Optimization of protein array pixelation

A number of parameters that could affect the performance of the protein array pixelation were examined to optimize the method (Fig. 2). The first consideration is the size of the pixel used for tryptic digestion. The smaller the pixel size, the more total samples will need to be analyzed by LC-ESI-MS/MS, thereby substantially increasing the total time needed to completely analyze a proteome. To determine the effect of pixel size, a test sample of nondepleted serum was loaded on multiple lanes of a 1-D gel and electrophoresed for the full distance. When the same 4 mm region of a gel lane was examined with pixel size of 1 mm (four pixels total), 2 mm (two pixels total), and 4 mm (one pixel total), the largest number of nonredundant proteins was identified from the four 1 mm pixels analyzed separately. When 2 mm pixels were used the number ofidentified proteins decreased moderately but when a 4 mm pixel was used the decrease was dramatic (Fig. 2A, columns 1-3). Even though the total analysis time decreased four-fold with the single 4 mm pixel analysis versus four 1 mm pixels, the 58% decrease in high-confidence protein identifications (>2 peptides) is clearly unacceptable. The 2 mm pixel size is a good compromise between the total analysis time and the number of proteins detected, since compared with 1 mm pixels, the analysis time was reduced by 50% and the high-confidence identifications were reduced by only 15% (Fig. 2A).

In these analyses, the number of protein identifications could be improved by increasing the sample injection volume from 2 to 4 mL (21% increase in high-confidence proteins; Fig. 2A, columns 3 and 4). Although this is a modest increase, it does not increase the analysis time and is therefore a positive factor. Increasing the RP-LC gradient time increased the high-confidence protein identification by 36% and the analysis time by 27% (Fig. 2A, columns 3 and 5). Hence, this change had a marginal advantage. A greater increase was observed when 10 mm C18 particle size POROS R2 resin was replaced with 5 mm MAGIC C18 particle size resin where a 57% increase in high-confidence proteins was obtained for a constant analysis time (Fig. 2A, columns 6 and 7). Extending the column length from 10 to 20 cm did not appreciably increase the number of proteins identified (6% increase in high-confidence proteins), but substantially improved protein coverage, since a 30% increase in the number of proteins with >3 peptides was observed with the 20 cm column (Fig. 2A, columns 7 and 8).

Fig. 2 Parameters affecting the efficiency of the protein array pixelation strategy. (A) Bar chart displaying the effect of pixel size (columns 13), sample injection volume (columns 3 and 4), RP-LCtime (columns 4 and 5), type ofC18 resin (columns 6 and 7), and column length (columns 7 and 8) on the number of nonredundant proteins identified. P, POROS R2 C18 10 pm; M, MAGIC C18 5 pm. (B) Bar charts showing the effect of gel separation distance and gasphase fractionation on the number of nonredundant proteins identified. Number of proteins identified from the human plasma F3

Fig. 2 Parameters affecting the efficiency of the protein array pixelation strategy. (A) Bar chart displaying the effect of pixel size (columns 13), sample injection volume (columns 3 and 4), RP-LCtime (columns 4 and 5), type ofC18 resin (columns 6 and 7), and column length (columns 7 and 8) on the number of nonredundant proteins identified. P, POROS R2 C18 10 pm; M, MAGIC C18 5 pm. (B) Bar charts showing the effect of gel separation distance and gasphase fractionation on the number of nonredundant proteins identified. Number of proteins identified from the human plasma F3

MicroSol-IEF fraction electro-phoresed for 1 cm (10 x 1 mm size pixel) and 4 cm (20 x 2 mm size pixel) on 1-D gels are shown. Gas-phase fractionation of the F3-7 pixel from the human plasma sample was analyzed using the full m/z range of 375-1600, or with three separate m/z ranges as indicated. Last column shows the combined number of nonredundant proteins identified from the three separate m/z ranges. Number of proteins identified by 1, 2, and >3 unique peptides are indicated by the white, black, and gray bars, respectively

Fig. 3 Major protein depletion and MicroSol-IEF separation of human plasma proteins.

(A) 1-D gel showing the plasma proteins [P] before depletion, and unbound [UB] and bound

[B] proteins from the MARS antibody column. Tr, transferrin; Alb, albumin; aT, antitrypsin; HC, Ig heavy chain; Hp, haptoglobin; LC, Ig light chain. (B) Seven MicroSol-IEF fractions ofthe depleted plasma proteins were subjected to 1-D gel separations for a total distance of 4 cm from the bottom ofthe wells. Separation of F3 fraction for 1 cm is shown in the right panel. Proteins were separated on 10% bis-Tris NuPage gels using MOPS buffer, and stained with Colloidal blue.

Fig. 3 Major protein depletion and MicroSol-IEF separation of human plasma proteins.

(A) 1-D gel showing the plasma proteins [P] before depletion, and unbound [UB] and bound

[B] proteins from the MARS antibody column. Tr, transferrin; Alb, albumin; aT, antitrypsin; HC, Ig heavy chain; Hp, haptoglobin; LC, Ig light chain. (B) Seven MicroSol-IEF fractions ofthe depleted plasma proteins were subjected to 1-D gel separations for a total distance of 4 cm from the bottom ofthe wells. Separation of F3 fraction for 1 cm is shown in the right panel. Proteins were separated on 10% bis-Tris NuPage gels using MOPS buffer, and stained with Colloidal blue.

We also examined the effect of 1-D gel separation distance on the number of proteins identified. For this analysis, a MicroSol-IEF fraction of the major protein-depleted plasma sample (F3, see below) was electrophoresed for a total distance of 4 or 1 cm (see Fig. 3B). The 4 cm lane was divided into 2 mm pixels for a total of 20 pixels, whereas the 1 cm lane was analyzed as 1 mm pixels for a total of 10 pixels. The total number of nonredundant proteins identified from the 4 cm lane was 56% greater than the 1 cm lane and the high-confidence identifications increased by 14% (Fig. 2B). Because longer gel separation distances are likely to increase the total number of analyses per proteome, the benefits of increased SDS gel separation distance are ambiguous. If a substantial number ofthe identifications based on one peptide are correct, the increased analysis time may be worthwhile.

A well-known major factor that limits peptide identification capability of complex peptide mixtures using LC-ESI-MS/MS is coelution of more peptides from the RP column than the mass spectrometer can analyze. One method of addressing this problem is gas-phase fractionation, where a single sample is repeatedly analyzed using different segments ofthe full m/z range in each run [25]. To test the utility of gas-phase fractionation in the current method, a 2 mm pixel (F3 pixel 7, depleted plasma sample; see below) was analyzed using the unsegmented m/z of 375-1600 approach and compared with gas-phase fractionation using three separate m/z segments of 375-780, 780-1200, and 1200-1600 (Fig. 2B). In the gas-phase frac-

tionation experiment, most proteins were identified using the m/z range of 7801200. In contrast, least proteins were identified using m/z of 1200-1600, and all proteins identified in this segment were also found in the other two segments (data not shown). However, peptides identified in the m/z 1200-1600 segment are important because they increased sequence coverage of many proteins. By combining the three m/z segments, the total number of nonredundant proteins identified increased by 47% compared to the single unsegmented analysis. However, the number of high-confidence proteins increased by a marginal 6%. Taking into consideration the three-fold increase in the analysis time, the segmented approach does not appear to be an efficient strategy for comprehensive proteome analysis using the protein array pixelation strategy.

Was this article helpful?

0 0

Post a comment