MIA PaCa-2 and PANC-1 – pancreas ductal adenocarcinoma cell lines with neuroendocrine differentiation and somatostatin receptors

Studies using cell lines should always characterize these cells to ensure that the results are not distorted by unexpected morphological or genetic changes possibly due to culture time or passage number. Thus, the aim of this study was to describe those MIA PaCa-2 and PANC-1 cell line phenotype and genotype characteristics that may play a crucial role in pancreatic cancer therapeutic assays, namely neuroendocrine chemotherapy and peptide receptor radionuclide therapy. Epithelial, mesenchymal, endocrine and stem cell marker characterization was performed by immunohistochemistry and flow cytometry, and genotyping by PCR, gene sequencing and capillary electrophoresis. MIA PaCa-2 (polymorphism) expresses CK5.6, AE1/AE3, E-cadherin, vimentin, chromogranin A, synaptophysin, SSTR2 and NTR1 but not CD56. PANC-1 (pleomorphism) expresses CK5.6, MNF-116, vimentin, chromogranin A, CD56 and SSTR2 but not E-cadherin, synaptophysin or NTR1. MIA PaCA-1 is CD24−, CD44+/++, CD326−/+ and CD133/1−, while PANC-1 is CD24−/+, CD44+, CD326−/+ and CD133/1−. Both cell lines have KRAS and TP53 mutations and homozygous deletions including the first 3 exons of CDKN2A/p16INK4A, but no SMAD4/DPC4 mutations or microsatellite instability. Both have neuroendocrine differentiation and SSTR2 receptors, precisely the features making them suitable for the therapies we propose to assay in future studies.


Results
Immunohistochemistry. Microscopic  Flow cytometry. Immunophenotyping of MIA PaCa-2 and PANC-1 cell lines revealed differences in morphology and immunohistochemistry. Flow cytometry was employed to clarify whether, in morphological terms, there were in fact two cell populations in MIA Paca-2 and three in PANC-1. This technique was also used to characterize these cell lines in terms of the presence of pancreatic stem cell markers, such as CD24, CD44, CD133/1 and CD326.
When performing the same study in the PANC-1 cell line (Fig. 4), while in optic microscopy and immunohistochemistry we observed three distinct populations, flow cytometry dot plot FSC/SSC revealed the existence of only one population of viable cells (P1). In fact, in 10,000 events, a population (P1) with 6,888 events was obtained (Fig. 4A), corresponding to 68.9% of total cells, with an MFI of 1,192, supporting the possibility of the existence of a pleomorphic cell line. To investigate the presence of stem cells using CD24 FITC, CD44 APC, CD133/1 PE and CD326 PE-Cy7, we have characterized this population (P1) in relation to each marker and performed the gates P2, P3, P4 and P5. P2 corresponds to a subpopulation expressing CD24 (Fig. 4B), comprising 23.4% of P1, with an MFI of 3,502; P3 corresponds to a subpopulation expressing CD133/1 (Fig. 4C), 0.7% of P1, with an MFI of 3,324; P4 corresponds to a subpopulation expressing CD326 (Fig. 4D), 49.1% of P1, with an MIF of 2,068; and P5 corresponds to a subpopulation expressing CD44 ( The MIA PaCa-2 cell line (Table 1 and Fig. 1) expresses CK5.6, AE1/AE3, E-cadherin, vimentin, chromogranin A, synaptophysin, SSTR2 and NTR1, but not CD56. These cells are epithelial as they express CK5.6, AE1/AE3 and E-cadherin, with mesenchymal characteristics because they express vimentin, neuroendocrine  Table 2. PANC-1 immunohistochemical profile. Cut off: − (0%); + (< 10%); + + (10-75%); + + + (> 75%); *polarized cytoplasm. differentiation since they express chromogranin A and synaptophysin, and hormonal receptors because they express SSTR2 and NTR1. The presence of SSTR2 and NTR1 may enable peptide receptor radionuclide therapy to be used in this cell line. PANC-1 cells (Table 2 and Fig. 2) do not express E-cadherin, synaptophysin or NTR1. They are cells with epithelial characteristics since they express CK5.6 and AE1/AE3, mesenchymal characteristics because they express vimentin, neuroendocrine differentiation as they express chromogranin A and CD56, and hormonal receptors because they express SSTR2. The presence of SSTR2 may allow peptide receptor radionuclide therapy to be used in this cell line.
If we compare the immunohistochemical pattern of these two cell lines (Tables 1 and 2), we notice that it is different, not only due to the absence of expression of certain markers and receptors, but also because of the variable expression of others. Thus, in the large cells of MIA PaCa-2 there is a more pronounced expression of CK5.6, AE1/AE3, E-cadherin, vimentin, synaptophysin, SSTR2 and NTR1. The large cells of PANC-1, on the other hand, have a more pronounced expression of chromogranin A and CD56. In the small cells of MIA PaCa-2 there is a more pronounced expression of CK5.6 and AE1/AE3, and to an even greater extent E-cadherin, chromogranin A and NTR1. In the small cells of PANC-1 there is a more pronounced expression of vimentin. This difference in cell marker and receptor expression between large and small cells in the same cell line, and between large and small cells in both cell lines, enhances the concept of population variability. Concerning NED, MIA PaCa-2 expressed chromogranin A and synaptophysin (Table 1), and PANC-1 expressed chromogranin A and CD56 (Table 2). In pancreatic carcinoma NED has been closely associated with tumor behavior: patients with tumors harboring NED have a better overall survival and NED seems to be an independent predictor of survival after surgery 7 . Both cell lines expressed chromogranin A, and this observation reveals NED to be associated with a better prognosis 32,33 .
Both MIA PaCa-2 and PANC-1 cells expressed epithelial markers CK5.6 and AE1/AE3, and the mesenchymal marker vimentin, allowing them to be characterized as epithelial-mesenchymal cells. The activation of an EMT program is the essential mechanism for the acquisition of a malignant phenotype by epithelial cancer cells, and the hallmark of EMT is the loss of the epithelial homotypic adhesion molecule E-cadherin and gain of mesenchymal marker vimentin 34 . PANC-1 had no E-cadherin expression. It has been demonstrated that EMT contributes to drug resistance in pancreatic cancer and that increased expression of E-cadherin is associated with improved survival in several tumor types 35 . In some reports of migration assays, it was demonstrated that migration of PANC-1 cells was greater than MIA PaCa-2 cells on transwell ® plates coated with collagen type I 36 . Although not conclusive, the invasive behavior through matrigel ® was also demonstrated for MIA PaCa-2 and PANC-1 36 . Our results indicate that MIA PaCa-2 and PANC-1 have EMT potential but that of PANC -1 is superior.
Concerning stem cell marker expression, both cell lines expressed population heterogeneity, with MIA PaCa-2 ( Fig. 3) cells expressing CD24 − CD44 +/++ CD326 −/+ CD133/1 − . It seems that the large cell population has the phenotype CD24 − CD44 ++ CD326 − CD133/1 − and the small cell population, CD24 − CD44 + CD326 −/+ CD133/1 − . We also found that MIA PaCa-2 had two relevant CD44 + CD326 − (48%) and CD44 + CD326 + (35.5%) populations (Fig. 3I). The PANC-1 cell line expressed the phenotype CD24 −/+ CD44 + CD326 −/+ CD133/1 − (Fig. 4). CD24 + CD44 + CD326 + cancer cells display the ability to self-renew, generate different progeny and recapitulate the phenotype of the tumor from which they were derived 37 . Cells negative for all three markers are not able to initiate pancreatic cancer until 10 4 or more cells are implanted 38 . Li et al. 38 demonstrated that cells expressing CD24 + CD44 + CD326 + formed 6 tumors in 12 (50%) NOD/SCID mice after they were injected with 100 cells subcutaneously in their flank. With the single marker CD44 + or with the dual markers CD44 + CD326 + , 4 out of 16 (25%) animals developed tumors when injected with as few as 100 cells, and when the number of cells injected was increased (500 > 10 3 > 10 4 ), the dual marker combination CD44 + CD326 + resulted in an enhanced tumorigenic potential compared with the single marker CD44 + . CD24 + cells are less tumorigenic than CD44 + cells, with 1 out of 16 (6.25%) animals developing tumors when injected with as few as 100 cells. With CD326, 8 out of 18 (44.4%) and 1 out of 18 (5.6%) animals developed tumors when injected with 500 CD326 + and 500 CD326 − cells respectively. 10 3 CD24 − cells were needed to generate a tumor when injected in 16 animals. We may presume that PANC-1 cells CD24 + CD44 + CD326 + are more tumorigenic than MIA PaCa-2 cells CD24 − CD44 + CD326 + , and that within MIA PaCa-2 the subpopulation CD44 ++ CD326 + is more tumorigenic than the subpopulation CD44 + CD326 − . The surface marker CD133 was first reported to have the properties of CSCs in brain and colon cancer 39 . Results from Hermann et al. 40 revealed that 500 CD133 + cells injected in NOD/SCID mice were able to generate tumors. Lee et al. 41 demonstrated that CD133 + cells had a higher tumorigenic potential and higher rates of metastasis to the lung than CD44 + cells, suggesting that CD133 is implicated in the aggressive behavior of pancreatic cancer. Neither MIA PaCa-2 nor PANC-1 expressed CD133, which seems to reduce their potential to metastasize. Despite these results, we should point out that the presence of stem cells cannot be inferred purely from surface marker expression. It is also known, for example, that the expression of the markers CD24, CD44 and CD326 varies with the local microenvironment or niche 42 . Functional analyses should therefore be performed, such as sphere formation assays 43 or tumorigenesis in a transplantation setting in immunocompromised mice 44,45 .
KRAS, CDKN2A/p16 INK4A , TP53, SMAD4/DPC4 and MSI analyses were used to throw light on the genetic carcinogenic mechanisms involved and identify possible sources of resistance to chemo and radiotherapy. Genotyping of MIA PaCa-2 (Fig. 5A) and PANC-1 (Fig. 6A) confirmed the presence of a homozygous (p.G12C; GGT > TGT) and a heterozygous missense mutation (p.G12D; GGT > GAT) in codon 12 of KRAS respectively. This is a hotspot codon well known for interfering with the KRAS GTPase function, keeping the molecule in its GTP-bound state, thus allowing uninterrupted activation of downstream effector pathways such as mitogen-activated protein kinase (MAPK) 46 . KRAS mutations confer drug resistance and lead to aggressive tumor growth and metastasis and a poor clinical outcome 47 . Moreover, tumor cells with a mutant KRAS are more radiation-resistant than are cells with the wild type of KRAS 48 . Some experiments have shown that farnesyltransferase inhibitor drugs, which prevent the post-translation processing of KRAS, essential for its appropriate cell-membrane localization, sensitize KRAS-mutated pancreatic cancer cells to radiation 49 .
CDKN2A, located on chromosome 9p21, encodes two proteins, P16 (INK4) and P14 (ARF). Both gene products have an independent first exon (exon 1-α and exon 1-β , respectively) but share exons 2 and 3, and are translated in different reading frames 50 . These proteins are involved in the negative control of cell proliferation: P16 (INK4) is a cyclin-dependent kinase inhibitor that induces G1 cell-cycle arrest by inhibiting Cdk4/6 and hence keeping the retinoblastoma protein (RB) in its active, dephosphorylated state; P14 (ARF) acts both at G1/S and G2/M phases in a TP53-dependent manner by destabilizing MDM2, the protein that maintains TP53 at low levels 51 . Concerning pancreatic tumors, the CDKN2A/p16 INK4A gene is inactivated in 95% of invasive carcinomas, 40% by homozygous deletion, 40% by an intragenic mutation coupled with loss of the second allele and 10-15% of cases by gene promoter hypermethylation. Tumors with CDKN2A deletion are larger and patients may have a survival period which is significantly shorter 52 . The genotyping of MIA PaCa (Fig. 5B) and PANC-1 (Fig. 5B) confirmed the presence of a homozygous deletion encompassing exons 1, 2 and 3 of the CDKN2A/p16 INK4A gene, resulting in the indirect impairment of two important tumor suppressor proteins, RB and TP53, thus reproducing the effect of inactivating mutations.
TP53 function also seems to be disrupted in the two cell lines by missense mutations. The MIA PaCa-2 cell line showed a pathogenic missense variant in a hotspot codon (p.R248W). The PANC-1 cell line had a common polymorphism in exon 4 (p.P72R; rs1042522), located in the proline-rich domain of TP53 and possibly associated with differences in apoptosis induction 53 , and a pathogenic missense variant in an exon 8 hotspot codon (p.R273H; rs28934576). The two pathogenic variants, in exons 7 and 8, correspond to the DNA binding domain of TP53 and are known to disrupt its function as a transcription activator 54,55 . There is also some evidence that some TP53 mutations, including R273H, may induce gain of function effects, as knock-in mice expressing this mutation showed increased metastases compared with TP53 knock-out mice 56 and a different cancer distribution. TP53 mutations were found in homozygosity in both cell lines, most probably due to the loss of heterozygosity (LOH) associated with chromosome 17p deletions including the wild-type allele, correlating with the loss of TP53 suppressor function.
No changes were found in MIA PaCa-2 or PANC-1 concerning SMAD4/DPC4. Furthermore, the absence of MSI from all of the five markers analyzed suggests that the DNA mismatch repair (MMR) system is preserved 57,58 in the MIA PaCa-2 and PANC-1 cell lines. In conclusion, MIA PaCa-2 expresses polymorphism and PANC-1 expresses pleomorphism. Polymophism and pleomorphism had not yet been described in previous studies, and they may introduce variations in the distribution of stem cell markers, making tumors more resistant to therapy. However, the association between stem cell markers and a specific type of cell morphology was not proved in this study. If this association exists, the treatment of these tumors may be even more complex than we currently assume, due to their heterogeneity.
It was confirmed that both cell lines have EMT differentiation but since PANC-1 has no E-cadherin, its behavior is more aggressive with a greater metastasizing potential.
Both cell lines express neuroendocrine differentiation and SSTR2, and this had not been established in previous studies. It is precisely these features which make them suitable for pancreatic cancer neuroendocrine chemotherapy and peptide receptor radionuclide therapy. Novocastra Laboratories Ltd., Newcastle, UK) was used as chromogen, according to the manufacturer's instructions. Hematoxylin was used to counterstain the slides, which were then dehydrated and mounted. As negative controls, we used positive controls without the addition of a primary antibody.
Flow Cytometry. Flow cytometry was used for identification and clarification of the cell population for each cell line. MIA Paca-2 and PANC-1 cell lines were dissociated using 0.25% trypsin/0. 53   Balanced Salt Solution (ATCC 30-2101) for 5 minutes at 37 °C. Trypsin was then inhibited with DMEM complemented with a 10% FBS culture medium (MIA PaCa-2 and PANC-1) and 2.5% HS (MIA PaCa-2). Approximately 0.5 × 10 6 cells were transferred to a 5 ml tube (BD Biosciences, San Jose, CA, USA) and washed twice with Dulbecco's Phosphate Buffered Saline (PBS). Cells were resuspended in 500 μ l PBS, and samples were acquired on a FACSCanto II flow cytometer (BD Biosciences) and analyzed with FACSDiva software (BD Biosciences). Thus, it was possible to obtain, select and represent the cells graphically in the form of a dot plot as a function of forward-scattered (FSC) and side-scattered light (SSC). We used the 7-AAD (7-amino-actinomycin-D) as a dead cell marker to confirm the presence of the non-viable population, whereby the livings cells were quantified by exclusion.
KRAS (codons 12, 13 and 61), TP53 (exons 4 to 8) and SMAD4/DPC4 (codifying exons 2 to 11) were studied by sequencing using primers and cycling conditions, as described in Table 3. Sequencing was performed using the ABI PRISM BigDye v1.1 Terminator Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA), the same primers as in PCR, and an ABI PRISM 3130 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Results were analyzed with sequencing analysis software v 5.2 (Applied Biosystems, Foster City, CA, USA). For CDKN2A/p16 INK4A , as the most frequent mutation is the homozygous deletion, we only proceeded to PCR and thus three exons, exons 1-α , 2 and 3, were amplified from the DNA of cell lines and of a healthy control using primers and the conditions described in Table 3. MSI testing was carried out with a multiplex PCR amplification of 5 quasi monomorphic mononucleotide repeats (BAT25, BAT26, NR21, NR22 and NR24) with DNA from the cell lines and from the healthy control. Briefly, 40 ng of genomic DNA were amplified in a 25 μ l reaction using primers at 0.16 μ M, MgCl 2 at 1.5 mM, dNTPs at 200 μ M and 0.2 U recombinant Taq DNA polymerase (Fermentas, Thermo Scientific, MA, USA). Primer sequences and PCR cycling conditions are described in Table 4. A volume varying from 0.1 to 1 μ l of amplified PCR product was added to 25 μ l of formamide and 1 μ l of internal fluorescence standard-sized GS500LIZ, incubated for 5 minutes at 95 °C and then applied to the ABI PRISM 3130 Genetic Analyzer using the POP7 polymer. Automatic fragment analysis was carried out with GeneMapper software v3.7 (Applied Biosystems, Foster City, CA, USA). The presence of three or more mutant alleles was considered to indicate the presence of MSI-high (MSI-H) 59 . The absence of MSI implies the absence of mutations in the MMR system.