Approaches to sample preparation depend, unsurprisingly, on the sample and the eventual analysis mode. In the case of tandem liquid chromatography (LC)-mass spectrometry (MS) methods, sample prep based on concentration, capture, filtration, adsorption/affinity, and flow-through enrich samples in target analytes while depleting species that potentially interfere with LC and MS.
Metabolomics studies present some of the most daunting challenges in sample preparation, particularly when the requirements of two methods (LC and MS) must be met. When a target or its molecular class is known or belongs to one general category such as proteins/peptides or genes/oligonucleotides, one can at least make educated guesses as to location, prevalence, and concentrations. That is not always true with metabolites, and never in de novo or untargeted metabolomics.
Almost any product or intermediate of metabolism falling under a molecular weight of about 1500 Da qualifies as a metabolite, including carbohydrates, lipids, terpenes, amino acids, vitamins, gene or protein fragments—anything. Among the 231 most common metabolites, the concentration dynamic range is approximately one million but that only tells part of this particular story. For all metabolites the gap widens to 1014, made even more challenging when typical metabolomic experiments might examine half a dozen targets.
Add to these concerns that the metabolome is constantly in flux, with concentrations of various species rising and falling according to the organism’s metabolic activity, and it becomes obvious why metabolomics is by far the hardest of the ‘omics in terms of drawing actionable conclusions.
The answer: Solid-phase extraction
Liquid chromatography-mass spectrometry (LC-MS) and nuclear magnetic resonance spectroscopy (NMR) are the two leading metabolomics methods. NMR is fast and cheap on a per-sample basis, and sample prep is straightforward or nonexistent; LC-MS provides broader and deeper metabolite coverage, higher sensitivity, and greater instrument affordability and accessibility, but at the cost of significantly more-involved sample preparation.
Solid-phase extraction (SPE) has emerged as a rapid, reproducible, versatile alternative to more traditional sample preparation methods regardless of endpoint analytical method, but especially for LC-MS workflows. SPE uses standard chromatography resins in small cartridges to concentrate targets, clear contaminants or interfering species, and fractionate targets by molecular class. Think of SPE as “micro-chromatography.”
Any resin and solvent system suited to conventional column chromatography works with SPE. Dozens of unique resins are commercially available in SPE format, with hundreds more possible through home-brewing. Researchers use them for every conceivable separation, including purification/cleanup through affinity, ion exchange, HILIC, reverse-phase, and mixed modes.
Standardizing workflows
In untargeted metabolomics, samples are first cooled to -80°C using cold methanol to quench metabolic activity and freeze the sample in its metabolic state. “Trying to extract all metabolites using any simple extraction method is inherently difficult, but made even harder due to the very broad physio-chemical characteristics of metabolites,” says Steve Fischer, Market Director for Academia and Government/Life Science Research Segment, at Agilent Technologies. “Researchers typically use a single-phase liquid extraction to remove the soluble metabolite fraction from the cellular debris. However, when using this approach, metabolite recovery usually shows poor reproducibility due to ion suppression caused by the complex matrix.”
For polar metabolites, HILIC liquid chromatography with electrospray ionization, mass spectrometry has become the method of choice. “LC-MS analysis is done in both positive and negative ion mode,” Fischer adds. “Typically, two separate runs are performed on the same sample since the optimal mobile-phase pH for separation and MS ionization for basic and acidic metabolites is very different.”
Lipidomics (the lipid fraction of the metabolome) is best tackled with reverse-phase LC, using two or more organic solvents in the mobile phase. The samples are then analyzed by both positive and negative ion mode LC-MS. Unlike polar metabolites, the mobile-phase pH is typically not a big factor in lipid separation or positive and negative mode ionization. There is, however, a strong lipid class response difference in positive and negative ion polarities thus requiring the use of both ion modes.
Because of these challenges, researchers have moved to selective metabolite recovery, which uses liquid-liquid extraction or SPE methods to separate metabolites into polar and non-polar fractions. For liquid-liquid extraction, this involves the use of two immiscible solvents in which the target polar metabolites ideally stay in the aqueous phase, and the non-polar metabolites are in the organic phase.
Alternatively, solid-phase extraction will separate polar and non-polar metabolites by retaining the non-polar components while allowing the polar metabolites to flow through. “The advantage of SPE over liquid-liquid extraction is better reproducibility, more tolerance of sample matrix variation, less labor and time to complete, and easier automation for high-throughput needs,” Fischer tells Biocompare. Highly selective SPE products such as Agilent Bond Elut Lipid Extraction have been developed that allow for the selective retention of lipids (an import metabolite subclass) from a sample.
In addition to simplifying analysis, sample preparation can reduce ion suppression in the LC-MS ion source. Some LC-MS ionization techniques are less prone to ion suppression than others, but none of them can be completely suppression-free. Ion suppression is a phenomenon that is caused by the presence of abundant compounds in the sample matrix that co-elute with analytes. The bottom line here is that ion suppression reduces sensitivity, which is a big deal when hunting down low-abundance metabolites.
“No LC-MS ion source exists that ionizes absolutely everything and is ion suppression-free,” Fischer says. “Plus, sending complex un-prepared samples into an LC-MS will foul the ion source. Researchers have tried flow-injection LC-MS analysis, but from samples that are already reasonably clean—low-matrix samples such as cell culture. But even there, ion suppression remains a problem. For more complex samples a researcher needs to either perform sample preparation or use 2D-LC chromatography to improve the separation of targets from unwanted matrix interferences, and ultimately obtain reproducible, accurate results.“
Arguing for automation
Metabolomic sample prep for LC-MS involves many uncertainties, which together affect reproducibility and general data trustworthiness. A case can therefore easily be made for automating SPE, particularly early during a research project, to eliminate variables associated with manual manipulations of samples and assay assets.
As Anis Fahandej-Sadi, a product specialist at Aurora Biomed, explains, the benefits of automating SPE are tangible at almost any throughput. “When I was studying at the University of Alberta, automation was routinely used for sample handling, and most of the mass spectrometers, chromatographs, and NMR instruments used autosamplers. Automation doesn’t just allow more walk-away time, it frees workers to be more scientifically productive. You don’t want a Ph.D. scientist doing things that a technician is capable of.”
For example, preparing libraries for next-generation sequencing typically involves multiple transfers of one to two microliters of reagents and/or sample for each well in a multiplexed experiment. “At those volumes, small mistakes have big consequences,” Anis says. “Even if you’re only running ten or twenty samples per day, automation will pay for itself.” Freeing lab workers to solve problems that are more complex and demanding of brainpower, he adds, “may cost money but in the long term will provide your research group or company with more value-added opportunities—even in academic groups.”