Label-Free Characterization and Quantification of Mucosal Inflammation in Common Murine Colitis Models With Multiphoton Imaging

Abstract

Background Clinical challenges in inflammatory bowel diseases require microscopic in vivo evaluation of inflammation. Here, label-free imaging holds great potential, and recently, our group demonstrated the advantage of using in vivo multiphoton endomicroscopy for longitudinal animal studies. This article extends our previous work by in-depth analysis of label-free tissue features in common colitis models quantified by the multiphoton colitis score (MCS). Methods Fresh mucosal tissues were evaluated from acute and chronic dextran sulfate sodium (DSS), TNBS, oxazolone, and transfer colitis. Label-free imaging was performed by using second harmonic generation and natural autofluorescence. Morphological changes in mucosal crypts, collagen fibers, and cellularity in the stroma were analyzed and graded. Results Our approach discriminated between healthy (mean MCS = 2.5) and inflamed tissue (mean MCS > 5) in all models, and the MCS was validated by hematoxylin and eosin scoring of the same samples (85.2% agreement). Moreover, specific characteristics of each phenotype were identified. While TNBS, oxazolone, and transfer colitis showed high cellularity in stroma, epithelial damage seemed specific for chronic, acute DSS and transfer colitis. Crypt deformations were mostly observed in acute DSS. Conclusions Quantification of label-free imaging is promising for in vivo endoscopy. In the future, this could be valuable for monitoring of inflammatory pathways in murine models, which is highly relevant for the development of new inflammatory bowel disease therapeutics.

SEMPAI: a Self‐Enhancing Multi‐Photon Artificial Intelligence for Prior‐Informed Assessment of Muscle Function and Pathology

“…As one limitation of this study, while intended as a general‐purpose tool, SEMPAI was only evaluated for muscle research. In the future, we plan to expand SEMPAI to other organ models, including existing gastroenterological [ 60 , 61 , 62 ] and pneumonological [ 63 ] MPM databases and respective priors. Further, SEMPAI did not yield a good predictive performance with 3D DL based on the underlying architecture.…”

“…We tested this approach with a large exclusive dataset of 3D SHG images of single muscle fibers with a multitude of pathologies and functional properties, which result from over a decade of experiments. The meta‐learning on a large database aims to build foundation models for different organs, which could find future application when translating from ex vivo [ 62 ] to in vivo experiments [ 60 , 61 ] or from animal models to humans. Both through the systematic analysis of differences and similarities between experiments and pathologies and the adaptation of the method by meta‐learning, as well as through the continuous expansion of its database, we expect a continuous self‐enhancement of the method.…”

Anti‐β7 integrin treatment impedes the recruitment on non‐classical monocytes to the gut and delays macrophage‐mediated intestinal wound healing

“…The optical filters were chosen to target second harmonic generation from collagen fibers (ET405/20, Chroma), and natural autofluorescence from NADH in mucosal epithelial cells (450/70 BrightLine HC, Semrock Inc.). 29 , 30 In addition, a third filter was used to target specific fluorescence from either GFP (525/50), tdTomato (572/35) or RFP (595/40) expressed by the respective reporter macrophage/monocyte subtype. MPM stacks were recorded at an axial spacing of 2 µm.…”

“…A water immersion objective (HC Fluortar L 16x/.6 W VISIR, Leica microsystems) was used on an upright multiphoton microscope (TriMScope II, LaVision BioTec) at a wavelength of 810 nm. The optical filters were chosen to target second harmonic generation from collagen fibers (ET405/20, Chroma), and natural autofluorescence from NADH in mucosal epithelial cells (450/70 BrightLine HC, Semrock Inc.) 29,30 . In addition, a third filter was used to target specific fluorescence from either GFP (525/50), tdTomato (572/35) or RFP (595/40) expressed by the respective reporter macrophage/monocyte subtype.…”