Integrative Structure Modeling: Overview and Assessment.

Integrative structure modeling computationally combines data from multiple sources of information with the aim of obtaining structural insights that are not revealed by any single approach alone. In the first part of this review, we survey the commonly used sources of structural information and the computational aspects of model building. Throughout the past decade, integrative modeling was applied to various biological systems, with a focus on large protein complexes. Recent progress in the field of cryo-electron microscopy (cryo-EM) has resolved many of these complexes to near-atomic resolution. In the second part of this review, we compare a range of published integrative models with their higher-resolution counterparts with the aim of critically assessing their accuracy. This comparison gives a favorable view of integrative modeling and demonstrates its ability to yield accurate and informative results. We discuss possible roles of integrative modeling in the new era of cryo-EM and highlight future challenges and directions.

CombFold: predicting structures of large protein assemblies using a combinatorial assembly algorithm and AlphaFold2.

Deep learning models, such as AlphaFold2 and RosettaFold, enable high-accuracy protein structure prediction. However, large protein complexes are still challenging to predict due to their size and the complexity of interactions between multiple subunits. Here we present CombFold, a combinatorial and hierarchical assembly algorithm for predicting structures of large protein complexes utilizing pairwise interactions between subunits predicted by AlphaFold2. CombFold accurately predicted (TM-score >0.7) 72% of the complexes among the top-10 predictions in two datasets of 60 large, asymmetric assemblies. Moreover, the structural coverage of predicted complexes was 20% higher compared to corresponding Protein Data Bank entries. We applied the method on complexes from Complex Portal with known stoichiometry but without known structure and obtained high-confidence predictions. CombFold supports the integration of distance restraints based on crosslinking mass spectrometry and fast enumeration of possible complex stoichiometries. CombFold's high accuracy makes it a promising tool for expanding structural coverage beyond monomeric proteins.

Discovering predisposing genes for hereditary breast cancer using deep learning.

Breast cancer (BC) is the most common malignancy affecting Western women today. It is estimated that as many as 10% of BC cases can be attributed to germline variants. However, the genetic basis of the majority of familial BC cases has yet to be identified. Discovering predisposing genes contributing to familial BC is challenging due to their presumed rarity, low penetrance, and complex biological mechanisms. Here, we focused on an analysis of rare missense variants in a cohort of 12 families of Middle Eastern origins characterized by a high incidence of BC cases. We devised a novel, high-throughput, variant analysis pipeline adapted for family studies, which aims to analyze variants at the protein level by employing state-of-the-art machine learning models and three-dimensional protein structural analysis. Using our pipeline, we analyzed 1218 rare missense variants that are shared between affected family members and classified 80 genes as candidate pathogenic. Among these genes, we found significant functional enrichment in peroxisomal and mitochondrial biological pathways which segregated across seven families in the study and covered diverse ethnic groups. We present multiple evidence that peroxisomal and mitochondrial pathways play an important, yet underappreciated, role in both germline BC predisposition and BC survival.

Structural basis for Mis18 complex assembly and its implications for centromere maintenance.

The centromere, defined by the enrichment of CENP-A (a Histone H3 variant) containing nucleosomes, is a specialised chromosomal locus that acts as a microtubule attachment site. To preserve centromere identity, CENP-A levels must be maintained through active CENP-A loading during the cell cycle. A central player mediating this process is the Mis18 complex (Mis18α, Mis18ß and Mis18BP1), which recruits the CENP-A-specific chaperone HJURP to centromeres for CENP-A deposition. Here, using a multi-pronged approach, we characterise the structure of the Mis18 complex and show that multiple hetero- and homo-oligomeric interfaces facilitate the hetero-octameric Mis18 complex assembly composed of 4 Mis18α, 2 Mis18ß and 2 Mis18BP1. Evaluation of structure-guided/separation-of-function mutants reveals structural determinants essential for cell cycle controlled Mis18 complex assembly and centromere maintenance. Our results provide new mechanistic insights on centromere maintenance, highlighting that while Mis18α can associate with centromeres and deposit CENP-A independently of Mis18ß, the latter is indispensable for the optimal level of CENP-A loading required for preserving the centromere identity.

ScanNet: an interpretable geometric deep learning model for structure-based protein binding site prediction.

Predicting the functional sites of a protein from its structure, such as the binding sites of small molecules, other proteins or antibodies, sheds light on its function in vivo. Currently, two classes of methods prevail: machine learning models built on top of handcrafted features and comparative modeling. They are, respectively, limited by the expressivity of the handcrafted features and the availability of similar proteins. Here, we introduce ScanNet, an end-to-end, interpretable geometric deep learning model that learns features directly from 3D structures. ScanNet builds representations of atoms and amino acids based on the spatio-chemical arrangement of their neighbors. We train ScanNet for detecting protein-protein and protein-antibody binding sites, demonstrate its accuracy-including for unseen protein folds-and interpret the filters learned. Finally, we predict epitopes of the SARS-CoV-2 spike protein, validating known antigenic regions and predicting previously uncharacterized ones. Overall, ScanNet is a versatile, powerful and interpretable model suitable for functional site prediction tasks. A webserver for ScanNet is available from http://bioinfo3d.cs.tau.ac.il/ScanNet/ .

Dual neutralization of influenza virus hemagglutinin and neuraminidase by a bispecific antibody leads to improved antiviral activity.

Targeting multiple viral proteins is pivotal for sustained suppression of highly mutable viruses. In recent years, broadly neutralizing antibodies that target the influenza virus hemagglutinin and neuraminidase glycoproteins have been developed, and antibody monotherapy has been tested in preclinical and clinical studies to treat or prevent influenza virus infection. However, the impact of dual neutralization of the hemagglutinin and neuraminidase on the course of infection, as well as its therapeutic potential, has not been thoroughly tested. For this purpose, we generated a bispecific antibody that neutralizes both the hemagglutinin and the neuraminidase of influenza viruses. We demonstrated that this bispecific antibody has a dual-antiviral activity as it blocks infection and prevents the release of progeny viruses from the infected cells. We show that dual neutralization of the hemagglutinin and the neuraminidase by a bispecific antibody is advantageous over monoclonal antibody combination as it resulted an improved neutralization capacity and augmented the antibody effector functions. Notably, the bispecific antibody showed enhanced antiviral activity in influenza virus-infected mice, reduced mice mortality, and limited the virus mutation profile upon antibody administration. Thus, dual neutralization of the hemagglutinin and neuraminidase could be effective in controlling influenza virus infection.

RhoMax: Computational Prediction of Rhodopsin Absorption Maxima Using Geometric Deep Learning.

Microbial rhodopsins (MRs) are a diverse and abundant family of photoactive membrane proteins that serve as model systems for biophysical techniques. Optogenetics utilizes genetic engineering to insert specialized proteins into specific neurons or brain regions, allowing for manipulation of their activity through light and enabling the mapping and control of specific brain areas in living organisms. The obstacle of optogenetics lies in the fact that light has a limited ability to penetrate biological tissues, particularly blue light in the visible spectrum. Despite this challenge, most optogenetic systems rely on blue light due to the scarcity of red-shifted opsins. Finding additional red-shifted rhodopsins would represent a major breakthrough in overcoming the challenge of limited light penetration in optogenetics. However, determining the wavelength absorption maxima for rhodopsins based on their protein sequence is a significant hurdle. Current experimental methods are time-consuming, while computational methods lack accuracy. The paper introduces a new computational approach called RhoMax that utilizes structure-based geometric deep learning to predict the absorption wavelength of rhodopsins solely based on their sequences. The method takes advantage of AlphaFold2 for accurate modeling of rhodopsin structures. Once trained on a balanced train set, RhoMax rapidly and precisely predicted the maximum absorption wavelength of more than half of the sequences in our test set with an accuracy of 0.03 eV. By leveraging computational methods for absorption maxima determination, we can drastically reduce the time needed for designing new red-shifted microbial rhodopsins, thereby facilitating advances in the field of optogenetics.

Targeted in situ cross-linking mass spectrometry and integrative modeling reveal the architectures of three proteins from SARS-CoV-2.

Atomic structures of several proteins from the coronavirus family are still partial or unavailable. A possible reason for this gap is the instability of these proteins outside of the cellular context, thereby prompting the use of in-cell approaches. In situ cross-linking and mass spectrometry (in situ CLMS) can provide information on the structures of such proteins as they occur in the intact cell. Here, we applied targeted in situ CLMS to structurally probe Nsp1, Nsp2, and nucleocapsid (N) proteins from severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and obtained cross-link sets with an average density of one cross-link per 20 residues. We then employed integrative modeling that computationally combined the cross-linking data with domain structures to determine full-length atomic models. For the Nsp2, the cross-links report on a complex topology with long-range interactions. Integrative modeling with structural prediction of individual domains by the AlphaFold2 system allowed us to generate a single consistent all-atom model of the full-length Nsp2. The model reveals three putative metal binding sites and suggests a role for Nsp2 in zinc regulation within the replication-transcription complex. For the N protein, we identified multiple intra- and interdomain cross-links. Our integrative model of the N dimer demonstrates that it can accommodate three single RNA strands simultaneously, both stereochemically and electrostatically. For the Nsp1, cross-links with the 40S ribosome were highly consistent with recent cryogenic electron microscopy structures. These results highlight the importance of cellular context for the structural probing of recalcitrant proteins and demonstrate the effectiveness of targeted in situ CLMS and integrative modeling.

A deep learning model for predicting optimal distance range in crosslinking mass spectrometry data.

Macromolecular assemblies play an important role in all cellular processes. While there has recently been significant progress in protein structure prediction based on deep learning, large protein complexes cannot be predicted with these approaches. The integrative structure modeling approach characterizes multi-subunit complexes by computational integration of data from fast and accessible experimental techniques. Crosslinking mass spectrometry is one such technique that provides spatial information about the proximity of crosslinked residues. One of the challenges in interpreting crosslinking datasets is designing a scoring function that, given a structure, can quantify how well it fits the data. Most approaches set an upper bound on the distance between Cα atoms of crosslinked residues and calculate a fraction of satisfied crosslinks. However, the distance spanned by the crosslinker greatly depends on the neighborhood of the crosslinked residues. Here, we design a deep learning model for predicting the optimal distance range for a crosslinked residue pair based on the structures of their neighborhoods. We find that our model can predict the distance range with the area under the receiver-operator curve of 0.86 and 0.7 for intra- and inter-protein crosslinks, respectively. Our deep scoring function can be used in a range of structure modeling applications.

Biallelic variants in the RNA exosome gene EXOSC5 are associated with developmental delays, short stature, cerebellar hypoplasia and motor weakness.

The RNA exosome is an essential ribonuclease complex required for processing and/or degradation of both coding and non-coding RNAs. We identified five patients with biallelic variants in EXOSC5, which encodes a structural subunit of the RNA exosome. The clinical features of these patients include failure to thrive, short stature, feeding difficulties, developmental delays that affect motor skills, hypotonia and esotropia. Brain MRI revealed cerebellar hypoplasia and ventriculomegaly. While we ascertained five patients, three patients with distinct variants of EXOSC5 were studied in detail. The first patient had a deletion involving exons 5-6 of EXOSC5 and a missense variant, p.Thr114Ile, that were inherited in trans, the second patient was homozygous for p.Leu206His and the third patient had paternal isodisomy for chromosome 19 and was homozygous for p.Met148Thr. The additional two patients ascertained are siblings who had an early frameshift mutation in EXOSC5 and the p.Thr114Ile missense variant that were inherited in trans. We employed three complementary approaches to explore the requirement for EXOSC5 in brain development and assess consequences of pathogenic EXOSC5 variants. Loss of function for exosc5 in zebrafish results in shortened and curved tails/bodies, reduced eye/head size and edema. We modeled pathogenic EXOSC5 variants in both budding yeast and mammalian cells. Some of these variants cause defects in RNA exosome function as well as altered interactions with other RNA exosome subunits. These findings expand the number of genes encoding RNA exosome subunits linked to human disease while also suggesting that disease mechanism varies depending on the specific pathogenic variant.

Formation of a repressive complex in the mammalian circadian clock is mediated by the secondary pocket of CRY1.

The basic helix-loop-helix PAS domain (bHLH-PAS) transcription factor CLOCK:BMAL1 (brain and muscle Arnt-like protein 1) sits at the core of the mammalian circadian transcription/translation feedback loop. Precise control of CLOCK:BMAL1 activity by coactivators and repressors establishes the ∼24-h periodicity of gene expression. Formation of a repressive complex, defined by the core clock proteins cryptochrome 1 (CRY1):CLOCK:BMAL1, plays an important role controlling the switch from repression to activation each day. Here we show that CRY1 binds directly to the PAS domain core of CLOCK:BMAL1, driven primarily by interaction with the CLOCK PAS-B domain. Integrative modeling and solution X-ray scattering studies unambiguously position a key loop of the CLOCK PAS-B domain in the secondary pocket of CRY1, analogous to the antenna chromophore-binding pocket of photolyase. CRY1 docks onto the transcription factor alongside the PAS domains, extending above the DNA-binding bHLH domain. Single point mutations at the interface on either CRY1 or CLOCK disrupt formation of the ternary complex, highlighting the importance of this interface for direct regulation of CLOCK:BMAL1 activity by CRY1.

Cross-activating c-Met/ß1 integrin complex drives metastasis and invasive resistance in cancer.

The molecular underpinnings of invasion, a hallmark of cancer, have been defined in terms of individual mediators but crucial interactions between these mediators remain undefined. In xenograft models and patient specimens, we identified a c-Met/ß1 integrin complex that formed during significant invasive oncologic processes: breast cancer metastases and glioblastoma invasive resistance to antiangiogenic VEGF neutralizing antibody, bevacizumab. Inducing c-Met/ß1 complex formation through an engineered inducible heterodimerization system promoted features crucial to overcoming stressors during metastases or antiangiogenic therapy: migration in the primary site, survival under hypoxia, and extravasation out of circulation. c-Met/ß1 complex formation was up-regulated by hypoxia, while VEGF binding VEGFR2 sequestered c-Met and ß1 integrin, preventing their binding. Complex formation promoted ligand-independent receptor activation, with integrin-linked kinase phosphorylating c-Met and crystallography revealing the c-Met/ß1 complex to maintain the high-affinity ß1 integrin conformation. Site-directed mutagenesis verified the necessity for c-Met/ß1 binding of amino acids predicted by crystallography to mediate their extracellular interaction. Far-Western blotting and sequential immunoprecipitation revealed that c-Met displaced α5 integrin from ß1 integrin, creating a complex with much greater affinity for fibronectin (FN) than α5ß1. Thus, tumor cells adapt to microenvironmental stressors induced by metastases or bevacizumab by coopting receptors, which normally promote both cell migration modes: chemotaxis, movement toward concentrations of environmental chemoattractants, and haptotaxis, movement controlled by the relative strengths of peripheral adhesions. Tumor cells then redirect these receptors away from their conventional binding partners, forming a powerful structural c-Met/ß1 complex whose ligand-independent cross-activation and robust affinity for FN drive invasive oncologic processes.

De novo STXBP1 mutation in a child with developmental delay and spasticity reveals a major structural alteration in the interface with syntaxin 1A.

STXBP1, also known as Munc-18, is a master regulator of neurotransmitter release and synaptic function in the human brain through its direct interaction with syntaxin 1A. STXBP1 binds syntaxin 1A is an inactive conformational state. STXBP1 decreases its binding affinity to syntaxin upon phosphorylation, enabling syntaxin 1A to engage in the SNARE complex, leading to neurotransmitter release. STXBP1-related disorders are well characterized by encephalopathy with epilepsy, and a diverse range of neurological and neurodevelopmental conditions. Through exome sequencing of a child with developmental delay, hypotonia, and spasticity, we found a novel de novo insertion mutation of three nucleotides in the STXBP1 coding region, resulting in an additional arginine after position 39 (R39dup). Inconclusive results from state-of-the-art variant prediction tools mandated a structure-based approach using molecular dynamics (MD) simulations of the STXBP1-syntaxin 1A complex. Comparison of the interaction interfaces of the wild-type and the R39dup complexes revealed a reduced interaction surface area in the mutant, leading to destabilization of the protein complex. Moreover, the decrease in affinity toward syntaxin 1A is similar for the phosphorylated STXBP1 and the R39dup. We applied the same MD methodology to seven additional previously reported STXBP1 mutations and reveal that the stability of the STXBP1-syntaxin 1A interface correlates with the reported clinical phenotypes. This study provides a direct link between the outcome of a novel variant in STXBP1 and protein structure and dynamics. The structural change upon mutation drives an alteration in synaptic function.

De novo, deleterious sequence variants that alter the transcriptional activity of the homeoprotein PBX1 are associated with intellectual disability and pleiotropic developmental defects.

We present eight patients with de novo, deleterious sequence variants in the PBX1 gene. PBX1 encodes a three amino acid loop extension (TALE) homeodomain transcription factor that forms multimeric complexes with TALE and HOX proteins to regulate target gene transcription during development. As previously reported, Pbx1 homozygous mutant mice (Pbx1-/-) develop malformations and hypoplasia or aplasia of multiple organs, including the craniofacial skeleton, ear, branchial arches, heart, lungs, diaphragm, gut, kidneys, and gonads. Clinical findings similar to those in Pbx mutant mice were observed in all patients with varying expressivity and severity, including external ear anomalies, abnormal branchial arch derivatives, heart malformations, diaphragmatic hernia, renal hypoplasia and ambiguous genitalia. All patients but one had developmental delays. Previously reported patients with congenital anomalies affecting the kidney and urinary tract exhibited deletions and loss of function variants in PBX1. The sequence variants in our cases included missense substitutions adjacent to the PBX1 homeodomain (p.Arg184Pro, p.Met224Lys, and p.Arg227Pro) or within the homeodomain (p.Arg234Pro, and p.Arg235Gln), whereas p.Ser262Glnfs*2, and p.Arg288* yielded truncated PBX1 proteins. Functional studies on five PBX1 sequence variants revealed perturbation of intrinsic, PBX-dependent transactivation ability and altered nuclear translocation, suggesting abnormal interactions between mutant PBX1 proteins and wild-type TALE or HOX cofactors. It is likely that the mutations directly affect the transcription of PBX1 target genes to impact embryonic development. We conclude that deleterious sequence variants in PBX1 cause intellectual disability and pleiotropic malformations resembling those in Pbx1 mutant mice, arguing for strong conservation of gene function between these two species.

CRB2 mutations produce a phenotype resembling congenital nephrosis, Finnish type, with cerebral ventriculomegaly and raised alpha-fetoprotein.

We report five fetuses and a child from three families who shared a phenotype comprising cerebral ventriculomegaly and echogenic kidneys with histopathological findings of congenital nephrosis. The presenting features were greatly elevated maternal serum alpha-fetoprotein (MSAFP) or amniotic fluid alpha-fetoprotein (AFAFP) levels or abnormalities visualized on ultrasound scan during the second trimester of pregnancy. Exome sequencing revealed deleterious sequence variants in Crumbs, Drosophila, Homolog of, 2 (CRB2) consistent with autosomal-recessive inheritance. Two fetuses with cerebral ventriculomegaly and renal microcysts were compound heterozygotes for p.Asn800Lys and p.Trp759Ter, one fetus with renal microcysts was a compound heterozygote for p.Glu643Ala and p.Asn800Lys, and one child with cerebral ventriculomegaly, periventricular heterotopias, echogenic kidneys, and renal failure was homozygous for p.Arg633Trp in CRB2. Examination of the kidneys in one fetus showed tubular cysts at the corticomedullary junction and diffuse effacement of the epithelial foot processes and microvillous transformation of the renal podocytes, findings that were similar to those reported in congenital nephrotic syndrome, Finnish type, that is caused by mutations in nephrin (NPHS1). Loss of function for crb2b and nphs1 in Danio rerio were previously shown to result in loss of the slit diaphragms of the podocytes, leading to the hypothesis that nephrosis develops from an inability to develop a functional glomerular barrier. We conclude that the phenotype associated with CRB2 mutations is pleiotropic and that the condition is an important consideration in the evaluation of high MSAFP/AFAFP where a renal cause is suspected.

FoXS, FoXSDock and MultiFoXS: Single-state and multi-state structural modeling of proteins and their complexes based on SAXS profiles.

Small Angle X-ray Scattering (SAXS) is an increasingly common and useful technique for structural characterization of molecules in solution. A SAXS experiment determines the scattering intensity of a molecule as a function of spatial frequency, termed SAXS profile. Here, we describe three web servers for modeling atomic structures based on SAXS profiles. FoXS (Fast X-Ray Scattering) rapidly computes a SAXS profile of a given atomistic model and fits it to an experimental profile. FoXSDock docks two rigid protein structures based on a SAXS profile of their complex. MultiFoXS computes a population-weighted ensemble starting from a single input structure by fitting to a SAXS profile of the protein in solution. We describe the interfaces and capabilities of the servers (salilab.org/foxs), followed by demonstrating their application on Interleukin-33 (IL-33) and its primary receptor ST2.

DksA regulates RNA polymerase in Escherichia coli through a network of interactions in the secondary channel that includes Sequence Insertion 1.

Sensing and responding to nutritional status is a major challenge for microbial life. In Escherichia coli, the global response to amino acid starvation is orchestrated by guanosine-3',5'-bisdiphosphate and the transcription factor DksA. DksA alters transcription by binding to RNA polymerase and allosterically modulating its activity. Using genetic analysis, photo-cross-linking, and structural modeling, we show that DksA binds and acts upon RNA polymerase through prominent features of both the nucleotide-access secondary channel and the active-site region. This work is, to our knowledge, the first demonstration of a molecular function for Sequence Insertion 1 in the ß subunit of RNA polymerase and significantly advances our understanding of how DksA binds to RNA polymerase and alters transcription.

Memdock: an α-helical membrane protein docking algorithm.

MOTIVATION: A wide range of fundamental biological processes are mediated by membrane proteins. Despite their large number and importance, less than 1% of all 3D protein structures deposited in the Protein Data Bank are of membrane proteins. This is mainly due to the challenges of crystallizing such proteins or performing NMR spectroscopy analyses. All the more so, there is only a small number of membrane protein-protein complexes with known structure. Therefore, developing computational tools for docking membrane proteins is crucial. Numerous methods for docking globular proteins exist, however few have been developed especially for membrane proteins and designed to address docking within the lipid bilayer environment. RESULTS: We present a novel algorithm, Memdock, for docking α-helical membrane proteins which takes into consideration the lipid bilayer environment for docking as well as for refining and ranking the docking candidates. We show that our algorithm improves both the docking accuracy and the candidates ranking compared to a standard protein-protein docking algorithm. AVAILABILITY AND IMPLEMENTATION: http://bioinfo3d.cs.tau.ac.il/Memdock/ CONTACTS: namih@tau.ac.il or wolfson@tau.ac.il SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Prediction of homoprotein and heteroprotein complexes by protein docking and template-based modeling: A CASP-CAPRI experiment.

We present the results for CAPRI Round 30, the first joint CASP-CAPRI experiment, which brought together experts from the protein structure prediction and protein-protein docking communities. The Round comprised 25 targets from amongst those submitted for the CASP11 prediction experiment of 2014. The targets included mostly homodimers, a few homotetramers, and two heterodimers, and comprised protein chains that could readily be modeled using templates from the Protein Data Bank. On average 24 CAPRI groups and 7 CASP groups submitted docking predictions for each target, and 12 CAPRI groups per target participated in the CAPRI scoring experiment. In total more than 9500 models were assessed against the 3D structures of the corresponding target complexes. Results show that the prediction of homodimer assemblies by homology modeling techniques and docking calculations is quite successful for targets featuring large enough subunit interfaces to represent stable associations. Targets with ambiguous or inaccurate oligomeric state assignments, often featuring crystal contact-sized interfaces, represented a confounding factor. For those, a much poorer prediction performance was achieved, while nonetheless often providing helpful clues on the correct oligomeric state of the protein. The prediction performance was very poor for genuine tetrameric targets, where the inaccuracy of the homology-built subunit models and the smaller pair-wise interfaces severely limited the ability to derive the correct assembly mode. Our analysis also shows that docking procedures tend to perform better than standard homology modeling techniques and that highly accurate models of the protein components are not always required to identify their association modes with acceptable accuracy. Proteins 2016; 84(Suppl 1):323-348. © 2016 Wiley Periodicals, Inc.

Structural characterization by cross-linking reveals the detailed architecture of a coatomer-related heptameric module from the nuclear pore complex.

Most cellular processes are orchestrated by macromolecular complexes. However, structural elucidation of these endogenous complexes can be challenging because they frequently contain large numbers of proteins, are compositionally and morphologically heterogeneous, can be dynamic, and are often of low abundance in the cell. Here, we present a strategy for the structural characterization of such complexes that has at its center chemical cross-linking with mass spectrometric readout. In this strategy, we isolate the endogenous complexes using a highly optimized sample preparation protocol and generate a comprehensive, high-quality cross-linking dataset using two complementary cross-linking reagents. We then determine the structure of the complex using a refined integrative method that combines the cross-linking data with information generated from other sources, including electron microscopy, X-ray crystallography, and comparative protein structure modeling. We applied this integrative strategy to determine the structure of the native Nup84 complex, a stable hetero-heptameric assembly (∼ 600 kDa), 16 copies of which form the outer rings of the 50-MDa nuclear pore complex (NPC) in budding yeast. The unprecedented detail of the Nup84 complex structure reveals previously unseen features in its pentameric structural hub and provides information on the conformational flexibility of the assembly. These additional details further support and augment the protocoatomer hypothesis, which proposes an evolutionary relationship between vesicle coating complexes and the NPC, and indicates a conserved mechanism by which the NPC is anchored in the nuclear envelope.