**4.3.1 Alternative digestion of samples**

346 Gel Electrophoresis – Advanced Techniques

One the contrary, Sypro ruby that binds to proteins through hydrophobic interactions stains hydrophilic glycoproteins quite poorly. Being most sensitive, silver staining is obviously very useful for proteomic approaches. However, the advantage with silver is hampered by its rather low linear range, making it less suitable for quantification than the other staining techniques. For plasma analyses we have therefore adopted a double staining strategy. As illustrated in figure 9, the 2-DE gel is first stained with Sypro Ruby and the proteins are quantified within a high dynamic range. The gel then can be destained and restained by silver to detect additional proteins. As these additional proteins are less abundant their intensities usually are within the limited linear range of the silver staining technique. The proteins may then be selected for MS analyses. As the sample preparation protocol is more time-consuming for silver stained gels than Sypro stained gels it is convenient to pick as many proteins as possible after the first staining step. In this step it is also important to check the optimal destaining time before MS analyses, with plasma samples at least 90 minutes (fig 9). If the

samples are not fully destained the signal to noise in the spectra are reduced.

Plasma proteins were separated with 2-DE and first stained with Sypro Ruby. The gel was then destained with 25 mM ammonium bicarbonate/50 % acetonitrile buffer and the efficiency of the process checked after different time intervals. Finally, the proteins were restained with silver. A: Proteins more stained by Sypro compared to silver. B: Glycoproteins more stained by silver. The rectangle shows an area containing additional proteins detected by silver. Residual Sypro staining of proteins after destaining are indicated with rings.

**4.3 Protein identification with peptide mass fingerprinting using MALDI-TOF MS** 

Peptide mass fingerprinting with MALDI-TOF MS is an excellent and robust technique for fast identification of plasma proteins after 2-DE (Lahm & Langen, 2000). However, several peptide peaks needs to be detected in the spectra with a high mass accuracy to avoid false

Fig. 9. Double staining of proteins

One of the most widely used ways to digest proteins before MS analyses is by trypsin, which cleaves C-terminally of lysine and arginine (not followed by proline). Although lysine and arginine often are distributed in the protein sequences in a way that provides sufficient number of peptides for identification after trypsin cleavage, this is not always so. Furthermore, it is sometimes necessary to use alternative digestion protocols in order to find specific peptides to e.g. characterize differences between protein isoforms. In these cases alternative enzymes, e.g. Asp-N (cleaves N-terminally of aspartic acid and cystein), Glu-C (C-terminally of glutamic acid) or chemical induced cleavage by CNBr that hydrolyzes Cterminally of methionine, is needed. In this study we used Asp-N as a complement to trypsin when identifying the plasma protein serum amyloid A4 (SAA4). This combined digestion approach generated almost 95 % sequence coverage of the protein (figure 10). SAA4 is a constitutively expressed protein which can be found in HDL as differently charged isoforms (fig 14, Karlsson et al., 2005). One explanation to these isoforms could be a small truncation of SAA4 in which one lysine and one tyrosine is removed C-terminally and thereby making the protein more acidic (Farwig et al., 2005). However, by using Asp-N we were able to detect both the intact C-terminal and the intact N-terminal peptide that were not possible after trypsin digestion, ruling out the presence of truncated SAA4 in our sample (figure 10). Besides the use of Asp-N to study SAA4, we have also used Glu-C to analyze apolipoprotein A-I (figure 6) and CNBr to study serum amyloid A-1/2 isoforms (figure 13).

Fig. 10. Alternative digestion to improve sequence coverage of Serum amyloid A4 (SAA4).

Sequence coverage obtained through peptide mass fingerprinting with MALDI-TOF mass spectrometry by the use of trypsin and endoproteinase Asp-N as indicated by the lines. Sequence coverage (without the signal peptide in position 1-18) was 79.5 % with trypsin and 57.1 % with Asp-N. The combined sequence coverage was 93.8%.

#### **4.3.2 Peptide sample cleaning**

Peptide samples after in-gel digestion can be cleaned by adsorption to C-18 containing pipette tips (ZipTip®). This clean-up procedure done manually is rather time-consuming but is absolutely necessary before electrospray-quadrupole MS. On the other hand, with a MALDI-TOF instrument, being more insensitive to salts and other contaminants, it is not that obvious. To investigate the possible advantage with ziptip cleaning before MALDI-TOF MS we picked 17 sypro stained proteins after 2-DE. The proteins were digested with trypsin and an aliquot of the obtained peptide solution was purified by ZipTip® (50% ACN elution solution according to the protocol recommended by the supplier), mixed with the matrix CHCA and spotted on the MALDI-target plate and another aliquot of the peptide solution was mixed directly with the matrix and spotted on the same plate. All samples were then

Two-Dimensional Gel Electrophoresis and Mass

**4.3.3 Choice of matrix** 

DE is recommended.

(DHB).

Spectrometry in Studies of Nanoparticle-Protein Interactions 349

To enhance the quality of the MALDI-TOF mass spectra and the number of desorbed peptides there are several matrices that could be considered to use. Different matrix compounds, both acidic and basic, have proved to work in sample preparation for MALDI mass analyzers. The far most commonly used matrix for peptide mass fingerprinting is CHCA, which is recommended for peptides with mass ions below 2500 Da (Beavis et al., 1992). Alternative matrices also used are sinapinic acid, mostly for masses higher than 25 kDa (Lewis et al., 2000) and DHB, originally suggested for glyco- or phosphopepides that are difficult to ionize (Strupat et al., 1991), but later also proven useful for silver stained proteins (Ghafouri et al., 2007). As illustrated in figure 11, CHCA and DHB have very different crystal structures on the target plate. Whereas CHCA usually has a homogeneously distributed spot appearance, making it ideal for automatic laser induced peptide desorption, the DHB crystals are needle shaped and often aggregated into fan-like structures directed from the outside towards the centre of the sample spot. Interestingly, we have found that peptides appear to be enriched in the base of the DHB structures (figure 11), significantly increasing the signal to noise ratio in spectra obtained from these areas. This is illustrated by the identification of transthyretin, one of the proteins that interact with silica nanoparticles and carbon nanotubes (figure 5). With DHB, the improved signal to noise in the spectrum displayed twice the number of peptides than with CHCA (figure 12). This dramatically increased the sequence coverage from 39 % obtained with CHCA to 72 % obtained with DHB. Thus, the use of DHB for low-abundant silver stained proteins from 2-

Fig. 11. Crystals of α-cyano-4-hydroxycinnamic acid (CHCA) and 2,5-dihydroxybenzoic acid

A; CHCA and B; DHB as matrix on a MALDI plate. The marked area indicates the position

One of the main challenges for human proteomics is to identify and characterize co- and post-translational modifications to be able to study their relevance and place in systems biology. Most human proteins are expressed as different isoforms often depending on posttranslational modifications. Two common modifications in plasma are truncations and glycosylations and here we have used the ability of 2-DE to separate such isoforms based on

of the laser where the best signal to noise was obtained with DHB.

**4.4 Separation and characterization of isoforms** 

differences in isoelectric point and molecular mass.

analyzed with MALDI-TOF MS with the same settings and with the laser induced collection of spectra in an automatic mode. The spectra were then used for NCBI database search with MS-fit using the same settings for all samples. All 17 proteins were identified with peptide mass fingerprinting in both ziptip cleaned and untreated samples. As shown in Table 2, ziptip cleaning significantly improved the peak intensities, signal to noise ratio and the mass accuracy. This illustrates that removal of salts and other contaminants that will compete with the peptides in the spectra increases the intensities of the peptide peaks and thereby increases the accuracy of the mass determinations and, as a consequence, also increases the reliability of the identifications. On the other hand, the number of peptides and sequence coverage found in ziptip cleaned samples were about the same as in the untreated samples (Table 2). In general, there was a clear tendency that in ziptip cleaned samples more peptides were detected in the lower mass region (<1000 m/z) while fewer peptides were detected in the higher mass region (>2000 m/z). This suggests that the removal of salt ions and low molecular chemicals with subsequent improved signal to noise increases the possibility to detect low molecular mass peptides but that this beneficial effect is counteracted by adsorption of larger peptides to the solid phase of the ziptip. To test this, 10 protein samples were sequential eluted from the ziptip with increasing ACN concentration, up to 90 %. Indeed, this procedure increased the number of peptides found and the sequence coverage increased from 52 +/- 16 % in the untreated samples to 58 +/- 16 in the ziptip cleaned samples (p<0.05). The effect varied among the different proteins but was in some samples quite profound, almost 2 times higher sequence coverage. It can be concluded from these experiments that purification of peptide samples with ziptip improves the results with MALDI-TOF MS. However, when it comes to the identification of proteins with peptide mass fingerprinting the beneficial effect of the cleaning procedure is quite limited as the number of peptides found, using the standard protocol, is not increased. Therefore, considering the work-load needed for the ziptip procedure, it is doubtful if it is practical to routinely clean samples with ziptip before MALDI-TOF analyses. However, for selected, low abundant, samples it can most likely make a significant difference for the identification. In these cases, sequential elution of the peptides from the ziptip with increasing acetonitrile concentrations is recommended.


Table 2. Influence of sample cleanup of in-gel digested proteins on peptide mass fingerprinting data obtained with MALDI-TOF MS.

Proteins were separated by 2-DE and in-gel digested by trypsin. The same peptide samples were then purified by ZipTip, mixed with the matrix and spotted on the MALDI plate or directly mixed with the matrix and spotted on the plate. Statistical interpretations were done by Wilcoxons signed rank sum test.
