HT29 colorectal cancer cells (ATCC, Rockville, MD) were maintained in Dulbecco’s Modified Eagle’s Medium/F-12 Nutrient Mixture Media with L-glutamine (1:1; 37⁰C; 5% CO_2; Invitrogen) containing 5% fetal calf serum (FCS; Invitrogen) and 1% penicillin/streptomycin. HT29-BR cells were generated and maintained as previously described 12. For all experimental work, cells were seeded at the appropriate densities and maintained in fresh media containing 3% FCS for 48 hrs prior to harvesting at 0, 0.5, 6, 15 and 48 hrs. Cell counts were performed with a haemocytometer following trypsinisation and staining with trypan blue. At 48 hrs, apoptosis was measured using the Apo-ONE homogeneous Caspase 3/7 Assay Kit (Promega) as previously described 12. Measurements were performed in triplicate and statistical analysis of the data was performed using Prism 5.0 Software (Graph Pad, San Diego, USA).

Cell culture experiments and subsequent gene expression analyses were performed in triplicate for each time point to minimize technical variability. At the appropriate time points following butyrate addition, HT29 and HT29-BR cells were harvested and RNA extracted. Untreated HT29 cells were used as the control group. Total RNA was prepared using the QIAGEN RNeasy Plus Mini Kit (QIAGEN, Valencia, CA) according to the manufacturer protocols. The RNA integrity and concentration for each sample was assessed using the Agilent BioAnalyzer. Samples with RNA integrity greater than 7 were used for analysis.

Human Exon 1.0ST arrays (Affymetrix Inc., Santa Clara, CA) were used for gene expression analysis and processed according to manufacturer protocols. Analysis of microarray data was performed using the Partek Genomics Suite (v6.6, Partek Inc., St Louis, MO) as previously described30. Differential gene expression was considered significant when the false discovery rate (FDR) <1% and when the fold change in expression was greater than ±2. Ingenuity Pathway Analysis (Ingenuity Systems, Inc., Redwood City, CA) was used to identify potential relationships between differentially expressed genes.

Butyrate-induced apoptosis was observed in the parental HT29 cells but not in the HT29-BR cells (data not shown), consistent with our previously published results 12 and confirms that the HT29-BR cell line is resistant to the apoptotic effects of butyrate. Principle component analysis of the array data revealed clear delineation of the transcriptional profiles between untreated HT29 cells and HT29-BR cells and indicated that changes in gene expression profiles occur most significantly with butyrate treatment of HT29 cells (Supplementary figure 1). In the untreated HT29 cells, no genes were identified as being differentially expressed at 0.5, 6 and 15 hrs when compared to 0 hr control. At 48 hrs, only 95 genes (0.4%) were identified as being differentially expressed indicating that observed effects were attributed to butyrate treatment (Supplementary table 1). The list of differentially expressed genes for HT29 cells treated with butyrate and HT29-BR cells at each time point is included as Supplementary tables 2 and 3 respectively.

ANOVA analysis revealed that for HT29-BR cells at 15 and 48 hrs, 17 genes and 227 genes were differentially expressed respectively when compared to HT29-BR cells at 0 hr (FDR <1%, fold change >±2) (Supplementary table 3). Comparison of the 0.5 and 6 hr time points to 0 hr, showed no significant differences in gene expression. These results indicate that the greatest observable difference in gene expression occurs between HT29 and HT29-BR cells (i.e. t=0 hr) and that minimal change in gene expression occur over time (i.e. up to t=48 hr) in the HT29-BR cells under the cell culture conditions used in this study (Supplementary figure 1).

Of the 17 genes differentially expressed at 15 hrs, 11 of these are involved in DNA replication and the DNA damage response, including 6 genes that are components of the minichromosome maintenance complex (MCM) (Figure 3). Other DNA repair genes identified include CDC45, CHEK1, TYMS, UNG and UHRF1. All 11 genes were down-regulated to a greater extent at 48 hrs in the HT29-BR cells with the exception of UNG where a minimal change in expression was detected.

In addition to those genes involved in DNA repair and replication at 15 hrs, the expression of three genes was up-regulated: GPR155 (up-regulated 2.1 fold at 15 hrs and 2.3 fold at 48 hrs), LAMP3 (up-regulated 2.1 fold at 15 hrs and 2 fold at 48 hrs) and SYT11 (up-regulated 2.1 fold at 15hrs and 2.6 fold at 48 hrs). Down-regulated genes include FAM111B (down-regulated 3.2 fold at 15 hrs and 4.1 fold at 48 hrs) and SLC26A9 (down-regulated 2.3 fold at 15 hrs and 1.8 fold at 48 hrs).

Of the 227 genes identified as being differentially expressed at 48 hrs, 137 (61%) were nuclear proteins, 42 genes (19%) localised to the cytoplasm, 16 genes (7%) were located at the plasma membrane, 13 genes (6%) were confined to the extracellular space and the remaining 16 genes (7%) could not be classified (Supplementary figure 3). Of the nuclear proteins, the expression of nine (6.5%) genes was up-regulated, and the remaining genes were down-regulated. Thirty-one genes (23%) were classified as enzymes, 14 genes (10%) were transcription regulators while 78 genes (57%) could not be classified into any functional category (Supplementary figure 4).

Further bioinformatic analysis also revealed that the activity of seven regulatory molecules to be potentially altered at 48 hrs (Table 1), including the transcription regulators TP53 and KDM5B, and UXT which were all activated. Another seven molecules were predicted to be inhibited including FOXM1, FOXO1, NCOA3, EIF4G1, S100A6, CCNK and CD24.

At 6 hrs, 476 genes were differentially expressed in response to butyrate (FDR <1%, fold change >±2), and over 1000 genes were differentially expressed over the 48 hrs time period. Based on gene expression changes, bioinformatic analysis predicted that the activity of 12 regulatory molecules was deregulated with butyrate treatment. Of note, the activity of MGEA5 was predicted to be inhibited early in the apoptotic process (6 hrs) only, i.e., its activity had returned to levels comparable with the control group at 15 hrs. MGEA5 is a highly conserved enzyme that catalyses the removal of the O-GlcNAc functional group from serine and threonine residues of proteins. Regulation of O-GlcNAcylation of proteins is highly dependent on the hexosamine biosynthetic pathway where it requires metabolic intermediates such as glutamine, glucose and acetyl-CoA to generate the donor sugar, UDP-N-acetylglucosamine, and it is therefore intimately linked with a cell’s nutritional status. Reduced activity of MGEA5, as observed at 6 hrs, may be indicative of low turnover of O-GlcNAcylated proteins. Perturbed cycling of protein O-GlcNAcylation has been implicated in the development of a number of malignancies, including CRC 3132. Many proto-oncogenes, tumour suppressors, cell cycle regulators and signal transducers are O-GlcNAcylated 33, and it has been recently shown that b-catenin is stabilised by O-GlcNAcylation 32. Aberrant Wnt/b-catenin signalling is a hallmark of colorectal tumorigenesis and elevated levels of b-catenin is associated with tumour progression. Additionally, cancer cells preferentially utilise glucose as an energy source (Warburg effect) and enhance the uptake of glucose by over-expressing glucose transporters such as GLUT1. Whether there is a link between this increase in glucose flux and altered cycling of O-GlcNAcylation in cancer cells to influence tumorigenesis or apoptosis presents a novel avenue for further investigation.

The activity of BRD4 was down-regulated by butyrate at 15 hrs specifically. BRD4 regulates gene transcription by binding to acetylated H3 and H4 histone proteins; however, its role in tumorigenesis is not understood. Inhibition of BRD4 by small molecule inhibitors has been shown to have anti-tumour effects and BRD4 has been identified as a potential therapeutic target in cancer 3435. Inhibition of BRD4 has been shown to reduce the expression and transcriptional activity of c-MYC in cancer cells 36. BRD4 has also been shown to stabilise and promote the transcriptional activity of NFkb in cancer cells by binding to the acetylated form of RelA, a subunit of NFkb, to maintain its nuclear localisation and prevent its degradation 37. Deregulation of BRD4, and the potential subsequent downstream effects on NFkb and MYC transcriptional targets, may represent novel mechanisms involved with the anti-tumorigenic effect of butyrate. Deregulated activity of CDKN1A, FOXM1, KDM5B and CCNK is possibly linked to regulation of cell cycle events, which is a well described effect of butyrate. FOXO1, EIF4G1, S100A6 and TGM2, however, have diverse roles in maintaining cellular function and their roles in tumorigenesis are not clearly defined.

Our analysis identified 942 genes as being differentially expressed in the HT29-BR cells that potentially contribute to the development of butyrate resistance (Supplementary table 3), including 40 genes specifically involved in CRC development (Supplementary table 4). Of particular interest is the possible activation of the EGF signalling network in the HT29-BR cell line which may also have implications in cancer therapeutics. Recent reports indicate potential synergistic effects of combined HDI and tyrosine kinase inhibitor therapy, including EGF receptor (EGFR) inhibitors, in cancer patients with EGFR-expressing tumours 3839. The role of EGF signalling in tumour cell proliferation and survival is well known, and over expression of EGFR has been documented in many different cancers although the mechanisms involved with its hyperactivation are not well understood. Reports on the effect of HDIs, including butyrate, on the expression and activity of EGF signalling are inconsistent and may be dependent on the in vitro model being investigated. For example, Chou et al reported that EGFR expression is decreased by HDI in CRC cells, and that this occurred at the transcriptional level 40. Conversely, Song et al reported enhanced activation of EGFR with HDI and that this was associated with concurrent phosphorylation and acetylation of EGFR at three specific lysine residues (K684, K836 and K843) 41. Our results are supportive of those reported by Song et al where EGF signalling by butyrate in HT29 cells likely leads to enhanced cell survival and development of butyrate resistance. Bongers et al have also described elevated EGFR signalling in colorectal polyps and adenomas in comparison to surrounding normal mucosa, and this may be particularly important in the development of polyps derived from the serrated pathway 42.

Our studies have also indicated an association between butyrate resistance and down-regulated expression of members of the MCM protein family. This is supported by a recent report indicating down-regulation of MCM proteins in CRC cell lines in response to trichostatin A 43. Low expression levels of these proteins are seen in quiescent, senescent or differentiated cells and deregulation of the MCM protein complex, composed of MCM2, 3, 4, 5, 6 and 7, is believed to result in genomic instability and ultimately lead to tumorigenesis 44. For example, cells that are deficient in MCM protein expression and/or activity have been associated with hypersensitivity to genotoxic stress and display higher intrinsic levels of DNA damage, even in the absence of genotoxic agents 45. Furthermore, MCM deficient mice have been shown to have a higher incidence of cancer that is believed to be result in higher levels of gene mutations, and mice that are deficient in MCM2 or MCM7 have been shown to develop tumours with high penetrance and that have short latency periods 45. CDC45, whose expression was down-regulated in the HT29-BR cell line, is also a critical part of this process as recruitment of this protein to the MCM complex is required for the initiation of DNA synthesis 46. Lower levels of MCM proteins and CDC45, coupled with reduced levels of CHK1, a protein that regulates the DNA damage response and which was also observed to be down-regulated in the HT29-BR cells, may provide a survival advantage to these cells that enables them to overcome apoptosis in the presence of butyrate.