Systems biology surveillance decrypts pathological transcriptome remodeling

A functional myocardium is the culmination of intricate dynamic signaling orchestrated by regulated processes [1, 2], and disruptions of cardiogenesis that lead to congenital cardiac defects occur in as high as 8 % of prenatal cases [3, 4]. Discrete molecular elements that give rise to inborn cardiac errors are numerous and precipitate developmental cardiopathology through multiple mechanisms including inhibition of cardiomyocyte proliferation and differentiation [5], abrogation of syncytial/electrical function [6], and disruption of nascent myocardial anatomy [7, 8]. Precise models of cardiac development that report phenotype derangement specific for a targeted gene have enhanced resolution of signaling axes that drive clinical pathology, yet these platforms are of limited use for anticipatory diagnosis of cardiopathology in advance of symptomatic presentation. Present methods to detect congenital heart defects are limited to time frames proximal to phenotype onset, which critically restricts opportunities for curative or palliative interventions.

Pre-manifestation in silico extrapolation of cardiac defects is feasible with the use of large volume, multidimensional biodata, as high throughput techniques generate profound datasets which could serve as the foundation for incisive dissection of complex molecular patterns [9]. Indeed, continuous development of bioinformatic tools expands the range of quantifiable properties of “omic” data, and provides the basis for advanced systems biology algorithms of deconvolution [10]. Resolution of the intricate molecular framework underlying phenotype dynamics using this interdisciplinary methodology has been successful [9, 11], yet parsing diverse and complex datasets remains an extant challenge.

Embryonic stem (ES) cells harbor a pluripotent genome with an engaged pre-differentiation molecular landscape [12, 13], and the naive transcriptome of an ES cell provides a unique source of material for exploration of the primordial gene network that supports normal development [14–17]. Disruptions within this principal transcriptome may ultimately drive pathological onset and disease progression, thus elucidation of primary transcriptome dynamics under normal and diseased conditions is required for advanced refinement of disease prediction [9, 18]. Case in point, ES cells that lack a functional calreticulin gene, i.e. calr ^−/− , abrogates calreticulin (CALR) expression, and provides a robust model for interrogation of a discrete mechanism of pathological cardiogenesis [19, 20]. Systems biology dissection of prioritized gene ontology remodeling in calr ^−/− ES cell lines revealed the underlying molecular network that drove ultrastructural, functional, and anatomical derangements in cardiac phenotype [18]. However, CALR is a Ca²⁺-handling chaperone [21, 22], and specific contributions of its particular functional domains, i.e. calcium binding versus protein folding, to final phenotype has remained uncharacterized.

Dissection of the potential contributions of each CALR domain to global phenotype generation would identify differential gene ontology remodeling specified by distinct regions within a multifunctional protein. This would refine and increase resolution of the primordial CALR-driven molecular atlas preceding organogenesis. Here, using an algorithm of bioinformatic deconvolution to unravel the intricate transcriptome networks harbored by calreticulin truncation variants, the contributions of the protein chaperoning domain (NP) versus the calcium handling (PC) region were investigated. Truncation variants, generated by transfection of CALR^−/− cell lines with constructs that exclusively expressed the protein chaperoning (CALR^−/−-NP) or the calcium handling (CALR^−/−-PC) domain, were used to determine differential Gene Ontology enrichments coded for by each CALR functional region. To build upon previous work, dysfunctional transcriptome recalibration preceding cardiogenesis was prioritized. The classification of Cardiovascular Development was enriched for by both truncation variants, though the down regulated cohort of genes in the transcriptome of the CALR^−/−-PC truncation variant exhibited specific prioritization of genes that promote Cardiovascular Disease. To determine functional interconnectivity of genes within this pathological category, molecular cartography was used to reveal a local scale-free network with significant prioritization of genes previously uncharacterized as part of CALR-governed cardiogenesis. Advanced systems biology-derived algorithms, combined with the molecular dynamics of pluripotent ES cells, thus facilitates decryption of covert gene network features and functional enrichment remodeling actuated by discrete regions within CALR that drive phenotype manifestation.

Anticipatory diagnosis of disease prior to symptomatic presentation is complicated by gene expression diversity that underlies pathology. To address this, stem cell transcriptomes were utilized as ad hoc multidimensional indices to interrogate molecular statuses of calreticulin-disrupted cell types, which identified novel molecular dynamics that enforced cardiopathological phenotype. Identification of prioritized gene expression network hubs and altered prioritization of global molecular functions reveals intricate systems biology phenomena that comprise pathological genome remodeling, critical for advanced understanding of disease etiology.

Snapshot overviews of mRNA expression changes that underlie cellular phenotypes define transcript expression differences, gene ontology re-organization, and gene network remodeling events that can be mapped in silico to extract novel molecular information [9]. Here, a broad bioinformatics overview identified multiple riboproteins within Eukaryotic Initiation Factor 2 (EIF2) Signaling as priority-regulated in gene lesion variant transcriptomes. This supports reported prevalence of altered translational machinery in CALR-disrupted transcriptomes and phenotypes [18]. CALR deletion leads to impaired system wide quality control of newly synthesized glycoproteins, as well as global accumulation of conformationally compromised substrates and general abrogation of calcium-dependent protein folding processes [25–27]. Accumulation of unfolded or misfolded proteins is salvaged in part by the unfolded protein response (UPR) of the endoplasmic reticulum [28], which enacts translational re-calibration in the event of stress stimuli. As part of this response, EIF pathways are targeted to facilitate restoration of protein homeostasis through prevention of misfolded protein accumulation that occurs under conditions of physiological and/or environmental duress, while concomitantly stimulating cells that contain conformationally defective proteins to initiate apoptosis [29–32]. Thus, global prioritization of EIF2 signaling components, in addition to enrichment within the down regulated transcript cohort of the PC variant, may reflect a cellular strategy of pre-emptive transcriptome recruitment to compensate for catastrophic system stress.

Global gene ontology remodeling precipitated by calr deficiency may reflect net effects of discrete contributions made by specific functional CALR regions. To assess this, CALR-truncated variants that exclusively expressed either the protein chaperoning or calcium binding domain were examined for transcriptome changes. Overall, multiple physiological categories were impacted, in line with previous observations reported for diverse gene ontology enrichments in the calr ^−/− condition [18]. Broad cellular processes were targeted, with the most enriched disease category for both truncation variants being Cancer. While curation bias is responsible for “Cancer” prioritization in multiple databases [24], functional distribution of CALR outside the endoplasmic reticulum, coupled with roles in cellular adhesion and migration, support the prioritized role of oncological development and progression in these CALR truncation mutants [33, 34].

Targeted focus on cardiac related processes for both CALR truncation variants revealed significant impacts on Cardiac Development and Cardiac Physiology themes, with Cardiovascular Disease uniquely enriched within down regulated genes of the PC variant, which lacks the chaperoning function of full length CALR. Jeopardized expression and/or function of genes within this CALR-truncated transcriptome in a cardiac context identified prioritized molecules that may underlie cardiac pathogenesis precipitated by CALR derangement. The present study identified Pitx2 as a prioritized gene network hub, i.e. specific promotion of Pitx2 in Cardiovascular Disease development was observed under conditions of CALR truncation/deficiency. This result suggests a critical role for Pitx2 within the CALR-regulated cardiogenic network. In the CALR-compromised sub-network, a high betweenness score indicates load and importance of Pitx2 to the sub-network, while high radiality demonstrates functional relevance and control that Pitx2 exerts on neighboring nodes [24, 35].

Specific promotion of Pitx2 in cardiac disease development under conditions of CALR deficiency suggests its critical involvement within the CALR-regulated cardiogenic network. Pitx2 has been reported to precipitate arrhythmogenesis upon dysregulated expression [36]. Previous studies have identified sinus bradycardia, complete heart block and sudden death associated with dysregulated CALR expression [37, 38]. Here, CALR-dependent remodeling of gene networks that contain Pitx2, and gene ontology prioritization of “Cardiac Arrhythmia”, provide bioinformatic evidence that suggest the arrhythmogenic phenotype generated by CALR dysregulation could be mediated by a CALR-PITX2 signaling pathway.

Further data for the critical cardiogenic role of Pitx2 have been reported in models that express hypomorphic Pitx2, which caused disorganization of the atrial septum, increased deposition of extracellular matrix in the atrial chambers, and elongation of the atrioventricular canal [39]. Importantly, Pitx2 determines right-left heart patterning [40, 41], thus complete loss of Pitx2 function causes embryonic lethality with associated catastrophic cardiovascular defects [42]. Interestingly, non-cardiac Pitx2 knockout phenotypes, such as omphalocele, are also present in calreticulin-deficient experimental animal models [42–44]. Together, these observations further support an integrated role for Pitx2 in CALR-defective phenotypes.

Molecular signatures of disease are presently resolved by static indices and molecular panels comprised of experimentally validated biomarkers that diagnose patho-susceptible cells or tissues. This approach is limited by marker selection bias that cannot account for, or provide insight into, global dynamics that sustain disease states. Here, high throughput profiling interrogates dynamic gene expression profiles that offer a depth of resolution unmatched by traditional biomarker panels. Stem cells are natural molecular registries that provide readily abstracted, information-rich transcriptomes that capture molecular underpinnings of discrete cellular phenotypes [18]. Leveraging modern bioinformatic analytics against intricate stem cell transcriptome readouts can extract systems biology features intractable to other approaches, and demonstrates a novel capacity for stem cells as advanced high resolution diagnostic tools (Fig. 6). Molecular cartography of transcript level differences between transcriptomes of pluripotent and phenotype-committed cells can thus prioritize genes critical to disease processes, identifying molecular candidates with potential roles in pathological etiology.

Pleiotropic responses to single gene lesions recruit a diversity of transcriptome elements that dictate developmental progress, and such dynamic expression changes creates a complex network that make identification of inconspicuous gene relationships challenging. Here, application of a systems biology-derived algorithm of bioinformatic dissection identified discrete gene ontology remodeling effects ascribed to either chaperone or calcium binding regions of CALR. Furthermore, expression changes of a Pitx2-inclusive cardiogenic sub-transcriptome were revealed downstream of calr deletion, providing novel bioinformatic support for cardiogenic disruption subsequent to calr derangement. This paradigm of systems deconvolution thus demonstrates that bioinformatic navigation of an intricate gene expression landscape identifies specific sub-transcriptome remodeling which enhances molecular resolution to expose novel elements of global pathways underlying development and disease processes.