Quantitative fluorescence microscopy studies of prostate cancer at the single-cell level using automated image analysis

Prostate cancer is associated with altered nuclear and nucleolar organisation, but fluorescence microscopy studies in prostate cell lines commonly use permeabilised immunofluorescence workflows and rarely include a non-permeabilised baseline. It is often unclear whether permeabilisation is a neutral preparation step or whether it changes the measurements used for interpretation. This thesis established a fluorescence microscopy workflow for preparation-aware, single-nucleus analysis in two prostate cell lines: LNCaP (malignant) and PNT2-C2 (non-malignant). Fixed cells were analysed under three matched conditions: non-permeabilised DAPI staining, permeabilised DAPI-only staining, and permeabilised DAPI plus anti-fibrillarin (DAPI+FIB) immunofluorescence. Nuclear morphology, DAPI-derived intensity measures, and fibrillarin-defined nucleolar measures were quantified per nucleus, with distribution-level analysis alongside mean values.

Convergence testing showed that descriptor estimates stabilised with increasing nucleus count, and that LNCaP required larger sample sizes than PNT2-C2, consistent with greater heterogeneity. In the non-permeabilised baseline, LNCaP nuclei were ∼ 45% larger, ∼ 20% more elongated, and showed greater dispersion in intranuclear DAPI staining/texture, with mean DAPI intensity coefficient of variation ∼ 40% higher and the kernel density estimate (KDE) mode ∼ 60% higher. However, permeabilisation materially altered several measurements. Most clearly, it reversed the cell-line ordering for nuclear area, and strongly impacted absolute DAPI intensity and density, showing that permeabilisation was not neutral for all descriptors. By contrast, nucleolar analysis under the DAPI+FIB workflow showed little between-line difference in nucleolar area, but clearer elevation of
nucleolar intensity in LNCaP, with mean nucleolar intensity ∼ 25% higher and the nucleoli:DAPI ratio ∼ 35% higher.

Overall, quantitative single-nucleus fluorescence microscopy identified robust differences between malignant and non-malignant prostate cell models, but only under a preparation-aware interpretation. The non-permeabilised condition was essential as both a rarely evaluated baseline, and the reference needed to show that permeabilisation can itself reshape nuclear readouts. Key limitations are preparation-sensitive fluorescence measurements and segmentation/thresholding errors.