Integrated stereological and biochemical studies on hepatocytic membranes. III. Relative surface of endoplasmic reticulum membranes in microsomal fractions estimated on freeze-fracture preparations.

Abstract

New methods are required for identifying membranes in subcellular fractions with respect to their origin, if such preparations are to be evaluated morphometrically. One method is freeze-fracturing which reveals intramembrane particles whose size, pattern, and numerical density differ for various membrane types. The question is examined whether the differences in numerical particle density per square micrometer of membrane (alpha) can be used to differentiate membrane vesicles found in microsomal fractions from liver cells with respect to their origin in the hepatocytes. It is found that the range of alpha for the protoplasmic face (PF) of endoplasmic reticulum (ER) membrane (1,900 less than alpha less than 3,250) is intermediate between those for plasma and mitochondrial membranes. Since PF(ER) should appear in the outer leaflet of microsomal vesicles, alpha was estimated on concave profiles of freeze-fracture preparations; the numerical frequency distribution of vesicles with respect to alpha was trimodal, with a major peak around 2,900/micrometer2 and 66% of the vesicles in the range determined for PF(ER). Using a new stereological method, it was calculated that 63% of the membrane surface in these microsomal fractions was of ER origin by this criterion. On the same preparations, an attempt was made to label the ER-derived membranes cytochemically for glucose-6-phosphatase. A line intersection count revealed 62% of the membrane surface to be of ER origin on the basis of marker enzyme labeling. These findings indicate a smaller part of ER membranes in microsomal fractions than would be predicted from biochemical data (77%). The possible reasons for such discrepancies are discussed; shifts in particle densities due to the preparation procedure could lead to an underestimate by freeze-fracturing, whereas the prediction from biochemical data could be overestimates if marker enzymes were not homogeneously distributed.