Mass spectrometry method development to investigate the ubiquitin-like protein modifier ISG15

This data collection results in more complete and increasingly more information rich datasets, requiring advanced software tools to analyse. Typically, this data collection is carried out through the generation of a spectral library that the experimental data is queried against. While many different forms of DIA exist, diaPASEF is a recently developed DIA scheme utilizing trapped ion mobility spectrometry (TIMS) to improve selectivity and sensitivity at acquisition.<br><br>First, we expanded on a gas phase fractionation method described by Searle and colleagues to improve the quality of spectral library generation without using offline fractionation methods.

We implemented a method that used coordination of narrow DIA windows with the trapped ion mobility dimension of diaPASEF to generate gas phase fractionation datasets (IM-GPF) to produce deeper spectral libraries, and then compared them with an existing offline fractionation dataset and generation of libraries by direct analysis of diaPASEF data [1]. We found that the IM-GPF library surpassed the direct analysis and approached the larger offline fractionation dataset in terms of raw identifications.

In terms of reliability and reproducibility, we found that quantitative data derived using the IM-GPF spectral library was more reproducible than the offline fractionation dataset.<br><br>Next, we systematically evaluated the quantitative performance of diaPASEF using 2 distinct approaches to response curve analysis. We studied the quantitative response behaviour of diaPASEF using 2 novel timsTOF platforms with design goals for high sensitivity or high dynamic range.

We used a matched matrix dilution series design for the generation of a quantitative benchmarking standard. This involved the creation of a Yeast: HeLa matched matrix in which known concentrations of both were mixed together to create a logarithmic response curve. These benchmarking samples were measured on the high dynamic range platform and modelled using a bilinear curve model previously applied to other DIA datasets.

In the second approach we used a set of 30 stable isotope labelled standards diluted into a HEK293 background to determine quantitative characteristics of the diaPASEF method on both platforms. In both experimental workflows we were able to identify figures of merit using bilinear fit models as previously described and determining limit of detection values and subsequently limit of quantification values for an array of peptides on each instrument using diaPASEF.<br><br>Having established improved methods for spectral library generation and quantitative benchmarking we next turned to methods for the analysis of the ubiquitin like modifier ISG15.

We investigated two separate experimental workflows. First, an affinity enrichment process was implemented using A549 cells expressing His-tagged ISG15 capable of conjugation or a separate cell line expressing His-ISG15 (ISG15.ΔGG) in which the glycine residues at the C-terminus had been mutated such that conjugation was impossible. We anticipated that the acquired data following enrichment would enable a comparison to suggest putative ISGylated proteins.

However, this approach was ultimately not successful as we could not reliably distinguish between potential ISGylated peptides due to a high background and reasoned that another approach should be taken. Following this, a K-ε-GG immunoprecipitation workflow was developed using a different set of A549 cells. An ISG15 knockout, USP18 (the de-ubiquitinase responsible for ISG15 deconjugation) knockout and inactivated USP18 mutant alongside wild-type A549 cells were stimulated with interferon-β and enriched using a PTM Scan Ubiquitin (K-ε-GG antibody) detection kit.

The samples were then acquired using an optimised diaPASEF method. Resulting data was analysed and yielded 222 potentially putative ISGylated proteins. While we believe it is challenging to conclude that they are bona fide ISGylation sites, we carried out a short investigation into the supporting literature and conclude that with future adjustments to improve the reproducibility of the workflow, a suitable model could be generated for infection experiments and investigation into the role of ISG15.<br>