Uplsapo 20

Peritoneal biopsies were formalin-fixed in 4% buffered formalin and paraffin-embedded following routine protocols [29 (link)]. Immunohistochemistry was performed as previously described [28 (link),30 (link),31 (link)]. Used antibodies: anti-TXNIP (ab188865; 1:200, Abcam, Cambridge, UK), anti-TRX (sc-271281; 1:100, Santa Cruz), anti- γH2AX (9718S; 1:100, CST).
For morphology, slides underwent hematoxylin and eosin (HE) staining. Tissue sections were examined with Olympus VS120 automated slide scanner equipped with a BX61VS microscope (objective: UPLSAPO 20× or UPLSAPO 2 40×, Olympus, Tokyo, Japan). Mesothelial cell loss was evaluated as: completely contained (0); partially lost (1); completely lost (2). Mean submesothelial thickness was evaluated by measuring the thinnest and thickest layer using the imaging software (OlyVIA V3.3, Olympus).

Multi-detector imaging represents a technique where emitted fluorescence from a sample exposed to one excitation wavelength is simultaneously detected by two or more photomultiplier tubes (PMTs). To enable this type of imaging, the confocal laser scanning microscope (Olympus, Stockholm, Sweden) was first loaded with the software Fluoview FV1000 (Olympus, Stockholm, Sweden) using factory pre-sets for Alexa Fluor 405, Alexa Fluor 488, Alexa Fluor 546 and Alexa Fluor 633. As this automatically loads the 405, 473, 559 and 635 lasers, bandwidth filters and PMTs into the ‘ON’ position, we manually adjusted or turned off these settings as required. By selecting Sequential acquisition by ‘line’ mode and placing the PMTs into ‘Group 1’, the resulting detection wavelengths were 430–455 nm, 490–540 nm, 575–620 nm and 655–755 nm. Unnecessary excitation wavelengths were then manually shut off in the ‘Acquisition settings’ panel. The same settings for Voltage, Gain and Offsets were applied to all detectors. Phase contrast images were collected in the transmitted light detector with excitation at 473 nm. Image stacks were collected using the lenses UPLSAPO 20 × NA0.75, UPLSAPO 40 × 2 NA0.95 and UPLSAPO 60XW NA1.2 (Olympus, Stockholm, Sweden).

Images were acquired on a Yokogawa CellVoyager 7000 automated microscope equipped with a CSU‐X1 spinning disc, Neo sCMOS cameras (Andor) and UPLSAPO 20× (NA 0.75, Olympus) lens. CellProfiler software was used for image analysis, cell segmentation and single‐cell feature quantification as described in Stoeger et al (2015). We segmented the nuclear periphery by expanding and shrinking the nucleus segmentation by 5 pixels. We segmented the cytoplasm by masking the cell segmentation by the expanded nucleus. The CellProfiler pipeline is available as Dataset EV2. We employed CellClassifier (https://www.pelkmanslab.org/?page_id=63) for data clean up and classification of transfected cells and cells in S‐phase of the cell cycle. We excluded missegmented cells, mitotic cells and cells displaying staining artefacts from further analysis (Stoeger et al, 2015). Computations were performed on the Brutus computing cluster (ETH Zürich) using the task manager iBRAIN.