Antagonistic effects of amino acids support abiotic nano-environments in clay

TL;DR

This study shows how amino acids interact with clay to create nano-environments that may have supported early life chemistry.

Contribution

The research reveals antagonistic effects of amino acids on clay structure, creating diverse abiotic nano-environments.

Findings

Lysine and arginine suppress clay swelling by intercalating between layers.

Gamma-aminobutyric acid causes layer distortion and nanocavity formation in clay.

The opposing effects of amino acids sustain diverse nano-environments in clay.

Abstract

Prebiotic chemistry in nano-environments confined within catalytic mineral media is emerging as a promising frontier in origins-of-life research. Such confined spaces exhibit physicochemical properties distinct from bulk conditions, enabling out-of-equilibrium processes such as condensation reactions in aqueous media. Here, we demonstrate that a yin-yang interplay of organo–clay interactions generates and supports a variety of confined geochemical nano-environments within clay layers and facilitates the persistence of partial exfoliation. We investigate how the structure of Ca-montmorillonite clay is affected by exposure to aqueous amino acid mixtures containing proteinogenic species (L-lysine or L-arginine) and their 1:40 mixture with the meteorite-common, non-proteinogenic \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\gamma$$\end{document}-aminobutyric acid. Using attenuated total reflectance Fourier transform infrared spectroscopy, X-ray diffraction, thermogravimetric analysis and electron microscopy, we reveal opposing effects of the different amino acids on the clay’s structure. Lysine and arginine intercalate between clay layers, bridge across them and suppress swelling, whereas \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\gamma$$\end{document}-aminobutyric acid induces pronounced layer distortion, exfoliation and nanocavity formation, even in the presence of the strongly-interacting species. The contrasting, complementary effects between fixation of the interlayer and formation of nano-compartments create and sustain diverse nano-environments. This balance could play a role in shaping nano-environments that facilitate prebiotic polymerisation and other chemistry central to life’s emergence on rocky bodies.

The online version contains supplementary material available at 10.1038/s41598-026-41502-6.

Figures

Schematic of Ca-montmorillonite (Ca-MMT) 2:1 aluminosilicate clay structure. The layered aluminosilicate units stack on top of each other to create a repeating structure with interlayer spaces, where (hydrated) cations reside and balance the negatively charged basal siloxane surfaces. The orientations of the Si–O–Al out-of-plane and Si–O–Si in-plane stretching vibrations in relation to directions within the crystal structure are also depicted.

Different spectral behaviour in normalised IR spectra of dehydrated Ca-MMT exposed to amino acids. In the organics fingerprint region 1300–1700 $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document}$ , spectra are shown for Ca-MMT control (dot-dashed grey) alongside samples exposed to a L-lysine (Lys) 0.75 mmol/g (green) and a mixture of Lys 0.75 mmol/g + GABA 30 mmol/g (light blue) or

Second-derivative curves of normalized IR spectra of dehydrated Ca-MMT exposed to amino acids, focusing on the high-frequency region of the Si–O stretching band envelope. Derivative spectra are shown for Ca-MMT interacted with MilliQ water (control, dot-dashed grey) and exposed to GABA 30 mmol/g (dotted purple, data from Ref. [16](#CR16)) alongside samples exposed to a L-lysine (Lys) 0.75 mmol/g (green) and a mixture of Lys 0.75 mmol/g + GABA 30 mmol/g (light blue) and b L-arginine (Arg) 0.75 mmol/g (red) and a mixture of Arg 0.75 mmol/g + GABA 30 mmol/g (blue). Each derivative spectrum encomp

Normalised XRD d(001) peaks of Ca-MMT exposed to amino acids. Plotted are XRD d(001) peaks of dehydrated Ca-MMT not exposed to amino acids (control, dot-dashed grey), exposed to GABA 30 mmol/g (dotted purple, data from ref. [16](#CR16)) alongside Ca-MMT exposed to a L-lysine (Lys) 0.75 mmol/g (green) and a mixture of Lys 0.75 mmol/g + GABA 30 mmol/g (light blue) b L-arginine (Arg) 0.75 mmol/g (red) and a mixture of Arg 0.75 mmol/g + GABA 30 mmol/g (blue). Further plotted are XRD d(001) peaks of rehydrated Ca-MMT samples, prepared by subsequent exposure of the same dehydrated samples to water v

TEM images of Ca-MMT’s ordered and distorted layered structures including nanocavities after exposure to amino acids. a Control sample showing characteristic layering; ordered yet with slight distortions, considered as ordered for this clay material. b Sample exposed to single-organic Lys showing ordered layering only. Clay samples exposed to organic mixtures of c Lys-GABA d Arg-GABA; showing highly-distorted layering and nanocavities with well-defined lattice fringes that are embedded within distorted layered structures. In all panels, ordered layering, distorted layering, and nanocavities ar

Full text

Introduction

Over the past decades, the enigma of the origin of life has largely been guided by contemporary biochemistry and constrained by our knowledge of past geological and geochemical conditions^1–3^. Prebiotic polymerisation of organic monomers, such as amino acids and nucleotides, stands centre stage alongside substantial focus on critical functions of life such as catalysis and containment^4–7^. In terms of the relevant prebiotic environments in which such chemistry may have taken place, hot springs undergoing wet-dry cycling^8,9^ and submarine alkaline hydrothermal systems^10,11^ are favourable scenarios. While the characteristics of the bulk environment are often considered, some publications have also recognised the unique conditions that arise under inorganic nanoconfinement and the possibilities that such nano-environments could offer, for example, for prebiotic condensation reactions in water^12–16^. In nanoconfinement, the behaviour of water is characterised by reduced dielectric permittivity and mobility, largely dictated by pore size^17–19^. The (inorganic) sorption behaviour in these spaces is further governed by surface chemistry and Gibbs free energy of solvation^20^. In nanopores, increased sorption of inorganic cations is observed alongside their dimerisation under partial dehydration^21,22^. Physicochemical conditions such as pH, heat capacity and conductivity, also differ significantly from the external environment^20,23,24^. Thus, nanoconfinement in inorganic media can facilitate a natural disequilibrium that is essential for origins-of-life processes. Furthermore, nanoconfinement is particularly compelling in view of a central challenge in prebiotic chemistry: emergent or amplifying processes that are promising in simple, few-component systems frequently do not withstand the combinatorial diversification that follows when more components are added^25^. Within nanoconfined environments, however, the survivability of these processes may be enhanced, even when alternative pathways are present. The perspectives of chemistry under confinement within origin-of-life scenarios give rise to important questions, including what is the relationship between the characteristics of the nanoconfined regions and the organics that are present?

Insight into prebiotic organo-mineral sorption interactions is fundamental to the understanding of the role of minerals in prebiotic processes^26^. The sorption of organic monomers onto and within catalytic clays is a crucial preceding step to polymerisation that has been widely studied; also specifically in the context of peptide-forming (proteinogenic) amino acids^27–32^. Clays that (reversibly) swell, such as montmorillonite (MMT) (Fig. 1), have received considerable attention for their capacity to catalyse polymerisation under wet-dry cycling conditions^33–38^. The mechanism of this process is not unequivocally determined, yet the confined interlayer region between the clay’s negatively charged silica sheets, which can incorporate organic molecules and gain/lose water depending on external conditions, is likely to play a role^13^. Absorption and/or intercalation of organics are reported to change the size of the interlayer space^39^, which in turn affects confined water properties. In single-organic studies, positively charged proteinogenic amino acids such as L-lysine (Lys) and L-arginine (Arg), are found to preferably intercalate within MMT’s interlayer galleries compared with other proteinogenic species over a wide pH range^40–42^. This is proposed to occur via electrostatic interactions and cationic exchange, as well as H-bonding and proton transfer processes^41–48^. The incorporation of these species into the interlayer space of MMT changes the spectral signature of clay^47^ and the interlayer response to stress, becoming stiffer in response to tensile forces^49^. It is also known that organo-MMT is generally harder to exfoliate under intense mechanical motion (i.e. high-energy ball milling) relative to unmodified clay^50,51^. These observations suggest that strongly-interacting amino acids exert a stabilising effect on the clay’s structure. Thus, the consequences of sorption processes are highly variable and can affect further monomer interactions and polymerisation. Non-proteinogenic amino acids that possess a different interaction potential with the clay can induce other effects within the interlayer space. Previous work from our group^16^ demonstrates that the meteorite-common \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\gamma$$\end{document} -aminobutyric acid (GABA) interacts in a profoundly different way with MMT clay compared with Lys and Arg, despite their structural similarity (GABA possesses a protonated amino group in its zwitterionic state at pH \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim$$\end{document} 4–10). The weakly-interacting GABA instead induces an irreversible distortion by partially exfoliating the mineral and forming nano-compartments within the interlayer space. Whilst these single-component studies demonstrate the variety of effects that different organic species can have on confined nano-environments within MMT, the prebiotic chemical environment is expected to be organically diverse. Given the differences in effects and interaction modes with the clays, this implies that different species may override the abilities of others to modify the interlayer space in an antagonistic manner when multiple organic species are present.Fig. 1. Schematic of Ca-montmorillonite (Ca-MMT) 2:1 aluminosilicate clay structure. The layered aluminosilicate units stack on top of each other to create a repeating structure with interlayer spaces, where (hydrated) cations reside and balance the negatively charged basal siloxane surfaces. The orientations of the Si–O–Al out-of-plane and Si–O–Si in-plane stretching vibrations in relation to directions within the crystal structure are also depicted.

Organic mixtures introduce the possibility not only of antagonistic interactions, but also synergistic interactions between the organic components and clay. These may suppress or enhance sorption, thereby influencing the confinement properties of the clay, with knock-on effects on the confined chemical environment and subsequent processes such as polymerisation^28^. Only a few studies have explored sorption processes in clays with more than one amino acid present in the solution. These demonstrate that organic selection can occur, driven by proteinogenic amino acid charge at particular pH, which dictates its ability to interact with the silica sheets and undergo subsequent polymerisation^52^. Prebiotic mixtures likely contained organics of varied functionality and origin; for example, meteoritic impacts may have seeded terrestrial ponds with extraterrestrial compounds, creating transiently higher organic concentrations^53^. Therefore, in addition to known proteinogenic organics, it is important to also consider space-born and non-proteinogenic species. One study compared the adsorption of proteinogenic and non-proteinogenic amino acids using predominantly isomeric mixtures, where the molecular structures differ only by the placement of a side group. That work reported a preference for adsorption of certain non-proteinogenic species^54^. In contrast, synergistic interactions that enhance the overall adsorption capacity of clays have been observed when minerals were pretreated with organics of lower surface affinity^55^. However, none of these studies assessed the effect of the organics on the clay structure, and moreover, did not regard the resiliency of different effects on the interlayer region in presence of other competing processes; an area of central importance to prebiotic chemistry.

In this work, we individually and simultaneously expose Ca-MMT to amino acids with differing strengths of interaction and opposing effects within the interlayer space, to study their modifications to the clay’s aluminosilicate structure and their robustness in more complex conditions. Our mixtures combined a proteinogenic amino acid with a meteorite-common, non-proteinogenic counterpart, using a 1:40 proteinogenic-to-meteoritic concentration ratio to mimic meteor impact-related concentration imbalances^56^. The exposure solutions consisted of either a single amino acid (Lys or Arg) or their mixture with excess GABA (Lys-GABA, Arg-GABA). Clay samples were treated with these solutions and subsequently characterised after washing and dehydration (‘exposed samples’). The same samples were further examined following rehydration with water vapour (‘rehydrated samples’). The chemical and structural responses of the clay’s aluminosilicate network to interaction with Lys/Arg and GABA were characterised by attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR), thermogravimetric analysis and powder X-ray diffraction (pXRD). Complementary transmission electron microscopy (TEM) imaging revealed the resulting physical change at the nanoscale. Together these analyses show that the antagonistic effects of amino acid interaction induce irreversible modifications of the interlayer spaces and sustain the formation of nano-compartments in MMT.

Results

Sorption behaviour of single-organic vs. mixture suspensions

Figure 2 shows the vibrational signatures of Ca-MMT clay and sorbed organics after the suspension exposure treatment. Spectra of the clay are shown exposed to water only (grey), Lys (green), Lys mixed with GABA (light blue), Arg (red) and Arg mixed with GABA (blue). All samples exposed to the proteinogenic amino acids exhibit absorption signatures, irrespective of the presence of GABA, as can be seen in Fig. 2a, b in the fingerprint region at 1300–1700 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} . The proteinogenic species Lys and Arg are clearly retained on the clay after washing; an indication of their strong interaction. We observe significant shifts in organic band-positions in the interfacial spectra relative to the free dissolved amino acids as present in the initial single-organic exposure solutions (Tables S1, S2). These shifts indicate a change in the coordination environment of the proteinogenic species when retained on/within MMT. In contrast, any organic bands related to GABA are completely absent from the interfacial spectra. This observation is consistent with our previous research^16^, where the weakly-interacting nature of GABA with MMT led to its easy removal by water washes. Interestingly, despite not being sorbed itself, exposure to GABA seems to affect the sorption of Lys and Arg. While sorbed band positions remain similar, a reduction in the relative contribution of the asymmetric \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {COO}^-$$\end{document} stretching band from Lys and Arg around 1595 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} is observed when GABA is incorporated in the exposure suspensions (blue curves). Other, weaker features also decrease in intensity, giving rise to a smoother curve with less pronounced organic signatures.Fig. 2. Different spectral behaviour in normalised IR spectra of dehydrated Ca-MMT exposed to amino acids. In the organics fingerprint region 1300–1700 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} , spectra are shown for Ca-MMT control (dot-dashed grey) alongside samples exposed to a L-lysine (Lys) 0.75 mmol/g (green) and a mixture of Lys 0.75 mmol/g + GABA 30 mmol/g (light blue) or b L-arginine (Arg) 0.75 mmol/g (red) and a mixture of Arg 0.75 mmol/g + GABA 30 mmol/g (blue). In the mineral’s absorbance region 450–1250 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} , spectra are shown for MMT control and c Lys and its mixture with GABA and d Arg and its mixture with GABA, with the respective colour code. Each spectrum is an average of three experimental repetitions (normalised to the Gaussian-fitted band around 517.7 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} marked with an asterisk), where the highlighted margins represent the corresponding standard deviations. Band assignments^41–45,47,48,57–60^ are shown using dashed vertical lines in positions found with derivative analysis of MMT control (black), single-organic Lys (green) and Arg (red). Str. = stretching, bend. = bending, sym. = symmetric, asym. = anti-symmetric, oop. = out-of-plane.

The IR signatures of the clay samples also exhibit notable differences after exposure to the single-organic amino acid and mixture solutions. This is most clearly evident in the absorbance region of the mineral at 450–1250 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} , where bands that originate from vibrations of the organic molecules themselves are not expected (Fig. 2c, d). The most significant alteration occurs in the contour of the broad Si–O stretching band envelope (950–1150 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} ), reflecting modifications in the aluminosilicate network. This envelope is generated by the presence of four main bands (the centres of which—for the control sample—are highlighted with dashed vertical lines) and relates to the structural state of the clay’s aluminosilicate network^61^, schematically shown in Fig. 1. Sorption of proteinogenic amino acids from the single-organic suspensions is found to slightly affect the contour of the band envelope, with Arg (red curve) showing a more pronounced effect than Lys (green curve). The shape of the broad feature is evidently changed after exposure to the GABA-containing solutions regardless of which proteinogenic amino acid was present. GABA exposure significantly increases the high-frequency bands that form the Si–O stretching band envelope and the bending vibration of Si–O–Si at 460 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} . GABA exposure also weakens the Al–OH–Al bending band at 916 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} and the Al–OH–Fe bending band at 886 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} . This behaviour is consistent with our previous work in which we report on MMT clay interacting with GABA alone^16^. It is clear that transiently and weakly-interacting GABA disrupts the clay’s structure, even in the presence of strongly-interacting, proteinogenic amino acids. To elucidate the combined impact of structural disruption and sorption on the clay, we performed a detailed examination of the Si–O stretching band envelope using second-derivative analysis.

Organo-clay interactions influence the structural state of clay

Figure 3 presents the second-derivative curves for the IR spectra from Fig. 2 within the high-frequency region of the Si–O stretching band envelope (1060–1130 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} ), highlighting the individual band positions identified from the local minima (listed in Table S3). The more pronounced the minimum in the second derivative, the sharper the signal in the original IR absorbance data^62^. The Si–O–Al out-of-plane stretching, a vibrational mode perpendicular to the layered structure (Fig. 1), is particularly sensitive to the state of the clay’s aluminosilicate network. We identify three Si–O–Al out-of-plane stretching frequencies, which are referred to as: I. sorption frequency; indication of sorbed molecular species around 1118 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} , II. dehydrated frequency around 1101 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} , and III. exfoliation frequency around 1085 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} . For Lys (green) and Arg (red) single-organic treatments, the strongest Si–O–Al out-of-plane mode is located at the sorption frequency I. In contrast, the water-exposed control shows a dominant feature at the dehydrated frequency II (dashed grey). Furthermore, the position of the Si–O–Si in-plane mode is shifted by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 20$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 1087$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} upon Lys sorption and to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 1084$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} upon Arg sorption. The dominance of the sorption frequency in the single-organic experiments points to absorption of the proteinogenic species as a key factor modifying the clay’s aluminosilicate network, consistent with previous work^47^. This is reflected in pronounced shifts of both out-of-plane and in-plane Si–O vibrations, leaving a distinct trace in the clay’s IR response.Fig. 3. Second-derivative curves of normalized IR spectra of dehydrated Ca-MMT exposed to amino acids, focusing on the high-frequency region of the Si–O stretching band envelope. Derivative spectra are shown for Ca-MMT interacted with MilliQ water (control, dot-dashed grey) and exposed to GABA 30 mmol/g (dotted purple, data from Ref. 16) alongside samples exposed to a L-lysine (Lys) 0.75 mmol/g (green) and a mixture of Lys 0.75 mmol/g + GABA 30 mmol/g (light blue) and b L-arginine (Arg) 0.75 mmol/g (red) and a mixture of Arg 0.75 mmol/g + GABA 30 mmol/g (blue). Each derivative spectrum encompasses an average of three experimental repetitions with their respective standard deviations displayed as highlighted margins. The local minima correspond to the band positions of the Si–O–Al out-of-plane stretching (oop.; I=sorption, II=dehydrated, III=exfoliation) and Si–O–Si in-plane stretching (ip.) vibrational modes of Ca-MMT based on second-derivative analysis (values are given in Table S3). The band positions are marked with a square for dehydrated, a circle for sorbed and a star for exfoliated (empty for oop. and filled for ip.), as given in the legend alongside each panel.

The second-derivative spectra of clays exposed to GABA-containing mixtures exhibit a markedly different shape (blue curves, Fig. 3). In these cases, the most prominent negative amplitude occurs at the exfoliation frequency III around 1085 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} , also present at a similar position for clay exposed to GABA-only from our previous work (purple curve)^16^. For the GABA-containing suspension exposures, the minima are four to five times greater than those of the single-organic exposures. Since the amplitude of the second derivative is inversely proportional to the square of the band’s full width at half maximum (FWHM)^62,63^, this indicates an additional signal in the clay exposed to both GABA-containing mixtures around 1085 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} , that is roughly twice as narrow as the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 1087$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 1084$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} in-plane bands of Lys-only and Arg-only exposures, respectively. The differing band widths distinguish contributions of two separate stretching modes: Si–O–Si in-plane at the sorption frequency and Si–O–Al out-of-plane at the exfoliation frequency III. Importantly, exfoliation frequency III is observed only after exposure to GABA-containing mixtures. Thus, the second-derivative analysis confirms the presence of (partially) exfoliated domains in clays treated with GABA-containing mixtures, even in the presence of absorbed proteinogenic amino acids.

Combined structural effect of exfoliation and intercalation

The effects of amino acid incorporation into the interlayer space of clays (intercalation) are typically assessed using X-ray diffraction^40,42^. We examined the Ca-MMT samples exposed to different proteinogenic acids and their mixtures with GABA using this technique, focusing on the analysis of the d(001) peak, shown in Fig. 4. The dehydrated Ca-MMT control exhibits a diminished and wide d(001) peak that corresponds to a basal spacing that is larger than the 2:1 layer ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 9.5$$\end{document}  Å^64^), consistent with collapsed layers upon partial dehydration due to water loss from the interlayer space characteristic to this material^65^. Clay samples exposed to GABA only, as recorded in our previous study^16^ and plotted here for comparison, demonstrate a further reduction in peak intensity which suggests reduced layer ordering relative to the control. In contrast, clay exposed to single-organic Lys (green curve, Fig. 4a) and Arg (red curve, Fig. 4b) solutions, retain their d(001) peaks. Both Lys and Arg remain intercalated within Ca-MMT’s interlayer space, reinforcing its layered structure even in the dehydrated state. Basal spacing values are smaller compared with the control and GABA-only exposures. The decrease is consistent with Ref. 43 who inferred a semi-perpendicular orientation of Lys molecules within the interlayer of Ca-MMT (for concentrations \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\le 25$$\end{document}  mmol/g). In addition, we observe a significantly increased intensity, suggesting a more ordered structure that must be induced by the retained Lys or Arg molecules. When the clay is exposed to Lys-GABA or Arg-GABA mixtures, the d(001) peak exhibits a reduced intensity (blue curves) compared with the single-organic proteinogenic exposures, consistent with reduced layer ordering. Basal spacing values of the dehydrated samples remain similar; around 12.6 Å and 12.9 Å for Lys, Lys-GABA and Arg, Arg-GABA, respectively. In agreement with the IR spectra, the pXRD findings show that intercalation stabilises the layered structure but cannot fully counteract the disruptive effect of GABA.Fig. 4. Normalised XRD d(001) peaks of Ca-MMT exposed to amino acids. Plotted are XRD d(001) peaks of dehydrated Ca-MMT not exposed to amino acids (control, dot-dashed grey), exposed to GABA 30 mmol/g (dotted purple, data from ref. 16) alongside Ca-MMT exposed to a L-lysine (Lys) 0.75 mmol/g (green) and a mixture of Lys 0.75 mmol/g + GABA 30 mmol/g (light blue) b L-arginine (Arg) 0.75 mmol/g (red) and a mixture of Arg 0.75 mmol/g + GABA 30 mmol/g (blue). Further plotted are XRD d(001) peaks of rehydrated Ca-MMT samples, prepared by subsequent exposure of the same dehydrated samples to water vapour (pale-blue background), using the same colour code for the control, GABA-only exposure and c Lys and its mixture with GABA and d Arg and its mixture with GABA. All diffractograms are normalised to the area of a dehydrated/rehydrated control sample’s diffraction peak in the range \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$34.5^\circ - 39^\circ \,( 2\theta )$$\end{document} . Each diffractogram represents the average of three replicates, with shaded margins indicating the corresponding standard deviation. Basal spacings and FWHM values calculated using fit parameters of asymmetric Lorentzian functions (see Methods) are listed in the order: control, GABA-only, Lys/Arg, Lys/Arg + GABA, with colour code of the corresponding curve.

Rehydration of the samples with water vapour served to evaluate the extent to which Ca-MMT’s swelling capacity can be restored (Fig. 4c, d). In the control sample, water intercalation fully restores the d(001) peak, consistent with the reversible swelling behaviour of MMT and its reported average basal spacing of 14.5–16  Å^64^. For samples intercalated with Lys (green curve) or Arg (red curve), the rehydrated d(001) peak is shifted relative to the control, indicating reduced interlayer spacings by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 1.5$$\end{document}  Å and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 1.8$$\end{document}  Å respectively. Furthermore, Lys intercalation produced a markedly lower diffraction peak intensity upon rehydration, whereas the effect for Arg is less pronounced. Comparison to their dehydrated counterparts shows that Lys intercalation permits only \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 1$$\end{document}  Å of swelling, while Arg allows merely \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 0.3$$\end{document}  Å of expansion upon interlayer water uptake. These results suggest that intercalated Lys and Arg restrict the swelling capacity of the clay by holding the clay layers together and limiting the influence of water molecules on the spacing of the basal planes. Thus, the amino acids act as bridging agents between the layers, binding the opposite silicate surfaces that define the interlayer space (Fig. 1).

Another observation is the increased width of the rehydrated d(001) peaks when Lys is intercalated, consistent with previous findings^42^. The broadness of the clay’s d(001) peak reflects the distribution of interlayer environments and, in this context, suggests greater variation in regions that are more or less affected by proteinogenic intercalation. On average, single-organic Lys exposure produced a broader peak (FWHM = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1.4\pm 0.1\,^\circ$$\end{document} ) than either Arg (FWHM = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1.0\pm 0.2\,^\circ$$\end{document} ) or water-only (FWHM = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1.1\pm 0.2\,^\circ )$$\end{document} exposures. This effect may arise from increased heterogeneity in the interlayer environment when both water and Lys are present, a phenomenon less evident in the cases of Arg and the control.

Rehydrated clays exposed to the GABA-containing mixtures exhibit reduced d(001) peak intensities for both Lys-GABA and Arg-GABA exposed samples relative to the single-organic proteinogenic counterparts (Fig. 4c, d). This indicates that interaction with GABA in the mixture-exposure experiments irreversibly alters the clay structure, producing a less-ordered structure consistent with our previous findings^16^. The analyses of both the dehydrated and rehydrated samples thus indicate that the stabilising effect of the proteinogenic amino acids is lessened by the irreversible layer distorting effect of GABA. In the case of the Lys-GABA mixture (light-blue curve), basal spacing is reduced by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 0.4$$\end{document}  Å relative to the Lys exposure (green curve), while the FWHM remains comparable but exhibits twice the standard deviation. For Arg intercalation, exposure to GABA has a much weaker effect on the basal spacing (blue compared with red curve). Since ATR-FTIR intensities are similar for the Lys and Lys-GABA samples (Fig. 2a), the reduced basal spacing arises from factors other than Lys absorption. As stated above, the rehydrated diffractogram of the clay exposed to single-organic Lys is more strongly affected by water sorption than that of clay exposed to single-organic Arg. When GABA is also present, this difference between Lys and Arg becomes even more pronounced. In the Lys-GABA case, the d(001) peak changes shape relative to the single-organic Lys exposure: the intensity at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 6^\circ \,(2\theta )$$\end{document} decreases relative to that at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 6.7^\circ \,(2\theta )$$\end{document} (Fig. 4c). This suggests that some interlayer regions are less affected by Lys intercalation and therefore more susceptible to alteration by GABA. Therefore, the findings suggest that Arg binds to the clay more uniformly than Lys.

While single-organic exposure experiments provide valuable insights into system behaviour, the mixture experiments amplify these differences and allow new perspectives. The XRD results are corroborated by thermogravimetric measurements (Fig. S1), which show decreased desorption/degradation temperatures for both proteinogenic species when sorbed to the clay from GABA-containing mixtures, where again, the effect of GABA is greater for Lys than for Arg. Overall, the XRD findings demonstrate that GABA can irreversibly disrupt the clay structure even in the presence of intercalated Lys or Arg when GABA is present in a large excess. In this chosen 1:40 proteinogenic to meteorite-common concentration ratio based on ref. 56, the exfoliation and intercalation effects act in competition and are of comparable magnitude. This highlights the importance of balance between complementary, opposing processes, which together enable the manifestation of both effects.

Confined prebiotic nano-environments in clay

To further investigate the nanoscale physical nature of the structural changes, TEM was employed to directly image of the clay’s layered structure and its modifications. Previous work has revealed that GABA interaction with Ca-MMT induces extensive layer distortion and the formation of nanocavities within the clay^16^. In this work, the same method is used to examine the rehydrated clays exposed to the single-organic and GABA-containing mixture solutions (see Methods for procedure, Table S4 for overview of the distorted layering analysis results and Figure S2 for the statistical analysis). Distorted layering is found in 36% of the instances where layered structures are also observed in the control sample and decreases slightly to 25% and 24% after exposure to Lys or Arg single-organic solutions, respectively, a change that remains within the experimental uncertainty (Table 1 and Fig. 5). Both the Ca-MMT control (Fig. 5a) and the Lys-exposed (Fig. 5b) samples exhibit mostly ordered layering. In contrast, clays treated with GABA-containing mixtures (Fig. 5c, d) show a statistically significant increase in distorted layering (highlighted by red arrows) relative to the control—from 36% to 70% for Lys-GABA and to 64% for Arg-GABA (Table 1, Figure S2). Each of these higher proportions each have a binomial p-value of 0.004 relative to the control sample. Furthermore, statistically significant increase in distorted layering is demonstrated for each of the mixture exposures relative to the proteinogenic-only exposures, with p-values of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$6*10^{-5}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1*10^{-5}$$\end{document} , for Lys/Lys-GABA and Arg/Arg-GABA pairs, respectively. GABA-induced distorted layering percentages in this work are similar to the 65% found in Ref. 16 regardless of the GABA to clay ratio being twice as high. The TEM analysis reinforces the XRD findings, which also indicate enhanced disorder in clays exposed to GABA-containing mixtures compared with the control and the corresponding single-organic proteinogenic exposure (Fig. 4), and further demonstrates that excess GABA induces nanoscale layer distortions in presence of strongly-interacting amino acids.Table 1TEM image analysis of layered structures and nanocavities in rehydrated Ca-MMT samples after exposure to amino acids.Exposure suspension \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$D_l$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$C_l$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$C_{avg}$$\end{document} control: MilliQ water36% (5)0% (0)–Lys 0.75 mmol/g25% (4)0% (0)–Lys 0.75 mmol/g + GABA 30 mmol/g70% (14)15% (3) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1.44\pm 0.18$$\end{document}   \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,nm}$$\end{document} Arg 0.75 mmol/g24% (4)0% (0)–Arg 0.75 mmol/g + GABA 30 mmol/g64% (18)14% (4) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1.82\pm 0.26$$\end{document}   \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,nm}$$\end{document} ^a^Where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$D_l$$\end{document} = distorted layered structures relative to all layered structures, in percentage (no. of instances), \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$C_l$$\end{document} = instances with cavities embedded within distorted structures relative to all layered structures, in percentage (no. of instances), and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$C_{avg}$$\end{document} = average cavity size and standard deviation. See Methods for detailed image analysis procedure and limitations.^a^One instance is a clear outlier and is excluded from the cavity count and size measurements. It exhibits an area with multiple cavities with an average cavity size of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$4.13\pm 1.04$$\end{document}   \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,nm}$$\end{document} , compatible with exposure to GABA only.^16^

Fig. 5TEM images of Ca-MMT’s ordered and distorted layered structures including nanocavities after exposure to amino acids. a Control sample showing characteristic layering; ordered yet with slight distortions, considered as ordered for this clay material. b Sample exposed to single-organic Lys showing ordered layering only. Clay samples exposed to organic mixtures of c Lys-GABA d Arg-GABA; showing highly-distorted layering and nanocavities with well-defined lattice fringes that are embedded within distorted layered structures. In all panels, ordered layering, distorted layering, and nanocavities are marked with white, red and black arrows, respectively. See Methods for detailed image analysis procedure and limitations.

However, the disruptive effect of GABA on the clay structure extends beyond layer distortion. TEM analysis strikingly reveals nanocavities embedded within distorted layers in 15% and 14% of instances for clays exposed to Lys-GABA and Arg-GABA mixtures, respectively (Table 1). The nanocavities (black arrows) are formed despite bridging and ordering effects of intercalated proteinogenic species. By contrast, no cavities are confirmed in clay samples exposed to single-organic Lys or Arg nor in the control sample, where ordered layered structures (white arrows) dominate. TEM findings further corroborate the ATR-FTIR (Figs. 2, 3) and XRD (Fig. 4) results, which point to partially-exfoliated domains in Ca-MMT after exposure to GABA-containing mixtures. TEM shows that not only does partial exfoliation take place, but also mid-layer nanocavities form in correlation with this process, even in the presence of strongly-interacting proteinogenic species.

Discussion

We reveal the nature of organo-clay interactions between amino acid mixtures containing species of different origins and functionalities and a layered clay structure. The strongly-interacting proteinogenic Lys and Arg, when present as single components in the exposure suspension, intercalate within the interlayers, restrict interlayer swelling, and impose a layer-ordering effect. In clays exposed to Lys/Arg mixtures with the meteorite-common GABA, which is known to induce layer distortion, exfoliation and nanocavity formation^16^, the ultimate structural response of the clay reflects both of the antagonistic effects of amino acids present. Distorted clay layering is observed in correlation with nanocavities that are embedded mid-layer in TEM images. For the Lys or Arg mixtures with GABA, the opposing effects seem to be competitive: one reinforces the layered structure, while the other separates it, and in this concentration ratio of 1:40 both act with comparable magnitude. We chose a proteinogenic amino acid to clay ratio that is similar to that used in a recent, prebiotically-relevant study examining lower-end ATR-FTIR detectable concentrations for Lys/Arg^42^. The same sorption mechanism governs these and lower Lys/Arg to clay concentrations^45^, thus it is reasonable to explore prebiotic sorption behaviour using these concentrations. Assuming that Lys/Arg and GABA form in planetary and meteoritic settings, correspondingly, their concentration ratio in a meteorite impact-influenced pond is expected to be between 1:20 and 1:100^56^, where our choice of 1:40 proteinogenic-to-meteoritic ratio represents an intermediate concentration ratio. Our findings thus suggest that the formation and survival of GABA-induced nano-compartments with different confining properties is expected to be regulated by the concentration ratios between the proteinogenic and meteorite-common amino acids. This yin-yang interplay can facilitate the persistence of confined nano-compartments within clay media.

This study shows that various confined nano-environments can occur simultaneously in the same crystalline clay domain. As the dimensions of confined interlayers and potential nanocavities differ, they possess different physicochemical properties and thus may facilitate different chemistry. The nano-compartments should retain amino acids with polymerisation potential in hydrated, confined environments whose conditions differ both from the bulk, external environment and from the interlayer spaces. Thermodynamic properties are known to be altered in similar aluminosilicate/silicate nanopores in ways that promote inorganic cation sorption and dimerisation under partial dehydration^21,22^. By analogy, we propose that the naturally-forming nanocavities in catalytic clay may promote organic polymerisation under partial dehydration alongside spatial confinement of the polymerisation products. Demonstration of the GABA-induced effects supports such confined geochemical nano-environments in the presence of strongly-interacting organics expected to polymerise, and opens up broad possibilities in origins-of-life chemistry. We hypothesise that rocky bodies, such as asteroids that contain hydrated phyllosilicate minerals and possess a carbonaceous composition (host proteinogenic and space-born amino acids)^66,67^, may also allow this type of organo-geochemical evolution.

Our results emphasise the importance of experimental studies that combine minerals and organic mixtures. We highlight the significance of the diverse origins and functionality of these organics, specifically focusing on abundant space-born organics. We demonstrate that GABA-induced partial exfoliation and nanocavity formation occurs in the presence of higher concentrations of strongly-interacting species than those expected under prebiotic conditions^68^. This further suggests that GABA-induced nanocavity formation is likely robust to the intercalation of other organic species with lower surface affinity at similar concentrations. These findings underline the importance of concentration ratios in the organic reservoir for shaping organo-geochemical evolution and ask us to explore the landscape of concentration ratio effects. Beyond prebiotic implications, harnessing this yin–yang interplay offers opportunities for material design and manipulation for multitude of potential applications, including drug delivery, development of fire retardants, energy storage, and plenty more.

Methods

The Ca-MMT clay “STx-1b” was obtained from the Clay Mineral Society, USA. The chemical formula of the clay is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(Si_{7.753}Al_{0.247})(Al_{3.281}Mg_{0.558}Fe_{0.136}Ti_{0.024}Mn_{0.002})(Ca_{0.341}Na_{0.039}K_{0.061})O_{20}(OH)_4$$\end{document} with a cation exchange capacity of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$66.1\pm 2.1$$\end{document} meq/100 gr^57^. Sedimentation of the clay in Milli-Q water (18.2 M \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Omega$$\end{document} *cm @25  \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} C, 1.5 ppb TOC) was performed by preparing a 20 wt% suspension in a clean tall beaker, stirring it for 2 hours at 21  \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} C at 600 rpm, after which the stirring was stopped and the clay grains in the suspension were allowed to sediment. The top layer ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim$$\end{document} cms) containing the fine clay fraction (particle diameter \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\le 2\,\mu$$\end{document} m) was collected in a clean beaker after a time ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim$$\end{document} hours) that corresponds to the settling velocity of these particles, as calculated according to Stokes law. For the calculation, a clay density of 2.3 g/cm \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^3$$\end{document} and water density and viscosity values at 21  \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} C were used. The fine fraction was allowed to sink to the bottom of the collection beaker for a few weeks, excess water was decanted, then the clay was dried in an oven for at least 10 days at 180  \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} C in air. The extracted clay fraction was then gently homogenised using a pestle and mortar for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1-2$$\end{document}  minutes prior to the experiments. L-lysine (crystallised, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge 98\%$$\end{document} ), L-arginine (free-base form, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge 98\%$$\end{document} ) and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\gamma$$\end{document} -aminobutyric acid ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge 99\%$$\end{document} ) were all purchased from Sigma-Aldrich and used without additional purification. Concentrated hydrochloric acid \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(37\%)$$\end{document} was also purchased from Sigma-Aldrich and used for pH adjustments of Lys and Arg solutions.

Experimental procedure

Stock solutions of Lys, Arg and GABA were prepared, these were used to prepare the amino acid solutions and mixtures examined in the experiments (‘exposure solutions’). The pH of all solutions was measured with a Mettler-Toledo probe (calibrated prior to measurement with standard solutions) at room temperature. The pH of the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2\,$$\end{document} M GABA stock solution was not adjusted resulting in all GABA stocks having pH values ranging between 7.59 and 7.85. The pH of the Lys and Arg stock solutions (naturally around pH 10–11) was adjusted using concentrated hydrochloric acid to reach a pH value of around 7.5. This was done by a gradual addition of 20–100  \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu$$\end{document} L of acid, stirring for 10–15 min to allow for equilibration after which a pH measurement was carried out. The final concentration of each of the Lys/Arg stock solutions was recalculated, accounting for the addition of acid volume. For Lys, two stocks were prepared, which had concentrations and pH values of 0.244 M and 7.46, 0.243 M and 7.52. For Arg, one stock solution was prepared, which had a concentration of 0.245 M and pH of 7.46. Using these stocks, the exposure solutions of single-organic 25 mM proteinogenic amino acids and their mixtures with 1 M GABA were prepared (one day prior to corresponding experiment); their final pH values are given in Table S5.

The experimental procedure is given in detail in a previous publication^16^, and here in short. Clay-amino acid exposure suspensions were prepared in 10 mL glass vials with a ratio of 100 mg clay per 3 mL amino acid solution (namely, 0.75 mmol/g Lys or Arg and in the mixture suspension also 30 mmol/g GABA). Control experiments with clay and 3 mL MilliQ water containing no organics were performed in triplicates. Blank experiments containing 3 mL organic exposure solution without clay were carried out once for each solution composition. The suspensions were held for 2 h at 80 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} C and stirred at a rate of 650 rpm. Then, the suspensions were allowed to cool and were centrifuged at 3000 rpm for 5 min followed by collection of the supernatant. The remaining solid clay was washed twice by adding about 5 mL MilliQ water, shaking at 1000 rpm for 15 min, centrifuging at 2500 rpm for 5 min and collecting the top liquid each time. This ensured that excess organics that were not interacting with the clay were removed to increase the signal related to organo-clay interactions and prevent polymerisation unrelated to the clay during further preparation and analyses. The washed clay samples were dehydrated in an oven under air flow at 120  \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} C for 163 h and 40 min, producing our ‘exposed samples’. Samples were dried to identify the exfoliated spectral signatures, which are obstructed in wet samples. All samples were treated in the same manner in preparation for analysis. All of the samples were gently, manually ground to a powder using a pestle and mortar for a similar period of time ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 15$$\end{document}  s) prior to ATR-FTIR and XRD measurements. ‘Rehydrated’ powder samples were produced by allowing some of the exposed sample from each experiment to sorb water vapour (obtained from the evaporation of MilliQ water) within a sealed box for 24 h, after which the weight gain due to sorbed water was calculated (Tables S6, S7). The rehydrated samples were gently and briefly ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim$$\end{document} 2–3 s), manually ground for homogenisation prior to ATR-FTIR, XRD and TEM measurements.

Attenuated total reflectance Fourier transform infrared spectroscopy

ATR-FTIR measurements were performed using a Perkin Elmer FT-IR spectrometer “Frontier” with a “GladiATR” mount (Pike technologies) which was connected to a nitrogen flow for at least one hour prior to measurements. Spectra were gathered in the range of 450–4000 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} at room temperature; solid samples acquired with a resolution of 2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} (data interval 0.5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} ) accumulated over 128 scans (new background before every sample), liquid samples with the same resolution and 8 scans (new background every 4–5 samples). Baseline subtraction was carried out on each spectrum by finding a linear fit that described the baseline in ranges in which no signals were observed, specifically 2400–2800 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} and 3800–4000 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} , and subtracting this from the entire spectrum. Full ATR-FTIR interfacial spectra of solid samples, both dehydrated and rehydrated, are given after baseline subtraction and normalisation (details provided below) in Figure S3. ATR-FTIR spectra of liquid samples after baseline subtraction and normalisation to the intensity of the OH stretching band of H \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}_2$$\end{document} O around 3300 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} ; original exposure solutions, blank experiments and water washes collected in one of the experimental repetitions, are given in Figure S4. Determination of peak positions was performed using derivative analysis for which smoothing of the data and first derivative is required^63^. Each interfacial spectrum was smoothed using a 21-point window Savitsky-Golay filter^69^, followed by derivation, then the first derivative was smoothed in the same manner and derived to produce the absorbance second derivative. Minima points in the spectrum of each solid clay sample were identified using a Matlab script in the ranges of 450–660 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} for normalisation and 885–1285 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} for analysis of the Si–O stretching band envelope. Normalisation is required in order to compare intensities between the different spectra. First, for normalisation to the Si–O–Al bending peak (around 518 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} for dehydrated and around 515 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} for rehydrated samples), a Gaussian fitting procedure was applied to five peaks identified in the range of 450–660 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} . These peak positions around 463, 518, 559, 600, 630 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} were used as starting points and fitted with a range of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm 6$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} . The widths of the peaks were also constrained to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm 2$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} around an optimal width recognised for each of the individual peaks (consistent throughout all of the samples). Each spectrum was then normalised to the fitted intensity of the Si–O–Al bending mode, as this mode was expected to remain most unchanged by interactions of the siloxane surfaces with respect to other vibrational modes. Spectra of three experimental repetitions were averaged and the standard deviation at each frequency was calculated. For the analysis of the Si–O stretching region, every spectrum had undergone the same procedure for peak finding as described above (with 21-point window smoothing) this time in the range of 885–1285 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} . The absorbance second derivative was averaged for the three experimental repetitions and the standard deviation was calculated. For the analysis of the second-derivative data in this region, we make use of the mathematical relation dictating that the intensity of the second derivative is inversely proportional to the square of the FWHM of the signal^62^. The same procedure was applied for interfacial spectra of single-organic 30 mmol/g GABA exposure from previous work^16^. Second-derivative analysis of the organic fingerprint region between 1250–1750 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,cm^{-1}}$$\end{document} , was carried out for the interfacial spectra by subtracting the normalised averaged spectra of the control sample from the normalised averaged interfacial spectra of each organic exposure treatment. Then, determination of the peak positions was carried out with the method described above, in this case with a 41-point smoothing window of the data sets and their first derivatives. The difference spectra of the initial amino acid solutions were created by subtracting the normalised spectrum of the second water wash of the control (no organics) from the respective single-component Lys and Arg 25 mM initial aqueous solutions normalised spectrum. This was followed by a peak position determination for each difference spectrum using a 61-point smoothing window of the data set and its first derivative.

Powder X-ray diffraction

pXRD measurements were conducted using a Rigaku MiniFlex II diffractometer with a Ni-filtered Cu K \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha$$\end{document} source ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda$$\end{document} =1.540562 Å) operated at voltage of 30 kV and current of 15 mA. A NaI(Tl) scintillation detector with a Be window of diameter \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\phi =23$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,mm}$$\end{document} and length of 80 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,mm}$$\end{document} was incorporated for detection of the X-rays. The powder samples were gently ground as described above and pressed on a low background sample holder made of single-crystal silicon with an 8 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,mm}$$\end{document} diameter, 0.2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {\,mm}$$\end{document} deep hole. Measurements were carried out in the range of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$3^\circ - 40^\circ \,(2\theta )$$\end{document} ; with resolution and scanning rate of 0.05 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} and 2.5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} /min in the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^\circ - 34^\circ \,(2\theta )$$\end{document} range, and 0.01 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} and 0.5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} /min in the ranges of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$3^\circ - 10^\circ \,(2\theta )$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$34^\circ - 40^\circ \,(2\theta )$$\end{document} . A background measurement with an empty sample holder was performed every day at the beginning of the work day, smoothed by a 21-point moving average and subtracted from each diffractogram. To increase the signal to noise ratio, diffraction data gathered at 0.01 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} resolution was binned to 0.05 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} . Area-based normalisation of the diffractograms was carried out using the MMT(105) peak in the range of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$34^\circ - 39.5^\circ \,(2\theta )$$\end{document} of one of the control samples for dehydrated and rehydrated samples. This particular peak retained its original shape and position throughout the differently-treated MMT samples and was therefore chosen for this normalisation method. Full XRD diffractograms, for both dehydrated and rehydrated samples, are given in Figure S5. Each one of the experimental repetitions’ MMT d(001) diffraction peak was fit to an asymmetric Lorentzian function of the form \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L(x) = \frac{A}{1 + \left( \frac{x - x_0}{\Gamma (x)} \right) ^2}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Gamma (x)$$\end{document} was defined as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Gamma (x) = \Gamma _L$$\end{document} for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$x < x_0$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Gamma (x) = \Gamma _R$$\end{document} for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$x > x_0$$\end{document} , in the range of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$4.5^\circ - 8.5^\circ \,(2\theta )$$\end{document} . Peak position was determined using fit parameter \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$x_0^\circ \,(2\theta )$$\end{document} and the corresponding d-spacing (Å) was calculated according to Bragg’s law^70^, explicitly: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d=2\pi /2k_0sin(^\circ 2\theta /2)$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$k_0=1/\lambda _0$$\end{document} ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda _0$$\end{document} is the X-ray source wavelength). For d-spacing error estimation we first calculated the total error of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$x_0^\circ \,(2\theta )$$\end{document} , then applied Bragg’s law. We did this by calculating the instrumental error according to uniform distribution with data resolution 0.05 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sigma _{\text {inst}}$$\end{document} ), the standard deviation of the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$x_0^\circ \,(2\theta )$$\end{document} position ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sigma _{\text {data}}$$\end{document} ) and then the total error according to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sigma _{\text {total}} = \sqrt{\sigma _{\text {inst}}^2 + \sigma _{\text {data}}^2}$$\end{document} . The total FWHM for each diffractogram was calculated using \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text {FWHM}_{\text {fit}} = \Gamma _L^{\text {fit}} + \Gamma _R^{\text {fit}}$$\end{document} , then the average FWHM for the experimental repetitions was calculated alongside its corresponding standard deviation. For the plots, diffractograms representing the average of the experimental repetitions were calculated with the associated standard deviation of each data point.

Thermogravimetric analysis

Thermogravimetric analysis coupled with differential scanning calorimetry (TGA/DSC) measurements were conducted using a NETZSCH STA 449 F3 Jupiter instrument. Each sample ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim 4.7-8.2$$\end{document}  mg) was placed in an alumina crucible and placed in the instrument alongside an empty reference crucible. One dehydrated sample of each treatment was measured, namely, MMT exposed to water (control) and MMT exposed to Lys, Lys-GABA, Arg or Arg-GABA solutions. Measurements were carried out in the temperature range of 26–325  \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} C at a heating rate of 2  \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^\circ$$\end{document} K/min under argon flow of 20 mL/min. Each TGA and DSC curve was smoothed using a 31 and 21-point window Savitsky-Golay filter^69^, respectively. The TGA derivative was smoothed by the same manner as the TGA curve and multiplied by -1 to produce the -dM/dT vs. T curve, for which the maximal value was manually identified.

Transmission electron microscopy

A Thermo Fischer TFS Talos F200X S(TEM) instrument operated at an accelerating voltage of 200kV was used for TEM imaging. For TEM analysis, rehydrated clay samples were prepared using the same procedure as described above the day before the measurements were conducted (Table S7). The powder clay samples were then loaded onto a 3 mm carbon film-supported copper 400 mesh TEM grid on the same day of the measurements. Imaging clay in the TEM is particularly challenging due to its high beam sensitivity. Therefore, fast identification of areas with an orientation in which the layers were visible was obtained by first randomly selecting a grain at the lowest magnifications possible. By following the grain to its edge we could identify locations that included several layers, of which a higher magnification image was obtained. Often several images of the same location could be obtained, where no change in the image was observed. After a certain time, complete amorphisation of the clay was observed under the beam. The time required and progression of amorphisation had no correlation with exposure of the clay to amino acids or not and with the presence of nanocavities in the imaged area. Had there been a localised effect of the beam on the cavities only, we would have expected it to be clearly visible. Such a direct effect of the beam on the cavities was not observed during the imaging process of Ca-MMT. Analysis of TEM images of the clay was conducted using the ImageJ program. Individual instances in which layered structures of at least 5–6 stacked layers were observed (not counting multiple images of the same region): 14 for control Ca-MMT clay sample, 16 for clay treated with Lys 0.75 mmol/g, 20 for clay treated with Lys 0.75 mmol/g + GABA 30 mmol/g, 17 for clay treated with Arg 0.75 mmol/g and 28 for clay treated with Arg 0.75 mmol/g + GABA 30 mmol/g. We classified whether each instance of the observed layered structures is mostly ordered or distorted by visual examination (this was independently repeated on several occasions to ensure a correct and consistent classification), as summarised in Table S4. To estimate the statistical significance of distorted layering percentages for each of the clay-exposure treatments relative to the Ca-MMT control, we tested a null hypothesis of distorted layering occurrence as observed for the control, assuming a binomial distribution. We then calculated the binomial p-value for each of the exposure treatments relative to the control, and of each mixture exposure relative to the corresponding proteinogenic-only exposure. The results of this statistical analysis are given in Figure S2. In the case of distorted layering, we also determined the occurrence of cavities within the clay layers, by implementing a conservative approach. We counted each instance of a cavity only if the suspected cavity was delineated by visible, well-defined lattice fringes that can be followed across its entire length on both sides (see Figure S6 for examples). In instances where more than one cavity was visible, this instance was counted only once in the statistical analysis (i.e. each instance is only classified with or without cavities). The size of the each confirmed cavity was measured in the d[001] direction (perpendicular to the layers) ten times using the ImageJ program (here all individual confirmed cavities were included). The average size of the confirmed cavities was calculated alongside its standard deviation. The challenging nature of the TEM measurements, especially the effects of the electron beam on the sample, may introduce an additional unknown uncertainty to the derived cavity sizes. However, as we did not observe any opening of the cavities during imaging nor were any effects found in the control samples, the associated error is likely small.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Information.