Technical Comments

Comment on “The Silicate-Mediated Formose Reaction: Bottom-Up Synthesis of Sugar Silicates”

See allHide authors and affiliations

Science  20 Aug 2010:
Vol. 329, Issue 5994, pp. 902
DOI: 10.1126/science.1188697


Lambert et al. (Reports, 19 February 2010, p. 984) reported that silicate ions catalyze the formation and stabilization of four- and six-carbon sugars from simple sugars, suggesting a possible prebiotic pathway for the synthesis of biologically important sugars. Here, we show that silicate has minimal impact in these respects, especially when compared to borate minerals.

Lambert et al. (1) recently suggested that aqueous sodium silicate mediates aldol reactions between simple C1 to C3 sugars (formaldehyde, glycolaldehyde, and glyceraldehyde) to produce tetroses (C4 sugars) (erythrose and threose) and pentoses (C5 sugars: arabinose, xylose, lyxose, and ribose) and other higher sugars. To support their suggestion, they generated product mixtures from these precursor carbohydrates, both with and without silicate. They then used the differences in the mass spectra of those mixtures to conclude that silicate influenced the relative amounts and stabilities of carbohydrate products formed. From this, they concluded that silicate can guide aldol reactions between these carbohydrates and that this guidance might have helped generate carbohydrates prebiotically on early Earth, where mineral silicate was undoubtedly present.

As part of work examining the role of minerals containing borate in guiding similar prebiotic chemistry (2), we performed essentially the same experiments some time ago. Our work was motivated in part by the knowledge that both silicate and aluminate are more abundant on Earth than borate (3), a 2004 report by Lambert et al. containing mass spectral evidence that certain cyclic carbohydrates bind silicate (4), and a 2003 report from Kinrade et al. showing that silicate binds ribose (5).

As analytical methods, we used 13C-labeled starting materials and measured the loss of their 13C resonances relative to 13C signals from standards within the sample. As discussed further below, these analytical methods are different from, but we contend superior to, those used by Lambert et al. (1). For further analysis, we isolated acetyl and acetonide derivatives of the products. Products containing C=O groups were also quantitated by high-performance liquid chromatography of their dinitrophenylhydrazone derivatives, using ultraviolet spectroscopy to determine their relative amounts (±5%). All of our reactions are run in buffer with controlled pH, because the rates of most reactions involving carbohydrates are sensitive to pH.

Because Lambert et al. (1) did not mention a pH for their reactions, we cannot be certain that the conditions that we used to explore the impact of silicate on aldol processes were exactly the same as theirs. However, we, like Lambert et al. (1), incubated glyceraldehyde (our starting material had a 3-13C label) with glycolaldehyde in the presence of silicate (220 mM, pH 11.8 ± 0.1). Further, we studied the stabilities of carbohydrates in the presence of silicate (and other mineral anions) over a wide range of conditions. We appeared to have used the same concentrations of silicate as they used, and tested both longer and shorter times, so that we could be certain that we did not miss anything relevant to the conclusions of interest to the prebiotic community.

Our results are incompatible with the results reported in (1). At both 65°C and room temperature, we found that the major product arising from reaction of 3-13C-glyceraldehyde with glycolaldehyde in the presence of silicate (220 mM) is arabinose, after 1 hour (at the higher temperature) and 4 days (at the lower temperature). A role for silicate in guiding arabinose formation is incompatible with these results, because they can be explained entirely by the fact that arabinose is simply the most stable of the four pentoses at pH 11.8. The formation of other pentoses, including ribose, in borate buffer is already documented (2); the observation that 2-hydroxymethylerythrose is formed in the presence of borate (but not silicate) is a key item of evidence showing that borate actually does guide aldol reactions but that silicate does not; 2-hydroxymethylerythrose may be an intermediate in the borate-moderated synthesis of pentoses.

We also asked whether silicate could stabilize preformed carbohydrates against decomposition in base. Again, we observed at most only a very small impact. For example, upon incubation of 5-13C-ribose in silicate, carbonate, and borate buffers (all 200 mM at pH 11.8), after 1 hour at 65°C, 40% of the ribose stabilized by silicate remained, whereas 24% of the ribose stabilized by carbonate remained. Essentially all of the ribose stabilized by borate remained in these experiments. Separate controls suggested that carbonate does not stabilize ribose against alkaline decomposition. This allows carbonate to be used as a buffer to reproduce Lambert et al.’s negative control, rather than measuring product ratios of carbonyl compounds in an unbuffered reaction. Thus, the differences in our results cannot be attributed to differences in procedures. Instead, they suggest that any stabilization of ribose by silicate is small. Similar experiments with silicate, arabinose, and 2-hydroxymethylerythrose produced similar results.

How do we account for the contradictions between our observations and those of Lambert et al. (1)? First, it appears that their control was a reaction run in unbuffered aqueous NaOH at the same pH as their reactions with silicate. As the rates of aldol reactions are quite sensitive to pH, neither rate constants nor product estimates for such reactions should be obtained in unbuffered media. Their reported impact of silicate on stability (“less than half”) might be consistent with our results had there been an excursion in pH by 0.2 units in their control.

We also question Lambert et al.’s use of electrospray mass spectrometry to assess the amounts of products in mixtures prepared with and without silicate. Inspection of figure 2 in (1) suggests that the ions being observed are different in the mixtures being compared; the first set of ions have silicate, whereas the second do not. A comparison of the intensities of different ions determined by mass spectrometry across two different ion systems cannot provide estimates of the actual ratios of products in the two product mixtures. Therefore, we disagree with Lambert et al. (1) that silicate mediates formose-like aldol reactions. Further, even if silicate does assist in stabilization of carbohydrates in alkaline solution, the assistance appears to be small compared with that offered by borate.

More positively, borate might be viewed as stabilizing carbohydrates too much, preventing them from reacting further when further reaction is desired. Silicate, through its weaker interaction, does not have this disadvantage. As the central issue in the prebiotic chemistry of carbohydrates arises from the need to have reactions occur when they are wanted (to form higher carbohydrates or convert them to nucleosides, for example) without having undesired reactions occur at the same time, Lambert et al.’s observations are important. Other mineral ions, including molybdate, might also help manage the need for some (but not too much) reactivity in prebiotic carbohydrate synthesis, as molybdate interconverts branched and unbranched carbohydrates without the need for intermediate aldol fragmentation.


View Abstract

Navigate This Article