366

Eur. J . Immunol. 1977. 7: 366-370

K.S. Zier, C. Huber and H. Braunsteiner

16 Goldschneider, I. and McCregor, D.D., J. Exp. Med. 1973. 138: 1443. 17 Ishii, Y., Koshiba, H.,Yamaoka, H. and Kikuchi, K., J. Immunol. 1976. I 17: 497. 18 Watanabe, T., Yagi, Y. and Pressman, D., J. Immunol. 1971. 106: 1213. 19 Yutoku, M.,Grossberg, A.L. and Pressman, D., J. Immunol. 1974. 112: 911.

Karen S. Zier, Ch. Huber and H. Braunsteiner Department of Internal Medicine, University of Innsbruck, lnnsbruck

20 Zeiller, K. and Pascher, G., Eur. J. Immunol. 1973. 3: 614. 21 Ochiai, T., Ahmed, A,, Strong, D.M., Scher, I . and Sell. K.W.. Transplantation 1975. 20: 198. 22 Trowbridge, IS., Ralph, P. and Bevan, M.J., Proc. Nut. Acad. Sci. US 1975. 75: 157. 23 Trowbridge, I.S. and Mazauskas, C., Eur. J. Immunol. 1976.6: 557 24 Barclay, A.N., Letarte-Muirhead, M. and Williams, A.F.. Biochem. J. 1975.151: 699.

Linear density gradient separation of human lymphocyte subsets

1. Analysis by mixed leukocyte culture and cell-mediated lympholysis responses" Linear density gradients were used t o separate either unsensitized human peripheral blood lymphocytes or cells sensitized in mixed leukocyte culture (MLC), at different time points in the immune response. Testing of cells from the various density fractions for their ability t o respond in MLC and cellmediated lympholysis (CML) revealed that (a) activity between individual fractions differed, as well as activity between individual fractions and a suspension of unfractionated cells, ( b ) although precursor cells o f MLC and CML were maximally enriched in closely associated light density regions, their respective distribution profiles were strikingly different, and (c) the density of cells responding in MLC and CML differed depending upon the point of the immune response a t which the cells were separated. Whereas effector cells were of light density relative t o the precursor cells, putative memory cells for CML exhibited a high density.

1. Introduction

The separation of a population of lymphocytes into fractions with different immunological reactivities precedes an understanding of the interrelationship between different subsets. Shortman et al. have reported that distinct subsets of lymphocytes differing in functional capacity and/or maturity can be separated o n linear density gradients (LDG) [ 1-31. Others have used such gradients to separate mouse o r rat cells that differ in their responses to mitogens [4], their ability t o function in mixed leukocyte culture (MLC) and graft-versus-host (GvH) reactions, o r t o serve as splenic colony-forming units [ 51, o r their ability to produce antibody [6]. Details concerning the conditions required for density separation, including those for human cells, have recently been published [7, 81.

in the MLC and cell-mediated lympholysis (CML) assays [ 9- I I ] , we decided t o use LDG separation to answer two main questions. (a) Do precursor cells of MLC and CML-reactive cells exhibit differences in their density distribution profile? (b) Does the density distribution profile of these populations change upon stimulation? Our results indicate that precursor cells for MLC and CML in the blood of unsensitized humans d o differ in their density distribution profiles. Moreover, marked changes in the density distribution profiles of these t w o cell populations have been observed following in vitro sensitization in MLC. Whereas precursor cells, as well as effector cells, for MLC and CML are enriched in the low density regions, by day 1 4 fractions exhibiting residual cytotoxic activity were only weakly proliferating and of high density.

Based upon results in both man and mouse which support the concept that different T cell subsets are primarily responsive

2. Materials and methods [I 16191

* This work was supported by Austrian Fonds ZUT Forderung der wissenschaftlichen Forschung, Project No. 3174. Correspondence: Karen Zier, Department of Internal Medicine, University of lnnsbruck, A-6020 lnnsbruck, Austria Abbreviations: LDG: Linear density gradient PBL: Peripheral blood lymphocytes MLC: Mixed leukocyte culture CML: Cell-mediated lyrnpholysis GvH: Craft-versus-host [SHldThd: Tritiated thymidine LU: Lytic unit

2.1. Enrichment of

T lymphocytes by column techniques

PBL were preselected by using standard techniques [ 121. In order t o study density distributions of T cells, B cells were removed either by passage of lymphocytes through Ig-anti-Ig columns, according t o the method of Wigzell [ 131, or by passage through nylon wool columns [ 141. The vast majority of the experiments involved the former method. In experiments using nylon wool-selected cells, distribution patterns

Eur. J. Immunol. 1977. 7: 366-370

similar to those obtained with Ig-anti-Ig columns were observed. The percentage of T cells was evaluated b y spontaneous E rosette formation with neuraminidase-treated (Behringwerke, Marburg, FRG) sheep red blood cells [ 151. This resulted in an average 25 % increase of E rosette-forming cells.

2.2. MLC Mixed cultures were performed by culturing an equal number of responding 'A' cells with allogeneic mitomycin-C-treated (Kyowa, Tokyo, Japan) 'Bm' cells in medium RPMI 1640 containing 25 mM HEPES buffer (GIBCO no. 240, Gibco, Grand Island, N.Y.)and supplemented with 10-20 % male pool serum and antibiotics. For cultures carried out in roundbottom microtiter plates (Creiner, Nurtingen, FRG), a total of 1 x 105 cellslwell were contained in a volume of 0.2 ml. Bulk cultures were performed in flasks (Falcon no. 301 3, Falcon Plastics, Oxnard, Calif.) containing a total of 20 x lo6 cells in 2 0 ml medium. The cells were incubated in 5 % COz at 37 'C. Cells in microtiter plates were labeled o n either day 5 or 6 with 2 pCi tritiated thymidine (['HIdThd) (specific activity 2 Ci/mmole, Radiochemical Center, Amersham, England) for 1 2 h t o assess the proliferative response as measured in MLC. Cells were precipitated o n t o glass fiber filters by means of a cell harvester (Otto Hiller, Madison,,Wisc.), and incorporated radioactivity assessed in a liquid scintillation counter (Packard, La Grange, Ill.). On days 6 and 14 sensitized cells from t h e flasks were separated on the gradient as detailed below.

2.3. CML Cells from day-6 microtiter cultures to be tested for killer activity were pooled, centrifuged and resuspended in the same amount of fresh medium for counting. Alternatively, sensitized cells from flasks were separated o n the gradient and the fractions then tested for killer activity. Serial dilutions were made and 0.1 ml 51 Cr-labeled target cells was added t o triplicate samples (NazCr04 in saline, 5 mCi/ml, specific activity 200-400 Ci/mmole, Eidg. lnstitut fiir Reaktorforschung, Wiirenlingen, Switzerland). Unstimulated lymphocytes cultured for 6 days served as target cells. The plates were centrifuged for 5 min at 50 x g and incubated 4-6 h in 5 % COz at 37 'C. For harvesting they were centrifuged at 400 x g for 10 min. The supernatant was aspirated and then counted in a well-type y-counter (Nuclear Chicago, Des Plaines, 111.). The cytotoxic activity was expressed as lytic units (LU) per fraction, the value of which is calculated from the formula [ 161: LU/culture =

total number of czlls/culture number of cells required for x % lysis

In gradient experiments the different fractions are regarded as representing individual cultures. The percent cytotoxicity was calculated from the formula: experimental release - spontaneous release x 100. maximum release - spontaneous release The experimental release represents cpm released from slCrlabeled target cells incubated with effector cells. The spontaneous and maximum releases equal the mean cpm released from triplicate samples of target cells incubated with medium or lysed with 2.5 % Nonidet-P4O (Shell), respectively.

Density heterogeneity of human lymphocyte subsets 2.4. LDG separation of unsensitized

367

PBL

Linear gradients were prepared from high (1.095 g/ml) and low (1.055 g/ml) density solutions of Ficoll-Ronpacon (CilagChemie, Vienna) in salt solution, using a modification of the method of Roos and Loos [8], described by Hayry and A n d e r s o n in detail [4]. Fifty x 1 O6 t o 100 x lo6 unsensitized cells were layered o n t o p of t h e LDG in 2 ml medium 199 supplemented with 1 % human albumin (Dade, Miami, Fla.). They were centrifuged (Hettich Rotanta/S, Tuttlingen, FRG) at either 400 or 2500 x g a t room temperature for 20-30 min. Ten min were allowed for deceleration. One ml fractions were collected into 15 ml conical-bottom, glass centrifuge tubes b y upward displacement of the gradient with a 50 % sucrose solution. Fractions were diluted in medium 199, washed and distributed into microtiter plates t o serve as responding cells in MLC.

2.5. LDG separation of MLC-sensitized lymphocytes ABm cells were cultured in upright flasks containing 2 0 ml medium. On days 6 and 14 part of the sensitized cells were pooled, counted and separated on the gradient as for unsensitized cells. Serial dilutions were made of cells from each fraction t o serve as effectors in CML. Samples of equal volume from these suspensions were transfered to microtiter plates and labeled with ['HIdThd overnight t o measure their prolifer at ive capacity. MLC samples of the cells o n both day 6 and 1 4 were compared o n a volume basis so as t o reflect the difference in proliferative activity between the fractions o n a population basis at t h e two time points of the response. Cytotoxic activity between individual fractions obtained o n either day 6 or 14 were compared t o determine areas enriched for killer cell activity.

3. Results 3.1. Density distribution profile of unsensitized and sensitized lymphocytes Unsensitized PBL, enriched for T cells, were separated o n LDG. The gradients were characterized b y refractive index determination, which established linearity (Fig. l ) , by centrifugation at different speeds, which demonstrated that equilibrium was reached at 400 x g, and b y rebanding isolated fractions, which confirmed the reproducibility of the separation (unpublished data). The results of one representative experiment showing the density distribution of unsensitized cells, are presented in Fig. 1. The density of the first two fractions did not lie on the linear portion of the curve, due t o admixture of the medium in which the cells were resuspended. In this experiment the majority of cells were recovered in fraction 9 , corresponding to a density of 1.070 g/ml, with cells ranging in density from 1.056 g/ml to 1.082 g/ml. In 8 o u t of 13 experiments the peak number of cells was recovered in fraction 9, 2 o u t of 1 3 in fraction 8 and 1 o u t of 13 in fraction 10. This gives a mean fraction of 8.62 -+ 0.87.

Eur. J. Immunol. 1977. 7: 366-370

K.S.Zier, C. Huber and H. Braunsteiner

368

.

Refractive l n b i

cells I lraction 151

im -

- 1.m

90-

m-

t3H]dThd incorporation per 5 x l o 4 cells cultured, or as total t3H]dThd incorporation per fraction. In order t o facilitate evaluation of the data, these have been normalized by setting the measured parameter (cpm, LU, o r the number of cells) obtained in the most active fraction equal to 100 %,thereby presenting an idea of both the most active fractions and the relationship between the different fractions.

mM Y)-

1

2

3

111619.11

4

5

6

1

8

P

i.om

I.w1

10 11 I2

13 I4

I5 16 I7 18 Iraction n u m b

1.080 density lglmll

Figure 1 . Density distribution of unsensitized PBL ( 0 - 0 ) . Results of cells/fractionare normalized in this and subsequent figures so that the number of cells recovered in the peak fraction is equal to 100 %. Refractive indices (x-x) were determined for each fraction to establish linearity of the gradient.

To compare the density distribution of unsensitized cells with that of cells separated following in vitro sensitization in MLC, ABm cells were first sensitized for 6 or 14 days and then separated on LDG. Fig. 2 demonstrates the appearance of light density cells after 6 days of sensitization. While the number of peaks obtained varies between experiments, the shift of t h e distribution t o lighter regions was seen in each of 5 separate experiments. It has been previously determined that b y day 1 4 cells in MLC exhibit minimal proliferative and cytotoxic activity [ 17, 181. When additional cells, set u p in MLC culture o n the same day as those separated on day 6, were separated o n day 14 the distribution was once more homogeneous. The very light peak detected on day 6 had disappeared and, while the majority of cells were as dense as when unsensitized cells were fractionated, there appeared t o be a slight enrichment of very dense cells (Fig. 2).

When equal numbers of cells from individual fractions were cultured in MLC they not only differed from one another in their ability t o incorporate [3H]dThd, but they also differed from an aliquot of cells (0 ABm) not separated o n the gradient (Fig. 3a). Both relative and total labeling intensities were greatest in the lighter half of the gradient. Preliminary experiments indicated that the cells were still in the exponential phase of growth when they were labeled. Therefore, differences in cpm between the various fractions should enable an evaluation of the numbers of precursor cells present in these suspensions. The density distribution of MLC-reactive cells in an unsensitized population which should reflect the distribution of MLC precursor cells, was compared with that of proliferating cells separated following in vitro sensitization. Fig. 3b demonstrates that o n day 6 the actively proliferating cells were found exclusively in the lighter half of the gradient. Neglgible activity was measurable in fraction 8, where the majority of the precursor activity was located. Additional cells from the same experiment were allowed to remain in culture until day 14 before being separated on LDG. The dotted line o n Fig. 3 b represents the cprn/fraction on day 14. The proliferative response in MLC at this time is activity I lraction (+I

cells I fraction 151

6l I \ 5

1

2

3

4 1.W

5

6

7

8

P 10 11 12 13 14 15 16 fraction number 1. om 1.m density lglmll

Figure 2. Density distribution of MLC-sensitized cells after 6 days ( 0 - 0 ) and 14 days (o---o) of culture.

1

2

3

4 5 1.w

6

7

8

P 10 11 I2 13 14 15 I6 1. om

lraction number

1.080 denrily lglmll

3.2. Density distribution of MLC-reactive cells separated before and after sensitization; functional analysis Unsensitized PBL were separated o n LDG as described in Sect. 2.4. From each fraction 5 x lo4 cells were then tested for their responsiveness in MLC. The results are expressed as

Figure 3. (a) MLC response of LDG-separated and unseparated unsen-

sitized PBL. Results are expressed as total cpm/fraction ( 0 - 0 ) and cpm/5 x lo4 cells (0-0).o ABm represents the proliferative response of an aliquot of cells not separated on LDG.(b) Density distribution of proliferating cells after 6 days ( 0 - 0 ) and 14 days (o---o) of in vitro sensitization. Results are expressed as cpm/fraction.

Density heterogeneity of human lymphocyte subsets

Eur. J. Immunol. 1977. 7: 366-370

extremely weak in each fraction (unseparated cells: AAm 726 f 31 1, ABm 1867 f 71 6) and no peaking of the values was seen.

3.3. Density distribution of CMLreactive cells When unsensitized T cells were separated before stimulation, the major area of CML precursor activity was a sharp peak in the lighter region of the gradient (Fig. 4a). Only a minor population of unsensitized cells, when compared with the number recovered in the peak fraction of the gradient, are of this density. While there are cells capable of a strong response in MLC, the total proliferative activity of this fraction richest in CML precursor activity is low. As with t h e MLC, fractions were found in which cells demonstrated enriched reactivity compared t o the unseparated cells (0 ABm). After 6 days of in vitro sensitization, cytotoxicity was most associated with newly appearing light density cells (Fig. 4b). Proliferative activity in this region was also strong. When residual cytotoxic activity was examined o n day 14, the detectable activity was much lower than a t t h e peak of the response (day 6 , 50: 1 effector t o target cell ratio = 42.6 %; day 14, 100: 1, 13.6 %). This small amount of remaining activity was a property of a very dense minority population of cells (Figs. 2 and 4b). Enrichment for memory activity was greatest in fraction 13, which contained 1.5 % of the total number of cells recovered. The proliferative ability of cells from these denser fractions was not greater than that of ,lighter cells (Fig. 3b). activity llraction IUI

0

4. Discussion

Recent evidence supports t h e hypothesis that MLC and CML responses are mediated by cooperation between more than one T cell subset [9-1 11. In the present study, we used a separation method based upon boyant density t o differentiate between T cell subsets mediating those responses and physically t o isolate T cell subsets. Density analyses were performed at three different time points during the MLC response. First, unsensitized lymphocytes were separated and subsequently tested in MLC and CML t o assess the density distribution of precursor cells. Second, MLC-sensitized cells were separated a t t h e time of peak responsiveness t o study MLC and CML effector cell distribution. Third, we attempted t o characterize putative memory cells by separating and testing long-term MLC cultures which showed background proliferation, but still detectable killing. We have observed density differences between cells separated either as functional precursors, as effectors a t t h e peak of the response, or as memory cells. It is also possible t o observe differences in the density distribution of cells reactive in MLC or CML within these three stages. The peaks of both MLC and CML precursor cells were lighter than the majority of peripheral T cells, though their respective density distribution profiles differed strikingly from one another. Thus, fractions were obtained which showed marked proliferation, but contained low numbers of LU. Upon sensitization, still lighter killer cells, either identical t o or associated with proliferating cells, appeared in culture. The majority of these effector cells detected at t h e peak of the response, belonged t o a density class which is only minimally represented in peripheral blood, As proliferation waned, a marked density change occurred in the remaining killer cells. These cells, which presumably carried the memory of that particular immune response, were small and heavy and contained within a population which exhibited a density distribution indistinguishable from that of circulating lymphocytes. These results, obtained with a technique presumably separating according t o the chemical composition of individual cells, lend further support t o the view that phenotypic differences exist between MLC and CML precursor cells [9- 111. Whereas precursor cells for specific killing in CML exhibited marked restriction in their density distribution, precursors of the proliferative response showed a strikingly heterogeneous distribution. These observations are consistent with data suggesting functional heterogeneity among cells responding in MLC [9- 1 11.

1 1

2

3

4

5

6

P

8

7

I I 12 I 3

10

14 15 fraction numbei

adiviiyl fraction IUI

im

369

-

90-

m-

ma50-

aa -

*." _._-.*

2010 0

-

-

-

I

2

3

4

1.w

5

6

7

8

9

10

Lorn

I

I1 I2 I 3 I4 I5

16 fraction number

1.080 dmily Iglml)

Figure 4. (a) CML response of either separated or unseparated unsensitized PBL on LDG. Results are expressed as LU/culture, where each fraction represents a culture, and are normalized; o ABm represents the response in CMLof an aliquot of ceUs not separated on the gradient. (b) Density distribution of killer cells sensitized for 6 days (0- 0 ) and 14 days (.-in MLC. -. Results ) are expressed as

LU/fraction and are normalized.

A further aspect of this study was the observation that killer cells underwent marked density changes upon stimulation. A relatively light precursor gave rise t o a rapidly expanding clone($ of very light cells. Without further antigenic stimulation, cytotoxicity was a property of dense cells which, with respect t o their low proliferative capacity and density, resembled small recirculating T lymphocytes. Similar density shifts were observed t o occur during in vivo differentiation of human lymphoid tumors*. Essentially similar findings have been obtained using velocity sedimentation t o separate mouse cytotoxic lymphocytes. Whereas at the peak of the

* Huber, Ch., Michlmayr, C., Zier, K. and Braunsteiner, H., manuscript in preparation.

370

Eur. J. Immunol. 1977. 7: 370-374

C.G. Fathman and M. Nabholz

response the killer activity was a property of rapidly sedimenting large cells, residual activity on day 1 4 was detected in slowly sedimenting small cells [ 19, 201. Based upon the results of this study and upon reports of others [l-71, we conclude that LDG enable t h e separation of functionally different subsets of T lymphocytes during various stages of immune activation. Experiments underway include examination of the fractions for (a) presence of surface markers, ( b ) ability to demonstrate either helper or suppressor effects in MLC and/or CML, and (c) the response following secondary restimulation in MLC. This technique may, therefore, serve as a useful tool for further characterization of distinct subsets of lymphocytes during different stages of in vitro cell-mediated immunity. The authors aregrateful to Drs. Dolores Schendel, Kirsten Linhhl, Rudolf Wankand Ammon Peck for thoughtful commentsand helpful discussions. We especkrlly thank Ms. A . Fodinger for b n g hours o f careful, expert techniml assistance.

Received December 15,1976; in revised form March 23,1977.

4 Hayry, P.H. and Andersson, L.C., in Natvig, J.B., Perlmann, P. and Wigzell, H. (Eds.), Lymphocytes, Isolation, Fractionation and Characterization, Scand. J. Immunol. Supplement no. 5, Universitetsforlaget,Osb 1976, p. 31. 5 El-Arini, M. and Osoba, D., J. Immunol. 1973.110: 1476. 6 Coraynski, R.M., Miller, R.G. and Phillips, R.A., Immunology 1970.19: 817. 7 Shortman, K., Aust. J. Exp. Biol. Med. Sci. 1968.46: 375. 8 Loos, J.A. and Roos, D., Exp. Cell Res. 1974.86: 333. 9 Cantor, H. and Boyse, E.A., J. Exp. Med. 1975.141: 1390. 10 Zier, K.S. and Bach, F.H., in Eijsvoogel, V. (Ed.), Proceedings

o f the X Leukocyte Culture Cbnference, Academic Press, New York 1976, in press. 11 Bach, F.H., Segall, M., Zier, K.S., Sondel, P.M., Alter, B.J. and Bach, M.L.,Science 1973.180: 403. 12 Boyum, A., Scand. J. Clin. Lab. Invest. Supplement no. 97. 1968.21: 31. 13 Wigzell, H., Huber, Ch. and Schirrmacher, V., Haematologim 1972. 6: 369. 14 Greaves, M.L. and Brown,G., J. Immunol. 1974.112: 420. 1 5 Michlmayr, G., Huber, Ch., Fink, U., Falkensammer, M. and Huber, H., Schweiz. Med. Wochenschr. 1974.104: 815.

16 Cerottini, J.-C., Engers, H.D., MacDonald, H.R. and Brunner, K.T., J. Exp. Med. 1974. 140: 71 3.

5. References 1 Shortman, K., von Boehmer, H., Lipp, J. and Hopper, K., 7Yansplant. Rev. 1975.25: 163. 2 Shortman, K., Byrd, W.J., Cerottini, J.C. and Brunner, K.T., Cell. Immunol. 1973. 6 : 24. 3 Shortman, K., Cerottini, J.C. and Brunner, K.T., Eur. J. Immunol. 1972.2: 313.

17 Hayry, P. and Andersson, L.C., Scand. J. Immunol. 1974. 3: 823. 18 MacDonald, R., Engers, H., Cerottini, J.-C. and Brunner, K.T., J. Exp. Med. 1974.140: 718. 19 Andersson, L.C. and Hayry, P., Scand. J. Immunol. 1974. 3: 461.

20 MacDonald, H.R., Cerottini, J.-C. and Brunner, K.T., J. Exp. Med. 1974.140: 1511.

C.G. Fathman and M. Nabholz'

ln v i m secondary mixed leukocyte reaction (MLR)

Basel Institute for Immunology, Basel

II. Interaction MLR determinants expressed by F1 cells T cells from strain A primed in vitro t o (C57BL/6 x A/J)Fl [(B6 x A)F1] cells, respond better t o restimulation by (B6 x A)FI than by B6 or a 1 :1 mixture of A and B6 cells. The increase in the response to F1 cells is specific and d u e t o MLR determinants present on (B6 x A)Fl cells but not on either of the parental cell types. (B6 x A)FI cells express more than one F1-specific MLR determinant, and this expression is dependent upon products of alleles of at least two loci within the major histocompatibility complex (MHC). Responsiveness t o these Fl-MLR determinants is apparently controlled by more than o n e locus within t h e MHC.

1. Introduction There have been several recent reports describing secondary in vitro responses of lymphocytes primed by previous in viva

or in virro exposure t o allogeneic leukocytes [ 1-81. We have

[I 16371

+ Present address: Markus Nabholz, I.S.R.E.C., Ch. de Boveresses, CH-1066 Epahges, Lausanne, Switzerland.

Correspondence: C.Garrison Fathman, Basel Institute for ImmunobgY, 487 Crenzacherstrasse, CH-4005 Basel 5 , Switzerland

recently described the characteristics of the proliferative response t o restimulation, of mouse cells surviving in culture several weeks, after a unidirectional mixed leukocyte reaction (MLR) [9]. We call these cells primed responder cells (PRC) and we have shown that there is a linear relationship between the number of PRC restimulated and the magnitude of their response. This allows a reliable quantitative comparison of the responses t o different stimulators. PRC are very much enriched for cells which respond specifically t o resiimulation with cells which share t h e I region of the H-2 complex with the priming stimulators. But we have also observed that such PRC react w i t h third party stimulators unrelated to the priming stimulator cell. Our results suggest that this response was due t o recognition of cross-reacting MLR determinants on the third party stimulator cell by the specifically primed cells in the PRC population. ~~

Abbreviations: M L R Mixed leukocyte reaction P R C Rimed responder cells FDA: Fluorescein diacetate [SHIdThd: Tritiated thymidine cpm: Counts/min

Linear density gradient separation of human lymphocyte subsets. I. Analysis by mixed leukocyte culture and cell-mediated lympholysis responses.

366 Eur. J . Immunol. 1977. 7: 366-370 K.S. Zier, C. Huber and H. Braunsteiner 16 Goldschneider, I. and McCregor, D.D., J. Exp. Med. 1973. 138: 144...
494KB Sizes 0 Downloads 0 Views