The origin of enzymatic activity from random sequences is a fundamental question in biology. Modern enzymes are highly evolved, large proteins with precise structures and functions. However, the transition from random polypeptides to functional enzymes during the early stages of life remains poorly understood. This research addresses this gap by investigating whether functional catalysts can be selected from a library of de novo designed proteins. The study employs a well-defined 4-helix bundle protein scaffold (S-824) as a starting point, introducing diversity through randomization of specific regions. Ultrahigh-throughput screening in microfluidic droplets significantly increases the chances of identifying rare active sequences. The use of a mixture of fluorogenic phosphodiester and phosphotriester substrates, along with common divalent metal cofactors, broadens the scope of potential catalysts that can be discovered, aiming to uncover metalloenzymes. The hypothesis is that biologically relevant catalysts can be isolated from collections of unevolved de novo designed sequences, even if such sequences are expected to be rare within the vastness of sequence space.

Previous research has demonstrated the challenges of obtaining functional proteins from random sequences. Studies have shown that random sequences rarely fold into stable structures or bind small molecules effectively. Keefe and Szostak showed the low frequency of ATP binding in a library of 80-residue polypeptides. The creation of catalysts is even more challenging. The concept of catalytic promiscuity has been suggested as a mechanism for the emergence of new functions after gene duplication. Metagenomic libraries have yielded catalysts for uncommon reactions, but eliciting function from entirely non-catalytic sequences requires extensive screening of large libraries. Computational protein design has shown promise, but routinely creating highly efficient catalysts rivalling evolved enzymes remains difficult. Dayhoff's hypothesis proposed that early functional proteins originated from short peptides, with their combination via duplication, diversification, and gene fusion events driving subsequent evolution. This research aims to test a broadened version of this hypothesis by identifying catalysts within a de novo protein library.

The study used a 102-residue de novo designed 4-helix bundle protein, S-824, as a scaffold. The apical loops and helix termini of S-824 were randomized, generating a library of approximately 1.7 million variants. Ultrahigh-throughput screening was performed using microfluidic droplets. Individual *E. coli* cells expressing a single library sequence were co-compartmentalized with fluorogenic phospho-di- and triesters, a mixture of MnCl2, ZnCl2, and CaCl2, and a lysis agent. Droplets were sorted based on fluorescence intensity using fluorescence-activated droplet sorting (FADS). Next-generation sequencing (NGS) was used to analyze the library before and after screening, identifying enriched sequences. Selected clones were further characterized. One highly active clone, mini-cAMPase, was purified and extensively characterized with respect to metal dependency, substrate specificity, kinetic parameters, structure (SEC, NMR, CD), and dynamic behavior (MD simulations). Control experiments were performed to rule out contamination by endogenous *E. coli* enzymes. The NGS data were used to understand the enrichment of truncated sequences and the positional distribution of truncations.

The screening yielded a truncated 59-residue enzyme, mini-cAMPase, that exhibited high catalytic activity. NGS analysis revealed a significant enrichment of truncated sequences after screening, suggesting that truncation contributes to the gain of phosphoesterase function. Mini-cAMPase was found to be manganese-dependent, showing activity towards a range of phosphodiesters and phosphonates, including cAMP and cGMP. Kinetic characterization revealed high catalytic proficiency (kcat/KM > 10¹⁴ M⁻¹s⁻¹ for bis(p-nitrophenyl) phosphate), comparable to much larger evolved enzymes. Structural analysis demonstrated that mini-cAMPase forms a dimeric structure, which is dynamic and less ordered than its parental S-824 sequence. MD simulations, NMR, and CD spectroscopy indicated increased structural dynamics in mini-cAMPase compared to its ancestor. Site-directed mutagenesis experiments showed that specific amino acid residues contribute to the catalytic activity. Control experiments confirmed that the observed activity is due to mini-cAMPase and not to contaminating *E. coli* enzymes. The enrichment of truncations in the library after screening suggests that shortening the protein structure is conducive to function.

The findings challenge the conventional view that functional enzymes emerge from incremental improvements of already active precursors. The study shows that a significant jump in sequence space, involving substantial truncation, can lead to the rapid acquisition of high catalytic activity. The high catalytic proficiency of mini-cAMPase, despite its small size and dynamic structure, supports the concept of catalytic promiscuity and the potential role of dynamic structures in the early stages of enzyme evolution. The results lend credence to Dayhoff's hypothesis concerning the emergence of function from short peptides or their assemblies. The truncation event in mini-cAMPase's evolution can be seen as analogous to the insertion and deletion events (InDels) that have been implicated as drivers of evolutionary innovation. The increased flexibility and conformational sampling observed in mini-cAMPase might facilitate functional diversification and evolvability, potentially providing advantages in the early stages of life when enzyme resources were likely limited. The study highlights the power of ultrahigh-throughput screening in uncovering unexpected solutions for functional acquisition in de novo protein design.

This research demonstrates the successful selection of a highly proficient phosphodiesterase (mini-cAMPase) from a library of de novo designed proteins. The acquisition of function involved a large jump in sequence space, characterized by significant truncations that led to a smaller, more dynamic, and dimeric enzyme. The high catalytic proficiency of mini-cAMPase, comparable to that of much larger evolved enzymes, supports the hypothesis that early enzymes may have been smaller and more promiscuous than their modern counterparts. The success of this screening strategy validates the concept of exploring novel pathways to enzyme evolution by combining de novo protein design with high-throughput screening. Future research could explore the evolutionary trajectory of mini-cAMPase, investigating how further mutations might enhance its catalytic activity and specificity.

The study screened a relatively small fraction of the vast sequence space. The specific mechanism by which truncation enhances catalytic activity requires further investigation. The MD simulations might not capture all relevant dynamic effects due to limitations in simulation timescales. Further studies are needed to completely understand the role of protein dynamics in the catalytic mechanism.