Search for: All records

Nonequilibrium error-correction mechanisms, such as kinetic proofreading, enable biological systems to amplify the discrimination among cognate and non-cognate substrates beyond what is possible at equilibrium. However, it remains unclear how such discrimination should be distributed over the underlying network to achieve the full nonequilibrium advantage of error reduction. Using a discrete-state stochastic framework, we first show that the Hopfield network, the seminal model of proofreading, with discrimination concentrated in two dissociation steps, displays distinct regimes of error reduction depending on the relative magnitudes of the rate constants of various steps. One such regime is actually anti-proofreading, showing no improvement in accuracy with increasing discrimination. In contrast, a biologically realistic model of the tRNA selection network in protein translation by the E. coli ribosome exhibits distributed discrimination. We demonstrate that the spread of discrimination across the entire network enables the system to achieve the full nonequilibrium advantage, even in the kinetic regime where the Hopfield network lies in the anti-proofreading zone at low discrimination strength. Our results further indicate that excessively strong discrimination adversely affects the system, eliminating the nonequilibrium advantage of error reduction without providing any additional gain in the speed of translation. 

Chemical inhibitors bind to enzymes, thereby inhibiting their catalytic activity. While many enzymes catalyze reactions with a single substrate, others, like DNA polymerase, can act on multiple related substrates. Substrate-selective inhibitors (SSIs) target these multisubstrate enzymes to modulate their specificity. Although SSIs hold promise as therapeutics, our theoretical understanding of how different inhibitors influence enzyme specificity remains limited. In this study, we examine enzyme selectivity within kinetic networks corresponding to known inhibition mechanisms. We demonstrate that competitive and uncompetitive inhibitors do not affect substrate specificity, regardless of rate constants. In contrast, noncompetitive and mixed inhibition can alter specificity and can lead to nonmonotonic responses to the inhibitor. We show that mixed and noncompetitive inhibitors achieve substrate-selective inhibition by altering the effective free-energy barriers of product formation pathways that are enabled by the inhibitor's presence. We then apply this framework to the Sirtuin-family deacylase SIRT2, showing that the suicide inhibitor thiomyristoyl lysine (TM) cannot influence substrate specificity unless there is a direct substrate exchange reaction or biochemical constraints are relaxed. These findings provide insights into engineering systems where cofactor binding modulates metabolic flux ratios. 

Transcription fidelity is inherently coupled to its strength, and highly expressed genes often exhibit elevated error rates. Epigenetic and structural factors, including histone modifications, DNA methylation, and nucleoid-associated proteins, modulate transcriptional output and, consequently, fidelity. However, the mechanistic origin of this fidelity-strength relationship remains poorly understood. Here, we propose that repulsive interactions among cotranscribing RNA polymerases (RNAPs) might explain these couplings. We develop a stochastic kinetic model of transcription elongation that incorporates both kinetic proofreading and repulsive forces generated through collisions between the neighboring RNAPs. In this framework, it is found that the collision forces accelerate leading RNAPs' elongation speed and impede their kinetic proofreading; the opposite trends occur for the trailing enzymes. As a result, interactions among multiple RNAPs at high initiation rates substantially elevate transcriptional error relative to isolated enzymes, with the magnitude of this increase determined by the intrinsic proofreading rate. In contrast, the mechanical partitioning of force between forward translocation and backtracking pathways primarily modulates elongation speed without altering fidelity. Together, our study provides a quantitative and mechanistic framework that links the collective dynamics of RNAPs to transcriptional errors, offering new physical insights into how transcriptional strength intrinsically compromises fidelity. 

Abstract Under starvation conditions, a spot of a few millionMyxococcus xanthuscells on agar will migrate inward to form aggregates that mature into dome-shaped fruiting bodies. This migration is thought to occur within structures called ‘streams,’ which are considered crucial for initiating aggregation. The prevailing traffic jam model hypothesizes that intersections of streams cause cell crowding and ‘jamming,’ thereby initiating the process of aggregate formation. However, this hypothesis has not been rigorously tested, in part due to the lack of a standardized, quantifiable definition of streams. To address this gap, we captured time-lapse movies and conducted fluorescent cell tracking experiments using wild-type and two motility-deficient mutantM. xanthusstrains. By quantitatively defining streams and developing a novel stream detection mask, we show that streams are not essential for nascent aggregate formation, though they may accelerate the process. Moreover, our results indicate that streaming has a genetic component: disrupting only one of the twoM. xanthusmotility systems hinders stream formation. Together, these findings challenge the idea that stream intersections are required to drive aggregate formation and suggest thatM. xanthusaggregation may be driven by mechanisms independent of streaming, highlighting the need for alternative models to fully explain aggregation dynamics. 

Under starvation,Myxococcus xanthusbacteria initiate a multicellular developmental program in which cells move to form fruiting bodies and differentiate into distinct cell types. Many genes affecting this process have been identified, and it is assumed that perturbing genes within the same pathway induces similar changes in the phenotype, although these changes may be subtle or obscured by pleiotropy. However, these pathways cannot be systematically mapped because there are no reliable methods for quantifying phenotype similarity. Here, we applied deep learning to quantify phenotype patterns and self-organization dynamics of 292 genetically distinct strains. We integrated ResNet and StyleGAN2 into a Variational Autoencoder and trained it together with a Siamese network that learns phenotypic similarity. This end-to-end system encoded high-resolution microscopy images into 13-dimensional feature vectors, effectively capturing variation in aggregation patterns across time and strains. Human evaluation confirmed that our model’s reconstructions were visually indistinguishable from real images and closely aligned with input phenotypes. Importantly, the feature space is interpretable: Individual dimensions correlate with biological features such as aggregate number and size, and extrapolation along these dimensions produces predictable morphological changes. Remarkably, our model revealed that developmental phenotypes and ultimate aggregation fate are predictable from the earliest images before visible aggregation begins. This predictability held across both genetic and environmental sources of variation, suggesting that subtle, early-stage phenotypic signatures carry critical information about developmental trajectories. These results demonstrate how machine learning can reveal hidden aspects of complex multicellular dynamics and provide methods for phenotypic analysis without manual annotation. 

The development of continuous-release devices or injectables for the long-term delivery of biologics is of great interest, especially monoclonal antibodies (mAbs) that require frequent, high-dose injections. Preclinical testing of these technologies in murine models is necessary for clinical translation; however, xenogeneic responses to the mAb and foreign body responses to the implants or injectables can confound results. Immune system knockout (KO) models that affect immune cells are often used in these experiments, but the effects of KO models on mAb pharmacokinetics (PK) are not well characterized. Here, we investigated the PK profile of the human mAb 3BNC117 after intravenous, subcutaneous, and intraperitoneal injections in four mouse strains: BL6, BCD, RAG2, and NSG mice. Noncompartmental analysis was used to quantify differences in PK between each mouse strain. Strikingly, both BL6 and NSG mice exhibited significantly higher mAb clearance compared to the other two strains. To better understand these differences, we developed a minimal physiologically based PK model of mAb PK incorporating Fc gamma R interaction in the peripheral tissue and the formation of anti-drug antibodies. The estimated model parameters demonstrate that the rapid clearance in the BL6 and NSG strains can be explained by the formation of anti-drug antibodies and increased Fc gamma R abundance in peripheral tissues, respectively. We then used simple allometric scaling relationships to assess which strains were reasonably predictive of human mAb PK. The scaled parameters obtained from BCD and RAG2 mice led to reasonably accurate human predicted PK, whereas the parameters obtained from NSG and BL6 mice did not. These results emphasize that the mechanistic differences influencing NSG and BL6 PK must be considered when assessing the translatability of data. 

ABSTRACT Quinones play a central role in maintaining redox balance and conserving energy but can trigger oxidative stress at high levels. However, the mechanisms by which microbes regulate quinone levels remain poorly understood, hindering effective metabolic engineering to modulate microbes for quinone production. Here, we show that the biosynthesis of the menaquinone precursor 1,4-dihydroxy-2-naphthoic acid (DHNA) in the lactic acid bacteriumLactococcus lactisis regulated by a combined genetic, enzymatic, and metabolic mechanism. Using synthetic biology approaches, we found that enzymes MenF and MenD both contribute to DHNA regulation, with MenD playing a more prominent role in controlling DHNA concentrations. A mathematical model elucidates a two-phase regulatory pattern resulting from the interplay of reversible flux and allosteric feedback inhibition, where either MenF or MenD can serve as the regulatory enzyme, depending on their relative expression ratio. In addition, the overproduction of DHNA is constrained by substrate availability, ensuring a sufficient but not excessive DHNA level to benefit cell growth while mitigating cytotoxicity. Collectively, these mechanisms maintain a fine-tuned physiological quinone level and suggest that modulating substrate supplement and MenF-to-MenD ratio could be keys for engineering DHNA production. IMPORTANCEQuinones are crucial molecules in cellular respiration, helping cells produce energy and maintain balance in their redox state. However, excessive quinone levels can be toxic, making it vital for microbes to tightly regulate their production. Our study uncovers howLactococcus lactis, a key food fermenting bacterium, uses a multi-layer mechanism to maintain optimal levels of the menaquinone precursor 1,4-dihydroxy-2-naphthoic acid (DHNA). By combining biosensors, genetic perturbations, and modeling, we show how cells balance the benefits and toxicity of quinones. These findings not only reveal fundamental microbial physiology but also provide strategies to engineer microbes for improved quinone production.