T cell-induced non-T cell cytotoxicity
Eur. J. Immunol. 1977. 7: 55-61 B.F. Mackler', Peggy A. ONeill' and M. Meistrich' Laboratory of Immunology, Dental Science Institute, The University of Texas Health Science Center at Houston' and Section of Experimental Radiotherapy, M.D. Anderson Hospital and Tumor Institute', Houston
55
T lymphocyte induction of non-T cell-mediated nonspecific cytotoxicity I. Introduction mechanisms* Mononuclear cells (MNC) from normal humans consistently failed t o give nonspecific cytotoxic responses. However, after removal of T cells by sheep erythrocyte (E) rosetting, the remaining non-RFC (rosette-forming cells) now gave significant nonspecific cytotoxic responses against both autologous and allogeneic target cells. Reconstitution experiments with T cell subpopulations failed t o suppress these nonspecific non-E-RFC-mediated cytotoxic responses. There was also no evidence to indicate the involvement of antibody in this nonspecific cytotoxicity. The cytotoxic cells were characterized as non-E-rosetting, non-phagocytic, and glass adherent lymphocytes; n o evidence of monocyte-macrophage participation was found. The inductive trigger of non-E-RFC-mediated cytotoxicity was found to be soluble factors released by T cells during E-rosette formation at 4 'C. Incubation of MNC with horse, marmoset and human erythrocytes under identical conditions failed t o trigger cytotoxicity. The incubation of quiescent MNC with E-rosetting supernatants (ERS) induced nonspecific cytotoxic responses equivalent t o those mediated by separated non-E-RFC. ERSactivated MNC destroyed both autologous and allogeneic target cells. The ERS supernatants themselves were not cytolytic. These findings suggested that cell separation procedures, and possibly in vivo events, which activate T cells may also induce non-T cell-mediated nonspecific cytotoxicity.
1. Introduction
Investigators utilizing various lymphocyte separation and purification procedures [ 1-31 have frequently observed nonspecific cytotoxicity with human cell populations. More recently, several laboratories [4-51 have reported, in addition t o immune T cell-mediated cytotoxicity, that non-T lymphocytes are capable of destroying tumor cells. Of particular interest is the occurrence of non-T cell nonspecific "spontaneous lymphocyte-mediated cytotoxicity" toward allogeneic and xenogeneic tumor cells [5]. These investigators postulated that a second cytotoxic lymphoid subpopulation distinct from T cells exists which mediates natural immune surveillance. Other workers [6-71 also described a non-T cell-mediated cytotoxicity response independent of antibody which involves complement (C3) receptor induction of lymphotoxin. More recently, the involvement of lymphotoxinlike substances has been suggested to mediate spontaneous cytotoxicity of non-T cells [8]. Given these observations, we sought to assess the nonspecific cytotoxic activities of normal human lymphoid subpopula[I 15221
* This work was supported by the National lnstitutesof Health, NIDR grants DE-04210 and DE-02232, General Research Support Grant DSI-29-74-01 and NCI grant CA-17364.
tions. The paper describes the cytotoxicity of mononuclear cells and subpopulations from normal healthy donors in direct microcytotoxicity assays using autologous and allogeneic target cells. In these studies, sheep erythrocyte (E)rosetting of T cells was found t o cause T cells to secrete mediators which induced non-T cell cytolytic responses.
2. Materials and methods 2.1. Separation of lymphoid subpopulations
Mononuclear cells (MNC) were separated from the blood of healthy donors by Ficoll(9 %)-Hypaque (33 %) gradient centrifugation (400 x g, 20 min, 25 "C) as previously described [6, 7,9]; recovery of MNC averaged 73 f 9.3 % SE. T lymphocytes were separated by rosetting with fresh sheep E in 2.5 % fetal calf serum (heat inactivated, E absorbed) for 1 h at 4 'C then centrifuged through Ficoll-Hypaque (400 x g, 20 min, 5 "C). In some experiments, high affinity E rosetting cells (ha-ERFC) were prepared by t h e Wybran method [ 101. Briefly, 1 O7 MNC/ml were incubated with a n equal volume of E-absorbed fetal calf s e p m for 1 h a t 37 'C then mixed with an equal volume of sheep E (4 x 1 07/ml) and sedimented at 350 x g for 5 min. The pelleted cells were gently resuspended, and rosetted cells were separated by Ficoll-Hypaque (400 x g, 20 min, 5 'C) gradient centrifugation.
Correspondence: Bruce F. Mackler, Immunology Lalpratory, Dental Science Institute, P.O.Box 20068, Houston, Texas 77025, USA
2.2. Macrophage depletion
Abbreviations: E: Erythrocyte EAC Sheep E coated with 19 S rabbit anti-E antibody and mouse complement components ERFC T cell rosettes E R S E-rosetting mixture supernatants haERFC High affinity T cell rosettes MEM: Eagle's minimum essential medium M N C Mononuclear cells non-ERFC Non E-rosetting cells (non-T cells) Smlg: Surface membrane-associated immunoglobulins FCS: Fetal calf serum
Non-ERFC remaining after removal of ERFC were depleted of phagocytic cells by the carbonyl iron method [l 11: 60-75 % of the original non-ERFC subpopulation was recovered after iron filing treatment. I n the recovered population, < 1 % phagocytic cells were evident as determined by 2 % neutral red dye uptake; phagocytic cells possessed red granules.
56
B.F. Mackler, P.A. ONeill and M. Meistrich
Eur. J. Immunol. 1977. 7:55-61
2.3. Lymphoid subpopulation markers
The use of automatic dispensing pipettes was found to increase reproducibility and is recommended.
2.3.1. Complement receptor-bearing cells (CRL) 2.4.2. Target cell lines
The percentage of CRL was quantitated using sheep E coated with 1 9 S rabbit anti-sheep E antibody (Cordis Laboratories, Miami, Florida) and components of fresh mouse complement, termed the EAC reagent. Lymphocytes and EAC were incubated a t 37 "C for 30 min with gentle agitation, and the percentage of EACRFC was quantitated. 2.3.2. Surface membrane immunoglobulins (SmIg)
Lymphocytes were preincubated before staining t o remove cytophilic antibody [ 121. Cells were suspended in a 1 % bovine serum albumin (BSA)-isotonic saline-azide buffer (4 "C), washed, then incubated 30 min at 4 "C with fluorescein-conjugated polyvalent goat anti-human immunoglobulin serum (Meloy Laboratories, Springfield, Va.). 2.3.3. Sheep erythrocyte rosette-forming cells (ERFC)
Cells ( 1 x 1 0 6 ) were mixed with an equal volume of fresh E (5 %) in RPMI 1640 medium (GIBCO, Grand Island, N.Y.) with 2.5 % fetal calf serum (FCS, heat inactivated, E absorbed). centrifuged (250 x g, 5 min at 4 "C), then incubated 1 h at 4 "C. The supernatant removed after incubation of MNC with sheep E was termed the E-rosetting supernatant (ERSup). The percentage of cells forming E rosettes was calculated after counting 200 nucleated cells. 2.3.4. Staining of monocytes
Monocytes were identified by acridine orange [ 131, nonspecific esterase [ 141, and peroxidase [ 151 cytochemical staining procedures.
Allogeneic target cells - human melanoma cell line SH-I [ 171, and lung fibroblast cell line WI-38 [ 181 - were cultured in Eagle's MEM plus 10 % fetal calf serum (heat inactivated), MEM vitamins, glutamine and antibiotics in a 5 70 COz in air atmosphere [6, 71. Autologous fibroblast cell lines were established from gingival specimens obtained during therapeutic periodontal surgery and were a gift from Dr. George Rose, The University of Texas Dental Branch. 2.5. Velocity sedimentation separation of non-ERFC
Non-ERFC were subfractionated by velocity sedimentation using a Beckman Elutriator rotor in a refrigerated 5-21 B centrifuge [19-211. The Elutriator system was flushed with 70 % ethanol and Hanks' balanced salt solution with 0.5 % BSA t o sterilize the system. All air bubbles were removed and the flow rate and rotor speed adjusted for the first fraction ( 1 1.3 f 0. I ml/min, 3000 rpm). Non-ERFC, depleted of ERFC, approximately 150 x 1 O6 - 50 x 1 O6 cells, were introduced into the chamber with constant flow t o collect the first fraction. The rotor speed was held at 3000 rpm while the flow rate was increased stepwise: 16.5 0.5 ml/min (fraction 2), 21.5 f 0.1 ml/min (fraction 3), 33.5 f 1.4 ml/ rnin (fraction 4); for fraction 5 the rotor was stopped and maximum flow rate was used t o remove the remaining cells from t h e chamber. The sedimentation rates were determined t o be: 5 2.4 mm/h/g(fraction 1); 2.4-3.5 mm/h/g(fraction 2); 3.5 -4.6 mm/h/g (fraction 3); 4 . 6 - 7 . 2 mm/h/g (fraction 4); and 2 7.2 mm/h/g (fraction 5). +_
3. Results 2.4. Cytotoxicity assay method 3.1. Cytotoxic activities by human lymphoid subpopulations
2.4.1. Microcytotoxicity assay
The lymphocyte cytotoxicity was assessed in a micro-tray method [ 161. Each well was seeded with approximately 200 target cells in 0.2 ml of Eagle's MEM plus 10 % fetal calf serum, MEM vitamins, 2 mM L-glutamine, penicillin and streptomycin using an automatic dispensing pipette (Schwarz/ Mann), then incubated overnight at 37 "C. At 24 h, lymphocytes were added t o each well in a 0.05ml volume using an automatic dispensing pipette and incubation continued for 48 h. The plates were then washed at t h e end of this second incubation, air dried and adherent target cells stained with Giemsa. Control target cell counts averaged 208.6 f 6.9 (SE) for allogeneic and 139.1 f 6.6 (SE) for autologous target cells. Trypan blue dye exclusion tests of non-adherent target cells showed that > 98 % of the cells in the supernatant were dead. Cytotoxicity was calculated by the formula: Avg. no. cells/well in test wells of cells
%Cytotoxicity = 1 -
+ lymphocytes
x 100
Avg. no. cells/well in control well of cells only
Averages were determined using 6 replicate wells for each test variable and statistical significance by Student's t-test.
Human lymphoid subpopulations separated by E rosetting were assayed for nonspecific cytotoxicity using allogeneic and autologous target cells. The results from multiple experiments using allogeneic melanoma (SH-I ) target cells in direct microcytotoxicity assays are shown in Figure 1. The MNC and ERFC were not cytotoxic even when tested at target cell: lymphocyte ratios of 1 : S O 0 and 1 :200, respectively. In contrast, the non-ERFC remaining after T cell removal had significant cytotoxic activity even a t ratios as low as 20: 1. The preparation of high affinity E rosettes (ha-ERFC) by a 5 min centrifugation step followed by a 30 min centrifugation on Ficoll-Hypaque t o separate lymphoid subpopulation induced similar levels of non-ERFC cytotoxicity (40.5 f 8.7 % S E at 1 :SO). Therefore, the ha-ERFC were actually in the presence of the non-ERFC for almost 3 5 min, which may account for the high activation of the non-ERFC by a relatively short high affinity 5 min rosetting period. In addition, extensive washing of non-ERFC, which presumably removed cytophilic antibodies, did not diminish non-ERFC-mediated cytotoxic activity (34.9 5.1 % SE at 1 :SO).
*
The contribution of tumor and histocompatibility antigens in this nonspecific cytotoxic phenomenon was assessed by using
T cell-induced non-T cell cytotoxicity
EM. J. Immunol. 1977. 7: 55-61
57
Table 1. Cytotoxicity of human lymphocyte populations toward autologous gingivalfibroblast target cells. The average number of target cellslwellare shown along their respective standard error (S.E.) and percentage (7%) cytotoxicity for each individual experiment. The average percent cytotoxicity of the pooled experiments is shown along with the SE. Exp.
Control
no.
Avg. no. cells/
well 1 2 3 4
f
SE
125.7 f 5 . 8 130.0 f 3.6 151.0 f 7.7 149.7 f 5.9
MNC (1 :100) Avg.no.celld well f S E
5% Tox.
123.3 f5.8 127.3 f 6.1 148.3f6.4 147.7 f 5.1
1.9 2.1 1.8 1.3 -
Average % cytotoxicity (Tox.) Among Expts. f SE
E-RFC (1: 100) Avg.no.alls/ % well f S E Tox. 122.9 f 8.1 127.9 f 2.1 149.8 f 6.7 146.9 f7.1
92.4 f 5.2 78.0 f 6.5 112.2 2.7 83.3 f 4.4
*
autologous gingival fibroblasts as target cells (Fig. 1 and Table 1). In these experiments, gingival fibroblasts cell lines were prepared from therapeutic gingival surgery specimens of 4 different patients, and these cells were used as target cells prior t o their 5th passage.
26.5 40.0 25.7 44.5 34.2 f 4 . 8
1.6 f 0.3
1.7 f 0.2
Figure 1 shows that experiments performed with autologous lymphocytes and target cells gave nonspecific cytotoxic responses consistent with t h e allogeneic studies. The individual autologous experiments are presented in Table 1 in terms of the actual target cell counts per sextruple replicates of each experimental variable and their respective percentage cytotoxicity. In terms of t h e actual cell numbers, gingival target cell fibroblasts in control cultures which lacked lymphocytes ranged between 125.7- IS1 with individual SE of 3.6- 7.7 %. The addition of autologous MNC failed t o cause target cell destruction; the percentage cytotoxicity in individual experiments ranged from 1.3 - 2.1 with a mean of 1.7 f 0.2 %. Similarly, autologous ERFC failed t o induce any appreciable cyt otoxicit y . However, autologous non-ER FC were cyt o t oxic four experiments ranging from 25*7-44*5 giving a mean in value of 34.2 k 4.8 % cytotoxicity.
2.2 1.6 0.8 1.9 -
Non-ERFC (1 :100) 7% well+fSE Tox.
Avg. no. cells/
50
t
?
30
2 Figure 2. The dose dependent cytotoxicity curve with increasing numbers of cytotoxic non-ERFC lymphocytes. Each point is the mean f SE of 4 experiments. The ratio of target cel1:lymphocytes is shown on the abscissa and percent cytotoxicity on the ordinate.
3.2. Reconstitutionof cytotoxic
with
cells
Since t h e removal of (high affinity a n d total) ERFC resulted in cytolytic activity by t h e remaining non-ERFC, efforts were made t o reverse this activation in T cell reconstitution experiments. In four experiments, t h e total ERFC or a subpopulation of ha-ERFC were recombined with non-ERFC at the same percentage found in MNC populations. The target cell: non-ERFC ratio was held constant (1 :20) while increasing numbers of either ERFC or ha-ERFC were added (Figure 3). The addition of these T cells neither significantly enhanced nor diminished the nonspecific cytotoxic responses of nonERFC.
Figure 1. Cytotoxicity of normal human lymphoid subpopulations for allogeneic and autologous target cells in direct miaocytotoxicity assays. Allogeneic human lung fibroblasts (Wl-38) or melanoma (SH-1) or autologous - human gingival fibroblast - target cells were used as
indicated. The results show as the mean k SE from multiple experiments (N) with allogeneic and autologous cell donors. The dose-dependent relationship of cytotoxic non-ERFC t o allogeneic target cells is shown in Figure 2. Nonspecific CYtotoxic activity was significant even at very low ratios Of target cells: lymphocytes (e.g., 1 : 1, 10: 1 and 20: 1).
j 3. Effect ~ of, reconstitution ~ with T cell subpopulations on non-ERFC-medhted cytotoxicity. NowERFC were reconstituted with increasing numbers of ERFC or ha-ERFC as indicated. (N) is the number of experiments performed.
~
Eur. J . Immunol. 1977. 7: 55-61
B.F. Mackler, P.A. ONeill and M. Meistrich
58
3.3. Characterization of cytotoxic non-ERFC effector cells 3.3.1. Glass adherence properties The cytotoxic non-ERFC effector cells were characterized and separated by glass adherence properties. In three experiments, the adherent cells represented 12.5 % o f the total nonERFC population and contained 5.9 f 4.2 % complement receptor-bearing cells (CRL), 22.2 f 1.6 % immunoglobulin staining cells (Smlg), 25.2 f 4.1 % neutral red and 34.4 f 6.7 % esterase positive cells. In contrast, the majority of cells was recovered in the nonadherent subpopulation (86.5 %) of which 37.2 k 1.4 7% were CRL, 32.7 f 1.2 % Smlg+, 16.8 f 3.6 % neutral red and 15.5 f 4.0 % esterase positive cells. The cytotoxic activities of these two cell fractions are shown in Table 2. Adherent and nonadherent cells were added t o target cells in numbers comparable t o their percentages in the non-ERFC population. The adherent cells gave cytotoxic responses that were equivalent t o the original non-ERFC population. The nonadherent cells gave much lower cytotoxic responses ( 1 5.9 f 6.1 %) which could be reduced even further by multiple sequential adherence of these cells to glass.
3.3.2. Phagocytic cells The role of phagocytic cells in the non-ERFC-mediated nonspecific cytotoxicity was assessed by removal of phagocytic cells using carbonyl iron. After removal of cells engulfing iron filings, the recovered 60-75 % of the original non-ERFC contained < 1 % phagocytic cells as determined by neutral red dye uptake. The phagocyte-depleted non-ERFC were tested for nonspecific cytotoxicity at target cel1:lymphocyte ratios equivalent t o the percentage of phagocytic and nonphagocytic cells in t h e original non-ERFC population (Table 2). The removal of phagocytic cells neither enhanced nor diminished the cytotoxic activity of the remaining non-ERFC. The separated phagocytic cells were found to have low cytotoxic
responses (23.2 f 5.5 %) which was presumed to represent the activity of macrophages activated by the iron filing removal method.
3.3. Velocity sedimentation separation of lymphocytes and monocy tes In order t o remove nonphagocytic monocytes and characterize further the cytotoxic effector cells, cytotoxic nonERFC were separated by velocity sedimentation into 5 fractions using a Beckman Elutriator rotor. This method separated the cells by volume with fraction 1 containing cells with the smallest volume and fraction 5 the largest volume. The total cell recovery of the original non-ERFC population applied t o the rotor ranged between 65-91 % (viability > 98 %) with t h e majority of cells being recovered in the 3rd and 4th fractions. Each fraction was assessed for nonspecific cytotoxic activity and for the presence of monocytes identifiable by acridine orange (AO), esterase (EST) and peroxidase (P)cytochemical staining (Figure 4). Fractions 1 and 2 were found t o have significant cytotoxic activities while containing rather negligible numbers of monocytes. The presence of increasing percentages of monocytes did not enhance the nonspecific cytotoxic activities of fractions 3 , 4 and 5 . The distribution of cytotoxic effector cells in t h e 5 fractions did not reflect monocyte distribution. These data suggested that monocytes were probably not the cytotoxic effector cells in non-ERFC populations.
Table 2. Characterization of cytotoxic non-ERFC effector cells by glass adherence and phagocytosis. The number o f experiments is designated by (N) and the target cells used were SH-1 Lymphoid su bpopula t ions
Target cell: lymphocyte ratio
c/;
(N)
Cytotoxicity f SII
M NC
1:lOO
(4)
1.2 f 0.8
Non-ERFC
1 : 200
(3) (4)
53.8 f 4.1 56.4 f 7.5 50.4 f 6.2
1:lOO 1:50
(4)
A. Ghs$ adhcrcncc Non-adherent nonI: R FC Adlicrcnt non-ERF('
15.9 f 6.1 13.7 f 5.5 51.x f 12.2 18.2 f 12.5
H. Iron filing trcatmcnt Non-phagocytic nonERFCd
I : 100 1:lOO
Plirgocyt ic non-ERFC
a) < 1 7; phagocytic cells.
1:50
(3) (3) (3)
70.4 f 14.1 66.7 f 6.6 48.0 f 3.7
1:50
( 3)
1:25
(3)
20.9 f 2.3 23.2 f 5.5
Figure 4. Velocity sedimentation separation of cytotoxic non-ERFC into subfractions. Fraction 1 contains cells with the smallest volume while fraction 5 has those with the largest volume. The perccnt cell recovery is the distribution of the total cells recovered in all five fractions.
3.4. The role of sheep E rosetting in the induction of nonERFC-mediated cytotoxicity Since t h e induction of non-ERFC-mediated cytotoxicity appeared t o occur after sheep E rosetting, the nature of this inductive mechanism was examined. To assess whether erythrocyte binding or the physical packing of cells by centrifugation triggered non-ERFC cytotoxicity, MNC ( I O7 cells) were mixed with similar concentrations ( 4 x 1 07)of either sheep, horse, marmoset or human erythrocytes in RPMl-I 640 plus 2.5 7% FCS (absorbed, heat inactivated) and gently packed by centrifugation (250 x g, 5 min at 4 "C). After 1 h incubation a t 4 OC, the number of ERFC was counted, and the nonERFC were separated by Ficoll-Hypaque density gradient cen-
T cell-induced non-T cell cytotoxicity
Eur. J. Immunol. 1977. 7: 55-61 trifugation. In several experiments, sheep E and MNC mixtures were incubated at 37 OC for 1 h. In Table 3, the mean results from multiple experiments are summarized. Cytotoxic non-ERFC were obtained only after sheep E rosette formation at 4 OC, but not at 37 OC. The MNC incubated with either horse, human or marmoset E formed negligible numbers of E rosettes, and the majority of cells was recovered in t h e non-ERFC population which was not activated. These findings suggested that FCS and the physical packing of MNC with erythrocytes were not sufficient to induce non-ERFC cytotoxicity. Similar sheep E-rosetting experiments performed in t h e absence of FCS eliminated any nonspecific involvement of FCS moieties. In these experiments, 43.0 f 9.6 % ERFC were obtained while the recovered non-ERFC had nonspecific cytotoxic activities of 38.6 f 6.4 % at 1 : 100. These results provided evidence suggesting that the actual E rosette formation was the inductive trigger for non-T cell-mediated nonspecific cytotoxicity. Table 3. MNC were rosetted with erythrocytes from various animals and incubated for 1 h. The number of ERFC was first determined, then the non-ERFC separated by Fimll-Hypaque gradients and nonERFC cytotoxicity assessed using SH-1 target cells. The number of experiments is designated by (N) and the mean shown f SE E
wurce -8)
Sheep Shcep Human Horw Marmoset
Incubation temperature (OC) 4 37 4 4 4
(N)
70 E rosetteforming cells
0 (11) (4) 45.7 f5.7 (3) 1.0 fO.5 0 (3) 3.3 f l . 9 (3) 0 (2)
3 Non-ERFC
cytotoxicity 13.7 f 4.7 60.3 f 13.6 7.0 f 1.6 5.3 f 3.1 3.9 f 3.9
0
a) Untreated MNC.
3.5. E-roset ting induction of non-ERFC-mediated cytotoxicit y by supernatant factors The E-rosetting experiments with various erythrocytes suggested that T cells induced non-ERFC-mediated cytotoxicity. To investigate this finding further, supernatant fluids from the E-rosetting mixture incubated at 4 OC for 1 h were assessed for their ability t o induce autologous MNC t o manifest nonspecific cytotoxic responses (Table 4). Control supernatants were prepared using similar concentrations of only sheep E and fetal calf serum incubated at 4 O C for 1 h. The quiescent MNC were incubated for 30 min in undiluted ER supernatant, washed twice with medium and plated in microcytotoxicity trays at 1 : 1 0 0 target cel1:lymphocyte ratios. I n repeated experiments, ER supernatants (ERSup) prepared with either home, human or marmoset E as well as control supernatant failed t o activate MNC. However, ERSup prepared with sheep E + MNC at 4 OC consistently induced autologous MNC (64.9 f 7.2 % at 1 : 100) as well as allogeneic MNC (59.4 f 8.1 % at 1 : 100) t o manifest cytotoxic responses. Although these ERSup induced non-T cell-mediated cytotoxicity, the ERSup did not contain any detectable cytolytic activity in itself. As controls, supernatants prepared with sheep
59
Table 4. E-rosetting supernatant (ERSup) induction of MNC cytotoxic activities. Supernatants from either E-rosetting mixtures or control supernatants (Sup) were incubated with autologous MNC, washed, then assessed for cytotoxicity. The number of experiments performed with cells from different donors is designated by (N) and the mean value is shown f SE
Effector cells
ERSup
'?Cytotouicit):
(N)
*
13.7 % 4.7 +1.3Xf0.21) 5.4 Q f 4 . 8
MNC
-
SUP SUP Human Horse Marmoset Sheep Sheep
-
-I- MNC 4- MNC 4- MNC
+
0.0
7.8 Y f 1.6 2 . 2 5 f 1.5 + 2.0 X f 0.61) 64.9 % f 7.2
MNC
-
-t MNC
a) The + 1.3 X and + 2.0 X indicate enhanced cell survival rather than % cytotoxicity.
E + MNC a t 37 OC were found t o lack the ability t o induce non-T cell-mediated cytotoxicity, an observation consistent with their inability t o form E rosettes.
T cell production of E-rosetting factors was also assessed after t h e removal of adherent erythrocytes. The experimental design is shown in Figure 5 along with the data from triplicate experiments. MNC were first rosetted with sheep E and an Erosetting supernatant (1 st ERSup) was obtained. The ERFC and non-ERFC were then separated by Ficoll-Hypaque gradient centrifugation. Consistent with previous findings, t h e 1 st ERSup induced autologous MNC t o become cytotoxic (36.1 % f 5.9) a t levels comparable t o the non-ERFC (32.1 % f 6.2). The 1st ERSup alone was not cytolytic but actually appeared t o enhance target cell growth (1.8 times). The ERFC sub population was treated with NH4Cl a t 37 OC t o lyze adherent sheep E which were removed by multiple washes with medium at 3 7 OC; t h e viability of the recovered T cells was > 9 5 %. The T cell cultures were incubated under similar conditions for 1 h, a n d a T cell supernatant was obtained. The T cell supernatants were neither directly cytolytic nor able t o induce MNC-mediated cytotoxicity. These results suggested E a p r i r e n t r l Design
1 Cytotoaicity t S.E.
Effector Cells
Suuernrtmt
-
1 s t ERSup
+1.u: 0.0
1 s t ERSUP
36.11 % 5.9
mc 5.n 1 s t ERSup a*-*
mc
+
-
)(on.E.RFC
+ T CELL -1 CULTURES
32.11
3.3
6.2
Non.E.RFC NH,CI CELLS
I
4.2% : 0.4
T Cells
-
-
T C e l l Sup
4.m
:3 . 5
T c e l l SUP
2.3X
2 0.9
.zx
0.0
mc
+
T CELL SUP
-
E ~ F C
mc
+
2nd ERSup
.I
2nd ERSUP
4 7 . s 5 5.4
Figure 5. T cell production of E-rosettirg supanatant (ERS) factors. The data shown is the mean f SE of 3 experiments The symbols + 1.8 X and + 1.2 X indicated enhanced cell survival rather than B cytotoxicity. Effector cells (*) were tested at 1:lOO target cell: lymphocyte ratio.
60
B.F. Mackler, P.A. OHeiU and M. Meistrich
that removal of adherent sheep E caused the cessation of T cell release of E-rosetting factors which induced non-ERFC cytotoxicity. However, re-rosetting with fresh sheep erythrocytes induced these same T cells t o again release factors (2nd E R Sup) capable of inducing M NC-media ted cy t o tox ic activity . T h e experimental design and results are given in Figure 5.
4. Discussion We have shown here that rosettingof human T cells with sheep erythrocytes caused the release of soluble factor(s) termed E-rosetting supernatant, (ERSup) which induced quiescent non-T cells t o mediate nonspecific cytotoxicity. Evidence was presented from several experimental approaches. First, mononuclear cells incubated with sheep E under rosetting conditions a t 3 7 OC failed t o form rosettes and failed t o produce ERSup. Second, T cells incubated with other heterologous erythrocytes were unable t o form rosettes and also failed t o produce ERSup. Third, removal of adherent sheep E from T cells producing ERSup caused the cessation of further ERSup release. In addition, control experiments eliminated t h e participation of (a) FCS, ( b ) sheep E eluates, and, (c) nonspecific activation of macrophages by physical packing during the rosetting process. The depletion of macrophages and monocytes from the mononuclear cell population prior t o rosetting was recently found neither t o enhance nor reduce the subsequent release of ERSup (O’Neill and Mackler, unpublished data). These findings suggested that E rosette-induced release of ERSup probably does not involve a monocytic cell population. The release of ERSup occurred rapidly, within 5 min during high affinity E-rosetting experiments and probably activated the non-ERFC during t h e subsequent 30 min period required f o r separation of t h e two lymphoid su bpopulations. We also concluded that this release could possibly represent either release of some preformed membrane-bound constituent or activation of some preformed intracellular molecule, rather than de novo protein synthesis. The ERSup-activated MNC gave cytotoxic responses equivalent to those mediated by E rosette-activated non-T cells. In addition, both activated lymphoid populations were equally cytotoxic for allogeneic and autologous target cell lines, suggesting the absence of histocompatibility restrictions. These results are consistent with recent evidence that non-T cell-mediated nonspecific cytotoxicity actually involves non-T cell production of lymphotoxin-like moieties [ 8 ] . Other studies have shown that ERSup, in addition to lacking lymphotoxin, also lacked mitogenic factors for lymphocytes and leukocyte migration inhibition activity. The ERSup failed t o potentiate mitogen stimulation of lymphocyte blastogenesis (Mackler a n d O’Neill, unpublished data), but did enhance the rate of fibroblast target cell division. These observations suggested that ERSup may contain either multiple fac:ors or one factor with multiple activities. Studies in our laboratory have now shown that t h e noncytolytic ERSup actually induced non-T lymphoid cells t o produce lymphotoxin-like products.* It is the lymphotoxin-like factors which nonspecifically lyze autologous a n d allogeneic
* O’Neill, P.A., publication.
Mackler, B.F. and Meistrich, M.L., submitted for
EM. J. Immunol. 1977. 7: 55-61 target cells. Other investigators have suggested that non-T cell-mediated nonspecific cytotoxicity for tumor cells involves lymphotoxin-like activity [ 8 ] . These findings also agreed with previous evidence using human non-T cells of normal donors [ 6 ] ,melanoma tumor-bearing patients [7],and patients with congenital agammaglobulinemia [ 221. Several lines of evidence suggested that t h e cytotoxic nonERFC were lymphocytes, and that t h e cytolysis did not involve either macrophages or monocytes. First, the removal of phagocytic cells neither enhanced nor diminished t h e nonspecific cytotoxicity of the remaining nonphagocytic cells. Second, velocity sedimentation experiments yielded a subfraction of cells containing negligible numbers of monocytes, as identified by acridine orange, nonspecific esterase, and peroxidase staining, which still mediated strong nonspecific cytotoxic responses. The presence of increasing numbers of monocytes in later subfractions did not enhance cytotoxic responses. These Elutriator experiments also suggested that the cytotoxic effector cells differed widely in size. The nature of these non-T cytotoxic effector cells remains to be fully defined; however, t h e lymphoid effector cells appeared to have complement (C3) receptors [23, 241. The identification of non-T lymphocytes as the effector cells mediating what has been termed “spontaneous cell-mediated cytotoxicity” agreed with other studies [4,51. Although non-T cells are involved in this cytotoxic phenomenon, n o evidence was found indicating that antibody was involved. On the contrary, there were several lines of evidence against the participation of antibody. First, MNC were not cytotoxic prior t o E rosetting. Second, the non-ERFC were equally cytotoxic toward autologous as well as allogeneic target cells. Third, the removal of cytophilic antibody from t h e non-ERFC by extensive washings did not diminish their cytotoxicity. These findings extended the studies of previous workers (4,5 1 in that we have presented evidence that activated T cells release factors (ERSup) which induce non-T cells, presumably bone marrow-derived lymphocytes, to produce lymphotoxin. Other reports have recently described the regulation of guinea pig B cell lymphokine responses by T cell factors. T cells were found t o release factor(s) which provided a second signal for antigen-stimulated B cells t o produce monocyte chemotactic factors [ 2 5 ] . Conversely, a T cell factor has been described which inhibits mitogen-induced macrophage migration inhibition factor (MIF) production by B cells [ 2 6 ] .Our observations suggested that E rosette activation of T cells may have fortuitously induced higher concentrations of activator (ERSup) rather than inhibitor factors.
One can speculate that other manipulations during cell separation may also activate T cells which induce the non-T cellmediated nonspecific cytotoxicity. Similar nonspecific activations may also occur in vivo from infections or responses t o tumors. This speculaticn is consistent with the high levels of nonspecific cytotoxicity of some mononuclear cells isolated from patients with tumors [2]. In addition, these observations raised serious questions concerning t h e use of E rosetting as a cell separation procedure t o obtain human non-T cells for subsequent functional assays.
Eur. J. lmmunol. 1977. 7: 55-61
We wish to thank Mr. Ron West, Ms. Sharon Edgeworth, Ms. Athena Herrin, Ms.Dora Woodsonand Ms.Patrikkr Trostle for excellent technicalassistance. We are grateful to Drs. Wesley Bullock and Philip Wydefor their comments and suggestions.
T cell-induced non-T cell cytotoxicity
61
10 Wybran, J., Carr, M.C. and Fudenberg, H.H., J. Clin. Invest. 1972. 51: 2537. 11 Butterworth, A.E., Cell Immunol. 1973. 7 357. 12 Winchester, R.J., Fu, SM., Hoffum, T.and Kunkel. H.G., J. Immunol. 1975. 114: 1210.
Received August 25, 1976; in final revised form January 3, 1977.
13 Golstein, P. and Blomgen, H., Cell. Immunol. 1973. 9: 127.
14 Yam, L.T., Li, C.Y. and Crosby, W.H., Am. J. ain. Pathol. 1971. 53: 283. 15 McJunkin, F.A., Anat. Rec. 1969.24: 67.
5. References 1 Takasugi, M., Ward, P.H., Mickey, M.R. and Terasaki, P.I., Nat. Concer Inst. Monogr. 1972.35: 251. 2 Pavie-Fisher, J.. Kourilsky, F.M., Picard, F., Banzet, P. and Puissant, A.. Clin. Exp. Immunol. 1975.21 : 430.
16 Hellstrom, I. and Hellstrom, K.E., in Bloom, B.R. and Glade, P.R. (Eds.) In Vitro Methods of Cell-MedkrtedImmunity, Academic Press, New York 1970, p. 409. 17 Giovanelk, B.E., Yim, S.O., Stehlin, J.S. and Williams, L.J., Jr., J. Nat. Concer Inst. 1972.48: 1531. 18 Hayflick, L. and Moorhead,
P.S., Exp. CellRes. 1961.25: 585.
3 Massey, RJ., Levin, A.C.. Johnson, D. and Deinhardt, F., Fed. Roc. 1975.34: 141.
19 Grabske, R.J., Lake, S., Gledhill, B.L.. and Meistrich, M.L., J. Cell. Physiol. 1975.86: 177.
4 Jondal, M. and Ross. H., Int. J. Concer 1975. 15: 596.
20 Meistrich, M.L., in Rescott, D.M. (Ed.) Separation of Spermatogenic Cells and Nuclei from Rodent Testes, Academic Press, New York 1976, in press.
5 Ross,H.F. and Jondal, M., CYin. Exp. Immunol. 1975.21: 226. 6 O'Neill, P.A., Mackler, B.F. and Wyde, P.R., Cell. Immunol. 1975. 20: 33. 7 O'NeilI, P.A., Mackler, B.F. and Romsdahl, M.M., J. Nat. cclncer Inst. 1976.57 431. 8 Peter. H.H., Eife, R.F. and Kalden, J.R.. J. Immunol. 1976.116: 342.
9 Mackler, B.F., Altman, L.C, Rosenstreich, D.L and Oppenheim, J.J., Nature 1974. 249: 834.
21 Miller, R.G. and Phillips, R.A., J. Cell. Physiol. 1969. 73: 931. 22 Mackler, B.F., O'Neill, P A , Richie, E., Mukhopadhyay, N. and Montgomery, J., Clin. Immunol. Immnopathol. 1976.6: 279. 23 Mackler, B.F. and O'Neill, P.A., Fed. Roc. 1976.35: 810. 24 O"eil1, P.A. and Mackler, B.F., Fed. Roc. 1976.35: 810. 25 Wahl, S.M. and Rosenstreich, D.L., Fed. Roc. 1976.35: 710. 26 Cohen, S. and Yoshida, T., Fed. Roc. 1976. 3 5 389.
Eur. J. Immunol. 1977. 7: 55-61 B.F. Mackler', Peggy A. ONeill' and M. Meistrich' Laboratory of Immunology, Dental Science Institute, The University of Texas Health Science Center at Houston' and Section of Experimental Radiotherapy, M.D. Anderson Hospital and Tumor Institute', Houston
55
T lymphocyte induction of non-T cell-mediated nonspecific cytotoxicity I. Introduction mechanisms* Mononuclear cells (MNC) from normal humans consistently failed t o give nonspecific cytotoxic responses. However, after removal of T cells by sheep erythrocyte (E) rosetting, the remaining non-RFC (rosette-forming cells) now gave significant nonspecific cytotoxic responses against both autologous and allogeneic target cells. Reconstitution experiments with T cell subpopulations failed t o suppress these nonspecific non-E-RFC-mediated cytotoxic responses. There was also no evidence to indicate the involvement of antibody in this nonspecific cytotoxicity. The cytotoxic cells were characterized as non-E-rosetting, non-phagocytic, and glass adherent lymphocytes; n o evidence of monocyte-macrophage participation was found. The inductive trigger of non-E-RFC-mediated cytotoxicity was found to be soluble factors released by T cells during E-rosette formation at 4 'C. Incubation of MNC with horse, marmoset and human erythrocytes under identical conditions failed t o trigger cytotoxicity. The incubation of quiescent MNC with E-rosetting supernatants (ERS) induced nonspecific cytotoxic responses equivalent t o those mediated by separated non-E-RFC. ERSactivated MNC destroyed both autologous and allogeneic target cells. The ERS supernatants themselves were not cytolytic. These findings suggested that cell separation procedures, and possibly in vivo events, which activate T cells may also induce non-T cell-mediated nonspecific cytotoxicity.
1. Introduction
Investigators utilizing various lymphocyte separation and purification procedures [ 1-31 have frequently observed nonspecific cytotoxicity with human cell populations. More recently, several laboratories [4-51 have reported, in addition t o immune T cell-mediated cytotoxicity, that non-T lymphocytes are capable of destroying tumor cells. Of particular interest is the occurrence of non-T cell nonspecific "spontaneous lymphocyte-mediated cytotoxicity" toward allogeneic and xenogeneic tumor cells [5]. These investigators postulated that a second cytotoxic lymphoid subpopulation distinct from T cells exists which mediates natural immune surveillance. Other workers [6-71 also described a non-T cell-mediated cytotoxicity response independent of antibody which involves complement (C3) receptor induction of lymphotoxin. More recently, the involvement of lymphotoxinlike substances has been suggested to mediate spontaneous cytotoxicity of non-T cells [8]. Given these observations, we sought to assess the nonspecific cytotoxic activities of normal human lymphoid subpopula[I 15221
* This work was supported by the National lnstitutesof Health, NIDR grants DE-04210 and DE-02232, General Research Support Grant DSI-29-74-01 and NCI grant CA-17364.
tions. The paper describes the cytotoxicity of mononuclear cells and subpopulations from normal healthy donors in direct microcytotoxicity assays using autologous and allogeneic target cells. In these studies, sheep erythrocyte (E)rosetting of T cells was found t o cause T cells to secrete mediators which induced non-T cell cytolytic responses.
2. Materials and methods 2.1. Separation of lymphoid subpopulations
Mononuclear cells (MNC) were separated from the blood of healthy donors by Ficoll(9 %)-Hypaque (33 %) gradient centrifugation (400 x g, 20 min, 25 "C) as previously described [6, 7,9]; recovery of MNC averaged 73 f 9.3 % SE. T lymphocytes were separated by rosetting with fresh sheep E in 2.5 % fetal calf serum (heat inactivated, E absorbed) for 1 h at 4 'C then centrifuged through Ficoll-Hypaque (400 x g, 20 min, 5 "C). In some experiments, high affinity E rosetting cells (ha-ERFC) were prepared by t h e Wybran method [ 101. Briefly, 1 O7 MNC/ml were incubated with a n equal volume of E-absorbed fetal calf s e p m for 1 h a t 37 'C then mixed with an equal volume of sheep E (4 x 1 07/ml) and sedimented at 350 x g for 5 min. The pelleted cells were gently resuspended, and rosetted cells were separated by Ficoll-Hypaque (400 x g, 20 min, 5 'C) gradient centrifugation.
Correspondence: Bruce F. Mackler, Immunology Lalpratory, Dental Science Institute, P.O.Box 20068, Houston, Texas 77025, USA
2.2. Macrophage depletion
Abbreviations: E: Erythrocyte EAC Sheep E coated with 19 S rabbit anti-E antibody and mouse complement components ERFC T cell rosettes E R S E-rosetting mixture supernatants haERFC High affinity T cell rosettes MEM: Eagle's minimum essential medium M N C Mononuclear cells non-ERFC Non E-rosetting cells (non-T cells) Smlg: Surface membrane-associated immunoglobulins FCS: Fetal calf serum
Non-ERFC remaining after removal of ERFC were depleted of phagocytic cells by the carbonyl iron method [l 11: 60-75 % of the original non-ERFC subpopulation was recovered after iron filing treatment. I n the recovered population, < 1 % phagocytic cells were evident as determined by 2 % neutral red dye uptake; phagocytic cells possessed red granules.
56
B.F. Mackler, P.A. ONeill and M. Meistrich
Eur. J. Immunol. 1977. 7:55-61
2.3. Lymphoid subpopulation markers
The use of automatic dispensing pipettes was found to increase reproducibility and is recommended.
2.3.1. Complement receptor-bearing cells (CRL) 2.4.2. Target cell lines
The percentage of CRL was quantitated using sheep E coated with 1 9 S rabbit anti-sheep E antibody (Cordis Laboratories, Miami, Florida) and components of fresh mouse complement, termed the EAC reagent. Lymphocytes and EAC were incubated a t 37 "C for 30 min with gentle agitation, and the percentage of EACRFC was quantitated. 2.3.2. Surface membrane immunoglobulins (SmIg)
Lymphocytes were preincubated before staining t o remove cytophilic antibody [ 121. Cells were suspended in a 1 % bovine serum albumin (BSA)-isotonic saline-azide buffer (4 "C), washed, then incubated 30 min at 4 "C with fluorescein-conjugated polyvalent goat anti-human immunoglobulin serum (Meloy Laboratories, Springfield, Va.). 2.3.3. Sheep erythrocyte rosette-forming cells (ERFC)
Cells ( 1 x 1 0 6 ) were mixed with an equal volume of fresh E (5 %) in RPMI 1640 medium (GIBCO, Grand Island, N.Y.) with 2.5 % fetal calf serum (FCS, heat inactivated, E absorbed). centrifuged (250 x g, 5 min at 4 "C), then incubated 1 h at 4 "C. The supernatant removed after incubation of MNC with sheep E was termed the E-rosetting supernatant (ERSup). The percentage of cells forming E rosettes was calculated after counting 200 nucleated cells. 2.3.4. Staining of monocytes
Monocytes were identified by acridine orange [ 131, nonspecific esterase [ 141, and peroxidase [ 151 cytochemical staining procedures.
Allogeneic target cells - human melanoma cell line SH-I [ 171, and lung fibroblast cell line WI-38 [ 181 - were cultured in Eagle's MEM plus 10 % fetal calf serum (heat inactivated), MEM vitamins, glutamine and antibiotics in a 5 70 COz in air atmosphere [6, 71. Autologous fibroblast cell lines were established from gingival specimens obtained during therapeutic periodontal surgery and were a gift from Dr. George Rose, The University of Texas Dental Branch. 2.5. Velocity sedimentation separation of non-ERFC
Non-ERFC were subfractionated by velocity sedimentation using a Beckman Elutriator rotor in a refrigerated 5-21 B centrifuge [19-211. The Elutriator system was flushed with 70 % ethanol and Hanks' balanced salt solution with 0.5 % BSA t o sterilize the system. All air bubbles were removed and the flow rate and rotor speed adjusted for the first fraction ( 1 1.3 f 0. I ml/min, 3000 rpm). Non-ERFC, depleted of ERFC, approximately 150 x 1 O6 - 50 x 1 O6 cells, were introduced into the chamber with constant flow t o collect the first fraction. The rotor speed was held at 3000 rpm while the flow rate was increased stepwise: 16.5 0.5 ml/min (fraction 2), 21.5 f 0.1 ml/min (fraction 3), 33.5 f 1.4 ml/ rnin (fraction 4); for fraction 5 the rotor was stopped and maximum flow rate was used t o remove the remaining cells from t h e chamber. The sedimentation rates were determined t o be: 5 2.4 mm/h/g(fraction 1); 2.4-3.5 mm/h/g(fraction 2); 3.5 -4.6 mm/h/g (fraction 3); 4 . 6 - 7 . 2 mm/h/g (fraction 4); and 2 7.2 mm/h/g (fraction 5). +_
3. Results 2.4. Cytotoxicity assay method 3.1. Cytotoxic activities by human lymphoid subpopulations
2.4.1. Microcytotoxicity assay
The lymphocyte cytotoxicity was assessed in a micro-tray method [ 161. Each well was seeded with approximately 200 target cells in 0.2 ml of Eagle's MEM plus 10 % fetal calf serum, MEM vitamins, 2 mM L-glutamine, penicillin and streptomycin using an automatic dispensing pipette (Schwarz/ Mann), then incubated overnight at 37 "C. At 24 h, lymphocytes were added t o each well in a 0.05ml volume using an automatic dispensing pipette and incubation continued for 48 h. The plates were then washed at t h e end of this second incubation, air dried and adherent target cells stained with Giemsa. Control target cell counts averaged 208.6 f 6.9 (SE) for allogeneic and 139.1 f 6.6 (SE) for autologous target cells. Trypan blue dye exclusion tests of non-adherent target cells showed that > 98 % of the cells in the supernatant were dead. Cytotoxicity was calculated by the formula: Avg. no. cells/well in test wells of cells
%Cytotoxicity = 1 -
+ lymphocytes
x 100
Avg. no. cells/well in control well of cells only
Averages were determined using 6 replicate wells for each test variable and statistical significance by Student's t-test.
Human lymphoid subpopulations separated by E rosetting were assayed for nonspecific cytotoxicity using allogeneic and autologous target cells. The results from multiple experiments using allogeneic melanoma (SH-I ) target cells in direct microcytotoxicity assays are shown in Figure 1. The MNC and ERFC were not cytotoxic even when tested at target cell: lymphocyte ratios of 1 : S O 0 and 1 :200, respectively. In contrast, the non-ERFC remaining after T cell removal had significant cytotoxic activity even a t ratios as low as 20: 1. The preparation of high affinity E rosettes (ha-ERFC) by a 5 min centrifugation step followed by a 30 min centrifugation on Ficoll-Hypaque t o separate lymphoid subpopulation induced similar levels of non-ERFC cytotoxicity (40.5 f 8.7 % S E at 1 :SO). Therefore, the ha-ERFC were actually in the presence of the non-ERFC for almost 3 5 min, which may account for the high activation of the non-ERFC by a relatively short high affinity 5 min rosetting period. In addition, extensive washing of non-ERFC, which presumably removed cytophilic antibodies, did not diminish non-ERFC-mediated cytotoxic activity (34.9 5.1 % SE at 1 :SO).
*
The contribution of tumor and histocompatibility antigens in this nonspecific cytotoxic phenomenon was assessed by using
T cell-induced non-T cell cytotoxicity
EM. J. Immunol. 1977. 7: 55-61
57
Table 1. Cytotoxicity of human lymphocyte populations toward autologous gingivalfibroblast target cells. The average number of target cellslwellare shown along their respective standard error (S.E.) and percentage (7%) cytotoxicity for each individual experiment. The average percent cytotoxicity of the pooled experiments is shown along with the SE. Exp.
Control
no.
Avg. no. cells/
well 1 2 3 4
f
SE
125.7 f 5 . 8 130.0 f 3.6 151.0 f 7.7 149.7 f 5.9
MNC (1 :100) Avg.no.celld well f S E
5% Tox.
123.3 f5.8 127.3 f 6.1 148.3f6.4 147.7 f 5.1
1.9 2.1 1.8 1.3 -
Average % cytotoxicity (Tox.) Among Expts. f SE
E-RFC (1: 100) Avg.no.alls/ % well f S E Tox. 122.9 f 8.1 127.9 f 2.1 149.8 f 6.7 146.9 f7.1
92.4 f 5.2 78.0 f 6.5 112.2 2.7 83.3 f 4.4
*
autologous gingival fibroblasts as target cells (Fig. 1 and Table 1). In these experiments, gingival fibroblasts cell lines were prepared from therapeutic gingival surgery specimens of 4 different patients, and these cells were used as target cells prior t o their 5th passage.
26.5 40.0 25.7 44.5 34.2 f 4 . 8
1.6 f 0.3
1.7 f 0.2
Figure 1 shows that experiments performed with autologous lymphocytes and target cells gave nonspecific cytotoxic responses consistent with t h e allogeneic studies. The individual autologous experiments are presented in Table 1 in terms of the actual target cell counts per sextruple replicates of each experimental variable and their respective percentage cytotoxicity. In terms of t h e actual cell numbers, gingival target cell fibroblasts in control cultures which lacked lymphocytes ranged between 125.7- IS1 with individual SE of 3.6- 7.7 %. The addition of autologous MNC failed t o cause target cell destruction; the percentage cytotoxicity in individual experiments ranged from 1.3 - 2.1 with a mean of 1.7 f 0.2 %. Similarly, autologous ERFC failed t o induce any appreciable cyt otoxicit y . However, autologous non-ER FC were cyt o t oxic four experiments ranging from 25*7-44*5 giving a mean in value of 34.2 k 4.8 % cytotoxicity.
2.2 1.6 0.8 1.9 -
Non-ERFC (1 :100) 7% well+fSE Tox.
Avg. no. cells/
50
t
?
30
2 Figure 2. The dose dependent cytotoxicity curve with increasing numbers of cytotoxic non-ERFC lymphocytes. Each point is the mean f SE of 4 experiments. The ratio of target cel1:lymphocytes is shown on the abscissa and percent cytotoxicity on the ordinate.
3.2. Reconstitutionof cytotoxic
with
cells
Since t h e removal of (high affinity a n d total) ERFC resulted in cytolytic activity by t h e remaining non-ERFC, efforts were made t o reverse this activation in T cell reconstitution experiments. In four experiments, t h e total ERFC or a subpopulation of ha-ERFC were recombined with non-ERFC at the same percentage found in MNC populations. The target cell: non-ERFC ratio was held constant (1 :20) while increasing numbers of either ERFC or ha-ERFC were added (Figure 3). The addition of these T cells neither significantly enhanced nor diminished the nonspecific cytotoxic responses of nonERFC.
Figure 1. Cytotoxicity of normal human lymphoid subpopulations for allogeneic and autologous target cells in direct miaocytotoxicity assays. Allogeneic human lung fibroblasts (Wl-38) or melanoma (SH-1) or autologous - human gingival fibroblast - target cells were used as
indicated. The results show as the mean k SE from multiple experiments (N) with allogeneic and autologous cell donors. The dose-dependent relationship of cytotoxic non-ERFC t o allogeneic target cells is shown in Figure 2. Nonspecific CYtotoxic activity was significant even at very low ratios Of target cells: lymphocytes (e.g., 1 : 1, 10: 1 and 20: 1).
j 3. Effect ~ of, reconstitution ~ with T cell subpopulations on non-ERFC-medhted cytotoxicity. NowERFC were reconstituted with increasing numbers of ERFC or ha-ERFC as indicated. (N) is the number of experiments performed.
~
Eur. J . Immunol. 1977. 7: 55-61
B.F. Mackler, P.A. ONeill and M. Meistrich
58
3.3. Characterization of cytotoxic non-ERFC effector cells 3.3.1. Glass adherence properties The cytotoxic non-ERFC effector cells were characterized and separated by glass adherence properties. In three experiments, the adherent cells represented 12.5 % o f the total nonERFC population and contained 5.9 f 4.2 % complement receptor-bearing cells (CRL), 22.2 f 1.6 % immunoglobulin staining cells (Smlg), 25.2 f 4.1 % neutral red and 34.4 f 6.7 % esterase positive cells. In contrast, the majority of cells was recovered in the nonadherent subpopulation (86.5 %) of which 37.2 k 1.4 7% were CRL, 32.7 f 1.2 % Smlg+, 16.8 f 3.6 % neutral red and 15.5 f 4.0 % esterase positive cells. The cytotoxic activities of these two cell fractions are shown in Table 2. Adherent and nonadherent cells were added t o target cells in numbers comparable t o their percentages in the non-ERFC population. The adherent cells gave cytotoxic responses that were equivalent t o the original non-ERFC population. The nonadherent cells gave much lower cytotoxic responses ( 1 5.9 f 6.1 %) which could be reduced even further by multiple sequential adherence of these cells to glass.
3.3.2. Phagocytic cells The role of phagocytic cells in the non-ERFC-mediated nonspecific cytotoxicity was assessed by removal of phagocytic cells using carbonyl iron. After removal of cells engulfing iron filings, the recovered 60-75 % of the original non-ERFC contained < 1 % phagocytic cells as determined by neutral red dye uptake. The phagocyte-depleted non-ERFC were tested for nonspecific cytotoxicity at target cel1:lymphocyte ratios equivalent t o the percentage of phagocytic and nonphagocytic cells in t h e original non-ERFC population (Table 2). The removal of phagocytic cells neither enhanced nor diminished the cytotoxic activity of the remaining non-ERFC. The separated phagocytic cells were found to have low cytotoxic
responses (23.2 f 5.5 %) which was presumed to represent the activity of macrophages activated by the iron filing removal method.
3.3. Velocity sedimentation separation of lymphocytes and monocy tes In order t o remove nonphagocytic monocytes and characterize further the cytotoxic effector cells, cytotoxic nonERFC were separated by velocity sedimentation into 5 fractions using a Beckman Elutriator rotor. This method separated the cells by volume with fraction 1 containing cells with the smallest volume and fraction 5 the largest volume. The total cell recovery of the original non-ERFC population applied t o the rotor ranged between 65-91 % (viability > 98 %) with t h e majority of cells being recovered in the 3rd and 4th fractions. Each fraction was assessed for nonspecific cytotoxic activity and for the presence of monocytes identifiable by acridine orange (AO), esterase (EST) and peroxidase (P)cytochemical staining (Figure 4). Fractions 1 and 2 were found t o have significant cytotoxic activities while containing rather negligible numbers of monocytes. The presence of increasing percentages of monocytes did not enhance the nonspecific cytotoxic activities of fractions 3 , 4 and 5 . The distribution of cytotoxic effector cells in t h e 5 fractions did not reflect monocyte distribution. These data suggested that monocytes were probably not the cytotoxic effector cells in non-ERFC populations.
Table 2. Characterization of cytotoxic non-ERFC effector cells by glass adherence and phagocytosis. The number o f experiments is designated by (N) and the target cells used were SH-1 Lymphoid su bpopula t ions
Target cell: lymphocyte ratio
c/;
(N)
Cytotoxicity f SII
M NC
1:lOO
(4)
1.2 f 0.8
Non-ERFC
1 : 200
(3) (4)
53.8 f 4.1 56.4 f 7.5 50.4 f 6.2
1:lOO 1:50
(4)
A. Ghs$ adhcrcncc Non-adherent nonI: R FC Adlicrcnt non-ERF('
15.9 f 6.1 13.7 f 5.5 51.x f 12.2 18.2 f 12.5
H. Iron filing trcatmcnt Non-phagocytic nonERFCd
I : 100 1:lOO
Plirgocyt ic non-ERFC
a) < 1 7; phagocytic cells.
1:50
(3) (3) (3)
70.4 f 14.1 66.7 f 6.6 48.0 f 3.7
1:50
( 3)
1:25
(3)
20.9 f 2.3 23.2 f 5.5
Figure 4. Velocity sedimentation separation of cytotoxic non-ERFC into subfractions. Fraction 1 contains cells with the smallest volume while fraction 5 has those with the largest volume. The perccnt cell recovery is the distribution of the total cells recovered in all five fractions.
3.4. The role of sheep E rosetting in the induction of nonERFC-mediated cytotoxicity Since t h e induction of non-ERFC-mediated cytotoxicity appeared t o occur after sheep E rosetting, the nature of this inductive mechanism was examined. To assess whether erythrocyte binding or the physical packing of cells by centrifugation triggered non-ERFC cytotoxicity, MNC ( I O7 cells) were mixed with similar concentrations ( 4 x 1 07)of either sheep, horse, marmoset or human erythrocytes in RPMl-I 640 plus 2.5 7% FCS (absorbed, heat inactivated) and gently packed by centrifugation (250 x g, 5 min at 4 "C). After 1 h incubation a t 4 OC, the number of ERFC was counted, and the nonERFC were separated by Ficoll-Hypaque density gradient cen-
T cell-induced non-T cell cytotoxicity
Eur. J. Immunol. 1977. 7: 55-61 trifugation. In several experiments, sheep E and MNC mixtures were incubated at 37 OC for 1 h. In Table 3, the mean results from multiple experiments are summarized. Cytotoxic non-ERFC were obtained only after sheep E rosette formation at 4 OC, but not at 37 OC. The MNC incubated with either horse, human or marmoset E formed negligible numbers of E rosettes, and the majority of cells was recovered in t h e non-ERFC population which was not activated. These findings suggested that FCS and the physical packing of MNC with erythrocytes were not sufficient to induce non-ERFC cytotoxicity. Similar sheep E-rosetting experiments performed in t h e absence of FCS eliminated any nonspecific involvement of FCS moieties. In these experiments, 43.0 f 9.6 % ERFC were obtained while the recovered non-ERFC had nonspecific cytotoxic activities of 38.6 f 6.4 % at 1 : 100. These results provided evidence suggesting that the actual E rosette formation was the inductive trigger for non-T cell-mediated nonspecific cytotoxicity. Table 3. MNC were rosetted with erythrocytes from various animals and incubated for 1 h. The number of ERFC was first determined, then the non-ERFC separated by Fimll-Hypaque gradients and nonERFC cytotoxicity assessed using SH-1 target cells. The number of experiments is designated by (N) and the mean shown f SE E
wurce -8)
Sheep Shcep Human Horw Marmoset
Incubation temperature (OC) 4 37 4 4 4
(N)
70 E rosetteforming cells
0 (11) (4) 45.7 f5.7 (3) 1.0 fO.5 0 (3) 3.3 f l . 9 (3) 0 (2)
3 Non-ERFC
cytotoxicity 13.7 f 4.7 60.3 f 13.6 7.0 f 1.6 5.3 f 3.1 3.9 f 3.9
0
a) Untreated MNC.
3.5. E-roset ting induction of non-ERFC-mediated cytotoxicit y by supernatant factors The E-rosetting experiments with various erythrocytes suggested that T cells induced non-ERFC-mediated cytotoxicity. To investigate this finding further, supernatant fluids from the E-rosetting mixture incubated at 4 OC for 1 h were assessed for their ability t o induce autologous MNC t o manifest nonspecific cytotoxic responses (Table 4). Control supernatants were prepared using similar concentrations of only sheep E and fetal calf serum incubated at 4 O C for 1 h. The quiescent MNC were incubated for 30 min in undiluted ER supernatant, washed twice with medium and plated in microcytotoxicity trays at 1 : 1 0 0 target cel1:lymphocyte ratios. I n repeated experiments, ER supernatants (ERSup) prepared with either home, human or marmoset E as well as control supernatant failed t o activate MNC. However, ERSup prepared with sheep E + MNC at 4 OC consistently induced autologous MNC (64.9 f 7.2 % at 1 : 100) as well as allogeneic MNC (59.4 f 8.1 % at 1 : 100) t o manifest cytotoxic responses. Although these ERSup induced non-T cell-mediated cytotoxicity, the ERSup did not contain any detectable cytolytic activity in itself. As controls, supernatants prepared with sheep
59
Table 4. E-rosetting supernatant (ERSup) induction of MNC cytotoxic activities. Supernatants from either E-rosetting mixtures or control supernatants (Sup) were incubated with autologous MNC, washed, then assessed for cytotoxicity. The number of experiments performed with cells from different donors is designated by (N) and the mean value is shown f SE
Effector cells
ERSup
'?Cytotouicit):
(N)
*
13.7 % 4.7 +1.3Xf0.21) 5.4 Q f 4 . 8
MNC
-
SUP SUP Human Horse Marmoset Sheep Sheep
-
-I- MNC 4- MNC 4- MNC
+
0.0
7.8 Y f 1.6 2 . 2 5 f 1.5 + 2.0 X f 0.61) 64.9 % f 7.2
MNC
-
-t MNC
a) The + 1.3 X and + 2.0 X indicate enhanced cell survival rather than % cytotoxicity.
E + MNC a t 37 OC were found t o lack the ability t o induce non-T cell-mediated cytotoxicity, an observation consistent with their inability t o form E rosettes.
T cell production of E-rosetting factors was also assessed after t h e removal of adherent erythrocytes. The experimental design is shown in Figure 5 along with the data from triplicate experiments. MNC were first rosetted with sheep E and an Erosetting supernatant (1 st ERSup) was obtained. The ERFC and non-ERFC were then separated by Ficoll-Hypaque gradient centrifugation. Consistent with previous findings, t h e 1 st ERSup induced autologous MNC t o become cytotoxic (36.1 % f 5.9) a t levels comparable t o the non-ERFC (32.1 % f 6.2). The 1st ERSup alone was not cytolytic but actually appeared t o enhance target cell growth (1.8 times). The ERFC sub population was treated with NH4Cl a t 37 OC t o lyze adherent sheep E which were removed by multiple washes with medium at 3 7 OC; t h e viability of the recovered T cells was > 9 5 %. The T cell cultures were incubated under similar conditions for 1 h, a n d a T cell supernatant was obtained. The T cell supernatants were neither directly cytolytic nor able t o induce MNC-mediated cytotoxicity. These results suggested E a p r i r e n t r l Design
1 Cytotoaicity t S.E.
Effector Cells
Suuernrtmt
-
1 s t ERSup
+1.u: 0.0
1 s t ERSUP
36.11 % 5.9
mc 5.n 1 s t ERSup a*-*
mc
+
-
)(on.E.RFC
+ T CELL -1 CULTURES
32.11
3.3
6.2
Non.E.RFC NH,CI CELLS
I
4.2% : 0.4
T Cells
-
-
T C e l l Sup
4.m
:3 . 5
T c e l l SUP
2.3X
2 0.9
.zx
0.0
mc
+
T CELL SUP
-
E ~ F C
mc
+
2nd ERSup
.I
2nd ERSUP
4 7 . s 5 5.4
Figure 5. T cell production of E-rosettirg supanatant (ERS) factors. The data shown is the mean f SE of 3 experiments The symbols + 1.8 X and + 1.2 X indicated enhanced cell survival rather than B cytotoxicity. Effector cells (*) were tested at 1:lOO target cell: lymphocyte ratio.
60
B.F. Mackler, P.A. OHeiU and M. Meistrich
that removal of adherent sheep E caused the cessation of T cell release of E-rosetting factors which induced non-ERFC cytotoxicity. However, re-rosetting with fresh sheep erythrocytes induced these same T cells t o again release factors (2nd E R Sup) capable of inducing M NC-media ted cy t o tox ic activity . T h e experimental design and results are given in Figure 5.
4. Discussion We have shown here that rosettingof human T cells with sheep erythrocytes caused the release of soluble factor(s) termed E-rosetting supernatant, (ERSup) which induced quiescent non-T cells t o mediate nonspecific cytotoxicity. Evidence was presented from several experimental approaches. First, mononuclear cells incubated with sheep E under rosetting conditions a t 3 7 OC failed t o form rosettes and failed t o produce ERSup. Second, T cells incubated with other heterologous erythrocytes were unable t o form rosettes and also failed t o produce ERSup. Third, removal of adherent sheep E from T cells producing ERSup caused the cessation of further ERSup release. In addition, control experiments eliminated t h e participation of (a) FCS, ( b ) sheep E eluates, and, (c) nonspecific activation of macrophages by physical packing during the rosetting process. The depletion of macrophages and monocytes from the mononuclear cell population prior t o rosetting was recently found neither t o enhance nor reduce the subsequent release of ERSup (O’Neill and Mackler, unpublished data). These findings suggested that E rosette-induced release of ERSup probably does not involve a monocytic cell population. The release of ERSup occurred rapidly, within 5 min during high affinity E-rosetting experiments and probably activated the non-ERFC during t h e subsequent 30 min period required f o r separation of t h e two lymphoid su bpopulations. We also concluded that this release could possibly represent either release of some preformed membrane-bound constituent or activation of some preformed intracellular molecule, rather than de novo protein synthesis. The ERSup-activated MNC gave cytotoxic responses equivalent to those mediated by E rosette-activated non-T cells. In addition, both activated lymphoid populations were equally cytotoxic for allogeneic and autologous target cell lines, suggesting the absence of histocompatibility restrictions. These results are consistent with recent evidence that non-T cell-mediated nonspecific cytotoxicity actually involves non-T cell production of lymphotoxin-like moieties [ 8 ] . Other studies have shown that ERSup, in addition to lacking lymphotoxin, also lacked mitogenic factors for lymphocytes and leukocyte migration inhibition activity. The ERSup failed t o potentiate mitogen stimulation of lymphocyte blastogenesis (Mackler a n d O’Neill, unpublished data), but did enhance the rate of fibroblast target cell division. These observations suggested that ERSup may contain either multiple fac:ors or one factor with multiple activities. Studies in our laboratory have now shown that t h e noncytolytic ERSup actually induced non-T lymphoid cells t o produce lymphotoxin-like products.* It is the lymphotoxin-like factors which nonspecifically lyze autologous a n d allogeneic
* O’Neill, P.A., publication.
Mackler, B.F. and Meistrich, M.L., submitted for
EM. J. Immunol. 1977. 7: 55-61 target cells. Other investigators have suggested that non-T cell-mediated nonspecific cytotoxicity for tumor cells involves lymphotoxin-like activity [ 8 ] . These findings also agreed with previous evidence using human non-T cells of normal donors [ 6 ] ,melanoma tumor-bearing patients [7],and patients with congenital agammaglobulinemia [ 221. Several lines of evidence suggested that t h e cytotoxic nonERFC were lymphocytes, and that t h e cytolysis did not involve either macrophages or monocytes. First, the removal of phagocytic cells neither enhanced nor diminished t h e nonspecific cytotoxicity of the remaining nonphagocytic cells. Second, velocity sedimentation experiments yielded a subfraction of cells containing negligible numbers of monocytes, as identified by acridine orange, nonspecific esterase, and peroxidase staining, which still mediated strong nonspecific cytotoxic responses. The presence of increasing numbers of monocytes in later subfractions did not enhance cytotoxic responses. These Elutriator experiments also suggested that the cytotoxic effector cells differed widely in size. The nature of these non-T cytotoxic effector cells remains to be fully defined; however, t h e lymphoid effector cells appeared to have complement (C3) receptors [23, 241. The identification of non-T lymphocytes as the effector cells mediating what has been termed “spontaneous cell-mediated cytotoxicity” agreed with other studies [4,51. Although non-T cells are involved in this cytotoxic phenomenon, n o evidence was found indicating that antibody was involved. On the contrary, there were several lines of evidence against the participation of antibody. First, MNC were not cytotoxic prior t o E rosetting. Second, the non-ERFC were equally cytotoxic toward autologous as well as allogeneic target cells. Third, the removal of cytophilic antibody from t h e non-ERFC by extensive washings did not diminish their cytotoxicity. These findings extended the studies of previous workers (4,5 1 in that we have presented evidence that activated T cells release factors (ERSup) which induce non-T cells, presumably bone marrow-derived lymphocytes, to produce lymphotoxin. Other reports have recently described the regulation of guinea pig B cell lymphokine responses by T cell factors. T cells were found t o release factor(s) which provided a second signal for antigen-stimulated B cells t o produce monocyte chemotactic factors [ 2 5 ] . Conversely, a T cell factor has been described which inhibits mitogen-induced macrophage migration inhibition factor (MIF) production by B cells [ 2 6 ] .Our observations suggested that E rosette activation of T cells may have fortuitously induced higher concentrations of activator (ERSup) rather than inhibitor factors.
One can speculate that other manipulations during cell separation may also activate T cells which induce the non-T cellmediated nonspecific cytotoxicity. Similar nonspecific activations may also occur in vivo from infections or responses t o tumors. This speculaticn is consistent with the high levels of nonspecific cytotoxicity of some mononuclear cells isolated from patients with tumors [2]. In addition, these observations raised serious questions concerning t h e use of E rosetting as a cell separation procedure t o obtain human non-T cells for subsequent functional assays.
Eur. J. lmmunol. 1977. 7: 55-61
We wish to thank Mr. Ron West, Ms. Sharon Edgeworth, Ms. Athena Herrin, Ms.Dora Woodsonand Ms.Patrikkr Trostle for excellent technicalassistance. We are grateful to Drs. Wesley Bullock and Philip Wydefor their comments and suggestions.
T cell-induced non-T cell cytotoxicity
61
10 Wybran, J., Carr, M.C. and Fudenberg, H.H., J. Clin. Invest. 1972. 51: 2537. 11 Butterworth, A.E., Cell Immunol. 1973. 7 357. 12 Winchester, R.J., Fu, SM., Hoffum, T.and Kunkel. H.G., J. Immunol. 1975. 114: 1210.
Received August 25, 1976; in final revised form January 3, 1977.
13 Golstein, P. and Blomgen, H., Cell. Immunol. 1973. 9: 127.
14 Yam, L.T., Li, C.Y. and Crosby, W.H., Am. J. ain. Pathol. 1971. 53: 283. 15 McJunkin, F.A., Anat. Rec. 1969.24: 67.
5. References 1 Takasugi, M., Ward, P.H., Mickey, M.R. and Terasaki, P.I., Nat. Concer Inst. Monogr. 1972.35: 251. 2 Pavie-Fisher, J.. Kourilsky, F.M., Picard, F., Banzet, P. and Puissant, A.. Clin. Exp. Immunol. 1975.21 : 430.
16 Hellstrom, I. and Hellstrom, K.E., in Bloom, B.R. and Glade, P.R. (Eds.) In Vitro Methods of Cell-MedkrtedImmunity, Academic Press, New York 1970, p. 409. 17 Giovanelk, B.E., Yim, S.O., Stehlin, J.S. and Williams, L.J., Jr., J. Nat. Concer Inst. 1972.48: 1531. 18 Hayflick, L. and Moorhead,
P.S., Exp. CellRes. 1961.25: 585.
3 Massey, RJ., Levin, A.C.. Johnson, D. and Deinhardt, F., Fed. Roc. 1975.34: 141.
19 Grabske, R.J., Lake, S., Gledhill, B.L.. and Meistrich, M.L., J. Cell. Physiol. 1975.86: 177.
4 Jondal, M. and Ross. H., Int. J. Concer 1975. 15: 596.
20 Meistrich, M.L., in Rescott, D.M. (Ed.) Separation of Spermatogenic Cells and Nuclei from Rodent Testes, Academic Press, New York 1976, in press.
5 Ross,H.F. and Jondal, M., CYin. Exp. Immunol. 1975.21: 226. 6 O'Neill, P.A., Mackler, B.F. and Wyde, P.R., Cell. Immunol. 1975. 20: 33. 7 O'NeilI, P.A., Mackler, B.F. and Romsdahl, M.M., J. Nat. cclncer Inst. 1976.57 431. 8 Peter. H.H., Eife, R.F. and Kalden, J.R.. J. Immunol. 1976.116: 342.
9 Mackler, B.F., Altman, L.C, Rosenstreich, D.L and Oppenheim, J.J., Nature 1974. 249: 834.
21 Miller, R.G. and Phillips, R.A., J. Cell. Physiol. 1969. 73: 931. 22 Mackler, B.F., O'Neill, P A , Richie, E., Mukhopadhyay, N. and Montgomery, J., Clin. Immunol. Immnopathol. 1976.6: 279. 23 Mackler, B.F. and O'Neill, P.A., Fed. Roc. 1976.35: 810. 24 O"eil1, P.A. and Mackler, B.F., Fed. Roc. 1976.35: 810. 25 Wahl, S.M. and Rosenstreich, D.L., Fed. Roc. 1976.35: 710. 26 Cohen, S. and Yoshida, T., Fed. Roc. 1976. 3 5 389.
