Maintenance of acinar cell organization is critical to preventing Kras-induced acinar-ductal metaplasia.

Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal cancers owing to a number of characteristics including difficulty in establishing early diagnosis and the absence of effective therapeutic regimens. A large number of genetic alterations have been ascribed to PDAC with mutations in the KRAS2 proto-oncogene thought to be an early event in the progression of disease. Recent lineage-tracing studies have shown that acinar cells expressing mutant Kras(G12D) are induced to transdifferentiate, generating duct-like cells through a process known as acinar-ductal metaplasia (ADM). ADM lesions then convert to precancerous pancreatic intraepithelial neoplasia (PanIN) that progresses to PDAC over time. Thus, understanding the earliest events involved in ADM/PanIN formation would provide much needed information on the molecular pathways that are instrumental in initiating this disease. As studying the transition of acinar cells to metaplastic ductal cells in vivo is complicated by analysis of the entire organ, an in vitro three dimensional (3D) culture system was used to model ADM outside the animal. Kras(G12D)-expressing acinar cells rapidly underwent ADM in 3D culture, forming ductal cysts that silenced acinar genes and activated duct genes, characteristics associated with in vivo ADM/PanIN lesions. Analysis of downstream KRAS signaling events established a critical importance for the Raf/MEK/ERK pathway in ADM induction. In addition, forced expression of the acinar-restricted transcription factor Mist1, which is critical to acinar cell organization, significantly attenuated Kras(G12D)-induced ADM/PanIN formation. These results suggest that maintaining MIST1 activity in Kras(G12D)-expressing acinar cells can partially mitigate the transformation activity of oncogenic KRAS. Future therapeutics that target both the MAPK pathway and Mist1 transcriptional networks may show promising efficacy in combating this deadly disease.

INTRODUCTION

With a 5-year survival rate of <5% for patients with pancreatic ductal adenocarcinoma (PDAC), the disease remains one of the most lethal malignancies primarily due to late diagnosis and ineffective chemo and radiation therapies (1). Recent identification of three distinct PDAC precursors - pancreatic intraepithelial neoplasia (PanIN), mucinous cystic neoplasm and intraductal papillary mucinous neoplasm - provides new hope for early detection and a much better understanding of the underlying mechanisms that are instrumental in the progression of the disease (2). Among these precursor lesions, PanINs are the best characterized (3). Indeed, a number of genetic mutations associated with PDAC have been detected in PanINs at various stages including mutations in the KRAS2, TP53 and SMAD4 genes which also are hallmarks of advanced PDAC, consistent with a PanIN → PDAC step-wise progression model (2). These observations are further supported by genetically engineered mouse models where PanIN/PDAC progression is initiated by single point mutations in the Kras gene and modified by additional genetic alterations in other oncogenes or tumor suppressors (4-10). However, the molecular mechanisms that govern each step of PanIN/PDAC progression remain elusive.

More controversial is the cellular origin of PDAC. PDAC has long been considered a disease of pancreatic ducts. Early efforts to model the disease by expressing Kras in pancreatic duct cells did not yield discenable pathology (11), although a recent study by Ray et al. (12) has shown that activation of endogenous Kras in large pancreatic ducts leads to rare early PanIN lesions. Despite this example, increasing evidence supports the idea that PanIN/PDAC can originate from differentiated acinar cells, which represent the major cellular component of the pancreas parenchyma (10, 13-15). The development of duct-like PanIN lesions from acinar cells necessitates massive remodeling of these cells, both morphologically and with respect to gene expression profiles. The transition from acinar to ductal cell properties has been termed acinar-ductal metaplasia (ADM) and lineage-tracing studies have confirmed that this process results from direct transdifferentiation of adult acinar cells that convert to a duct cell phenotype upon Kras expression (10, 13, 16). The relevance of ADM to PDAC is also supported by the observation that ADM is frequently associated with human PanIN lesions from PDAC patients (17, 18). Additionally, ADM development has been shown to precede PanIN formation in mouse Kras models (17), suggesting that ADM represents the initial stage of PDAC development.

Although it is clear that Kras expression in adult acinar cells can generate ADM lesions that progress to PanINs and PDAC, the molecular signaling pathways that are instrumental in these conversion events remain poorly defined. Indeed, Kras expression alone does not guarantee ADM/PanIN development because in mouse PDAC models where ~95% of acinar cells express Kras only a small cohort of cells generate detectable ADM/PanINs after a significant latent period (months), demonstrating that differentiated acinar cells are not universally responsive to KRAS signaling. This observation is compounded by the complexity of following ADM/PanIN development within the context of the intact organ. In an effort to develop an in vitro system to study the very earliest events of ADM formation, we focused on the initial transition of acinar cells converting to ductal cells upon Kras expression in a 3D culture model. We found that Kras-expressing acinar cells in 3D culture rapidly converted to ductal cells that mimicked properties associated with in vivo ADM and PanIN lesions, including silencing of acinar genes and expression of duct genes. Analysis of RAS signaling components demonstrated that activation of the Raf/MEK/ERK pathway was essential for Kras-induced ADM. Furthermore, the ability of acinar cells to undergo a transdifferentiation process to ductal cells was significantly attenuated by forced expression of the acinar-restricted transcription factor Mist1, suggesting that maintaining acinar cell identity mitigates the strong oncogenic potential of Kras expression. These results provide possible new therapeutic approaches to treat early ADM/PanIN cases by simultaneously targeting the MAPK pathway and Mist1 transcriptional networks.

RESULTS

KrasG12D expression converts pancreatic acini to ductal cysts in vitro

Kras mutations are virtually universal in human PDAC, although identification of the initial adult cell (duct, acinar, centroacinar) that acquires the mutation and initiates the transformation cascade remains unknown. Recent studies examining pancreatic cancer in mouse models have shown that Kras oncogene expression in adult acinar cells leads to PDAC that fully mimics the human disease (10, 13, 15, 16). Lineage-tracing of Elastase-CreER; LSL-Kras mice confirmed that activation of Kras in adult acinar cells induces acinar-ductal metaplasia (ADM), which can progress to pancreatic intraepithelial neoplasia (PanIN) and on to PDAC (Figure 1A).

Figure 1

Acinar-specific Kras → PanIN → PDAC. (A) Adult Elastase-CreER; LSL-Kras and Elastase-CreER; LSL-KrasG12D/+; LSL-Trp53R172H; R26RmTmG mice were given corn oil (control) or TM and pancreas sections were processed for standard histology after 1-5 months. The earliest lesions to appear are ADM (inset) followed by PanINs and then PDAC. R26 lineage tracing confirms that ADM, PanINs and PDAC are derived from Kras-expressing acinar cells (green). Red signal represents unrecombined R26 stromal cells. arrows - ADM and PanIN lesions; asterisks - adjacent stromal tissue. (B) Mist1 acinar cells were isolated from 7 day post-treated mice and individual acini were cultured for 5 days in a 3D collagen matrix. The majority of control acinar cells undergo extensive cell death by 5 days, whereas Kras-expressing acinar cells rapidly transdifferentiate into ductal cysts (H&E sections) that retain β-gal expression (whole mount X-gal staining). Identical results were obtained with Elastase-CreER; LSL-Kras mice. (C) Whereas corn oil treated control Mist1 acinar cells generate very few ductal cysts, TM-treated acinar cells generate a large number of ductal cysts in 3D collagen matrix.

Despite this fundamental process of initiating a transformation pathway, it has been difficult to study the intricacies of ADM → PanIN → PDAC in the intact animal. In an effort to dissect the initial Kras-dependent pathways in vitro, we asked if ADM could be modeled in tissue culture. Previous studies showed that wildtype acinar cells treated with TGFα and maintained in a 3D collagen matrix converted to duct-like cysts that activated EGFR downstream signaling pathways, including Ras (4, 19). To determine if constitutive KRAS activity could similarly lead to ductal cyst formation, Mist1 or Elastase-CreER; LSL-Kras mice were treated with corn oil (control) or tamoxifen (TM) to activate acinar-specific Kras expression. Pancreata were then harvested 7 days post-treatment, digested with collagenase and individual acini clusters placed in a 3D collagen matrix. As expected, control and Kras-expressing acinar cells initially retained their normal apical-basal polarity with the rER accumulating at the basal aspect and zymogen granules clustered at the apical pole (Figure 1B). However, by 5 days in 3D culture there was a dramatic difference in cell viability between the Kras-expressing and non-expressing cell populations. Control acinar cells rapidly became disorganized and underwent massive cell death. In contrast, a high proportion (~70-90%) of Kras-expressing acinar cells survived and converted to ductal cyst structures that were comprised of a single layer of epithelial cells surrounding an empty luminal space, resembling the metaplastic ADM/PanIN lesions observed in vivo (Figure 1A,B). Interestingly, these metaplastic cysts were virtually identical to structures formed by primary pancreatic duct cells (20). Lineage tracing using the R26R reporter allele confirmed that the ductal cysts were derived via transdifferentiation of mature Kras-expressing acinar cells (Figure 1B). In agreement with a recent study by Heid et al. (21), ductal cyst conversion was dependent on Kras activity as the vast majority of control acinar cells (-TM) failed to convert to ductal cysts over the duration of the experiment (Figure 1C).

The profound morphological changes in the 3D cultures were reminiscent of the in vivo changes that take place in the whole organ upon Kras activation. Indeed, a dramatic switch from acinar to ductal gene expression profiles occurred during the 5 day culture period. At day 0, individual acini expressed high levels of amylase, elastase, carboxypeptidase and lipase, with virtually no detectable expression of duct gene products including keratin 19 (K19), cystic fibrosis transmembrane conductance regulator (CFTR) and carbonic anhydrase II (CAII) (Figure 2; 3), confirming the purity of the starting acinar cell preparation. As expected, these cells also expressed high levels of the acinar-specific transcription factors Ptf1a and Mist1 (Figure 2; 3), with Mist1 being essential to maintaining an adult acinar cell phenotype (22). However, by day 5 Kras-induced ductal cysts exhibited a dramatic shift in their gene expression profiles that mirrored the expression pattern of PanIN lesions. Ductal cysts and PanINs uniformly silenced acinar genes and activated expression of duct-specific genes, including CFTR, K19, CAII and the duct-restricted transcription factors Sox9 and HNF6 (Figure 2; 3; S1; S2). These cells also expressed Nestin (Figure 3; S2), a marker of ductal cells in early PanINs (23). As predicted, PanINs in Mist1 pancreata were highly proliferative and a similar proliferation index was observed with the Kras-induced ductal cysts (Figure 2; S3). The parallel between ductal cyst formation in vitro and ADM/PanIN genesis in vivo suggests that the 3D acinar culture faithfully mimics the intracellular pathways that regulate acinar-ductal metaplasia.

Figure 2

KrasG12D-expressing acinar cells undergo conversion to a ductal phenotype when maintained in 3D collagen matrix

Fluorescence and light microscopy reveals that Kras acinar cells express acinar gene products at day 1 but rapidly convert to a ductal cell phenotype with activation of K19 and Sox9 expression. Inset for day 1 Mist1 panel shows the Dapi stained nuclei for this acinus. Note that a single centroacinar cell (arrow) in an isolated acinus is SOX9 positive. Pancreas serial sections from experimental littermates confirm the absence of AMYLASE and MIST1 and the expression of SOX9 and K19 in acinar-derived PanINs (arrows). Both PanINs and ductal cysts are also Ki67 positive (arrows). asterisks - adjacent acinar tissue.

Figure 3

KrasG12D-induced ductal cyst conversion leads to repression of the acinar gene program and activation of the ductal gene program

(A,B) RT-qPCR of day 0 and day 5 cultures confirms that Kras acinar cells express high levels of acinar gene products at day 0 but rapidly down-regulate acinar genes and up-regulate duct genes by day 5, a time that corresponds to maximum ductal cyst formation.

MAPK activity is elevated in PanINs and is essential for ductal cyst formation

It is well established that Kras expression induces cellular transformation by activating downstream signaling events, including the Raf/MEK/ERK, PI3K/AKT and RalGDS/Ral pathways (Figure 4A). Although these pathways have been the subject of potential therapeutic benefit against pancreatic tumors (24, 25), little is known about the role they have in the earliest transformation events of ADM and PanIN formation. In an effort to dissect the importance of the PI3K/AKT and Raf/MEK/ERK pathways prior to tumor formation, we examined Mist1 (HET/Kras) pancreata 3 months post-TM to ascertain if ADM/PanIN regions exhibited elevated KRAS signaling. Interestingly, despite the fact that ≥90% of adult acinar cells express Kras, only regions of ADM/PanIN exhibited sufficiently high pAKT and pERK levels that could be detected by immunohistochemistry (Figure 4B). Thus, expression of Kras in acinar cells (for up to 3 months) is not sufficient to uniformly activate the PI3K/AKT and Raf/MEK/ERK arms of the KRAS pathway in all cells.

Figure 4

The Raf/MEK/ERK pathway is essential to early ADM/PanIN formation in vitro and in vivo

(A) Canonical upstream and downstream pathways associated with Kras. (B) ADM/PanIN lesions (arrows) in the HET/Kras pancreas exhibit elevated levels of pAKT and pERK. Note that the majority of acinar cells (asterisks) in HET/Kras pancreata are KrasG12D positive but pAKT/pERK negative. (C) TGFα treatment of control wildtype acinar cells leads to ductal cyst formation in 3D culture. As expected, inhibition of EGFR activity (PD153035 treatment) blocks TGFα-dependent ductal cyst formation but has no effect on the downstream Kras pathway of HET/Kras acinar cells. (D) Inhibition of PI3K by LY294002 blocks pAKT activity but does not inhibit ductal cyst formation. (E) The MEK inhibitor PD98059 efficiently blocks pERK activity in HET/Kras acinar 3D cultures. Inhibition of MEK also leads to complete inhibition of Kras-induced ductal cyst formation.

To examine the importance of the pAKT and pERK pathways in converting acinar cells to ADM, we investigated these KRAS downstream pathways in the 3D culture model. As an initial test, we examined the importance of the upstream EGFR pathway with respect to Kras-induced ductal cyst formation. As previously reported (4, 19, 21), wildtype acinar cells placed in 3D culture exhibited only background levels of ductal cysts. However, when cells were provided TGFα, rapid ductal cyst formation occurred (Figure 4C). As expected, inhibition of EGFR by the EGFR tyrosine kinase inhibitor PD153035 completely blocked ductal cyst formation of wildtype acinar cells but had no effect on HET/Kras ductal cyst formation, confirming that KRAS functions downstream of EGFR (Figure 4A,C).

We next examined the importance of the PI3K/AKT and Raf/MEK/ERK pathways for ductal cyst formation in Het/Kras mice. As predicted, elevated pAKT was observed in Kras-isolated acinar cells when compared to control cells isolated from corn oil treated littermates (Figure 4D). Treatment of the cultures with the PI3K inhibitor LY294002 led to a ~70% reduction in pAKT levels but had no significant effect on ductal cyst formation in the 3D culture model. Likewise, 90% reduction of pAKT levels by the PI3K inhibitor VIII produced only modest decreases in ADM conversion (Figure S4). Together, these results suggest that the PI3K/AKT pathway is not required for, nor essential to, acinar-ductal metaplasia.

A similar analysis was performed on the Raf/MEK/ERK pathway. Whereas control cells exhibited low levels of pERK, Het/Kras acinar cells had elevated pERK levels (Figure 4E). Treatment of cells with the MEK inhibitor PD98059 efficiently blocked accumulation of pERK at concentrations as low as 20 μM. Interestingly, inhibition of MEK activity perfectly correlated with decreased ductal cyst formation (Figure 4E). We propose that activation of the Raf/MEK/ERK pathway in a subset of Kras-expressing acinar cells is required to initiate an ADM response in mouse models of PDAC.

Sustained Mist1 expression suppresses ductal cyst formation in vitro and ADM/PanIN formation in vivo

Kras-expressing acinar cells rapidly lose their acinar characteristics as they acquire a duct-like phenotype (13, 17). An early event of this switch is silencing the Mist1 locus, which encodes a basic helix-loop-helix transcription factor critical to acinar cell differentiation (4, 17). Mist1 is highly expressed in pancreatic acinar cells and its expression correlates with cellular differentiation and the establishment of the secretory machinery (22, 26). Previous studies have shown that human acinar cells lose Mist1 expression upon ADM/PanIN formation (4, 17), and in mouse models devoid of the Mist1 gene, Kras-expressing acinar cells rapidly undergo ADM and develop extensive PanIN lesions through a pathway that involves elevated pERK activity (4, 27). Therefore, we hypothesized that MIST1 promotes acinar cell differentiation and suppresses the oncogenic activity of Kras by keeping pERK levels low. Given that ADM is dependent on MAPK activity (Figure 4E), we examined if sustained Mist1 expression could restrict the conversion of Kras-expressing acinar cells to ductal cysts.

To test this concept, a transgenic mouse line (LSL-Mist1) that produces constitutive Mist1 expression upon Cre-mediated recombination was generated (Figure 5A). Transgene expression of myc-tagged Mist1 was driven by a CMV early enhancer/chicken β-actin hybrid promoter, whose activity is independent of the differentiation status of cells (28). When Mist1 (KO/Kras/Mist1myc) mice were treated with TM, Mist1 expression was observed in >90% acinar cells with no expression detected in duct cells or in corn oil treated (-TM) control mice (Figure 5B). Next, we examined MAPK activity in pancreata +/- Kras and Mist1 expression. As shown in Figure 5C, loss of Mist1 (KO) led to elevated pERK levels, even in the absence of Kras expression. However, sustained Mist1 expression in the KO/Kras/Mist1myc pancreata generated lower pERK levels when compared to KO/Kras littermates (Figure 5C). We then examined how constitutive Mist1 expression influenced ductal cyst conversion in 3D cultures. As expected, Mist1+/- (HET) and Mist1-/- (KO) acini generated few ductal cysts when placed in collagen matrix for 5 days (Figure 5D). In contrast, TM-treated Mist1 (HET/Kras) acini rapidly converted to ductal cysts and the percentage of conversion increased when Mist1 (KO/Kras) acinar cells were tested (Figure 5D; S5). Importantly, when KO/Kras/Mist1myc acini were examined, ductal cyst formation was reduced to levels approaching those observed in Kras-negative (Mist1+/-, Mist1-/-) cells (Figure 5D; S5). Similar results were obtained with the TGFα model where Mist1 expression blocked TGFα-induced ductal cyst formation of Mist1-/- acinar cells (Figure S6). These results demonstrate that sustained Mist1 expression lowers pERK levels and protects acinar cells from TGFα- or Kras-induced ADM formation in vitro.

Figure 5

Sustained Mist1 expression suppresses KrasG12D-induced ductal cysts

(A) Schematic diagram of the LSL-Mist1myc transgene. (B) Anti-myc immunohistochemistry of KO/Kras/Mist1myc pancreata reveals extensive acinar-specific expression of LSL-Mist1myc in tamoxifen-treated mice (arrows). (C) Immunoblot of pERK activity in HET vs. KO and KO/Kras vs. KO/Kras/Mist1myc pancreata. The absence of Mist1 leads to increased pERK activity while sustained Mist1 expression reduces pERK levels in KO/Kras/Mist1myc pancreata. (D) Quantification of ductal cyst formation in control and Kras-expressing cells. Mice were treated with TM and acinar cells placed in 3D culture as described in Materials and Methods. Sustained Mist1 expression dramatically inhibits Kras-induced ductal cyst formation.

Finally, to determine if sustained Mist1 expression could similarly deter ADM/PanIN development in vivo, adult KO/Kras/Mist1myc mice and their littermate controls were treated with TM and pancreata were examined three months post-TM for the presence of ADM and PanIN lesions. HET/Kras mice developed rare focal areas of ADM and PanIN lesions at 3 months and in all cases the Mist1 locus was transcriptionally silenced in acinar-derived PanINs and in advanced ADM lesions (Figure 6A,B). As expected, KO/Kras pancreata developed organ-wide extensive ADM/PanINs owing to the absence of Mist1 prior to Kras activation (Figure 6A,B). In contrast, sustained Mist1 expression greatly attenuated ADM/PanIN development in the KO/Kras/Mist1myc mice as the majority of the organ consisted of normal appearing acinar tissue with only isolated areas of ADM/PanINs (Figure 6A,B). Indeed, the number of PanINs that developed in KO/Kras/Mist1myc mice and reported in Figure 6B actually represent an upper limit because not all PanINs (~18%) were MIST1myc positive, presumably due to selected Cre-mediated recombination of the LSL-Kras locus and not the LSL-Mist1 transgene. Thus, a subset of Kras-expressing acinar cells did not have an opportunity to respond to sustained MIST1 protein. However, despite the clear protective effect of Mist1 expression, MIST1myc protein per se did not fully inhibit Kras-induced PanINs as MIST1myc+ PanINs were still generated, albeit at a greatly reduced level (Figure 6A,B). The MIST1myc-dependent reduction in ADM/PanIN formation was also reflected in the expression profiles of key acinar vs. ductal genes where MIST1myc significantly reduced ductal gene activity while maintaining acinar gene expression to levels approaching, or exceeding, those obtained with HET/Kras mice (Figure 6C). These results are consistent with MIST1 exerting an anti-ADM/PanIN activity in the intact pancreas and support the concept that constitutive Mist1 expression mitigates the activity of oncogenic Kras in acinar cells, leading to a reduction in ADM/PanIN formation.

Figure 6

Sustained Mist1myc expression inhibits ADM and PanIN formation in vivo

(A) PanIN formation (arrows) is greatly accelerated in KO/Kras pancreata but suppressed by sustained Mist1 expression in KO/Kras/Mist1myc mice. Acinar-derived PanINs are Mist1 negative in HET/Kras mice. Interestingly, induced Mist1 expression alone is not sufficient to inhibit all PanIN formation since some MIST1myc+ PanINs still develop in KO/Kras/Mist1myc pancreata. asterisks - adjacent acinar tissue. (B) Quantification of ADM and PanIN lesions in HET/Kras, KO/Kras and KO/Kras/Mist1myc samples. Note that constitutive Mist1 leads to significant reductions in ADM/PanIN formation. (C) Analysis of gene expression profiles by RT-qPCR confirms the reduced ADM/PanIN formation in KO/Kras/Mist1myc samples when compared to KO/Kras littermates. All values were normalized to the control HET samples, which were set to 1.0.

DISCUSSION

The role of acinar cells in pancreatic tumorigenesis has long been suspected, but definitive evidence had been lacking until lineage tracing studies established that PanINs and PDAC can arise from adult acinar cells (10, 13, 16). These studies also implicated the process of acinar-ductal metaplasia as a bridge between normal acinar cells and duct-like PanINs. Consistent with this hypothesis, ADM occurrence precedes PanIN development in PDAC mouse models (17). Despite the central role for Kras in inducing ADM, Kras mutations are not a prerequisite. Mutant Kras is often detected in ADM and low-grade PanIN lesions in human diseased organs but these mutations are not universally present (2, 29), suggesting that additional underlying events can influence ADM formation. Indeed, exposure to EGFR growth factors leads to metaplastic lesions in vitro and in vivo in the absence of active Kras (30, 31). Similarly, EGFR activity is a hallmark of Kras-induced ADM/PanIN lesions (17), implying a central role for this pathway in Kras-dependent and Kras-independent ADM. ADM is also commonly associated with acute and chronic pancreatitis in both humans and rodents (18, 32, 33), suggesting that inflammation and/or cell damage can simultaneously generate an ADM response without Kras activity. Although a number of studies have shown that metaplastic cells in pancreatitis settings eventually resume an acinar cell phenotype (33, 34), the presence of Kras mutations efficiently redirects metaplastic cells toward PanIN development (15, 35). These results, together with the observation that chronic pancreatitis increases the risk for PDA (36), underscore the relevance of ADM in pancreatic tumorigenesis and the need to elucidate the molecular mechanisms that regulate the fate of metaplastic cells.

In this study, we adopted a collagen matrix 3D culture assay to model the conversion of mature acinar cells to ductal metaplastic cells within a compressed timeframe (5 days vs. months). As expected, expression of the Kras oncogene promoted formation of duct-like cysts, with an efficacy similar to treatment of wildtype acinar cells with exogenous TGFα (4, 19, 21). Further analysis confirmed that the starting cell preparation was highly enriched for amylase-expressing acinar cells, which converted to K19-expressing ductal cells in the presence of Kras. Using inhibitors that specifically blocked individual signaling components, we also demonstrated that Raf/MEK/ERK represent a critical downstream effector pathway through which KRAS operates to induce ADM. Interestingly, Braf and Kras, the predominant raf and ras gene mutations, respectively, in pancreatic cancer operate in a mutually exclusive pattern in pancreatic tumors, suggesting that upstream (Kras) or downstream (Braf) effectors of the Raf/MEK/ERK pathway exhibit interchangeable functions in converting acinar cells to ADM lesions (37). Recent lineage-tracing studies of primary human acinar cells also revealed that surviving acinar cells undergo an ADM response that is dependent on MAPK activity (38). Likewise, we have found that TGFα-induced ADM is also dependent on the Raf/MEK/ERK pathway (unpublished results), revealing that these downstream effectors represent a common regulator of ADM in mouse and human Kras-dependent and Kras-independent settings. Future studies will test if individual dominant-negative mutants of the MAPK pathway similarly block ADM in models of tumorigenesis and pancreatitis.

While these studies were in progress, Heid et al. (21) reported that acinar cells isolated from Ptf1a mice were also capable of producing ductal cysts in 3D culture. Additionally, conditional deletion of Rac1 significantly blocked ductal cyst formation in vitro and PanIN development in vivo. This was a surprising finding as Rac1 is often thought to be activated by the PI3K pathway which we showed by selective inhibition to be nonessential for ADM formation. However, a number of studies have shown that Rac can also be activated in a PI3K-independent fashion via the Raf-specific guanine exchange factor Tiam1 (39, 40). Importantly, Tiam1 null mice are resistant to Ras-induced tumor formation (41), confirming the importance of the Tiam1-Rac1 axis for the Ras signaling pathway. Thus, our study and the work by Heid et al. (21) strongly suggest that there are two key pathways (MAPK and Rac1) that are instrumental in Kras-induced ADM and that both pathways are essential to converting normal acinar cells into ADM and PanIN lesions. Future PDAC therapeutic strategies that target both pathways are likely to have higher efficacy than approaches targeting only a single effector.

Are there other ways to block ADM? We can find some leads in studies of pancreatitis, where most ADM spontaneously heals and re-differentiates in the absence of Kras signaling. For example, β-catenin activation following acute pancreatitis is required for acinar cell regeneration, while stabilized β-catenin protects acinar cells against Kras--induced PanIN formation (16). Similarly, Mist1 exhibits a protective role not only in Kras-induced ADM but also in ADM associated with acute pancreatitis, where Mist1 null pancreata endure more severe damage following caerulein treatment (42). At this time, it is unclear how Mist1 influences ADM decisions. One possibility is that MIST1 protects acinar cells from metabolic stress associated with pancreatitis or Kras expression by regulating aspects of the MAPK pathway. Indeed, Mist1 is silenced in pancreatitis-associated acinar cells (unpublished results) and Mist1 acinar cells exhibit increased MEK and ERK phosphorylation (4). Given the central role of the MAPK pathway in ADM and the observation that pancreatitis enhances Kras-induced tumorigenesis (15, 35, 43), it is likely that Mist1 inhibits ADM by maintaining basal levels of these MAPK effectors. Since Mist1 is influential in converting embryonic stem cells to an acinar cell fate (44), sustained Mist1 expression also likely keeps key acinar characteristics intact in Kras cells. Further investigation into the mechanisms of this regulatory network and identification of MIST1 target genes should lead to better therapeutic interventions and improved clinical outcomes for PDAC patients.

Materials and Methods

Mouse strains and genotyping

Mist1CreER/+, LSL-KrasG12D/+, R26RLacZ and R26RmTmG reporter mouse lines have been described previously (4, 13, 45). Elastase mice were generated according to standard protocols using a previously reported Elastasepr-Mist1MB construct where the Mist1MB coding region was replaced by a CreERT2 coding region to drive acinar-specific expression of CreERT2 (46). LSL-Mist1 transgenic mice were generated using the pCAG-LoxP-CAT-LoxP-LacZ construct in which the LacZ gene was replaced by a myc-tagged Mist1 coding sequence (47). Induction of CreERT2 activity was accomplished by providing adult mice (6-9 wk) tamoxifen (TM, 4 mg/mouse/day) for 2-3 consecutive days. Genotyping primer sets are listed in Supplemental Table I. All animal studies were conducted in compliance with NIH and the Purdue University IACUC guidelines.

3D acinar cell culture

Mice were treated with TM 7 days prior to harvesting the pancreas. Immediately after sacrificing, pancreata were rinsed in cold 1X HBSS (Invitrogen, Carlsbad, CA) and cut into small pieces with scissors. Primary acini were released by collagenase P (Roche Applied Science, Mannheim, Germany) digestion (200 μg/ml in 1x HBSS) for 10-20 minutes at 37°C. Isolated acini were washed 3 times in cold 5% FBS, 1x HBSS and then filtered sequentially through 500 μm and 105 μm nylon meshes (Spectrum Laboratories, Rancho Dominguez, CA). Cell suspensions were carefully layered on top of 30% FBS, 1x HBSS and acini were collected by centrifugation (1000 rpm, 2 min at 4°C) and then resuspended in 8 ml of 3D culture base medium (RPMI, 10% fetal bovine serum, 0.1 mg/ml soybean trypsin inhibitor, 1 μg/ml dexamethasone and antibiotics). 24-well tissue culture plates were coated with a 250 μl/well collagen layer (100 μl 10x RPMI, 900 μl 3mg/ml collagen, neutralized with either 4.2% NaHCO3 or 0.34N NaOH) at least one hr prior to acini isolation. The cell suspension was then mixed with collagen 1:1 and plated (0.5 ml per well). The cell-collagen mix was allowed to solidify for 1 hr at 37°C before adding 1 ml 3D culture media. Media was changed on days 1 and 3. On day 5 the percentage of acini that converted to ductal cysts was calculated by counting individual clusters in all wells. Collagen matrix was then digested for 10-15 min at 37°C with 200 μg/ml collagenase P in 1x HBSS to release cells. RNA was prepared using the E.Z.N.A. total RNA kit (Omega Bio-Tek, Norcross, GA). Alternatively, collagen discs were fixed in 4% formaldehyde, embedded in paraffin and 5 μm sections were prepared for standard immunohistochemistry as described below.

Histology and immunohistochemistry

Mouse pancreata were fixed in 10% neutral buffered formalin, paraffin embedded, sectioned to 5 μm and stained using conventional hematoxylin and eosin. Sections were deparaffinized and retrieved using the 2100-Retriever (PickCell Laboratories, Amsterdam, The Netherlands) and antigen unmasking solution (Vector Laboratories, Burlingame, CA). Samples were blocked using the MOM blocking reagent (Vector Laboratories) and incubation of primary antibodies was conducted for 1 hr at room temperature or overnight at 4°C. Biotinylated secondary antibodies were applied for 10 min at 25°C. Visualization was accomplished via 3,3'-Diaminobenzidine peroxidase staining or tertiary avidin-conjugated fluorescent antibodies. Primary antibodies and conditions are provided in Supplemental Table II.

Protein immunoblots

Twenty μg of whole cell protein extracts were separated on 12% acrylamide gels, transferred to PVDF membranes and incubated with primary antibodies (antibody conditions are provided in Supplemental Table III). Immunoblots were developed using an ECL kit (Thermo Scientific, Waltham, MA).

RT-qPCR gene expression analysis

Pancreas RNA or RNA from 3D cultures was isolated using the E.Z.N.A. total RNA kit and reverse transcribed using the iScript™ cDNA synthesis kit (Bio-Rad, Hercules, CA). Equal amounts of cDNA reactions were amplified with FastStart Universal SYBR Green Master (Roche Applied Science, Indianapolis, IN) using the primer sets listed in Supplemental Table IV. Target sequences were amplified with 95°C/30 sec, 59°C/60 sec, 72°C/30 sec conditions. Fold changes in gene expression between TM-treated and control animals or between day 0 and day 5 3D cultures were calculated using the comparative 2-ΔΔCt method. Error bars represent the S.E.M.