Pancreatic acinar cells that generate pancreatic enzymes display the fastest protein synthesis among normal cell types (Leblond et al., 1957). To meet this need the exocrine pancreas accumulates amino acids (AAs) to a greater extent than most other tissues (Longnecker, 1977). This affinity to AAs may have application in the development of pancreas-specific drugs. At the same time, pancreatic toxicity has been observed in animals administered some unnatural AAs or large amounts of several natural AAs. High doses of basic AAs (most often L-arginine) are widely used in animal models of acute pancreatitis (AP), although the mechanisms of the pancreatic damage caused by AAs are not well understood. It has been reported that calcium independent mitochondrial injury precedes other early events in the induction of AP by L-lysine, L-ornithine, and L-arginine (Biczo et al., 2011b, 2018; Kui et al., 2014). However, in some reports the ultrastructural changes in response to toxic doses of AAs were detected earlier in the endoplasmic reticulum (ER) than mitochondrial abnormalities (Kishino and Kawamura, 1984; Kubisch et al., 2006). In a range of experimental models of AP, including caerulein hyperstimulation, ductal infusion of bile acid, or fatty acid plus ethanol, it has been firmly established that sustained cytosolic calcium elevation initiates deleterious events in cells (Ward et al., 1996; Perides et al., 2010; Huang et al., 2014; Wen et al., 2015). Caffeine, which in pancreatic acinar cells blocks calcium signaling mediated by IP₃ receptors, has been found to be a strong protector in these non-AA models of AP (Huang et al., 2017). Downstream of cytosolic calcium overload, inhibition of the mitochondrial permeability transition pore (MPTP) by genetic deletion or pharmacological inhibition of peptidyl prolyl isomerase D (cyclophilin D, encoded by Ppif gene) prevents ATP loss and improves biochemical, immunological and histopathological parameters in non-AA models of AP (Shalbueva et al., 2013; Mukherjee et al., 2016). Recently Ppif^-/- has been shown to reduce the severity of AP induced by L-arginine through reduction of ATP synthesis not dependent on calcium overload (Biczo et al., 2018). The effect of Ppif^-/-, however, has not been tested in acinar cells and mice challenged by other AAs.

L-AAs serve as ligands for family C (or class 3) G-protein coupled receptors (GPCR), including CaSR (calcium sensing receptor), and GPRC6A (G protein-coupled receptor family C group 6 member A) (Conigrave and Hampson, 2006; Wellendorph et al., 2009). These two receptors have been reported to be expressed in the pancreas (Bruce et al., 1999; Racz et al., 2002; Wellendorph and Brauner-Osborne, 2004); CaSR mutations have been linked to chronic pancreatitis (Muddana et al., 2008). In other tissues these receptors respond to tens and hundreds of micromolar concentrations of AAs, with CaSR activated primarily by aromatic AAs such as phenylalanine, tryptophan, and histidine while GPRC6A more sensitive to arginine and lysine (Conigrave and Hampson, 2006; Wellendorph et al., 2009). In the experimental conditions of AP the concentration of AA exceeds 10 mM. The role of CaSR and GPRC6a, however, has never been assessed in relation to AA-induced pancreatic damage.

Since it has been noted that conversion of L-arginine to L-ornithine is an important step in the induction of AP by L-arginine (ARG-AP) (Biczo et al., 2010), we have compared the effects of caffeine, modulators of necroptosis, cyclophilin D knock-out, and allosteric modulators of GPCR class C (CaSR and GPRC6A) on pancreatic acinar cell death and AP induced by three basic AAs L-arginine, L-ornithine, and L-histidine, including on a new model of L-histidine-induced AP (HIS-AP).

Our data demonstrate that in vitro responses to various basic AAs have some important differences. The experiments with caffeine, which in pancreatic acinar cells blocks intracellular calcium signaling mediated by IP₃ receptors (Huang et al., 2017), indicate involvement of calcium signaling in the in vitro responses to L-arginine. However, that does not seem to be relevant to ARG-AP. Surprisingly, L-histidine-induced cell death was exacerbated by caffeine in vitro, yet cyclophilin D knock-out showed protection against this AA both in vitro and in vivo. Altogether these data are consistent with the recent finding (Biczo et al., 2018) that cytosolic calcium was not central in AA-induced toxicity in the pancreas, although caffeine does not inhibit all components of calcium signaling or overload. In our hands, cyclophilin D knock-out could not ameliorate ARG-AP. While in their study (Biczo et al., 2018) the authors used the 3.3 × 3 g/kg regimen, we used a standard 2 × 4 g/kg protocol, as 3 × 3 g/kg regimen led to near 100% mortality in our mice. It has been described previously that rats injected with high doses of L-arginine fall into three groups of weak, normal, and strong responders (Bohus et al., 2008). We believe that such a distribution of responders may be differently affected in Ppif^-/- and Wt mice by unknown factors and eventually create differences in the results with respect to cyclophilin D knock-out. For instance, while doing our experiments, we noticed that the C57BL/6J mice that weighted less than 20 g were predominantly weak responders to L-arginine. Therefore, we decided to use only mice above 25 g in our work. One more factor that could underlie the lack of protection against L-arginine and L-ornithine in Ppif^-/- mice in our experiments was the reported ability of L-ornithine to produce reactive oxygen species in the acinar cells (Chvanov et al., 2015). The effects of oxidative stress on bioenergetics in these cells was later shown to be independent of cyclophilin D (Armstrong et al., 2018). The variability in results produced in AA based models of AP is not uncommon and may come down to the fact that the dose of AAs sufficient to trigger AP is very close to the dose of AAs that cause systemic toxicity and mortality, so that the disease is difficult to control (Kui et al., 2014). As a result the AA dose-dependence of disease severity is quite steep and for some of the parameters is bell-shaped (Tashiro et al., 2001), leading to over-sensitivity of the results to minor changes in parameters and protocols.

We found that in vitro necrotic cell death pathway activation of freshly isolated pancreatic acinar cells by L-arginine can be reduced by caffeine, a calcilytic NPS-2143, a calcimimetic calindol and a GPRC6A specific antagonist Cpd1. At the same time, none of these compounds could rescue pancreatic tissue from damage by L-arginine in vivo. This in vivo pattern can be matched by that of L-ornithine in vitro, where damage induced by L-ornithine is equally insensitive to any of these chemicals. The temporal metabolomics study in rats (Bohus et al., 2008) showed that during the first 8 h of ARG-AP there is substantial conversion of L-arginine to L-ornithine and urea. It has also been shown that inhibition of arginase, which converts L-arginine to L-ornithine, ameliorates ARG-AP in rats (Biczo et al., 2010). Thus our data support and provide further evidence that L-arginine to L-ornithine conversion, which is part of the urea cycle, is an important step in the induction of ARG-AP. It has also been documented previously that L-ornithine elicits mitochondrial depolarization and potentially toxic reactive oxygen species production in pancreatic acinar cells (Chvanov et al., 2015). Individuals who lack ornithine transcarbamylase (OTC), functional enzyme that converts L-ornithine into L-citrulline in the mitochondria, are at risk of AP (Anadiotis et al., 2001; Prada et al., 2012; Machado et al., 2013). It is also notable that the mitochondrial isoform of arginase (Arg2) plays a critical role in obesity-associated pancreatic cancer (Zaytouni et al., 2017). These data may signify specific importance of L-ornithine elevation in pancreatic mitochondria in a range of pancreatic pathologies.

It has been known for the last 60 years that non-natural or excessive doses of natural AAs, including ethionine, methionine, azaserine, cycloleucine, β-3-thienyl-DL-alanine, β-3-furyl-DL-alanine and, more recently, L-arginine, L-ornithine and L-lysine can cause lesions of the exocrine pancreas. Longnecker Longnecker (1977) stated that “the pancreas has been shown to concentrate both normal and several abnormal amino acids to a greater degree than most other tissues”. Such a propensity to concentrate AAs may be utilized in drug development by attaching an AA to an active small molecule drug destined for the pancreas. Thus, it is important to understand the mechanisms of AA toxicity, when it arises. Pancreatic acinar cells show the fastest protein synthesis among many cell types (Leblond et al., 1957). Mechanistically, the putative damage to protein synthesis by excess dietary or exogenous AAs was postulated to fall into three categories: imbalance, toxicity, and antagonism (Harper, 1956). Plasma AA imbalance in patients undergoing L-asparaginase chemotherapy (Minowa et al., 2012) may contribute to the high risk of AP associated with this treatment, in addition to the identified calcium overload (Peng et al., 2016). It is feasible that in these patients as well as in the animal models the damage to acinar protein synthesis plays a significant role in the initiation of the AP via an unfolded protein response and ER stress. An antibiotic puromycin, which blocks protein synthesis and generates ER stress, causes necrosis specifically in pancreatic acinar cells and only a few other cell types when injected intraperitoneally in rats (Longnecker and Farber, 1967; Longnecker et al., 1968). In relation to the AA experimental models of AP, ER stress was described as a starting event in ARG-AP in rats (Kubisch et al., 2006). By contrast, another recent comprehensive work identified mitochondrial injury (without prior calcium overload) as an initiator of damage (Biczo et al., 2018). The same mechanism was also suggested in choline deficient, ethionine-supplemented diet-induced AP.

The differences between the patterns of toxicity in the pancreas caused by different AAs have been known previously. For example, an ultrastructural study of pancreatic damage caused by L-arginine in rats highlighted the changes in the ER as preceding those in the mitochondria (Kishino and Kawamura, 1984). By contrast, upon injection of L-lysine in rats the same research group observed the reverse sequence of events (Kitajima and Kishino, 1985). In that respect, ethionine is closer to L-arginine than to L-lysine, because the ultrastructural changes in the ER after ethionine administration come earlier than those in mitochondria (Herman and Fitzgerald, 1962). While the damage caused by ethionine is augmented by a choline deficient diet, malnourishment of the experimental animals increases the pancreatic injury brought about by L-arginine (Takama and Kishino, 1985). However, whereas L- and D-ethionine are equally effective (Wachstein and Meisel, 1953), only the L-enantiomer of arginine has been shown to cause AP (Dawra et al., 2007). The fine differences between the effects of various AAs may indicate existence of individual AA-specific components of pancreatic toxicity in addition to a general one, which presumably unifies L-arginine, L-lysine, L-ornithine and ethionine and is mitochondria-linked (Biczo et al., 2018). Excessive histidine (Frezza, 2018), choline-deficient, ethionine-supplemented diet (Longnecker, 2002) and L-arginine (through NO synthesis) are known to reduce tetrahydrofolate availability. This would interfere with cytosolic and mitochondrial steps of one-carbon metabolism in the pancreas (Balaghi and Wagner, 1995; Horne and Holloway, 1997) and as such represents a shared pathway for these AAs. At the same time, it is feasible that some AA-specific pathways are related to protein synthesis (Weaver et al., 1994). Despite necroptosis having been shown to be an important mode of acinar cell death in AP models induced by CER (Wang et al., 2016) and bile acid (Louhimo et al., 2016), our findings rule out necroptosis as a significant contributor to the necrotic cell death caused by basic AAs.

We were successful in generating AP in mice injected with 2 × 4 g/kg L-histidine free base. Previous assessment of this AA for effects on rats was done with a lower dose of 1 × 3 g/kg L-histidine-HCl and reported no damage to the pancreas (Biczo et al., 2011a). That report highlighted the difficulty of working with L-histidine in vivo arising from its poor solubility, leading to introduction a lot of fluid into the animals. Indeed, we had to injection large volumes of freshly prepared 7% stock solution by microwaving and cooling to physiological temperatures. The model HIS-AP was different from ARG-AP as the severity of the disease caused by L-histidine was markedly decreased in Ppif^-/- mice. The cell death in vitro and in vivo observed in the presence of high levels of L-histidine adds to safety concerns of using histidine-tryptophan-ketoglutarate (HTK) solution for pancreas transplant preservation (Alonso et al., 2008; Troppmann, 2010). HTK solution is normally used cold, and we see no cell death in vitro in the presence of 20 mM L-histidine in our experiments if we drop the temperature from 37°C to room temperature (22–24°C, the data not shown). However, the very high concentration of histidine in HTK solution (300 mM) may still pose a risk for the transplants once the circulation is restored and body temperature reached.

The members of GPCR class C, CaSR, and GPRC6A, in addition to being sensors of the extracellular calcium level, also serve as nutrient (such as L-AA and polyamine) sensors. The former receptor responds to aromatic L-AAs phenylalanine, tryptophan and histidine with EC₅₀ at low micromolar concentrations, while the latter receptor senses L-arginine and L-lysine (Conigrave and Hampson, 2006; Wellendorph et al., 2009). CaSR is expressed in human pancreatic exocrine (including acinar) and endocrine cells (Racz et al., 2002). In rats around 50% of pancreatic acinar cells show functional responses to CaSR stimulation by inorganic cations Gd³⁺ or 8 mM extracellular Ca²⁺, however, pancreatic acinar cells do not respond to millimolar concentrations of an organic polycation neomycin known to activate CaSR elsewhere (Bruce et al., 1999). It has also been reported that intravenous AAs, which should in theory activate CaSR and GPRC6a, do not stimulate exocrine pancreatic secretion (Stabile et al., 1984). Using RT-PCR, GPRC6A expression has been detected in the pancreas, although the main signal may come from endocrine cells (Wellendorph and Brauner-Osborne, 2004). We confirmed the presence of CaSR and GPRC6A receptors in murine and human pancreatic acinar cells by immunocytochemistry. We investigated whether these receptors play any role in ARG-AP but reached a negative conclusion, based on no effect of NPS-2143 (CaSR and to less extent GPRC6A inhibitor), calindol (CaSR activator), or Cpd1 (GPRC6A inhibitor). The in vitro cell death caused by L-arginine was, however, dampened by NPS-2143, Cpd1 calindol, and R-568. The in vitro responses to L-histidine were inhibited by Cpd1, while responses to L-ornithine were insensitive to any of these GPCR modulators.

In contrast to ARG-AP, NPS-2143 administration substantially ameliorated the severity of CER-AP. This is an encouraging finding, because it may have some therapeutic promise for the human disease. From one point of view, it is known that AP is associated with hyperparathyroidism (and hence hypercalcaemia) (Biondi et al., 2011) or post-operative hypercalcaemia following liver transplantation (Chung et al., 2011). On the other hand, severe AP is associated with “true” hypocalcaemia and low levels of parathyroid hormone (McMahon et al., 1978). NPS-2143 has a long half-life in the circulation (Gowen et al., 2000) and may therefore represent a lead compound as a treatment option to restore systemic calcium post-induction of AP. NPS-2143 is an R-enantiomer belonging to the biaryloxypropanol class of molecules. Calcilytics NPS-2143, encaleret, and ronacaleret are closely structurally related to several third generation adrenergic antagonists, such as (in order of decreasing similarity) bucindolol, carvedilol, naftopidil, and nebivolol. Carvedilol (a non-selective β blocker/α1 blocker), nebivolol (β1 receptor blocker), naftopidil (α1 blocker), and a closely linked non-AR drug ranolazine are used clinically as racemates of R- and S-enantiomers. Whereas anti-β-adrenergic activity of NPS-2143 and other biaryloxypropanol calcilytics is recognized (notably for the S-enantiomers) (Marquis et al., 2009; Widler, 2011), the interaction between the third generation beta antagonists and CaSR/GPRC6A has never been a subject of a scrutiny. However SAR studies suggest that transformations from NPS-2143 to S-bucindolol and S-carvedilol should generally retain calcilytic activity (Yang et al., 2005; Balan et al., 2009; Shinagawa et al., 2010; Kiefer et al., 2011; Shinagawa et al., 2011). Indeed, the reported IC50 for the CaSR of the Cpd9 variant of NPS-2143 (bucindolol) is 100 nm, not that different from NPS-2143 itself at 50 nm (Gavai et al., 2005). Since carvedilol and nebivolol are prescribed and deemed to be clinically safe (despite some anti-hERG activity), they may offer an alternative to NPS-2143 and may be worthy of further investigation in AP.

Our data provide evidence for the role of L-arginine to L-ornithine conversion in the pathogenesis of ARG-AP. The role of systemic factors in this conversion, which is part of the urea cycle and ammonia detoxification, requires further elucidation. Our results also point to little role of intracellular calcium signaling or surface GPCR class C in ARG-AP. The fine differences between the pharmacological profiles of the effects of various basic AAs suggest additional non-mitochondrial mechanisms of pancreatic toxicity that may occur via disruption of protein synthesis or ER stress.

XZ, TJ, NS, WH, and MC performed the in vitro and in vivo experiments, analyzed and interpreted the data. LY, XY, CH, and DD involved in molecular biology experiments. LW participated in caerulein-induced acute pancreatitis experiment in mice. PS, TL, QX, and DC supervised the students and involved in data interpretation. RM provided the Ppif^-/- mice and involved in data interpretation. WH, MC, and RS obtained the funding, designed the study, supervised the students, drafted and critically revised the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.