Despite ongoing developments in analytical technology, the detection and identification of lipids remains a significant challenge.
Waters Corporation has successfully streamlined the data acquisition and analysis process via a workflow that combines the attributes of the Xevo™ MRT Mass Spectrometer and the data capabilities of the Lipostar2 software (Mass Analytica).
Lipidomics data were derived by analyzing plasma samples from colorectal cancer (CRC) patients and healthy control plasma to demonstrate the benefits of this approach.
Data was acquired in DIA mode, leveraging many of the key features and benefits of the waters_connect™ software platform combined with the Lipostar2 software. This data highlighted dysregulation of the lipid metabolism pathways based on cancer type.
The example presented here involved coupling the Xevo™ MRT MS (Figure 1) to an ACQUITY™ Premier LC. The mobile phase A contained 10 mM Ammonium Formate in Acetonitrile:Water and B contained Ammonium Formate in IPA:Acetonitrile. A CSH™ C18 (2.11 x 100 mm) column and a 12-minute (50-99% B) reverse phase gradient were also employed.
Figure 1. The Xevo MRT Mass Spectrometer. Image Credit: Waters Corporation
The waters_connect™ software platform was used to collect data before transferring this to Lipostar2 software via the UNIFI™ application programming interface (API) and as mzML files. The Lipostar2 software was then used to perform peak picking, alignment, identification, statistical analysis, and pathway analysis.
Data transfer from waters_connect™ platform to Lipostar2
The data generated using high-resolution instruments on waters_connect™ applications allows data to be directly imported into the most recent version of Lipostar2 via its API.
Many labs will likely have existing informatics and data pipelines in place, however, so users are afforded added flexibility thanks to the option to convert data to the mzML file format for use in popular third-party software such as MZmine#, MS-Dial*, XCMS™, and Skyline.
Data can be converted at the acquisition point via the Acquisition Method Editor (AME) or after it has been acquired via DATA Convert Applications. The key advantage of mzML is that it is a universal file format. Figure 2 describes this workflow.
Figure 2. The data transfer workflow of data from waters_connect to Mass Analytica software and other third-party solutions. Image Credit: Waters Corporation
Results and discussion
Figure 3 shows validated supervised models of controls versus CRC samples. These results are shown both with and without QC samples included, as well as with stratification based on CRC type, colon, and rectum. The QCs are highlighted in red, and they tightly cluster in the center of the model, showing good reproducibility over the course of the analysis.
Figure 3. PLS-DA models of Controls vs CRC patients, including NIST and study reference QCs (A) and with QCs removed from the model (B). PLS-DA models of Controls vs Colon vs Rectum patients, including NIST and study reference QCs (C), and with QCs removed from the model (D). Image Credit: Waters Corporation
It was possible to observe a clear separation of healthy controls and CRC cohorts when the QC samples (including NIST plasma) were removed. It was also possible to separate colon cancer plasma from rectum cancer and healthy controls if SR QC samples are excluded from the model generation.
It was noted that colon and rectum cancer samples clustered together if QC groups were included, however. This was because PLS-DA aims to find the maximum difference between groups.
The resulting multivariate statistical analysis (MVA) data was interrogated to identify the most significant lipid. This was done using the loading plot, S-Plot, and VIP scores (Figure 4).
Figure 4. Various tools are available in Lipostar to interrogate the most relevant features that drive clustering of sample cohort groups. Image Credit: Waters Corporation
Variables were color-coded according to class identifications, with the 50 most relevant variables shown via the separation observed in the PLS-DA model highlighted in the loading plot. These were displayed as a list of mass/charge–retention time pair features (m/z@tR).
Applying the associated identification information is also possible when approved assignments or compounds are selected. This will also appear in S plots with VIP score tables. In a VIP score plot, results greater than one are generally regarded as important to group separation.
Other advanced tools are available for visualizing and interpreting statistical analysis data, including the ANOVA T-Test table with p-values/fold changes. For example, a t-test can ascertain whether a statistically significant difference exists between the means of two groups.
Figure 3 illustrates how the Cer(36:1) adduct mean significantly differs between the control and CRC samples.
Accurately identifying putative lipids tends to be exceptionally challenging, relying on the extensive use of databases. Goracci et al. (2017) describe the Lipostar2 identification workflow, which sees Lipostar2 employ a rule-based approach to generate theoretical fragments for lipid structures.
For instance, it is possible to label a fragmentation rule as “mandatory.” This was the case in the choline head group fragment with m/z 184.074 for the protonated phosphocholines or m/z 264.270 for protonated ceramides.
It is also possible to cluster adducts of a particular compound based on retention time and m/z values. Identified lipids can be linked to associated pathways, a phenomenon that has been explored within peer-reviewed literature. For example, Figure 5 shows a Cer 42:1 assigned green due to its low mass accuracy error and high fragment score, as well as three associated adducts at the same retention time of 7.92 minutes.
Figure 5. Example Cer(42:1) identification at 7.92 min as in Lipostar2 (A) and waters_connect Platform (B). Image Credit: Waters Corporation
Figure 4B highlights how the Lipostar2 identification can be confirmed using the waters_connect™ platform. The extracted chromatogram at RT 7.92 minutes has been identified as Cer 42:1, while the long-chain base fragment has been observed with low mass error.
Figure 6 features box and whisker plots of ceramide and lysophoscholine lipids. The findings (acquired via MVA) are linked to CRC dysregulation, and various peer-reviewed publications agree.
It is important to note that some ceramide species are increased in CRC patients versus healthy controls, much like the pancreatic cancer pathway already discussed. However, no significant difference is observed between the two sample cohorts when the whole class is considered.
However, LPCs are downregulated in CRC samples compared to healthy controls. This occurs on a class basis. There is also potential for further investigation using larger sample cohorts and targeted lipidomic approaches.
Figure 6. Example box and whisker plots of ceramides species and LPC species from the CRC sample cohort. Image Credit: Waters Corporation
Lipids can also be quantified in Lipostar2 using the calibration curves of standards spiked in plasma, as the Xevo™ MRT MS boasts a linear dynamic range over five orders of magnitude.
An increasing number of researchers cite concentration values rather than relative responses when transferring data between labs, geographies, and the instruments used for collection.
Figure 7 features a ceramide standard calibration curve. This curve can be assigned to lipids of the sample class and adduct to generate concentration values.
Figure 7. Example Calibration curves of Ceramide standard spiked in human plasma that can be used to calculate concentrations of lipid species if the same class. Image Credit: Waters Corporation
Conclusions
The examples presented here highlighted a robust combination of reversed–phase UPLC™, Xevo™ MRT MS, and intelligent workflow-driven software. This approach was leveraged in the straightforward, reliable, rapid, and accurate lipidomic analysis of biological samples.
The Lipostar2 software facilitates the accurate identification and quantification of compounds within data sets acquired in either DDA or DIA mode. The accessible, easy-to-use workflow discussed enabled the expedited processing of the lipidomic data generated for the CRC study described.
Data generated from the pilot study were subjected to a comprehensive statistical analysis via Lipostar2 software to enable biological interpretation.
Acknowledgments
Produced from materials originally authored by Nyasha Munjoma, Lee A. Gethings, Richard Lock, and Jayne Kirk from Waters Corporation; from Paolo Tiberi from Mass Analytica Ltd.; and Laura Goracci from the Department of Chemistry, Biology and Biotechnology, University of Perugia.
About Waters Corporation
At Waters, we unlock the potential of science by solving problems that matter. Our software and instruments ensure the safety of the medicines we take, the purity of the food we eat and the water we drink, and the quality and durability of products we use every day. Together with our customers, in labs around the world, we deliver scientific insights to improve human health and well-being, helping to leave the world better than we found it.
Sponsored Content Policy: News-Medical.net publishes articles and related content that may be derived from sources where we have existing commercial relationships, provided such content adds value to the core editorial ethos of News-Medical.Net which is to educate and inform site visitors interested in medical research, science, medical devices and treatments.