Age: 41 years Diagnosis: myelodysplastic syndrome Donor type: matched related donor HLA match: 6/6 Conditioning: full intensity Blood drawn post BMT: 26 days GVHD stage Skin-Liver-Gut: 1-0-3 Followup: 53 days Outcome: dead
Extracted molecule
protein
Extraction protocol
Twenty milliliters of blood was collected from each patient. Blood was collected in EDTA-containing Vacutainers (Becton Dickinson, Franklin Lakes, NJ) to prevent clotting. Plasma was obtained by Ficoll (Amersham, Piscataway, NJ) gradient centrifugation of whole blood at 1,600 rpm for 30 min within 3 h of collection. The plasma was carefully removed from mononuclear cells, aliquoted without the addition of protease inhibitors into cryovials and stored at -80 degrees C until use.
Label
Fluorescein-ULS
Label protocol
4 ul aliquots of each of 21 GVHD+ and 21 GVHD- plasma samples were combined to form a reference pool, which was labeled with biotin-ULS for 2 hours at 42 degrees C, and a second aliquot of each individual sample labeled with Fluorescein-ULS. Then 5 ul of 10X KREAstop was added for 5 minutes to each of the reaction tubes and free-ULS- reagent was removed using spin columns.
Twenty milliliters of blood was collected from each patient. Blood was collected in EDTA-containing Vacutainers (Becton Dickinson, Franklin Lakes, NJ) to prevent clotting. Plasma was obtained by Ficoll (Amersham, Piscataway, NJ) gradient centrifugation of whole blood at 1,600 rpm for 30 min within 3 h of collection. The plasma was carefully removed from mononuclear cells, aliquoted without the addition of protease inhibitors into cryovials and stored at -80 degrees C until use.
Label
Biotin-ULS
Label protocol
4 ul aliquots of each of 21 GVHD+ and 21 GVHD- plasma samples were combined to form a reference pool, which was labeled with biotin-ULS for 2 hours at 42 degrees C, and a second aliquot of each individual sample labeled with Fluorescein-ULS. Then 5 ul of 10X KREAstop was added for 5 minutes to each of the reaction tubes and free-ULS- reagent was removed using spin columns.
Hybridization protocol
Arrays were incubated with 500 ul of protein array blocking buffer for 30 minutes on an orbital shaker prior to hybridizations. The eluents containing the labeled plasma proteins from one individual sample and one aliquot of the reference pool were hybridized to each of the arrays overnight. After repeated washes, the chips were probed with the fluorescent reporter molecules (streptavidin-DY647 and anti-fluorescein-DY547) for 1 hour at room temperature, washed, and dried.
Scan protocol
The slides were scanned within 24-48 hours with an Axon 4000A scanner, using settings recommended for Cy3 and Cy5 detection, to obtain images representing the individual sample (the 532 channel, channel 1) and the reference pool (the 635 channel, channel 2).
Description
top array on slide, batch 2
Data processing
Dot intensity measures were performed with Genepix 3.0, and the “median of ratios” (the median of the intensity ratios at each pixel in a dot) used as the initial measure of the ratio of individual sample to reference pool. A total of only 81 dots were flagged as unusable in all of the arrays. Each antibody is dotted in triplicate on the arrays, and we computed the geometric mean of the 3 ratios from these replicate dots as our unadjusted values for each antibody (with now just one piece of missing data). A major difficulty of the analysis was the large non-specific signal in the 532 channel, with intense signals even in the two control arrays made in batch 2, one of which used unlabeled sera only, and the other which used a biotin-ULS labeled reference pool sample versus an unlabeled sample. Let Di be the average of the 532 channel intensities for these two test arrays for the i-th antibody, and let Mi be the minimum of the 532 channel intensities for the i-th antibody for the 42 arrays of major interest (which was sometimes less than Di). We used a conservative estimate of non-specific 532 channel intensity of Ci = min(Di, Mi-200) to adjust the ratios as follows. Let Rij be the unadjusted average ratio and Sij the average 635-channel intensity for the i-th antibody on the j-th array. We adjusted the ratios by using Rij – (Ci/Sij), and then took base-2 logarithms to obtain the adjusted log-ratios. The adjusted log-ratios were normalized to an artificial standard computed by median centering these values for each array, and then averaging across arrays for each antibody. Using the median centered array data we selected the 60 antibodies with smallest standard deviation across arrays, and forced the mean for these 60 antibodies on each array to agree with the same mean for the standard. We excluded data from 2 hybridizations that had very bright dots and appeared as outliers in preliminary principal components analysis (samples 47 and 77). We then fit a two-way ANOVA model with terms for batch, sample type (+GVHD or –GVHD), and their interactions (no interactions gave p<.01). In the matrix we have removed batch effects by forcing the the mean log-ratio in each batchs to be equal so that the data can be more easily compared. We also provide a supplemental data file holding the data where batch effects were not removed.
