NEW YORK – A team led by scientists at Helmholtz Munich and the Max Planck Institute of Biochemistry has developed a workflow combining tissue clearing, robotic tissue extraction, and mass spectrometry to generate deep proteomic profiles of biological features while retaining 3D spatial information.
Detailed in a study published in Cell in December, the approach could let researchers better study the initial stages of disease development, said Matthias Mann, head of the department of proteomics and signal transduction at Max Planck and one of the senior authors on the paper. The technique's combination of tissue clearing and high-sensitivity mass spec allows for identification and proteomic profiling of affected cells early in a disease process, Mann said.
Tissue clearing is a sample processing method that uses various chemical techniques to turn tissue sections or even whole organisms transparent, allowing researchers to better examine features within the sample. In recent years, different clearing approaches have been combined with high-resolution microscopy to examine specimens at the cellular and subcellular level. Scientists have also added tools like fluorescent tagging and AI-based image analysis to further pinpoint cellular and structural abnormalities of potential interest.
Ali Ertürk, director of the Helmholtz Munich Institute for Tissue Engineering & Regenerative Medicine and senior author on the study, is one of the developers of the DISCO (3D imaging of solvent-cleared organs) tissue clearing approach, which the researchers employed in the Cell paper. One of the first questions about the workflow, Mann noted, was whether it was compatible with mass spec-based proteomics, a prospect about which he admitted he was initially skeptical.
"This [cleared tissue] is as hard as a rock," he said. "So how can we get proteins out of it?"
Mann's former graduate student Andreas-David Brunner (who has since taken position as a scientist at drug company Boehringer Ingelheim) worked out protocols for extracting and analyzing proteins from the cleared samples. To Mann's surprise, the approach "worked like a charm."
Tissue clearing, in fact, decreased the depth of proteome coverage accessible in the samples by only a small amount, he said, while having little impact on quantitation. Comparing measurements made in fresh frozen tissue to those made in cleared tissue, the researchers found that they could measure roughly 7,800 proteins and that the measurements made in cleared tissue correlated strongly with those made in fresh frozen tissue.
"That was a very big surprise, indeed," Mann said.
Having determined that the DISCO treatment was compatible with mass spec-based proteomics, the researchers were able to incorporate proteomic measurements into a workflow developed by Ertürk's lab that uses laser capture microdissection (LCM) to extract tissue regions of interest and then builds 3D models of these regions by combining data from the multiple 2D LCM tissue sections.
Because tissue clearing allows researchers to see into the depths of a sample with high resolution, it makes it possible to identify and therefore investigate the earliest manifestations of a disease, Mann said. He cited the study's examination of small amyloid plaques in the brains of mice developing Alzheimer's.
"It turns out that these plaques form at an even earlier stage than people thought," he said. "And we can now go in and harvest these plaques and see what is happening in the very early stages. If one didn't have this clearing, we wouldn't know where these early plaques are, and so one would have to study the disease at a much later time point when they were much bigger."
Examining tissue regions impacted by these early-stage amyloid plaques, the researchers identified a number of proteins linked to Alzheimer's disease. Comparing these regions to adjacent sections without plaques, they identified 43 differentially regulated proteins including ones linked to vesicle fusion, vesicle mediated transport, and secretory pathways and showing significant quantitative change.
Comparing the proteomic signatures of these early-stage plaques to those of later-stage plaques, they identified 70 proteins that differed significantly in expression levels. They also observed proteome differences between the dorsal and ventral subiculum regions of the hippocampus, "indicating region and time-specific changes in the proteomic landscape" of the plaque microenvironments.
Harsharan Bhatia, spatial omics team leader at Helmholtz, said that he and his colleagues believe the approach could let them better investigate the heterogeneity of disease processes like plaque formation in Alzheimer's.
"Normally what we do is look at cellular interactions in 2D," he said. "But the behavior and the microenvironment of [for example] a plaque, is very heterogenous. Understanding the molecular mechanisms and molecules around theses plaques in 3D space will give us better [insights]."
Bhatia suggested that, for instance, more thorough characterization of the heterogeneity of early-stage Alzheimer's plaques and their microenvironments could help drug companies better understand why their anti-amyloid therapies have been less effective than hoped.
"We can learn better, and accordingly, we can engineer our drugs better," he said.
The researchers also demonstrated the utility of the technique for studying the development of calcified atherosclerotic plaques in a cleared human heart. Mann also highlighted potential applications in areas like cancer where he said the technique could, for instance, allow researchers to identify and isolate micro metastases and analyze their proteomes to identify molecules driving cancer development.
In addition to combining proteomics with its tissue clearing approach, Ertürk's lab developed a robotic tissue extraction system (called DISCO-bot) that uses robot-guided needle biopsies to extract tissue regions of interest, which the authors said will allow them to scale their experiments to look at larger samples. Ertürk has founded a company, Deep Piction, to commercialize the robotic extraction technology.
Tracy Liu, an assistant professor at the West Virginia University School of Medicine whose research focuses on the use of molecular imaging to study cancer, said the approach is "an exciting step towards being able to do 3D analysis," adding that, particularly in cancer, "we are seeing that the spatial distribution of your cells within the tumor microenvironment is very important."
"To be able to get 3D information and to understand the spatial distribution of this stuff is very important," said Liu, who was not involved in the study.
She noted, however, that at least in the case of cancer, she believes the technique needs to offer single-cell resolution to be most useful.
"For me, it's very important to understand for a specific cancer cell what are the specific cells in proximity to it," she said.
She noted that in the Cell study, the researchers were able to analyze individual regions as small as 60 cells.
"That's interesting, but not being able to identify the differences within those 60 cells could be the difference between a cancer cell that may be able to metastasize," she said. "Or within those 60 cells, if just one of those cells has a gene signature or proteomic signature that is resistant to a treatment, that one cell that you can't see within the cluster of 60 could be what causes a person to relapse after treatment."
Bhatia said that he and his colleagues are working to bring their analyses down to the single-cell level, though he noted they are not yet there, citing several challenges including miniaturizing the cell lysis and protein digestion steps in cleared samples and demonstrating that they can still achieve substantial proteome coverage in such samples.
While Ertürk's group leveraged the Mann lab's proteomic expertise in the Cell paper, Bhatia said the group has since established its own mass spec setup, purchasing a Bruker timsTOF SCP instrument, which is intended for analysis of very small sample sizes, down to single cells.