Figure 1.
Chromatographic run order for MAF and LC-MS analysis.
Duplicate aliquots of CSF (denoted by ‘a’ and ‘b’) from each of six cognitively normal individuals (numbered 1 through 6) and fourteen aliquots of CSF (labeled P1 through P14) pooled from 100 individuals with mild [n = 9], very mild [n = 33], and no [n = 58] dementia of the Alzheimer type were selected for single block proteomic analysis. The order that samples underwent MAF (Y-axis) and then LC-MS analysis (X-axis) were independently randomized. A single sample (P7) that did not pass the 1D SDS-PAGE quality assessment (Fig. S1) and was not analyzed by LC-MS is not represented.
Figure 2.
Data processing and symmetrical matrix for all sample pairwise comparisons of log2 annotated peptide intensities.
A. Data processing steps in the visual script used within Rosetta Elucidator™ software. The intensities from the aligned peptide chromatograms were normalized and concatenated to sum signals from all charge states, isotope groups (Steps 1 through 3). Peak intensities of the isotope groups that were assigned to unique peptide sequences within each sample were summed (Step 4) for Pearson correlation coefficients (PCC). Common laboratory contaminants (e.g. keratin) and residual proteins from the MAF procedure (summarized in Table S4) were removed in Step 5. Methionine-containing peptides were removed in Step 6. ‘Early-eluting’ peptides were removed in Step 7. At each step the data were exported from the software and imported into DAnTE-R for further analysis using pair-wise correlations and scatter plots of the log2 transformed intensity data. B. Symmetrical matrix/non-clustering heatmap of PCC values from all pairwise comparisons from the annotated peptide intensity data (center), using a colorimetric scale ranging from black (low correlation, 0.65 and below), to red, orange, yellow, and white (high correlation, maximum 1). Self-pairwise comparisons, which yield a PCC equal to 1.0, appear as the diagonal of white squares. C, D, E, F. Representative scatter plots of all aligned charge group intensities from paired samples: P13b vs P5; P5 vs P6; 1a vs 1b; and P3 vs P4. Units of X- and Y-axes both represent log2 transformed charge group intensities.
Figure 3.
Correlations of pairwise sample comparisons after ‘removal’ of process contaminant, methionine-containing and early-eluting low-intensity peptides.
A. Non clustering heat map of Pearson correlation values, after removal of pooled sample 13, calculated from all annotated peptides; B. Heat map after removal of contaminant and residual MAF-related peptides; C. Heat map after removal of Met-containing peptides from the contaminant-minus and MAF-minus set; D. Heat map after removal of low intensity, early eluting peptides (retention time = 20–42 min) from the contaminant/MAF/methionyl-peptide minus set. Colorimetric scale as depicted and described in Figure 2B. The ‘removed’ peptides and intensities are summarized in Table S4. Pearson correlation matrices are represented numerically in Table S5.
Figure 4.
Annotated peptide intensity scatter-plots from select pairwise comparisons with removal of methionine-containing, early-eluting peptides.
A. Scatter plot of annotated peptide intensities from individual sample aliquots 1a and 1b (process contaminant peptides already removed); B. Scatter plot of annotated peptide intensities from 1a and 1b after removal of Met-containing peptides from the contaminant-minus set; C. Scatter plot of annotated peptide intensities from pooled sample aliquots P3 and P4 (process contaminant and methionyl peptides already removed); D. Scatter plot of annotated peptide intensities from P3 and P4 after removal of early-eluting peptides. Colorimetric scale (in A) represents range of peptide intensities along x-axis, ranging from black (log2 intensity = 11), through red, orange and yellow to white (log2 intensity = 28). The ‘removed’ peptides and intensities are summarized in Table S4.
Figure 5.
Coefficient of variation for each of 81 proteins.
Coefficients of variation were calculated using values from all contributing peptides from pooled sample replicates. Numerical values in Table S6.
Figure 6.
Subject variance for each of 81 proteins.
Subject variance was calculated using values from all contributing peptides from paired replicates of individual samples. Numerical values in Table S6.
Figure 7.
Biological variability and technical variability of 81 proteins, represented by all contributing peptides.
Box and whiskers plot. All proteins ranked by mean of values from pooled sample replicates, left to right. Bar indicates median, box indicates 25th to 75th percentile, and whiskers indicate 10th to 90th percentile. Values derived from individual samples (n = 6; n = average of each pair of aliquots) are indicated in blue; those from pooled sample aliquots (n = 12), in red. Numerical values in Table S6.
Figure 8.
Unsupervised hierarchical clustering of duplicate samples from 6 cognitively normal individuals and all 81 proteins.
All proteins were represented by two or more peptides. Data from all contributing peptides were used to calculate protein abundance. Samples are represented by columns as indicated by lettered and numbered colored blocks, below; proteins are represented by rows, as indicated by gene symbols on the right. Normalized protein abundance values (Z-scores) are indicated colorimetrically for each protein in each sample; red = high, black = mean value, green = low.
Figure 9.
Unsupervised clustering of duplicate samples from 6 individuals, replicates of pooled CSF, and 81 proteins.
Formatted as described for Fig. 8.
Figure 10.
Unsupervised clustering of pooled and individual CSF replicates, limited to proteins with subject variance>60%.
Formatted as described for Fig. 8.