetate, δ 5.89 (d, 1H, J = 3.0 Hz); glycolic acid acetate, δ 4.53 (s, 2H).

Molecular energy diagram

The geometries of the aldoses, ketoses, and enols were optimized by means of the DFT method using the B3LYP exchange–correlation functional and the 6-31G* basis set with the aid of the Gaussian 09 W software package. During the optimization, solvent effects were incorporated using the polarizable continuum model with the keyword “SCRF = (solvent = water)” by which all the parameters for cavity creation are optimized for water solution.

Conflicts of interest

There are no conflicts to declare.


We wish to thank Dr Hirohiko Houjou and Makoto Isogai (UTokyo) for advice regarding the calculation of molecular energies. This work was supported by a Grant-in-Aid for Scientific Research on Innovative Areas “Hadean Bioscience” (15H01058) of The Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.

Notes and references

  1. M. A. Cordiner, A. J. Remijan, J. Boissier, S. N. Milam, M. J. Mumma, S. B. Charnley, L. Paganini, G. Villanueva, D. Bockelée-Morvan, Y.-J. Kuan, Y.-L. Chuang, D. C. Lis, N. Biver, J. Crovisier, D. Minniti and I. M. Coulson, Astrophys. J. Lett., 2014, 792, L2 Search PubMed .
  2. J. K. Jørgensen, C. Favre, S. E. Bisschop, T. L. Bourke, E. F. van Dishoeck and M. Schmalzl, Astrophys. J. Lett., 2012, 757, L4 Search PubMed .
  3. A. Butlerow, Justus Liebigs Ann. Chem., 1861, 120, 295–298 Search PubMed .
  4. R. Breslow, Tetrahedron Lett., 1959, 1, 22–26 Search PubMed .
  5. G. Harsch, H. Bauer and W. Voelter, Liebigs Ann. Chem., 1984, 623–635 Search PubMed .
  6. A. Ricardo, M. A. Carrigan, A. N. Olcott and S. A. Benner, Science, 2004, 303, 196 Search PubMed .
  7. F. R. Bowler, C. K. W. Chan, C. D. Duffy, B. Gerland, S. Islam, M. W. Powner, J. D. Sutherland and J. Xu, Nat. Chem., 2013, 5, 383–389 Search PubMed .
  8. S. Becker, I. Thoma, A. Deutsch, T. Gehrke, P. Mayer, H. Zipse and T. Carell, Science, 2016, 352, 833–836 Search PubMed .
  9. B. J. Cafferty, D. M. Fialho, J. Khanam, R. Krishnamurthy and N. V. Hud, Nat. Commun., 2016, 7, 1–7 Search PubMed .
  10. F. Zapata and R. N. Roy, Use of phosphate rocks for sustainable agriculture, FAO Fertilizer and Plant Nutrition Bulletin 13, FAO, Rome, Italy, 2004 Search PubMed .
  11. E. H. Roux, D. H. de Jager, J. H. du Plooy, A. Nicotra, G. J. van der Linde and P. de Waal, J. South. Afr. Inst. Min. Metall., 1989, 89, 129–139 Search PubMed .
  12. C. Reid and L. E. Orgel, Nature, 1967, 216, 455 Search PubMed .
  13. A. W. Schwartz and R. M. de Graaf, J. Mol. Evol., 1993, 36, 101–106 Search PubMed .
  14. A. N. Simonov, O. P. Pestunova, L. G. Matvienko, V. N. Snytnikov, O. A. Snytnikova, Y. P. Tsentalovich and V. N. Parmon, Adv. Space Res., 2007, 40, 1634–1640 Search PubMed .
  15. H.-J. Kim, A. Ricardo, H. I. Illangkoon, M. J. Kim, M. A. Carrigan, F. Frye and S. A. Benner, J. Am. Chem. Soc., 2011, 133, 9457–9468 Search PubMed .
  16. S. A. Benner, H.-J. Kim and M. A. Carrigan, Acc. Chem. Res., 2012, 45, 2025–2034 Search PubMed .
  17. K. Momma and F. Izumi, Crystallogr. Comput., IUCr Newslett., 2006, 7, 106–119 Search PubMed .
  18. K. Sudarsanan and R. A. Young, Acta Crystallogr., 1969, B25, 1534–1543 Search PubMed .


Electronic supplementary information (ESI) available: Mass and NMR charts. See DOI: 10.1039/c7ob02051a

This journal is © The Royal Society of Chemistry 2017