Pathway analysis identifies perturbation of genetic networks induced by butyrate in a bovine kidney epithelial cell line
- First Online:
- Cite this article as:
- Li, C., Li, R.W., Wang, Y. et al. Funct Integr Genomics (2007) 7: 193. doi:10.1007/s10142-006-0043-2
- 301 Views
Ruminant species have evolved to metabolize the short-chain volatile fatty acids (VFA), acetate, propionate, and butyrate, to fulfill up to 70% of their nutrient energy requirements. The inherent VFA dependence of ruminant cells was exploited to add a level of increased sensitivity to the study of the role of butyrate gene-response elements in regulatory biochemical pathways. Global gene expression profiles of the bovine kidney epithelial cells regulated by sodium butyrate were investigated with high-density oligonucleotide microarrays. The detailed mechanisms by which butyrate induces cell growth arrest and apoptosis were analyzed using the Ingenuity Pathways Knowledge Base. The functional category and pathway analyses of the microarray data revealed that four canonical pathways (Cell cycles: G2/M DNA damage checkpoint, and pyrimidine metabolism; G1/S checkpoint regulation and purine metabolism) were significantly perturbed. The biologically relevant networks and pathways of these genes were also identified. IGF2, TGFB1, TP53, E2F4, and CDC2 were established as being centered in these genomic networks. The present findings provide a basis for understanding the full range of the biological roles and the molecular mechanisms that butyrate may play in animal cell growth, proliferation, and energy metabolisms.
KeywordsApoptosisButyrateCell cycleGenetic network
Short-chain volatile fatty acids (VFA; acetate, propionate, and butyrate), nutrients especially critical to ruminant mammals, are formed during the microbial fermentation of the dietary fiber in the gastrointestinal tract of mammalian species and are directly absorbed at the site of production (Bugaut 1987). Ruminant species have evolved to metabolize the short-chain VFA, acetate, propionate, and butyrate to fulfill up to 70% of their nutrient energy requirements. Because ruminants are so dependent on low-molecular-weight VFA like acetate, propionate, and butyrate for metabolic needs (Bergman 1990), cells derived from ruminant species are presumed to be equipped to respond to VFA with increased sensitivity compared to other species (Bergman 1990). Whereas acetate and propionate hold a prominent position in providing energy to the ruminant metabolism, butyrate, low in its relative concentrations, appears to be involved in metabolism beyond its role as a nutrient. Butyrate was the most potent among short-chain fatty acids in the induction of apoptosis and inhibition of cell proliferation (Emenaker et al. 2001; Hague and Paraskeva 1995). Roles for butyrate have been established in cell differentiation, proliferation, motility, and in particular, the induction of cell cycle arrest and apoptosis (Chen et al. 2003; Gassull and Cabre 2001; Scheppach et al. 1995). Apoptosis is a genetically regulated cellular suicide mechanism that plays a crucial role in development and in the maintenances of homeostasis of animals (Johnson 2002). The mechanism(s) by which butyrate induces cellular differentiation and suppresses growth has not been elucidated. Similarly, the mechanism through which butyrate and other short-chain fatty acids induce the cell cycle regulatory and apoptotic effects and the mechanism by which the decision between cell death and survival is arbitrated are poorly understood. Few definitive studies in cattle, if any, have addressed the capabilities for nutrients to modulate gene expression and proteomic outcomes as a means of arresting metabolic stress. Cell cycle regulatory and apoptotic effects of butyrate at the cellular and molecular levels in normal bovine cells have not been studied thus far.
In a previous study (Li and Elsasser 2005), potential biological roles of butyrate were investigated using the established Madin–Darby bovine kidney (MDBK) epithelial cell line. The study focused on determining whether normal bovine cells in a standard cell culture condition were sensitive to the growth inhibitory effects of butyrate. Butyrate not only induced apoptosis but also induced cell cycle arrest at the G1/S boundary and M/G2 in MDBK cells. The cell responses were concentration-dependent. More recently, in looking into the possible mechanisms for the apoptosis and cell cycle arrest induced by butyrate, global gene expression profiles of the bovine kidney epithelial cells regulated by sodium butyrate were investigated with high-density oligonucleotide microarrays (Li and Li 2006). The gene expression profiling generated by the microarray technique revealed that butyrate induces many significant changes in the gene expression in bovine kidney epithelial cells. More recently, pathway analysis technologies allow for the mapping of gene expression data into relevant pathways based on their functional annotation and known molecular interactions. To examine the molecular functions and genetic networks, the data that we generated from the microarray was explored using ingenuity pathways analysis (Ingenuity® Systems, http://www.ingenuity.com), a web-delivered application that enables the discovery, visualization, and exploration of molecular interaction networks in gene expression data. Ingenuity pathways analysis (Ingenuity® Systems) is a software application that enables biologists and bioinformaticians to identify the biological mechanisms, pathways, and functions most relevant to their experimental datasets or genes of interest (Abdel-Aziz et al. 2006; Calvano et al. 2005; Mayburd et al. 2006; Pospisil et al. 2006; Su et al. 2006). In this present research, we report our findings regarding the functional category and pathway analysis of differential expressed genes in MDBK cells (the MDBK epithelial cell line) treated with butyrate.
Cell culture and butyrate treatments
The MDBK epithelial cells (American Type Culture Collection, Manassas, VA, catalog no. CCL-22) were cultured in Eagle’s minimal essential medium and supplemented with 5% fetal bovine serum (Invitrogen, Carlsbad, CA) in 25 cm2 flasks as described in our previous report (Li and Elsasser 2005). At approximately 50% confluence (during the exponential phase), the cells were treated with 10 mM of sodium butyrate for 24 h (Calbiochem, San Diego, CA). A concentration of 10 mM of butyrate was selected in our experiments for reasons explained in our previous report (Li and Elsasser 2005). Three replicate flasks of cells for both treatment and control groups (a total of six samples) were used for the microarray experiments.
Oligonucleotide microarray, hybridization, image acquisition, and data analysis
The bovine microarray platform used was described previously (Li et al. 2006). A total of 86,191 unique 60mer oligonucleotides were designed and synthesized in situ using photo deprotection chemistry (Singh-Gasson et al. 1999). Each unique oligonucleotide was repeated four times on the array (a total of approximately 340,000 features). These oligonucleotides represented 45,383 unique bovine sequences/genes, including 40,808 tentative consensus sequences from The Institute for Genomic Research (TIGR) Bos taurus gene index (http://www.tigr.org) and 4,575 singletons.
Hybridization, image acquisition, and data analysis were described previously (Li et al. 2006). The microarrays scanned using an Axon GenePix 4000B scanner (Molecular Devices, Union City, CA) at 5-μM resolution. The data was extracted from the raw images using NimbleScan software (NimbleGen, Madison, WI). The control and butyrate treatment each had three replicates, and a total of six microarrays were used in the experiment (Gene Expression Omnibus, GEO, accession GSE3970).
Support trees for hierarchical clustering
Hierarchical clustering (Eisen et al. 1998) has been a common tool used for microarray data visualization. In this study, we used the support trees for hierarchical clustering method from the TIGR MeV v3.0 (http://www.tigr.org). This method adds the statistical support for the nodes of the trees based on resampling the normalized input data. The microarray raw data processing was described (Li and Li 2006) briefly. Relative signal intensities (log2) for each feature were generated using the robust multi-array average algorithm (Irizarry et al. 2003a,b). The data was processed based on the quantile normalization method (Bolstad et al. 2003) using the R package (http://www.bioconductor.org). This normalization method aims to make the distribution of intensities for each array in a set of arrays the same. The method assumes that a quantile–quantile plot of two data vectors with the same distribution will have a straight diagonal line. The method performed better in dealing with bias and reducing variability across arrays compared to other methods (Bolstad et al. 2003). The background-adjusted, normalized, and log-transformed intensity values were then analyzed using the significance analysis of microarrays (SAM) method (Tusher et al. 2001) with a two-class unpaired design (SAM version 2.20 from http://www-stat.stanford.edu/∼tibs/clickwrap/sam/academic/). SAM is the most popular method for microarray analysis with 635 citations of the original publication as of October 2004 (Larsson et al. 2005). SAM ranks genes based on a modified t-test statistic. The unique features of SAM include implementing permutation testing and the ability to estimate a global false discovery rate (FDR, an expected percentage of false positives among the claimed positives) and a gene error chance (q value). A sequence was declared to be significant when it met a stringent median FDR cut-off at 0% (Li and Li 2006). A basic local alignment search tool search was conducted for all sequences that met the threshold to remove possible redundancies. When a gene was represented by multiple sequences, the fold change with a q value of only one sequence was selected to represent this gene. The selected genes that meet this stringent significance threshold, especially those genes homologous to their respective human gene counterparts approved by the Human Organisation (HUGO) Gene Nomenclature Committee (http://www.gene.ucl.ac.uk) and/or with known functions and pathways, were subject to the clustering analysis. The parameters used to generate signal transduction clusters are as follows: For both gene trees and experiment trees, the bootstrap resampling with 1,000 iterations was used. Linkage methods to calculate cluster-to-cluster distance was complete linkage; that is, the distances were measured between each member of one cluster to each member of other clusters, and the maximum of these distances is considered the cluster-to-cluster distance. The distance metrics used is Euclidean distance.
Canonical pathway analysis of datasets
The analysis of canonical pathways identified the pathways from the Ingenuity Pathways Analysis library of canonical pathways that were most significant to the dataset. Genes from the dataset that were associated with a canonical pathway in the Ingenuity Pathways Knowledge Base were considered for the analysis. The significance of the association between the dataset and the canonical pathway was measured in two ways: (1) A ratio of the number of genes from the dataset that map to the pathway divided by the total number of genes that map to the canonical pathway is displayed. (2) Fischer’s exact test was used to calculate a p value determining the probability that the association between the genes in the dataset and the canonical pathway is explained by chance alone.
Functional analysis of datasets
The functional analysis identified the biological functions and/or diseases that were most significant to the dataset. Genes from the datasets that were associated with biological functions and/or diseases in the Ingenuity Pathways Knowledge Base were considered for the analysis. Fischer’s exact test was used to calculate a p value determining the probability that each biological function and/or disease assigned to that dataset was due to chance alone.
Pathways analysis and network generation
A dataset containing gene identifiers and corresponding expression values was uploaded into in the application. Each gene identifier was mapped to its corresponding gene object in the Ingenuity Pathways Knowledge Base. These genes, called focus genes, were overlaid onto a global molecular network developed from information contained in the Ingenuity Pathways Knowledge Base. Networks of these focus genes were then algorithmically generated based on their connectivity.
Functional analysis of a network
The functional analysis of a network identified the biological functions and/or diseases that were most significant to the genes in the network. The network genes associated with biological functions and/or diseases in the Ingenuity Pathways Knowledge Base were considered for the analysis. Fischer’s exact test was used to calculate a p value determining the probability that each biological function and/or disease assigned to that network was due to chance alone.
Network/pathways graphical representation
A network pathway is a graphical representation of the molecular relationships between genes/gene products. Genes or gene products are represented as nodes, and the biological relationship between two nodes is represented as an edge (line). All edges are supported by at least one reference from the literature, from a textbook, or from canonical information stored in the Ingenuity Pathways Knowledge Base. The intensity of the node color indicates the degree of up- (red) or down- (green) regulation. Nodes are displayed using various shapes that represent the functional class of the gene product. Edges are displayed with various labels that describe the nature of the relationship between the nodes (e.g., P for phosphorylation, T for transcription).
Canonical pathways analysis
The significance calculated for each canonical pathway is a measurement of the likelihood that the pathway is associated with the dysregulated genes by random chance. Interestingly, among four significantly perturbed canonical pathways, three of them (Cell cycles: G2/M DNA damage checkpoint, and pyrimidine metabolism; and G1/S checkpoint regulation) showed significant down-regulation, and only the purine metabolism pathway showed significant up-regulation. In addition, cell cycle checkpoint pathways (G2/M DNA damage checkpoint and G1/S checkpoint regulation) are impaired due to the treatment of butyrate. For the first time, two pathways for critical regulation of purine and pyrimidine metabolism were linked to the butyrate induced biological effects. These findings underlie mechanisms that butyrate induced cell arrest in cell cycle progression. Global functional analysis (Fig. 2b) also indicated that the levels of many genes involved in cell cycle; cellular movement; DNA replication, recombination, and repair; and cellular growth and proliferation are significantly and consistently dysregulated. The major dysregulated functions induced by butyrate are illustrated in Fig. 2b, and the major dysregulated functions and genes involved are listed in the Electronic Supplementary Material Table S2.
The biologically relevant networks and pathways
This study, being the first global expression profiling and pathway analysis on butyrate-induced gene perturbation in MDBK cells, has generated comprehensive information on the experimental system that can be used in many functional genomic studies in bovine. This study suggests underlying mechanisms for biological effects that we previously observed such as cell cycle arrest and appoptosis in MDBK cells induced by butyrate (Li and Elsasser 2005). Overall, our data confirmed that butyrate induces profound changes in gene expression related to multiple signal pathways and genomic networks in bovine kidney epithelial cells. To the best of our knowledge, this is the first study that has been carried out using the pathway analysis software to identify the influences of butyrate on genetic networks in normal bovine cell line.
Many previously published studies focused on the biological effects of butyrate on cancer cells. As a result, there is a wealth of knowledge on butyrate as a histone deacetylase (HDAC) inhibitor, the role of aberrant histone acetylation in tumorigenesis, and the potential for cancer chemoprevention and therapy (Dashwood et al. 2006; Myzak and Dashwood 2006; Myzak et al. 2006). There is little, if any, information about the biological effects of butyrate in “normal” cell lines. An excellent and recent study using microarray and computational gene network analyses to identify the detailed mechanisms by which butyrate induces cell growth arrest and the differentiation of mouse colonic epithelial MCE301 cells was reported (Tabuchi et al. 2006). Some findings in our present study, such as IGF2 genes being up regulated and centered in major genetic networks, are consistent with what has been reported (Tabuchi et al. 2006). However, our results extended far beyond their findings. A large number of these differentially expressed genes were newly identified, and their relationship to butyrate had not been reported before. The functional category and pathway analyses of the microarray data revealed that four canonical pathways (Cell cycles: G2/M DNA damage checkpoint, and pyrimidine metabolism; G1/S Checkpoint Regulation and purine metabolism) were significantly perturbed. More biologically relevant networks and pathways of these genes were also identified. IGF2, TGFB1, TP53, E2F4, and CDC2 were established as being centered in these functional genetic networks. Those extended findings are perhaps the results of a much higher density microarray platform used in our experiment. It could also be the reflection of the difference in the response to butyrate between cell types.
Butyrate exerts a very broad range of effects on many biological pathways via its inhibitory ability on HDAC in MDBK cell line (Li and Elsasser 2005; Li et al. 2004; Tabuchi et al. 2006). The genes we identified (Supplementary Table S1) that fall within a broad range of functional categories appeared to provide the molecular basis for its pleiotropic effects. The pathway analysis on the butyrate-induced genes in MDBK cells using the Ingenuity Pathways Analysis tool successfully identified remarkable changes in genetic networks related to cell cycle, cell death, and DNA replication. The perturbation of the G2/M DNA damage checkpoint and the G1/S checkpoint pathways are likely responsible for the cell cycle arrest at the G1 and G2/M stages as the result of butyrate treatment on MDBK cells (Li and Elsasser 2005).
Moreover, for the first time, pyrimidine and purine metabolism pathways have been linked to the biological effects of butyrate. Nucleic acids are important intracellular signaling molecules and coenzymes, which are the single most important means of coupling endergonic to exergonic reactions. Nucleic acids are also the storage of genetic information in the form of DNA and RNA. RNA is composed of nucleotides containing a phosphoribosyl component and one of the aromatic bases: adenine (A), guanine (G), cytosine (C), and uracyl (U). These bases can be distinguished by their nitrogen containment and aromatic ring structures and come in two forms—pyrimidines and purines. Because both pyrimidines and purines are the building blocks for DNA and RNA, the perturbation of their metabolisms may explain our previous observations that, after being treated with butyrate, cells take much longer time to re-enter the S phase than the cells that were arrested by serum starvation (Li and Elsasser 2006).
Apoptosis is induced by both extrinsic and intrinsic pathways initiated by the activation of death receptors and stress-inducing stimuli (Adams 2003; Danial and Korsmeyer 2004). IGFBP-3 can enhance both extrinsic and intrinsic apoptotic pathways (Butt and Williams 2001). In agreement with a previous report (Chiba et al. 2004; Tabuchi et al. 2006), our study found that IGF2 and IGFBP3 are significantly up-regulated and associated with the network involving in cellular functions such as cellular movement and cell death. There is also growing evidence that IGFBP3 can have IGF-independent effects on cell growth (Williams et al. 2006). Our microarray and real-time RT-PCR results also confirmed that IGFBP6 was down regulated by butyrate (Tabuchi et al. 2006). In addition, our results found that TGFB1 was significantly up regulated. Studies have shown that TGFB1 regulates anterior pituitary cell proliferation and hormone secretion. There is also a report that indicates that butyrate and TGFB1 inhibit pituitary cell proliferation and regulate the expression of 7B2, PC1, and PC2 in a cell culture model of pituitary tumors (Kobayashi et al. 2003). However, to our best knowledge, there is no relevance between butyrate and the expression of TGFB1 reported before. TGFB1 is a multifunctional peptide that controls proliferation, differentiation, and other functions in many cell types. TGFB1 acts synergistically with TGFA in inducing transformation. It also acts as a negative autocrine growth factor. The dysregulation of TGFB1 activation and signaling may result in apoptosis (Bommireddy et al. 2003; Soulitzis et al. 2006; Vozenin-Brotons et al. 1999). We speculate that an up-regulated expression of both IGF2 and TGFB1 may have a synergistic effect in inducing cell apoptosis.
TP53 and a transcription factor, E2F4, are centered in one of the down-regulated genomic networks in this study. P53 has been extensively studied on its function and involvement in butyrate-induced biological effects (Jung et al. 2005; Shi et al. 2006; Watson 2006). Butyrate suppresses the growth of WT-p53-containing cells more efficiently. However, butyrate treatment leads to a major G2/M arrest of cells in the presence of p53, whereas cells without the wild-type p53 accumulated mainly in the G1 phase of the cell cycle. Furthermore, apoptosis induction by butyrate is greatly reduced in the absence of p53, suggesting that a p53 pathway mediates, in part, growth suppression by butyrate, and the p53 status may be an important determinant of chemosensitivity to butyrate (Joseph et al. 2005). Our data also indicate that the TP53 genes may have a different response and different roles in normal and transformed cells. Apparently, more studies are still required to understand the exact roles that TP53 plays in butyrate-induced biological effects.
In the present report, genes for MCM proteins 2, 3, 4, 5, and 6, as well as origin recognition complex largest subunit (Orc1), are significantly down regulated. This finding indicates that, in some ways, butyrate treatment directly targets these genes, and down-regulated the genes that are essential for the initiation of DNA replication. The eukaryotic ORC selects the genomic sites where pre-replication complexes are assembled and DNA replication begins. Orc1 has been identified in the previous studies as a primary control point in regulating the assembly of pre-replication complexes on mammalian chromosomes. First, ubiquitination of Orc1 was found as an important regulation mechanism of the initiation of DNA replication (Li and DePamphilis 2002). And second, the role for cyclin-dependent protein kinase 1 (Cdk1)/cyclin A in preventing the mammalian Orc1 from binding to chromatin during mitosis was identified (Li et al. 2004). More intriguing, our results from this study indicate that there is another mechanism for the regulation of the function of ORC1 at the level of RNA transcription.
The extensive repression of cyclin-dependent kinases as well as cell-cycle-related genes, such as CDC2/CDK1, CDC20, CDC 25A, CCNG1, CCNB1, CCNB2, CCNA2, CCNG1, and PCNA, may be closely associated with the cell growth arrest induced by butyrate. They are consistent with the growing body of evidence suggesting that the disruption of the coordinate between the regulation of DNA synthesis and cyclin-dependent kinase activity is an important feature of apoptosis. It is important for us to understand the mechanism(s) of how butyrate targets these genes and causes these changes in gene expressions. However, such an understanding will require great effort and is certainly beyond the scope of this report.
In conclusion, global gene expression profiling and computational pathway analyses provide a detailed knowledge of changes in gene expression induced by butyrate. This detailed knowledge will provide a basis for understanding the molecular mechanisms of the biological effects of butyrate in normal bovine kidney epithelial cells.