Human CPA1 mutation causes digestive enzyme misfolding and chronic pancreatitis in mice.

NC_000072.6 , Mus musculus strain C57BL/6J chromosome 6, GRCm38.p2 C57BL/6J; NM_025350.3 , Mus musculus carboxypeptidase A1 (Cpa1) mRNA. Mice were on the C57BL/6N genetic background. The human CPA1 mutation p.N256K was knocked-in to the mouse Cpa1 locus using homologous recombination in C57BL/6 embryonic stem (ES) cells (Cyagen, Santa Clara, CA). The mouse Cpa1 gene is located on chromosome 6; it spans ~8.5 kb and comprises 10 exons. The targeting vector contained the mouse Cpa1 gene with the p.N256K mutation in exon 7 and a 1,941 nt sequence including a neomycin resistance gene flanked by loxP sites in intron 6, which served as a positive selection marker ( Supplementary Fig S1 ). Correctly targeted ES cell clones were identified by long range PCR followed by Southern blot verification. Mutant ES cells were injected into mouse embryos (blastocysts), which were implanted into pseudopregnant females. The resulting chimeras were bred with wild-type C57BL/6N mice to achieve germline transmission of the mutant allele. This allele was designated as CPA1 null , because the presence of the neomycin cassette resulted in diminished Cpa1 mRNA and CPA1 protein expression. To remove the neomycin resistance gene, the mutant mice harboring the CPA1 null allele were bred with a Cre-deleter strain that expresses the Cre recombinase in the early mouse embryo (B6.FVB-Tg(EIIa-cre)C5379Lmgd/J; Jackson Laboratories). The final CPA1 N256K knock-in allele carried the p.N256K mutation in exon 7 and a 144 nt residual sequence in intron 6 containing a single loxP site ( Supplementary Fig S2 ). C57BL/6N control mice were obtained from Charles River Laboratories (Wilmington, MA) or produced in our breeding facility from the same stock. CPA1 N256K and CPA1 null mice were maintained in homozygous state. Both male and female animals were studied. To genotype CPA1 N256K mice, we used primers that amplified exon 7 with parts of the flanking introns. The amplicon size from the wild-type allele was 646 bp, whereas the mutant allele yielded a 790 bp product due to the presence of the residual sequence in intron 6 ( Supplementary Fig S2 ). The primer sequences were as follows. CPA1 N256K F1: 5’-AGT GGG ATG TAC CTT TGA GC-3’; CPA1 N256K R1: 5’-AGG CCC AAG TCC CTG AGT GT-3’. CPA1 null mice were first genotyped for the presence of the neomycin resistance gene using the CPA1 Neo F1 (5’-GAG CTT GCG GAA CCC TTA AT-3’) and CPA1 N256K R1 primers. Positive mice yielded a 606 bp amplicon. Homozygosity was then verified by direct sequencing of the site of the p.N256K mutation after PCR amplification with the CPA1 N256K F2 (5’-TGA AGG TGT TGA TAG GTT ACT AG-3’) and CPA1 N256K R1 primers which yielded a 509 bp product ( Supplementary Fig S1 ). All other experimental procedures are described in the online Supplementary Methods file.

The aim of this study was to generate a novel knock-in mouse strain harboring the most frequently found human CPA1 mutation p.N256K in the native mouse Cpa1 gene. To demonstrate feasibility of this approach, first we had to confirm that the p.N256K mutation in the context of mouse CPA1 induces misfolding and ER stress in the same manner as it was described for human CPA1. Therefore, we measured CPA1 secretion and ER stress markers in HEK 293T cells transfected with wild-type and mutant CPA1 expression plasmids. We found that the p.N256K CPA1 mutant was not secreted into the conditioned medium ( Fig 1A,B ), but was retained intracellularly ( Fig 1B ) and induced ER stress, as judged by increased XBP1 splicing, and elevated mRNA levels for the ER chaperone HSPA5 (BiP) and ER-stress associated pro-apoptotic transcription factor DDIT3 (CHOP) ( Fig 1C ). These results indicate that we can successfully model the effects of human CPA1 mutation p.N256K in mouse Cpa1 . A targeting vector containing the mouse Cpa1 gene with the p.N256K mutation in exon 7 and a neomycin resistance cassette flanked by loxP sites in intron 6 was used to introduce the mutation into the mouse genome by homologous recombination in C57BL/6 embryonic stem cells ( Fig 2A ). Mice carrying this targeted allele expressed diminished levels of Cpa1 mRNA ( Fig 2B ) and CPA1 protein ( Fig 2C ) and were, therefore, designated as a CPA1 null strain and used as an additional control for our experiments. This control was important to distinguish between the effects due to loss of CPA1 protein/activity or to CPA1 misfolding. To obtain CPA1 N256K mutant mice, the neomycin cassette was excised by breeding with a Cre-deleter strain. The final CPA1 N256K strain harbored the p.N256K mutation in exon 7 and a residual loxP site in the neighboring intron. Although human CPA1 mutations are found in the heterozygous state, in our experiments we used homozygous mice, which exhibited more robust and earlier changes in pathology. To characterize Cpa1 mRNA expression in the pancreas of CPA1 N256K and CPA1 null strains, we prepared pancreatic cDNA from 1-month-old animals ( Fig 2B ). Analysis with qPCR revealed that in the CPA1 N256K strain Cpa1 mRNA expression was reduced by 30%, relative to C57BL/6N controls, while in the CPA1 null strain it was diminished by 95%. The slight decrease in mRNA expression in the CPA1 N256K animals was likely due to the negative impact of the remaining single loxP site in intron 6. Western blot analysis of pancreatic homogenates from 1-month-old CPA1 N256K mice demonstrated reduced but still significant CPA1 protein content, compared to C57BL/6N control mice ( Fig 2C ). As expected, we did not detect CPA1 protein expression in the pancreas of CPA1 null mice ( Fig 2C ). We isolated pancreatic acini from 6-week-old CPA1 N256K and C57BL/6N control mice and measured spontaneous and cerulein-induced CPA1 secretion to the culture medium ( Fig 3 ). Relative to the wild-type CPA1 protein, secretion of the CPA1 p.N256K mutant was hardly detectable ( Fig 3A ). In contrast, western blot analysis of acinar cell lysates revealed high levels of mutant CPA1 protein, indicating intracellular retention. As seen with pancreas homogenates (cf Fig 2C ), intracellular levels of the N256K mutant were somewhat reduced relative to the wild-type CPA1 content, indicating that some of the misfolded and retained mutant protein becomes degraded. We note that compared to cell culture experiments, degradation of mutant CPA1 protein is less efficient in mice. The two novel CPA1 strains showed no apparent phenotypic alterations and bred normally. When followed up to 6 months, CPA1 N256K and CPA1 null mice gained weight with similar kinetics as the C57BL/6N controls ( Fig 4A ). Macroscopic pancreas morphology was comparable in CPA1 null and C57BL/6N control mice. In contrast, we observed smaller pancreata in CPA1 N256K mice starting at 3 months of age (not shown). Pancreas weight measurements confirmed atrophy in the Cpa1 N256K strain, which corresponded to ~35–40% decrease at 6 months of age ( Fig 4B ). To identify the histological changes underlying pancreatic atrophy in CPA1 N256K mice, we stained pancreas sections of 1, 3, 6 and 12 months old mice and compared these to the corresponding CPA1 null and C57BL/6N sections ( Fig 5 ). Pancreas histology of CPA1 null mice was similar to C57BL/6N controls and showed no discernable changes over time. In contrast, we observed progressive pathological changes on sections from CPA1 N256K mice, which included loosely packed tissue architecture due to loss of acini, infiltration of inflammatory cells and the presence of pseudotubular complexes ( Fig 5A,B ). Increased acinar cell vacuolization was also apparent at later time points. Quantification of the acinar cell area indicated ~30% loss by 12 months of age in CPA1 N256K mice ( Fig 5C ). The histological picture was consistent with the development of spontaneous, progressive chronic pancreatitis in mice carrying the homozygous CPA1 p.N256K mutation. This conclusion was further corroborated by Masson’s trichrome staining, which revealed widespread fibrotic changes among the remaining acini in CPA1 N256K mice ( Fig 6A ). Furthermore, pancreata from CPA1 N256K mice contained significantly higher hydroxyproline levels than those from CPA1 null or C57BL/6N control mice ( Fig 6B ). Acinar-ductal metaplasia (pseudotubular complexes) was also visualized by immunohistochemical staining for cytokeratin 19 and Sox9, which yielded strong positivity in CPA1 N256K sections ( Fig 6C ). To evaluate possible ultrastructural changes in acinar cells from CPA1 N256K mice, we examined pancreas tissue by electron microscopy ( Fig 7A ) from 4-months-old mice. Zymogen granules showed comparable numbers and morphology in acinar cells from CPA1 N256K and C57BL/6N control mice. In CPA1 N256K mice, we observed some vacuolization and the presence of concentric arrangements of ER stacks (ER whorls) without visible ER dilations. Electron micrographs also showed pseudotubular complexes and macrophages between acini. Infiltration of macrophages was further confirmed by immunohistochemistry for the marker F4/80 in the pancreas of CPA1 N256K mice ( Fig 7B ). We examined blood levels of the pancreatitis marker amylase in CPA1 N256K, CPA1 null and C57BL/6N control mice at ages of 1 month, 3 months and 6 months. We detected a significant increase in plasma amylase in 1-month-old CPA1 N256K mice relative to the two other strains ( Fig 8A ). Consistent with subsequent acinar atrophy, no increase in plasma amylase was apparent at 3 and 6 months in CPA1 N256K mice. We also found a slight but significant decrease in plasma amylase in 1-month-old CPA1 null mice compared with C57BL/6N controls. This was likely due to the lower tissue content of amylase in CPA1 null mice at this age (not shown). Intra-pancreatic trypsin activation was evaluated in pancreas homogenates from 1 month, 3 months and 6 months old mice. Remarkably, we observed significant elevations in trypsin activity in pancreata from CPA1 N256K mice relative to CPA1 null and C57BL/6N controls ( Fig 8B ). High trypsin activity persisted and even increased during the disease course. Finally, to determine whether the histological changes in the CPA1 N256K strain are associated with elevated ER stress markers, we measured pancreatic mRNA expression for the chaperone Hspa5 (BiP) and the transcription factor Ddit3 (CHOP) at 1 month ( Fig 9A ), 3 months ( Fig 9B ) and 12 months ( Fig 9C ) of age. Compared to C57BL/6N and CPA1 null controls, in CPA1 N256K mice BiP showed a consistent but modest upregulation, which was apparent as a trend at 1 month and 12 months but reached statistical significance at 3 months. In contrast, significant CHOP upregulation was observed in the pancreas of CPA1 N256K mice at all time points ( Fig 9 ). We also measured Hspa5 mRNA from isolated acinar cells of 6-week-old mice and found a 1.5-fold upregulation in cells from CPA1 N256K mice versus C57BL/6N controls (not shown). Finally, staining for markers of apoptosis (TUNEL, cleaved caspase 3) revealed very few positive cells at any given time tested (not shown); consistent with the relatively low levels of chronic ER stress observed and the slowly progressive nature of parenchymal atrophy.

Our thinking about the pathogenic mechanism of genetic risk factors in chronic pancreatitis has been shaped by biochemical and cell biological experiments while animal models have been conspicuously missing. This unique experimental trajectory of our field is partly to blame for the scarcity of novel therapeutic approaches, as the unavailability of pre-clinical models has hindered drug development. Genetic risk is mediated via two seemingly unrelated pathological pathways; the trypsin-dependent and the misfolding-dependent mechanisms.[ 2 , 8 ] While the causative role of premature intra-pancreatic trypsin activation in pancreatitis onset has been appreciated for some time, the notion that mutation-induced misfolding may cause pancreatic pathology is relatively novel. The concept was first proposed in 2009, based on the observation that certain PRSS1 mutations caused misfolding and ER stress in cell culture experiments.[ 5 ] More compelling evidence for ER-stress related pathogenesis came from our 2013 study that identified the association of loss-of-function variants in CPA1 with early-onset chronic pancreatitis.[ 7 ] The majority of functionally defective CPA1 variants exhibited the ‘misfolding phenotype’ of diminished secretion with intracellular retention and degradation. The most frequently found variant p.N256K was also shown to induce ER stress in AR42J rat acinar cells transfected with recombinant adenovirus. Recent studies demonstrating association of the p.L104P PRSS1 variant and a novel p.S282P CPA1 variant with autosomal dominant hereditary pancreatitis provided a convincing indication that misfolding variants can have strong, highly-penetrant genetic effects.[ 6 , 9 , 10 ] Importantly, both variants induced significant ER stress in cell culture, further corroborating the presumed link between misfolding, ER stress and chronic pancreatitis. However, despite the accumulation of highly suggestive evidence for the pathogenic role of the misfolding pathway, animal models that recapitulate CPA1 −associated chronic pancreatitis, both phenotypically and mechanistically, remained a critical unmet challenge. In the present study, we set out to address this problem by knocking in the most frequent human CPA1 mutation p.N256K into the mouse Cpa1 gene. First, we verified in cell culture experiments that the p.N256K mutation exerts the same effect on mouse CPA1 as it does on human CPA1. Indeed, the mutated mouse CPA1 exhibited the expected misfolding phenotype (diminished secretion, intracellular retention and degradation) and caused significant ER stress in cells, as judged by elevated BiP and CHOP mRNA levels and increased splicing of the XBP1 mRNA. Next, the novel knock-in mouse strain was generated using a homologous recombination approach. To replicate the exact human situation, we created a constitutive Cpa1 p.N256K allele expressed throughout the mouse’s life. In contrast to humans where CPA1 mutations are heterozygous, we bred the mice to homozygosity to obtain stronger and earlier responses. It is important to note that heterozygous mutant mice developed similar pancreatic lesions as homozygous animals but on a slower time-scale and to a smaller extent, which was impractical for experimental studies (not shown). In our study design, we also included a CPA1 null strain to demonstrate that the effects of the p.N256K mutation are not due to loss of CPA1 protein or activity. Arguably, our most exciting finding is the progressive histological damage observed in the pancreas of CPA1 N256K mutant mice. Thus, starting at the age of 1 month, these animals exhibit classic signs of chronic pancreatitis including pancreatic atrophy due to loss of acini, fibrosis, tubular complexes (acinar-ductal metaplasia) and inflammatory cell infiltration. Approximately a third of the acinar parenchyma becomes ablated with no apparent change in islets. These observations clearly establish that the p.N256K mutation and, by extension, misfolding CPA1 variants in general, can directly cause chronic pancreatitis. Because no such changes were observed in the CPA1 null strain, the results also demonstrate that loss of CPA1 protein or activity per se are not harmful to the pancreas. To examine whether the development of the progressive morphological changes were driven by mutation-induced misfolding, we investigated signs of ER stress in the CPA1 N256K strain. Interestingly, as revealed by electron microscopy, there were no apparent changes in ER morphology with the exception of a few ER whorls. Consistent with this picture, only a modest increase in the mRNA levels of the master chaperone BiP was observed. On the other hand, we found marked upregulation of the pro-apoptotic transcription factor CHOP. Taken together, the observations demonstrate that in CPA1 N256K mice chronic, relatively low-level ER stress results in CHOP-dependent acinar cell loss. In turn, acinar cell death triggers the other hallmarks of chronic pancreatitis such as inflammatory cell infiltration, fibrosis and acinar-to-ductal metaplasia. An unexpected observation was the high trypsin activity present in the pancreata of the CPA1 N256K mice throughout the disease course. This finding raises the possibility that ER-stress somehow triggers trypsin-dependent pathological mechanisms and disease onset and progression may be ultimately driven by trypsin. Alternatively, high trypsin activity may be a marker of parenchymal damage rather than the cause of it. Finally, it is important to highlight the potential pre-clinical utility of this new model of misfolding-induced chronic pancreatitis. The CPA1 N256K strain develops, gains weight and breeds normally. These properties together with the convenient time line of the development of the pancreatic lesions make this an ideal model to test the effects of various environmental insults (alcohol, smoking, high-fat diet) or pharmaceuticals on the progression of chronic pancreatitis. In conclusion, we demonstrated that the human CPA1 mutation p.N256K induces misfolding of mouse CPA1 and causes spontaneous and progressive chronic pancreatitis in the CPA1 N256K knock-in mouse, associated with the upregulation of the ER-stress related pro-apoptotic transcription factor CHOP. No pancreatic pathology was observed in CPA1 null mice, indicating that loss of CPA1 function per se does not lead to acinar cell damage. The observations offer convincing evidence for the pathogenic role of a misfolding-dependent pathway in chronic pancreatitis and set the stage for interventions targeting ER-stress associated mechanisms in this disease.

The inflammatory disease continuum of the pancreas includes the clinical diagnoses of acute pancreatitis, recurrent acute pancreatitis and chronic pancreatitis.[ 1 ] Onset and progression of a single episode of acute pancreatitis towards chronic pancreatitis is often determined by underlying genetic risk factors associated with digestive proteases or their inhibitor. Mutations in PRSS1 (cationic trypsinogen), SPINK1 (serine protease inhibitor Kazal type 1), CTRC (chymotrypsin C) increase activation of trypsinogen to trypsin in the pancreas and thereby result in acinar cell damage and inflammation.[ 2 ] A protective mutation in PRSS2 (anionic trypsinogen) and an inversion at the CTRB1-CTRB2 (chymotrypsin B1–B2) locus also modify trypsin levels and pancreatitis risk.[ 3 , 4 ] Importantly, genetic risk in chronic pancreatitis can be mediated by mechanisms that are unrelated to trypsin activity. Thus, a subset of PRSS1 mutations has no impact on trypsin activity but causes enzyme misfolding and elicits endoplasmic reticulum (ER) stress in cell culture experiments.[ 5 , 6 ] Similarly, CPA1 (carboxypeptidase A1) mutations cause enzyme misfolding in vitro , resulting in diminished CPA1 secretion, intracellular retention and degradation, and associated ER stress.[ 7 , 8 ] CPA1 mutations that elicit misfolding are highly overrepresented in pediatric cases of chronic pancreatitis and may be associated with autosomal dominant hereditary pancreatitis, indicating that these risk factors have a strong, essentially disease-causing effect.[ 7 , 9 ] While the in vitro observations support a pathogenic role for mutation-induced misfolding and ER stress; evidence from appropriate animal models has been lacking and the disease mechanism of CPA1 mutations has remained hypothetical. The unavailability of animal models not only limited our understanding of the disease mechanism but also hindered pre-clinical testing of novel therapeutics. Here we addressed this knowledge gap by generating a novel knock-in mouse strain carrying the human CPA1 mutation p.N256K in the mouse Cpa1 locus.