T-ce 11-mediated cytolysis: analysis of killer and target cell deformability

and deformation during conjugate formation

COLETTE FOA\ JEAN-LOUIS MEGE2, CHRISTIAN CAPO2, ANNE-MARIE BENOLIEL2,

JEAN-REMY GALINDO3 and PIERRE BONGRAND2*1 Jeitne Equipe CNRS', MP 031 353, 27 Bd Lei Rome, 13009, Marseille, France2Laboraloire d'lmmniiologie, Hopital de Sainte-Marguerite, BP 29, 13277 Marseille Cedex 09, France3INSERM U 119, 27 Bd Lei Rome, 13009 Marseille, France

* Author for correspondence

Summary

T-cell-mediated cytolysis is initiated by the for-mation of strong adhesions between killer andtarget cells. The present work was aimed atdetermining whether T lymphocytes might exertsome mechanical stress on targets during thebinding process.

Target S194 myeloma cells were thus conju-gated to cy to toxic T lymphocytes (CTLs) raisedby mixed lymphocyte culture or a cloned lymph-oid line that was no longer cytolytic (TG2OUA2).After incubation periods of various lengths, con-jugates were processed for electron microscopyand micrographs were digitized for computer-ized analysis: the cell surface curvature (at themicrometre level) and rugosity (at the submicro-metre level) were quantified in free and adhesion-involved regions. Also, the size of cell interactionareas and the distribution of intermembrane dis-tances were measured. Finally, TG2OUA2 andtarget cells were aspirated into glass micropip-ettes with calibrated pressure in order to assaytheir resistance to deformation by mechanicalforces. The following conclusions were sugges-ted.

(1) Formation of extensive contact zones (witha linear size of several micrometres) with tight

intermembrane adhesion (more than 30 % of themembrane contours in adhesive zones wereseparated by an apparent distance lower than500 A) was essentially completed within less thanone minute.

(2) CTLs or TG2OUA2 cells were more villousthan their targets, and they seemed to deform inadhesive zones in order to adapt to the targetcontour, rather than imposing some deformationon the target. This may be a general feature ofactively adherent cells.

(3) Some CTLs, but no TG2OUA2 cells,exhibited conspicuous protrusions extendingtowards the bound target. In this case, the targetcell but not the CTL displayed markedlyincreased roughness in the adhesion area.

(4) TG2OUA2 cells were more resistant tomechanical deformation than S194 target cells, inaccordance with previous reports suggesting thatthe membrane of CTLs was more resistant thanthat of target tumour cells.

Hence, CTLs might transiently impose mech-anical stress on the target membrane during thecourse of lethal-hit delivery.

Key words: cell adhesion, cell deformability, T-cell-mediated cytolysis, cell surface.

Introduction

Lymphocyte-mediated cytotoxicity is an importantimmunological process as well as an attractive modelfor studying cell interactions. Many authors havestudied the mechanism of target cell lysis by cytotoxicJournal of Cell Science 8Printed in Great Britain (

561-573 (1988)) The Company of Biologists Limited 1988

T lymphocytes (CTLs) or natural killer cells, andmuch recent evidence supports the involvement of asecretory process in the lytic event. Indeed, electron-microscopic studies revealed the presence of con-spicuous granules in cytotoxic but not other lymphoidcells (Timonen et al. 1981; Grossi et al. 1983; Henkart,

561

1985) and purified granules were lytic to several targetcells. Further, these granules induced on the targetmembrane a deposition of cylindrical pore-like struc-tures that had been previously detected on the surfaceof target cells pre-exposed to CTLs (Dennert &Poddack, 1983; Henkart et al. 1984) or natural killercells (Poddack & Dennert, 1983). These results were inline with earlier reports describing the deposition ofunidentified material (Bykovskaya et al. 1983) orlysosomal enzymes (David et al. 1979) on the interac-tion area between target and killer cells.

However, the secretion theory does not account forall experimental results on cell-mediated cytotoxicity.Indeed, it is not clear why cytotoxic T lymphocytes arenot damaged when they kill their target (Golstein,1974; Martz, 1976), whereas they are not resistant to T-cell-mediated cytolysis (Golstein, 1974). Also, severalelectron-microscopic studies revealed various morpho-logical peculiarities of the killer-target interaction area,such as erythrocyte deformation by killer lymphocytes(Biberfeld & Perlmann, 1975; Tonietti et al. 1970),formation of finger-like projections of the killer cellsapproaching the target nucleus (Sanderson & Glauert,1979; Rosen et al. 1981; Carpen et al. 1982), andconcentration of actin filaments in the killer cytoplas-mic region near the contact area (Ryser et al. 1982).Further, Grimm et al. (1979) provided electron-micro-scopic images that strongly suggested the occurrence ofa mechanical tearing of the target membrane by CTLs.Finally, it was shown that the mechanical strength oftarget-CTL adhesions was often higher than that ofthe target membrane (Bongrand & Golstein, 1983;Bongrande/ al. 1983).

In view of the above results, it appeared of interest toperform an extensive analysis of killer-target interac-tion areas with objective quantification procedures inorder to address the following questions. (1) Does thekiller cell exert any mechanical effort on the target afterbond formation? (2) Does the apposition between killerand target cell membranes exhibit any remarkablefeature as compared to other models of membrane-membrane adhesion?

Our work was guided by two recent advances in thefield of Cell Biology: First, several groups (Meiselmanet al. 1984) made systematic use of micromanipulationand micropipette aspiration techniques to study themechanical properties of nucleated cells. Peripheralblood leucocytes were thus found to behave as liquiddroplets surrounded by a tensile membrane (with atension of about 102 dyne cm"1 (Evans & Kukan,1984)). Measuring the deformation of erythrocytesbrought in contact with an adhesive surface permittedthe quantitative determination of the free energies ofadhesion between erythrocyte membranes bound bypolysaccharides (BuxbaumeJ al. 1982) or lectins (Sung

et al. 1985). Second, other authors described a quanti-tative method for studying biological adhesion byanalysing cell contours on electron micrographs andevaluating intermembrane distance distribution lawsand membrane folding (Mege et. al. 1986; Foa et al.1988).

It is thus possible to find a relationship between bulkcell deformations (detected by light-microscopicexamination of cells exposed to mechanical or adhesivestimuli) and smoothing of the cell surfaces (detectedwith electron microscopy). The underlying concept isthat the cell surface microvilli act as reserve cellmembrane permitting departure from a spherical shapewith an apparent increase in area (Enckson &Trinkaus, 1976).

This report describes a study made on the interac-tion between cytotoxic T lymphocytes and targettumour cells by combining the approaches describedabove. Individual target cells or CTLs were subjectedto micropipette aspiration for determination of theirresponse to mechanical forces. Conjugates were thenprepared and studied by electron microscopy for deter-mination of membrane folding in free and bound areasof attached cells. The results were used to estimate theaffinity between target cells and CTLs, and estimatethe possible mechanical strain exerted by interactingcells.

Materials and methods

CellsTarget cells (TG) were S194 murine myeloma cells of Balb/corigin (H-2d).

Two populations of cytotoxic cells were used: cytotoxic Tlymphocytes (CTLs) were raised by mixed lymphocyteculture of spleen cells from CS7B1/6 mice (H-2b) andirradiated stimulatory spleen cells (H-2cl). We also made useof the lymphoid cell line TG2OUA2 (Nabholz et al. 1978)that was kindly supplied by M. Nabholz (Lausanne, Switzer-land). This line was derived from a cloned CTL (B6, H-2b)and was first highly cytolytic against S194. However, it wasno longer cytolytic when our experiments were performed,although it retained full ability to bind target cells (Foa et al.1985).

Conjugate formationTarget cells (107) and 2X107 effector cells were suspended in2 ml Hepes-buffered Dulbecco's modified minimum essentialmedium (DMEM) at pH7-2. Cells were centrifuged {250 g,5 min) in conical plastic tubes, then incubated for variousperiods of time at 37 °C before being transferred into an ice-cold bath and processed for electron microscopy. Sampleswere resuspended and deposited in a haemocytometer formicroscopic determination of conjugate formation. In somecases (see Table 4), conjugate formation and glutaraldehydefixation were done at 37°C.

562 C. Foa et al.

Electron microscopyConjugated cells were fixed with glutaraldehyde (4%), thencut, and stained with uranyl acetate and lead citrate beforebeing observed in a Jeol (Japan) 100 C electron microscope.Discrimination between TGs and CTLs or TG2OUA2 cellswas easily achieved by morphological criteria (Foa et al.1985) and only C T L - T G or TG2OUA2-TG conjugateswere retained. All samples were photographed at the samemagnification (X5000) and the same final enlargement of15 000 was obtained on printed micrographs. About 150electron micrographs were studied.

Use of published electron micrographsIt appeared of interest to compare our micrographs withpublished data in order to assess the generality of ourconclusions. Whenever necessary, a pantograph was used toobtain images of about 15 000 times the final magnification forcomputerized analysis.

Analysis of electron micrographsOur procedure has been described (Mege et al. 1986; Foa etal. 1988). Briefly, cell contours were digitized with a Calcomp2000 digitizer (Calcomp, CA, USA) connected to a CBM 64computer (Commodore, PA, USA). The mean distancebetween consecutive stored coordinate pairs was 11 mm,corresponding to an actual distance of 0075ftm. Repeateddigitization of a test contour showed that the digitizationprocedure was reproducible within about 014 mm (corre-sponding to 0-01 j

L=\-2um, corresponding to » = 1 7 , was found areasonable compromise: too-low values of n would result inneglect of the contribution of medium and large asperities tothe cell membrane area excess. Too-high values would makeit impossible to analyse limited regions of cell contours (sinceL represents a minimum for the length of a cell contour thatcan be analysed).

Another parameter of simpler significance was the meancurvature, C; namely, the mean reciprocal radius of thereference circular line used to calculate parameter e. A localflattening of the cell surface is expected to result in a decreaseof the mean curvature, C.

Characterization of the tightness of cell adhesionThe tightness of adhesion between apposed membranes on aCTL-TG conjugate was quantified by calculating the fre-quency distribution of distances separating these membranes.This was performed as follows (Fig. 2): the distance d;between each point vV, of the target contour and the CTLcontour (M\...Mn) was calculated as the minimal distancebetween iV, and any point of segments M\Mi,MzM$.. .M,,-\M,,. The shadowed area in Fig. 2 is the set ofpoints separated from M\.. .Ms by a distance shorter than d.Distances d, were then divided into 25 intervals of 100 Awidth [0,100], [100,200],... [2400,2500] and the total lengthIk of the target contour falling in interval [(k — l)Xl00,6x100] (/=£6«25) was calculated as:

Characterization of individual contoursAn objective parameter called the 'relative length excess', e,was defined to account for the intuitive concept of membranerugosity, or folding (Mege et al. 1986). The basic idea wasthat each cell has more membrane than is required to encloseits volume. Since the surface area of a solid of fixed volume isminimal when this solid is spherical, the 'relative membraneexcess' of a cell may be defined as:

Cell surface area/surface area of a sphere of the same volume — 1.

The problem was to find an objective way of evaluating alocal cell membrane area excess on electron micrographs.The trick was to compare the length of a sample segment of acell contour with the length of a 'reference circle' that wasconstructed as shown in Fig. 1: The reference circular lineM\MN was chosen to enclose the same area with the straightsegment M\Mn as the cell contour M\M2- • -M,,. The relativelength excess, e, is:

e = L/Lo-l,

where L is the length of the cell contour (calculated asM\Mx+ • • .+M,,-[M,,, and Lo is the length oiM\Mn, and E isthe mean value of e calculated for a given cell contour.

The significance of the relative length excess was assessedby studying model contours (Mege et al. 1986). It wasconcluded that e is the relative excess of membrane areacontributed by asperities with a characteristic length com-prised between the average distance between consecutivepoints of digitized contours (i.e. 0-075um with the magnifi-cation used) and the selected length L. The value of

\

\

Fig. 1. Quantitative analysis of digitized cell contours.Scanning the actual cell contour (thin continuous line) onelectron micrographs yielded a digitized contour consistingof a set of straight segments (broken line, M\Mz- • .M).This digitized contour was divided in sequential testsegments of n points (Mi. . .M,,, il72.. .A/,,+ i,. • •) that werecompared to circular reference lines (thick continuous line)enclosing the same area as these digitized contours with thestraight line joining their extremities.

Killer T lymphocyte-target adhesion 563

where the summation is extended to all points A', such that dfalls in the interval [ (6- l )X 100,/eX 100]. The arbitrary limitof 2500 A apparent distance was chosen to discriminatebetween adherent and non-adherent zones, which proved tobe a reasonable cutoff value.

MicromanipulationOur procedure has been described (Mege et al. 1985).Briefly, cells were suspended in a medium of high viscosityand density made of 80 % lymphocyte separation medium(MSL, Eurobio), 10% Hepes-buffered RPMI medium(Gibco) and 10 % foetal calf serum (Gibco). A hanging dropstuck to a glass coverslip was examined with a Reichert Jung-Microstar microscope (this is a conventional microscope withthe peculiarity that the stage is fixed and objectives aremoving during focussing). Individual cells were aspiratedwith a glass micropipette of about 4jUm inside diameter,handled with a DeFonbrune micromanipulator (Beaudouin,Paris) connected to a glass U tube and a 30 ml syringethrough a three-way stopcock. The cell was subjected tocontrolled aspiration (between —5 and —25 cm water) andthe length of the obtained protrusion was measured after 15-swait, using a eyepiece micrometer (Fig. 3). The results wereused to estimate the cell surface tension for a given defor-mation as explained in Appendix I.

Results

Significance of contour analysis

A sample photograph representing a C T L - T G conju-gate is shown in Fig. 4, together with digitized con-tours. The mean relative length excess, E, was calcu-lated using different values for the length of the testsegment (see Materials and methods for definition) andthe results are shown in Table 1. These results areconsistent with the view that the parameter E rep-resents the contribution to the total membrane area ofcell surface asperities with a characteristic length oibetween about 0-075 im (i.e. the average distancebetween consecutive digitized points) and the length oitest segments (Foa et al. 1988). As shown in Table 1,

Fig. 2. Intermembrane distance distribution. The lengthof the digitized contour A'i.. .A',, of a bound cell that wasseparated from the contour M\.. .M,, of another cell by adistance less than d was calculated by determining whichpoints A'I, were in the shaded zone and ascribing to each ofthese points i\\ the length (A^_iAr/,+A'^VA+1)/2. This wasperformed for 25 different values of d ranging between 100and 2500 A.

the asperities of the CTL-free region were essentially'microvilli' with a size between about 0-3 (im and 1 ^m,since E did not increase when the test segment lengthwas increased from 1-2/im to 6-2 jum. On the contrary,the bound area displayed larger 'macrovilli', whichwere responsible for the increase in E when the lengthof test segments was increased to several micrometres.

Typical micrographs representing fresh C T L - T Gand TG2OUA2-TG conjugates are shown in Figs 5and 6, respectively. The contact lengths depend onboth the actual contact area and the level of the section.Therefore, only mean values can be compared. Theapparent contact zone was defined as the region wherethe distance between killer and target cells was lowerthan 2500 A, corresponding approximately to the con-tact area visualized by optical microscopy. The close-contact zone was defined as the region where theapparent membrane distance between the two conju-gated cells was less than 500 A, corresponding to anactual distance of about 250 A, as shown in AppendixII.

Fig. 3. Cell aspiration into a micropipette. A S194 (target)cell was sucked into a glass micropipette of 3 fim internaldiameter with a negative pressure of —2cm water. Thelength of the obtained protrusion is 7 fim.

564 C. Foa et al.

i i |

TC

Fig. 4. Significance of contour analysis. A target cell (TG) is bound to a cytotoxic T lymphocyte (CTL). The electronmicrograph (A) is shown together with digitized contours (B). Free (F) and bound (Bo) segments of both cell contours (B).Free (F) and bound (Bo) segments of both cell contours were analysed. Bar, 1/Jm.

Table 1. Significance of the relative length excess

Contour length (ftm)E (0-3\vm test s)E (1-2/mi test s)E (6-2,um test s)

Target

Freecontour

3-90-0060-092

cell

Boundcontour

14-90-0140-1760-513

Freecontour

16-80-0140-1860-162

Killer cell

Boundcontour

15-400160-2030-691

The four cell contours shown in Fig. 4 were digitized and divided into overlapping segments of 0-3/im (5 points), 1-2/mi (17 points) or6-2/(m (82 points) length for determination of the mean relative length excess (E) corresponding to each test segment (test s) length.

Killer T lymphocyte-target adhesion 565

5A

B

Fig. 5. Interacting zones of fresh CTLs and S194 target conjugated cells. Both types of cell are easily recognizable becauseof their general aspect. The CTLs are generally small with a round nucleus containing dense euchromatin. Depending onthe section, the cytoplasm exhibits a few granules. Targets are larger and their nuclei display dispersed heterochromatin.The two photographs show various aspects of the interacting areas, the size of which depends on the section. The contactlasted 1 min (A) or 5 min (B). X9300.

566 C. Foa et al.

^ 7

B

Fig. 6. Interaction areas between cultured TG2OUA2 and S194 cells. TG2OUA2 cells are characterized by a cytoplasmfilled with numerous heterogeneous granules. Most of them contain a granular material and a few display a densehomogeneous core (arrowhead). We only retained the conjugates where TG2OUA2 cells displayed these granules andtargets were free of these granules. Contact lasted 1 min (A) or 5 min (B). X9000.

Killer T lymphocyte-target adhesion 567

Table 2. Kinetics of conjugate formation

% Conjugated cellsCell combination

Contact duration(min) TG2OUA2/S194 CTL/S194

15

IS

30

60

16

26

65

SO

33

68SO4538

TG2OUA2 or CTLs and target cells were mixed andcentrifuged, then incubated for various periods of time beforegentle resuspension and processing for electron microscopy.Samples were deposited into a haemocytometer for opticaldetermination of the percentage of bound cells. At least 200 cellswere counted in each sample.

Time dependence of interaction parametersTarget cells and CTLs or TG2OUA2 cells were fixedafter a period of contact ranging between 1 min and60 min. Samples were gently resuspended and exam-ined microscopically for determination of the percent-age of conjugated cells (Table 2). Other samples wereprocessed for electron microscopy and quantitativeanalysis.

The time dependence of the apparent length ofbound contours and the percentage of contours withintermembrane distance shorter than 500 A are shownin Fig. 7. Clearly, no significant evolution of bindingparameters was found, suggesting that bond formationwas completed within a few minutes. Therefore, itappeared reasonable to pool the results correspondingto different incubation periods in order to increase thesignificance of the experimental findings.

I0-6

0-4

0-2-

25 50Time (min)

u 25 50Time (min)

Fig. 7. Kinetics of contact formation. S194 target cells andCTLs ( ) or TG2OUA2 cells ( ) were incubated forvarious amounts of time before processing for electronmicroscopy. A. The length of apparent contact(corresponding to an intermembrane distance less than2500 A) was measured on random sections. Eachexperiment was repeated twice and a total of 173 electronmicrographs were examined. Each point representsbetween 7 and 26 determinations. Vertical bar length is2XS.E. B. The fraction of bound contour lengthcorresponding to an apparent intermembrane distance lessthan 500 A (this was called tight contact) was calculated onthe same photographs.

Study of cell surfaces

Electron micrographs representing conjugates betweenCTLs and S194 cells or TG2OUA2 and S194 cellswere used to evaluate membrane curvature and relativelength excess in free regions or regions involved inadhesion. The results are summarized in Table 3.

First, TG2OUA2 and CTLs appeared more villousthan TGs, since their relative length excess was signifi-cantly higher than that of TGs in free regions(TG2OUA2-TG and C T L - T G conjugates) and inbound regions (TG2OUA2-TG conjugates), at the0-01 confidence level according to Student's /-test.

Second, no significant difference was found betweenfree and bound regions of conjugated CTLs and TGs(see the first two columns of Table 3).

Third, marked differences were found between thefree and adherent regions of bound TG2OUA2 andS194 cells: indeed, both the relative length excess andthe curvature of the TG2OUA2 cell surface were lowerin adhesive regions (P<0-05 according to Student's /test). Also, the curvature of the target cell was signifi-cantly lower in the adhesion areas.

The overall conclusion suggested by these results isthat the target surface was minimally affected byconjugate formation, suggesting that TG2OUA2 orCTLs managed to fit to the target surfaces rather thanto deform them.

A possible problem with our results is that cellfixation was done at 4°C; therefore, rapid cell chillingmight alter the plasma membrane morphology. Inorder to address this point, conjugates were madebetween CTLs and TGs at 4°C or 37°C, and were thenfixed at the same temperature. As shown in Table 4,when cells were processed at 37 °C, CTLs appeared lessvillous in contact regions than in binding areas(P<0-05), which was consistent with the above con-clusion.

568 C. Foa et al.

It seemed of interest to apply our analytic procedureto electron micrographs obtained by different authorswith various experimental models. Some examples areshown in Table 5. Although wide differences werefound between different cell types, target cells seemedless villous in bound areas. Interestingly, CTLs alsoappeared smoother in the contact areas, except on thephotographs published by Sanderson & Glauert(1979), where the emphasis was on projections of

cytotoxic T cells towards the target. Hence, we re-examined all our photographs for the presence ofconspicuous protrusions of TG2OUA2 cells or CTLsin adhesion areas, as exemplified in Fig. 4.Interestingly, no such protrusion was found on the 55conjugates made with the TG2OUA2 line, which wasno longer cytolytic, whereas six typical images werefound on a series of 89 random conjugates made withCTLs. The roughness parameters of the free and

Table 3. Comparison of free and bound cell suifaces

CTL killer S194 target TG2OUA2 SI94 target

Free regionE0(111111-')

Bound region/ • :

C(mm""')

0-15 (±0-018)571 (±41)

014 (±0011)479 (±45)

0-07 (±0-017)369 (±29)

0-07 (±0-009)379 (±33)

0-19 (±0020)630 (±60)

011 (±0014)420 (±44)

0-12 (±0-016)554 (±49)

010 (±0019)369 (±35)

Conjugates made between CTLs and S194 cells (first two columns) or TG2OUA2 and S194 cells (last two columns) were analysed fordetermination of the mean relative excess (/_;') and curvature (C) on free cell contours or regions involved in killer to target binding. Eachresult shown is a mean of 72-87 separate results (standard error is given in parenthesis).

Table 4. Effect of temperature on the morphological features of CTL—target cell interaction

Mean relative length excess (/'")Binding and fixation at 4°C Binding and fixation at 37°C

CTL killer S194 target CTL killer S194 target

Free region

Bound region

0-09 ±0-03

(9)0-09 ± 0 0 5

(8)

0-04 ± 0 0 1

(11)0-07 ±0-02

(10)

0-13 ±003(?)

0-04 ±001(6)

0-06 ±0-02(7)

0-04 ±0-02(5)

Cytotoxic T lymphocytes were raised by mixed lymphocyte culture in spleen cells of C57BL/6 origin and conjugated with specific S194target cells at 4°C or 37°C; they were then fixed with glutaraldehvde at the same temperature and examined by electron microscopy for aquantitative analysis of the roughness of the free and adherent membrane region of interacting cells. Means are shown ± standard error(number of experiments is given in parenthesis).

Table 5. Analysis of published micrographs of conjugates made between target and killer cells

Mean relative length excess (/'.')Target cell Killer cell

Free area Bound area Bound area Target/killer Reference

0-56 ±0-04

0-35 ±0-05

014 ± 0 0 10-42 ±0-020-05 ±0-0020-51 ±0-06

0-23 ±0-020-24 ± 0-020-17 ±0-030-21 ±0-010-19 ±0-030-12 ± 0-01005 ± 0-002006 ±0-0050-10 ±0001

0-08 ±0-01

0-38 ±0-010-29 ± 0-05

0-07 ±0-0030-12± 0 0 10-10 ± 0-01

0-45 ±0-040-34 ±0-030-30 ±0-030-26 ± 0 0 10-16±0-030-05 ±0-0020-03 ±0-0020-08 ±0-01009 ±0-02

P815/C57BL10P815/C57BL10P815/C57BL10EL4/Balb/cEL4/Balb/cEL4/Balb/cEL4/Balb/cP815/C57BLNatural killer/K562

Sanderson & Glauert (1979), fig. 1Sanderson & Glauert (1979), fig. 9Sanderson & Glauert (1979), fig. 4Kalina & Berke (1976), fig. 2AKalina & Berkc (1976), fig. 2BGrimm & Bonavida (1979), fig. 3AGrimm & Bonavida (1979), fig. 6AKorene/ at. (1973), fig. 3Carpenef at. (1982), fig. 5

Published micrographs representing CTLs-targct cell conjugates were digitized for quantitative determination of the relative lengthexcess of free and bound zones of interacting cells.

Killer T lymphocyte—target adhesion 569

Table 6. Comparison of free and bound suifaces in conjugates betiveen CTLs and S194 cells with conspicuousinterdigitations in adhesion areas

Relative length excess (E)

Target ccIncubation time(min)

155

606060

MeanStandard deviation

Electron micrographs representing conjugates made with CTLs and S194 target cells were selected for presence of conspicuousprotrusions sent by the CTL towards the target. The mean relative length excesses of free and bound zones on each cell contour areshown.

Target

Free area

0-020-020-020-050-040-02

0-030-005

cell

Bound area

0080-180-170-330-090-23

0-180-038

Killer

Free area

0-350-120-380-000140-03

0-170-065

cell

Bound area

0-070-330-280-140-230-23

0-210 039

adherent areas of these conjugates are shown in Table6. In contrast with the findings shown in Table 3, thetarget area was less villous in the adhesion area(P<0'01) , whereas CTLs displayed similar (and rela-tively high) roughness in free and bound areas.

A final point concerning the possibility that killercells might inflict mechanical stress on target cells wasto determine which cell was stiffer in conjugates. Thisquestion was addressed in the following set of exper-iments.

Defonnability of TG2OUA2 and SJ94 cellsTG2OUA2 and S194 cells were subjected to micropip-ette aspiration with calibrated pressure and the lengthof the induced protrusions was recorded. The relativearea increase, /, induced by cell deformations was thencalculated together with the membrane tension, T, asdescribed in Appendix I.

As shown in Fig. 8, S194 cells exhibited more-extensive deformations than TG2OUA2 cells in re-sponse to mechanical stress.

Discussion

The aim of this work was to investigate the possibilitythat cytotoxic T lymphocytes might exert some mech-anical stress on bound target cells. For this purpose, weused an objective way of quantifying cell surfacerugosity at the submicrometre level (by calculating theexcess area contributed by microvilli) and curvature atthe micrometre level. These determinations were per-formed on electron micrographs and a high number ofrandom sections had to be examined to obtain signifi-cant information. The conclusions reached after ana-lysing more than 150 conjugates are as follows.

1

0-8J

0-6

0-4

0-2

0" 0-1 0-2 0-3 0-4 0-5

Fig. 8. Deformability of TG2OUA2 and S194 cells. Cellswere subjected to calibrated aspiration (—2, —5 and— 10 cm water) with a glass micropipette of 3 [im internaldiameter. The length of the induced protrusion wasmeasured after 15-s equilibration, which permittedquantitative determination of apparent cell area increase (/)and membrane tension (T) as shown in Appendix I. Eachpoint represents a mean of 12-17 separate measurementsobtained with different cells subjected to the samepressure. ( ) TG2OUA2; ( ) S194. Bar length is2XS.E.

(1) When CTLs or non-cytolytic TG2OUA2 cellswere coincubated with targets for 1-60 min, theyestablished extensive contact areas (with linear dimen-sions of the order of several micrometres; see Fig. 7)with fairly tight interactions, and this process seemedto be completed fairly rapidly, since both the size ofadhesive zones (Fig. 7A) and the tightness of interac-tion (Fig. 7B) were maximal after incubation for 1 min.

(2) The establishment of these contacts did notresult in any measurable stretching or deformation ofthe target cell microvilli when CTLs raised by mixedlymphocyte culture were used as killer cells. Further,

570 C. Foa et al.

when TG2OUA2 cells were used (or CTLs whenfixation was performed at 37°C), they exhibited sub-stantial (30-40%) and significant (P<0-01) decreasesin their area excess and surface curvature, whereastarget cells displayed similar decrease in curvaturewithout any significant decrease in their excess area.The simplest interpretation of the results displayed inTable 3 might be that the killer cells tended to fit theirsurface to the target membrane, in order to generate anextensive contact area. Indeed, it may be suggestedthat adaptation to an adjacent surface is a commonfunction of cells engaged in active adhesion. Obviously,such adaptation is an absolute requirement for thebinding of rigid surfaces such as foreign particles orsurrounding tissues. This was recently reported in astudy of the interaction between macrophages anderythrocytes (Mege et al. 1987).

(3) Whereas the results mentioned above are prob-ably relevant to adhesion rather than cytolysis, onlyactively cytolytic CTLs, not the non-cytolyticTG2OUA2 line, exhibited large protrusions indentingthe target surface, as reported by Sanderson & Glauert(1979). On the corresponding micrographs, boundareas were much more villous than the free regions ofthe surface of the target cells, in contrast to that of theCTLs. This finding is consistent with the possibilitythat the CTLs might exert some mechanical stress onthe membrane, with a traction and concomitantsmoothing of the unbound area according to the'contractile carpet' model described by Evans andKukan (1984). It is not surprising that these con-clusions are not apparent in Table 3, since the inter-digitating areas amounted to less than a small percent-age of the total length of the analysed contours. Thislow frequency of such events may be due to theirtransient nature as well as to a fairly poor cytolyticactivity of splenic CTLs as compared to other CTLpopulations.

(4) The above concept is consistent with furtherstudies done by the micropipette aspiration techniques.Indeed, TG2OUA2 cells were more resistant to mech-anical stress than S194 cells (Fig. 8).

It must be pointed out that the mechanical analysiswas not performed on CTLs, owing to the heterogen-eity of this population. However, it may be recalledthat former studies of C T L - T G conjugates (Bongrand& Golstein, 1983; Bongrand et al. 1983) suggested thatthe CTL membrane was more resistant than that of theTG, since the latter could be disrupted when CTLsand TGs were separated by hydrodynamic forces.

Hence, our results are consistent with the possibilitythat CTLs might exert some local stress on the targetmembrane, which might play an adjuvant role in thelytic process and be related to the transient immunityof the killer cell to lysis that must occur during lethalhit delivery (Golstein, 1974). However, more work is

Fig. 9. Geometrical model for micropipette-induced celldeformations. A. Cell in suspension; B, deformed cell.L. Length of protrusion; p, atmospheric pressure.

required before we can fully interpret our data, and thepresent study emphasize our lack of knowledge of themechanisms that govern cell shape at the micrometreand submicrometre levels. For example, it is notknown whether membrane folding is the mere conse-quence of the presence of contractile elements belowthe plasma membranes or whether microvilli areendowed with intrinsic mechanical resistance due tothe presence of axial microfilament bundles (Loor,1977). It is planned to address this question bystudying the deformations of lymphocytes subjected tocentrifugal forces at both the micrometre and sub-micrometre levels. This problem may be of importancein view of the probable existence of a close relationshipbetween cell surface shape and behaviour (Folkman &Moscona, 1978; Haeuptle et al. 1983).

Appendix I

Derivation of effective cell membrane tension frommicropipette aspiration data

Cells were modelled as liquid droplets of constantvolume V, surrounded by a tensile membrane ofvariable area A and tension T, as proposed by Evans &Kukan (1984). Ao is the initial area (corresponding to asphere of radius Ro; Fig. 9A). Sucking a cell into amicropipette of inner radius a with negative pressure— A/> generates a protrusion of length L (Fig. 9B) withconcomitant area increase from / lo to A. The relativearea increase / is:

I=(A-AO)/AO (Al)

and the membrane tension 7"(I) corresponding to thisarea increase may be easily derived using the Laplacelaw:

T=Ap/(2(\/a-\/R)), (A2)

where R is the radius of the spherical part of the cell(see Fig. 9B). R and A were obtained using thefollowing elementary geometrical formulae:

V = AnRl/Z = JIR3(8 + 9cos6» - cos30)/

12 + naz{L -a)+ 2na2/3 (A3)

P- Ap

Killer T lymphocyte—target adhesion 571

(this is the volume conservation condition).

6= sin'1 (a/R) (A4)

Assuming that 6 was small, equation (A3) yields:

R = (Rl + a31A - 3a2L/4)*. (A6)

Finally, the surface area of the deformed cell is:

A = 2JIR2( 1 + cos0) + lna{L - a) + 2naz, (A7)

since a, Ro and Ap are known, R was derived usingequation (A6) and T was then obtained using equation(A2). Finally, / was calculated using equations (Al)and (A7).

Appendix II

Relationship betzveen actual and apparentintennembrane distance measured on planar cellsections

Let us consider an elementary double-membrane patchin the area of interaction between two cells. This maybe modelled as two parallel circular discs of radius aseparated by distance d. We assume that d is muchsmaller than a. The apparent intermembrane distancemeasured on a section P is:

dt\ = d/cosd, (A8)

where 6 is the angle between P and the disc axis.Further, the probability of a plane of random locationand fixed orientation containing the considered mem-brane patch is proportional to 2a-cos6. Hence, themean value of the apparent intermembrane gap dA

measured on electron micrographs is:

<dA>=\ {2acosd)dQ/\ (2acosd)dQ, (A9)

where dQ. = 2jtsin&/0 is an elementary solid angle.Finally, we obtain:

<dA> = Id. (A10)

Hence, the actual intermembrane distance isabout half the mean apparent intermembrane dis-tance evaluated on electron micrographs (assumingthat the sample processing for electron microscopydoes not alter this distance).

This work was supported by a grant from the INSERM(CRE no. 862005). We thank Dr M. Nabholz for giving usthe TG2OUA2 CTL line.

References

BIBERFELD, P. & PERLMANN, P. (1970). Morphologicalobservations on the cytotoxicity of human blood

lymphocytes for antibody-coated chicken erythrocytes.Expl Cell Res. 62, 433-440.

BONGRAND, P. & GOLSTEIN, P. (1983). Reproducibledissociation of cellular aggregates with a wide range ofcalibrated shear forces: application to cytolyticlymphocyte-target cell conjugates. .7- ininiuii. Metlt. 58,209-224.

BONGRAND, P., PIERRES, M. & GOLSTEIN, P. (1983). T-cell

mediated cytolysis: on the strength of effector-target cellinteraction. Eur.J. Immiiii. 13, 424-429.

BUXBAUM, K., EVANS, E. & BROOKS, D. E. (1982).

Quantitation of surface affinities of red blood cells indextran solutions and plasma. Biochemistry 21,3235-3239.

BYKOVSKAYA, S. N., RYTENKO, A. N., RAUSHCHENBACH,

M. O. & BYKOVSKY, A. F. (1978). Ultrastructuralalteration of cytolytic T lymphocytes following theirinteraction with target cells. III. Plasmalemma"membranosomes". Cell hiiiinin. 42, 197-207.

CARPEN, O., VIRTANEN, I. & SAKSELA, I. (1982).

Ultrastructure of human natural killer cells: nature ofthe cytolytic contacts in relation to cellular secretion. J.Immun. 128, 1159-1162.

DAVID, A., BERNARD, J., THIERNESSE, N., NICOLAS, G.,

CERROTTINI, J. C. & ZAGURY, D. (1979). Le processusd'exocytose lysosomale localisee est-il rcsponsable deFaction cytolytique des lymphocyte T tueurs? C. r. hebd.Seanc. Acad. Set., Paris, ser D 288, 442-444.

DENNERT, G. & PODDACK, E. (1983). Cytolysis by H2-specific T killer cells. Assembly of tubular complexes ontarget membranes. J . exp. Med. 157, 1483-1495.

ERICKSON, E. A. & TRINKAUS, J. P. (1976). Microvilli andblebs as sources of reserve membrane during cellspreading. Expl Cell Res. 99, 375-384.

EVANS, E. & KUKAN, B. (1984). Passive material behaviourof granulocytes based on large deformation and recoveryafter deformation tests. Blood 64, 1028-1035.

FOA, C , BONGRAND, P., GALINDO, J. R. & GOLSTEIN, P.

(1985). Unexpected cell surface labeling in conjugatesbetween cytotoxic T lymphocytes and target cells. J.Histochem. Cytochem. 33, 647-657.

FOA, C , MEGE, J. L., CAPO, C , BENOLIEL, A. M.,

GALINDO, R. & BONGRAND, P. (1988). Electronmicroscopic study of the surface topography of isolatedand adherent cells. In Physical Basis of Cell—CellAdhesion (ed. P. Bongrand). Boca Raton: CRC Press (inpress).

FOLKMANN, J. & MOSCONA, A. (1978). Role of cell shapein growth control. Nature, Land. 273, 345-349.

GOLSTEIN, P. (1974). Sensitivity of cytotoxic T cells to Tcell-mediated cytotoxicity. Nature, Land. 252, 81-83.

GRIMM, E., PRICE, Z. & BONAVIDA, B. (1979). Studies on

the induction and expression of T cell-mediatedimmunity. III. Effector-target junctions and target cellmembrane disruption during cytolysis. Cell Immun. 46,77-99.

GROSSI, C. E., ZICCA, A., CADONI, A., MINGARI, M. C ,

MORETTA, A. & MORETTA, L. (1983). Ultrastructuralcharacteristics of human T cell clones with variouscytolytic activities. Eur.J. Immun. 13, 670-677.

572 C. Foa et al.

HAEUPTLE, M. T., SUARD, Y. L. M., BORGENMANN, E.,

RECCIO, H., RACINE, L. & KRAEHENBUL, J. P. (1983).

Effect of cell shape change on the function anddifferentiation of rabbit mammary cells in culture. J. CellBiol. 96, 1425-1434.

HENKART, P. A. (1985). Mechanism of lymphocyte-mediated cytotoxicity. A. Rev. Immun. 3, 31-58.

HENKART, P. A., MILLARD, P. J., REYNOLDS, C. W. &

I-IENKART, M. P. (1984). Cytolytic activity of purifiedcytoplasmic granules from cytotoxic rat LGL tumors, jf.exp. Med. 160, 75-93.

KALINA, M. & BERKE, G. (1976). Contact regions of

cytotoxic T lymphocyte-target cell conjugates. CellImmun. 25, 41-51.

KOREN, II. S., Ax, W. & FREUND-MOELBERT, E. (1973).

Morphological observations on the contact induced lysisof target cells. Eur.J. Immun. 3, 32-37.

LOOR, F. (1977). Structure and dynamics of thelymphocyte surface. In B and T Cells in ImmuneRecognition (ed. F. Loor & G. E. Roelants), pp.153-190. New York: J. Wiley.

MARTZ, E. (1976). Multiple target cell killing by thecytolytic T lymphocyte and the mechanism ofcytotoxicity. Transplant!! 2A, 5—11.

MEGE, J. L., CAPO, C , BENOLIEL, A. M. & BONGRAND, P.

(1987). Use of cell contour analysis to evaluate theaffinity between macrophages and glutaraldehyde-treatederythrocytes. Biophys.J. 52k, 177-186.

MEGE, J. L., CAPO, C , BENOLIEL, A. M., FOA, C. &

BONGRAND, P. (1985). Study of cell deformability by asimple method. jf. immun. Meth. 82, 3-15.

MEGE, J. L., CAPO, C , BENOLIEL, A. M., FOA, C ,

GALINDO, R. & BONGRAND, P. (1986). Quantification of

cell surface roughness: a method for studying cellmechanical and adhesive properties. .7. theor. Biol. 119,147-160.

MEISELMAN, H. J., LICHTMAN, M. A. & LACELLE, P. L.

(1984). White cell mechanics. In Basic Science andClinical Aspects. New York: A. R. Liss.

NABHOLZ, M., ClANFRIGLIA, M., ACRITO, D. ,

CONZELMANN, A., HAAS, W., VON BOEHMER, H. ,

MACDONALD, H. R., POHLIT, H. & JOHNSON, J. P.

(1978). Cytolytically active murine T-cell hybrids.Nature, bond. 287, 437-439.

PODDACK, E. & DENNERT, G. (1983). Cell mediated

cytolysis: assembly of two types of tubules with putativecytolytic function by cloned natural killer cells. Nature,Land. 302, 442-446.

ROSEN, D., FISHELSON, Z. & BERKE, G. (1981). The role

of cytotoxic T lymphocyte projections in target cell lysis.Transplant!! Proc. XIII, 1073-1078.

RYSER, J. E., RUNGGER-BRANDLE, E., CHAPONNIER, C ,

GABBIANI, G. & VASSALLI, P. (1982). The area of

attachment of cytotoxic T lymphocytes to their targetcells shows high motility and polarization of actin, butnot myosin. J. Immun. 128, 1159-1162.

SANDERSON, C. J. & GLAUERT, A. M. (1979). The

mechanism of T-cell mediated cytotoxicity. VI. T-cellprojections and their role in target cell killing.Immunology 36, 119—129.

SUNG, L. A., RABAT, E. A. & CHIEN, S. (1985).

Interaction energies in lectin-induced erythrocyteaggregation. J. Cell Biol. 101, 652-659.

TIMONEN, T., ORTALDO, J. R. & HERBERMAN, R. B.

(1981). Characteristics of human large granularlymphocytes and relationship to natural killer and K.cells. J. exp. Med. 153, 569-582.

TONIETTI, G., PECCI, G., LIBERATORE, C , SALSANO, F. &

FONTANA, L. (1975). Infrastructure of lymphocyte-mediated cytotoxicity. I. Interaction betweenphytohemagglutinin-stimulated human lymphocytes andchicken erythrocytes. Br.J. Haemat. 30, 71-74.

(Received 30 April 1987 - Accepted, in revised fonn,4Januaiy 19SS)

Killer T lymphocyte—target adhesion 573