Shuyi Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Shuyi Zhang has a total of 73 co-authors. Here are the top co-authors by number of shared publications:
- Donald A. Bryant
Department of Biochemistry and Molecular Biology, The Pennsylvania State University
6 shared publications Donald A. Bryant
Department of Biochemistry and Molecular Biology The Pennsylvania State University, University Park Pennsylvania
5 shared publications- Gerard Charles Dismukes
Waksman Institute of Microbiology, Rutgers University, Piscataway, New Jersey 08854, United States
3 shared publications Xiao Qian
Waksman Institute of Microbiology, Rutgers, The State University of New Jersey, Piscataway NJ, USA
3 shared publications- Christopher A. Voigt
Massachusetts Institute of Technology
2 shared publications G. Kenchappa Kumaraswamy
Waksman Institute Rutgers University New Brunswick New Jersey
2 shared publications
Latest Publications
- Open Access
- 4 March 2026
- Published by Springer Nature in Research Square
Accelerating protein directed evolution via reinforcement learning
With the advancement of artificial intelligence, the protein fitness landscape becomes predictable, providing reliable guidance in the selection of advantageous mutations for the directed evolution of proteins. In practice, however, simply combining a small number of advantageous single mutations is unlikely to produce variants with global superiority, while exhaustive exploration of the astronomical mutational combinations is highly challenging. In this study, we introduce a virtual directed evolution pipeline, RelaVDEP, to facilitate functional optimization of the target protein in silico. By developing a reward model to balance the efficiency and accuracy of protein functional prediction and by designing a model-based reinforcement learning framework to explore the vast combinatorial space of protein mutations, this pipeline is capable of automatically identifying diversified multiple mutational variants with notable improvement in desired functional properties for a broad spectrum of proteins (including highly engineered targets like eGFP and PETase), as evidenced by experimental validation
- 4 March 2026
- Published by Springer Nature in Research Square
Open AccessAccelerating protein directed evolution via reinforcement learning
With the advancement of artificial intelligence, the protein fitness landscape becomes predictable, providing reliable guidance in the selection of advantageous mutations for the directed evolution of proteins. In practice, however, simply combining a small number of advantageous single mutations is unlikely to produce variants with global superiority, while exhaustive exploration of the astronomical mutational combinations is highly challenging. In this study, we introduce a virtual directed evolution pipeline, RelaVDEP, to facilitate functional optimization of the target protein in silico. By developing a reward model to balance the efficiency and accuracy of protein functional prediction and by designing a model-based reinforcement learning framework to explore the vast combinatorial space of protein mutations, this pipeline is capable of automatically identifying diversified multiple mutational variants with notable improvement in desired functional properties for a broad spectrum of proteins (including highly engineered targets like eGFP and PETase), as evidenced by experimental validation
- 19 November 2025
- Published by Springer Nature in Nature Chemical Engineering
- Vol. 2 (11) , 685-698
An industrial automated laboratory for programmable protein evolution
- 19 November 2025
- Published by Springer Nature in Nature Chemical Engineering
- Vol. 2 (11) , 685-698
An industrial automated laboratory for programmable protein evolution
- Top Cited PapersOpen Access
- 21 February 2025
- Published by Springer Nature in Nature Communications
- Vol. 16 (1) , 1-14
Oral delivery of therapeutic proteins by engineered bacterial type zero secretion system
Genetically engineered commensal bacteria are promising living drugs, however, their therapeutic molecules are frequently confined to their colonization sites. Herein, we report an oral protein delivery technology utilizing an engineered bacterial type zero secretion system (T0SS) via outer membrane vesicles (OMVs). We find that OMVs produced in situ by Escherichia coli Nissle 1917 (EcN) can penetrate the intact gut epithelial barrier to enter the circulation and that epithelial transcytosis involves pinocytosis and dynamin-dependent pathways. EcN is engineered to endogenously load various enzymes into OMVs, and the secreted enzyme-loaded OMVs are able to stably catalyze diverse detoxification reactions against digestive fluid and even enter the circulation. Using hyperuricemic mice and uricase delivery as a demonstration, we demonstrate that the therapeutic efficacy of our engineered EcN with a modified T0SS outperforms that with a direct protein secretion apparatus. The enzyme-loaded OMVs also effectively detoxify human serum samples, highlighting the potential for the clinical treatment of metabolic disorders.
- 21 February 2025
- Published by Springer Nature in Nature Communications
- Vol. 16 (1) , 1-14
Top Cited PapersOpen AccessOral delivery of therapeutic proteins by engineered bacterial type zero secretion system
Genetically engineered commensal bacteria are promising living drugs, however, their therapeutic molecules are frequently confined to their colonization sites. Herein, we report an oral protein delivery technology utilizing an engineered bacterial type zero secretion system (T0SS) via outer membrane vesicles (OMVs). We find that OMVs produced in situ by Escherichia coli Nissle 1917 (EcN) can penetrate the intact gut epithelial barrier to enter the circulation and that epithelial transcytosis involves pinocytosis and dynamin-dependent pathways. EcN is engineered to endogenously load various enzymes into OMVs, and the secreted enzyme-loaded OMVs are able to stably catalyze diverse detoxification reactions against digestive fluid and even enter the circulation. Using hyperuricemic mice and uricase delivery as a demonstration, we demonstrate that the therapeutic efficacy of our engineered EcN with a modified T0SS outperforms that with a direct protein secretion apparatus. The enzyme-loaded OMVs also effectively detoxify human serum samples, highlighting the potential for the clinical treatment of metabolic disorders.
- Open Access
- 16 January 2025
- Published by Springer Nature in Nature Communications
- Vol. 16 (1) , 1-14
Predictive genetic circuit design for phenotype reprogramming in plants
Plants, with intricate molecular networks for environmental adaptation, offer groundbreaking potential for reprogramming with predictive genetic circuits. However, realizing this goal is challenging due to the long cultivation cycle of plants, as well as the lack of reproducible, quantitative methods and well-characterized genetic parts. Here, we establish a rapid (~10 days), quantitative, and predictive framework in plants. A group of orthogonal sensors, modular synthetic promoters, and NOT gates are constructed and quantitatively characterized. A predictive model is developed to predict the designed circuits’ behavior accurately. Our versatile and robust framework, validated by constructing 21 two-input circuits with high prediction accuracy (R2 = 0.81), enables multi-state phenotype control in both Arabidopsis thaliana and Nicotiana benthamiana in response to chemical inducers. Our study achieves predictable design and application of synthetic circuits in plants, offering valuable tools for the rapid engineering of plant traits in biotechnology and agriculture.
- 16 January 2025
- Published by Springer Nature in Nature Communications
- Vol. 16 (1) , 1-14
Open AccessPredictive genetic circuit design for phenotype reprogramming in plants
Plants, with intricate molecular networks for environmental adaptation, offer groundbreaking potential for reprogramming with predictive genetic circuits. However, realizing this goal is challenging due to the long cultivation cycle of plants, as well as the lack of reproducible, quantitative methods and well-characterized genetic parts. Here, we establish a rapid (~10 days), quantitative, and predictive framework in plants. A group of orthogonal sensors, modular synthetic promoters, and NOT gates are constructed and quantitatively characterized. A predictive model is developed to predict the designed circuits’ behavior accurately. Our versatile and robust framework, validated by constructing 21 two-input circuits with high prediction accuracy (R2 = 0.81), enables multi-state phenotype control in both Arabidopsis thaliana and Nicotiana benthamiana in response to chemical inducers. Our study achieves predictable design and application of synthetic circuits in plants, offering valuable tools for the rapid engineering of plant traits in biotechnology and agriculture.
- 11 November 2024
- Published by Springer Nature in Nature Methods
- Vol. 22 (1) , 102-112
EvoAI enables extreme compression and reconstruction of the protein sequence space
Designing proteins with improved functions requires a deep understanding of how sequence and function are related, a vast space that is hard to explore. The ability to efficiently compress this space by identifying functionally important features is extremely valuable. Here we establish a method called EvoScan to comprehensively segment and scan the high-fitness sequence space to obtain anchor points that capture its essential features, especially in high dimensions. Our approach is compatible with any biomolecular function that can be coupled to a transcriptional output. We then develop deep learning and large language models to accurately reconstruct the space from these anchors, allowing computational prediction of novel, highly fit sequences without prior homology-derived or structural information. We apply this hybrid experimental–computational method, which we call EvoAI, to a repressor protein and find that only 82 anchors are sufficient to compress the high-fitness sequence space with a compression ratio of 1048. The extreme compressibility of the space informs both applied biomolecular design and understanding of natural evolution.
- 11 November 2024
- Published by Springer Nature in Nature Methods
- Vol. 22 (1) , 102-112
EvoAI enables extreme compression and reconstruction of the protein sequence space
Designing proteins with improved functions requires a deep understanding of how sequence and function are related, a vast space that is hard to explore. The ability to efficiently compress this space by identifying functionally important features is extremely valuable. Here we establish a method called EvoScan to comprehensively segment and scan the high-fitness sequence space to obtain anchor points that capture its essential features, especially in high dimensions. Our approach is compatible with any biomolecular function that can be coupled to a transcriptional output. We then develop deep learning and large language models to accurately reconstruct the space from these anchors, allowing computational prediction of novel, highly fit sequences without prior homology-derived or structural information. We apply this hybrid experimental–computational method, which we call EvoAI, to a repressor protein and find that only 82 anchors are sufficient to compress the high-fitness sequence space with a compression ratio of 1048. The extreme compressibility of the space informs both applied biomolecular design and understanding of natural evolution.
- Open Access
- 1 August 2024
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 52 (16) , 10005-10016
CRISPR perfect adaptation for robust control of cellular immune and apoptotic responses
A central challenge in the quest for precise gene regulation within mammalian cells is the development of regulatory networks that can achieve perfect adaptation—where outputs consistently return to a set baseline post-stimulus. Here, we present such a system that leverages the CRISPR activation (CRISPRa) and anti-CRISPR proteins as two antithetic elements to establish perfect adaptation in mammalian cells and dynamically regulate gene expression. We demonstrate that this system can maintain stable expression levels of target genes in the face of external perturbations, thus providing a robust platform for biological applications. The versatility of our system is further showcased through its integration with endogenous regulatory mechanisms in T cells, such as the NF-κB-mediated immune response, and its ability to program apoptosis responses for precise spatial and temporal control of cellular growth and death. This study not only advances our understanding of gene regulation in mammalian cells but also opens new avenues for therapeutic intervention, particularly in diseases characterized by dysregulated gene expression.
- 1 August 2024
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 52 (16) , 10005-10016
Open AccessCRISPR perfect adaptation for robust control of cellular immune and apoptotic responses
A central challenge in the quest for precise gene regulation within mammalian cells is the development of regulatory networks that can achieve perfect adaptation—where outputs consistently return to a set baseline post-stimulus. Here, we present such a system that leverages the CRISPR activation (CRISPRa) and anti-CRISPR proteins as two antithetic elements to establish perfect adaptation in mammalian cells and dynamically regulate gene expression. We demonstrate that this system can maintain stable expression levels of target genes in the face of external perturbations, thus providing a robust platform for biological applications. The versatility of our system is further showcased through its integration with endogenous regulatory mechanisms in T cells, such as the NF-κB-mediated immune response, and its ability to program apoptosis responses for precise spatial and temporal control of cellular growth and death. This study not only advances our understanding of gene regulation in mammalian cells but also opens new avenues for therapeutic intervention, particularly in diseases characterized by dysregulated gene expression.
- Open Access
- 15 March 2022
- Published by Frontiers Media SA in Frontiers in Molecular Biosciences
- Vol. 9, 882743
Editorial: Synthetic Live Biotherapeutic Products for Diseases
Editorial: Synthetic Live Biotherapeutic Products for Diseases
- 15 March 2022
- Published by Frontiers Media SA in Frontiers in Molecular Biosciences
- Vol. 9, 882743
Open AccessEditorial: Synthetic Live Biotherapeutic Products for Diseases
Editorial: Synthetic Live Biotherapeutic Products for Diseases
- Open Access
- 5 October 2018
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 46 (20) , 11115-11125
Engineered dCas9 with reduced toxicity in bacteria: implications for genetic circuit design
Large synthetic genetic circuits require the simultaneous expression of many regulators. Deactivated Cas9 (dCas9) can serve as a repressor by having a small guide RNA (sgRNA) direct it to bind a promoter. The programmability and specificity of RNA:DNA basepairing simplifies the generation of many orthogonal sgRNAs that, in theory, could serve as a large set of regulators in a circuit. However, dCas9 is toxic in many bacteria, thus limiting how high it can be expressed, and low concentrations are quickly sequestered by multiple sgRNAs. Here, we construct a non-toxic version of dCas9 by eliminating PAM (protospacer adjacent motif) binding with a R1335K mutation (dCas9*) and recovering DNA binding by fusing it to the PhlF repressor (dCas9*_PhlF). Both the 30 bp PhlF operator and 20 bp sgRNA binding site are required to repress a promoter. The larger region required for recognition mitigates toxicity in Escherichia coli, allowing up to 9600 ± 800 molecules of dCas9*_PhlF per cell before growth or morphology are impacted, as compared to 530 ± 40 molecules of dCas9. Further, PhlF multimerization leads to an increase in average cooperativity from n = 0.9 (dCas9) to 1.6 (dCas9*_PhlF). A set of 30 orthogonal sgRNA–promoter pairs are characterized as NOT gates; however, the simultaneous use of multiple sgRNAs leads to a monotonic decline in repression and after 15 are co-expressed the dynamic range is <10-fold. This work introduces a non-toxic variant of dCas9, critical for its use in applications in metabolic engineering and synthetic biology, and exposes a limitation in the number of regulators that can be used in one cell when they rely on a shared resource.
- 5 October 2018
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 46 (20) , 11115-11125
Open AccessEngineered dCas9 with reduced toxicity in bacteria: implications for genetic circuit design
Large synthetic genetic circuits require the simultaneous expression of many regulators. Deactivated Cas9 (dCas9) can serve as a repressor by having a small guide RNA (sgRNA) direct it to bind a promoter. The programmability and specificity of RNA:DNA basepairing simplifies the generation of many orthogonal sgRNAs that, in theory, could serve as a large set of regulators in a circuit. However, dCas9 is toxic in many bacteria, thus limiting how high it can be expressed, and low concentrations are quickly sequestered by multiple sgRNAs. Here, we construct a non-toxic version of dCas9 by eliminating PAM (protospacer adjacent motif) binding with a R1335K mutation (dCas9*) and recovering DNA binding by fusing it to the PhlF repressor (dCas9*_PhlF). Both the 30 bp PhlF operator and 20 bp sgRNA binding site are required to repress a promoter. The larger region required for recognition mitigates toxicity in Escherichia coli, allowing up to 9600 ± 800 molecules of dCas9*_PhlF per cell before growth or morphology are impacted, as compared to 530 ± 40 molecules of dCas9. Further, PhlF multimerization leads to an increase in average cooperativity from n = 0.9 (dCas9) to 1.6 (dCas9*_PhlF). A set of 30 orthogonal sgRNA–promoter pairs are characterized as NOT gates; however, the simultaneous use of multiple sgRNAs leads to a monotonic decline in repression and after 15 are co-expressed the dynamic range is <10-fold. This work introduces a non-toxic variant of dCas9, critical for its use in applications in metabolic engineering and synthetic biology, and exposes a limitation in the number of regulators that can be used in one cell when they rely on a shared resource.
- Open Access
- 9 December 2016
- Published by Frontiers Media SA in Frontiers in Microbiology
- Vol. 7, 1972
Natural and Synthetic Variants of the Tricarboxylic Acid Cycle in Cyanobacteria: Introduction of the GABA Shunt into Synechococcus sp. PCC 7002
For nearly half a century, it was believed that cyanobacteria had an incomplete tricarboxylic acid (TCA) cycle, because 2-oxoglutarate dehydrogenase (2-OGDH) was missing. Recently, a bypass route via succinic semialdehyde (SSA), which utilizes 2-oxoglutarate decarboxylase (OgdA) and succinic semialdehyde dehydrogenase (SsaD) to convert 2-oxoglutarate (2-OG) into succinate, was identified, thus completing the TCA cycle in most cyanobacteria. In addition to the recently characterized glyoxylate shunt that occurs in a few of cyanobacteria, the existence of a third variant of the TCA cycle connecting these metabolites, the γ-aminobutyric acid (GABA) shunt, was considered to be ambiguous because the GABA aminotransferase is missing in many cyanobacteria. In this study we isolated and biochemically characterized the enzymes of the GABA shunt. We show that N-acetylornithine aminotransferase (ArgD) can function as a GABA aminotransferase and that, together with glutamate decarboxylase (GadA), it can complete a functional GABA shunt. To prove the connectivity between the OgdA/SsaD bypass and the GABA shunt, the gadA gene from Synechocystis sp. PCC 6803 was heterologously expressed in Synechococcus sp. PCC 7002, which naturally lacks this enzyme. Metabolite profiling of seven Synechococcus sp. PCC 7002 mutant strains related to these two routes to succinate were investigated and proved the functional connectivity. Metabolite profiling also indicated that, compared to the OgdA/SsaD shunt, the GABA shunt was less efficient in converting 2-OG to SSA in Synechococcus sp. PCC 7002. The metabolic profiling study of these two TCA cycle variants provides new insights into carbon metabolism as well as evolution of the TCA cycle in cyanobacteria.
- 9 December 2016
- Published by Frontiers Media SA in Frontiers in Microbiology
- Vol. 7, 1972
Open AccessNatural and Synthetic Variants of the Tricarboxylic Acid Cycle in Cyanobacteria: Introduction of the GABA Shunt into Synechococcus sp. PCC 7002
For nearly half a century, it was believed that cyanobacteria had an incomplete tricarboxylic acid (TCA) cycle, because 2-oxoglutarate dehydrogenase (2-OGDH) was missing. Recently, a bypass route via succinic semialdehyde (SSA), which utilizes 2-oxoglutarate decarboxylase (OgdA) and succinic semialdehyde dehydrogenase (SsaD) to convert 2-oxoglutarate (2-OG) into succinate, was identified, thus completing the TCA cycle in most cyanobacteria. In addition to the recently characterized glyoxylate shunt that occurs in a few of cyanobacteria, the existence of a third variant of the TCA cycle connecting these metabolites, the γ-aminobutyric acid (GABA) shunt, was considered to be ambiguous because the GABA aminotransferase is missing in many cyanobacteria. In this study we isolated and biochemically characterized the enzymes of the GABA shunt. We show that N-acetylornithine aminotransferase (ArgD) can function as a GABA aminotransferase and that, together with glutamate decarboxylase (GadA), it can complete a functional GABA shunt. To prove the connectivity between the OgdA/SsaD bypass and the GABA shunt, the gadA gene from Synechocystis sp. PCC 6803 was heterologously expressed in Synechococcus sp. PCC 7002, which naturally lacks this enzyme. Metabolite profiling of seven Synechococcus sp. PCC 7002 mutant strains related to these two routes to succinate were investigated and proved the functional connectivity. Metabolite profiling also indicated that, compared to the OgdA/SsaD shunt, the GABA shunt was less efficient in converting 2-OG to SSA in Synechococcus sp. PCC 7002. The metabolic profiling study of these two TCA cycle variants provides new insights into carbon metabolism as well as evolution of the TCA cycle in cyanobacteria.
- 24 December 2015
- Published by Wiley in Biotechnology and Bioengineering
- Vol. 113 (7) , 1448-1459
Consequences of ccmR deletion on respiration, fermentation and H 2 metabolism in cyanobacterium Synechococcus sp . PCC 7002
CcmR, a LysR‐type transcriptional regulator, represses the genes encoding components of the high‐affinity carbon concentration mechanism in cyanobacteria. Unexpectedly, deletion of the ccmR gene was found to alter the expression of the terminal oxidase and fermentative genes, especially the hydrogenase operon in the cyanobacterium Synechococcus sp . PCC 7002. Consistent with the transcriptomic data, the deletion strain exhibits flux increases (30–50%) in both aerobic O 2 respiration and anaerobic H 2 evolution. To understand how CcmR influences anaerobic metabolism, the kinetics of autofermentation were investigated following photoautotrophic growth. The autofermentative H 2 yield increased by 50% in the CcmR deletion strain compared to the wild‐type strain, and increased to 160% (within 20 h) upon continuous removal of H 2 from the medium (“milking”) to suppress H 2 uptake. Consistent with this greater reductant flux to H 2 , the mutant excreted less lactate during autofermentation (NAD(P)H consuming pathway). To enhance the rate of NADH production during anaerobic metabolism, the ccmR mutant was engineered to introduce GAPDH overexpression (more NADH production) and LDH deletion (less NADH consumption). The triple mutant ( ccmR deletion + GAPDH overexpression + LDH deletion) showed 6–8‐fold greater H 2 yield than the WT strain, achieving conversion rates of 17 nmol 10 8 cells −1 h −1 and yield of 0.87 H 2 per glucose equivalent (8.9% theoretical maximum). Simultaneous monitoring of the intracellular NAD(P)H concentration and H 2 production rate by these mutants reveals an inverse correspondence between these variables indicating hydrogenase‐dependent H 2 production as a major sink for consuming NAD(P)H in preference to excretion of reduced carbon as lactate during fermentation. Biotechnol. Bioeng. 2016;113: 1448–1459.
- 24 December 2015
- Published by Wiley in Biotechnology and Bioengineering
- Vol. 113 (7) , 1448-1459
Consequences of ccmR deletion on respiration, fermentation and H 2 metabolism in cyanobacterium Synechococcus sp . PCC 7002
CcmR, a LysR‐type transcriptional regulator, represses the genes encoding components of the high‐affinity carbon concentration mechanism in cyanobacteria. Unexpectedly, deletion of the ccmR gene was found to alter the expression of the terminal oxidase and fermentative genes, especially the hydrogenase operon in the cyanobacterium Synechococcus sp . PCC 7002. Consistent with the transcriptomic data, the deletion strain exhibits flux increases (30–50%) in both aerobic O 2 respiration and anaerobic H 2 evolution. To understand how CcmR influences anaerobic metabolism, the kinetics of autofermentation were investigated following photoautotrophic growth. The autofermentative H 2 yield increased by 50% in the CcmR deletion strain compared to the wild‐type strain, and increased to 160% (within 20 h) upon continuous removal of H 2 from the medium (“milking”) to suppress H 2 uptake. Consistent with this greater reductant flux to H 2 , the mutant excreted less lactate during autofermentation (NAD(P)H consuming pathway). To enhance the rate of NADH production during anaerobic metabolism, the ccmR mutant was engineered to introduce GAPDH overexpression (more NADH production) and LDH deletion (less NADH consumption). The triple mutant ( ccmR deletion + GAPDH overexpression + LDH deletion) showed 6–8‐fold greater H 2 yield than the WT strain, achieving conversion rates of 17 nmol 10 8 cells −1 h −1 and yield of 0.87 H 2 per glucose equivalent (8.9% theoretical maximum). Simultaneous monitoring of the intracellular NAD(P)H concentration and H 2 production rate by these mutants reveals an inverse correspondence between these variables indicating hydrogenase‐dependent H 2 production as a major sink for consuming NAD(P)H in preference to excretion of reduced carbon as lactate during fermentation. Biotechnol. Bioeng. 2016;113: 1448–1459.