NEW YORK(GenomeWeb) – Researchers at Vanderbilt University have devised a method for combining information from imaging mass spectrometry and optical microscopy experiments.
Detailed in a study published this week in Nature Methods, the approach brings together the spatial resolution of optical microscope and the molecular information generated by imaging mass spectrometry, allowing researchers to examine the same piece of tissue with high molecular specificity and high spatial resolution, Richard Caprioli, a Vanderbilt researcher and senior author on the paper, told GenomeWeb.
He noted that the approach could prove particularly useful in areas like anatomical pathology, especially as clinicians begin to use multiple molecular markers and protein signatures in the assessment of biopsies.
Additionally, the technique is applicable to other imaging methods such as MRI, CT, and PET. Currently, the Vanderbilt researchers are using it to combine imaging mass spec data with MRIs, Caprioli said.
The method is based on a large-scale regression analyses linking imaging mass spec measurements to optical microscopy measurements. Using partial least-squares regression, the researchers related the mass spec ion intensity measurements to various microscopy-based measurements to create a model that enabled them to take microscopy data and produce a prediction of the corresponding IMS data.
In this way, Caprioli said, he and his colleagues were able to generate predictive images with mass spec's rich molecular information and optical microscopy's high spatial resolution.
"It's a case where the sum of the two methods is much, much better than either one of them alone," he said.
Optical microscopes can offer spatial resolution in the range of low hundreds of nanometers, while current imaging mass spec platforms top out at around 1,000 nanometers, or 1 micron, Caprioli said.
The resolution of imaging mass spec instruments continues to improve, he said, noting that it could in the future draw closer to that of optical microscopy. But, he added, microscopy will likely retain its advantage in resolving power.
Microscopy, on the other hand, is extremely limited in terms of the molecular information it can offer. Typically, proteins are detected in tissue samples using antibody staining, which allows for little multiplexing.
Additionally, Caprioli noted, the technique is poorly suited to discovery research "because you have to know in advance the molecule you're looking for."
However, "the mass spectrometer doesn't require any of this because it doesn't use a surrogate marker for the antigen, it looks at it directly," he said. "So you don't have to know what you're looking for in advance to find something interesting."
The authors also noted in the Nature Methods paper that combining the two modalities can improve discovery by helping researchers better distinguish between signal and noise. Features that might go undetected or be passed off as noise using one method can take on potential significance when they are corroborated using the second method, they wrote.
For instance, using the combined approach to study a rat kidney section, the researchers detected tissue features that, in the imaging mass spec data, appeared to be simply matrix noise, and in the microscopy data were only faintly detectable. However, these features were retained in the final fused image, indicated that they were, in fact, real.
Caprioli said that in the past he and his team had tried combining images from mass spec imaging and optical microscopy by more or less manually overlaying them using software like Photoshop – an approach that he said was "not very productive."
Ultimately, the idea of mathematically combining the two data sources came to them from outside the world of proteomics. Similar approaches have been used for years in fields like satellite imaging, Caprioli noted. "We got the idea that maybe that would work in this case, and it turns out that it works very well."
While the underlying math is not new, Caprioli said that to his knowledge this was the first time it had been applied to linking imaging mass spec to microscopy data.
Last year, Shimadzu launched its iMScope TRIO instrument, which combines an optical microscope with MALDI TOF and ion trap mass spectrometry with the similar aim of providing high spatial resolution and detailed molecular information. However, the device combines these levels of information using an overlaying-based approach, as opposed to the regression analysis employed by the Vanderbilt researchers.
Caprioli said he sees significant value for his team's approach in analytical pathology applications, particularly as clinicians begin to measure larger numbers of proteins in making their assessments.
"What is happening is that a number of people are measuring signatures of disease now, not just one or two proteins but, say, 10 or 15 that altogether represent a signature of disease." he said. "Now that is going to be really tough to do by antibodies. But imagine if you could just scan over [a biopsy] with a laser and get all 10 of them at the same time. Now blend that with the high resolution of the microscope image, and you have the best of both worlds."
In the meantime, though, some parties are pursuing highly multiplexed antibody-based approaches. Perhaps most notably, GE Healthcare in 2013 launched its MultiOmyx protein detection system, which can multiplex dozens of proteins in a single biopsy while offering subcellular spatial resolution.
The platform uses antibodies conjugated to fluorescent dyes to stain proteins of interest in batches of two to four at a time. Researchers then image the stained tissue and deactivate the fluorescent dyes via a proprietary process. They can then stain the tissue with the next round of antibodies, multiplexing in an iterative fashion. In experiments to test for background resulting from multiple staining cycles, GE researchers have found no added background after as many as 100 cycles.
Fluidigm is also developing a new imaging platform using its CyTOF mass cytometry technology. Targeted for commercial launch this year, the system, which is based on research by University of Zurich researcher Bernd Bodenmiller and Swiss Federal Institute of Technology Zurich researcher Detlef Günther, allows for the multiplexing of dozens of proteins at 1 micron resolution.
The lab of Stanford University researcher Garry Nolan has also developed a technique using mass spectrometry and metal-conjugated antibodies for highly multiplexed protein analyses at the subcellular level. That method uses secondary ion mass spectrometry on a magnetic sector mass spectrometer to measure multiple proteins with a resolution of around 200 nanometers.