Global Differences in RBC Membrane Protein Expression Between Normal and PNH Individuals.

Abstract 1986

Poster Board I-1008

Paroxysmal nocturnal hemoglobinuria (PNH) is an acquired hemolytic anemia that results from the deficiency of glycosyl phosphatidylinositol (GPI)-linked proteins on the surface of blood cells carrying a PIGA gene mutation. The absence of 2 GPI-linked complement regulatory proteins, CD59 and CD55, on PNH red blood cells (RBCs) accounts for their increased susceptibility to complement-mediated lysis. The pathophysiology of other PNH complications, including thrombophilia and the association with aplastic anemia, is less well understood. We hypothesized that the differences between PNH and normal RBCs are not restricted to the lack of GPI-linked proteins. To test this hypothesis we have developed a label-free proteomics platform to identify global differences in RBC membrane protein expression between normal and PNH individuals. Six control samples and 17 PNH samples from 13 different patients were analyzed. PNH patients were categorized as high (>80%), medium (between 20% and 80%), and low (<20%), based on their combined percentage of type II and type III PNH RBCs. RBC membranes were prepared, and proteins were extracted and subjected to trypsin digest. Peptides were then analysed by nano-liquid chromatography coupled to tandem mass spectrometry (nanoLC-MS/MS). All 23 nanoLC-MS/MS runs (6 control and 17 PNH) were aligned, and peptide ion currents were extracted and normalized for relative quantitation of peptide abundances and inference of relative protein abundance. ANOVA analyses of the different sample classes were performed in order to distinguish differences in the relative abundances at the peptide level (from a dataset containing ∼30,000 peptides). Comparison of control vs. high PNH samples yielded a set of 142 peptides showing patterns of expression that clearly separated the two groups. Targeted nanoLC-MS/MS identified the differential expression of peptides from 5 of the 8 known GPI-linked proteins expressed on human RBCs. As expected, the relative abundances of these peptides were inversely related to the percentage of PNH RBCs in the samples. In addition, we found peptides from complement protein C3 increased on PNH RBCs from patients receiving the complement C5 inhibitor eculizumab. We have developed a statistical model that uses peptide ion intensities to infer relative abundance at the protein level, and allows for significance testing between healthy and disease groups. RBC membranes from the 6 normal control and 7 high PNH samples were processed using MS with ‘data-dependant acquisition', which is biased towards annotation of the more abundant peptides. 1676 peptides from 119 proteins were annotated with high confidence. Nine of these showed a significant difference between control and PNH, including CD59 and semaphorin-7A, the 2 known GPI-linked proteins annotated in this experiment. The remaining 7 non-GPI-linked proteins were peroxiredoxin-2 (PXDN2), valosin-containing protein, gamma-actin, catalase (CATA), heat shock protein 90A, flotilin-1, and RAP1A. Several of these non-GPI-linked proteins are of potential interest to the PNH phenotype or PNH pathophysiology. The anti-oxidative enzymes CATA and PXDN2 both showed significantly lower levels associated with PNH RBC membranes compared to control, and may contribute to an increased susceptibility of the PNH RBC plasma membrane to oxidative damage. The small G-protein RAP1A showed a significantly elevated expression in PNH RBC membrane preparations. RAP1A activation mediates increased sickle cell RBC adhesion to laminin (Blood, 105: 3322), suggesting that elevated RAP1A levels may also contribute to the increased propensity of thrombosis in PNH patients. In summary, differential protein expression analysis indicates that PNH and normal RBCs differ by more than only the lack of GPI-linked proteins. Differences in protein expression are likely to contribute to disease manifestations and are likely to provide insights into additional unexpected underlying disease processes. Future studies will investigate longitudinal changes in the RBC proteome during the course of the disease and in response to treatment, potentially providing new biomarkers that help to distinguish patients at risk for complications and patients likely to benefit from specific forms of treatment. Our comparative label-free proteomics platform should be broadly applicable to the analysis of other RBC disorders.