Pit cells — a new type of cell first described here and so named because they contain highly characteristic granules — are situated in the wall of rat liver sinusoids, and have hyaloplasmic pseudopodia intermingling with the microvilli of the parenchymal cells. The characteristic granules are mainly situated at one side of the nucleus, the other side showing organelle-free hyaloplasm. Pit cells are also found in portal tracts and in granuloma-like cellular aggregates. They also occur in rat peripheral blood, although there are morphological differences between cells in these two sites. Pit cells can be regarded as regular inhabitants of the sinusoidal wall, and therefore belong to the series of sinusoidal cells, i.e., the endothelial (Wisse, 1972), Kupffer (Widmann et al., 1972; Wisse and Daems, 1970; Wisse, 1974a,b), and fat-storing cells (Ito, 1973). Pit cells do not phagocytose and do not react to a great number of experimental conditions, to which endothelial and Kupffer cells do react (Wisse, 1972, 1974b). Mitosis has been observed in a pit cell.

Kupffer cell fine structure was studied under a number of conditions, i.e., after i.v. injections with Thorotrast (from 3 minutes to 7 days), splenectomy, RES stimulation with Zymosan, bile duct ligation, and partial hepatectomy, as well as in Gunn rat livers. After Thorotrast injection, many thrombocytes were seen in the vicinity of the Kupffer cells; wormlike structures and bristle-coated micropinocytotic vesicles were found to play only a minor role in endocytosis; strong phagocytosis occurred after intravenous aggregation of the particles; chylomicrons were trapped within these aggregates; and at 3 hours phagocytosed Thorotrast-containing degranulated neutrophils were observed. After 2 days, newly formed dense bodies contained less Thorotrast. Granuloma formation was observed after the injection of both Thorotrast and Zymosan; the Zymosan particles were taken up exclusively by Kupffer cells. After Zymosan, splenectomy, bile duct ligation, and partial hepatectomy, as well as in Gunn rats, there were only minor changes in Kupffer cell morphology except for mitotic figures occurring after Thorotrast, Zymosan, and partial hepatectomy. This phenomenon is interpreted as contradicting the hypothesis that Kupffer cells derive from other cells, including endothelial cells and monocytes. Certain differences between monocytes and Kupffer cells with respect to phagocytic capacity, morphology, origin, and cytochemistry are discussed. No evidence was obtained against the postulate that Kupffer cells are able to divide and give rise to daughter cells.

Continuation of the investigation of sinusoidal lining cells in rat liver (120–124) has led to a description of the morphology and peroxidase cytochemistry of normal Kupffer cells. Kupffer cells were found to have a variable but often stellate shape and a surface fuzzy coat, visible after "direct" osmium fixation. Worm-like structures were restricted to Kupffer cells, as were the annulate lamellae. The dense bodies varied in shape, diameter, and density. Kupffer cells may rest on or be inserted in, but are not continuous with, the fenestrated endothelial lining (120). No gaps were observed between Kupffer cells and endothelial cells. These morphological features make it possible to distinguish the Kupffer cells from other sinusoidal lining cells. The Kupffer cells did not show fat droplets, glycogen particles, autophagic vacuoles, or multivesicular bodies. After incubation by either perfusion or immersion for the demonstration of peroxidatic activity in livers fixed for 40 seconds, only the Kupffer cells showed a positive reaction in the nuclear envelope, the RER, and annulate lamellae. This specific staining reaction therefore permits the light-microscopical recognition of Kupffer cells. Time-series incubation of Vibratome sections showed that the peroxidatic reaction in Kupffer cells is sensitive to the mode of fixation, H2O2, pH 9.0, 2 × 10−2 M AT, 10−2 M NaN3, boiling, and the omission of DAB or H2O2 from the medium. No significant retardation of the reaction occurred in the presence of 10−2 M KCN. It is concluded that a peroxidase rather than a catalase is responsible for the reaction with DAB.

The fenestrated endothelial lining of rat liver sinusoids has been described elsewhere (90). The continuity of this fenestrated lining with a certain cell permitted identification of the latter as an endothelial cell (90–92). Among other criteria, the morphology and dynamics of the vacuolar apparatus before and after injection of colloidal Thorotrast particles, the morphological specializations of the endoplasmic reticulum, and the presence of a sphaeridium (nuclear body) serve to characterize the liver sinusoidal endothelial cell and to distinguish it from the Kupffer cell. The endothelial cells could be recognized easily in all livers studied under both normal and experimental conditions, e.g., partial hepatectomy, Thorotrast or Zymosan injections, and splenectomy. No transitional stages between endothelial and Kupffer cells, bearing one or more characteristic features of both cells, were observed. It is therefore concluded that the present data do not support the hypothesis that endothelial cells can develop into Kupffer cells.

Animal cells have an external surface layer closely bound to the plasma membrane, called the cell coat (1) or glycocalyx (for a recent survey see ref.2). This coat is usually composed of acidic glycoproteins or mucopolysaccharides (3), and is thought to play an important role in surface interactions between cells and transport phenomena across the plasma membrane.

After perfusion fixation the sinusoidal lining of rat liver is composed mainly of endothelial cell processes bearing interextending sieve plates. The pore size of the fenestrations, was measured to be 0.1 μ; no larger openings occur in the lining. Rat liver sinusoids can therefore be regarded as fenestrated capillaries differing from other fenestrated capillaries by the absence of a diaphragm within the fenestrations and the absence of a basement membrane surrounding the capillary. Concomitant experiments showed that the fenestrations could be demonstrated irrespective of the method used for preparation, i.e., chemical fixation or freeze etching. With chemical fixation the fenestrations could be made visible irrespective of the fixing compound (glutaraldehyde or osmium) or of the method of application of the fixative solution (perfusion or immersion). The failure to observe the fenestrations after immersion fixation of tissue blocks could not be explained by factors such as (postmortem) autolysis or mechanical deformation, and from this evidence it is concluded that the absence of fenestrations in routine immersion fixation is due to an unknown effect occurring during or caused by the penetration of the fixative into tissue blocks. On the basis of morphological criteria a highly selective filtering effect can be expected in the lining in vivo, especially with regard to the passage of particulate material of varying size, for instance chylomicrons. The connection of the fenestrated lining to a certain type of cell present in the sinusoidal lining can be used to identify this cell as an endothelial cell. This criterion is a new contribution to the morphological distinction between endothelial and Kupffer cells.

Since the first descriptions of the ultrastructure of mitochondria by Palade and Sjöstrand, the attachment of the cristae has been a subject of discussion. Palade considered the cristae to be platelike infoldings of the inner mitochondrial membrane, whereas Sjöstrand described them as septa, only seldom attached to the inner mitochondrial membrane. Andersson-Cedergren was the first to report an alternative way of attachment of the cristae mitochondriales. In the present study, based on serial sections, the attachment of the cristae mitochondriales was investigated in mouse hepatic parenchymal cell mitochondria. The appearance of the cristae mitochondriales was studied in three directions perpendicular to each other. It could be demonstrated that the cristae mitochondriales are attached to the inner mitochondrial membrane by a varying number of round stems, here called the pediculi cristae. Probably in other types of cells the attachment of the cristae is similar to that in mouse hepatic parenchymal cells.