Dynamic changes in viscoelastic properties in cytotoxic T-lymphocyte-mediated killing

The immune system accomplishes cell lysis by utilizing humoral factors, including complement components, and the cytolytic killer cells. The molecular mechanisms leading to complement lysis are fairly well understood (Tranum-Jensen et al. 1978; Podack & Tschopp, 1982; Tschopp et al. 1982; Ramm et al. 1982). Less is known about the molecular mechanisms of cell-mediated cy-tolysis (Martz & Benaceraff, 1976; Kung et al. 1979; Berke, 1983; Henkart, 1985). Long-term allospecific human cytolytic T lymphocyte (CTL) clones have been developed and characterized, and they are useful for the study of the mechanism of killing at the molecular and cellular levels (Reiss et al. 1980; Krensky et al. 1982, 1983). We have recently reported the direct measurement of the junction avidity of single pairs of cloned CTLs and their target cells (TCs) (Sung et al. 1986). Electron microscopy and immunofluorescence microscopy have been performed to examine the interdigitated plasma membranes at the contact region between a CTL and its TC (Sanderson & Glauert, 1979; Rosen et al. 1981), the centrioles and the Golgi system of CTL (Bykovskaja et al. 1978), and the microfilamental-microtubular networks (Geiger et al. 1982; Ryser et al. 1982). TC lysis involves damage of the membrane, submembrane cortical components and intracellular components. The intracellular damage incurred in the TC that has received a lethal hit after CTL-TC conjugation involves DNA fragmentation (Sanderson, 1981; Duke et al. 1983) and bubbling of the nucleus, and the subsequent release of DNA from the nucleus. Specific antigen-receptor bridging between CTL and TC leads to reorientation of actin microfilaments and microtubules, which can be inhibited by monoclonal antibodies (Springer et al. 1982; Goldstein et al. 1982). The damage to the cell membrane probably results from the insertion of tubular complexes by the CTL (Dennert & Podack, 1983; Young et al. 19866; Liu et al. 1986) and the loss of osmotic integrity.

The aims of this study are to use the micromanipulation technique to elucidate the intracellular damage initiated by a lethal hit through the junction zone. Our findings indicate that the lysis of the TC involves complex biophysical events and provide new information on the biophysical aspects of the cytolytic process in relation to the biochemical events.

JY (HLA-A2, 2; B7, 7; DR4, w6) is an Epstein-Barr virus-transformed B lymphoblastoid cell line. JY cells were maintained in RPMI-1640 medium (M. A. Bioproducts, Bethesda, MD) supplemented with 10% heat-inactivated foetal calf serum (M. A. Bioproducts), 2mM-L-glutamine (GIBCO, Grand Island, NY), penicillin at 100 units ml⁻¹ and streptomycin at 100μgml⁻¹ (GIBCO), 100mM-Hepes buffer (M. A. Bioproducts), and 25 μM-2-mercaptoethanol (Eastman Organic Chemicals, Rochester, NY).

Mononuclear cells obtained from the peripheral blood of a normal donor, M.P. (HLA-A11, Aw32; B27, Bw51; Cw2, DR7, 7; Genex 3.53 ⁻), were separated on a Ficoll-Hypaque gradient (1ymphocyte Separation Medium, Bionetics, Kensington, MD) at 600g, washed repeatedly and stimulated with irradiated JY cells (10000R) as described (Reinherz et al. 1980). Bulk CTL cultures were maintained in supplemented RPMI-1640 medium as described above for 3–4 weeks with weekly JY stimulation, and then cloned by limiting dilution at less than one cell per well in flat-bottomed 96-well microtitre plates (1INBRO, Flow Laboratories, Inc., McLean, VA) with 1×10 ⁵ to 2×10⁵ irradiated JY cells per well and human conditioned medium containing 10% interleukin 2 (IL-2). Clones were expanded in 16mm wells (1INBRO), characterized for cytolytic function and specificity, and subcloned. The CTL clone with the highest killing efficiency, Fl, was chosen for biophysical study.

Assays for cytotoxicity were performed in triplicate in V-bottom 96-well microtitre plates (1INBRO). Target cells (JY), which had been labelled with ^3lCr (Na₂⁵¹CrO₄, New England Nuclear, Boston, MA), were plated at 1×10³ cells per well with varying numbers of effector cells (Fl) in a total volume of 150 μl of RPMI-1640 medium supplemented with 10% foetal calf serum, antibiotics, Hepes buffer, and 2-mercaptoethanol. The assay plates were centrifuged at 200 g for 5 min and incubated at 37°C. After a 4h incubation, the plates were again centrifuged, and 100-μl samples of cell-free supernatants were counted in a gamma counter to determine the amount of radioactivity released. Cytolytic activity was calculated as the percentage of specific release (SR) of ⁵¹Cr: SR = 100×[(E–C)/(T–C)], where E, C and T are the radioactivities (ctsmin⁻¹) released from the TC by incubation with the effector CTL cells, by incubation with medium alone, and by mixing with 5% Triton X-100 detergent, respectively.

About 0·5 ml of JY cell suspension (10³ cells ml⁻¹) was loaded in a small round chamber located on the sample stage of an inverted microscope. With the use of a ×100 objective (NA 1·25, oil immersion) and a ×20 eyepiece, the viewing field was displayed on a video monitor through a video camera and tape recorder system; final magnification on a television monitor was ×9000, and the system was calibrated using a microscale (50 μm × 2 μm). The video system was equipped with a video timer that recorded the date and time. About 30 min after loading the JY cells, approximately 200 F1 cells suspended in culture medium without IL-2 or lectins were loaded into the same chamber. The F1 and JY cells were each held at the tip of a micropipette by using a small aspiration pressure (approx. l·5mmH₂O). A pair of F1 and JY cells were brought to close proximity and aligned by manipulating the pipettes. After conjugation, the holding pipettes were removed and the morphological changes of a single pair of cells following a controlled onset of their conjugation were observed dynamically. These studies were performed at room temperature.

The biophysical properties of the individual cytotoxic T lymphocytes (Fl) and their target cells (JY) in the resting state (no conjugation) and their dynamic changes during the killing process were determined at room temperature in the same culture medium without IL-2 or lectins by using the micropipette aspiration technique (Sung et al. 1982, 1985; Chien et al. 1984). The internal radii of the micropipette, R_p, were 3·6–4·0μm for F1 and 4·8–5·2μm for JY. With the use of a hydraulic micromanipulator (Narishige Scientific Instrument Lab., Tokyo, Japan), the pipette tip was positioned near the surface of a cell. A negative pressure, SP, was applied to the micropipette as a step function to aspirate a small portion of the cell into the micropipette. The time course of deformation of the cell was continuously recorded on the video recorder. At a later time, sequential photographs were taken from the recorded video image during single-frame replay on the video monitor. The degree of cell deformation into the micropipette as a function of time was obtained with the aid of an EyeCom II video-digitizer (model 109 PT; Spatial Data System Inc., Goleta, CA) and a PDP 11/23 Mine Microcomputer (Digital Equipment Corp., Marlboro, MA). The time history of deformation typically showed an initial rapid phase followed by a slow creep, similar to the behaviour of peripheral blood leucocytes studied previously (Sung et al. 1982; Schmid-Schonbein et al. 1981). When small deformations are studied, as in this case, the response primarily reflects the intrinsic rheological properties of the cell, e.g. the influence of the cytoskeletal apparatus (Chien & Sung, 1984). In order to summarize and compare the results, the experimental curve on the time-dependent deformation of the cell was fitted by a viscoelastic model. In contrast to the red blood cells, which respond to micropipette aspiration by simple exponential expansion with time (Chien et al. 1978), the CTL and TC are similar to peripheral blood leucocytes (Schmid-Schonbein et al. 1981) in that they exhibit an initial rapid deformation essentially synchronous with the applied pressure prior to the timedependent slow creep. The modelling of this type of behaviour requires the three-element viscoelastic model shown schematically in Fig. 1. This model, which is described in detail in the Appendix, consists of two parallel arms, one of which is an elastic element K₁, the other contains a second elastic element K₂ and a viscous element μ in series. The elastic elements and K₁, which do not possess time-dependent characteristics by themselves, determine the degree of the initial rapid deformation of the cell, which varies with [(1/K₁) + (1/K₂)]. The viscous element μ provides the time-dependent behaviour; the combination of μ with the two elastic elements determines the rate of creep during the slow phase of deformation (Schmid-Schonbein et al. 1981), with the time constant given by μ/K₁+ K₁)/K₁K₂. The maximum deformation for long periods of time is proportional to 1/K₁.

The time-dependent deformation of an F1 cell in response to a step aspiration pressure is shown in the photomicrographs of Fig. 2. A plot of the time course of the deformation of F1 and JY cells is shown in Fig. 3, where the continuous lines represent the best fit of the theoretical model (Fig. 1) to the experimental data, and the coefficients K₁, K₂ and μ are listed in the figure. As indicated in Materials and methods, μ is a viscous element, which provides the time-dependent behaviour; K₁ and K₂ are two elastic elements with K₂ being in series with μ to form one arm of the three-element model (Fig. 1) and K₁ forming another arm parallel to K₂—μ. Compared to F1 cells, JY cells had higher values of K₁ and K₂, while the values of μ were not significantly different. These results indicate that JY cells have greater stiffness than the F1 cells, as reflected by the higher elastic coefficients, but similar viscosity. The experimental results obtained for other F1 cells (w = 23) and for JY cells (n = 32) were analysed in the same manner, and the degree of fitting by the model is similar to that shown in Fig. 3. The mean±s.E.M. values for the viscoelastic coefficients (K₁, K₂ and μ) of F1 and J Y cells are summarized in Table 1. The elastic coefficients (K₁, and K₂) of JY cells were approximately twice as large as those of F1 cells, while the cytoplasmic viscosity (μ) was nearly identical for J Y and F1 cells.

The viscoelastic constants for F1 cells in the resting state were similar to our results on other T lymphocytes (Sung et al. 1982). The viscoelastic constants of JY cells were similar to those for other lymphoblastoid cell lines (unpublished observations).

The CTL-TC interaction can be divided into five stages on the basis of changes of the biophysical properties of the JY target cell during the cytolytic process. The viscoelastic coefficients of JY cells during the killing process are summarized in Table 2. The morphological data are summarized in Table 3.

By manipulating the two holding micropipettes, an F1 cell and a J Y cell were brought into close proximity under microscopic observation. The pressure in the Fl-holding micropipette was released and the pipette was moved away from the cell. When this did not lead to the separation of the pair, conjugation was taken to have begun (Fig. 4A).

This first stage, which started from the beginning of Fl-JY conjugation, lasted about 10min. During this time the length of the physical conjugation increased to 2–3 μm. The biophysical coefficients K₂ and μ of the JY cell increased by 34 and 47%, respectively, as compared to the resting state, but K₁ remained essentially constant.

During this state, the F1 cell sent out pseudopods (Fig. 4B), but the F1 did not move around the JY surface. At the end of this period, the activity of F1 pseudopodia decreased (Fig. 4C). By focusing up and down, one could visualize the interdigitation of the plasma membranes of the two cells at the interface. The remainder of the JY cell surface began to exhibit numerous small convolutions (<1 μm in diameter and length).

From 10–60 min after cell conjugation, the JY cell regained its values of K₁, K₂ and μ found in the resting state (Table 2). At the end of this stage, the cell diameter of JY increased slightly, while that of F1 remained essentially unchanged (Fig. 5; Table 3).

Between 60 and 70 min after conjugation, the JY cell diameter increased by about 6–7%. The granules in the cytoplasm showed a marked decrease in movements and became closer to the nucleus. The K₁, K₂ and μ values of JY cells decreased by 26, 65 and 58%, respectively, as compared to the resting state.

During this stage, the JY cell sent out one small protrusion (budding) from its cytoplasmic membrane adjacent to the conjugated area (Fig. 4D), and granules became clustered around the nucleus.

During this stage, between 70 and 140 min, the JY cell volume increased by more than 200% of the original value, and the cell diameter increased by −50%. The JY cell membrane was apparently stretched taut, and no convolutions remained on the surface (Fig. 6). The K₁ and K₂ values increased to levels higher than those in the resting state. The μ value rose to that of the resting state (Table 1).

Bubbles were formed in the nucleus, with bubble diameters developing to 4–6 μm (Fig. 7). The development and rupture of each bubble lasted about 9 min. (The frequency of nuclear bubbling was approximately 0·11 bubble min⁻¹.) After a number of nuclear bubblings the nucleus was locally swollen, with a protrusion (Fig. 7); there was also a cytoplasmic protrusion through the plasma membrane. Granules were clustered near the nucleus and were absent in the swollen, dilated region.

In this final state (Figs 8 and 9), the JY cell swelled to become a smooth sphere. In response to aspiration pressure, the cell acted as a viscous material that lacked elastic properties. The JY cell was no longer viable as judged by Trypan Blue exclusion. By the end of the killing process (fifth stage), the JY cell volume was approximately the same as in the fourth stage. Continued observations for an additional 60 min showed no change in behaviour.

The present study was designed to elucidate the structure-function relationship in T-cell killing by studying the morphological and biophysical changes in this biological system. Morphological changes in single pairs of a cytotoxic T cell (Fl) and its target cell (JY) have been studied dynamically during the various stages of the killing process by using video microscopy. The stressstrain characteristics of the TC during the killing process have been investigated by using the micropipette test, and the viscoelastic coefficients of the cell computed with the aid of a theoretical model. The micropipette test involves small deformations and the results primarily reflect the behaviour of the submembrane cell content, especially the cytoskeletal apparatus (Sung et al. 1982; Chien & Sung, 1984; Schmid-Schonbein et al. 1981).

CTL–TC conjugation is the first step of the cell-mediated killing process. The present study, performed at room temperature, shows that Fl-JY conjugation persists throughout the entire process. The cells may still be conjugated, even after JY lysis. Earlier reports on experiments measuring ⁵¹Cr release from polyclonal CTL cultures indicated that the CTL was able to detach after 3-10 min of initial contact with the TC. The difference between these and the present results may be related to differences in experimental conditions, e.g. temperature, cell environment, etc. Although these studies were performed at room temperature, previous reports (Sung et al. 1982) indicate that the biophysical properties of cells do not show significant alterations with changes in temperature from 20–37°C. Furthermore, the F1 CTL clone used in these experiments was highly cytolytic. This enabled us to isolate single CTL–TC pairs and to observe killing at room temperature. Our microscopic observations on CTL–TC interactions agree with the previous reports (Sanderson & Glauert, 1979) that the conjugated area has membrane interdigitation. There may also be CTL cytoplasmic rearrangement. These changes probably result from the interaction between the surface receptor and their ligands and may allow the CTL to release or deliver cytolytic materials to the TC (Henkart, 1985; Podack & Dennert, 1983; Liu et al. 1986). We observed that one CTL could conjugate with two TCs at the same time, and it may be possible for one CTL to kill two TCs (either sequentially or concurrently) without detaching from either one.

Our microscopic observations indicate that the JY lysis phenomenon is different from simple osmotic lysis. Osmotic imbalance causes cells to swell or shrink in all directions, but during the beginning of the killing process the JY cell is swollen in a specific area, rather than in all directions. This local membrane protrusion could result from the insertion of channel complexes. A cytolytic pore-forming protein (perforin) has recently been isolated from the granules of H-2-restricted CTL, which can form structural and functional channels in the TC membrane (Podack & Dennert, 1983; Young et al. 1986b). The incorporation of this channel-forming protein into the membrane may cause disturbances in the integrity of the membrane structure and thus lead to local membrane weakness. Ultrastructural examination of these channels indicated that perforin proteins associated with lipids from tubular polymers similar to those in the ninth component of complement, C9 (Podack & Tschopp, 1982; Podack & Dennert, 1983; Young et al. 1986a).

The existence of perforin channel formation is further supported by our observations of only localized nuclear swelling. The nuclear bubbling occurs after conjugation but before rupture of the cell membrane; this is different from the swelling without bubbling seen in a hypo-osmotic medium where the water influx causes uniform stretching of the nuclear membrane. Our unpublished observations indicate that the nucleus of the lymphocyte has a very high viscosity (about 15 times higher than the cytoplasmic matrix) and acts almost like a purely elastic material. This elastic nucleus is packaged with histones and non-histone chromosomal proteins associated with nuclear DNA. In the killing process, this elastic, packed material is likely to be fragmented (DNA fragmentation), thus allowing the nuclear material to undergo bubbling and rupture. Recently, Liu et al. (1987) have shown that murine cytotoxic T lymphocytes contain, in addition to the cytotoxic pore-forming protein perforin, another cytolytic factor found in both the cytoplasm and granules, which causes DNA fragmentation in several target cells.

In the resting stage, the JY cell has a larger diameter (≃ 15 μm) than the F1 cell. The elastic coefficients of JY cells are twice as large as those of F1 cells, i.e. the initial deformation and the maximum deformation in response to a step aspiration pressure are smaller in JY than in F1 cells; the viscosity values are about equal. These results suggest the existence of differences in the cytoskeletal structure between JY and F1 cells. The dynamics at the cell surface, including the capping process (Loor et al. 1972; Sanderson & Glauert, 1979), lymphocyte mobility (Yefenof & Klein, 1976), and the interaction of the lymphocyte with a target cell, are all under the control of cytoskeletal structures in the cortical cytoplasm of the cell (Hepler & Palevitz, 1974; Olmsted et al. 1974; Pollard & Weihing, 1974). The biophysical behaviour of this cytoplasmic matrix is altered by several physicochemical treatments, such as changes in temperature, pH and osmolality (Sung et al. 1982), as well as by colchicine (Chien & Sung, 1984), which disrupts the organization of microtubules, and cytochalasin B, which disrupts the microfilaments.

On the basis of changes in the viscoelastic properties of the JY target cell, five distinct stages are observed in this complex killing process. In the first stage, the chemical recognition of CTL receptors and TC ligands leads to CTL-TC binding, and physical conjugation occurs. The area of conjugation increases with time during the first 15 min after the two cells have conjugated. The increase in the conjugated area may involve receptor migration, microvilli stretching and cytoskeletal rearrangement. The viscoelastic properties determined during this stage indicate that K₂ and μ increase significantly, while k₁ remains essentially the same as in the resting state. Since colchicine treatment has been found to cause a decrease in the values of K₂ and μ, but not k₁ (Schmid-Schonbein et al. 1981), cell conjugation at this stage may possibly be associated with a change in the organization of microtubules, which become more resistant to a deforming force.

In the second stage, the CTL is still conjugated firmly with TC, and the conjugation area remains the same as in the first stage; however the values of K.₂ and μ decrease to those found in the resting stage. This suggests the relaxation of the microtubule organization of the TC involving a recovery from the stiff configuration seen in the first stage or the delivery of the lethal hit from CTL to TC. In this second stage increases slightly, which may reflect membrane digitation and local pH change, as suggested by the slight swelling of TC at the end of this stage.

In the third stage, the TC shows a marked increase in cell volume and dramatic decreases in k₁, k₁ and μ. In this stage, the cytoskeletal proteins are diluted, and the actin microfilaments are probably disrupted. The chemi-osmotic injury is manifested by the nuclear bubbling and rupture seen under the microscope. The duration of the third stage is relatively short, but the cell damage is severe.

In the early part of the fourth stage, nuclear bubbling and rupture is continuous. The volumes of the nucleus and the cell are doubled or tripled in comparison to the original volume. The convolutions and microvilli of cell membranes have been stretched taut and disappear. The dramatic increases in the values of k₁, k₁ and μ in this stage reflect the loss of excess surface area as the membrane is stretched taut.

In the fifth and last stage, in response to a constant aspiration pressure applied via the micropipette, the deformation of the TC continuously increases with time, and it is not possible to discover its mechanical properties by the micropipette technique (Fig. 8). In addition to cytoskeletal alterations, the membrane may also be damaged by the mechanical stretching due to the extreme expansion of cell volume and membrane semipermeability function. The cell behaves as a viscoplastic material that has lost its elastic properties, and the aspirated portion maintains a deformed shape even after its release from the pipette (Fig. 9). In this stage, the TC cell is probably completely damaged by the CTL.

The present investigation was performed on CTL-TC conjugated pairs, with specific emphasis on the viscoelastic properties of the TC during the killing process. The biophysical behaviour of cells is a functional manifestation of their biochemical composition and molecular organization. The process of TC lysis mediated by CTL involves complex steps. The elucidation of changes of the viscoelastic behaviour of the TC in relation to the biochemical events in the killing process may increase our understanding of the chemical-mechanical transduction process in cell killing by CTL. Additional investigations on the biophysical properties of CTL during this process, and the effects of different chemical agents and monoclonal antibodies would help to elucidate further the biochemical basis of the biophysical events in cell killing.

In this equation k₁ and k₂ represent two elastic coefficients measured in dyncm⁻², and μ is a viscous coefficient measured in dynscm⁻² (1 dyn = 10⁻⁵N). The dots on σ_ij and e_ij indicate the time derivatives of the stress and strain deviators, i.e. ∂ σ₁/∂_t and ∂e_ij/∂_t.

The equation must be solved subject to the following stress boundary conditions. Let A_g be the area of contact (a concentric ring) between the glass pipette and the JY cell, A_i the area inside the pipette, and A_o the remainder of the cell surface outside the pipette. A schematic drawing of the JY cell with a micropipette is shown in Fig. A1. Let the pressure acting on these three surfaces of the cell be P_g, P_t and P_o, respectively (Fig. A2). Then ΔP _i (t) = P_i(t) — P_o is the applied aspiration pressure as a function of time. P_o is constant and equal to the hydrostatic fluid pressure, which is close to atmospheric pressure. P_g is computed from the condition of equilibrium for the whole cell expressed and is dependent on A_g, which in turn depends on the wall thickness and diameter of the pipette (d_p).

In other words, there exists at any instant a linear relation between the displacement D_p(t) and the pressure amplitude ΔP. This assumption has been proved satisfactorily (Chien & Sung, 1984; Schmid-Schonbein et al. 1981; Sung et al. 1982).

The time constant (τ) of the JY cell deformation process can be calculated as:

The standard solid model is the simplest viscoelastic model that fits the experimental results. For small strains the experimental results agreed very well with those predicted from the proposed model.

This work was supported by US Public Health Service grants R23 CA-37955 and R01-CA38955 (K.-L.P.S.) and CA-13429 (S.J.B.) from the National Cancer Institute, and P01 HL 16851 (S.C.) from the National Heart, Lung and Blood Institute. The authors are grateful to Dr Patrick C. Kung and Dr Timothy A. Springer for their valuable advice and discussions, and to Dr Samuel C. Silverstein for his valuable critical review of this manuscript. We are indebted to Ms Huei-ming Tsai, Mr Michael Longo and Mr Harry Spanglet for their technical assistance, Mr Gerard Norwich for his photography, and Mrs Micheline Faublas for her secretarial and editorial assistance.