Synthesis of carbohydrates in mineral-guided prebiotic cycles

Article abstract of DOI:10.1021/ja201769f

One present obstacle to the "RNA-first" model for the origin of life is an inability to generate reasonable "hands off" scenarios for the formation of carbohydrates under conditions where they might have survived for reasonable times once formed. Such scenarios would be especially compelling if they deliver pent(ul)oses, five-carbon sugars found in terran genetics, and exclude other carbohydrates (e.g., aldotetroses) that may also be able to function in genetic systems. Here, we provide detailed chemical analyses of carbohydrate premetabolism, showing how borate, molybdate, and calcium minerals guide the formation of tetroses (C₄H₈O₄), heptoses (C₇H₁₄O₇), and pentoses (C ₅H₁₀O₅), including the ribose found in RNA, in "hands off" experiments, starting with formaldehyde and glycolaldehyde. These results show that pent(ul)oses would almost certainly have formed as stable borate complexes on the surface of an early Earth beneath a humid CO₂ atmosphere suffering electrical discharge. While aldotetroses form extremely stable complexes with borate, they are not accessible by pathways plausible under the most likely early Earth scenarios. The stabilization by borate is not, however, absolute. Over longer times, material is expected to have passed from borate-bound pent(ul)oses to a branched heptulose, which is susceptible to Cannizzaro reduction to give dead end products. We show how this fate might be avoided using molybdate-catalyzed rearrangement of a branched pentose that is central to borate-moderated cycles that fix carbon from formaldehyde. Our emerging understanding of the nature of the early Earth, including the presence of hydrated rocks undergoing subduction to form felsic magmas in the early Hadean eon, may have made borate and molydate species available to prebiotic chemistry, despite the overall "reduced" state of the planet.

Full text of DOI:10.1021/ja201769f

ARTICLE
pubs.acs.org/JACS
Synthesis of Carbohydrates in Mineral-Guided Prebiotic Cycles
Hyo-Joong Kim,† Alonso Ricardo,‡ Heshan I. Illangkoon,§ Myong Jung Kim,† Matthew A. Carrigan,†
Fabianne Frye,^ and Steven A. Benner*,†
†Foundation for Applied Molecular Evolution, Westheimer Institute for Science and Technology, P.O. Box 13174, Gainesville,
Florida 32604, United States
‡Ra Pharmaceuticals, One Kendall Square, Suite B14301, Cambridge, Massachusetts 02139, United States
§Department of Chemistry, University of Florida, Gainesville, Florida 32611, United States
^Department of Cell Biology, Harvard Medical School, 225 Longwood Avenue, Boston, Massachusetts 02115, United States
S Supporting Information
b
ABSTRACT: One present obstacle to the “RNA-first” model
for the origin of life is an inability to generate reasonable “hands
off” scenarios for the formation of carbohydrates under condi-
tions where they might have survived for reasonable times once
formed. Such scenarios would be especially compelling if they
deliver pent(ul)oses, five-carbon sugars found in terran genet-
ics, and exclude other carbohydrates (e.g., aldotetroses) that
may also be able to function in genetic systems. Here, we
provide detailed chemical analyses of carbohydrate premetabo-
lism, showing how borate, molybdate, and calcium minerals
guide the formation of tetroses (C4H8O4), heptoses (C7H14O7),
and pentoses (C5H10O5), including the ribose found in RNA, in
“hands off” experiments, starting with formaldehyde and glyco-
laldehyde. These results show that pent(ul)oses would almost
certainly have formed as stable borate complexes on the surface
of an early Earth beneath a humid CO2 atmosphere suffering
electrical discharge. While aldotetroses form extremely stable
complexes with borate, they are not accessible by pathways plausible under the most likely early Earth scenarios. The stabilization by
borate is not, however, absolute. Over longer times, material is expected to have passed from borate-bound pent(ul)oses to a
branched heptulose, which is susceptible to Cannizzaro reduction to give dead end products. We show how this fate might be
avoided using molybdate-catalyzed rearrangement of a branched pentose that is central to borate-moderated cycles that fix carbon
from formaldehyde. Our emerging understanding of the nature of the early Earth, including the presence of hydrated rocks
undergoing subduction to form felsic magmas in the early Hadean eon, may have made borate and molydate species available to
prebiotic chemistry, despite the overall “reduced” state of the planet.
’ INTRODUCTION
Recently, minerals have been found that cause otherwise
stable organic precursors to react to yield biomolecules.8 This
increases the diversity of organic species that might have been
present on the early Earth. In contrast with this have been studies
with minerals that prevent biomolecules (once formed) from
reacting further.9,10 This, in principle, would decrease the diversity
of organic species present on early Earth.
Which kinds of minerals prove to be more productive as we
struggle to develop our understanding of life’s origins depends on
which of two competing views of “origins” is more compelling:
(a) the view that holds that life is more likely to arise in more
complex mixtures,11 or (b) the view that increasing complexity is
Prebiotic chemistry embodies two themes. One, exemplified
by Powner et al. and M€uller et al.,1,2 explores prebiotic reactivity
by performing sequential steps of chemistry under human
control in the laboratory. The other, exemplified by historic
work of L€ow,3 L€ob,4 and Miller,5 seeks to create laboratory
conditions that resemble early Earth, hoping to constrain how
“chemical evolution” might have proceeded in these environ-
ments without continuous human intervention.
Those following the first theme must manage challenges
directed against the relevance of their work.6 Those following
the second must also do this, as well as address imperfect models
for early Earth environments, the need to observe in days what
might have taken millennia to occur naturally, and the propensity
of organic species, left unattended, to form unproductive “tars”.7
Received: February 25, 2011
Published: May 09, 2011
r
2011 American Chemical Society
9457
dx.doi.org/10.1021/ja201769f J. Am. Chem. Soc. 2011, 133, 9457–9468
|

Journal of the American Chemical Society
ARTICLE
likely to increase the number of inhibitors of life.7 The second
implies that biofriendly environments must manage reactivity to
allow some reactions to occur while preventing others.
This paper addresses these two contrasting needs in prebiotic
chemistry, for “reaction” and, in the same environment, for “no
reaction”. We focus on carbohydrates (CnH2nOn), whose in-
trinsic reactivity makes this contrast especially severe. Indeed, the
intrinsic instability of many carbohydrates under “prebiotic
conditions” has encouraged some to preclude carbohydrates as
parts of the very first genetic molecules.12
Here, we show how borate-, molybdate-, and calcium-containing
minerals guide the formation of tetroses (C4H8O4), heptoses
(C7H14O7), and pentoses (C5H10O5) in “hands off” experi-
ments. These products include the aldopentose ribose found in
RNA, the carbohydrate hypothesized by the “RNA first” model
for the origin of life to be important to start Darwinian
chemistry.13 Detailed chemical studies are presented to show
how these minerals might have guided prebiotic pathways
and cycles to give such carbohydrates as metastable mineral
complexes in a “carbohydrate world” (or one of its variant
appellations).14
turns yellow, then brown. Earlier studies19 showed that during
this lag period, C4, C5, C6, and C7 carbohydrates accumulate.
Yellowing occurs only after all HCHO is removed. Further
incubation yields mixtures that progressively become more
like “tar”.
Some have speculated that the formose process requires an
impurity in either the HCHO or the Ca(OH)2 to be initiated.
Several lines of evidence suggest that this is not the case. In
particular, HCHO obtained in two ways (from commercial
formalin and via the depolymerization of sublimed paraformal-
dehyde) and Ca(OH)2 obtained in three ways (commercial,
dissolution of calcium metal in water, and double decomposition
of CaCl2 and NaOH) gave indistinguishable lag kinetics (data
not shown).
During the lag period run in D2O, the initially formed C4, C5,
C6, and C7 carbohydrates incorporate essentially no deuterium.
This implies that once a carbonyl compound involved in their
formation containing n carbon atoms enolizes in the presence of
HCHO, it never “ketonizes” to restore the CdO species but is
rather carried on to form an nþ1 carbohydrate. Consistent with
this view, only after all of the HCHO is consumed, products
containing deuterium begin to be formed.19 These facts, plus
general principles, suggest an overall mechanistic hypothesis for
the formose process, represented by the scheme in Figure 1.
Here, a very small amount of glycolaldehyde (C2a, C2H4O2),
formed by the very rare coupling of two HCHO molecules,
begins the process of “fixing” many more HCHO molecules.
These cycles repeat just three types of reactions characteristic of
compounds having a CdO (carbonyl) group: (i) removal of a
proton (Hþ) from a carbon next to the CdO (“enolization”) to
give an enediolate, (ii) attack of the resulting enediolate nucleo-
phile on a CdO electrophile to form a new carbonꢀcarbon bond
(“aldol addition”), and (iii) retroaldol fragmentation of higher
species to generate lower carbohydrates.
These data also support four rules that unify all of the data,
including new data reported here: (a) Reverse enolization
restoring Hþ to an enediolate does not compete with aldol
addition of HCHO if HCHO is present. (b) Retroaldol reactions
extruding HCHO proceed at negligible rates. (C) Carbohydrates
that can form cyclic hemiacetals are less reactive than those that
do not. (d) Carbohydrates that can neither enolize nor cyclize are
susceptible to reduction via a Cannizzaro reaction with HCHO,
with formate being the second product.
’ RATIONALE, EXPERIMENTAL DESIGN, AND RESULTS
Model for the Prebiotic Environment. We begin by assum-
ing that formaldehyde (C1H2O1 or HCHO), the simplest
carbohydrate, was generated within a humid CO2-rich atmo-
sphere by the action of light and electrical discharge.15 This
assumption appears to be robust regardless of whether the atmo-
sphere was dominated by reduced or oxidized carbon (e.g., CH4
or CO2), having been observed in many laboratories over the
past century.4,5,15
Our model further assumes that HCHO rained from the
atmosphere onto the surface of early Earth in aqueous streams
that eroded igneous rocks containing serpentine minerals (e.g.,
peridotite). The resulting streams would have been alkaline,16
although their pH might be lowered through buffering by CO2
derived from the atmosphere.
If they were also hot (60ꢀ80 ꢀC) and contained calcium
(Ca2þ), these erosion mixtures would have inevitably provided
conditions compatible with the conversion of formaldehyde to
higher carbohydrates via the “formose process”.3,17 Ca2þ appears
essential for the first step in the formose process, which converts
two molecules of HCHO into glycolaldehyde, which subse-
quently enolizes and reacts further with formaldehyde to yield,
transiently, carbohydrates as intermediates on the way to more
complex, tarry mixtures (Figure 1).
Our model also assumes that igneous rocks exposed to erosion
contained tourmalines, a mineral that contains borate. Tourma-
lines appear in basalts as one of their lithologies.18 Tourmalines
are well known to erode easily to deliver borate to an erosion
aquifer. Here, borate would have been concentrated by evapora-
tion in the hydrosphere on any early Earth that contained
unsubmerged land surfaces.
The cause of the late rapid decline in HCHO is attributed to
the formation of dihydroxyacetone (C3k) via retroaldol frag-
mentation (dotted arrows, Figure 1) of higher carbohydrates that
are formed during the lag period. Retroaldol fragmentation is
expected to be faster in highly substituted β-hydroxycarbonyl
species. Further, under the rules summarized above, dihydroxy-
acetone cannot be formed in any way other than via retroaldol
fragmentation.
The principal competing reaction to consume formaldehyde
unproductively is the Cannizzaro reaction, which converts
HCHO into essentially unreactive methanol (CH3OH) and
formate (HCOOꢀ) at high pH. On early Earth, methanol and
formate were presumably recycled by evaporation and photo-
chemical oxidation to carbon dioxide, which can again be reduced
to HCHO.
General Features of the Formose Process. The formose
process has been much discussed,17 and recent work continues to
add to its characterization.19 One feature of the formose process
itself is the initial slow conversion of formaldehyde, creating a lag
period whose length depends on the concentration of HCHO
and temperature (Supporting Information, Figure S1). During
this period, the mixture remains uncolored. This lag period is
then followed by a rapid loss of HCHO, after which the material
Mineral Species That Inhibit Undesired Reactions. The
formose process exploiting the reactivity of compounds contain-
ing CdO units has been extensively discussed as a prebiotic
source of carbohydrates, including ribose.17,19 Unfortunately, the
9458
dx.doi.org/10.1021/ja201769f |J. Am. Chem. Soc. 2011, 133, 9457–9468

Journal of the American Chemical Society
ARTICLE
Figure 1. The complexity of the classical formose process arises from just the two reaction types, enolization and aldol addition, repeated again and
again, to give a complex manifold of reaction possibilities. To systematize these, carbohydrates (C) are labeled by the number of carbon atoms they
contain, followed by a letter representing whether they are aldehydes (a), ketones (k), enediolates (e), or branched (b) or linear (l) isomers. In the
classical formose process,14 two formaldehyde (HCHO) molecules give single molecules of glycolaldehyde (C2a, upper left), which initiate a cascade of
reactions that fix more HCHO to give higher carbohydrates. Open horizontal black arrows represent enolizations. Heavy vertical or diagonal black
arrows represent aldol additions at less hindered centers of an enediolate. Light vertical or diagonal black arrows represent aldol additions at more
hindered centers. Dotted red arrows indicate retroaldol reactions that fragment higher carbohydrates to give lower carbohydrates. Compounds in
magenta are “dead ends”, not capable of either retroaldol reaction or enolization. Compounds in blue and magenta have been prepared separately in
authentic form to support the analysis of this complexity. Dihydroxyacetone (C3k) is the key to the late, rapid consumption of HCHO. The red dotted
box encloses the proposed borate-moderated cycle, shown in detail in Figure 2.
desired carbohydrate products arising from the formose process
also have CdO groups. This makes them also reactive: unstable
against further processing and decomposition at the high pH and
high temperatures required to initiate the formose process. Thus,
while the high pH’s characteristic of serpentinizing solutions are
essential for the formose process, highly alkaline conditions cause
the destruction of any interesting carbohydrates formed. There-
fore, we must find components of prebiotic environment that
might confer nonreactivity selectively on carbohydrates after they
are formed, especially pentoses and pentuloses (“pent(ul)oses”).
Various mineral species might be considered for this role,
including borates and silicates. Both borate20 and silicate21 bind
to 1,2-diols. This binding is especially tight when the ꢀOH
groups are held in a cis orientation in a five-membered ring, which
is possible for carbohydrates that form cyclic hemiacetals. Cyclic
hemiacetals are possible only for carbohydrates that have at least
four carbon atoms (Figure S2). As complexes of cyclic hemi-
acetals have no CdO group, they are essentially unreactive at
high pH.
borate, in contrast, is significantly greater. Long know known to
bind to carbohydrates, borate was shown to stabilize ribose and
other pentoses formed by the reaction of glycolaldehyde (C2a)
and glyceraldehyde (C3a).23,24 To confirm these reports, we
quantitated this stabilization (Tables S1 and S2); ribose gives
the thermodynamically most stable pentoseꢀborate complex
(Figure S3).
Mitigating the Borate Inhibition: Excess Glycolaldehyde.
Borate therefore might offer the nonreactivity needed to capture
pentoses once they are formed. The nonreactivity caused by
borate minerals has a downside, however. The formose reaction
cycles that fix more HCHO to give higher carbohydrates have
intermediates that also have 1,2-dihydroxyl units that might bind
to borate and prevent their reaction. Thus, borate at 100 mM
prevents formose cycling essentially completely (data not
shown). Instead of being fixed, HCHO disproportionates under
these conditions via the Cannizzaro reaction to give formate and
methanol. Even at 6 mM, borate significantly slows formose
cycling.
Silicate as a stabilizing species has been examined elsewhere
and found to offer relatively little stabilization of carbohydrates
that are able to form cyclic hemiacetals.22 Stabilization of these by
Obviously, the power of any mineral to stabilize carbohydrates
(nonreaction) is useless if the mineral prevents their formation
(reaction). Two approaches might resolve this paradox.
9459
dx.doi.org/10.1021/ja201769f |J. Am. Chem. Soc. 2011, 133, 9457–9468

Journal of the American Chemical Society
ARTICLE
Figure 2. Proposed borate-constrained abiotic metabolic cycle to fix HCHO to form pentoses (including ribose) and pentuloses. Compounds in green
are available prebiotically from meteorites, electrical discharge, photochemistry, minerals, or the interstellar nebula. Certain carbon atoms that have been
C-13 labeled by chemical synthesis are indicated. Species are labeled on the basis of the number of carbon atoms they contain (e.g., C2 has two carbon
atoms) and whether they are aldehydes (a), enediols (e), ketones (k), branched (b), or linear (l).
The first reflects the well-known formation of glycolaldehyde
by electrical discharge through humid CO2.4 We asked whether
adding borate to a mixture of HCHO and glycolaldehyde (C2a,
Figure 2) in a ratio of 1:2 might give borate-stabilized pent-
(ul)oses in the absence of cycling. Here, the proposed pathway
requires first the enolization of glycolaldehyde to give C2e, then
the aldol adition of HCHO to C2e to give glyceraldehyde (C3a),
and finally the reaction of glyceraldehyde (as an electophile or as
its enolate) with the second molecule of glycolaldehyde (as its
enolate or as an electrophile) to give pentoses or pentuloses
(respectively).
or silicate (pH 11.9) gave primarily arabinose. Arabinose is the
pentose that is the most stable under these conditions. Thus, this
result is consistent with the hypothesis that arabinose predomi-
nates not because of any interaction between it and the mineral
species, but because it is simply the most stable pentose.22b
These results were then examined without labeling using
standard preparative chemistry (Table 1). Ribulose (ca. 3%,
not easily observed by NMR), xylulose (ca. 12%), and arabinose
(ca. 7%) were isolated as acetonides following incubation of
HCHO and glycolaldehyde (1:2) in borate buffer. Also iso-
lated as acetonides were threose and erythrose (ca. 30 and 9%),
arising from (2 þ 2) aldol reactions. Other species were not
identified.
These results provide estimates for the relative rate constants
of enolization of glycolaldehyde and glyceraldehyde and the
relative reactivity of HCHO and C2a as electrophiles under these
conditions. Qualitatively, formaldehyde and C2a participate
approximately equally in product formation, despite HCHO
being present overwhelmingly as its hydrate. This illustrates
the high electrophilicity of free HCHdO.
Indeed, adding 1 equiv of H13CHO to 2 equiv of unlabeled
glycolaldehyde in “borate buffer” (1.1 M sodium carbonate, pH
10.4, [boron] = 0.28 M) gave 5-13C-ribose (61.18 ppm), 5-13C-
arabinose (63.33 ppm), and 1-13C-xylulose (64.73 ppm) as
major products (Figure 3). The formation of ribose was con-
firmed by superimposition of the signals from authentic material.
The ribose and arabinose products arose through the enolization
of C2a to give C2e, the aldol addition of HCHO to give C3a (2 þ
1 = 3), and the aldol addition of C3a with another C2e molecule
to give the pentoses. These results are consistent with the (3 þ
2 = 5) reaction observed by Ricardo et al.10 Xylulose is formed by
the reaction of C3e and C2a in a (2 þ 3 = 5) reaction.
Another source of a C3 species might be glycerol, abundant in
meteorites but lacking a CdO unit and therefore quite stable in
non-oxidizing environments.25 Glycerol yields dihydroxyacetone
(with a CdO unit, C3k, Figures 1 and 2) via oxidation in the
presence of iron(II)26 minerals by H2O2, coming via electrical
Once formed, the C5 species bind borate, and nonreaction
ensues. The analogous reaction with borate replaced by carbonate
9460
dx.doi.org/10.1021/ja201769f |J. Am. Chem. Soc. 2011, 133, 9457–9468

Journal of the American Chemical Society
ARTICLE
Figure 3. 13C NMR spectra showing the formation of riboseꢀborate from glycolaldehyde and H13CHO in the presence of borate. Top: Glycolaldehyde
(100 mM) was incubated (65 ꢀC, 1 h) with H13CHO (50 mM) in borate buffer (1.1 M sodium carbonate, pH 10.4, [boron] = 0.278 M, D2O). CH3OH
served as an internal standard (49.50 ppm). Bottom: The same 13C NMR spectrum, with added authentic 5-13C-ribose (Omicron). In separate
experiments (data not shown), assignments of arabinose and xylulose were confirmed by superimposition of their signals upon signals from authentic
samples.
discharge or photochemistry in moist air.27 As noted above,
Table 1. Fraction of Species Isolated (by Their Acetonide
Form) from the Reaction of Glycolaldehyde (0.24 g, 4 mmol)
and HCHO (0.06 g, 2 mmol) in Borate Buffer at 65 ꢀC for 1 h
dihydroxyacetone causes rapid consumption of HCHO in the
classical formose process. Of course, the enediol obtained from
dihydroxyacetone is the same as the enediol obtained from
glyceraldehyde.
Interestingly, without HCHO, borate and glycolaldehyde give
tetroses, which are strongly stabilized by borate as their cyclic
hemiacetals. Thus, glycolaldehyde (C2a) in borate buffer (65 ꢀC
for 1 h or room temperature, overnight) gave threose and
erythrose in a 3:1 ratio (86% yield by HPLC of their dinitrophe-
nylhydrazones, Figure 4). Consistent with their ability to form
cyclic hemiacetals that coordinate borate, threose and erythrose
are quite stable in these buffers at pH 10.4. The formation of
threose is significant because threose can replace ribose in the
backbone of an RNA-like genetic molecule.28,29
Demonstration of Cycles in the Presence of Borate. While
these studies show the near inevitability of formation of pent-
(ul)oses in the presence of borate if glycolaldehyde is produced
in a prebiotic atmosphere in large amounts relative to formal-
dehyde, it is likely that HCHO was far more abundant than
glycolaldehyde on the early Earth. Hence, we examined how
borate might guide and possibly inhibit cycles that fix HCHO
on a large scale, where prebiotic glycolaldehyde would be a
catalyst.
9461
dx.doi.org/10.1021/ja201769f |J. Am. Chem. Soc. 2011, 133, 9457–9468

Journal of the American Chemical Society
ARTICLE
Figure 4. Glycolaldehyde (C2a) in borate buffer. (a) HPLC analysis of the reaction of glycolaldehyde in borate buffer in the absence of HCHO (65 ꢀC,
1 h) by dinitrophenylhydrazones derivatization. (b) Natural abundance 13C NMR analysis of the reaction of glycolaldehyde in borate buffer in the
absence of HCHO (room temperature, overnight): 13C NMR spectra of (A) reaction products, (B) authentic threose in borate buffer, (C) samples A
and B mixed, and (D) authentic erythrose in borate buffer. Signal at 49.50 ppm is reference CH3OH. Erythrose and threose do not undergo reaction in
borate buffer, even after many days.
We began with the hypothesis of a borate-constrained
premetabolic cycle that fixes HCHO, consisting of the reac-
tions within the red dotted box in Figure 1, extracted in
Figure 2. Here, the enediol C3e obtained via enolization of
glyceraldehyde or dihydroxyacetone should fix a single HCHO
molecule to give erythrulose (C4k, 3 þ 1 = 4). Erythrulose
cannot form a cyclic hemiacetal (and is therefore reactive) but
can bind borate weakly through its 3,4-diol unit. This binding
should direct the enolization of erythrulose away from the
borate to give the 1,2-enediol (C4e), proceeding clockwise
around Figure 2. This enediol should fix a second HCHO to
give either the linear (C5l) or branched (C5b) C5 species at
the top and top-right of Figure 2 by reaction (respectively) at
the less hindered or more hindered enediol carbon of C4e (4 þ
1 = 5 reactions).
The branched pentoses C5b do not have enolizable hydrogens
and, therefore, cannot react further as nucleophiles. They can,
however, undergo retroaldol fragmentation to generate glycolal-
dehyde C2a and the enediol of glyceraldehyde C3e (5 = 2 þ 3).
The enediol of glycolaldehyde (C2e) should react with another
molecule of HCHO to form C3a (2 þ 1 = 3 reaction), which
should enolize to form a second molecule of C3e, continuing the
cycle two-fold. With each cycle, the amount of HCHO would be
fixed, leading to large amounts of fixed carbon.
To obtain experimental support for this cycle, we synthesized
(Figure S4) or obtained each of the compounds shown
in blue and magenta in Figure 1. We then incubated these
in borate buffer at 65 ꢀC (except as noted), following the
progress of the reaction using carbon-13 label, introduced
from H13CHO, 13C-glycolaldehyde, 13C-glyceraldehyde, and/or
9462
dx.doi.org/10.1021/ja201769f |J. Am. Chem. Soc. 2011, 133, 9457–9468

Journal of the American Chemical Society
ARTICLE
Figure 5. (a) HPLC analysis of the reaction of dihydroxyacetone (33 mM) with HCHO (100 mM) in borate buffer (65 ꢀC, 3 days) by derivatization of
dinitrophenylhydrazones. (b) Incubation of dihydroxyacetone (C3k) with HCHO. Dihydroxyacetone (33 mM) was incubated with H13CHO
(100 mM) for 3 days at 65 ꢀC in borate buffer (1.1 M sodium carbonate, [boron] = 0.278 M, pH 10.4). Peak assignments: 2-hydroxymethylerythrose
(C5b) of three different borate complexes, 74.01 (C4), 69.233 (C4), and 63.832 ppm (C20); 2-hydroxymethylthreose, 73.487 ppm (C5b); glycolate,
61.825 ppm (C2).
various 13C-labeled compounds synthesized in this work. The
chemical shift of various carbohydrates was influenced by borate,
requiring us to make 13C NMR measurements in borate buffer.
Alternatively, dinitrophenylhydrazone, acetonide, and acetate
derivatives were prepared and resolved by HPLC or column
chromatography.
With Excess HCHO, Glyceraldehyde Gives Branched Tetrose,
Two Stereoisomeric Branched Pentoses, and (Presumably)
Linear 3-Ketopentulose. With these tools, each step in the
cycle was demonstrated. First, incubation of dihydroxyacetone
C3k with HCHO gave C5b branched pentoses in the presence of
borate (3 þ 1 þ 1 = 5 reaction, Figure 5) in about 11% yield, as
determined by preparative chemistry and HPLC of the
dinitrophenylhydrazone derivatives (Table 2 and Figure 5a).
This was essentially the same result as observed starting with
glyceraldehyde C3a, which gives the same enediol C3e inter-
mediate as dihydroxyacetone C3k. A parallel experiment with
H13CHO generated labeled C5b (erythro > threo, Figure 5b).
The analogous procedure with silicate or carbonate without
borate gave primarily “dead end” carbohydrates C4b and C7k (or
their reduction products), which cannot enolize.22b This implied
that productive carbohydrates either are not formed or decom-
pose in the absence of guiding or stabilizing borate.
Reaction of H13CHO with Erythulose Gives Product Pen-
tose Mixtures Similar to Those Obtained with Glyceralde-
hyde or Dihydroxyacetone. Erythrulose is the presumptive
9463
dx.doi.org/10.1021/ja201769f |J. Am. Chem. Soc. 2011, 133, 9457–9468

Journal of the American Chemical Society
ARTICLE
intermediate between C3e and C5b. To confirm this, erythrulose
(C4k) was incubated with H13CHO at 65 ꢀC in borate buffer. By
NMR, the same mixture of diastereomeric erythro and threo
branched pentoses (C5b) was observed. This provides evidence
that erythrulose is a common intermediate between glyceralde-
hyde and dihydroxyacetone, establishing nearly all of the cycle.
Search for Leakage Products C4b, C5l, C6b1, and C7k
(Figures 1 and 2). In addition to producing C5b, the proposed
cycle (Figure 2) suggests two paths that might remove material
from the cycle. The first involves addition of HCHO to the more
hindered center of C3e to give branched tetrose C4b. Authentic
branched tetrose was synthesized (Figure S4e), and found to be
difficult to observe by NMR in borate buffer, presumably because
it forms many complexes undergoing dynamic exchange.
Further, C4b was found to be sensitive to Cannizzaro reduction,
which removes it CdO unit and makes it undetectable as a
hydrazone. This sensitivity is consistent with its inability to form
a cyclic hemiacetal.
Preparative studies therefore recovered reduced C4b product
as its acetate. This allowed detection of reduced C4b (Table 2),
whose structure was confirmed by comparison with a synthetic
compound (Figure S4e). This implies that about 10% of the
material proceeding through the cycle might be diverted to C4b.
Absent a retroaldol reaction to extrude HCHO, this is a dead end.
Following Cannizzaro reduction, it is a dead end in any case.
A second leakage product might arise from the attack of the
enediol of erythrulose (C4e) at its less hindered center on
HCHO to give C5l. To explore the fate of material that leaked
from the cycle in this way, synthetic 1-13C-C5l was prepared
(Figure S4d)30 and incubated in borate buffer at room tempera-
ture without HCHO. Consistent with the fact that C5l cannot
form a cyclic hemiacetal to bind borate, it reacted with a half-life
of 4ꢀ5 days at room temperature to give 1- and 5-13C-xylulose
(90%, both sites are labeled due to the symmetry of the
intermediate) and 1- and 5-13C-ribulose (10%, Figure 6). These
pentuloses are stabilized by binding to borate; ribulose is
further converted to ribose in the presence of borate.20,31 Thus,
this leakage was “productive”: it gave linear pent(ul)oses.
In the presence of HCHO, a second leakage product, heptu-
lose C7k, was detected via acetate derivatization (Table 2), with
its structure being confirmed by the comparison with a synthetic
compound (Figure S4f). This heptulose is also known as a
product of formose reaction cycles without borate.32 Absent a
retroaldol reaction to extrude HCHO, C7k is also a dead end.
Unable to form a cyclic hemiacetal, C7k also proved to be
sensitive to Cannizzaro reduction, generating a polyol. C7k
presumably arises via enolization of C5l, followed by reaction
with HCHO at its less hindered center to give C6bk2, which
enolizes and reacts again with HCHO at its less hindered center
(Figure 1).
Table 2. Fraction of Species Isolated (by Their Acetate
Form) from the Reaction of Dihydroxyacetone (0.18 g, 2
mmol) and HCHO (0.18 g, 6 mmol) in Borate Buffer at
60ꢀ65 ꢀC for 2 Days
These observations led us to seek another product expected
from the enolization of C5l, this arising from attack of HCHO on
the other, slightly more hindered secondary center to give
C6bk1 (Figure 1).33 Preparative reaction of HCHO and
Figure 6. 13C NMR spectrum showing conversion of 1-13C-labeled 1,2,4,5-tetrahydroxypentan-3-one (linear 3-pentulose, C5l) to ribulose and xylulose
upon incubation for 14 days (25 ꢀC) in borate buffer. In the presence of borate, ribulose equilibrates with ribose and arabinose, while xylulose equilibrates
with xylose and lyxose.
9464
dx.doi.org/10.1021/ja201769f |J. Am. Chem. Soc. 2011, 133, 9457–9468

Journal of the American Chemical Society
ARTICLE
Table 3. Fraction of Species Isolated (by Their Acetonide
Form) from the Reaction of Dihydroxyacetone (0.18 g, 2
mmol) and HCHO (0.18 g, 6 mmol) in Borate Buffer at
60ꢀ65 ꢀC for 2 Days
dihydroxyacetone and analysis of its product by acetonide
derivatization allowed detection of some C6bk1 (∼5% yield,
Table 3). The structure of C6bk1 was unequivocally established
by a comparison with the synthetic compound (Figure S4g).
Branched Pentose C5b Is Stabilized by Borate. Material
that does not leak from the cycle in Figure 2 ends up as branched
pentoses C5b, which can complete the cycle if they undergo
retroaldol cleavage to generate C2a and C3e. The retroaldol
cleavage was readily demonstrated in the absence of borate; ca.
90% of C5b is gone in 1 h at 65 ꢀC in 200 mM carbonate (pH
11.8), giving product mixtures too complex to analyze. The
branched pentoses C5b are only slightly stabilized by 200 mM
silicate (also pH 11.8), with ca. 75% gone in the same time.22b
However, consistent with the ability of C5b to form cyclic
hemiacetals that can bind borate (Figure S2), both stereoiso-
meric branched pentoses were stabilized by borate, with stability
increasing with borate concentrations (Table S2). At standard
borate concentrations (278 mM), the rate of retroaldol reaction
was too slow to conveniently measure. With just 20 mM borate in
the presence of H13CHO to give (65 ꢀC, 1 day) the retroaldol
products, reduced 13C-labeled C4b and C7k (and its Cannizzaro-
reduced alcohol) were seen. These structures were proven by
comparison with authentic material obtained by direct chemical
synthesis (Figure S4). The non-dead-end products presumably
moved to more complex products at these low concentrations of
borate.
Mineralogical Solutions to the “Dead End Problem”.
Borate stabilization of the branched pentoses C5b creates a
problem for the cycle. As long as it is bound to borate, C5b does
not fragment to produce C2a and C3e that might capture more
HCHO and continue around the cycle. Further, we were unable
to find concentrations of borate that were low enough to allow
the cycle to proceed via fragmentation of the branched pentoses
yet high enough to allow for product analysis (other than of the
dead end species C4b and C7k) over periods of time convenient
for laboratory study.
Accordingly, we looked for other ways to allow these branched
species to be processed productively. Two were found, both
involving mineral species, and both related to the Bilik reaction,34
which rearranges a branched carbohydrate (such as C4b and
C5b) to give a linear carbohydrate (C4k and C5k, respectively).
At high pH, Ca2þ catalyzes a Bilik reaction. Thus, treating
20-13C-2-hydroxymethylerythrose (C5b) with Ca(OH)2 in the
absence of borate gave (presumably without retroaldol reaction)
xylulose (Figure 7). Adding borate slowed the rate.
Figure 7. Bilik reaction of branched pentose (C5b) catalyzed by
calcium. 13C NMR spectrum of the reaction of 20 mM C5b (erythro,
5-13C labeled) in the presence of 20 mM calcium chloride and 20 mM
sodium hydroxide at room temperature for 30 min.
At lower pH, the Bilik reaction can be catalyzed by molybdate
minerals. Thus, incubating (65 ꢀC, 24 h) labeled C5b (erythro,
50 mM) in the presence of sodium molybdate (Na2MoO4 2H2O,
3
2 mM) at pH 5.9 leads to an equilibrium mixture of C5b starting
material and linear xylulose (a desired product), with some linear
pentulose 1,2,4,5-tetrahydroxypentan-3-one C51. The rearrangement
was stereospecific: threo C5b gave ribulose (Figure 8). Xylulose and
ribulose equilibrate slowly with Mo6þ to give xylose and ribose.31
Thus, molybdate minerals can generate free ribose and xylose
from C5b branched pentoses.
’ DISCUSSION
These results establish the cycle in Figure 2 as a laboratory
reality. Further, they provide an indication of how borate is able
to direct organic species derived from HCHO and glycolalde-
hyde along just a few of many alternative paths that are
conceivable (Figure 1). For example, borate helps direct attack
of HCHO at the more hindered center of the enediol of
eryrthulose, C4e, to give the branched pentoses C5b. Further,
these study show that borate hinders the retroaldol reaction of
C5b, permitting it to accumulate.
Since HCHO and glycolaldehyde are generated by electrical
discharge through humid CO2 atmospheres, and since some
borate must be present in leaching igneous rocks, it appears hard
to avoid the conclusion that any subaerial evaporite region on
early Earth exposed to a humid CO2 atmosphere suffering
electrical discharge accumulated some branched pentose C5b
as its borate complex. Only the amount is at issue. This amount
would be determined by the excess of carbon over boron, the pH,
and the temperature, factors not well constrained in models for
early Earth.
These results also support rules that allow the interpretation of
such processes generally. In the preparative work reported here,
approximately 50% of the carbon originating in HCHO can be
accounted for in various products. Further, as long as HCHO is
present, the mixture never turns yellow. Thus, as long as HCHO
is present to capture carbon fragments arising from retroaldol
9465
dx.doi.org/10.1021/ja201769f |J. Am. Chem. Soc. 2011, 133, 9457–9468

Journal of the American Chemical Society
ARTICLE
Figure 8. Bilik reaction of branched pentoses (C5b) catalyzed by molybdate. (a) 13C NMR spectrum of the reaction of 50 mM C5b (erythro, 5-13C-
labeled) in the presence of 2 mM sodium molybdate (Na2MoO4 2H2O, pH 5.9) at 65 ꢀC for 24 h. (b) 13C NMR spectrum of the reaction of 50 mM C5b
3
(threo, 5-13C-labeled) in the presence of 2 mM sodium molybdate (Na2MoO4 2H2O, pH 5.9) at 65 ꢀC for 24 h.
3
fragmentation of C5b that escapes its borate complex, pent-
(ul)oses will accumulate, simply because they are the first
products accessible in this system that are able to form cyclic
hemiacetals that can coordinate borate strongly.
These results also suggest a choice among mineral species that
might have been productively involved in the prebiotic transfor-
mation of HCHO and its derived carbohydrates. For example,
Lambert and co-workers have suggested that silicate might be
useful. In the Earth's crust, silicate is far more abundant than
borate.22a In aqueous solution, however, especially at neutral pH
and below, silicate precipitates to give SiO2 (opal or quartz),
while borate remains in solution. Further, borate stabilizes cyclic
pentoses more than silicate, whose stabilizing effect, although
detectable, is small.
was almost certainly available to prebiotic chemistry. Calcium
borate minerals (colemanite and ulexite, for example) are
relatively soluble in water. Indeed, the special need for Ca2þ in
the formose process (see above) may arise because it can catalyze
Bilik-like rearrangements.
The disadvantage of Ca2þ as a prebiotic Bilik catalyst is that it
operates only at high pH, conditions where the carbohydrate
products require borate stabilization. This is not true for molyb-
date, which catalyzes the formation of linear pent(ul)oses at near-
neutral pH and moderate temperatures. The pent(ul)oses
formed do not require borate stabilization under these condi-
tions; indeed, at lower pH, the borate complexes are less stable,
making their carbohydrates accessible to molybdate.
While the model that we have used here for the prebiotic
atmosphere is currently the consensus, what is the likelihood
that the other constraints are met? Borate minerals are known
from 3.8 Ga old rocks in the Isua supracrustal belt of western
These results also show how other minerals might determine
the fate of C5b. Both Ca2þ minerals and molybdate minerals can
convert C5b to linear pent(ul)oses via the Bilik reaction. Ca2þ
9466
dx.doi.org/10.1021/ja201769f |J. Am. Chem. Soc. 2011, 133, 9457–9468

Journal of the American Chemical Society
ARTICLE
Greenland,35 where they may have arisen from evaporite basins
of the sort needed for the cycles proposed here. Borate
minerals are also associated with stromatolites in the Barberton
greenstone belt in South Africa, where they seem almost
certain to have arisen from evaporites.18 While some have
questioned whether the early Earth was sufficiently differen-
tiated to have allowed borate concentration in the geosphere,36
borate is enriched in the residual melt of any igneous species,
from which borates are easily weathered, allowing them to be
concentrated in the hydrosphere even if they are not concen-
trated in the lithosphere. This makes it difficult to argue against
borate-rich evaporates in any early Earth scenario that includes
dry land.
Molybdate presents more of a challenge to the prebiotic
chemist, as it is oxidized relative to the redox state of early
Earth, as presently modeled (MoO2 þ H2O þ 2Fe3þ f MoO3 þ
2Hþ þ 2Fe2þ = þ236 mV). However, a “planetary redox
potential” is unlikely to be relevant to the existence of such
minerals. Emerging models for early Earth suggest that con-
tinents and their associated subduction zones were present as
early as 4.4ꢀ4.5 Ga.37 These would have generated felsic
magmas that would have included minerals that are more
oxidized than the terran surface as a whole, including sulfate,
borate, and molybdate (S. J. Mojzsis, personal communication).
These results highlight the potential of minerals to provide
simultaneously both the desired reactivity and the desired
nonreactivity within a “prebiotic soup”. Further, they offer a
“vestigiality” explanation for why pent(ul)oses, including ribose,
are found in genetic material. With excess HCHO, pentoses are
the first species that can be formed by a cycle with excess HCHO
that have available a hemiacetal form that can be bound and
stabilized by borate. Once stabilized, they react no further to give
hexoses, heptoses, and higher sugars, even with excess HCHO.
While aldotetroses can form cyclic ligands that are extremely
stable in borate, they are not accessible by the pathway where
HCHO is in excess, and therefore they are not formed.
Further, this work illustrates how fluctuating conditions might
support transformations of prebiotic organic molecules, includ-
ing changes in pH or the relative amounts of stabilizing mineral
species and organic species needing stabilization. For example,
for those concerned that it might be difficult to release ribose
from its borate complex,38 simply lowering the pH through
buffering from atmospheric CO2 can do this. At pH 7, ribose is
released from borate to nearly neutral conditions, where it is
quite stable against enolization and aldol reactions that lead to
the destruction of carbonyl compounds at high pH. There, ribose
is available to be phosphorylated by prebiotic mechanisms
developed in other laboratories.39
As a final word, although the work here is driven by the “RNA-
first” hypothesis for the origin of life, the cycles described here
share some of the features proposed for cycles hypothesized for
“metabolism-first” models.40 Although “genetics-first” and “me-
tabolism-first” models for the origin of life are currently being
presented as adversaries,41 no logic compels them to be. It is
nearly certain that chemical processes that might be likened to
metabolism occurred on Earth before genetics was established in
its macromolecular form. These processes may have provided the
components of whatever genetic system did first emerge. While it
is difficult to know whether borate-moderated formaldehyde-
fixation cycles meet criteria required by advocates of a “metabo-
lism-first” scenario, Figure 2 represents a metabolic cycle resem-
bling those found in contemporary terran life.
’ EXPERIMENTAL SECTION
General Methods. 13C-labeled carbohydrates (arabinitol, arabinose,
lyxose, ribose, xylose, ribulose, xylulose, glycolaldehyde, glyceraldehydes)
were obtained from Omicron Bio. H13CHO and labeled paraformalde-
hyde were obtained from Cambridge Isotopes. All other reagents were
obtained from Sigma-Aldrich and were used without purification. Flash
column chromatography was carried out using Merck 9385 silica gel 60
(230ꢀ400 mesh). NMR spectroscopy was carried out on a Varian
Mercury 300 NMR spectrometer.
Typical Procedure for Dinitrophenylhydrazone Forma-
tion and Analysis by HPLC. A mixture of dihydroxyacetone (0.18 g,
2 mmol) and formaldehyde (0.18 g, 6 mmol) in borate buffer (1100 mM
carbonate and 278 mM borate, made by dissolving 4.68 g of Na2CO3 and
0.688 g of H3BO3 in 40 mL of H2O) was stirred at 60ꢀ65 ꢀC for 2 days
under an Ar atmosphere.
To 20 μL of the above reaction mixture were added 300 μL of TFA
solution (2% TFA in MeOH, v/v) and 200 μL of 2,4-dinitrophenylhy-
drazine solution (1.5% DNP in dimethoxyethane, w/v). The mixture
was heated at 65 ꢀC for 90 min and then cooled to room temperature,
and 400 μL of acetone was added. After evaporation to dryness, the
residue was treated with 400 μL of 5% triethylamine in methanol and
evaporated. This residue was dissolved in 80 μL of 1,2-dimethoxyethane
and treated with 500 μL of water, and the resultant suspension was
centrifuged (10 000 rpm, 2 min). The aliquot was injected into the
HPLC (column, Waters Nova-Pak HR C18 6 μm, 60 Å, 7.8 ꢁ 300 mm
Prep Column; eluents, A = 0.02% TFA in water, B = CH3CN, gradient
from 15 to 25% B in 60 min, flow rate 1 mL/min). The peaks of the
DNPꢀsugar derivatives eluted were detected by their absorbance at
360 nm.
Typical Procedure for Acetate Derivatization. A mixture of
dihydroxyacetone (0.18 g, 2 mmol) and formaldehyde (0.18 g, 6 mmol)
in borate buffer was heated to 60ꢀ65 ꢀC for 2 days under Ar atmo-
sphere. After cooling to room temperature, the mixture was neutralized
by acidic resin (Dowex), filtered, and lyophilized. The resulting solid was
dissolved in methanol (20 mL), evaporated on a rotary evaporator
(repeated three times), and further dried under high vacuum to give a
reddish brown solid. It was then treated with acetic anhydride (3 mL),
DMAP (50 mg), and pyridine (20 mL) and stirred at room temperature
for 24 h. It was evaporated and separated by silica gel column
chromatography (EtOAc:Hex = 1:2 to 2:1) to give 11 crude fractions.
1
After evaporation, each fraction was weighed and characterized by H
NMR. The identity of the each fraction was confirmed by comparison
with the authentic material.
Typical Procedure for Acetonide Derivatization. A mixture
ofdihydroxyacetone (0.18 g, 2 mmol) and formaldehyde(0.18g, 6 mmol)
in borate buffer was heated to 60ꢀ65 ꢀC for 2 days under Ar atmo-
sphere. After cooling to room temperature, the mixture was neutralized
by acidic resin (Dowex), filtered, and lyophilized. The resulting solid was
dissolved in methanol (20 mL), evaporated on a rotary evaporator
(repeated three times), and further dried under high vacuum to give a
reddish brown solid. It was then treated with acetone (50 mL) and
sulfuric acid (1 mL) and stirred at room temperature for 1.5 h. It was
neutralized by sodium bicarbonate and evaporated and separated by
silica gel column chromatography (EtOAc:Hex = 1:2 to 2:1) to give five
crude fractions. After evaporation, each fraction was weighed and
characterized by 1H NMR. The identity of the each fraction was
confirmed by comparison with the authentic material.
’ ASSOCIATED CONTENT
S
Supporting Information. Relative stability of pent-
b
(ul)oses (Tables S1 and S2 and Figure S3), consumption of
formaldehyde in formose reaction (Figure S1), and synthetic
9467
dx.doi.org/10.1021/ja201769f |J. Am. Chem. Soc. 2011, 133, 9457–9468

Journal of the American Chemical Society
ARTICLE
details of various suspected carbohydrate intermediates with 13
label at specific sites (Figure S4). This material is available free of
C
(32) Shigemasa, Y.; Ueda, T.; Saimoto, H. Bull. Chem. Soc. Jpn. 1990,
63, 389–394.
(33) Shigemasa, Y.;Hamada, T.;Hirabayashi, M.;Waki, E.;Nakashima,
R.; Harada, K.; Takeda, N.; Suzuki, M. Chem. Lett. 1981, 899.
(34) Petrus, L.; Petrusovꢀa, M.; Hricovíniovꢀa, Z. Glycoscience 2001,
215, 15.
charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
(35) (a) Appel, P. W. U. Precambrian Res. 1995, 72, 227–234.
(b) Chaussidon, M.; Appel, P. W. U. Chem. Geol. 1997, 136, 171–180.
(36) Grew, E. S.; Bada, J. L.; Hazen, R. M. Origins Life Evol. Biosph.
2011, DOI 10.1007/s11084-010-9233-y.
Corresponding Author
sbenner@ffame.org
(37) Harrison, T. M.; Blichert-Toft, J.; Muller, W.; Albarede, F.;
Holden, P.; Mojzsis, S. J. Science 2005, 310, 1947–1950.
(38) Anastasi, C.; Crowe, M. A.; Sutherland, J. D. J. Am. Chem. Soc.
2007, 129, 24–25.
(39) Pasek, M. A.; Dworkin, J. P.; Lauretta, D. S. Geochim. Cosmo-
chim. Acta 2007, 71, 1721–1736.
(40) (a) Dekker, P. Ann. N.Y. Acad. Sci. 1979, 316, 236–250.
(b) Copley, S. D.; Smith, E.; Morowitz, H. J. Bioorg. Chem. 2007, 35, 430.
(41) Orgel, L. E. PLOS Biol. 2008, 6, e18.
’ ACKNOWLEDGMENT
This paper is dedicated to the memories of Leslie Orgel and
Stanley Miller. We are indebted to the NASA Astrobiology/
Exobiology program (NNX-08AO23G), the National Science
Foundation Chemical Bonding Program (CHE-0434507), and
the John A. Templeton Foundation for support of this work.
’ REFERENCES
(1) Powner, M. W.; Gerland, B.; Sutherland, J. D. Nature 2009,
459, 239–242.
(2) M€uller, D.; Pitsch, S.; Kittaka, A.; Wagner, E.; Wintner, C. E.;
Eschenmoser, A. Helv. Chim. Acta 1990, 73, 1410–1469.
(3) L€ow, O. J. Prakt. Chem. (Leipzig) 1915, 92, 133–135.
(4) L€ob, W. Ber. Deutch. Chem. Ges. 1913, 46, 684–697.
(5) Miller, S. L. Science 1953, 117, 528–529.
(6) Shapiro, R. Sci. Am. 2007, 296, 46–53.
(7) Benner, S. A. Life, the Universe and the Scientific Method;
Foundation Press: Gainesville, FL, 2009; p 312.
(8) Wachtershauser, G. Prog. Biophys. Mol. Biol. 1992, 58, 85–201.
(9) Prieur, B. E. C. R. Acad. Sci. Paris, Chimie 2001, 4, 667.
(10) Ricardo, A.; Carrigan, M. A.; Olcott, A. N.; Benner, S. A. Science
2004, 303, 196.
(11) (a) Kauffman, S. A. At Home in the Universe; Oxford University
Press: Oxford, 1995. (b) Schwartz, A. W. Chem. Biodivers. 2007,
4, 656–664.
(12) Larralde, R.; Robertson, M. P.; Miller, S. Proc. Natl. Acad. Sci.
U.S.A. 1995, 92, 8158.
(13) Joyce, G. F. Origins Life Evol. Biosphere 2006, 36, 202.
(14) Weber, A. L. Origins of Life Evol. Biosphere 2002, 32, 333–357.
(15) Cleaves, H. J. Precambrian Res. 2008, 164, 111–118.
(16) Moody, J. B. Lithos 1976, 9, 125.
(17) Breslow, R. Tetrahedron Lett. 1959, 1, 22–26.
(18) Byerly, G. R.; Palmer, M. R. Contrib. Mineral. Petrol. 1991,
107, 387–402.
(19) Ricardo, A.; Frye, F.; Carrigan, M. A.; Tipton, J. D.; Powell,
D. H.; Benner, S. A. J. Org. Chem. 2006, 71, 9503.
(20) Chapelle, S.; Verchere, J. A. Tetrahedron 1988, 44, 4469.
(21) Lambert, J. B.; Lu, G.; Singer, S. R.; Kolb, V. M. J. Am. Chem.
Soc. 2004, 126, 9611–9625.
(22) (a) Lambert, J. B.; Gurusamy-Thangvelu, S. A.; Ma, K. Science
2010, 327, 984–986. (b) Kim, H. J.; Benner, S. A. Science 2010, 329, 902.
(23) Mendicino, J. F. J. Am. Chem. Soc. 1960, 82, 4975.
(24) Campello, J. P.; Veiga, L. A. Ann. Acad. Brazil Cienc. 1973,
45, 309.
(25) Cooper, G.; Kimmich, N.; Belisle, W.; Sarinana, J.; Brabham,
K.; Garrel, L. Nature 2001, 414, 879.
(26) Fenton, H. J. H.; Jones, H. O. J. Chem. Soc. 1900, 77, 69.
(27) Liang, M.-C.; Hartman, H.; Kopp, R. E.; Kirschvink, J. L.; Yung,
Y. L. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 18896.
(28) Ebert, M. O.; Mang, C.; Krishnamurthy, R.; Eschenmoser, A.;
Jaun, B. J. Am. Chem. Soc. 2008, 130, 15105–15115.
(29) Horhota, A.; Zou, K.; Ichida, J. K.; Yu, B.; McLaughlin, L. W.;
Szostak, J. W.; Chaput, J. C. J. Am. Chem. Soc. 2005, 127, 7427.
(30) Linclau, B.; Boydell, A. J.; Clarke, P. J.; Horan, R.; Jacquet, C.
J. Org. Chem. 2003, 68, 1821.
(31) Angyal, S. J. Glycoscience 2001, 215, 1.
9468
dx.doi.org/10.1021/ja201769f |J. Am. Chem. Soc. 2011, 133, 9457–9468