Pancreatic cancer cells express high amounts of AChE
The role of ACh as a neurotransmitter in the parasympathetic nervous system (PSNS) is well-known, but its non-neuronal function as a local signaling molecule that influences basic cell functions such as apoptosis and proliferation, is often underestimated [9]. The cholinergic system comprising ACh and its synthesizing enzyme, choline acetyltransferase (ChAT), as well as its degrading enzyme, acetylcholinesterase (AChE), have been detected in several cancer entities, such as colon, liver and lung cancer [13,11,19]. However, the amount of AChE expression in PCa has not been genuinely analyzed before. By employing immunostaining against AChE, we demonstrated mild to weak staining in normal acinar cells, pancreatic islets and intrapancreatic nerves (Figures 1a-c). Notably, immunostaining increased in precursor lesions of PCa, i.e. pancreatic intraepithelial neoplasia (PanIN) lesions (Figures 1d-f). The strongest staining was found in areas of invasive ductal adenocarcinoma with prominent staining localized in the peri- and sub-membranous cell compartment of cancer cells (Figures 1g-i). Combined staining of AChE and the PCa cell marker Cytokeratin 19 (CK19) using double-immunofluorescence (IF) staining confirmed strong co-localization of CK19 with AChE (Figures 1g-l), underlining the specificity of AChE expression in cancer cells.
In order to further quantify the expression of AChE in PCa cells (PCC), we isolated dorsal root ganglia (DRG) from postnatal C57BL/6J mice. DRG neurons incorporate afferent neuronal signals of different qualities and express AChE [20]. Here, immunoblot analysis of commonly used human PCC lines, T3M-4, SU.86.86 and Panc-1, as well as colon carcinoma cell lines SW620 and DLD-1, and the glioblastoma cell line LN229, revealed high levels of AChE expression when compared to DRG of C57BL/6J mice (Figures 1m-n). Collectively, these results demonstrate that PCCs express high amounts of AChE in a specific manner.
AChE inhibition suppresses PCC growth in vitro
Next, we sought to explore the effect of AChE inhibition on cancer cell growth. Administration of physostigmine or pyridostigmine, two commonly used AChE inhibitor, led to significant inhibition of viability in T3M-4, but not in SU86.86 cells (T3M-4-physostigmine: 189±99% with 0.1ng/µl vs. 217±110% with solvent with 0.1ng/µl; T3M-4-pyridostigmine: 191±90% with 3ng/µl vs. 220±117% with solvent, Figure 2a-f). Similarly, following administration of acetylcholine (ACh), or of carbachol, a direct muscarinic receptor agonist, a dose-dependent growth reduction was evident for both cell lines (ACh-T3M-4: 242±165% with 1,000µM vs. 384±374% with solvent; ACh-SU86.86: 212±116% ng/µl with 1,000µM vs. 297±220% with solvent; Carbachol-T3M-4: 348±285% with 100µM vs. 511±391% with solvent, Carbachol-SU86.86: 260±145% with 100µM vs. 421±270% with solvent, Figures 2c&f-h). To exclude a difference in the basal AChE activity of the cell lines SU86.86 and T3M4, we performed a colorimetric AChE activity assay based on the Ellman method, which revealed no difference in the basal AChE activity of the two cell lines (Figure 2i). These results confirmed that non-neuronal cholinergic signaling decelerates PCC growth in vitro, which was achieved through increasing ACh availability, through AChE inhibition, or through direct muscarinic stimulation.
Cholinergic activation inhibits PCC invasion in vitro and in vivo
In order to test the effect of AChE inhibition in PCCs on their invasive potential, we performed Matrigel-based invasion assays with SU86.86 cells (Figure 2j-l). Also here, in a dose dependent manner, physostigmine inhibited invasion when applied at the concentration of 30ng/µL (70.4±21.6% of control, Figure 2i). Pyridostigmine inhibited PCC invasion at intermediate and high concentrations of 10ng/µL (71.4±23.4% of control) and 30ng/µL (45.5±13.0% of control), respectively (Figure 2k), and ACh limited invasion significantly at the higher concentration of 1000µM (43.2±26.6% of control, Figure 2l). Collectively, these findings indicated that both AChE inhibition and direct cholinergic activation through ACh inhibit PCa cell invasion in vitro.
In order to test whether indirect cholinergic activation also reduces pancreatic cancer growth in vivo in animals with intact vagal innervation, we utilized a xenograft mouse model, which was generated by injecting PCC into Crl:NMRI-Foxn1 nu/nu nude mice. The effect of AChE inhibition was investigated in a prophylactic and a therapeutic treatment group (Figure 3a). Both groups received daily subcutaneous injections of low- or high-dose physostigmine or pyridostigmine. The prophylactic treatment arm started simultaneously with tumor induction, whereas therapeutic treatment started 1 week after tumor induction. After a 4-week injection period, tumor size was assessed. A significant decrease in the tumor size was observed in animals that received physostigmine or pyridostigmine prophylactically (p) in a dose-dependent manner (Control /saline=22.0±5.7mm vs. p-physo-high=14.5±1.3mm vs. p-pyrido-high = 11.8±2.5mm, Figure 3b). However, therapeutic administration of these indirect parasympathomimetics to established xenograft tumors did not influence tumor size over the course of the treatment (Figure 3c). In order to evaluate the invasive potential of PCC in vivo, we assessed the proportion of xenografted mice that showed penetrating tumor growth into neighboring organs, i.e. kidneys and lungs, following therapeutic or prophylactic treatment with AChE-blockers. In animals in which AChE was blocked prophylactically, only 18% of the specimens showed penetrating tumor growth, whereas 80% of the control group showed tumor infiltration into neighboring organs (Figure 3d, Supplementary figure 1). These findings suggested a partially tumor-suppressing effect of non-neuronal cholinergic activation in vivo, yet only in the context of developing tumors.
Indirect parasympathomimetic agents suppresses immune cell infiltration by tumor-associated macrophages (TAM) and reduce serum cytokine levels in xenografted PCa mice
Non-neuronal cholinergic signaling is also involved in the regulation of the immune system as most immune cells express ACh, AChE, and muscarinic receptors [21]. In this context, we aimed to analyze if indirect cholinergic activation not only has a direct cancer-suppressive effect, but also modulates the immune response in the tumor microenvironment. Therefore, we quantified tumor-associated macrophage (TAM) amounts in pancreatic tumors of the xenograft mouse model. Tumor-associated macrophages are a subpopulation of cytokine-secreting monocytes and have been implicated in playing an important role in the tumor microenvironment (TME). Upon activation, TAM differentiate into M1 or M2 polarized macrophages and release abundant cytokines [22]. Here, we performed double-IF for the murine macrophage marker f4/80 and CD45 (Figure 3e). Our analysis demonstrated a reduction of CD45+/f4/80+ -TAM infiltration in murine tumors of physostigmine or pyridostigmine treatment groups (Saline-prophl.: 6.6±0.8 cells/m2, Physostigmine-high: 2.7±0.4 cells/m2, Pyridostigmine-high: 2.9±0.7 cells/m2, Figure 3e). As cholinergic activation is known to exert a systemic anti-inflammatory effect (“the cholinergic anti-inflammatory pathway”), we then assessed the serum levels of the cytokines interleukin 6 (IL6), interleukin 10 (IL10) and tumor necrosis factor-alpha (TNFalpha) in the xenografted mice (Figure 3f). Here, we detected a massive suppression of the levels of all these cytokines in all treated groups, regardless of the dosage of treatment, when compared to saline-treated controls (IL6 control: 582.8±428.0 pg/ml, IL6 treated: 25.4±21.5.0 pg/ml; IL10 control: 488.2±371.6 pg/ml, IL10 treated: 19.4±26.1 pg/ml; TNFalpha control: 540.2±398.4 pg/ml, TNFalpha treated: 10.4±29.3 pg/ml, Figure 3F). Collectively, these data suggested a prominent suppression of tumor-associated local and systemic inflammation markers in the xenografted PCa mice upon physostigmine or pyridostigmine treatment.
Cholinergic activation leads to intracellular p-ERK1/2 and p-p38 MAPK inhibition and induces cell cycle arrest
In order to determine molecular mechanisms responsible for growth and invasion inhibition in PCC upon AChE inhibition, we performed a phospho-kinase antibody array for screening that enables the profiling of 43 different human kinases in two experimental arms (Figure 4a-b). Here, we compared intensity of phosphorylation of these multiple kinases in T3M4 cells treated with either physostigmine or left untreated, and used the clues from this initial screen for subsequent validation analyses. Among well described mitogen-activated protein kinases (MAPK) that are known to be widely expressed in PCa and involved in cell proliferation, invasion, cell-survival and cell cycling, we found extracellularly regulated kinase 1 and 2 (ERK1/2), p38, proto-oncogene tyrosine-kinase Src (Src) and 5’-AMP-activated protein kinase α (AMPKα) to be altered under the treatment (Figure 4a-c) [23]. In validation immunoblots with T3M-4 cells, we confirmed the decrease in phosphorylated ERK1/2 (pERK) levels, particularly after treatment with high-dose pyridostigmine (61.5±13.9% of control). This effect was more pronounced for SU86.86 cells, which, after treatment with physostigmine or pyridostigmine, exhibited even more obviously diminished pERK1/2 levels in a dose-dependent manner at both mid-level and high concentrations (physostigmine-mid: 80.5±2.6% of control, physostigmine-high: 69.2±7.9% of control, pyridostigmine-mid: 70.2±8.2% of control, pyridostigmine-high: 60.3±11.8% of control, Figure 4e). As an essential component of the MAPK signal transduction pathway, p38 reacts to extracellular stimuli and mediates cellular responses [24,25]. In our experiments, phosphorylation of p38 was abolished upon administration of low (43.7±8.2% of control), mid- (53.4±17.3% of control) and high (69.3±10.9% of control) physostigmine concentrations, but not via pyridostigmine (Figure 4f). However, following treatment with either of these drugs, there was no significant change in the amount of intracellular p-Src nor p-AMPKα (Figure 4g-h). In summary, our experiments demonstrated that inhibition AChE reduced ERK phosphorylation.
To investigate whether cell cycle progression of PCC is also affected by AChE inhibition, we performed a propidium iodide-(PI-) based, flow cytometric cell cycle analysis (Figure 4i-j). After administering ACh, significantly more PCCs were observed in the G1/0 phase (ACh: 56.1±3.2% vs. Control: 47.8±0.9%) and fewer cells in the S-phase (ACh: 10.0±0.1% vs. Control: 13.0±0.8%, Figure 4i-j). Physostigmine did not alter G1/0-phase amount, but reduced S-Phase cell-count (8.4±1.7%), and pyridostigmine enhanced G1/0-phase count (55.3±1.5%), but did not alter the S-Phase cell count (Figure 4i-j). No significant difference in cell count was noted for cells in G2/M-Phases, however following all treatments a trend towards lower cell counts was observed. Overall, we thus detected a cell cycle arrest in G1/0 phase following AChE inhibition.
Adjuvant indirect cholinergic treatment does not impact survival in a resectable PCa mouse model
In order to translate our findings into a clinically relevant setting, we used a novel R0-resectable, genetically induced PCa mouse model [15]. In this model, plasmids containing the Sleeping Beauty (SB) transposase SB13, a Kras-G12V encoding transposon, and the Cre recombinase were injected and electroporated into the pancreatic tail of p53floxed mice (p53fl/fl) via mini-laparotomy [15]. Upon activation of the Cre recombinase, tumor formation was initiated in a local fashion (the Pfl model), which is in contrast with the multilocular tumor growth of classical genetically induced mouse models of PCa (Figure 5a-b). Three weeks after the tumor induction, the animals developed macroscopically visible tumors. After pancreatic tail resection, mice received adjuvant chemotherapy with gemcitabine. Here, mice of the Pfl-genotype exhibited a median survival of 41 days (Figure 5c). When adjuvant gemcitabine treatment was combined with physostigmine, median survival was 32 days. Combinational therapy with gemcitabine and pyridostigmine was associated with a median survival to 39 days (Figure 5c-d). The most common reason for death was combined local and distant (hepatic or peritoneal) recurrence. Thus, the AChE inhibitors did not generate any additive survival benefit in this innovative, adjuvant therapy setting.
Correlation of tissue AChE and ChAT expression with clinicopathological variables in human PCa
Lastly, to compare our findings from this translational mouse model to human PCa, we analyzed the expression patterns of AChE and the ACh synthesizing enzyme ChAT in human PCa tissues (n=39) by semiquantitive immunostaining scores and correlated to clinicopathological variables of the corresponding patients. Furthermore, high vs. low scores of AChE immunostaining (separated by the median score) did not result in any difference in the overall survival rate of the PCa patients (Figure 5e). Accordingly, tissue expression scores of AChE did not associate with different UICC tumor stages (Figure 5f). Interestingly, higher tumor grades, i.e. a poor tumor differentiation, were associated with significantly lower AChE IHC scores (G1: 1.5±0.2, G2: 1.0±0.2, G3: 1.0±0.4, Figure 5g), suggesting a spontaneous loss of AChE in increasingly aggressive PCa. This was surprising, as we originally hypothesized that high AChE expression, i.e. diminished cholinergic input, would be associated with worse survival and poor differentiation. As this was not the case, we quantified the expression level of choline acetyltransferase enzyme (ChAT), which catalyzes the formation of acetylcholine, in the nerves of these tissues via immunohistochemical scoring. Correlation of ChAT expression levels to tumor stage revealed that high expression levels ChAT were indeed associated with low tumor stages (r2: 0.20, p=0.049, Figure 5h). Hence, we concluded that advanced tumor stages were characterized by low cholinergic input due to low ChAT expression, and yet also by suppression of the degrading enzyme AChE. This simultaneous suppression of the ACh-synthesizing and ACh-degrading intrinsic mechanisms in PCa may explain the lack of a prognostic effect of tumor AChE levels in established mouse and human PCa.