betaFeedback

WorksComputer Science

A Copper-Inducible Yeast Biosensor with Bioluminescent Readout: Mathematical Modeling, Experimental Validation, and Photodetector Coupling for Molecular Communication in IoBNT

Abstract

Engineered living biosensors constitute promising model organisms for bringing Molecular Communications (MC) closer to practical applications. However, challenges associated with their development and characterization have resulted in the relevant literature being limited in the MC domain, . In this paper, we contribute in this direction by leveraging engineered yeast as a basis to detect heavy metals such as copper. The biosensor's development is supported by a validated mathematical characterization and analysis for possible integration in an MC system. Different from the existing literature, mathematical representations of the major biological processes within the yeast copper detoxification pathway are integrated into a comprehensive framework that predicts the time evolution of the CUP1 gene, which is treated as the system output. In the engineered circuit, CUP1 was replaced with GFP, therefore making GFP fluorescence the final system output. Experimental analysis including mRNA transcript quantification and fluorometric measurements, validate the GFP reporting model to a satisfactory degree. This establishes the proposed model as a useful tool for virtual experimentation and the developed testbed as a suitable platform for further engineering of yeast-based copper sensors. The theoretical analysis is further extended by incorporating a self-sustained bioluminescent reporter for yeast, replacing GFP and enabling optical output without external light excitation. The optical response is coupled to a photodetector model, rendering the system suitable for integration into mobile monitoring devices such as Smart Pills. The presented characterization of the yeast receiver as a copper biosensor, constitutes the first step towards the development and characterization of a complete end-to-end MC based system for gut theranostics after coupling with suitable engineered probiotic senders.

Introduction

Over the past decade, MC has attracted increasing attention in the scientific community, driven by advances in both theory and experimentation. To date, the literature has predominantly focused on theoretical foundations, including computational frameworks and modeling techniques for channel characterization, localization, and information rate optimization. In parallel, MC has been positioned as an enabling layer for applications ranging from industrial scenarios such as IoT-based inspection and monitoring in pipelines [1] to healthcare, including the Internet of Bio-Nano Things (IoBNT) for disease monitoring and treatment [2], intra-vessel monitoring [3], and particulate targeted drug delivery for tumors [4].

Nevertheless, the translation of the solid theoretical base into realistic settings requires experimental testbeds that can reproduce key propagation and signal transduction mechanisms at relevant scales. Consequently, a growing body of work has developed MC testbeds at both the macroscale and microscale. Microscale implementations frequently incorporate biological entities as transmitters and receivers coupled through a fluidic channel, thereby enabling molecular propagation in controlled environments. However, among the reported biological testbeds, prokaryotic organisms are mostly commonly adopted, whereas only a limited number of studies consider eukaryotic communication units [5, 6, 7, 8]. Importantly, irrespective of organism class, microorganisms can exhibit measurable responses to external molecular stimuli. Such an attribute, naturally connects MC testbeds to biosensing applications and in particular with living biosensors.

Bacteria and yeast constitute notable model organisms which have been leveraged for living biosensor development, however, their integration into the MC framework remains limited with a notable bacteria based example reported in [9]. The co-integration of living sensing components with embedded electronics for in vivo biomarker monitoring (with the most notable enabling technology being Smart Pills [10]) underscores the practical relevance of biosensing-based receivers and motivates the potential of MC integration. Moreover, living biosensors can also be envisioned as MC senders e.g. in the context of Organ-on-Chip (OoC) integration [11], further strengthening their potential to serve the MC domain.

Yeast-based biosensors offer advantages over bacterial systems due to their closer cellular and molecular similarity to higher eukaryotes. In particular, Saccharomyces cerevisiae (S. cerevisiae hereafter) shares key features with mammalian cells, making it a relevant model for sensing human-related biomarkers. Combined with advances in synthetic biology and genome editing, yeast can be readily engineered to sense targets either through its native pathways or by incorporating heterologous genes. The unique properties of yeast underscore its biosensing potential, particularly for detecting short-lived disease biomarkers in humans [7]. However, the capability of MC integrated into yeast biosensors has so far remained largely untapped, therefore limiting their ability to play a pivotal role in human healthcare.

Previous work on a yeast-based MC testbed [7], [8] developed a mathematical framework for pheromone-mediated yeast cell-to-cell (cell2cell) communication, and was supported by extensive experimental trials to validate theoretical findings. Subsequent research focused on noise analysis tailored for eukaryotic systems, led to the development of Maximum-Likelihood schemes for signal detection. In this work, departing from previous biosensor designs within the MC literature, a yeast biosensor that incorporates the unicellular S. cerevisiae as the sensing unit for the detection of cytotoxic copper accumulation in the gut is developed. A comprehensive mathematical model describing the major biological processes and mechanistic interactions involved in the copper detoxification pathway is developed. Particular emphasis is placed on the dynamics of the underlying biochemical reactions, and the interplay among the involved chemical compounds. In response to excessive copper accumulation in their vicinity, S. cerevisiae cells are induced to express a detoxifying gene, CUP1, which is treated as the system output as well as an indication of potentially toxic heavy-metal levels. Building upon the yeast MC testbed reported in [7], a new testbed is developed and is utilized as a validation tool for the mathematical analysis. Moreover, the theoretical characterization of the biosensor is developed with Smart Pill integration as a guiding objective. Overall, the envisioned MC system includes the application of living senders that can detect and transmit molecular signals in areas such as the gut, and encapsulated MC receivers that are able to forward the detected information to external monitor devices, thereby aiming to advance the IoBNT toward intra-body diagnostics and therapeutics. The contributions of this paper can be summarized as follows:

  • Dynamic integration of the major reactions involved in the copper detoxification pathway into a system of coupled nonlinear Ordinary Differential Equations (ODEs). To the best of the authors’ knowledge, copper uptake, intracellular dissemination through the metallochaperone pathways, copper-dependent activation of transcriptional regulators, and gene expression have not previously been jointly represented within a dynamic receiver model.

  • Development of the yeast module of the biosensor as an MC receiver using the new testbed developed. Experimental trials via fluorometric measurements characterize the yeast’s response, in both mRNA and protein level. mRNA and fluorescence measurements are also employed to assess the specificity of the proposed biosensor against another chemically related metal, namely zinc.

  • Cross matching between experimental results and theoretical findings for the GFP reporter, reveals satisfying concordance between the two across multiple concentrations. This renders the mathematical tool presented here as a solid base for further research.

  • Extension of the receiver model to include bioluminescence as an appropriate reporting output, coupled to a photodiode response. Toward this goal, the GFP reporter models are replaced by Lux-based models. The resulting Lux response is treated as a predictive extension that is not experimentally validated in the present work.

The modeling effort here has been centered around the theoretical characterization of the yeast biosensor as a copper receiver, where a direct copper input is assumed. The receiver model developed in this work however, may provide the basis for its integration with a transmission entity and a channel sub-system into a complete MC framework. Communication performance metrics such as achievable data rates and error analysis will also be able to be investigated.

The rest of the paper is organized as follows: In Section II the current state of the art in biosensors is presented. In Section III the overall system under consideration is described, and the design principles of the biosensor are outlined. The mathematical characterization of the yeast MC receiver followed by integration to a photodetector is discussed in detail in Section IV. In Section V, the experimental platform is described, and the methods and materials used for the experiments are provided. In Section VI both theoretical and experimental results are presented, and an attempt to validate the GFP predictions is reported. Moreover, the models derived for the extension to a luminescent response, are used to characterize the biosensor’s output and conduct a sensitivity analysis to investigate the robustness of the mathematical approach. Finally, Section VII summarizes the paper’s findings and provides some directions for future work.

Reported biosensors

A yeast-derived biosensor for copper monitoring in the gut is presented, with Smart Pill integration considered as a target deployment scenario. Prominent Smart Pill biosensors suitable for in vivo operation have been reported in [10] and [12]. In [10], probiotic bacteria were engineered to detect biomarkers associated with Inflammatory Bowel Disease (IBD), including nitric oxide, hydrogen peroxide (H2O2\mathrm{H}_{2}\mathrm{O}_{2}), tetrathionate, and thiosulfate. In vivo validation was performed in both mice and pigs, and GFP readouts indicated that adequate sensitivity and specificity were achieved for the targeted analytes. The calibrated bacterial sensors were subsequently encapsulated, and GFP was replaced with the luxABCDE operon to enable autonomous bioluminescent reporting. With the exception of the thiosulfate sensor, specificity and sensitivity were largely preserved after encapsulation. Analyte information was successfully transmitted to external units, thereby reducing the need for direct sampling from the gastrointestinal tract or from stool. In [12], swallowable bacterial sensors designed to respond to heme within the intestinal environment were reported. Magnetic particles were incorporated to facilitate retrieval from stool, while sodium alginate droplets were used for encapsulation to enhance survival through digestion. Performance was assessed in a colitis mouse model, where microsphere recovery and analysis were completed in approximately 25 minutes. The reported response scaled with disease severity.

Additional pill-form biosensors have been reported in [13, 14, 15, 16]. In [13], a glucose-monitoring capsule was reported in which two electrodes (forming a glucose biofuel cell (BFC)), nano-circuitry, and a magnetic human-body communication (mHBC) antenna operating from 40 to 200 MHz were integrated for in-gut sensing. Energy harvesting was enabled through glucose-driven BFC operation, thereby reducing reliance on battery sources. The electronic components were insulated using silicone/polyurethane coatings, while an enteric coating was employed to protect the electrodes from the stomach environment. Signal transmission was achieved through voltage-to-frequency conversion, whereby glucose concentration was mapped to the carrier frequency. Sensor calibration was performed in vitro, whereas in situ validation in pigs demonstrated frequency variations in response to different glucose concentrations through supplementation of saline samples. Specificity was assessed by applying alternative chemical inputs, for which baseline frequency operation was maintained. In [14], a capsule biosensor targeting Reactive Oxygen Species (ROS) associated with IBD was presented, where a blue dye readout was released upon ROS exposure. A modified dextran coating was employed and was dissolved in the presence of ROS, enabling dye release and subsequent stool-based readout. Evaluation in a colitis mouse model yielded 72% sensitivity and 78% specificity. In [15], a multi-analyte capsule platform was reported. It comprised an array of sensing modalities including aptamer-based voltammetric serotonin sensors, enzymatic amperometric glucose sensors, ion-selective potentiometric pH sensors, and impedimetric ionic-strength sensors. In vivo experiments were conducted in mice and rabbits, where fasting protocols were applied prior to administration of dietary samples to trigger secretion of the targeted compounds. In [16], a gas-sensing capsule was reported in which an oxygen-responsive membrane was utilized as the sensing element. Power was supplied using coin-cell batteries, while fluorescence emission was generated through membrane-oxygen interaction. Optical excitation was provided by integrated LEDs, and the resulting signal was transmitted via Bluetooth to an external device. Capsule localization was achieved using an external planar wearable coil, through magnetic-field excitation of an embedded magnetic sensor. Validation was performed in vitro using GI tract fluid simulants, while promising results were additionally reported in tests conducted in cow intestines.

Recruitment of biological entities in place of purely artificial sensing components has been considered advantageous, as improved biocompatibility, smoother long-term operation, and reduced health risk due to fewer synthetic materials may be achieved. In addition, enhanced energy efficiency may be enabled, since substantially lower power requirements are typically imposed by biological transduction compared to fully electronic sensing chains. Although living entities in the form of bacteria have been investigated in [10] and [12], MC techniques have not been explicitly integrated to characterize and optimize the communication aspects of the reported systems. Receiver-oriented modeling and analysis have, however, been pursued in a number of bacteria-based MC biosensing studies. In [17], bacteria were engineered to respond to L-rhamnose and to produce GFP as an output. The underlying signal-transduction pathway was explicitly modeled and fitted to experimental measurements. A peak response was reported in less than one hour, therefore demonstrating the feasibility of time-resolved biosensing with engineered cells. In [18], a BioLAN network was reported in which an electronically encoded input was transduced into hydrogen peroxide via electrochemically controlled oxygen reduction. This biomolecular input was then used to trigger bacteria that produced acyl homoserine lactone (AHL), which subsequently diffused toward two additional bacterial populations. In one population, β\beta-galactosidase production was induced and was converted into an electronic readout through electrochemical oxidation and external detection. In the other population a red fluorescent protein was expressed and was used as a transmission confirmation signal. In [9], a bacteria biosensor was analyzed from an MC perspective, where S. mutans was considered to respond to the XIP peptide through an innate signal-transduction pathway, thereby indicating its presence. The luminescent response was simulated and was validated using experimental data. Steady-state and dynamic sensitivity analyses were performed, through which a tradeoff between increased sensitivity and output-response strength was revealed under parameter variation. In [19], Streptococcus pneumoniae was proposed as a sensing entity for quorum-sensing molecules that promote pathogenic colonization and can lead to severe host complications. The bacteria were engineered to respond to AI-2 and CSP molecules. Luminescent reporting was enabled through expression of the firefly luciferase gene.

In contrast to prokaryotic biosensors, a broad range of yeast-based biosensors has been reported in the literature, spanning diverse applications and introducing distinct challenges and opportunities [20]. In [21], a whole-cell yeast biosensor was engineered for melatonin detection in a fluidic medium through modification of native G Protein Coupled Receptors (GPCR). Evaluation was performed in growth medium supplemented with melatonin as well as in wine samples. There, a tradeoff between detection range and signal-to-noise ratio was identified and specificity was demonstrated. A related GPCR-based design was reported in [22], where the yeast mating pathway was engineered to express colorimetric reporters (lycopene, proviolacein, and prodeoxyviolacein). Based on experimental trials under α\alpha-factor stimulation, lycopene was found to provide the most robust response, thereby suggesting a practical alternative to fluorescent and luminescent reporters. Colorimetric reporting through lycopene was also employed in [23]. On-site pathogen surveillance was enabled by introducing a lycopene production pathway into S. cerevisiae and embedding the engineered cells in a dipstick format suitable for direct contact with the liquid under test. Stool-based biosensing was investigated in [24], where doxycycline presence was reported using Saccharomyces boulardii. Specifically, the ADE2 locus was positioned between two Cre recognition sites, such that Cre expression in response to doxycycline resulted in a visible pigmentation switch from white to red. Successful recovery of red-pigmented yeast from stool was demonstrated in mouse trials. Environmental sensing has also been reported; in [25], S. cerevisiae was engineered to produce luciferase in response to the fungicide tebuconazole in water. Yeast sensing platforms targeting metal detection have additionally been demonstrated. In [26], a prototype capable of sensing a range of copper concentrations was presented, where engineered S. cerevisiae cells were interfaced with a screen-printed electrode in an electrolyte medium. Copper-induced disruption of glucose metabolism was translated into a reduction of intracellular current density, which was captured by the electrodes and transmitted via Bluetooth to an external unit for processing and visualization. A field-portable and storable S. cerevisiae biosensor that has progressed to a commercial product was described in [27], where detection of endocrine-disrupting chemicals was enabled (YestroSens).

It is noted that the efforts in [26] and [27] are distinguished by the inclusion of real-time transmission of detection events to external units, which constitutes a key requirement for practical yeast biosensing and aligns with IoBNT-oriented operation. Nevertheless, the yeast biosensors summarized above have largely been developed as laboratory or application-driven prototypes, with the mathematical characterization that can support subsequent analysis and integration in an MC context largely missing. This work overcomes these limitations, by offering a flexible practical implementation platform for MC receiver/sender development, as well the assorted mathematical characterization which can base MC-type analysis, design and implementation. The flexibility stems from the genetic amenability of yeast which can support a wide range of biomarker detection, as well as multiple modes of integration in existing systems. This is crucial in the realization of MC use cases in practical applications.

System Description

The system under consideration is based on the copper detoxification pathway of S. cerevisiae cells [28]. Copper, which is naturally encountered across biological environments, is essential for multiple cellular processes, serving as a co-factor in enzymatic reactions and contributing to the structural stabilization of various proteins. However, excessive copper accumulation and unregulated uptake can impair protein function and promote the generation of ROS, thereby leading to cellular toxicity. Owing to this dual role, mechanisms for maintaining intracellular copper homeostasis have evolved across life forms, including regulated copper transport, chelation, and efflux.

In the extracellular environment, copper ions are predominantly found in the cupric form. Upon reaching the vicinity of S. cerevisiae cells, cupric ions are reduced to cuprous ions by the yeast reductases Fre1 and Fre2. Following internalization, copper ions are distributed via metallochaperones, through which copper is delivered to specific copper-binding proteins within the cell. Under elevated copper conditions, the copper-detoxification pathway culminates in the upregulation of genes that preserve homeostasis, primarily by limiting further uptake and enabling copper sequestration, thereby preventing toxicity. Detoxification is mainly mediated by the metallothionein CUP1, through which copper is sequestered and the pool of free intracellular copper ions is reduced.

The system considered in this work comprises a whole-cell yeast copper receiver, as illustrated in Figure 1a, within a biosensing scenario in which reporting elevated copper levels can serve as an indicative tool. The yeast testbed developed in this work is utilized for direct Cu stimulation of yeast cells, thereby enabling quantification of the detoxification response in wild-type (WT) cells through replacement of the endogenous CUP1 gene with a reporter protein. The theoretical analysis reported herein follows the same framework, where a direct copper stimulus is assumed by the mathematical model and GFP dynamics are computed for validation purposes.

Natural response of a yeast cell to copper and genetically modified response for copper sensing with GFP as the output

Figure 1. Natural response of a yeast cell to copper (figure above) and genetically modified response for copper sensing with GFP as the output (figure below).

Although GFP reporting provides a straightforward means for quantifying gene expression and, consequently, for integrating MC principles into the overall system, fluorescence measurements typically require external excitation light sources. In mobile MC applications where miniaturization is critical, such as Smart Pills, reliance on additional optical components is preferably minimized, while self-sustained operation is desired. To this end, a self-contained bioluminescent reporting scheme is considered here as a replacement for the GFP reporter employed previously. This enables direct coupling of the yeast optical output to the photodetector response.

Bioluminescence in living cells is generated by luciferase systems encoded in the organism’s genome, where expression can be constitutive or induced under specific stimuli. In such systems, light emission is produced through an enzymatic reaction between a luciferase enzyme and one or more chemical substrates. Although Lux systems are naturally present in several bacterial species, this capability is absent in wild-type yeast. This fact motivates genetic engineering when self-sustained optical reporting is sought for MC-integrated biosensing applications.

Among the luciferase systems considered for yeast, integration of the NanoLuc luciferase gene (yNlucyNluc) and of the bacterial Lux cassette has been investigated in [29] and [30], respectively. The approach in [29] is characterized by relatively simple biochemical processes and straightforward integration methods; however, external supplementation of the substrate (furimazine) is required, which would hinder truly standalone operation if such a reporter was adopted in a yeast-derived biosensor.

In contrast, although a higher level of complexity is involved, the Lux system described in [30] is self-sufficient, as the required substrates can be produced internally and recycled. In that work, expression of the luxABCDEluxABCDE cassette from Photorhabdus luminescens was implemented, comprising the enzymes LuxA, LuxB, LuxC, LuxD, and LuxE, while co-expression of the flavin oxidoreductase gene FRP from Vibrio harveyi was additionally employed. Collectively, these enzymes enable the bioluminescent pathway described in [31], relying solely on the cell’s resources; however, several steps of the overall process are still not fully understood. The genes luxAluxA and luxBluxB, which encode the luciferase subunits, are commonly treated as a functional dimer and are referred to as luxABluxAB in the literature.

The theoretical analysis conducted in this paper is extended to include replacement of GFP with LuxAB, where expression is driven by the CUP1CUP1 promoter and is mediated by the transcription factor Ace1. As a result, the copper detoxification pathway is coupled to the bioluminescent reporter, thereby yielding a more complete receiver model for the mathematical characterization of a yeast-derived biosensor. Key components of the proposed biosensing platform include yeast cells suspended in growth medium, coupled with a photodetector and an antenna for wireless transmission of detected events, as illustrated in Figure 2.

Schematic view of the biosensor

Figure 2. Schematic view of the biosensor.

System Model

The system is comprised of three main components: the yeast receiver, the photodetector, and an antenna that enables information exchange with the external environment. Interaction between the receiver and copper ions is governed by a sequence of intracellular processes, which may be classified into (i) uptake, (ii) intracellular transport, (iii) metalation and (iv) transcriptional induction. Extension of the mathematical framework to include a luminescent reporter enables an optical output to be generated. This optical signal is converted into photocurrent by a photodiode, thereby enabling subsequent wireless transmission through an antenna unit.

The primary objective is the formulation of a tractable mathematical representation of the dominant biological processes driving the receiver response, based on mechanistic information available in the literature. For the photodiode stage, established device models are adopted to characterize the output current. In vitro experiments conducted on the developed testbed consider direct Cu stimulation of yeast cells and quantify the resulting fluorescent response. As agreement between experimental measurements and theoretical predictions is achieved to a satisfactory extent, the computational model is positioned as a simulation platform through which this biosensing system may be studied from an MC perspective. In addition, virtual experiments may be enabled in cases where laboratory cost or time constraints are limiting, while mechanistic insight into the underlying biological processes may be provided.

Nevertheless, since such a nuanced model of this pathway has not been documented previously to the best of the authors’ knowledge, the presented analysis is intended to serve as a basis for further refinement. The proposed framework takes the form of coupled differential equations that characterize the time evolution of the involved chemical compounds, including the system output.

Receiver

Model construction is initiated by identifying the dominant interactions among the chemical compounds participating in the copper detoxification pathway, as described in [28]. Copper is assumed to be available in the cuprous form (Cu(I)) following reduction by the plasma-membrane reductases Fre1 or Fre2. Uptake is facilitated by two copper transporters (Ctr) located in the plasma membrane, namely Ctr1 and Ctr3. These form homo-trimeric, high affinity Cu-selective pores across the membrane. After internalization, Cu(I) is bound by glutathione (GSH) in the cytosol. Acting as a native chelator, GSH is then involved in copper distribution toward metallochaperones, including Atx1, Ccs1, and Cox17, through which copper is delivered to specific intracellular targets.

Ccs1 is responsible for escorting copper to Sod1 in the cytoplasm, which is associated with detoxification and redox regulation. Through Atx1, copper is delivered to Ccc2, by which copper is loaded onto additional cuproproteins (CuPrs) within the Golgi apparatus. Additional transport routes mediated by GSH include copper delivery to mitochondrial metallochaperones (e.g., Cox17, Cox11, and Sco1/2) for subsequent loading to cytochrome c oxidase (CcO), copper sequestration into vacuoles under elevated conditions, and handoff to scavenger proteins that contribute to buffering of free copper ions.

To preserve intracellular homeostasis and prevent toxicity, transcriptional regulators are activated and mediate expression of genes involved in copper uptake and detoxification. In the system considered here, the output is provided by the expression of the CUP1 gene, which encodes the metallothionein protein Cup1. Its induction is mediated by the transcriptional regulator Ace1, following copper-dependent activation in the nucleus. Copper uptake-related genes, including CTR1 as well as the reductases FRE1/FRE2, are induced by the transcriptional regulator Mac1 in an analogous regulatory role. A detailed description of the copper detoxification pathway is provided in [28].

The exact mechanism through which Cu ions activate Ace1 and Mac1 via GSH remains incompletely understood.

However, in [28] it is noted that Cu(I)-GSH complexes have been detected in the nucleus, which suggests that a fraction of intracellular copper may be transferred directly from GSH to Ace1. Owing to the limited mechanistic detail currently available, direct loading of copper from GSH to Ace1 is assumed in this work. Cu transfer to Ace1 leads to Ace1 metallation and subsequent CUP1 induction.

Transport of copper to the vacuole and mitochondria is also not fully characterized and is therefore excluded from the present analysis. Moreover, transcriptional regulation mediated by Mac1 is omitted, while Ctr1/3 production is approximated using zero-order kinetics. As illustrated in Figure 3, the reaction set considered in this work includes Ctr1/Ctr3, GSH, Atx1, Ccs1, Sod1, Ccc2, Ace1, and CUP1.

Schematic of reaction processes within the copper detoxification pathway

Figure 3. Schematic of reaction processes within the copper detoxification pathway.

Chemical reactions within the copper detoxification pathway considered here are grouped into five main reaction types: uptake, trimerization, transport (handoff), metallation, and transcriptional regulation. The chemical equations summarized in Table I, together with the reaction diagram in Figure 3, provide the foundation upon which the reported mathematical models are derived. In the subsequent analysis, concentration species are denoted in brackets for readability.

Chemical reactionChemical Equation
Ctr1/3 trimerizationCtr1/3+Ctr1/3+Ctr1/3⇌kbindkunbindCtr1/3trimCtr1/3 + Ctr1/3 + Ctr1/3 \xrightleftharpoons[k_{\mathrm{bind}}]{k_{\mathrm{unbind}}} Ctr1/3_{\mathrm{trim}}
Ctr1/3Ctr1/3 trimer loaded with CuCtr1/3trim+Cu→kuptCtr1/3trimCuCtr1/3_{\mathrm{trim}} + Cu \xrightarrow{k_{\mathrm{upt}}} Ctr1/3_{\mathrm{trim}}Cu
Cu handoff from Ctr1/3 to GSHCtr1/3trimCu+GSH→ktranCtr1/3trimCu−GSH→ktranCtr1/3trim+GSHCu\begin{aligned} Ctr1/3_{\mathrm{trim}}Cu + GSH &\xrightarrow{k_{\mathrm{tran}}} Ctr1/3_{\mathrm{trim}}Cu-GSH \\ &\xrightarrow{k_{\mathrm{tran}}} Ctr1/3_{\mathrm{trim}} + GSHCu \end{aligned}
Cu handoff from GSH to Ccs1GSHCu+Ccs1→ktranGSHCu−Ccs1→ktranGSH+Ccs1Cu\begin{aligned} GSHCu + Ccs1 &\xrightarrow{k_{\mathrm{tran}}} GSHCu-Ccs1 \\ &\xrightarrow{k_{\mathrm{tran}}} GSH + Ccs1Cu \end{aligned}
Cu handoff from Ccs1 to Sod1Ccs1Cu+Sod1→ktranCcs1−Sod1Cu→ktranCcs1+Sod1Cu\begin{aligned} Ccs1Cu + Sod1 &\xrightarrow{k_{\mathrm{tran}}} Ccs1-Sod1Cu \\ &\xrightarrow{k_{\mathrm{tran}}} Ccs1 + Sod1Cu \end{aligned}
Cu handoff from GSH to Atx1GSHCu+Atx1→ktranGSHCu−Atx1→ktranGSH+Atx1Cu\begin{aligned} GSHCu + Atx1 &\xrightarrow{k_{\mathrm{tran}}} GSHCu-Atx1 \\ &\xrightarrow{k_{\mathrm{tran}}} GSH + Atx1Cu \end{aligned}
Cu handoff from Atx1 to Ccc2Atx1Cu+Ccc2→ktranAtx1Cu−Ccc2→ktranAtx1+Ccc2Cu\begin{aligned} Atx1Cu + Ccc2 &\xrightarrow{k_{\mathrm{tran}}} Atx1Cu-Ccc2 \\ &\xrightarrow{k_{\mathrm{tran}}} Atx1 + Ccc2Cu \end{aligned}
Cu handoff from GSH to Ace1GSHCu+Ace1→ktranGSHCu−Ace1→kmetalAce1act\begin{aligned} GSHCu + Ace1 &\xrightarrow{k_{\mathrm{tran}}} GSHCu-Ace1 \\ &\xrightarrow{k_{\mathrm{metal}}} Ace1_{\mathrm{act}} \end{aligned}
CUP1CUP1 transcriptionAce1Cu+D→ktranscrCUP1Ace1Cu + D \xrightarrow{k_{\mathrm{transcr}}} CUP1

Table I. Chemical Equations governing the various chemical reactions within the copper detoxification pathway.

Copper uptake is characterized by saturable kinetics [32], approaching a plateau at elevated extracellular copper concentrations. Accordingly, the uptake rate is modeled using a Michaelis–Menten-type expression:

vuptake∝Vmax⁡[S]KD+[S](1)v_{\mathrm{uptake}} \propto\frac{V_{\max}[S]}{K_D+[S]} \tag*{(1)}

The trimerization of the copper transporters is modeled as a third order reaction of the form:

Ctr1/3+Ctr1/3+Ctr1/3⇌kb1kd1Ctr1/3trimer(2)\mathrm{Ctr1/3}+\mathrm{Ctr1/3}+\mathrm{Ctr1/3}\underset{k_{d1}}{\overset{k_{b1}}{\rightleftharpoons}}\mathrm{Ctr1/3}_{\mathrm{trimer}} \tag*{(2)}

The binding rate constant of the Ctr molecules is denoted by kb1k_{b1}. A reverse reaction is also considered, through which trimer dissociation occurs with a rate constant kd1k_{d1}, thereby reforming the initial reactants. Copper binding has been reported to induce conformational changes that disrupt Ctr-mediated transport [33]. In particular, copper-dependent destabilization has been observed for Ctr1 [34]. Since the corresponding dose-response kinetics have not been fully characterized, this additional loss in transportation activity is represented empirically by the Hill-type degradation rate kCu−degk_{\mathrm{Cu-deg}} defined in Appendix A-C. Because Ctr1 and Ctr3 are represented as a common species, applying this Ctr1-derived behavior to both transporters constitutes a simplifying modeling assumption.

From a Systems Biology perspective, Cu-handoff reactions are treated as distinct from simple binding processes, since different reaction mechanics are implied by their chemical formulation. In particular, the Cu transport (handoff) process may be represented as a short chain of reactions in which the Cu carrier (denoted ACuACu) initially binds to the Cu recipient (denoted BB), followed by dissociation of the transient complex and formation of AA and BCuBCu. Such transport is suggested to be facilitated by protein–protein interactions [35, 36]. An example is provided in Eq. (3):

ACu+B→ktranAACu−B→ktranBA+BCu(3)ACu + B \xrightarrow{k_{\mathrm{tranA}}} ACu-B \xrightarrow{k_{\mathrm{tranB}}} A + BCu \tag*{(3)}

In this work, the Cu-handoff reactions are treated as irreversible, such that copper is assumed to be transferred unidirectionally from one chaperone to the next. This approximation is deemed reasonable, since in [36] copper binding rates to the involved chaperones were reported to be substantially higher than the corresponding unbinding rates, thereby suggesting predominantly forward flux. Moreover, in [35] it was observed that formation of the intermediate complex during handoff occurs rapidly, while formation of the Cu–recipient complex constitutes the rate-limiting step. Given the limited mechanistic characterization available for yeast, the same assumption is adopted here, and the intermediate complex is neglected to maintain reasonable model complexity without sacrificing biological relevance.

Mechanistic information regarding metalation of the transcription factor (TF) Ace1 remains limited. Nevertheless, in [37] Ace1 binding to CUP1CUP1 promoters was reported to exhibit switch-like behavior and saturation at elevated copper concentrations, resulting in a sigmoidal response. Accordingly, the metalation rate is modeled using a Hill-type expression of the form:

fmet∝[Cudonor]nKD,Ace1n+[Cudonor]n(4)f_{\mathrm{met}} \propto\frac{[Cu_{\mathrm{donor}}]^{n}}{K_{D,Ace1}^{n} + [Cu_{\mathrm{donor}}]^{n}} \tag*{(4)}

Where KD,Ace1K_{D,Ace1} is the half-maximum constant for the metalation of Ace1 by Cu ions.

Transcriptional regulation is initiated through binding of the TF Ace1 to the CUP1CUP1 promoter, thereby enabling gene induction. Derivation of the transcription rate is based on a quasi-steady-state (QSS) assumption for binding and unbinding of Ace1 molecules to the promoter. Since Ace1 binds to the promoter in monomeric form [38], a Hill-type expression with n=1n=1 would be obtained for the fractional transcription level which implies the probability that the promoter is occupied by the TF. However, Ace1 is known to bind to four Metal Response Elements (MREs) within the CUP1CUP1 promoter. Therefore, following an approach similar to [39], the activation function based on the transcription probability is modeled as:

fact∝1∑n=14(KD,mRNA/[Ace1])n+1(5)f_{\mathrm{act}} \propto\frac{1}{\sum_{n=1}^{4}(K_{D,mRNA}/[Ace1])^{n}+1} \tag*{(5)}

Where KD,mRNAK_{D,mRNA} is the TF-promoter dissociation factor. The derivation of Eq. (5) is given in Appendix A-A in more detail.

Lux Model

The LuxAB enzyme catalyzes the light-emitting reaction by utilizing reduced flavin mononucleotide (FMNH2_2), molecular oxygen (O2_2), and a long-chain aldehyde (RCHO) to produce oxidized FMN, a fatty acid (RCOOH), water, and light, as shown below:

FMNH2+RCHO+O2→LuxABFMN+RCOOH+H2O+Light(6)FMNH_{2} + RCHO + O_{2} \xrightarrow{\mathrm{LuxAB}} FMN + RCOOH + H_{2}O + \mathrm{Light} \tag*{(6)}

The substrates required for LuxAB catalysis are shown on the left-hand side of Eq. (6). Oxidized Flavin Mononucleotide (FMN) participates in the reaction catalyzed by the Flavin Reductase P (FRP) in the presence of the cofactors Nicotinamide adenine dinucleotide (NADH) or Nicotinamide adenine dinucleotide phosphate (NADPH), through which FMNH2_2 is produced. In parallel, the LuxCDE multi-enzyme reductase complex converts fatty acyl-ACP into RCHO required by LuxAB. In this process, LuxD acts as a transferase that generates fatty acids from either fatty acyl-ACP or acyl-CoA. The resulting fatty acid is ATP-activated to form acyl-AMP, which is subsequently attached to LuxE to form an acyl intermediate. It is believed that this intermediate is channeled to LuxC, where an additional acyl intermediate is formed and is reduced by NADPH to yield the aldehyde.

The light-emission pathway proceeds through multiple intermediate stages, beginning with binding of FMNH2_2 and oxygen to LuxAB and formation of the Lux:FMNHOOH complex (C4a-hydroperoxyflavin). As discussed in [31], deprotonation of this C4a adduct leads to formation of C4a-peroxyflavin, which can attack the aldehyde and yield the Lux:FMNHOH:RCOOH complex, also referred to as excited C4a-hydroxyflavin. This intermediate is commonly treated as the light-emitting species and relaxes to the ground state through photon emission at approximately 490 nm. The resulting non-excited hydroxyflavin undergoes dehydration, after which fatty acid and FMN are released.

Recycling of the two key substrates required for LuxAB catalysis enables self-sustained bioluminescent reporting without external substrate supplementation. This is suitable for MC-integrated biosensing applications targeting health monitoring through Smart Pill technologies. The light pathway incorporated into the mathematical model is illustrated schematically in Figure 4.

Light pathway

Figure 4. Light pathway.

The enzymatic reactions described above and depicted in Figure 4 involve more than one substrate, and therefore depart from classical kinetic formulations such as Michaelis–Menten models. For the derivation of rate equations that are suitable for ODE models describing each enzymatic reaction, the King–Altman schematic method is adopted [40]. The King–Altman approach is applicable to mechanisms in which an arbitrary number of substrates bind to an enzyme in various arrangements, while a steady-state assumption is applied to the enzyme forms that are created. Multi-enzyme reactions can also be handled using this method.

Starting from basic principles, a single enzyme–multiple substrate reaction is considered with nn substrates available. The nn different enzyme forms that are created are denoted as E1,E2,...EnE_1, E_2, ...E_n. The rate of change of a species EiE_i can be given by

d[Ei]dt=k1i[E1]+k2i[E2]+...+kni−∑kij[Ei].\frac{d[E_i]}{dt}=k_{1i}[E_1]+k_{2i}[E_2]+...+k_{ni}-\sum k_{ij}[E_i].

Here, the rate constant for Ei→EjE_i \rightarrow E_j is denoted by kijk_{ij}, and the rate constant for Ej→EiE_j \rightarrow E_i is denoted by kjik_{ji}. Moreover, the total enzyme concentration is conserved:

[E1]+[E2]+...+[En]=[Etot](7)[E_1]+[E_2]+...+[E_n]=[E_{tot}] \tag*{(7)}

We solve for the mm-th enzyme form concentration in a system of equations where the mm-th equation is replaced with Eq. (7):

−∑k1j[E1]+k21[E2]+...+km1[Em]+...+kn1[En]=0-\sum k_{1j}[E_1]+k_{21}[E_2]+...+k_{m1}[E_m]+...+k_{n1}[E_n]=0
k12[E1]−∑k2j[E2]+...+km2[Em]+...+kn2[En]=0k_{12}[E_1]-\sum k_{2j}[E_2]+...+k_{m2}[E_m]+...+k_{n2}[E_n]=0
[E1]+[E2]+...+[Em]+...+[En]=[Etot]k1n[E1]+k2n[En]+...+kmn[Em]+...−∑knj[En]=0(8)\begin{aligned} [E_1]+[E_2]+...+[E_m]+...+[E_n]&=[E_{tot}]\\ k_{1n}[E_1]+k_{2n}[E_n]+...+k_{mn}[E_m]+...-\sum k_{nj}[E_n]&=0 \tag*{(8)} \end{aligned}

To solve for [Em][E_m], Cramer’s rule is applied, resulting in the following expression:

[Em]=Nm[Etot]D(9)[E_m]=\frac{N_m[E_{tot}]}{D} \tag*{(9)}

Where the numerator NmN_m is given by the following determinant:

Nm=∣−∑jk1jk21⋯kn1k12−∑jk2j⋯kn2⋮⋮⋱⋮k1nk2n⋯−∑jknj∣(10)N_m= \left| \begin{array}{ccccc} -\sum_j k_{1j} & k_{21} & \cdots& k_{n1}\\ k_{12} & -\sum_j k_{2j} & \cdots& k_{n2}\\ \vdots& \vdots& \ddots& \vdots\\ k_{1n} & k_{2n} & \cdots& -\sum_j k_{nj} \end{array} \right| \tag*{(10)}

And the denominator DD:

D=∣k12−∑jk2j⋯km2⋯kn2⋮⋮⋮⋮11⋯1⋯1⋮⋮⋮⋮k1nk2n⋯kmn⋯−∑jknj∣(11)D= \left| \begin{array}{ccccc} k_{12} & -\sum_j k_{2j} & \cdots& k_{m2} & \cdots& k_{n2}\\ \vdots& \vdots& & \vdots& & \vdots\\ 1 & 1 & \cdots& 1 & \cdots& 1\\ \vdots& \vdots& & \vdots& & \vdots\\ k_{1n} & k_{2n} & \cdots& k_{mn} & \cdots& -\sum_j k_{nj} \end{array} \right| \tag*{(11)}

Expansion of the determinant in Eq. (10) reveals some key properties which form the basis for the King–Altman schematic method. Every product in the expansion of $N_m contains (n−1)(n-1) constants kijk_{ij}, essentially forming reaction “paths” among the different enzyme forms. The index mm cannot occur as a first index in any kijk_{ij} within a product. Every other index can occur once as first index in every product, while cycles of indices are canceled within the expansion. Ultimately, every product that is left in NmN_m has the same sign, where mm occurs at least once as a second index (every ”reaction path” that survive in NmN_m contain EmE_m as a ”sink”).

It is assumed in the general derivation that all possible enzyme forms react directly with each other; however, such full connectivity is typically not realized in practice. Non-feasible reactions are eliminated by setting the corresponding rate constants to zero, whereby the associated paths are removed from NmN_m. The denominator DD is obtained as the sum of all numerators, i.e., D=N1+N2+...Nm+...NnD=N_1+N_2+...N_m+...N_n. Further theoretical details on the principles summarized above are provided in [40].

Application of the principles outlined above to enzymatic reactions such as those depicted in Figure 4 yields schematic representations that are subsequently used for derivation of rate equations to be included in the receiver ODE model. Enzymatic reaction schematics are patterns that connect the different enzyme forms according to the binding and release events dictated by the underlying physical mechanism. In multi-substrate reactions, different substrate-binding schemes may arise, primarily depending on the order of binding events [40], with common mechanisms including compulsory-order and random-order binding. Based on the available information, the enzymatic reactions comprising the light pathway are assumed here to follow compulsory-order kinetics.

The reaction between FRP, FMN, and NADPH is modeled as ternary-complex formation involving two substrate-binding steps, namely EFRP+F⇌k−1k1EFRPFE_{FRP}+F \mathrel{\mathop{\rightleftharpoons}^{k_1}_{k_{-1}}} E_{FRP}F and EFRPF+N⇌k−2k2EFRPFNE_{FRP}F+N \mathrel{\mathop{\rightleftharpoons}^{k_2}_{k_{-2}}} E_{FRP}FN, followed by two product-release steps, EPQ⇌k−3k3EQ+PEPQ \mathrel{\mathop{\rightleftharpoons}^{k_3}_{k_{-3}}} EQ+P and EQ⇌k−4k4E+QEQ \mathrel{\mathop{\rightleftharpoons}^{k_4}_{k_{-4}}} E+Q. Here, FMN and NADPH are denoted by F and N, respectively. The intermediate EFRPFNE_{FRP}FN is transformed to EPQEPQ through EFRPFN⇌EPQE_{FRP}FN \rightleftharpoons EPQ, where PP and QQ denote the two products, namely FMNH2_2 and NADP+^+, which are released. Due to limited information regarding intermediate states between EFRPFNE_{FRP}FN and EPQEPQ, these two complexes are treated as a single species and are collapsed into a single complex EXEX. As part of the King–Altman method, reactions are expressed using pseudo first-order kinetics, as illustrated in Figure 5 (e.g., the bimolecular rate constant k1k_1 is replaced by the pseudo first-order rate constant k1Fk_1F). The next step of the schematic method is to derive all possible schemes based on the master pattern. The velocity of the first product release (P), namely FMNH2_2, is in this case:

Master pattern for sequential binding of FMN and NADPH to FRP

Figure 5. Master pattern for sequential binding of FMN and NADPH to FRP.

d[P]dt∝vFMNH2=k3[EX]−k−3p[EQ]=k3NEX[EtotFRP]D−k−3pNEQ[EtotFRP]D(12)\frac{d[P]}{dt} \propto v_{FMNH_2}=k_3[EX]-k_{-3p}[EQ]=k_3\frac{N_{EX}[E_{tot}^{FRP}]}{D}-k_{-3p}\frac{N_{EQ}[E_{tot}^{FRP}]}{D} \tag*{(12)}

Product inhibition is not considered in this work; therefore, product release is treated as non-reversible (i.e., k−3k_{-3} and k−4k_{-4} are set to zero), and the corresponding rate derivations pertain to initial-rate conditions. In this case, it is easy to prove that dPdt=dQdt\frac{dP}{dt}=\frac{dQ}{dt} (equal product release rate), and thus vFMNH2=vNADPHv_{FMNH_2}=v_{NADPH}, using the steady state property for enzyme forms in a King-Altman setting. Based on the patterns depicted in Figure 6, together with the principles outlined above, the legitimate King-Altman patterns for this reaction are obtained as shown in Figure 7.

All possible King-Altman Patterns

Figure 6. All possible King-Altman Patterns.

Figure 7. Legitimate King-Altman patterns for the flavin reduction reaction.

The only pattern that does not contain product inhibition, is the one depicted in Figure 7b. Therefore, it contributes to NEXN_{EX} which is given by the following expression:

NEX=k4k1[F] k2[N](13)N_{EX}=k_4k_1[F]\,k_2[N] \tag*{(13)}

By deriving the legitimate King–Altman patterns for the remaining enzyme forms involved in this reaction, namely EFRPE_{FRP}, EFRPFE_{FRP}F, EFRPFNE_{FRP}FN, and EQEQ, the denominator DFRPD_{FRP} can be constructed and the full expression in Eq. (12) can be obtained. Subsequently, Cleland’s notation is adopted to group rate constants, as described in [40], [41], thereby transforming the result into a more convenient form:

vFMNH2=VFRP+[F][N]DFRP(14)v_{FMNH_2}=\frac{V_{FRP}^{+}[F][N]}{D_{FRP}} \tag*{(14)}

And,

DFRP=KD,FKm,N+Km,N[F]+Km,F[N]+[F][N](15)D_{FRP}=K_{D,F}K_{m,N}+K_{m,N}[F]+K_{m,F}[N]+[F][N] \tag*{(15)}

Where the forward catalytic rate is defined as VFRP+=kcat[FRP]V^{+}_{FRP}=k_{cat}[FRP], the Michaelis constants are denoted by Km,FK_{m,F} and Km,NK_{m,N} for FMN and NADPH respectively, and KD,FK_{D,F} represents the dissociation constant of FMN. The derivation of Eq. (14) is provided in Appendix A-B, together with the rest of the enzymatic rates presented here. The conversion of fatty acyl-ACP to aldehyde by the reductase LuxCDE is treated in a similar fashion. It is further assumed that fatty acid is recycled entirely as one of the products of the reaction between LuxAB, FMNH2_2, RCHO, and O2_2; therefore, the aldehyde supply chain is assumed to be realized solely through LuxE and LuxC. Hence, the acyl-transferase LuxD is omitted by the mathematical model. The velocity of the reaction between RCOOH, ATP and LuxE for the production of acyl-LuxE (LuxE-RCO), can be given in a similar fashion as follows:

vLuxE−RCO=VLuxE+[RCOOH][ATP]DLuxE(16)v_{LuxE-RCO}=\frac{V^{+}_{LuxE}[RCOOH][ATP]}{D_{LuxE}} \tag*{(16)}

Where

DLuxE=KD,RCOOHKm,ATP+Km,ATP[RCOOH]+Km,RCOOH[ATP]+[RCOOH][ATP](17)D_{LuxE}=K_{D,RCOOH}K_{m,ATP}+K_{m,ATP}[RCOOH]+K_{m,RCOOH}[ATP]+[RCOOH][ATP] \tag*{(17)}

The velocity of the reaction between LuxC, LuxE-RCO and NADPH for the production of aldehyde (A), is given by:

vA=VLuxC+[LuxE−RCO][N]DLuxC(18)v_A=\frac{V^{+}_{LuxC}[LuxE-RCO][N]}{D_{LuxC}} \tag*{(18)}

Where

DLuxC=KD,LuxE−RCOKm,N2+Km,N2[LuxE−RCO]+Km,LuxE−RCO[N]+[LuxE−RCO][N](19)D_{LuxC}=K_{D,LuxE-RCO}K_{m,N2}+K_{m,N2}[LuxE-RCO]+K_{m,LuxE-RCO}[N]+[LuxE-RCO][N] \tag*{(19)}
Km,N2[LuxE−RCO]+Km,LuxE−RCO[N]+[LuxE−RCO][N](20)K_{m,N2}[LuxE-RCO]+K_{m,LuxE-RCO}[N]+[LuxE-RCO][N] \tag*{(20)}

Catalysis by LuxAB is described by the processes shown in Figure 4. Beyond substrate-binding and product-release events, multiple intermediate stages are involved, which play a critical role in both light emission and substrate recycling [31]. Owing to the distinct nature of this light pathway, the corresponding King–Altman schematic becomes more complex and requires special treatment.

As shown in Figure 8, the first enzyme–substrate complex is denoted by ELuxFHE_{Lux}FH, where FH represents the substrate FMNH2_2. Upon oxygen binding, a ternary complex is formed and the intermediate EXEX is generated. Here, O2_2 is assumed to be constant and is therefore absorbed into the pseudo first-order constant rate KOK_O. The intermediate EXEX represents a critical state of the system, since it corresponds to the enzyme form that undergoes deprotonation prior to binding of the aldehyde substrate (denoted by AA). Since protonation and deprotonation typically occur on fast timescales, thermodynamic equilibrium between the protonated and deprotonated state of EXEX (see Figure 4) is commonly assumed [40]. As such, the aldehyde-binding rate constant kbAk_{bA} is replaced by:

King-Altman pattern for the catalysis of LuxAB and light emission

Figure 8. King-Altman pattern for the catalysis of LuxAB and light emission.

k2=kbAKEKE+[h](21)k_2=k_{bA}\frac{K_E}{K_E+[h]} \tag*{(21)}

where the equilibrium constant is denoted by KEK_E and the cytosolic hydrogen concentration by [h]. The complex denoted by EYEY corresponds to the aldehyde-bound intermediate (flavin C4a-hydroperoxymicetal), which is subsequently transformed into EZEZ. The complex EZEZ represents the excited C4a-hydroxyflavin state and is treated as the light-emitting species. The intermediate ELEL corresponds to the dehydrated/de-excited complex following photon emission. Here, H2_2O release is assumed to be constant and its effect is absorbed into k4k_4, leading to EKEK, where fatty-acid release has occurred. Subsequent release of FMN returns the system to the free enzyme form ELuxE_{Lux}. Photon emission is given by:

[Ph]=kL[EZ](22)[Ph] = k_L [EZ] \tag*{(22)}

Where kLk_L is the photon emission rate constant in min−1min^{-1}, [Ph][Ph] is computed in μmol\mu mol photons L−1min−1L^{-1} min^{-1}, and [EZ][EZ] is computed in μmolL−1\mu mol L^{-1}. The derivation of [EZ][EZ] is provided in Appendix A-B. The number of photons per minute can be given by the following formula:

nL=10−6NA[Ph]Vcell(23)n_L = 10^{-6}N_A[Ph]V_{cell} \tag*{(23)}

where NAN_A is the Avogadro number and Vcell=1.13⋅10−13V_{cell} = 1.13 \cdot10^{-13} liters denotes the yeast cell volume considered here [42]. The full ODE model, together with the parameter values and initial concentrations used in simulations, is provided in Appendix A-C, where all concentration values are expressed in μM\mu M and time is depicted in minutes.

Due to the novelty of the proposed model and the scarcity of yeast-specific quantitative data, parameter values were selected from the yeast literature whenever possible. When relevant yeast data were unavailable, parameter values reported for bacterial processes were adopted. When no information was available, reasonable estimates were assumed based on acceptable value ranges for the corresponding processes. The initial concentration values that are not mentioned in Appendix A-C, are set to zero.

Photodetector

The photocurrent output of a photodetector coupled to the receiver model described in Section IV-A is derived using the photon emission rate from the Lux reporter, as defined in Section IV-A1. In this work, a CMOS p+/n-well/psub photodiode akin to the one used in [43] is considered as the photodetector element, consistent with sensing circuitry employed in prominent biosensors such as Smart Pills [10]. The corresponding photocurrent response can be expressed as follows:

Ipd=RpdPL(24)I_{pd} = R_{pd} P_L \tag*{(24)}

where RpdR_{pd} denotes the photodiode responsivity in Amperes per Watt (A/WA/W) and PLP_L denotes the optical power generated by yeast bioluminescence in WW. The optical power is obtained by combining the photon number (per second) with the Planck constant hph_p, the speed of light CC and the emission wavelength λ\lambda (in meters) as follows:

PL=nL60hpCλ(25)P_L = \frac{n_L}{60}\frac{h_p C}{\lambda} \tag*{(25)}

For the Lux-mediated light emission, λ=490\lambda= 490 nm.

Experimental Setup

Reliability of the mathematical tools presented in Section IV depends on the availability of an experimental platform through which theoretical findings can be validated. In the context of advancing MC toward biosensing applications, such a setup provides a foundation for development of yeast-based MC receivers that are suitable for integration within biosensors. Moreover, the regulatory networks governing metal homeostasis in S. cerevisiae can be leveraged as a basis for a broad class of engineered yeast biosensors. Accordingly, the testbed presented in this work is informed by the extensively studied yeast response to copper stress, with particular emphasis placed on sequestration and detoxification mechanisms under heavy-metal accumulation [44].

Testbed Description

As described in Section III, WT yeast cells feature an intrinsic biological property that tightly controls intracellular copper concentrations, thereby maintaining metal homeostasis under physiological conditions. Copper, despite being an essential micronutrient for the yeast physiology can become toxic at excessive concentrations [45], and triggers inherent detoxification mechanisms which restore cellular homeostasis [46]. Translocation of the copper-responsive transcriptional regulator Ace1 controls the induction of the CUP1 gene which encodes the metal-binding Cup1 metallothionein to discharge the cell from cytotoxic buildup [46], [47].

The experimental platform reported here leverages the well-established pathway of copper-inducible CUP1 gene activation, to engineer a fluorescent yeast ”receiver” capable of sensing the presence of accumulated copper. This has been implemented by replacing the CUP1 gene with the GFP gene, enabling the expression of a green fluorescent protein under the control of the copper-responsive promoter pCUP1 (Section III, Figure 1b). To this end, the engineered receiver is predicted to emit a green signal when exposed to high copper concentrations, providing a measurable output for MC. The overall system is foreseen to deepen our understanding on the dissemination of information through molecular signaling.

Materials and Methods

Yeast strains and cultivation conditions

1) Yeast strains and cultivation conditions: All yeast strains in this study were originated from S. cerevisiae BY4741 (’wild type’, genotype MATa) and transformed with PCR products by the lithium acetate method [48]. The cup1Δ::sfGFP reporter strain was constructed by genetic replacement of the CUP1-2 gene via homologous recombination of a PCR product containing the super-folder GFP (sfGFP) fused to a CLN2 PEST sequence and the natMX cassette for selection. The CUP1-1 gene locus remained intact in the yeast genome. The constructed strain was verified through DNA sequencing. The PCR product was amplified from the Gal1pORF-sfGFPdeg-natMX strain [49]. All oligonucleotides used for PCR-amplifications are listed in Table II. Yeast cells were cultivated to saturation overnight at 30 °C in selective YPAD medium (0.12 g/L adenine hemisulfate salt, 10g/L yeast extract, 20 g/L peptone and 20 g/L glucose) supplemented with appropriate antibiotics. The next day, cells were diluted to OD600=0.2\mathrm{OD}_{600} = 0.2 and grown for a few hours until OD600\mathrm{OD}_{600} reached approximately 0.6, prior to experimental induction.

A. For PCR amplification
Yeast StrainPrimer pair
cup1Δcup1\Delta::sfGFPF: 5’- AAGAAAAACAAACTGTACAATCAATCAATCAATCATCACATAAAatgtccaagggtgaagagct -3’<br>R: 5’- TTAAAAACAGCAAATAGTTAGATGAATATATTAAAGACTATTCGTTcagtatagcgaccagcattc -3’
B. For quantitative PCR
GenePrimer pair
sfGFPF: 5’- CCATTTTGGTAGAACTGGAC -3’<br>R: 5’- CATATGGTCTGGGTATCTTG -3’
β\beta-actinF: 5’- CATCTTCCATGAAGGTCAAG -3’<br>R: 5’- CTTGTGGTGAACGATAGATG -3’

Table II. Oligonucleotide primers used in the study.

Copper induction and cell harvesting

Yeast cells at the logarithmically growing phase (OD600_{600} ∼\sim 0.6) were stimulated with either 1, 10, 100 or 500 μ\muM of copper sulfate (Sigma-Aldrich) and samples of cell suspension were harvested at specific time points following initial stimulation. For the RNA assay, aliquots of ∼\sim 3 ⋅\cdot 10^{6}$ cells were harvested per each time point, washed with sterile water and stored at -80 °C before proceeding with RNA isolation. For fluorometric measurements, stimulated cells were adjusted to the exponential phase of growth, washed once with sterile water and aliquots of ∼\sim 3 ⋅\cdot 10^{6}$ cells were transferred to a 96-well microplate (Thermo Scientific) to monitor fluorescence yield. For the specificity assay, yeast cells were stimulated with 100 μ\muM zinc chloride (Sigma-Aldrich). Green emission was determined in arbitrary units (AU) using the Infinite F200 fluorescence microplate reader (Tecan Trading Ltd, Switzerland) with excitation, 485/20 nm; emission, 510/20 nm; manual gain, 100. The presented data from the fluorescence assays were obtained from at least three independent experiments.

RNA isolation, reverse transcription and quantitative PCR

Total RNA was isolated from frozen samples using the hot phenol method [50] and cDNA synthesis was performed with the PrimeScript RT Reagent Kit (Perfect Real Time, TaKaRa Bio). Quantitative real-time PCR was performed on a CFX96 thermal cycler (BioRad) using the KAPA SYBR FAST qPCR Master Mix (KAPA Biosystems). Actb (β\beta)-actin was used as a housekeeping gene and all oligonucleotide primers used in the assay are listed in Table II. Relative changes in gene expression were analyzed using the 2−ΔΔC[T]2^{-\Delta\Delta C[T]} algorithm. All samples and standards were run in triplicates and the presented results were obtained from at least two independent experiments.

Results

In this section, results obtained from both experimental measurements and computational analysis are presented. Experimental trials incorporate GFP-based quantification at the transcript (mRNA) and protein levels, where copper is externally supplied to the growth medium. Joint observation of these two response modalities provides insight into the dynamics of the yeast receiver. Transcript induction is being characterized across multiple copper concentrations in order to observe any dose-dependent differences in promoter activity, while fluorometric measurements are limited to a single input concentration. Protein-level dynamics are inherently slower due to additional post-transcriptional mechanisms involved and therefore, can potentially misrepresent nuances from diverse input conditions.

Theoretical results are reported for single copper-stimulation scenarios and are presented for both the GFP and Lux reporters. GFP-based computations are included primarily for validation, whereas Lux-based computations enable coupling of the receiver model to the photodetector response, thereby supporting biosensing scenarios such as event reporting in Smart Pill applications. A direct Cu input is assumed in the current model, and molecular propagation effects are therefore neglected; incorporation of propagation phenomena is left for future work.

GFP abundance

GFP fluorescence was monitored in cup1Δcup1\Delta::sfGFP cells at regular intervals (as indicated on the graph) for a total duration of 6h after initial stimulation with 100 μ\muM of copper sulfate. As shown in Figure 9, a significant increase of green signal was detectable on a fluorescence reader as early as 15 minutes. Maximum fluorescence was reached at 300 minutes, followed by a gradual decrease in signal intensity, indicating the onset of protein turnover. The translation delay time together with the relatively slow decay of GFP due its stability, results in a slow overall response.

GFP fluorescence over time of induction

Figure 9. Expression of enhanced green fluorescent protein in cup1Δcup1\Delta::sfGFP cells following induction with 100 μ\muM of copper sulfate. Green fluorescence intensity was monitored in copper-induced cells for 360 minutes following initial induction. Fluorescence values at each time point were normalized to those not exposed to copper sulfate. Values from three independent experiments are presented as mean ±\pm SEM. AU, arbitrary units.

RNA levels of GFP

Although the GFP reporter provides a measurable output of the yeast receiver in real time, the protein-level dynamics are comparatively slow due to translation and maturation delays. The demonstrated stability can provide a robust readout and reduce susceptibility to transient fluctuations. However, the overall slow nature of protein abundance reporting can limit the achievable data rate in MC scenarios where repeated stimulation of the receiver is required.

To better elaborate on the promoter activity during induction, GFP mRNA levels were additionally measured. The single concentration used for protein abundance measurements is extended to a set of values, aiming to characterize the promoter’s activity across different input conditions. Therefore, the investigation focuses on mRNA responses for copper concentrations of 1, 10, 100 and 500 μM\mu\mathrm{M}. The distinct dynamics shown in Figure 10 indicate that substantially faster responses can be achieved at the transcript level compared to protein abundance.

GFP mRNA levels

Figure 10. GFP transcript levels were determined by quantitative PCR following the induction of cup1Δ::sfGFP with 1, 10, 100 and 500 μM\mu\mathrm{M} of copper sulfate as indicated in the charts. RNA levels were monitored for 120’ following initial induction at the indicated time points. Across all copper concentrations, gene expression peaked at 10 minutes, with higher copper concentrations eliciting a more robust response. GFP expression levels decreased at subsequent time points, indicating a transient induction following copper exposure. Values were normalized to the expression of β\beta-actin and are presented as fold change over non-induced cells. Data represent mean ±\pm SEM from two independent experiments (N=2). Statistical significance was determined using a two-tailed Student’s t-test. *p < 0.05, **p < 0.01, ***p < 0.001, ns – non significant.

Following stimulation with 1 μM\mu\mathrm{M}, mRNA transcripts are barely detectable-yet peak at 10 minutes, while 10 μM\mu\mathrm{M} leads to notable upregulation increases at both 3 and 10 minutes after stimulation, without altering the peak time of the response. Copper inputs of 100 and 500 μM\mu\mathrm{M}, cause a notably faster activation of the CUP1 promoter leading to detectable transcripts within 3 minutes. Expression levels peak at 10 minutes followed by a sharp decrease. Overall, the detoxification response maintains a similar transient pattern across all tested concentrations, being particularly fast-acting at higher copper dosages. The natural dynamics of the activity at the promoter level are indicative of the cellular protection against elevated copper concentrations. Real-time RNA sequence detection has been reported in the literature [51], [52]. These studies indicate that genetically encoded RNA aptamers capable of specifically binding an exogenously supplied fluorophore have emerged as a promising strategy for the precise tracking and quantification of transcripts in living cells. Among them, the Spinach aptamer has been successfully applied to label RNA molecules in living S. cerevisiae cells. More specifically, the Spinach sequence can be cloned between the coding region and the 3-UTR of a target gene, producing green fluorescence upon interaction with DFHBI, a GFP chromophore analogue. In parallel, advanced imaging technologies have been developed to provide real-time spatiotemporal and quantitative characterization of both highly and weakly expressed transcripts. This is of great importance, as it points to the potential of leveraging the rapid RNA response of yeast cells to achieve high event rates.

Biosensor Selectivity

To evaluate potential cross-reactivity with other metal ions, cells were exposed to ZnCl2\mathrm{ZnCl}_{2}, a chemically related metal that could potentially interfere with biosensor activity, under the same induction conditions used for copper stimulation. RT-qPCR analysis revealed no significant increase in GFP transcript levels following ZnCl2\mathrm{ZnCl}_{2} treatment compared with non-induced controls, demonstrating that the biosensor is selectively activated by copper and not by zinc (Figure 11). Consistent with these findings, monitoring of GFP fluorescence over time showed that ZnCl2\mathrm{ZnCl}_{2} failed to induce reporter expression, whereas CuSO4\mathrm{CuSO}_{4} elicited a robust and time-dependent increase in GFP fluorescence (Figure 12). Together, these results demonstrate that the biosensor responds specifically to copper and is not activated by zinc under the conditions tested.

GFP transcript levels in cup1∆::sfGFP cells

Figure 11. GFP transcript levels in cup1∆::sfGFP cells following exposure to 100 μ\muM ZnCl2_2 for the indicated induction times, determined by quantitative PCR. Zinc treatment did not induce GFP transcript expression above basal levels. Values were normalized to the expression of β\beta-actin and are presented as fold change over non-induced cells. Data represent mean ±\pm SEM from two independent experiments (N=2). Statistical significance was determined using a two-tailed Student’s t-test. *p < 0.05, ns – non significant.

GFP fluorescence in cup1∆::sfGFP cells

Figure 12. GFP fluorescence in cup1∆::sfGFP cells following induction with 100 μ\muM CuSO4_4 or 100 μ\muM ZnCl2_2. While CuSO4_4 induced a marked increase in GFP fluorescence, ZnCl2_2 failed to elicit a detectable response throughout the induction period, demonstrating the specificity of the CUP1/Ace1 regulatory system for copper under the conditions tested. Data represent mean ±\pm SEM from three independent experiments (N=3).

Validation

In this section, the potential of a simulation platform for characterizing the response of the yeast receiver to copper stimuli, in a manner aligned with the experiments in Section VI-A and Section VI-B, is demonstrated. Such a tool can complement experimental procedures by informing the design of subsequent trials and by providing mechanistic insight into the underlying biological processes. This is particularly important in settings where laboratory cost constitutes a limiting factor.

To this end, a scenario involving a single copper stimulation is considered as the input to the mathematical model of the copper detoxification pathway described in Section IV and specified further in Appendix A. The GFP reporter is treated as the model output, and theoretical mRNA transcript and protein abundance levels are quantified as a time-evolution profile. Model versatility is examined by juxtaposing theoretical predictions under 1, 10, 100 and 500 μ\muM of copper input, with the GFP transcript measurements reported in Section VI-B at corresponding time points. For protein abundance levels reported in Section VI-A, GFP predictions are limited to a single copper input of 100 μ\muM. The resulting validation comparisons are presented in Figure 13 and Figure 14 respectively.

Figure 13. Comparison between theoretical and experimental results across four different copper inputs. mRNA levels are reported.

Figure 14. Comparison between theoretical and experimental results for a single copper input. Protein abundance levels are reported.

To facilitate comparison of the time-evolution profiles, measured and simulated results were normalized to their respective maximum values. Differences in absolute amplitude between theoretical and experimental traces are

commonly attributed to discrepancies in quantification procedures, since the computational methods embedded in laboratory instrumentation are not readily replicated through simplified mathematical representations.

As shown in Figure 13, satisfactory concordance is observed between the computed and measured profiles overall. RNA expression peaks at approximately 10 minutes and is followed by rapid degradation.

The observed disparities are primarily associated with the initial transcript levels and the tail end of the responses. For copper inputs below 100 μ\muM (blue and red curves), comparatively low activation is observed at 3 minutes, resulting in closer agreement with the model predictions. In contrast, higher mRNA levels are reached within the first 3 minutes for the measured profiles that correspond to 100 and 500 μ\muM (yellow and purple curves), whereas lower mRNA abundance is predicted by the model during this early phase. In addition, lingering transcripts are observed beyond 45 minutes in the experimental data, which are not captured by the current

E1mRNAE_{1}^{mRNA}E10mRNAE_{10}^{mRNA}E100mRNAE_{100}^{mRNA}E500mRNAE_{500}^{mRNA}
0.270.110.150.2

caption: tbl:3: RMSE between model predictions and mRNA measurements for the GFP reporter.

formulation. To quantify the observed differences, the Root Mean Square Error (RMSE) between the computed and measured profiles is reported in tbl:3, where E1mRNAE_{1}^{mRNA}, E10mRNAE_{10}^{mRNA}, E100mRNAE_{100}^{mRNA} and E500mRNAE_{500}^{mRNA} correspond to the errors for each concentration. Albeit the reported deviations, the agreement in peak timing, together with the overall similarity in dynamics, supports the value of the proposed model as a simulation platform for describing mRNA expression, while acknowledging its limitations.

Notable discrepancies are also observed between the theoretical and experimental time-course protein profiles in Figure 14. Although the peak time coincides in both cases, the predicted GFP trajectory exhibits a smooth, gradual increase toward the maximum at t=300t = 300 minutes. In contrast, the experimental recordings show a steeper rise with noticeable fluctuations, followed by a comparatively rapid decline in fluorescence. Such a fast decay is not captured by the current model, for which slower degradation kinetics are predicted. The kinetic parameters adopted in the GFP model are mentioned in detail in Appendix A-C. It ought to be noted however, that the destabilized form of the sfGFP used for the experiments was not possible to be reflected in the model parameters. As a result, a wider pulse is obtained in the computed profile; nevertheless, the overall trend remains consistent, with maximum abundance reached at approximately the same time as in the measurements. The mismatch observed in Figure 14 is again quantified through the RMSE, which is found to be 0.24.

The observed differences in both graphs may be attributed to biological nuances that are not represented by the mathematical model, since emphasis has been placed on processes deemed the most relevant. Additional regulatory effects and system-specific complexities pertaining to gene expression, protein folding and stability have been neglected. It is emphasized that the presented model is intended to serve as a basis for further refinement, and is able to characterize practical response profiles of the yeast receiver to a satisfying extent.

Bioluminescence Computations

Here, we present results obtained from an extended version of the model used for validation, which also includes its current response to copper-based stimulation. Constructing a fully developed model tailored to yeast biosensors is crucial for creating a framework that supports the incorporation of yeast MC receivers into applications in the IoBNT field.

The models describing the Lux reporter are integrated with the mathematical framework previously established for the copper detoxification pathway, and the resulting comprehensive model is employed to quantify the photocurrent response of the biosensor. The majority of the model parameters, as well as the initial concentrations of the chemical species considered in the simulations, were compiled from multiple literature sources, primarily studies on Vibrio harveyi and Photohabdus luminescens. Additional kinetic investigations are necessary to obtain more accurate, yeast-specific parameter values; nevertheless, the full model presented here provides a foundational framework for subsequent refinement and optimization.

The copper input mobilizes the chemical reactions within the copper detoxification pathway, which culminate in the upregulation of the luxAB gene, which is translated to the Lux enzyme. The latter triggers light emission. As in Section VI-A, a single simulation is considered at the receiver, where 100 μ\muM of copper is directly supplied without the consideration of molecular propagation nuances. The photocurrent profile is shown in Figure 15.

Figure 15. Photocurrent response for 100 μ\muM copper stimulation.

Although experimental data are not available for validation of the Lux reporter considered here, the response shown in Figure 15 provides some useful qualitative insights. The time evolution of the output exhibits a peak at approximately 49 minutes and maintains a bell-shaped pulse profile with long sustain, due to the long half-life of the luciferase in Photohabdus luminescens [53]. The p+/n-well/psub CMOS photodiode considered in the computations has an active diffusion area of 40 ×\times 40 μ\mum2^{2} and has the highest responsivity near λ=490\lambda= 490 nm among the other designs reported in [43]. The dark current can be estimated from the measured dark-voltage variation as follows:

Idark=CpddVdarkdt,(26)I_{\mathrm{dark}} = C_{\mathrm{pd}} \frac{dV_{\mathrm{dark}}}{dt}, \tag*{(26)}

where Cpd=1352.7C_{\mathrm{pd}} = 1352.7 [fF] and dVdarkdt=18\frac{dV_{\mathrm{dark}}}{dt} = 18 [mV/s]. The resulting dark current is about 24.3 [fA], while the predicted maximum photocurrent is approximately 31 [fA]. Therefore, the predicted photocurrent is approximately 1.3 times larger than the estimated dark current. This comparison provides an initial indication that the photocurrent may be distinguishable from the detector baseline under ideal conditions such as those considered here. However, a complete feasibility analysis should also consider dark-current shot noise, thermal noise, detector bandwidth, integration time, and optical collection and propagation losses, which are expected to lead to a notably weaker signal and a higher noise background. Stochastic effects associated with molecular propagation and transcription-factor binding may introduce additional variability. Furthermore, the present single-cell system should be extended to account for population-level emission in a bulk cell culture medium, which may lead to a more robust photocurrent response overall.

The faster response peak compared to the GFP measurements reported in Section VI-A is attributed primarily to the maturation time required for GFP to reach its fluorescent form, as well as to its relatively long lifetime, which can result in lingering fluorescent molecules within the cell. In contrast, luciferase reporters are typically characterized by minimal maturation time [54], [55], therefore enabling more prompt protein-level responses and a shorter-lived output. Such effects have been incorporated into the model in order for the corresponding physical behavior to be reflected.

To further assess the proposed model, a sensitivity analysis is conducted in order to quantify changes in the biosensor response under parameter perturbations. In this manner, robustness of the predicted dynamics to specific parameter variations is examined, while response trends under alternative parameter sets can also be explored. Such an analysis can be useful for identifying dominant mechanisms and, in principle, for guiding the design of modified biosensors that incorporate ’mutant’ receiver configurations.

Several parameters from the equations in Appendix A-C were varied, including the maximum copper uptake rate constant (Eq. (113)), the Ace1–DNA promoter dissociation constant (Eq. (116)), the luxAB mRNA degradation rate (Eq. (96)), the LuxC forward catalytic rate (Eq. (118)), the LuxAB enzyme degradation rate (Eq. (98)), and the hydroxyflavin de-excitation rate (Eq. (34)). In each case, only one parameter was varied while all remaining parameters were kept fixed. As in the previous simulations, a single copper stimulation of 100 μM100\,\mu\mathrm{M} is considered and the resulting time-course response is evaluated across the tested parameter values.

Inspection of the plots in Figure 16 indicates that parameter variations primarily affect the amplitude and peak time, depending on the parameter being perturbed, while pulse width was less altered. As shown in Figure 16a, larger values of the maximum copper uptake constant result in increased current amplitude, albeit at the expense of a delayed peak and slightly larger pulse width. This behavior can be attributed to enhanced copper influx over the course of the simulated experiment, which prolongs the buildup of downstream species in the reaction chain.

Parameter sweeps for six model parameters

Figure 16. Parameter sweeps for 6 model parameters. (a) Maximum copper uptake rate constant sweep. (b) Dissociation constant for the binding of Ace1 to the CUP1 DNA promoter sweep. (c) luxAB mRNA degradation rate sweep. (d) Forward catalysis rate of LuxC sweep. (e) LuxAB protein abundance degradation rate sweep. (f) De-excitation and de-hydration rate sweep of hydroxyflavin, after light emission.

A less pronounced increase in amplitude is observed in Figure 16d, where increasing the forward catalytic rate of LuxC produces a slightly stronger current response and

a markedly earlier peak. However, the peak amplitude approaches a plateau at higher catalytic rates, suggesting that LuxC is no longer the rate-limiting step and that other processes within the Lux pathway limit the response. The higher current output observed for values of VLuxC+V^{+}_{\mathrm{LuxC}} above 0.08 μM min−10.08\,\mu\mathrm{M}\,\mathrm{min}^{-1} can be attributed to increased aldehyde production, which increases substrate availability for LuxAB-driven light emission.

In contrast, in Figure 16b and Figure 16f, parameter perturbations introduce stronger variation in pulse amplitudes. Reduced dissociation constant values for Ace1 binding to DNA promoters correspond to stronger binding affinity, thereby increasing transcript production in the nucleus and ultimately increasing LuxAB abundance and photon generation directly. Notable increase of the dissociation constant may result in faster time peaks by a very small margin, albeit at the expense of a dramatic decrease of the output current. The pulse width of the response remains largely unaffected. Altering the de-excitation rate in Figure 16f also results in much varied responses inferring high response sensitivity: faster de-excitation of the hydroxyflavin intermediate (see Figure 4) reduces photon yield, whereas smaller values lead to brighter emission and to a severely increased photodetector response, without influencing peak time and pulse width.

The influence of Lux degradation rates on the current output is examined in Figure 16c and Figure 16e for luciferase mRNA and protein, respectively. In both cases, higher degradation rates yield diminished responses with narrower pulse widths, whereas lower degradation rates produce more intense and long-lasting responses. Current amplitude changes are more pronounced for mRNA degradation variations, while pulse width is markedly altered at higher protein degradation values. Peak instance is similarly affected in both cases. This suggests that incorporation of protein-degradation tags, through which accelerated turnover can be introduced, can enable responses that more closely track promoter activity and partially counteract translation delay. However, a tradeoff is introduced between detection rate and detection reliability in an MC system, since faster but weaker output pulses are more susceptible to noise and losses, thereby increasing the probability of erroneous detection.

The effects discussed above are summarized in Table IV, where the current pulse metrics (amplitude, peak time, and pulse width), are provided for every plot of Figure 16. Pulse width is defined here as the time interval during which the signal remains above of the 50% of its maximum current amplitude. If the signal does not fall below the 50% threshold within the simulated time window, the pulse end time is set equal to the final simulation time point. Moreover, the corresponding variances are reported in Table V.

parameterkupt,maxk_{upt,max}KDmRNAK_{DmRNA}kdeg,mRNAk_{deg,mRNA}VLuxCV_{LuxC}kdeg,LuxABk_{deg,LuxAB}k4k_4
Peak1 [fA] (blue)0.00230.1567.225.6134.231
Peak2 [fA] (orange)316.55201.1430.14313.1
Peak3 [fA] (yellow)66.71.72313119.420.062
Peak4 [fA] (purple)93.150.763.3430.124.620.016
PeakT1 [min] (blue)46.7350.41360112.37550.41
PeakT2 [min] (orange)50.447.4193.2350.2350.4150.4
PeakT3 [min] (yellow)59.444750.4150.229.9250.33
PeakT4 [min] (purple)694723.9150.219.550.3
Width1 [min] (blue)275275.2-301.75341.5275.2
Width2 [min] (orange)275.2275308.95275275.2275.2
Width3 [min] (yellow)277.3275275.18275.150.73275.2
Width4 [min] (purple)280.65275245.65275.1422.11275.14

Table IV. Amplitude, Peak Time and Pulse Width values (parameter sweep).

parameterPeakVarPeakTimeVarWidthVar
kupt,maxk_{upt,max}10032.252.1
KDmRNAK_{DmRNA}97.670.1
kdeg,mRNAk_{deg,mRNA}99.4293.3623.1
VLuxCV_{LuxC}15.155.218.94
kdeg,LuxABk_{deg,LuxAB}86.57493.52
k4k_41000.210.01

Table V. Amplitude, Peak Time and Pulse Width variances (%, parameter sweep).

To complement the sensitivity analysis with respect to internal model parameters, the influence of the initial amounts of three chemical compounds is additionally examined, namely the transporters Ctr1/3, the cofactor FMN, and intracellular fatty acids, as shown in Figure 17. In addition, a simulated dose-response graph is provided, where the output current is reported for different copper input levels.

Different initial concentrations for Ctr transporters, FMN and fatty acids, and copper dose-response curves

Figure 17. Different initial concentrations for Ctr transporters, FMN and fatty acids. A copper dose-response is also included.

Pronounced sensitivity is observed with respect to the initial concentration of the Ctr transporters. As shown in Figure 17a, increased transporter abundance results in a substantial increase in current amplitude. The output pulse also widens as the Ctr concentration increases up to 1 μ\muM, while the current peak is progressively delayed. In contrast, the output current remains largely unaffected by variations in the initial FMN concentration, as shown in Figure 17b. This suggests that FMN is not a limiting substrate within the tested concentration range and may already be present at effectively saturating levels.

In Figure 17c, low initial fatty-acid concentrations, such as 0.08 μ\muM, substantially attenuate the current output without noticeably affecting the pulse width. Higher fatty-acid concentrations support increased aldehyde production and, consequently, greater substrate availability for LuxAB-driven light emission, resulting in stronger current responses. A sharp increase in current amplitude is observed when the RCOOH concentration is increased

from 8 to 100 μ\muM. However, the highest concentration also introduces a delay in the current peak and a modest increase in pulse width. Finally, the simulated copper dose-response curves in Figure 17d show that different copper input levels produce distinct photocurrent profiles that differ mainly in amplitude. Changes in peak time and pulse width remain comparatively limited, while the current amplitude approaches saturation at the higher copper concentrations.

In addition to the profiles presented in Figure 17, the current-pulse metrics are again summarized in Table VI, while the corresponding variances are reported in Table VII.

chemical compoundCtr1/3FMNRCOOHCu input
Peak1 [fA] (blue)1530.20.30.3
Peak2 [fA] (orange)30.230.23.0224.43
Peak3 [fA] (yellow)12230.230.1630.2
Peak4 [fA] (purple)268.330.2313.530.81
PeakT1 [min] (blue)4750.2350.2247.16
PeakT2 [min] (orange)50.2350.2650.2249.6
PeakT3 [min] (yellow)80.8350.250.2250.23
PeakT4 [min] (purple)156.550.2610850.3
Width1 [min] (blue)274.92275.1275.14275
Width2 [min] (orange)275.14275.12275.14275.03
Width3 [min] (yellow)285.6275.1275.14275.14
Width4 [min] (purple)287.16275.1292.5275.14

Table VI. Amplitude, Peak Time and Pulse Width values (initial concentration sweep).

chemical compoundPeakVarPeakTimeVarWidthVar
Ctr1/394.5370.234.26
FMN0.090.140.01
RCOOH99.953.45.92
Cu996.20.1

Table VII. Amplitude, Peak Time and Pulse Width variances (%, initial concentration sweep).

Although individual parameter perturbations provide useful insight into the influence of key model parameters on the output current, evaluating the biosensor under simultaneous parameter changes is also important. Since the model includes several parameters obtained from non-yeast systems, varying multiple parameters together can provide a broader representation of parameter uncertainty. For this purpose, the selected parameters include values that were either obtained from non-yeast systems or assumed due to the lack of available data. These parameters are the LuxE forward catalytic rate (VLuxE+V^{+}_{LuxE}, Eq. (117)), the FMN dissociation constant (KD,FK_{D,F}, Eq. (15)), the oxygen reaction rate constant (kOk_O, Eq. (34)), the Michaelis constant of ATP (Km,ATPK_{m,ATP}, Eq. (17)), the dissociation constant of fatty acids (KD,RCOOHK_{D,RCOOH}, Eq. (17)), the Ctr trimerization rate (kb1k_{b1}, Eq. (65)), and the binding rate constant of the substrate FMNH2_2 (k1k_1, Eq. (34)).

The corresponding parameter sets are provided in Table VIII. The results shown in Figure 18 provide further insight into how combinations of parameter changes affect the photocurrent profile.

Figure 18. Combined parameter sweep.

Out of the parameters tested, the binding rate kb1k_{b1} appears to exert the strongest influence on the output current. Higher values of this parameter across the different sets, consistently resulted in increased amplitude. This result is consistent with those shown in Figure 16a and Figure 17a, indicating that copper transport through the Ctr proteins are among the strongest determinants of the current amplitude. The combination of higher VLuxE+V^{+}_{LuxE} and k1k_1 also stands out, as it is associated with a steeper increase in photocurrent during the rising phase in sets 2, 4, and 5. In contrast, lower values of these two rates are generally associated with more gradual responses, as observed in sets 1 and 3. The suggested influence of VLuxE+V^{+}_{LuxE} is also consistent with the results shown in Figure 16d, which indicate that the forward catalytic rates of both enzymes involved in aldehyde production affect the slope of the response.

It is imperative to note that, the temporal characteristics of the biosensor’s response are expected to play a pivotal role in the achievable data rate of the envisioned communication system. When viewed from an MC perspective, lower response latency combined with smaller output pulse width will contribute to real-time, fast detection of molecular events. In case of re-stimulation capability of the system under consideration, wider pulses may cause overlap between output pulses as observed in previous work [7], undermining detection performance.

Overall, the proposed mathematical framework is able to adapt to parameter variations that reflect different experimental conditions to a satisfactory extent. This is crucial, due to the scarcity of reported values for yeast systems in the literature. Since the model includes a large number of parameters that couple the copper detoxification pathway to the Lux reporter—many of which have not been quantitatively determined for eukaryotic systems—further experimentation, together with refinement of modeling assumptions, is expected to yield a more nuanced theoretical tool. In this manner, a digital-twin representation of a yeast-derived biosensor can be progressively approached.

ParameterSet1Set2Set3Set4Set5
VLuxEV_{LuxE} [μM min−1\mu\mathrm{M}\ \mathrm{min}^{-1}]0.1100.051.2100
KD,FK_{D,F} [μM\mu\mathrm{M}]0.00171.70.170.0171.7⋅10−41.7 \cdot10^{-4}
kOk_O [min−1\mathrm{min}^{-1}]210000210002102100700
Km,ATPK_{m,ATP} [μM\mu\mathrm{M}]0.0020.020.22⋅10−42 \cdot10^{-4}2⋅10−52 \cdot10^{-5}
KD,RCOOHK_{D,RCOOH} [μM\mu\mathrm{M}]0.0110.110−310^{-3}10−410^{-4}
kb1k_{b1} [μM−2 min−1\mu\mathrm{M}^{-2}\ \mathrm{min}^{-1}]0.11000100101
k1k_1 [μM−1 min−1\mu\mathrm{M}^{-1}\ \mathrm{min}^{-1}]102010210.210200102000

Table VIII. Parameter sets used for a mixed parameter sweep.

Conclusion

In this work, S. cerevisiae is leveraged for the development of an MC receiver capable of reporting copper presence in its vicinity. A theoretical framework is presented for characterization of the dominant chemical reactions within the yeast copper detoxification pathway, while the system output is quantified through a coupled non-linear ODE model. Validation attempts of GFP model predictions against experimental trials conducted on a dedicated MC testbed, supports the proposed computational approach. Hence, it provides the basis for a simulation platform suitable for virtual experimentation.

The framework is further extended by replacing the GFP reporter with an auto-luminescent system capable of generating optical output without external light excitation, thereby enabling coupling of the yeast receiver to a photodetector unit and supporting integration into biosensing devices such as Smart Pills. The resulting model is used to characterize the photocurrent response of the proposed yeast-derived biosensor, while sensitivity analysis provides insight into key mechanisms underlying the light pathway.

Overall, this paper constitutes a step toward establishing MC as a key contributor for IoBNT applications. Future work will focus on the implementation of a new testbed that integrates the Lux system in yeast cells, and experimental validation of the Lux model extension to further strengthen its predictive value. Moreover, a transmission model coupled with the molecular propagation channel that accounts for the effects of the physical Tx-Rx separation, will be integrated together with the receiver into a comprehensive communication framework. Noise analysis pertaining to all different parts of the communication model will support error performance characterization and detection methods. Finally, investigation of the restimulation capabilities of the system both physically and computationally will be pursued, and achievable data rates will be assessed.

Receiver Model

Transcriptional activation function fact

In this work, MRE occupancy is assumed to pertain to two distinct occupation states (OS), namely OS0OS_0 when there is no TF bound and OS1OS_1 when TF molecules are attached to the MREs of the promoter. The fractional transcriptional level (FTL) for a single MRE that is bound by Ace1 is given as follows:

FTL=OS1OS0+OS1(27)FTL = \frac{OS_1}{OS_0 + OS_1} \tag*{(27)}

Binding creates the transition from OS0OS_0 to OS1OS_1 and unbinding reverts back to OS0OS_0. By considering rapid binding/unbinding of Ace1 to and from the MRE, and the fact that Ace1 binds to the promoter as a monomer [38], it is straightforward to express the ratio OS0OS1\frac{OS_0}{OS_1} as follows:

OS0OS1=KD,mRNA[Ace1Cu](28)\frac{OS_0}{OS_1} = \frac{K_{D,mRNA}}{[Ace1Cu]} \tag*{(28)}

Where the dissociation constant KD,mRNAK_{D,mRNA} is defined as the ratio of the unbinding constant rate of Ace1 from the OS1OS_1 conformation towards the binding rate of Ace1 to the OS0OS_0 conformation.

Thus, the FTL can be written as:

FTL=1KD,mRNA[Ace1Cu]+1(29)FTL = \frac{1}{\frac{K_{D,mRNA}}{[Ace1Cu]} + 1} \tag*{(29)}

Moreover, Ace1 binds to four MREs [38], which we assume that contribute to the gene induction in an independent manner. To reflect all possible contributions of MRE occupancy within the DNA promoter, Eq. (29) is modified accordingly to:

FTL=1∑n=14(KD,mRNA[Ace1Cu])n+1(30)FTL = \frac{1}{\sum_{n=1}^{4}\left(\frac{K_{D,mRNA}}{[Ace1Cu]}\right)^n + 1} \tag*{(30)}

As FTL scales the maximum transcriptional induction possible depending on the possible conformations available for the system under consideration, the final form of the transcriptional activation function factf_{act} takes the form:

fact=ktr,maxFTLf_{act} = k_{tr,max} FTL

With ktr,maxk_{tr,max} being the maximum transcription rate constant. This approach matches the approach also followed in [39] to a large extent.

Rate equations based on the King-Altman schematic method

Denominator 'D' in (12)

1) Denominator ’D’ in (12): As mentioned in Section IV-A1, the denominator D comprises all products stemming from the valid King-Altman patterns for every enzyme form in the reaction between FRP, FMN and NADPH. Since there are four legitimate patterns for every enzyme form and the patterns for EX have already been derived, we proceed by outlining the rest of the 12 diagrams that use Figure 6 as a template. Irreversible ends_mid=1 product release is maintained for all diagrams. Hence, for the enzyme form EFRPFE_{FRP}F, the legitimate diagrams are shown in Figure 19.

Legitimate King-Altman patterns for the flavin reduction reaction for the enzyme form $E_{FRP}F$

Figure 19. Legitimate King-Altman patterns for the flavin reduction reaction for the enzyme form EFRPFE_{FRP}F.

The legitimate patterns for the enzyme form EQEQ are the following, shown in Figure 20:

Legitimate King-Altman patterns for the flavin reduction reaction for the enzyme form $EQ$

Figure 20. Legitimate King-Altman patterns for the flavin reduction reaction for the enzyme form EQEQ.

And finally, for the free enzyme form EFRPE_{FRP}, the patterns are shown in Figure 21.

Legitimate King-Altman patterns for the flavin reduction reaction for the free enzyme form $E_{FRP}$

Figure 21. Legitimate King-Altman patterns for the flavin reduction reaction for the free enzyme form EFRPE_{FRP}.

Out of all the diagrams presented in Figure 19, Figure 20 and Figure 21, the valid ones that contribute to every enzyme form will only be those shown in Figure 19c, Figure 19d, Figure 20c, Figure 21a, Figure 21b, Figure 21d, due to the exclusion of product inhibition. Together with the pattern of Figure 7b, these diagrams provide the products for the denominator DFRP=DEFRP+DEFRPF+DEX+DEQD_{FRP}=D_{E_{FRP}}+D_{E_{FRP}F}+D_{EX}+D_{EQ}. Each ’D’ simply contains the products of the reaction rates depicted on the paths in the corresponding

legitimate patterns. After some simple algebraic manipulation, the common denominator DFRPD_{FRP} can be written as:

DFRP=k4k−1(k−2+k3)+k1k4(k−2+k3)[F]+k4k3k2[N]+k1k2(k4+k3)[F][N](31)D_{FRP}=k_4k_{-1}(k_{-2}+k_3)+k_1k_4(k_{-2}+k_3)[F]+k_4k_3k_2[N]+k_1k_2(k_4+k_3)[F][N] \tag*{(31)}

The same procedure is followed to construct all the legitimate patterns for the enzymatic reactions involving LuxE and LuxC, and their derivations are omitted. Although the kinetic constants involved in the King-Altman formulations describe the chemical interactions that contribute to every enzyme form in more detail, using Cleland’s form is often more convenient as comprehensive parameters like the Michaelis constant or dissociation factors are typically more often mentioned in the literature. Using both kinds of formulations can also aid in parameter estimation, simply by expressing the group parameters of Cleland’s formulation as functions of the kinetic constants from the original King-Altman formulation, and proceed with proper algebraic manipulations [40].

Rate of LuxAB catalysis

The master pattern based on which the King-Altman patterns are drawn, is shown in Figure 8. Light emission is assumed to be generated from the enzyme form EZEZ, and the King-Altman patterns for this case are shown in Figure 22.

Figure 22. Legitimate King-Altman patterns for the light-emitting enzyme form EZEZ (excited hydroxyflavin).

Reverse binding of products and oxygen is not considered in this cycle, nor inverse reactions of intermediates that result from chemical conformation changes. Thus, reversibility is considered only for ELuxFHE_{Lux}FH and EYEY. Thus, the only valid pattern is the one depicted in Figure 22e. The concentration of EZEZ is given as:

[EZ]=[EtotLux]NEZDLux(32)[EZ]=\frac{[E_{\mathrm{tot}}^{Lux}]N_{EZ}}{D_{Lux}} \tag*{(32)}

With NEZN_{EZ} given by:

NEZ=k5k6k1k0k2k3[FH][A](33)N_{EZ}=k_5k_6k_1k_0k_2k_3[FH][A] \tag*{(33)}

The constant rate k2k_2 is defined by Eq. (21). The derivation of DLuxD_{Lux} requires all possible patterns to be drawn out for all possible enzyme forms. As this is a long and repetitive-yet straightforward procedure, we proceed by directly providing DLuxD_{Lux} as follows:

DLux=k2k3k4k5k6(kO+k−1)[A]+kOk1k4k5k6(k3+k−2)[FH]+k1k2(k3k4k5k6+kOk4k5k6+kOk3k5k6+kOk3k4k6+kOk3k4k5)[FH][A](34)\begin{aligned} D_{Lux} ={}& k_2 k_3 k_4 k_5 k_6 (k_O + k_{-1})[A] \\ &+ k_O k_1 k_4 k_5 k_6 (k_3 + k_{-2})[FH] \\ &+ k_1 k_2(k_3 k_4 k_5 k_6 + k_O k_4 k_5 k_6 + k_O k_3 k_5 k_6 + k_O k_3 k_4 k_6 + k_O k_3 k_4 k_5)[FH][A] \tag*{(34)} \end{aligned}

Photon emission results from Eq. (22). We considered the inclusion of key intermediate states within the light pathway to be of great importance as they can reveal important insights on the kinetic mechanisms that participate in this cycle. Also, many of the kinetic parameters were able to be adopted or inferred from the literature.

ODE model

All species involved in the copper detoxification pathway are explicitly modeled. In the Lux pathway, FRP is treated as constant. LuxE and LuxC concentrations are incorporated into VLuxEV_{LuxE} and VLuxCV_{LuxC}, and are thus treated similarly. RCOOH, FMN and NADPH are modeled, yet

their equations do not contain a source term. This is due to the assumption of their replenishment internally, at the end of the simulated response. Unlike those of the copper detoxification pathway, non-assumed Lux parameters are adopted from bacterial systems.

d[Cu]dt=p0−n0(35)\frac{d[Cu]}{dt}=p0-n0 \tag*{(35)}
d[Ctr1/3]dt=p1−n1(36)\frac{d[Ctr1/3]}{dt}=p1-n1 \tag*{(36)}
d[Ctr1/3trim]dt=p2−n2(37)\frac{d[Ctr1/3_{trim}]}{dt}=p2-n2 \tag*{(37)}
dCtr1/3trimCudt=p3−n3(38)\frac{dCtr1/3_{trim}Cu}{dt}=p3-n3 \tag*{(38)}
d[GSH]dt=p4−n4(39)\frac{d[GSH]}{dt}=p4-n4 \tag*{(39)}
d[GCu]dt=p5−n5(40)\frac{d[GCu]}{dt}=p5-n5 \tag*{(40)}
d[Ccs1]dt=p6−n6(41)\frac{d[Ccs1]}{dt}=p6-n6 \tag*{(41)}
d[Ccs1Cu]dt=p7−n7(42)\frac{d[Ccs1Cu]}{dt}=p7-n7 \tag*{(42)}
d[Sod1]dt=p8−n8(43)\frac{d[Sod1]}{dt}=p8-n8 \tag*{(43)}
d[Atx1]dt=p9−n9(44)\frac{d[Atx1]}{dt}=p9-n9 \tag*{(44)}
d[Atx1Cu]dt=p10−n10(45)\frac{d[Atx1Cu]}{dt}=p10-n10 \tag*{(45)}
d[Ccc2]dt=p11−n11(46)\frac{d[Ccc2]}{dt}=p11-n11 \tag*{(46)}
d[Ace1]dt=p12−n12(47)\frac{d[Ace1]}{dt}=p12-n12 \tag*{(47)}
d[Ace1Cu]dt=p13−n13(48)\frac{d[Ace1Cu]}{dt}=p13-n13 \tag*{(48)}
d[GFPmRNA]dt=p14−n14(49)\frac{d[GFP_{mRNA}]}{dt}=p14-n14 \tag*{(49)}
d[GFPnascent]dt=p15−n15(50)\frac{d[GFP_{nascent}]}{dt}=p15-n15 \tag*{(50)}
d[GFPmature]dt=p16−n16(51)\frac{d[GFP_{mature}]}{dt}=p16-n16 \tag*{(51)}
d[luxABmRNA]dt=pluxABmRNA−nluxABmRNA(52)\frac{d[luxAB_{mRNA}]}{dt}=p_{luxAB_{mRNA}}-n_{luxAB_{mRNA}} \tag*{(52)}
d[LuxAB]dt=pLuxAB−nLuxAB(53)\frac{d[LuxAB]}{dt}=p_{LuxAB}-n_{LuxAB} \tag*{(53)}
d[FRP]dt=pFRP−nFRP(54)\frac{d[FRP]}{dt}=p_{FRP}-n_{FRP} \tag*{(54)}
d[RCOOH]dt=pRCOOH−nRCOOH(55)\frac{d[RCOOH]}{dt}=p_{RCOOH}-n_{RCOOH} \tag*{(55)}
d[LuxE−RCO]dt=pLuxERCO−nLuxERCO(56)\frac{d[LuxE-RCO]}{dt}=p_{LuxERCO}-n_{LuxERCO} \tag*{(56)}
d[A]dt=pA−nA(57)\frac{d[A]}{dt}=p_A-n_A \tag*{(57)}
d[F]dt=pF−nF(58)\frac{d[F]}{dt}=p_F-n_F \tag*{(58)}
d[FH]dt=pFH−nFH(59)\frac{d[FH]}{dt}=p_{FH}-n_{FH} \tag*{(59)}
d[N]dt=pN−nN(60)\frac{d[N]}{dt}=p_N-n_N \tag*{(60)}
p0=0(61)p_0=0 \tag*{(61)}
n0=kupt[Ctr1/3trim](62)n_0=k_{upt}[Ctr1/3_{trim}] \tag*{(62)}
p1=SCtr1/3+kdis1[Ctr1/3trim](63)p_1=S_{Ctr1/3}+k_{dis1}[Ctr1/3_{trim}] \tag*{(63)}
n1=kb1[Ctr1/3][Ctr1/3][Ctr1/3]+kdeg1[Ctr1]+kCu−deg[Ctr1/3](64)n_1=k_{b1}[Ctr1/3][Ctr1/3][Ctr1/3]+k_{deg1}[Ctr1]+k_{Cu-deg}[Ctr1/3] \tag*{(64)}
p2=kb1[Ctr1/3][Ctr1/3][Ctr1/3]+ktran1[Ctr1/3trimCu][GSH](65)p_2=k_{b1}[Ctr1/3][Ctr1/3][Ctr1/3]+k_{tran1}[Ctr1/3_{trim}Cu][GSH] \tag*{(65)}
n2=kdis1[Ctr1/3trim]+kupt[Ctr1/3trim](66)n_2=k_{dis1}[Ctr1/3_{trim}]+k_{upt}[Ctr1/3_{trim}] \tag*{(66)}
p3=kupt[Ctr1/3trim](67)p_3=k_{upt}[Ctr1/3_{trim}] \tag*{(67)}
n3=ktran1[Ctr1/3trimCu][GSH](68)n_3=k_{tran1}[Ctr1/3_{trim}Cu][GSH] \tag*{(68)}
p4=SGSH+ktran2[GCu][Ccs1]+ktran3[GCu][Atx1]+kmet[Ace1](69)p_4=S_{GSH}+k_{tran2}[GCu][Ccs1]+k_{tran3}[GCu][Atx1]+k_{met}[Ace1] \tag*{(69)}
n4=kdeg2[GSH]+ktran1[GSH][C1trimCu](70)n_4=k_{deg2}[GSH]+k_{tran1}[GSH][C1_{trim}Cu] \tag*{(70)}
p5=ktran1[GSH][C1trimCu](71)p_5=k_{tran1}[GSH][C1_{trim}Cu] \tag*{(71)}
n5=ktran2[GCu][Ccs1]+ktran3[GCu][Atx1]+kmet[Ace1](72)n_5=k_{tran2}[GCu][Ccs1]+k_{tran3}[GCu][Atx1]+k_{met}[Ace1] \tag*{(72)}
p6=SCcs1+ktran5[Ccs1Cu][Sod1](73)p_6=S_{Ccs1}+k_{tran5}[Ccs1Cu][Sod1] \tag*{(73)}
n6=kdeg3[Ccs1]+ktran2[GCu][Ccs1](74)n_6=k_{deg3}[Ccs1]+k_{tran2}[GCu][Ccs1] \tag*{(74)}
p7=ktran2[GCu][Ccs1](75)p_7=k_{tran2}[GCu][Ccs1] \tag*{(75)}
n7=ktran5[Ccs1Cu][Sod1](76)n_7=k_{tran5}[Ccs1Cu][Sod1] \tag*{(76)}
p8=SSod1(77)p_8=S_{Sod1} \tag*{(77)}
n8=kdeg4[Sod1]+ktran5[Ccs1Cu][Sod1](78)n_8=k_{deg4}[Sod1]+k_{tran5}[Ccs1Cu][Sod1] \tag*{(78)}
p9=SAtx1+ktran6[Atx1Cu][Ccc2](79)p_9=S_{Atx1}+k_{tran6}[Atx1Cu][Ccc2] \tag*{(79)}
n9=ktran3[GCu][Atx1]+kdeg5[Atx1](80)n_9=k_{tran3}[GCu][Atx1]+k_{deg5}[Atx1] \tag*{(80)}
p10=ktran3[Atx1][GCu](81)p_{10}=k_{tran3}[Atx1][GCu] \tag*{(81)}
n10=ktran6[Atx1Cu][Ccc2](82)n_{10}=k_{tran6}[Atx1Cu][Ccc2] \tag*{(82)}
p11=SCcc2(83)p_{11}=S_{Ccc2} \tag*{(83)}
n11=kdeg6[Ccc2]+ktran6[Atx1Cu][Ccc2](84)n_{11}=k_{deg6}[Ccc2]+k_{tran6}[Atx1Cu][Ccc2] \tag*{(84)}
p12=SAce1(85)p_{12}=S_{Ace1} \tag*{(85)}
n12=kdeg7[Ace1]+kmet[Ace1](86)n_{12}=k_{deg7}[Ace1]+k_{met}[Ace1] \tag*{(86)}
p13=kmet[Ace1](87)p_{13}=k_{met}[Ace1] \tag*{(87)}
n13=ktran7[Ace1Cu](88)n_{13}=k_{tran7}[Ace1Cu] \tag*{(88)}
p14=fact(89)p_{14}=f_{act} \tag*{(89)}
n14=kdeg,GFPmRNA[GFPmRNA](90)n_{14}=k_{deg,GFP_{mRNA}}[GFP_{mRNA}] \tag*{(90)}
p15=ktl,GFP[GFPmRNA](91)p_{15}=k_{tl,GFP}[GFP_{mRNA}] \tag*{(91)}
n15=kmat[GFPnascent](92)n_{15}=k_{mat}[GFP_{nascent}] \tag*{(92)}
p16=kmat[GFPnascent](93)p_{16}=k_{mat}[GFP_{nascent}] \tag*{(93)}
n16=kdeg,GFPmat[GFPmature](94)n_{16}=k_{deg,GFP_{mat}}[GFP_{mature}] \tag*{(94)}
pluxABmRNA=fact(95)p_{luxAB_{mRNA}}=f_{act} \tag*{(95)}
nluxABmRNA=kdeg,luxABmRNA[luxABmRNA](96)n_{luxAB_{mRNA}}=k_{deg,luxAB_{mRNA}}[luxAB_{mRNA}] \tag*{(96)}
pLuxAB=ktl,LuxAB[luxABmRNA](97)p_{LuxAB}=k_{tl,LuxAB}[luxAB_{mRNA}] \tag*{(97)}
nLuxAB=kdeg,LuxAB[LuxAB](98)n_{LuxAB}=k_{deg,LuxAB}[LuxAB] \tag*{(98)}
pFRP=0(99)p_{FRP}=0 \tag*{(99)}
nFRP=0(100)n_{FRP}=0 \tag*{(100)}
pRCOOH=fLuxABk4(101)p_{RCOOH}=f_{LuxAB}k_4 \tag*{(101)}
nRCOOH=fLuxE(102)n_{RCOOH}=f_{LuxE} \tag*{(102)}
pLuxERCO=fLuxE(103)p_{LuxERCO}=f_{LuxE} \tag*{(103)}
nLuxERCO=fLuxC,A(104)n_{LuxERCO}=f_{LuxC,A} \tag*{(104)}
pA=fLuxC,A(105)p_A=f_{LuxC,A} \tag*{(105)}
nA=fLuxABk4(106)n_A=f_{LuxAB}k_4 \tag*{(106)}
pF=fLuxABk4(107)p_F=f_{LuxAB}k_4 \tag*{(107)}
nF=fFRP,EX(108)n_F=f_{FRP,EX} \tag*{(108)}
pFH=fFRP,EX(109)p_{FH}=f_{FRP,EX} \tag*{(109)}
nFH=fLuxABk4(110)n_{FH}=f_{LuxAB}k_4 \tag*{(110)}
pN=kh(fFRP,EQ+fLuxC,N)(111)p_N=k_h(f_{FRP,EQ}+f_{LuxC,N}) \tag*{(111)}
nN=fFRP,EQ+fLuxC,N(112)n_N=f_{FRP,EQ}+f_{LuxC,N} \tag*{(112)}
kupt=kupt,max[Cu]KM,Cu+[Cu](113)k_{upt}=\frac{k_{upt,max}[Cu]}{K_{M,Cu}+[Cu]} \tag*{(113)}
kCu−deg=kdeg,max[Cu]2KM,deg2+[Cu]2(114)k_{Cu-deg}=\frac{k_{deg,max}[Cu]^2}{K_{M,deg}^2+[Cu]^2} \tag*{(114)}
kmet=kmet,max[GCu]2KD,Ace12+[GCu]2(115)k_{met}=\frac{k_{met,max}[GCu]^2}{K_{D,Ace1}^2+[GCu]^2} \tag*{(115)}
fact=ktr,max∑n=14(KD,mRNA[Ace1Cu])n+1(116)f_{act}=\frac{k_{tr,max}}{\sum_{n=1}^{4}\left(\frac{K_{D,mRNA}}{[Ace1Cu]}\right)^n+1} \tag*{(116)}
fLuxE=VLuxE+[RCOOH][ATP]DLuxE(117)f_{LuxE}=\frac{V^+_{LuxE}[RCOOH][ATP]}{D_{LuxE}} \tag*{(117)}
fLuxC,A=fLuxC,N=VLuxC+[LuxE−RCO][N]DLuxC(118)f_{LuxC,A}=f_{LuxC,N}=\frac{V^+_{LuxC}[LuxE-RCO][N]}{D_{LuxC}} \tag*{(118)}
fFRP,EX=fFRP,EQ=VFRP+[F][N]DFRP(119)f_{FRP,EX}=f_{FRP,EQ}=\frac{V^+_{FRP}[F][N]}{D_{FRP}} \tag*{(119)}
fLuxAB=k5k6k1kOk2k3[FH][A][LuxAB]DLux(120)f_{LuxAB}=\frac{k_5k_6k_1k_Ok_2k_3[FH][A][LuxAB]}{D_{Lux}} \tag*{(120)}
DFRP=KD,FKm,N+Km,N[F]+Km,F[N]+[F][N]D_{FRP}=K_{D,F}K_{m,N}+K_{m,N}[F]+K_{m,F}[N]+[F][N]
DLuxE=KD,RCOOHKm,ATP+Km,ATP[RCOOH]+Km,RCOOH[ATP]+[RCOOH][ATP]D_{LuxE}=K_{D,RCOOH}K_{m,ATP}+K_{m,ATP}[RCOOH]+K_{m,RCOOH}[ATP]+[RCOOH][ATP]
DLuxC=KD,LuxE−RCOKm,N2+Km,N2[LuxE−RCO]+Km,LuxE−RCO[N]+[LuxE−RCO][N]D_{LuxC}=K_{D,LuxE-RCO}K_{m,N2}+K_{m,N2}[LuxE-RCO]+K_{m,LuxE-RCO}[N]+[LuxE-RCO][N]
DLux=k2k3k4k5k6(kO+k−1)[A]+kOk1k4k5k6(k3+k−2)[FH]+k1k2(k3k4k5k6+kOk4k5k6+kOk3k5k6+kOk3k4k6+kOk3k4k5)[FH][A]\begin{aligned} D_{Lux}={}&k_2k_3k_4k_5k_6(k_O+k_{-1})[A]\\ &+k_Ok_1k_4k_5k_6(k_3+k_{-2})[FH]\\ &+k_1k_2(k_3k_4k_5k_6+k_Ok_4k_5k_6+k_Ok_3k_5k_6+k_Ok_3k_4k_6\\ &\quad+k_Ok_3k_4k_5)[FH][A] \end{aligned}
ParameterDescriptionValueReference
kupt,maxk_{\mathrm{upt,max}}Maximum Cu uptake rate0.0033 min−10.0033\ \mathrm{min}^{-1}approximated [32]
kMCuk_{M\mathrm{Cu}}Michaelis Constant for copper uptake4.4 μM4.4\ \mu\mathrm{M}[32]
kdeg,maxk_{\mathrm{deg,max}}Maximum Ctrs degradation dependent on Cu0.01 min−10.01\ \mathrm{min}^{-1}inferred [34]
kM,degk_{M,\mathrm{deg}}Half-maximum activation constant for copper-dependent degradation of Ctr proteins20 μM20\ \mu\mathrm{M}inferred [34]
kb1k_{b1}Binding rate of Ctr monomers10 μM−2min−110\ \mu\mathrm{M}^{-2}\mathrm{min}^{-1}assumed
kdis1k_{dis1}Unbinding rate of Ctr trimers1 min−11\ \mathrm{min}^{-1}assumed
kmet,maxk_{\mathrm{met,max}}Maximum Ace1 metalation rate due to Cu binding to Ace10.1 min−10.1\ \mathrm{min}^{-1}inferred [37]
kD,Ace1k_{D,\mathrm{Ace1}}Half maximum activation constant of Ace10.1 μM0.1\ \mu\mathrm{M}[37]
ktr,maxk_{\mathrm{tr,max}}Maximum GFP transcription rate from CUP1CUP1 promoters2.5 min−12.5\ \mathrm{min}^{-1}approximated [56]
kD,mRNAk_{D,\mathrm{mRNA}}Dissociation constant for Ace1 binding to the MREs of the DNA promoter10−9 μM10^{-9}\ \mu\mathrm{M}assumed
ktran1k_{\mathrm{tran1}}Cu handoff rate between Ctr1/3 and GSH0.4 μM−1min−10.4\ \mu\mathrm{M}^{-1}\mathrm{min}^{-1}approximated [35]
ktran2k_{\mathrm{tran2}}Cu handoff between GSH and Ccs10.4 μM−1min−10.4\ \mu\mathrm{M}^{-1}\mathrm{min}^{-1}approximated [35]
ktran3k_{\mathrm{tran3}}Cu handoff rate between GSH and Atx10.4 μM−1min−10.4\ \mu\mathrm{M}^{-1}\mathrm{min}^{-1}approximated [35]
ktran4k_{\mathrm{tran4}}Cu handoff rate between GSH and Ace10.4 μM−1min−10.4\ \mu\mathrm{M}^{-1}\mathrm{min}^{-1}approximated [35]
ktran5k_{\mathrm{tran5}}Cu handoff rate between Ccs1 and Sod10.4 μM−1min−10.4\ \mu\mathrm{M}^{-1}\mathrm{min}^{-1}approximated [35]
ktran6k_{\mathrm{tran6}}Cu handoff rate between Atx1 and Ccc20.4 μM−1min−10.4\ \mu\mathrm{M}^{-1}\mathrm{min}^{-1}approximated [35]
ktran7k_{\mathrm{tran7}}Transport rate of metalated Ace1 towards the DNA promoter0.4 min−10.4\ \mathrm{min}^{-1}assumed
kdeg1k_{\mathrm{deg1}}Ctr1/3 monomer degradation rate0.077 min−10.077\ \mathrm{min}^{-1}approximated [34, 57, 58]
kdeg2k_{\mathrm{deg2}}GSH degradation rate0.0012 min−10.0012\ \mathrm{min}^{-1}approximated [57, 58]
kdeg3k_{\mathrm{deg3}}Ccs1 degradation rate0.0014 min−10.0014\ \mathrm{min}^{-1}[59]
kdeg4k_{\mathrm{deg4}}Sod1 degradation rate0.001 min−10.001\ \mathrm{min}^{-1}[59]
kdeg5k_{\mathrm{deg5}}Atx1 degradation rate0.0011 min−10.0011\ \mathrm{min}^{-1}[59]
kdeg6k_{\mathrm{deg6}}Ccc2 degradation rate0.006 min−10.006\ \mathrm{min}^{-1}[60]
kdeg7k_{\mathrm{deg7}}Ace1 degradation rate0.04 min−10.04\ \mathrm{min}^{-1}inferred [61]
kdeg,GFPmRNAk_{\mathrm{deg,GFPmRNA}}GFP mRNA degradation rate0.12 min−10.12\ \mathrm{min}^{-1}[56, 62]

Table IX. Model Parameters.

ParameterDescriptionValueReference
ktl,GFPk_{\mathrm{tl,GFP}}Average translation rate in yeast1.76 min−11.76\ \mathrm{min}^{-1}[63]
kmatk_{\mathrm{mat}}GFP maturation rate0.0058 min−10.0058\ \mathrm{min}^{-1}approximated from [64]
kdeg,GFPmatk_{\mathrm{deg,GFPmat}}mature GFP degradation rate0.0022 min−10.0022\ \mathrm{min}^{-1}assumed (long half-life)
SCtr1/3S_{\mathrm{Ctr1/3}}Constitutive Ctr1/3 synthesis rate0.0001 μM min−10.0001\ \mu\mathrm{M}\ \mathrm{min}^{-1}assumed
SGSHS_{\mathrm{GSH}}Constitutive GSH synthesis rate0.004 μM min−10.004\ \mu\mathrm{M}\ \mathrm{min}^{-1}assumed
SCcs1S_{\mathrm{Ccs1}}Constitutive Ccs1 synthesis rate0.0001 μM min−10.0001\ \mu\mathrm{M}\ \mathrm{min}^{-1}assumed
SAtx1S_{\mathrm{Atx1}}Constitutive Atx1 synthesis rate0.001 μM min−10.001\ \mu\mathrm{M}\ \mathrm{min}^{-1}assumed
SAce1S_{\mathrm{Ace1}}Constitutive Ace1 synthesis rate0.0003 μM min−10.0003\ \mu\mathrm{M}\ \mathrm{min}^{-1}assumed
SSod1S_{\mathrm{Sod1}}Constitutive Sod1 synthesis rate0.0001 μM min−10.0001\ \mu\mathrm{M}\ \mathrm{min}^{-1}assumed
SCcc2S_{\mathrm{Ccc2}}Constitutive Ccc2 synthesis rate0.00001 μM min−10.00001\ \mu\mathrm{M}\ \mathrm{min}^{-1}assumed
kdeg,luxABmRNAk_{\mathrm{deg,luxABmRNA}}luxAB mRNA degradation rate0.09 min−10.09\ \mathrm{min}^{-1}assumed
ktl,LuxABk_{\mathrm{tl,LuxAB}}average translation rate in yeast1.76 min−11.76\ \mathrm{min}^{-1}[63]
kdeg,LuxABk_{\mathrm{deg,LuxAB}}degradation rate of LuxAB0.003 min−10.003\ \mathrm{min}^{-1}[53]
VLuxE+V^{+}_{\mathrm{LuxE}}forward catalysis rate for LuxE1 μM min−11\ \mu\mathrm{M}\ \mathrm{min}^{-1}approximated [65]
VLuxC+V^{+}_{\mathrm{LuxC}}forward catalysis rate for LuxC1 μM min−11\ \mu\mathrm{M}\ \mathrm{min}^{-1}[65]
KD,RCOOHK_{D,\mathrm{RCOOH}}Dissociation constant of fatty acids0.01 μM0.01\ \mu\mathrm{M}assumed
[ATP][\mathrm{ATP}]ATP concentration2500 μM2500\ \mu\mathrm{M}[66]
Km,ATPK_{m,\mathrm{ATP}}Michaelis constant for ATP0.002 μM0.002\ \mu\mathrm{M}[65]
Km,RCOOHK_{m,\mathrm{RCOOH}}Michaelis constant for fatty acids0.04342 μM0.04342\ \mu\mathrm{M}[65]
Km,LuxE−RCOK_{m,\mathrm{LuxE-RCO}}Michaelis constant for the acyl-LuxE intermediate complex0.2 μM0.2\ \mu\mathrm{M}assumed
KD,LuxE−RCOK_{D,\mathrm{LuxE-RCO}}Dissociation constant for the acyl-LuxE intermediate complex0.001 μM0.001\ \mu\mathrm{M}assumed
Km,N2K_{m,\mathrm{N2}}Michaelis constant for NADPH (LuxC reaction)5 μM5\ \mu\mathrm{M}[65]
Km,NK_{m,\mathrm{N}}Michaelis constant for NADPH (FRP reaction)9 μM9\ \mu\mathrm{M}[67]
Km,FK_{m,\mathrm{F}}Michaelis constant for FMN0.1117 μM0.1117\ \mu\mathrm{M}calculated [68, 40]
KD,FK_{D,\mathrm{F}}Dissociation constant of FMN0.017 μM0.017\ \mu\mathrm{M}[68]
kcatk_{\mathrm{cat}}Constant rate for FRP catalysis260 min−1260\ \mathrm{min}^{-1}calculated [68, 40]
khk_hConstant rate for re-formation of NADP to NADPH after re-cycling11assumed
KEK_EEquilibrium constant for protonation deprotonation1.6⋅10−8 μM1.6 \cdot10^{-8}\ \mu\mathrm{M}[69]
[H][\mathrm{H}]hydrogen concentration0.16 μM0.16\ \mu\mathrm{M}[42], BNID 107762
kOk_Oreaction rate constant of oxygen21000 min−121000\ \mathrm{min}^{-1}[31]
k1k_1binding rate of FMNH2_21020 μM−1min−11020\ \mu\mathrm{M}^{-1}\mathrm{min}^{-1}[70]

Table X. Model Parameters (continued).

k−1k_{-1}unbinding rate of FMNH2_27200 min−17200\ min^{-1}[70]
kbAk_{bA}binding rate of aldehyde600 μM−1 min−1600\ \mu M^{-1}\ min^{-1}[70]
k−2k_{-2}Unbinding rate for aldehyde18000 min−118000\ min^{-1}[70]
k3k_3hydroxyflavin formation rate15 min−115\ min^{-1}approximated [31]
k4k_4hydroxyflavin de-excitation and de-hydration rate0.0001 min−10.0001\ min^{-1}approximated [31, 70]
k5k_5fatty acid RCOOH release rate46 min−146\ min^{-1}[70]
k6k_6FMN release rate36 min−136\ min^{-1}inferred [70]
kLk_Lphoton emission rate144 min−1144\ min^{-1}[53]
RpdR_{pd}photodiode responsivity0.185 A/W0.185\ A/W (at λ=490 nm\lambda= 490\ nm)[43]

Table XI. Model Parameters (continued).

Chemical CompoundInitial Concentration (μM\mu M)Reference
Ctr1/30.1[59]
Ctr1/3trim_{\mathrm{trim}}0.03assumed
GSH300[71]
Ccs10.25approximated [59]
Sod12.2approximated [59]
Atx10.22approximated [59]
Ccc20.034approximated [59]
Ace10.0428approximated [59]
FRP1000assumed
RCOOH8assumed
F500saturating concentration [67, 68]
N500saturating concentration [67, 68]

Table XII. Initial values of the various chemical compounds within the Copper Uptake Pathway.

Acknowledgments

This research has received support from the projects BRIDGE2HORIZON/0823D/004 (YeasPSense) and EX-CELLENCE/0524/0595 (YeCIE) which is implemented under the Cohesion Policy Funds “THALIA 2021-2027” with EU co-funding.

References

References

  1. [1]M. M. Al-Zubi and M.-S. Alouini, “Macroscale molecular communication in iot-based pipeline inspection and monitoring applications: Preliminary experiment and mathematical model,” IEEE Open Journal of the Communications Society, vol. 6, pp. 7541–7554, 2025. [Online]. Available: https://api.semanticscholar.org/CorpusID:27781076arxiv.org/abs/2504.09325
  2. [2]P. K. Bulasara, S. R. Sahoo, N. Gupta, Z. Han, and N. Kumar, “The internet of bio-nano things with insulin-glucose, security and research challenges: A survey,” ACM Computing Surveys, vol. 57, pp. 1–42, 2024. [Online]. Available: https://api.semanticscholar.org/CorpusID:27454414
  3. [3]L. Felicetti, M. Femminella, and G. Reali, “Molecular communications in blood vessels: Models, analysis, and enabling technologies,” Communications of the ACM, vol. 68, pp. 60–69, 2025. [Online]. Available: https://api.semanticscholar.org/CorpusID:276323946DOI
  4. [4]I. R. R. Kamal, S. M. A. El-atty, S. F. El-Zohgdy, and R. F. Soliman, “Intelligent deep learning model for targeted cancer drug delivery,” Scientific Reports, vol. 15, 2025. [Online]. Available: https://api.semanticscholar.org/CorpusID:279029667DOI
  5. [5]L. Felicetti, M. Femminella, G. Reali, P. Gresele, M. Malvestiti, and J. N. Daigle, “Modeling cd40-based molecular communications in blood vessels,” IEEE Transactions on NanoBioscience, vol. 13, pp. 230–243, 2014. [Online]. Available: https://api.semanticscholar.org/CorpusID:10687659
  6. [6]T. Nakano, Y.-H. Hsu, W. C. Tang, T. Suda, D. Lin, T. Koujin, T. Haraguchi, and Y. Hiraoka, “Microplatform for intercellular communication,” 2008 3rd IEEE International Conference on Nano/Micro Engineered and Molecular Systems, pp. 476–479, 2008. [Online]. Available: https://api.semanticscholar.org/CorpusID:3294201DOI
  7. [7]N. Netsikas, S. Kyriakoudi, A. Kirmizis, B. D. Unluturk, A. Pitsillides, I. F. Akyildiz, and M. Lestas, “Engineering yeast cells to facilitate information exchange,” IEEE Transactions on Molecular, Biological and Multi-Scale Communications, vol. 10, pp. 2–20, 2024. [Online]. Available: https://api.semanticscholar.org/CorpusID:267211917
  8. [8]N. Netsikas, S. Kyriakoudi, A. Kirmizis, I. Krikidis, I. F. Akyildiz, and M. Lestas, “Noise characterization and robust signal detection in yeast pheromone molecular communication,” IEEE Transactions on Molecular, Biological, and Multi-Scale Communications, vol. 11, pp. 218–227, 2025. [Online]. Available: https://api.semanticscholar.org/CorpusID:277933680
  9. [9]F. Anderl, G. Salvadori, M. Veletić, F. C. Petersen, and I. Balasingham, “Quantitative aspects, engineering and optimization of bacterial sensor systems,” IEEE Transactions on Molecular, Biological, and Multi-Scale Communications, vol. 10, pp. 517–533, 2024. [Online]. Available: https://api.semanticscholar.org/CorpusID:270711191
  10. [10]M. E. Inda-Webb, M. Jimenez, Q. Liu, N. V. Phan, J. Ahn, C. Steigeer, A. J. Wentworth, A. Riaz, T. Zirtiglou, K. fu Katherine Wong, K. Ishida, N. J. Fabian, J. Jenkins, J. Kuosmanen, W. A. M. Madani, R. McNally, Y. Lai, A. M. Hayward, M. Mimee, P. M. Nadeau, A. P. Chandrakasan, G. Traverso, R. T. Yazicigil, and T. K. Lu, “Sub-1.4 cm3 capsule for detecting labile inflammatory biomarkers in situ,” Nature, vol. 620, pp. 386–392, 2023. [Online]. Available: https://api.semanticscholar.org/CorpusID:260202114
  11. [11]S. Ling, C. Xu, Y. Jiang, F. Xu, H. Jin, H. Chen, S. Chen, Z. Ding, and D. Sun, “Toward continuous monitoring systems: emerging trends of on-chip sensors in organ-on-a-chip,” Microsystems & Nanoengineering, vol. 12, 2026. [Online]. Available: https://api.semanticscholar.org/CorpusID:285543398DOI
  12. [12]C.-Y. Xu, J. Mi, J.-F. Zhao, Y.-L. Zhu, Y.-H. Wu, C.-R. Zhang, Z. Zou, B. Ye, and Y. Zhou, “Magnetic hydrogel: Enhanced bacterial biosensor for speedy gut disease detection.” ACS sensors, 2025. [Online]. Available: https://api.semanticscholar.org/CorpusID:283107712DOI
  13. [13]E. D. la Paz, N. H. Maganti, A. Trifonov, I. Jeerapan, K. Mahato, L. Yin, T. Sonsa-ard, N. Ma, W. Jung, R. Burns, A. Zarrimpar, J. Wang, and P. P. Mercier, “A self-powered ingestible wireless biosensing system for real-time in situ monitoring of gastrointestinal tract metabolites,” Nature Communications, vol. 13, 2022. [Online]. Available: https://api.semanticscholar.org/CorpusID:254100372DOI
  14. [14]Z. Zhuang, L. L. Huang, B. C. Seo, S. Pandey, J. M. Karp, Y. Lee, and C. L. Maiakawa, “A radically simple, ingestible colorimetric biosensor pill for cost-effective, non-invasive monitoring of intestinal inflammation,” Device, vol. 3, 2025. [Online]. Available: https://api.semanticscholar.org/CorpusID:280507202DOI
  15. [15]J. Min, H. Ahn, H. Lukas, X. Ma, R. Bhansali, S. Sunwoo, C. Wang, Y. Xu, D. R. Yao, G. Kim, Z. Li, T. K. Hsiai, A. Emami, H. Jung, and W. Gao, “Continuous biochemical profiling of the gastrointestinal tract using an integrated smart capsule,” Nature Electronics, vol. 8, pp. 844–855, 2025. [Online]. Available: https://api.semanticscholar.org/CorpusID:281088776
  16. [16]A. Abdigazy, M. Arfan, J. Shao, M. S. Islam, M. F. Hassan, and Y. Khan, “3d gas mapping in the gut with ai-enabled ingestible and wearable electronics,” Cell Reports Physical Science, 2024. [Online]. Available: https://api.semanticscholar.org/CorpusID: 270457015
  17. [17]A. Amerizadeh, A. Mashhadian, M. Farahnak-Ghazani, H. Arjmandi, M. A. Rad, A. Shamloo, M. Vosoughi, and M. Nasiri-Kenari, “Bacterial receiver prototype for molecular communication using rhamnose operon in a microfluidic environment,” IEEE Transactions on NanoBioscience, vol. 20, pp. 426–435, 2021. [Online]. Available: https://api.semanticscholar.org/CorpusID:235597581
  18. [18]J. L. Terrell, T. Tschirhart, J. P. Jahnke, K. T. Stephens, Y. Liu, H. Dong, M. M. Hurley, M. Pozo, R. Mckay, C.-Y. Tsao, H.-C. Wu, G. J. Vora, G. F. Payne, D. N. Stratis-Cullum, and W. E. Bentley, “Bioelectronic control of a microbial community using surface-assembled electro genetic cells to route signals,” Nature Nanotechnology, vol. 16, pp. 688 – 697, 2021. [Online]. Available: https://api.semanticscholar.org/CorpusID:232415045DOI
  19. [19]G. Salvadori, F. Anderl, M. Veletic, F. Petersen, and I. Blasingham, “Synergistic interplay of quorum sensing molecules in streptococcus pneumoniae: Promising prospects for quorum-sensing based bionanosensor design.”
  20. [20]E. Wahid, O. B. Ocheja, E. Marsili, C. Guaragnella, and N. Guaragnella, “Biological and technical challenges for implementation of yeast-based biosensors,” Microbial Biotechnology, vol. 16, pp. 54 – 66, 2022. [Online]. Available: https://api.semanticscholar.org/CorpusID:253800447DOI
  21. [21]R. Bisquert, A. Guillén, S. Muñiz-Calvo, and J. M. Guillamón, “Engineering a gprc-based yeast biosensor for a highly sensitive melatonin detection from fermented beverages,” Scientific Reports, vol. 14, 2024. [Online]. Available: https://api.semanticscholar.org/CorpusID:271644221DOI
  22. [22]A. C. Lind, F. David, and V. Swiewers, “Evaluation and comparison of colorimetric outputs for yeast-based biosensors in laboratory and point-of-use settings,” FEMS Microbiology Letters, vol. 371, 2024. [Online]. Available: https://api.semanticscholar.org/CorpusID:269986415DOI
  23. [23]N. Ostrov, M. Jimenez, S. Billerbeck, J. Brisbois, J. Matragrano, A. Ager, and V. W. Cornish, “A modular yeast biosensor for low-cost point-of-care pathogen detection,” Science Advances, vol. 3, 2017. [Online]. Available: https://api.semanticscholar.org/CorpusID:10053602
  24. [24]L. C. Dacquay, B. Samman, M. P. Newman, D. J. Philpott, and D. R. McMillen, “Engineering a colorimetric yeast-based assay to detect small molecules in the gastrointestinal system,” Frontiers in Synthetic Biology, 2024. [Online]. Available: https://api.semanticscholar.org/CorpusID:270809521
  25. [25]F. Mendes, E. Miranda, L. Amaral, C. Carvalho, B. B. Castro, M. J. Sousa, and S. R. Chaves, “Novel yeast-based biosensor for environmental monitoring of tebucanazole,” Applied Microbiology and Biotechnology, vol. 108, 2024. [Online]. Available: https://api.semanticscholar.org/CorpusID:266744147DOI
  26. [26]E. Wahid, O. B. Ocheja, A. Longo, E. Marsili, M. Trotta, M. Grattieri, C. Guaragnella, and N. Guaragnella, “A bluetooh-enabled electrochemical platform based on saccharomyces cerevisiae yeast cells for copper detection,” Biosensors, vol. 15, 2025. [Online]. Available: https://api.semanticscholar.org/CorpusID:281237455DOI
  27. [27]N. Lobsiger, J. Venetz, M. Gregorini, M. Christen, B. Christen, and W. J. Stark, “T yestrosens, a field-portable s. cerevisiae biosensor device for the detection of endocrine-disrupting chemicals: Reliability and stability,” Biosensors & bioelectronics, vol. 146, p. 111710, 2019. [Online]. Available: https://api.semanticscholar.org/CorpusID:203943151DOI
  28. [28]H. Shi, Y. Jiang, Y. Yang, Y. Peng, and C. Li, “Copper metabolism in saccharomyces cerevisiae: an update,” BioMetals, vol. 34, pp. 3 – 14, 2020.DOI
  29. [29]A. E. Masser, G. Kandasamy, J. M. Kaimal, and C. Andréasson, “Luciferase nanoluc as a reporter for gene expression and protein levels in saccharomyces cerevisiae,” Yeast (Chichester, England), vol. 33, pp. 191 – 200, 2016. [Online]. Available: https://api.semanticscholar.org/CorpusID:13391DOI
  30. [30]R. K. Gupta, S. S. Patterson, S. Ripp, M. L. Simpson, and G. S. Sayler, “Expression of the photorhabdus luminescens lux genes (luxa, b, c, d, and e) in saccharomyces cerevisiae,” FEMS yeast research, vol. 4 3, pp. 305–313, 2003. [Online]. Available: https://api.semanticscholar.org/CorpusID:16729312pubmed.ncbi.nlm.nih.gov/14654435
  31. [31]R. Tinikul, P. Chuntaboon, J. Phonbupha, and T. Paladkong, “Chapter fourteen - bacterial luciferase: Molecular mechanisms and applications,” in Flavin-Dependent Enzymes: Mechanisms, Structures and Applications, ser. The Enzymes, P. Chaivien and F. Tamanoi, Eds. Academic Press, 2020, vol. 47, pp. 427–455. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S1874604720300172DOI
  32. [32]C.-M. Lin and D. J. Kosman, “Copper uptake in wild type and copper metallothionein-deficient saccharomyces cerevisiae kinetics and mechanism.” Journal of Biological Chemistry, vol. 265, no. 16, pp. 9194–9200, 1990.pubmed.ncbi.nlm.nih.gov/2188974
  33. [33]X. Wu, D. Sinani, H. Kim, and J. Lee, “Copper transport activity of yeast ctr1 is down-regulated via its c terminus in response to excess copper*,” Journal of Biological Chemistry, vol. 284, pp. 4112 – 4122, 2009. [Online]. Available: https://api.semanticscholar.org/CorpusID:10649691DOI
  34. [34]M. M. O. Peña, S. T. Puig, and D. J. Thiele, “Characterization of the saccharomyces cerevisiae high affinity copper transporter ctr3*,” The Journal of Biological Chemistry, vol. 275, pp. 33 244 – 33 251, 2000. [Online]. Available: https://api.semanticscholar.org/CorpusID:32045123pubmed.ncbi.nlm.nih.gov/10924521
  35. [35]K. N. Chacón, J. Perkins, Z. Mathe, K. B. Alwan, E. N. Ho, M. N. Ucisik, K. M. Merz, and N. J. Blackburn, “Trapping intermediates in metal transfer reactions of the cuscaf export pump of escherichia coli,” Communications Biology, vol. 1, 2018. [Online]. Available: https://api.semanticscholar.org/CorpusID:53306729DOI
  36. [36]Y. Hatori and S. Lutsenko, “The role of copper chaperone atox1 in coupling redox homeostasis to intracellular copper distribution,” Antioxidants, vol. 5, 2016. [Online]. Available: https://api.semanticscholar.org/CorpusID:17515278DOI
  37. [37]P. Fürst and D. Hamer, “Cooperative activation of a eukaryotic transcription factor: interaction between cu (i) and yeast ace1 protein.” Proceedings of the National Academy of Sciences, vol. 86, no. 14, pp. 5267–5271, 1989.pubmed.ncbi.nlm.nih.gov/2664778
  38. [38]M. M. O. Peña, K. A. Koch, and D. J. Thiele, “Dynamic regulation of copper uptake and detoxification genes in saccharomyces cerevisiae,” Molecular and cellular biology, vol. 18, no. 5, pp. 2514–2523, 1998.pubmed.ncbi.nlm.nih.gov/9599102
  39. [39]T. M. Mitre, M. C. Mackey, and A. Khadra, “Mathematical model of galactose regulation and metabolic consumption in yeast,” Journal of theoretical biology, vol. 407, pp. 238–258, 2016. [Online]. Available: https://api.semanticscholar.org/CorpusID:7238466arxiv.org/abs/1602.03862
  40. [40]A. Cornish-Bowden, Fundamentals of enzyme kinetics. John Wiley & Sons, 2013.
  41. [41]H. L. Segal, “Enzyme kinetics. behavior and analysis of rapid equilibrium and steady-state enzyme systems,” 1976.DOI
  42. [42]R. Milo, P. Jorgensen, U. Moran, G. M. Weber, and M. Springer, “Binumbers—the database of key numbers in molecular and cell biology,” Nucleic Acids Research, vol. 38, pp. D750 – D753, 2009. [Online]. Available: https://api.semanticscholar.org/CorpusID:1678494DOI
  43. [43]G. Koklu, R. Etienne-Cummings, Y. Leblebici, G. D. Micheli, and S. Carrara, “Characterization of standard cmos compatible photodiodes and pixels for lab-on-chip devices,” 2013 IEEE International Symposium on Circuits and Systems (ISCAS2013), pp. 1075–1078, 2013. [Online]. Available: https://api.semanticscholar.org/CorpusID:6337273DOI
  44. [44]H. Tohoyama, M. Inouhe, M. Joho, and T. Murayama, “Production of metallothionein in copper- and cadmium-resistant strains of saccharomyces cerevisiae,” Journal of Industrial Microbiology, vol. 14, pp. 126–131, 1995. [Online]. Available: https://api.semanticscholar.org/CorpusID:22957597DOI
  45. [45]B.-E. Kim, T. Nevitt, and D. J. Thiele, “Mechanisms for copper acquisition, distribution and regulation.” Nature chemical biology, vol. 4 3, pp. 176–85, 2008. [Online]. Available: https://api.semanticscholar.org/CorpusID:15864537DOI
  46. [46]T. R. Butt, E. J. Sternberg, J. A. Gorman, P. E. Clark, D. Hamer, M. Rosenberg, and S. T. Crooke, “Copper metallothionein of yeast, structure of the gene, and regulation of expression.” Proceedings of the National Academy of Sciences of the United States of America, vol. 81 11, pp. 3332–6, 1984. [Online]. Available: https://api.semanticscholar.org/CorpusID:5735684pubmed.ncbi.nlm.nih.gov/6374656
  47. [47]D. J. Thiele, “Acel regulates expression of the saccharomyces cerevisiae metallothionein gene,” Molecular and cellular biology, vol. 8, no. 7, pp. 2745–2752, 1988.
  48. [48]R. J. D. R. R. A. Woods, “Transformation of yeast by lithium acetate/single-stranded carrier dna/polyethylene glycol method.” Methods in enzymology, vol. 350, pp. 87–96, 2002. [Online]. Available: https://api.semanticscholar.org/CorpusID:23912282
  49. [49]P. Bheda, D. Aguilar-Gómez, N. B. Becker, J. Becker, E. Stavrou, I. Kukhtevich, T. Höfer, S. J. Maerkl, G. Charvin, C. Marr, A. Kirmizis, and R. Schneider, “Single-cell tracing dissects regulation of maintenance and inheritance of transcriptional reinduction memory.” Molecular cell, 2020. [Online]. Available: https://api.semanticscholar.org/CorpusID:218600292DOI
  50. [50]M. E. Schmitt, T. A. Brown, and B. L. Trumpower, “A rapid and simple method for preparation of rna from saccharomyces cerevisiae.” Nucleic acids research, vol. 18 10, pp. 3091–2, 1990. [Online]. Available: https://api.semanticscholar.org/CorpusID:26807437pubmed.ncbi.nlm.nih.gov/2190191
  51. [51]D. Guet, L. T. Burns, S. K. Maji, J. Boulanger, P. Hersen, S. R. Wente, J. Salamero, and C. Dargemont, “Combining spinach-tagged rna and gene localization to image gene expression in live yeast,” Nature Communications, vol. 6, 2015. [Online]. Available: https://api.semanticscholar.org/CorpusID:3545627DOI
  52. [52]I. Kukhtevich, M. Rivero-Romano, N. Rakesh, P. Bheda, Y. Chadha, P. Rosales-Becerra, S. Hamperl, D. Bureik, S. Dornauer, C. Dargemont, A. Kirmizis, K. M. Schmoller, and R. Schneider, “Quantitative rna imaging in single live cells reveals age-dependent asymmetric inheritance.” Cell reports, vol. 41 7, p. 111656, 2022. [Online]. Available: https://api.semanticscholar.org/CorpusID:253578911DOI
  53. [53]R. Tinikul, K. Thotsaporn, W. Thaveekarn, S. Jitrapakdee, and P. Chaiyen, “The fusion vibrio campbellii luciferase as a eukaryotic gene reporter.” Journal of biotechnology, vol. 162 2-3, pp. 346–53, 2012. [Online]. Available: https://api.semanticscholar.org/CorpusID:19276367DOI
  54. [54]A. Mazo-Vargas, H. Park, M. Aydin, and N. E. Buchler, “Measuring fast gene dynamics in single cells with time-lapse luminescence microscopy,” Molecular Biology of the Cell, vol. 25, pp. 3699 – 3708, 2014. [Online]. Available: https://api.semanticscholar.org/CorpusID:8700797
  55. [55]T. O. Baldwin, J. A. Christopher, F. M. Raushel, J. F. Sinclair, M. M. Ziegler, A. J. Fisher, and I. Rayment, “Structure of bacterial luciferase.” Current opinion in structural biology, vol. 5 6, pp. 798–809, 1995. [Online]. Available: https://api.semanticscholar.org/CorpusID:483991pubmed.ncbi.nlm.nih.gov/8749369
  56. [56]G. Mehta, D. A. Ball, P. R. Eriksson, R. V. Chereji, D. J. Clark, J. G. McNally, and T. S. Karpova, “Single-molecule analysis reveals linked cycles of rsc chromatin remodeling and ace1p transcription factor binding in yeast.” Molecular cell, vol. 72 5, pp. 875–887.e9, 2018. [Online]. Available: https://api.semanticscholar.org/CorpusID:52978523DOI
  57. [57]A. Belle, A. Tanay, L. Bitincka, R. Shamir, and E. K. O’Shea, “Quantification of protein half-lives in the budding yeast proteome,” Proceedings of the National Academy of Sciences, vol. 103, pp. 13004 – 13009, 2006. [Online]. Available: https://api.semanticscholar.org/CorpusID:17808863pubmed.ncbi.nlm.nih.gov/16916930
  58. [58]R. Christiano, N. Nagaraj, F. Fröhlich, and T. C. Walther, “Global proteome turnover analyses of the yeasts s. cerevisiae and s. pombe.” Cell reports, vol. 9 5, pp. 1959–1965, 2014. [Online]. Available: https://api.semanticscholar.org/CorpusID:491352DOI
  59. [59]S. R. Engel, S. A. Aleksander, R. S. Nash, E. D. Wong, S. Weng, S. R. Miyasato, G. Sherlock, and J. M. Cherry, “Saccharomyces genome database: Advances in genome annotation, expanded biochemical pathways, and other key enhancements,” bioRxiv, 2024. [Online]. Available: https://api.semanticscholar.org/CorpusID:272820048DOI
  60. [60]M. Martín-Pérez and J. Villén, “Feasibility of protein turnover studies in prototroph saccharomyces cerevisiae strains.” Analytical chemistry, vol. 87 7, pp. 4008–14, 2015. [Online]. Available: https://api.semanticscholar.org/CorpusID:41958300
  61. [61]D. Das, S. Dey, R. C. Brewster, and S. Choubey, “Effect of transcription factor resource sharing on gene expression noise,” PLoS Computational Biology, vol. 13, 2017. [Online]. Available: https://api.semanticscholar.org/CorpusID:3734698DOI
  62. [62]M. M. C. Hyde, L. Block-Alper, J. A. Felix, P. Webster, and D. I. Meyer, “Induction of secretory pathway components in yeast is associated with increased stability of their mrna,” The Journal of Cell Biology, vol. 156, pp. 993 – 1001, 2002. [Online]. Available: https://api.semanticscholar.org/CorpusID:2570723pubmed.ncbi.nlm.nih.gov/11901166
  63. [63]B. Futcher, G. I. Latter, P. J. Monardo, C. S. McLaughlin, and J. I. Garries, “A sampling of the yeast proteome,” Molecular and Cellular Biology, vol. 19, pp. 7357 – 7368, 1999. [Online]. Available: https://api.semanticscholar.org/CorpusID:4977099pubmed.ncbi.nlm.nih.gov/10523624
  64. [64]R. Heim, A. B. Cubitt, and R. Tsien, “Improved green fluorescence,” Nature, vol. 373, pp. 663–664, 1995. [Online]. Available: https://api.semanticscholar.org/CorpusID:40179694pubmed.ncbi.nlm.nih.gov/7854443
  65. [65]P. A. Welham and D. J. Stekel, “University of birmingham mathematical model of the lux luminescence system in the terrestrial bacterium photorhabdus luminescens,” 2008. [Online]. Available: https://api.semanticscholar.org/CorpusID:21574557DOI
  66. [66]V. C. Özalp, T. R. Pedersen, L. J. Nielsen, and L. F. Olsen, “Time-resolved measurements of intracellular atp in the yeast saccharomyces cerevisiae using a new type of nanobiosensor*.” The Journal of Biological Chemistry, vol. 285, pp. 37 579 – 37 588, 2010. [Online]. Available: https://api.semanticscholar.org/CorpusID:3461932DOI
  67. [67]H. Chung and S. C. Tu, “Structure-function relationship of vibrio harveyi nadph-flavin oxidoreductase frp: essential residues lys167 and arg15 for nadph binding.” Biochemistry, vol. 51 24, pp. 4880–7, 2012. [Online]. Available: https://api.semanticscholar.org/CorpusID:5141215DOI
  68. [68]J. M. Musila, D. L. Forbes, and H. R. Ellis, “Functional evaluation of the π-helix in the nad(p)h:fmn reductase of the alkanesulfonate monooxygenase system.” Biochemistry, vol. 57 30, pp. 4469–4477, 2018. [Online]. Available: https://api.semanticscholar.org/CorpusID:49701052DOI
  69. [69]R. Tinikul, N. Lawan, N. Akarethatanap, P. Pimpiriyakul, W. Chaiantunya, C. Suadee, J. Sucharitakul, P. Chenprakhon, D. P. Ballou, B. Entsch, and P. Chaiyen, “Protonation status and control mechanism of flavin–oxygen intermediates in the reaction of bacterial luciferase,” The FEBS Journal, vol. 288, 2020. [Online]. Available: https://api.semanticscholar.org/CorpusID:227949077DOI
  70. [70]H. M. Abu-Soud, L. S. Mullins, T. O. Baldwin, and F. M. Raushel, “Stopped-flow kinetic analysis of the bacterial luciferase reaction.” Biochemistry, vol. 31 15, pp. 3807–13, 1992. [Online]. Available: https://api.semanticscholar.org/CorpusID:9454005
  71. [71]T. Tahmasebi, R. Nosrati, H. Zare, H. Saderi, R. Moradi, and P. Owlia, “Isolation of indigenous glutathione producing saccharomyces cerevisiae strains,” Iranian Journal of Pathology, vol. 11, pp. 354 – 362, 2016. [Online]. Available: https://api.semanticscholar.org/CorpusID:23055183pubmed.ncbi.nlm.nih.gov/28855927

Paper details

vlnc:W_9qsjhw5y
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY-SA 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

Not recorded for this paper

View statement
Looking for related papers…
Contents