Combined molecular dynamics and neural network method for predicting protein antifreeze activity

Kozuch, Daniel J.; Stillinger, Frank H.; Debenedetti, Pablo G.

doi:10.1073/pnas.1814945115

Cited by 42 publications

(49 citation statements)

References 57 publications

(38 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(18) TH generally increases with the area of the protein IBS and the strength of binding to ice, but it is also strongly modulated by properties of the non-ice binding site and the concentration of the AFPs in solution. (13,18,(36)(37)(38) Interestingly, experiments indicate that the TH of TmAFP is a non-monotonous function of the number of icebinding loops, with optimum activity for N = 9 TxT coils. (30) This indicates that the same properties have distinct effects on the ice nucleation and antifreeze efficiencies of ice-binding molecules.…”

Section: Discussionmentioning

confidence: 99%

“…(30) This indicates that the same properties have distinct effects on the ice nucleation and antifreeze efficiencies of ice-binding molecules. Combined with theories and heuristic approaches that predict TH, (18,37,38) the results of this work could be used to engineer at the molecular level responsive materials that can reconfigure through assembly and disassembly of individual ice-binding blocks to act as ice nucleants or antifreeze in response to environmental cues.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

How Do Size and Aggregation of Ice-binding Proteins Control their Ice Nucleation Efficiency

Qiu¹,

Hudait²,

Molinero³

2018

Preprint

View full text Add to dashboard Cite

Organisms that thrive at cold temperatures have evolved ice-binding proteins to manage the nucleation and growth of ice. Bacterial ice-nucleating proteins (INP) and insect hyperactive antifreeze proteins (AFP) bind ice through the same amino acid motifs, despite their opposite functions. AFPs are generally small, while INPs are long and aggregate in the cell membrane. It is not yet understood to which extent the size and aggregation determine the temperature Thet at which proteins nucleate ice. Here we address this question using molecular simulations and nucleation theory. The simulations indicate that the 2.5 nm long Tm AFP nucleates ice at 2±1 ° C above the homogeneous nucleation temperature. Addition of ice-binding loops to Tm AFP increases Thet until the length of the binding-site becomes ~4 times its width, beyond which Thet plateaus. We calculate that the INP of Ps. syringae, Ps INP, reaches its maximum Thet = -26 ° C when its binding site has 16 ice-binding loops, in excellent agreement with Thet = -25 ±1 ° C measured for an engineered 16-loop fragment of Ps INP. To further increase Thet , the proteins must aggregate. We predict Thet per number of Ps INP in the aggregate, and conclude that assemblies with 34 INP already reach Thet = -2 ° C characteristic of this bacterium. Interestingly, we find that Thet of aggregates is a non-monotonic and strongly varying function of the distance between proteins. We conclude that to achieve maximum freezing efficiency, bacteria must exert exquisite, sub-angstrom control of the distance between INP in their membrane.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

How Do Size and Aggregation of Ice-binding Proteins Control their Ice Nucleation Efficiency

Qiu¹,

Hudait²,

Molinero³

2018

Preprint

View full text Add to dashboard Cite

show abstract

“…Composition of k-spaced amino acid pairs. Several machine-learning approaches have been utilized to perform the prediction task for AFPs 28,42 . The fundamental task in developing a computation-based classification model is the translation of protein sequences to interpretative encoded numerical features.…”

Section: Methods Evaluation Parametersmentioning

confidence: 99%

AFP-LSE: Antifreeze Proteins Prediction Using Latent Space Encoding of Composition of k-Spaced Amino Acid Pairs

Usman

Khan

Lee

2020

Sci Rep

View full text Add to dashboard Cite

Species living in extremely cold environments resist the freezing conditions through antifreeze proteins (Afps). Apart from being essential proteins for various organisms living in sub-zero temperatures, Afpshave numerous applications in different industries. They possess very small resemblance to each other and cannot be easily identified using simple search algorithms such as BLAST and PSI-BLAST. Diverse AFPs found in fishes (Type I, II, III, IV and antifreeze glycoproteins (AFGPs)), are sub-types and show low sequence and structural similarity, making their accurate prediction challenging. Although several machine-learning methods have been proposed for the classification of AFPs, prediction methods that have greater reliability are required. In this paper, we propose a novel machine-learning-based approach for the prediction of AFP sequences using latent space learning through a deep auto-encoder method. For latent space pruning, we use the output of the auto-encoder with a deep neural network classifier to learn the non-linear mapping of the protein sequence descriptor and class label. The proposed method outperformed the existing methods, yielding excellent results in comparison. A comprehensive ablation study is performed, and the proposed method is evaluated in terms of widely used performance measures. In particular, the proposed method demonstrated a high Matthews correlation coefficient of 0.52, F-score of 0.49, and Youden's index of 0.81 on an independent test dataset, thereby outperforming the existing methods for Afp prediction.In Antarctic fish, a survival mechanism that prevented them from freezing in seawater at sub-zero temperatures was observed, which led to the discovery of antifreeze proteins (AFP) 1 . AFPs have been identified as a crucial substance for resisting a freezing environment in various species including plants, bacteria, fungi, insects, and animals 2 . Ice exists in different geometric shapes due to the varying arrangements of oxygen atoms; therefore, the structural and sequential arrangements of AFPs largely vary to accommodate this heterogeneity of ice molecules 3 . Ice also exhibits the property of recrystallization, by which small ice crystals bind to the water molecules, thus becoming a large ice lattice, causing severe damage to the cell membrane, which, in some cases, may be lethal 4 . AFPs are commonly categorized into glycoproteins (AFGPs) and non-glycoproteins (AFPs) 5 . They protect the organisms using two mechanisms: (i) thermal hysteresis (TH), by which the freezing point of water is depressed to a few degrees by the adsorption-inhibition effect without altering the melting point 6 ; (ii) ice crystal inhibition, by which the AFP sites bind to the surfaces of ice and inhibit their growth to become a larger ice lattice, developing either small harmless ice crystals or forming a needle-shaped lattice, thus diminishing the recrystallization property of ice 2 .AFPs are indispensable in organisms such as fish 7 , fungi 8 , bacteria 9 , plants 10 , and insects 11 . Furthermo...

show abstract

“…Kozuch and collaborators combined molecular dynamics of water molecules near an AFP surface and geometric structure of the IBS with neural networks aiming to predict the magnitude of the thermal hysteresis gap of AFPs [106]. To build the predictive model, the team used 17 AFP structures desposited in the Protein Data Bank and their corresponding TH activity as a function of AFP concentration, c AFP .…”

Section: Production Of Af(g)ps and Synthetic Analogues: Design Andmentioning

confidence: 99%

Peptidic Antifreeze Materials: Prospects and Challenges

Surís-Valls

Voets

2019

IJMS

View full text Add to dashboard Cite

Necessitated by the subzero temperatures and seasonal exposure to ice, various organisms have developed a remarkably effective means to survive the harsh climate of their natural habitats. Their ice-binding (glyco)proteins keep the nucleation and growth of ice crystals in check by recognizing and binding to specific ice crystal faces, which arrests further ice growth and inhibits ice recrystallization (IRI). Inspired by the success of this adaptive strategy, various approaches have been proposed over the past decades to engineer materials that harness these cryoprotective features. In this review we discuss the prospects and challenges associated with these advances focusing in particular on peptidic antifreeze materials both identical and akin to natural ice-binding proteins (IBPs). We address the latest advances in their design, synthesis, characterization and application in preservation of biologics and foods. Particular attention is devoted to insights in structure-activity relations culminating in the synthesis of de novo peptide analogues. These are sequences that resemble but are not identical to naturally occurring IBPs. We also draw attention to impactful developments in solid-phase peptide synthesis and ‘greener’ synthesis routes, which may aid to overcome one of the major bottlenecks in the translation of this technology: unavailability of large quantities of low-cost antifreeze materials with excellent IRI activity at (sub)micromolar concentrations.

show abstract

Combined molecular dynamics and neural network method for predicting protein antifreeze activity

Cited by 42 publications

References 57 publications

How Do Size and Aggregation of Ice-binding Proteins Control their Ice Nucleation Efficiency

How Do Size and Aggregation of Ice-binding Proteins Control their Ice Nucleation Efficiency

AFP-LSE: Antifreeze Proteins Prediction Using Latent Space Encoding of Composition of k-Spaced Amino Acid Pairs

Peptidic Antifreeze Materials: Prospects and Challenges

Contact Info

Product

Resources

About