Optimization of immunolocalization of cell cycle proteins in human corneal endothelial cells

En face observation of corneal endothelial cells (ECs) using flat-mounted whole corneas is theoretically much more informative than observation of cross-sections that show only a few cells. Nevertheless, it is not widespread for immunolocalization (IL) of proteins, probably because the endothelium, a superficial monolayer, behaves neither like a tissue in immunohistochemistry (IHC) nor like a cell culture in immunocytochemistry (ICC). In our study we optimized IL for ECs of flat-mounted human corneas to study the expression of cell cycle-related proteins.

We systematically screened 15 fixation and five antigen retrieval (AR) methods on 118 human fresh or stored corneas (organ culture at 31 °C), followed by conventional immunofluorescence labeling. First, in an attempt to define a universal protocol, we selected combinations able to correctly localize four proteins that are perfectly defined in ECs (zonula occludens-1 [ZO-1] and actin) or ubiquitous (heterogeneous nuclear ribonucleoprotein L [hnRNP L] and histone H3). Second, we screened protocols adapted to the revelation of 9 cell cycle proteins: Ki67, proliferating cell nuclear antigen (PCNA), minichromosome maintenance protein 2 (MCM2), cyclin D1, cyclin E, cyclin A, p16(Ink4a), p21(Cip1) and p27(Kip1). Primary antibody controls (positive controls) were performed on both epithelial cells of the same, simultaneously-stained whole corneas, and by ICC on human ECs in in vitro non-confluent cultures. Both controls are known to contain proliferating cells. IL efficiency was evaluated by two observers in a masked fashion. Correct localization at optical microscopy level in ECs was define as clear labeling with no background, homogeneous staining, agreement with previous works on ECs and/or protein functions, as well as a meaningful IL in proliferating cells of both controls.

The common fixation with 4% formaldehyde (gold standard for IHC) failed to reveal 12 of the 13 proteins. In contrast, they were all revealed using either 0.5% formaldehyde at room temperature (RT) during 30 min alone or followed by AR with sodium dodecyl sulfate or trypsin, or pure methanol for 30 min at RT. Individual optimization was nevertheless often required to optimize the labeling. Ki67 was absent in both fresh and stored corneas, whereas PCNA was found in the nucleus, and MCM2 in the cytoplasm, of all ECs. Cyclin D1 was found in the cytoplasm in a paranuclear pattern much more visible after corneal storage. Cyclin E and cyclin A were respectively nuclear and cytoplasmic, unmodified by storage. P21 was not found in ECs with three different antibodies. P16 and p27 were exclusively nuclear, unmodified by storage.

IL in ECs of flat-mounted whole human corneas requires a specific sample preparation, especially to avoid overfixation with aldehydes that probably easily masks epitopes. En face observation allows easy analysis of labeling pattern within the endothelial layer and clear subcellular localization, neither of which had previously been described for PCNA, MCM2, or cyclin D1.