NUP155 insufficiency recalibrates a pluripotent transcriptome with network remodeling of a cardiogenic signaling module

Abstract

Background: Atrial fibrillation is a cardiac disease driven by numerous idiopathic etiologies. NUP155 is a nuclear pore complex protein that has been identified as a clinical driver of atrial fibrillation, yet the precise mechanism is unknown. The present study employs a systems biology algorithm to identify effects of NUP155 disruption on cardiogenicity in a model of stem cell-derived differentiation.

Methods: Embryonic stem (ES) cell lines (n = 5) with truncated NUP155 were cultured in parallel with wild type (WT) ES cells (n = 5), and then harvested for RNAseq. Samples were run on an Illumina HiSeq 2000. Reads were analyzed using Strand NGS, Cytoscape, DAVID and Ingenuity Pathways Analysis to deconvolute the NUP155-disrupted transcriptome. Network topological analysis identified key features that controlled framework architecture and functional enrichment.

Results: In NUP155 truncated ES cells, significant expression changes were detected in 326 genes compared to WT. These genes segregated into clusters that enriched for specific gene ontologies. Deconvolution of the collective framework into discrete sub-networks identified a module with the highest score that enriched for Cardiovascular System Development, and revealed NTRK1/TRKA and SRSF2/SC35 as critical hubs within this cardiogenic module.

Conclusions: The strategy of pluripotent transcriptome deconvolution used in the current study identified a novel association of NUP155 with potential drivers of arrhythmogenic AF. Here, NUP155 regulates cardioplasticity of a sub-network embedded within a larger framework of genome integrity, and exemplifies how transcriptome cardiogenicity in an embryonic stem cell genome is recalibrated by nucleoporin dysfunction.

Keywords: Atrial fibrillation; Embryonic stem cells; NUP155; Network bioinformatics; RNAseq.

Fig. 1

Nucleoporins in cardiac differentiation. a Nup153, Nup155, Nup85, Rae1, and Tpr expression profiles during cardiac specification. Undifferentiated stem cells (ES LIF+); early differentiated stem cells (ES LIF-); cardiac precursors (CP), cardiomyocytes (CM). b ExAC constraint metrics and pLI values for Nup153, Nup155, Nup85, Rae1, and Tpr. c Venn diagrams intersect discrete gene groups independently parsed from the same high throughput dataset to identify nucleoporins (nup) that emerge as robust candidates involved in cardiogenesis. Nup155 is the highest priority molecule (p < 0.001). SOM = self-organizing map; QC = quality control

Fig. 2

Contractile NUP155 deficient embryoid bodies exhibit dysrhythmia. a Embryoid bodies (EBs) were loaded with Fluo4-AM to visualize Ca²⁺ handling during systolic and diastolic phases of contractile cycling, as indicated. Color legend to left of image identifies fluorescence of regions that range from high Ca²⁺ concentration in red, medium concentration in cyan, to low concentration in blue. b Ca²⁺ handling for each region of interest (ROI), with timescale in seconds (s) and range of fluorescent intensity units (i.u.) indicated in lower right. c, d Visualization and measurement of contractile EBs following treatment with 10 mM isoproterenol (Iso). e, f Untreated beating areas in NUP155 deficient EBs demonstrates increased frequency of contractile cycling, with variable Ca²⁺ handling. g, h Agonist treatment of NUP155 deficient EBs aggravates the irregular contractile cycling observed in unstimulated controls that ranged in severity from hypercontractility to loss of rhythmic contraction. I Measurement of intervals between peaks highlight a significant decrease of time between Ca²⁺ waves in NUP155^+/− EBs compared to controls, independent of isoproterenol treatment (n = 3, *p < 0.05 vs WT Ctrl, **p < 0.05 vs WT Iso). j Changes in mean amplitude of Ca²⁺ waves. WT Iso treated, and NUP155^+/− EBs with and without Iso treatment did not show significant differences, but were significantly decreased compared to WT Ctrl. (n = 3, *p < 0.05 vs WT Ctrl)

Fig. 3

Molecular signature of NUP155 truncation in a pluripotent genome. Deep transcriptome profiling of WT and NUP155^+/− embryonic stem cells was performed using RNAseq. a Principal component analysis (PCA) revealed distinct hallmark gene expression profiles with clear segregation of WT from NUP155^+/− transcriptomes. Filled circles represent ES (dark grey) and NUP155^+/− (light green) transcriptomes of distinct biological replicates, plotted in a three dimensional volumetric space. Axes: X – PC1 (24.08%), Y – PC2 (13.2%), Z – PC3 (10.38%). b Pairwise correlation of samples reveals reproducible clustering of discrete up and downregulated gene expression patterns that define ES and NUP155^+/− populations. Lower right: colorscale indicates normalized intensity, where red, yellow, and blue represent upregulated, no change, and downregulated trends, respectively. c Volcano plot of gene expression changes to enumerate up and downregulated mRNA in the NUP155^+/− transcriptome, according to criteria of absolute Fold Change (FC) > 2.0 and p < 0.05. Filled circles in red represent up-regulated genes that meet this criteria, while blue circles represent downregulated transcripts. Circles in grey indicate genes that fall below the indicated threshold values. A total of 326 genes were identified that met the filtering parameters, with 176 up and 150 downregulated, respectively. PC – Principal Component

Fig. 4

Network cartography of a nucleoporin-disrupted transcriptome. (a) The collective transcriptome inclusive of up and downregulated transcripts were analyzed by Ingenuity Pathways Analysis to identify experimentally observed interactions among the 326 genes. The network generated from this data was visualized in a circular layout that positions nodes circumferentially with their connections (edges) plotted diametrically. Singletons are network nodes with only one connection to the larger network and are arrayed on the outside of the circle plot. This layout emphasizes nodes that have high edge density, seen in this network on the right. Right panel: Magnification of network arc with high edge density. Nodes were colored properties according to degree, or number of connections, where high degree is represented by dark red and low degree in white, shown here in the colorscale above panel. b Topological analysis revealed a clustering coefficient distribution associated with hierarchical network structure. Inset: Degree distribution demonstrates a power law relationship indicative of scale-free architecture (c) Neighborhood connectivity plot identifies dissortative nature of the network, where highly connected nodes tend to connect to nodes with a lower number of edges. d Nodes with high betweenness centrality are critical to maintaining network integrity as they connect other regions of the network to one another. APP, HNF4A, TP53, NTRK1/TRKA, and CTNNB1 possessed distinct betweenness centrality scores that segregated them from other nodes in the network. Inset: Legend identifies genes with the topmost betweenness centrality scores, ranked in order from highest to lowest, and are colored to facilitate identification within the plot. e Closeness centrality scores are important for speed of informational transmission within a network. Here, nodes with the highest closeness centrality clustered together. The nodes prioritized for high betweenness centrality measures were identical to the molecules with critical closeness centrality scores. Inset: Legend depicts nodes rank ordered from high to low. Identity of nodes with discrete centrality metrics are preserved as identical, yet distinct reprioritization of those molecules is observed on comparison of closeness versus betweenness

Fig. 5

Deconvolution of modular functional enrichment within a sub-network of the NUP155^+/− transcriptome. a Sub-networks that comprise the larger network possess characteristic identities at the mesoscopic level. Identification of the most significant sub-network revealed robust and consistent functional enrichment in Cardiovascular System Development, and incorporated a variety of up and down-regulated genes. Magnification of hubs NTRK1/TRKA and SRSF2/SC35 shown on the right. b, c Betweenness and closeness centrality plots of this small network confirm the same molecule, NTRK1/TRKA, as critical for integration and information transmission within the module, labeled in both plots and highlighted in red. SRSF2/SC35 possessed the next highest betweenness and closeness centrality measures, labeled and highlighted in green. d, e RNAseq abundances for NTRK1/TRKA and SRSF2 confirmed significant expression changes for both transcripts. Normalized intensities shown on y-axis