Systematic protein array pixelation of the human serum proteome

Protein array pixelation of the HUPO serum sample, BDCA02-Serum, was performed using a method similar to that used for the plasma sample, except this method incorporated several refinements to further improve coverage of the pro-teome (Fig. 5A). The major protein depletion was performed on 415 mL (35.3 mg) of serum using a dual MARS column. A total of 4.3 mg of unbound proteins were recovered, indicating that approximately 88% of the total serum protein content was removed in this sample compared with the removal of 83% of total plasma proteins. This difference was at least partially due to more effective removal of targeted major proteins using the dual column compared with the single MARS column depletion of the plasma. This is consistent with the number of albumin peptides observed in both samples after depletion, where 9.2% sequence coverage of albumin was obtained from the depleted serum compared to 59.1% sequence coverage from the depleted plasma sample. Similarly, serotransferrin was identified with 40.7% sequence coverage in the depleted plasma sample but was not detected in the depleted serum sample. Since minor amounts of major proteins such as albumin (~40 mg/mL concentration) are still major components of the sample, they will still interfere with the overall analysis. Hence, it is better to use a longer antibody column and under-load the column to ensure the most effective depletion of targeted proteins as possible.

Following reduction and alkylation of the depleted serum proteins, the sample was fractionated into five pH fractions by MicroSol-IEF (Fig. 5B). The fractionation was performed using the commercially available pH membrane partitions, which greatly simplify the MicroSol-IEF procedure. To compensate for the reduced MicroSol-IEF fractions, the majority of the fractions (F2-F5) were separated on 1-D gels for a total distance of 6 cm. Compared to the 4 cm separation of the plasma sample, the longer separation distance should allow for increased sample loading (up to 50%) without overloading the gels. Due to the lower amount of proteins in F1, this fraction was concentrated by acetone precipitation and electrophoresed for only 2 cm to minimize empty regions in the gel lane (Fig. 5B). In addition, we also

Fig. 5 Protein array pixelation of the HUPO serum sample (BDCA02-Serum). (A) Diagram showing the improved methodologies for analysis of the human serum sample. Steps identical to those used for analysis of the plasma sample are shown in gray. (B) 1-D gel showing the five MicroSol-IEF fractions (F1-F5) of major protein-depleted human serum. Ml shows protein extracted from the pH 3.0 membrane. Separation distances for each fraction are as indicated. Proteins were separated on 10% bis-Tris NuPage gels using MOPS buffer, and stained with Colloidal blue.

Fig. 5 Protein array pixelation of the HUPO serum sample (BDCA02-Serum). (A) Diagram showing the improved methodologies for analysis of the human serum sample. Steps identical to those used for analysis of the plasma sample are shown in gray. (B) 1-D gel showing the five MicroSol-IEF fractions (F1-F5) of major protein-depleted human serum. Ml shows protein extracted from the pH 3.0 membrane. Separation distances for each fraction are as indicated. Proteins were separated on 10% bis-Tris NuPage gels using MOPS buffer, and stained with Colloidal blue.

extracted proteins from the membrane partition (M1) between the anode buffer and F1 to detect proteins that might be trapped in the membrane. The Ml fraction was also concentrated by acetone precipitation prior to 1-D gel separation for 2 cm (Fig. 5B).

Following gel electrophoresis, the gel lanes containing Ml and F1 were analyzed as 2 mm size pixels. In the initial optimization studies presented above, more unique proteins were identified from four 1 mm size pixels than a single 4 mm pixel (Fig. 2A). To potentially increase the number of proteins identified, gel lanes containing the F2-F5 fractions were pixelated in a variable manner (1-4 mm size pixel) depending on the band intensity. Regions of the gel with intense staining were analyzed as 1 mm pixels, and regions without much staining were analyzed as 4 mm pixels (Figs. 5B, 6A). For a direct comparison of pixel size using current methods, fraction F3 was also reanalyzed as uniform 2 mm pixels. In total, 159 pixels were generated for tryptic digestion and analyzed by LC-ESI-MS/MS

Fig. 6 Result from analysis ofthe human serum proteome. (A) Heat map showing distribution of the identified serum proteins in the 2-D protein array. Redundant proteins among pixels were not eliminated in this data set. Total number of proteins identified was 11 656. Total number of nonredundant proteins was 4377, which were defined by 9393 nonredundant peptides. (B) Comparison ofthe number of nonredundant proteins identified from the F3

fraction using the fixed pixelation strategy (F3f) and variable pixelation strategy (F3v). (C) Comparison of the number of nonredundant proteins identified from the Ml and F1 fractions. Proteins unique to a single data set are indicated in the "only" columns, and proteins present in both data sets are indicated in the "common" columns. Number of proteins identified by 1, 2, and >3 unique peptides are indicated by the yellow, purple, and blue bars, respectively.

Fig. 6 Result from analysis ofthe human serum proteome. (A) Heat map showing distribution of the identified serum proteins in the 2-D protein array. Redundant proteins among pixels were not eliminated in this data set. Total number of proteins identified was 11 656. Total number of nonredundant proteins was 4377, which were defined by 9393 nonredundant peptides. (B) Comparison ofthe number of nonredundant proteins identified from the F3

fraction using the fixed pixelation strategy (F3f) and variable pixelation strategy (F3v). (C) Comparison of the number of nonredundant proteins identified from the Ml and F1 fractions. Proteins unique to a single data set are indicated in the "only" columns, and proteins present in both data sets are indicated in the "common" columns. Number of proteins identified by 1, 2, and >3 unique peptides are indicated by the yellow, purple, and blue bars, respectively.

with a total analysis time of 13.4days (121 min RP-LC total run time per sample). Depending upon protein concentration, between 1.1 and 5.2% (2.5% average) of each soluble MicroSol-IEF fraction was used for gel pixelation. The average is equivalent to 10.4 mL (885 mg) ofthe original serum sample. After tryptic digestion, 23.3% of the digested material was injected and analyzed by LC-ESI-MS/MS. Therefore, protein identification was performed using an amount equivalent to approximately 2.4 mL (204 mg) ofthe original serum sample.

All samples from this serum analysis were analyzed using a Thermo Electron linear IT LTQ mass spectrometer, which is more sensitive and has a faster scan rate than the LCQ Deca XP+ [29]. The number of proteins identified for all 159 pixels is shown as a heat map in the 2-D array (Fig. 6A). Each pixel contained between 13 and 199 proteins that pass the XCorr/ACn/Sf criteria defined above. Comparison ofthe uniform (F3f) and variable (F3v) pixelation methods ofthe F3 fraction indicated that the variable pixelation method did not offer any improvement over the uniform pixelation method (Fig. 6B). In fact, uniform pixelation identified 6.9% more high-

confidence proteins compared to the variable pixelation method. In addition, the uniform pixelation method is easier and quicker to perform, as there is no need to correlate the pixel size with band intensity. The total number of nonredundant proteins identified from the 159 pixels was 4377 from a total of 9393 nonredundant peptides. Of these, 752 proteins (17.2%) were identified as high-confidence and the majority (82.8%) was single-peptide proteins (Tab. 1). Therefore, the overall improvements due to further refinement ofthe method and, most importantly, use ofthe highly sensitive LTQ mass spectrometer, resulted in about fourfold increase in the number of high-confidence proteins identified in serum compared to the plasma analysis.

The establishment ofthe pH gradient during MicroSol-IEF is dependent on the membrane partitions between each fraction [21]. During MicroSol-IEF, some proteins can be partially or completely trapped in the membrane partitions and are therefore excluded from the soluble fractions. To investigate this possibility, proteins were extracted from the five membrane partitions and analyzed on 1-D gels (data not shown). Except for M1 (pH 3.0 membrane partition between anode buffer and F1), all protein bands from the other membrane partitions appeared to have corresponding protein bands in adjacent soluble fractions. Furthermore, the protein bands from the membrane partitions are much less intensely stained than their soluble fractions counterparts. The only exception is apolipoprotein B-100 which is found predominately in the membrane partition between F3 and F4, presumably due to its large size of ~540 kDa [28]. The tentative conclusion that most proteins trapped in membrane partitions were also partially recovered in adjacent fractions was further supported by parallel analysis of membrane and soluble fractions from similar serum separations on 2-D gels (data not shown). In M1, however, two apparently unique protein bands were observed in the 1-D gel analysis (Fig. 5B shows the concentrated M1 proteins). Pixelation ofthe M1 fraction and comparison with F1 fraction indicated that 42 high-confidence proteins were identified in M1, but only 7 were not identified in F1 (Fig. 6C). Out of these seven high-confidence M1 proteins, six were not identified elsewhere in the entire serum pro-teome analysis. Hence, a few acidic proteins were found exclusively in the pH 3 membrane partition, and it is likely that a small number of unique proteins are in other membrane partitions and could be detected if higher sensitivity methods like LC-ESI-MS/MS are used instead of 1- or 2-D gels. Therefore, it may be advantageous to include membrane partition extracts in the analyses to provide more comprehensive coverage of proteins.

0 0

Post a comment