Characterisation of 3D tissue microstructures and cell distributions in human lung tissue using correlative micro-CT and immunofluorescence

Abstract

Micro-CT (μCT) is a non-destructive three-dimensional (3D) imaging technique used for imaging microscopic structures. The human lungs contain a highly specialised and complex series of 3D networks and structures used for the transport of air and blood that drives gas exchange. The lungs also contain an extensive cellular population of structural and immune cells critical to normal lung function which are widely studied using destructive two-dimensional (2D) histological imaging techniques.
In order to study lung microstructure the 3D imaging provided by μCT was combined with the specific localisation of cell antigens provided by immunofluorescence (IF) to develop a correlative imaging workflow. Automated co-registration of 2D IF images to μCT volumes was possible by the novel use of tissue autofluorescence to automatically identify matching structural features between the μCT and fluorescence images. This was used to inform image transformation tools which overcame sectioning artefacts to enable the direct localisation of specific cell types within the μCT tissue volume, which was only possible by correlating the images. Previous localisation of cell types and features in μCT has mostly relied on time consuming methods of manual segmentation. The interconnected 3D network of pulmonary microvasculature, down to a 25 μm thickness, was segmented semi-automatically from co-registered μCT datasets. The developed workflow automated previous bottlenecks of image registration and segmentation to significantly decrease the total time taken to generate correlative imaging data.
The correlative μCT and IF imaging workflow was used to investigate chronic obstructive pulmonary disease (COPD) as an exemplar respiratory disease with known effects on lung tissue microstructures and cell populations. A progressive decline in lung function occurs in COPD which has been well-studied using clinical 3D imaging techniques, but the 3D microstructure has been less studied. Eleven surgically resected human lung tissue biopsies, fixed in formalin and embedded in paraffin (FFPE), were split into two age-matched clinical groups (6 non-COPD and 5 mild-moderate COPD) based on lung function scores. The blood vessel network in each of the tissue samples was semi-automatically segmented for comparative assessments. The volume and size of the blood vessels did not differ significantly between the two groups. The networks were also assessed by generating a skeletonised blood vessel network which, despite no significant differences being recorded, indicated a potential reduction in the average number of blood vessel branches from 5.3 per mm3 in non-COPD tissue to 1.4 per mm3 in mild-moderate COPD tissue.
IF staining of cytokeratin-18 (Ck18) was used to identify airway epithelium within the same grouped lung tissue volumes as the blood vessels. Airway epithelium damage and loss is reported in COPD however the number of tissues containing airways (n=8) was not sufficient to determine
any biologically significant changes between the groups. IF staining was also used to localise mast cells and macrophages identified using AA1 and CD68 respectively. As exemplar infiltrating cells with a known role in many respiratory diseases, these were used to demonstrate the versatility of the correlative imaging workflow to identify individual cell types which were not visible at the resolution of the μCT. Automated cell counts of whole tissue sections, utilising DAPI nuclei staining, ranged between 100 and 1,000 cells per mm2 of tissue cross-sections. The segmented individual cells were used to provide a novel distribution of mast cells in relation to the 3D networks in the grouped tissue volumes. This showed that the vast majority of mast cells were located within 2 mm of the nearest blood vessel with a peak number of mast cells located around 0.5 mm from the nearest blood vessel. Preliminary work on macrophages showed a higher number of cells per section and a higher degree of cell clustering.
Overall the development and application of the correlative μCT and IF imaging workflow to FFPE lung tissue samples demonstrated the capabilities of this work for investigating 3D tissue microstructure and cell types. These were used on a pilot study for investigating microstructural networks and infiltrating cells in mild-moderate COPD, which can be used as the basis for larger studies in the future. The total time to generate the data is significantly faster (<2 weeks) compared to existing manual techniques (several months) and it is applicable to the huge range of FFPE tissues held in biobanks and archives for future studies.

More information

Identifiers

ORCID for Matthew John Lawson: orcid.org/0000-0003-0115-1698

Catalogue record

Export record