Immunohistochemistry (IHC)

From formalin-fixed paraffin-embedded tissue blocks, 4 μm thick paraffin sections were immunohistochemically stained for CXCR4 (dilution 1:500, clone UMB2, Cat# ab124824, Abcam, UK), CXCL12 (dilution, 1:500, clone EPR1216, Cat# ab155090, Abcam, UK), E-Cadherin (epithelial marker, dilution 1:1000, clone EP700Y, Cat# ab40772, Abcam, UK) and CD45 (myeloid marker, dilution 1:1000, clone MEM-28, Cat# ab8216, Abcam, UK). Antigen retrieval was performed in pH 6 citrate buffer (Sigma, Australia). After incubation with DakoEnVision secondary reagents (Dako, Australia), positive staining was visualized for brightfield using diaminobenzidine (Dako, Australia). Sections were counterstained in Mayer’s hemotoxylin (Sigma, Australia). All sections for brightfield analysis were stained at the same time. Sections were scanned using an Aperio Scanscope AT Turbo (Leica Biosystems, Australia) and images were captured at a resolution of 0.25 μm/pixel. The extent of overall CXCR4 expression was semi-quantified by rating the intensity and presence of CXCR4 staining across the entire section on a scale of 0 (absent), 1 (low), or 2 (medium – high) by a researcher blinded to the diagnosis of the section.

All sections for multiplex staining in both panels were stained at the same time. For multiplex staining, tyramide signal amplification was performed using Perkin-Elmer’s OPAL dye reagents (diluted 1:100 and incubated for 6 min) in 3 spectrally distinct fluorophores that have non-overlapping fluorescent spectra, allowing the individual antigens to be separated from a single image (Additional Figure 1).

Whole sections were scanned at 4× magnification using the Vectra 3 Quantitative Pathology Imaging System and the entire section was encircled to define the area to be analysed. Regions of interest (ROIs) were then automatically generated that encapsulated the area of the tissue section for subsequent analysis at 20× magnification (with a resolution 0.5 μm) using Phenochart (v 1.0, Perkin Elmer) software (Additional Figure 2). Exposure settings and emission/excitation wavelengths for each of the channels and OPAL dyes are listed in Additional Figure 2.

Sections were sequentially stained with primary antibodies in two panels and cell nuclei was stained with Hoechst in both. Panel 1 consisted of CXCR4, E-Cadherin and CXCL12. Panel 2 consisted of CXCR4, CD45 and CXCL12. The sequence of targets and dye in Panel 1 was CXCR4 (OPAL 520), e-cadherin (OPAL 570), and CXCL12 (OPAL 690). The sequence of targets and dye in Panel 2 was CXCR4 (OPAL 520), CD45 (OPAL 570) and CXCL12 (OPAL 690). Hoechst dye was applied as the last step in both panels.

Multiplex analysis was performed on tissue sections from 7 NDC donors and 7 patients with IPF using HALO imaging processing software (Indica Labs Highplex FL version 3.0) on a total of 2867 images (ROIs) (Table 3). There were 1458 ROIs captured in the e-cadherin panel and 1409 ROIs in the CD45 panel. Additional File 3 details the thresholds used for positive cell quantification. Cells in lung tissue were identified using Hoechst staining and thresholds for positivity of the panel markers were manually set by examining 2 randomly chosen ROIs from a NDC donor and IPF patient (total of 4 images). Additional File 3 shows two ROIs quantified for Panel 1 and Additional File 4 shows two ROIs quantified for Panel 2. Thresholds and corresponding cell phenotype combinations are then applied to all ROIs used in the analysis. Quantification is performed on all the ROIs from each tissue section and data is presented as percentage of Hoechst+ cells counted.

Summary of quality control parameters for multiplex immunohistochemsitry analysis

Quality control parameters were quantified from the regions of interest (ROIs) used in the multiplex analysis presented in Table ​Table33 and Figs. 4 and ​and5.5. The parameters of normal (n = 7) and idiopathic pulmonary fibrosis (IPF, n = 7) lung tissue was compared using Mann-Whitney U test. Effect size (Cohen’s D) was calculated for comparisons that were significantly different (*p < 0.05, **p < 0.01). Avg, average

Multiplex quality control variables for each of the panels are detailed in Table ​Table3.3. There were no differences in cell area, cytoplasm area, nucleus area, nucleus perimeter or total area analysed between NDC and IPF tissues. Average nucleus roundness was smaller statistically (less round) in IPF compared to NDC tissues but this parameter does not affect the number of counted cells and is likely a reflection of minor cell compression in fibrotic tissue [16].