Quantifying immunogold labelling patterns of cellular compartments when they comprise mixtures of membranes (surface-occupying) and organelles (volume-occupying).

In quantitative immunoelectron microscopy, subcellular compartments that are preferentially labelled with colloidal gold particles can be identified by estimating labelling densities (LDs) and relative labelling indices (RLIs). Hitherto, this approach has been limited to compartments which are either surface occupying (membranes) or volume occupying (organelles) but not a mixture of both (membranes and organelles). However, some antigens are known to translocate between membrane and organelle compartments and the problem then arises of expressing gold particle LDs in a consistent manner (e.g., as number per compartment profile area). Here, we present one possible solution to tackle this problem. With this method, each membrane is treated as a volume-occupying compartment and this is achieved by creating an acceptance zone at a fixed distance on each side of membrane images. Gold signal intensity is then expressed as an LD within the membrane profile area so created and this LD can be compared to LDs found in volume-occupying compartments. Acceptance zone width is determined largely by the expected dispersion of gold labelling. In some cases, the zone can be applied to all visible membrane images but there is a potential problem when image loss occurs due to the fact that membranes are not cut orthogonal to their surface but are tilted within the section. The solution presented here is to select a subset of clear images representing orthogonally sectioned membranes (so-called local vertical windows, LVWs). The fraction of membrane images forming LVWs can be estimated in two ways: goniometrically (by determining the angle at which images become unclear) or stereologically (by counting intersections with test lines). The fraction obtained by either method can then be used to calculate a factor correcting for membrane image loss. In turn, this factor is used to estimate the total gold labelling associated with the acceptance zone of the entire (volume-occupying) membrane. However calculated, the LDs over the chosen (membrane and organelle) compartments are used to obtain observed and expected gold particle counts. The observed distribution is determined simply by counting gold particles associated with each compartment. Next, an expected distribution is created by randomly superimposing test points and counting those hitting each compartment. LDs of the chosen compartments are used to calculate RLI and chi-squared values and these serve to identify those compartments in which there is preferential labelling. The methods are illustrated by synthetic and real data.