Membrane proteins are extremely difficult to isolate and characterise due to their association with the lipid bi-layer or the peptidoglycan and relatively lower abundance when in comparison with the whole cell complex. Established methods for the extraction and characterisation of membrane proteins that are commonly used include sodium carbonate precipitation, sucrose density gradients and the use of detergents to selectively solubilise and enrich the sample in favour of membrane proteins . However these methods each have their own caveats. Detergent based methods use reagents that are often directly incompatible with downstream analytical techniques and so further clean up steps are required, resulting in a lengthy workflow [12, 21] while sucrose density gradient and sodium carbonate precipitation face problems when resolubilising the membrane protein enriched fraction.
Here, we attempted to characterise the surface proteome of S. Typhimurium using Lipid-based Protein Immobilisation technology in the form of LPI™ FlowCells. The LPI™ FlowCell system provides a novel platform for the identification and characterisation of membrane proteins. No detergents are required and no sample clean up is needed prior to downstream analysis. The immobilised proteins can be digested with proteases in multiple steps to increase sequence coverage, and the peptides eluted can be characterised directly using LC-MS/MS.
Initial work highlighted the need to incorporate a wash step during the production of the intact membrane vesicles to minimise the carryover of contaminating cytosolic proteins that can potentially mask the lower abundant OMPs. The results generated showed that washing the membrane vesicles with a high pH sodium carbonate solution lowered the amount of non membrane proteins identified, and so enriching the vesicle preparation in favour of outer membrane proteins.
We have shown that a multi-step digest protocol can also be effectively used to increase total sequence coverage of proteins and to generate a list of outer membrane proteins identified with a greater confidence. However, even after incorporating a second digestion step, 17 outer membrane proteins were still only identified with one peptide hits, which is probably due to them being of low abundance. The addition of the acid cleavable mass spectrometry compatible detergent PPS Silent® was incorporated into the work flow to try and improve the solubilisation and in-solution enzymatic protein digestions of hydrophobic proteins with trypsin. Results indicated that the addition of PPS Silent® increased the total amount of different proteins identified with one peptide hits. However when counting just confident protein identifications (two or more peptide hits) this increase is less pronounced. Looking at confident protein identifications with PPS Silent®, the total number of outer membrane proteins increased from 38 to 42. However, PPS Silent® appears to enhance detection of non-membrane proteins over outer membrane proteins as the proportion of non-membrane proteins increased marginally, while the proportion of outer membrane proteins decreased in the samples subjected to PPS Silent®. This suggests that outer membrane proteins are relatively resistant to solubilising in PPS Silent®, while non-membrane associated proteins solubilise more readily.
When comparing the data generated from this study with previously published work by Coldham & Woodward, more OMPs (total of 54) were identified here in comparison to 34 reported in their study. However, there were proteins that were not identified by using the LPI™ FlowCell. Coldham & Woodward identified 34 outer membrane proteins using a method based on fractionating the whole cell lysate into its various intracellular parts coupled with two dimensional HPLC-mass spectrometry (2D-LC-MS/MS). Of the 34 outer membrane proteins identified, just over half (18) were found in our dataset. Overall there were 36 S. typhimurium OMPs identified in our dataset that were not reported previously  (Additional file 2). Some of these differences may be due to the use of different strains and variation in microbial culture conditions between both studies which will be reflected in their protein expression profiles. In addition, since the method used by Coldham & Woodward relied on multiple fractionation steps of the whole cell lysate, potential loss of outer membrane proteins, especially lower abundant ones could have occurred at each step in their workflow. Furthermore, it has been reported that results generated from mass spectrometry vary depending on the database search algorithm used to identify proteins . The work carried out by Coldham &Woodward used the search algorithm SEQUEST, while in this study the search algorithm MASCOT was used. Therefore, the differences observed between the two methods could also be attributed to the database search algorithms and parameters used. Previous work carried out by Molloy et al  identified 30 outer membrane proteins from Escherichia coli (E. coli) which is closely related to S. Typhimurium using a method based on the enrichment of outer membrane proteins using sodium carbonate washes and incorporating the detergent ASB-14 to aid in solubilising them prior 2D GE. This study manages to identify 15 of the 30 outer membrane proteins. A further 15 outer membrane proteins reported by Molloy et al were not seen in this study while 39 outer membrane proteins were identified in this study that was not reported by Molloy et al. Some of these differences may be attributed again to the different strains and growth conditions used as well as the different instrument used to identify the proteins and bioinformatics tools used to confirm the presence of outer membrane proteins. These results show there is no real consensus of proteins identified between the LPI™ FlowCell method and more established methods such as 2D GE and 2D-LC-MS/MS (Additional file 2). Instead these methods complement each other and therefore when designing experiments to identify outer membrane proteins it is important to try a range of approaches to maximise the coverage of OMPs detected.
Finally, when collating the results from both digests performed in this study, different classes of membrane proteins with varying functions were also identified. A total of 69 proteins were identified as being outer membrane proteins of which 54 were identified with two or more peptide hits (Additional file 1). Using the database UniProtKB http://www.uniprot.org some of the functions of the outer membrane proteins were deduced. These included the transporters BtuB which is responsible for the uptake of vitamin B12, LamB which is involved in the uptake of maltose and maltodextrins and LolB which is involved in the incorporation of lipoproteins in the outer membrane. Other biologically significant proteins identified included the enzymes MltC which may play a role in cell elongation and division and NlpD which is involved in catabolic processes in cell wall formation as well as proteins involved in virulence such as Lpp1, Lpp2 and OmpX. To further verify the functions of the outer membrane proteins identified in the present study, manual mining of the data, which involved searching through literature containing information on the proteins of interest, was also undertaken. This approach shed further light on outer membrane proteins identified that were not apparent using UniProtKB, a shortcoming of using a single approach to verify the functions of proteins . These included membrane-bound lytic murein transglycosylase (MltB and MltC) which is important for cell growth , conjugal transfer surface exclusion protein (TraT) which is responsible for resistance to bacterial killing by serum  and RcsF protein which is part of the Rcs phosphorelay signalling pathway responding to peptidoglycan damage by regulating colanic acid capsular exopolysaccharide synthesis, and has also been seen to enhance bacterial survival in the presence of antibiotics .