Periodic Reporting for period 5 - ABIOS (ABIOtic Synthesis of RNA: an investigation on how life started before biology existed)
Reporting period: 2024-08-01 to 2024-12-31
The emergence of life is one of the most fascinating and yet largely unsolved questions in the natural sciences, and thus a significant challenge for scientists from many disciplines. There is growing evidence that ribonucleic acid (RNA) polymers, which are capable of genetic information storage and self-catalysis, were involved in the early forms of life. But despite recent progress, RNA synthesis without biological machineries is very challenging. The current project aims at understanding how to synthesize RNA in abiotic conditions. We will address problems associated with three critical aspects of RNA formation that I will rationalize at a molecular level: (i) accumulation of precursors, (ii) formation of a chemical bond between RNA monomers, and (iii) tolerance for alternative backbone sugars or linkages. These questions range from the formation of chemical bonds up to the stability of large biopolymers, and we thus need a multi-scale approach combining techniques that range from quantum calculations to large-scale all-atom simulations, employed together with efficient enhanced-sampling algorithms, forcefield improvement, cutting-edge analysis methods and model development.
Objectives
1 • To explain why the poorly-understood thermally-driven process of thermophoresis can contribute to the accumulation of dilute precursors.
2 • To understand why linking RNA monomers with phosphoester bonds is so difficult, to understand the molecular mechanism of possible catalysts and to suggest key improvements.
3 • To rationalize the molecular basis for RNA tolerance for alternative backbone sugars or linkages that have probably been incorporated in abiotic conditions.
Conclusions:
1. We established a clear protocol for studying thermophoresis using atomistic simulations in dilute aqueous solutions. However, the molecular mechanism underlying this phenomenon remains elusive. We successfully ruled out popular theories that fail to explain the accumulation of molecular solutes under thermal gradients.
2. By employing cutting-edge, AI-enhanced reactive molecular simulations, we investigated the uncatalyzed reaction and provided a rationale for the low reactivity of phosphate in phosphoester bond formation. We also developed a general approach to study chemical reactivity in the condensed phases using neural network potentials obtained through active learning. Ongoing studies within our group are exploring potential strategies to accelerate this reaction. Additionally, we examined the same reaction in the context of a plausible autocatalytic network involving small self-reproducing RNAs (ribozymes).
3. We offered a rationale for the temperature-induced separation of RNA duplexes, which are the end-products of template-based RNA replication. We also developed methods to test alternative sugar backbones, though our studies remain inconclusive regarding their effects on strand separation.
Objectives
1 • To explain why the poorly-understood thermally-driven process of thermophoresis can contribute to the accumulation of dilute precursors.
2 • To understand why linking RNA monomers with phosphoester bonds is so difficult, to understand the molecular mechanism of possible catalysts and to suggest key improvements.
3 • To rationalize the molecular basis for RNA tolerance for alternative backbone sugars or linkages that have probably been incorporated in abiotic conditions.
Conclusions:
1. We established a clear protocol for studying thermophoresis using atomistic simulations in dilute aqueous solutions. However, the molecular mechanism underlying this phenomenon remains elusive. We successfully ruled out popular theories that fail to explain the accumulation of molecular solutes under thermal gradients.
2. By employing cutting-edge, AI-enhanced reactive molecular simulations, we investigated the uncatalyzed reaction and provided a rationale for the low reactivity of phosphate in phosphoester bond formation. We also developed a general approach to study chemical reactivity in the condensed phases using neural network potentials obtained through active learning. Ongoing studies within our group are exploring potential strategies to accelerate this reaction. Additionally, we examined the same reaction in the context of a plausible autocatalytic network involving small self-reproducing RNAs (ribozymes).
3. We offered a rationale for the temperature-induced separation of RNA duplexes, which are the end-products of template-based RNA replication. We also developed methods to test alternative sugar backbones, though our studies remain inconclusive regarding their effects on strand separation.
Progress Along the Three Research Axes
1 • Thermophoresis
We established a robust strategy to reproduce thermophoretic settings using atomistic molecular dynamics simulations, which we now routinely apply to various small, neutral solutes in aqueous solutions. Our data collection covered many systems, allowing us to assess how the thermophoretic behavior of a molecule depends on factors such as size, water affinity, mass, and concentration. These observations suggested that existing models and theories may not accurately describe molecular solutes in water under thermal gradients. While we attempted to develop new models to predict molecular behavior in these conditions, this work remains ongoing and inconclusive thus far.
2 • Phosphoester Bond Formation
We addressed the complex issue of phosphoester bond reactivity—a reaction with unfavorable energetics and slow kinetics. Initially, we employed a semi-empirical strategy for atomic interaction modeling. After significant effort in designing reaction coordinates and sampling methods, we applied this setup to a simple model reaction involving a small alcohol and phosphate. We identified two distinct reaction pathways and mechanisms but also uncovered key limitations in our approach.
To overcome these limitations, we shifted to machine-learned interatomic potentials, which are increasingly popular in molecular simulations. However, challenges remained in selecting chemical structures for the training dataset and identifying relevant collective variables for sampling. We successfully addressed both issues and studied the mechanism of uncatalyzed phosphoester bond formation in unprecedented detail. We are currently investigating how pH and activating leaving groups affect the reaction.
Additionally, we developed ArcaNN, a freely available active learning protocol and software, which is now widely transposable to other chemical reactions. Lastly, we initiated a related research axis on phosphoester bond formation in autocatalytic RNA networks involving small ribozymes.
3 • RNA Duplex Stability
We focused on the chemical alterations of RNA sugar backbones, beginning with one of RNA’s most crucial properties under abiotic conditions—duplex stability during the final stages of RNA replication. Without enzymes, strand separation is challenging and typically requires high temperatures. To explore how chemical modifications could later influence strand separation, we first developed a robust strategy to assess RNA duplex separation as temperature increases.
Building on a method we had previously applied to proteins, we observed exciting results on small model strands of varying sequences. Our findings showed that the widely-accepted two-state, all-or-nothing model (in which strands are either fully associated or fully separated) needs refinement. Instead, the separation mechanism is more gradual, with significant base fraying at duplex ends well before full unfolding. We characterized the sequence- and size-dependence of this fraying effect.
Finally, we developed forcefield parameters to study chemical modifications of RNA duplex backbones. Although investigations into how these modifications impact separation mechanisms and melting temperatures are ongoing, early results are promising.
1 • Thermophoresis
We established a robust strategy to reproduce thermophoretic settings using atomistic molecular dynamics simulations, which we now routinely apply to various small, neutral solutes in aqueous solutions. Our data collection covered many systems, allowing us to assess how the thermophoretic behavior of a molecule depends on factors such as size, water affinity, mass, and concentration. These observations suggested that existing models and theories may not accurately describe molecular solutes in water under thermal gradients. While we attempted to develop new models to predict molecular behavior in these conditions, this work remains ongoing and inconclusive thus far.
2 • Phosphoester Bond Formation
We addressed the complex issue of phosphoester bond reactivity—a reaction with unfavorable energetics and slow kinetics. Initially, we employed a semi-empirical strategy for atomic interaction modeling. After significant effort in designing reaction coordinates and sampling methods, we applied this setup to a simple model reaction involving a small alcohol and phosphate. We identified two distinct reaction pathways and mechanisms but also uncovered key limitations in our approach.
To overcome these limitations, we shifted to machine-learned interatomic potentials, which are increasingly popular in molecular simulations. However, challenges remained in selecting chemical structures for the training dataset and identifying relevant collective variables for sampling. We successfully addressed both issues and studied the mechanism of uncatalyzed phosphoester bond formation in unprecedented detail. We are currently investigating how pH and activating leaving groups affect the reaction.
Additionally, we developed ArcaNN, a freely available active learning protocol and software, which is now widely transposable to other chemical reactions. Lastly, we initiated a related research axis on phosphoester bond formation in autocatalytic RNA networks involving small ribozymes.
3 • RNA Duplex Stability
We focused on the chemical alterations of RNA sugar backbones, beginning with one of RNA’s most crucial properties under abiotic conditions—duplex stability during the final stages of RNA replication. Without enzymes, strand separation is challenging and typically requires high temperatures. To explore how chemical modifications could later influence strand separation, we first developed a robust strategy to assess RNA duplex separation as temperature increases.
Building on a method we had previously applied to proteins, we observed exciting results on small model strands of varying sequences. Our findings showed that the widely-accepted two-state, all-or-nothing model (in which strands are either fully associated or fully separated) needs refinement. Instead, the separation mechanism is more gradual, with significant base fraying at duplex ends well before full unfolding. We characterized the sequence- and size-dependence of this fraying effect.
Finally, we developed forcefield parameters to study chemical modifications of RNA duplex backbones. Although investigations into how these modifications impact separation mechanisms and melting temperatures are ongoing, early results are promising.
1 • We have developed a unique strategy to address thermophoresis in all-atom molecular dynamics simulations, and could demonstrate that existing theoretical models failed to explain the thermophoretic properties of small molecular solutes.
2 • We have a thorough understanding of phosophoester bond formation for the model reaction between phosphate and a small alcohol molecule using state-of-the-art artificial intelligence-based approaches. We developed a robust protocol that could be of general use to study chemical reactivity in the condensed phases.
3 • We could provide a full molecular picture of the nucleic acid duplexes separation upon thermal denaturation, nuancing the commonly-accepted two-state picture.
2 • We have a thorough understanding of phosophoester bond formation for the model reaction between phosphate and a small alcohol molecule using state-of-the-art artificial intelligence-based approaches. We developed a robust protocol that could be of general use to study chemical reactivity in the condensed phases.
3 • We could provide a full molecular picture of the nucleic acid duplexes separation upon thermal denaturation, nuancing the commonly-accepted two-state picture.