The ultrafiltration barrier of Mytilus edulis, Chlamys hastata, and Mercenaria mercenaria was studied (1) to resolve whether the basal lamina or the substructure of the ultrafiltration slits (slit diaphragm) is the principal filter and (2) to address the question of permselective behavior of the ultrafiltration barrier. The ultrafiltration slits in M. edulis are 15-22 nm wide. In grazing sections of the filtration barrier, projections from adjoining pedicels protrude into the ultrafiltration slits. These projections have a periodicity of about 20 nm and occur in both staggered and opposing patterns. However, considerable variability characterizes the organization of the slit diaphragm. Tracer experiments (mostly in Chlamys hastata and Mercenaria mercenaria) revealed that horseradish peroxidase (HRP, 40 kDa) is passed through the ultrafiltration barrier; colloidal gold (16 nm) and native ferritin (400 kDa), on the other hand, are retained at the level of the basal lamina. In addition, the distribution of HRP indicates that podocyte and pedicel surfaces are negatively charged. Highly anionic sites in the basal lamina were detected with ruthenium red. We conclude that (1) the basal lamina is the principal filter, and (2) the cutoff range for ultrafiltration is larger than 40 but smaller than 400 kDa. We hypothesize that the negative charges are important in filtration barrier function and maintenance. Finally, comparing bivalve ultrafiltration and mammalian glomerular filtration, we conclude that they share important characteristics: the basal lamina/basement membrane is the principal ultrafiltration barrier and the basal lamina/basement membrane and podocyte/pedicel surfaces are negatively charged. Additional key words: podocyte, ultrastructure, permselectivity Ultrafiltration, the initial step of primary urine formation, is of fundamental importance for the excretory and osmoregulatory performance of animals. Detailed physiological (Hevert 1984; Robinson & Morse 1994) and ultrastructural evidence (Pirie & George 1979; Meyhofer et al. 1985; Khan et al. 1988; Andrews & Jennings 1993) has firmly established that ultrafiltration in bivalves takes place within the pericardial glands. Similar understanding has been gained about the ultrafiltration process in other molluscs (see reviews by Martin 1983; Andrews 1988; Morse & Reynolds, in press). Despite this insight into understanding the cytomorphology of the sites of ultrafiltration, little is known about the chemical or structural basis for ultrafiltration or the molecular size range of materials in bivalves or molluscs in general. Bivalve pericardial glands, which include the auricular glands of Andrews & Jennings (1993), are brownish in color and are associated with the surface of the auricle, or with the pericardial cavity lining, or are found as glandular outpockets from the pericardial wall, lying in the hemocoel between the outer mantle epithelium and the pericardial epithelium. Interdigitating basal extensions of pericardial gland epithelial cells, the podocytes, along with their underlying basal lamina, form the ultrafiltration barrier that separates two cavities, the hemocoel and the coelomic space or urinary space; this latter space empties via the coelomic kidney into the mantle cavity (Meyhofer et al. 1985; Morse & Zardus, in press). Two structural components of the ultrafiltration barrier, the substructure in the filtration slits and the basal lamina, have been implicated as the principal filter in gastropod molluscs. Boer & Sminia (1976) studied the three-dimensional substructure in the ultrafiltration slits of the gastropod Viviparus viviparus and presented evidence that this substructure effectively produced a 9-by 11-nm sieve or diaphragm. They concluded, This content downloaded from 157.55.39.105 on Fri, 07 Oct 2016 04:29:16 UTC All use subject to http://about.jstor.org/terms Bivalve ultrafiltration supported by evidence of Rodewald & Karnovsky (1974) from the mouse and rat, that the principal filtration barrier in primary urine formation is the substructure in the ultrafiltration slits. Andrews (1979), however, questioned this conclusion, based on her observations of a wider spacing of the substructure in the ultrafiltration slits of Viviparus contectus and V. japonicus; instead she suggested that the basal lamina is the principal filter. Electron-microscopical studies with various tracer molecules on the vertebrate glomerular ultrafilter presented strong evidence that the basement membrane is the principal filter (for reviews see Farquhar 1982; Kanwar et al. 1991). Also in vertebrates, the sizeand charge-selective properties of the ultrafilter and the composition and structural properties of the basement membrane have been studied in considerable