Open Database Searching Enables the Identification and Comparison of Bacterial Glycoproteomes without Defining Glycan Compositions Prior to Searching

Abstract

Mass spectrometry has become an indispensable tool for the characterisation of glycosylation across biological systems. Our ability to generate rich fragmentation of glycopeptides has dramatically improved over the last decade yet our informatic approaches still lag behind. While glycoproteomic informatics approaches using glycan databases have attracted considerable attention, database independent approaches have not. This has significantly limited high throughput studies of unusual or atypical glycosylation events such as those observed in bacteria. As such, computational approaches to examine bacterial glycosylation and identify chemically diverse glycans are desperately needed. Here we describe the use of wide-tolerance (up to 2000 Da) open searching as a means to rapidly examine bacterial glycoproteomes. We benchmarked this approach using N-linked glycopeptides of Campylobacter fetus subsp. fetus as well as O-linked glycopeptides of Acinetobacter baumannii and Burkholderia cenocepacia revealing glycopeptides modified with a range of glycans can be readily identified without defining the glycan masses prior to database searching. Utilising this approach, we demonstrate how wide tolerance searching can be used to compare glycan utilisation across bacterial species by examining the glycoproteomes of eight Burkholderia species (B. pseudomallei; B. multivorans; B. dolosa; B. humptydooensis; B. ubonensis, B. anthina; B. diffusa; B. pseudomultivorans). Finally, we demonstrate how open searching enables the identification of low frequency glycoforms based on shared modified peptides sequences. Combined, these results show that open searching is a robust computational approach for the determination of glycan diversity within bacterial proteomes.

What Are We Missing by Using Hydrophilic Enrichment? Improving Bacterial Glycoproteome Coverage Using Total Proteome and FAIMS Analyses

“…ubonensis [44]. Consistent with our observations within B. cenocepacia both stepped FAIMS and unenriched samples revealed multiple unique glycopeptides/glycoproteins absent within ZIC-HILIC enrichments (Figure 5A to B; Supplementary Tables 9-13).…”

“…Analysis of the physiochemical properties of glycopeptides observed using each approach highlights that the glycopeptides observed within B. cenocepacia with stepped FAIMS and unenriched samples are largely similar, favouring shorter and more aliphatic glycopeptides compared to the larger and less aliphatic glycopeptides observed with ZIC-HILIC enrichment ( Figure 4D, Supplementary Table 8). The modest overlap of unique glycopeptides observed with stepped FAIMS and analysis without FAIMS fractionation illustrates that the glycoproteome of B. cenocepacia is larger than initially thought [31,44], and that ZIC-HILIC enrichment fails to capture a significant proportion of the Burkholderia glycoproteome.…”

“…The previously published ZIC-HILIC glycopeptide datasets [44] were used to compare the differences in glycopeptides observed with and without ZIC-HILIC enrichment. It should be noted that the proteome digests within this work are the same digests used for ZIC-HILIC enrichments in [44]. Analysis and visualization was undertaken using R (https://www.r-project.org/) with the glycopeptide physicochemical properties determined using the Peptides package [70] and data visualized using ggplot2 [71] and ggpubr packages.…”

“…To date, studies of bacterial glycoproteomes have extensively used ZIC-HILIC enrichment [29][30][31][32][40][41][42][43] to characterise glycosylation substrates and glycan diversity. Using ZIC-HILIC enrichment, we have previously shown over 100 proteins are glycosylated within Burkholderia species [44]. Despite this insight, the complete repertoire of most Burkholderia glycoproteomes are still unknown, hampering our understanding of the impacts observed when glycosylation is disrupted [31,45,46].…”

“…The short length of these two glycans raises notable concerns about potential bias in the observable glycoproteome within ZIC-HILIC enrichments [23,34]. This concern, coupled with our recent identification of extensive formylation artefacts as a result of ZIC-HILIC enrichment [44], suggests alternative approaches -such as orthogonal fractionation -are required for the analysis of bacterial glycopeptides.…”