Analog Noise Reduction in Enzymatic Logic Gates

In this work we demonstrate both experimentally and theoretically that the analog noise generation by a single enzymatic logic gate can be dramatically reduced to yield gate operation with virtually no input noise amplification. This is achieved by exploiting the enzyme’s specificity when using a co-substrate that has a much lower affinity than the primary substrate. Under these conditions, we obtain a negligible increase in the noise output from the logic gate as compared to the input noise level. Experimental realizations of the AND logic gate with the enzyme horseradish peroxidase using hydrogen peroxide and two different co-substrates, 2,2’-azino-bis(3-ethylbenzthiazoline-6-sulphonic acid) (ABTS) and ferrocyanide, with vastly different rate constants confirmed our general theoretical conclusions.

💡 Research Summary

The paper addresses a fundamental obstacle in the practical deployment of enzymatic logic gates: the amplification of analog noise inherent to the highly nonlinear kinetics of enzyme‑catalyzed reactions. The authors propose a design principle that deliberately exploits the disparity in binding affinities between the primary substrate (hydrogen peroxide, H₂O₂) and a co‑substrate (the second input). By selecting a co‑substrate whose Michaelis constant (Kₘ) is much larger than that of the primary substrate, the second reaction step becomes rate‑limiting, thereby reducing the gain with which input fluctuations are transferred to the output.

A theoretical framework is constructed by extending the classic Michaelis–Menten model to a two‑step cascade: (1) rapid formation of the enzyme–H₂O₂ complex, followed by (2) slower binding of the co‑substrate to this complex. The authors derive an expression for the noise‑transfer coefficient η = ΔY_out / ΔX_in, where ΔX_in represents the standard deviation of the input concentrations (both H₂O₂ and co‑substrate) and ΔY_out denotes the standard deviation of the measured output (optical absorbance or electrochemical current). The analysis shows that η < 1—i.e., no amplification of input noise—can be achieved when the co‑substrate’s Kₘ is sufficiently high, effectively flattening the response surface of the gate.

Experimental validation is performed using horseradish peroxidase (HRP) as the catalytic enzyme to implement an AND gate. Two co‑substrates are examined: (i) ABTS (2,2′‑azino‑bis(3‑ethylbenzthiazoline‑6‑sulphonic acid)), which has a relatively high catalytic turnover (k_cat) and moderate affinity, and (ii) ferrocyanide (Fe(CN)₆⁴⁻), which exhibits a much lower k_cat and a markedly larger Kₘ. By varying the concentrations of H₂O₂ and the co‑substrate, as well as the reaction time, the authors record the output signals and compute η for each system.

The results confirm the theoretical predictions. The ABTS‑based gate shows η ≈ 1.3, meaning that input noise is amplified by roughly 30 %. In contrast, the ferrocyanide‑based gate yields η ≈ 1.02, essentially eliminating noise amplification. Moreover, the authors identify an optimal reaction time of about 60 seconds, at which η reaches its minimum and the signal‑to‑noise ratio (SNR) is maximized.

In the discussion, the authors argue that the same affinity‑tuning strategy can be generalized to other enzymatic platforms (e.g., glucose oxidase, alanine transaminase) and to larger networks of gates where cumulative noise becomes a critical issue. They suggest that each gate in a cascade could be engineered with a co‑substrate selected to keep η ≤ 1, thereby preserving signal fidelity throughout the network. Potential applications include bio‑electronic hybrid circuits, neuromorphic computing elements, and point‑of‑care diagnostic sensors where low power consumption and biocompatibility are essential.

In conclusion, the study demonstrates that by deliberately choosing a co‑substrate with low affinity for the enzyme, the analog noise generated by a single enzymatic logic gate can be dramatically suppressed. This simple yet powerful approach paves the way for more reliable and scalable biochemical computing architectures, bridging the gap between laboratory demonstrations and practical bio‑computing technologies.