Reconstructing networks of pathways via significance analysis of their intersections

GraphPath: a graph attention model for molecular stratification with interpretability based on the pathway-pathway interaction network

MOTIVATION

Studying the molecular heterogeneity of cancer is essential for achieving personalized therapy. At the same time, understanding the biological processes that drive cancer development can lead to the identification of valuable therapeutic targets. Therefore, achieving accurate and interpretable clinical predictions requires paramount attention to thoroughly characterizing patients at both the molecular and biological pathway levels.

RESULTS

Here, we present GraphPath, a biological knowledge-driven graph neural network with multi-head self-attention mechanism that implements the pathway-pathway interaction network. We train GraphPath to classify the cancer status of patients with prostate cancer based on their multi-omics profiling. Experiment results show that our method outperforms P-NET and other baseline methods. Besides, two external cohorts are used to validate that the model can be generalized to unseen samples with adequate predictive performance. We reduce the dimensionality of latent pathway embeddings and visualize corresponding classes to further demonstrate the optimal performance of the model. Additionally, since GraphPath's predictions are interpretable, we identify target cancer-associated pathways that significantly contribute to the model's predictions. Such a robust and interpretable model has the potential to greatly enhance our understanding of cancer's biological mechanisms and accelerate the development of targeted therapies.

AVAILABILITY AND IMPLEMENTATION

https://github.com/amazingma/GraphPath.

Framework for classification of cancer gene expression data using Bayesian hyper-parameter optimization

Computational classification of cancers is an important research problem. Gene expression data has 1000s of features, very few samples, and a class imbalance problem. In this paper, we have proposed a framework for the classification of cancer gene expression profiles. The framework consists of a pipeline of methods for data pre-processing, feature selection, and classification. Data pre-processing is done by standard scaling and normalization of the features. The feature selection is performed in two steps. First, recursive feature elimination (RFE) is used; then, a genetic algorithm is applied only in case RFE results in a feature subset of size more than a specific threshold. Next, is a meta-pool of diverse, individual as well as ensemble classifiers. Hyper-parameters of each member in the meta-pool are optimized using Bayesian Optimization. An algorithm is developed to select the best classifier from the meta-pool based on classification accuracy and computation time taken. We evaluated the framework on 6 publicly available microarray datasets and the PAN-Cancer RNA Sequencing dataset. We found that the classifier selected by the proposed framework produced significant improvement in classification accuracy and computation time required to predict labels for test datasets. A detailed comparison with the state-of-the-art methods shows that the proposed framework outperforms all of them.

Rewired Pathways and Disrupted Pathway Crosstalk in Schizophrenia Transcriptomes by Multiple Differential Coexpression Methods

Transcriptomic studies of mental disorders using the human brain tissues have been limited, and gene expression signatures in schizophrenia (SCZ) remain elusive. In this study, we applied three differential co-expression methods to analyze five transcriptomic datasets (three RNA-Seq and two microarray datasets) derived from SCZ and matched normal postmortem brain samples. We aimed to uncover biological pathways where internal correlation structure was rewired or inter-coordination was disrupted in SCZ. In total, we identified 60 rewired pathways, many of which were related to neurotransmitter, synapse, immune, and cell adhesion. We found the hub genes, which were on the center of rewired pathways, were highly mutually consistent among the five datasets. The combinatory list of 92 hub genes was generally multi-functional, suggesting their complex and dynamic roles in SCZ pathophysiology. In our constructed pathway crosstalk network, we found "Clostridium neurotoxicity" and "signaling events mediated by focal adhesion kinase" had the highest interactions. We further identified disconnected gene links underlying the disrupted pathway crosstalk. Among them, four gene pairs (PAK1:SYT1, PAK1:RFC5, DCTN1:STX1A, and GRIA1:MAP2K4) were normally correlated in universal contexts. In summary, we systematically identified rewired pathways, disrupted pathway crosstalk circuits, and critical genes and gene links in schizophrenia transcriptomes.

Machine learning-based integrative analysis of methylome and transcriptome identifies novel prognostic DNA methylation signature in uveal melanoma

Uveal melanoma (UVM) is the most common primary intraocular human malignancy with a high mortality rate. Aberrant DNA methylation has rapidly emerged as a diagnostic and prognostic signature in many cancers. However, such DNA methylation signature available in UVM remains limited. In this study, we performed a genome-wide integrative analysis of methylome and transcriptome and identified 40 methylation-driven prognostic genes (MDPGs) associated with the tumorigenesis and progression of UVM. Then, we proposed a machine-learning-based discovery and validation strategy to identify a DNA methylation-driven signature (10MeSig) composing of 10 MDPGs (AZGP1, BAI1, CCDC74A, FUT3, PLCD1, S100A4, SCN8A, SEMA3B, SLC25A38 and SLC44A3), which stratified 80 patients of the discovery cohort into two risk subtypes with significantly different overall survival (HR = 29, 95% CI: 6.7-126, P < 0.001). The 10MeSig was validated subsequently in an independent cohort with 57 patients and yielded a similar prognostic value (HR = 2.1, 95% CI: 1.2-3.7, P = 0.006). Multivariable Cox regression analysis showed that the 10MeSig is an independent predictive factor for the survival of patients with UVM. With a prospective validation study, this 10MeSig will improve clinical decisions and provide new insights into the pathogenesis of UVM.

Identification of key pathways and genes in lung carcinogenesis

The present study aimed to identify key pathways and genes in the pathogenesis of lung cancer. The GSE10072 dataset was downloaded from the Gene Expression Omnibus database. Protein-protein interaction data were collected from Human Protein Reference Database, and 201 pathways were downloaded from the Kyoto Encyclopedia of Genes and Genomes database. Signaling network impact analysis was performed to identify enriched pathways, followed by the construction of a pathway-pathway crosstalk network. Benzopyrene was used to treat normal human lung cells at concentrations of 0.01, 0.1, 1 and 10 µM, and cell viability was measured. Furthermore, growth arrest and DNA damage inducible β (GADD45B), p53, cyclin B, Akt and nuclear factor (NF)-κB protein levels were also measured via western blotting. Impact analysis identified 11 enriched lung cancer-associated KEGG pathways, including 'complement and coagulation cascades', 'ECM-receptor interaction', 'P53 signaling pathway', 'cell adhesion molecules' and 'focal adhesion'. In addition, cell cycle, 'drug metabolism-cytochrome P450', 'metabolic pathways', 'pathways in cancer', 'focal adhesion' and 'antigen processing and presentation' were central in the pathway-pathway cross-talk network. Furthermore, the upregulated gene GADD45B was associated with three of the pathways, including an activated pathway ('MAPK signaling pathway') and two repressed pathways ('cell cycle' and 'P53 pathway'). Western blotting demonstrated that the expression of NF-κB, Akt and GADD45B increased over time in lung cells treated with benzopyrene, whereas the expression levels of cyclin B and P53 decreased. In conclusion, GADD45B may contribute to lung carcinogenesis via affecting the MAPK, P53 signaling and cell cycle pathways.

Identification of potential tissue-specific cancer biomarkers and development of cancer versus normal genomic classifiers

Machine learning techniques for cancer prediction and biomarker discovery can hasten cancer detection and significantly improve prognosis. Recent “OMICS” studies which include a variety of cancer and normal tissue samples along with machine learning approaches have the potential to further accelerate such discovery. To demonstrate this potential, 2,175 gene expression samples from nine tissue types were obtained to identify gene sets whose expression is characteristic of each cancer class. Using random forests classification and ten-fold cross-validation, we developed nine single-tissue classifiers, two multi-tissue cancer-versus-normal classifiers, and one multi-tissue normal classifier. Given a sample of a specified tissue type, the single-tissue models classified samples as cancer or normal with a testing accuracy between 85.29% and 100%. Given a sample of non-specific tissue type, the multi-tissue bi-class model classified the sample as cancer versus normal with a testing accuracy of 97.89%. Given a sample of non-specific tissue type, the multi-tissue multi-class model classified the sample as cancer versus normal and as a specific tissue type with a testing accuracy of 97.43%. Given a normal sample of any of the nine tissue types, the multi-tissue normal model classified the sample as a particular tissue type with a testing accuracy of 97.35%. The machine learning classifiers developed in this study identify potential cancer biomarkers with sensitivity and specificity that exceed those of existing biomarkers and pointed to pathways that are critical to tissue-specific tumor development. This study demonstrates the feasibility of predicting the tissue origin of carcinoma in the context of multiple cancer classes.

PINTnet: construction of condition-specific pathway interaction network by computing shortest paths on weighted PPI

Background

Identifying perturbed pathways in a given condition is crucial in understanding biological phenomena. In addition to identifying perturbed pathways individually, pathway analysis should consider interactions among pathways. Currently available pathway interaction prediction methods are based on the existence of overlapping genes between pathways, protein-protein interaction (PPI) or functional similarities. However, these approaches just consider the pathways as a set of genes, thus they do not take account of topological features. In addition, most of the existing approaches do not handle the explicit gene expression quantity information that is routinely measured by RNA-sequecing.

Results

To overcome these technical issues, we developed a new pathway interaction network construction method using PPI, closeness centrality and shortest paths. We tested our approach on three different high-throughput RNA-seq data sets: pregnant mice data to reveal the role of serotonin on beta cell mass, bone-metastatic breast cancer data and autoimmune thyroiditis data to study the role of IFN- α. Our approach successfully identified the pathways reported in the original papers. For the pathways that are not directly mentioned in the original papers, we were able to find evidences of pathway interactions by the literature search. Our method outperformed two existing approaches, overlapping gene-based approach (OGB) and protein-protein interaction-based approach (PB), in experiments with the three data sets.

Conclusion

Our results show that PINTnet successfully identified condition-specific perturbed pathways and the interactions between the pathways. We believe that our method will be very useful in characterizing biological mechanisms at the pathway level. PINTnet is available at http://biohealth.snu.ac.kr/software/PINTnet/.

Inflammaging and human longevity in the omics era

Inflammaging is a recent theory of aging originally proposed in 2000 where data and conceptualizations regarding the aging of the immune system (immunosenescence) and the evolution of immune responses from invertebrates to mammals converged. This theory has received an increasing number of citations and experimental confirmations. Here we present an updated version of inflammaging focused on omics data - particularly on glycomics - collected on centenarians, semi-supercentenarians and their offspring. Accordingly, we arrived to the following conclusions: i) inflammaging has a structure where specific combinations of pro- and anti-inflammatory mediators are involved; ii) inflammaging is systemic and more complex than we previously thought, as many organs, tissues and cell types participate in producing pro- and anti-inflammatory stimuli defined "molecular garbage"; iii) inflammaging is dynamic, can be propagated locally to neighboring cells and systemically from organ to organ by circulating products and microvesicles, and amplified by chronic age-related diseases constituting a "local fire", which in turn produces additional inflammatory stimuli and molecular garbage; iv) an integrated Systems Medicine approach is urgently needed to let emerge a robust and highly informative set/combination of omics markers able to better grasp the complex molecular core of inflammaging in elderly and centenarians.

A method for developing regulatory gene set networks to characterize complex biological systems

Background

Traditional approaches to studying molecular networks are based on linking genes or proteins. Higher-level networks linking gene sets or pathways have been proposed recently. Several types of gene set networks have been used to study complex molecular networks such as co-membership gene set networks (M-GSNs) and co-enrichment gene set networks (E-GSNs). Gene set networks are useful for studying biological mechanism of diseases and drug perturbations.

Results

In this study, we proposed a new approach for constructing directed, regulatory gene set networks (R-GSNs) to reveal novel relationships among gene sets or pathways. We collected several gene set collections and high-quality gene regulation data in order to construct R-GSNs in a comparative study with co-membership gene set networks (M-GSNs). We described a method for constructing both global and disease-specific R-GSNs and determining their significance. To demonstrate the potential applications to disease biology studies, we constructed and analysed an R-GSN specifically built for Alzheimer's disease.

Conclusions

R-GSNs can provide new biological insights complementary to those derived at the protein regulatory network level or M-GSNs. When integrated properly to functional genomics data, R-GSNs can help enable future research on systems biology and translational bioinformatics.

Identification of a DNA methylation signature in blood cells from persons with Down Syndrome

Down Syndrome (DS) is characterized by a wide spectrum of clinical signs, which include segmental premature aging of central nervous and immune systems. Although it is well established that the causative defect of DS is the trisomy of chromosome 21, the molecular bases of its phenotype are still largely unknown. We used the Infinium HumanMethylation450 BeadChip to investigate DNA methylation patterns in whole blood from 29 DS persons, using their relatives (mothers and unaffected siblings) as controls. This family-based model allowed us to monitor possible confounding effects on DNA methylation patterns deriving from genetic and environmental factors. Although differentially methylated regions (DMRs) displayed a genome-wide distribution, they were enriched on chromosome 21. DMRs mapped in genes involved in developmental functions, including embryonic development (HOXA family) and haematological (RUNX1 and EBF4) and neuronal (NCAM1) development. Moreover, genes involved in the regulation of chromatin structure (PRMD8, KDM2B, TET1) showed altered methylation. The data also showed that several pathways are affected in DS, including PI3K-Akt signaling. In conclusion, we identified an epigenetic signature of DS that sustains a link between developmental defects and disease phenotype, including segmental premature aging.

Systems biology and brain activity in neuronal pathways by smart device and advanced signal processing

Contemporary biomedicine is producing large amount of data, especially within the fields of “omic” sciences. Nevertheless, other fields, such as neuroscience, are producing similar amount of data by using non-invasive techniques such as imaging, functional magnetic resonance and electroencephalography. Nowadays a big challenge and a new research horizon for Systems Biology is to develop methods to integrate and model this data in an unifying framework capable to disentangle this amazing complexity. In this paper we show how methods from genomic data analysis can be applied to brain data. In particular the concept of pathways, networks and multiplex are discussed. These methods can lead to a clear distinction of various regimes of brain activity. Moreover, this method could be the basis for a Systems Biology analysis of brain data and for the integration of these data in a multivariate and multidimensional framework. The feasibility of this integration is strongly dependent from the feature extraction method used. In our case we used an “alphabet” derived from a multi-resolution analysis that is capable to capture the most relevant information from these complex signals.

Discovering pathway cross-talks based on functional relations between pathways

Background

In biological systems, pathways coordinate or interact with one another to achieve a complex biological process. Studying how they influence each other is essential for understanding the intricacies of a biological system. However, current methods rely on statistical tests to determine pathway relations, and may lose numerous biologically significant relations.

Results

This study proposes a method that identifies the pathway relations by measuring the functional relations between pathways based on the Gene Ontology (GO) annotations. This approach identified 4,661 pathway relations among 166 pathways from Pathway Interaction Database (PID). Using 143 pathway interactions from PID as testing data, the function-based approach (FBA) is able to identify 93% of pathway interactions, better than the existing methods based on the shared components and protein-protein interactions. Many well-known pathway cross-talks are only identified by FBA. In addition, the false positive rate of FBA is significantly lower than others via pathway co-expression analysis.

Conclusions

This function-based approach appears to be more sensitive and able to infer more biologically significant and explainable pathway relations.

Ranking differential hubs in gene co-expression networks

Identifying the genes that change their expressions between two conditions (such as normal versus cancer) is a crucial task that can help in understanding the causes of diseases. Differential networking has emerged as a powerful approach to detect the changes in network structures and to identify the differentially connected genes among two networks. However, existing differential network-based methods primarily depend on pairwise comparisons of the genes based on their connectivity. Therefore, these methods cannot capture the essential topological changes in the network structures. In this paper, we propose a novel algorithm, DiffRank, which ranks the genes based on their contribution to the differences between the two networks. To achieve this goal, we define two novel structural scoring measures: a local structure measure (differential connectivity) and a global structure measure (differential betweenness centrality). These measures are optimized by propagating the scores through the network structure and then ranking the genes based on these propagated scores. We demonstrate the effectiveness of DiffRank on synthetic and real datasets. For the synthetic datasets, we developed a simulator for generating synthetic differential scale-free networks, and we compared our method with existing methods. The comparisons show that our algorithm outperforms these existing methods. For the real datasets, we apply the proposed algorithm on several gene expression datasets and demonstrate that the proposed method provides biologically interesting results.

Identifying dysfunctional crosstalk of pathways in various regions of Alzheimer's disease brains

Background

Alzheimer's disease (AD) is a major neurodegenerative disorder leading to amnesia, cognitive impairment and dementia in the elderly. Usually this type of lesions results from dysfunctional protein cooperations in the biological pathways. In addition, AD progression is known to occur in different brain regions with particular features. Thus identification and analysis of crosstalk among dysregulated pathways as well as identification of their clusters in various diseased brain regions are expected to provide deep insights into the pathogenetic mechanism.

Results

Here we propose a network-based systems biology approach to detect the crosstalks among AD related pathways, as well as their dysfunctions in the six brain regions of AD patients. Through constructing a network of pathways, the relationships among AD pathway and its neighbor pathways are systematically investigated and visually presented by their intersections. We found that the significance degree of pathways related to the fatal disorders and the pathway overlapping strength can indicate the impacts of these neighbored pathways to AD development. Furthermore, the crosstalks among pathways reveal some evidence that the neighbor pathways of AD pathway closely cooperate and play important tasks in the AD progression.

Conclusions

Our study identifies the common and distinct features of the dysfunctional crosstalk of pathways in various AD brain regions. The global pathway crosstalk network and the clusters of relevant pathways of AD provide evidence of cooperativity among pathways for potential pathogenesis of the neuron complex disease.

Gene pathways and subnetworks distinguish between major glioma subtypes and elucidate potential underlying biology

Molecular diagnostic tools are increasingly being used in an attempt to classify primary human brain tumors more accurately. While methods that are based on the analysis of individual gene expression prove to be useful for diagnostic purposes, they are devoid of biological significance since tumorgenesis is a concerted deregulation of multiple pathways rather than single genes. In a proof of concept, we utilize two large clinical data sets and show that the elucidation of enriched pathways and small differentially expressed sub-networks of protein interactions allow a reliable classification of glioblastomas and oligodendrogliomas. Applying a feature selection method, we observe that an optimized subset of pathways and subnetworks significantly improves the prediction accuracy. By determining the enrichment of altered genes in pathways and subnetworks we show that optimized subsets of genes rarely seem to be a target of genomic alteration. Our results suggest that groups of genes play a decisive role for the phenotype of the underlying tumor samples that can be utilized to reliably distinguish tumor types. In the absence of enrichment of genes that are genomically altered we assume that genetic changes largely exert an indirect rather than direct regulatory influence on a number of tumor-defining regulatory networks.

Viral organization of human proteins

Although maps of intracellular interactions are increasingly well characterized, little is known about large-scale maps of host-pathogen protein interactions. The investigation of host-pathogen interactions can reveal features of pathogenesis and provide a foundation for the development of drugs and disease prevention strategies. A compilation of experimentally verified interactions between HIV-1 and human proteins and a set of HIV-dependency factors (HDF) allowed insights into the topology and intricate interplay between viral and host proteins on a large scale. We found that targeted and HDF proteins appear predominantly in rich-clubs, groups of human proteins that are strongly intertwined among each other. These assemblies of proteins may serve as an infection gateway, allowing the virus to take control of the human host by reaching protein pathways and diversified cellular functions in a pronounced and focused way. Particular transcription factors and protein kinases facilitate indirect interactions between HDFs and viral proteins. Discerning the entanglement of directly targeted and indirectly interacting proteins may uncover molecular and functional sites that can provide novel perspectives on the progression of HIV infection and highlight new avenues to fight this virus.

Functional genomics and networks: new approaches in the extraction of complex gene modules

The engine that makes the cell work is made of an intricate network of molecular interactions. Nowadays, the elements and relationships of this complex network can be studied with several types of high-throughput techniques. The dream of having a global picture of the cell from different perspectives that can jointly explain cell behavior is, at least technically, feasible. However, this task can only be accomplished by filling the gap between data and information. The availability of methods capable of accurately managing, integrating and analyzing the results from these experiments is crucial for this purpose. Here, we review the new challenges raised by the availability of different genomic data, as well as the new proposals presented to cope with the increasing data complexity. Special emphasis is given to approaches that explore the transcriptome trying to describe the modules of genes that account for the traits studied.

Regulation of gene expression in hepatic cells by the mammalian Target of Rapamycin (mTOR)

Background

We investigated mTOR regulation of gene expression by studying rapamycin effect in two hepatic cell lines, the non-tumorigenic WB-F344 cells and the tumorigenic WB311 cells. The latter are resistant to the growth inhibitory effects of rapamycin, thus providing us with an opportunity to study the gene expression effects of rapamycin without confounding effects on cell proliferation.

Methodology/Principal Findings

The hepatic cells were exposed to rapamycin for 24 hr. Microarray analysis on total RNA preparations identified genes that were affected by rapamycin in both cell lines and, therefore, modulated independent of growth arrest. Further studies showed that the promoter regions of these genes included E-box-containing transcription factor binding sites at higher than expected rates. Based on this, we tested the hypothesis that c-Myc is involved in regulation of gene expression by mTOR by comparing genes altered by rapamycin in the hepatic cells and by c-Myc induction in fibroblasts engineered to express c-myc in an inducible manner. Results showed enrichment for c-Myc targets among rapamycin sensitive genes in both hepatic cell lines. However, microarray analyses on wild type and c-myc null fibroblasts showed similar rapamycin effect, with the set of rapamycin-sensitive genes being enriched for c-Myc targets in both cases.

Conclusions/Significance

There is considerable overlap in the regulation of gene expression by mTOR and c-Myc. However, regulation of gene expression through mTOR is c-Myc-independent and cannot be attributed to the involvement of specific transcription factors regulated by the rapamycin-sensitive mTOR Complex 1.

Rapamycin response in tumorigenic and non-tumorigenic hepatic cell lines

BACKGROUND

The mTOR inhibitor rapamycin has anti-tumor activity across a variety of human cancers, including hepatocellular carcinoma. However, resistance to its growth inhibitory effects is common. We hypothesized that hepatic cell lines with varying rapamycin responsiveness would show common characteristics accounting for resistance to the drug.

METHODOLOGY/PRINCIPAL FINDINGS

We profiled a total of 13 cell lines for rapamycin-induced growth inhibition. The non-tumorigenic rat liver epithelial cell line WB-F344 was highly sensitive while the tumorigenic WB311 cell line, originally derived from the WB-F344 line, was highly resistant. The other 11 cell lines showed a wide range of sensitivities. Rapamycin induced inhibition of cyclin E-dependent kinase activity in some cell lines, but the ability to do so did not correlate with sensitivity. Inhibition of cyclin E-dependent kinase activity was related to incorporation of p27(Kip1) into cyclin E-containing complexes in some but not all cell lines. Similarly, sensitivity of global protein synthesis to rapamycin did not correlate with its anti-proliferative effect. However, rapamycin potently inhibited phosphorylation of two key substrates, ribosomal protein S6 and 4E-BP1, in all cases, indicating that the locus of rapamycin resistance was downstream from inhibition of mTOR Complex 1. Microarray analysis did not disclose a unifying mechanism for rapamycin resistance, although the glycolytic pathway was downregulated in all four cell lines studied.

CONCLUSIONS/SIGNIFICANCE

We conclude that the mechanisms of rapamycin resistance in hepatic cells involve alterations of signaling downstream from mTOR and that the mechanisms are highly heterogeneous, thus predicting that maintaining or promoting sensitivity will be highly challenging.

Overcoming Resistance to Conventional Drugs in Ewing Sarcoma and Identification of Molecular Predictors of Outcome

Purpose

The improvement of Ewing sarcoma (EWS) therapy is currently linked to the discovery of strategies to select patients with poor and good prognosis and of modified treatment regimens. In this study, we analyzed the molecular factors governing EWS response to chemotherapy to identify genetic signatures to be used for risk-adapted therapy.

Patients and Methods

Microarray technology was used for profiling 30 primary tumors and seven metastases of patients who were classified according to event-free survival. For selected genes, real-time polymerase chain reaction was applied in 42 EWS primary tumors as validation assay. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide test was used to evaluate in vitro drug sensitivity.

Results

We identified molecular signatures that reflect tumor resistance to chemotherapy. Annotation analysis was applied to reveal the biologic functions that critically influenced clinical outcome. The prognostic relevance of glutathione metabolism pathway was validated. The expression of MGST1, the microsomal glutathione S-transferase (GST), was found to clearly predict EWS prognosis. MGST1 expression was associated with doxorubicin chemosensitivity. This prompted us to assess the in vitro effectiveness of 6-(7-nitro-2,1,3-benzoxadiazol-4-ylthio)hexanol (NBDHEX), a new anticancer agent that efficiently inhibits GST enzymes. Six cell lines were found to be sensitive to this new drug.

Conclusion

Classification of EWS patients into high- and low-risk groups is feasible with restricted molecular signatures that may have practical value at diagnosis for selecting patients with EWS who are unresponsive to current treatments. Glutathione metabolism pathway emerged as one of the most significantly altered prognosis-associated pathway. NBDHEX is proposed as a new potential therapeutic possibility.

Displayed correlation between gene expression profiles and submicroscopic alterations in response to cetuximab, gefitinib and EGF in human colon cancer cell lines

Background

EGFR is frequently overexpressed in colon cancer. We characterized HT-29 and Caco-2, human colon cancer cell lines, untreated and treated with cetuximab or gefitinib alone and in combination with EGF.

Methods

Cell growth was determined using a variation on the MTT assay. Cell-cycle analysis was conducted by flow cytometry. Immunohistochemistry was performed to evaluate EGFR expression and scanning electron microscopy (SEM) evidenced the ultrastructural morphology. Gene expression profiling was performed using hybridization of the microarray Ocimum Pan Human 40 K array A.

Results

Caco-2 and HT-29 were respectively 66.25 and 59.24 % in G0/G1. They maintained this level of cell cycle distribution after treatment, suggesting a predominantly differentiated state. Treatment of Caco-2 with EGF or the two EGFR inhibitors produced a significant reduction in their viability. SEM clearly showed morphological cellular transformations in the direction of cellular death in both cell lines treated with EGFR inhibitors. HT-29 and Caco-2 displayed an important reduction of the microvilli (which also lose their erect position in Caco-2), possibly invalidating microvilli absorption function. HT-29 treated with cetuximab lost their boundary contacts and showed filipodi; when treated with gefitinib, they showed some vesicles: generally membrane reshaping is evident. Both cell lines showed a similar behavior in terms of on/off switched genes upon treatment with cetuximab. The gefitinib global gene expression pattern was different for the 2 cell lines; gefitinib treatment induced more changes, but directly correlated with EGF treatment.

In cetuximab or gefitinib plus EGF treatments there was possible summation of the morphological effects: cells seemed more weakly affected by the transformation towards apoptosis. The genes appeared to be less stimulated than for single drug cases.

Conclusion

This is the first study to have systematically investigated the effect of cetuximab or gefitinib, alone and in combination with EGF, on human colon cancer cell lines. The EGFR inhibitors have a weaker effect in the presence of EGF that binds EGFR. Cetuximab treatment showed an expression pattern that inversely correlates with EGF treatment. We found interesting cyto-morphological features closely relating to gene expression profile. Both drugs have an effect on differentiation towards cellular death.