Deciphering c-MYC-regulated genes in two distinct tissues

Background The transcription factor MYC is a critical regulator of diverse cellular processes, including both replication and apoptosis. Differences in MYC-regulated gene expression responsible for such opposing outcomes in vivo remain obscure. To address this we have examined time-dependent changes in global gene expression in two transgenic mouse models in which MYC activation, in either skin suprabasal keratinocytes or pancreatic islet β-cells, promotes tissue expansion or involution, respectively. Results Consistent with observed phenotypes, expression of cell cycle genes is increased in both models (albeit enriched in β-cells), as are those involved in cell growth and metabolism, while expression of genes involved in cell differentiation is down-regulated. However, in β-cells, which unlike suprabasal keratinocytes undergo prominent apoptosis from 24 hours, there is up-regulation of genes associated with DNA-damage response and intrinsic apoptotic pathways, including Atr, Arf, Bax and Cycs. In striking contrast, this is not the case for suprabasal keratinocytes, where pro-apoptotic genes such as Noxa are down-regulated and key anti-apoptotic pathways (such as Igf1-Akt) and those promoting angiogenesis are up-regulated. Moreover, dramatic up-regulation of steroid hormone-regulated Kallikrein serine protease family members in suprabasal keratinocytes alone could further enhance local Igf1 actions, such as through proteolysis of Igf1 binding proteins. Conclusions Activation of MYC causes cell growth, loss of differentiation and cell cycle entry in both β-cells and suprabasal keratinocytes in vivo. Apoptosis, which is confined to β-cells, may involve a combination of a DNA-damage response and downstream activation of pro-apoptotic signalling pathways, including Cdc2a and p19Arf/p53, and downstream targets. Conversely, avoidance of apoptosis in suprabasal keratinocytes may result primarily from the activation of key anti-apoptotic signalling pathways, particularly Igf1-Akt, and induction of an angiogenic response, though intrinsic resistance to induction of p19Arf by MYC in suprabasal keratinocytes may contribute.


Background
The c-MYC proto-oncogene encodes a transcription factor, c-MYC (MYC), which regulates the expression of cellular targets involved in a wide range of diverse cellular functions, including cell growth, proliferation, loss of cell-cell contact, loss of differentiation and angiogenesis [1][2][3][4][5][6][7][8]. While the predominant role of physiological MYC in most tissues is to promote G1/S transition in the cell cycle (and thereby proliferation) [1,9,10] and inhibit differentiation [11][12][13], deregulated MYC (oncogenic) can lead to uncontrolled proliferation and tumour growth [3]. Paradoxically though, MYC is able to act as its own tumour suppressor, as deregulated MYC activity can also promote apoptosis (both in vitro [14][15][16] and in vivo [17]) and senescence [18,19]. See [20] for a recent review of the MYC field. Such linkage between seemingly opposing functions -proliferation and apoptosis -is also found in other cell-cycle-associated genes, such as E2f, E1a and c-Fos [21].
The mechanisms by which MYC elicits the vast host of biological responses for which it appears to be responsible are not yet fully understood. Currently, around 1,700 genes have been classified as putative MYC targets [22,23] using methods such as serial analysis of gene expression (SAGE) [24], DNA microarrays [3] and subtractive hybridization [25]. It has been hypothesized that MYC may have the potential to regulate up to 15% of the entire genome [26], leading to it being described as a 'master regulator' of gene expression.
Regulatable transgenic mouse models have allowed controlled activation of a modified MYC-containing chimaeric transcription factor (MYC-ER TAM ) in distinct cell populations in adult mice, such as the pancreatic islet βcells [27] and suprabasal keratinocytes (SBK) of skin epidermis [28]. Our previous work has shown that continuous activation of MYC-ER TAM in these diverse tissues exposes the dual potential of MYC to activate pathways involved in cell replication and cell death under differing environmental conditions. In suprabasal epidermis, MYC promotes entry of post-mitotic keratinocytes into the cell cycle, concomitant with loss of differentiation and increased vascularization leading to formation of precancerous papillomas [28]. In contrast, although MYC promotes rapid cell cycle entry of pancreatic β-cells, these cells quickly proceed to undergo apoptosis leading to severe cell depletion and diabetes [27]. This indicates a crucial role for tissue context and the surrounding microenvironment in determining cell fate.
The divergence of MYC-induced phenotypes between these two tissues has enabled us to compare MYC-regulated gene expression patterns over a time course of MYC-ER TAM activation, by employing high-throughput transcriptome analysis using microarrays. Comparison of the transcriptional response between the two tissues identified potential signalling pathways which may promote apoptosis of β-cells and prevent apoptosis in SBK; the DNA-damage response pathway, and the Insulin-like growth factor 1 (Igf1) signalling pathway, respectively. In addition, up-regulation of angiogenesis-related genes and of those encoding members of the steroid hormone-regulated Kallikrein serine protease family was found in SBK but not in β-cells. Kallikreins may increase availability and action of Igf1 through proteolysis of Igf1 binding proteins. Together with angiogenesis, Kallikreins may provide a local tissue-specific regulatory mechanism for determining ultimate MYC function in vivo.

Results and Discussion
Activation of MYC-ER TAM mediates transcription of genes involved in a wide range of cellular functions Time courses were set up following activation of MYC in β-cells (pancreas) and SBK (skin) via administration of 4-hydroxytamoxifen (4OHT) for 4 hrs, 8 hrs, 16 hrs and 32 hrs as described (Methods). Vehicle-treated samples acted as direct time point controls for 4OHT-treated samples. Laser capture microdissection (LCM) was utilized to allow isolation of pancreatic islet tissue. Significant gene expression changes for the main experimental conditions (4OHT treatment, tissue type and time point following initial treatment) and their interactions, as well as information on the effects of further covariates such as batch effects and RNA quality, were identified using a custom R package, Envisage (available on request).
Analysis of gene expression for 12,349 curated probe sets (Methods) identified 6,633 unique genes as being significantly altered following activation of MYC with a false discovery rate of 5%. 1,615 genes showed significant effects for the joint effects of 4OHT treatment, time and tissue type; 2,015 genes showed significant effects for the interaction between 4OHT treatment and tissue type; 2,221 genes showed significant effects for the interaction between 4OHT treatment and time; and 1,843 genes showed significant effects for the main effect of 4OHT treatment only. Of the MYC-responsive genes, the expression levels of 1,199 were altered greater than 2-fold (up-or down-regulated) after only 4 hours of 4OHT treatment for the pancreatic β-cells, while only 530 were similarly affected for SBK. However, at 8 hours following initial 4OHT treatment, the expression levels of 1,905 and 1,882 were altered greater than 2fold for the pancreas and the skin respectively ( Figure  1A). This suggests a more prominent initial response to MYC in β-cells compared to SBKs.
Gene ontology (GO) enrichment analysis using the DAVID functional annotation tool [29] identified enrichment of genes involved in myriad cellular functions following activation of MYC. The majority of MYCresponsive genes were involved in metabolic, transcriptional, transportational and signal transduction pathways ( Figure 1B; green bars). Genes involved in post-transcriptional modification (e.g. RNA processing, RNA splicing) and post-translational modification (e.g. protein catabolic process, protein folding, protein localization and ubiquitin-dependent catabolic process) were also significantly enriched. Similarly, a significant enrichment of genes relating to ribosome biogenesis was detected, suggestive of MYC's recently elucidated role as a regulator of ribosome biogenesis and protein synthesis [30]. As expected given the role of MYC in proliferation, genes involved in cell cycle progression were amongst the most significantly enriched (p = 1.21 × 10 -17 ). Genes involved in apoptosis and DNA-damage checkpoint pathways were also enriched (p = 7.81 × 10 -3 and p = 5.07 × 10 -5 respectively), along with genes involved in related functions such as cellular response to stress (p = 1.56 × 10 -8 ) and cytoskeleton organisation (p = 9.12 × 10 -5 ).
Enrichment of GO terms for MYC-responsive genes showing early changes in expression (≥ 2-fold change within the first 8 hours of MYC activation) for the pancreas ( Figure 1B; blue bars) and skin ( Figure 1B; red bars) identified similar numbers for both tissues, with the exception of genes relating to DNA-damage and DNA replication, where a larger number of genes are detected for the pancreas than for the skin. These results indicate that whilst expression of genes relating to multiple cellular functions is common to both tissues, expression of genes involved in DNA-damage and replication is more specific to the β-cells.

Expression of putative MYC target genes following MYC-ER TAM activation
The MYC Target Gene Database [31] currently identifies 1,697 genes as putative MYC targets [22]. Of these, 13.4% and 19.2% were found to be both MYC-responsive and show a 2-fold change in expression in the skin and pancreas respectively within 8 hours (Additional file 1, Table S1). The predominant role for these genes was in DNA replication, biosynthesis, metabolism, cell cycle, cell division and other related functions. Cellular functions relating to apoptosis and cell death were also highly enriched, although to a lesser degree than those relating to cellular proliferation. These data suggest that activation of the MYC-ER TAM protein in vivo leads to a rapid change in the expression of a large number of putative MYC targets. However, known target genes represent only a small fraction of detected MYC-responsive genes, indicating that the majority of these observed expression changes may be downstream of direct MYCinduced transcription.
To identify the level of overlap between the genes classed as significantly altered in this study and those Figure 1 Transcriptional response to MYC-ER TAM activation. A) Activation of MYC-ER TAM led to a significant change in expression of a large number of genes within only 4 hours for the pancreas (≥ 2-fold change, p ≤ 0.05). A similar transcriptional response was not detected until 8 hours post-MYC-ER TAM activation for SBK. B) Analysis of 6,633 genes, whose expression was altered within 8 hours upon activation of MYC-ER TAM in either pancreas or skin (green bars), identified enrichment of genes in a wide range of MYC-mediated functions. Analyses of genes whose expression was altered greater than 2-fold within 8 hours for the SBK (red bars) and the pancreatic β-cells (blue bars) showed a similar level of response to MYC-ER TAM activation in both tissues within 8 hours, with a slight increase in the level of response for the pancreas as compared to the skin, in particular for DNA-replication and response to DNA-damage. p-values represent significance of enrichment of GO classes for the 6,633 genes. C) Genes identified as showing significant differential expression due to the joint effects of MYC-ER TAM activation and tissue (with or without an interaction with the time point) were clustered to identify functionally related genes (red, up-regulated; black, no change; green, down-regulated). The blow-up identifies a group of functionally related genes showing increased expression in SBK but not islet β-cells.
identified in previous analyses, we utilised the Gene Set Enrichment Analysis (GSEA) program developed by the Broad Institute [32]. This allowed us to identify gene sets in which significant differentially expressed genes are enriched. Gene sets were taken from the Molecular Signatures Database (MSigDB), as well as from additional published datasets. Additional file 1, Table S7 and Additional file 1, Table S8 show the results from GSEA (FDR < 0.01) for the genes showing significant expression at the early time points (4 hours and 8 hours) for the pancreas and skin, respectively. These results indicate a clear enrichment in the pancreas of genes annotated as being related to MYC function from previous publications, including MYC target genes from the MYC Target Gene Database [22], genes indicative of a MYC-induced oncogenic signature from Blid et al. [33] and genes up-regulated in the studies of Coller et al., Schumacher et al., Yu et al., and Lee et al. [3,[34][35][36]. Similarly, down-regulated genes showed enrichment with down-regulated genes from the study of Yu et al. [36]. In comparison, genes up-regulated in the skin are enriched for genes also found to be up-regulated in the studies of Coller et al., Yu et al. and Zeller et al. [3,22,36], while MYC target genes from the MYC Target Gene Database [22], genes indicative of a MYC-induced oncogenic signature from Blid et al. [33] were enriched with an FDR value < 0.017. This suggests that the gene expression signature identified in both the pancreas and skin is indicative of changes in expression related to MYC function.

Cell cycle response following MYC activation
One of the key functions of the MYC onco-protein is promotion of cell cycle progression, particularly G1/S phase transition [1,2,37]. Activation of MYC in suprabasal keratinocytes (SBK) and pancreatic β-cells has been previously shown to initiate G1/S transition in target cells [27,28,38], and this was seen through immunohistological staining with anti-Ki67 antibodies (Figure 2A), as well as in the response detected in genes relating to cell cycle progression by gene ontology classification ( Figure  1B and Additional file 1 Table S2). A subset of some interesting genes from this list is shown in Table 1.
Activation of MYC in the β-cells resulted in a change in expression of 213 cell cycle-and proliferation-related genes within 8 hours of MYC activation, with 116 genes up-regulated and 101 genes down-regulated (Additional file 1, Table S2). Pcna, a MYC target gene associated with the cell cycle, was expressed in both pancreatic βcells and SBK throughout most of the time-course, with Cdc25a expressed only in the former. In β-cells, cyclin genes Ccnd1, Ccnd2 and Ccne2, whose products are necessary for G 1 /S-phase transition in the cell cycle, were up-regulated within 4 hours of MYC activation.
Cyclin genes Ccna2, Ccnb1 and Ccne1, whose products are involved in later G 1 /S-phase and G 2 /M-phase cell cycle events, were up-regulated greater than 3-fold subsequently at 8 hours (Table 1 and Figure 2C).
Activation of MYC in the SBK resulted in a less prominent cell cycle response compared to β-cells, with a change in expression of 144 cell cycle-and proliferation-related transcripts within 8 hours of MYC activation -73 genes up-regulated and 74 genes downregulated (Additional file 1, Table S2). Of G 1 /S-phase cell cycle genes, Ccnd2 and Ccnd3 showed a ≥ 2-fold increase in expression at 8 hours. In contrast to β-cells, later cell cycle genes such as Ccna2 and Ccne2 were either down-regulated or unchanged in SBK. Interestingly, the Ccnb1 gene whose product, cyclin B1, is involved predominantly in later cell cycle events (G2/M transition), was down-regulated 2-fold in SBK early in the time-course. Cyclin B1 interacts with Cdc2a (Cdk1) to form the mitotic initiation complex, which was also down-regulated by~2-fold at 4 hrs in epidermis. This contrasts with β-cells, in which both Ccnb1 and Cdc2a showed significant up-regulation. Although speculation here, these findings may suggest SBK are able to enter the G1/S phase of the cell cycle, which is a known function of MYC, but not progress through the G2/M phase. Interestingly, it was previously shown that over-expression of MYC causes a P53-dependent G2 arrest in normal fibroblasts [39]. Such cells may then be enforced by MYC to reinitiate DNA replication, resulting in aneuploidy.
Of these cell cycle-related genes, Ccna2 [40] and Ccnd1 [41,42] have been previously designated as putative MYC targets through high-throughput screening, and Ccnb1 [43] and Ccnd2 [44][45][46] have been previously confirmed as direct transcriptional targets of MYC through the use of chromatin Immunoprecipitation (ChIP) analysis. The cyclin D2-related kinase Cdk4, also a previously characterized direct MYC target [47], showed increased expression after 4 hours of MYC activation in the pancreas, with a 6-fold increase detected subsequently at 16 hours. Cdk4 was also found to be highly up-regulated at 8 hours in the skin, with a fold change of almost 12 (Table 1). No significant change was detected for the cyclin E-associated CDK gene Cdk2 in either the skin or the pancreas, However Cdk7, which has a role in both activating cyclin complexes and regulating transcription, was up-regulated at 8 hours in the skin.
Down-regulation of another known MYC target gene, the cyclin dependent kinase (CDK) inhibitor Cdkn1b (p27 Kip1 ), which inhibits G 1 /S-phase transition by association with the cyclin E-Cdk2 complex [48], was detected for both the skin and the pancreas. Also, the expression of Cks2, a MYC target gene whose product is  B) The change in expression of genes relating to cell cycle following activation of MYC-ER TAM is represented as a heatmap (red, upregulated; black, no change; green, down-regulated). Significant change in expression is detected for key early G1-S phase transition genes such as Ccnd1, Ccnd2 and Ccne2, indicating activation of cell cycle progression, particularly for the pancreas. C) qRT-PCR validation for gene expression changes in genes relating to cell cycle progression indicated a clear proliferative response in the pancreatic β-cells following MYC-ER TAM activation. In particular, genes relating to G1/S-phase transition such as Ccnd1, Ccnd2 and Ccne2 showed early changes in expression (8 hours), whilst genes whose products relate to later cell cycle events such as Ccna2 and Ccnb1 showed increased expression at later time points (32 hours). Few significant changes were detected for these key proliferation markers in SBK. This may be due to the relatively low proportion of keratinocytes that are responsive to Myc-induced proliferation at these early time-points, whilst differentiated keratinocytes of the granular layer are refractory to the proliferative influence of MYC. involved in degradation of p27 Kip1 , increased from 8 hours following MYC activation in the pancreas. Interestingly, the Cdc2a gene, whose product Cdk1 is essential for mammalian cell division [49,50], was also found to be highly up-regulated in β-cells (Table 1). Cdk1 has been found to substitute for other CDKs to drive cell cycle progression [51], and is particularly associated This may indicate a significant role for Cdk1 in the promotion of cell cycle progression following MYC activation in β-cells. Alternatively, it has been shown that premature activation of Cdk1 can lead to mitotic catastrophe in G 2 /M-phase and apoptosis (upstream of p53induced mitochondrial outer membrane permeabilization) in neurons [53]. Given that this CDK was also detected at later time points, this may indicate a possible role for Cdk1 in the MYC-induced apoptosis pathways. In addition to this, the CDK inhibitor Cdkn2c (p18 Ink4c ), which inhibits G 1 /S-phase transition via interactions with Cdk4 [54] and Cdk6 [55], was down-regulated early in the pancreas. However, by 16 hours expression levels had risen dramatically, which may be indicative of cell cycle arrest prior to apoptosis. In addition, the CDK inhibitor Cdkn1a (p21 Cip1 ) -a downstream target of the tumour suppressor p53 -was upregulated at 8 hours (Table 1). Growth arrest specific gene 6 (Gas6) was up-regulated in SBK (Table 1), and has been found to have different roles dependant on tissue location, including promotion of cell cycle progression in fibroblasts and Schwann cells [56,57], and has been shown to play a role in cellular survival by inducing signalling through the PI3K/Akt pathway leading to phosphorylation of the forkhead box protein Foxo1 in endothelial cells [58] and in oligodendrocytes of the central nervous system [59]. Interestingly, the growth arrest and DNA-damage-inducible 45 gamma gene, Gadd45g, a proposed MYC target whose product is involved in growth arrest at the G 2 /M DNA-damage checkpoint [60], showed increased expression at 4 hours in SBK, and remained 3fold up-regulated throughout the time course, whilst down-regulation at 8 hours was detected in β-cells. This suggests potential activation of pathways to limit unchecked proliferation in the keratinocytes (Table 1).
Genes relating to increased cellular mass, cytoskeleton organization and DNA replication were also detected for SBK (Additional file 1, Table S3), including the membrane skeletal proteins Adducin 1 (Add1) and Pdlim3, the actin-modulating protein Cofilin 1 (Cfl1), members of the kinesin family of microtubule motor proteins, members of the myosin superfamily of actin binding motor proteins (e.g. Myh4, Myoc, Myo1f), and members of the tubulin family of microtubule proteins (Tuba3a, Tubb2c, Tubb6). Plectin 1 (Plec1), one of the main components of the cytoskeleton, showed an increase in expression of roughly 3-to 4-fold throughout the early stages of the time course (Table 1). This increased activity of microtubule formation and actin formation for both the pancreas and skin is indicative of increased cellular turnover in both tissues.

Apoptotic response following MYC activation
The ultimate phenotypic response to activation of MYC in pancreatic β-cells is apoptosis [27]. Immunohistological staining for Caspase 3, an early marker for initiation of apoptosis pathways, indicated an apoptotic response to MYC activation in the β-cells but not in the SBK ( Figure 3A). In contrast, MYC-activated SBK that have begun a process of terminal differentiation, re-enter the cell cycle but are protected from conventional apoptosis [33,38]. These cells will ultimately be shed and removed from the surrounding micro-environment thus ridding the host of potentially harmful pre-cancerous cells [61]. Our array data confirm a large transcriptional response detected in genes relating to apoptosis and survival by gene ontology classification in both tissues ( Figure 3B and Additional file 1 Table S4). A subset of important genes from this list is shown in Table 2.
Activation of MYC in pancreatic β-cells identified a significant change in expression for 92 genes relating to cell death and apoptosis. Of these, 42 genes showed an increase and 50 genes showed a decrease in expression (Additional file 1, Table S4). Early activation of key regulators of apoptosis featured prominently in these data. Activation of MYC in SBK resulted in significant changes in expression for 66 genes relating to apoptosis and cell survival, including 37 genes showing an increase Selected MYC-responsive genes relating to cell growth and cell proliferation. 'MYC-response p-value' is the p-value identified for the highest-order interaction of the MYC activation variable and represents the significance of this term within the selected model. Flags represent contrast p-values comparing 4OHT-treated and vehicle-treated samples at specific time points ('*', p ≤ 0.05; '**' p ≤ 0.01). Cells are colour-coded based on a detected fold-change greater than 1.5-fold (red, up-regulated; blue, down-regulated) and 29 genes showing a decrease in expression (Additional file 1, Table S4). The tumour suppressor Cdkn2a, which encodes for the CDK inhibitor p16 Ink4a and the alternative reading frame tumour suppressor p19 Arf (p14 Arf in humans), was up-regulated 2-fold at 4 hours and remained at an elevated level throughout the time course for β-cells, but remained unchanged in SBK (Table 2 and Figure  3C). Previous studies have identified a possible role for the p19 Arf /p53/Mdm2 tumour suppressor pathway in  black, no change; green, down-regulated). C) qRT-PCR validation for gene expression changes in genes relating to apoptosis and cell survival confirms disparate changes in gene expression between skin and pancreas. Genes relating to DNA-damage in particular, such as Atr and Chk2, show a clear increase in expression for pancreatic β-cells, which confirms changes seen in microarray data, as did apoptosis genes Cycs and Endog. As with the microarray data, up-regulation of the Igf1 gene was detected from 8 hours for the skin following MYC-ER TAM activation, with an increase in expression also detected for the pancreas at the later 32 hour time point.  MYC-mediated apoptosis [62,63]. Importantly, data from the Evan lab has elegantly shown that high levels of MYC-ER TAM activation present in this transgenic model, led to expression of p19 Arf concomitant with apoptosis [64]. In contrast, low levels of MYC-ER TAM activation did not induce apoptosis or p19 Arf but rather β-cell proliferation. However, Finch et al. previously showed that loss of p19 Arf in β-cells of the MYC-ER TAM transgenic model resulted in mainly increased proliferation, not suppression of apoptosis [65]. Thus the role of p19 Arf in defending against aberrant oncogenic MYCinduced hyper-proliferation may be related to cell cycle arrest, and not directly to apoptosis pathways in β-cells.
The anti-apoptotic function of Bclx L seen in Rip7-Bclx L /pIns-MYC-ER TAM (RM) double transgenic mice indicates that MYC-induced apoptosis is related to the Bax/Bak-mediated intrinsic mitochondrial pathway [27,66]. Activation of this intrinsic apoptotic pathway was evident at the transcriptional level in β-cells by continued 2-fold increased expression of Bax and the somatic Cytochrome c gene (Cycs) from 16 hours onwards after MYC activation, and up-regulation of the mitochondrial respiratory gene for Endonuclease G, Endog, after only 4 hours of MYC activation. Both Bax [67,68] and Cycs [42,69] have been previously shown to be putative direct MYC targets due to the presence of non-canonical E-box MYC-Max binding sites, and association of MYC with the Bax promoter has been previously demonstrated through ChIP [68].
Gene expression profiling of pancreatic β-cells identified a strong and rapid induction of the DNA-damage response pathway (Table 2). A large increase in expression was detected after 8 hours of MYC activation for Rad51 and H2afx, previously identified MYC targets whose protein products are involved in homologous recombination and repair of DNA [70,71]. Also, significant up-regulation of Hus1 and Rad1 -whose products form the 9-1-1 DNA-damage sensing machinery with Rad9 -indicated that oncogenic stress through deregulation of MYC resulted in the induction of DNA double strand breaks. The gene for the DNA-damage mediator Atr was also found to be up-regulated by 2-fold throughout the time course from 4 hours, and the associated checkpoint kinases Chk1 and Chk2 were up-regulated 2-fold from 8 hours (Table 2 and Figure 3C). The gene for the double-strand break-related DNA-damage mediator Atm showed 2-fold down-regulation at 8 hours, although it has been shown that Atm plays a significant role in MYC-induced apoptosis in lymphomagenesis in mice [72]. The genes Cdkn2a (p16/p19 Arf ) [2,62] and Atr [73] are previously categorized MYC target genes. The checkpoint kinase genes Chk1 and Chk2 have not been previously classified as MYC target genes, indicating that the observed changes in expression may occur downstream of Atr. Interestingly, as mentioned previously, expression of Cdkn2c (p18 Ink4c ), which inhibits Cdk4/6, was significantly up-regulated (6-fold) in βcells at 16 hrs. p18 Ink4c interacts directly with Atm/Atr leading to an increase in p53 protein, and promotion of growth arrest and/or cell death, suggesting a link between p18 Ink4c and the DNA-damage response pathways [74]. Comparatively few DNA-damage related genes showed significant changes in expression for SBK, although we cannot exclude the possibility that DNAdamage proteins may be regulated by MYC through non-transcriptional means (e.g. protein phosphorylation).
Of particular interest was the survival factor gene Igf1, whose product has been shown to inhibit MYC-induced apoptosis in vitro by blocking Cytochrome c release from the mitochondria through the Akt1 tumour suppressor pathway [75]. Igf1 was up-regulated throughout much of the time-course for SBK, and this was confirmed using qRT-PCR ( Figure 3C). Consistent with this finding, Akt1 was up-regulated 3-fold at 8 hours and 2fold at 32 hours. A similar pattern was observed for Akt2, a second member of the Akt protein kinase family. This strongly supports the view that activation of the Igf-Akt pathway may contribute to suppression of MYC-related apoptosis. Interestingly, Igf1, Akt1 and Akt2 were up-regulated at later time points in β-cells. It is therefore possible that the Igf-Akt pathway is activated in both tissues, but that the β-cells may be able to bypass these signals.
The pro-apoptotic Bcl2 family member Pmaip1 (Noxa) and the gene encoding inhibitor of growth protein p47 Ing3 (Ing3) both showed a decrease in expression of 2-fold at 8 hours in SBK, but exhibited no change in β-cells (Table 2). Over-expression of p47 Ing3 results in cell cycle arrest and apoptosis in cancer cell lines [76], and reduced expression and loss of heterozygosity of the p47 Ing3 locus have been found to be associated with human head and neck squamous carcinomas [77]. Also seen were several members of the TNF superfamily of apoptosis-inducing receptors. Tnfrsf12a, which has been found to be involved in inducing both apoptosis and angiogenesis [78,79], showed an increase in expression in the skin of 3-fold at 8 hours. Tnfrsf4, whose product has been implicated in promoting survival through induction of Bcl2 and Bcl XL expression in CD4 T cells [80], similarly showed an increase in expression of 2fold at 8 hours.

A pro-angiogenic response in suprabasal epidermis
The placental growth factor gene, Pgf¸showed a marked increase from 8 hours throughout the time course for the skin (Table 3). In contrast, a 2-fold down-regulation was detected for the pancreas throughout much of the time-course. Pgf is a member of the vascular endothelial growth factor (VEGF) family, and has been shown to result in increased numbers, branching and size of dermal blood vessels following over-expression in basal keratinocytes of adult mice [81]. This may indicate a role in the development of neovasculature seen following MYC activation in the SBK [28]. Vegfa, a further member of the VEGF family important in angiogenesis, was downregulated in the pancreas but showed up-regulation in the skin at 8 hours. In contrast, Vegfc, important in the growth of lymphatic vessels, showed down-regulation in the skin at early time points, but was up-regulated almost 4-fold in the pancreas at the later 32 hour timepoint. Vegfb showed a 2-fold increase from 16 hours in pancreas compared to no change in the skin. These results indicate a transcriptional response upon MYC activation for genes relating to neovascular growth.

Activation of MYC leads to loss of differentiation
Activation of MYC is often associated with loss of differentiation of cells and has been found to block terminal differentiation in a variety of cell types [11,82,83]. Activation of MYC in the pancreatic β-cells resulted in down-regulation of Ins1 and Ins2 by > 5-fold at 4 hours, indicating acute loss of Insulin production within a short time period following MYC-deregulation. However, the expression levels of both subsequently increased dramatically showing almost 10-fold up-regulation from 16 hours (Table 3). Since the islet area used for RNA extraction was roughly identical for each sample, this indicates acute increase in the levels of Insulin production within the β-cells in response to continuous MYC activation, and not in response to an increase in β-cell mass. Although perhaps a paradox at first glance, this response may be the result of a positive feedback loop due to increased Insulin release into the bloodstream immediately following MYC activation as we have previously shown [23]. Observation of transcript levels of both mRNAs at a later time point (72 hours; data not shown) indicated that this period of high Insulin production is limited, as gene expression subsequently returned to lower levels indicative of loss of βcell differentiation. This indicates a short window within the first two days where a balance is struck between an increased rate of Insulin production and the simultaneous loss of cells due to MYC-driven apoptosis. Members of the homeodomain transcription factor family, Pdx1, Pax4, Hb9, Nkx2.2 and Nkx6.1, are essential in pancreatic development [84]. Probe-sets for the pancreatic and duodenal homeobox gene Pdx1 (or Ipf1), whose product activates transcription of the Insulin gene as well as a number of genes involved in glucosesensing [85], showed a significant loss (> 3-fold) in expression at 8 hours following MYC activation (as well as potentially significant loss in expression < 2-fold at all other time points), which correlated with the early reduction seen in Insulin production ( Table 3). The transcription factor gene Nkx6.1, whose product is essential for β-cell differentiation [86], also showed significant down-regulation in the early stages of MYC activation, although expression of this gene was shown to increase during later stages. Further Pdx1-regulated  genes Slc2a2 (Glut2; previously classed as a putative MYC target gene) and Gck (Glucokinase), both part of the glucose-sensing machinery and involved in membrane transport and phosphorylation of glucose respectively, also followed similar expression profiles (Table 3). Gu et al. identified a group of 217 mature islet-specific genes [87], 63 of which showed a significant response following MYC activation in our analysis (Additional file 1, Table S6). The majority of these showed down-regulation at early time points. These data indicate a loss in β-cell differentiation and carbohydrate metabolism function following activation of MYC. Activation of MYC in the SBK resulted in significant changes in expression of many genes relating to differentiation (Additional file 1, Table S5). In particular, it was clear that the primary result of MYC activation on these genes was down-regulation, with 199 differentiation-related genes showing a loss of expression compared to only 112 showing up-regulation. In addition to these general differentiation markers, activation of MYC led to down-regulation of several key keratinocyte differentiation genes (Table 3). Most notable was a significant 3-fold decrease in expression for the Involucrin gene, Ivl, after only 4 hours that was maintained throughout much of the time course. Involucrin is a key factor in the progression of differentiation of keratinocytes which works together with its substrate transglutaminase to cross-link with membrane proteins and provide support to the cell [88].
Cornifin (Sprr1b), a precursor to the epidermal cornified envelope, is a further keratinocyte differentiation markers that has been shown to affect the number of distinct layers of differentiated keratinocytes [89,90]. As with Ivl, Sprr1b showed consistently marked down-regulation throughout the time course (Table 3). Similarly, Cystatin A (Csta), a cysteine protease inhibitor that is found expressed in keratinocytes as the precursor of the cornified cell envelope [91,92], showed > 2-fold downregulation throughout much of the time course. Up-regulation of α and β integrin genes such as Itga7, Itga9, Itgb2, Itgb3 and Itgb6, particularly at later time points, suggests altered adhesion of SBK with surrounding cells and the extracellular matrix following MYC activation (Table 3). Also, expression changes were detected for several Keratin genes, including up-regulation of the suprabasal-specific Krt1 and the basal-specific Krt14 at 8 hours (Table 3), which encode fibrous structural proteins in keratinocytes.
Previous findings from the Watt group in which MYC is targeted to basal keratinocytes has, in contrast, shown that activation of MYC promotes an increase in the number of proliferating keratinocytes concomitant with promotion of terminal differentiation of epidermal stem cells [89]. In the microarray experiment of Frye et al. [90], gene expression signatures were compared between whole skin sections from 4OHT-treated K14-MYC-ER TAM mice and 4OHT-treated WT mice to identify cellular networks involved in the promotion of terminal differentiation of epidermal stem cells at the expense of hair lineages [90]. Activation of MYC for 4 days was sufficient to cause hyperproliferation of the interfollicular epidermis, with increased expression of genes relating to both proliferation (e.g. Ki67, Krt17, Krt6, Cdc2, Cdc20) and interfollicular epidermis differentiation (Filaggrin and Cornifin/Sprp1). In contrast, expression of genes relating to early G1/S-phase cell cycle progression are more prominent when MYC is activated in SBK, whilst keratinocyte-specific differentiation genes such as Sprp1 generally show down-regulation. This suggests that activation of MYC suppresses differentiation in keratinocytes already undergoing terminal differentiation (i.e. SBK) to allow proliferation.
In a more recent study, low, intermediate, or high levels of MYC activity were induced in basal keratinocytes, and showed that MYC drives proliferation at all levels, but at high levels MYC promotes keratinocyte differentiation [93]. Promotion of differentiation may therefore act as a fail-safe mechanism against neoplastic conversion of epidermal stem cells.

The transcriptome fingerprint of MYC activation is varied in distinct tissues: pancreatic islets and skin
Clustering of genes whose expression changed significantly due to the joint effects of MYC activation and the tissue type identified genes whose expression profiles were correlated across the time courses for the two tissues, indicating possible co-regulation and functional similarity. Genes relating to DNA replication, DNAdamage checkpoint and cell cycle showed tight co-regulation, and showed significantly increased expression in the pancreatic β-cells. The expression profiles of minichromosome maintenance (MCM) deficient genes Mcm2, Mcm5, Mcm6 and Mcm7, whose products make up part of the MCM complex involved in DNA unwinding [94], the Cdt1 gene, whose product is involved in association of the MCM complex with chromatin, and various helicase related genes were found to be closely related, indicated co-expression of genes relating primarily to DNA replication. Close correlation with these genes was also seen in the pancreas for DNA-damage checkpoint related genes Atr and Chk1. These genes showed consistently high levels of expression in the βcells, indicating a key role for DNA-damage response and repair in MYC-induced apoptosis. Conversely, no significant change was detected for these genes in the SBK.
In SBK, close correlation was detected for genes involved in proteolysis, particularly members of the Kallikrein family of serine proteases. Expression of Kallikrein genes (Klk1, Klk9, Klk21, Klk24 and Klk27) was found to increase significantly within 8 hours following activation of MYC, and continued to increase dramatically throughout the time course (Table 3 and Figure  1C). Klk1 and Klk9 have previously been found to be expressed throughout the epidermis of normal human skin [95], and play a role in degradation of the extracellular matrix (ECM) and loss of squamous cells during differentiation. Deregulated expression of Kallikreins has also been implicated in many cancer types [96][97][98][99][100][101]. The mouse Kallikreins Klk21, Klk24 and Klk27 have also been shown to be functionally active within the testes, both in degradation of the extra-cellular matrix and initiation of survival through degradation of Igf1bp3 [102][103][104]. Kallikrein proteins have been associated with angiogenesis. Growth of new vasculature is facilitated by regulating the tissue micro-environment, particularly through degradation of the extra-cellular matrix often via activation of matrix metalloproteinases (MMPs). This indicates a functional relationship between Kallikreins and the vascularisation-related gene, placental growth factor Pgf (a member of the vascular endothelial growth factor family of angiogenesis-related genes) and the Thrombospondin 2 gene Thbs2 (an anti-angiogenic factor which has been linked to extra-cellular matrix remodelling through modulation of matrix metalloproteinases [105]) which showed close correlation with these genes. Given the prominent angiogenic phenotype that develops in these pre-cancerous skin lesions following MYC activation [33], it is reasonable to speculate here that Kallikreins and Pgf may be involved.

Conclusions
Deregulation or over-expression of the MYC onco-protein is a frequent feature of human cancers, which attests to the pleiotropic role that ectopic MYC plays in cellular function. However, oncogenic MYC can also trigger activation of intrinsic tumour suppressor programs such as p19 Arf /p53, which serve to limit propagation of such harmful cells by inducing growth arrest or apoptosis. While much is known about the mechanisms of MYC functions, the pathways responsible for deciding the ultimate fate of the cell between 'life' (uncontrolled proliferation) and 'death' (i.e. apoptosis) in vivo are not yet clear. The decision for a cell to become apoptotic depends on the complex interactions of many pro-and anti-apoptotic factors. Different tissues may exhibit varying levels of these factors ultimately determining the fate of a cell. However, tissue-specific environmental characteristics can also affect the interaction between these factors, having a decisive effect on cell fate. The MYC-ER TAM transgenic mouse model allows controlled overexpression of MYC in distinct adult tissues, SBK (epidermis) and pancreatic β-cells, enabling tracking of early changes downstream of aberrant MYC activity. In this study, high-throughput transcriptional profiling was used to identify transcriptional events that may provide clues to explain the disparity in the phenotypic response to MYC activation in SBK (proliferation and survival) and pancreatic β-cells (proliferation and apoptosis).
Whilst expression of genes relating to multiple cellular functions (eg. metabolic, transcriptional, transportational, ribosome biogenesis) is common to both tissues, expression of genes involved in DNA-damage and replication is enriched to β-cells. Consistent with early increased expression of Ki67 in β-cells [27], key G 1 /Sphase genes (e.g. those encoding Cyclin D2, Cyclin E and Cdk4) showed early changes in expression, whilst G 2 /M-phase genes (e.g. those encoding Cyclin A and Cyclin B) showed changes at later time points. Downregulation of the cyclin-dependent kinase inhibitor (CDKI) Cdkn1b (p27 Kip1 ) gene in β-cells was evident as previously shown in other cell types [48]. The CDKI Cdkn2c (p18 Ink4c ), which inhibits Cdk4 and Cdk6, was also down-regulated, consistent with G 1 /S phase transition. The short time period over which these changes were seen supports the idea that MYC-induced cell cycle progression occurs through direct activation of key cell cycle genes such as Ccnd1, Ccnd2 and Ccne2.
In SBK, although there were changes in cell cycle related genes, the response was much less prominent in comparison with β-cells. Gene expression changes included up-regulation of Krt6a, Pcna, Ccnd3, Cdk4, and the key G1/S phase cyclin gene Ccnd2, accompanied by significant down-regulation of the CDKI gene Cdkn1b (p27 Kip1 ) throughout the time course. However, further cell cycle related genes (e.g. Ccna1, Ccne2 and Cdk2) did not appear to be measurably affected by MYC activation in SBKs. It is possible that SBK may at some point struggle to progress through the G 2 /M phase, which may be indicated by down-regulation of Ccnb1 and Cdc2a, whose products are essential in later cell cycle stages. This contrasts with pancreatic β-cells, in which we found both Ccnb1 and Cdc2a significantly upregulated. The less pronounced cell cycle response in skin may be due to the relatively low proportion of keratinocytes that are responsive to Myc-induced proliferation at these early time-points. It has previously been shown that there is only a narrow window when very early suprabasal cells that have migrated out of the basal layer, are capable of responding to Myc-induced cell cycle entry [28,105]. The more differentiated keratinocytes of the granular layer are refractory to the proliferative influence of MYC.
Gene expression profiling of the pancreatic β-cells identified the DNA-damage checkpoint pathway as a likely route by which MYC mediates apoptosis in this system, leading to downstream activation of p53 and Bax-mediated release of Cytochrome c from the mitochondria. In addition, close correlation was seen in the pancreas for DNA-damage checkpoint related genes Atr and Chk1, and members of the MCM complex, Mcm2, Mcm5 and Mcm7. The change in expression for these genes following MYC activation was consistently high in the β-cells, suggesting a key role for DNA-damage response and repair in MYC-induced apoptosis. Conversely, no significant change was detected for these genes in the SBK. Recent evidence strongly suggests that deregulated MYC induces rapid accumulation of DNAdamage, which is the primary cause of activation of the Atm/Atr-dependant checkpoint [106,107]. The study of Dominguez-Sola et al. (2007), in particular, suggests that the pleiotropic role of MYC is due not only to transcriptional regulation of downstream genes, but also due to direct interactions with the DNA. This study also showed that over-expression of MYC results in DNAdamage and checkpoint activation. Consequently, the activation of DNA-damage response (DDR) pathways results in ultimate destruction of the offending cell.
Whilst direct control of these genes by MYC is not discernible from these data, it is clear that MYC deregulation induces a transcriptional response representative of cells undergoing DNA repair, which is a likely explanation for activation of the intrinsic apoptotic pathway. These results fit with the hypothesis that deregulated MYC leads to oncogenic stress and DNA-damage, although whether this is direct or indirect remains to be seen.
With regard to events downstream of the DDR, we found an increase in expression of genes associated with activation of mitochondrial outer membrane permeabilisation in pancreatic β-cells. These included the p53-target Bax, and the pro-apoptotic mitochondrial factors Cycs and Endog, which may indicate replenishment of proteins lost from the mitochondria during apoptotic signalling. In addition, increased expression of the tumour suppressor Cdkn2a (p19 Arf ) was also detected, which may have been involved in stabilization of the p53 tumour suppressor by inhibition of Mdm2, or in promoting cell cycle arrest prior to apoptosis together with up-regulation of the p53-target gene Cdkn1a (p21 Cip1 ). In contrast, SBK showed no change in these genes, but instead a decrease in expression of the proapoptotic Bcl2 family member Pmaip1 (Noxa) and the inhibitor of growth gene Ing3 (p47 Ing3 ). Previous work has linked over-expression of p47 Ing3 with apoptosis [76], and reduced expression with human head and neck squamous carcinomas [77]. SBK may be protected from apoptosis in vivo by the Igf1-Akt survival pathway. Of particular interest was the early induction of Igf1, Akt1 and Akt2 in the SBK, and the tight correlation seen between the three. Expression of these three genes is also seen at later time points in the pancreas, indicating that the Igf1-Akt pathway may be activated in both tissues but is somehow bypassed in the β-cells.
One key difference between the two systems appears to be the presence of members of the Kallikrein serine protease family, which were dramatically up-regulated throughout the time-course for SBK. Kallikrein proteins have been linked to many forms of cancer, and of particular note is their role in the Igf1-Akt survival pathway. Klk1, Klk21, Klk24 and Klk27 have been linked to Igf1regulated tumour survival through degradation of the Igf binding protein Igfbp3 in humans. This prevents Igfbp3 from antagonizing Igf1-Igf1r interactions, allowing Igf1 to bind to its receptor and initiate survival via the Akt pathway [102][103][104]108]. Interestingly, the highest expression in a similar study from Frye et al. using a basal keratinocyte model for MYC activation [90] was for the brain and skin protease gene Bssp1, also known as Kallikrein 6 (Klk6). This gene remained low in WT treated mice, but was significantly increased following MYC expression (129-fold up-regulated). In both models, MYC activation drives vastly increased Kallikrein expression, so it is possible that these proteins play similar roles in cell survival in both systems. Increased expression of Kallikrein genes in SBK following MYC activation may therefore create an environment that favours survival over apoptosis.
In addition to the increased cell proliferation in both tissues, our data indicate a loss of differentiation in both pancreatic β-cells and SBK in response to activation of MYC. In pancreatic β-cells, we found down-regulation of genes that are essential in pancreatic development, such as Pdx1 and Nkx6.1, as well as genes involved in glucosesensing such as Slc2a2 (Glut2) and Gck, both putative MYC-targets. In SBK, many significant changes were detected for genes relating to keratinocyte differentiation that generally pointed to a loss or delay in differentiation -an observation that was previously noted in this mouse model [28]. Interestingly, activation of MYC in the basal layer of the epidermis actually promotes keratinocyte differentiation [93], a mechanism which is thought to prevent epidermal stem cells from becoming cancerous. One possible explanation for the contrasting behaviour in suprabasal epidermis, is that SBKs are post-mitotic and have already entered a terminal differentiation process. Subsequent activation of MYC in early SBK may promote loss of differentiation to enable cell cycle entry. Since MYC-activated SBKs are already migrating upwards towards the skin surface, they are less likely to pose a cancer risk to the host given that they will ultimately be sloughed off as previously shown [28].
We have previously shown that MYC activation in SBK results in a prominent angiogenic phenotype [28]. From our microarray data, potential candidates that may promote such a response include Pgf, a member of the VEGF family, which was highly up-regulated in skin but in fact down-regulated in pancreas. We also found Vegfa up-regulated in skin but not in pancreas. Interestingly Vegfc, which is important in growth of lymphatic vessels, was down-regulated in skin but up-regulated in pancreas. However, given that prominent angiogenesis is not detected in the skin until 3-4 days following MYC activation [28], it is possible that the short time course considered here is too early to identify a transcriptional response in all relevant genes relating to vascularisation. Kallikrein proteins have also been implicated in facilitating angiogenesis via degradation of the cellular matrix [109] and our data showed co-regulation of Pgf and Kallikrein genes. These data suggest that the local tissue microenvironment in SBK that promote angiogenic growth may be linked to survival pathways that protect the cells against apoptosis.
In summary, activation of MYC results in cell growth, loss of differentiation and cell cycle entry in both β-cells and SBK in vivo. Apoptosis, which is confined to β-cells, involves a combination of a DNA-damage response and downstream activation of pro-apoptotic signalling pathways, including Cdc2a and p19 Arf /p53, and downstream targets. Conversely, avoidance of apoptosis in SBK may result primarily from the activation of key anti-apoptotic signalling pathways, particularly those involved in the Igf1-Akt pathway, and induction of an angiogenic response (Pgf, Vegfa and Klks). A contributory role for intrinsic resistance to the induction of DNA-damage and the p19 Arf tumour suppressor pathway by MYC in SBK is also possible. A possible mechanism whereby tissue-specific environmental factors may influence cell fate following MYC deregulation has also been proposed, hypothesising that the decision to live or die may relate to the local tissue-specific microenvironment. However, this remains speculation as the approach taken here gives an insight into only one aspect of the changes occurring within the cell in response to MYC deregulation. Much remains to be learnt from analysis of protein-level changes, post-translational modifications, or epigenetic modifications of DNA. This study has identified several new lines of investigation for future analysis into the dual roles of MYC in apoptosis and survival. It is hoped that such studies will prove fruitful and provide further insight into the complex role of this enigmatic protein.

Tissue Sample Preparation
Transgenic mice expressing switchable pIns-MYC-ER TAM in pancreatic β-cells or inv-MYC-ER TAM in SBK have been previously characterized and described [27,28]. [8][9][10][11][12] week old male mice were treated with 4OHT or vehicle for 4, 8, 16 or 32 hours. Triplicate samples were collected for each time point for each of the four conditions; pIns-MYC-ER TAM 4OHT treated/ MYC-ER TAM active ("Panc T"), pIns-MYC-ER TAM vehicle treated/MYC inactive ("Panc U"), Inv-MYC-ER TAM 4OHT treated/MYC active ("Skin T"), Inv-MYC-ER TAM vehicle treated/MYC inactive ("Skin U"). All mice were housed and treated in accordance with protocols and regulations sanctioned by the Home Office under the Animals Act of 1986.

RNA Isolation and Microarray Hybridization
A modified LCM protocol was devised to preserve RNA integrity. Fresh frozen pancreas sections were cut to a thickness of 15 μm, bound to a MMI MembraneSlide (Molecular Machines and Industries, Rockledge, FL) and fixed in ice-cold 100% ethanol for 2 mins. Sections were stained briefly (10 secs) with a 1% Toluidine Blue dye in 100%. Stained sections were dehydrated in 2 changes of 100% ethanol and 2 changes of xylene for 1 minute each, airdried for 2 minutes and finally left in a vacuum dessicator for 5 minutes before transportation to the laser capture platform. The SL μCut LCM system (Molecular Machines and Industries, Rockledge, FL) was used to isolate islets of Langerhans from surrounding exocrine tissue. RNA was collected using the RNA Microkit protocol (Qiagen, Valencia, CA). The laser capture procedure was repeated on freshly cut pancreas sections to collect a total area of islet cells equal to roughly 1.5 × 10 6 μm 2 for each sample. Due to the thinness of murine suprabasal epidermis, isolation of sufficient good quality RNA for microarray hybridization from LCM of SBK proved difficult, a problem also noted by Agar et al. [110]. RNA was instead collected from 5 fresh frozen skin sections (20 μm) collected across several levels of the tissue. A modified version of the Affymetrix GeneChip 2-cycle target labelling in vitro transcription (IVT) protocol was used, incorporating double volumes of polyA controls and reagents in the first round cRNA synthesis stage to increase the yield. 10 μg double-amplified biotin-labelled cRNA were hybridized to Affymetrix MOE430 2.0 GeneChips (Affymetrix, Santa Clara, CA) as described in the Affymetrix GeneChip Expression Analysis Technical Manual.

Data Analysis
Data were normalized across all samples at the probelevel using GC-RMA [111]. Analysis of gene expression data was performed using the Bioconductor microarray analysis packages in R [112] and GeneSpring GX 7.3.1 (Agilent, Santa Clara, CA). 4OHT-treated samples were normalized to their respective vehicle-treated counterparts to ensure that normalized values corresponded to the fold-change in gene expression due to activation of MYC-ER TAM . Removal of control and non-responsive probes identified 12,349 probe sets. The custom R package Envisage was used to identify probe-sets showing significant differential expression after MYC-ER TAM activation, across time, and between pancreas and skin tissue. p-values were calculated for each model term and their interactions, and were corrected for multiple testing [113]. Only the highest-order interaction p-value was considered for genes where the selected model contains multiple terms relating to the MYC-ER TAM activation state. Contrast p-values were calculated for each condition by applying an unpaired t-test comparing 4OHT-treated samples with vehicle-treated samples within each of the 8 groups (4 hrs, 8 hrs, 16 hrs and 32 hrs for skin or pancreas). Significant early-changing (≥ 2-fold within 8 hours) probe sets were compared with known gene-ontologies using the DAVID functional annotation tool [29]. Quality threshold (QT) partitional clustering [114] was used to identify genes showing similar expression profiles, using the Pearson cross-correlation coefficient with a minimum correlation of 0.9 and a minimum cluster size of 14.
Quantitative Real-Time reverse transcriptase PCR TaqMan qRT-PCR was performed on original total RNA samples for genes of interest. 20 ng total RNA was reverse transcribed to cDNA using a high-capacity cDNA reverse transcription kit (Applied Biosystems, Foster City, CA). cDNA transcripts were pre-amplified prior to the qRT-PCR reaction in a multiplexed reaction using TaqMan preAmp mastermix (Applied Biosystems, Foster City, CA), with pooled TaqMan qRT-PCR assays (Applied Biosystems, Foster City, CA) at a concentration of 0.2X in 1X TE buffer. qRT-PCR was performed using an ABI Prism 7000 scanner (Applied Biosystems, Foster City, CA), with an 18s rRNA endogenous control probe (Applied Biosystems, Foster City, CA). qRT-PCR was performed for skin and pancreas 4OHT-and vehicletreated RNA samples for early time-points 4 hrs and 8 hrs, and for the later 32 hrs time-point. As with microarray analysis, quantitative measures of gene expression upon MYC-activation were calculated by comparing vehicle-and 4OHT-treated samples directly for each condition.