GPI-anchor and GPI-anchored protein expression in PMM2-CDG patients

Congenital disorders of glycosylation (CDG) are caused by defective glycosylation of glycoproteins and glycolipids. Up to date more than 70 CDG have been reported [1‐3]. These disorders show an extremely broad clinical spectrum involving nearly all organs, with degrees of severity that range from early death to very mildly affected adults [4]. The number of inborn defects of glycosylation is expected to further increase rapidly [1]. The most prevalent N-linked disorder, CDG-Ia, first reported in 1980, is an autosomal recessive defect caused by deficient phosphomannomutase (PMM) activity [5] due to mutations in PMM2[6]. Thus, this disorder is now called PMM2-CDG [7]. PMM2 catalyzes the conversion of mannose-6 phosphate to mannose-1-phosphate, an early step in the assembly of the lipid-linked oligosaccharide (LLO) precursor required for N-glycosylation (Figure 1). The most widely used method to screen for N-glycan synthesis defects is isoelectric focusing of serum transferrin (Tf), showing a type 1 pattern in PMM2-CDG (increase of asialo- and disialotransferrin) [8].

Two other metabolic pathways also require mannose-1-phosphate, although with lower requirements than N-glycosylation (nine mannose residues): glycosylphosphatidyl inositol (GPI) anchor biosynthesis (three mannose residues), and O-mannosylation synthesis (one mannose) (Figure 1) [9]. PMM2-CDG patients have no symptoms of α-dystroglycan hypomannosylation. However, the identification of mutations in the genes for the 3 subunits of dolichol-phosphate-mannose (DPM) synthase in patients with N-glycosylation defects and muscular dystrophy-dystroglycanopathy syndrome [10] prompt further studies to address whether impaired phosphomannomutase activity might also affect other metabolic pathways requiring dolichol phosphate-mannose, like O-mannosylation. Moreover, as far as we know, there is no study evaluating GPI-anchor and GPI-anchored protein expression in PMM2-CDG patients. Since GPI-anchored proteins are involved in multiple pathways, defects in these proteins may contribute to the complex phenotype of this disease. For example, the tendency towards thrombosis in patients with PMM2-CDG could be multifactorial, and potentially (at least in part), due to deficiency of GPI-anchored complement inhibitors on the surface of circulating erythrocytes and other cells [11]. In this context, it has to be noted that two GPI-anchor synthesis defects, paroxysmal nocturnal haemoglobinuria (PNH), and PIGM-CDG (Figure 1), exhibit a high rate of occurrence of severe thrombotic events due to complement-mediated hemolysis resulting from the deficiency of complement regulatory proteins, CD55 and CD59 [12, 13].

This study aimed to evaluate the expression of GPI-anchor and GPI-anchored proteins in different cell types from PMM2-CDG patients.

The characteristics of the PMM2-CDG patients included in this study are shown in Table 2.

Since mannose-1-phosphate is required for the synthesis of both LLO and GPI-anchor, one would suspect that patients with congenital defects affecting enzymes involved in the early and common steps of these pathways might have defects in both the glycan content of N- glycoproteins and the GPI-anchor/GPI-anchored protein expression (Figure 1). Actually, GPI-anchored protein deficiencies are included among the congenital disorders of glycosylation [1, 12, 22‐24]. However, as far as we know, no study has evaluated any potential effect of congenital disorders of glycosylation on GPI-anchor and GPI-anchored protein expression. In this study we have addressed this issue in the most prevalent CDG: PMM2-CDG. We have analyzed a broad range of GPI-anchored proteins in four different blood populations from 12 PMM2-CDG patients with different ages, gender, PMM2 mutations and clinical expression of disease. Our study suggests that in contrast to the significant effect on N-glycosylation, mutations in PMM2 have negligible effects on GPI-anchor and GPI-anchored protein expression in different blood cells. Several hypotheses can be proposed to explain this observation: i, The synthesis of GPI-anchors requires a smaller number of mannoses (three) than the synthesis of LLO (nine) (Figure 1). Thus, the residual PMM activity of PMM2-CDG patients [13, 22] could be sufficient to satisfy the requirements for normal GPI-anchor synthesis. ii, It is possible that when GDP-Man and Dol-P-Man are in limited supply, they are used more efficiently in other pathways, so that GPI anchoring occurs but N-glycosylation is incomplete, as suggested by Thomas and coworkers [25]. iii, A different turnover time of these mannosylated structures. iv, Elements with the ability to modulate PMM2 defects by different mechanisms (PMM1 by redundancy, or PGM and PMI by affecting the balance of complementary pathways) might generate mannose-1-phosphate specifically used in the GPI-anchor synthesis. Further studies are required to validate these hypotheses. Independently of this, our study shows that neither the GPI-anchor nor the GPI-anchored protein expressions are affected in PMM2-CDG patients, with the restriction that we have not tested whether the GPI-anchors of these patients are qualitatively affected or show heterogeneity.

According to these results, quantitative defects of GPI-anchor/GPI-anchored proteins are unlikely to be involved in the multisystemic clinical expression of PMM2-CDG patients. We were particularly interested to investigate whether a potential CD59 defect in erythrocytes of PMM2-CDG patients might contribute to their increased incidence of thrombosis [11, 15, 26]. This defect is usually evaluated through the complement-mediated lysis of erythrocytes, similar to patients with acquired (PNH) or with congenital GPI deficiency (PIGM-CDG) [27]. The normal expression of CD59 in RBC of PMM2-CDG individuals observed by flow cytometry analysis, the negative results observed in the Ham’s and sucrose tests, and the absence of anemia in PMM2-CDG patients [4] argue against quantitative and qualitative CD59 defects in these patients. Therefore, in PMM2-CDG patients, mechanism(s) different from complement-mediated hemolysis, seem to be involved in the disturbance of the hemostatic system leading to increased risk of thrombosis. This may be associated with the significant deficiencies of relevant anticoagulant elements (antithrombin, protein C and protein S) although they also have deficiencies of procoagulant factors (especially Factor IX and XI) associated with a significant deficiency of relevant anticoagulant elements (antithrombin, protein C and protein S) and procoagulant factors (factor IX and XI) [28].

Additionally, CD16 and CD14 immunostaining of neutrophils and monocytes, respectively, was found to be significantly reduced in PMM2-CDG patients by flow cytometric analysis with 3G8 and KD1 (CD16) and 61D3 (CD14) monoclonal antibodies. CD16, the low affinity Fcγ receptor III for IgG (FcγRIII) exists as a polypeptide-anchored form (FcγRIIIA, or CD16-A) in human natural killer cells and monocyte/macrophages, and as a GPI-anchored form in human neutrophils (FcγRIIIB or CD16-B) [29]. CD14 is a component of the innate immune system that along with the Toll-like receptor 4 and MD-2, acts as a co-receptor for the detection of bacterial lipopolysaccharide and also exists in two forms: either anchored into the membrane by a GPI tail (mCD14), or present in a soluble form (sCD14) [30]. Reduced CD16 expression is likely to contribute to the impaired clearance of immune complexes [31]. Moreover, a CD14 deficient animal model exhibits higher bacterial disease severity [32]. Consequently, the decreased CD16 and CD14 expression could potentially contribute to explain the recurrent infections described in PMM2-CDG patients [33, 34]. However, several lines of evidences from our study suggest that these patients express normal levels of CD16 and CD14 in neutrophils and monocytes, respectively. The apparent deficiency detected by 3G8, KD1, and 61D3 can be explained by the ability of these antibodies to recognize an epitope influenced by varying proportions of mannose contents, resulting from the PMM2 mutations of these patients. Indeed, a previous study has shown that the immunoreactivity of the 3G8 epitope on CD16 expressed on neutrophils is dependent primarily on a high mannose-type oligosaccharide, according to the sensitivity to the endo H digestion of the receptor [35], and KD1 may recognize a very close epitope (Additional file 2: Figure S2). Moreover, it has been shown that the FcγIII receptors of natural killer cells and neutrophils differ in their affinity to IgG because of primary sequence differences between the two receptor isoforms, and also by cell type-specific glycosylation of the receptor [36], which justifies the different CD16 expression on lymphocytes and neutrophils of PMM2-CDG patients. Further studies must investigate if the abnormal glycosylation of CD16 in neutrophils and CD14 in monocytes of PMM2-CDG patients affects the function of these immune receptors, as suggested for CD16 [35], which might potentially be implicated in the recurrent infections of these patients.

The reduced staining of CD16 on neutrophils (detected by 3G8 or KD1 binding) was striking in children, but not in the three adult PMM2-CDG patients analyzed. Interestingly, significant correlations between both CD16 and asialotransferrin with age were observed (Figure 7). These results agree with the suggested stabilization of glycosylation defects with age described in a previous study [37], and might explain the progressive normalization of the transferrin profile in patients who initially present with an abnormal pattern [38], although further studies including serial analysis of additional patients comparing values at different points in time should be performed to fully validate it. Indeed, age-related changes of glycoproteins currently represent an active research area [39]. It remains to be determined whether these changes might also have prognostic relevance in PMM2-CDG patients, as most of the PMM2-CDG patients commonly present with a severe clinical phenotype including recurrent infections, in the first years of life and show a relatively steady clinical picture by puberty [40].

In conclusion, PMM2 mutations do not impair GPI-anchor or GPI-anchored protein expression. However, the glycosylation anomalies caused by PMM2 mutations might affect the immunoreactivity of monoclonal antibodies and thus lead to incorrect conclusions about the expression of different proteins, including GPI-anchored proteins. Our results also suggest that neutrophils and monocytes are sensitive to PMM2 mutations, and relevant immunologic molecules could have abnormal proportion of mannose content with potential functional consequences that might contribute to explain the recurrent infections of PMM2-CDG patients. Finally, our study confirms the stabilization of glycosylation defects with age.