Enrichment Strategies and Global Analysis of Sulfation
Post-translationally modified peptides represent a small percentage of the digested proteomes, even when the level of modification within specific proteins is relatively high. Thus, for systems biology studies, peptides bearing a particular PTM have to be selectively isolated or at least enriched. In addition, diagnostic mass spectrometric features that enable selective detection/more reliable assignment of the modification of interest are extremely valuable even in the analysis of mixtures rich in specifically modified sequences.
To date, no large-scale, high-throughput sulfation studies have been performed. This can be explained by the mass spectrometry behavior of sulfopep- tides as presented earlier. In addition, no reliable and sufficiently selective enrichment method has been established for sulfated sequences yet.
Large-scale PTM identification studies are frequently aided by specific antibodies. The generation of sulfo-Tyr recognizing antibodies has been reported [56, 57], and the latter group also used their antibody for the enrichment of modified proteins . A sulfo-Tyr-specific antibody is also marketed by multiple vendors (Abcam, Millipore). However, no large-scale sulfotyrosine studies using any of these antibodies have been reported. Thus, we cannot draw any conclusions about their specificity and binding affinity.
Phosphopeptides are often enriched based on their strong affinity to a variety of metal ions. Immobilized metal ion affinity chromatography (IMAC) with Ga(III) was reported to enrich sulfopeptides from the skin secretion of a frog, however, with very modest selectivity . The utility of a weak anion exchanger at a close to neutral pH has been demonstrated on model sulfopeptides and bovine fibrinogen . Following trypsin digestion, C-terminal basic residues were removed by carboxypeptidase B to aid the isolation of sulfated peptides. While this treatment enables the enrichment of the peptides (unless there are additional basic residues within the sequence), it is also counterproductive for the following mass spectrometric analysis. Other chromatographic fractionation methods, such as electrostatic repulsion hydrophilic interaction chromatography (ERLIC) [60, 61] and strong cation exchange (SCX) separations , were relatively successful for the enrichment of phosphopeptides. These techniques also take advantage of the acidity of the modifying group, but at a low enough pH where all basic residues are protonated, and only the strong acidic group bears a negative charge. These techniques may also be tested for the enrichment of the more acidic sulfopeptides; however, the elimination of the interfering phospho- peptides by alkaline phosphatase treatment would be recommended.
Even with a certain degree of selective enrichment, finding post-translation- ally modified peptides in such a complex mixture as, for example, a plasma protein digest is not a simple task to tackle. The similarity of phospho- and sulfopeptides further complicates the matter. Theoretically, sulfopeptides could be identified by accurate mass measurements. However, one usually has to rely only on the mass accuracy of the precursor ion since none of the CID- generated fragment ions retain the modification. Thus, the mass error must not be higher than 2 ppm in order to differentiate between phospho- and sul- fopeptides up to 2500 Da.
The characteristic 80 Da loss may be used to our advantage. For example, one can use neutral loss analysis to identify the sulfopeptides [38, 63], or alternatively the diagnostic fragment ion, SO3- (m/z 80), may be used either in precursor ion scans in negative mode or in a “ping-pong style” acquisition setup, where in-source fragmentation is monitored in negative mode, while the full scan mass measurements and data-dependent MS/MS acquisitions are performed in positive mode . Obviously, the latter approach will indicate which MS surveys contain sulfopeptides, but will not directly identify them.
Conventionally, large-scale proteomic analyses are performed in positive ion mode. The generally used activation methods are ion trap CID (resonance activation), beam-type CID (also sometimes referred to as higher-energy C-trap dissociation (HCD)), and ETD. During database searches, phosphorylation is frequently considered as variable modification, while sulfation is usually not. However, even if it were included, unless very good mass accuracy is specified for the precursor ion (see preceding text), both modifications may fit within the permitted mass error. Based on our experience, most search engines identify sulfopeptides as phosphorylated sequences and even assign the residue modified. As far as we know, only ProteinProspector permits the computer “to admit” that there is no sufficient information to distinguish between positional isomers and displays this automatically in the search results . ProteinProspector also permits searching for fragile modifications as neutral losses; that is, there is no obligatory site localization .
The difficulty of sulfopeptide identification is beautifully illustrated by a study initiated by the ABRF Proteome Informatics Research Group . Seventy synthetic, PTM-bearing peptides, among them seven sulfopep- tides, were mixed with a yeast proteome tryptic digest. High-quality beam- type CID data were acquired, and identical peak lists and a database were sent out to the study participants. As it turned out, sulfopeptides represented the biggest challenge - “Lack of knowledge of how this modification behaves in CID (either by software or by user) led to many people either completely missing these peptides or reporting them as phosphopeptides instead” .