The Immunological Synapse and CD28-CD80 Interactions Shannon K

The immunological synapse and CD28-CD80 interactions Shannon K. Bromley1,Andrea Iaboni2, Simon J. Davis2,Adrian Whitty3, Jonathan M. Green4, Andrey S. Shaw1,ArthurWeiss5 and Michael L. Dustin5,6 Published online: 19 November 2001, DOI: 10.1038/ni737

According to the two-signal model of T cell activation, costimulatory molecules augment T cell receptor (TCR) signaling, whereas adhesion molecules enhance TCR–MHC-peptide recognition.The structure and binding properties of CD28 imply that it may perform both functions, blurring the distinction between adhesion and costimulatory molecules. Our results show that CD28 on naïve T cells does not support adhesion and has little or no capacity for directly enhancing TCR–MHC- peptide interactions. Instead of being dependent on costimulatory signaling, we propose that a key function of the immunological synapse is to generate a cellular microenvironment that favors the interactions of potent secondary signaling molecules, such as CD28.

The T cell receptor (TCR) interaction with complexes of peptide and as CD2 and CD48, which suggests that CD28 might have a dual role as major histocompatibility complex (pMHC) is central to the T cell an adhesion and a signaling molecule4. Coengagement of CD28 with response. However, efficient T cell activation also requires the partici- the TCR has a number of effects on T cell activation; these include pation of additional cell-surface receptors that engage nonpolymorphic increasing sensitivity to TCR stimulation and increasing the survival of ligands on antigen-presenting cells (APCs). Some of these molecules T cells after stimulation5. CD80-transfected APCs have been used to are involved in the “physical embrace” between T cells and APCs and assess the temporal relationship of TCR and CD28 signaling, as initiat- are characterized as adhesion molecules. With modified APCs, a dis- ed by natural ligands, and showed that CD28 interactions precede or tinct activity was identified that was required for effective T cell stim- occur simultaneously to TCR interactions6. There are a number of pos- © http://immunol.nature.com Group 2001 Nature Publishing ulation; it was termed costimulation. T cells stimulated in the absence sibilities for the mechanism by which costimulation is initiated by of costimulation were rendered nonresponsive to subsequent antigen CD28. CD28 recruits phosphatidylinositol-3-kinase (PI3K)7 and the stimulation. These observations led to the “two-signal model” of T cell pleckstrin homology domainÐcontaining tyrosine kinase Itk8 and can activation, in which signal one is transduced by the TCR and enhanced activate the Src family kinase Lck through different motifs in the cyto- by adhesion molecules and signal two is generated by costimulatory plasmic domain of CD289. These biochemical activities are not unique molecules on the surface of the APC1,2. to CD28; they are shared by the TCR, which leads to the concept that The first costimulatory receptor to be identified was CD283. CD28 is CD28 intensifies and prolongs biochemical signals that are normally a homodimeric type 1 transmembrane protein that is a member of the generated by the TCR. CD28 and interleukin 1 signaling, which leads immunoglobulin (Ig) superfamily and has a single Ig-like domain. to enhanced cytokine production by T cells, is associated with CD28 interacts with CD80 (also known as B7-1) and CD86 (B7-2), enhanced activation of NF-κB10. which are expressed on the APC in response to activating signals that Compartmentalization of receptors at the interface between the T cell result from, for example, CD40 engagement. The CD28-ligand interac- and APC is correlated with T cell activation. The immunological tion is topologically similar to that of many adhesion molecules, such synapse is characterized by a ring of lymphocyte functionÐassociated 1Ðintercellular adhesion molecule 1 (LFA-1ÐICAM-1) interactions that surround a central cluster of TCR-pMHC interactions11,12; these are 20 pMHC referred to as a central supramolecular activation cluster (cSMAC)11. pMHC + B7 Figure 1. Costimulatory effect of GPI- CD28 is unusual among the nonpolymporphic receptors tested in that 15 CD80 on transgenic T cell activation. 2B4 T cells were incubated for 48 h on pla- its interaction with CD80 occupies the central cluster of the immuno- 10 12

3 nar phospholipid bilayers that contained 160 logical synapse and is colocalized with the engaged TCR . Thus, one molecules/µm2 GPI–ICAM-1 and various mechanism by which CD28 may enhance T cell activation is to stabi- 5 densities of GPI–I-Ek–MCC(91–103), with 3 lize the immunological synapse. Engagement of CD28 promotes the 2

10 cpm [ H]thymidine µ or without 160 molecules/ m of GPI- 13 0 CD80.T cells were then pulsed with 0.4 µCi cytoskeletal-dependent recruitment of cell surface proteins and lipid 0.01 0.1 1 10 100 14 of [3H]thymidine and collected 12 h later. rafts rich in kinases and adaptor proteins , which contribute to building I-E k -MCC(91–103) (molecules/µm2 ) Data are mean±s.e.m. of triplicate wells. the immunological synapse15.

1Department of Pathology and Immunology and 4Department of Medicine,Washington University School of Medicine, 660 S. Euclid Ave, St. Louis MO, USA. 2Nuffield Department of Medicine,The University of Oxford, Oxford, UK. 3Biogen Inc, Cambridge, MA. 5Howard Hughes Medical Institute, Division of Rheumatology, Department of Medicine, University of California at San Francisco, San Francisco, CA, USA. 6Department of Pathology and the Program in Molecular Pathogenesis, New York University School of Medicine and The Skirball Institute of Biomolecular Medicine, 540 First Ave, New York, NY, USA. Correspondence should be addressed to M. L. D. ([email protected]).

Figure 2. Naïve T cells do not adhere complex formed between I-Ek and the peptide formed by amino acids 80 to CD80 substrates. 2B4 T cells were (aa) 91Ð103 of moth cytochrome c, termed MCC(91Ð103) (which is settled onto planar phospholipid bilayers recognized as an agonist pMHC by T cells expressing the 2B4 TCR)12. 60 that contained 500 molecules/µm2 of CD48 or GPI-CD80 preincubated at 37 In this system the live T cells from 2B4 TCRÐtransgenic mice interact 40 °C. Images of the input cells were with the functional planar bilayer substrate, which essentially replaces 12 Adhesion (%) 20 acquired. After 20 min, the bilayers were the APC . Previously in vitroÐprimed effector T cells were used for gently washed with a laminar flow and the immunological synapse formation studies with bilayers12,21. 0 percentage of input cells that adhered to CD48 CD80 Immunological synapse formation in naïve T cells has not been report- Bilayer the bilayers was determined. Data are representative of two experiments. ed with the minimal bilayer system; however, optimal proliferation of naïve T cells in response to planar bilayers that contained agonist pMHC required CD28 cross-linking with anti-CD2822. CD28 could alternatively enhance TCR signaling through its func- We examined here the role of CD28-CD80 interactions in TCR- tion as an adhesion molecule within the immunological synapse16. The pMHC recognition and immunological synapse formation. When TCR must engage rare antigenic ligands presented on the surface of an CD28 was highly expressed, as it was on some Jurkat subclones, it APC, but this interaction is of low affinity. In addition, the TCR and mediated adhesion with a high 2D affinity but unexpectedly low max- MHC molecules are small compared to surrounding abundant cell sur- imum binding. When CD28 was presented on the cell and CD80 was face molecules such as integrins, CD43 and CD45. The distance presented in the planar bilayer substrate, only one-third of the CD28 spanned by the interaction of CD28 with CD80 or CD86 (∼15 nm), interacted with CD80 in the contact area. In comparison, when CD2 which occurs between the T cell and APC membranes, is similar to that is presented on Jurkat cells and CD58 is presented on the bilayer, over spanned by the TCR-pMHC interaction4,17. Thus, the interaction two-thirds of the CD2 interacts with CD58 in the contact area18. Low between CD28 and CD80 could generate the appropriate spacing for lateral mobility of CD28 was confirmed by fluorescence photo- TCR to efficiently interact with pMHC4. The interaction of CD2 on T bleaching recovery experiments and attributed to sequences in the cells with CD58 on the APC has a high two-dimensional (2D) affini- cytoplasmic domain of CD28. In cells that expressed physiological ty4,18. This is consistent with the creation of an ordered contact area in amounts of CD28, this low mobility prevented the effective interac- which CD2-CD58 interactions hold the membranes at a uniform dis- tion of CD28 with CD80. However, CD28-CD80 interactions were tance and effectively concentrate the respective binding sites in an atto- detected in contacts formed by CD2-CD48 and were focused within liter (10Ð18 l) volume. Thus, CD2 promotes TCR engagement in a size- the central cluster of immunological synapses formed by naïve T dependent manner, which suggests that topological mechanisms might cells. In contrast, engagement of CD28 was not required for synapse be important in T cell stimulation19. Whether CD28-CD80 interactions formation nor did it determine the density of MHC molecules that enhance TCR interactions with pMHC as part of its activity has not accumulated within the central cluster of the synapse. Rather than been tested directly. regulating the TCR-MHC interaction in the manner of an adhesion © http://immunol.nature.com Group 2001 Nature Publishing Immunological synapses can be studied with the supported planar molecule, it appeared instead that TCR engagement and immunolog- bilayers system20. The advantage of this system is that the interaction of ical synapse formation triggered the local interaction of CD28 with cell surface receptors with fluorescently labeled ligands in the planar CD80. Our results suggested that formation of the immunological bilayer on a microscope coverglass substrate can be directly measured synapse by naïve T cells requires only pMHC (signal one) and adhe- by fluorescence microscopy18. Formation of immunological synapses sion, but allows integration of pMHC (signal one) and CD80 (signal and proliferation are stimulated by bilayers that contain ICAM-1 two)Ðtriggered signals by establishing a physical environment in (which is bound by the T cell adhesion molecule LFA-1) and a pMHC which CD28-CD80 interactions are facilitated.

b c d e 240 a 50 15 8000 12 40 6000 10 160 2 10 30 m ) 8 RU µ 4000 6 20 B/F 5 80 2000 4 10 area ( 2 Average contact Cell adhesion (%) 0 0 CD80 bound/contact 0 0 0 0 100 200300 400 500 0 100 200300 400 500 0 100 200300 400500 024681012 02040 GPI-CD80 GPI-CD80 GPI-CD80 B*p Time (s) (molecules/µm2 ) (molecules/µm2 ) (molecules/µm2 )

Figure 3. CD28-CD80 binding parameters. Jurkat cells were injected into a flow cell and f 300 g 150 incubated for 30 min at 37 °C on planar phospholipid bilayers that contained Cy3-CD80 and Cy5- 200 CD48. (a) The percentage of adherent cells was determined by comparing the interference reflec- 100 K = 1.8 µM tion and transmitted light images. (b) Average contact area. (c) Average number of CD80 mole- d µ 100 K = 2.0 M Bound/Free cules bound per contact. (d) 2D Kd plot. B/F denotes [bound CD80]/[free CD80]; B*p denotes d 50 Bound (RU) [bound CD80]*(contact area/cell area).At least 50 cells were analyzed for each datapoint. 2D Kd µ 2 0 values from these experiments were 0.6–0.8 molecules/ m . Data are representative of three 0 experiments. (e–g) 3D binding affinity of murine sCD80 to immobilized hCD28-Fc. (e) Injections 0 61218 0 100 200 300 Concentration (µM) Bound (RU) of sCD80 at 37 °C started at 16 µM and were followed by seven twofold dilutions, flowing over hCD28-Fc immobilized at a concentration of 3000 response units (RU).The curves represent total specific binding after subtraction of the background responses observed in a control flow cell with immobilized SLAM protein.A nonlinear fit of the Langmuir binding isotherm53

(f) and the linear analysis of a Scatchard plot (g) yielded a Kd of 2 µM. Comparison of the membrane and solution affinities yielded a confinement region of 3 nm.

Figure 4. Low CD28 mobility reg- ulated by its cytoplasmic tail. a b c Photobleaching recovery of CD28 on 125 80 Extracellular TM Cytoplasmic Jurkat and peripheral blood T cells.(a) CD2 100 CD8 Peripheral blood T cells were labeled CD28 60 at 4 °C with FITC–anti-CD2 or FITC– 75 40 CD28 anti-CD28. Cells were left to settle 50 CD8 onto coverslips coated with 10 µg/ml 25 20 CD28 of the nonactivating anti–LFA-1. In 0 0 terms of the ability to attach the T Fluorescence recovery (%) 0 100 200 300 400 CD2CD28CD2CD28 020406080 Fluorescence recovery (%) Jurkat PBL cells to the coverslip, similar results Time (s) Recovery (%) were obtained with poly(L-lysine), but anti–LFA-1 provided a more secure attachment in protein that contained media. Photobleaching was done at 24 °C. Bleaching recovery kinetics are shown as the percentage recovery of fluorescence after photobleaching of CD28 or CD2. Data are representative of at least eight cells in two independent experiments. (b) Average recovery of CD2 and CD28 on Jurkat cells and peripheral blood T cells (PBLs).The lateral mobility was calculated 300 s after bleaching. (c) Fluorescence recovery after photobleaching of a CD8-CD28 chimera and of truncated CD8 and CD28 expressed in HUT cells. Cells were labeled at 4 °C with FITC–anti-CD8 or FITC–anti-CD28. Data are representative of at least ten cells.

naïve T cells (Fig. 2). Naïve T cells also adhered poorly to substrates Results with 500 molecules/µm2 of ICAM-1, but adhesion in this case was GPI-CD80 costimulates naïve T cell activation increased by treatment of cells with phorbol myristate acetate (PMA) To examine the role of CD80 in adhesion and immunological synapse (S. Bromley and M. Dustin, unpublished data). In contrast, treatment of formation we generated a glycosylphosphatidylinositol-anchored naïve T cells with PMA had no effect on adhesion to a CD80-contain- CD80 (GPI-CD80) and labeled it with the fluorescent dye Cy3 to ing substrate (data not shown). allow visualization by fluorescence microscopy. To confirm the cos- To better understand this lack of adhesion of naïve T cells to CD80- timulatory activity of GPI-CD80, planar phospholipid bilayers that containing bilayers, we determined the amount of CD28 expressed on contained GPIÐI-EkÐMCC(91Ð103), ICAM-1 and CD80 were pre- naïve T cells. Naïve T cells expressed ∼1500 molecules of CD28 per pared; ICAM-1 and CD80 were both present at densities of 160 mol- cell (∼6 molecules/µm2), a much lower amount than for CD2 and LFA- ecules/µm2. I-EkÐMCC(91Ð103) was titrated from densities of 70Ð0.07 1, which are both present at ∼20,000 molecules per naïve T cell. Human molecules/µm2. T cells were purified from 2B4 TCRÐtransgenic mice, Jurkat T cells, however, expressed ∼22,000 CD28 molecules per cell settled onto the bilayers and their activation was measured by tritiated (∼30 molecules/µm2), so these were used to evaluate the 2D affinity of [3H]thymidine incorporation. T cells proliferated in response to bilay- the interaction between CD28 and CD80. This functional compatibili- ers that contained ICAM-1 and I-EkÐMCC(91Ð103) alone. Inclusion of ty of the mouse and human CD28-CD80 interaction systems was in CD80 in the bilayers enhanced T cell activation in response to densi- agreement with published data25. © http://immunol.nature.com Group 2001 Nature Publishing ties of I-EkÐMCC(91Ð103) as low as 0.7 molecules/µm2 (Fig. 1). This result was consistent with published analyses of APCs that expressed CD28-CD80 achieves a high 2D affinity CD8023,24. We therefore deduced that the GPI-CD80 in the planar To determine the membrane affinity of the CD28-CD80 interaction, bilayer system was biologically active. Jurkat subclone JE2.1 T cells were left to interact for 20 min with pla- Because CD80 in planar bilayers costimulated naïve T cell prolifer- nar phospholipid bilayers that contained increasing densities of Cy3- ation, we next determined whether naïve T cells could adhere to CD80- conjugated GPI-CD80. Within 20 min, the CD28-CD80 interaction containing bilayers. As a positive control, naïve T cells were tested for reached binding equilibrium (data not shown). We measured the per- adhesion to bilayers that contained 500 molecules/µm2 of CD48. As centage of input Jurkat T cells that adhered at each initial CD80 den- expected, a high percentage of naïve T cells adhered to the CD48 bilay- sity in the bilayer. As expected, adhesion was dose-dependent and very ers. In contrast, we found that planar phospholipid bilayers that con- efficient at 500 molecules/µm2 of CD80 (Fig. 3a). This result was also tained CD80 at 500 molecules/µm2 did not support the adhesion of used to normalize the bound CD80 density in the cell contact areas based on the assumption that cells with the highest CD28 expression a b will preferentially adhere to bilayers that contain low initial CD80 densities18. Similarly, the average area of cell contact with the bilayer 100 increased with increasing initial CD80 densities in the bilayer (Fig. 75 3b). The number of CD80 molecules bound in the cell contact area also increased with increased CD80 density (Fig. 3c). Further analysis 50 of this data, with a linearized plot or nonlinear fitting method, yielded 25 a membrane affinity of 0.8 molecules/µm2 and maximal binding of Counts per channel 7300 molecules per cell (Fig. 3d), a relatively high affinity that was Adhesion (% of input) 0 40 80 120 160 200 0 1 10 10 2 10 3 10 4 Truncated CD28 WT CD28 similar to that observed for CD2-CD58 interactions in the planar 18 RFI bilayer system . The CD28-CD80 interaction had similar dimensions to the TCR- Figure 5.The CD28 cytoplasmic domain of CD28 limited interaction with pMHC interaction, prompting the proposal that CD28-CD80 interac- CD80. (a) Expression of wild-type CD28 (dashed line) and truncated CD28 (solid tions might hold APC and T cell membranes at a uniform distance ideal line) in HUT cells. Isotype-matched control mAb staining of HUT cells (dash-dotted for enhancing “scanning” TCR interactions with pMHC. The effective line). (b) HUT cells that expressed either wild-type CD28 or truncated CD28 were intercellular volume for the interactions of membrane proteins was cal- left to interact for 20 min at 37 °C with bilayers that contained 500 molecules/µm2 of CD28.The percentage of input cells that adhered to the bilayers was calculated as culated from the 2D and 3D (solution) affinities for receptor-ligand in Fig. 2. Data are representative of two experiments. pairs. Surface plasmon resonanceÐbased analysis of the specific binding

Figure 6. CD28 engagement had 250 was then irreversibly photobleached, and the fluorescence recovery no effect on I-Ek or ICAM-1 monitored over time to determine the percentage of laterally mobile engagement within the immuno- 200

) molecules. For both the Jurkat and peripheral blood T cells, 30% of logical synapse. Naïve 2B4 T cells 2 m 150 were settled onto bilayers that con- µ the CD28 was laterally mobile (Fig. 4a,b). In contrast, 60% of the tained Cy5-conjugated GPI–ICAM-1 100 CD2 was laterally mobile (Fig. 4a,b), which was consistent with pub- or Oregon green–conjugated GPI–I-Ek lished data27. (0.7 molecules/µm2), with or without 50 To determine whether the CD28 cytoplasmic domain was responsi- Accumulated receptors Cy3-conjugated GPI-CD80. After 20 0 ble for its limited lateral mobility, equivalent photobleaching experi-

min of interaction, images were (molecules/ I-Ek ICAM I-Ek ICAM acquired. The density of accumulated CD80 None ments were done on HUT cells transfected with truncated CD28 (in ICAM-1 (solid bars) and I-Ek (hatched which the 40-aa cytoplasmic domain was deleted) and CD28-CD8 bars) was determined.To calculate the chimeric receptors (Fig. 4c). Truncated CD8 receptors (which also density of bound I-Ek, the density of accumulated I-Ek in the area of LFA-1 accumula- lacked their cytoplasmic domain) showed 50% lateral mobility. tion was subtracted from the total accumulation in the central cluster of the immunological synapse. Data are representative of two experiments. Addition of the CD28 cytoplasmic domain to the CD8 receptor (CD28- CD8 chimeras) reduced the mobility to 20%, which showed that the cytoplasmic domain was sufficient to impart reduced mobility. of soluble mouse CD80 to immobilized human CD28 yielded a solution Removal of the cytoplasmic domain from CD28 increased lateral

Kd of 2 µM (Fig. 3eÐg). Comparison of this 3D Kd with the 2D Kd pre- mobility to 60%, which showed that the cytoplasmic domain of CD28 dicted that the CD28-CD80 interaction occurred in a volume equivalent was necessary for the low mobility of CD28. to the contact area multiplied by a height of 3 nm (see Web Table 1 on To examine the effect of the CD28 cytoplasmic domain on the the supplementary information page of Nature Immunology on line), CD28-CD80 interaction, we measured the adhesion of HUT cells that reflecting a high degree of order in the contact area. Therefore, the expressed either wild-type CD28 or truncated CD28, to CD80-contain- CD28-CD80 interaction could, in principle, establish ideal conditions ing bilayers. The HUT cells expressed similar amounts of CD28 com- for scanning TCR-pMHC interactions. However, the reverse scenario, pared to resting human peripheral blood T cells or naïve mouse T cells that is, that TCR-pMHC and/or CD2-CD48 interactions might enhance (data not shown) and the two different CD28 receptors were similarly CD28-CD80 binding, was also possible. expressed on the HUT cells (Fig. 5a). As with resting human peripheral blood T cells, <10% of HUT cells that expressed wild-type CD28 Restrictions on CD28 mobility receptors bound to the CD80-containing bilayers. In contrast, >80% of The capture of cell surface receptors by their ligands in the interface HUT cells that expressed truncated CD28 adhered to the bilayers (Fig. formed between interacting cells depends on the mobility of the recep- 5b). Thus, the cytoplasmic domain of CD28 profoundly inhibited adhe- tors26. We analyzed the number of bound CD80 molecules in the con- sion mediated by CD28-CD80 interactions. tact areas formed by Jurkat cells that were interacting with GPI- © http://immunol.nature.com Group 2001 Nature Publishing CD80Ðcontaining bilayers. These data suggested that only a fraction of CD28-CD80 and the immunological synapse the CD28 receptors was free to diffuse into the contact area to interact The presence of CD80 in the planar bilayer enhanced the proliferative with ligand, as only 7,300 of the 22,000 molecules on the T cell surface response of naïve 2B4 T cells to planar bilayers that also contained were engaged (Fig. 3d). agonist pMHC and ICAM-1 (Fig. 1). This could have been due to To directly characterize the lateral mobility of CD28 on T cells flu- enhanced immunological synapse formation or enhanced signaling orescence, photobleaching recovery experiments were done. Jurkat or after immunological synapse formation. To investigate CD80 depen- human peripheral blood T cells were labeled with fluorescein isoth- dence on TCR ligation by pMHC, we determined the overall frequen- iocyanate (FITC)-conjugated anti-CD28, initially at 4 ¼C to limit cy of mature immunological synapse formation and the extent of internalization of the antibody. After labeling, the cells were settled pMHC and ICAM-1 engagement within the synapses formed between onto coverslips coated with TS2/4, a nonactivating antiÐLFA-1. A 1 naïve 2B4 T cells and bilayers that contained ICAM-1 and I-Ek in the µm2 area of fluorescently labeled CD28 or CD2 on the T cell surface presence or absence of CD80. A mature immunological synapse was

a I-Ek CD80 ICAM-1 Overlay b

) 200 200 200 2

CD80 m 150 150 150 µ 100 100 100 50 50 50 Receptors 0 0 0 (molecules/ 0 5 10 15 20 0 5 10 15 20 0 5 10 15 20 No CD80 Time (min)

Figure 7. CD80 engagement in the central cluster was not required for formation of the immunological synapse. 2B4 T cells were settled onto bilayers that contained ICAM-1 or I-Ek–MCC(91–103), with or without CD80. (a) Accumulation of ICAM-1 (blue), I-Ek–MCC(91–103) (green) and CD80 (red) in the pattern of a mature immunological synapse are shown, as is the overlay. Data are representative of two experiments. (b) Time-course of ICAM-1, CD80 and I-Ek accumulation.The density of accumulated ICAM-1 (blue circles), CD80 (red triangles) and I-Ek (green squares) was determined.The density of bound CD80 and I-Ek was determined as in Fig. 2. Data are representative of five cells; three separate experiments are shown.

defined as a central cluster of accu- Table 1.The effect of adhesion molecules and immunological synapse formation on mulated pMHC complexes sur- CD28-CD80 interaction rounded by a ring of accumulated ICAM-112. The bilayers contained Bilayers Density of CD28-CD80 Contact area Total CD28-CD80 Predicted CD28-CD80 I-EkÐMCC(91Ð103) at 0.7 mole- interactions (µm2) interactions interactions cules/µm2, this was the lowest den- (molecules/µm2) (molecules) (molecules) sity of agonist pMHC that promot- ICAM-1 + CD80 0 7.2 0 0 ed formation of mature immuno- CD48 + CD80 15 1.3 19 260 logical synapses (data not shown) ICAM-1 + pMHC + CD80 31 0.6 19 120 and T cell proliferation in the presence of CD80 (Fig. 1). Unexpect- The experimentally determined density, contact areas and total CD28 engagement were measured at an initial CD80 density of 160 molecules/µm2 in the presence of ICAM-1, CD48 or ICAM-1 + agonist pMHC. Data for CD28+/+ and CD28–/– cells edly, CD80 did not change the fre- were collected, averaged and the difference between CD28-CD80 interactions determined as the difference in CD80 density quency of immunological synapse between the CD28+/+ and CD28–/– cells.This value was compared to the predicted engagement of CD28, based on a simple formation by the naïve T cells, mass action model, using the measured contact area and the 2D Kd value measured with Jurkat cells.The equation used to × × µ 2 which remained at 12% of the input calculate this was: theoretical number = (area [free CD28] [free CD80]) ÷ Kd. Free CD28 = 1 molecule/ m , free CD80 2 2 = 160 molecules/µm and the 2D Kd = 0.8 molecules/µm .The 2D Kd values for the human CD28–mouse CD80 and the cells that formed mature immuno- mouse CD28–mouse CD80 interactions were essentially identical (A. Iaboni and S. J. Davis, unpublished observations).The logical synapses. In addition, equi- observed engagement was less than predicted. valent amounts of ICAM-1 (150 molecules/µm2) and pMHC (50 molecules/µm2) were engaged in the presence and absence of CD80 Cy5ÐICAM-1, Oregon greenÐconjugated I-EkÐMCC(91Ð103) and (Fig. 6). Thus, CD28 engagement did not enhance either immunolog- Cy3-conjugated CD80. Upon contacting their T cell ligands, ICAM-1 ical synapse formation or the interaction of the TCR with pMHC. and I-EkÐMCC(91Ð103) relocalized within the interface to form SMACs. Two patterns of ICAM-1 engagement by the naïve T cells CD28-CD80 interactions in close contact areas were seen. Either ICAM-1 was engaged in a peripheral ring that sur- CD80 enhanced naïve T cell responses to pMHC and ICAM-1 that were rounded the engaged pMHC or ICAM-1 showed punctuated engage- coinserted in the planar bilayers. We, therefore, determined the effects ment throughout cell contact with multiple regions of ICAM-1 exclu- on CD28-CD80 interactions of antigen-dependent and -independent T sion. Engaged I-EkÐMCC(91Ð103), on the other hand, accumulated into cell contacts with the bilayers. As with human peripheral blood T cells a small central cluster, as described for effector T cells12. Invariably, (data not shown), we did not detect adhesion of naïve murine T cells to CD80 localized with engaged I-EkÐMCC(91Ð103) in the central cluster bilayers that contained CD80 alone (Fig. 2). To detect very weak CD28 of the immunological synapse (Fig. 7a). CD80 and I-EkÐMCC(91Ð103) interactions in the context of other adhesion molecules, we compared were engaged with similar kinetics. Whereas ICAM-1 receptors were © http://immunol.nature.com Group 2001 Nature Publishing the contacts formed by CD28+/+ and CD28Ð/Ð T cells with bilayers that engaged within seconds of contact with the planar bilayers, CD80 and contained CD80 and/or CD48, ICAM-1 and pMHC. T cells in CD28Ð/Ð I-EkÐMCC(91Ð103) interactions with their respective ligands took min- mice develop normally28. Therefore, it was likely that the only difference utes (Fig. 7b). The number of CD80 molecules engaged within the cen- in mature T cells from CD28+/+ and CD28Ð/Ðmice was in expression of tral cluster of the immunological synapse was still small (with an aver- CD28. Thus, contacts of CD28Ð/Ð T cells provided an ideal background age of ∼20, Table 1). But the density of engaged CD80 molecules (31 for calculating CD28 engagement by CD28+/+ T cells. The determination molecules/µm2) within the immunological synapse was more than dou- of background-free CD80 was critical, as a variable amount of CD80 ble that seen in contacts enhanced by CD2-CD48 interactions (Table (10Ð20%) was excluded from the contact area formed by CD2-CD48 1). This implied that conditions for CD28-CD80 binding within the interaction. With this sensitive approach we did not detect interaction of immunological synapse were enhanced over those in CD2-CD48 con- CD28 with CD80 in bilayers that contained ICAM-1 and CD80 (Table tacts. Finally, although CD80 and I-EkÐMCC(91Ð103) localized to the 1). However, CD28-CD80 interaction was detectable in contacts formed same region of the immunological synapse and were engaged with sim- by naïve T cells that interacted with bilayers containing CD48 and CD80 ilar kinetics, with regard to the movement and degree of engagement of (Table 1). This result was consistent with colocalization of CD28-CD80 I-EkÐMCC(91Ð103) and ICAM-1, the synapses that formed in the pres- interactions with CD2-CD58 interactions in contacts between Jurkat ence of CD80 engagement were indistinguishable from those that cells and planar bilayers containing ICAM-1, CD58 and CD80 (R. formed in its absence (Fig. 7a). Burack, unpublished observation). The stimulation of CD28-CD80 interaction by adhesion molecules was sensitive to molecular size, as Discussion CD28-CD80 interactions were not enhanced by the interaction of LFA- CD28 was hypothesized to have a dual function as both an adhesion 1 with ICAM-1, which spans a longer distance between cells than and signaling molecule4. CD28 engagement triggers cytoskeletal and CD28-CD80, but were enhanced by CD2-CD48, which spans a similar membrane rearrangements that result in stabilization of the immuno- distance to CD28-CD804,29. Nevertheless, in this context, CD80 engage- logical synapse13,14. We found that CD28-CD80 interactions were able

ment was 10% of that predicted by the 2D Kd measured with the Jurkat to generate a high 2D affinity through the precise alignment of cell cells (Table 1). This suggested that CD2-CD48 contacts were not per- membranes. In addition, engaged CD28 accumulated in the central fectly compatible with the CD28-CD80 interaction. cluster of the immunological synapse. Thus, the CD28-CD80 interaction appears well suited to enhancing TCR engagement with pMHC in CD28-CD80 location in the immunological synapse the manner of other adhesion molecules, such as CD2. Contrary to this To localize engaged CD80 in relation to other well characterized com- hypothesis4, however, we found that CD28 engagement did not facili- ponents of the immunological synapse, we left naïve 2B4 T cells to tate TCR-MHC interactions and was not required for immunological interact with planar phospholipid bilayers that contained GPI-anchored synapse formation. Instead, the adhesive capacity of the CD28-CD80

interaction appeared to be negligible. In addition, we observed that higher than the actual numbers for a specific contact area, then the impli- topologically equivalent adhesive interactions, such as those involving cation is that the contact area is not ideal for CD28-CD80 interaction. CD2 and CD48, were required for detectable CD28-CD80 engagement Although the CD2-CD48 interaction enhanced CD28-CD80 interactions,

to take place and that these effects were augmented by immunological it was at only 7% of that predicted by the 2D Kd. Even in the mature synapse formation. The interaction of CD28 with CD86 was likely even immunological synapse where CD28-CD80 engagement was higher, it more dependent on immunological synapse formation than the interac- was still only 16% that expected in a self-assembled contact area. These tion of CD28 with CD80, as the affinity of the former pair was 10Ð20% few engaged CD28 molecules had very little physical impact on the inter- that of the latter (A. Iaboni and S. J. Davis, unpublished data). actions of the TCR, which was present at much higher densities and in Therefore, rather than enhancing TCR engagement and initiating vast excess compared to ligand. The small number of engaged CD28 synapse formation, we concluded that the immunological synapse more molecules emphasized the potency of costimulatory signaling. likely established a microenvironment in which functional CD28-CD80 The results we obtained with CD28hi Jurkat cells and the truncated or CD28-CD86 interactions were initiated and/or enhanced. CD28 HUT cell line showed that the limited CD28 engagement was a It is possible that immunological synapse formation requires other product of the low density of CD28 on naïve T cells and a cytoplasmic signals, such as stimulation by chemokines. However, we found that domainÐdependent restriction in its lateral mobility. It is also possible, compared to 30% of effector T cells21, 12% of freshly isolated L- however, that the intermembrane distance generated by CD2-CD48 selectin+ splenic T cells were able to form immunological synapses in interactions or within the immunological synapse were not ideal for the bilayer with only ICAM-1 and pMHC. Although chemokines make CD28-CD80 interactions. This seems unlikely, however, given recent this process more efficient, they are not essential (S. K. Bromley and structural analyses of cytotoxic T lymphocyte associated antigen 4 M. L. Dustin, unpublished observations). We found that CD28 engage- (CTLA-4)ÐCD80 and CTLA-4ÐCD86 complexes17,34. Another possibil- ment was not required for immunological synapse formation by naïve ity is that crowding effects impeded the movement of free CD28 and cells. Thus, immunological synapse formation may be synonymous CD80 or that incompatible membrane microdomains were recruited to with successful generation of signal one for naïve T cell activation. the contact area. For example, engaged CD2 and TCR are partially Other studies have suggested that CD28 costimulation could be due localized to the liquid-ordered phase lipid structures known as rafts35Ð37, to CD28-mediated enhancement of adhesion between T cells and whereas CD28 is not37. Regardless of the detailed mechanism, our APCs. Chinese hamster ovary (CHO) cells transfected with CD28 results showed that maximum CD28 engagement was likely strongly adhere to activated B cells, and this adhesion is blocked by anti-CD8030. favored by formation of the synapse and that it occurred subsequent to, Similar studies with conjugates formed between T cells and CHO cells and was dependent upon, initial TCR signaling. transfected with CD80 show that CD28 engagement induces focal Why is a large fraction of CD28 immobile and apparently inactive? adhesion-like contacts at sites of Rac and talin accumulation31. Two methods of measuring mobility—2D binding analysis of CD28- However, all these results were obtained with the CD28 or CD80 over- CD80 interactions and fluorescence recovery after photobleaching— expression systems. Our results showed that naïve T cells did not both indicated that ∼70% CD28 was immobile and therefore did not © http://immunol.nature.com Group 2001 Nature Publishing adhere to substrates that contained physiological amounts of CD80 but migrate to the contact area. It is probably important that CD28 signaling could be engaged in contacts formed via the antigen-independent inter- is not activated in a capricious manner. The ligation of CD28 can acti- action of the adhesion molecules CD2 and CD48 and, at higher densi- vate Lck and PI3K pathways and, in some cases, fully activate T cells. ty, in the immunological synapse. Although we found that physiologi- Studies that used mitogenic antibodies to cross-link CD28 showed that cal engagement of CD28 takes place in the center of the synapse, we CD28 can transduce signals that stimulate T cells to proliferate inde- cannot completely rule out a role for CD28-CD80 interactions outside pendently of TCR signal transduction38. In accordance with our findings, the immunological synapse of an antigen-stimulated T cell. the activity of this mitogenic antibody is dependent on cytoskeletal These results emphasize a new view of costimulation, but are rearrangements induced by TCR signaling: in the absence of TCR trig- nonetheless compatible with a role for CD28 in immunological synapse gering, 100-fold more antibody is required for mitogenic activity. formation13,14. Although those studies focused on the delivery of rafts or We propose that, when T cells make contact with APCs in the membrane molecules to the synapse13,14, we examined receptor engage- absence of antigen, immobile CD28 limits the degree of CD28 signal- ment and immunological synapse formation in a direct manner with the ing and renders CD28 signaling dependent upon formation of the passive planar-bilayer system. Our results showed that CD28-CD80 synapse. We had no evidence that TCR engagement and signaling interactions promoted T cell proliferation without altering the initial induced a global change in CD28 mobility, as appears to be the case for TCR-pMHC interactions or the gross organizational properties of the the regulation of LFA-1 mobility39, because the overall number of nascent immunological synapse. It seems likely, however, that the engaged CD28 molecules in CD2-CD48Ðinduced contacts and in the synapse is a dynamic, evolving structure and that the initial stages of its synapse were the same. However, we cannot rule out the possibility that development—which are characterized by TCR and adhesion molecule TCR engagement triggered local changes in CD28 mobility or even movements—might well be followed by the costimulation-dependent active clustering of CD28 at the center of the synapse in order to direct- accumulation of membrane components and kinases that stabilize the ly enhance CD28-CD80 interactions. synapse or lead to its differentiation. Our results also supported pub- Our results strongly suggest that signal one and signal two might be lished data that showed that CD28 engagement has no effect on the rate sequentially processed by T cells, prompting a new view of costimula- of TCR down-regulation5,32 (a surrogate marker for TCR engagement33). tion. According to this new view, the T cell initially focuses on the qual- Our study raises the question of how TCR engagement is so effective- ity and quantity of pMHC on the APC surface and, if these exceed some ly insulated from the process of CD28 engagement. A possible explana- preset threshold, the T cell forms the synapse as an active process in the tion arises from the quantitative analysis of interactions in the synapse. course of signal one initiation. Once the T cell embarks on this program,

Based on the law of mass action, the 2D Kd for CD28-CD80 interactions the nascent immunological synapse generates a cellular microenviron- allows the predicted amounts of CD28-CD80 interaction in different ment that expressly favors signaling via several potent secondary costim- types of contact areas to be calculated. If these predicted numbers are ulatory molecules. These molecules amplify TCR signaling and initiate

the process of cell commitment. We have examined CD28 here, but it is buffer) for 1 h at 4 ¡C. The lysates were then sonicated to complete solubilization of the likely that the interactions of other costimulatory molecules—such as GPI-anchored proteins. The lysate was ultracentrifuged at 100,000g for 1 h and the super- natant was passaged through a 0.22-µm membrane (Corning, Corning, NY). The clarified those of the tumor necrosis factor (TNF)-TNF receptor and cytokine- lysate was then incubated with the appropriate mAbs (1610A1 for CD80, YN1/1 for ICAM- cytokine receptor families—are also favored by conditions within the 1 and 14-4-4 for I-Ek), which were covalently coupled to agarose at 4 ¡C overnight so that immunological synapse. We propose that adhesion-facilitated TCR trig- the ligand-binding sites of these proteins were protected. The mAb agarose with bound antigens was then washed extensively with 0.1 M NaHCO3 pH 9, 0.1% Triton X-100 (labeling gering, immunological synapse formation and synapse-facilitated cos- buffer) before the bound proteins were fluorescently labeled, following the manufacturer’s timulatory signaling represent three key steps in a series of obligatory protocols, with Cy3 (for CD80) (Amersham Pharmacia), Cy5 (for ICAM-1 and CD48) checkpoints that lead to full T cell activation and commitment. (Amersham Pharmacia) and Oregon green (for I-Ek) (Molecular Probes, Eugene, OR). The free dye was extensively washed from the agarose with Tris saline with 1% Triton X-100 and then Tris saline with 1% n-octyl β-D-glycopyranoside. These proteins were then eluted from the sepharose column with 50 mM glycine pH 3, 0.15 M NaCl, 1% n-octyl β-D-glu- Methods copyranoside. The purified protein was then reconstituted into liposomes by detergent dial- Cells and monoclonal antibodies (mAbs). Jurkat cells selected for the ability to produce ysis with 0.4 mM (final concentration) egg phosphatidylcholine55. Planar bilayers were interleukin 2 in response to TCR cross-linking were provided by A. Chan (Washington formed by sandwiching liposome droplets between glass coverslips in a parallel plate-flow University, St. Louis, MO). Jurkat cells that were expressing full-length murine CD28 or a cell (Bioptech, Butler, PA). I-Ek was loaded with MCC(91Ð103) at 37 ¡C for 24 h. The truncated form of murine CD28 (in which the last 16 cytoplasmic aa were deleted) were as bilayers were then blocked with 5% nonfat dried milk for 20 min at room temperature. The described9. HUT cells that were expressing full-length human CD28, truncated human fraction of mobile receptors within the bilayers was determined by monitoring fluorescence CD28, truncated human CD8 (which was also missing the cytoplasmic domain) or chimeric recovery after photobleaching. CD28-CD8 (in which the ectodomain and transmembrane domain of CD8 were linked to the CD28 cytoplasmic domain) were as described40. 2B4 TCRÐtransgenic T cells were puri- T cell activation assays. Planar phospholipid bilayers with GPIÐI-Ek (at various densities) fied from the spleen of transgenic mice41 with nylon wool columns, after which and GPI-CD80 (170 molecules/µm2) were formed on 5-mm glass coverslips. I-Ek was macrophages were depleted by adherence to plastic. loaded with agonist peptide as described12. Briefly, I-Ek was pulsed with 100 µM The following mAbs were used TS2/4 (antiÐhuman LFA-1)42, 1610A1 (antiÐmurine MCC(91Ð103) at 37 ¡C for 24 h. Lower densities of agonist-loaded complexes were formed CD80)43, YN1/1 (antiÐmurine ICAM-1)44, 14-4-4 (antiÐmurine I-Ek)45, CD2.1 (antiÐhuman by mixing in the null peptide in which residue 99 of MCC(91Ð103) is changed from a K to CD2)46, D4 (antiÐI-Ek, from M. Davis, Stanford, CA)47, CD28.2 (antiÐhuman CD2848), RPA- an A (MCC99A) while maintaining the total peptide concentration at 100 µM. Molecular T8 (antiÐhuman CD8, BD PharMingen, San Diego, CA) and OX78 (antiÐmouse CD48)49. densities of CD80 and I-EkÐMCC(91Ð103) were determined by radioimmunoassay with iodinated 1610A1 and D4 mAbs, respectively, for detection. T cells (5×104) purified from Expression of GPI-anchored murine CD80. To obtain DNA encoding the extracellular 2B4 mice were added to each well in 100 µl RPMI 10% fetal bovine serum (FBS). T cells portion of murine CD80, PCR was done with the vector pCDM8.mCD8050 as a template were incubated on the substrates for 48 h and then pulsed with 0.4 µCi [3H]thymidine. After (from G. Freeman, Dana Farber Cancer Institute, Boston, MA). The 5′ primer 12 h, cells were collected onto filter mats and titrated [3H]thymidine incorporation was (GATTCCTGGataTCCCCATCA) introduced an EcoRV site (restriction site underlined) quantified by scintillation counting. into the CD80 5′-untranslated region. The 3′ primer (AAGTGaagCTTGCTATCAGG) spanned the 3′ end of the extracellular portion of CD80 and the 5′ end of the transmembrane Two-dimensional affinity analysis. Two-dimensional affinity analysis was modified from region, introducing a HindIII site at the juncture. Lower case letters indicate the bases that a described procedure18. Planar bilayers that contained Cy3-conjugated GPI-CD80 and Cy5- were changed from the native murine CD80 sequence. The PCR product was digested with conjugated murine CD48 were formed in a parallel plate-flow chamber. Murine CD48, HindIII and EcoRV. DNA encoding the GPI signal sequence from decay-accelerating fac- which does not bind human CD2 on Jurkat cells, was included in the bilayers to monitor tor in pBlueSKÐ was digested with HindIII and EcoRI51. The CD80 PCR product was then molecular exclusion. The dimensions of CD48 are similar to those of CD80, so the CD48 ligated to the GPI signal sequence in pBlueSKÐ. should have been excluded to a similar degree as free CD80. Adhesion assay medium was © http://immunol.nature.com Group 2001 Nature Publishing For expression, the DNA encoding GPI-CD80 was digested from pBlueSKÐ with EcoRI pH 7.4 Hepes-buffered saline, which contained 1% human serum albumin. Adhesion assays and XhoI, the ends blunted with the Klenow fragment of DNA polymerases I and the DNA were done at 37 ¡C. Cell interaction with the GPI-CD80 bilayer was imaged 30 min after was then subcloned into the vector pMON 3360b52. The construct was sequenced with injection of cells into the flow chamber. At this time, CD28-CD80 interaction equilibrium Applied Biosystems DNA sequencing system (model 377). BHK cells were transfected and was reached. Images of cells (transmitted light), contact areas (interference reflection selected with 453 U/ml hygromycin B (Calbiochem, San Diego, CA). Twenty clones were microscopy), Cy3-CD80 accumulation and Cy5-CD48 exclusion (epi-fluorescence illumi- selected and CD80 protein expression was analyzed by flow microfluorimetry analyses with nation) were done on an inverted microscope (Yona Microscopes, Silver Spring, MD). Data mAb 1610A1. was obtained at four densities of CD80 in the bilayers. The fluorescence values were con- verted to CD80 site-density by subtracting the fluorescence intensity values from back- Surface plasmon resonance. A BIAcore 3000 instrument (Biacore AB, Uppsala, Sweden) ground images (no fluorescent bilayer, with cells present or absent as appropriate) value, was used to determine the affinity of the interaction between soluble murine CD80 (sCD80) then dividing by the fluorescence-specific activity (fluorescence U/molecule). The specific and immobilized human CD28-Fc (hCD28-Fc). Mouse sCD80 was generated by PCR with activity was determined by measuring fluorescence values from images of bilayers with dif- the 5′ primer TAGTAGAAGCTTTCCCCATCCGCTCAAGCAGGCCACCATGGCTTG- ferent CD80 densities, where bilayers of known area were also characterized by parallel CAATTGTCAGTTGATGC, which was complementary to murine CD80 leader sequence immunoradiometric analysis with iodinated anti-CD80. The density of CD80 interacting in

and inserted the 5′ untranslated sequence of rat CD4, and the 3′ primer CTACTATCTA- the contact area (CD80I) and the free CD80 in the contact area (CD80F) were determined as: GATTAGTGATGGTGATGGTGATGGTCTTCTGGGGGTTTTTCCCAGGT, which trun- 240 cated the translation product at Asp , and added bases encoding six histidine residues fol- CD80F = CD80B × (CD48C/CD48B) (1)

lowed by a stop codon. The protein was expressed with the glutamine synthetaseÐbased CD80I = CD80C Ð CD80F (2) gene expression system and purified with metal chelate chromatography and gel filtration, 53 as described . The extinction coefficient at 280 nm of sCD80 was determined, by amino where CD80C was the density of CD80 in the contact area, CD80B was the density of CD80

acid analysis, to be 1.80 ml/mg. The purity and activity of sCD80 was established by in the surrounding bilayer, CD48C was the specific fluorescence of Cy5-CD48 in the con- 53 immunoprecipitation with CTLA-4ÐFc and antibodies, as described . A chimeric form of tact area and CD48B was the specific fluorescence of Cy5-CD48 in the surrounding bilayer. CD28, which consisted of the extracellular region of CD28 and murine Fc, was produced The number of CD28 molecules on the Jurkat cells was quantified with iodinated CD28.2. essentially as described54. The hCD28-Fc construct was transfected into 293T cells and The Jurkat cell surface area was measured from transmitted light images as 750 µm2 (ref.

expressed transiently over 4 days. The protein was purified on a protein A sepharose col- 18). The 2D Kd was determined with the Golan-Zhu relationship in the linearized form: umn followed by size-exclusion chromatography on a Superdex S200 HR column (Amersham Pharmacia Biotech, Piscataway, NJ) immediately before use on the BIAcore. y = mx + b (3) hCD28-Fc was directly coupled to the CM5 sensor chip with the BIAcore Amine Coupling

kit (Biacore AB), along with a control protein (SLAM) in an adjacent flow cell. Mouse where y = CD80I/CD80F, m = Ð1/Kd, x = CD80F × (AC/AT) and b = CD28T /(ACKd). In addi- µ sCD80 was injected at a flow rate of 20 l/min for 15 s at 37 ¡C. The specific equilibrium- tion, AC was contact area, AT was the total cell surface area and CD28T was the total num- binding response was determined by subtracting the control flow cell signal from the ber of CD28 molecules on the surface. To compare cells with the same average number of hCD28-Fc cell signal. Binding data were analyzed to give affinity constants, as described53. CD28 molecules per cell at all CD28 densities, data was normalized to a single percentage 18 binding value, based on ranking by CD80I × AC, as described . In this form, the negative

Planar bilayers. GPI-CD80 and GPIÐICAM-1 expressed in BHK cells, CD48 in reciprocal of the slope was the Kd and the x intercept was CD28T/AT. The data could also M12.C3F6 cells and GPIÐI-Ek expressed in CHO cells were grown in cell factories (Nunc, be analyzed by nonlinear regression with a nonlinear form of the same model (A. Whitty, Rochester, NY) or spinner flasks. The cell pellets were washed once with Tris saline (25 unpublished data). This method was more accurate because it avoided the biased weight- 56 mM Tris pH 8.0, 150 mM NaCl, 0.025% NaN3) and snap frozen. Briefly, 3 g of cell pellets ing of datapoints that characterized Scatchard-type linearizations . In the nonlinear analy-

were thawed in Tris-saline with 1% Triton X-100, 100 mM phenylmethylsulfonylfluoride, sis, the average number of bound CD80 molecules per contact (that is, CD80I × AC) was

0.5 mM EDTA, 50 millitrypsin inhibitor U/ml aprotonin and 5 mM iodoacetamide (lysis plotted as a function of CD80F × (AC/AT). The resulting fit was strictly hyperbolic, with

half-maximal binding directly giving the 2D Kd and maximal binding defining the average receptor-ligand complex. J. Exp. Med. 190,31–41 (1999). number of CD80 molecules bound per contact at saturation. For the data we describe here, 20. McConnell, H. M.,Watts,T. H.,Weis, R. M. & Brian,A.A. Supported planar membranes in studies of the results of the nonlinear analysis were essentially identical to those obtained from cell-cell recognition in the immune system. Biochim. Biophys.Acta 864,95–106 (1986). 21. Johnson, K. G., Bromley, S. K., Dustin, M. L. & Thomas, M. L.A supramolecular basis for CD45 tyrosine analysis with the linearized equation. phosphatase regulation in sustained T cell activation. Proc. Natl.Acad. Sci. USA 97, 10138–10143 (2000). σ The ratio of the 2D and 3D Kd was the confinement region ( ): 22. Sagerström, C. G., Kerr, E. M.,Allison, J. P.& Davis, M. M.Activation and differentiation requirements of primary T cells in vitro. Proc. Natl.Acad. Sci. USA 90, 8987–8991 (1993).

σ = 2D Kd / 3D Kd (4) 23. Harding, F.A., McArthur, J. G., Gross, J.A., Raulet, D. H. & Allison, J. P.CD28-mediated signalling co-stim- ulates murine T cells and prevents induction of anergy in T-cell clones. Nature 356, 607–609 (1992). The confinement region provided an estimate of third dimension for molecular interactions 24. Doty, R.T. & Clark, E.A. Subcellular localization of CD80 receptors is dependent on an intact cytoplas- 57 mic tail and is required for CD28-dependent T cell costimulation. J. Immunol. 157, 3270–3279 (1996). in the cell-bilayer interface . This distance was related to fluctuations in the intermembrane 25. McHugh, R. S.,Ahmed, S. N.,Wang,Y. C., Sell, K.W. & Selvaraj, P.Construction, purification and func- distance that controlled the encounters between the membrane-associated receptors. tional incorporation on tumor cells of glycolipid-anchored human B7-1 (CD80). Proc. Natl.Acad. Sci. USA 92, 8059–8063 (1995). Fluorescence photobleaching recovery. Peripheral blood T cells or Jurkat cells were 26. Chan, P.Y. et al. The influence of receptor lateral mobility on adhesion strengthening between mem- stained with the FITC-conjugated mAbs CD28.2 or CD2.1 for 30 min at 4 ¡C. In other branes containing LFA-3 and CD2. J. Cell. Biol. 115, 245–255 (1991). experiments, HUT cells were stained with FITC-conjugated anti-CD28 or anti-CD8. The 27. Liu, S. J., Hahn,W. C., Bierer, B. E. & Golan, D. E. Intracellular mediators regulate CD2 lateral diffusion and cytoplasmic calcium ion mobilization upon CD2-mediated T cell activation. Biophys. J. 68, cells were immobilized with coverslips coated with nonactivating antiÐLFA-1 coated at 200 459–470 (1995). 2 molecules/µm (mAbs TS2/4 or H155). For photobleaching, an image of a group of anti- 28. Shahinian,A. et al. Differential T cell costimulatory requirements in CD28-deficient mice. Science body-labeled cells was acquired to determine the prebleach intensity, then a 1.5 µm2 spot on 261, 609–612 (1993). the edge of the cell was exposed to a 100 ms pulses of the laser at 488 nm to irreversibly 29. van der Merwe, P.A., McNamee, P.N., Davies, E.A., Barclay,A. N. & Davis, S. J.Topology of the CD2- photobleach the fluorophores. Within 1 s of bleaching, an image was acquired that could be CD48 cell-adhesion molecule complex: implications for antigen recognition by T cells. Curr. Biol. 5, used to determine the initial post-bleach intensity. A series of images was then acquired that 74–84 (1995). 30. Linsley, P.S., Clark, E.A. & Ledbetter, J.A.T-cell antigen CD28 mediates adhesion with B cells by ended when the fluorescence intensity in the spot reached a steady state (10Ð20 min), at interacting with activation antigen B7/BB-1. Proc. Natl.Acad. Sci. USA 87, 5031–5035 (1990). which point an image was acquired and used determine the final post-bleach intensity. 31. Kaga, S., Ragg, S., Rogers, K.A. & Ochi,A. Stimulation of CD28 with B7–2 promotes focal adhesion-like Fluorescence recovery was monitored by periodic imaging of the cell after bleaching58. The cell contacts where Rho family small G proteins accumulate in T cells. J. Immunol. 160,24–27 (1998). percentage lateral mobility was calculated as [(prebleach intensity Ð final post-bleach inten- 32. Bachmann, M. F. et al. Distinct roles for LFA-1 and CD28 during activation of naïve T cells: adhesion sity)/(prebleach intensity Ð initial post-bleach intensity)]×100. versus costimulation. Immunity 7, 549–557 (1997). 33. Valitutti, S., Dessing, M.,Aktories, K., Gallati, H. & Lanzavecchia,A. Sustained signalling leading to T cell activation results from prolonged T cell receptor occupancy. Role of T cell actin cytoskeleton. J. Exp. Note: Supplementary information can be found on the Nature Immunology website Med. 181, 577–584 (1995). http://immunol.nature.com/supp-info/ 34. Stamper, C. C. et al. Crystal structure of the B7-1/CTLA-4 complex that inhibits human immune responses. Nature 410, 608–611 (2001). Acknowledgements 35. Brown, D.A. & London, E. Structure and function of sphingolipid- and cholesterol-rich membrane rafts. J. Biol. Chem. 275, 17221–17224 (2000). We thank H. Reiser and M. Naquet for cells producing the 1610A1 and H155 antibodies, 36. Xavier, R., Brennan,T., Li, Q., McCormack, C. & Seed, B. Membrane compartmentation is required for respectively; G. Freeman for the mCD80 cDNA;A. Holdorf for Jurkat cells expressing efficient T cell activation. Immunity 8, 723–732 (1998). truncation mutants of murine CD28; R. Burack for sharing his unpublished results on 37. Yashiro-Ohtani,Y. et al. Non-CD28 costimulatory molecules present in T cell rafts induce T cell cos- cooperation of CD2 and CD28; R. Houdei for cell and antibody production; and R. timulation by enhancing the association of TCR with rafts. J. Immunol. 164, 1251–1259 (2000). Barrett for preparation of the manuscript. Supported by funds from the NIH, HHMI, the 38. Bischof,A., Hara,T., Lin, C. H., Beyers,A. D. & Hunig,T.Autonomous induction of proliferation, JNK and NF-κB activation in primary resting T cells by mobilized CD28. Eur. J. Immunol. 30, 876–882 (2000). Whitaker Foundation and Irene Diamond. 39. Kucik, D. F., Dustin, M. L., Miller, J. M. & Brown, E. J.Adhesion-activating phorbol ester increases the mobility of leukocyte integrin LFA-1 in cultured lymphocytes. J. Clin. Invest. 97, 2139–2144 (1996). Received 17 July 2001; accepted 9 October 2001. 40. Stein, P.H., Fraser, J. D. & Weiss,A.The cytoplasmic domain of CD28 is both necessary and sufficient for costimulation of interleukin-2 secretion and association with phosphatidylinositol 3′-kinase. Mol. © http://immunol.nature.com Group 2001 Nature Publishing 1. Babcock, S. K., Gill, R. G., Bellgrau, D. & Lafferty, K. J. Studies on the two-signal model for T cell acti- Cell. Biol. 14, 3392–3402 (1994). vation in vivo. Transplant. Proc. 19, 303–306 (1987). 41. Berg, L. J. et al. Expression of T-cell receptor α-chain genes in transgenic mice. Mol. Cell. Biol. 8, 2. Jenkins, M. K. & Schwartz, R. H.Antigen presentation by chemically modified splenocytes induces 5459–5469 (1988). antigen-specific T cell unresponsiveness in vitro and in vivo. J. Exp. Med. 165, 302–319 (1987). 42. Krensky,A. M. et al. The functional significance, distribution, and structure of LFA-1, LFA-2, and LFA- 3. Thompson, C. B. et al. CD28 activation pathway regulates the production of multiple T-cell-derived 3: cell surface antigens associated with CTL-target interactions. J. Immunol. 131, 611–616 (1983). lymphokines/cytokines. Proc. Natl.Acad. Sci. USA 86, 1333–1337 (1989). 43. Razi-Wolf, Z. et al. Expression and function of the murine B7 antigen, the major costimulatory mole- 4. Shaw,A. S. & Dustin, M. L. Making the T cell receptor go the distance: a topological view of T cell cule expressed by peritoneal exudate cells. Proc. Natl.Acad. Sci. USA 89, 4210–4214 (1995). activation. Immunity 6, 361–369 (1997). 44. Takei, F. Inhibition of mixed lymphocyte response by a rat monoclonal antibody to a novel murine 5. Viola,A. & Lanzavecchia,A.T cell activation determined by T cell receptor number and tunable lymphocyte activation antigen (MALA-2). J. Immunol. 134, 1403–1407 (1985). thresholds. Science 273, 104–106 (1996). 45. Ozato, K., Mayer, N. & Sachs, D. H. Hybridoma cell lines secreting monoclonal antibodies to mouse 6. Tuosto, L. & Acuto, O. CD28 affects the earliest signaling events generated by TCR engagement. Eur. H-2 and Ia antigens. J. Immunol. 124, 533–540 (1980). J. Immunol. 28, 2131–2142 (1998). 46. Olive, D. et al. Anti-CD2 (sheep red blood cell receptor) monoclonal antibodies and T cell activation 7. Pagès, F. et al. Binding of phosphatidyl-inositol-3-OH kinase to CD28 is required for T cell signalling. I. Pairs of anti-T11.1 and T11.2 (CD2 subgroups) are strongly mitogenic for T cells in presence of Nature 369, 327–329 (1994). 12-O-tetradecanoylphorbol 13-acetate. Eur. J. Immunol. 16, 1063–1068 (1986). 8. August,A. et al. CD28 is associated with and induces the immediate tyrosine phosphorylation and 47. Reay, P.A. et al. Determination of the relationship between T cell responsiveness and the number of activation of the Tec family kinase ITK/EMT in the human Jurkat leukemic T-cell line. Proc. Natl.Acad. MHC-peptide complexes using specific monoclonal antibodies. J. Immunol. 164, 5626–5634 (2000). Sci. USA 91, 9347–9351 (1994). 48. Nunes, J. et al. CD28 mAbs with distinct binding properties differ in their ability to induce T cell 9. Holdorf,A. D. et al. Proline residues in CD28 and the Src homology (SH)3 domain of Lck are activation: analysis of early and late activation events. Int. Immunol. 5, 311–315 (1993). required for T cell costimulation. J. Exp. Med. 190, 375–384 (1999). 49. Kato, K. et al. CD48 is a counter-receptor for mouse CD2 and is involved in T cell activation. J. Exp. 10. Kalli, K., Huntoon, C., Bell, M. & McKean, D. J. Mechanism responsible for T-cell antigen receptor- and Med. 176, 1241–1249 (1992). CD28- or interleukin 1 (IL-1) receptor-initiated regulation of IL-2 gene expression by NF-κB. Mol. 50. Freeman, G. J. et al. Structure, expression, and T cell costimulatory activity of the murine homologue Cell. Biol. 18, 3140–3148 (1998). of the human B lymphocyte activation antigen B7. J. Exp. Med. 174, 625–631 (1991). 11. Monks, C. R., Freiberg, B.A., Kupfer, H., Sciaky, N. & Kupfer,A.Three-dimensional segregation of 51. Coyne, K. E., Crisci,A. & Lublin, D. M. Construction of synthetic signals for glycosyl-phosphatidylinos- supramolecular activation clusters in T cells. Nature 395,82–86 (1998). itol anchor attachment.Analysis of amino acid sequence requirements for anchoring. J. Biol. Chem. 12. Grakoui,A. et al. The immunological synapse: a molecular machine controlling T cell activation. 268, 6689–6693 (1993). Science 285, 221–227 (1999). 52. Warren,T. G. et al. High-level expression of biologically active, soluble forms of ICAM-1 in a novel 13. Wülfing, C. & Davis, M. M.A receptor/cytoskeletal movement triggered by costimulation during T mammalian-cell expression system. Protein Express. Purif. 5, 498–508 (1994). cell activation. Science 282, 2266–2269 (1998). 53. van der Merwe, P.A., Bodian, D. L., Daenke, S., Linsley, P.& Davis, S. J. CD80 (B7-1) binds both CD28 14. Viola,A., Schroeder, S., Sakakibara,Y. & Lanzavecchia,A.T lymphocyte costimulation mediated by and CTLA-4 with a low affinity and very fast kinetics. J. Exp. Med. 185, 393–403 (1997). reorganization of membrane microdomains. Science 283, 680–682 (1999). 54. Linsley, P.S. et al. Binding of the B cell activation antigen B7 to CD28 costimulates T cell prolifera- 15. Dustin, M. L. & Shaw,A. S. Costimulation: Building an immunological synapse. Science 283, 649–650 (1999). tion and interleukin 2 mRNA accumulation. J. Exp. Med. 173, 721–730 (1991). 16. Michel, F. & Acuto, O. Induction of T cell adhesion by antigen stimulation and modulation by the 55. Dustin, M. L., Bromley, S. K., Kan, Z., Peterson, D.A. & Unanue, E. R.Antigen receptor engagement coreceptor CD4. Cell. Immunol. 173, 165–175 (1996). delivers a stop signal to migrating T lymphocytes. Proc. Natl.Acad. Sci. USA 94, 3909–3913 (1997). 17. Schwartz, J. C., Zhang, X., Fedorov,A.A., Nathenson, S. G. & Almo, S. C. Structural basis for co-stimu- 56. Munson, P.J. & Rodbard, D. Number of receptor sites from Scatchard and Klotz graphs: a construc- lation by the human CTLA-4/B7-2 complex. Nature 410, 604–608 (2001). tive critique. Science 220, 979–981 (1983). 18. Dustin, M. L. et al. Low affinity interaction of human or rat T cell adhesion molecule CD2 with its 57. Bell, G. I., Dembo, M. & Bongrand, P.Cell adhesion: Competition between nonspecific repulsion and ligand aligns adhering membranes to achieve high physiological affinity. J. Biol. Chem. 272, specific binding. Biophys. J. 45, 1051–1064 (1984). 30889–30898 (1997). 58. Axelrod, D., Koppel, D. E., Schlessinger, J., Elson, E. L. & Webb,W.W. Mobility measurement by analysis 19. Wild, M. K. et al. Dependence of T cell antigen recognition on the dimensions of an accessory of fluorescence photobleaching recovery kinetics. Biophys. J. 16, 1055–1069 (1976).

1166 nature immunology • volume 2 no 12 • december 2001 • http://immunol.nature.com