Project Summary
The human nuclear pore complex (NPC) consists of !1000 proteins of !30 different types. It is embedded in
the nuclear envelope, and controls macromolecular transport between the nucleus and cytoplasm. Mutations
in constituent proteins of the NPC have been connected to various human diseases including autoimmune dys-
functions, neurological diseases, cardiovascular disorders, and cancer. While static structures of the NPC have
been modeled, additional research is necessary to understand the assembly process of the NPC. An improved
understanding of NPC assembly pathways may elucidate the crucial role the NPC has played in eukaryotic evo-
lution and eventually decode the link between genetic alterations in NPC associated proteins and a variety of
human diseases. However, the size of the NPC complex and timescales approaching one hour for assembly
completion have posed challenges to previous studies. To overcome these difficulties, we propose to compute
a spatiotemporal integrative model of human NPC assembly in collaboration with experimental characterization
from Jan Ellenberg at the European Microbiology Laboratory, Heidelberg. To build our spatiotemporal integrative
model of NPC assembly, we will model static snapshots of intermediate states throughout interphase NPC as-
sembly, connect snapshots to produce assembly trajectories, and refine our model based on new experimental
characterization. Our initial snapshots of NPC assembly will be modeled by a Bayesian posterior model density
that explicitly scores our model based on its agreement with both experimental data and physical theories. Our
model at each time step will be informed by the coarse-grained structure of the fully assembled NPC, atomic struc-
tures of its components, physical theories, and cryo-electron tomograms and fluorescence correlation spectra of
the intermediate states. Next, we will consider implicit connections between these modeled intermediate states
to form trajectories. We will score this ensemble of trajectories based on both the probability of static structures
sampled along the trajectory and the probability of the implicit transitions between static structures. Finally, we
will verify our model through experimental collaboration with Dr. Ellenberg. Specifically, we will predict proteins
crucial for rate limiting steps in NPC assembly, and then test our model’s predictions with new auxin-degradation
experiments and correlative light-electron microscopy. If our model cannot sufficiently describe NPC assembly,
we can refine our model by reflecting this new data as a new likelihood function, incorporating this likelihood
into our Bayesian posterior model density, and reweighting our integrative model. This entire process will be re-
peated iteratively until we reach sufficient agreement between simulation and experiment. The end result will be
an experimentally informed and validated molecular model of NPC assembly. Moreover, we will have produced
a general methodological framework, implemented in our open-source Integrative Modeling Platform (IMP), to
enable characterization of dynamic biomolecular processes in general.
