Biology subjects

Miller, M. A.

Publications and source records attributed to Miller, M. A..

7 recordsLinked to original sources

bioRxiv2026

MicroGrowAgents: An Agentic AI System for Microbial Cultivation Engineering

Microbial cultivation optimization remains labor-intensive and inefficient, requiring extensive experimental screening to identify suitable growth conditions. Traditional one-factor-at-a-time approaches are particularly ineffective for exploring complex, multidimensional nutrient parameter spaces. We present MicroGrowAgents, an agent-assisted system for auditable design of candidate growth media through integration of knowledge graphs, metabolic modeling, and optimal experimental design. The system comprises 29 documented agents implementing 58 skills across seven functional categories that query structured biological knowledge (KG-Microbe: 864,363 validated species), Bakta-annotated genomes (667,000+ annotated features), curated organism-specific FACTS sheets, and a DOI-linked publication corpus, and apply this evidence to three ends: recommending candidate ingredients and concentration ranges, specifying the factors of a statistically optimal MaxPro experimental design, and interpreting cultivation outcomes against known biochemistry. We applied the approach to Methylorubrum extorquens AM1 (formerly Methylobacterium extorquens; reclassified per LPSN) by cultivating 69 designed media plus a default media baseline (70 total tested conditions) in quadruplicate and assessing two concurrent objectives: biomass turbidity (740 nm) and apparent residual-Nd depletion (residual Nd by arsenazo III). Monte-Carlo resampling of the replicate-level uncertainty (1000 iterations) identified MPOB_058 as the single MC-stable Pareto-optimal condition (membership frequency 0.922); paired-control biology analysis flags MPOB_058 as chemistry-confounded and nominates MPOB_008 as the cleaner biological-signal anchor (its lower abiotic-drift contribution to the residual-Nd measurement makes it the better candidate for confirming biological lanthanide handling in a follow-up round), with MPOB_019 borderline-stable, providing a prioritized anchor set for confirmation in subsequent rounds rather than a single declared optimum. The integration of chemical similarity search (208,000+ embeddings), metabolic gap analysis, and multi-modal reasoning enables evidence-based hypothesis generation that reduces experimental burden while accelerating discovery of growth-promoting conditions. On the Biolog Odin platform, the MC-stable composite candidate MPOB_058 grew 79% more integrated biomass (area under the 740 nm growth curve) and 46% faster (maximum specific growth rate _max; Gompertz fits, R{superscript 2} > 0.99) than the standard MP base medium for this organism (the unsupplemented starting-point recipe carried as the on-plate control). This biomass and growth-rate advantage is a direct kinetic measurement; the conditions apparent residual-Nd depletion ranking, by contrast, awaits confirmation in a subsequent round, because MPOB_058s apparent Nd depletion is partly abiotic. Unlike general-purpose AI co-scientists, MicroGrowAgents grounds every recommendation in inspectable evidence -- structured provenance manifests with per-session input checksums, schema- and ontology-validated outputs, and 90.5% literature citation coverage (143 of 158 curated DOIs with evidence-supports-claim verification) -- separating deterministic design and analysis from agentic interpretation so that recommendations are transparent, explainable, and auditable, while passing 7 of 9 bbop-skills agentic-system criteria in a Claude-Code self-audit (the two unmet criteria are MCP-standardized tool exposure and full input-data cryptographic hashing, both tracked roadmap items). The Bigger PictureIdentifying growth conditions for poorly characterized microbes is one specific instance of a recurring problem in the natural sciences: how to integrate heterogeneous prior knowledge -- structured databases, published literature, mechanistic models -- with statistically efficient experimental design so that limited wet-lab time is spent where it most reduces the search. We show that a multi-agent AI system can perform that integration from organism input to wet-lab design, with every recommendation traceable from its source evidence through to the experimental design that tested it. The same architecture (specialized agents over structured knowledge bases, mechanistic models, and statistical design, with checksummed provenance recorded as the unit of evidence) is in principle transferable to other DBTL loops against under-explored search spaces such as catalyst discovery, materials formulation, cell-line engineering, drug-combination screens, or environmental remediation. We demonstrate it here only for microbial cultivation and flag cross-domain transfer as future work rather than a demonstrated capability. Our cultivation results on Methylorubrum extorquens AM1 are a concrete demonstration that this pattern works on real laboratory data: an uncertainty-aware shortlist of candidate media emerged from Monte-Carlo stability analysis of two competing objectives, with the leading composite candidate growing 79% more integrated biomass and 46% faster than the standard MP base medium for this organism; this biomass and growth-rate advantage is directly measured, while its apparent residual-Nd depletion advantage awaits confirmation in a subsequent round, since the apparent Nd depletion is partly abiotic. But the lesson we want to port across research domains is the architecture and its provenance discipline, not the specific organism. MicroGrowAgents is released under BSD-3 with checksummed YAML provenance manifests so that any group adopting these agents inherits a workflow designed to be re-runnable from its recorded inputs.

synthetic biology↗

bioRxiv2025

Enhanced mRNA delivery using ultrasound-delivered click reactive anchors

Therapeutic nucleic acid delivery has many potential applications, but it remains challenging to target extrahepatic tissues in a flexible and image-guided manner. To address this issue, we report a bioorthogonal pre-targeting strategy that uses focused ultrasound to promote the delivery of mRNA-loaded lipid nanoparticles (mRNA-LNP). We synthesized amphiphilic click reactive anchors (ACRAs) consisting of a phospholipid PEG-conjugate functionalized with transcyclooctene (TCO) or its companion reactive partner methyltetrazine (mTz), yielding ACRA-TCO and ACRA-mTz. ACRA derivatives were screened for cellular activity, yielding functionalized DOPE-PEG (1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N- (polyethylene glycol)) derivatives outperforming those containing saturated lipid or branched PEG. Nanobubbles encapsulating ultrasound-responsive gas precursor delivered ACRA-TCO to targeted cells and tissues using focused ultrasound, and this pre-targeting promoted the subsequent delivery of mRNA- LNP functionalized with companion ACRA-mTz. In cell cultures and in mice, ultrasound pre-targeting enhanced the accumulation of mTz-functionalized small molecule and nanoparticle compounds by 75% and 3.6-fold, respectively, and increased gene expression using mRNA-LNP in vivo. Taken together, this report presents a modular, ultrasound-enabled strategy for enhancing nucleic acid delivery in targeted tissues.

bioengineering↗

bioRxiv2021

Circadian Proteins Cry and Rev-erb Deepen Cellular Quiescence by Down-regulating Cyclin D and Cdk4,6

The proper balance and transition between cellular quiescence and proliferation are critical to tissue homeostasis, and their deregulations are commonly found in many human diseases, including cancer and aging. Recent studies showed that the reentry of quiescent cells to the cell cycle is subjected to circadian regulation. However, the underlying mechanisms are largely unknown. Here, we report that two circadian proteins, Cryptochrome (Cry) and Rev-erb, deepen cellular quiescence in rat embryonic fibroblasts, resulting in stronger serum stimulation required for cells to exit quiescence and reenter the cell cycle. This finding was opposite from what we expected from the literature. By modeling a library of possible regulatory topologies linking Cry and Rev-erb to a bistable Rb-E2f gene network switch that controls the quiescence-to-proliferation transition and by experimentally testing model predictions, we found Cry and Rev-erb converge to downregulate Cyclin D/Cdk4,6 activity, leading to an ultrasensitive increase of the serum threshold to activate the Rb-E2f bistable switch. Our findings suggest a mechanistic role of circadian proteins in modulating the depth of cellular quiescence, which may have implications in the varying potentials of tissue repair and regeneration at different times of the day.

systems biology↗

bioRxiv2021

Virulence shift in a sexual clade of Type X Toxoplasma infecting Southern Sea Otters

How virulent parasites are maintained in nature is an important paradigm of eukaryotic pathogenesis. Here we used population genetics and molecular methods to study the evolution and emergence of genetic variants of the protozoan parasite Toxoplasma gondii, referred collectively as Type X (HG12), recovered from a threatened marine mammal species. Specifically, 53 T. gondii strains were isolated from southern sea otters (SSO) that stranded between 1998-2004 with T. gondii infection (ranging from chronic incidental infections to fatal encephalitis). Over 74% of these SSO, collected throughout their geographic range, were infected with Type X, based on multi-locus PCR-DNA sequencing. Depending on the locus investigated, Type X strains possessed one of three allelic types that had independently assorted across the strains examined; either genetically distinct alleles, referred to as "{psi}" or "8", or a Type II allele. Phylogenetic incongruence among locus-specific trees, genome-wide CGH array and WGS analyses confirmed that Type X is a sexual clade of natural recombinants that resemble F1 progeny from a genetic cross between Type II and a mosaic of two distinct "{psi}" or "{delta}" ancestries. A single Type X genotype (19/53; 36%) that largely caused subclinical chronic infections in SSO, was highly pathogenic to mice (LD100= 1 parasite). To determine whether murine virulence genes could be mapped within this population of natural isolates, we performed a genome scan and identified four QTLs with LOD scores greater than 4.0. Targeted disruption of ROP33, the strongest candidate from among 16 genes within the highest QTL on Chromosome VIIa established ROP33 as a murine virulence locus. The ability of this highly pathogenic mouse-virulent T. gondii clone to expand its environmental niche and infect a majority of SSO supports a virulence shift model whereby generalist pathogens like T. gondii utilize their sexual cycles to produce new strains that possess an expanded biological potential. Such a trait enables pathogens to extend their host range or be naturally selected within their vast intermediate host range to maximize transmission. Our work establishes a rationale for how virulent strains can be maintained cryptically in nature across a pathogens broad host range, and act as potential reservoirs for epidemic disease. ImportanceWaterborne outbreaks of protozoal parasites are capable of causing fatal disease in a wide range of animals, including humans. Population expansion of felids in addition to anthropogenic changes near marine estuarine environments may facilitate marine wildlife exposure to highly infectious Toxoplasma gondii oocysts shed in felid feces. Infected cats shed millions of environmentally-resistant T. gondii oocysts that can be widely dispersed by storm events. In North America T. gondii is thought to possess a highly clonal population structure dominated by 4 clonal lineages (I, II, III, and X). Population genetic analysis of 53 T. gondii isolates collected longitudinally from SSO infected with T. gondii that stranded between 1998-2004 identified 74% of otters infected with Type X T. gondii, and that Type X is not a clonal lineage, but rather a recombinant clade of at least 12 distinct strains consistent with a recent genetic cross. Importantly, one Type X haplotype was isolated from 36% of southern sea otters (Enhydra lutris neries) across their geographic range in California. This haplotype was highly pathogenic to mice but caused relatively benign infections in SSO. A genome scan was performed to identify a virulence locus; a secreted serine threonine kinase (ROP33) that enhanced pathogenicity in laboratory mice, but not sea otters. Our data support a virulence shift model whereby generalist pathogens like T. gondii utilize their sexual cycles to produce virulent strains that can be maintained cryptically in nature, according to their differential capacity to cause disease within the pathogens broad intermediate host range. This type of "host selection" has important public health implications. Strains capable of causing fatal infections can persist in nature by circulating as chronic infections in resistant intermediate host species that act as reservoirs for epidemic disease.

molecular biology↗

bioRxiv2021

A minimal rDNA array replicates early, delays genome replication, and uncouples anaphase entry from S phase completion

Eukaryotes maintain hundreds of copies of ribosomal DNA (rDNA), many more than required for ribosome biogenesis, suggesting a yet undefined role for large rDNA arrays outside of ribosomal RNA synthesis. We demonstrate that reducing the Saccharomyces cerevisiae rDNA array to 35 copies, which is sufficient for ribosome function, shifts rDNA from being the latest replicating region in the genome to one of the earliest. This change in replication timing results in delayed genome-wide replication and classic replication defects. We present evidence that the requirement for rDNA to replicate late, which is conserved among eukaryotes, also coordinates the completion of genome replication with anaphase entry through the proper sequestration of the mitotic exit regulator Cdc14p in the rDNA-containing nucleolus. Our findings suggest that, instead of being a passive repetitive element, the large late-replicating rDNA array plays an active role in genome replication and cell cycle control.

molecular biology↗

bioRxiv2021

An RNAi screen in C. elegans for genes that play a role in secretion and cleavage of VAPB MSP domain.

VAPB (VPR-1 in C. elegans) is a type-II ER transmembrane protein whose N-terminal Major Sperm Protein domain (MSPd) is cleaved and secreted. Mutations in the MSPd of human VAPB cause impaired secretion and are associated with Amyotrophic Lateral Sclerosis (ALS). In C. elegans, the secreted MSPd signals non-cell-autonomously to regulate striated muscle mitochondrial morphology and gonad development. As VAPB/VPR-1 does not have a signal peptide and its MSPd extends into the cytosol, it is unclear how the protein is proteolytically cleaved and secreted. To identify genes that are involved in VPR-1 cleavage and unconventional secretion, we performed an RNA interference (RNAi) screen in C. elegans. Worms null for vpr-1 are sterile and have striated muscle mitochondrial abnormalities. These defects can be rescued by vpr-1 expression in the neurons, germline, or intestine, implying that these three tissues share a common machinery to cleave and secrete the MSPd. Examination of shared gene expression in these tissues revealed a list of 422 genes, which we targeted with RNAi. vpr-1 null worms expressing vpr-1 from intestine were used in the screen, and the brood size of these worms after RNAi knockdown was scored. Disruption of factors involved in VPR-1 MSPd processing and/or secretion should revert fertility phenotypes in these worms. We identified many genes that induce compromised fertility when knocked down in these but not wild type worms, including a V-SNARE, several proteasome components, stress response molecules, and mitochondrial genes. Our screen thus identified many potential players involved in MSPd processing and/or secretion. SummaryThe MSP domain (MSPd) of a type-II ER transmembrane protein called VAPB is cleaved and secreted to function as a non-cell-autonomous signal. The topology of VAPB positions MSPd in the cytosol. It is thus unclear how MSPd is cleaved from VAPB and released extracellularly. Using C. elegans, we screened 422 genes by RNAi to identify potential candidates regulating MSPd cleaving and secretion. We identified the Golgi v-SNARE YKT-6 and several components of the 20S and 19S proteasome that may mediate MSPd trafficking and cleaving, respectively. These results have promising implications in advancing our understanding of MSPd signaling.

cell biology↗

bioRxiv2020

Cell shape, and not 2D migration, predicts ECM-driven 3D cell invasion in breast cancer

Metastasis, the leading cause of death in cancer patients, requires the invasion of tumor cells through the stroma in response to migratory cues, such as those provided by the extracellular matrix (ECM). Recent advances in proteomics have led to the identification of hundreds of ECM proteins which are more abundant in tumors relative to healthy tissue. Our goal was to develop a pipeline to easily predict which of these ECM proteins is more likely to have an effect on cancer invasion and metastasis. We evaluated the effect of 4 ECM proteins upregulated in breast tumor tissue in multiple human breast cancer cell lines in 3 assays. We found there was no linear relationship between the 11 cell shape parameters we quantified when cells adhere to ECM proteins and 2D cell migration speed, persistence or 3D invasion. We then used classifiers and partial-least squares regression analysis to identify which metrics best predicted ECM-driven 2D migration and 3D invasion responses. ECM-driven 2D cell migration speed or persistence did not correlate with or predict 3D invasion in response to that same cue. However, cell adhesion, and in particular cell elongation and irregularity accurately predicted the magnitude of ECM-driven 2D migration and 3D invasion in all cell lines. Testing predictions revealed that our models are good at predicting the effect of novel ECM proteins within a given cell line, but that ECM responses are cell-line specific. Overall, our studies identify the cell morphological features that determine 3D invasion responses to individual ECM proteins. This platform will help provide insight into the functional role of ECM proteins abundant tumor tissue and help prioritize strategies for targeting tumor-ECM interactions to treat metastasis. FundingThis work was supported by the National Institutes of Health [R00-CA207866-04 to M.J.O.]; Tufts University [Start-up funds from the School of Engineering to M.J.O.] and funds from NSF REU to A.W. Conflict-of-interest: None. Insight BoxMetastasis, the dissemination of tumor cells, is driven by the interaction of invading tumor cells with their local environment, in particular with the ECM, which provides structure and support to our tissues. This study presents an integrated approach to predict the effect of individual ECM proteins on 3D invasion and metastasis based on simple adhesion assays which quantify cell shape. Machine learning classification and partial-least squares regression models reveal that ECM-driven 2D cell migration metrics are not predictive of 3D invasion, and that cell shape of cells adhered to ECM can predict that proteins effect on 3D invasion. These data provide a pipeline for predicting the effect of ECM proteins on breast cancer cell invasion and metastasis.

bioengineering↗