The mechanism of K-cell (antibodydependent) mediated cytotoxicity: III. the ultrastructure of K cell projections and their possible role in target cell killing

The name K cell has been suggested for the effector cells in antibody-dependent, cell-mediated cytotoxicity (Anon, 1973). The term is often applied loosely to include any antibody-dependent ‘killer’ cell, including granulocytes. In the present study the term is restricted to lymphoid cells capable of ‘killing’ antibody-coated nucleated mammalian cells (Sanderson & Taylor, 1976), and for convenience lymphoid cells forming contacts with antibody-coated target cells are referred to as K cells in the morphological descriptions in this paper. The evidence associating these cells with cytotoxic activity is considered in the Discussion.

Although little is known about the origin or biological function of K cells, their possible role in defence mechanisms has led to considerable interest in their mode of action and in their distribution. It has been shown that the kinetics of release of target cell components is correlated with morphological changes observable by timelapse cinématography (Sanderson & Thomas, 1977 a, b), and it has been demonstrated that there are many similarities between K cells and cytotoxic T cells. For example, K cells share some physical and morphological properties of T cells, and in both systems the target cell dies in a burst of membrane blebbing (zeiosis) which is quite different from the changes seen when a cell is lysed by antibody and complement. There are, however, important differences between the T and K cell systems. First, T cells appear as a result of antigenic stimulation and possess specific receptors for antigen, whereas K cells are present in normal animals and attach to target cells by means of receptors for the Fc pieces of immunoglobulins. Secondly, whereas with T cells target cell death occurs as a random event in time after contact, varying from seconds to several hours (Sanderson, 1976 a, b), K cells appear to bring about target cell death within about 15 min of contact (Sanderson & Thomas, 1977 a, b).

Ina study of T cell cytotoxicity by electron microscopy, Sanderson & Glauert (1977) observed projections from T cells which had pushed into target cells. These projections were only rarely seen and so it was not possible to come to any conclusions as to their role in the cytotoxic process. It seemed possible that the difficulty of finding these projections in electron micrographs might be related to their short time span, and the study reported in this paper was carried out because the kinetics of K cell cytotoxicity indicated that there would be a better chance of observing the cytolytic event, by fixing material during the first 15 min after contact, than there would be in the T cell system. The results show that K cells do produce long projections and the possibility that these are related to the mechanism of target cell killing is discussed. As a result of these findings we have been encouraged to return to the T cell system and have found that projections are frequently observed in this system as well when the experimental parameters are adjusted to maximize the number of cytolytic events occurring at a certain time. These findings have been reported fully elsewhere (Sanderson & Glauert, 1979).

The mouse mastocytoma (P815) cell line of DBA/2 origin was used as target cells. The cells were grown in stationary culture in RPMI-1640 medium supplemented with 10% foetal calf serum. In each experiment the target cells were labelled with “chromium so that cytotoxicity could be estimated by measuring the release of isotope (Sanderson & Thomas, 1977a).

Agus rat spleen was used as a source of K cells. The population of K cells in the spleen cell suspensions was enriched by Ficoll-triosil separation (by taking the lymphoid cells at the interface), followed by passage through columns of nylon wool (taking the non-adherent cells). Full details of the procedure for the preparation of suspensions enriched in K cells have been described previously (Sanderson & Thomas, 1977b).

Antisera were obtained 14 days after intraperitoneal injection of 3 × 10⁷ P815 cells into rats. The antisera were heat inactivated and used at a final dilution of 1:1000.

The cytotoxic reactions were carried out in 0 · 75 ml polypropylene tubes with pointed bottoms (Walter Sarstedt Ltd, Leicester, England). 2 · 5 × 10⁶ cells from the suspensions enriched in K cells were mixed with 10⁶ P815 cells, in the presence or absence of anti-P8is antiserum. The effector and target cells were prewarmed to 37 °C and were centrifuged at 37 °C immediately after mixing by bringing the centrifuge up to 400 r.c.f. and then applying the brake. The cells were then resuspended on a vortex mixer and centrifuged a second time.

After incubation at 37 °C for 4–15 min from the commencement of the first centrifugation, the cell pellets were fixed by carefully removing the supernatant medium and replacing it with prewarmed fixative, consisting of 2·5 % glutaraldehyde in 0·09 M cacodylate buffer, pH 7·2, containing 2·5 mM calcium chloride. After 15 min at 37 °C, the pellets were kept at 22 °C for a further 45 min. The fixative was then removed and replaced with cacodylate buffer, and the fixed pellets were stored in buffer at 4 °C. Subsequently the pellets were removed from the centrifuge tubes and each pellet was divided into a number of small pieces. These pieces were then postfixed in 1 % osmium tetroxide in 0·1 M cacodylate buffer, pH 7·2, containing 2·5 mM calcium chloride, for 1 h, rinsed briefly in distilled water, stained with 0·5 % aqueous uranyl acetate for 1 h, dehydrated in ethanol and embedded in Araldite by standard techniques. Pellets which showed a tendency to disperse in buffer after the primary fixation in glutaraldehyde were encapsulated in a small drop of 2 % agar (Oxoid, London) before postfixation in osmium tetroxide and subsequent processing.

The Araldite blocks were sectioned on an LKB Ultrotome III or a Cambridge Huxley Mark 2 ultramicrotome, and thin sections were stained with lead citrate and examined in an AEI-EM6B electron microscope operating at 60 kV with a 50-μm objective aperture.

Cytotoxicity was tested by assaying the release of ⁵¹chromium from labelled P815 cells (Sanderson & Thomas, 1977a) after incubation with effector cells for 15 min at 37 °C. Isotope release was determined in two separate, but similar, experiments in which the effector cell to target cell ratio was 25:1. In the first experiment 30% specific release of isotope was obtained, and in the second, 11%. This illustrates the large differences between cell preparations which is a common feature of this type of experiment. The experiment with the highest levels of chromium release gave the greatest number of effector-to-target cell interactions, and the majority of the micrographs illustrating this paper come from this experiment.

Control cultures, in the absence of antibody, showed no close contacts between lymphoid cells and P815 cells.

In the presence of antibody, survey electron micrographs showed that some P815 cells are surrounded by a number of lymphoid cells (Fig. 1), while elsewhere a single cell had made contact with more than one tumour cell (Fig. 2). The cells in contact with P815 cells have the typical morphology of lymphoid cells, with large, indented nuclei surrounded by a thin layer of cytoplasm (Figs. 1, 2), and will be referred to as K cells.

Two types of initial contact could be distinguished between K cells and P815 cells. In the first, and less frequent type, there were large areas in which the membranes of the effector and target cells were in close contact (Fig. 3). In the second, and most frequent type, there were only point contacts separated by areas in which there was a wide space of varying width between the two cells (Fig. 4). The projections from the K cells which are described below appear to develop from these point contacts.

Many of the K cells in contact with P815 cells had surface projections. The production of projections did not appear to be restricted to any particular part of the K cell membrane. For example, projections have developed in all three areas of contact of the K cell with tumour cells in Fig. 2. Furthermore, projections are mainly seen in these areas of contact and not on other regions of the K cell surface (Figs. 1, 2), suggesting that the formation of the projections may be initiated by localized contact between the K and P815 cells. This suggestion is supported by the observation that the tips of the projections are often in close contact with the target cell membrane (Fig, 5). The formation of projections is an early event in the interaction with target cells, since they were most frequently observed in preparations fixed 4-5 min after contact between the effector and target cells had been initiated by centrifugation.

The projections from K cells in contact with P815 cells vary in shape; some are straight (Figs. 2, arrow, and 6), while others are curved (Fig. 5) or branched with complex outlines. The tips of the projections are usually pointed, although a few have blunt ends (Fig. 2, double arrow). When seen in cross-section (Fig. 5, arrow) the projections appear approximately round and, except at the tip, there is a space between the limiting membrane of the projection and the tumour cell membrane (Fig. 5). The projections contain a meshwork of filamentous material (Fig. 6) to the exclusion of ribosomes and other organelles.

In regions the surfaces of the K cell and the target cell appear to have interdigitated (Fig. 6). It is noticeable that the tumour cell interdigitations contain ribosomes, while those from the K cell do not.

Some of the projections have pushed deeply into the cytoplasm of a P815 cell. The full depth of these projections is difficult to determine because of the unlikelihood of sectioning a projection along its full length, but cross-sections of what appear to be parts of projections are observed deep in the cytoplasm of some target cells (Fig. 7, arrows). The plasma membranes of both cells have remained intact.

Observations by time-lapse cinématography have shown that K cells are always close to the target cell nucleus immediately before the onset of changes which lead to target cell death (Sanderson & Thomas, 1977b). More recently (Sanderson, unpublished observations) K cells have been observed to push repeatedly into the nuclear regions of target cells and to distort the outline of the nucleus. It was therefore of interest to note that some of the long cytoplasmic projections from K cells seen in electron micrographs had pushed through the cytoplasm of the target cell and had nearly reached the nucleus (Figs. 6, 8). The nucleus was often indented at the tip of such projections (Fig. 6), although the nuclear membrane, as well as the plasma membranes of the K and target cells, were still intact.

Following K cell contact and interaction with a target cell in the region of the target cell nucleus, the target cell undergoes violent blebbing, or zeiosis (Sanderson & Thomas, 1977b), during which the cell takes on very bizarre shapes. Cells in zeiosis were easily identified in electron micrographs (Figs. 9, 10), and some of these cells still had K cells attached by means of short projections from the K cell (Fig. 9).

At the final stage of lysis the target cell returns to a normal spherical shape (Sanderson & Thomas, 1977b), but becomes swollen and loses most of its cytoplasmic contents (Fig. 11). Cell lysis was observed in material fixed only 15 min after cell contact had been initiated by centrifugation, illustrating the rapidity of the cytotoxic process.

While many of the characteristics of T cell-mediated cytotoxicity have been described (see reviews by Berke, 1977, and Kimura & Wigzell, 1977), the K cell system has not been examined in the same detail. While cytotoxic systems with erythrocyte targets probably involve granulocyte effector cells (for example, see Greenberg, Shen & Roitt, 1973; Frye & Friou, 1975; Golstein & Gomperts, 1975), in systems with nucleated mammalian target cells the most efficient effector cell (K cell) is a low density, non-adherent cell (Sanderson, Clark & Taylor, 1975; Sanderson & Thomas, 1978) and in the present study a cell separation technique has been used to enrich this population of cells. Although these procedures result in a considerable increase in cytotoxic activity (Sanderson & Thomas, 1977a), the cell suspension is by no means a pure preparation of K cells. It is important therefore to consider the possibility that the cells observed in contact with target cells in the present study are unrelated to the cytotoxicity observed in ⁶¹chromium release assays.

Various observations suggest that the lymphoid cells in contact with target cells in our preparations are K cells. Cell contacts are not observed in the absence of antibody, indicating that the initial interaction is antibody-dependent, and most of the other cell types known to have Fc receptors (granulocytes and B cells, for example) are removed by the cell separation procedures used. In addition, analysis of timelapse films of similar preparations has indicated that contacts that do not result in target cell death are rare (Sanderson & Thomas, 1977b), and that the cells responsible for target cell lysis have the morphology typical of lymphoid cells. Furthermore, cells of similar morphology are observed attached to target cells undergoing zeiosis (e.g. Fig. 9).

A few K cells and target cells formed contacts in which the membranes of the two cells were in close, parallel contact with each other over a wide area (Fig. 3), as originally described by Biberfeld & Johannson (1975). More frequently point contacts, separated by regions with a considerable intercellular space, were observed (Fig. 4). These point contacts appeared to be the precursors to the projections from K cells, and it is not clear whether the larger areas of close, parallel contact play an essential role in the cytotoxic process.

The most striking feature of the interaction between K cells and target cells observed in the present study was the formation of long projections from the K cells which pushed into the target cells. These projections were only seen frequently in preparations fixed within 5 min of the initiation of contact between the effector and target cells and could thus be easily missed. In addition, it is important to fix the preparations at the temperature of the incubation (37 °C), since a reduction in temperature may well lead to a retraction of such slender structures.

The long projections from lymphoid K cells which pushed into Pi85 cells caused deep infoldings of the plasma membrane of the target cell and distortions of the nucleus, but there were no other detectable changes in the target cell, and the nuclear and plasma membranes remained intact. There is thus no direct proof that the projections are the cause of target cell death, although two types of evidence suggest that they do play an important role in the cytotoxic process. The rapid movement of K cells near the target cell nucleus just before the onset of cell lysis (Sanderson & Thomas, 1977b) and the accompanying cytoplasmic and nuclear distortions of the target cell are observed under the same conditions under which the projections are most readily found in electron micrographs. In addition, similar projections have been observed during interaction of cytotoxic T cells from immunized mice with target cells (Sanderson, 1977b; Sanderson & Thomas, 1977b).

The formation of projections from K cells as a result of contact with antibody-coated target cells provides an analogy with the membrane movements accompanying the engulfment of particles by phagocytic cells. In both systems the movement of the membrane of the effector cell is stimulated by contact of its receptors with a surface. Although the type of membrane movement is different, it appears that in both systems the receptors provided a signal which stimulates changes in the underlying cytoskeletal system of the cell and the consequent production of projections in the form of filopodia or lamellipodia which contain filamentous material to the exclusion of other cytoplasmic organelles (Hartwig, Davies & Stossel, 1977). In phagocytic cells, projections are formed which move over the surface of the particle that is being engulfed, while cytotoxic cells push out projections into the substance of the target cell with sufficient force to distort the target cell membrane and to dent the nucleus.

The mechanism by which these projections might cause target cell death is not yet clear. On the one hand, the fact that zeiosis occurs simultaneously over the whole cell surface suggests that some type of widespread change has occurred within the cell, and not just local membrane damage. On the other hand, there are no obvious changes in the cytoplasmic or nuclear structure of the target cell, even in the early stages of zeiosis. The possibility cannot be ruled out, however, that the projections produce lesions in the target cell which are too small to be visible in electron micrographs of thin sections, but which are sufficient to allow changes in the distribution of soluble factors and ions and which are lethal to the cell. It seems likely that these lesions occur in the nuclear area of the cell and perhaps in the nuclear membrane itself. Observations which provide an experimental analogy to the possible effect of the projections were reported by Munro & Daniel (1965). These authors found that cells recovered rapidly from major disruption of their plasma membranes, but that a micro-electrode introduced into the nuclear region led to cell death, preceded by violent zeiosis.

We acknowledge the support of the Sir Halley Stewart Trust (to A.M.G.) and we are very grateful to R. A. Parker and Janet Atherton for skilled technical assistance and to the Wellcome Trust for the loan of the AEI-EM6B electron microscope.