Formation of Activated Biomolecules by Condensation on Mineral Surfaces – A Comparison of Peptide Bond Formation and Phosphate Condensation

Abstract

Many studies have reported condensation reactions of prebiotic molecules, such as the formation of peptide bonds between amino acids, to occur to some degree on mineral surfaces. We have studied several such reactions on the same divided silica. When drying steps are applied, the equilibria of peptide formation from glycine, and polyphosphate formation from monophosphate, are displaced to the right because these reactions are dehydrating condensations, accompanied by the emission of water. In contrast, the equilibrium of AMP dismutation is not significantly favored by drying. The silica surface plays little role (if any) in the thermochemistry of the condensation reactions, but is does play a significant kinetic role by acting as a catalyst, lowering the condensation temperatures with respect to bulk solids. Of course, the surface also catalyzes the inverse hydrolysis reactions.

Introduction

Interesting results have appeared since the 1970s involving “interfacial” prebiotic scenarii, where two molecules or more are made to react together after adsorption on a mineral surface, yielding activated molecules such as peptides from amino acids ((Lambert 2008) and references therein). However, progress in this field has been irregular, largely a hit-or-miss affair, and fundamental questions raised early about these scenarii remain unanswered (de Duve and Miller 1991). This is in part because the thermodynamic and kinetic aspects of the problem are not clearly distinguished and we will therefore start by recalling a few basic facts.

Many thermodynamically uphill prebiotic reactions are in fact condensation reactions accompanied by the elimination of one water molecule, of the general form:

$$ \mathrm{A}+\mathrm{B}=\mathrm{A}-\mathrm{B}+{\mathrm{H}}_2\mathrm{O} $$

(1)

Frequently, such a reaction is unfavorable, which means that its equilibrium constant K is very low, and therefore that its standard Gibbs free energy of reaction is significantly positive, i.e., Δ _r G° > 0 (remember that $K={e}^{\frac{-{\varDelta}_r G \mathit{^{\circ}}}{ RT}}$). Now this conclusion of course depends on what thermodynamic phase the reaction is occuring in. Biochemists and geochemists usually have in mind reactions in aqueous solution so that reaction (1) should be specified as:

$$ {\mathrm{A}}_{\mathrm{aq}}+{\mathrm{B}}_{\mathrm{aq}}=\mathrm{A}-{\mathrm{B}}_{\mathrm{aq}}+{\mathrm{H}}_2{\mathrm{O}}_{\mathrm{liq}} $$

(2)

Why would things be different at the interface? The first suggestion that comes to mind is that some or all of the molecules involved may have different energy (or rather, Gibbs free energy) levels than in the solution. Without putting too fine a point on it, it might be acceptable to define an “adsorbed phase” (subscript “ads”), in which the corresponding reaction:

$$ {\mathrm{A}}_{\mathrm{ads}}+{\mathrm{B}}_{\mathrm{ads}}=\mathrm{A}-{\mathrm{B}}_{\mathrm{ads}}+{\mathrm{H}}_2{\mathrm{O}}_{\mathrm{ads}} $$

(3)

is thermodynamically favored, i.e. (Δ _r G°) _ads ≠ (Δ _r G°) _aq , and (Δ _r G°) _ads < 0. This is perfectly conceivable. It will be the case if the products interact with the surface significantly more strongly than the reagents, and/or if higher effective concentrations are achieved in the adsorbed phase as compared to the bulk solution. Presumably, it was the original idea of Bernal who first proposed a prebiotic role of mineral surfaces (Bernal 1951). The question of whether the reaction products of (3) are more stabilized by adsorption than its reagents was clearly stated by de Duve and Miller in 1991 (de Duve and Miller 1991). However, it is only in 2010 that the idea has been rigorously tested—with a negative result. Marshall-Bowman et al. (2010) studied the equilibrium between glycine and diglycine (A = B = Gly, as well as related oligomerization equilibria) in aqueous solution with and without added minerals, and found no significant difference between the reaction equilibria in the “adsorbed phase” and the solution. The dispersed solids tested included several of the most likely prebiotic minerals, among them montmorillonite clays and silica.

This begs the question of why many studies observed significant polymerization in the same systems (Gly on silica and Gly on montmorillonite), even up to the observation of pentapeptides. In fact, all of these studies involve wetting-and-drying cycles, first used by Lahav et al. (1978): for significant times in the procedure, the Gly-mineral samples are activated at rather high temperatures in the dry state, i.e. the interface is not between the mineral and an aqueous solution, but between the mineral and the gas phase. In these conditions water may desorb and the reaction has to be rewritten as:

$$ {\mathrm{A}}_{\mathrm{ads}}+{\mathrm{B}}_{\mathrm{ads}}=\mathrm{A}-{\mathrm{B}}_{\mathrm{ads}}+{\mathrm{H}}_2{\mathrm{O}}_{\mathrm{gas}} $$

(4)

This makes a difference, because the Gibbs free energy of reaction has both an enthalpic and an entropic part (Δ _r G° = Δ _r H°−T Δ _r S°). Even in the hypothesis that the reaction enthalpy is exactly the same in the adsorbed phase as in solution, reaction (4) may become favorable because its standard entropy will be very positive if water has a low activity in the gas phase, i.e. if the conditions are very dry. More intuitively, this is an application of Le Chatelier’s principle: eliminating H₂O, one of the reaction products, drives the equilibrium to the right. Numerically, to keep the reaction quotient constant:

$$ {Q}_{eq}=\frac{a_{AB}{a}_{H_2O}}{a_A{a}_B} $$

while decreasing a _H2O , one has to increase the dimer/monomer ratio.

Thus, the mineral surface most likely plays little role in reaction thermodynamics. The same argument would be valid if A and B were precipitated in a bulk solid phase instead of being adsorbed—and indeed studies of prebiotic reactions in the solid state have drawn renewed interest recently (Kolb 2012).

When it comes to kinetics the mineral surface does matter however. Polymerization reactions may be theoretically favored in bulk solids but they are certainly slow since e.g. bulk glycine is indefinitely stable at room temperature. The fact that they have been reported to occur at a measurable rate for adsorbed glycine at temperatures as low as 80 °C means that some surface groups can have a significant catalytic effect on condensation reactions. Due to a basic principle of kinetics, they must also have a catalytic effect on the inverse reaction, namely the hydrolysis of A–B (in fact, this has been evidenced in the already mentioned study (Marshall-Bowman et al. 2010) that excluded a significant thermodynamic effect of interfaces).

In what follows we intend to illustrate the above general considerations by their application to three different systems of prebiotic interest: glycine/silica, monophosphate/silica, and AMP/silica.

In previous studies, high-surface amorphous silica has proved to constitute a good test material for prebiotic condensation studies (Meng et al. 2004; Lambert et al. 2009; Lopes et al. 2009; Bouchoucha et al. 2011)—not only do molecules deposited on this mineral support show interesting reactivity, but their reactions can be characterized rather easily by thermogravimetry and NMR for technical reasons. Thus, they were selected for a comparison of the reactivity of different biomolecules on the same support. This eliminates a source of variability that is often neglected: the differences between the surfaces of several minerals with the same bulk composition may be much more important than is usually realized.

Experimental

Materials

Potassium dihydrogenophosphate (KH₂PO₄), 5′ adenosine monophosphate sodium salt (AMP), 5′ adenosine diphosphate disodium salt (ADP) and 5′adenosine triphosphate disodium salt (ATP), were purchased from Sigma-Aldrich.

The mineral phases used for deposition were commercial fumed silicas that are quite well-known because they are often used in heterogeneous catalyst preparation studies. Aerosil 380 was provided by Evonik; it has a BET surface aera of 380 m²/g. Aerosil 150 was provided by Degussa and has a surface area of 150 m²/g.

Deposition Procedure

The deposition procedures used in the present paper included both impregnation and selective adsorption. In an impregnation procedure aqueous solutions of the relevant (bio)molecules are contacted with the solid silica support, and after some equilibration time the whole system is dried without previous phase separation, so that the entire amount of introduced biomolecules remains in the final solid sample. Thus, for inorganic phosphates (Pi)/SiO₂ and AMP/SiO₂ preparation, 10 mL of either KH₂PO₄ or AMP solution of the desired concentration were added dropwise to 400 mg of silica (Aerosil 150 in the first case, Aerosil 380 in the second case). The mixture was a fluid dispersion; it was stirred for 30 min at RT and dried overnight in an oven at 70 °C.

A similar procedure was used to prepare 6 % Gly-Gly/SiO₂. However the Gly/SiO₂ samples discussed in this paper was prepared by a selective adsorption procedure. A 0.03 M solution of glycine isotopically enriched in ¹³C on the carbonyl was contacted with Aerosil 380 (silica concentration 30 g.L⁻¹). After 4 h under stirring, the solid phase was separated by centrifugation. Our previous results indicated that the nature of the deposition procedure is not crucial; samples having the same Gly content exhibit similar behaviors, irrespective of whether they were prepared by impregnation or selective adsorption.

Solid samples were stored in a desiccator containing silica gel before analysis, except when the objective was to test the effect of high water activity. In this case they were stored in a sealed container containing a beaker full of distilled water.

Thermogravimetric Analyses

Thermogravimetric analysis (TGA) of the samples was carried out on a TA Instruments Waters LLC, with a SDT Q600 analyzer, using variable heating rates β (standard: 5 °C/min), under dry air flow (100 mL/min).

Solid state-NMR

Solid-state ¹³C MAS NMR spectroscopy was carried out at room temperature on a Bruker Avance 400 spectrometer operating with a field of 9.4 T, equipped with a 4 mm MAS probe with a spinning rate of 12 kHz. Spectra were recorded using a proton cross-polarization (CP) sequence with a pulse length of 3 μs (for a π/2 pulse of about 5 μs), a data acquisition time of 154 ms, and a recycle delay of 5 to 10s. The chemical shifts were determined by reference to an external adamantane sample (δ = +38.52 ppm),

Solid-state ³¹P MAS NMR spectra were recorded at room temperature with a Bruker Avance 500 spectrometer with a field of 14.0 T, equipped with a 4 mm MAS probe with a spinning rate of 10 kHz. We used a simple 1-pulse sequence sequence with a pulse length of 1.25 μs, a data acquisition time of 30 ms, and a recycle delay of 5 s.

MALDI-TOF

Mass spectra were generated using a 4700 Proteomic Analyzer MALDI-TOF/TOF (Applied Biosystems) fitted with a Nd:YAG laser (λ = 355 nm, pulse duration 4 ns, repetition rate 200 Hz). The organic matter contained in the activated Gly/SiO₂ samples was desorbed with distilled water, and 1 μL of the desorption solution was mixed with 1 μL of matrix (CHCA, alpha-cyano-4-hydroxycinnamic acid, ~5 mg/mL in 1/1 acetonitrile/water 0.1%TFA) All MALDI-TOF spectra, resulting from the average of a few tens or hundred laser shoots, were obtained in positive and negative ion reflector mode in the (10–4000) m/z range. The final solution was vortexed for 1 min. at high speed prior to deposition on the MALDI plates. It was checked that the desorption procedure used was quantitative for these samples, i.e., no residual adsorbed organic matter was detected by TG.

Results

Glycine on Silica

Silica is among the supports that have been studied in early work on amino acids polymerization using wetting and drying cycles, including the simplest AA, glycine (Gromovoy et al. 1991; Bujdák and Rode 1997, 1999).

The reaction that takes place upon moderate temperature activation of these systems is actually a little different from Eqs. (1–4) stoechiometrically, since it is a cyclodimerization yielding the heterocycle diketopiperazine or DKP (Basiuk et al. 1991, 1992; Basiuk and Gromovoy 1993) with the elimination of two water molecules:

$$ {\mathrm{Gly}}_{\mathrm{ads}}+{\mathrm{Gly}}_{\mathrm{ads}}={\mathrm{DKP}}_{\mathrm{ads}}+2{\mathrm{H}}_2{\mathrm{O}}_{\mathrm{gas}} $$

(5)

However this does not fundamentally modify the basic arguments put forward in the introduction. Reaction (5) is easily observed on silica by thermogravimetry—differential thermoanalysis (TG-DTA) and its product has been identified by several spectroscopic techniques (Meng et al. 2004). Gas phase elimination of water molecules formed along this reaction is detected around 150–160 °C (in the fast heating conditions we used; see Fig. 1a, b). It is markedly endothermic, indicating that the enthalpic part of Δ _r G° remains unfavorable in the adsorbed state (Fig. 1c, d). Yet the reaction is possible because of the highly favorable entropic contribution as discussed above.

Fig. 1

DTG/DTA of glycine on silica under air flow, β = 5 °C/min, zoom on the region corresponding to the thermal condensation to DKP. Trace a: differential thermogravimetry (DTG) of 5 % Gly/Aerosil 380; trace b: corresponding blank (DTG of bare Aerosil 380); trace c: heat flow signal for 5 % Gly/Aerosil 380; trace d: corresponding blank

As regards the kinetics, we must underline that bulk α-glycine also dimerizes to DKP—but at much higher temperatures, i.e., 240 °C in the same conditions. At these temperatures dimerization is immediately followed by decomposition to ill-characterized polymers. Thus, the silica surface is quite efficient for catalyzing reaction (5).

Our TG experiments were conducted under dry air, and in these conditions quantification of the DTG signal of water usually indicates (at least for low Gly loadings, <5 %) that Gly condensation to DKP is 100 % complete, or very close.

Under higher water activities, the conclusion is different, however. We have previously published ¹³C NMR results indicating the different outcomes of the thermal activation of Gly/SiO₂ as a function of water activity in the gas phase (Lambert et al. 2009). Similar data are shown in Fig. 2 (see (Lopes et al. 2009) for a justification of NMR peak assignments): while heating at 160° in dry argon flow results in quantitative cyclodimerization to DKP, heating under wet argon flow (i.e. containing a water vapor pressure of 3200 Pa) almost completely inhibits polymerization reactions.

Fig. 2

¹³C MAS-NMR, in the carbonyl region, of ¹³C enriched Gly/Aerosil 380 (obtained by selective adsorption from a 0.03 M solution), after heating up to 160 °C, a. in dry argon flow, b. in a flow of Ar previously saturated with H₂O at 25 °C (P_H2O = 3.2 kPa)

This allows to better understand data previously published by other teams. Bujdák and Rode (Bujdák and Rode 1997) carried out cycles consisting in wetting with distilled water, followed by drying at 80 °C for 24 h, on Gly/SiO₂ (1 mmol.g⁻¹, or 7.5 % by weight), a procedure inspired by the early work of Lahav et al. on clays (Lahav et al. 1978; Lahav and White 1980). They observed small amounts of DKP and linear Gly-Gly, increasing as a function of the number of cycles—the maximum amount converted after 7 cycles was 0.4 % of the total glycine. The first question these data raise is to know why the overall yield was so low, as compared to 100 % conversion in a single step in our experiments. This could of course be due to the lower activation temperature (80 °C vs. 160 °C). However it is a basic fact of kinetics that lower reaction temperatures may be traded for longer reaction times. Indeed the DKP formation peak in TG may be displaced to lower temperatures by using smaller heating rates β, as indicated in Fig. 3. A Kissinger analysis (Kissinger 1956, 1957) of these data yielded an activation enthalpy for glycine condensation of 96 kJ.mol⁻¹ (Lambert et al. 2013); therefore, lowering the temperature from 160 to 80 °C should slow the reaction down by a factor of 420, and since it proceeds to completion in a few minutes at 160 °C, it should be almost complete after a day and a half at 80 °C.

Fig. 3

Effect of heating rate on the temperature of the first DTG peak (T_max) of system 2 % Gly/Aerosil 380

We think that the main reason for the very different condensation yields between both studies is actually the water activity in the gas phase during the drying step. There is no indication that precautions were taken to control the partial pressure of water in the atmosphere in contact with the sample upon drying in the studies of Bujdák et al. Previous articles indicate that the drying step was carried out “in a heating box” (Bujdák et al. 1995), presumably in contact with ambient air, and thus containing a rather high and variable water vapor pressure. As seen from Fig. 2, this should severely hinder glycine condensation.

The next question is why wetting-and-drying cycles produce linear peptides (polyglycine) while a single thermal activation step only yields the cyclic dimer DKP. First it must be noted that the formation of long linear chains of glycine does not require more free energy expenditure than that of DKP, if expressed per amino acid residue. Available thermodynamic data suggest that the Gibbs free energy cost of every peptide bond is approximately constant, at +13 to +15 kJ.mol⁻¹, and the number of peptide bonds per amino acid residue is actually less in a peptide of length n (where it is equal to (n−1):n) than in DKP (1:1). In particular, opening DKP to linear Gly-Gly is a hydrolysis reaction and thermodynamically favorable in solution (6), at the silica/water interface, or even at the silica/gas interface in the presence of a high activity of water (7):

$$ {\mathrm{DKP}}_{\mathrm{aq}}+{\mathrm{H}}_2{\mathrm{O}}_{\mathrm{liq}}=\mathrm{Gly}-{\mathrm{Gly}}_{\mathrm{aq}} $$

(6)

$$ {\mathrm{DKP}}_{\mathrm{ads}}+{\mathrm{H}}_2{\mathrm{O}}_{\mathrm{liq}/\mathrm{gas}}=\mathrm{Gly}-{\mathrm{Gly}}_{\mathrm{ads}} $$

(7)

Thus, if DKP forms during the drying steps of wetting-and-drying cycles, it could hydrolyze during the subsequent rehydration steps. The latter are carried out at room temperature, but the silica surface, as mentioned above, is expected to catalyze hydrolysis reactions. We actually demonstrated this, although not for the hydrolysis of DKP_ads, but rather for that of Gly-Gly_ads. In TG, the latter species cyclizes to DKP at a temperature lower by 10 °C than monomeric Gly molecules: in this way, TG allows to quantify Gly-Gly_ads vs. Gly_ads when both are present together. As shown in Fig. 4, when 6 % Gly-Gly/Aerosil 380 was stored under room humidity, a significant hydrolysis to monomeric glycine had occured after 13 days at RT. In summary, DKP could well be an intermediate in the synthesis of linear peptides instead of constituting a dead end.

Fig. 4

DTG of a. 6 % Gly-Gly/Aerosil 380 immediately after preparation, b. the same after 13 days ageing at RT under 100 % humidity, air flow, β = 0.5 °C.min⁻¹

To check this idea, we reproduced Bujdák and Rode’s wetting-and-drying cycles on 1 mmol.g⁻¹ Gly/Aerosil samples. The samples were heated in a drying oven up to either 85 or 135 °C for one day, then wetted with 10 mL water: this constituted a wetting-and-drying cycle, which was repeated 7 times. The organic matter was then analyzed by MALDI-TOF. As shown in Fig. 5, the amounts of condensed molecules were too low to draw useful conclusions for activation at 85 °C, probably due to sluggish kinetics. When activation was carried out at 135 °C, on the other hand, the amount of DKP was observed to stabilize after one cycle, while the amount of linear Gly-Gly kept increasing. This behavior suggests that DKP is an intermediate in the formation of linear Gly-Gly from monomeric Gly.

Fig. 5

MALDI-TOF analysis of dimers formed from 7.5 % Gly/Aerosil 380 upon successive wetting-and-drying cycles. The evolution of the peak intensity ratioes between the two dimers and the monomer is shown; absolute values of these ratioes have no quantitative meaning

While our data on the evolution of these two molecules are compatible with those of Bujdák and Rode (1997), the sensitive MALDI-TOF technique also allowed us to identify longer peptides, chiefly tetraglycine which might form along, e.g.:

$$ \mathrm{Gly}-{\mathrm{Gly}}_{\mathrm{ads}}+{\mathrm{DKP}}_{\mathrm{ads}}=\mathrm{Gly}-\mathrm{Gly}-\mathrm{Gly}-{\mathrm{Gly}}_{\mathrm{ads}} $$

(8)

an “enthalpically neutral” reaction that does not involve water among the reagents, or the products. This oligopeptide was observed by Bujdák and Rode, but only when starting from preformed dimers, not from monomers.

Inorganic Phosphates on Silica

In this series of experiments inorganic phosphates have been deposited on an Aerosil 150 silica from aqueous solutions of KH₂PO₄ (a more complete series of experiments on Aerosil 380 has recently been decribed (Georgelin et al 2013)). Inorganic phosphate condensation can yield the diphosphate (pyrophosphate) along:

$$ {\left({\mathrm{H}}_2{\mathrm{P}\mathrm{O}}_4\right)}_{\mathrm{aq}}^{-}+{\left({\mathrm{H}}_2{\mathrm{P}\mathrm{O}}_4\right)}_{\mathrm{aq}}^{-}={\left({\mathrm{H}}_2{\mathrm{P}}_2{\mathrm{O}}_7\right)}_{\mathrm{aq}}^{2-}+{\mathrm{H}}_2{\mathrm{O}}_{\mathrm{liq}} $$

(9)

Condensation may also proceed further, yielding longer polyphosphates:

$$ \mathrm{n}{\left({\mathrm{H}}_2{\mathrm{PO}}_4\right)}_{\mathrm{aq}}^{-}=\mathrm{H}{\left({\mathrm{PO}}_3\right)}_{\mathrm{n}}{\left(\mathrm{OH}\right)}_{\mathrm{aq}}^{\mathrm{n}-}+\left(\mathrm{n}-1\right){\mathrm{H}}_2{\mathrm{O}}_{\mathrm{liq}} $$

(10)

Once formed, these polyphosphates might have phosphorylated organic biomolecules (Schwartz 2006); pyrophosphates (diphosphates) have been proposed as an early alternative to ATP (Holm and Baltscheffsky 2011).

The samples have been investigated by TG and ³¹P solid-state NMR. Bulk KH₂PO₄ exhibits a single peak at +4.1 ppm (Fig. 6a).

Fig. 6

³¹P NMR of a. bulk KH₂PO₄; b. 10%Pi/Aerosil 150; c. 10%Pi/Aerosil 150 after thermal activation at 160 °C under vacuum; d. the same after rehydration under air for one month; e. 10%Pi/Aero150 after thermal activation at 160 °C under ambient air

After adsorption on silica (10%Pi/Aero150, Fig. 6b), the main signal is narrow and centered at +1.1 ppm, which corresponds to (H₂PO₄)⁻ ions in aqueous solution in equilibrium with their conjugate base, (HPO₄)²⁻, as reported by several authors (Crutchfield et al. 1962; Yoza et al. 1994; Turner et al. 2003; Cade-Menun 2005). The weak peak at 4.1 ppm corresponds to a minor amount of bulk KH₂PO₄. A second signal is present at −8.8 ppm and corresponds to a minor amount of diphosphate already formed at 70 °C (see below).

After thermal activation at 160 °C of 10%Pi/Aerosil150 under vacuum, i.e., in conditions of low water activity (Fig. 4c), one observes a very broad peak with a maximum at about −6.25 ppm, and a composite signal with two maxima at −18.2 ppm and −19.8 ppm. The former falls in the chemical shift range of either diphosphate, or terminal phosphate groups in polyphosphates, while the latter two correspond to the “middle” groups of linear polyphosphates, based on solution NMR data (Yoza et al. 1994). In the case of linear polyphosphates longer than the tetraphosphate, one indeed expects several chemically non-equivalent phosphate groups with slightly different chemical shifts; the existence of two signals in this region of the spectrum indicates that polymerization proceeds up to rather long chains.

10%Pi/Aero150 activated under vacuum shows signals of roughly the same intensity in the “terminal phosphate” and “middle phosphate” ranges, but the monophosphate signal has disappeared.

After reexposure to room atmosphere for one month (Fig. 4d), a signal appears again in the monophosphate range (at +1.1 ppm). Another one is still present in the terminal phosphate range, but much narrower than before rehydration (at +7.7 ppm). In the middle phosphate range, only one signal remains (the one at −19.8 ppm); it is also narrowed, and its intensity is much reduced with respect to the state before rehydration.

Thermal activation at the same temperature of 160 °C, but under room atmosphere in an oven, led to a spectrum rather similar to that of the rehydrated sample (compare Fig. 4e with d), with sharp signals in the monophosphate and terminal phosphate ranges, the middle phosphate one being only of minor intensity. It is particularly instructive to compare the spectra of the oven-dried (4.e) and vacuum-dried (4.c) samples. The importance of the water activity upon heating is obvious, just as it was for Gly/SiO₂. Heating the supported monophosphate in an oven results in very little P-O-P bond formation, in stark opposition to heating under low water activity.

As regards the kinetics of the condensation reaction, catalysis by the silica surface is manifest for the Pi/SiO₂ system, as it was for Gly/SiO₂. Once again this is visible when comparing to the reactivity of the same compound in bulk form. Indeed, bulk KH₂PO₄ also polymerizes upon heating (probably even more extensively than when supported, since it seems to go to the stoechiometry KPO₃), but only at higher temperatures (between 230 and 340 °C).

AMP on Silica

AMP may not be an obvious prebiotic starting molecule because it already has one C-O-P bond. However its reactions are interesting to study in the present context.

Our initial results indicate that when bulk AMP (actually a monohydrate, Na₂AMP.H₂O) is heated in the 160–200° range, it first loses its hydration water and is then partially transformed into adenine, ADP and a minor amount of ATP. Adenine and ATP may be detected by XRD; ³¹P NMR spectra shown in Fig. 7 (compare 7 a and 7b) have signals attributable to Q² phosphorus in O₂P(OP)(OC) and O₂P(OP)(OC) environments, respectively, at −10 and −23 ppm (weak). This is consistent with the presence of AMP, ADP and a minor amount of ATP. The dismutation reaction yielding ADP can be written as:

Fig. 7

³¹P NMR of a. bulk AMP; b. bulk AMP, after thermal activation up to 200 °C; c. 5%AMP/Aerosil 380, after thermal activation up to 200 °C

$$ 2{\mathrm{AMP}}_{\mathrm{solid}}={\mathrm{Adenosine}}_{\mathrm{solid}}+{\mathrm{ADP}}_{\mathrm{solid}} $$

and it does not involve water, as a reagent or as a product. The corresponding reaction in solution is weakly favored with a ΔrG° of −20 kJ.mol⁻¹ at pH 7 and an entropic driving force (Alberty and Goldberg 1992). Thus, in opposition to the former two examples, it seems that drying does not involve a spectacular shift in reaction equilibrium. The same can be said for the reaction forming ATP, formally written as:

$$ 3{\mathrm{AMP}}_{\mathrm{solid}}=2{\mathrm{Adenosine}}_{\mathrm{solid}}+{\mathrm{ATP}}_{\mathrm{solid}} $$

For silica-supported AMP, ³¹P NMR shows that the same dismutation reactions are taking place, as the spectrum of thermally activated 5 % AMP/Aerosil380 (Fig. 7c) is very similar to that of bulk AMP activated in the same conditions. In contrast with the cases of Gly/silica and Pi/silica, there is no clear evidence that the tranformation occurs at a lower temperature, and thus no evidence that the surface plays a catalytic role. This absence of catalysis is not straightforwardly rationalized. Indeed, even if the overall reaction is not a condensation or a hydrolysis, it’s likely elementary steps most probably involve such a reactivity, as e.g. the mechanism shown in Fig. 8 that goes through the formation of diadenosine diphosphate (Ap2A) as an intermediate, and we would have expected the silica surface to interfere with one of these steps, or both.

Fig. 8

One possible mechanism for AMP dismutation to adenine and ADP. Note that the first step (formation of Ap2A) evolves one water molecule, while the second step consumes one

Discussion

In discussions of the interfacial reactivity of adsorbed biomolecules in prebiotic scenarii, many papers cursorily mention “a catalytic effect” of mineral surfaces, e.g. on amino acids condensation. We believe that a sound evaluation of these scenarii first necessitates a clear distinction between the thermodynamic and the kinetics aspects of the prebiotic reaction under study, catalysis being part of the latter.

Let us consider the thermochemistry first. The condensation of amino acids to peptides is an example of a reaction that is forbidden in aqueous solution but has been observed in the adsorbed state in many studies. This is because it involves the formation of one water molecule, and thus can be made exergonic simply by lowering the activity of water, i.e., by drying. In those studies where substantial yields of oligomerization have been observed, it was the drying, not the surface, that made this reaction thermodynamically possible. As we have seen in the present paper, the same holds for the formation of polyphosphates from monophosphates on silica, since this is also a condensation accompanied by the elimination of water. Here too, controling the water activity upon drying is crucial as thermal treatment in an oven is inefficient as compared to treatment under dry air flow. These findings can probably be generalized to other such reactions.

The polymerization of nucleotides to RNA by formation of phosphodiester bonds is one other condensation with water elimination. Note that in the often quoted works by Ferris and Ertem on nucleotide/clay systems (Ferris et al. 1989; Ferris and Ertem 1993; Ferris et al. 1996; Ertem and Ferris 1997, 1998, 2000), the thermodynamic difficulty was bypassed by the use of activators such as carbodiimides (Ferris et al. 1989) or phosphorimidazoles (Ferris and Ertem 1993) which make polymerization favorable even in solution. To the best of our knowledge, the unactivated polymerization of nucleotides on clays has not been attempted. In our hands, the reaction of AMP on silica did only result in a dismutation to adenosine and ADP (and minor ATP), a reaction which neither produces nor consumes water, and for which the effect of dehydration is not straightforward, but the results on this system are only preliminary.

Then comes the question of reaction kinetics, and to study this point polymerization in the adsorbed state can be compared to polymerization in the bulk solid. A catalytic effect of the silica surface is evident as the polymerization temperature of glycine is lowered by 80 °C when adsorbed; interesting molecular modeling studies cast some light on the molecular mechanism of this catalytic effect (Rimola et al. 2006). For adsorbed phosphates too, the polymerization temperature is lowered, by more than 50 °C as compared to the bulk. Thus, it is perfectly correct to say that e.g. silica (and no doubt other minerals) catalyzes the dehydrating polymerization of biomolecules, as long as one does not forget to address the thermodynamical aspect first.

One practical conclusion of the present results for further studies is that greater care must be taken to control water activity in the gas phase during drying treatments. The critical water activity could be determined by studies in controlled atmosphere with different water partial pressures. At the same time, knowledge of apparent activation energies can tell what treatment duration in the “dry phase” is needed as a function of the treatment temperature. Together, these elements will allow to determine if polymerization is a likely outcome of realistic climatic variations.

Finally, it must be noted that drying is not the only way of decreasing water activity. It also decreases in solutions with a high ion strength, and especially in the presence of complexing ions, as is the case in the salt-induced peptide formation reaction (SIPF, e.g. Plankensteiner et al 2004). In the same way, the phosphorylation of biomolecules by inorganic phosphate may be favored when both are dissolved in a non-aqueous solvent (Schoffstall 1976).

Conclusion

Several biomolecule condensation reactions involving the elimination of water can be induced by adsorption on silica followed by drying in rather mild conditions, provided the water activity is low enough. The result is twofold: the storage of chemical energy than can later be released by hydrolysis if the adsorbed molecules are lixiviated into the solution, and an ascent toward the complexity of biopolymers. It must be remembered that the surface helps in speeding up the condensation kinetics, but it is the drying step that makes this reaction possible at all. Controlling the drying conditions allows to obtain quantitative condensation and study the reaction products more easily. Further studies are underway to evaluate the potential for chemical selectivity in more complex systems.