MECOM permits pancreatic acinar cell dedifferentiation avoiding cell death under stress conditions

Maintenance of the pancreatic acinar cell phenotype suppresses tumor formation. Hence, repetitive acute or chronic pancreatitis, stress conditions in which the acinar cells dedifferentiate, predispose for cancer formation in the pancreas. Dedifferentiated acinar cells acquire a large panel of duct cell-specific markers. However, it remains unclear to what extent dedifferentiated acini differ from native duct cells and which genes are uniquely regulating acinar cell dedifferentiation. Moreover, most studies have been performed on mice since the availability of human cells is scarce. Here, we applied a non-genetic lineage tracing method of human pancreatic exocrine acinar and duct cells that allowed cell-type-specific gene expression profiling by RNA sequencing. Subsequent to this discovery analysis, one transcription factor that was unique for dedifferentiated acinar cells was functionally characterized. RNA sequencing analysis showed that human dedifferentiated acinar cells expressed genes in “Pathways of cancer” with a prominence of MECOM (EVI-1), a transcription factor that was not expressed by duct cells. During mouse embryonic development, pre-acinar cells also transiently expressed MECOM and in the adult mouse pancreas, MECOM was re-expressed when mice were subjected to acute and chronic pancreatitis, conditions in which acinar cells dedifferentiate. In human cells and in mice, MECOM expression correlated with and was directly regulated by SOX9. Mouse acinar cells that, by genetic manipulation, lose the ability to upregulate MECOM showed impaired cell adhesion, more prominent acinar cell death, and suppressed acinar cell dedifferentiation by limited ERK signaling. In conclusion, we transcriptionally profiled the two major human pancreatic exocrine cell types, acinar and duct cells, during experimental stress conditions. We provide insights that in dedifferentiated acinar cells, cancer pathways are upregulated in which MECOM is a critical regulator that suppresses acinar cell death by permitting cellular dedifferentiation.


Introduction
Few cancers have such poor survival rates as pancreatic ductal adenocarcinoma (PDAC) [1]. Understanding the molecular mechanisms that are at play in early-stage cancer formation is important to improve disease outcomes, but Edited by M Bianchi. clinical samples of early disease, in this case, are rare. Hence, genetically modified mouse models are pivotal to provide us insight into the contribution of the different pancreatic cell types to tumor formation [2][3][4]. Accumulating evidence points to acinar cells at the origin of pancreatic tumor development. In order to become susceptible to tumorigenesis, acinar cells have to lose their differentiated phenotype to some extent. Consequently, tumorigenesis is restrained by mechanisms that control acinar cell differentiation [5][6][7][8][9][10]. Loss of differentiation can result from repetitive acute or chronic pancreatitis whereby dedifferentiated acinar cells acquire a panel of markers that are typical for (embryonic) duct cells [11][12][13]. Notably, duct cells can also give rise to tumors but likely under different conditions and through other mechanisms [14][15][16].
To date, knowledge is lacking on the mechanisms that govern the duct-like dedifferentiated acinar cell phenotype versus the "native" duct cell phenotype, the latter being more resistant to stress-or inflammation-associated tumorigenesis. Previously, we established in vitro experimental models that recapitulate human [12] and rodent [13] acinar dedifferentiation. Specifically, upon suspension culture acinar cells lose their acinar-specific markers while gaining (embryonic) duct-like features, thereby mimicking chronic pancreatitis [12,13]. Here, we investigated the differential gene expression in adult human dedifferentiated acinar cells compared to native duct cells from the same cell preparations. The transcription factor MECOM (EVI-1) was found to be uniquely re-expressed-since embryonic development -in dedifferentiated acinar cells.
Although the MECOM transcription factor is a known oncogene with a role in apoptosis [17][18][19] and pancreatic tumor formation [20,21], its function in acinar cell (de) differentiation was unknown. We analyzed its role in the context of acinar cell dedifferentiation by caerulein-induced pancreatitis modeling in transgenic Evi-1 knock-out mice. We provide mechanistic insights that mouse acinar cells are dependent on MECOM expression to prevent cell death, allowing them to acquire a dedifferentiated cell state.

Materials and methods
Human donor material and cell culture Human healthy donor pancreatic exocrine cell preparations as residual material from a pancreatic islet donor program were processed by the Beta Cell Bank of the JDRF Center for Beta Cell Therapy in Diabetes (Brussels, Belgium), affiliated to the Eurotransplant Foundation (Leiden, The Netherlands). Consent for the use of residual donor material for research was obtained according to the legislation in the country of organ procurement. This project was approved by the Medical Ethical Committee of UZ Brussel-Vrije Universiteit Brussel (B.U.N. 143201732260). Human exocrine cells were cultured as previously described [11,12].
All animal experiments were approved by the Ethical Committee for Animal Testing at the Vrije Universiteit Brussel (#17-637-1).

Fluorescent activated cell sorting (FACS)
FITC-conjugated UEA-1 (Ulex Europaeus Agglutinin-1, Sigma-Aldrich) lectin labeling of acinar cells within the human exocrine cell fraction (isolated from healthy donor pancreata) was performed as in [12]. On day 4 of suspension culture, cell clusters were dissociated following the protocol of [11]. Analysis and cell sorting were performed on a BD FACSAria (BD Biosciences, Franklin Lakes, NJ, USA). Viable, single cells were gated based on forward and side scatter.

Cell lines and culture
The mouse partially differentiated acinar cell line 266-6® CRL-2151 was purchased from American Type Culture Collection (ATCC, Manassas, Virginia, USA). Cells were maintained in DMEM + GlutaMax medium, supplemented with 10% heat-inactivated fetal bovine serum and 1% penicillin-streptomycin (all from Gibco) in a humidified atmosphere at 37°C and 5% CO 2 . All cell lines tested negative for mycoplasma contamination. The absolute cell number was determined by seeding 50,000 cells per well of a 6-well plate on day 0 and counted 96 h after seeding. CellTiter Glo assay (Promega, Madison, Wisconsin, USA) was carried out according to the manufacturer's instructions. In short, cells were seeded in triplo in opaque 96-well plates and cell viability was measured 96 h later using the GloMax Discover Microplate reader (Promega). Luciferase reporter assay (Promega) was carried out according to the manufacturer's instructions. Briefly, 1,000 cells were seeded in triplo in opaque 96-well plates and transfected 24 h later with a total of 200 ng DNA (pcDNA3.1-Sox9 only, pGL3-E1Bp-MECenh (a kind gift by R. Cullum, BC Cancer Research Center, Vancouver, Canada) only or cotransfection with both). Luciferase activity was measured 48 h later using the GloMax ® Discover Microplate reader.

Lentiviral transduction and selection
In total, 266-6 cells were seeded at a density of 50,000 cells per well of a 6-well plate. Twenty-four hours later, cells were transduced with lentiviral supernatants supplemented with protamine sulphate (10 µg/mL, Sigma-Aldrich) for 48 h. After splitting, cells were selected for at least 1 week in 0.4 µg/mL puromycin (Invitrogen) before any further analysis was performed. Human exocrine cells were transduced with lentiviral shRNA-Evi1 vectors (MOI 10) on the day of isolation for 16 h at a density of 200,000 cells in advanced RPMI supplemented with 5% heat-inactivated FBS, 1% GlutaMax, and 1% Pen/Strep supplemented with protamine sulphate (10 µg/mL, Sigma-Aldrich). Cells were then washed with PBS and further cultured for 4-7 days in Advanced RPMI supplemented with 5% heat-inactivated FBS, 1% GlutaMax, and 1% Pen/Strep.

Immunostainings
Pancreas tissue and cell pellets were fixed overnight at room temperature in 10% neutral-buffered formalin (Kli-niPath, Olen, Belgium). Next, samples were washed twice with PBS, dehydrated, and embedded in paraffin.

RNA in situ hybridization
RNA in situ hybridization (RISH) was performed on formalin-fixed, paraffin-embedded tissue sections, according to the manufacturer's protocols for manual RNAscope ® 2.5 HD Assay-RED (Bio-techne). Images were acquired using an EVOS FL1 auto digital microscope (Thermo Fisher Scientific). Immunostainings and average mRNA dots (RISH) were determined using HALO ® image analysis platform (Indica Labs) or FIJI ® imaging software.

Quantitative reverse transcription-polymerase chain reaction (qRT-PCR)
Total RNA was isolated from cells using NucleoSpin RNA isolation kit (Macherey-Nagel, Düren, Germany) or GenElute Mammalian total RNA Miniprep kit (Sigma-Aldrich). RNA concentration was measured by NanoDrop2000 (Thermo-Fisher). cDNA was prepared using the GoScript Reverse Transcription System (Invitrogen). qPCR was performed using FastSYBRGreen 5× MasterMix on a QuantStudio 6 (Invitrogen). The analysis was done using the ΔΔCt method using HPRT as the housekeeping gene.

RNA sequencing and data analysis
In total, 75 bp single-end RNAseq of purified human dedifferentiated acinar and duct cells was performed on an Illumina NextSeq 500 platform at VIB, Nucleomics Core, Leuven Belgium with TruSeq library prep. Low quality ends (<Q20) were trimmed using FastX 0.0.13 (http://ha nnonlab.cshl.edu/fastx_toolkit/index.html). Reads shorter than 35 bp after trimming were removed. Using FastX 0.0.13 and ShortRead 1.16.3 (Bioconductor), polyA reads (more than 90% of bases equal A), ambiguous reads (containing N), low quality reads (more than 50% of the bases <Q25), and artifacts reads (all but three bases in the read equal one base type) were removed. The preprocessed reads were aligned to the reference genome of Homo sapiens (GRCh3773) with Tophat v2.0.13 [25]. Withinand between-sample normalization was corrected for using full quantile normalization with the EDASeq package from Bioconductor. The number of reads was counted in the alignment that overlapped with gene features using fea-tureCounts 1.4.6 [26]. FPKM values were determined by dividing for each sample the normalized counts by the total number of counts (in millions). Then for each gene, the scaled counts were divided by the gene length (in kbp). As such we got the number of Fragments Per Kilobase of gene sequence and per Million fragments of library size. Data will be publicly available on EBI (accession number E-MTAB-9386). In total, 100 bp paired-end RNA sequencing of Mecom KD versus wild-type 266-6 cells was performed using Kappa HMR Library prep at BrightCore, UZ Brussel, Brussels, Belgium on an Illumina Novaseq 6000. The read quality of the raw data was evaluated using FASTQC (http://www.bioinformatics.ba braham.ac.uk/projects/fastqc/), which flags any potential abnormalities that may have occurred during library preparation or sequencing. The raw reads were mapped against the Mus Musculus genome (version GRCm38-83) using the RNA-seq aligner STAR [27]. The mapped reads were then translated into gene counts with the open-source tool (HTSeq) [28]. Following quantification of expression levels, we carried out a differential expression analysis for both RNAseq datasets between experimental groups using the DESeq2 package in R [29]. KEGG pathway analysis was performed using the KOBAS gene set enrichment tool. Transcription factors were annotated using the animal transcription factor database of human transcription factors.

Biospecimens and clinical data
Biospecimens, transcriptomic and clinical data were collected by the Australian pancreatic cancer genome initiative (APGI), and informed consent was obtained from all patients. Use of clinical samples and data were in accordance with national ethical guidelines and regulations in Australia (HREC/11/RPAH/329-Sydney Local Health District-RPA Zone, protocol X11-0220) and in Belgium (UZ Brussel, B.U.N.143201732468).

Statistics
Sample sizes were determined based on the type of sample, being three independent biological repeats for cell lines and three to six independent biological repeats for mice experiments. Specifically for the in vivo experiments with ElaCreERT and ElaCreERT;Mecom f/f mice, the sample size was calculated based on power analysis with the power of 0.8 at a p = 0.05 using G*Power [30]. Statistical analysis was performed using Prism v8.0 (GraphPad Software, La Jolla, CA, USA). Differences among groups were tested for statistical significance by applying paired or unpaired twotailed parametric Student's t-test, or one-way or two-way ANOVA with Bonferroni post-hoc test to correct for multiple comparisons. For each experiment, the specific statistical test is mentioned in the figure legend. The data met the assumptions of the test and the variance between groups that were being statistically compared was similar. The number of independent repeats is indicated as N.

MECOM is a cancer-related transcription factor uniquely expressed in human dedifferentiated acinar cells
Given the difference in propensity of dedifferentiated acinar cells and duct cells for pancreatitis-associated tumor development [5], we were specifically interested in the differential gene expression of these two populations. A mixed human acinar and duct cell fraction was obtained as residual material from a pancreatic islet donor program. The cells were fluorescently labeled with the acinar-specific FITC-UEA-1 lectin and cultured in suspension, as described previously [11,12]. After culture, we FACS-purified dedifferentiated acinar cells and duct cells, based on FITC-UEA-1 and the duct cell marker CA19.9, respectively [11,12] (Fig. 1a). We confirmed the mutually exclusive expression pattern of CD142, a progenitor marker [11] in the UEA1+ dedifferentiated acinar cells, and of CA19.9 in the UEA1− duct cells (Fig. 1b). We validated the different expression levels of the acinar cell markers MIST-1 and chymotrypsin, the progenitor marker CD142, and the duct-cell marker CK19 in the original mixed exocrine cell fraction and in the purified dedifferentiated acinar cells (UEA1+/CA19.9−) and native duct cells (UEA1−/CA19.9+) (Fig. 1c). Comparing dedifferentiated acinar cells with native duct cells by RNAseq analysis highlighted 7953 differentially expressed genes (Adj. p < 0.01). Of these, 1219 genes were expressed higher in dedifferentiated acinar cells (log fold change (FC) of ≤−2; Adj. p < 0.01) (Supplementary Table 1) and were enriched for the KEGG pathways (Adj. p < 0.01): "Protein digestion and absorption" and "Pancreatic secretion". This list included genes such as carboxypeptidase A and amylase, pointing to gene expression reminiscent of the abundant acinar cell transcriptome [31]. In addition, "Pathways in cancer" belonged to the top enriched pathways such as those of the WNT signaling pathway, which is wellknown for its function in pancreatic tumorigenesis. Other enriched pathways included "PI3K-AKT signaling", which is involved in acinar cell dedifferentiation and pancreatic tumorigenesis [32][33][34], and "Complement and coagulation cascades" encompassing genes belonging to the tissue damage pathway (Fig. 1e) and that code for F8, fibrinogen, and F3 (CD142), the marker used above [11] (Fig. 1b).
Dedifferentiated acinar and duct cells did have an equal expression of the embryonic precursor gene PDX1, confirming our previous work [13]. PTF1A, NR5A2, RBPJL, FOXA3, and GATA4 were expressed higher in the dedifferentiated acini, reminiscent of their fully differentiated state. Of note, GATA6 expression, which is required for embryonic acinar cell differentiation [35,36] and is a suppressor of pancreatic carcinogenesis [37,38], was not different from the duct cells.
Conversely, duct cell-enriched genes included CFTR, SOX9, NKX6.1, HES1, HNF1B, and ONECUT2 (HNF6B), established markers of duct cell differentiation. The most significant duct cell-enriched gene was DCDC2 (Doublecortin Domain Containing 2), demonstrated to bind tubulin and enhance microtubule polymerization [39] (Fig. 1d). In the duct cell fraction, we found KEGG pathway enrichment for "ECM receptor pathway interaction" and "O-linked glycosylation". Also, "Pathways in cancer" featured here, but the genes in this pathway differed from the ones in the dedifferentiated acinar cell fraction (Supplementary Table 1).
There were 71 genes encoding transcription factors within the differentially expressed genes. The most significantly enriched transcription factor in the dedifferentiated acinar cell fraction was MECOM (log2FC = −6.09; Adj. p = 1.04E−123) (Fig. 1d, Supplementary  Table 1), a known oncogene [40][41][42]. It has been described as an activator of KRAS and glypican-1 in pancreatic cancer [20,21] and, very recently, it was reported in relation to acinar cell-derived tumorigenesis [43]. Hence, we confirmed MECOM expression in human PDAC samples (N = 3) (Supplementary Fig. 1a). We also used RNAseq analysis of stage 1 and 2 PDAC (N = 103) to differentiate low and high MECOM expression. MECOM expression correlated positively with disease-free survival ( Supplementary Fig. 1b). In the same cohort, we previously reported a similar association for SOX9 and showed that SOX9 associated with the classical subtype, a molecular subset of PDAC that has a better prognosis. Interestingly, in that same study, we had found a correlation in expression between SOX9 and MECOM (Pearson R=0.49) [23]. When now we performed pathway analysis of genes correlating with MECOM ( Supplementary  Fig. 1c), indeed, similar pathways as those reported in our study on SOX9 were enriched, among which ErbB signaling was highlighted [23].
In conclusion, we FACS-purified human dedifferentiated acinar and duct cells and analyzed their transcriptome. The dedifferentiated acinar cell fraction showed a signature reminiscent of the acinar cell identity and specifically expressed MECOM, a transcription factor that featured in the pathway "Pathways of cancer". MECOM was expressed in PDAC, where it correlated with SOX9 and ErbB/ERK signaling. We hypothesized that MECOM plays a role during acinar cell dedifferentiation, an event that initiates pancreatic tumor development, and this in connection to SOX9 and ERK signaling.
MECOM is expressed in developing acinar cells and is re-expressed during pancreatitis-associated acinar cell dedifferentiation We comprehensively assessed the expression of MECOM during acinar cell differentiation or loss thereof.
First, we confirmed the differential expression of MECOM in the purified human dedifferentiated acinar and duct cells from the experiments above (Fig. 2a). MECOM mRNA appeared at low levels in the acinar compartment of the normal human pancreas (0.24 ± 0.09 mRNA dots/ cell; N = 3) but increased in samples of chronic pancreatitis (1.46 ± 0.10 mRNA dots/cell; N = 3; *p = 0.02) (Fig. 2b, c).
Next, we assessed the expression pattern ( Fig. 2f-i) and levels (Fig. 2k) of Mecom mRNA in the murine pancreas. In contrast to the adult mouse pancreas, where Mecom was expressed at very low levels, a significantly increased expression was noted in the E16.5 pancreas where it localized to pro-acinar tip cells [44,45]. Alike the human duct cells, we did not detect Mecom in CK19+ mouse duct cells (Fig. 2j). Also similar to humans, Mecom became reexpressed during acute and chronic pancreatitis in mice (Fig. 2h, i, k).
Altogether, these data show that Mecom is transiently expressed in developing acinar cells and becomes re-expressed during pancreatitis in dedifferentiated acinar cells.

SOX9 regulates Mecom expression in dedifferentiated acinar cells
SOX9 is an important marker of acinar cell dedifferentiation and is indispensable for tumor formation in the pancreas [23]. SOX9 correlates with MECOM in PDAC patients, as referred above, an observation also reported in other tissues [46,47]. Hence, we investigated whether these two transcription factors might correlate in the context of acinar cell dedifferentiation. Indeed, we observed a strong correlation between MECOM and SOX9 (Pearson R = 0.83; *p = 0.015; N = 6) (Fig. 3a, b) in the human exocrine cell cultures in which acinar cells gradually dedifferentiate, a finding which was confirmed at the protein level (Fig. 3c).
In conclusion, we showed that Mecom is regulated by SOX9 in the pancreatic acinar cell context.

Mecom maintains cell adhesion and cell viability in a partially differentiated acinar cell line and in human and mouse acinar cell cultures
We engaged in a lentiviral all-in-one CRISPR/Cas9-mediated approach with three different gRNAs designed to knock down Mecom in a partially differentiated acinar cell line (266-6) to study the role of Mecom in acinar cell dedifferentiation. Only gRNA3 provided evaluable knockdown (KD) (Supplementary Fig. 2) and was used further.  ElaCreERT;Sox9 f/f versus control (ElaCreERT) littermates by administering tamoxifen three times over a period of 5 days, followed by a 2-week washout after the start of tamoxifen treatment. Acute pancreatitis is induced by administering caerulein by 7 hourly injections for 3 days over a period of 5 days. Mice are sacrificed on day 4 after caerulein administration. Immunohistochemical staining and for SOX9 and RNA in situ hybridization of Mecom mRNA expression and quantification in mouse pancreatic tissue after experimentally induced acute pancreatitis in control (ElaCreERT) and acinar-specific Sox9 knockout (ElaCreERT;Sox9 f/f ) mice. Scale = 50 µM. (mean ± SD; N = 3; **p < 0.01; unpaired two-tailed t-test).
RISH for Mecom in this loss-of-function condition (Fig. 4a) confirmed a significant reduction of mRNA expression (0.48 ± 0.06 Mecom mRNA dots/cell in Mecom KD cells compared to 0.20 ± 0.05 Mecom mRNA dots/cell in control cells; 55.50 ± 14.43 % reduction; **p = 0.0033). In the Mecom KD condition, the cells were not able to organize j Collagen IV immunofluorescent staining in red on 266-6 NTgRNA and gRNA3-transduced cultured cells. k Immunoblot for phospho-ERK and total ERK and quantifications of pERK relative to total protein stain. (mean ± SD; N = 3; *p < 0.05; Unpaired two-tailed t-test).
into the typical "acinar dome-like" structures and seemed to lack the capacity to attach properly to the culture dish compared to the cells transduced with non-targeting gRNA (NTgRNA) (Fig. 4b), suggestive of loss of proper cell-cell contacts. We also observed a decrease in absolute cell number (Fig. 4c), an overall diminished cell viability (Fig. 4d), and a significant increase in cleaved caspase 3+ cells (Fig. 4e), all indicative of increased cell death and loss of adhesive capacity.
We further studied the effects of Mecom KD in this cell line by RNAseq analysis. Differential gene expression analysis (Fig. 4f) revealed downregulation of genes enriched in "Focal adhesion", "ECM receptor interaction" and "Pathways in cancer" (Fig. 4g), underscoring our observations of diminished dome formation. We also compared the differentially expressed genes from the Mecom KD condition with a publicly available MECOM ChIP-Seq dataset [49] and found 9.89% overlapping genes (Fig. 4h) to make an informed decision on which genes to consider further. In particular, we assessed integrins that link the extracellular matrix (ECM) components such as collagen to the cytoskeleton, hypothesizing that Mecom might compromise cell-cell interaction by diminished Collagen IV-Integrin a2 interaction and downstream ERK-mediated signaling (Fig. 4i), the latter considered crucial in acinar cell dedifferentiation [50]. Diminished expression of collagen type IV and reduced ERK signaling upon Mecom loss-of-function was confirmed by immunofluorescent staining (Fig. 4j, k).
In conclusion, Mecom is indispensable to maintain cell adhesion and cell viability in a partially differentiated acinar cell line.
Next, we crossed ElaCreERT deletor mice with mice carrying a floxed exon 4 of Evi, an exon which is conserved in all transcript variants [24] (Fig. 5a). We studied effects in acinar cells isolated from these mice using an established cell culture model of acinar cell dedifferentiation, same as the human cell culture model described above [13]. Immunohistochemical staining for MECOM showed 42.71 ± 13.42% fewer MECOM positive cells in freshly isolated (day 0) MECOM KO exocrine cells, consistent with previous reports on the recombination efficiency of the Ela-CreERT strain [51][52][53]. Interestingly, by day 4, we detected only 23.89 ± 13.42% fewer MECOM positive cells in the KO condition suggesting selective loss of MECOM deficient cells (Supplementary Fig. 3).
We observed that after cell isolation the control cells remained organized in small acinar spheroid-like structures while in the MECOM KO condition they often presented as single cells. After 4 days of culture, clusters did form in the ElaCreERT;Mecom f/f mouse cells but cluster diameter was significantly smaller compared to those of ElaCreERT control mice (57.21 ± 15.73% reduction in average diameter) (Fig. 5b,  e), suggestive of diminished cell adhesion and cell death of the single cells. In our experience, isolated acinar cells cannot survive as single cells (e.g., after FACS purification).
These observations were confirmed by shRNA-mediated knockdown in cultured human exocrine cells in culture. At 24 h, 4, and 7 days after transduction, average cluster diameter was significantly decreased (Supplementary Fig. 4), underscoring the compromised cell-cell adhesion.
In conclusion, MECOM is indispensable for the maintenance of cell-cell adhesion and cell viability in primary cultures of mouse and human acinar cells.

Mice with Mecom-deficient acinar cells show more acinar cell death and prolonged immune infiltration after acute pancreatitis
Since the primary cell culture models above recapitulate acinar cell dedifferentiation as it occurs during pancreatitis [11][12][13], we subjected to control and MECOM KO mice to caerulein-induced acute pancreatitis. In wild-type mice, acute pancreatitis peaks around day 4 and resolves by day 11 (Fig. 6a). We observed no macroscopic differences in the pancreas nor in weight of the pancreas or bodyweight when comparing MECOM KO with control mice ( Supplementary  Fig. 5). Haematoxylin-eosin staining however showed more inter-acinar spaces at day 11 in the MECOM KO mice (Fig. 6b, f). Similar to the phenotype observed in the 266-6 cells (Fig. 4) and the ex vivo cultures (Fig. 5), this was preceded by an increased incidence of apoptosis ( Fig. 6c, g). Since we observed more inter-acinar space in the MECOM KO mice, we analyzed immune cell infiltration. At day 11, CD3 positive area in MECOM KO mice remained at a level comparable to the acute phase of pancreatitis (day 4), whereas in the control mice it was significantly decreased (Fig. 6e, i). F4/80+ macrophages showed a similar trend, and CD19+ B-cells were only present in the surrounding lymph nodes ( Supplementary  Fig. 6). Remarkably, no difference in Masson's Trichrome blue staining could be detected between control and MECOM KO mice (Fig. 6d, h), indicating that the interacinar space was not filled with collagen. The histological pattern showing the accumulation of excessive liquid in the interstitial space was consistent with that of edema. Also, we did not observe any statistically significant changes in acinar-to-ductal transdifferentiation markers at day 11 (not shown).
Altogether, these results indicate that MECOM-deficiency in pancreatic acinar cells under the experimental stress of acute pancreatitis leads to increased acinar cell death with ensuing immune cell infiltration and edema.

Discussion
When subjected to stress, such as occurring during chronic pancreatitis, acinar cells dedifferentiate and acquire a ductal cell-like phenotype, rendering them prone to Kras induced transformation, in contrast to the native duct cells which seem more resistant to neoplastic development. Where studies in the past focused on the acinar cell plasticity and on how the dedifferentiated acinar cells become duct-like [13], the inherent differences between the two cell populations that may underly a different propensity to tumor formation remained unclear. Therefore, we profiled, for the first time to our knowledge, the transcriptome of human dedifferentiated acinar cells versus their native duct cell counterpart. This comparative transcriptomic analysis provided several new insights, including the fact that both cell populations were enriched in "Pathways of cancer" although the related genes differed in both fractions. We identified MECOM, a transcription factor known to be involved in (pancreatic) cancer, as the most uniquely expressed gene in the dedifferentiated acinar cells.
MECOM is known to regulate many aspects of tumorigenesis including differentiation, proliferation, and apoptosis [17][18][19]. It is a regulator of stomach-specific genes, some of which become upregulated in dedifferentiated acinar cells [54], and was identified in a specific cluster of acinar cells during PDAC development in a scRNAseq experiment recently published [55]. Previously, a role for MECOM in pancreas cancer was proposed [20,21,43]. Here, we focused on its role in acinar cell dedifferentiation. We uncovered that MECOM is uniquely expressed in dedifferentiated mouse and human acinar cells and absent from native duct cells, both in cell cultures and in tissue samples. Notably, MECOM expression levels appeared even higher in embryonic mouse pancreas, specifically in the pro-acinar tip cells again suggesting a relation to an immature acinar cell phenotype. Fig. 6 Mecom depletion in acinar cells in vivo causes pronounced cell death during recovery after caeruleininduced acute pancreatitis. a Acinar-specific deletion of MECOM is performed by administering tamoxifen three times over a period of 5 days, followed by a 2-week washout period after the start of tamoxifen treatment. Acute pancreatitis is induced by administering caerulein by 7 hourly injections for 3 days over a period of 5 days. Mice are sacrificed at day 4 and day 11 after caerulein administration. b, f H&E staining, and quantification of acinar/nonacinar area ratio relative to day 4. c, g Cleaved caspase 3 immunohistochemical staining and quantification at day 4. d, h Collagen staining and quantification relative to day 4. Scale = 50 µM. e, i CD3 staining and quantification relative to day 4 for ElaCreERT and ElaCreERT;Mecom f/f mice. Scale = 50 µM. (mean ± SD; N = 6; **p < 0.01; ns = nonsignificant. two-way ANOVA with post-hoc Bonferroni correction).
Interestingly, SOX9 is a known regulator of acinar cell dedifferentiation and a driver of pancreatic tumor formation [23]. Here, SOX9 expression correlated with that of MECOM. An interplay of MECOM and SOX9 has been described in other tissues [46,47] and we confirmed that SOX9 transactivates the Mecom enhancer in an acinar cell context. Still, it requires further investigation why SOX9 does not transactivate MECOM expression in duct cells where SOX9 expression is the highest. SOX9 likely needs other binding partners, that are absent in duct cells, to activate MECOM expression.
We observed a clear phenotype when culturing MECOM deficient acinar cells (266-6 cells and primary mouse and human acinar cells), i.e., impaired formation of acinar domes or cell clusters, suggesting impaired cell adhesion and higher rates of apoptosis. We showed that MECOM depletion reduced the cell-cell interactions through Itga2 and its ligands such as Col4a2. In our Mecom KD cells, the expression of several integrins (Itga2, Itga4, Itga5, and Itga9) was decreased. Importantly, available ChIP-Seq data of MECOM in human ovarian carcinoma cells also showed direct binding to the promotor regions of Col4a2 and Itga2 [49]. We confirmed this in Mecom KD cells, where reduced activity of the ERK signaling pathway could be directly linked to reduced acinar cell dedifferentiation, a process for which this signaling pathway is critical [12,50]. In a caerulein-induced acute pancreatitis model in MECOMdeficient mice, we again make similar observations and demonstrated a prolonged immune cell infiltration, similar as reported recently by Ye et al. [43].
We propose that normal acinar cells have two options when they are subjected to stress, either to subside in an "incognito" reversible dedifferentiated state, or to die. MECOM seems indispensable for the former option. Ye et al., who, like us, did not provide formal proof by lineage tracing, claimed that MECOM-deficient acinar cells transdifferentiate more extensively during pancreatitis [43], while we suggest that MECOM-deficient acinar cells undergo apoptosis and therefore cannot contribute any longer to the CK19+ transdifferentiated cell pool. This is supported by our observations of increased inter-acinar spaces filled with immune infiltrates and fluid (edema) in the acute pancreatitis model. Such phenotype has also been reported in EGF knockout mice [56]. We also did not detect any increase in collagen deposition around the recovering acini, which clearly differs from acinarto-ductal transdifferentiation that is accompanied by typical collagen deposition [57], further supporting our conclusion.
Altogether, these results show that dedifferentiated acinar cells have a clearly distinct gene expression profile compared to native duct cells. Under stress, MECOM expression is turned on, downstream of SOX9, where it prevents acinar cell death and allowing cellular dedifferentiation.
Author contributions EB, EW, IH, LB, and IR conceived and designed the research; EB, EW, JB, EM, MVB, YH, and IH performed experiments; EB, EW, IH, LB, and IR drafted the manuscript; DLDP, MK, SG, and PJ contributed to animals and human donor material.
Funding This work was supported by Prijs Kankeronderzoek-Oncologisch Centrum VUB to EB, the Research Foundation Flanders (FWO) (Odysseus project and FWO research project to IR (FWOODYS12 and FWOAL931) and postdoctoral fellowship to IH (FWOTM652)) and European Commission (H2020 681070), granted to the Diabetes Research Center of Vrije Universiteit Brussel.

Compliance with ethical standards
Ethics approval Biospecimens, transcriptomic and clinical data were collected by the Australian Pancreatic Cancer Genome Initiative (APGI, www.pancreaticcancer.net.au), which is supported by an Avner Pancreatic Cancer Foundation Grant (www.avnersfoundation. org.au). Informed consent was obtained from all patients. Use of clinical samples and data were in accordance with national ethical guidelines and regulations in Australia (HREC/11/RPAH/329-Sydney Local Health District-RPA Zone, protocol X11-0220) and in Belgium (UZ Brussel, B.U.N.143201732468). Consent for the use of residual donor material for research was obtained according to the legislation in the country of organ procurement. All animal experiments were approved by the Ethical Committee for Animal Testing at the Vrije Universiteit Brussel (#17-637-1).

Conflict of interest The authors declare no competing interests.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.