A robotic prebiotic chemist probes long-term reactions of complexifying mixtures

Understanding the transition from abiotic chemistry to living systems on early Earth, a process spanning approximately 100 million years, presents a significant challenge. Traditional laboratory experiments exploring prebiotic chemistry are typically limited to short durations (hours or days), hindering the investigation of complex, multicomponent systems requiring extended reaction times. The exploration of such systems is further hampered by the analytical complexity of resulting product mixtures, often containing numerous unknown compounds. Existing methodologies, rooted in synthetic organic chemistry, often intentionally constrain chemical space to facilitate analysis. However, to emulate the conditions of early Earth and adequately probe the origin of life, experiments need to move beyond single-flask approaches, encompassing dynamic environmental factors like mineral surfaces, variable temperature, pH, and redox conditions. While previous work has explored the cyclical reaction of simple mixtures, the vast number of potential reactions and the extended timeframes required are substantial hurdles. The current field lacks an experimental framework capable of testing competing hypotheses over extended timescales, given the enormous size of the relevant chemical search space. This paper addresses this gap by introducing a 'robotic prebiotic chemist,' an automated system designed to conduct unconstrained multicomponent chemistry experiments over extended periods with integrated analytical measurements and decision-making capabilities.

The authors review existing literature highlighting the limitations of short-term experiments in understanding prebiotic chemistry and the need for long-term, automated approaches. They cite studies on prebiotic network evolution, reaction parameters, and the challenges posed by intractable mixtures in origin-of-life research. They reference various theories on the emergence of life from non-living substrates, emphasizing the lack of testable hypotheses over the relevant geological timescales. The paper discusses current research focusing on prebiotic plausibility and its constraints imposed by our incomplete geochemical knowledge. It also points to previous work demonstrating the potential of cyclical reactions to diversify product space, but notes the limitations of existing methodologies in dealing with the vast chemical space relevant to the emergence of life. This sets the stage for the introduction of the robotic prebiotic chemist as a novel solution to these challenges.

The robotic prebiotic chemist platform is a closed-loop system designed to execute unconstrained multicomponent chemistry experiments on mineral surfaces in a cyclical manner. The system incorporates five pumps (four syringe pumps and one peristaltic pump), four valves for liquid handling, 18 reagents in aqueous solution, a reactor vessel under nitrogen with a reflux condenser, a magnetic stirrer hot plate, a sampling loop for HPLC-MS, and a sample wheel. The system is largely automated, with only manual vial changes in the sample wheel required every 3 days. The platform's automation extends to modifying experimental conditions and inputs based on data acquired during the experiment. Each experiment involves 60–150 cycles, each lasting 3–12 hours, beginning with a clean reactor charged with a mineral mixture (quartz, ulexite, pyrite). Three randomly selected input solutions are added at the start of each cycle. The system then stirs and heats the mixture at 70°C and 300 rpm. After each cycle, a sample is taken for analysis, and 70% of the mixture is removed, before replenishing with fresh input solutions for the next cycle. The platform utilizes a Python-based control system and data analysis pipeline. The decision-making algorithm uses a 'Mass Index' metric based on mass spectrometry data to assess the complexity of the product mixture. This metric calculates the difference between the highest and lowest peaks in the spectrum, divided by the total number of peaks above a threshold, providing a rapid assessment for the algorithm to adapt input compositions. If the slope of the Mass Index values decreases below a defined threshold, the algorithm randomly changes the input solution composition. To test reproducibility, experiments were repeated by manually inputting the previously recorded sequence of input solutions, eliminating the algorithm's decision-making aspect. The system is equipped with both an online benchtop ESI-MS and an offline high-resolution Orbitrap ESI-MS for data acquisition and analysis. The HPLC-DAD (Diode Array Detector) was also utilized.

The study successfully demonstrated the autonomous operation of the robotic platform for extended periods (over 30 days), executing numerous consecutive cycles with varying chemical compositions. The experiments revealed that the Mass Index, a heuristic for assessing mixture complexity, did not always show a progressive increase over cycles, suggesting a complex interplay of factors influencing the system's evolution. Some experiments exhibited periods of progressive Mass Index increases, indicating the formation of heavier product species or a reduction in overall species number. In contrast, other runs displayed static or fluctuating Mass Index values. Reproducibility tests using predetermined input trajectories revealed differences in product distribution between replicates, highlighting the inherent variability of the system, despite using identical input sequences. Analysis of the highest m/z (mass-to-charge ratio) values across different runs did not reveal a consistent increase with cycle number, further confirming the complexity of the system's behavior. However, many of the highest m/z species were heavier than the heaviest starting material. The highest observed Mass Index was found in run A, correlating with a large number of unique product species, suggesting that the random selection of input reagents resulted in significant complexification of the product mixture. The data highlights that the Mass Index can increase even without a dramatic increase in the mass of the heaviest product, indicating its utility in handling complex mixtures. The study demonstrated that the system could generate up to 5256 unique product species in a single cycle, showcasing its capacity to explore a vast chemical space.

The results demonstrate the feasibility of conducting long-term, autonomous prebiotic chemistry experiments to explore complex chemical spaces. The use of a robotic system addresses the limitations of traditional methods and allows for the investigation of dynamic, multicomponent reactions. The findings highlight the inherent complexity and variability of such systems, emphasizing the importance of systems-level approaches rather than focusing solely on individual components. The Mass Index, though a simplification, proved a valuable heuristic for algorithmic control of the experiment and enabling real-time adjustments based on data analysis. The observation of varying trends in the Mass Index across different experiments underscores the need for further refinement of algorithmic approaches to capture the full complexity of these dynamic systems. The reproducible, yet not identical results obtained from repeating the experiments with the same input sequence demonstrate the need to factor in experimental variations despite using the same mineral origin and washing procedures. The study establishes a new experimental paradigm for prebiotic chemistry research, enabling the exploration of complex mixtures and providing new insights into the emergence of life from simpler systems.

The development of the automated robotic prebiotic chemist platform represents a crucial step in enabling long-term, unconstrained experiments in prebiotic chemistry. The platform successfully generated highly complex mixtures, showcasing the potential for algorithmic control of such experiments. The use of a simple heuristic, the Mass Index, allowed for real-time adaptation of experimental parameters based on ongoing data analysis. Future research could focus on developing more sophisticated algorithms to capture the full complexity of these systems, exploring different metrics and analytical approaches. Additionally, investigating the role of mineral surfaces through detailed characterization techniques would enhance our understanding of the system's behavior. This platform opens new avenues for exploring long-standing hypotheses about the emergence of life and the transition from prebiotic chemistry to biological systems.

The study acknowledges that the Mass Index, while useful as a heuristic for algorithmic control, is a simplification that may not capture the full complexity of the chemical mixtures. The reproducibility tests, while demonstrating the capacity of the system for repeat experiments, also revealed variations between replicate runs, likely due to small variations in the mineral environment and instrument fluctuations. The selection of the 18 initial input reagents, while deliberately chosen for its diversity, may have introduced a bias in the exploration of chemical space. Further investigation is needed to explore a broader range of starting materials and conditions to more fully represent the chemical environment of early Earth.