139, 198-207 ( 1992)

of Tumor Necrosis Factor-a! by Naive or Memory T Lymphocytes Activated via CD28

V. VON FLIEDNER,*"*S. MIESCHER,*J. GI~RAIN,I.H. GALLATI,$ C. BARRAS,?D. HEUMANN,~ AND J-C. CEROTTINI* *Ludwig Institute for Cancer Research, Lausanne Branch, Epalinges, Switzerland; tDivision of Infectious Diseases, University Hospital, Lausanne, Switzerland; and $5 Hoffman-LaRoche, Easel, Switzerland Received June 21, 1991; accepted August 16. 1991

While it is well establishedthat activated T cells can produce tumor necrosisfactor alpha (TNFa), it is less clear whether this function is confined to a given subset, e.g., memory cells. To approach this question, we investigated the production of TNF-(U by human peripheral blood T lymphocytes activated with anti-CD28 mAb since this activation pathway is known to potentiate cytokine production. Under the culture conditions used, the amount of TNF-cu produced was markedly enhanced compared to that obtained after activation with immobilized anti-CD3 mAb. The enhancement of TNF-(Y production was already apparent after incubation of T cells for 6 hr. Up to 5 rig/ml of TNF-a was measured on Day 2 in supematants of cultures of lo4 T lymphocytes. To determine the source of the cells producing high amounts of TNF-(r, T lymphocytes were separated into two subpopulations, namely naive cells (expressing the CD45RA isoform) and memory cells (expressing the CD45RO isoform). While both subpopulations proliferated equally well after stimulation with anti-CD28 mAb, up to 90% of the TNF-n produced under theseconditions originated from memory T cells. Theseresultsthus document that T cell activation via CD28 results in a marked increase in TNF-ol production without affecting the functional disparity that exists between naive and memory T cells. 0 1992 Academic PKSS,IIIC.


Tumor necrosis factor alpha (TNF-a) is a pleiotropic mediator of many inflammatory responses.While monocytes and macrophagesappear to be a major source of TNF-(u, there is evidence that this cytokine can be produced, at least in vitro, by other cells, including granulocytes (l), NK cells (2), and T cells (3, 4). In the latter cells, TNF-cu can be identified, like in monocytes, not only as a soluble, secreted product but also as a surface membrane-associated polypeptide (5-8). Production of TNF-a by T cells is not constitutive, but can be transiently induced, like that of other lymphokines, by various stimuli, including those provided by mitogenic lectins, phorbol ester combined with ionomycin, or antibodies directed against the T cell receptor (TCR)-CD3 molecular complex or CD28 (3, 4, 9, 10). The latter is a 44-kDa homodimeric glycoprotein expressedon the surface of about 80% of peripheral blood T lymphocytes (1 l- 13). Recent studies suggestthat CD28 may partici To whom correspondenceshould be addressedat Unite d’Onco-Immunologie Clinique, Ludwig InstituteCHUV-BH 19-602, 1011 Lausanne, Switzerland. 198 0008-8749192$3.00 Copyright 0 1992 by Academic Press,Inc. All rights of reproduction in any form reserved.





ipate in a T cell activation pathway which is distinct from that triggered via the TCRCD3 complex or CD2 (14, 15). In particular, it has been reported that production of several lymphokines, including TNF-(u, can be increased in cultures of T lymphocytes optimally activated with anti-CD3 mAb by the addition of anti-CD28 mAb (10). Moreover, the enhancement of lymphokine production observed under these conditions has been shown to be relatively resistant to the effect of cyclosporine, in contrast to that induced by antiCD3 mAb alone (10, 16, 17). The primary mechanism by which anti-CD28 mAb increaseslymphokine production in activated T cells involves an inhibitory effect on the degradation of the lymphokine mRNA (18). In addition, a possible effect of the CD28-dependent activation pathway on lymphokine gene transcription has been suggested recently ( 19). There is increasing evidence that production of lymphokines is not uniform among T lymphocytes. In particular, earlier studies from our laboratory showed that in the mouse, memory T cells produced greater amounts of IFN-y than naive T cells when stimulated with mitogens (20, 21). In contrast, IL-2 production was comparable in the two subpopulations. Similar results have been obtained in human studies using either mitogenic lectins or anti-CD3 mAb to activate peripheral blood T lymphocytes (22,23). In the human studies, advantage has been taken of the possibility to separate naive T cells from memory T cells on the basis of the CD45 isoform expressedon the cell surface (24, 25). The CD45 family consists of five isoforms which are produced by alternative splicing of three exons near the N-terminus of the molecule (26). It is now well accepted that the CD45RA isoform is expressed on naive T cells, whereas memory T cells express the CD45RO isoform (27). Given these results, we were interested in determining whether the marked production of TNF-(U observed after activation of T lymphocytes via CD28 reflected the stimulation of lymphokine production in all T cells or, alternatively, the enhancement of lymphokine production in memory T cells. In the present work, we show that memory T cells are the major source of TNF-cx (and IFN-7) which is produced in large amounts upon activation via CD28. MATERIALS AND METHODS Isolation of T Lymphocytes

Peripheral blood was obtained from normal donors and the mononuclear cells separated on Ficoll-Hypaque gradients. A purified peripheral blood T lymphocyte population (thereafter referred to as PBL-T) was prepared by using multiple adherence to plastic in 5% pooled human serum (Swiss Red Cross Center, Bern, Switzerland), followed by two passagesover nylon wool columns (Fenwal, Travenol, IL). The final preparation contained less than 0.5% monocytes as assessedby May-GrunwaldGiemsa and nonspecific esterasestainings and furthermore did not respond to PHA stimulation. The PBL-T were seededat IO4 or lo5 cells/well and stimulated as indicated, and then supernatants were harvested at different time intervals and assayedfor the production of TNF-a and INF-7. In parallel experiments cells were assessedfor proliferation by pulsing for the last 6 hr of incubation with [3H]thymidine (Amersham International plc, Zurich, Switzerland), specific activity 2.5 Ci/mmol, added at 1 &i/ well. Cells were harvested using a PHD cell harvester (Cambridge Technology Inc., Cambridge, MA) and counted in a liquid scintillation counter.




Antibodies The following mAbs were employed in thesestudies: antiCD45RO (UCHL 1, IgG2a, Dakopatts, Instrumenten Gesellschaft AG, Zurich, Switzerland) purified anti-CD3 (Lau-T3, IgGl, from Dr. S. Carrel, Ludwig Institute, Lausanne, Switzerland); antiCD28 mAb (CK248, IgM; Ref. (28); donated by Dr. A. Moretta, Institut Ludwig, Lausanne, Switzerland). A goat anti-mouse IgG/M F(ab’)2 antiserum labeled with FITC (GAM-FITC) (Tago, Opopharma, Zurich, Switzerland) was used as a second layer in immunofluorescence staining of cells before sorting on a FACS II (Becton-Dickinson, Sunnyvale, CA). Stimulation of Cells Immobilized CD3 mAb was prepared by coating 96-well U-bottom Greiner plates with Lau-T3 mAb (purified by caprylic acid precipitation) at a supraoptimal dose of 30 &ml in phosphate-buffered saline (PBS), 100 &well for 4-6 hr at 37°C followed by three washeswith PBS. Titration was done at dosesranging from 0.3 to 30 pg/ml by measuring proliferation at 48 hr. Phorbol-I 2-myristate- 12-acetate(PMA) obtained from Sigma, Basel, Switzerland, was used at 1 rig/ml. Anti-CD28 mAb, an IgM-K, was obtained from supernatant by ammonium sulfate precipitation; it was also titrated to obtain maximal proliferation in the presenceof 1 rig/ml PMA or 30 pg/ml immobilized anti-CD3 mAb and a supraoptimal dose of 30 pg protein/ml was selected (maximal proliferation also around 20 pg protein/ml). ImmunoJluorescence and Cell Sorting Sorting experiments were performed on PBL-T stained with the mAb UCHLl (20 &4 X lo6 cells in a volume of 200 ~1 RPM1 medium; Seromed, Basel, Switzerland) containing 10%fetal calf serum (FCS, Seromed, Base&Switzerland) for 30 min on ice followed by two washeswith RPM1 medium. The second layer GAM-FITC was also incubated for a further 30 min on ice followed by two further washesin RPM1 medium. The stained cells were resuspended in a final volume of 0.5 ml 10% FCS in RPM1 and immediately sorted. Before sorting cells, the sample tubing of the cell sorted was boiled, followed by rinsing with a solution of gentamycin in PBS. The gating limits for sorting were set according to three parameters: forward and right angular light scatter and logarithmically amplified green fluorescence intensity. The light scatter parameters were used to define a previously standardized lymphocyte region and the green fluorescence gate was set such that the negative and strongly positive cells were sorted. A small percentage(2-3%) of weakly positive cells were thus eliminated. Sorted cells were collected in sterile plastic tubes coated with sterile FCS. Purity was controlled by repassing sorted cells on the FACS II (>99% purity). After sorting, the cells were immediately washed and resuspendedin I ml of RPM1 containing 10%FCS. Viability was greater than 99% after each sorting. ELBA for Human TNF-a Reagents including recombinant human TNF-a, a mouse mAb to human TNF-a (clone 6B), and a rabbit anti-human TNF-ar were developed at Hoflinan-LaRoche, Basel. Ninety-six-well flat-bottomed polystyrene plates (Dynatech, Embrach, Switzerland) were coated with 200 &well mouse mAb to human TNF-a (clone 6I3) diluted



at 5 pg/ml in 100 rrnJ4phosphate buffer, pH 6.0. The plates were incubated overnight at room temperature, followed by washing twice with water. The plates were then filled with 200 mM Tris-HCl, pH 7.5, containing 10 g/liter bovine serum albumin and 0.25 g/liter thimerosal for a minimum of 24 hr at 4°C. Before testing, buffer was discarded without washing. A standard preparation of recombinant human TNF-a was titrated in doubling dilutions in a volume of 100 &well. Test samples were appropriately diluted and assayedin duplicate. The dilution buffer was 100 mA4 TrisHCl, pH 9.0, containing 1 g/liter phenol and 10%FCS. Fifty microliters of an appropriate dilution of a rabbit anti-human TNF-a coupled to peroxidase was added to each well and the plates were incubated at 4°C for 16-24 hr. Plates were carefully washed using successivelytwo washings in water, two washings in PBS containing 0.5 ml/liter of Tween 20. Plates were again washed in water and the peroxidase activity was revealed by incubating the plates for 15 to 30 min in 2 mM H202 in 100 mM glycin-HCl buffer, pH 4.0, containing 1 mJ4 tetramethyl benzidin as chromogen, 20 vol buffer to 1 vol chromogen. Chromogen (20 n&Q was dissolved in 10% aceton and 90% methanol. Peroxidase activity was blocked by the addition of 0.5 M H2S04 followed by reading at 450 nm. This assay is specific for human TNF-a, does not crossreact with human TNF-P, and is sensitive to approximately 30 pg/ml. ELISA for Human IFN-y Reagents used were recombinant human IFN-7, two mouse mAb directed against IFN-7, clone 69 coated on microplates, and clone 123 labeled with peroxidase (29). Ninety-six-well round-bottom polystyrene plates (Dynatech) were coated with 200 pl/ well mouse mAb to human IFN-7 (clone 69) at 5 pg/ml in 50 mA4 phosphate buffer, pH 7.5. The plates were incubated at room temperature for 24 hr followed by washing three times with water, then filled with 200 mM Tris/HCl, pH 7.5, containing 10 g/ liter bovine serum albumin and 0.25 g/liter thimerosal and finally were incubated for 24 hr at room temperature. Before testing, buffer wasdiscarded. Samplesand a standard human r-IFN-y were titrated in doubling dilutions in a final volume of 50 &well in 100 mM phosphate buffer, pH 6.5, containing 5 g/liter bovine serum albumin. To each well 200 ~1 of the second mouse mAb directed to human IIN- (Clone 123) conjugated with peroxidase was added. Plates were incubated for 16-24 hr at 4°C followed by five washeswith water. The peroxidase activity was revealed as described for TNF-a. The sensitivity was approximately 50 pg/ml using human r-IFN-y as a standard. RESULTS AND DISCUSSION TNF-a! Production by T Cells Stimulated with CD28 mAb CK248 It has been reported previously that the anti-CD28 (IgG2a) mAb 9.3 can enhance the production of lymphokines by normal human T cells that are activated by antiCD3 mAb or PMA ( 10). In a first series of experiments, we investigated whether CK248, a anti-CD28 IgM mAb, produced a similar effect. To this purpose, peripheral blood T cells ( lo5 cells/well) were incubated with mAb CK248 in the presence or absenceof immobilized Lau-T3, a anti-CD3 mAb, or PMA ( 1 rig/ml). The production of soluble TNF-a was measured after incubation for 6, 24, and 48 hr. As shown in Table 1, no production of TNF-a was detectable after incubation of T cells with either


VON FLIEDNER ET AL. TABLE 1 TNF-a Production Induced by CD28 Costimulation of Total PBL-T TNF-ol production (&ml) Stimulation

Hours of incubation:

None PMA Anti-CD28 Anti-CD3 Soluble anti-CD28 + immobilized anti-CD3 Soluble anti-CD28 + PMA





Production of tumor necrosis factor-alpha by naive or memory T lymphocytes activated via CD28.

While it is well established that activated T cells can produce tumor necrosis factor alpha (TNF-alpha), it is less clear whether this function is con...
761KB Sizes 0 Downloads 0 Views