438
P. M. Chisholm and W. L. Ford
Patricia M. ChisholmO and W. L. Ford Department of Pathology, University of Manchester Medical School
Eur. J. Immunol. 1978.8: 438-445
Selection of antigen-specific cells by adherence to allogeneic cell monolayers: cytolytic activity, graft-vs.-host activity and numbers of adherent and nonadherent cells" Rat lymph node cells taken at the peak of cytolytic activity following a skin allograft were separated into adherent and nonadherent fractions by incubation o n monolayers of thoracic duct lymphocytes either of the same strain as the graft donor or of an Ag-B different strain. In t h e face of a 3-fold enrichment of cytolytic activity in the adherent cells and a 3-fold depletion in the nonadherent cells there was no detectible partition of graft-vs.-host (GVH) activity. Supplementary experiments supported the simplest interpretation of this finding, namely that the antigen receptors on GVH-reactive cells did not influence their adherence in this system. Similarly, there was no partition of the GVH activity of nonimmune lymph node cells by adherence. Labeling lymph node cells with either radioactive uridine or thymidine in vitro, suggested that about 20 % o f DNA-synthesizing cells in the immune population adhered because .of antigen recognition.
1 Introduction
Since Dutton and Mishell in 1967 eliminated the subset of lymphocytes reactive t o sheep erythrocyte antigens [ 11 several techniques have been described by which subpopulations of lymphocytes can be either enriched for o r depleted of antigensensitive cells t o a particular antigen. Starting with either immune or nonimmune populations, positive and negative selection have been achieved for b o t h T and B cells. In the case of T cells reactive t o those alloantigens determined by the major histocompatibility complex (MHC), the situation is of particular interest because of t h e well-known anomalies of these reactions, especially the high proportion (around 5 %) [ 2 , 31 of nonimmune lymphocytes proliferating in response to each MHC and because immunization in vivo, while provoking cytolytic activity, fails t o increase more than marginally t h e number of cells which react to a second antigenic stimulus by proliferation [4, 51. The cytolytic T cells in an alloimmune population can be selectively depleted by exploiting their increased adhesion t o cells bearing the immunizing alloantigen as has frequently been described following t h e original observations of Brondz [6]. The principle involved is that of the Wigzell column [7] in which antigen is immobilized, and an antigen-specific subset is retained because of binding brought about by surface receptors for antigen. The same principle apparently operates [I 20421
in vivo when antigen-specific T lymphocyte; are selectively retained in lymphoid tissue during passage from blood t o lymph in an F1 hybrid [8] or allogeneic recipient [ 91. However, it has been anomalous and in a practical sense disappointing that attempts t o deplete mixed lymphocyte reaction (MLR) or GVH activity in vitro by adhesion t o immobilized antigenic cells have often been unsuccessful [ 10- 121. The most promising results have been obtained by permitting reactive populations t o adhere for several hours to fibroblast monolayers. Both GVH-reactive cells and the precursors of cytolytic cells were depleted in the nonadherent fraction [ 13- 16). However, the impression from the considerable literature on the use of cell monolayers as immunoabsorbants is that cytolytic cells are easier to partition than GVH-reactive cells. This was supported by measuring each function in samples of nonadherent cells in which it was found that cytolytic activity but not GVH activity was reduced [ l o ] . The main object of our experiments was to make quantitative estimates of the GVH and cytotoxic activities of both adherent and nonadherent cells in a system in which immune (or nonimmune) lymph node (LN) cells were incubated on allogeneic monolayers in order t o test the suspicion that the partition of cytolytic activity would be greater than the partition of GVH activity. A second object was t o estimate the proportions of immune and nonimmune T cells which adhered t o allogeneic monolayers through the agency of their antigen receptors. This was done by application of a radioisotopic labeling technique that has already been used to estimate the number of T cells responding in a systemic GVH reaction [ 171.
* Most of the work was supported by a grant from the Medical
'
Research Council G972/455B. Present address: Department of Immunology, Royal Postgraduate Medical School, Hammersmith Hospital, Du Cane Road, London W12, GB. P.M.C. was supported by a grant from the Manchester Central District Research Grants Committee.
Correspondence: William L. Ford, Department of Pathology, Stopford Building, Oxford Road, Manchester M13 9PL, GB Abbreviations: GVH: Graft-vs.-host MLR: Mixed lymphocyte reaction (responder activity) TDL: Thoracic duct lymphocytes MHC: Major histocompatibility complex LN:Lymph node PBS: Phosphate-bufferedsaline FCS: Fetal calf serum Urd: Uridine
dThd: Thymidine [ *25I]dUrd: 1xI-labeled deoxyuridine dpm: Disintegrations per minute
2 Materials and methods
2.1 Rats and their immunization Three highly inbred strains were used - A 0 (Ag-B2), HO (Ag-BS, synonym PVG/c) and DA (Ag-B4), [ 181. Appropriate F1 hybrids were used for the assay of GVH activity. HO rats were immunized b y single allografts of A 0 o r DA skin applied t o the flank. DA rats were similarly grafted with A 0 or HO skia. The draining axillary and brachial LN were removed at the eight-day peak of cytolytic activity and teased into a single cell suspension.
Eur. J. Immunol. 1978.8: 438-445
Cytolytic and graft-us.-hostactivities of alloimmune cells
439
2.2 Cytotoxic assay
3 Results
The cytotoxicity of immune LN cells was measured against the WF/C1 lymphoma cell line which, like t h e A 0 strain and th e strain of origin (Wistar Furth), expressed t h e Ag-B-2 MHC' [ 18, 191. A standard number ( 2 5 x 1 0 3 )of target cells which had previously been labeled with ['"I]dUrd were incubatcd with graded numbers of attacker cells for 1 8 h in Linbro microtiter plates. Details of the assay and evidence that it measured T cell-mediated cytolysis directed predominantly against Ag-B antigens have been published [ 191.
3.1 Cytotoxic activity of adherent and nonadherent cells
2 .3 Graft-vs.-host activity This was measured by the popliteal LN assay [ 201. Four hind feet o f F hybrids were injected with parental strain lymphocytes at each of two dose levels. The draining LN were removed and weighed 7 days after injection.
2.4 Monolayer preparation Thoracic duct lymphocytes (TDL) were collected by t h e method of Gowans [21]. An excess of lymphocytes ( 7 5 x l o 6 ) were allowed to settle o n a Falcon polystyrene tissue culture petri dish (Falcon Plastics, Oxnard, CA, 60 x 1 5 mm) which had been pretreated with a solution of poly-t-lysine (Sigma Chemical Co. St. Louis, MO) at 5 0 pg/ml as described by Stulting and Berke [22]. Rinsing off the surplus cells left a n unbroken monolayer from which few cells were lost during the subsequent fractionation procedure. Each monolayer consisted of about SO x 1 O6 lymphocytes.
By measuring the release of '''1 from a constant number of WF/GI target cells after incubation with 5 graded doses of immune LN cells it was possible to quantitatively compare the cytolytic activity of adherent and nonadherent cells with a sample of the starting population. The principle of a parallelline assay [ 261 was applied using the parallel segments of the release) regreslog dose response (percentage of specific ''I sion lines as exemplified in Fig. 1. In all experiments, the cytolytic activity of the nonadherent cells was reduced, and the activity of the adherent cells was increased as has been found by many others (for review see [27]. In the first six experiments, the cytolytic activity of the adherent cells was equivalent to that of 3.1 (f 0.55) times as many unfractionated cells, and the activity of nonadherent cells was equivalent to less than half as many (0.45 f 0.05) unfractionated cells. In the next four experiments, samples of the adherent, nonadherent and unfractionated populations were assayed simultaneously for both cytolytic and GVH activities. The relative cytolytic activities of the adherent and nonadherent populations in these four experiments were 3.0 0.05 and 0.28 f 0.10, respectively (Mean f S.E.).
/
2.5 Partition of adherent and nonadherent cells A suspension of 15 x 1 O6 LN cells in 2.5 ml of RPMI medium 1640 (Gibco, Grand Island, NY) with 10 % fetal calf serum (FCS) was incubated o n each monolayer for 30 min at 37 O C . Then the plates were centrifuged at 130 x g for 5 min. [23] and nonadherent cells were removed by shaking [24]. Most of the adherent cells were removed by addition o f 5 m~ EDTA in phosphate-buffered saline (PBS) with 1 0 % FCS followed by gentle shaking and rinsing.
2.6 Radioactive labeling In some experiments, LN cells were labeled in vitro with either [3H]Urd ( 7 pCi/ml) o r [14C]Urd (1 pCi/ml), and in other experiments they were labeled with either [3H]dThd (0.6 pCi/ml) or [14C]dThd (0.4 pCi/ml). The method of labeling and the choice of isotopes and spec. act. were precisely as previously described [ 17, 211. T h e principle of alternatively (3H/14C) labeling a population immune t o the antigens expressed by the monolayer and a population immune t o other antigens has been described [ 17, 251. The 3H and 14C activities of nonadherent and adherent cells detached after EDTA treatment were estimated in a Beckman scintillation counter [21]. Radioactivity in the cells persistently adherent t o t h e monolayer after EDTA treatment was estimated by digesting t h e residual cells in th e petri dish by direct addition of 2 N sodium hydroxide.
1.6
3.2
6.25
12.5
IMMUNE CELLS I x 10'
25.0
I
Figure 1. The specificity of cytolytic cell adherence. HO anti-A0 lymphocytes were incubated on either A 0 or DA monolayers, and
the adherent and nonadherent cells were tested for cytolytic activity. Unfractionated cells (0-0); cells adherent to specific (AO) monolayers ( 0 - 0 ) ; cells adherent to third-party (DA) monolayers ( 0 - - 0 ) ; cells nonadherent to specific monolayers (A-A); cells nonadherent to third-party monolayers (A- -A).
.
.
Neither the threefold enrichment of activity in t h e adherent population nor the two to fourfold depletion in the nonadherent population was improved by applying a number of variations of the technique including the use of allogeneic fibroblast monolayers [24]. However, the possibility that a minority of cytolytic cells are exceptional in that they are
Eur. J. Immunol. 1978.8: 438-445
P. M. Chisholm and W. L. Ford
440
not removable by adherence t o antigenic monolayers, was discounted by experiments of which an example is illustrated in Fig. 2 . Either 2 o r 3 successive incubations of the nonadherent cells always reduced cytolytic activity t o an undetectable level suggesting that most o r all cytolytic cells express receptors which can bind them t o allogeneic cells. Other workers have generally found that t h e depletion of cytolytic activity after a single adhesion step t o a n antigenic monolayer is only partial for reasons which remain unclear [e.g. 10, 231.
The specificity of the partition was further confirmed by the finding that the small minority of LN cells which were adherent t o the DA monolayer did not show the augmented activity of the population adherent t o A 0 monolayers; in fact, the cytolytic activity of these cells was reduced. It might be expected that the minority adhering for reasons other than antigen recognition would include most of the macrophages and damaged cells. However, subsequent experiments suggested that this might not be the only explanation for the reduced cytolytic activity.
60
40
20
W v)
2 1.25
2.5
5.0
10.0
20.0
Since the monolayers used as immunoabsorbants for cytolytic lymphocytes were of the same Ag-B type as the target cells, the possibility had t o be considered that the depletion of cytolytic activity in the nonadherent population might be partly attributable t o competition during the cytolytic assay of cells which had become detached from the monolayer. By prelabeling the TDL used t o make the monolayer with [3H]Urd in vitro, the proportion of detached monolayer cells in the nonadherent and adherent populations was estimated on autoradiographic counts t o be approximately 3 % and 15 %, respectively. The possibility that the 3 % contamination of the nonadherent fraction might account for the specific reduction in cytolytic activity was tested by measuring the inhibition of target cell lysis with either lymphocytes bearing the same Ag-B complex as the target cells (AO) o r unlabeled tumor target cells (WF/G1). As expected, it was found that the lysis of labeled WF/G1 target cells progressively decreased as increasing numbers of unlabeled tumor cells were added (Fig. 3). However, even when A 0 TDL were added t o the labeled target cells at a ratio of 4: 1 , no inhibition of lysis was detected (Fig. 4). The surface area of the tumor cells was nearly four times that of TDL, but this would only partly account for the difference in competitive inhibition. It was
(b)
Ab ( 2.3
4:l )
2:l 1
1:l I
I
l:Z )
7
1.25
2.5 5.0 10.0 20.0 IMMUNE CELLS ( x 10 5 ,
Figure 2. Successive adsorption of cytolytic lymphocytes o n specific allogeneic monolayers. Nonadherent cells were recovered from a monolayer and incubated on fresh monolayers for a second, then a third time. The three populations of nonadherent cells (a) [ 1, 2, 31 and adherent cells (b) [ 1 , 2, 31 were assayed for cytolytic activity. Unfractionated cells (0-0); nonadherent cells (A--A); adherent cells ( 0 - 0 ) . After 2 cycles of fractionation no activity remained in the nonadherent population.
The specificity of t h e absorption of cytolytic cells was tested by incubating lymphocytes from HO rats which had been immunized with A 0 skin grafts o n DA monolayers. The cells which were nonadherent t o DA monolayer-s had identical cytolytic activity t o that of the starting population (Fig. 1).
(1:4)
1.6
13.2402-
3.2
6.25
12.5
IMMUNE CELLS ( x 10
25.0
)
Figure 3. Inhibition of cytolysis by unlabeled tumor target cells. Incubation mixtures consisted of HO anti-WF LN cells, 25 x lo3 1251labeled WF/Gl tumor cells and variable numbers of unlabcled WF/GI cells. The ratio of labeled to unlabeled tumor cells is given in parentheses. Immune lymphocytes and labeled tumor cells (0-0); immune lymphocytes, labeled tumor cells and unlabeled tumor cells (0-
0).
Eur. J . Irnmunol. 1978.8: 438-445 0
Cytolytic and graft-us.-hostactivities of alloimmune cells
441
CYTOTOXIC ACTIVITY
UNLABELLEDTDL UNLABELLED TUMOUR CELLS
CONTROL
114
III Ii
0.62
1.25
24
50
UNLABELLED TARGET CELLS
I
1o.O
x 104
7
i
1.6
J
3.2
Figure 4. Inhibition of cytolysis by unlabeled target cells. The bars represent mean l z S I release r S.E. of triplicates after 18 h. Incubation mixtures consisted of 25 x l o 5 LN cells fIom HO rats immunized with WF ,;kin grafts, 25 x 103 l251-labeled WF/GI lymphoma cells and variahle numbers of unlabeled WF/G1 lymphoma cells ( M ) or
WI:.TDL
6.25
12.5
IMMUNE CELLS I I 10' I
G V H ACTIVITY
(El).
clear that the minor contamination of nonadherent cells with cells from the monolayer could not account for the specific reduction of cytolytic activity by "cold target inhibition"
3.2 Graft-us.-host activity of adherent and nonadherent cells In the face of a tenfold difference in cytolytic activity there was n o difference in GVH activity of adherent and nonadherent subsets, and the activity of each was at t h e same level as that of the starting population (Fig. 5 ) . Th e same result was found consistently, and when popliteal LN weights obtained in 4 experiments were pooled and composite dose response regression lines were calculated, the plots for adherent, rionadherent and unfractionated populations were almost exactly coincident ( n o t illustrated). There appeared to be n o partition of GVH-reactive lymphocytes under these experimental conditions. Since each plot was derived from 3 2 LN responses, a minor difference either way, about 10 %, would have been detectable. The feasibility of partitioning the GVH-reactive cells o f a nonimmune population on a monolayer was also tested because of the possibility that the immune population might include suppressor cells which could specificially inhibit the GVH-reactive cells. This suppressor activity could conceivably be a property of the cytolytic cells o r could be attributable t o a separate class of cells which also partitioned according t o antigen recognition. Thus, there might be numerical enrichment and depletion of GVH-reactive cells in the adherent and nonadherent fractions, but this could be masked by t h e increased inhibitory effect of the enriched suppressor cells and the decreased inhibitory effect due t o suppressor cell depletion. In fact, there was again n o detectable difference in the GVH activity of adherent, nonadherent and unfractionated cells from nonimmune LN (Fig. 6). In these circumstances this could not have been due t o an interaction with cytolytic cells since none were present.
1
0.5
1- I
1.0
2.0
3.0
NUMBER OF CELLS INJECTED l x 1 0 ' 1
Figire 5. Cytolytic and GVH activity of lymphocytes adherent and nonadherent to allogeneic cell monolayers. LN cells from HO rats immunized with A 0 skin grafts were incubated on AO monolayers. Adherent and nonadherent cells were tested both for cytolytic activity (a) and GVH activity (b) against the immunizing MHC. Unfractionated adherent cells cells (0-0); nonadherent cells (A-A);
(0-O);syngeneic
Fl hybrid cells
A.
Another objection t o the conclusion that the monolayer system does not discriminate between GVH-reactive cells recognizing different antigens is that there may have been some enrichment in the adherent population, but this was masked by the dilution effect of an increased representation of,macrophages, B cells or dead cells. However, when nonimmune A 0 LN cells were separated on HO monolayers and then tested by GVH assay in ( A 0 x D A )FI hybrids, there was again no definite difference in GVH activity between adherent and nonadherent cells t o the "third party" antigenic monolayer, just as was the case when the test F1 hybrids expressed the same MHC as the monolayer (Fig. 6). Similarly, when immune A 0 LN cells were partitioned by adherence t o syngeneic monolayers, the GVH activity of the adherent and nonadherent fractions was equal, as it was in the case of antigenic mono-
442
Eur. J . immunol. 1978.8: 438-445
P. M. Chisholm and W. L. Ford
the lodgement of antigen-reactive cells in the spleen and their consequent loss from the recirculating lymphocyte pool. The physical separation of antigen-sensitive and other cells can be detected by labeling reactive and nonreactive populations with radioisotopes. By 24 h after injection, a substantial surplus of the reactive population is found in the spleen, and a corresponding deficit is obvious in the thoracic duct lymph of the recipient [ 171.
1.5
This technique was adapted t o study the partition of antigenreactive cells o n a monolayer. LN cells from HO rats which had been immunized with either A 0 or DA skin grafts were labeled separately in vitro with either [ 3H] Urd o r [ 14C]Urd, washed and then combined as shown in Scheme 1. Cell mixtures were then incubated on A 0 monolayers, and adherent and nonadherent cells were recovered as before for radioactive counting. The monolayers were also digested and counted so that the immune cells persistently adherent after EDTA treatment could also be examined.
1.o
-
-B I-
r
2
A 0 b I A 0 x HO I F1 0.5
P
E Y
CI
z
Scheme 1. Mixtures of different immune cell populations, labeled
n 2
F
1.0
0.6
3.0
2.0
4.0
2
5
t n
g
separately with 3H and 14C,were incubated on allogeneic monolayers. The proportions of HO anti-A0 and HO anti-DA cells which adhered to A 0 monolayers were indicated by measurements of the two isotopes in the nonadherent, adherent and persistently adherent (monolayer) fractions.
z
:
+
HO a n t i - A 0 CELLS
1.5
f
W
s
f
f
’4cx
’H
I
HO anti-DA C E L L S
v
’H
t
14c
H-HO a m - A 0
+ l4
C.HO anti-DA
I
[
’‘ C-HO anii.AO
MIXTURE A
t 3
H.HO antt.DA
I
MIXTURE B
1.0
0.5
A O F I A0 IDA
P
I
F1
NONADHERENT
ADHERENT
NONADHERENT
ADHERENT
CELLS
CELLS
CELLS
CELLS
COUNT’H 0.0
)Izocz.slI
1.o
A0
MONOLAYER
A0
2.0
$0
i o
NUMBER O F CELLS INJECTED I x 10‘ I
FCpure 6. Lack of specificity of nonimmune GVH-reactive cell adherence. LN cells from nonimmunized A 0 rats were incubated on HO monolayers, and adherent and nonadherent cells were tested for
GVH activity against the monolayer alloantigens (a) and against “third-party” alloantigens (b). Unfractionated cells ( 0 -0); nonadherent cells (A-4; adherent cells ( 0 - 0 . ) . There was no significant difference in the GVH activity of the different cell populations in either the ( A 0 x HO)F1 or ( A 0 x DA)F1 hybrids. layers. Thus, many o r all of t h e GVH-reactive cells which adhered to the monolayers did so for reasons which did n o t involve their antigen-specific receptor.
3.3 Proportions of radioactively labeled lymphocytes which adhered to allogeneic cell monolayers An early event in the systemic GVH reaction induced by intravenous injection of parental strain cells into an F , hybrid is
A N D 1 4 C IN
UNFRACTIONATED C E L L S NON - A D H E R E N T CELLS ADHERENT CELLS
MONOLAVER C E L L FRACTION
A balance sheet of the 3H and 14C counts present in different cell fractions of each mixture in a representative experiment is presented in Table 1A. As expected, more HO anti-A0 cells adhered t o the A 0 monolayer than did HO anti-DA cells, n o matter which were labeled with 3H. The anti-A0 cells were in surplus in both the fraction eluted with EDTA and the fraction which persistantly adhered t o the monolayer. All these points were consistently noted in each of the four experiments referred t o in Table 1 B. The mean surplus of immune cells in the adherent population accounted for 4.1 % of the radioactivity in the starting population. The same principle was applied t o alternatively labeled nonimmune LN cells which were either allogeneic or syngeneic for the monolayer o n which they were incubated. In contrast
Eur. J . Immunol. 1978.8: 438-445
Cytolytic and graft-vs.-host acitivities of alloiinmune cells
443
Table 1. Adherence of Urd-labeled immune LN cells to monolayers Mixture I
A. Balance sheet of typical experiment
Mixture I1 ‘4C-labeled HO anti-A0 cellsa) + 3H-labeled HO anti-DA cells
3H-labeled HO anti-A0 cellsa) + “%2-labeled HO anti-DA cells 3H dpm I4C dpm 1 346 000 (87.5) 125 000 Adherent cells (EDTA-eluted) (B) (%of A + B + C) (8.2) Adherent cells (retained on monolayers) (C) 66 440 (%of A + B + C ) (4.3) Total adherent cells 12.5 ( B + C) as %of (A + B + C )
Nonadherent cells (A) (%of A + B + C)
B. Summary of 4 similar experiments
Mean
k
S.E.
dpm
153 200 (91.6) 8 280 (4.9) 5 840 (3.5)
180 200 (87.1) 9 570 (4.6) 17 210 (8.3)
8.4
12.9
3H dpm
1 970 000 (94.4) 53 630 (2.6) 63 140 (3.0) 5.6
Radioactivity in fraction as %of total recovered HO anti-A0 cells adherent to A 0 monolayer (A)
HO anti-A0 cells adherent to DA monolayer (B)
10.8 * 0.9
6.7 f 0.7
Specifically adherent population (A-B)
4.1
f
Specific adherence index = A/B
*
0.6
1.63 0.10
a) LN cells from HO rats immunized with A 0 or DA skin pafts 8 days previously were incubated on A 0 TDL monolayers
Table 2. Adherence of Urd-labeled nonimmune LN cells to monolayersa)
t o t h e immune cells, the partition of t h e potentially reactive population and t h e nonreactive population between the nonadherent, adherent (EDTA-eluted) a n d persistently adherent fractions were identical (Table 2). We returned to immune LN cell, and repeated t h e experiments described in Fig. 7 with t w o i m m u n e populations labeled in v i m with L3H]dThd or [ 14C]dThd. This was intended t o confine t h e label t o t h e reacting population to a greater e x t e n t than was the case with [ 3H] Urd. The balance sheet of a typical experiment (Table 3A) and t h e summarized results of 4 experiments (Table 3B) showed that this intention was realized since t h e ratio of i m m u n e t o “nonimmune“ adhesion was increased from 1.7 t o a b o u t 3. T h e surplus o f adherent cells in the population which was immune t o t h e MHC of the mono-
Mean
f
Adherent to allogeneic monolayers
Adherent to syngeneic monolayers
(%)
(%)
4.5 8.0 6.3 5.2.f 1.5
S.E.
4.5 7.6 6.1 5.0 f 1.4
a) Mixtures of A 0 and HO LN cells were labeled with I3H] Urd and [I4C]Urd, then incubated on AO. TDL monolayers. Each figure is the sum of the EDTA-eluted and persistently adherent fractions and is calculated as the mean of a double experiment with 3H/14C swapping.
Table 3. Adherence of dThd-labeled immune LN cells to monolayers A. Balance sheet of typical experiment
Mixture I 3H-labele4i HO anti-A0 cellsa) + “JC-labeled HO anti-DA cells 3H dpm I4C dpm
Mixture I1 14C-labeled HO anti-A0 cellsa) + 3H-labeled HO anti-DA cells 1%
dpm
3H dprn
113 700 (73.5)
1 1 260 (94.7)
9 440 (76.3)
164 300 (94.5)
26 900 Adherent cells (EDTA-eluted) (B) (17.4) (%of A + B + C ) Adherent cells (retained o n monolayers) ( C ) 14 170 (9.1) (%of A + B + C) Total adherent cells 26.5 (B + C) as %of ( A + B + C)
4 00 (3.4) 227 (1.9)
1 820 (14.7) 1105 (9.0)
4 650 (2.6) 5 070 (2.9)
23.7
5.5
Nonadherent cells (A) (%of A + B + C )
B. Summary of 4 similar experiments
5.3
Radioactivity in fraction as %of total recovered HO anti-A0 cells Specifically adherent to DA adherent population monolayer (B) (A-B)
HO anti-AO cells adherent t o A 0 monolayer (A) Mean
f
S.E.
29.6 f 5.4
10.0 * 2.7
19.6 f 3.6
Specific adherence index = A/B
3.36 1: 0.5 8
a) see Table 1.
444
Eur. J. Immunol. 1978.8: 438-445
P. M. Chisholm and W. L. Ford
layer, accounted for almost 20 %.of the radioactive dThd within the whole population (adherent and nonadherent).
4 Discussion
By fractionating alloimmune o r nonimmune LN cells o n the basis of their adherence t o a monolayer of allogeneic lymphocytes, a systematic study has been made of GVH activity, cytolytic activity and cell numbers as reflected by radioactive labeling in both adherent and nonadherent fractions. The simplest explanation of the results (summarized in Table 4) requires three assumptions: (a) in an immune population, GVH-reactive cells and cytolytic cells belong t o separate subsets; (b) although GVH-reactive cells may display surface receptors for alloantigens, these are not able to bring about partition in the in uitro conditions described here; and (c) the cytolytic and GVH-reactive cells do not interact in the functional tests. These assumptions are consistent with a great deal of recent work o n T cell subsets. In particular, the first assumption has been established by fractionating mouse T cells according to their Ly phenotype which has indicated that cytolytic cells and their precursors belong t o the Ly-2,3 subset, whereas GVH-reactive cells belong t o the Ly-1 and Ly-1,2,3 subsets [28]. Moreover, there is substantial evidence that the former recognize serologically defined antigens [ 291, whereas the latter respond t o other (lymphocyte-defined) antigens specified by genes in the I region of the H-2 complex (for review see [ 281). The successful partition of suppressor T cell activity (Ly-2,3 cells) by immobilized antigens, while helper T cell activity (Ly-1 cells) is unaffected under the same conditions, is closely analogous to the present observations [30, 311. Table 4. Summary of main results Immune LN cells Nonimmune LN cells Adherent Nonadherent Adherent Nonadherent cells cells cells cells (jitolytic acitivtya)
G V H activity&) Cells labeled withb) radioactive Urd
Cells labeled withb)
radioactive dThd
t
4
N.C.
N.C.
f t
4 4
N.D.~) N.C.d)
N.D. N.C.
N.C.
N .C.
N.D.
N.D.
a) Increased or decreased relative to unfractionated cells. b) Numbers increased or decreased relative to a nonreactive reference
population. c) N.D. = not done. d) N.C. = no change.
The explanation for the failure to partition GVH activity cannot be that it is impossible, because it has been established that different lymphocyte subsets respond to different MHC (a) by inducing suicide of DNA-synthesizing cells [ 321 with [3H]dThd, (b) by passaging lymphocytes from blood t o lymph through an F1 hybrid o r allogeneic recipient [8, 91 and (c) by using anti-idiotypic sera [ 331. Early in the systemic GVH initiated by injecting parental strain lymphocytes into F l hybrids, there is an efficient separation of the reacting minority of T lymphocytes which settle in the spleen, bone
marrow and LN and the nonreacting cells which continue t o recirculate. This separation is thought t o be a consequence of the engagement of receptors for antigen with the high concentration of antigenic cells in lymphoid tissue. Why does this not happen in uitro o n monolayers of TDL which express antigens determined by the MHC, including antigens like Ia, since B cells comprise one-third t o one-half of the TDL? The possibility of separating GVH-reactive cells by adherence has been pursued by several workers. At least two groups [ 13, 341 have noted some depletion of activity in the nonadherent population, but it has been much less profound than the depletion achieved in vivo, and some have applied less precise GVH assay systems than the one used here. Recently, Neefe and Sachs [35] observed that the nonadherent subset of immune cells were depleted of cytolytic activity while retaining those antigen-reactive cells which can acquire cytolytic activity in culture. An overall view is that under certain undefined conditions, fractionation of cells by adherence t o monolayers may weakly partition GVH activity, but under other conditions such as those described here, it is possible to partition cytolytic activity with moderate efficiency while not affecting GVH activity at all. The possible explanations for this are (a) the cytolytic cell expresses receptors which individually have a higher affinity for antigen than the GVH-reactive cell; (b) the cytolytic cell may have a greater avidity for the target cell because it has a higher density of antigen receptors o r because of the “stabilization of receptor units“ [ 111; (c) the cytolytic cell reacts t o the engagement of its receptors by a nonspecific increase in its adherence, whereas the GVH-reactive cells do not react this way at least under these conditions; Id) the difference lies in some property of the different antigens thought t o be recognized by t h e two types of reactive cells, and (e) the cytolytic cell is more adherent before encountering antigen, and antigen-specific binding supplements this t o make the total adherence t o allogeneic cells greater than is the case with GVH-reactive cells. The last can almost be ruled out because in the fraction of cells nonspecifically adherent t o monolayers, cytolytic activity was reduced, while GVH activity was not indicating that cytolytic cells were on average less adherent than were GVH-reactive cells. (This is not necessarily true of other antigenic cells which may be involved in uivo). That an active response by the cytotoxic cell increases its adherence seems very likely [36], but acceptance of this notion would not exclude t h e other possibilities. The results of the partition of cell-associated radioactivity fitted in well with the functional studies (Table 4). There was no evidence of any partition of nonimmune cells by a sensitive method which easily detects the sequestration of GVH-reactive cells in uivo [ 171. By contrast, among the majority of cells in the immune population which labeled with [3H]Urd there was a subset which adhered t o the monolayer because of antigen recognition. This subset accounted for about 4 % of the radioactivity in the population, but it must be emphasized that this cannot be equated with the proportion of the specifically adherent cells, because radioactive Urd labels small B lymphocytes very lightly and large lymphocytes somewhat more intensely than small T lymphocytes [ 371. When DNAsynthesizing cells were studied using radioactive dThd, the specifically adherent cells accounted for about 20 % of the label in the population. Unless the avidity of the responding cells for the monolayer varies around S phase, this suggests
Eur. J . Immunol. 1978.8: 438-445 that about 20 % of these cells are capable of adhesion. T h e cells were isolated from LN soon after the peak o f DNA synthesis in response t o a skin graft [ 381, and the 4 % difference in Urd-labeled cells may possibly be entirely accounted for by the partition of DNA-synthesizing cells. Cytolytic cells may only be a subset of specifically adherent cells since memory cells - defined as noncytolytic cells which can differentiate without proliferation into cytolytic cells within 2 3 h of an antigenic stimulus [39] - may also be specifically adherent. Since the specifically adherent T cells are llkely t o be more heavily labeled with radioactive Urd than t h e average rionadherent T cell, these results suggest that a m a x i m u m of 4 % o f T cells are specifically adherent, which is compatible with estimates of t h e frequency of cytolytic cells by other means [40-421. Henney argues convincingly that after optimal in vivo immunization t h e true frequency is only 1-2 % (421. This underlines the paradox that even after the proliferation involved in generating cytolytic cells during immunization, the frequency is somewhat lower than the frequency of GVH o r MLC-reactive cells in a nonimmune population. Other work has supported the simplest explanation for this, namely that the precursors of cytolytic cells in a nonimmune population are only a small proportion of t h e cells which proliferate in response t o t h e MHC (for review see [ 281).
Cytolytic and graft-vs.-host activities of alloimmune cells
445
10 Mage, M. G. and McHugh, L. L., J. Zmmunol. 1973.111: 652. 11 Clark, W. R. and Kimura, A. K., Transplantation 1973. 16: 110.
1 2 Rubin, B., Clin. Exp. Immunol. 1975. 20: 513. 1 3 Lonai, P., Eliraz, A., Wekerle, H. and Feldman, M., Transplantation 1973.15: 368. 14 Altman, A., Cohen, I. R. and Feldman, M., Cell. Immunol. 1973. 7: 134. 15 Wekerle, H., Lonai, P. and Feldman, M., Proc. Nut. Acad. Sci US 1972. 69: 1620. 16 Bonavida, B. and Kedar, E.,Nature 1974. 249: 658. 17 Atkins, R. C. and Ford, W. L., J. Exp. Med. 1975. 141: 664. 18 Festing, M. and Staats, J., Transplantation 1973. 1 6 : 221. 19 Chisholm, P. M., Smith, M. E., Sparshott, S. M. and Ford, W. L., Transplantation 1977.23: 470. 20 Ford, W. L., Burr, W. and Simonsen, M., Transplantation 1970. 10: 258. 21 Ford, W. L. in Weir, D. M. (Ed.), Handbook o f Experimental Immunology, 3rd Edit., Blackwell, Oxford 1978, chapter 23. 22 Stulting, R. D. and Berke, G., J. Exp. Med. 1973. 137: 932. 23 Kedar, E., Bonavida, B., Ortiz De Landazurai, M. and Fahey, J. L., J. Immunol. Methods 1974. 5: 97. 24 Golstein, P., Svedmyr, E. A. J . and Wigzell, H . , J. Exp. Med. 1971. 134: 1385. 25 Thursh, D. R. and Emeson, E. E., J. Exp. Med. 1972.135: 754.
The observation that in a responding LN, three times as many DNA-synthesizing cells adhere t o the homologous monolayer as adhere to a heterologous monolayer, suggests this as a method for confirming t h e antigenic specificity of a n ongoing immune response. The test is quicker t o perform than a cytoly tic test, and the selective adhesion may conceivably be detectable earlier after immunization than is cytolytic activity. In many clinical situations including renal transplantation [43] and several autoimmune processes [44], an increased number of activated lymphocytes are found in t h e blood. I t would be of obvious interest t o determine their antigenic specificity by selective adhesion.
26 Michie, D., in Wir, D. M. (Ed.), Handbook of Experimental Immunology, 2nd Edit., Blackwell, Oxford 1973, chapter 30.
We thank Ms. T. Aslan for excellent technical assistance and Dr. E. B. Bell for his advice.
33 Binz, H. and Wigzell, H., J. Exp. Med. 1975.142: 1231.
27 Binz, H. and Wigzell, H., a n t e m p . Top. Zmmunobiol. 1977. 7: 113. 28 Cantor, H. and Boyse, E., Contemp. Top. Immunobiol. 1977. 7: 47. 29 Alter, B. J., Schendel, D. J., Bach, M. L., Bach, F. H. and Klein, J., J. Exp. Med. 1971.137: 1303. 30 Okumura, K., Takemori, T., Tokuhisa, T. and Tada, T., J. Exp. Med. 1977.146: 1234. 31 Taniguchi, M. and Miller, J. F. A. P., J. Exp. Med. 1977. 146: 1450. 32 Zosche, D. C. and Bach, F. H., Science 1971. 172: 1350. 34 Mage, M. G. and McHugh, L. L., J. Immunol. 1975.115: 911.
Received February 11, 1978.
35 Neefe, J. R. and Sachs, D. H., J. Exp. Med. 1976.144: 996. 36 Henney, C. S. and Bubbers, J. E., J. Immunol. 1973. I1 I : 85.
5 References
37 Howard, J. C., Hunt, S. V. and Gowans, J. L., J. Exp. Med. 1972. 135: 200.
1 Dutton, R. W. and Mishell, R. I.,J Exp. Med. 1967. 126: 443.
38 Tilney, N. L. and Ford, W. L., Transplantation 1974. 1 7 : 12.
2 Ford, W. L., Simmonds, S. J . and Atkins, R. C., J. Exp. Med. 1975. 141: 681. 3 Binz, H . and Wigzell, H.,J. Exp. Med., 1975. 142: 1218. 4 Ford, W. L. and Simonsen, M., J. Exp. Med. 1971.133: 938. 5 Wilson, D. B. and Nowell, P. C., J. Exp. Med. 1971. 133: 442. 6 Brondz, B. D. Folk Biol. (Prague) 1968 14: 115.
39 MacDonald, H. R., Sordat, B., Cerottini, J. C. and Brunner, K. T., J. Exp. Med. 1975.142: 622. 40 Berke, G., Gabison, D. and Feldman, M., Eur. J. Immunol. 1975. 5 : 813. 41 Martz, E.,J. Zmmunol. 1975.125: 261. 42 Henney, C. S . , Contemp. Top. Immunobiol. 1977. 7: 245.
I Wigzell, H. and Anderson, B., J. Exp. Med. 1969.129: 23. 8 Ford, W. L. and Atkins, R. C.,Nature-New Biol. 1971. 234: 178. 9 Dorsch, S. and Roser, B. J., Aust. J. Exp. Biol. Med. Sci. 1974. 5 2 : 45.
44 Blarney, R. W., Bennett, R. C., Singer, A. M., Nicol, R. and Harrison, J., Br. J.Surg. 1975. 62: 863.
43 Bacon, P. A., Sewell, R. L. and Crowther, D., Clin. Exp. Zmmunol. 1975. 1 9 : 201.
P. M. Chisholm and W. L. Ford
Patricia M. ChisholmO and W. L. Ford Department of Pathology, University of Manchester Medical School
Eur. J. Immunol. 1978.8: 438-445
Selection of antigen-specific cells by adherence to allogeneic cell monolayers: cytolytic activity, graft-vs.-host activity and numbers of adherent and nonadherent cells" Rat lymph node cells taken at the peak of cytolytic activity following a skin allograft were separated into adherent and nonadherent fractions by incubation o n monolayers of thoracic duct lymphocytes either of the same strain as the graft donor or of an Ag-B different strain. In t h e face of a 3-fold enrichment of cytolytic activity in the adherent cells and a 3-fold depletion in the nonadherent cells there was no detectible partition of graft-vs.-host (GVH) activity. Supplementary experiments supported the simplest interpretation of this finding, namely that the antigen receptors on GVH-reactive cells did not influence their adherence in this system. Similarly, there was no partition of the GVH activity of nonimmune lymph node cells by adherence. Labeling lymph node cells with either radioactive uridine or thymidine in vitro, suggested that about 20 % o f DNA-synthesizing cells in the immune population adhered because .of antigen recognition.
1 Introduction
Since Dutton and Mishell in 1967 eliminated the subset of lymphocytes reactive t o sheep erythrocyte antigens [ 11 several techniques have been described by which subpopulations of lymphocytes can be either enriched for o r depleted of antigensensitive cells t o a particular antigen. Starting with either immune or nonimmune populations, positive and negative selection have been achieved for b o t h T and B cells. In the case of T cells reactive t o those alloantigens determined by the major histocompatibility complex (MHC), the situation is of particular interest because of t h e well-known anomalies of these reactions, especially the high proportion (around 5 %) [ 2 , 31 of nonimmune lymphocytes proliferating in response to each MHC and because immunization in vivo, while provoking cytolytic activity, fails t o increase more than marginally t h e number of cells which react to a second antigenic stimulus by proliferation [4, 51. The cytolytic T cells in an alloimmune population can be selectively depleted by exploiting their increased adhesion t o cells bearing the immunizing alloantigen as has frequently been described following t h e original observations of Brondz [6]. The principle involved is that of the Wigzell column [7] in which antigen is immobilized, and an antigen-specific subset is retained because of binding brought about by surface receptors for antigen. The same principle apparently operates [I 20421
in vivo when antigen-specific T lymphocyte; are selectively retained in lymphoid tissue during passage from blood t o lymph in an F1 hybrid [8] or allogeneic recipient [ 91. However, it has been anomalous and in a practical sense disappointing that attempts t o deplete mixed lymphocyte reaction (MLR) or GVH activity in vitro by adhesion t o immobilized antigenic cells have often been unsuccessful [ 10- 121. The most promising results have been obtained by permitting reactive populations t o adhere for several hours to fibroblast monolayers. Both GVH-reactive cells and the precursors of cytolytic cells were depleted in the nonadherent fraction [ 13- 16). However, the impression from the considerable literature on the use of cell monolayers as immunoabsorbants is that cytolytic cells are easier to partition than GVH-reactive cells. This was supported by measuring each function in samples of nonadherent cells in which it was found that cytolytic activity but not GVH activity was reduced [ l o ] . The main object of our experiments was to make quantitative estimates of the GVH and cytotoxic activities of both adherent and nonadherent cells in a system in which immune (or nonimmune) lymph node (LN) cells were incubated on allogeneic monolayers in order t o test the suspicion that the partition of cytolytic activity would be greater than the partition of GVH activity. A second object was t o estimate the proportions of immune and nonimmune T cells which adhered t o allogeneic monolayers through the agency of their antigen receptors. This was done by application of a radioisotopic labeling technique that has already been used to estimate the number of T cells responding in a systemic GVH reaction [ 171.
* Most of the work was supported by a grant from the Medical
'
Research Council G972/455B. Present address: Department of Immunology, Royal Postgraduate Medical School, Hammersmith Hospital, Du Cane Road, London W12, GB. P.M.C. was supported by a grant from the Manchester Central District Research Grants Committee.
Correspondence: William L. Ford, Department of Pathology, Stopford Building, Oxford Road, Manchester M13 9PL, GB Abbreviations: GVH: Graft-vs.-host MLR: Mixed lymphocyte reaction (responder activity) TDL: Thoracic duct lymphocytes MHC: Major histocompatibility complex LN:Lymph node PBS: Phosphate-bufferedsaline FCS: Fetal calf serum Urd: Uridine
dThd: Thymidine [ *25I]dUrd: 1xI-labeled deoxyuridine dpm: Disintegrations per minute
2 Materials and methods
2.1 Rats and their immunization Three highly inbred strains were used - A 0 (Ag-B2), HO (Ag-BS, synonym PVG/c) and DA (Ag-B4), [ 181. Appropriate F1 hybrids were used for the assay of GVH activity. HO rats were immunized b y single allografts of A 0 o r DA skin applied t o the flank. DA rats were similarly grafted with A 0 or HO skia. The draining axillary and brachial LN were removed at the eight-day peak of cytolytic activity and teased into a single cell suspension.
Eur. J. Immunol. 1978.8: 438-445
Cytolytic and graft-us.-hostactivities of alloimmune cells
439
2.2 Cytotoxic assay
3 Results
The cytotoxicity of immune LN cells was measured against the WF/C1 lymphoma cell line which, like t h e A 0 strain and th e strain of origin (Wistar Furth), expressed t h e Ag-B-2 MHC' [ 18, 191. A standard number ( 2 5 x 1 0 3 )of target cells which had previously been labeled with ['"I]dUrd were incubatcd with graded numbers of attacker cells for 1 8 h in Linbro microtiter plates. Details of the assay and evidence that it measured T cell-mediated cytolysis directed predominantly against Ag-B antigens have been published [ 191.
3.1 Cytotoxic activity of adherent and nonadherent cells
2 .3 Graft-vs.-host activity This was measured by the popliteal LN assay [ 201. Four hind feet o f F hybrids were injected with parental strain lymphocytes at each of two dose levels. The draining LN were removed and weighed 7 days after injection.
2.4 Monolayer preparation Thoracic duct lymphocytes (TDL) were collected by t h e method of Gowans [21]. An excess of lymphocytes ( 7 5 x l o 6 ) were allowed to settle o n a Falcon polystyrene tissue culture petri dish (Falcon Plastics, Oxnard, CA, 60 x 1 5 mm) which had been pretreated with a solution of poly-t-lysine (Sigma Chemical Co. St. Louis, MO) at 5 0 pg/ml as described by Stulting and Berke [22]. Rinsing off the surplus cells left a n unbroken monolayer from which few cells were lost during the subsequent fractionation procedure. Each monolayer consisted of about SO x 1 O6 lymphocytes.
By measuring the release of '''1 from a constant number of WF/GI target cells after incubation with 5 graded doses of immune LN cells it was possible to quantitatively compare the cytolytic activity of adherent and nonadherent cells with a sample of the starting population. The principle of a parallelline assay [ 261 was applied using the parallel segments of the release) regreslog dose response (percentage of specific ''I sion lines as exemplified in Fig. 1. In all experiments, the cytolytic activity of the nonadherent cells was reduced, and the activity of the adherent cells was increased as has been found by many others (for review see [27]. In the first six experiments, the cytolytic activity of the adherent cells was equivalent to that of 3.1 (f 0.55) times as many unfractionated cells, and the activity of nonadherent cells was equivalent to less than half as many (0.45 f 0.05) unfractionated cells. In the next four experiments, samples of the adherent, nonadherent and unfractionated populations were assayed simultaneously for both cytolytic and GVH activities. The relative cytolytic activities of the adherent and nonadherent populations in these four experiments were 3.0 0.05 and 0.28 f 0.10, respectively (Mean f S.E.).
/
2.5 Partition of adherent and nonadherent cells A suspension of 15 x 1 O6 LN cells in 2.5 ml of RPMI medium 1640 (Gibco, Grand Island, NY) with 10 % fetal calf serum (FCS) was incubated o n each monolayer for 30 min at 37 O C . Then the plates were centrifuged at 130 x g for 5 min. [23] and nonadherent cells were removed by shaking [24]. Most of the adherent cells were removed by addition o f 5 m~ EDTA in phosphate-buffered saline (PBS) with 1 0 % FCS followed by gentle shaking and rinsing.
2.6 Radioactive labeling In some experiments, LN cells were labeled in vitro with either [3H]Urd ( 7 pCi/ml) o r [14C]Urd (1 pCi/ml), and in other experiments they were labeled with either [3H]dThd (0.6 pCi/ml) or [14C]dThd (0.4 pCi/ml). The method of labeling and the choice of isotopes and spec. act. were precisely as previously described [ 17, 211. T h e principle of alternatively (3H/14C) labeling a population immune t o the antigens expressed by the monolayer and a population immune t o other antigens has been described [ 17, 251. The 3H and 14C activities of nonadherent and adherent cells detached after EDTA treatment were estimated in a Beckman scintillation counter [21]. Radioactivity in the cells persistently adherent t o t h e monolayer after EDTA treatment was estimated by digesting t h e residual cells in th e petri dish by direct addition of 2 N sodium hydroxide.
1.6
3.2
6.25
12.5
IMMUNE CELLS I x 10'
25.0
I
Figure 1. The specificity of cytolytic cell adherence. HO anti-A0 lymphocytes were incubated on either A 0 or DA monolayers, and
the adherent and nonadherent cells were tested for cytolytic activity. Unfractionated cells (0-0); cells adherent to specific (AO) monolayers ( 0 - 0 ) ; cells adherent to third-party (DA) monolayers ( 0 - - 0 ) ; cells nonadherent to specific monolayers (A-A); cells nonadherent to third-party monolayers (A- -A).
.
.
Neither the threefold enrichment of activity in t h e adherent population nor the two to fourfold depletion in the nonadherent population was improved by applying a number of variations of the technique including the use of allogeneic fibroblast monolayers [24]. However, the possibility that a minority of cytolytic cells are exceptional in that they are
Eur. J. Immunol. 1978.8: 438-445
P. M. Chisholm and W. L. Ford
440
not removable by adherence t o antigenic monolayers, was discounted by experiments of which an example is illustrated in Fig. 2 . Either 2 o r 3 successive incubations of the nonadherent cells always reduced cytolytic activity t o an undetectable level suggesting that most o r all cytolytic cells express receptors which can bind them t o allogeneic cells. Other workers have generally found that t h e depletion of cytolytic activity after a single adhesion step t o a n antigenic monolayer is only partial for reasons which remain unclear [e.g. 10, 231.
The specificity of the partition was further confirmed by the finding that the small minority of LN cells which were adherent t o the DA monolayer did not show the augmented activity of the population adherent t o A 0 monolayers; in fact, the cytolytic activity of these cells was reduced. It might be expected that the minority adhering for reasons other than antigen recognition would include most of the macrophages and damaged cells. However, subsequent experiments suggested that this might not be the only explanation for the reduced cytolytic activity.
60
40
20
W v)
2 1.25
2.5
5.0
10.0
20.0
Since the monolayers used as immunoabsorbants for cytolytic lymphocytes were of the same Ag-B type as the target cells, the possibility had t o be considered that the depletion of cytolytic activity in the nonadherent population might be partly attributable t o competition during the cytolytic assay of cells which had become detached from the monolayer. By prelabeling the TDL used t o make the monolayer with [3H]Urd in vitro, the proportion of detached monolayer cells in the nonadherent and adherent populations was estimated on autoradiographic counts t o be approximately 3 % and 15 %, respectively. The possibility that the 3 % contamination of the nonadherent fraction might account for the specific reduction in cytolytic activity was tested by measuring the inhibition of target cell lysis with either lymphocytes bearing the same Ag-B complex as the target cells (AO) o r unlabeled tumor target cells (WF/G1). As expected, it was found that the lysis of labeled WF/G1 target cells progressively decreased as increasing numbers of unlabeled tumor cells were added (Fig. 3). However, even when A 0 TDL were added t o the labeled target cells at a ratio of 4: 1 , no inhibition of lysis was detected (Fig. 4). The surface area of the tumor cells was nearly four times that of TDL, but this would only partly account for the difference in competitive inhibition. It was
(b)
Ab ( 2.3
4:l )
2:l 1
1:l I
I
l:Z )
7
1.25
2.5 5.0 10.0 20.0 IMMUNE CELLS ( x 10 5 ,
Figure 2. Successive adsorption of cytolytic lymphocytes o n specific allogeneic monolayers. Nonadherent cells were recovered from a monolayer and incubated on fresh monolayers for a second, then a third time. The three populations of nonadherent cells (a) [ 1, 2, 31 and adherent cells (b) [ 1 , 2, 31 were assayed for cytolytic activity. Unfractionated cells (0-0); nonadherent cells (A--A); adherent cells ( 0 - 0 ) . After 2 cycles of fractionation no activity remained in the nonadherent population.
The specificity of t h e absorption of cytolytic cells was tested by incubating lymphocytes from HO rats which had been immunized with A 0 skin grafts o n DA monolayers. The cells which were nonadherent t o DA monolayer-s had identical cytolytic activity t o that of the starting population (Fig. 1).
(1:4)
1.6
13.2402-
3.2
6.25
12.5
IMMUNE CELLS ( x 10
25.0
)
Figure 3. Inhibition of cytolysis by unlabeled tumor target cells. Incubation mixtures consisted of HO anti-WF LN cells, 25 x lo3 1251labeled WF/Gl tumor cells and variable numbers of unlabcled WF/GI cells. The ratio of labeled to unlabeled tumor cells is given in parentheses. Immune lymphocytes and labeled tumor cells (0-0); immune lymphocytes, labeled tumor cells and unlabeled tumor cells (0-
0).
Eur. J . Irnmunol. 1978.8: 438-445 0
Cytolytic and graft-us.-hostactivities of alloimmune cells
441
CYTOTOXIC ACTIVITY
UNLABELLEDTDL UNLABELLED TUMOUR CELLS
CONTROL
114
III Ii
0.62
1.25
24
50
UNLABELLED TARGET CELLS
I
1o.O
x 104
7
i
1.6
J
3.2
Figure 4. Inhibition of cytolysis by unlabeled target cells. The bars represent mean l z S I release r S.E. of triplicates after 18 h. Incubation mixtures consisted of 25 x l o 5 LN cells fIom HO rats immunized with WF ,;kin grafts, 25 x 103 l251-labeled WF/GI lymphoma cells and variahle numbers of unlabeled WF/G1 lymphoma cells ( M ) or
WI:.TDL
6.25
12.5
IMMUNE CELLS I I 10' I
G V H ACTIVITY
(El).
clear that the minor contamination of nonadherent cells with cells from the monolayer could not account for the specific reduction of cytolytic activity by "cold target inhibition"
3.2 Graft-us.-host activity of adherent and nonadherent cells In the face of a tenfold difference in cytolytic activity there was n o difference in GVH activity of adherent and nonadherent subsets, and the activity of each was at t h e same level as that of the starting population (Fig. 5 ) . Th e same result was found consistently, and when popliteal LN weights obtained in 4 experiments were pooled and composite dose response regression lines were calculated, the plots for adherent, rionadherent and unfractionated populations were almost exactly coincident ( n o t illustrated). There appeared to be n o partition of GVH-reactive lymphocytes under these experimental conditions. Since each plot was derived from 3 2 LN responses, a minor difference either way, about 10 %, would have been detectable. The feasibility of partitioning the GVH-reactive cells o f a nonimmune population on a monolayer was also tested because of the possibility that the immune population might include suppressor cells which could specificially inhibit the GVH-reactive cells. This suppressor activity could conceivably be a property of the cytolytic cells o r could be attributable t o a separate class of cells which also partitioned according t o antigen recognition. Thus, there might be numerical enrichment and depletion of GVH-reactive cells in the adherent and nonadherent fractions, but this could be masked by t h e increased inhibitory effect of the enriched suppressor cells and the decreased inhibitory effect due t o suppressor cell depletion. In fact, there was again n o detectable difference in the GVH activity of adherent, nonadherent and unfractionated cells from nonimmune LN (Fig. 6). In these circumstances this could not have been due t o an interaction with cytolytic cells since none were present.
1
0.5
1- I
1.0
2.0
3.0
NUMBER OF CELLS INJECTED l x 1 0 ' 1
Figire 5. Cytolytic and GVH activity of lymphocytes adherent and nonadherent to allogeneic cell monolayers. LN cells from HO rats immunized with A 0 skin grafts were incubated on AO monolayers. Adherent and nonadherent cells were tested both for cytolytic activity (a) and GVH activity (b) against the immunizing MHC. Unfractionated adherent cells cells (0-0); nonadherent cells (A-A);
(0-O);syngeneic
Fl hybrid cells
A.
Another objection t o the conclusion that the monolayer system does not discriminate between GVH-reactive cells recognizing different antigens is that there may have been some enrichment in the adherent population, but this was masked by the dilution effect of an increased representation of,macrophages, B cells or dead cells. However, when nonimmune A 0 LN cells were separated on HO monolayers and then tested by GVH assay in ( A 0 x D A )FI hybrids, there was again no definite difference in GVH activity between adherent and nonadherent cells t o the "third party" antigenic monolayer, just as was the case when the test F1 hybrids expressed the same MHC as the monolayer (Fig. 6). Similarly, when immune A 0 LN cells were partitioned by adherence t o syngeneic monolayers, the GVH activity of the adherent and nonadherent fractions was equal, as it was in the case of antigenic mono-
442
Eur. J . immunol. 1978.8: 438-445
P. M. Chisholm and W. L. Ford
the lodgement of antigen-reactive cells in the spleen and their consequent loss from the recirculating lymphocyte pool. The physical separation of antigen-sensitive and other cells can be detected by labeling reactive and nonreactive populations with radioisotopes. By 24 h after injection, a substantial surplus of the reactive population is found in the spleen, and a corresponding deficit is obvious in the thoracic duct lymph of the recipient [ 171.
1.5
This technique was adapted t o study the partition of antigenreactive cells o n a monolayer. LN cells from HO rats which had been immunized with either A 0 or DA skin grafts were labeled separately in vitro with either [ 3H] Urd o r [ 14C]Urd, washed and then combined as shown in Scheme 1. Cell mixtures were then incubated on A 0 monolayers, and adherent and nonadherent cells were recovered as before for radioactive counting. The monolayers were also digested and counted so that the immune cells persistently adherent after EDTA treatment could also be examined.
1.o
-
-B I-
r
2
A 0 b I A 0 x HO I F1 0.5
P
E Y
CI
z
Scheme 1. Mixtures of different immune cell populations, labeled
n 2
F
1.0
0.6
3.0
2.0
4.0
2
5
t n
g
separately with 3H and 14C,were incubated on allogeneic monolayers. The proportions of HO anti-A0 and HO anti-DA cells which adhered to A 0 monolayers were indicated by measurements of the two isotopes in the nonadherent, adherent and persistently adherent (monolayer) fractions.
z
:
+
HO a n t i - A 0 CELLS
1.5
f
W
s
f
f
’4cx
’H
I
HO anti-DA C E L L S
v
’H
t
14c
H-HO a m - A 0
+ l4
C.HO anti-DA
I
[
’‘ C-HO anii.AO
MIXTURE A
t 3
H.HO antt.DA
I
MIXTURE B
1.0
0.5
A O F I A0 IDA
P
I
F1
NONADHERENT
ADHERENT
NONADHERENT
ADHERENT
CELLS
CELLS
CELLS
CELLS
COUNT’H 0.0
)Izocz.slI
1.o
A0
MONOLAYER
A0
2.0
$0
i o
NUMBER O F CELLS INJECTED I x 10‘ I
FCpure 6. Lack of specificity of nonimmune GVH-reactive cell adherence. LN cells from nonimmunized A 0 rats were incubated on HO monolayers, and adherent and nonadherent cells were tested for
GVH activity against the monolayer alloantigens (a) and against “third-party” alloantigens (b). Unfractionated cells ( 0 -0); nonadherent cells (A-4; adherent cells ( 0 - 0 . ) . There was no significant difference in the GVH activity of the different cell populations in either the ( A 0 x HO)F1 or ( A 0 x DA)F1 hybrids. layers. Thus, many o r all of t h e GVH-reactive cells which adhered to the monolayers did so for reasons which did n o t involve their antigen-specific receptor.
3.3 Proportions of radioactively labeled lymphocytes which adhered to allogeneic cell monolayers An early event in the systemic GVH reaction induced by intravenous injection of parental strain cells into an F , hybrid is
A N D 1 4 C IN
UNFRACTIONATED C E L L S NON - A D H E R E N T CELLS ADHERENT CELLS
MONOLAVER C E L L FRACTION
A balance sheet of the 3H and 14C counts present in different cell fractions of each mixture in a representative experiment is presented in Table 1A. As expected, more HO anti-A0 cells adhered t o the A 0 monolayer than did HO anti-DA cells, n o matter which were labeled with 3H. The anti-A0 cells were in surplus in both the fraction eluted with EDTA and the fraction which persistantly adhered t o the monolayer. All these points were consistently noted in each of the four experiments referred t o in Table 1 B. The mean surplus of immune cells in the adherent population accounted for 4.1 % of the radioactivity in the starting population. The same principle was applied t o alternatively labeled nonimmune LN cells which were either allogeneic or syngeneic for the monolayer o n which they were incubated. In contrast
Eur. J . Immunol. 1978.8: 438-445
Cytolytic and graft-vs.-host acitivities of alloiinmune cells
443
Table 1. Adherence of Urd-labeled immune LN cells to monolayers Mixture I
A. Balance sheet of typical experiment
Mixture I1 ‘4C-labeled HO anti-A0 cellsa) + 3H-labeled HO anti-DA cells
3H-labeled HO anti-A0 cellsa) + “%2-labeled HO anti-DA cells 3H dpm I4C dpm 1 346 000 (87.5) 125 000 Adherent cells (EDTA-eluted) (B) (%of A + B + C) (8.2) Adherent cells (retained on monolayers) (C) 66 440 (%of A + B + C ) (4.3) Total adherent cells 12.5 ( B + C) as %of (A + B + C )
Nonadherent cells (A) (%of A + B + C)
B. Summary of 4 similar experiments
Mean
k
S.E.
dpm
153 200 (91.6) 8 280 (4.9) 5 840 (3.5)
180 200 (87.1) 9 570 (4.6) 17 210 (8.3)
8.4
12.9
3H dpm
1 970 000 (94.4) 53 630 (2.6) 63 140 (3.0) 5.6
Radioactivity in fraction as %of total recovered HO anti-A0 cells adherent to A 0 monolayer (A)
HO anti-A0 cells adherent to DA monolayer (B)
10.8 * 0.9
6.7 f 0.7
Specifically adherent population (A-B)
4.1
f
Specific adherence index = A/B
*
0.6
1.63 0.10
a) LN cells from HO rats immunized with A 0 or DA skin pafts 8 days previously were incubated on A 0 TDL monolayers
Table 2. Adherence of Urd-labeled nonimmune LN cells to monolayersa)
t o t h e immune cells, the partition of t h e potentially reactive population and t h e nonreactive population between the nonadherent, adherent (EDTA-eluted) a n d persistently adherent fractions were identical (Table 2). We returned to immune LN cell, and repeated t h e experiments described in Fig. 7 with t w o i m m u n e populations labeled in v i m with L3H]dThd or [ 14C]dThd. This was intended t o confine t h e label t o t h e reacting population to a greater e x t e n t than was the case with [ 3H] Urd. The balance sheet of a typical experiment (Table 3A) and t h e summarized results of 4 experiments (Table 3B) showed that this intention was realized since t h e ratio of i m m u n e t o “nonimmune“ adhesion was increased from 1.7 t o a b o u t 3. T h e surplus o f adherent cells in the population which was immune t o t h e MHC of the mono-
Mean
f
Adherent to allogeneic monolayers
Adherent to syngeneic monolayers
(%)
(%)
4.5 8.0 6.3 5.2.f 1.5
S.E.
4.5 7.6 6.1 5.0 f 1.4
a) Mixtures of A 0 and HO LN cells were labeled with I3H] Urd and [I4C]Urd, then incubated on AO. TDL monolayers. Each figure is the sum of the EDTA-eluted and persistently adherent fractions and is calculated as the mean of a double experiment with 3H/14C swapping.
Table 3. Adherence of dThd-labeled immune LN cells to monolayers A. Balance sheet of typical experiment
Mixture I 3H-labele4i HO anti-A0 cellsa) + “JC-labeled HO anti-DA cells 3H dpm I4C dpm
Mixture I1 14C-labeled HO anti-A0 cellsa) + 3H-labeled HO anti-DA cells 1%
dpm
3H dprn
113 700 (73.5)
1 1 260 (94.7)
9 440 (76.3)
164 300 (94.5)
26 900 Adherent cells (EDTA-eluted) (B) (17.4) (%of A + B + C ) Adherent cells (retained o n monolayers) ( C ) 14 170 (9.1) (%of A + B + C) Total adherent cells 26.5 (B + C) as %of ( A + B + C)
4 00 (3.4) 227 (1.9)
1 820 (14.7) 1105 (9.0)
4 650 (2.6) 5 070 (2.9)
23.7
5.5
Nonadherent cells (A) (%of A + B + C )
B. Summary of 4 similar experiments
5.3
Radioactivity in fraction as %of total recovered HO anti-A0 cells Specifically adherent to DA adherent population monolayer (B) (A-B)
HO anti-AO cells adherent t o A 0 monolayer (A) Mean
f
S.E.
29.6 f 5.4
10.0 * 2.7
19.6 f 3.6
Specific adherence index = A/B
3.36 1: 0.5 8
a) see Table 1.
444
Eur. J. Immunol. 1978.8: 438-445
P. M. Chisholm and W. L. Ford
layer, accounted for almost 20 %.of the radioactive dThd within the whole population (adherent and nonadherent).
4 Discussion
By fractionating alloimmune o r nonimmune LN cells o n the basis of their adherence t o a monolayer of allogeneic lymphocytes, a systematic study has been made of GVH activity, cytolytic activity and cell numbers as reflected by radioactive labeling in both adherent and nonadherent fractions. The simplest explanation of the results (summarized in Table 4) requires three assumptions: (a) in an immune population, GVH-reactive cells and cytolytic cells belong t o separate subsets; (b) although GVH-reactive cells may display surface receptors for alloantigens, these are not able to bring about partition in the in uitro conditions described here; and (c) the cytolytic and GVH-reactive cells do not interact in the functional tests. These assumptions are consistent with a great deal of recent work o n T cell subsets. In particular, the first assumption has been established by fractionating mouse T cells according to their Ly phenotype which has indicated that cytolytic cells and their precursors belong t o the Ly-2,3 subset, whereas GVH-reactive cells belong t o the Ly-1 and Ly-1,2,3 subsets [28]. Moreover, there is substantial evidence that the former recognize serologically defined antigens [ 291, whereas the latter respond t o other (lymphocyte-defined) antigens specified by genes in the I region of the H-2 complex (for review see [ 281). The successful partition of suppressor T cell activity (Ly-2,3 cells) by immobilized antigens, while helper T cell activity (Ly-1 cells) is unaffected under the same conditions, is closely analogous to the present observations [30, 311. Table 4. Summary of main results Immune LN cells Nonimmune LN cells Adherent Nonadherent Adherent Nonadherent cells cells cells cells (jitolytic acitivtya)
G V H activity&) Cells labeled withb) radioactive Urd
Cells labeled withb)
radioactive dThd
t
4
N.C.
N.C.
f t
4 4
N.D.~) N.C.d)
N.D. N.C.
N.C.
N .C.
N.D.
N.D.
a) Increased or decreased relative to unfractionated cells. b) Numbers increased or decreased relative to a nonreactive reference
population. c) N.D. = not done. d) N.C. = no change.
The explanation for the failure to partition GVH activity cannot be that it is impossible, because it has been established that different lymphocyte subsets respond to different MHC (a) by inducing suicide of DNA-synthesizing cells [ 321 with [3H]dThd, (b) by passaging lymphocytes from blood t o lymph through an F1 hybrid o r allogeneic recipient [8, 91 and (c) by using anti-idiotypic sera [ 331. Early in the systemic GVH initiated by injecting parental strain lymphocytes into F l hybrids, there is an efficient separation of the reacting minority of T lymphocytes which settle in the spleen, bone
marrow and LN and the nonreacting cells which continue t o recirculate. This separation is thought t o be a consequence of the engagement of receptors for antigen with the high concentration of antigenic cells in lymphoid tissue. Why does this not happen in uitro o n monolayers of TDL which express antigens determined by the MHC, including antigens like Ia, since B cells comprise one-third t o one-half of the TDL? The possibility of separating GVH-reactive cells by adherence has been pursued by several workers. At least two groups [ 13, 341 have noted some depletion of activity in the nonadherent population, but it has been much less profound than the depletion achieved in vivo, and some have applied less precise GVH assay systems than the one used here. Recently, Neefe and Sachs [35] observed that the nonadherent subset of immune cells were depleted of cytolytic activity while retaining those antigen-reactive cells which can acquire cytolytic activity in culture. An overall view is that under certain undefined conditions, fractionation of cells by adherence t o monolayers may weakly partition GVH activity, but under other conditions such as those described here, it is possible to partition cytolytic activity with moderate efficiency while not affecting GVH activity at all. The possible explanations for this are (a) the cytolytic cell expresses receptors which individually have a higher affinity for antigen than the GVH-reactive cell; (b) the cytolytic cell may have a greater avidity for the target cell because it has a higher density of antigen receptors o r because of the “stabilization of receptor units“ [ 111; (c) the cytolytic cell reacts t o the engagement of its receptors by a nonspecific increase in its adherence, whereas the GVH-reactive cells do not react this way at least under these conditions; Id) the difference lies in some property of the different antigens thought t o be recognized by t h e two types of reactive cells, and (e) the cytolytic cell is more adherent before encountering antigen, and antigen-specific binding supplements this t o make the total adherence t o allogeneic cells greater than is the case with GVH-reactive cells. The last can almost be ruled out because in the fraction of cells nonspecifically adherent t o monolayers, cytolytic activity was reduced, while GVH activity was not indicating that cytolytic cells were on average less adherent than were GVH-reactive cells. (This is not necessarily true of other antigenic cells which may be involved in uivo). That an active response by the cytotoxic cell increases its adherence seems very likely [36], but acceptance of this notion would not exclude t h e other possibilities. The results of the partition of cell-associated radioactivity fitted in well with the functional studies (Table 4). There was no evidence of any partition of nonimmune cells by a sensitive method which easily detects the sequestration of GVH-reactive cells in uivo [ 171. By contrast, among the majority of cells in the immune population which labeled with [3H]Urd there was a subset which adhered t o the monolayer because of antigen recognition. This subset accounted for about 4 % of the radioactivity in the population, but it must be emphasized that this cannot be equated with the proportion of the specifically adherent cells, because radioactive Urd labels small B lymphocytes very lightly and large lymphocytes somewhat more intensely than small T lymphocytes [ 371. When DNAsynthesizing cells were studied using radioactive dThd, the specifically adherent cells accounted for about 20 % of the label in the population. Unless the avidity of the responding cells for the monolayer varies around S phase, this suggests
Eur. J . Immunol. 1978.8: 438-445 that about 20 % of these cells are capable of adhesion. T h e cells were isolated from LN soon after the peak o f DNA synthesis in response t o a skin graft [ 381, and the 4 % difference in Urd-labeled cells may possibly be entirely accounted for by the partition of DNA-synthesizing cells. Cytolytic cells may only be a subset of specifically adherent cells since memory cells - defined as noncytolytic cells which can differentiate without proliferation into cytolytic cells within 2 3 h of an antigenic stimulus [39] - may also be specifically adherent. Since the specifically adherent T cells are llkely t o be more heavily labeled with radioactive Urd than t h e average rionadherent T cell, these results suggest that a m a x i m u m of 4 % o f T cells are specifically adherent, which is compatible with estimates of t h e frequency of cytolytic cells by other means [40-421. Henney argues convincingly that after optimal in vivo immunization t h e true frequency is only 1-2 % (421. This underlines the paradox that even after the proliferation involved in generating cytolytic cells during immunization, the frequency is somewhat lower than the frequency of GVH o r MLC-reactive cells in a nonimmune population. Other work has supported the simplest explanation for this, namely that the precursors of cytolytic cells in a nonimmune population are only a small proportion of t h e cells which proliferate in response t o t h e MHC (for review see [ 281).
Cytolytic and graft-vs.-host activities of alloimmune cells
445
10 Mage, M. G. and McHugh, L. L., J. Zmmunol. 1973.111: 652. 11 Clark, W. R. and Kimura, A. K., Transplantation 1973. 16: 110.
1 2 Rubin, B., Clin. Exp. Immunol. 1975. 20: 513. 1 3 Lonai, P., Eliraz, A., Wekerle, H. and Feldman, M., Transplantation 1973.15: 368. 14 Altman, A., Cohen, I. R. and Feldman, M., Cell. Immunol. 1973. 7: 134. 15 Wekerle, H., Lonai, P. and Feldman, M., Proc. Nut. Acad. Sci US 1972. 69: 1620. 16 Bonavida, B. and Kedar, E.,Nature 1974. 249: 658. 17 Atkins, R. C. and Ford, W. L., J. Exp. Med. 1975. 141: 664. 18 Festing, M. and Staats, J., Transplantation 1973. 1 6 : 221. 19 Chisholm, P. M., Smith, M. E., Sparshott, S. M. and Ford, W. L., Transplantation 1977.23: 470. 20 Ford, W. L., Burr, W. and Simonsen, M., Transplantation 1970. 10: 258. 21 Ford, W. L. in Weir, D. M. (Ed.), Handbook o f Experimental Immunology, 3rd Edit., Blackwell, Oxford 1978, chapter 23. 22 Stulting, R. D. and Berke, G., J. Exp. Med. 1973. 137: 932. 23 Kedar, E., Bonavida, B., Ortiz De Landazurai, M. and Fahey, J. L., J. Immunol. Methods 1974. 5: 97. 24 Golstein, P., Svedmyr, E. A. J . and Wigzell, H . , J. Exp. Med. 1971. 134: 1385. 25 Thursh, D. R. and Emeson, E. E., J. Exp. Med. 1972.135: 754.
The observation that in a responding LN, three times as many DNA-synthesizing cells adhere t o the homologous monolayer as adhere to a heterologous monolayer, suggests this as a method for confirming t h e antigenic specificity of a n ongoing immune response. The test is quicker t o perform than a cytoly tic test, and the selective adhesion may conceivably be detectable earlier after immunization than is cytolytic activity. In many clinical situations including renal transplantation [43] and several autoimmune processes [44], an increased number of activated lymphocytes are found in t h e blood. I t would be of obvious interest t o determine their antigenic specificity by selective adhesion.
26 Michie, D., in Wir, D. M. (Ed.), Handbook of Experimental Immunology, 2nd Edit., Blackwell, Oxford 1973, chapter 30.
We thank Ms. T. Aslan for excellent technical assistance and Dr. E. B. Bell for his advice.
33 Binz, H. and Wigzell, H., J. Exp. Med. 1975.142: 1231.
27 Binz, H. and Wigzell, H., a n t e m p . Top. Zmmunobiol. 1977. 7: 113. 28 Cantor, H. and Boyse, E., Contemp. Top. Immunobiol. 1977. 7: 47. 29 Alter, B. J., Schendel, D. J., Bach, M. L., Bach, F. H. and Klein, J., J. Exp. Med. 1971.137: 1303. 30 Okumura, K., Takemori, T., Tokuhisa, T. and Tada, T., J. Exp. Med. 1977.146: 1234. 31 Taniguchi, M. and Miller, J. F. A. P., J. Exp. Med. 1977. 146: 1450. 32 Zosche, D. C. and Bach, F. H., Science 1971. 172: 1350. 34 Mage, M. G. and McHugh, L. L., J. Immunol. 1975.115: 911.
Received February 11, 1978.
35 Neefe, J. R. and Sachs, D. H., J. Exp. Med. 1976.144: 996. 36 Henney, C. S. and Bubbers, J. E., J. Immunol. 1973. I1 I : 85.
5 References
37 Howard, J. C., Hunt, S. V. and Gowans, J. L., J. Exp. Med. 1972. 135: 200.
1 Dutton, R. W. and Mishell, R. I.,J Exp. Med. 1967. 126: 443.
38 Tilney, N. L. and Ford, W. L., Transplantation 1974. 1 7 : 12.
2 Ford, W. L., Simmonds, S. J . and Atkins, R. C., J. Exp. Med. 1975. 141: 681. 3 Binz, H . and Wigzell, H.,J. Exp. Med., 1975. 142: 1218. 4 Ford, W. L. and Simonsen, M., J. Exp. Med. 1971.133: 938. 5 Wilson, D. B. and Nowell, P. C., J. Exp. Med. 1971. 133: 442. 6 Brondz, B. D. Folk Biol. (Prague) 1968 14: 115.
39 MacDonald, H. R., Sordat, B., Cerottini, J. C. and Brunner, K. T., J. Exp. Med. 1975.142: 622. 40 Berke, G., Gabison, D. and Feldman, M., Eur. J. Immunol. 1975. 5 : 813. 41 Martz, E.,J. Zmmunol. 1975.125: 261. 42 Henney, C. S . , Contemp. Top. Immunobiol. 1977. 7: 245.
I Wigzell, H. and Anderson, B., J. Exp. Med. 1969.129: 23. 8 Ford, W. L. and Atkins, R. C.,Nature-New Biol. 1971. 234: 178. 9 Dorsch, S. and Roser, B. J., Aust. J. Exp. Biol. Med. Sci. 1974. 5 2 : 45.
44 Blarney, R. W., Bennett, R. C., Singer, A. M., Nicol, R. and Harrison, J., Br. J.Surg. 1975. 62: 863.
43 Bacon, P. A., Sewell, R. L. and Crowther, D., Clin. Exp. Zmmunol. 1975. 1 9 : 201.
