Mix’n’match: HuBMAP Centers Develop and Refine Registered, Mixed-Imaging Methods

Mix’n’match: HuBMAP Centers Develop and Refine Registered, Mixed-Imaging Methods

HuBMAP SciTech Webinar, October 11, 2021

One of HuBMAP’s central goals is to visualize different biological signals and structures in an integrated, high-definition way that enables study of interacting life processes. At this month’s SciTech Webinar, two HuBMAP collaborations presented their work integrating imaging methods that provide complimentary information.

Imaging of Proteoforms in Tissues Using nano-DESI Mass Spectrometry Imaging

Jeannie Camarillo and Pei Su of Northwestern University’s RTI described their group’s collaboration with the Purdue University TTD and the Vanderbilt University TMC in combining individual ion mass spec (I2MS) with nanoDESI in the human kidney. I2MS, Camarillo said, is a type of “top-down” mass spectrometry (MS) that provides molecular weight information/identification of intact biochemical species, in contrast to “bottom up” MS that acquires information from fragments of proteins and other biomolecules. In addition to providing label-free imaging, the method commonly enables molecular identification of greater than 10,000 biomolecules in a single sample.

NanoDESI (nanospray desorption electrospray ionization) is a complementary method that enables ionization and capture of biomolecules in situ from biological samples. Combining the two methods (PiMS, positive-ion MS), according to Su, promises lower bias for proteins, identification of whole proteoforms and fine spatial resolution.

Conventional MS is typically limited to molecules up to 20 kDa, Su added; to date, the group has used PiMS to identify proteoforms with a limit of about 100 kDa. The method identifies protein molecular weights with fine distributions, enabling detection of relative abundance of more than 250 different proteoforms in different anatomical locations. The group has validated their proteoform identifications by two independent methods: their Bioinformatics Engine compares masses with the Proteoform Atlas (>250 identified to date); on-tissue, top-down MS obtains both sequencing information and post-translational modifications (24 to date).

The group showed how PiMS images (color-coded for different molecules of interest), autofluorescence and optical images can be displayed to reveal anatomical structure with excellent alignment. One such difference that they demonstrated was a single-amino-acid substitution of threonine for serine in the GST A2 protein — a molecular mass difference of only 14 Daltons, and one that an antibody-based labeling method probably would have missed.

The group has so far identified proteoforms from 3 to 80 kDa, roughly five-fold higher than previously possible. The method currently has a spatial resolution of 60-80 microns, with a goal of 10-20 microns for the next stage of development. Proteins identified by PiMS align well with kidney structures, vasculature and functional tissue units, enabling visualization and study of those units.

Integration of multi-parametric, multi-modality & multi-resolution data acquired using MALDI and Cell DIVE

A collaboration between the Vanderbilt TMC and GE Research RTI presented their proof-of-concept integration of MALDI-MS and Cell DIVE, technologies that visualize complementary populations of biomolecules. MALDI Imaging MS, with which the Vanderbilt group has extensive experience, has a unique capability of detecting lipid compounds and other metabolites that are imaged poorly or not at all by other methods; the antibody-based Cell DIVE, pioneered by the GE group, generates spatial proteomic data for over 60 proteins per sample.

While the information obtainable by combining the methods would be valuable, integrating them is challenging. One major obstacle is that Cell DIVE is performed on formalin-fixed, paraffin-embedded tissues, which severely affects the sensitivity of MALDI-MS, typically performed on fresh-frozen samples. Cell DIVE works by sequentially staining the sample with antibody probes, requiring sustained tissue adherence to the slide that is difficult to sustain in tissue damaged by the laser-driven liberation of molecules required for DESI. Finally, MALDI-MS’s 10-micron resolution matches poorly with the sub-micron resolution possible with Cell DIVE.

To date the group has successfully adopted the Vanderbilt matrix removal, fixation and pemeabilization process for fresh frozen tissues from non-used donor kidneys, optimizing tissue handling to avoid adherence issues. While the standard 1-hour Cell DIVE staining was not always successful for visualizing some proteins with the mixed method, an overnight staining protocol overcame this problem in most cases.

The group has also studied the effects of laser power and resolution in the MALDI-MS step on staining performance in the subsequent Cell DIVE step. One positive of the laser damage was that the laser ablation marks helped register the MALDI-MS, Cell DIVE and autofluorescence images. They have also worked to understand the impact of tissue quality on performance. Finally, they have also used Cell DIVE-only labeling and autofluorescence of serial sections to help identify and untangle nonlinear deformations of the samples.

Next steps for the project include using a deep-learning model to complete their segmentation workflow; characterizing and optimizing the impact of MALDI-MS parameters and matrix composition on the quality of the integrated data; processing more kidney samples as well as expanding the method to skin tissue samples; and implementing computational approaches for correlative analysis of the data.