หน้าหลัก

3. Yet, this number is still 1010 times bigger than the number of 3. Yet, this number is still 1010 times bigger than the number of proteins in the remaining blood proteome, and the a priori probability to detect by MS an antibody molecule with a given sequence is vanishingly small. However, recent studies have revealed that antigen specific antibody homology is more frequent than would be expected by pure chance14?1. Indeed, when the immune system in different individuals is challenged by the same antigen, the antibodies raised against this challenge should bind to it efficiently, which puts restraint on sequence variability of these antigen-specific Igs. In a homogeneous group of patients, the abundance(s) of peptides from the homologous Ig variable region with binding affinities to disease-specific antigen(s) may even be high enough to be detected by MS and may correlate with the disease strongly enough to be useful as biomarker(s). Since the Ig sequences of interest are unlikely to be found in standard sequence databases, analysis of the hidden blood proteome requires de novo polypeptide sequencing. Here, we introduce the SpotLight approach to the analysis of the hidden blood proteome. Given that a majority of Title Loaded From File polymorphism within the blood proteome is derived from antibodies, the SpotLight approach includes a simple enrichment step for polyclonal Immunoglobulin G (IgGs) using Melon Gel (MG). MG enrichment is not based on Fc-region specificity and certain blood proteins (herein referred to as MG proteins) are also co-enriched. To produce and annotate a database of IgG and other de novo sequences, SpotLight employs several important steps prior to regular standardized label-free proteomics database search and quantitation (Fig. 1). The MG-enriched fraction is digested and analyzed by LC-MS/MS using two complementary fragmentation techniques. Both MS and MS/MS data are acquired with high resolution, which is a pre-requisite for reliable de novo sequencing22. Newly derived sequences are analyzed by BLAST in terms of homology to either IgG or other proteins. In any case, their sequences are inserted in the sequence database, together with the tentative assignment. Next, the LC-MS/MS datasets of both intact and MG-enriched proteomes are processed using our novel DeMix-Q label-free workflow for peptide identification and quantification as well as for protein inference23,24. To test the SpotLight approach, we selected a cohort of early stage patients diagnosed with similar neurodegeneration disorders: Alzheimer's disease (AD) and Dementia with Lewy Bodies (DLB). These disorders have similarities in pathology, and their differentiation is nontrivial25 but important due to the differences in prognosis and treatment response. Clinical criteria have good specificity but relatively low sensitivity, and there is a need for accurate, cheap, and easily available biomarkers. Due to overlap in pathology, CSF and MRI-based biomarkers are not sufficiently accurate25. The 144 patients (97 AD and 47 DLB; see Table 1) were separated into a homogeneous Group A (24 AD and 24 DLB) that was used for multivariate (MV) statistical analysis and model building, and a heterogeneous Group B (remaining patients) that was employed for model verification. Subsets of the generated data, (i.e. intact and MG-enriched proteomes; proteins and peptides, IgG and non-IgG peptides; known and new peptides) were tested according to the quality factor Q2 of the model, the p-value of AD/DLB separation and theScientific RepoRts | 7:41929 | DOI: 1.