Article Text

PDF

Original article
c-Myc downregulation is required for preacinar to acinar maturation and pancreatic homeostasis
  1. Victor J Sánchez-Arévalo Lobo1,
  2. Luis César Fernández1,
  3. Enrique Carrillo-de-Santa-Pau1,
  4. Laia Richart1,
  5. Isidoro Cobo1,
  6. Jaroslaw Cendrowski1,
  7. Ulisses Moreno1,
  8. Natalia del Pozo1,
  9. Diego Megías2,
  10. Bernardette Bréant3,
  11. Christopher V Wright4,
  12. Mark Magnuson4,
  13. Francisco X Real1,5
  1. 1Epithelial Carcinogenesis Group, Cancer Cell Biology Programme, Spanish National Cancer Research Center-CNIO, Madrid, Spain
  2. 2Confocal Microscopy Unit, Biotechnology Programme, Spanish National Cancer Research Center-CNIO, Madrid, Spain
  3. 3Centre de Recherche des Cordeliers, INSERM, Paris, France
  4. 4Department of Cell & Developmental Biology, Vanderbilt Center for Stem Cell Biology, Vanderbilt University, Nashville, Tennessee, USA
  5. 5Departament de Ciències Experimentals i de la Salut, Universitat Pompeu Fabra, Barcelona, Spain
  1. Correspondence to Dr Francisco X Real or Dr Víctor J Sánchez-Arévalo Lobo, Epithelial Carcinogenesis Group, Cancer Cell Biology Programme, Spanish National Cancer Research Center-CNIO, Melchor Fernandez Almagro 3, Madrid 28029, Spain; freal{at}cnio.es or vjsanchez{at}cnio.es

Abstract

Background and aims c-Myc is highly expressed in pancreatic multipotent progenitor cells (MPC) and in pancreatic cancer. The transition from MPC to unipotent acinar progenitors is associated with c-Myc downregulation; a role for c-Myc in this process, and its possible relationship to a role in cancer, has not been established.

Design Using coimmunoprecipitation assays, we demonstrate that c-Myc and Ptf1a interact. Using reverse transcriptase qPCR, western blot and immunofluorescence, we show the erosion of the acinar programme. To analyse the genomic distribution of c-Myc and Ptf1a and the global transcriptomic profile, we used ChIP-seq and RNA-seq, respectively; validation was performed with ChIP-qPCR and RT-qPCR. Lineage-tracing experiments were used to follow the effect of c-Myc overexpression in preacinar cells on acinar differentiation.

Results c-Myc binds and represses the transcriptional activity of Ptf1a. c-Myc overexpression in preacinar cells leads to a massive erosion of differentiation. In adult Ela1-Myc mice: (1) c-Myc binds to Ptf1a, and Tcf3 is downregulated; (2) Ptf1a and c-Myc display partially overlapping chromatin occupancy but do not bind the same E-boxes; (3) at the proximal promoter of genes coding for digestive enzymes, we find reduced PTF1 binding and increased levels of repressive chromatin marks and PRC2 complex components. Lineage tracing of committed acinar precursors reveals that c-Myc overexpression does not restore multipotency but allows the persistence of a preacinar-like cell population. In addition, mutant KRas can lead to c-Myc overexpression and acinar dysregulation.

Conclusions c-Myc repression during development is crucial for the maturation of preacinar cells, and c-Myc overexpression can contribute to pancreatic carcinogenesis through the induction of a dedifferentiated state.

  • PANCREATIC CANCER
  • PANCREATIC DISEASE
  • PANCREATIC ENZYMES
  • ONCOGENES

Statistics from Altmetric.com

Significance of this study

What is already known on this subject?

  • In mouse embryonic pancreas, multipotent progenitors express high levels of c-Myc.

  • c-Myc is a major human oncogene.

  • c-Myc is amplified and/or overexpressed in a large fraction of pancreatic ductal adenocarcinomas (PDAC).

What are the new findings?

  • c-Myc interacts with Ptf1a, the master regulator of acinar differentiation.

  • During pancreatic development, c-Myc downregulation is required for the full maturation of acinar cells.

  • In mice, c-Myc overexpression leads to repression of the acinar programme, upregulated expression of ductal markers and ectopic activation of a liver differentiation programme.

  • These effects are associated with the acquisition of repressive chromatin marks at the promoter of acinar genes and with PRC2 complex upregulation.

  • Expression of mutant KRas in vitro and in vivo leads to c-Myc overexpression and acinar dysregulation.

How might it impact on clinical practice in the foreseeable future?

  • c-Myc is one of the most important human oncogenes, and our findings provide clues to better understand its role in the early stages of PDAC development.

  • There is increasing evidence that c-Myc is druggable, and, recently, new strategies to block its oncogenic activity have been identified.

Introduction

Pancreatic development and differentiation involve sequential and orderly changes in the activity of transcription factors that selectively control the maturation of specific cell types.1 Endoderm-derived multipotent progenitor cells (MPC) giving rise to all epithelial pancreatic cell types are present from E9.5 to E12.5 and are defined as Ptf1a+, Mychigh, Cpa+.2 Lineage tracing has shown that, after E14.5, MPC are absent, c-Myc is repressed and two types of committed progenitors arise: bipotent progenitors giving rise to endocrine and ductal cells, and unipotent acinar progenitors. Their emergence is established through the activation of alternative differentiation programmes relying on the cross-regulation of Ptf1a and Nkx6.2 ,3

The acquisition of acinar identity relies on high expression of Ptf1a, a tissue-specific basic helix-loop-helix (bHLH) transcription factor. Ptf1a forms heterotrimers with a class B ubiquitous bHLH protein (Tcf3 or Tcf12) and with a third component, Rbpj or Rbpjl, to generate the transcriptionally active pancreas transcription factor 1 (PTF1) complex.4 Early during acinar differentiation, an embryonic PTF1 complex containing Rbpj (PTF1–J) autoinduces Ptf1a and activates Rbpjl. High-level Rbpjl expression leads to the replacement of Rbpj to generate the PTF1–L complex, essential for acinar maturation.5 ,6 Additional transcriptional regulators are required to complete acinar differentiation and sustain this programme during adulthood.7–11

c-Myc plays a broad role in the regulation of cell differentiation, including reprogramming.12–16 Expression of high c-Myc levels in MPC has been known for several years, but the role for c-Myc in these cells, and in the pancreas, is incompletely elucidated. Conditional pancreatic c-Myc inactivation at E10.5 leads to pancreatic hypoplasia at birth and to progressive acinar-to-adipocyte transdifferentiation in mice 2–10 months old of age,17 indicating that c-Myc is essential for the expansion and maintenance of acinar cells. Apc inactivation during development leads to pancreatomegaly in a c-Myc-dependent manner.18 The mechanisms through which c-Myc participates in these processes are poorly understood. To bind DNA, c-Myc heterodimerises with the ubiquitous protein Max. The recruitment of the Myc–Max complex to target genes depends on E-box (CACGTC) recognition, protein–protein interactions and the epigenetic context.19 ,20

We identified Myc proteins as Ptf1a partners through a yeast two-hybrid screen, leading to hypothesise that both proteins might act coordinately to regulate pancreatic differentiation during development and play a role in cancer. Here, we analyse the role of c-Myc in acinar differentiation and homeostasis. We show that c-Myc downregulation is required for preacinar to acinar maturation and that mutant KRas upregulates c-Myc expression in acinar cells. Our study sheds light on the mechanisms contributing to the oncogenic role of c-Myc in the pancreas.

Results

c-Myc binds and represses Ptf1a transcriptional activity

To identify new Ptf1a interactors that might modulate its transcriptional activity, we used four yeast two-hybrid baits coding for proteins comprising different Ptf1a regions (N-terminus and C-terminus, with or without the bHLH domain).21 Among the interacting cDNAs identified were Rbpj, Tcf3/E47 and Tcf12/HeLa E-box binding factor (HEB), known to bind Ptf1a, thus confirming the efficacy of the strategy. Using the C-terminus of Ptf1a lacking the bHLH domain as bait, and the E9.5 embryo library, we identified one clone corresponding to Bmyc (8NP_075815), coding for a 20 kDa protein displaying high similarity to the N-terminal Myc-boxes of c-Myc but lacking a bHLH domain. Bmyc has been proposed to function as a dominant negative protein as it has antiproliferative activity22 ,23 and is upregulated in adult mice.17 The direct interaction of Ptf1a with murine Bmyc was validated using pull-down experiments (not shown).

We subsequently focused on c-Myc because there is no human orthologue of Bmyc, and c-Myc is coexpressed with Ptf1a in MPC. Haemagglutinin tag-Pancreas transcription factor 1 alpha (HA–Ptf1a) and Flag–Myc were found in the same complex in lysates of transfected HEK293 cells immunoprecipitated with anti-Flag followed by immunoblotting with anti-HA; the results were confirmed by reverse coimmunoprecipitation (figure 1A and online supplementary figure S1A). To determine whether Max was present in the same complex, lysates of HEK293 transfected with Flag–Myc and HA–Ptf1a plasmids were immunoprecipitated with anti-Max antibodies: Flag–Myc- but not Ptf1a -was coimmunoprecipitated with Max (figure 1B), indicating that Max and Ptf1a are not present in the same complex. In PTF1, Ptf1a binds Rbpjl through its C-terminus, and Tcf3 or Tcf12 through its bHLH domain. We hypothesised that c-Myc might compete with the binding of Tcf3 to Ptf1a. In HEK293 transfected with plasmids coding for green fluorescent protein (GFP)–Tcf3, Flag–Ptf1a and HA–Myc, Tcf3 outcompeted c-Myc (figure 1C). By contrast, in cells transfected with plasmids coding for Flag–Ptf1a, HA–Myc and Rbpjl, c-Myc did not displace Rbpjl from PTF1 (figure 1D).

Figure 1

c-Myc interacts with Ptf1a and represses its transcriptional activity in vitro. HEK293T cells were transiently cotransfected with plasmids coding for Flag–Myc and HA–Ptf1a, alone or in combination. (A) Proteins immunoprecipitated (IP) from whole cell lysates (total extract) with anti-Flag beads were analysed by immunoblotting with the indicated antibodies. (B) IP proteins from whole cell lysates with anti-Max antibodies were analysed by immunoblotting with the indicated antibodies. (C) IP proteins from whole cell lysates with anti-Flag beads were analysed by immunoblotting with the indicated antibodies. (D) IP proteins from whole cell lysates with anti-Flag beads were analysed by immunoblotting with the indicated antibodies. (E) HEK293 cells were transfected with the 6xA-Luc reporter plasmid alone or together with expression plasmids coding for Flag–Myc, HA–Ptf1a or Flag–Rbpjl, as indicated. The transcriptional activity of the PTF1 complex was assayed measuring the activity of the luciferase gene under the control of a 6-mer of the A element.5 Renilla (pRL-TK) was cotransfected to allow for normalisation of transfection efficiency. (F) Nucleolin promoter–reporter assay shows no change in activity upon c-Myc and Ptf1a overexpression. The histograms represent the means±SEM of three independent experiments. p value was determined using an unpaired Mann-Whitney test.

c-Myc can interact and negatively modulate the activity of several transcription factors.24–26 In transient transfection assays with HEK293 cells, c-Myc significantly reduced the transcriptional activity of a luciferase reporter construct driven by PTF1-responsive sequences (6xA26, a 6-mer of the A element of PTF1 target genes)27 ,28 (figure 1E). PTF1 acquires highest transcriptional activity through the recruitment of the acetyltransferase P300/CBP-associated factor (P/CAF),29 and c-Myc overexpression also reduced the transcriptional activity driven by Ptf1a-P/CAF (see online supplementary figure S1B). By contrast, the activity of a promoter–reporter construct of nucleolin was not affected by Ptf1a overexpression (figure 1F).

c-Myc downregulates the acinar programme in adult mice

To determine the effects of persistent high c-Myc levels in vivo, we used Ela1-Myc mice expressing c-Myc under the control of the elastase-1 promoter.30 Ela1-Myc transgenic mice have been used to study pancreatic carcinogenesis.30–32 In these mice, c-Myc is expressed after E14.5, when elastase expression is upregulated during acinar differentiation. Starting at day 1 after birth (P1) until 8 weeks, prior to tumour development, Ela1-Myc mice display progressive acinar dilation, acinoductal metaplasia (ADM) and acinar dysplasia. By 5 months, all mice develop cancer with mixed areas of acinar and ductal differentiation.30 ,33 Expression of the c-Myc transgene led to a threefold upregulation of total c-Myc mRNA at all stages analysed with reduced levels of endogenous c-Myc mRNA likely resulting from a negative feedback regulation34 (figure 2A). Persistent c-Myc expression in acinar cells severely compromised their postnatal acinar maturation: at P1, a modest reduction in the levels of Ela1, Amy2a and Ctrb1 transcripts was observed, whereas carboxypeptidase 1 (Cpa1) transcripts were unaffected (figure 2B and online supplementary figure S2). Cpa1 is one of the earliest exocrine genes activated during pancreatic development/differentiation and a marker of acinar precursors, suggesting that the early acinar programme is not affected at this stage.2 ,35 At P11 and 8 weeks, all digestive enzyme transcripts and proteins were markedly reduced (figure 2B and online supplementary figure S2), indicating a severe disruption of the acinar programme. To determine the molecular mechanisms involved, we analysed the expression of transcripts coding for the PTF1 components. Ptf1a and Rbpjl mRNAs were expressed at similar levels in Ela1-Myc mice and their wild-type (WT) counterparts at P1, but they were subsequently downregulated, as were the PTF1 targets (figure 2B). Total levels of c-Myc mRNA were not significantly changed from P1 to 8 weeks, despite the continuous decrease in the levels of the endogenous transcript (figure 2A), likely due to the low but persistent activity of PTF1 driving the elastase transgene.

Figure 2

The acinar genetic programme is progressively repressed in Ela1-Myc mice. RT-qPCR analysis of gene expression in the pancreas of mice of the indicated genotypes and age; transcript levels were normalised against Hprt. (A) Expression of total (left) and endogenous (right) c-Myc in wild-type (WT) (n=3) and Ela1-Myc (n=3) mice at P1, P11 and 8 weeks of age. Fold change taking WT mice as reference. Data are expressed as the mean±SEM. p value was determined using an unpaired Mann-Whitney test. (B) Expression of transcripts of digestive enzymes (left) or PTF1 components (right) in WT (n=3) and Ela1-Myc (n=3) mice at P1, P11 and 8 weeks of age. Data are expressed as the mean±SEM. p value was determined using an unpaired Mann-Whitney test.

In 8-week-old Ela1-Myc pancreata, Tcf3 was downregulated and c-Myc coimmunoprecipitated with Ptf1a (figure 3A), supporting that the two proteins interact in vivo; decreased expression of digestive enzymes was confirmed by western blotting (figure 3B). Concomitantly, interaction of endogenous Tcf3 and Ptf1a was reduced, as shown by coimmunoprecipitation (figure 3A). Accordingly, lower Ctrb1 expression was observed by immunofluorescence in P1 Ela1-Myc pancreata (figure 3C). We conclude that in Ela1-Myc mice, c-Myc binds to Ptf1a and represses the PTF1-driven programme due to elevated c-Myc expression and Tcf3 downregulation.

Figure 3

Reduced postnatal expression of digestive enzymes in the pancreas of Ela1-Myc mice. (A) Ptf1a immunoprecipitation followed by western blot with the indicated antibodies. Ten per cent of the lysate was used as total extract and blotted with specific antibodies. (B) Total extracts of pancreata from 8-week-old Ela1-Myc and wild-type (WT) mice were blotted with the indicated antibodies. (C) Representative images of c-Myc, Ptf1a and Ctr1b expression in P1 WT and Ela1-Myc pancreata.

Genome-wide effects of persistent c-Myc overexpression

To assess the genome-wide effects of persistent c-Myc expression, we analysed RNA from 8-week-old Ela1-Myc pancreata using RNA-seq. At this age, focal dysplastic lesions—but no tumours—are observed. In Ela1-Myc pancreata, a global transcriptome amplification is accompanied by the overactivation of a c-Myc signature36 and the downregulation of an acinar signature containing multiple digestive enzymes, such as Cela2a and Cpa14 (Gene Set Enrichment Analysis (GSEA), false discovery rate (FDR)<0.0001) (see online supplementary figure S3A–D). We validated the downregulation of eight genes included in this signature by RT-qPCR (see online supplementary figure S3E).

Because Ptf1a and c-Myc interact in vitro and in vivo, we analysed their genome-wide distribution in 8-week-old Ela1-Myc pancreata using chromatin immunoprecipitation followed by massive parallel sequencing (ChIP-Seq). Peak detection analysis using model-based analysis of ChIP-Seq (MACS) defined 5464 c-Myc-enriched and 17632 Ptf1a-enriched regions. A large proportion of c-Myc-binding regions (2947 peaks, 53.95%) overlapped with Ptf1a-binding regions (figure 4A). Genome sequences with common peaks were preferentially located at ±1 kb from the closest transcription start site (TSS) (figure 4B). The density profiles of c-Myc and Ptf1a peaks showed a large degree of overlap (figure 4C) in association with global transcriptional amplification in Ela1-Myc pancreata (figure 4E). Motif analysis of peaks common to c-Myc and Ptf1a showed an enrichment of E-boxes recognised by bHLH proteins (c-Myc and Tcf3/Tcf12) (p=4.4×10−286). Moreover, there was an enrichment of the TC-box characteristic of PTF1 (p=1.5×10−42) (figure 4D). Access to full motif analyses is provided in online supplementary material. Motif spacing analysis of the regions containing both sequences showed TC-boxes preferentially locating close to E-boxes (between 5 and 16 bp), as described for PTF1 targets (see online supplementary figure S4A). We then interrogated whether c-Myc and Ptf1a colocalise in the promoter (TSS+1 kb) of acinar-related genes described by Masui et al.4 In 12 out of 23 genes, both proteins bound to the same regions at repressed genes (figure 4F).

Figure 4

Genome-wide distribution of Ptf1a and c-Myc in the pancreas of 8-week-old Ela1-Myc mice. (A) Overlap of the binding sites identified by ChIP-seq using antibodies against c-Myc and Ptf1a. (B) Genomic distribution of common peaks between c-Myc and Ptf1a relative to the gene body of Refseq-annotated transcripts (left). Histogram with the distance of the peak summit to the transcription start site (TSS) (right). (C) Density plot of c-Myc and Ptf1a ChIP-seq data on binding site loci with a 5 kb region centred on the peak summits. c-Myc-exclusive peaks (left); common peaks (middle); Ptf1a-exclusive peaks (right). (D) Multiple em for motif elicitation (MEME) motif prediction of DNA sequences enriched in the common peaks from c-Myc and Ptf1a ChIP-seq. (E) Heatmap showing the expression of the complete set of genes ascribed to common or exclusive peaks for c-Myc, Ptf1a. (F) Enrichment analysis for acinar genes (set described by Masui et al) for gene promoters with c-Myc and Ptf1a overlapping peaks (TSS±1 kb) and expression heatmap of genes with overlapping peaks.

To determine whether c-Myc and Ptf1a bind to well-established PTF1 regulatory sequences, we performed ChIP-qPCR on the promoter of Cpa1 and Cela2a, two bona fide PTF1 targets (figure 5 and online supplementary figure S5); Cpa1, nucleolin (Ncl) and the acetylcholine receptor (AchR) were used as controls (see online supplementary figure S5A and B). The ChIP-Seq analysis revealed overlapping binding of Ptf1a and c-Myc in the proximal promoter of Cpa1; the corresponding regions contained three PTF1-binding motifs including an E-box and a TC-box at positions −140 (E-box site 1), −589 (E-box site 2) and −1104 (E-box site 3) with respect to the TSS (figure 5B). Of note, site 1 corresponds to the well-characterised PTF1-binding site in the promoter of digestive enzyme genes.28 In WT pancreata, Ptf1a bound strongly to sites 1 and 2 and—to a lesser extent—to site 3 using ChIP-qPCR (figure 5C, upper panel); c-Myc was undetectable at these sites due to its low expression levels. In Ela1-Myc mice, binding of Ptf1a to site 1 was reduced, and no significant enrichment could be detected at sites 2 and 3. By contrast, c-Myc was found at sites 2 and 3 but not at the canonical site 1. Max was also found at sites 2 and 3, indicating that binding to these regions involves the Myc–Max complex (figure 5C, lower panel). The analysis of the promoter of Cela2a yielded similar results: Ptf1a and c-Myc bound to 3 regions 5′ from the TSS (see online supplementary figure S5C). We further confirmed binding to the putative Ptf1a-binding sites in the two distal sites (−1697 and −2085) using ChIP-qPCR. Ptf1a—but not c-Myc—bound to both regions in WT mice. By contrast, in Ela1-Myc mice, significant enrichment was observed for c-Myc and Max but not for Ptf1a.

Figure 5

c-Myc invades E-boxes in the proximal promoter of Cpa1 in 8-week-old Ela1-Myc mice. (A) UCSC browser screen capture of the c-Myc and Ptf1a ChIP-seq and RNA-seq results obtained at Cpa1 locus in Ela1-Myc mice. (B) Control ChIP-qPCR experiments show weak Myc–Max binding to the promoter of the established Myc target gene nucleolin in wild-type (WT) mice and increased binding in Ela1-Myc mice. (C) Ptf1a binding to the E-box-containing regions in the promoter of Cpa1 in WT and Ela1-Myc mice (n=5 mice/group) (upper). c-Myc and Max binding to the E-box-containing regions in the promoter of Cpa1 in WT and Ela1-Myc mice (lower). ChIP-qPCR values are expressed as median of percentage chromatin immunoprecipitated, normalised to the isotype-matched control. p value was determined using an unpaired Mann-Whitney test.

Overall, we do not find evidence for colocalisation of Ptf1a and c-Myc in the same E-boxes; instead, the Myc–Max complex invades PTF1-binding sites on the tested target genes.

Late c-Myc repression of the acinar programme and PRC2 complex activation

c-Myc can also regulate gene expression indirectly, through the induction of the PRC2 complex that represses gene expression through the deposition of the H3K27me3 mark.37 Immunostaining of H3K27me3 in P1, P11 and 8-week-old pancreata revealed a strong and sustained expression only in Ela1-Myc mice (figure 6A). RT-qPCR expression analysis of PRC2 complex components (Ezh2, Suz12 and Eed) showed elevated levels exclusively in 8-week-old Ela1-Myc pancreata (figure 6B). We detected c-Myc binding to the promoter of PRC2 members, supporting a direct regulation by c-Myc (figure 6C). These changes were associated with increased H3K27me3 at the promoter of Cela2a and Cpa1 (figure 6D). These results strongly suggest that the repression of the acinar programme in adult mice caused by c-Myc results from increased PRC2 activity at the promoter of target genes.

Figure 6

Increased levels of H3K27me3 at acinar gene promoters is associated with c-Myc-mediated regulation of PRC2 components in adult Ela1-Myc mice. (A) Immunostaining of pancreas from P1, P11 and 8-week-old wild-type (WT) and Ela1-Myc mice with anti-H3K27me3. (B) Expression of PRC2 components in WT (n=3) and Ela1-Myc (n=3) mice at P1, P11 and 8 weeks of age, determined by RT-qPCR. (C) c-Myc binding to the E-box-containing regions in the promoter of Ezh2, Suz12 and Eed in WT and Ela1-Myc mice (n=5 mice/group). ChIP-qPCR values are expressed as median of percentage of chromatin immunoprecipitated, normalised to isotype-matched control IgG. (D) H3K27me3 enrichment on the promoter of Cela2a and Cpa1 in WT and Ela1-Myc mice (n=5 mice/group). ChIP-qPCR values are expressed as median of percentage of chromatin immunoprecipitated, normalised to the isotype-matched control IgG and total histone H3. Data are expressed as the mean±SEM. p value was determined using an unpaired Mann-Whitney test.

High c-Myc levels block preacinar cell maturation

We then assessed the effects of persistent c-Myc expression on the fate of preacinar cells. ADM is a common process resulting from altered acinar differentiation.38 In agreement with the described erosion of the acinar programme, the mRNAs of the ductal markers Krt7 and Krt19 were overexpressed in Ela1-Myc pancreata (figure 7A). Ectopic activation of a liver signature was also detected in Ela1-Myc pancreata, as reported for Rbpjl-null pancreata4 (figure 7B).

Figure 7

c-Myc downregulation is required for full acinar maturation. (A) Expression of ductal cell keratin mRNA using transcriptome sequencing (fragments per kilobase of exon per million (FKPM) values) in wild-type (WT) and Ela1-Myc mice (n=3). (B) Gene Set Enrichment Analysis of a gene set defined in Rbpjl pancreas-null mice comparing Ela1-Myc versus WT mice.4 (C) Lack of coexpression of GPF, Sox9, Nkx6.1 in control (WT, Ptf1aCreERT2;Rosa26YFP) and Ela1-Myc (Ela1-Myc;Ptf1aCreERT2;Rosa26YFP) mice upon Tamoxifen (TMX) treatment at E15.5 and lineage tracing analysis at E18.5. Arrowheads indicate GFP+ cells; arrows indicate Ptf1a+ cells; bold arrows indicate double-positive cells (GFP+/Ptf1a+). (D) Ptf1a and Gata6 expression in E18.5 WT and Ela1-Myc pancreata; the corresponding quantifications of Ptf1a fluorescence are shown in the right panel. Merged images shown in online supplementary figure S6.

The fate of c-Myc overexpressing acinar progenitors was analysed by lineage tracing. Ela1-Myc;Ptf1aCreERT2;Rosa26YFP and control Ptf1aCreERT2;Rosa26YFP mice received tamoxifen at E15.5—when preacinar cells are committed to become acinar cells; pancreata were analysed at E18.5. In control mice, only acinar cells were GFP+ (figure 7C and online supplementary table S1). Similarly, in Ela1-Myc;Ptf1aCreERT2;Rosa26YFP mice, acinar cells expressed GFP; we did not find coexpression of GFP with Sox9 or Nkx6.1—markers of ductal and endocrine cells, respectively (see online supplementary table S1). Fluorescence quantification analysis showed reduced expression of Ptf1a in Ela1-Myc;Ptf1aCreERT2;Rosa26YFP E18.5 acinar precursors compared with WT littermates: GFP-labelled cells from Ela1-Myc mice were enriched in Ptf1alow cells (not shown). In addition, reduced expression of Ptf1a and Gata6 was demonstrated in acinar precursors at E18.5 (figure 7D and online supplementary figure S6): 96.2% and 89.7% of Ptf1ahigh and Ptf1alow cells, respectively, showed low Gata6 expression, indicating incomplete differentiation.

Therefore, persistent c-Myc overexpression leads to a blockade in acinar cell maturation accompanied by liver mis-specification.

KRas activation leads to c-Myc overexpression and acinar dysregulation

To determine whether the effects of mutant KRas in pancreatic cells might be mediated by c-Myc, we expressed G12V-KRas in 266.6 transformed acinar cells through retroviral transduction. KRas overexpression led to a modest increase of c-Myc (figure 8A, B)—similar to the one detected in vivo in the pancreas—and to reduced expression of Ptf1a and of multiple digestive enzyme transcripts (figure 8C). This was accompanied by a marked upregulation of Krt19 and Krt7 transcripts (figure 8D), suggesting the activation of ADM. In addition, c-Myc was consistently overexpressed in mutant KRas-driven PanINs and pancreatic ductal adenocarcinomas (PDAC) (figure 8E), supporting the notion that c-Myc induced by active KRas may contribute to pancreatic carcinogenesis through dysregulated differentiation and facilitation of ADM.

Figure 8

Mutant KRas regulates c-Myc expression and differentiation. (A–D) Retroviral expression of mutant KRas in 266.6 acinar cells leads to reduced Ptf1a expression at both protein (A) and mRNA (B) levels. This is accompanied by reduced expression of digestive enzyme transcripts (C) and upregulation of ductal keratin markers (D), using RT-qPCR. (E) c-Myc expression is upregulated in the pancreas of Ptf1aCre; KRasLSL-G12V mice, as detected using immunohistochemistry. (a) normal pancreas; (b) low-grade PanIN lesions; (c) pancreatic ductal adenocarcinoma; (d) lesional pancreas of an Ela1-Myc mouse.

Discussion

We show that c-Myc downregulation is required for complete differentiation of pancreatic acinar cells. c-Myc misexpression leads to erosion of the acinar programme, activation of ectopic differentiation programmes and pancreatic tumours. Our findings provide support to the connection between altered pancreatic differentiation and carcinogenesis and provide clues to an improved understanding of the mechanisms through which mutant KRas initiates pancreatic cancer.

Our finding that Ptf1a and Myc proteins interact suggested a contribution to the regulation of acinar differentiation. In vitro, we confirmed that c-Myc ectopic overexpression led to reduced PTF1 activity that is exquisitely required for acinar differentiation. To explore a role in vivo, we performed gain-of-function experiments using Ela1-Myc mice in which sustained c-Myc expression is achieved starting at the time of initiation of acinar differentiation. In the absence of c-Myc downregulation, postnatal acinar maturation is impaired, and a dramatic erosion of the acinar programme precedes tumour development.

In Ela1-Myc mice, multiple mechanisms contribute to the blockade of acinar differentiation. The genome-wide expression analysis of adult pancreata revealed that, upon sustained c-Myc expression, the acinar programme was repressed while ductal and hepatocyte markers were upregulated. These findings are reminiscent of the phenotype of mice in which Rbpjl is inactivated in the embryonic pancreas.4 Persistent c-Myc expression led to reduced PTF1 transcriptional activity. Newborn Ela1-Myc mice displayed a downregulation of a subset of digestive enzymes, mainly those expressed late during acinar maturation, without major changes in the expression levels of PTF1 components. However, at later time points, we demonstrate a dramatic blockade of the acinar programme, concomitant with reduced expression of both pancreas-specific (Ptf1a and Rbpjl) and ubiquitous (Tcf3 and Tcf12) PTF1 components.

In adult Ela1-Myc mice, these changes allow c-Myc–Ptf1a complex formation and downregulation of all digestive enzymes. The c-Myc–Ptf1a complex was not found at bona fide PTF1-binding sites, suggesting a mechanism of sequestration independent of chromatin binding, leading to reduced recruitment of Ptf1a to chromatin. Concomitantly, we found c-Myc–Max invasion of putative Ptf1a-binding sites at promoters of acinar genes.4 These findings are in agreement with the reports of invasion of active promoters and distal elements by c-Myc when expressed at high levels.39 In this context, c-Myc binds low-affinity binding sites, non-canonical sequences and enhancers. These effects occur concomitant with the global transcriptional amplification observed in Ela1-Myc mice, as described by Lin et al.40 The location of c-Myc–Max at these genomic sites in proximal promoters leads to the recruitment of histone modifiers, including the PRC2 complex, resulting in gene repression. Hdac3 or Dmnt3a could play additional roles in this process.41 ,42

The loss of acinar differentiation is often associated with activation of ADM,38 thought to contribute to the development of pancreatic cancer.33 The incomplete acinar differentiation at late embryonic stages raised the question of whether high c-Myc expression provides a permissive genomic context allowing the acquisition of alternative differentiation potentials. Using lineage tracing, we failed to find evidence that preacinar cells become endocrine/ductal precursors. Persistent overexpression of c-Myc starting at earlier embryonic stages might impact on allocation to this lineage.

Our finding that c-Myc overexpression in acinar cells leads to a blockade in acinar differentiation may be relevant to understand the molecular pathogenesis of PDAC. Genetic mouse models have shown that expression of mutant KRas in mouse acinar cells can lead to bona fide ductal tumours43 ,44 and that embryonic preacinar cells—but not adult acinar cells—are sensitive to the oncogenic effects of mutant KRas,43 ,44 underlining the importance of a detailed understanding of acinar maturation to identify the early mechanisms involved in tumour development. In addition, incomplete acinar differentiation resulting from genetic inactivation of Gata6 and Ptf1a in the pancreas cooperates with mutant KRas in PDAC development.45 ,46 KRas-activating mutations lead to EGFR–c-Myc cross-regulation and increased c-Myc expression47–49 to levels similar to those of Ela1-Myc mice, and c-Myc is required for mutant KRas-driven tumours in the pancreas.50 c-Myc can also participate at later stages of PDAC progression, mainly through gene amplification leading to its overexpression.51 ,52

Altogether, we provide evidence that c-Myc downregulation is crucially required for the maturation of preacinar to acinar cells and that c-Myc upregulation can thus contribute to pancreatic tumour development.

Material and methods

Mice

Ela1-Myc mice30 were provided by E. Sandgren (U. Wisconsin). WT and Ela1-Myc mice were bred in a mixed B6 background and maintained under sterile and pathogen-free conditions. Unless otherwise indicated, experiments were conducted with P1, P11 and 8-week-old mice. Procedures were approved by the Animal Ethical Committee of Instituto de Salud Carlos III and performed following Guidelines for Ethical Conduct in the Care and Use of Animals as stated in the International Guiding Principles for Biomedical Research Involving Animals, developed by the Council for International Organisations of Medical Sciences.

Reagents

Details on plasmids and antibodies are provided as online supplementary information.

Cell culture, reporter assays, immunoprecipitation and western blotting

See online supplementary information.

Immunofluorescence analysis

See online supplementary information.

ChIP-Seq

DNA (20 ng) was quantified, resolved by electrophoresis, and fractions of 50–250 bp were extracted. Input samples correspond to balanced blends of inputs from selected samples. Fractions were processed through end-repair, dA-tailing and ligation to adapters following Illumina's ‘TruSeq DNA Sample Preparation Guide’ (part # 15005180 Rev. C). Adapter-ligated libraries were amplified by limited-cycle PCR with Illumina paired end (PE) primers (12 cycles). The resulting purified DNA library was applied to an Illumina flow cell for cluster generation (TruSeq cluster generation kit V.5) and sequenced on the Genome Analyzer IIx with sequencing by synthesis (SBS) TruSeq V.5 reagents following manufacturer's protocols.

RNA-Seq

Total RNA (1 µg) was spiked with external RNA controls consortium dUTP-deoxy Uridine TriPhosphate (ERCC) ExFold RNA spike-In mixes (Life Technologies). Quality was assessed on an Agilent 2100 Bioanalyzer; samples with a RNA integrity number >8.5 were used. PolyA+ fractions were purified, randomly fragmented, converted to double stranded cDNA, and processed through end-repair, dA-tailing and adapter ligation following Illumina’s ‘TruSeq Stranded mRNA Sample Preparation Part # 15031047 Rev. D’ (this kit incorporates dUTP during second strand cDNA synthesis, which implies that only the cDNA strand generated during first strand synthesis is eventually sequenced). Adapter-ligated libraries were generated by PCR with Illumina PE primers (eight cycles). The resulting cDNA libraries were applied to an Illumina flow cell for cluster generation (TruSeq cluster generation kit V.5) and sequenced.

Bioinformatics analyses

ChIP-Seq and RNA-Seq data processing, peak annotation and density plot analyses, motif enrichment analyses and GSEA are described in detail in the online supplementary information file.

Statistical analyses

Data are provided as mean±SEM. Comparison of data that did not follow a normal distribution was performed using Mann-Whitney test. Significance was considered for p<0.05.

Acknowledgments

The authors thank the investigators cited in the text for providing reagents; ML Campos for contributions to the yeast two-hybrid assay; Y Cecilia for animal care; the core facilities of CNIO for support; members of the Epithelial Carcinogenesis Group and R MacDonald for valuable discussions; and P Martinelli, I Moreno de Alborán and A Sabó for comments to a previous version of the manuscript.

References

View Abstract

Footnotes

  • Twitter Follow Luis Fernández @luiscefd, Ulisses Moreno @Celismoreno and Victor Javier Sanchez-Arevalo @vjsanchez

  • Contributors VJS-AL designed and performed the majority of the experiments. LCF and DM performed the immunofluorescence experiments and quantification in TMX-treated mice. EC-d-S-P was responsible for bioinformatics analysis. LR performed reporter assays. IC performed endogenous coimmunoprecipitations with pancreatic tissue. JC contributed to the RT-qPCR and immunofluorescence experiments. UM performed ChIP to show the co-occupancy between c-Myc and Ptf1a. NdP contributed to the mouse experiments. BB, CVW and MM provided important reagents for the study. VJS-AL and FXR designed and supervised the overall conduct of the study. FXR obtained financial support.

  • Funding This work was supported, in part, by grants from Ministerio de Economía y Competitividad, Madrid, Spain (SAF2007-60860, SAF2011-29530 and ONCOBIO Consolider), Instituto de Salud Carlos III, Madrid, Spain (RTICC RD12/0036/0034, partially funded by the European Regional Fund), Comunidad Autónoma de Madrid (grant CEL-DD) and European Union Seventh Framework Programme (grants 256974 and 289737) to FX.R. LCF was supported by a Marie Curie Training Grant (FP7-PEOPLE-2010-IEF, project 274946). IC was recipient of a Beca de Formación del Personal Investigador (MINECO, Madrid, Spain). JC was recipient of a grant from the La Caixa International PhD programme. UM was supported by a Beca Mixta CONACyT México, MZO2015.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement All data generated through the work reported in this manuscript are fully available either through public databases or upon request to the authors.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.