henhaobio
Sign in

CellBiophysicsPeer-reviewed journal article

Bringing the Genetically Minimal Cell to Life on a Computer in 4D

Zane R. ThornburgiD, Andrew K. MaytiniD, Jiwoong KwoniD, Troy A. BrieriD, Benjamin R. GilbertiD, Enguang FuiD, Yang-le GaoiD, Jordan QuennevilleiD, Tianyu WuiD, Henry Li, Talia Long, Weria PezeshkianiD, Lijie Sun, DANIELA MATIAS DE CARVALHO BITTENCOURTiD, John I. GlassiD, Angad P. MehtaiD, Taekjip HaiD, Zaida Ann Luthey-SchulteniD

PDFDOIPreprint

In-depth summary

Claude Opus 5.5 · xhigh effort · 5 Nov 2026 · from the full text of version 1 (15 Jun 2025)

A hybrid 4D whole-cell simulation of the minimal bacterium JCVI-syn3A integrates spatial reaction-diffusion, metabolism, polymer chromosome dynamics, and division across a complete 105-minute cell cycle.

Previous whole-cell models either treated cells as well-stirred reactors lacking spatial organization or simulated 3D spatial dynamics for only the first 20 minutes of a static cell before DNA replication. Coupling GPU-accelerated lattice reaction-diffusion, deterministic metabolic kinetics, and Brownian dynamics chromosome modeling simulates all 493 genes, ribosome biogenesis, DNA replication, and division across 50 JCVI-syn3A cell cycles.

Key results

  1. ShownThe 4D whole-cell model recovers the experimental 105-minute cell doubling time and origin-to-terminus DNA ratio from emergent lipid/membrane-protein synthesis and dNTP-dependent replication rates.

    Across 50 simulated replicate cells, surface area doubled in 105 minutes on average (matching the 105 min experimental doubling time) and chromosomes replicated in 46 minutes (B period ~5 min, C period 46 min, D period 54 min), yielding a simulated ori:ter ratio of 1.28 versus 1.21 in exponential-phase whole-genome sequencing (replicates 1.20, 1.21, 1.21 vs stationary-phase 0.98, 0.99, 0.97). · 4D whole-cell modeling reflects experiment; DNA coverage analysis; Figure legends

  2. ShownAiryscan 3D super-resolution fluorescence microscopy of JCVI-syn3B confirms symmetric cell division via prolate and dumbbell-shaped intermediates that constrain the 4D division geometry.

    Three-color Airyscan imaging (35 nm xy pixel, 130 nm z-step) of 1,319 analyzed JCVI-syn3B cells expressing FtsZ-mCherry (out of 2,968 segmented cells) showed spherical cells (mean diameter 0.89 +/- 0.37 um) transitioning through prolate (diameter > 0.6 um, aspect ratio < 0.8) and dumbbell shapes, alongside ~15% cell-in-cell structures in late exponential phase. · 4D whole-cell modeling reflects experiment; Airyscan Imaging; Figure legends

  3. ShownSpatially resolved 3D gene expression and replication initiation kinetics are sensitive to DnaA-ssDNA binding affinity and mRNA competition between ribosomes and membrane-tethered degradosomes.

    Initiating replication within 2-46 minutes (20-23 DnaA monomers) required shifting DnaA-ssDNA on/off rates from 100 mM^-1s^-1 / 0.55 s^-1 to 140 mM^-1s^-1 / 0.42 s^-1; scaling mRNA-ribosome binding 1.3-fold and mRNA-degradosome binding 0.3-fold gave active fractions of 55% for ribosomes (881), 70% for RNAP (176), and 10% for degradosomes (192), while 81 genes (<60 copies) went untranscribed in 1-3 cells. · Formation of macromolecular complexes; Activity of gene expression complexes and…; Replication initiation kinetics are sensitive in…; Figure legends

  4. ShownMetabolic NTP fluctuations throttle transcription elongation prior to volume doubling, and random 3D diffusion produces near-binomial partitioning of macromolecules to daughter cells.

    Depletion of UTP and GTP pools to averages around 0.3 mM reduced average dnaA transcription elongation rates to ~55% of maximum (~11 nt/s, range 30-80%) until cell volume doubled (radius growing from 200 to 250 nm), while post-division distributions of ribosomes, degradosomes, PtsG, and GapDH across daughter cells followed unbiased binomial-like distributions. · 4D whole-cell modeling reflects experiment; Genetic information reactions and metabolism are…; Partitioning of macromolecules to daughter cells…

Methods, data and reproducibility

The authors built a hybrid 4D whole-cell model of JCVI-syn3A (543 kbp genome, 493 genes) coupling a 10 nm lattice reaction-diffusion solver in Lattice Microbes v2.5 (hooked every 12.5 ms) with 1 s Gillespie CME transcription, 1 s LSODA ODE metabolism, and 4 s GPU LAMMPS/Kokkos Brownian dynamics of a 10 bp/bead chromosome. Predictions were validated by Illumina MiSeq sequencing and 3D Airyscan imaging.

  • Simulated 4D whole-cell trajectories of JCVI-syn3A across a complete cell cycle · JCVI-syn3A (synthetic minimal bacterium) · n = 50 replicate simulated cells · CP016816.2
  • Illumina MiSeq paired-end whole-genome DNA sequencing of JCVI-syn3A-mCherry in exponential (8 h) and stationary (21 h) growth phases · JCVI-syn3A · n = 3 biological replicates per growth phase · PRJNA1257452
  • Three-color 3D Airyscan super-resolution confocal fluorescence microscopy of fixed JCVI-syn3B cells expressing FtsZ-mCherry · JCVI-syn3B · n = 2,968 segmented cells (1,319 analyzed; 1,649 excluded)
  • Bacterial replisome structure used to parameterize minimum DnaA filament length for DnaB helicase loading · bacteria · n = 1 structure · PDB: 5IKN
Methods & tools
Lattice Microbes (v2.5), pyLM, jLM, LAMMPS (with Kokkos), btree chromo, odeCELL, ODEPACK (LSODA), cobrapy, FreeDTS, Bowtie 2 (v2.5.4), samtools (v1.21), VMD
Reagents & environment
Lattice Microbes v2.5 (pyLM, jLM), LAMMPS with custom Kokkos fix brownian/fix addforce, btree chromo, sc chain generation, odeCELL, ODEPACK LSODA, cobrapy, FreeDTS, Bowtie 2 v2.5.4, samtools v1.21, bcl2fastq v2.20, MultiQC, ZEN, VMD
Assay conditions & compute
4 to 6 days on 2 NVIDIA A100 GPUs per simulated cell cycle (~15,000 GPU hours total for 50 replicate cells); single full-timestep trajectory estimated at >80 TB
Reproducibility gaps
Model source code and 50-cell trajectory archives are stated as to be released on GitHub and Zenodo upon publication acceptance rather than via active URLs in the text.
Limitations (4)
  • Each simulated cell cycle requires 4 to 6 days on two NVIDIA A100 GPUs (~15,000 GPU hours for 50 cells), restricting population sample sizes, and RDME frames are recorded only every 1 s because saving every timestep would exceed 80 TB per trajectory. · stated by the authors
  • The model omits polysomes (allowing only one ribosome per mRNA), polycistronic operon co-transcription, coupled transcription-translation, and coupled membrane translation-translocation, resulting in underproduction of long proteins (>3 kb). · stated by the authors
  • Cell division morphology is prescribed geometrically as two overlapping spheres without a mechanistic kinetic and mechanical model of FtsZ ring polymerization and constriction. · stated by the authors
  • Daughter chromosome segregation relies on an uncalibrated 12 pN repulsive force, simplified one-sided SMC loop extrusion with a fixed 4 s dwell time, and omits DNA torsional supercoiling potentials. · stated by the authors

Read this if you develop whole-cell kinetic or spatial reaction-diffusion simulations, polymer models of bacterial chromosomes, or systems biology models of minimal synthetic cells.

Citations
44
FWCI
88.24
Field percentile
top 1%

Abstract

We present a whole-cell spatial and kinetic model for the 100 minute cell cycle of the genetically minimal bacterium, JCVI-syn3A. This is the first simulation of a complete cell cycle in 4D including all genetic information processes, metabolic networks, growth, and cell division. Integrating hybrid computational methods, dynamics of the morphological transformations were achieved. Growth is driven by synthesis of lipids and membrane proteins and constrained by new fluorescence imaging data. Chromosome replication and segregation is controlled by essential SMC and topoisomerase proteins in Brownian dynamics simulations with replication rates responding to dNTP pools from metabolism. The model captures the origin to terminus ratio measured in our DNA sequencing and recovers other experimental measurements like doubling time, mRNA half-lives, protein distributions, and ribosome counts. Because of stochasticity, each replicate cell is unique. Not only do we predict average behavior for partitioning to daughter cells, we predict the heterogeneity among them.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISPelletier '21Witz '19Colin '21Ahmadian '20Micali '18Taoma '24
62% semantic matchSynthetic Biology
James F. Pelletier, Lijie Sun, Kim S. Wise+9 more·Cell·Mar 29, 2021· 142 citations