Poster K1 - Automated Brain Region Dissection for MAPseq using Laser Dissection Microscopy

Thursday, October 24, 2024

10:00 AM - 11:00 AM EST

Location: Exhibit Hall E, F & G

Presenting Author(s)

Rishi Koneru

Student
Johns Hopkins University
Leander, Texas, United States

Co-Author(s)

MA

Manjari Anant

PhD Student
Johns Hopkins University, United States
HK

Hyopil Kim

Postdoctoral fellow
Biomedical Engineering
Baltimore, Maryland, United States

Primary Investigator(s)

JK

Justus Kebschull, PhD

Assistant Professor
Johns Hopkins University, United States

Introduction: Mapping neuronal projections at single-cell resolution is critical for understanding the intricate wiring of the brain. Multiplexed Analysis of Projections by Sequencing (MAPseq) enables this by injecting unique RNA barcodes into a source brain region, allowing these barcodes to be trafficked down axonal processes, and subsequently sequencing the barcodes at target regions. The precision of target region dissection is essential to accurately evaluate projection patterns. Recently, laser microdissection microscopy (LMD) has been employed to enhance the accuracy of target region dissection compared to manual dissection. LMD's ability to precisely excise imported shape boundaries presents an opportunity for automating the process of dissecting brain slices.

Materials and

Methods: MAPseq tissue sections were obtained from 150um coronal sections of the 12-week-old C57BL6/J mouse brain. The brain was fixed in PFA and sectioned using a cryostat. The sections were then stained with 0.05% Toluidine Blue solution and subsequently mounted on PEN membrane metal frames (Leica Microsystems). The mounted slice was imaged using the Leica Microsystems LMD7 laser microdissection microscopes under a 2.5x lens.

These tissue sections were aligned to a 3D mouse brain atlas (Staeger et al) using a Large Deformation Diffeomorphic Metric Mapping (LDDMM) function obtained from the STalign (Clifton, Anant et al) library. Before performing the alignment, the target image was gray-scaled, normalized, and down sampled to the same pixel size as the atlas. Then, 3D-2D LDDMM was used to produce a transform from the 3D atlas to the MAPseq tissue section. This transform was then applied to the atlas annotations to create a labeled target image. The boundaries for each labeled region were computed using a concave hull, implemented using the Shapely library. Finally, the boundaries were exported to an XML format to be uploaded to the LDM.

Results, Conclusions, and Discussions: We developed software to automatically detect brain regions within MAPseq tissue sections for subsequent LMD dissection. This software aligns MAPseq tissue samples to a 3D common coordinate brain atlas containing detailed brain region annotations, and imports region boundaries into LMD software to facilitate automated brain region detection and dissection. This streamlined system reduces contamination from adjacent regions and decreases operator bias, enabling more reliable and detailed mapping of neuronal projections.

Acknowledgements (Optional):