Use of Periodic Acid-Schiff (PAS) Stained Kidney Sections for Deep Visual Proteomics (DVP) Studies

Background: PAS stain has many desirable characteristics for artificial intelligence guided segmentation of biopsy specimens but PAS-stained tissue sections have been avoided for proteomic studies. We conducted studies on commercially purchased human kidney wedge specimens to address the hypothesis that PAS would be equivalent to Hematoxylin & Eosin (H&E) for DVP studies. Methods: Five 5um serial sections from 3 different kidney samples on polyethylene terephthalate membrane frame slides were deparaffinized, rehydrated to water and either unstained or PAS, Periodic Acid (PA)-treated, Hematoxylin (H) stained, or H&E-stained. 20 glomeruli from each kidney sample were annotated on the first serial section and tracked through whole slide images of the remaining serial sections for collection by laser capture microdissection. Samples were prepared for LC-MS analysis using a data dependent analysis (DDA) approach (Merchant ML et al PMID: 32561683). Proteomic differences across the five test conditions were compared by ANOVA (p<0.05) and post-hoc t-tests compared total protein area (TPA) for PAS, PA, HE and H with areas normalized to no stain. Results: 533 proteins were identified across the 15 glomerular samples. Based on TPA, no significant differences were observed across the glomerular digests for the three wedge biopsies comparing no stain, PAS, PA, HE, and H staining. Comparing individual proteins, only 3 of 533 proteins were significantly different by ANOVA. Integrin beta 1 (ITB1) was significantly increased in H versus H&E, PAS, or PA. Elongation factor 2 (EF2) was significantly increased in PAS and PA versus HE or H. Fibulin-1 (FBN1) was significantly increased in H versus PAS, PA or H&E. No significant difference was observed for total peptide area (TPA) across the glomerular digests comparing no stain, PAS, PA, HE, and H. Conclusion: PAS is a suitable stain to guide DVP studies with no significant inferiority to HE, H alone, PA or no staining. Overall, the use of a stain seems to have some benefit for proteomic studies. This benefit is speculated to be related to the process of deparafinization and heating prior to the proteomic processing. We anticipate the use of PAS-stained tissues will open the door for directly integrating AI-guided pathologic assessment of images with DVP analysis of the primary stained tissue. Funding: NIDDK Support