CD28, IL-2-Independent Costimulatory Pathways for CD8 T Activation Homero Sepulveda, Adelheid Cerwenka, Tammy Morgan and Richard W. Dutton This information is current as of October 1, 2021. J Immunol 1999; 163:1133-1142; ; http://www.jimmunol.org/content/163/3/1133

References This article cites 69 articles, 37 of which you can access for free at: Downloaded from http://www.jimmunol.org/content/163/3/1133.full#ref-list-1

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision http://www.jimmunol.org/

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: by guest on October 1, 2021 http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 1999 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. CD28, IL-2-Independent Costimulatory Pathways for CD8 T Lymphocyte Activation

Homero Sepulveda,1* Adelheid Cerwenka,2† Tammy Morgan,† and Richard W. Dutton3†

We investigate, here, the mechanism of the costimulatory signals for CD8 activation and confirm that costimulation signals via CD28 do not appear to be required to initiate proliferation, but provide survival signals for CD8 T cells activated by TCR ligation. We show also that IL-6 and TNF-␣ can provide alternative costimulatory survival signals. IL-6 and TNF-␣ costimulate naive CD8 T cells cultured on plate-bound anti-CD3 in the absence of CD28 ligation. They act directly on sorted CD8-positive T cells. They also costimulate naive CD8 T cells from Rag-2-deficient mice, bearing transgenic TCRs for HY, which lack memory cells, a potential source of IL-2 secretion upon activation. IL-6 and TNF-␣ provide costimulation to naive CD8 T cells from CD28, IL-2, or IL-2R␣-deficient mice, and thus function in the absence of the -CD28 and IL-2 costimulatory pathways. The CD8 T cell generated via the anti-CD3 plus IL-6 and TNF-␣ pathway have effector function in that they express strong cytolytic activity on Ag-specific targets. They secrete only very small amounts of any of the cytokines tested upon restimulation with peptide-loaded Downloaded from APC. The ability of the naive CD8 T cells to respond to TCR ligation and costimulatory signals from IL-6 and TNF-␣ provides a novel pathway that can substitute for signals from CD4 helper cells or professional APC. This may be significant in the response to viral Ags, which can be potentially expressed on the surface of any class I MHC-expressing cell. The Journal of Immunology, 1999, 163: 1133–1142.

uch attention has been paid to the way in which lym- A number of candidates have been suggested for the nature of http://www.jimmunol.org/ phocytes do or do not become activated when they the second costimulatory signal, signal two. These have included M encounter Ag. The original two-signal model of B7.1 (8, 9) and B7.2 (10–15), which interact with CD28 (16) and Bretscher and Cohn (1) was put forward to explain peripheral un- CTLA-4 (17), the CD2 ligand (18–21), the CD30 ligand (22), responsiveness of B cells to self Ag. In their hypothesis, signal CD40 (23, 24), the heat stable Ag ligand (25–27), CD137, the one, resulting from the interaction of the Ag receptor with Ag, 4-1BB ligand (28–30), cytokines IL-1 (31–35), IL-2, and IL-6 (32– induced unresponsiveness if delivered alone, but induced prolifer- 36), TNF-␣ (36, 37), TGF-␤ (38), and various others, including ad- ation if accompanied by signal two. The model was later adapted hesion molecules, such as ICAM-1 (39–43) and LFA-3 (44). to apply to the activation of cytotoxic T cells to alloantigens (2).

Interest has focused most strongly on B7.1 and B7.2, which by guest on October 1, 2021 Experimental studies by Schwartz and his colleagues (3–5) with have been shown to interact with CD28 and induce signaling path- CD4 T cell clones showed that signal one alone induced an anergic ways that end in the stabilization of IL-2 message (45–50). Inter- state, and signal one plus signal two induced the production of IL-2 action of these same receptors with CTLA-4 provides inhibitory and the expression of the IL-2 receptor. signals (51–56). In the initial models, the nature of the two signals was not pre- The way that most other costimulatory signals function is less cisely defined. We now see signal one resulting from the engage- well established. Some have been seen as growth factors, others ment of the TCR with the peptide MHC ligand. It is clear that both survival factors, while yet others have been presumed to work by the density of the peptide MHC ligand and the affinity of the TCR increasing the adhesion between T cells and APC (57). In partic- for the ligand must, in some way, affect the strength of the signal ular, it was not established whether some of these other costimu- that is delivered. The issue of how this might affect the require- ment for signal two has been raised by Goldstein et al. (6) and by lators acted directly on the responding T cells or whether they Cai and her colleagues (7), who showed that T cells can respond induced one of the other known costimulators (such as B7) on the to signal one alone if it is very strong, but it has not been defin- APC (58). itively answered. There has also been a gradual appreciation that signal two is involved in survival of the responding cells (36, 37, 59), but, based on the studies with T cell clones, the concept that costimulation is required in addition for the initial activation is still widely ac- *Molecular Pathology Program, University of California at San Diego, La Jolla, CA cepted. This issue is not easily resolved, and the apparent result 92093; and †Trudeau Institute, Saranac Lake, NY 12983 depends on the assay used and the interpretation of the result. Received for publication August 6, 1998. Accepted for publication May 6, 1999. Thus, absence of thymidine incorporation is taken as evidence that The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance the cells did not divide. However, as we show here, the cells can with 18 U.S.C. Section 1734 solely to indicate this fact. divide rapidly but die before or during the thymidine uptake 1 Current address: R. W. Johnson Pharmaceutical Research Institute, San Diego, CA period. 92121. There are a number of additional caveats that should be kept in 2 Current address: Department of Immunobiology, DNAX Research Institute, Palo mind when reviewing the experimental basis for these models. Alto, CA 94304. First, it is important to remember that the initial proposal of 3 Address correspondence and reprint requests to Dr. Richard W. Dutton, Trudeau Institute, P.O. Box 59, Saranac Lake, NY 12983. E-mail address: dutton@ Bretscher and Cohn (1) was a theory designed to provide an northnet.edu explanation for a regulatory step in the activation of naive B cells.

Copyright © 1999 by The American Association of Immunologists 0022-1767/99/$02.00 1134 ALTERNATIVE COSTIMULATORY PATHWAYS FOR CD8 T CELLS

Much of the later work concerned the conditions for the activation obtained from Dr. Drew Pardoll (Baltimore, MD) and have been bred in of T cell clones, not naive T cells, and it is clear that cells in a T our facility. cell clone may be in a very different developmental state from that Cell preparations of the immediately ex vivo. It is also very clear that the activation requirements for naive cells are very much more CD8 T cells isolated from the spleen and lymph nodes were enriched by stringent than those for effector cells or resting memory cells. Sag- passing through nylon wool and treating with anti-CD4 (RL172.4), anti- heat stable Ag (J11D), anti-class II MHC (D3.137, M5114, CA4) mAbs, erstrom et al. (60) have shown that naive cells rapidly lose the and anti-NK 1.1 (DX-5) and complement. Small resting CD8 T cells were requirement for a second signal as they progress into their re- harvested from the bottom interphase of a four-layer Percoll gradient (Sig- sponse. Similar findings were seen for stimulation via CD2 (19), ma, St. Louis, MO). Naive CD8 populations were typically 90% pure. and studies by our colleagues have shown that effector cells need APC much lower numbers of TCR ligand interactions and have little requirement for costimulation (61). It is also to be noted that the P815 cells were used as APC in a small number of experiments with pep- costimulatory requirements for proliferation and for T cell func- tide-stimulated CD8 T cells from Clone 4 mice and were incubated with HA peptide at 11 ␮M for 30 min and washed before use. In other exper- tions, such as cytotoxic activity or cytokine secretion, can be very iments, we used “APC blasts,” T cell-depleted splenocytes, which were different (H. Sepulveda, unpublished observations). prepared using anti-Thy-1.2 (and F7D5), anti-CD4 (RL172.4), anti-CD8 Finally, most of the studies have concerned the stimulation of (3.155) mAbs, and complement, and stimulated with LPS (25 ␮g/ml) and CD4 T cells (41), with only a few studies involving CD8 (6, 7, 42, Dextran sulfate (25 ␮g/ml) for 48 h. 43, 59, 66). It is not clear how far the conditions for activation of Preparation of Tc1 and Tc2 effector cells the two lineages may differ. In the current studies, we have been concerned only with the CD8 T cells from the spleens and lymph nodes of Clone-4 TCR-transgenic Downloaded from costimulatory requirements for populations of naive CD8 T cells mice were prepared as above. APC were loaded with the HA peptide (11 ␮M) at 37°C for 30 min, treated with mitomycin C (50 ␮g/ml, Sigma) at immediately ex vivo and have utilized extensively purified popu- 37°C for 40 min, and washed three times before use. CD8 T cells were lations of CD8 T cells from TCR transgenic mice specific for a cultured in RPMI 1640 (Irvine Scientific, Santa Ana, CA) supplemented known peptide, in which most of the T cells are naive. We utilized with penicillin, streptomycin, glutamine, 2-ME, HEPES, and 10% FCS a number of T cell markers, such as CD44, CD25, and Ly6C, to (HyClone Laboratories, Logan, UT). For effector generation, CD8 T cells ϫ 5 from the Clone-4 transgenic mice (2 10 cells/ml) were stimulated with http://www.jimmunol.org/ distinguish between the phenotype of naive, effector, and memory HA peptide-loaded APCs (2 ϫ 105 cells/ml) in the presence of IL-2 (20 cells. In our initial experiments, we have confirmed that CD8 cells U/ml, supernatant from the X63Ag.IL-2 murine cell line), IL-12 (9.2 U/ml, are activated and divide in response to TCR ligation in the appar- kindly provided by Dr. Stanley Wolf (Genetics Institute, Cambridge, ent absence of costimulation and that the role of costimulation is to MA)), and anti-IL-4 (10 ␮g/ml, 11B11) for Tc1 cultures and in the pres- provide survival signals. We have stimulated the CD8 T cells with ence of IL-2 (20 U/ml), IL-4 (200 U/ml, X63.Ag.IL-4 supernatant), and anti-IFN-␥ mAb (XMG1.2, 20 ␮g/ml) for Tc2 cultures. On day 4 of cul- plate-bound Ab to the TCR complex to eliminate the contribution ture, effectors were 99% CD8ϩV␤8ϩ. of APC and have examined the effect of costimulation on the re- sponding T cells. We were thus able to show that there are alter- Flow cytometry nate costimulator molecules to B7.1 and ICAM-1, including IL-6 The following mAbs were used for immunofluorescent staining: Cy- by guest on October 1, 2021 and TNF-␣, that they act directly on the responding CD8 T cells, Chrome anti-CD8 (PharMingen, San Diego, CA), anti-v␤8 PE (PharMin- and that it is naive CD8 cells that respond to these cytokines, and gen; clone MR5-2), FITC anti-CD62L (PharMingen; clone MEL-14), that these cytokines can provide costimulatory signals in the ab- FITC anti-CD44 (PharMingen; clone IM7), FITC anti-CD45RB (Phar- Mingen; clone 23G2), FITC anti-CD25 (PharMingen; IL2R␣-chain, clone sence of the CD28 and IL-2 pathways. CD8 T cells stimulated with 3C7), anti-Ly6C (PharMingen; clone AL-21), PE anti-CD54 (PharMingen; plate-bound anti-CD3 plus IL-6 and TNF-␣ are physiologically clone 3E2), FITC anti-CD80 (PharMingen; clone 16-10A1), and FITC anti- active in that they develop into effector cells that are cytotoxic on CD86 (PharMingen; clone GL1). After staining with the appropriate Abs, Ag-specific targets, but have little ability to secrete cytokines. samples were analyzed using a FACScan (Becton Dickinson, Mountain These costimulatory molecules can thus provide a mechanism View, CA), and data processed using CellQuest (Becton Dickinson) soft- ware. For the sorting of the CD8 T cells, CyChrome anti-CD8 (PharMin- for CD8 T cell activation in the absence of professional APC. gen; clone 53-6.7) was used. This may be important in the response to viral infection in which the viral peptide may be expressed on any class I MHC- Cell cultures expressing cell. Cells were cultured in RPMI 1640 (Irvine Scientific), supplemented with penicillin, streptomycin, glutamine, 2-ME, HEPES, and 10% FCS (Hy- Clone Laboratories). Cultures were set up in 96-well plates (Costar, Cam- Materials and Methods bridge, MA) in triplicate, in 0.2-ml volumes at a concentration of 2 ϫ Mice 105/ml. For stimulation with anti-CD3, 96-well plates were coated with 0.1 ml of the hamster Ab 145-2C11 (American Type Culture Collection Mice were purchased from the Animal Breeding Facility at the Trudeau (ATCC), Manassas, VA) at 10 ␮g/ml for2hat37°C. Then, plates were Institute (Saranac Lake, NY). Clone-4 v␤8.2/v␣10 TCR-transgenic mice washed twice with 0.3 ml of PBS per well. When anti-CD28 is used, the were kindly provided by Dr. Linda Sherman (Scripps Research Institute, ascites of the cell line 37N51 (kindly provided by Dr. J. P. Allison, Uni- La Jolla, CA). The Clone-4 TCR-transgenic mice bear the ␣- and ␤-chains veristy of California, Berkeley, CA) was added at 1:50, a dilution found to of the Clone-4 CTL specific for the transmembrane peptide, residues 518– provide optimal costimulation. Cultures were supplemented with recom- 528 (IYSTVASSL) of hemagglutinin (HA)3 2 on H-2Kd. The Clone-4 binant murine IL-6, TNF-␣ (R&D Systems, Minneapolis, MN), either TCR-transgenic mice were backcrossed for eight generations with B10.D2. alone or in combination, and were added at a concentration of 5 ng/ml and CD28-deficient, IL-2-deficient, and IL-2R␣-deficient mice were bred at the 20 ng/ml, respectively. For the cell counts used to determine cell recov- Trudeau Institute from stock kindly provided by Drs. Carl June (62), eries, cultures were set up in 48-well plates (Costar, Cambridge, MA) in I. Horak (63), and Laurie Davidson (64), respectively. HY-specific 5 Ϫ Ϫ duplicates in 1-ml volumes at 2 ϫ 10 /ml and counted using trypan blue TCR-transgenic Rag-2 / mice were purchased from Taconic Farms exclusion to differentiate between live and dead cells. (Germantown, NY). Anti-HY TCR-transgenic H-2b SCID mice were Proliferation ϳ 3 Abbreviations used in this paper: HA, hemagglutinin; CFSE, 5-(and-6)-carboxy- The proliferation of CD8 T cells was measured by the 16 h uptake of 3 fluorescein diacetate, succinimidyl ester; FS, forward scatter; SS, side scatter; PI, [ H]thymidine (ICN Biomedicals, Irvine, CA) added at 0.2 ␮Ci/well, 48 h propidium iodide; Tc1 and Tc2, type 1 and type 2 CD8 effector cytotoxic T cells. from the start of culture. The Journal of Immunology 1135

Analysis of cytokine production Enriched CD8 T cells (2 ϫ 105 cells/ml) were stimulated with mitomycin- treated P815 cells (1.2 ϫ 106/ml) loaded or not with the HA peptide. IFN-␥, IL-4, and IL-5 were measured by specific ELISAs as described (65)

Cytotoxicity assay P815 cells were used as targets and loaded or not with the indicated con- centrations of the HA peptide for 30 min at 37°C. Subsequently, target cells (1 ϫ 106 cells/400 ␮l RPMI medium containing 1% FCS) were incubated with 3.7 mBq 51Cr (sp. act. 1.85 TBq/g; NEN Life Science Products, Bos- ton, MA) for1hat37°C. Labeled targets were washed three times before use. CD8 T cells were set up at the indicated ratios with the labeled targets (104 targets/well) at 37°C. Supernatants were collected after 4 h, and ra- dioactivity was detected by ␥-counting. Means and SEs of duplicate cul- tures are shown. The percentage of cytotoxicity was calculated using the formula: 100 ϫ [(cpm experimental Ϫ cpm spontaneous)/(cpm total Ϫ cpm spontaneous)]. Spontaneous release was typically 10–15% of the maximum release. Downloaded from Results Costimulation provides survival signals The CD8 cells were purified from spleen and lymph nodes of Clone 4 TCR transgenic mice specific for the peptide IYS- TVASSL, as described in Materials and Methods. They were phe- low high notypically naive, CD44 , CD62L , CD25 negative with low http://www.jimmunol.org/ forward scatter (FS) and side scatter (SS). The CD8 cell prepara- tions responded very poorly (as measured by [3H]TdR incorpora- tion at 48–60 h) to plate-bound anti-CD3 and strongly to anti-CD3 plus CD28, indicating that they were functionally naive. (Fig. 1A). The cell recoveries at days 2 and 4 for each of the stimulatory conditions reflected the results obtained by thymidine incorpora- FIGURE 1. A, Naive CD8 T cells require costimulation for the prolif- tion. In the absence of costimulation, the CD8 cell recoveries fell erative response. CD8 T cells were prepared from HA TCR transgenic mice and were cultured on plate-bound anti-CD3 in the presence or ab- in the second 2-day period, while the CD8 cultured with anti-CD28 sence of anti-CD28. Proliferation was assayed by the uptake of tritiated by guest on October 1, 2021 continued to proliferate and expanded manyfold (Fig. 1B). The thymidine present in culture from 48 to 64 h. B, Expansion of cell numbers thymidine uptake and cell recovery responses of CD8 T cells stim- is dependent on costimulation. CD8 T cells were cultured on plate-bound ulated with peptide-loaded APC were comparable to that seen with anti-CD3 plus (ࡗ) or minus anti-CD28 (Ⅺ). Cultures were harvested at the plate-bound anti-CD3 plus anti-CD28 (data not shown). times indicated and the number of live cells determined.

5-(and-6)-Carboxyfluorescein diacetate succinimidyl ester days 1, 2, 3, and 4 (right panels of each pair, Fig. 2), it can be seen (CFSE) and propidium iodide (PI) analyses of cell proliferation that all of the cells undergo cell division on TCR ligation, in the and survival presence (anti-CD3 plus anti-CD28) or absence of costimulation To get a more comprehensive picture of the response of the naive (anti-CD3 alone). Cells cultured alone do not divide and survive CD8 cells stimulated under different conditions, we used the fol- poorly (CD8 only). There are no cells left with the original CFSE lowing dye-label technique. Cells were first stained with the cy- intensity at the later time points in the stimulated cultures, even in toplasmic vital dye, CFSE, before in vitro stimulation. The inten- cultures stimulated with anti-CD3 alone. Second, the cells progress sity of dye label fell rapidly to 50% in the first few hours of culture through multiple rounds of division whether the costimulation is (data not shown) and then stabilized in the absence of cell division. strong (with anti-CD28) or absent or weak (plate-bound anti-CD3 The subsequent division of the CFSE-labeled cells was monitored alone), and the cells appear to divide at the same rate. Under con- by the reduction of dye intensity to half after each cell division. ditions of strong costimulation, the cells survive (left panel of anti- CD8 T cells were cultured either alone, on plate-bound anti-CD3, CD3 plus anti-CD28) and can be seen to undergo as many as eight or on plate-bound anti-CD3 plus anti-CD28. The cells were har- rounds of division (right panel), and few cells become PI-positive. vested at day 1, 2, 3, or 4 and stained for CD8 and V␤8, and PI was It is important to note that if all the cells had survived, there would added, as described in Materials and Methods. In the subsequent be a much greater increase in total cell number by day 4 than is four-color flow cytometric analysis, cells were gated for cells that actually observed, and it is clear that, even under optimal condi- were CD8ϩ,V␤8ϩ and were analyzed for the level of PI and CFSE tions, cells are being lost. The PI CFSE plot for each day showed label. Representative data are presented in Fig. 2, as a two-color that a variable number of the CFSE-labeled cells become PI-pos- plot for PI and CFSE (left panels), and as the histogram for CFSE itive at each time point and after each cell division. In the absence (right panels) for each of the culture conditions, for days 1, 2, 3, of costimulation, a major fraction of the cells die at each stage (as and 4 of culture. seen in the left panels for anti CD3 alone). Under conditions of This analysis reveals many features of the response not readily strong costimulation, this number is markedly reduced. It is im- discernible by the thymidine incorporation data shown for the portant to note that cells that are seen as already PI-positive at one same experiment. First, looking only at the CFSE histograms for time point have disintegrated by the next time point and no longer 1136 ALTERNATIVE COSTIMULATORY PATHWAYS FOR CD8 T CELLS Downloaded from

FIGURE 2. Cells proliferate but die in the absence of costimulation. CD8 T cells (as in Fig. 1) were stained with CFSE and cultured alone (left hand pairs) or on plate-bound anti-CD3 plus (right hand pairs) or minus (center pairs) anti-CD28. The cells were harvested at day 1 (top row), day 2 (second row), day 3 (third row), or day 4 (bottom row) and analyzed for PI and CFSE stain, gating on CD8 cells. Left panel (of each pair), The plot of PI vs CFSE. Right panel (of each pair), The CFSE histogram for the PI-negative cells. http://www.jimmunol.org/ score in the gated plot. They can be seen, however, for an extra day We therefore investigated the ability of IL-6 and TNF-␣ to sub- in the FS, SS plot (data not shown), but are gated out of the further stitute for B7 and ICAM-1 costimulation. Purified naive CD8 T analysis. cells were cultured on plate-bound anti-CD3 in the presence and We were concerned that small numbers of memory CD8 T cells absence of varying concentrations of IL-6 or TNF-␣. It can be seen derived from cells expressing endogenous TCR chains might con- that either cytokine provided strong proliferative stimuli in a dose- tribute IL-2 upon anti-CD3 stimulation and costimulatory signals. dependent manner, approaching the maximum at ϳ3–5 ng/ml for We therefore repeated the analysis with CD8 T cells from anti-HY each cytokine (Fig. 4). The cell recoveries at days 3 and 4, for by guest on October 1, 2021 TCR transgenic RAG-2Ϫ/Ϫ mice, as shown in Fig. 3A. It can be seen same factors, reflect the cumulative effect of IL-6 and TNF-␣ on that cells still divided in response to anti-CD3 alone, as judged by cell proliferation and survival (Table I). progressive loss of CFSE stain, as shown in the upper panel, and went through as many divisions as did cells costimulated with anti-CD28 IL-6 and TNF-␣ act directly on CD8 T cells (lower panel). This experiment has been conducted three times with Ͼ ϩ ␤ ϩ comparable results, twice with CD8 T cells from anti-HY Rag-2- Although the CD8 populations are 90% CD8 ,V 8 ,itwas deficient mice, and once with cells from anti-HY SCID mice. Cells possible that the cytokines acted indirectly by inducing costimu- stimulated with plate-bound anti-CD3 alone increased in size, and all latory molecules, such as B7, on some contaminating cell popu- showed up-regulated expression of CD69 as seen in Fig. 3B. lation. We stained the purified CD8 T cells with anti-CD8 and sorted for CD8-positive cells. The resulting population consisted of 99% CD8 T cells, which were CD44low, CD62Lhigh, high Costimulation in the absence of anti-CD28 CD45RB , CD25-negative, and showed both low and high staining with anti-Ly6C (data not shown). We cultured three pop- In other experiments (data not shown), we found that the CD8 T ulations: 1) the original purified, but unsorted, CD8 cells; 2) the cell populations we have used consist of naive cells, require co- unsorted CD8 reacted with anti-CD8 but not sorted; and 3) the stimulation to proliferate at day 2 and expand in number to day 4, sorted CD8 cells on plate-bound anti-CD3, in the presence and and that they can receive costimulatory signals from ICAM-1 and absence of the same cytokines and measured the [3H]thymidine B7, as shown by others for CD4 (40, 41) and CD8 T cells (42, 43). uptake (Fig. 5). The pattern of the response of sorted and unsorted We found, however, that different APC differ widely in their abil- populations (regardless of whether the latter had been reacted with ity to stimulate proliferation, even though they express the same the anti CD8 or not (data not shown)) to the costimulatory cyto- known costimulatory molecules in comparable amounts. We kines was very similar, indicating that the effect was directly on the sought to discover the factors responsible for this difference. We CD8 cells themselves and not via some contaminating population. showed that APC, such as P815, A20 line, and LPS/dextran sulfate- stimulated B blasts, made large amounts of message for ␣ ␤ ␤ IL-6 and TNF-␣ enhance the anti-CD3 stimulation of CD8 T IL-6, TNF- , TNF- , and LT- (data not shown), which were Ϫ/Ϫ potential candidates for costimulatory molecules. Of these, we fo- cells from anti-HY TCR transgenic Rag-2 mice cused on IL-6 and TNF-␣, as they had been previously reported to Memory CD8 T cells secrete IL-2 upon restimulation in vitro (65), enhance the CD4 T cells response to antigenic stimulation (35–37) and it was possible that contaminating memory cells could supply and because murine TNF-␤ was not available and LT-␤ is not a small amounts of this costimulatory factor. To confirm that the secreted molecule. IL-6 and TNF-␣ acted on a totally naive population of CD8 T cells, The Journal of Immunology 1137

FIGURE 4. The proliferative response of naive CD8 T cells from Clone-4 mice to varying concentrations of TNF-␣ and IL-6 in the presence of anti- CD3. HA-CD8 T cells were cultured (2 ϫ 105/ml) in the presence of plate- Downloaded from bound anti-CD3(10 ␮g/ml) and the indicated concentrations of TNF-␣ or IL-6. Cultures were incubated for 48 h, [3H]thymidine was added and incubated for an additional 16 h. Upper panel, TNF-␣; Lower panel, IL-6

and TNF-␣ could provide synergistic costimulation to plate-bound Ϫ/Ϫ anti-CD3-stimulated CD8 T cells from CD28 mice (Fig. 7). http://www.jimmunol.org/ The response to costimulation by IL-6 or TNF-␣ alone was less marked than with CD8 T cells from the Clone-4 HA mice, and strong synergy was seen when both cytokines were added. IL-2 provides strong proliferative signals to CD8 T cells acti- vated by plate-bound anti-CD3. In other experiments (data not shown), we found that the IL-6 or TNF-␣ induced the secretion of small amounts of IL-2 into the supernatants early in the response, but the amount was small (100 U) compared with that induced with anti-CD28 costimulation (550 U). It was possible, however, by guest on October 1, 2021 that the amount secreted was enough to account for the costimu- FIGURE 3. A, CD8 T cells respond to anti-CD3 alone and divide re- latory effect, and we next examined whether or not IL-6 or TNF-␣ Ϫ/Ϫ peatedly. CFSE-labeled CD8 T cells from Rag-2 anti-HY TCR trans- costimulation was dependent on the production or response to genic mice were cultured on plate-bound anti-CD3 in the presence or ab- IL-2. It can be seen in Fig. 8 that IL-6 and TNF-␣ could indeed sence of anti-CD28. Cells were harvested at day 1 or 3 for anti-CD3 alone provide costimulation to plate-bound anti-CD3-stimulated CD8 T and on day 1, 2, or 3 for anti-CD3 plus anti-CD28. CD8 gated T cells were Ϫ/Ϫ ␣Ϫ/Ϫ analyzed for level of CFSE and PI expression. B, CD8 T cells respond to cells from either IL-2 or IL-2R mice. The proliferative re- anti-CD3 alone and increase in size and up-regulate CD69. The same CD8 sponses were reproducibly lower than those seen with cells from wild- T cells from Rag-2Ϫ/Ϫ anti-HY TCR transgenic mice were cultured on type mice, especially in the case of the IL-2R␣-deficient mice. plate-bound anti-CD3 and were harvested at the time indicated and ana- ␣ lyzed for FS and SS and for the level CD69 expression. This experiment is CD8 effectors generated in the presence of IL-6 and TNF- representative of three experiments of similar design costimulation have effector function Finally, we investigated whether CD8 populations generated with alternate costimulation had effector function. Four-day effectors

Ϫ/Ϫ we prepared CD8 T cells from anti-HY TCR transgenic Rag-2 Table I. Cell recoveries in the presence or absence of costimulation mice, which can be assumed to have zero or much reduced num- with IL-6 or TNF-␣a bers of memory CD8 T cells. It can be seen in Fig. 6 that these cells gave little response to plate-bound anti-CD3 alone, but re- Fold Expansion Over sponded to costimulation with TNF-␣, IL-6, or both at levels com- Initial Cell Number parable to that seen with anti-CD28. It was of interest that under Costimulation Conditions Day 3 Day 4 these conditions IL-6 and TNF-␣ act synergistically. Anti-CD3 2.0 1.4 Anti-CD3 plus IL-6 2.8 8.5 ␣ ␣ Anti-CD3 plus TNF- 3.8 4.8 IL-6 and TNF- act via a CD28, IL-2-independent pathway Anti-CD3 plus IL-6 and TNF-␣ 3.8 7.5 In other experiments (data not shown) we found that TNF-␣, but Anti-CD3 plus anti-CD28 5.5 12.0 not IL-6, could up-regulate B7.1, but not B7.2, on anti- a CD8 T cells from HA-specific Clone-4 (2 ϫ 105/ml) transgenic mice were cul- CD3-stimulated CD8 T cells, and it was thus possible that TNF-␣, tured on plate-bound anti-CD3 Ϯ IL-6 (5 ng/ml), TNF-␣ (20 ng/ml), or anti-CD28. Cells were harvested after 3 or 4 days of culture, and cell counts were expressed at least, could still act via the CD28 pathway, providing costimu- as fold expansion over the initial number. The results are representative of three lation by T cell-T cell interactions. We showed, however, that IL-6 experiments. 1138 ALTERNATIVE COSTIMULATORY PATHWAYS FOR CD8 T CELLS

FIGURE 5. IL-6 and TNF-␣ are costimulatory for proliferation of FIGURE 7. IL-6 and TNF-␣ costimulate the response of CD8 T cells sorted CD8 T cells. Naive HA-CD8 T cells from Clone-4 mice were sorted Ϫ/Ϫ Ϫ/Ϫ using anti-CD8 CyChrome (PharMingen) on a FACStar cell sorter (Becton from CD28 mice to plate-bound anti-CD3. CD8 T cells from CD28 Dickinson). Sorted and not sorted CD8 T cells (2 ϫ 105/ml) were placed mice were cultured in the presence or absence of plate-bound anti- ␣ ␣ in 96-well plates coated with anti-CD3 (10 ␮g/ml). IL-6 (5 ng/ml) and/or CD3, IL-6, and TNF- at four different concentrations: 1) TNF- (20 ng/ ␣ ␣ TNF-␣ (10 ng/ml), or anti-CD28 (1:50) were added and cultured for 48 h. ml), IL-6 (5 ng/ml); 2) TNF- (5 ng/ml), IL-6 (2 ng/ml); 3) TNF- ␣ Proliferation was measured by the incorporation of [3H]thymidine between (2 ng/ml), IL-6 (1 ng/ml); and 4) TNF- (0.5 ng/ml), IL-6 (0.5 ng/ml). No Downloaded from 48 and 72 h. responses to cytokines were seen in the absence of anti-CD3. were prepared by stimulation with plate-bound anti-CD3 and IL-6 Discussion and TNF-␣ as costimulatory molecules. Tc1 and Tc2 populations Our principal findings can be summarized as follows. First, we

with type 1 or type 2 profiles of cytokine secretion were prepared http://www.jimmunol.org/ have confirmed that CD8 T cells respond to TCR ligation alone (or as previously described (65) from the same pool of naive CD8 T at least when costimulation is reduced to the minimum possible) cells for the purpose of comparison. The resulting cells were tested by cell division, but die rapidly. This is in agreement with the work in a 4-h radiolabeled chromium release assay. The IL-6/TNF-␣ of others who have also seen proliferative responses in the absence effectors (which we designate TcIL-6/TNF-␣) were shown to have of costimulation when the TCR signal was strong (6, 7, 66). Co- the same level of cytotoxic activity as Tc1 (Fig. 9A). Tc2 effectors stimulation by the B7/CD28 pathway provides survival signals that were more cytotoxic, as we have observed in our previous studies keep the cells alive, as previously shown (59). The use of the (65). The same effector cell populations were tested for cytokine production when restimulated with peptide and APC. Tc1 and Tc2 made the cytokines expected, IFN-␥ for Tc1 and IL-4 and IL-5 for by guest on October 1, 2021 Tc2. IL-6/TNF-␣ (TcIL-6/TNF-␣) effectors, however, made rela- tively small amounts of IFN-␥, and no measurable amounts of IL-4 or IL-5 (Fig. 9B). The same supernatants were also negative for IL-10 and IL-2 (data not shown).

FIGURE 6. IL-6 and TNF-␣ enhance the proliferation of naive CD8 T cells from anti-male Ag (HY) TCR transgenic Rag-2Ϫ/Ϫ mice. Naive anti- FIGURE 8. IL-6 and TNF-␣ costimulate the response of CD8 T cells HY CD8 T cells (2 ϫ 105/ml) were cultured with plate-bound anti-CD3 from IL-2 and IL-2R␣Ϫ/Ϫ mice to plate-bound anti-CD3. CD8 T cells from (coated at 10 ␮g/ml) in the presence or absence of TNF-␣ (20ng/ml) or the cytokine or cytokine receptor-deficient mice were cultured in the pres- IL-6 (5ng/ml). Cultures were incubated for 48 h, and proliferation was ence or absence of plate-bound anti-CD3, IL-6, and TNF-␣. No responses measured by the incorporation of [3H]thymidine between 48 and 64 h. to cytokines were seen in the absence of anti-CD3. The Journal of Immunology 1139 Downloaded from http://www.jimmunol.org/

FIGURE 9. A, Anti-CD3 plus IL-6/TNF-␣-generated effectors (Tc6/TNF-␣) have cytolytic activity against peptide-loaded targets. CD8 T cells from Clone 4 mice were cultured in the presence or absence of plate-bound anti-CD3, IL-6, and TNF-␣ (Tc6/TNF-␣), or IL-2 plus IL-12, anti IL-4, and peptide-loaded APC (Tc1) or IL-2, plus IL-4, anti-IFN-␥, and peptide-loaded APC (Tc2). The 4-day effectors were assayed for cytolytic activity in a 4-h radiolabeled chromium release assay. B, IL-6/TNF-␣ generated effectors (Tc6/TNF-␣) make small amounts of IFN-␥. CD8 T cells from Clone 4 mice were cultured in the presence or absence of plate-bound anti-CD3, IL-6, and TNF-␣ (Tc6/TNF-a), or IL-2 plus IL-12, anti IL-4, and peptide-loaded APC (Tc1) or IL-2, plus IL-4, anti-IFN-␥, and peptide-loaded APC (Tc2) as for Fig. 9A. The 4-day effectors were restimulated with peptide-loaded APC and the supernatants collected after 24 h and assayed for the level of IFN-␥, IL-4, and IL-5. by guest on October 1, 2021

tracker dye, CFSE, coupled with PI allowed us to follow the fate cells, and that they act on naive CD8 T cells via a pathway that is of the stimulated cells. In all cases, the CD8 T cells divided in independent of CD28 and IL-2. The resulting effectors have sim- response to T cell ligation, and no cells were left that had not ilar effector functions to cells stimulated with B7-positive APC in divided. When strong costimulation was provided, a major fraction that they are strongly cytotoxic but they secret little or no cytokine of the dividing cells remained alive and made it through multiple upon restimulation. divisions. When less costimulation was provided, a much higher In the studies reported, we set out to define the costimulatory proportion of cells died after each division. When steps were taken signals that are required for the Ag-driven proliferative response of to reduce costimulatory signals to the minimum by using CD8 T naive peripheral CD8 T cells. We used purified T cells from cells from TCR transgenic, Rag-2-deficient, or SCID mice, devoid “Clone-4” mice bearing the ␣- and ␤-chain of the TCR specific for of memory T cells, most of the cells died as judged by PI-positive the HA peptide, IYSTVASSL. The majority of the cells have not stain and by low cell recovery. Nevertheless, some cells that had encountered Ag, and the CD8 cells from these mice are largely been through multiple division could always be detected. One in- naive, as shown by flow cytometric analyses of cell surface mark- terpretation, which we favor, is that CD8 T cells divide in response ers and by the fact they do not respond to TCR ligation alone, as to T cell ligation alone, but die rapidly. In this interpretation, co- judged by [3H]thymidine incorporation in the 48–60 h interval of stimulatory molecules provide survival signals that keep the re- culture. A small but variable number of memory cells are present, sponding cells alive. Another interpretation is that the division seen with anti-CD3 alone is due to cells receiving residual co- and their presence can affect the response to other cytokines, as stimulatory signals from small numbers of contaminating APC in will be discussed below. the CD8 T cell preparation. We have repeated this experiment a In earlier studies (our unpublished observations), we had made number of times with similar results, but varying degrees of cell an analysis of costimulatory action using CL7 fibroblasts trans- death. We have not pursued this further, since, if all costimulation fected with known costimulatory molecules. It was noteworthy, is removed, all the cells die, either before or after division, and however, that although these transfectants expressed adequate lev- there is no good experimental finding to discriminate conclusively els of class I MHC, B7.1, and ICAM-1, they were less stimulatory betwen the two models. than other APC that we tested. When we screened the various APC Second, we have shown that there are alternate costimulator for the presence of cytokine mRNA, however, we found that the molecules to ICAM-1 and B7.1 and B7.2, including IL-6 and RNA from the more effective APC contained message for a num- TNF-␣, that the latter two act directly on the responding CD8 T ber of potential stimulatory factors, including IL-6 and TNF-␣, 1140 ALTERNATIVE COSTIMULATORY PATHWAYS FOR CD8 T CELLS which had been previously shown to enhance T cell response to used H-2Ld loaded with peptide at high density, and found the TCR ligation (36, 37), while the transfected CL7 fibroblast line CD8 T cells from 2C mice could respond without any costimula- was negative. These two cytokines were then tested in the presence tion at all. In our studies, we showed that the costimulatory cyto- and absence of stimulation by plate-bound anti-CD3, and it was kines were effective at both high (anti-CD3) and low (the CL7 found that they could deliver these survival signals in the absence transfectants) TCR ligand cross-linking (data not shown). More of costimulation via CD28. Thus, [3H]thymidine incorporation recently, Cai and her colleagues (7) have shown that the mobility was enhanced, and cell recovery at day 4 was as good with IL-6 or of the ligand for the TCR is also a factor, providing a much stron- TNF-␣ as that with anti-CD28. Titrations of the effect of the cy- ger and sufficient signal when the ligand for the TCR is tokines showed that the response became almost maximal by 5 immobilized. ng/ml. IL-6 and TNF-␣ together were somewhat more stimulatory The experiments described in Figs. 5–8 did nothing to reveal the than either alone, but, in general, with cells from the Clone-4 mice, mechanism of the costimulatory effect of the IL-6 nor TNF-␣, there did not seem to be much evidence that the effects of the two although the fact that the two cytokines enhance the response to cytokines were additive. the B7.1-ICAM-1-transfected fibroblasts (data not shown) sug- Similar experiments with other cytokines (data not shown) gested that they may act by some pathway other than that used by showed that IL-1 was without effect and that TGF-␤ was strongly these two costimulatory molecules. We have shown that IL-6 and inhibitory. (Other candidate molecules could not be tested, murine TNF-␣ induce only very modest amounts of IL-2, compared with TNF-␤/LT-␣, because only the human cytokine is available, LT␤ anti-CD28, also suggesting a different pathway. Experiments with because it is not secreted). CD28 knockout mice revealed that IL-6 and TNF-␣ costimulation We next asked whether these factors acted directly on CD8 T was still observed and must act via some pathway that did not Downloaded from cells or via some intermediate cell present in the purified but un- involve CD28. The experiments with CD8 T cells from the IL-2 sorted CD8 T cell population. Purified CD8 T cells were reacted and IL-2R␣-deficient mice showed that the two cytokines did not with anti-CD8 Ab and positively selected in the cell sorter. We require the induction of, or a response to, IL-2 and more probably 3 found that the level of [ H]thymidine incorporation and the yield acted via a signal pathway that led directly to enhanced CD8 sur- of cells at day four was just as high for sorted CD8 T cells as for vival. It is possible that the IL-6 and TNF-␣ act directly to elevate the unsorted culture. We concluded that the action of the cytokines survival or down-regulate death gene expression, http://www.jimmunol.org/ did not depend on contaminating cells and that they must act di- but the nature of the signaling pathway has yet to be determined. rectly on the CD8 T cells. The response of the cells from the IL-2-deficient mice, and espe- We have shown elsewhere that IL-2 is produced by memory cially from the IL-2R␣-deficient mice, were much lower than that CD8 T cells when they are restimulated (65) and it was thus pos- of cells from the wild-type animal. The significance of the ob- sible that anti-CD3 induced IL-2 secretion in the memory cells, served lower response is not clear, but CD8 T cells from these which allowed the response of the naive cells to proceed. It was mice display a memory phenotype (data not shown) and have been clear, however, that the two cytokines can act directly on naive shown to be relatively unresponsive to antigenic stimulation (68). CD8 T cells, without “help” from memory cells, as the same co- It was important to determine whether activated CD8 T cells stimulatory effect was observed when CD8 T cells from TCR generated with IL-6/TNF-␣ costimulation, which we designated by guest on October 1, 2021 transgenic anti-HY Rag-2Ϫ/Ϫ mice were used. T cells from the Tc6/TNF-␣, had effector functions. They were indeed as cytotoxic anti-HY Rag-2Ϫ/Ϫ mice are unable to express endogenous ␣-or as Tc1 effectors generated with IL-2, IL-12, anti-IL-4, and peptide- ␤-chains and therefore generate little or no memory T cells in the loaded APC. They were less cytotoxic than Tc2 effectors generated absence of the male Ag. The response was somewhat lower than with IL-2, IL-4, anti-IFN-␥, and peptide-loaded APC, and they that of the wild-type mice, presumably because there was no IL-2 made only very modest amounts of IFN-␥ and no detectable IL-2, production from memory T cells to enhance the expansion of the IL-4, IL-5, or IL-10. The absence of significant cytokine secretion responding cells. Joseph et al. (36) have argued that IL-6 and suggests that different subsets of effectors may arise as a result of TNF-␣ are only costimulatory if IL-2 has already been induced. It different sources of costimulatory signals. We are currently en- is still possible that anti-CD3 alone induces some IL-2 in the naive gaged in a more systematic investigation of the range of effector cells, but there was almost no proliferation to anti-CD3 alone. It functions. was noteworthy that, in these experiments and in other experi- ments in which the number of memory cells and hence the poten- Finally, it should be noted that the ability of cytokines to sub- tial for IL-2 production was minimized, that IL-6 alone was barely stitute for costimulation via B7 may be important in allowing the stimulatory and TNF-␣ alone was less strongly costimulatory, but activation of the class I MHC-restricted CD8 T cells, in the ab- that there was now a marked synergy when the two cytokines were sence of help from IL-2-secreting CD4 T cells and by nonprofes- added together. We hypothesize that each of the two cytokines are sional APC lacking B7 or other cell surface costimulatory mole- costimulatory in the presence of small amounts of IL-2, but that cules. This may be crucial in the response to viruses that do not both must be added when IL-2 is present in very small amounts (as initially infect APC and that are noncytopathic. Both cytokines can with cells from the Rag-2Ϫ/Ϫ mice or the CD28Ϫ/Ϫ mice) or is totally be derived from multiple sources. IL-6 is made by monocytes, absent, as in the case of the IL-2 knockout). The responses are also fibroblasts, endothelial cells, macrophages, and also by T and B ␣ larger when IL-2 is present, since IL-2 is a T cell growth factor. and other cells after activation. TNF- is secreted by Several investigators (36, 66, 67) have suggested that the re- macrophages, neutrophils, and NK cells, following stimulation by quirement for costimulation is reduced when the TCR signal is IFNs, LPS, or other bacterial products and by activated T cells. strong. Other studies have shown requirements for B7 and The cytokines could also provide an APC-independent alternative ICAM-1 costimulation using Drosophila cells (42). Here, the TCR to so-called “licensed” APC (69–71) to activate CD8 T cells. ligand is presumably present at high density, and yet costimulation was required. In the study of Cai and Sprent (66), RMAS cells were used, which only express cell surface class I MHC when References stabilized by peptide. RMAS is a thymoma and is potentially a 1. Bretscher, P., and M. Cohn. 1970. A theory of self-nonself discrimination. Sci- source of costimulatory cytokines. Goldstein et al. (6), however, ence 950:1042. The Journal of Immunology 1141

2. Lafferty, K. J., S. J. Prowse, C. J. Simeonovic, and H. S. Warren. 1983. Immu- 29. Pollok, K. E., Y. J. Kim, Z. Zhou, J. Hurtado, K. K. Kim, R. T. Pickard, and nobiology of tissue transplantation: a return to the passenger leukocyte concept. B. S. Kwon. 1993. Inducible T cell 4–1BB: analysis of expression and Annu. Rev. Immunol. 1:143. function. J. Immunol. 150:771. 3. Jenkins, M. K., D. M. Pardoll, J. Mizuguchi, H. Quill, and R. H. Schwartz. 1987. 30. DeBenedette, M. A., A. Shahinian, T. W. Mak, and T. H. Watts. 1997. Costimu- T-cell unresponsiveness in vivo and in vitro: fine specificity of induction and lation of CD28- T lymphocytes by 4–1BB ligand. J. Immunol. 158:551. molecular characterization of the unresponsive state. J. Exp. Med. 165:302. 31. Williams, J. M., D. Deloria, J. A. Hansen, C. A. Dinarello, R. Loertscher, 4. Jenkins, M. K., and R. H. Schwartz. 1987. Antigen presentation by chemically H. M. Shapiro, and T. B. Strom. 1985. The events of primary T cell activation can modified splenocytes induces antigen-specific T cell unresponsiveness in vitro be staged by use of Sepharose-bound anti-T3 (64. 1) monoclonal antibody and and in vivo. J. Exp. Med. 165:302. purified 1. J. Immunol. 135:2249. 5. Mueller, D. L., M. K. Jenkins, and R. H. Schwartz. 1989. Clonal expansion versus 32. Houssiau, F. A., P. G. Coulie, D. Olive, and J. Van Snick. 1988. Synergistic functional clonal inactivation: a costimulatory signaling pathway determines the out- activation of human T cells by interleukin 1 and . Eur. J. Immunol. come of T cell antigen receptor occupancy. Annu. Rev. Immunol. 7:445. 18:653. 6. Goldstein, R. G., T. Chen, M. Brunswick, H. Mostowsky, and S. Kozlowski. 33. Houssiau, F. A., P. G. Coulie, and J. Van Snick. 1989. Distinct roles of IL-1 and 1998. Purified MHC class I peptide complexes activate naive CD8ϩ T cells IL-6 in human T cell activation. J. Immunol. 143:2520. independently of the CD28/B7 and LFA-1/ICAM-1 costimulatory interactions. 34. Vink, A., C. Uyttenhove, P. Wauters, and J. Van Snick. 1990. Accessory factors J. Immunol. 160:3180. involved in murine T cell activation: distinct roles of interleukin 6, interleukin 1 7. Luxembourg, A. T., A. B., Brunmark, Y. Kong, M. R. Jackson, P. A. Peterson, and tumor necrosis factor. Eur. J. Immunol. 20:1. 35. Holsti, M. A., J. McArthur, J. P. Allison, and D. H. Raulet. 1994. Role of IL-6, J. Sprent, and Z. Cai. 1998. Requirements for stimulating naive CD8 T cells via ϩ signal one alone. J. Immunol. 161:5226. IL-1, and CD28 signaling in responses of mouse CD4 T cells to immobilized anti-TCR monoclonal antibody. J. Immunol. 152:1618. 8. Freeman, G. J., A. S. Freedman, J. M. Segil, G. Lee, J. F. Whitman, and 36. Joseph, S. B., K. T. Minor, and M. Croft. 1998. Augmentation of naive, Th1 and L. M. Nadler. 1989. B7, a new member of the Ig superfamily with unique ex- Th2 effect CD4 response by IL-6, IL-1 and TNF. Eur. J. Immunol. 28:277. pression on activated and neoplastic B cells. J. Immunol. 143:2714. 37. Vella, A. T., T. Mitchell, B. Groth, P. S. Linsley, J. M. Green, C. B. Thompson, 9. Freeman, G. J., G. S. Gray, C. D. Gimmi, D. B. Lombard, L. J. Zhou, M. White, J. W. Kappler, and P. Marrack. 1997. CD28 engagement and proinflammatory J. D. Fingeroth, J. G. Gribben, and L. M. Nadler. 1991. Structure, expression, and cytokines contribute to T cell expansion and long-term survival in vivo. J. Im- T cell costimulatory activity of the murine homologue of the human B lympho-

munol. 158:4714. Downloaded from cyte activation antigen B7. J. Exp. Med. 174:625. ϩ 38. Lee, H., and S. Rich. 1993. Differential activation of CD8 T cells by transform- 10. Azuma, M., D. Ito, H. Yagita, K. Okumura, J. H. Phillips, L. L. Lanier, and ing growth factor-␤1. J. Immunol. 151:668. C. Somoza. 1993. B70 antigen is a second ligand for CTLA-4 and CD28. Nature 39. Kuhlman, P., V. T. Moy, B. A. Lollo, and A. A. Brian. 1991. The accessory 366:76. function of murine intercellular adhesion molecule-1 in T lymphocyte activation: 11. Freeman, G. J., F. Borriello, R. J. Hodes, H. Reiser, J. G. Gribben, J. W. Ng, contributions of adhesion and co-activation. J. Immunol. 146:1773. J. Kim, J. M. Goldberg, K. Hathcock, G. Laszlo, et al. 1993. Murine B7-2, an 40. Dubey, C., M. Croft, and S. L. Swain. 1995. Costimulatory requirements of naive alternative CTLA4 counter-receptor that costimulates T cell proliferation and CD4ϩ T cells: ICAM-1 or B7-1 can costimulate naive CD4 T cell activation but production. J. Exp. Med. 178:2185. both are required for optimum response. J. Immunol. 155:45. http://www.jimmunol.org/ 12. Freeman, G. J., F. Borriello, R. J. Hodes, H. Reiser, K. S. Hathcock, G. Laszlo, 41. Dubey, C., and M. Croft. 1996. Accessory molecule regulation of naive CD4 T A. J. McKnight, J. Kim, L. Du, D. B. Lombard, et al. 1993. Uncovering of cell activation. Immunol. Res. 15:114. functional alternative CTLA-4 counter-receptor in B7-deficient mice. Science 42. Cai, Z., A. Brunmark, M. R. Jackson, D. Loh, P. A. Peterson, and J. Sprent. 1996. 262:907. Transfected Drosophila cells as a probe for defining the minimal requirements for 13. Freeman, G. J., J. G. Gribben, V. A. Boussiotis, J. W. Ng, V. A. Restivo, Jr., stimulating unprimed CD8ϩ T cells. Proc. Natl. Acad. Sci. USA 93:14736. L. A. Lombard, G. S. Gray, and L. M. Nadler. 1993. Cloning of B7-2: a CTLA-4 43. Deeths, M. J., and M. F. Mescher. 1997. B7-1-dependent co-stimulation results counter-receptor that costimulates human T cell proliferation. Science 262:909. in qualitatively and quantitatively different responses by CD4ϩ and CD8ϩ T 14. Hathcock, K. S., G. Laszlo, H. B. Dickler, J. Bradshaw, P. Linsley, and cells. Eur. J. Immunol. 27:598. R. J. Hodes. 1993. Identification of an alternative CTLA-4 ligand costimulatory 44. Parra, E., A. G. Wingren, G. Hedlund, T. Kalland, and M. Dohlsten. 1997. The for T cell activation. Science 262:905. role of B7-1 and LFA-3 in costimulation of CD8ϩ T cells. J. Immunol. 158:637. 15. Ranger, A. M., M. P. Das, V. K. Kuchroo, and L. H. Glimcher. 1996. B7-2 45. Linsley, P. S., W. Brady, L. Grosmaire, A. Aruffo, N. K. Damle, and

(CD86) is essential for the development of IL-4-producing T cells. Int. Immunol. J. A. Ledbetter. 1991. Binding of the B cell activation antigen B7 to CD28 by guest on October 1, 2021 8:1549. costimulates T cell proliferation and interleukin 2 mRNA accumulation. J. Exp. 16. June, C. H., J. A. Ledbetter, P. S. Linsley, and C. B. Thompson. 1990. Role of Med. 173:721. the CD28 receptor in T-cell activation. Immunol. Today 11:211. 46. Harding, F. A., J. G. McArthur, J. A. Gross, D. H. Raulet, and J. P. Allison. 1992. 17. Linsley, P. S., W. Brady, M. Urnes, L. S. Grosmaire, N. K. Damle, and CD28-mediated signaling co-stimulates murine T cells and prevents induction of J. A. Ledbetter. 1991. CTLA-4 is a second receptor for the B cell activation anergy in T-cell clones. Nature 356:607. antigen B7. J. Exp. Med. 174:561. 47. Harding, F., and J. Allison. 1993. CD28–B7 interactions allow the induction of 18. Meuer, S. C., R. E. Hussey, M. Fabbi, D. Fox, O. Acuto, K. A. Fitzgerald, CD8 cytotoxic lymphocytes in the absense of exogenous help. J. Exp. Med. J. C. Hodgdon, J. P. Protentis, S. F. Schlossman, and E. L. Reinherz. 1984. An 177:1791. alternative pathway of T-cell activation: a functional role for the 50 kd T11 sheep 48. Norton, S. D., L. Zuckerman, K. B. Urdahl, R. Shefner, J. Miller, and erythrocyte receptor . Cell 36:897. M. K. Jenkins. 1992. The CD28 ligand, B7, enhances IL-2 production by pro- 19. Byrne, J. A., J. L. Butler, E. L. Reinherz, and M. D. Cooper. 1989. Virgin and viding a costimulatory signal to T cells. J. Immunol. 149:1556. memory T cells have different requirements for activation via the CD2 molecule. 49. Lindstein, T., C. H. June, J. A. Ledbetter, G. Stella, and C. B. Thompson. 1989. Int. Immunol. 1:29. Regulation of lymphokine messenger RNA stability by a surface-mediated T cell 20. Kato, K., M. Koyanagi, H. Okada, T. Takanashi, Y. W. Wong, A. F. Williams, activation pathway. Science 244:339. K. Okumura, and H. Yagita. 1992. CD48 is a counter-receptor for mouse CD2 50. Linsley, P. S., and J. A. Ledbetter. 1993. The role of the CD28 receptor during and is involved in T cell activation. J. Exp. Med. 176:1241. T cell responses to antigen. Annu. Rev. Immunol. 11:191. 51. Krummel, M. F., and J. P. Allison. 1995. CD28 and CTLA-4 have opposing 21. Holter, W., M. Schwarz, A. Cerwenka, and W. Knapp. 1996. The role of CD2 as effects on the response of T cells to stimulation. J. Exp. Med. 182:459. a regulator of human T-cell cytokine production. Immunol. Rev. 153:107. 52. Krummel, M. F., and J. P. Allison. 1996. CTLA-4 engagement inhibits IL-2 22. Bowen, M. A., R. K. Lee, G. Miragliotta, S. Y. Nam, and E. R. Podack. 1996. accumulation and cell cycle progression upon activation of resting T cells. J. Exp. Structure and expression of murine CD30 and its role in cytokine production. Med. 183:2533. J. Immunol. 156:442. 53. Chambers, C. A., M. F. Krummel, B. Boitel, A. Hurwitz, T. J. Sullivan, 23. Grewal, I. S., and R. A. Flavell. 1996. The role of CD40 ligand in costimulation S. Fournier, D. Cassell, M. Brunner, and J. P. Allison. 1996. The role of CTLA-4 and T-cell activation. Immunol. Rev. 153:85. in the regulation and initiation of T-cell responses. Immunol. Rev. 153:27. 24. Yang, Y., and J. M. Wilson. 1996. CD40 ligand-dependent T cell activation: 54. Boussiotis, V. A., G. J. Freeman, J. G. Gribben, and L. M. Nadler. 1996. The role requirement of B7-CD28 signaling through CD40. Science 273:1862. of B7-1/B7-2:CD28/CLTA-4 pathways in the prevention of anergy, induction of 25. Liu, Y., B. Jones, A. Aruffo, K. M. Sullivan, P. S. Linsley, and C. A. Janeway, productive immunity and down-regulation of the immune response. Immunol. Jr. 1992. Heat-stable antigen is a costimulatory molecule for CD4 T cell growth. Rev. 153:5. J. Exp. Med. 175:437. 55. Sperling, A. I., and J. A. Bluestone. 1996. The complexities of T-cell co-stimu- 26. Liu, Y., B. Jones, W. Brady, C. A. Janeway, Jr., P. S. Linsley, and P. S. Linley. lation: CD28 and beyond. Immunol. Rev. 153:155. 1992. Co-stimulation of murine CD4 T cell growth: cooperation between B7 and 56. Waterhouse, P., L. E. Marengere, H. W. Mittrucker, and T. W. Mak. 1996. CTLA-4, heat-stable antigen. Eur. J. Immunol. 22:2855. a negative regulator of T-lymphocyte activation. Immunol. Rev. 153:183. 27. Liu, Y., R. H. Wenger, M. Zhao, and P. J. Nielsen. 1997. Distinct costimulatory 57. Sprent, J. 1990. Antigen-presenting cells for CD8ϩ T cells. Immunol. Rev. 177: molecules are required for the induction of effector and memory cytotoxic T 213. lymphocytes. J. Exp. Med. 185:25. 58. Valle, A., J. P. Aubry, I. Durand, and J. Banchereau. 1991. IL-4 and IL-2 up- 28. Goodwin, R. G., W. S. Din, T. D. Smith, D. M. Anderson, S. D. Gimpel, regulate the expression of antigen B7, the B cell counterstructure to T cell CD28: T. A. Sato, C. R. Maliszewski, C. I. Brannan, N. G. Copeland, N. A. Jenkins, et an amplification mechanism for T-B cell interactions. Int. Immunol. 3:229. al. 1993. Molecular cloning of a ligand for the inducible T cell gene 4–1BB: a 59. Sperling, A. I., J. A. Auger, B. D. Ehst, I. C. Rulifson, C. B. Thompson, and member of an emerging family of cytokines with homology to tumor necrosis J. A. Bluestone. 1996. CD28/B7 interactions deliver a uniques signal to naive T factor. Eur. J. Immunol. 23:2631. cells that regulates cell survival but not early proliferation. J. Immunol. 157:3909. 1142 ALTERNATIVE COSTIMULATORY PATHWAYS FOR CD8 T CELLS

60. Sagerstrom, C. G., E. M. Kerr, J. P. Allison, and M. M. Davis. 1993. Activation 66. Cai, Z., and J. Sprent. 1996. Influence of antigen dose and costimulation on the and differentiation requirements of primary T cells in vitro. Proc. Natl. Acad. Sci. primary response of CD8ϩ T cells in vitro. J. Exp. Med. 183:2247. USA 90:8987. 67. Bluestone, J. A. 1995. New perspectives of CD28–B7-mediated T cell costimu- 61. Dubey, C., M. Croft, and S. L. Swain. 1996. Naive and effector CD4 T cells differ lation. Immunity 2:555. in their requirements for T cell receptor versus costimulatory signals. J. Immunol. 157:3280. 68. Willerford, D., J. Chen, J. Ferry, L. Davidson, A. Ma, and F. Alt. 1995. Inter- 62. Shahinian, A., K. Pfeffer, K. P. Lee, T. M. Kundig, K. Kishihara, A. Wakeham, leukin-2 receptor ␣ chain regulates the size and content of the peripheral lym- K. Kawai, P. S. Ohashi, C. B. Thompson, and T. W. Mak. 1993. Differential T phoid compartment. Immunity 3:521. cell costimulatory requirements in CD28-deficient mice. Science 261:609. 69. Ridge, J. P., F. Di Rosa, and P. Matzinger. 1998. A conditioned can 63. Horak, I. 1996. Interleukin-2-knockout mice: a new model to study autoimmunity be a temporal bridge between a CD4ϩ T-helper and T-killer cell. Nature 393:474. and self-tolerance. Sb. Lek. 97:25. 70. Bennett, S. R. M., F. R. Carbone, F. Karamlis, R. A. Flavell, J. F. A. P. Miller, 64. . Willerford, D. M., J. Chen, I. A. Ferry, L. Davidson, A. Ma, and F. W. Alt. 1995. and W. R. Heath. 1998. Help for cytotoxic-T-cell responses is mediated by CD40 Interleukin-2 receptor ␣ chain regulates the size and content of the peripheral lymphoid compartment. Immunity 3:521. signaling. Nature 393:478. 65. Cerwenka, A., L. L. Carter, J. B. Reome, S. L. Swain, and R. W. Dutton. 1998. 71. Schoenberger, S. P., R. E. M. Toes, E. I. H. Van der Voort, R. Offringa, and In vivo persistence of CD8 polarized T cell subsets producing type 1 or type 2 C. J. M. Melief. 1998. T-cell help for cytotoxic T lymphocytes is mediated by cytokines. J. Immunol. 161:97. CD40-CD40L interactions. Nature 393:480. Downloaded from http://www.jimmunol.org/ by guest on October 1, 2021