Matthew A. Spence
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Matthew A. Spence has a total of 147 co-authors. Here are the top co-authors by number of shared publications:
- Colin J. Jackson
Research School of Chemistry Australian National University Canberra Australian Capital Territory Australia
17 shared publications - Ashley M. Buckle
Department of Biochemistry and Molecular Biology, Biomedicine Discovery Institute, Monash University, Clayton, Australia
4 shared publications - Dana S. Matthews
Research School of Chemistry, The Australian National University, Canberra, ACT 2601, Australia
4 shared publications Mahakaran Sandhu
Research School of Chemistry, The Australian National University, Canberra, ACT 2601, Australia
4 shared publications- Audrey A. Burnim
Department of Chemistry and Chemical Biology, Cornell University
3 shared publications - Da Xu
Johns Hopkins University
3 shared publications
Latest Publications
- 11 November 2026
- Published by Springer Nature in Nature Chemical Biology
Temperature dependence of binding entropy drives evolution in LacI family proteins
Proteins operate through ligand and solvent interactions governed by thermodynamics, yet the enthalpy-entropy trade-offs that guide functional evolution remain unclear. Although the evolution of ligand specificity has been well studied, the thermodynamic determinants that underlie changes in specificity are not well understood. Here we reveal a shift from entropy-driven binding in the most distant ancestor to the LacI/GalR family (LGF) of transcription factors, to enthalpy-driven binding in the most recent ancestor and extant LacI. The most distant ancestor is characterized by the ability to bind its ligand in an open and dynamic conformation. We propose that entropically driven binding is allowed by the presence of entropic reservoirs. This thermodynamic binding trade-off between the most distant and most recent ancestor is in accordance with the concept that ancient life existed in a hot environment, where higher temperatures enhanced entropically driven binding. This suggests that molecular binding mechanisms evolved not just for ligand specificity but to adapt to environmental pressures such as cooling temperatures where enthalpic binding modes are favored.
- 11 November 2026
- Published by Springer Nature in Nature Chemical Biology
Temperature dependence of binding entropy drives evolution in LacI family proteins
Proteins operate through ligand and solvent interactions governed by thermodynamics, yet the enthalpy-entropy trade-offs that guide functional evolution remain unclear. Although the evolution of ligand specificity has been well studied, the thermodynamic determinants that underlie changes in specificity are not well understood. Here we reveal a shift from entropy-driven binding in the most distant ancestor to the LacI/GalR family (LGF) of transcription factors, to enthalpy-driven binding in the most recent ancestor and extant LacI. The most distant ancestor is characterized by the ability to bind its ligand in an open and dynamic conformation. We propose that entropically driven binding is allowed by the presence of entropic reservoirs. This thermodynamic binding trade-off between the most distant and most recent ancestor is in accordance with the concept that ancient life existed in a hot environment, where higher temperatures enhanced entropically driven binding. This suggests that molecular binding mechanisms evolved not just for ligand specificity but to adapt to environmental pressures such as cooling temperatures where enthalpic binding modes are favored.
- Open Access
- 4 September 2026
- Published by Portland Press Ltd. in Biochemical Journal
- Vol. 483 (10) , 1891-1913
Graph-based representations in modern protein science
For more than 50 years, the linear sequence and the multiple sequence alignment have been the foundational data structures of protein science, and they remain central to homology search, phylogenetic inference, covariance-based contact prediction, and modern protein language models. However, relational and graph-based representations are increasingly being adopted alongside sequence-based methods to capture biological relationships that linear data structures express only implicitly. Proteins fold as three-dimensional residue interaction networks, evolve through high-dimensional genotype networks defined by mutational connectivity, and operate within cellular protein–protein interaction graphs. Here, we review how graph theory is being used to describe and understand these relationships across protein science, with an emphasis on what these methods offer biochemists working on enzyme superfamilies, protein engineering, drug targets, and functional annotation. We trace the development of these ideas from early theoretical topologies, through statistical coupling and the structural network analyses, to the geometric and graph-like representations used in recent machine-learning-driven advances. Throughout, we emphasise that graphs do not replace sequences or MSAs but provide a complementary representation for biochemical relationships that are difficult to express in one dimension.
- 4 September 2026
- Published by Portland Press Ltd. in Biochemical Journal
- Vol. 483 (10) , 1891-1913
Open AccessGraph-based representations in modern protein science
For more than 50 years, the linear sequence and the multiple sequence alignment have been the foundational data structures of protein science, and they remain central to homology search, phylogenetic inference, covariance-based contact prediction, and modern protein language models. However, relational and graph-based representations are increasingly being adopted alongside sequence-based methods to capture biological relationships that linear data structures express only implicitly. Proteins fold as three-dimensional residue interaction networks, evolve through high-dimensional genotype networks defined by mutational connectivity, and operate within cellular protein–protein interaction graphs. Here, we review how graph theory is being used to describe and understand these relationships across protein science, with an emphasis on what these methods offer biochemists working on enzyme superfamilies, protein engineering, drug targets, and functional annotation. We trace the development of these ideas from early theoretical topologies, through statistical coupling and the structural network analyses, to the geometric and graph-like representations used in recent machine-learning-driven advances. Throughout, we emphasise that graphs do not replace sequences or MSAs but provide a complementary representation for biochemical relationships that are difficult to express in one dimension.
- Open Access
- 16 May 2025
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 11 (20) , eads8318
Ancestral reconstruction of polyethylene terephthalate degrading cutinases reveals a rugged and unexplored sequence-fitness landscape
The use of protein engineering to generate enzymes for the degradation of polyethylene terephthalate (PET) is a promising route for plastic recycling, yet traditional engineering approaches often fail to explore protein sequence space for optimal enzymes. In this work, we use multiplexed ancestral sequence reconstruction (mASR) to address this, exploring the evolutionary sequence space of PET-degrading cutinases. Using 20 statistically equivalent phylogenies of the bacterial cutinase family, we generated 48 ancestral sequences revealing a wide range of PETase activities, highlighting the value of mASR in uncovering functional variants. Our findings show PETase activity can evolve through multiple pathways involving mutations remote from the active site. Moreover, analyzing the PETase fitness landscape with local ancestral sequence embedding (LASE) revealed that LASE can capture sequence features linked to PETase activity. This work highlights mASR’s potential in exploration of sequence space and underscores the use of LASE in readily mapping the protein fitness landscapes.
- 16 May 2025
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 11 (20) , eads8318
Open AccessAncestral reconstruction of polyethylene terephthalate degrading cutinases reveals a rugged and unexplored sequence-fitness landscape
The use of protein engineering to generate enzymes for the degradation of polyethylene terephthalate (PET) is a promising route for plastic recycling, yet traditional engineering approaches often fail to explore protein sequence space for optimal enzymes. In this work, we use multiplexed ancestral sequence reconstruction (mASR) to address this, exploring the evolutionary sequence space of PET-degrading cutinases. Using 20 statistically equivalent phylogenies of the bacterial cutinase family, we generated 48 ancestral sequences revealing a wide range of PETase activities, highlighting the value of mASR in uncovering functional variants. Our findings show PETase activity can evolve through multiple pathways involving mutations remote from the active site. Moreover, analyzing the PETase fitness landscape with local ancestral sequence embedding (LASE) revealed that LASE can capture sequence features linked to PETase activity. This work highlights mASR’s potential in exploration of sequence space and underscores the use of LASE in readily mapping the protein fitness landscapes.
- Open Access
The composition of biophysical constraints generates complex, rugged regions of protein fitness landscapes
Epistasis, in which the effect of a mutation depends on the genetic background it is introduced into, drives protein evolution and design by shaping the fitness landscape. Quantifying how constrained by epistasis (and thus rugged) a fitness landscape is when data is combinatorially incomplete, however, remains challenging. Here, we propose a spectral graph theory approach to measure fitness landscape ruggedness as the characteristic time step of a heat diffusion process. Our approach is conceptually analogous to modelling a fitness signal as heat and measuring the ruggedness of the fitness landscape as the length of time that signal has diffused over a network graph for. Through theory and simulation, we show that our diffusion-based approach describes the global ruggedness of a fitness landscape, even when only a fraction of the fitness map is experimentally characterized. Applying this approach to empirical fitness landscapes that encompass deep mutational scans, combinatorially complete datasets, and sparse sampling of homologous sequences, we find large variation in ruggedness both among and within different protein folds and functions. Finally, we use spectral graph theory to demonstrate that ruggedness is intimately related to the fraction of sequence space that maintains sufficient fitness to overcome purifying selection, which we term the solution set of the fitness function. These findings offer a theoretically grounded explanation for why complex biophysical traits often exhibit strong epistasis, while simpler traits, such as fold stability, evolve over smooth fitness landscapes. Our results underscore how spectral methods can help unify and quantify the interplay between fitness constraints and evolutionary outcomes in protein sequence space.
Open AccessThe composition of biophysical constraints generates complex, rugged regions of protein fitness landscapes
Epistasis, in which the effect of a mutation depends on the genetic background it is introduced into, drives protein evolution and design by shaping the fitness landscape. Quantifying how constrained by epistasis (and thus rugged) a fitness landscape is when data is combinatorially incomplete, however, remains challenging. Here, we propose a spectral graph theory approach to measure fitness landscape ruggedness as the characteristic time step of a heat diffusion process. Our approach is conceptually analogous to modelling a fitness signal as heat and measuring the ruggedness of the fitness landscape as the length of time that signal has diffused over a network graph for. Through theory and simulation, we show that our diffusion-based approach describes the global ruggedness of a fitness landscape, even when only a fraction of the fitness map is experimentally characterized. Applying this approach to empirical fitness landscapes that encompass deep mutational scans, combinatorially complete datasets, and sparse sampling of homologous sequences, we find large variation in ruggedness both among and within different protein folds and functions. Finally, we use spectral graph theory to demonstrate that ruggedness is intimately related to the fraction of sequence space that maintains sufficient fitness to overcome purifying selection, which we term the solution set of the fitness function. These findings offer a theoretically grounded explanation for why complex biophysical traits often exhibit strong epistasis, while simpler traits, such as fold stability, evolve over smooth fitness landscapes. Our results underscore how spectral methods can help unify and quantify the interplay between fitness constraints and evolutionary outcomes in protein sequence space.
- 25 March 2025
- Published by American Chemical Society (ACS) in Biochemistry
- Vol. 64 (8) , 1673-1684
Computational and Experimental Exploration of Protein Fitness Landscapes: Navigating Smooth and Rugged Terrains
Proteins evolve through complex sequence spaces, with fitness landscapes serving as a conceptual framework that links sequence to function. Fitness landscapes can be smooth, where multiple similarly accessible evolutionary paths are available, or rugged, where the presence of multiple local fitness optima complicate evolution and prediction. Indeed, many proteins, especially those with complex functions or under multiple selection pressures, exist on rugged fitness landscapes. Here we discuss the theoretical framework that underpins our understanding of fitness landscapes, alongside recent work that has advanced our understanding─particularly the biophysical basis for smoothness versus ruggedness. Finally, we address the rapid advances that have been made in computational and experimental exploration and exploitation of fitness landscapes, and how these can identify efficient routes to protein optimization.
- 25 March 2025
- Published by American Chemical Society (ACS) in Biochemistry
- Vol. 64 (8) , 1673-1684
Computational and Experimental Exploration of Protein Fitness Landscapes: Navigating Smooth and Rugged Terrains
Proteins evolve through complex sequence spaces, with fitness landscapes serving as a conceptual framework that links sequence to function. Fitness landscapes can be smooth, where multiple similarly accessible evolutionary paths are available, or rugged, where the presence of multiple local fitness optima complicate evolution and prediction. Indeed, many proteins, especially those with complex functions or under multiple selection pressures, exist on rugged fitness landscapes. Here we discuss the theoretical framework that underpins our understanding of fitness landscapes, alongside recent work that has advanced our understanding─particularly the biophysical basis for smoothness versus ruggedness. Finally, we address the rapid advances that have been made in computational and experimental exploration and exploitation of fitness landscapes, and how these can identify efficient routes to protein optimization.
- Open Access
The temperature dependence of binding entropy is a selective pressure in protein evolution
Proteins operate through ligand and solvent interactions governed by thermodynamics, yet the enthalpy-entropy trade-offs that guide their functional evolution remain poorly understood. The LacI/GalR family (LGF) of transcription factors provides a system for examining how these trade-offs evolve over billions of years but has seldom been studied from a full thermodynamic perspective. While the evolution of ligand specificity has been well-studied, the thermodynamic determinants underlying the changes in specificity is not well understood – especially how proteins alter thermodynamic strategies to optimize affinity. By reconstructing LGF ancestors, we reveal a shift from entropy-driven binding in the most distant ancestor, to enthalpy-driven binding in the most recent ancestor and extant LacI. The most distant ancestor is characterized by the ability to bind its ligand in an open and dynamic conformation, and we propose that entropically-driven binding is driven by the presence of entropic reservoirs. This thermodynamic binding trade-off between the most distant and most recent ancestor is in accordance with the concept of ancient life that existed in a hot Earth environment, where higher temperatures enhanced entropically-driven binding. This suggests that molecular binding mechanisms evolved not just for ligand specificity, but to adapt to environmental pressures such as cooling Earth temperatures where enthalpic binding modes are favored.
Open AccessThe temperature dependence of binding entropy is a selective pressure in protein evolution
Proteins operate through ligand and solvent interactions governed by thermodynamics, yet the enthalpy-entropy trade-offs that guide their functional evolution remain poorly understood. The LacI/GalR family (LGF) of transcription factors provides a system for examining how these trade-offs evolve over billions of years but has seldom been studied from a full thermodynamic perspective. While the evolution of ligand specificity has been well-studied, the thermodynamic determinants underlying the changes in specificity is not well understood – especially how proteins alter thermodynamic strategies to optimize affinity. By reconstructing LGF ancestors, we reveal a shift from entropy-driven binding in the most distant ancestor, to enthalpy-driven binding in the most recent ancestor and extant LacI. The most distant ancestor is characterized by the ability to bind its ligand in an open and dynamic conformation, and we propose that entropically-driven binding is driven by the presence of entropic reservoirs. This thermodynamic binding trade-off between the most distant and most recent ancestor is in accordance with the concept of ancient life that existed in a hot Earth environment, where higher temperatures enhanced entropically-driven binding. This suggests that molecular binding mechanisms evolved not just for ligand specificity, but to adapt to environmental pressures such as cooling Earth temperatures where enthalpic binding modes are favored.
- 18 December 2024
- Published by Springer Nature in Nature Machine Intelligence
- Vol. 6 (12) , 1542-1555
Leveraging ancestral sequence reconstruction for protein representation learning
- 18 December 2024
- Published by Springer Nature in Nature Machine Intelligence
- Vol. 6 (12) , 1542-1555
Leveraging ancestral sequence reconstruction for protein representation learning
- Open Access
- 16 November 2024
- Published by Oxford University Press (OUP) in Journal of Experimental Botany
- Vol. 76 (4) , 1099-1115
Accounting for the impact of genotype and environment on variation in leaf respiration of wheat in Mexico and Australia
An approach to improving radiation use efficiency (RUE) in wheat is to screen for variability in rates of leaf respiration in darkness (Rdark). We used a high-throughput system to quantify variation in Rdark among a diverse range of spring wheat genotypes (301 lines) grown in two countries (Mexico and Australia) and two seasons (2017 and 2018), and in doing so quantify the relative importance of genotype (G) and environment (E) in influencing variations in leaf Rdark. Through careful design, residual (unexplained) variation represented less than 10% of the total observed. Up to a third of the variation in Rdark (and related traits) was under genetic control. This suggests opportunities for breeders to use Rdark as a novel selection tool. In addition, E accounted for more than half of the total variation in area-based rates of Rdark. Here, the day of measurement was crucial, suggesting that day-to-day variations in the environment influence rates of Rdark measured at a common temperature. Overall, this study provides new insights into the role G and E play in determining variation in rates of leaf Rdark of one of the most important cereal crops, with implications for future improvements in carbon use efficiency and yield.
- 16 November 2024
- Published by Oxford University Press (OUP) in Journal of Experimental Botany
- Vol. 76 (4) , 1099-1115
Open AccessAccounting for the impact of genotype and environment on variation in leaf respiration of wheat in Mexico and Australia
An approach to improving radiation use efficiency (RUE) in wheat is to screen for variability in rates of leaf respiration in darkness (Rdark). We used a high-throughput system to quantify variation in Rdark among a diverse range of spring wheat genotypes (301 lines) grown in two countries (Mexico and Australia) and two seasons (2017 and 2018), and in doing so quantify the relative importance of genotype (G) and environment (E) in influencing variations in leaf Rdark. Through careful design, residual (unexplained) variation represented less than 10% of the total observed. Up to a third of the variation in Rdark (and related traits) was under genetic control. This suggests opportunities for breeders to use Rdark as a novel selection tool. In addition, E accounted for more than half of the total variation in area-based rates of Rdark. Here, the day of measurement was crucial, suggesting that day-to-day variations in the environment influence rates of Rdark measured at a common temperature. Overall, this study provides new insights into the role G and E play in determining variation in rates of leaf Rdark of one of the most important cereal crops, with implications for future improvements in carbon use efficiency and yield.
- Open Access
Ancestral reconstruction of polyethylene terephthalate degrading cutinases reveals a rugged and unexplored sequence-fitness landscape
The enzymatic degradation of polyethylene terephthalate (PET) is a promising method of advanced plastic recycling. Traditional protein engineering methods often fall short in exploring protein sequence space for optimal enzymes due to structural and rational design limitations. Our study addresses this by using multiplexed ancestral sequence reconstruction (mASR) to explore the evolutionary sequence space of PET-degrading cutinases. With a dataset of 397 cutinase sequences, we created a diverse library of ancestral sequences. Experimental characterization of 48 ancestral sequences revealed a wide range of PETase activities, highlighting the value of mASR in uncovering functional variants when compared to traditional ASR. Our results showed that PETase activity in cutinases evolved through diverse pathways involving distal mutations to the active site, and is readily accessible within this family. Additionally, our analysis of the PETase fitness landscape using one-hot encoding (OHE) and local ancestral sequence embedding (LASE) highlighted the effectiveness of LASE in capturing sequence features relevant to activity. This work emphasizes the utility of mASR as a protein engineering tool for identifying enhanced PET-degrading enzymes, and the advantages of the LASE embedding scheme in mapping the PETase fitness landscape.
Open AccessAncestral reconstruction of polyethylene terephthalate degrading cutinases reveals a rugged and unexplored sequence-fitness landscape
The enzymatic degradation of polyethylene terephthalate (PET) is a promising method of advanced plastic recycling. Traditional protein engineering methods often fall short in exploring protein sequence space for optimal enzymes due to structural and rational design limitations. Our study addresses this by using multiplexed ancestral sequence reconstruction (mASR) to explore the evolutionary sequence space of PET-degrading cutinases. With a dataset of 397 cutinase sequences, we created a diverse library of ancestral sequences. Experimental characterization of 48 ancestral sequences revealed a wide range of PETase activities, highlighting the value of mASR in uncovering functional variants when compared to traditional ASR. Our results showed that PETase activity in cutinases evolved through diverse pathways involving distal mutations to the active site, and is readily accessible within this family. Additionally, our analysis of the PETase fitness landscape using one-hot encoding (OHE) and local ancestral sequence embedding (LASE) highlighted the effectiveness of LASE in capturing sequence features relevant to activity. This work emphasizes the utility of mASR as a protein engineering tool for identifying enhanced PET-degrading enzymes, and the advantages of the LASE embedding scheme in mapping the PETase fitness landscape.
- Open Access
Leveraging ancestral sequence reconstruction for protein representation learning
Protein language models (PLMs) convert amino acid sequences into the numerical representations required to train machine learning (ML) models. Many PLMs are large (>600 M parameters) and trained on a broad span of protein sequence space. However, these models have limitations in terms of predictive accuracy and computational cost. Here, we use multiplexed Ancestral Sequence Reconstruction (mASR) to generate small but focused functional protein sequence datasets for PLM training. Compared to large PLMs, this local ancestral sequence embedding (LASE) produces representations 10-fold faster and with higher predictive accuracy. We show that due to the evolutionary nature of the ASR data, LASE produces smoother fitness landscapes in which protein variants that are closer in fitness value become numerically closer in representation space. This work contributes to the implementation of ML-based protein design in real-world settings, where data is sparse and computational resources are limited.
Open AccessLeveraging ancestral sequence reconstruction for protein representation learning
Protein language models (PLMs) convert amino acid sequences into the numerical representations required to train machine learning (ML) models. Many PLMs are large (>600 M parameters) and trained on a broad span of protein sequence space. However, these models have limitations in terms of predictive accuracy and computational cost. Here, we use multiplexed Ancestral Sequence Reconstruction (mASR) to generate small but focused functional protein sequence datasets for PLM training. Compared to large PLMs, this local ancestral sequence embedding (LASE) produces representations 10-fold faster and with higher predictive accuracy. We show that due to the evolutionary nature of the ASR data, LASE produces smoother fitness landscapes in which protein variants that are closer in fitness value become numerically closer in representation space. This work contributes to the implementation of ML-based protein design in real-world settings, where data is sparse and computational resources are limited.