Quantitative correlation of ENPP1 pathogenic variants with disease phenotype

Abstract

Ectonucleotide pyrophosphatase/phosphodiesterase 1 (ENPP1) codes for a type 2 transmembrane glycoprotein which hydrolyzes extracellular phosphoanhydrides into bio-active molecules that regulate, inter alia, ectopic mineralization, bone formation, vascular endothelial proliferation, and the innate immune response. The clinical phenotypes produced by ENPP1 deficiency are disparate, ranging from life-threatening arterial calcifications to cutaneous hypopigmentation. To investigate associations between disease phenotype and enzyme activity we quantified the enzyme velocities of 29 unique ENPP1 pathogenic variants in 41 patients enrolled in an NIH study along with 33 other variants reported in literature. We correlated the relative enzyme velocities with the presenting clinical diagnoses, performing the catalytic velocity measurements simultaneously in triplicate using a high-throughput assay to reduce experimental variation. We found that ENPP1 variants associated with autosomal dominant phenotypes reduced enzyme velocities by 50 % or more, whereas variants associated with insulin resistance had non-significant effects on enzyme velocity. In Cole disease the catalytic velocities of ENPP1 variants associated with AD forms trended to lower values than those associated with autosomal recessive forms - 8-32 % vs. 33 % of WT, respectively. Additionally, ENPP1 variants leading to life-threatening vascular calcifications in GACI patients had widely variable enzyme activities, ranging from no significant differences compared to WT to the complete abolishment of enzyme velocity. Finally, disease severity in GACI did not correlate with the mean enzyme velocity of the variants present in affected compound heterozygotes but did correlate with the more severely damaging variant. In summary, correlation of ENPP1 enzyme velocity with disease phenotypes demonstrate that enzyme velocities below 50 % of WT levels are likely to occur in the context of autosomal dominant disease (due to a monoallelic variant), and that disease severity in GACI infants correlates with the more severely damaging ENPP1 variant in compound heterozygotes, not the mean velocity of the pathogenic variants present.

Keywords: ARHR2; Cole disease; ENPP1; FGF23; GACI; OPLL; Pyrophosphate.

Figure 1:. Schematic of the location of ENPP1 pathogenic variants on enzymatic domains.

Figure 2:. Enzymatic velocity and disease phenotypes of ENPP1 variants displayed by primary sequence.

The Enzyme velocities for 62 ENPP1 pathogenic variants associated with disease phenotypes were determined simultaneously in triplicate in two separate experiments. The data are displayed as box-plots with whiskers representing the 25–75 quartiles and min-max values, respectively. The individual measurements are displayed as symbols colored by the disease phenotype defined in the legend. Protein domains are boxed at the top of the figure, with boundaries between domains denoted by vertical lines. ENPP1 pathogenic variants in GACI, ARHR2, and PXE exhibit an autosomal recessive (AR) inheritance pattern, while those in Cole disease exhibit an autosomal dominant (AD) inheritance pattern except for the C120R pathogenic variant, which is autosomal recessive. The ENPP1 pathogenic variants in early onset osteoporosis, hypophosphatemia, DISH, OPLL, and childhood obesity/stroke all induce disease in an autosomal dominant manner. While it is recognized that ENPP1 pathogenic variants may induce either GACI or PXE phenotypes, the ENPP1 pathogenic variant G586R has only been described in a PXE patient as a (biallelic, monoallelic, compound heterozygous). Finally, while most, if not all, surviving GACI patients will go on to develop ARHR2 [42], the two ARHR2 pathogenic variants denoted above were only described in ARHR2.

Figure 3:. Enzymatic velocities of ENPP1 variants in GACI.

The data are displayed as in Figure 1, for only GACI patients. Measurements below the dotted line are statistically significant relative to WT (p<0.05), and measurement above the dotted line are not (p>0.05, ANOVA multiple comparison with Brown-Forsythe and Welch tests).

Figure 4:. A. Enzyme velocities of ENPP1 variants associated with autosomal dominant disease. B. Enzymatic velocities of ENPP1 variants in Cole Disease. C. Enzymatic velocities of ENPP1 pathologic variants in disorders other than GACI and Cole disease.

Data are displayed as in figure 1. Statistical significance determined by ANOVA Brown-Forsythe and Welch tests, with significance explicitly stated when 0.05>p>0.0001; ****p<0.0001. SSA: sickle cell anemia.

Figure 5:. A. Enzymatic velocity of the paired ENPP1 pathogenic variant in compound heterozygous GACI patients where the corresponding allele does not significantly compromise enzymatic velocity.

There were 7 ENPP1 pathogenic variants observed in GACI which did not significantly reduce enzyme velocity below WT levels, 5 in the catalytic domain and 2 in the SMB2 domain. These variants were S162G, G186R, G342V, R349K, R481W, R481Q, and Y570C. The enzymatic velocities of the corresponding alleles in compound heterozygous patients were identified except for patients with R349K and Y570C. In every case the corresponding allele reduced the enzymatic velocity below 50% of WT except patients with G186R, which occurred in siblings with biallelic, homozygous G186R variants. B. Linear regression of the relative catalytic velocity of the most compromised ENPP1 variant with disease severity in compound heterozygous GACI patients. A linear regression analysis found that the GACI disease severity score correlated inversely with the relative catalytic velocity of the most severe ENPP1 variant, demonstrating that disease severity in GACI is dependent on degree to which the catalytic velocity is compromised in the most severely affected allele. F=4.9, DFn=1, DFd=136, R²=0.35, and p=0.029 (simple linear regression analysis). Individual values of disease severity are represented by filled circles, with whiskers representing the standard deviation in each measurement. In most cases the standard deviation does not exceed the symbol width, and so the whiskers are not apparent.