Machine-Assisted Discovery of Chondroitinase ABC Complexes toward Sustained Neural Regeneration

Abstract

Among the many molecules that contribute to glial scarring, chondroitin sulfate proteoglycans (CSPGs) are known to be potent inhibitors of neuronal regeneration. Chondroitinase ABC (ChABC), a bacterial lyase, degrades the glycosaminoglycan (GAG) side chains of CSPGs and promotes tissue regeneration. However, ChABC is thermally unstable and loses all activity within a few hours at 37 °C under dilute conditions. To overcome this limitation, the discovery of a diverse set of tailor-made random copolymers that complex and stabilize ChABC at physiological temperature is reported. The copolymer designs, which are based on chain length and composition of the copolymers, are identified using an active machine learning paradigm, which involves iterative copolymer synthesis, testing for ChABC thermostability upon copolymer complexation, Gaussian process regression modeling, and Bayesian optimization. Copolymers are synthesized by automated PET-RAFT and thermostability of ChABC is assessed by retained enzyme activity (REA) after 24 h at 37 °C. Significant improvements in REA in three iterations of active learning are demonstrated while identifying exceptionally high-performing copolymers. Most remarkably, one designed copolymer promotes residual ChABC activity near 30%, even after one week and notably outperforms other common stabilization methods for ChABC. Together, these results highlight a promising pathway toward sustained tissue regeneration.

Keywords: chondroitinase ABC; data-driven design; glial scar degradation; machine learning; polymer-enzyme complexes; protein stabilization.

Figure 1.. Workflow schematic.

Automated polymer synthesis was conducted using a Hamilton Microlab Starlet. Initially, a library with over 500 unique copolymers were quickly evaluated for their ability to retain ChABC activity at 37°C for 24 hrs. Quantified retained enzyme activity (REA) information along with polymer composition and chain length were used to train a Gaussian Process Regression model (GPR) that predicts REA based on copolymer chemistry and chain length. Using active learning, 24 new polymers were proposed based on GPR modeling, and these polymers were subsequently synthesized and tested to provide additional data for GPR. After three of these active learning iterations, a few of the most promising designs were selected for further characterization

Figure 2.. Generational improvement for PECs.

A) Comparison of distributions of REA for ChABC-PECs for the seed database and three iterations of active learning. The distributions are represented as violins colored by REA as well as black candlesticks delimiting the second and third quartiles of the data, meaning that 50% of all data points in each iteration is included in the candle. The gray dash indicates the median REA, and the white circle indicates the average. The transparent gray shading is a guide to the eye to follow changes in summary statistics. B) Comparison of average compositions of copolymers in the top quartile for the seed dataset and polymers proposed in the three iterations of active learning. C) Comparison of average compositions of copolymers in the top quartile for the seed dataset and polymers proposed in the three iterations of active learning. In 2B) and 2C), the width of horizontal bars indicate the fraction of incorporation for each monomer; bars are organized in decreasing order of octanol-water partition coefficient, such that more hydrophobic monomers are at the left and more hydrophilic monomers are at the right.

Figure 3.. Retained enzyme activity (REA).

A) REA of ChABC in the presence of polymer at different concentrations at 37°C. PECs at different concentrations increased activity of enzyme at t=0 at least 2–3 fold and maintained high levels of enzyme for the initial few days. B) Comparison of PEC with common enzyme stabilizers Trehalose and Sucrose. C) PEC retained >100% activity while native ChABC lost all activity within 24 hrs. D) Activity of native ChABC and ChABC-PEC (12.5 μM) at varying substrate concentrations. Data represented here as mean±SD, n=3 for all experiments.

Figure 4.

Dynamic light scattering. Biophysical characterization of ChABC, copolymer, and ChABC-PEC using DLS.

Figure 5.. Toxicity and activity in vitro.

A) No cytotoxity was observed when astrocytes were treated with heteropolymer at various concentrations. B) TNF-α secretion by astrocytes treated with heteropolymers. No inflammation was observed in the presence of our copolymer constructs compared to control group. C) No cytotoxicity was observed when PECs were treated with astrocytes at two different concentrations. Data presented as mean±SD, n=6–9, P-values calculated using paired comparison plot with Tukey analysis.