Biology subjects

Maddamsetti, R.

Publications and source records attributed to Maddamsetti, R..

At least 19 recordsLinked to original sources

bioRxiv2026

Gene circuit-driven amplified selection enables evolution of fast-growing Escherichia coli

The laboratory Escherichia coli K-12 strain has doubled no faster than [~]20 minutes for decades. This plateau could reflect a biophysical limit or simply the way batch culture selects on growth rate. Here we show it can be broken through amplified selection with a Red Queen gene circuit, which takes advantage of growth rate heterogeneity in monoclonal populations to selectively suppress slow-growing cells and creates a tunable mapping from intrinsic growth rate to survival. After 70 days ([~]1,000 generations) of amplified selection in MG1655+FHr and subsequent removal of the circuit, a top evolved clone (RQ70) reached a maximum specific growth rate of 2.61 h-{superscript 1} in shake-flask culture. This corresponds to a doubling time of 15.9 minutes, to our knowledge the shortest reported for E. coli K-12, against 18.1 minutes for evolved controls and 20.3 minutes for the ancestor. The gain came at the cost of a [~]3-fold increase in lag time, indicating that the 20-minute plateau is a multi-trait optimum under conventional batch selection rather than an absolute constraint. We argue that synthetic gene circuits can therefore reshape the evolutionary process itself, pushing performance beyond apparent physiological limits.

synthetic biology↗

bioRxiv2026

HaloMPNN: retraining ProteinMPNN on halophilic proteomes for salt-tolerant enzyme design

Machine learning-guided protein sequence redesign is now routinely used to optimize multiple properties relevant for protein engineering, most prominently thermostability and recombinant expression levels. Salt tolerance is a valuable property for "blue-biotechnology"- enabled biomanufacturing, yet no generative computational method exists to redesign proteins for increased salt tolerance. We hypothesized that a training dataset heavily biased toward salt-adapted proteomes would yield a model capable of designing proteins with halophilic properties. To test this, we retrained the sequence redesign model ProteinMPNN on proteins from "salt-in" extreme halophiles such as Haloarcula marismortui, a Dead Sea archaeon that grows optimally near 3-4 M NaCl, roughly six times the salinity of seawater, and accumulates molar concentrations of salts in its cytoplasm. Our model, HaloMPNN, redesigns non-halophilic proteins so that their properties shift towards those of natural halophilic proteins: lower predicted isoelectric point, greater surface acidity, and reduced surface and core hydrophobicity. Redesigning a broad range of non-halophilic proteins with SolubleMPNN, ProteinMPNN, and HyperMPNN shows that this shift is specific to HaloMPNN rather than a generic consequence of sequence redesign. HaloMPNN therefore offers both a route to designing candidate salt-tolerant enzymes and a means of identifying the characteristics that underlie halophilic adaptation.

bioengineering↗

bioRxiv2026

Parallel evolution of chimeric genes in microbial evolution experiments

Recombination can generate new or improved proteins by merging pieces of pre-existing genes into a new whole. Here, when we reanalyzed genomic data from several laboratory evolution experiments with Escherichia coli, evidence for the convergent evolution of chimeric genes became apparent. In these experiments, a pair of paralogous genes were recombined into a single hybrid copy by large genomic deletions. In one case, the excision of a cryptic e14 prophage occurred in 6 out of 8 replicates of a 22-day fluoroquinolone-resistance evolution experiment. These parallel prophage excisions recombined an icd isocitrate dehydrogenase gene with a homologous icdC C-terminal fragment pseudogene. In the other case, convergent 23 kB deletions recombined cpsG paralogs across replicate populations of the Lenski long-term evolution experiment with Escherichia coli (LTEE), generating a chimeric phosphomannomutase. Together, these results indicate that chimeric genes evolve rapidly in bacteria and are more common than previously believed, because they are easily missed by standard genomic analyses. Experiments are needed to determine whether these chimeric proteins are adaptive, are by-products of adaptive genomic deletions, or both. Data SummaryAll data used in this study are publicly available. All data, code, and results for this study are available on Zenodo (DOI: 10.5281/zenodo.20692138). Genomic data published by Mohiuddin et al. are available in the European Nucleotide Archive (Study ERP166963 with Study Accession PRJEB83321), and genomic data published by Tenaillon et al. are available from the NCBI BioProject database under accession number PRJNA294072. Impact StatementGenomic deletions can generate chimeric proteins by splicing together parts of existing genes. However, the extent to which chimeric proteins form, and the extent to which they are adaptive are not well-understood. Complex structural variants (such as chimeric protein formation) are often missed or incompletely annotated by standard genomic pipelines. Furthermore, there is an ongoing need for new methods to detect selection on complex structural variants. In this work, we report evidence of chimeric gene formation in laboratory evolution experiments with Escherichia coli. In both experiments, chimeric genes evolved in parallel across independent replicate populations, which is an unambiguous signal of positive selection (although it remains unclear whether selection is operating on the chimeric gene, the genomic deletion creating the chimeric gene, or both). The observation of chimeric gene formation in laboratory experiments indicate that chimeric genes evolve rapidly in bacteria and are more common than previously believed. Therefore, this work points to an important new direction for microbial genomics research: what is the extent of chimeric gene formation in natural bacterial populations, and what consequences might chimeric genes have for bacterial adaptation?

microbiology↗

bioRxiv2026

Zero-shot ecological annotation of microbial genomes with myLLannotator accelerates scientific discovery

Large language models (LLMs) are promising scientific assistants, but many models remain inaccessible for resource-constrained scientists. To address this, we present myLLannotator, a python package for metadata annotation built on llama3.2-3B. myLLannotator labels 18,000+ genomes in ~2 hours on a laptop, reproducing a discovery involving substantial manual annotation. We use myLLannotator to discover that duplicated genes are depleted in endosymbiotic bacteria, demonstrating its power to accelerate hypothesis testing and discovery.

bioinformatics↗

bioRxiv2025

A foundation model for microbial growth dynamics

Microbial growth dynamics contain rich information about microbial populations, which support applications from antibiotic testing to microbiome engineering. However, the high dimensionality of growth data and the scarcity of large, task-specific datasets have limited generalizable modeling analysis across systems. Here, we develop a foundation model for microbial growth dynamics. It is a large-scale, self-supervised representation model trained on [~]370,000 experimental and simulated growth curves spanning diverse microbial species, environmental conditions, and community contexts. The model learns lower-dimensional latent embeddings that capture essential dynamical features of raw growth data and enable accurate reconstruction of these data. The concise representations enhance predictive performance in diverse downstream applications. Using these embedding, we achieve few-shot learning for antibiotic classification and concentration prediction, accurate forecasting of simulated and experimental communities, and inference of total abundance from relative-abundance data. By extracting transferable representations from heterogeneous datasets, our model provides a general framework for analyzing and predicting microbial community dynamics from limited measurements.

synthetic biology↗

bioRxiv2025

Integrating theory and machine learning to reveal determinants of plasmid copy number

Plasmids are extrachromosomal mobile genetic elements whose copy numbers (PCNs) critically influence microbial evolution, antibiotic resistance and pathogenicity. Despite their importance and immense diversity, the ecological, evolutionary and molecular factors determining PCN remain poorly understood. Here, we present a theoretical model to explain the empirical power-law relationship between plasmid size and copy number, one of the fundamental quantitative principles governing PCN control. However, this relationship alone has limited predictive power. To improve PCN prediction, we introduce a data-driven approach incorporating diverse features. Trained on >10,000 plasmids, our machine learning model achieves significantly enhanced accuracy, with plasmid-encoded protein domains emerging as key predictors. Applying this framework, we conduct the first comprehensive analysis of PCN distributions across hundreds of thousands of metagenomic plasmids (IMG/PR database) and tens of thousands of clinical isolates, uncovering niche specific taxonomic PCN hotspots and ecological adaptations. These results provide critical insights into plasmid ecology, ARG surveillance and shed lights on the gut plasmidome, a "dark matter" in human microbiome.

microbiology↗

bioRxiv2025

Emergence of population-level feedback control by transposon-plasmid coevolution

The origins of adaptive functions remain poorly understood, despite considerable interest1-5. Here, we report the de novo evolution of population-level feedback control in clonal Escherichia coli strains containing high-copy plasmids, enabling these strains to express green fluorescent protein (GFP) in response to tetracycline. Selection maintains tetA+ and tetA- plasmids within single cells, due to negative feedback from the toxic effects of a tetA-gfp tetracycline resistance transposon. At high plasmid copy numbers, the intracellular equilibrium of tetA+ and tetA- plasmids robustly responds to tetracycline through rapid evolutionary dynamics6-8. Theory predicts that the GFP response to tetracycline is determined by the covariance between GFP expression and bacterial fitness. Our findings show that small mutational changes can result in large qualitative changes to population-level behavior. Here, the evolution of polymorphic intracellular populations of mobile genetic elements allows host populations to dynamically respond to antibiotic in the environment.

systems biology↗

bioRxiv2025

Transposon-plasmid nesting enables fast response to fluctuating environments

Mobile genetic elements (MGEs) play a critical role in shaping the response and evolution of microbial populations and communities. Despite distinct maintenance mechanisms, different types of MGEs can form nested structures. Using bioinformatics analysis of 14,338 plasmids in the NCBI RefSeq database, we found transposons to be widespread and significantly enriched on plasmids relative to chromosomes, highlighting the prevalence of transposon-plasmid nesting. We hypothesized that this nested structure provides unique adaptive advantages by combining transposition-driven genetic mobility with plasmid-mediated copy number amplification. Using engineered transposon systems, we demonstrated that nesting enables rapid and tunable responses of transposon-encoded genes in fluctuating environments. Specifically, transposition maintains a reservoir of the encoded genes, while plasmid copy number fluctuations further amplify the dynamic range of gene dosage, thus enhancing the response speed and stability of transposon-encoded traits. Our findings demonstrate an adaptive benefit of transposon-plasmid nesting and provide insights into their ecological persistence and evolutionary success.

systems biology↗

bioRxiv2024

Scaling laws of plasmids across the microbial tree of life

Plasmids play a critical role in shaping the dynamics and evolution of microbial communities. The capacity of a plasmid to express genes is constrained by two parameters: length and copy number. However, the interplay between these parameters and their constraints on plasmid evolution have remained elusive due to the absence of comprehensive quantitative analyses. To address this gap, we developed Probabilistic Iterative Read Assignment (PIRA), a new computational method that overcomes previous computational bottlenecks, enabling rapid and accurate determination of plasmid copy numbers at an unprecedented scale. Applying PIRA to all microbial genomes in the NCBI RefSeq database with linked short-read sequencing data in the Sequencing Read Archive (SRA), we analyzed 4,317 bacterial and archaeal genomes encompassing 11,338 plasmids, spanning the microbial tree of life. Our analysis reveals three scaling laws of plasmids: first, an inverse power-law correlation between plasmid copy number and plasmid length; second, a positive linear correlation between protein-coding genes and plasmid length; and third, a positive correlation between metabolic genes per plasmid and plasmid length, particularly for large plasmids. These scaling laws imply fundamental constraints on plasmid evolution and functional organization, indicating that as plasmids increase in length, they converge toward chromosomal characteristics in copy number and functional content. Our findings not only advance the understanding of plasmid dynamics but also have implications for microbial evolution, biotechnology, and the design of synthetic plasmids. SignificanceBy discovering universal scaling laws and developing a new computational method to compute plasmid copy numbers across the microbial tree of life, we show that as plasmids increase in length, they converge to chromosomes in their copy number and their coding and metabolic properties. This insight reveals fundamental principles governing plasmid evolution and has implications for biotechnology and medicine.

microbiology↗

bioRxiv2024

Decoding pattern formation rules by integrating mechanistic modeling and deep learning

Predictive programming of self-organized pattern formation using living cells is challenging in major part due to the difficulty in navigating through the high-dimensional design space effectively. The emergence and characteristics of patterns are highly sensitive to both system and environmental parameters. Often, the optimal conditions able to generate patterns represent a small fraction of the possible design space. Furthermore, the experimental generation and quantification of patterns is typically labor intensive and low throughput, making it impractical to optimize pattern formation solely based on trials and errors. To this end, simulations using a well-formulated mechanistic model can facilitate the identification of optimal experimental conditions for pattern formation. However, even a moderately complex system can make these simulations computationally prohibitive when applied to a large parameter space. In this study, we demonstrate how integrating mechanistic modeling with machine learning can significantly accelerate the exploration of design space for patterning circuits and aid in deriving human-interpretable design rules. We apply this strategy to program self-organized ring patterns in Pseudomonas aeruginosa using a synthetic gene circuit. Our approach involved training a neural network with simulated data to predict pattern formation 10 million times faster than the mechanistic model. This neural network was then used to predict pattern formation across a vast array of parameter combinations, far exceeding the size of the training dataset and what was computationally feasible using the mechanistic model alone. By doing so, we identified many parameter combinations able to generate desirable patterns, which still represent an extremely small fraction of explored parametric space. We next used the mechanistic model to validate top candidates and identify coarse-grained rules for patterning. We experimentally demonstrated the generation and control of patterning guided by the learned rules. Our work highlights the effectiveness in integrating mechanistic modeling and machine learning for rational engineering of complex dynamics in living cells.

synthetic biology↗

bioRxiv2024

Circular RNA Obelisk-S.s is highly abundant in Streptococcus sanguinis SK36

A new class of viroid-like circular RNAs, called Obelisks, was recently reported by Zheludev et al.1. They identify a specific 1137 nt Obelisk, called Obelisk-S.s, in monoculture transcriptomes of Streptococcus sanguinis SK36, a commensal bacterium of the human oral microbiome. Here, we report that Obelisk-S.s. is highly abundant in SK36, despite its absence from the SK36 genome (i.e., as DNA). In 11 out of 17 monoculture SK36 RNA-seq datasets examined, Obelisk-S.s. is more abundant than any mRNA. Given its abundance, we hypothesized that multiple Obelisk-S.s variants could coexist within SK36. We found three Obelisk-S.s mutations at 5-10% allele frequency in some samples: a R162R synonymous mutation in one set of replicate transcriptomes, and an I48I synonymous mutation and an intergenic mutation in another set of replicate transcriptomes. A simple mathematical model shows how high Obelisk abundance can transiently stabilize intracellular Obelisk populations, and how extreme Obelisk abundances may stabilize intracellular Obelisk populations indefinitely. Evolution experiments with SK36 could test this theory and could shed light on how Obelisks function and evolve within their microbial hosts.

microbiology↗

bioRxiv2023

Programming dynamic division of labor using horizontal gene transfer

The metabolic engineering of microbes has broad applications, including in biomanufacturing, bioprocessing, and environmental remediation. The introduction of a complex, multi-step pathway often imposes a substantial metabolic burden on the host cell, restraining the accumulation of productive biomass and limiting pathway efficiency. One strategy to alleviate metabolic burden is division of labor (DOL), in which different subpopulations carry out different parts of the pathway and work together to convert a substrate into a final product. However, the maintenance of different engineered subpopulations is challenging due to competition and convoluted inter-strain population dynamics. Through modeling, we show that dynamic division of labor (DDOL) mediated by horizontal gene transfer (HGT) can overcome these limitations and enable the robust maintenance of burdensome, multi-step pathways. We also use plasmid genomics to uncover evidence that DDOL is a strategy utilized by natural microbial communities. Our work suggests that bioengineers can harness HGT to stabilize synthetic metabolic pathways in microbial communities, enabling the development of robust engineered systems for deployment in a variety of contexts.

synthetic biology↗

bioRxiv2021

Selection maintains protein interactome resilience in the long-term evolution experiment with Escherichia coli

Most cellular functions are carried out by a dynamic network of interacting proteins. An open question is whether the network properties of protein interactomes represent phenotypes under natural selection. One proposal is that protein interactomes have evolved to be resilient, such that they tend to maintain connectivity when proteins are removed from the network. This hypothesis predicts that interactome resilience should be maintained by natural selection during long-term experimental evolution. I tested this prediction by modeling the evolution of protein-protein interaction (PPI) networks in Lenskis long-term evolution experiment with Escherichia coli (LTEE). In this test, I removed proteins affected by nonsense, insertion, deletion, and transposon mutations in evolved LTEE strains, and measured the resilience of the resulting networks. I compared the rate of change of network resilience in each LTEE population to the rate of change of network resilience for corresponding randomized networks. The evolved PPI networks are significantly more resilient than networks in which random proteins have been deleted. Moreover, the evolved networks are generally more resilient than networks in which the random deletion of proteins was restricted to those disrupted in LTEE. These results suggest that evolution in the LTEE has favored PPI networks that are, on average, more resilient than expected from the genetic variation across the evolved strains. My findings therefore support the hypothesis that selection maintains protein interactome resilience over evolutionary time. Significance StatementUnderstanding how protein-protein interaction (PPI) networks evolve is a central goal of evolutionary systems biology. One property that has been hypothesized to be important for PPI network evolution is resilience, which means that networks tend to maintain connectivity even after many nodes (proteins in this case) have been removed. This hypothesis predicts that PPI network resilience should be maintained during long-term experimental evolution. Consistent with this prediction, I found that the PPI networks that evolved over 50,000 generations of Lenskis long-term evolution experiment with E. coli are more resilient than expected by chance.

evolutionary biology↗

bioRxiv2021

Idiosyncratic purifying selection on metabolic enzymes in the long-term evolution experiment with Escherichia coli

Bacteria, Archaea, and Eukarya all share a common set of metabolic reactions. This implies that the function and topology of central metabolism has been evolving under purifying selection over deep time. Central metabolism may similarly evolve under purifying selection during longterm evolution experiments, although it is unclear how long such experiments would have to run (decades, centuries, millennia) before signs of purifying selection on metabolism appear. I hypothesized that central and superessential metabolic enzymes would show evidence of purifying selection in the long-term evolution experiment with Escherichia coli (LTEE). I also hypothesized that enzymes that specialize on single substrates would show stronger evidence of purifying selection in the LTEE than generalist enzymes that catalyze multiple reactions. I tested these hypotheses by analyzing metagenomic time series covering 62,750 generations of the LTEE. I find mixed support for these hypotheses, because the observed patterns of purifying selection are idiosyncratic and population-specific. To explain this finding, I propose the Jenga hypothesis, named after a childrens game in which blocks are removed from a tower until it falls. The Jenga hypothesis postulates that loss-of-function mutations degrade costly, redundant, and nonessential metabolic functions. Replicate populations can therefore follow idiosyncratic trajectories of lost redundancies, despite purifying selection on overall function. I tested the Jenga hypothesis by simulating the evolution of 1,000 minimal genomes under strong purifying selection. As predicted, the minimal genomes converge to different metabolic networks. Strikingly, the core genes common to all 1,000 minimal genomes show consistent signatures of purifying selection in the LTEE. Significance StatementPurifying selection conserves organismal function over evolutionary time. However, few studies have examined the role of purifying selection during adaptation to novel environments. I tested metabolic enzymes for purifying selection in an ongoing long-term evolution experiment with Escherichia coli. While some populations show signs of purifying selection, the overall pattern is inconsistent. To explain these findings, I propose the Jenga hypothesis, in which loss-of-function mutations first degrade costly, redundant, and nonessential metabolic functions, after which purifying selection begins to dominate. I then tested several predictions of the Jenga hypothesis using computational simulations. On balance, the simulations confirm that we should find evidence of purifying selection on the metabolic pathways that sustain growth in a novel environment.

evolutionary biology↗

bioRxiv2020

Universal constraints on protein evolution in the long-term evolution experiment with Escherichia coli

Although it is well known that abundant proteins evolve slowly across the tree of life, there is little consensus for why this is true. Here, I report that abundant proteins evolve slowly in the hypermutator populations of Lenskis long-term evolution experiment with Escherichia coli (LTEE). Specifically, the density of all observed mutations per gene, as measured in metagenomic time series covering 60,000 generations of the LTEE, significantly anti-correlates with mRNA abundance, protein abundance, and degree of protein-protein interaction. The same pattern holds for nonsynonymous mutation density. However, synonymous mutation density, measured across the LTEE hypermutator populations, positively correlates with protein abundance. These results show that universal constraints on protein evolution are visible in data spanning three decades of experimental evolution. Therefore, it should be possible to design experiments to answer why abundant proteins evolve slowly. Significance StatementA universal evolutionary pattern is that highly abundant and highly interacting proteins evolve slowly. This pattern was discovered in analyses that cover millions of years worth of sequence variation, so it is not clear how long it takes (decades, centuries, millennia) for such patterns to emerge. Here, I report that this universal evolutionary pattern emerges in metagenomic data that cover just 30 years of experimental evolution.

evolutionary biology↗

bioRxiv2020

Maintenance of metabolic plasticity despite relaxed selection in a long-term evolution experiment with Escherichia coli

AO_SCPLOWBSTRACTC_SCPLOWTraits that are unused in a given environment are subject to processes that tend to erode them, leading to reduced fitness in other environments. Although this general tendency is clear, we know much less about why some traits are lost while others are retained, and about the roles of mutation and selection in generating different responses. We addressed these issues by examining populations of a facultative anaerobe, Escherichia coli, that have evolved for >30 years in the presence of oxygen, with relaxed selection for anaerobic growth and the associated metabolic plasticity. We asked whether evolution led to the loss, improvement, or maintenance of anaerobic growth, and we analyzed gene expression and mutational datasets to understand the outcomes. We identified genomic signatures of both positive and purifying selection on aerobic-specific genes, while anaerobic-specific genes showed clear evidence of relaxed selection. We also found parallel evolution at two interacting loci that regulate anaerobic growth. We competed the ancestor and evolved clones from each population in an anoxic environment, and we found that anaerobic fitness had not decayed, despite relaxed selection. In summary, relaxed section does not necessarily reduce an organisms fitness in other environments. Instead, the genetic architecture of the traits under relaxed selection and their correlations with traits under positive and purifying selection may sometimes determine evolutionary outcomes.

evolutionary biology↗

bioRxiv2020

Divergent evolution of mutation rates and biases in the long-term evolution experiment with Escherichia coli

All organisms encode enzymes that replicate, maintain, pack, recombine, and repair their genetic material. For this reason, mutation rates and biases also evolve by mutation, variation, and natural selection. By examining metagenomic time series of the Lenski long-term evolution experiment (LTEE) with Escherichia coli (Good, et al. 2017), we find that local mutation rate variation has evolved during the LTEE. Each LTEE population has evolved idiosyncratic differences in their rates of point mutations, indels, and mobile element insertions, due to the fixation of various hypermutator and antimutator alleles. One LTEE population, called Ara+3, shows a strong, symmetric wave pattern in its density of point mutations, radiating from the origin of replication. This pattern is largely missing from the other LTEE populations, most of which evolved missense, indel, or structural mutations in topA, fis, and dusB-- loci that all affect DNA topology. The distribution of mutations in those genes over time suggests epistasis and historical contingency in the evolution of DNA topology, which may have in turn affected local mutation rates. Overall, the replicate populations of the LTEE have largely diverged in their mutation rates and biases, even though they have adapted to identical abiotic conditions.

evolutionary biology↗

bioRxiv2020

A Simple Test to Infer Mode of Selection in Metagenomics Time Series of Evolving Asexual Populations

A general method to infer both positive and purifying selection during the real-time evolution of hypermutator pathogens would be broadly useful. To this end, we introduce a simple test to infer mode of selection (STIMS) from metagenomic time series of evolving microbial populations. We test STIMS on metagenomic data generated by simulations of bacterial evolution, and on metagenomic data spanning 62,750 generations of Lenskis long-term evolution experiment with Escherichia coli (LTEE). In both cases, STIMS recovers signals of positive and purifying selection on gold standard sets of genes. Using STIMS, we find strong evidence of ongoing positive selection on key regulators of the E. coli gene regulatory network, even in some hypermutator populations. STIMS also detects positive selection on regulatory genes in hypermutator populations of Pseudomonas aeruginosa that adapted to subinhibitory concentrations of colistin - an antibiotic of last resort - for just twenty-six days of laboratory evolution. Our results show that the fine-tuning of gene regulatory networks is a general mechanism for rapid and ongoing adaptation. The simplicity of STIMS, together with its intuitive visual interpretation, make it a useful test for positive and purifying selection in metagenomic data sets that track the evolution of hypermutator populations in real-time.

evolutionary biology↗