Functional predictors of pathogenicity of missense CPA1 variants in chronic pancreatitis

To the Editor:

We read with great interest the recent article by Zhu et al. [1], demonstrating that misfolding of pancreatic lipase due to an inborn mutation causes chronic pancreatitis (CP). Variants in the CPA1 gene, encoding procarboxypeptidase A1, can elicit CP by the same mechanism, however, not all variants are alike, and correct prediction of benign versus pathogenic behavior has been challenging.

In 2013, rare heterozygous variants in CPA1 were identified in multiple cohorts of CP cases [2]. Functional analysis indicated that patients preferentially carried loss-of-function CPA1 variants, most of which were poorly secreted or not secreted at all from transfected cells. When examined in aggregate, loss-of-function CPA1 variants were significantly overrepresented in cases versus controls, with higher prevalence in early-onset disease. Individually, however, only three loss-of-function missense variants (p.N256K, p.Y308H, p.R382W) showed significant enrichment in CP. Disease-association could not be ascertained for the majority of variants due to their ultra-low frequency.

Variant p.N256K was characterized functionally in detail using HEK 293T cells, AR42J cells, and a knock-in mouse model [2–4]. The experiments revealed that p.N256K triggered CP in mice and caused loss of proenzyme secretion, intracellular retention, and increased endoplasmic reticulum (ER) stress in cellular models and mice. The findings suggested that CP develops in carriers of p.N256K due to mutation-induced proenzyme misfolding and consequent ER stress. Subsequent reports described rare CPA1 variants p.V251M, p.S282P, and p.K374E in hereditary CP, confirming their pathogenic nature and high penetrance [5–7]. All three disease-causing variants showed the same functional characteristics as the p.N256K variant. It was proposed that CPA1 variants with a “misfolding phenotype” that mimicked that of p.N256K should be designated pathogenic [4, 8]. However, a quantitative relationship between pathogenicity and the extent of functional impairment has not been elucidated, therefore, the clinical significance of many CPA1 variants with various degrees of secretion defect has remained unclear.

Here, we used transfected HEK 293T cells to characterize 50 missense CPA1 variants, from CP patients and controls, with respect to cellular secretion and mRNA expression of the ER stress marker BiP (Supplementary Table S1). As benchmarks, we included 6 pathogenic (p.V251M, p.N256K, p.S282P, p.Y308H, p.K374E, p.R382W), and 5 benign (p.Q94R, p.T124I, p.A137G, p.G166D, p.A208T) variants, for which clinical significance could be assigned based on genetic evidence. When BiP levels were plotted as a function of secretion, a seemingly inverse linear relationship became apparent, i.e. low secretion was associated with high BiP (Figure 1, Supplementary Table S2). All known pathogenic variants exhibited <10% of wild-type CPA1 secretion and BiP levels that were similar or higher than those of the p.N256K pathogenic reference variant. Thus, variants with these properties can be classified as pathogenic with high confidence. In contrast, known benign variants were secreted to levels >20% of wild type, and cellular BiP levels were variable with some nearing or exceeding the putative pathogenic range. This observation indicates that variants that are secreted above the 20% threshold should be categorized as benign, irrespective of BiP levels. A few variants exhibited borderline properties (secretion levels in the 10-20% range); these remain of uncertain significance for now (Supplementary Table S2). The outlier signal-peptide variant p.L5P is designated as likely benign because it impairs ER entry and, therefore, causes no ER stress.

Figure 1.

Effect of 50 missense CPA1 variants on procarboxypeptidase A1 secretion and cellular BiP mRNA levels in transiently transfected HEK 293T cells. Some symbols may overlap. See Supplementary Table S1 for a complete list of variants. Transfections (n=4-13) were carried out as described previously [4]. Each 6-well transfection plate included a pcDNA3.1(−) vector control, wild-type CPA1, the p.N256K pathogenic reference variant, and 3 missense variants. Conditioned media and cells were harvested after 48 h. Media were analyzed by SDS-PAGE and Coomassie Blue staining, and bands were quantified by densitometry. Secreted proenzyme levels were expressed as percent of the wild type. Total RNA was isolated from cells and BiP mRNA levels were measured by reverse-transcription quantitative PCR, as reported earlier [4]. To minimize inter-experimental variability, BiP levels were first calculated as fold change over vector, and then expressed as percent of the p.N256K value within the same experiment. For clarity, error bars have been omitted; standard deviation values are listed in Supplementary Table S2.

Limitations of our study include the use of a cellular model that differs from pancreatic acinar cells, and the reliance on a single ER stress marker. Nevertheless, the findings are compelling, and clearly indicate that the best predictor of pathogenicity of missense CPA1 variants is severely reduced secretion with high BiP expression. Conversely, measurable secretion, even at low levels, predicts benign clinical behavior regardless of BiP elevation. This data set could serve as a useful reference for the assignment of clinical significance of novel CPA1 variants identified in patients with CP.