A journey through mass spectrometry imaging with Shane Ellis, PhD

Tell us a little bit about yourself.

I received my PhD from the University of Wollongong in ambient mass spectrometry and lipidomics in 2013. My PhD supervisors Prof Stephen Blanksby and A/Prof. Todd Mitchell are the primary reason I was and still am working in lipidomics. In 2012, I moved to FOM-AMOLF in Amsterdam to take up a postdoc position in the group of Prof. Ron Heeren. Here I worked on the development of active pixel detectors for mass spectrometry applications, including for high mass detection and stigmatic imaging using MALDI. In 2014, together with Prof. Heeren and several other colleagues, we moved to Maastricht to establish the Maastricht Multimodal Molecular Imaging Institute. Here I held an Assistant Professor position (receiving tenure in 2018) and led a group focusing on both instrument development for mass spectrometry imaging (MSI) and developing innovative applications for lipid imaging. In 2020, after the birth of our first child, we moved back to Wollongong to take up the position at UOW.

What drew you to spatial lipidomics as opposed to "traditional" lipidomics?

I always found MSI very interesting and was lucky enough to undertake an undergraduate research project on setting up a DESI source. Here I took my very first MSI image, which was very old school as we only had a 1D stage, so I had to manually position the stage at the end of each line between scans. I was then hooked, and my PhD incorporated both imaging and lipidomics, and I have been doing it ever since. MSI was still quite new at the time, and I enjoyed the instrumentation side of things, so it seemed the ideal field to continue in. Imaging is also great in that you generate very interesting images that even people not familiar with the method can readily appreciate. Since then my interest in understanding lipid biology to help interpret MSI data has grown significantly, so in a ways MSI has also focused my interest on lipid biology.

Much of your work relies on matrix assisted laser desorption ionization (MALDI). What advantages do see with MALDI over desorption electrospray ionization (DESI)?

I believe both MALDI and DESI, along with many other MSI methods, each have their own place and advantages. My group focuses on MALDI as I believe it is still the most versatile and sensitive method for lipid imaging. The rapid development in matrix application technologies means this is not so much a disadvantage anymore, at least in my opinion. A tissue section can be coated in matrix in 10-15 minutes, which on the timescale of the other steps in the MSI workflow is not overly significant. MALDI also offers much higher spatial resolution than DESI resolution - nowadays as low as 1 micron – and is comparable to some cluster SIMS methods but with MALDI yielding much less fragmentation. Furthermore, with the development of post-ionization methods like MALDI-2 that so far are mostly coupled with MALDI, the breadth of lipid classes that can be detected using MALDI is greater than conventional techniques like DESI.

Mass spectrometry imaging can generate huge amounts of data. How do you handle this much information and make sense of it?

The huge amounts of data we generate is a challenge we all face. A typical MSI can experiment on one tissue section can easily be several 10s of GB. Commercial software solutions such as SCiLS and LipostarMSI have made handling this data quite accessible, so for many projects, we use these. We are lucky enough to have good contacts with developers to help build in some features that we require. However, some cases require custom software to be developed, for example, integrating data from different omics techniques such as spatial transcriptomics or multiplexed quantitative imaging. In these cases we either try and develop approaches ourselves or are lucky to have several exciting collaborations, for example, with Aspect Analytics, to help develop data analysis strategies. We are also working with groups in the field of machine learning to develop new ways to extract out the meaningful patterns in MSI data to help guide later analysis. Multidisciplinary collaborations are also key to ensuring we can properly understand our MSI data, for example, with pathologists and biologists.

Data interpretation in terms of understanding the biological origins and significance of lipid MSI data is another challenge and an area that as a field we have a lot to learn. We can only now scratch the surface of the biochemical information and meaning behind the lipid distributions we observed with MSI. One challenge here is we can often identify lipids to a higher degree of specificity then they are specified in some pathway databases, so how do we understand our data in terms of pathways? With improved informatics tools, lipid identification methods and pathway understanding it is likely we will learn a lot about localized lipid metabolism in tissues. One exciting area we and others are exploring is whether lipid distributions, particularly when we can resolve individual isomers, can be used as a readout of localized enzyme activity.

In your research, you have looked at a variety of tissues including brain, liver, and kidney. Are there any organs or tissues you have not worked with yet that you are particularly excited about analyzing via mass spec imaging?

We are increasingly applying our MSI approaches to various cancer tissues to understand how drugs can alter lipid metabolism within cancers and to study the lipidomics signatures of metastasis. We, like many other groups, are also applying MSI to perform lipidomics on single cells. This is a new area for us and one I am very excited about. We expect here the post-ionization techniques we are working on will help expand the types of lipids we can detect from single cells. One area we are applying this is in the study of the lipid phenotypes of neurons derived from the stem cells of Parkinson’s patients and how the lipid profiles and heterogeneity amongst cell populations is influenced by different genetic mutations associated with Parkinson’s. As a logical next step, we are also very keen to see how MSI can be used to study lipid distributions at the sub-cellular level.

What is your vision for mass spectrometry imaging in the future of medicine? Do you see it being used routinely in clinics, or do you think it will be a specialized technique only for use under certain circumstances?

I think there is a good chance in the future lipid-based MSI will play a significate role in medicine. One only needs to look at the success of intraoperative MS such as the I-knife, Mass Spec Pen and Spidermass to see the potential for MSI to guide surgery in real-time. All these methods rely on the altered lipid profiles of different tissue types. Many groups have shown MSI to be very powerful to disease classification and even for helping guide treatment decisions based on the molecular signatures of an individual cancer. As the robustness and ease of use of MSI instrumentation continues to evolve the likelihood of them being used, for example in pathology, is high. Of course, a challenge is generating the right databases to guide the MS data interpretation and ensuring these databases are transferrable between labs. These methods can possibly find success in areas where current diagnostic approaches are limited, especially as the drive towards personalized medicine increases and the unique molecular phenotype of one’s biopsy can be best exploited.

Lipids (at least some classes) are usually the easiest thing to detect from tissues, and lipid metabolism is altered in many types of diseases so they present the ideal analyte class for these applications.

Are there any particular lipid classes which are more difficult to work with in MSI than others?

Conventional MSI can detect many phospholipid and sphingolipid classes. However, we know many lipids are often not detected, either due to poor ionization efficiency of low abundance (the same issue also largely holds in shotgun lipidomics). Both post-ionization and derivatization methods are rapidly evolving and have already significantly increased the types of lipids that can be studied with MSI, for example, many glycosphingolipids and sterols benefit substantially from post-ionization. One example we are applying this is mapping the distribution of glucosylsphingosine (GluSph) in brain tissue, especially in brains carrying a GBA1 mutation in the context of Parkinson’s disease. GluSph is present at very low concentrations and so far, we have only been able to detect it using post-ionization methods. This highlights the close relationship between technology developments and our ability to learn more about lipids in complex biological systems.

What do you consider to be the greatest breakthrough in lipid research in recent years?

Lipid MSI has evolved significantly in recent years. I see a few breakthroughs that will likely have significant impact.

• With advances in sensitivity and resolution, MSI is on the verge of being able to study lipid distributions at the sub-cellular level which opens up many new and exciting biological questions that we can think about addressing.
• Isomer-resolved lipid imaging using methods such as OzID, Paterno-Buchi reactions and UVPD are rapidly growing in popularity. These will have a tremendous impact for precisely mapping lipids throughout biological systems. Many lipids are detected as a population of isomers, and using these techniques, it is becoming possible to image structurally-defined lipid species in the absence of isomeric interference. Not only is this important for understanding the spatial distributions of defined lipids, but with isomer-level structural fidelity, one can begin to rationalize lipid distributions in terms of enzymatic processes as desaturation, elongation and acyl chain remodeling.
• The combination of lipid MSI with isotope labelling is very exciting and is beginning to be utilized by more groups. This enables mapping of the dynamics of lipid synthesis and turnover throughout heterogeneous tissues providing added information to the processes and kinetics of lipid metabolism.
• More generally, the availability of many standards, including isomer pure standards by companies such as Avanti has had a huge role in the development and application of quantitative lipidomics workflows. Without these standards, many studies would not be possible and the field likely would not have seen the growth it has. We are especially excited to utile the MSI Splash mix for quantitative lipid MSI of many different tissues.

What is the best piece of advice you have ever received from anyone?

Science should be fun and to never expect new experiments to work first try!