Method optimization

Prior to analysing the milk samples from the challenge study, a label-free quantitative proteomics method was optimized using milk samples collected from cows. Eight milk samples (sample no. 1 to 8) were aseptically collected from different quarters of four cows that were referred to the Scottish Centre for Production Animal Health & Food Safety, School of Veterinary Medicine, University of Glasgow, UK and used for method optimization.

1-D electrophoresis

In high-throughput proteomics analysis using LC-MS/MS, a few high-abundant proteins in skimmed milk mask the quantitation of low-abundant proteins so it is necessary to deplete caseins in skimmed milk that constitute approximately 80% of total proteins in skimmed milk and lactoglobulins in whey that constitute approximately 50% of the total whey protein to accurately quantitate low- abundant proteins (Alonso-Fauste et al., 2012, Baeker et al., 2002, Boehmer et al., 2008, Hogarth et al., 2004, Smolenski et al., 2007, Smolenski et al., 2014). Skimmed milk is the milk fraction that is obtained after removing cream (fat pellet) from milk while whey is the fluid milk fraction that is left from milk following the precipitation of caseins (Hogarth et al., 2004, Reinhardt et al., 2013). In order to obtain whey for this study, caseins but not lactoglobulins were depleted in skimmed milk as globulins were considered to be an important family of proteins in this study. While there are multiple methods available for caseins depletion, ultracentrifugation was used for this purpose (Alonso-Fauste et al., 2012, Baeker et al., 2002, Boehmer et al., 2008, Hogarth et al., 2004, Smolenski et al., 2007, Smolenski et al., 2014, Yamada et al., 2002) in this study. To test the efficiency of ultracentrifugation and the addition of calcium chloride (CaCl2)

to skimmed milk in depleting caseins, a 1-D gel electrophoresis of samples was performed on a gel (Criterion Precast Gels, Bio-Rad Laboratories) before and after ultracentrifugation of skimmed milk, and of skimmed milk samples to which various amounts of CaCl2 had been added (Figure 2.2). Similarly, to examine the

effects of dilution of skimmed milk samples with PBS, 1-D electrophoresis was performed on a gel using whey obtained from the skimmed milk samples that were either undiluted or diluted with PBS at 1:2 and 1:4 concentrations prior to the separation of whey.

Figure 2.4: Gel showing 1-D electrophoresis of milk samples used in the method optimization before and after ultracentrifugation, with various amount of addition of CaCl2 and with different amounts of dilution with PBS.

For 1-D electrophoresis, two milk samples (sample 1 and 4) were used as biological replicates. Bands labelled as kappa-casein and alpha-S2-casein (subsequently identified by LC-MS/MS) show darker bands before ultracentrifugation. Visually, there are small differences with the addition of CaCl2 or dilution with PBS. Lane 1

- protein molecular weight reference markers, lane 2 and 7 (labels 1 and 4) – milk samples before ultracentrifugation, lane 3 and 8 (labels 1A and 4A) – after ultracentrifugation without the addition of CaCl2, lane 4 and 9 (labels 1B and 4B)

- after ultracentrifugation with 20 mM CaCl2, lane 5 and 10 (labels 1C and 4C) -

after ultracentrifugation with 40 mM CaCl2, lane 6 and 11 (labels 1D and 4D) -

after ultracentrifugation with 60 mM CaCl2, lane 12 and 14 (labels 1A(1:2) and

4A(1:2)) - after diluting skimmed milk with PBS at 1:2 ratio and then ultracentrifugation without the addition of CaCl2, and lane 13 and 15 (labels

1A(1:4) and 4A(1:4)) - after diluting skimmed milk with PBS at 1:4 ratio and then ultracentrifugation without the addition of CaCl2.

For 1-D electrophoresis, two milk samples (sample no. 1 and 4) were used as biological replicates. 1-D electrophoresis results showed separation of milk proteins on the gel (Figure 2.4). Bands that were subsequently identified using LC- MS/MS as caseins were highly dense before ultracentrifugation and very light after ultracentrifugation that showed depletion of caseins. Visually, there were little differences with the addition of CaCl2 for depletion of caseins or bias due to

dilution with PBS. In order to identify proteins, the protein bands in the 1-D electrophoresis gel were excised and subjected to trypsin digestion, and the proteins in these bands were identified using LC-MS/MS at Glasgow Polyomics, College of Medical, Veterinary and Life Sciences, University of Glasgow, UK.

Mascot Exponentially Modified Protein Abundance Index (emPAI) scores (Ishihama et al., 2005) for the proteins identified from the excised bands were compared between the samples that underwent ultracentrifugation with or without the addition of CaCl2, and with or without dilution with PBS. The results showed no significant difference in depletion of caseins in samples either with the addition of CaCl2 or those samples without the addition of CaCl2, and no significant difference between the samples that were either diluted with PBS or those samples not diluted with PBS. So, as informed by the 1-D electrophoresis results and to keep the pre-processing of the milk samples to as minimum as possible to avoid any technical bias, ultracentrifugation without the addition of CaCl2 was

selected as the preferred method to obtain whey in this study and used in subsequent analyses.

Label-free quantitative LC-MS/MS optimization

LC-MS/MS data was generated from the samples described in this section (2.3.2.1) to which various amounts of CaCl2 had been added (samples 1B, 1C, 1D, 4B, 4C

and 4D), and the skimmed milk samples diluted with PBS (samples 1A(1:2), 1A(1:4), 4A(1:2), 4A(1:4)) as previously described in this section. Separation of whey, Bradford protein assay, whey protein extraction and salt removal, protein quantity normalization and trypsin digestion were performed as described in sections 2.3.2.2, 2.3.2.3, 2.3.2.4, 2.3.2.5 and 2.3.2.6 respectively. The LC-MS/MS analysis was performed in Bruker Amazon mass spectrometer using two different gradients (60-minutes and 120-minutes gradients) for optimization. The data were pre-processed using MZmine (Pluskal et al., 2010). The Bruker Amazon files in the proprietary ‘.yep’ file format were converted to ‘.mzXML’, an open data format files, and the quality of the raw data was visually assessed for consistency between the samples and chromatographic shifts by generating 2D and 3D plots from MS1 spectra using MZmine. Performance of software for identification and quantitation of proteins were compared to optimize data analysis. The quantitation and identification software used for optimization include MaxQuant (Cox and Mann, 2008), ProteoWizard (Chambers et al., 2012), Trans-Proteomic Pipeline (Deutsch et al., 2010), OpenMS (Sturm et al., 2008) and Mascot Distiller (Matrix Science, 2014).

Compared with a previous published method (Reinhardt et al., 2013), many refinements that might improve the data generation and data analysis were made and are described in detail in the following sections. While precipitation of proteins with acetone is a common method to extract proteins (especially water- soluble proteins), the efficiency with which different proteins precipitate might differ. For example, proteins with high hydrophilicity, more acidic pH or larger size (higher molecular weight) are readily precipitated by acetone (Crowell et al., 2013, Thongboonkerd et al., 2002). To overcome bias in total protein quantity introduced in extraction of proteins using acetone that could be propagated downstream, normalization of total proteins after acetone precipitation was performed by use of the Bradford protein assay. In the preparation of trypsin digests, sodium deoxycholate (C24H39NaO4) was used in addition to acetonitrile to

improve complete digestion of proteins. Sodium deoxycholate (SDC) is an ionic detergent surfactant, and is compatible with tryptic digestion up to 5% concentration (Lin et al., 2008). SDC is acid insoluble and its aqueous solutions tend to precipitate as the pH is lowered to 6.5. This property is being used in removing SDC after trypsin digestion from the protein digest potentially without detrimental loss of peptides. Previous studies comparing the use of SDC with other compounds that are used for enhancing protein denaturation for trypsin digestion showed SDC at 1% concentration improved trypsin digestion efficiency by almost 5-fold (Leon et al., 2013, Masuda et al., 2008, Proc et al., 2010, Zhou et al., 2006).