Discussion
Within this work, we demonstrate the utility of FAIMS based fractionation and DIA analysis for probing bacterial glycosylation systems revealing glycosylation changes observed in response to chimericpglO alleles in N. gonorrhoeae occur largely independent of changes in glycoprotein levels. Benchmarking the use of FAIMS fractionation for N. gonorrhoeae we show the glycoproteome ofN. gonorrhoeae can effectively be studied without the need for bespoke affinity reagents and that novel glycopeptides can be readily identified using this approach. Importantly this work shows that despite the absence of protein abundance changes in glycoproteins alterations within the broader proteome of N. gonorrhoeae are readily identifiable in response to changes in glycosylation patterns.This work builds on our previous exploration of PglO targeting specificities [52] and supports that PglO protein substrate recognition is discrete yet reprogrammable by the manipulation of PglO. While early studies have noted differences in the compatibility of O- oligosaccharyltransferases to target substrates [74], our observations here expand on these findings demonstrating this also occurs outside of heterologous expression systems. Our work here supports that differences in PglO targeting ranges may be more common than previously thought which has important ramifications for predicting the targets of bacterial glycosylation across genera as well as unlocking the potential of these enzymes for glycoengineering [75].
While the use of FAIMS fractionation has been noted to be advantageous for glycoproteomics typically improving glycoprotein/glycopeptide coverage by as much as 75% [35, 38], within this study we observed a >200% increase in the number of identified glycoproteins in N. gonorrhoeae compared to earlier studies [26, 48]. While it is likely that this increase is due to multiple factors including the use of current generation MS instrumentation, as well as the glycosylation of proteins not natively glycosylated within N. gonorrhoeae by chimericpglO alleles, it is noteworthy that many novel glycoproteins identified here correspond to integral membrane proteins containing multiple transmembrane domains (Supplementary table 4). Previous studies exploring the N. gonorrhoeae glycoproteome have traditionally used protein centric approaches including protein level antibody-based enrichment [26] or 2D gel electrophoresis coupled with antibody-based detection [48] which while effective are known to be incompatible with poorly soluble hydrophobic membrane proteins [76, 77]. The use of peptide-centric analysis, as undertaken here, is known to enhance both the detection as well as the characterization of modifications within bacterial membrane glycoproteins [24]. While most previously known N. gonorrhoeae glycoproteins were identified using FAIMS fractionation, glycopeptides derived from four previously identified glycoproteins failed to be detected with two of these proteins (NGO_1237 and NGO_0994) detected by DIA analysis and the remaining two glycoproteins either not detected (NGO_0983) or absent due to being disrupted within these backgrounds (PilE, NGO_1177). Importantly, while NGO_1237 and NGO_0994 are both known to be glycosylated the precise location of glycosylation within NGO_0994 is yet to be confirmed being predicted to be located in a region lacking Lysines / Arginines [52] and glycosylation within NGO_1237 requiring alternative enzymes to Trypsin to be accessible [48, 78]. Combined, these results support that for the Trypsin accessible glycoproteome, FAIMS based analysis provides broad coverage of the known glycoproteome and provides access to novel membrane glycoproteins.
The identification that changes in glycosylation patterns impact theN. gonorrhoeae proteome contributes to a growing body of work linking proteome alterations with changes in glycosylation occupancy across a range of Gram-negative species [79-81]. Interestingly while previous studies have explored the impact of the abolishment of glycosylation few studies have assessed how changes in glycosylation occupancy or glycosylation patterns impact bacterial proteomes. Recently it was demonstrated that within Burkholderia species the silencing of glycosylation using CRISPRi reduced occupancy but resulted in only modest proteomic impacts [82] yet in contrast our findings here support the N. gonorrhoeae proteome is extensively impacted by changes in glycosylation patterns. WhileO- linked glycosylation within B. cenocepacia has been shown to be important for the stability of multiple proteins [81, 83] we observed no change in the protein abundance of glycoproteins within this work suggesting that within N. gonorrhoeae O- linked glycosylation may exert its effects by modulating protein interactions or protein activities. Consistent with this previous work has shown for N. gonorrhoeaePilE glycosylation is not required for Pilin formation yet appears to enhance the stability of this structure [84] and for other forms of bacterial glycosylation, such as N-linked glycosylation within C. jejuni,glycosylation has also been shown to stabilise protein complexes with this effect occurring independent of changes in protein abundance [85]. Thus, while we find the abundance of glycoproteins are unaffected future work may seek to confirm if the glycosylation events identified here also play roles in modulating protein functions and/or the stabilization of protein complexes driving the proteomic effects observed here.
Within this study, our implementation of DIA analysis allowed access to proteome as well as glycosylation occupancy information yet, it is important to note that many glycopeptides observed by FAIMS fractionation are not observable within our DIA dataset due to the imperfect overlap in the peptides observed between these two approaches. As previously noted, FAIMS fractionation preferentially enables access to large (>1000 m/z), low charge density glycopeptides [35] with a similar trend also observed for N. gonorrhoeae (Supplementary Figure 1B). Due to this bias, most N. gonorrhoeae glycopeptides fall outside the m/z range utilised in our DIA analysis (350 to 951 m/z) which has been tailored to maximize the proteome coverage [53]. While DIA methods compatible with glycopeptides are emerging [86, 87], these methods consciously utilize wider DIA m/z ranges which improve access to glycopeptides yet this can lead to more congested MS/MS spectra reducing DIA depth [53]. While enzymatic [88, 89] and genetic [90] approaches can be used to reduce glycan sizes and heterogeneity to enhance the detection of both glycosylated and non-glycosylated forms of peptides the size of the native N. gonorrhoeae trisaccharide (552Da) limits the feasibility of using identical analysis approaches. Despite this caveat we find the non-glycosylated forms of 50 peptides subjected to glycosylation could be observed using DIA supporting that while not optimized to allow coverage of both the glycoproteome and proteome information on glycosylation occupancy is still readily accessible. Interestingly within this analysis, we observed the changes in the non-glycosylated forms of peptides is generally modest, less than a <40% reduction in intensity with only a few examples of peptides unable to be identified between different strains (Figure 5 and Supplementary Figure 7 to 10). Thus, this relatively small impact on abundance suggests that the overall glycosylation occupancy may be modest for many N. gonorrhoeae glycoproteins observed here.
In summary, this work furthers our understanding of the utility of FAIMS based fractionation for bacterial glycoproteomics and expanse the known glycoproteome of N. gonorrhoeae . This work contributes to a growing body of studies that demonstrates that bacterial oligosaccharyltransferases possess discrete targeting ranges and that the manipulation of these enzymes can vary the observed glycosylation pattern within bacterial systems. Our identification that changes inO -linked glycosylation patterns leads to changes in the proteome further supports that glycosylation plays multiple roles in N. gonorrhoeae biology. Finally, this work demonstrates how glycoproteomic insights and peptide centric DIA information can be integrated to provide support for changes in occupancy across N. gonorrhoeaestrains expressing different pglO chimeras. These changes in glycosylation patterns largely appear independent of changes in glycoprotein levels and peptide analysis supports that glycosylation occupancy is generally low within most observed glycoproteins.