This study aimed to investigate the efficiency of a range of potentially useful detergents to increase the ion abundance and ion population from fingermark protein-originating peptides, within a shotgun proteomic approach involving in situ proteolysis prior to MALDI MS analysis. In the initial stage of method development, proteolysis was performed with and without the presence of surfactants. In the former case, trypsin solutions containing 0.5% of either of the non-ionic detergents, OcGlu (conventionally used detergent since the work of Djidja et al. ), OcThio, DDM, and MEGA-8, were spotted onto ungroomed fingermarks. Mass spectra generated from the direct profiling of digested fingermarks are displayed in Figure 2, showing that the addition of any of the detergents used results in higher ion abundance and ion population, compared with trypsin alone (Figure 2a), in agreement with previous observations for on-tissue digests. Generally, OcGlu and OcThio exhibit similar peptide peaks, albeit in higher intensities within the OcGlu digest (Figure 2b and c, respectively) and this could be expected because of the near-identical chemical structures (Figure 1a and b).
From manual spectral interrogation, OcGlu and OcThio appear to yield a greater number of tryptic peptides in the lower mass region, whereas DDM (Figure 2d) shows a richer peptide ion population throughout the mass range acquired. For a more objective and systematic evaluation, the majority of the in silico digests of the known protein targets reported were examined; it was observed that apart from exceptions (i.e., m/z 2038.0161 QEIECQNQEYSLLLSIK from Keratin-9), the m/z of peptides fall within the fractional range 0.4–0.8 (in line with what the proteomic community considers the fractional range of peptides to fall within). Therefore, only the m/z signals falling within this range were considered as peptides in the spectra; these peptide signals were used (their number and relative intensity) to allow a uniform relative comparison between the performance of the different detergents. Figure 2f shows the median values with a range for the number of peptides between 0.4 and 0.8 for each detergent. All the detergents yielded a comparable number of peptides (Figure 2f). OcGlu, followed by OcThio, produced the largest range comparing data from the three repeat experiments; conversely, the small range on the DDM and MEGA-8 bars demonstrated reproducibility.
The MALDI mass spectrum generated by employing MEGA-8 as a detergent has been used to search for proteins that have been shown to be present in fingermarks or sweat by others using different non-in situ approaches [19, 22]. As the mass accuracy of the instrument upon calibration was 10 ppm, putative assignments, reported in Table 1, were made on the basis of the peptide maps using in silico digestions using the bioinformatic tools hosted by the Expasy portal (http://www.expasy.org/). Interestingly, psoriasin, an antimicrobial protein, putatively detected by Ferguson et al.  in the intact analysis, was also detected here through a peptide at nominal m/z 2750.2882 with a relative error of 9 ppm. Taking stock from these observations, further method development involved in situ digestions performed by spraying the trypsin solution instead of spotting it, with the view to enable MALDI MS imaging to be applied to fingermark specimens as well as to other tissues in the future. Given their performance in the spotting experiments, OcGlu, DDM, and MEGA-8 were employed in two separate experiments: (1) application of the trypsin solution containing either of these three detergents through spray-coat and (2) spray coat application of a trypsin solution containing all of the three detergents in order to maximize coverage and encompass the benefits of each detergent from the spotting experiments. OcGlu has been published as a proven additive to enhance proteolytic digests [7, 12]; although current investigations show that this detergent produced a large range of peptide numbers across the three repeat experiments (Figure 2f), the inclusion of this detergent was used for comparative purposes.
In order to determine whether the overall intensity of peptides changes across the spectra generated by the different detergents, the ratio between the sum of peptide intensities against a single matrix peak intensity (m/z 1072.1047, [CHCA – 4H + 4Na + 1 K]+) was calculated (Figure 2g). Taking into account three repeats, OcGlu produced spectra with the highest peptide:matrix intensity ratio (average of 46.5:1); MEGA-8 consistently produced spectra containing peptides of an overall higher relative intensity (average of 29.8:1) than DDM-generated peptides (average of 12.7:1 respectively). OcThio was not carried forward into the second set of experiments as, in terms of peptide:matrix intensity ratio in the spotting experiments, the range of values across the three mass spectra replicates was much higher than that yielded by OcGlu (that is, the use of OcThio gives rise to hot spots) (Figure 2g).
In experiment (1), on-tissue digests by trypsin spray-coating were carried out on ungroomed fingermarks using either 0.5%, 1%, or 2% detergent. Figure 3a shows the number of peptides with a fractional mass between 0.4 and 0.8 represented as median values with range. Of the three detergents, MEGA-8 produced reproducible peptide numbers across the three experimental repeats as demonstrated by the narrow ranges. It is apparent that the 2% MEGA-8 digest produces the highest number of peptides (Figure 3a), which corroborates the detailed visual inspection of the mass spectral data when zooming in the spectrum (data not shown). The number of peptides varied across replicates for both OcGlu and DDM at all three concentrations, as indicated by the ranges. Interestingly, for all of the three detergents at 0.5% concentration, generally the peptide intensity:matrix intensity ratio is actually higher when the trypsin solution is sprayed compared with when it is spotted. This could be due to the low abundance of proteins in this particular specimen investigated (lower than in classic tissue sections) and simply, when spraying, the ratio detergent/endogenous proteins and/or trypsin/endogenous proteins, could be more favorable than those from spotting experiments. For method development, this is an encouraging observation towards future imaging experiments as it shows that signal intensity is not compromised when trypsin is sprayed. Second panel on the right hand side of Figure 3 shows representative peptide profiles from the three detergents when mixed together and added to 20 μg/mL trypsin solution at 0.5% concentration (OcGlu:DDM:MEGA-8) at various relative ratios of (c) 1:1:1, (d) 1:2:1, and (e) 1:1:2. Trypsin containing at any one time the three detergents at 1% yielded very similar distribution of peptides to those obtained using the detergents in 0.5% concentration (Supplementary Figure S1D–F). However, notably, all three detergents at 1% concentration yielded overall much lower intensities for the same peptides. In theory, especially from the 1% detergent concentration experiments, it could be expected that an even higher concentration of detergent may result in lower trypsin efficiency because of possible denaturation of the enzyme or peptide ion suppression. However, this was not the case for DDM or MEGA-8. Remarkably, using 2% MEGA-8, MALDI spectra contain both the highest number of peptides and peptide signals exhibiting the highest intensity overall (Figure 3a and b). From conducting this systematic study it is apparent that it is reasonable to measure the efficiency of a detergent based on three criteria: the ability to (a) yield the highest number of peptides, (b) yield the highest peptide:matrix intensity ratio overall, and (c) to reproducibly yield (a) and (b) across three replicates.
Whilst DDM at 2% produced the highest peptide:matrix intensity ratio (Figure 3b), the large range across replicates and considerably low number of peptides produced in comparison to MEGA-8 at 2% (Figure 3a) would indicate that overall MEGA-8 was the more efficient detergent. In contrast to the spotting experiments, these spectra allowed peptide mass fingerprinting to be undertaken by assigning multiple peptides for many of the tentatively assigned proteins. For example, keratin type 1, previously putatively identified by assigning one peptide, through spotting a trypsin solution containing 2% MEGA-8, is now identified through two peptides at lower m/z (m/z 1060.5605 and 1323.6727) in the spraying experiments. Interestingly, dermcidin, a small antimicrobial protein, putatively detected by Ferguson et al. , was also detected here through five peptide assignments at m/z 676.3829, 725.3932, 1128.5365, 1459.7622, 1466.7872. Here it is important to actually specify that this assignment could be made to dermcidin or a dermcidin-derived species originating from the in vivo cleavage of this small protein. However, at this stage, this accurate information, possibly requiring the use of a second enzyme to confirm, is outside the scope of this paper. Another antimicrobial species, LL-37, the C-terminal part of the only human cathelicidin identified to date called human cationic antimicrobial protein (hCAP), was also putatively detected through the peptide at m/z 1365.6610. The peptide assignments for this “spray-coat” 2% MEGA-8 experiment are shown in Table 2. The spraying method holds the promise to be able to identify lower abundance species since both dermcidin and LL-37, absent in the spotting experiments, are present in sweat at a lower abundance than serum albumin and keratin (previously identified in the spotting experiments). Also, it was observed that in the spraying experiments, protein identifying peptides were generally of lower m/z than those in spotting experiments; just as one example, psoriasin was putatively identified through the peptide at m/z 2750.2882 in the spotting experiments and through the peptide at m/z 1385.7194 in spraying experiments. We speculate that this is due to a higher efficiency of analyte extraction in the spotting experiments, especially with regards to higher m/z species. However, identifications through lower m/z peptide maps are advantageous as these species would be easier to fragment and have generally higher signal intensities. It should be taken into account that the overall proteolysis efficiency (number of peptides generated and intensities) is not solely dependent on the intrinsic efficiency of the detergents but also on the nature and abundance of the proteins . As different detergents may denature specific classes of proteins, it can be hypothesized that mixing the detergents in appropriate ratios could allow the benefits of each to be combined to increase the number and types of peptides and proteins identified .Therefore, detergents were mixed and tested in experiment (2) at different relative ratios: (a) OcGlu:DDM:MEGA-8 at 1:1:1, (b) 1:2:1, and (c) 1:1:2. The surfactants were used at different concentrations and, in the mixed detergents experiments, the 0.5% concentration worked best (data not shown). When examining the resulting MALDI mass spectra, in Supplementary Figure S1A–I, the ion population appears to be confined to <m/z 1800 for the individual detergents, whereas the spectra originating from using the mixed detergent “formulation” produces an even spread of signals from m/z 600 through to m/z 2800 (Figure 3c–e), thus verifying the initial hypothesis. As anticipated, the benefits encompassed from the three detergents can be seen with ion population spread across the entire mass range regardless of the differences in detergent proportions, though, amongst the trialled ratios, the 1:1:2 ratio yielded the highest peak intensity and the richest ion population. Though the entire mass range was more ion populated, the ion intensity was generally lower than that observed for the application of the individual detergents. Therefore, to further improve the peptide intensities, future work should be carried out testing the same ratio albeit with a higher surfactant concentration (from 0.5% to 2%), in particular of MEGA-8, given that this detergent showed to be working at its best when present in 2% concentration.
Another surfactant of interest in these investigations was RapiGest SF. MALDI MS spectra resulting from the addition of RapiGest SF to trypsin sprayed onto an ungroomed fingermark shows an improvement in the number of tryptic peptides produced and increase in ion intensity (Figure 4a) compared against trypsin alone (Figure 4b). In particular, there are peptide peaks between m/z 1000 and 2000 Da, present in the RapiGest SF generated peptide profile that are absent in the tryptic digest alone. RapiGest SF was used at 0.1%, which is the optimum concentration suggested by the manufacturer Waters for in solution experiments. Additionally, results indicate that the inclusion of 0.1% RapiGest SF produces a similar number of ion signals as for the 2% MEGA-8 digests, albeit of lower intensity, except in the 1000–1200 Da mass region, showing good scope for further experiments for the optimization of the RapiGest SF concentration. However, the novel in situ use of RapiGest SF reported in this study may require further development in order to determine the optimal concentration for this kind of proteolysis.
In view of making this alternative protocol useful for other biological specimens, additional experiments were carried out on fresh frozen rat brain tissue comparing the performances of sprayed OcGlu and MEGA-8 detergents (the most efficient detergents) at 0.5% concentration. The spectra generated several hundred peaks with S/N ratio above 3.5. Most of the peaks were considered to be from the most abundant proteins such as the putatively identified myelin basic protein (MBP); these ion signals were assigned using a 10 ppm mass tolerance and included m/z 615.2892 (DSHTR, 7.6 ppm error), 643.3698 (RPSQR, 9.9 ppm error), 705.3278 (SGSPMAR, 9.9 ppm error), 709.3462 (FSWGGR 6.4 ppm, MBP isoform 4), 726.4108 (HGFLPR, 8.7 ppm error), 811.4387 (TPPPSQGK, 9.7 ppm error) and 1339.7151 (HRDTGILDSIGR, 5.6 ppm error) which acted as a reference for evaluation of the different detergent performances. This comparative strategy, was informed by the work of the European Union COST Action Programme BM1104 "Mass Spectrometry Imaging: New Tools for Healthcare Research" in which the authors are participants. This program has networked 24 countries having, amongst others, excellent expertise in in situ tryptic proteolysis; after carefully designing a round robin study (agreed by the 24 countries), MBP (an extremely abundant protein in brain tissue) was selected as a target to compare trypsin digestion conditions for rat brain tissue and draw conclusions on best practice and most efficient protocols. The paper is in preparation (under revision of the co-authors) and will be soon submitted.
One additional finding was that under closer examination of the higher mass regions, peak intensities originating from the use of MEGA-8 are clearly superior to those of OcGlu as shown in Supplementary Figure S2, in agreement with the corresponding observations for fingermark digests. Also, less peptides could be putatively assigned to MBP when using 0.5% OcGlu in the proteolytic solution (m/z 615.2835, 643.3671, 709.3480, 1460.7118).
In Figure 5, a representative example of the above observations is reported in which the peptide at m/z 726.4108 is absent in the 0.5% OcGlu spectrum digest (Figure 5a), is instead clearly present at high intensity within the 0.5% MEGA-8 spectrum digest (Figure 5b). A rat brain tissue section was also submitted to MALDI MS imaging using 0.5% MEGA-8 in the proteolytic solution showing the opportunity for efficient chemical mapping (Figure 5c).
The mixed detergent trypsin solution previously trialled using fingermarks was also employed to digest rat brain tissue using a 0.5% concentration for each detergent at the same ratios previously tested. This time OcGlu:DDM:MEGA-8 1:1:1, and not the 1:1:2 ratio, produced the best peptide profile of the three mixed detergent digestion (Supplementary Figure S3), and tentative identifications were made as to the presence of MBP peptides (m/z 726.328 and 1339.556). However, in this case the mixed detergent tryptic solution produced an overall lower spectral intensity than OcGlu and MEGA-8 applied separately. This suggests that it cannot be assumed that one protocol is directly transferable from one sample type to another, and protocol refinements are indeed needed when developing a mixed detergent method for a different tissue model. Finally, rat brain tissue was digested with trypsin containing 0.1% RapiGest SF to evaluate transferability of this novel application method for this detergent to other specimen. The resulting mass spectrum exhibited peptides of high intensities in the low to mid mass range (m/z 600–1500), as shown in Supplementary Figure S4. Peptide peaks were generally of higher intensity compared with those generated from the mixed detergents proteolytic solution, though, as observed above, further optimization is needed both for the RapiGest SF and the mixed detergents when applied to rat brain tissue in this case.