$$\rightleftharpoonup{xx}$$
$$\longleftharp{xx}$$,
$$\longrightharp{xx}$$,
The automated proteomics sample preparation workflow on the automated workstation was adapted from our previous automated protocol with a robust LC-SRM-MS acquisition method1 for albumin, the selected plasma protein, and β-galactosidase (β-gal), an exogenous protein used for quality control. After processing, the samples were run on a triple quadrupole LC-MS in an SRM assay targeting serum albumin, β-galactosidase. The coefficient of variation (CV) of SRM signal for each transition was used to monitor the reproducibility of automated digestion protocol.
We automated the reagent addition and mixing steps for a digestion of 5 μL plasma samples with a 2-hour on-deck-incubator trypsin incubation. To determining the reproducibility, 5 μL of a plasma pool was pipetted into multiple wells of a reaction plate with a multichannel head (96 pin). To monitor for consistency, we added β-gal protein before the reduction and alkylation reactions. The automated proteomics sample preparation workflow was tested with a robust LC-SRM-MS acquisition method including albumin, the highest abundance plasma protein, and β-gal protein, used for quality control. Three β-gal peptides and two albumin peptides were monitored from processed plasma albumin proteins and spiked β-gal protein (Table 4).
In an effort to save time and simplify the procedure, we reduced the liquid transferring steps of adding/mixing/incubation of reagent and reaction mixture with the workstation from a nine-step workflow to a six-step workflow (Figure 1). The total proteomic workflow was comprised of two experimental components: automated sample preparation and LC-MS/MS. Firstly, we evaluated the precision of LC-MS/MS SRM data acquisition by eight consecutive injections from the same digestion well of the autosampler plate. The precision of the automated sample preparation workflow was calculated by the percent of coefficient of variance (%CV) of total proteomic SRM workflow minus%CV of the LC-MS/MS (Figure 9B). With the streamlined plasma digestion procedure on the automated workstation, the experimental precision of automated samples preparation was less than 11.4% for exogenous spiked β-gal proteins (10.0% as average), and less than 14.9% for the most abundant human serum albumin (9.9% as an average) (Figure 9). Good signal intensities were observed for both human serum albumin and β-gal proteins as expected (Figure 9C).
For each pipetting and liquid transferring step, techniques were specifically optimized. To monitor the precision of liquid transferring steps, we spiked stable isotope-labeled (SIL) peptide standards for endogenous human serum albumin protein and exogenous β–gal protein in three independent reagent transferring steps: Reaction Mix 1, Cysteine Blocker, and Reaction Mix 2 (Figure 10). MRM signals from these five SIL peptides were acquired to monitor the precision of automated liquid transferring steps. The average %CV for peptides, DDNPNLPR^, and GDFQFNISR^ (^ represents the N15 labeled amino acid) from Reaction Mix 1 step ranged from 1.8% to 11.2%. The average %CV of one peptide (IDPNAWVER^) from the Cysteine Blocker step ranged from 6.6 to 8.8%. The average %CV of two peptides (WVGYGQDSR^ and LVNEVTEFAK^) ranged from 6.2% to 11.9% (Figure 10).
To validate the automated proteomics sample preparation workflow, we evaluated reproducibility across multiple proteins and multiple days for human serum albumin, exogenous β–gal, and 40 additional plasma proteins. We processed 21 replicate samples (pooled normal human plasma), well location shown in Figure 11A, on three different days. Intra-day CVs were calculated from 21 wells prepared on the same day. The mean intra-day %CVs for 40 proteins ranged from 4% - 20% (Figure 11B). To evaluate the edge effect of the plate based automated workflow, %CV was calculated from specific wells within designated columns and rows (Figure 12A for column and row map). MRM signals intensities were similar in all column and row configurations with %CV ranging from 3% - 22% (Figure 12B).
In summary, the optimized automated workflow yields 96 uniformly-processed samples in less than five hours with excellent experimental precision. For compatibility with an automated workflow, we selected reagents that have negligible non-specific side reactions, are stable in ambient light, are LC-MS/MS friendly, and can be stored as frozen aliquots.

Figure 1: Schema of sample preparation workflow. The main 6 liquid transferring steps are listed. Please click here to view a larger version of this figure.

Figure 2: Tip loading Script. The VB Script details are shown in here. The script specifies tip loading conditions. Please click here to view a larger version of this figure.

Figure 3: Automated workstation Deck layout. The deck consists of 1x1 ALPs, Tip Loading ALPs, Trash, Tip wash, Peltier and an incubator. Please click here to view a larger version of this figure.

Figure 4: Properties of the reagent plate. Shown are properties needed for the Reagent Plate labware when accessed using the Guided Labware Setup. Select the corresponding column and type in the variables as indicated in the figure. Please click here to view a larger version of this figure.

Figure 5: Tip counting script. This script helps to keep track of number of tips on the deck. Please click here to view a larger version of this figure.

Figure 6: Overview of the method for digesting and aliquoting plasma samples. Steps for reagent calculations, labware setup, and liquid manipulations in the liquid handler’s method. Please click here to view a larger version of this figure.

Figure 7: Layout of the reagent plate. Shown are the chemical reagents needed for plasma digestion and autosampler preparation and distributed across the reagent plate labware. This figure has been modified from a technical note10. Please click here to view a larger version of this figure.

Figure 8: Layout for the labware on the deck of the automated workstation. Shown is the deck layout for the plasma digestion method for the liquid handler. This figure has been modified from a technical note10. Please click here to view a larger version of this figure.

Figure 9: The precision of total proteomic workflow is comprised of workstation CV and targeted LC MS/MS CV.
Five peptides from β-gal and albumin were monitored, the chromatograms and peptide retention time for each peptide is shown (A), Precision was determined from 30 wells/samples processing representative experiment, CVs% for total proteomic workflow, LC MS/MS analysis and automated sample processing were calculated (B). Overall, the digested peptides showed good signals ranged from 1x105 up to 1x108. Please click here to view a larger version of this figure.

Figure 10: Determination of precision for liquid transferring steps. Synthetic peptides were spiked in step specific reagents, and automated liquid transferring was determined by total%CV. From 30 wells/samples experiment minus% CV LC MSMS (determined by 8 repeat LC MSMS injections. Please click here to view a larger version of this figure.

Figure 11: Multi-days reproducibility of automated proteomic sample preparation workflow with 42 protein MRM analysis. (A) Reaction plate map for each of three days is show here, 5 µL plasma was added to each of 21 wells. 3 wells received 5 ul water were used as negative/blank controls. (B) The average intensities of 190 transitions comprise of 75 peptides and 42 proteins (left) and%CV for each MRM transition was calculated from 21 wells digestion for each day (right). Please click here to view a larger version of this figure.

Figure 12: Reproducibility of specific locations of wells (position of columns and rows). (A) Columns and Rows location with a plate map are shown here. (B) Column and row location specific MRM signals of average intensities from specified wells (left) and cv% (right) from a single plate digestion. Please click here to view a larger version of this figure.

Figure 13: Screen shot of technique editor. For each Pipetting template, define the properties of liquid level sensing, Clot detection, Piercing, Liquid Type, General, Aspirate, Dispense, Mix and Calibration. The template and techniques used in this protocol are shown in Supplemental Table 2). Please click here to view a larger version of this figure.
| Variable | Variable | Value | Description |
| Autosampler | Boolean | TRUE | Use autosampler |
| Betagal | Integer | 5 | Betagal volume |
| BG1 | Integer | 0.8 | BG1 volume |
| BG2 | Integer | 0.8 | BG2 volume |
| BG3 | Integer | 0.8 | BG3 volume |
| CysteinBlocker | Integer | 1.25 | Cysteine blocker volume |
| CysteineBuffer | Integer | 0.45 | Cysteine buffer volume |
| Denaturant | Integer | 5 | Denaturant volume |
| DigestTransfer | Integer | 10 | Digest transfer volume |
| First Buffer | Integer | 25.9 | First buffer volume |
| First Column | Integer | 1 | First Column |
| HSA1 | Integer | 0.8 | HSA1 volume |
| HSA2 | Integer | 0.8 | HSA2 volume |
| lastcolumn | Integer | 12 | Last Column |
| MobilePhase | Integer | 90 | Mobile phase volume |
| Quench | Integer | 10 | Quench column |
| ReducingAgent | Integer | 5 | Reducing agent column |
| Sample | Integer | 5 | Sample column |
| SamplePlate | Boolean | TRUE | Use sample plate |
| SecondBuffer | Integer | 58.4 | Second buffer volume |
| Trypsin | Integer | 10 | Trypsin column |
Table 1: Start step variables
| Value | Variable |
| =FirstBuffer+Denaturant+ReducingAgent+Betagal+BG1+HSA1 | FirstMix |
| =CysteineBlocker+CysteineBuffer+BG2 | CysteineMix |
| =SecondBuffer+BG3+HSA2 | SecondMix |
| =(FirstMix*Columns)+30 | FirstMixWell |
| = (CysteineMix*Columns)+20 | CysteineWell |
| =(SecondMix*Columns)+10 | SecondMixWell |
| =(Trypsin*Columns)+10 | TrypsinWell |
| =(Quench*Columns)+10 | QuenchWell |
| =(FirstMixWell*8)+100 | FirstMixStock |
| =CysteineWell*8+20 | CysteineMixStock |
| =(SecondMixWell*8)+100 | SecondMixStock |
Table 2: Volume mix variables
| Type | Name | Position | Depth | Properties | Use? |
| BCDeep96Round | Reagent Plate | P5 | 1(top) | # | =not SamplePlate |
| BCDeep96Round | Reagent Plate | P5 | 1(top) | # | =SamplePlate |
| BCDeep96Round | Reaction Plate | P11 | 1(top) | | TRUE |
| Bio_RadPCR96* | Samples | P9 | 1(top) | | =SamplePlate |
| Bio_RadPCR96* | Autosampler Plate | P10 | 1(top) | | =Autosampler |
| BC90 | Empty | | 1(top) | | TRUE |
| BC90 | | | 1(top) | | TRUE |
| BC90 | | | 1(top) | | TRUE |
| BC230 | | | 1(top) | | =Autosampler |
| BC90 | | | 1(top) | | =Columns>1 |
| BC90 | | | 1(top) | | =Columns>3 |
| BC90 | | | 1(top) | | =Columns>5 |
| BC90 | | | 1(top) | | =Columns>7 |
| BC90 | | | 1(top) | | =Columns>9 |
Note:
*: Corresponds to the 96 well Bio-Rad plate.
#: Click properties and then enter the volume variables as indicated in Figure 6. |
Table 3: Setting up the guided setup
| Protein ID | Peptide sequence | Q1 Mass (Da) | Q3 Mass (Da) | Time (min) | Fragment Ion | Declustering Potential | Collision energy | Collision Cell Exit Potential |
| sp|P00722|BGAL_ELOCI | GDFQFNISR | 542.3 | 262.1 | 19.7 | +2y2 | 61 | 21 | 8 |
| 542.3 | 636 | 19.7 | +2y5 | 61 | 25 | 12 |
| 542.3 | 764.2 | 19.7 | +2y6 | 61 | 25 | 18 |
| IDPNAWVER | 550.3 | 436.1 | 18.1 | +2y7+2 | 61 | 23 | 8 |
| 550.3 | 871.2 | 18.1 | +2y7 | 61 | 25 | 18 |
| 550.3 | 774.2 | 18.1 | +2y6 | 61 | 33 | 8 |
| WVGYGQDSR | 534.3 | 782.1 | 12.1 | +2y7 | 51 | 25 | 6 |
| 534.3 | 562.1 | 12.1 | +2y5 | 51 | 27 | 6 |
| 534.2 | 505.2 | 12.1 | +2y4 | 90 | 25 | 8 |
| sp|P02768|ALBU_HUMAN | DDNPNLPR | 470.8 | 596.2 | 9.2 | +2y5 | 61 | 27 | 16 |
| 470.8 | 499.3 | 9.2 | +2y4 | 61 | 27 | 18 |
| 470.8 | 710.4 | 9.2 | +2y6 | 61 | 27 | 18 |
| LVNEVTEFAK | 575.3 | 694.4 | 18.2 | +2y6 | 73.1 | 29.6 | 18 |
| 575.3 | 937.5 | 18.2 | +2y8 | 73.1 | 29.6 | 18 |
| 575.3 | 823.4 | 18.2 | +2y7 | 73.1 | 29.6 | 18 |
Table 4: MRM parameters
Supplemental Table 1: Reagent plate Please click here to download this table.
Supplemental Table 2: Protocol template Please click here to download this table.