Does In-Source Fragmentation Require a Soft Touch?

A point of contention in the ongoing debate over the role of in-source fragmentation in untargeted metabolomics is the extent to which the issue can be minimized with instrument tuning. Pieter Dorrestein and Yasin El Abiead believe it can be; Martin Giera and Gary Siuzdak argue that sensitivity issues rule this option out in the real world. But what if the problem lies earlier in the workflow – with the ionization source?

Here, we speak with Jan-Christoph Wolf, CTO of Plasmion GmbH, who believes that softer ionization sources such as SICRIT can allow for a wider range of options for chromatographic coupling, thereby directly addressing the problem of ISF through chromatographic separation. “If we rely exclusively on (n)ESI for metabolomics, we have a dark metabolome by design,” he says.

The debate over the dark metabolome hinges on whether many signals in LC-MS data are real metabolites or analytical artifacts. How do you interpret this controversy? 

It is a necessary scientific debate. As scientists, the number one rule is always to question your results from every possible angle.  However, due to the complexity and amount of spectra and data generated from biological samples, this has become almost impossible on a single component level.

In my view, the current discussion around in-source fragmentation (ISF) is somewhat limited. We should begin earlier in the analytical workflow. One key – and rarely addressed – aspect here is whether ESI or nano-ESI ionizes all molecules of interest. The answer is (clearly) no. This implies that a part of the dark metabolome may consist of compounds that remain undetected simply because they are not ionized efficiently. In addition, the ionization process and the MS inlet can introduce further complexity – through the formation of oxidation products caused by the heat applied during electrospray or within the inlet, for example. Furthermore, different ion forms such as M+NH4⁺, M+Na⁺ or M+K⁺ can originate from a single compound. These variants differ in their stability and may produce different fragmentation patterns. These factors are often overlooked. The result is that both the extent of the dark metabolome and the number of false positives resulting from artifacts and fragments may be significantly larger than is currently assumed.

In your view, how much of the “dark metabolome” could be attributed to ISF versus true biological complexity?

It would be easy to suggest arbitrary percentages, but in my view, such speculation does not meaningfully advance the discussion. What is clear is that both in-source artifacts and true biological signals contribute to the dark metabolome. Rather than focusing on assigning proportions, the more productive path is to ask how we can systematically address this challenge. What is needed are rigorous, data-driven methodological studies that can disentangle analytical artifacts from genuine metabolic features by means that are reproducible.

Some argue that instrumentation tuning already minimizes ISF in real-world workflows. Do you believe current practices go far enough?

There are limits to what can be achieved through instrumental tuning, and in my view, the primary cause of ISF is not the instrument itself, but the underlying ion chemistry. A simple example is cholesterol, which has a protonated mass of 387.36. If you search any LC-MS database, you will consistently find a fragment at 369.35, corresponding to the loss of a water molecule, regardless of the instrument settings. In other words, ISF can be minimized, but it cannot be completely avoided. However, this is not necessarily a problem. From an analytical perspective, we have additional tools at our disposal: chromatography, for example, can provide an extra dimension to help resolve these cases.

How can soft ionization methods like SICRIT address in-source fragmentation – how do they alter the picture we get from typical metabolomics experiments?

As mentioned above, ESI does not ionize all components of interest. For very non-polar compounds (e.g. some lipids) in particular, ESI struggles. Here, alternative ionization methods like APCI and SICRIT can be used. Both expand the range of detectable analytes; however, SICRIT has some clear advantages over APCI, including reduced formation of ISF and a broader ionization range. It can even ionize perfluoro-alkanes. But this comes at a cost. Due to its plasma chemistry, we must in some cases expect and consider other forms of ionization, such as M+ or an M-H+ species, simply because not every molecule can accept a proton.

The true strength of SICRIT, however, is that it enables the coupling of any chromatography (GC, LC, SFC, CE) or even metabolomic imaging experiments with a single ionization source. This opens up new metabolomic workflows, for example on GC or SFC, that directly address the problem of ISF through chromatographic separation.

Have you evaluated SICRIT’s impact on ISF?

We constantly evaluate the softness of SICRIT in both LC and GC experiments. It proves to be comparable to ESI, but I would like to highlight some independent research here.

A recent review from Emilie Bertrand and Valérie Gabelica investigated the ionization energy of SICRIT and estimated it to be roughly 130 percent of ESI, making it the softest commercially available plasma ionization technology. Since it is a fundamentally different technique to ESI, ISF may also depend on different factors. SICRIT is not spray-based, but rather a gas phase ionization technique. It is highly complementary to ESI as demonstrated in a recent metabolomic networking study. In this study, Allyson McAtammey and her coworkers applied ESI and SICRIT to a standard metabolomic workflow for biofilm forming microbes. They detected a comparable number of features (~1,800) for both techniques, but with an overlap of only 1,205 features. Through library matching, they identified different compound classes that are either not accessible or only marginally detectable by ESI.

These and other studies support my earlier point: if we rely exclusively on (n)ESI for metabolomics, we have a dark metabolome by design!

Have you experienced or do you anticipate any barriers to more widespread adoption of soft ionization methods? 

Indeed, the adoption of such new technology faces several barriers. A major challenge is the need to reconsider and potentially disrupt workflows that have been established for decades and are still widely taught in academic settings. One clear example is the traditional separation between the GC-MS and LC-MS communities. Using GC instead of LC in combination with soft ionization on an (LC-)MS platform can offer superior performance in terms of separation power and sensitivity. However, many practitioners tend to stick with familiar methods and are hesitant to explore alternatives.

I hope the ongoing discussion around the dark metabolome will help open both doors and minds to new technologies and workflows that could bring previously undetectable compounds into view.

What is your view on the current state of untargeted metabolomics, given the ongoing debate over the dark metabolome, and are you optimistic about the future of the field?

Currently, we are confronted with what we don’t see, or what we believe we see. I think this has always been the ideal starting point for science: curiosity.

Newsletters

Receive the latest analytical science news, personalities, education, and career development – weekly to your inbox.

I have read and understand the Privacy Notice*

Subscribe