Shoulders, M. D. | Search bioRxiv

bioRxiv2026

Machine learning-assisted directed evolution of plant Rubisco

Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) is foundational to life on Earth, catalyzing carbon dioxide (CO2) fixation to generate biomass. However, Rubisco is a slow and inefficient enzyme that has proven challenging to engineer. We applied the structure-informed machine learning (ML) model ESM-IF1 to identify plausible amino acid sites in the large subunit of Nicotiana tabacum Rubisco to target for directed evolution. ML-assisted library design followed by selection in Rubisco-dependent Escherichia coli identified multiple enriched variants displaying improved catalytic efficiency. Several improved variants carried amino acid changes not found in the evolutionary lineage of plants, despite being assembly competent in plant chloroplasts, demonstrating that ML-assisted protein design can explore functional sequence space beyond what is observed from natural sequence diversity. Most prominently, the T391I substitution improved carboxylation rate by 29% and aerobic carboxylation efficiency by 43%. Our findings demonstrate the utility of ML-assisted evolution for engineering Rubisco with improved carboxylation efficiency and potential for enhancing crop productivity.

plant biology↗

bioRxiv2026

Synthesis of Ultra-Large Fibrous Proteins from Bacteria via a Looped-Translation System

High molecular weight fibrous proteins such as silk, elastin, and collagens, are fundamental for providing shape to macroscopic biological structures, yet their recombinant production remains challenging because of their extreme size and sequence repetitiveness. Here, we report a circular RNA-based ribosome translation platform that enables iterative ribosome synthesis of fibrous proteins through continuously "looped" translation. To promote efficient circularization of repetitive fibrous protein transcripts, we combined a synonymous codon locker sequence strategy with RNA circularization chaperones. Guided by a ribosome traffic model, we further optimized the translation bottlenecks within the circular RNA, substantially improving translation yields. The established looped translation platform is applicable to at least six classes of fibrous proteins and generated products with molecular weight exceeding titin at 3.8 MDa. The synthesized polypeptides were characterized through electron microscopy, bulk material fabrication, and mechanical analysis, demonstrating properties associated with ultra-high molecular weight polypeptides. Finally, we coupled looped translation to secretion through a programmed ribosomal frameshift, enabling export of fibrous protein across cellular membranes in both Escherichia coli and Bacillus subtilis. We envision that the genetic tools presented here could find a range of applications in bioplastics and engineered living materials.

synthetic biology↗

bioRxiv2025

Anti-CRISPR-mediated continuous directed evolution of CRISPR-Cas9 in human cells

Engineering CRISPR-Cas systems for improved or altered function is central to both research and therapeutic applications. Unfortunately most optimization, especially directed evolution in bacterial hosts, fails to capture the functional requirements of the complex mammalian cellular milieu, where activity is usually required. Robust strategies to enable continuous directed evolution of genome-targeting agents directly in human cells remain lacking. Here, we introduce CRISPR-MACE (Mammalian cell-enabled Adenovirus-assisted Continuous Evolution) as a foundational technology to address this need. CRISPR-MACE integrates virus-based continuous evolution with anti-CRISPR-based tunable selection to generate novel Streptococcus pyogenes Cas9 variants with both increased and decreased DNA binding capacity and nearly 1000-fold-enhanced resistance to AcrIIA4, the strongest known inhibitor of SpCas9. Notably, across independent evolution campaigns the same Cas9 gatekeeper mutation reproducibly emerged first, enabling subsequent adaptive steps along two interdependent axes of Cas9 function. In addition to advancing CRISPR technologies, this work establishes key principles and synthetic circuits for continuously evolving CRISPR-Cas systems directly in human cells. SIGNIFICANCE STATEMENTCRISPR technologies are typically engineered in bacteria, even though they must function in the far more complex environment of human cells. This gap has limited the discovery of variants with improved DNA recognition or with resistance to inhibitors that operate differently in mammalian systems. Here we establish CRISPR-MACE, a continuous evolution platform that leverages pressure from anti-CRISPR proteins to select Cas9 variants directly in human cells that have novel functions. Evolved variants show improvements in DNA binding strength and residence time, as well as striking escape from the potent Cas9 inhibitor AcrIIA4. Many anti-CRISPR proteins use distinct mechanisms, so our strategy can drive future continuous evolution campaigns in mammalian cells that expand the functional properties of genome-targeting agents.

synthetic biology↗

bioRxiv2025

In vivo directed evolution of an ultra-fast RuBisCO from a semi-anaerobic environment imparts oxygen resistance

Carbon dioxide (CO2) assimilation by the enzyme Ribulose-1,5-bisphosphate Carboxylase/Oxygenase (Rubisco) underpins biomass accumulation in photosynthetic bacteria and eukaryotes. Despite its pivotal role, Rubisco has a slow carboxylation rate [Formula] and is competitively inhibited by oxygen (O2). These traits impose limitations on photosynthetic efficiency, making Rubisco a compelling target for improvement. Interest in Form II Rubisco from Gallionellaceae bacteria, which comprise a dimer or hexamer of large subunits, arises from their nearly 5-fold higher [Formula] than the average Rubisco enzyme. As well as having a fast [Formula] (25.8 s-1 at 25 {degrees}C), we show that Gallionellaceae Rubisco (GWS1B) is extremely sensitive to O2 inhibition, consistent with its evolution under semi-anaerobic environments. We therefore used a novel in vivo mutagenesis-mediated screening pipeline to evolve GWS1B over six rounds under oxygenic selection, identifying three catalytic point mutants with improved ambient carboxylation efficiency; Thr-29-Ala (T29A), Glu-40-Lys (E40K) and Arg-337-Cys (R337C). Full kinetic characterization showed that each substitution enhanced the CO2 affinity of GWS1B under oxygenic conditions by subduing oxygen affinity, leading to 25% (E40K), 11% (T29A) and 8% (R337C) enhancements in carboxylation efficiency under ambient O2 at 25 {degrees}C. By contrast, under the near anaerobic natural environment of Gallionellaceae, the carboxylation efficiency of each mutant was impaired [~]16%. These findings demonstrate the efficacy of artificial directed evolution to access novel regions of catalytic space in Rubisco. SignificanceGiven Rubiscos crucial role in carbon dioxide assimilation, addressing its slow carboxylation rate and oxygen inhibition is a significant challenge. Utilizing one of the fastest known, yet also highly oxygen-sensitive, Rubisco - from the bacteria Gallionellaceae - we applied a novel in vivo directed evolution pipeline in Escherichia coli to discover mutations that specifically enhance carboxylation efficiency under ambient oxygen, a condition distinct from Gallionellaceaes natural semi-anaerobic environment. Our findings underscore the potential of directed evolution to unlock new catalytic capabilities for Rubisco, with implications for both fundamental research and practical agricultural applications.

biochemistry↗

bioRxiv2024

Human cartilage model of the precocious osteoarthritis-inducing COL2A1 p.Arg719Cys reveals pathology-driving matrix defects and a failure of the ER proteostasis network to recognize the defective procollagen-II

Mutations in the COL2A1 gene, encoding procollagen-II, cause various chondrodysplasias, including precocious osteoarthritis with mild spondyloepiphyseal dysplasia engendered by the p.Arg719Cys substitution. The molecular mechanisms underlying these disorders remain incompletely understood, largely owing to the absence of models faithfully recapitulating the human disease. Here, we developed an in vitro human cartilage model using isogenic induced pluripotent stem cell (iPSC) lines carrying either wild-type or Arg719Cys COL2A1. Directed differentiation into chondrocytes yielded cartilage tissues that were analyzed by immunohistochemistry, electron microscopy, SDS-PAGE, and RNA-sequencing. Tissues derived from Arg719Cys heterozygotes displayed a deficient matrix, closely reflecting the human disease phenotype. Arg719Cys procollagen-II was excessively post-translationally modified and partially retained within the endoplasmic reticulum (ER), leading to ER distention. Notably, despite introduction of an aberrant cysteine residue--expected to engage redox-sensitive folding and quality control pathways--Arg719Cys procollagen-II was not detectably recognized by the ER proteostasis network. The resulting inability to mount a quality control response, including activation of the unfolded protein response, indicates a failure in cellular surveillance. As a result, malformed procollagen-II both accumulates intracellularly and is secreted, contributing to the deposition of a structurally compromised extracellular matrix that drives disease pathology. The iPSC-derived cartilage model presented here provides a genetically defined and expandable, human-based system for dissecting the mechanisms of failed proteostasis in collagenopathies. These findings shed light on the types of substitutions in procollagen that cells can or cannot recognize, and underscore the therapeutic potential of targeting cellular surveillance and collagen quality control pathways in COL2A1-related disorders and beyond. SIGNIFICANCEThe p.Arg719Cys substitution in COL2A1 is known to cause dominantly inherited precocious osteoarthritis with mild spondyloepiphyseal dysplasia, but its molecular pathogenesis remains poorly understood. Prior work suggests disrupted fibrillogenesis as the molecular etiology, yet phenotypically relevant human models are lacking. Here, we use human iPSC-derived cartilage to uncover how the Arg719Cys substitution alters procollagen-II folding, modification, trafficking, and matrix deposition. We observe hallmark disease phenotypes, including a sparse collagen-II matrix and distended endoplasmic reticulum (ER), consistent with intracellular accumulation of procollagen-II. Surprisingly, the mutant procollagen-II does not appear to be differentially engaged by the ER proteostasis network, despite the incorporation of an aberrant and apparently non-disulfide bond-forming Cys residue, suggesting a fundamental failure of cellular surveillance. This unrecognized misfolding leads to secretion of defective collagen culminating in matrix dysfunction. These findings establish a mechanistic basis for disease etiology and highlight collagen quality control as a potentially tractable therapeutic target. The iPSC-based model provides a scalable, human-relevant platform for dissecting disease mechanisms and developing therapies for COL2A1-related disorders and other collagenopathies.

cell biology↗

bioRxiv2024

Dominant-negative TP53 mutations potentiated by the HSF1-regulatedproteostasis network

Protein mutational landscapes are shaped by how amino acid substitutions affect stability and folding or aggregation kinetics. These properties are modulated by cellular proteostasis networks. Heat shock factor 1 (HSF1) is the master regulator of cytosolic and nuclear proteostasis. Chronic HSF1 ac-tivity upregulation is a hallmark of cancer cells, potentially because upregulated proteostasis factors fa-cilitate the acquisition and maintenance of oncogenic mutations. Here, we assess how HSF1 activation influences mutational trajectories by which p53 can escape cytotoxic pressure from nutlin-3, an inhibitor of the p53 regulator MDM2. HSF1 activation broadly increases the fitness of dominant-negative p53 substitutions, particularly non-conservative, biophysically unfavorable amino acid changes within buried regions of the p53 DNA-binding domain. These findings demonstrate that HSF1 activation reshapes the oncogenic mutational landscape by preferentially supporting the emergence and persistence of bio-physically disruptive, cancer-associated p53 substitutions, linking proteostasis network activity directly to oncogenic evolution.

molecular biology↗

bioRxiv2023

Chondrodysplasia-inducing COL2A1 p.Gly1170Ser causes an ER storage defect without associated unfolded protein response in a human cartilage model

Mutations in the COL2A1 gene, encoding procollagen-II, cause various chondrodysplasias, including precocious osteoarthritis with mild spondyloepiphyseal dysplasia engendered by the p.Arg719Cys substitution. The molecular mechanisms underlying these disorders remain incompletely understood, largely owing to the absence of models faithfully recapitulating the human disease. Here, we developed an in vitro human cartilage model using isogenic induced pluripotent stem cell (iPSC) lines carrying either wild-type or Arg719Cys COL2A1. Directed differentiation into chondrocytes yielded cartilage tissues that were analyzed by immunohistochemistry, electron microscopy, SDS-PAGE, and RNA-sequencing. Tissues derived from Arg719Cys heterozygotes displayed a deficient matrix, closely reflecting the human disease phenotype. Arg719Cys procollagen-II was excessively post-translationally modified and partially retained within the endoplasmic reticulum (ER), leading to ER distention. Notably, despite introduction of an aberrant cysteine residue--expected to engage redox-sensitive folding and quality control pathways--Arg719Cys procollagen-II was not detectably recognized by the ER proteostasis network. The resulting inability to mount a quality control response, including activation of the unfolded protein response, indicates a failure in cellular surveillance. As a result, malformed procollagen-II both accumulates intracellularly and is secreted, contributing to the deposition of a structurally compromised extracellular matrix that drives disease pathology. The iPSC-derived cartilage model presented here provides a genetically defined and expandable, human-based system for dissecting the mechanisms of failed proteostasis in collagenopathies. These findings shed light on the types of substitutions in procollagen that cells can or cannot recognize, and underscore the therapeutic potential of targeting cellular surveillance and collagen quality control pathways in COL2A1-related disorders and beyond. SIGNIFICANCEThe p.Arg719Cys substitution in COL2A1 is known to cause dominantly inherited precocious osteoarthritis with mild spondyloepiphyseal dysplasia, but its molecular pathogenesis remains poorly understood. Prior work suggests disrupted fibrillogenesis as the molecular etiology, yet phenotypically relevant human models are lacking. Here, we use human iPSC-derived cartilage to uncover how the Arg719Cys substitution alters procollagen-II folding, modification, trafficking, and matrix deposition. We observe hallmark disease phenotypes, including a sparse collagen-II matrix and distended endoplasmic reticulum (ER), consistent with intracellular accumulation of procollagen-II. Surprisingly, the mutant procollagen-II does not appear to be differentially engaged by the ER proteostasis network, despite the incorporation of an aberrant and apparently non-disulfide bond-forming Cys residue, suggesting a fundamental failure of cellular surveillance. This unrecognized misfolding leads to secretion of defective collagen culminating in matrix dysfunction. These findings establish a mechanistic basis for disease etiology and highlight collagen quality control as a potentially tractable therapeutic target. The iPSC-based model provides a scalable, human-relevant platform for dissecting disease mechanisms and developing therapies for COL2A1-related disorders and other collagenopathies.

genetics↗

bioRxiv2023

The Immune-Evasive Proline 283 Substitution in Influenza Nucleoprotein Increases Aggregation Propensity Without Altering the Native Structure

Nucleoprotein (NP) is a key structural protein of influenza ribonucleoprotein complexes and is central to viral RNA packing and trafficking. In human cells, the interferon induced Myxovirus resistance protein 1 (MxA) binds to NP and restricts influenza replication. This selection pressure has caused NP to evolve a few critical MxA-resistant mutations, particularly the highly conserved Pro283 substitution. Previous work showed that this essential Pro283 substitution impairs influenza growth, and the fitness defect becomes particularly prominent at febrile temperature (39 {degrees}C) when host chaperones are depleted. Here, we biophysically characterize Pro283 NP and Ser283 NP to test if the fitness defect is owing to Pro283 substitution introducing folding defects. We show that the Pro283 substitution changes the folding pathway of NP without altering the native structure, making NP more aggregation prone during folding. These findings suggest that influenza has evolved to hijack host chaperones to promote the folding of otherwise biophysically incompetent viral proteins that enable innate immune system escape. TeaserPro283 substitution in flu nucleoprotein introduces folding defects, and makes influenza uniquely dependent on host chaperones.

biophysics↗

bioRxiv2023

Epistasis meets pleiotropy in shaping biophysical protein subspaces associated with antimicrobial resistance

Protein space is a rich analogy for genotype-phenotype maps, where amino acid sequence is organized into a high-dimensional space that highlights the connectivity between protein variants. It is a useful abstraction for understanding the process of evolution, and for efforts to engineer proteins towards desirable phenotypes. Few framings of protein space consider how higher-level protein phenotypes can be described in terms of their biophysical dimensions, nor do they rigorously interrogate how forces like epistasis--describing the nonlinear interaction between mutations and their phenotypic consequences--manifest across these dimensions. In this study, we deconstruct a low-dimensional protein space of a bacterial enzyme (dihydrofolate reductase; DHFR) into "subspaces" corresponding to a set of kinetic and thermodynamic traits [(kcat, KM, Ki, and Tm (melting temperature)]. We then examine how three mutations (eight alleles in total) display pleiotropy in their interactions across these subspaces. We extend this approach to examine protein spaces across three orthologous DHFR enzymes (Escherichia coli, Listeria grayi, and Chlamydia muridarum), adding a genotypic context dimension through which epistasis occurs across subspaces. In doing so, we reveal that protein space is a deceptively complex notion, and that the process of protein evolution and engineering should consider how interactions between amino acid substitutions manifest across different phenotypic subspaces.

systems biology↗

bioRxiv2023

Evolutionary druggability: leveraging low-dimensional fitness landscapes towards new metrics for antimicrobial applications

The term "druggability" describes the molecular properties of drugs or targets in pharmacological interventions and is commonly used in work involving drug development for clinical applications. There are no current analogues for this notion that quantify the drug-target interaction with respect to a given target variants sensitivity across a breadth of drugs in a panel, or a given drugs range of effectiveness across alleles of a target protein. Using data from low-dimensional empirical fitness landscapes composed of 16 {beta}-lactamase alleles and seven {beta}-lactam drugs, we introduce two metrics that capture (i) the average susceptibility of an allelic variant of a drug target to any available drug in a given panel ("variant vulnerability"), and (ii) the average applicability of a drug (or mixture) across allelic variants of a drug target ("drug applicability"). Finally, we (iii) disentangle the quality and magnitude of interactions between loci in the drug target and the seven drug environments in terms of their mutation by mutation by environment (G x G x E) interactions, offering mechanistic insight into the variant variability and drug applicability metrics. Summarizing, we propose that our framework can be applied to other datasets and pathogen-drug systems to understand which pathogen variants in a clinical setting are the most concerning (low variant vulnerability), and which drugs in a panel are most likely to be effective in an infection defined by standing genetic variation in the pathogen drug target (high drug applicability).

systems biology↗

bioRxiv2021

The Host Cell's Endoplasmic Reticulum Proteostasis Network Profoundly Shapes the Protein Sequence Space Accessible to HIV Envelope

The sequence space accessible to evolving proteins can be enhanced by cellular chaperones that assist biophysically defective clients in navigating complex folding landscapes. It is also possible, however, for proteostasis mechanisms that promote strict quality control to greatly constrain accessible protein sequence space. Unfortunately, most efforts to understand how proteostasis mechanisms influence evolution rely on artificial inhibition or genetic knockdown of specific chaperones. The few experiments that perturb quality control pathways also generally modulate the levels of only individual quality control factors. Here, we use chemical genetic strategies to tune proteostasis networks via natural stress response pathways that regulate levels of entire suites of chaperones and quality control mechanisms. Specifically, we upregulate the unfolded protein response (UPR) to test the hypothesis that the host endoplasmic reticulum (ER) proteostasis network shapes the sequence space accessible to human immunodeficiency virus-1 (HIV) envelope (Env) protein. Elucidating factors that enhance or constrain Env sequence space is critical because Env evolves extremely rapidly, yielding HIV strains with antibody and drug escape mutations. We find that UPR-mediated upregulation of ER proteostasis factors, particularly those controlled by the IRE1-XBP1s UPR arm, globally reduces Env mutational tolerance. Conserved, functionally important Env regions exhibit the largest decreases in mutational tolerance upon XBP1s activation. This phenomenon likely reflects strict quality control endowed by XBP1s-mediated remodeling of the ER proteostasis environment. Intriguingly and in contrast, specific regions of Env, including regions targeted by broadly neutralizing antibodies, display enhanced mutational tolerance when XBP1s is activated, hinting at a role for host proteostasis network hijacking in potentiating antibody escape. These observations reveal a key function for proteostasis networks in decreasing instead of expanding the sequence space accessible to client proteins, while also demonstrating that the host ER proteostasis network profoundly shapes the mutational tolerance of Env in ways that could have important consequences for HIV adaptation.

biochemistry↗

bioRxiv2021

XBP1s-Mediated Endoplasmic Reticulum Proteostasis Network Enhancement Can Selectively Improve Folding and Secretion of an Osteogenesis Imperfecta-Causing Collagen-I Variant

Osteogenesis imperfecta (OI) is typically caused by autosomal dominant mutations in genes encoding collagen type-I, most commonly resulting in Gly[->]Ser triple-helical domain substitutions that disrupt collagen folding and/or stability. Here, we test the hypothesis that upregulating the endoplasmic reticulum (ER) proteo-stasis network via the unfolded protein response (UPR) can improve the folding and secretion of the clinically severe, prototypical OI-causing COL1A1 p.G425S collagen-1(I) variant. We first show that small molecules that activate the entire UPR by causing global ER protein misfolding stress severely ablate collagen-I secretion from both G425S Col1(I)- and wild-type (WT) Col1(I)-expressing primary fibroblasts. In contrast, stress-independent, specific induction of just the UPRs XBP1s transcriptional response can enhance collagen-I secretion from G425S Col1(I) patient primary fibroblasts up to ~300% of basal levels. Notably, the effect is selective - collagen-I secretion from WT Col1(I)-expressing healthy donor primary fibroblasts is unaltered by XBP1s. XBP1s pathway activation appears to post-translationally enhance the folding/assembly and secretion of G425S Col1(I), as only modest impacts on collagen-I transcription or synthesis are observed. Consistent with this notion, we find that the stable, triple-helical collagen-I secreted by XBP1s-activated G425S 1(I) patient fibroblasts includes a higher proportion of the mutant 1(I) polypeptide than the collagen-I secreted under basal ER proteostasis conditions. We note that consistent reproducibility of these results is dependent on as yet unascertained experimental variables. Still, these promising observations suggest the potential for ER proteo-stasis network modulation to improve mutant collagen proteostasis in the collagenopathies, motivating further investigation of the effects generality, underlying mechanism, and potential therapeutic benefits.

cell biology↗