Automated, image-based quantification of peroxisome characteristics with perox-per-cell

Abstract Summary perox-per-cell automates cumbersome, image-based data collection tasks often encountered in peroxisome research. The software processes microscopy images to quantify peroxisome features in yeast cells. It uses off-the-shelf image processing tools to automatically segment cells and peroxisomes and then outputs quantitative metrics including peroxisome counts per cell and spatial areas. In validation tests, we found that perox-per-cell output agrees well with manually quantified peroxisomal counts and cell instances, thereby enabling high-throughput quantification of peroxisomal characteristics. Availability and implementation The software is coded in Python. Compiled executables and source code are available at https://github.com/AitchisonLab/perox-per-cell.

For all experiments, cells were grown in synthetic defined medium (SD: 6.7 g/L Yeast nitrogen base without amino acids + 0.79 g/L CSM) with 2% Dextrose in flask cultures shaken at 250 rpm at 30 °C until log phase after which they were pelleted and resuspended in 50 µg/ml calcofluor white stain (Sigma, Cat No. 18909) for 5-10 min followed by imaging at room temperature.3D images consisting of 26 XY images with a Z-slice spacing of 0.204 µm (total Z-stack thickness 5.1 µm) were acquired at 100× magnification using a fluorescence microscope (Axioskop 2 MOT plus, Carl Zeiss, Inc.) equipped with a Plan Apochromat 100×/1.4Oil DIC objective, an Axio Cam HRm camera and an HBO 100 Mercury lamp.Identical exposure times (50 ms) were used to acquire the green channel images whereas the exposure time for blue channel was adjusted for individual images based on the intensity of calcofluor staining.
For manual counting, images were deconvolved with theoretically generated point spread functions using Axiovision software V4.9.1 SP2 followed by the generation of maximum intensity Z-projections (MIP) of both green and blue channels.All the deconvolved MIP images from WT and mutant strains were blinded and labelled as '1-44', and their grey levels were set to 'best fit' in the Axiovision software prior to providing them to two individuals who manually counted peroxisomes in individual cells using the 'measure events' tool in Axiovision.Unlike manual counting, no post-processing was performed on the raw images used as the input for perox-per-cell.

Strain construction
All strains used in this study (Table S1) were constructed using lithium acetate-PEG-mediated transformations.For this, the parental wild type (WT) strain (BY4742) or those lacking PEX30, VPS1, PEX3 and INP1 ORFs [obtained from the MATα deletion library (Giaever et al., 2002)] were transformed with a linear construct expressing GFP with Enhanced Peroxisome Targeting Sequence [ePTS1 (DeLoache et al., 2016)] at its 3′ end from the TDH3 promoter and followed by the PGK1 terminator along with the natMX6 selection cassette and targeted after the STOP codon of the UBC9 ORF to fluorescently tag peroxisomes.INP2 was deleted in the WT strain expressing GFP-ePTS1 using the hphMX6 cassette flanked by ~600bp sequences upstream and downstream of the INP2 ORF.

Statistical methods
To choose microscopy images that best represented the number of peroxisomes per cell or peroxisomal areas among a group of images, we first pooled all perox-per-cell results from those images, then performed Wilcoxon rank-sum tests to compare each individual image's results to the pool.The image with the least significant P-value was selected as the most representative.

Fig. S2 .
Fig. S2.Number of cells counted either manually or using perox-per-cell across 44 test images.