Immunological synapse arrays: Patterned proteinsurfaces that modulate immunological synapsestructure formation in T cellsJunsang Doh* and Darrell J. Irvine†‡

*Department of Chemical Engineering and †Department of Materials Science and Engineering, and Biological Engineering Division, Massachusetts Instituteof Technology, 77 Massachusetts Avenue, Cambridge, MA 02139

Edited by Dan R. Littman, New York University Medical Center, New York, NY, and approved February 28, 2006 (received for review October 28, 2005)

T cells are activated by recognition of foreign peptides displayedon the surface of antigen presenting cells (APCs), an event thattriggers assembly of a complex microscale structure at the Tcell–APC interface known as the immunological synapse (IS). Itremains unresolved whether the unique physical structure of thesynapse itself impacts the functional response of T cells, indepen-dent of the quantity and quality of ligands encountered by the Tcell. As a first step toward addressing this question, we createdmulticomponent protein surfaces presenting lithographically de-fined patterns of tethered T cell receptor (TCR) ligands (anti-CD3‘‘activation sites’’) surrounded by a field of tethered intercellularadhesion molecule-1 (ICAM-1), as a model substrate on which Tcells could be seeded to mimic T cell–APC interactions. CD4� T cellsseeded on these surfaces polarized and migrated; on contact withactivation sites, T cells assembled an IS with a structure modulatedby the physical pattern of ligand encountered. On surfaces pat-terned with focal spots of TCR ligand, T cells stably interacted withactivation sites, proliferated, and secreted cytokines. In contrast, Tcells interacting with activation sites patterned to preclude cen-tralized clustering of TCR ligand failed to form stable contacts withactivation sites, exhibited aberrant PKC-� clustering in a fraction ofcells, and had significantly reduced production of IFN-�. Theseresults suggest that focal clustering of TCR ligand characteristic ofthe ‘‘mature’’ IS may be required under some conditions for full Tcell activation.

protein patterned surface � T cell activation

M icroscale patterns of proteins immobilized on surfaces canbe used to dissect the role of spatial organization in the

signals transferred to cells from the extracellular matrix or othercells (1, 2). By presenting ligands in a spatially defined mannerfrom a synthetic substrate, cell functions such as life or death (3),adhesion and migration (4, 5), receptor clustering and membranecompartmentalization (6), and differentiation (7) can be con-trolled, and the role of physical patterns of cell- or extracellularmatrix-derived signals on cell responses can be elucidated.Studies of this type have to date primarily focused on cellresponses to a single signaling or adhesion protein patterned intodefined regions, surrounded by a ‘‘background’’ that lacks pro-tein (3–7). However, surfaces comprising multiple signalingproteins patterned into distinct regions on cellular and subcel-lular length scales would be useful for the study of the complex,spatially organized receptor–ligand interactions that occur inmany cell–cell and cell–extracellular matrix contacts (8–12).

The interactions between T cells and antigen-presentingcells (APCs) during T cell activation provide a stunningexample of such complexity. T cells are activated when theirT cell receptors (TCRs) recognize and engage antigenicpeptides displayed on the surface of APCs. This initial eventtriggers receptor pairs in the membranes of the two cells toassemble an organized structure at the cell–cell contact sitetermed an immunological synapse (IS) (8, 12). The ‘‘mature’’IS formed by T cells encountering high densities of agonist

peptides is composed of a central cluster of TCRs engagingforeign peptide–MHC molecules on the APC surface, sur-rounded by a concentric ring of T cell integrins, particularlylymphocyte function-associated antigen-1 (LFA-1), bindingAPC intercellular adhesion molecule-1 (ICAM-1) adhesionreceptors in the periphery of the contact region (8, 12).Intriguingly, this mature IS structure is only one of severalsupramolecular organizations observed in T cell–APC com-munication: inverse patterns of receptor clustering (i.e., sig-naling receptors clustered peripherally around a central accu-mulation of adhesion receptors) have been observed in the firstfew minutes of T cell activation before formation of a matureIS (8, 13); T cells encountering low (but fully activating)densities of foreign peptides show diffuse receptor clusteringin the interface (14–16); immature T cells exhibit multifocalclusters of receptors during interactions with APCs duringpositive selection (17, 18); and naive T cell–dendritic cellconjugates have been reported to form only nanoscale clustersof receptors in their synapse (19).

Motivated by this phenomenological diversity in synapsestructures, there is great interest in understanding how differentpatterns of receptor and intracellular signaling molecule clus-tering at immune cell–cell interfaces may impact lymphocytefunctional responses (20, 21). For example, Mossman et al. (22),using patterned lipid bilayers where T cell ligands were confinedto �m-scale ‘‘corrals,’’ showed that initial TCR signaling couldbe altered by preventing central clustering of TCR (22). As analternative strategy, we fabricated engineered surfaces designedto ‘‘replace’’ the APC and present multiple protein ligands to Tcells in fixed physical patterns, mimicking (or not) the organi-zation observed in native synapses, as a tool to dissect the roleof this structure in directing T cell functions. We developed anapproach to pattern arrays of ‘‘activation sites’’ containingimmobilized TCR ligands in defined geometries, surrounded byimmobilized adhesion proteins in an ‘‘adhesion field’’ (Fig. 1).We hypothesized that T cells seeded onto such a surface wouldinitially polarize and migrate on the adhesion field, mimickingtheir migration within lymph nodes in search of antigen (23, 24).T cells encountering an activation site would be presented witha defined physical distribution of ligand; for the mature ISprotein pattern shown in Fig. 1, the responding cell would ‘‘see’’a concentric distribution of TCR ligands and adhesion ligandsmimetic of the receptor organization on the APC surface duringnative T cell triggering. We found that such ‘‘immunological

Conflict of interest statement: No conflicts declared.

This paper was submitted directly (Track II) to the PNAS office.

Abbreviations: APC, antigen-presenting cell; IS, immunological synapse; ICAM-1, intercel-lular adhesion molecule-1; LFA-1, lymphocyte function-associated antigen-1; PNMP,poly(o-nitrobenzyl methacrylate-r-methyl methacrylate-r-poly(ethylene glycol) methacry-late); SAv, streptavidin; TCR, T cell receptor.

‡To whom correspondence should be addressed at: Massachusetts Institute of Technology,Room 8-425, 77 Massachusetts Avenue, Cambridge, MA 02139. E-mail: djirvine@mit.edu.

© 2006 by The National Academy of Sciences of the USA

5700–5705 � PNAS � April 11, 2006 � vol. 103 � no. 15 www.pnas.org�cgi�doi�10.1073�pnas.0509404103

Dow

nloa

ded

by g

uest

on

May

12,

202

0

synapse arrays’’ supported robust T cell migration. On contactwith activation sites, T cells responded in a manner characteristicof live APC–T cell interactions: T cells stopped migration,elevated intracellular calcium, and accumulated receptors andsignaling molecules at the interface in a manner dependent onthe pattern of ligands presented from activation sites. Synapsearrays mimicking the structure of the mature IS triggered full Tcell activation, including IL-2 and IFN-� secretion. Interestingly,we found that alteration of the structure of activation sites notonly altered the physical pattern of receptor and intracellularsignaling molecule assembly in T cells, but also influencedendpoint functional responses of the responding cells.

Results and DiscussionFabrication of Immunological Synapse Arrays. We first sought tocreate arrays of ‘‘focal’’ T cell activation sites, surrounded bya field of tethered adhesion molecules, as schematically illus-trated in Fig. 1. An antibody against the CD3� chain of theTCR complex (25) was tethered to surfaces within the acti-vation sites, as a commonly used surrogate for the T cellreceptor’s native ligand (peptide–MHC complexes) (26, 27),whereas recombinant ICAM-1 was immobilized in the adhe-sion field. We hypothesized that such surfaces would allow Tcells engaging an activation site to ligate LFA-1 in a peripheraldistribution around the central patch of TCR ligands, mim-icking the microscale organization of the mature immunolog-ical synapse (8, 12). To create these surfaces, we first extendedan approach we previously reported for patterning using abiotinylated photoresist copolymer, poly(o-nitrobenzylmethacrylate-r-methyl methacrylate-r-poly(ethylene glycol)methacrylate) (PNMP) (28), to achieve multicomponent pat-terning of commercially available, multibiotinylated proteins.The assembly procedure and resulting surface structures areschematically illustrated in Fig. 2A. This process is enabled bythe polyelectrolyte structure of UV-irradiated PNMP: on UVexposure, the o-nitrobenzyl groups of the photoresist arecleaved to create carboxylic acids (chemical structure shown inFig. 7, which is published as supporting information on thePNAS web site). When immersed in aqueous buffer solutionsat near-neutral pH, films of the UV-exposed resist immedi-ately dissolve. However, when exposed PNMP films are im-mersed in aqueous solutions at slightly reduced pH (�6.5), afraction of the carboxylate groups of the UV-exposed copol-ymer protonate, hydrogen bonding among the PNMP chainsoccurs, and the film remains intact (28). To create multicom-ponent protein patterns, PNMP thin films cast on cationicaminosilane glass substrates were UV exposed through aphotomask to define the activation sites (Fig. 2 Ai), followed bydissolution of the bulk of the film in the exposed regions bywashing with PBS (pH 7.4; Fig. 2 Aii). In this step, a thin layerof the exposed polyelectrolyte resist remains bound to thecationic substrate to present biotin for further protein assem-bly (Fig. 7). In Fig. 2 Aiii, the film is exposed to UV withouta mask, priming the background portions of the film (whichwill become the adhesion field) for dissolution. A first ligandis then immobilized over the entire surface (Fig. 2 Aiv) byincubation in a pH 6.0 solution; the exposed photoresist filmmasking the adhesion field regions does not dissolve at this

slightly reduced pH. The first ligand immobilization is followedby a gentle ‘‘erasure’’ of the remaining masking film by washingin PBS at pH 7.4 (Fig. 2 Av). As in Fig. 2 Aii, a thin molecularlayer of biotinylated PNMP is retained on the substrate,providing fresh biotin groups for a final step of backfilling witha second ligand (Fig. 2 Avi). This procedure allowed segregatedpatterning of two ligands onto surfaces, keeping the immobi-lized proteins fully hydrated in near-neutral pH buffers duringthe surface processing. Biotinylated anti-CD3 in the activationsites was tethered via a streptavidin (SAv) bridge, whereasrecombinant ICAM-1�Fc fusion proteins were immobilized bymeans of a biotinylated anti-Fc antibody and SAv (illustratedin Fig. 2 Aiv Inset and Avi Inset).

By blocking excess biotin groups of patterned anti-CD3 withAlexa Fluor 647-conjugated SAv and by detecting immobilizedICAM-1 with an FITC-conjugated anti-ICAM-1 antibody, thefinal spatial distribution of the two ligands on the patternedPNMP surface was visualized. Representative fluorescence im-ages depicting the spatial patterning of anti-CD3 (far redfluorescence) and the adhesion ligand ICAM-1 (green fluores-cence) in a square array of circular activation sites 6 �m indiameter are shown in Fig. 2B. As shown by the single-color andoverlay images, the pattern fidelity and segregation of ligands totheir respective domains by this approach was excellent. Al-though ICAM-1�Fc molecules were immobilized here by meansof noncovalent antibody binding to the Fc portion of the fusionprotein, the half-life for release of immobilized ICAM-1�Fcfrom the surface was �55 h (Fig. 8, which is published assupporting information on the PNAS web site), indicating thatthe ligand density changed only very slowly over the time coursesstudied in the experiments described below.

T Cell Responses to Immunological Synapse Arrays. Motile T cellsthat encounter antigen-loaded APCs receive a ‘‘stop signal’’

Fig. 1. Schematic of immunological synapse array surface pattern.

Fig. 2. Fabrication of immunological synapse arrays. (A) Schematic proce-dure of surface patterning. A PNMP thin film is irradiated through a photo-mask (Ai), developed with PBS (Aii), and then re-exposed to UV in the absenceof a mask (Aiii). A first ligand is deposited (Aiv), followed by washing with pH7.4 buffer (Av), and finally the second ligand is immobilized by using the newlyexposed biotin groups (Avi). (Aiv and Avi Insets) The structure of proteinligand linkages to the surface. The chemical structure of the PNMP photoresistand film after UV exposure and development (Aii) is further illustrated in Fig.7. (B) Fluorescence micrographs taken from one field of a two-componentpatterned surface. SAv-Texas red detecting biotinylated �-CD3 (Left), �-ICAM-1-FITC detecting ICAM-1 (Center), and their overlay (Right).

Doh and Irvine PNAS � April 11, 2006 � vol. 103 � no. 15 � 5701

APP

LIED

BIO

LOG

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

May

12,

202

0

provided by TCR triggering (29), which is closely followed by anincrease in intracellular calcium levels driven by TCR signaling(30–32). We confirmed that synthetic synapse arrays elicited asimilar sequence of early T cell responses by fluorescencevideomicroscopy analysis of murine primary CD4� T cell blastsinteracting with surfaces patterned as shown in Fig. 2B. Impor-tantly, the discrete presentation of TCR ligand in defined areasallowed T cells landing on the adhesion field to adopt a polar-ized, motile pre-antigen-contact state before TCR triggering onactivation sites (Fig. 9, which is published as supporting infor-mation on the PNAS web site). Representative time-lapseimages of single T cell responses to synapse surfaces presentinganti-CD3 or an isotype control IgG from the activation sites areshown in Fig. 3 A and B (and, respectively, in Movies 1 and 2,which are published as supporting information on the PNAS website). T cells migrating on control surfaces passed throughactivation sites presenting isotype control antibodies withoutstopping or altering intracellular calcium levels (Fig. 3A andMovie 1). In contrast, migrating T cells that encountered anactivation site presenting anti-CD3 stopped migration, changedfrom a polarized to rounded morphology, and elevated intra-cellular calcium (Fig. 3B and Movie 2). Quantitation of theresponse of T cells to contact with activation sites by calculationof the cells’ instantaneous velocity and average fura ratio vs. timeshowed the temporal proximity of the halt in migration andcalcium elevation that occurs coincident with activation sitecontact (illustrated in Fig. 10, which is published as supportinginformation on the PNAS web site, for two representative singlecells). Although the diameter of the anti-CD3 activation sitesexamined here (and in the majority of the studies describedbelow) was significantly larger than the typical dimensions ofcentral supramolecular activation clusters (cSMACs) formed byT cells (8, 12), anti-CD3 activation sites with smaller diameters(4 �m, 3 �m, or 2 �m) triggered qualitatively similar migration�halt and calcium responses (data not shown). Thus, the presen-tation of segregated patterned protein signals from these sub-strates elicited dynamic T cell migration and calcium signalingbehavior mirroring responses observed with live T cell–APCinteractions (30).

In addition to controlling T cell activation state, the segre-gated presentation of ‘‘stop’’ (TCR ligand) and ‘‘go’’ (ICAM-1)signals from patterned surfaces led to self-assembly of T cells onthe array sites. T cells migrated randomly on synapse surfacesuntil activation sites were encountered; once triggered by anactivation site, responding cells generally centered themselvesover the activation site and maintained a long-lasting (�7–17 h)contact. Within 30 min at optimal T cell seeding densities (�1.5cells seeded on the surface per activation site), the majority ofthe activation sites were occupied by single T cells (Fig. 11, which

is published as supporting information on the PNAS web site).Such cellular self-organization dictated by surface patterns couldbe used to prevent or promote cell–cell contacts as well ascell–substrate contacts.

Synapse Arrays Drive Full T Cell Activation. We next asked whetherpatterned anti-CD3�ICAM-1 surfaces elicited full activation ofT cells, including proliferation and cytokine production. Becausethe photolithographic process used allows entire large-areasubstrates to be accurately patterned, single-cell as well aspopulation assays are possible with these array surfaces. By usingvideomicroscopy, T cell division triggered by activation sites wasdirectly observed, as illustrated by the time-lapse image se-quence of a T cell dividing on one of the array activation sites�20 h postseeding (Fig. 4A and Movie 3, which is published assupporting information on the PNAS web site). Interestingly, weobserved that daughter cells formed after cell division wereignorant of activation sites and rapidly migrated through TCRligand-presenting regions with only transient pauses (�10 min)or without stopping at all for up to 3 h after cell division(unpublished data). Unlike uniformly coated anti-CD3 sub-strates commonly used to elicit polyclonal T cell activation, hereT cells can migrate away from the activation sites, as they canwhen interacting with discrete live APCs in vitro or in vivo.

At the population level, we measured the secretion of the Tcell growth factor IL-2 and the effector cytokine IFN-� by T cellsinteracting with synapse arrays presenting patterned anti-CD3�ICAM-1 or control ligands (anti-CD3�SAv or isotype IgG�ICAM-1) in the activation site�adhesion field regions (Fig. 4 Band C). Higher levels of IL-2 and IFN-� production were seenon the anti-CD3�ICAM-1 surfaces than on those of anti-CD3�SAv, possibly due to rapid T cell motility and subsequentencounter of T cells with activation sites and�or costimulationdelivered from LFA-1 engagement of ICAM-1 (33, 34). Negli-gible amounts of cytokine were produced by T cells cultured onisotype IgG�ICAM-1-patterned surfaces. Cytokine productionwas also triggered by 3-�m-diameter activation sites, and titra-tion of the density of tethered anti-CD3 within 6-�m-diameteractivation sites revealed a dose-dependent response of IL-2production with anti-CD3 density (Fig. 12, which is published assupporting information on the PNAS web site).

Altogether, these results demonstrate that patterned synapsearrays can elicit full T cell functional responses, and allowsingle-cell dynamics of T cells over extraordinary time intervalsand at times as late as the onset of cell division to be tracked.

Fig. 3. Cell morphology and calcium signaling (tracked by fura fluorescenceratio) as single OT-II CD4� T cells contact a control (A) or anti-CD3-presenting(B) activation site of a synapse array.

Fig. 4. Synapse arrays induce proliferation and cytokine production inprimed CD4� T cell blasts. (A) Time-lapse image sequence depicting a 5C.C7CD4� T cell undergoing cell division on a synapse array surface. Arrows trackthe original cell and resulting two daughter cells. (B and C) IL-2 (B) and IFN-�(C) production by OT-II CD4� T cell blasts incubated on surfaces for 48 h (shownare average � SEM for one representative experiment of three).

5702 � www.pnas.org�cgi�doi�10.1073�pnas.0509404103 Doh and Irvine

Dow

nloa

ded

by g

uest

on

May

12,

202

0

Modulation of Cell Surface and Intracellular Protein Clustering bySynapse Array Microstructure. In live APC–T cell interactions,TCR triggering is followed by the clustering of receptor-ligandpairs in the T cell–APC interface, forming an immunologicalsynapse. Depending on the activation conditions, �m-scalephysical patterning of a number of cell surface and intracellularsignaling molecules accompanies this process. To determinewhether a similar IS structure would assemble in response to thepresentation of tethered patterned ligands, we performed im-munostaining on CD4� T cell blasts fixed after 20 min ofinteraction with patterns of focal anti-CD3 spots surrounded byICAM-1 (Fig. 5 A and B). In synapses formed between T cellsand live B cells, the cytoskeletal protein talin enriches in aperipheral ring surrounding a central accumulation of the criticalsignaling molecule PKC-� (12). In a similar manner, we observedfocal clustering of PKC-� centered over sites of patternedanti-CD3, surrounded by a ring-like accumulation of talin (Fig.5A). On the cell surface, LFA-1 and TCR have been shown tocluster in a pattern similar to the concentric arrangement of talinand PKC-� inside the cell (12); we also observed this concentricarrangement of TCR and LFA-1 for T cells interacting withactivation sites of synapse arrays (Fig. 5B). Similar structureswere observed for activation sites displaying 25-fold lowerdensities of anti-CD3 (data not shown), and with smaller-diameter activation sites (as shown in Fig. 13, which is publishedas supporting information on the PNAS web site, for 2-�m focalsites).

Clearly, a significant difference between T cell–live APCinteractions and T cell–synapse array interactions rests with themobility of the ligands presented; ICAM-1 and peptide–MHCsdiffuse laterally in the membrane of the live APC but here wehave immobilized these ligands on the surface by short moleculartethers. However, a powerful feature of this approach is thepotential to use the fixed distribution of ligands to template anarbitrary organization of T cell surface receptors and thus probethe importance of specific physical patterns of receptor cluster-ing on T cell functions, independent of changes in the totalamount or quality of ligands presented to the T cell. We thusexamined T cell responses to two additional types of immuno-logical synapse arrays: ‘‘multifocal’’ patterns of anti-CD3 pat-terned as four 2-�m circles placed at the corners of a square 6

�m on a side, and ‘‘annular’’ patterns of anti-CD3 with an outerdiameter of 8 �m and inner diameter of 4 �m (Fig. 5 C and D).These activation site geometries were chosen to template T cellsurface receptor clustering in patterns mimicking intermediatesynapse structures observed before the formation of a mature IS(8, 13, 35). In each case, the adhesion field was composed ofimmobilized ICAM-1. Immunostaining of T cells 20 min afterseeding onto these altered synapse arrays revealed that theaccumulation of both cell surface receptors and intracellularsignaling molecules was impacted by the surface ligand pattern.As shown in Fig. 5C, PKC-� clustered over each 2-�m patch ofTCR ligand on ‘‘quad’’ patterns, with LFA-1 accumulatingaround each of these patches. Such multifocal PKC-� clusteringwas observed in 53% of cells (n � 30) on quad patternedsurfaces. On activation sites with an annular geometry, PKC-�clustered in coincidence with the annular activation patch in 25%of cells in full contact with the surface pattern (discussed furtherbelow). Strikingly, LFA-1 accumulated outside the anti-CD3ring but was virtually excluded from the region of the interfaceinside the ring of surface anti-CD3, despite the presence ofICAM-1 in this region (Fig. 5D).

Modulation of T Cell Responses by Altering Surface Ligand Patterns.Having observed that surface ligand patterns could modulate Tcell synapse assembly, we further investigated the behavior of Tcells on focal and annular anti-CD3 patterned surfaces. First, werecorded the dynamics of T cell–surface interactions by video-microscopy 30 min after seeding cells on focal and annulararrays. All T cells interacting with focal patterns of anti-CD3centered themselves for full contact with the anti-CD3 circle,exhibited a rounded morphology, and made minimal membraneextensions away from the contact site (Movie 4, which ispublished as supporting information on the PNAS web site). Incontrast, the majority of T cells interacting with annular patternsmade only partial contact with the anti-CD3 ring, exhibited apartially polarized morphology, and made dynamic membraneextensions away from the activation site: many cells continuouslychanged directions while maintaining a partial contact with theanti-CD3 ring (Movie 5, which is published as supportinginformation on the PNAS web site). This unusual motion of Tcells could be correlated with their polarization by performingtubulin immunostaining on fixed T cells. Fig. 6 A and B showsrepresentative micrographs of T cells interacting with focal andring anti-CD3-patterned surfaces, respectively. For both pat-terns, colocalization of the microtubule-organizing center(MTOC) and PKC-� over the anti-CD3 pattern was observed in�90% of cells that exhibited significant clustering of PKC-�(�80% of cells) at the cell–substrate interface (Fig. 6 A and B).Although T cells contacting focal patterns of anti-CD3 alwayscovered the entire area of the activation site, the majority of Tcells (64%) on ring patterns made only partial contact withactivation sites. The morphology of PKC-� accumulated by Tcells contacting different patterns of anti-CD3 is summarized inFig. 6C. Most T cells that had partial contact with anti-CD3 ringsshowed focal clustering of PKC-� colocalized with the MTOC attheir points of contact with the activation site, as shown in Fig.6B Left. For a fraction of cells (25%) fully contacting anti-CD3rings, PKC-� assembled in an annular structure, as shown in Fig.6B Right. Altogether, annular patterns of anti-CD3 seem toperturb the stable polarization of T cells. Helper T cells polarizePKC-� and their MTOC toward APCs presenting agonist ligand(36), and they can also rapidly change the direction of polariza-tion when a nearby stronger stimulus is detected (37). It ispossible that semimotile T cells making only partial contact withannular activation sites may be in the process of continuouslyrepolarizing their TCRs and signaling machinery around theannulus of ligand, seeking a focus of maximal stimulus. Alter-natively, the inability of T cells spread over the annular activation

Fig. 5. Synapse array patterns template T cell surface receptor and intracel-lular signaling molecule accumulation at the cell–surface contact site. Shownin each panel are schematics of the anti-CD3�ICAM-1 substrate pattern andrepresentative immunofluorescence images at the cell–substrate contactplane of OT-II CD4� T cells fixed 20 min after seeding on synapse surfaces. (Aand B) Immunostaining of PKC-� (green) and talin (red) (A) or TCR and LFA-1on focal anti-CD3 patterns (B). (C and D) Immunostaining of PKC-� (green) andLFA-1(red) on multifocal patterns (each anti-CD3 spot 2 �m in diameter) (C) orannular anti-CD3 patterns (D). (Scale bars: 5 �m.)

Doh and Irvine PNAS � April 11, 2006 � vol. 103 � no. 15 � 5703

APP

LIED

BIO

LOG

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

May

12,

202

0

sites to focally cluster ligated TCRs may have frustrated thenormal program of T cell activation.

To determine whether the altered early response of T cells toannular ligand patterns affected the functional outcome of T cellactivation, several measures of the T cell response to annularactivation site patterns were compared with those obtained on6-�m focal patterns. IL-2 production (Fig. 6D) and calciumsignaling (average fura ratio over 20 min, data not shown) werenot significantly affected by changing the physical display ofanti-CD3 from focal sites to ring structures. Likewise, colocal-ization of phosphotyrosine with TCR, as a measure of TCR-associated signaling (22), was similar for T cells responding tofocal or annular activation sites (Fig. 14, which is published assupporting information on the PNAS web site). In contrast,IFN-� secretion by T cells cultured on annular anti-CD3 patternswas greatly reduced compared with focal anti-CD3 patterns (P �0.02; Fig. 6D). Notably, the surface area of focal and annularactivation sites (and thus the total amount of anti-CD3 encoun-tered by the T cells) was comparable (ring patterns tested herehad a 1.3-fold larger activation site area); thus, the reduction inIFN-� production cannot be ascribed to a lower density ofavailable ligand on the annular pattern. The exact mechanism ofthis effect on cytokine production remains to be determined, butthese results demonstrate that, at least for the 5C.C7 transgenicCD4� T cells tested here, encounter with a nonfocal display ofTCR ligand may alter the program of T cell activation. Alter-natively, annular activation sites may have selectively activated a

subpopulation of the primed T cell population; for example, Th1and Th2 cells form different synapse structures and have dif-ferent requirements for activation (38). Although this alternativecannot be formally excluded, we found that, on restimulationwith anti-CD3-coated plates, a significant fraction of primed5C.C7 T cells (as used for all of our studies) produced theTh1-associated cytokine IFN-�, but only �1% produced theTh2-associated cytokine IL-4 (data not shown), in agreementwith earlier studies on 5C.C7 T cells (39).

In this article, we have created surfaces presenting segregatedpatterns of two protein ligands, which mimic the microscaleorganization of ligands on APCs observed during the assemblyof an immunological synapse. The ability to pattern commer-cially available proteins into defined, segregated regions whileretaining activity makes the patterning strategy described hereimmediately applicable to a broad range of readily availableprotein ligands of interest to many problems in cell biology.Using these patterned surfaces, we have shown that T cellactivation events, and in particular the molecular assembly of Tcell synapses, can be modulated simply by changing the mi-croscale organization of stimuli. This model system allows T cellresponses to be monitored from initial signaling events occurringwithin seconds to proliferation and cytokine production occur-ring 20–30 h after the onset of activation, at the population orsingle-cell levels.

Materials and MethodsMaterials Used in Surface Fabrication. A photoresist copolymer,PNMP, was synthesized, biotinylated, and characterized as de-scribed (28). This resist material is a random terpolymer withcomposition o-nitrobenzyl methacrylate (o-NBMA):methylmethacrylate (MMA):poly(ethylene glycol) methacrylate(PEGMA) � 36:37:27 by weight, number average molecularweight 6,500 g�mol, and polydispersity index of 1.78. Detailedinformation on proteins used for surface patterning and char-acterization is provided in Supporting Text, which is published assupporting information on the PNAS web site.

Fabrication of Immunological Synapse Arrays. Glass coverslips (24 �50 mm; VWR Scientific) were cleaned by 10 M NaOH (10 minwith sonication) and silanized by 3-aminopropyl triethoxysilane(APTS) following a published procedure (40) to create a posi-tively charged surface layer. Biotinylated PNMP was dissolved in1,4-dioxane (3 wt�vol %) and spincoated on APTS-modifiedcoverslips to obtain �130 nm-thick-films. Immunological syn-apse arrays were prepared on PNMP-coated coverslips by usinga photolithography-based technique we previously reported (28)for patterning multiple monobiotinylated proteins. The arraysare based on segregated patterns of immobilized biotinylatedanti-CD3 and ICAM-1�Fc fusion proteins. The patterning ap-proach largely follows our previous report (28), except that,because we sought to use commercially available biotinylatedantibodies that typically have 4–12 biotin groups per molecule(unpublished data), steps to block excess biotin groups on thebiotinylated anti-CD3 molecules patterned in the first step of theimmobilization procedure were required (41). Briefly, PNMPthin films on cationic glass substrates prepared as describedabove were first exposed to UV irradiation through a photomaskfor 20 min (Fig. 2 Ai) and developed by rinsing with pH 7.4 PBSto define the regions that would become activation sites (Fig.2Aii). Next, the substrate was reexposed to UV without aphotomask, priming the background regions of the resist film fordissolution (Fig. 2 Aiii). The removable plastic sidewalls of acommercial eight-well chambered coverslip (well area, 0.8 cm2;Lab-Tek Permanox slides; Nunc) were then attached to thepartially processed patterned coverslip by using Superflex ClearRTV Silicone (Henkel Loctite, Rocky Hill, CT) and cured for24 h at 20°C, to create culture wells with the patterned glass

Fig. 6. T cell responses to focal vs. annular anti-CD3 activation sites. (A andB) Representative micrographs of 5C.C7 CD4� T cells interacting with focaland annular activation sites, respectively. Overlays show DIC�activation sitefluorescence (Upper Left), PKC-��activation site fluorescence (UpperRight), PKC-��tubulin (Lower Left), and tubulin�activation site fluores-cence (Lower Right). (Scale bars: 5 �m.) (C) Quantification of PKC-� clus-tering morphologies. (D) IL-2 and IFN-� secreted by 5C.C7 T cells cultured onfocal vs. annular activation site patterns after 48 h, normalized by cytokineconcentrations secreted on focal patterns (average � SEM from threeindependent experiments).

5704 � www.pnas.org�cgi�doi�10.1073�pnas.0509404103 Doh and Irvine

Dow

nloa

ded

by g

uest

on

May

12,

202

0

substrate serving as a base. Before ‘‘erasing’’ the backgroundfilm region, SAv (10 �g�ml) and biotinylated anti-CD3 (orbiotinylated isotype control antibody, 5 �g�ml) were sequentiallyincubated over the surface for 30 min each in PBS (pH 6.0) at4°C, binding to the available surface-tethered biotin groups (Fig.2Aiv). (As an estimate of the binding capacity of biotinylatedPNMP films, the density of SAv coupled to PNMP surfacesmeasured by using 125I-labeled SAv was �4,000 molecules per�m2; however, this value likely overestimates the density ofactive anti-CD3 and ICAM-1 sites on the surfaces.) Excess freebiotin groups remaining on the tethered anti-CD3 were blockedby sequential incubation with SAv (10 �g�ml, 30 min) andbiotin-PEO-amine (10 �g�ml, 30 min, at 4°C; Pierce EZ-link).The exposed PNMP film masking the background (and proteinbound to it) was then dissolved by immersing the protein-conjugated surface in pH 7.4 PBS (Fig. 2 Av). Although the bulkof the masking PNMP film dissolved, the cationic substrateelectrostatically retained a thin layer of biotinylated PNMP onthe surface. By using these freshly exposed biotin groups, theadhesion field was functionalized by sequential assembly of SAv(10 �g�ml, 30 min), biotinylated anti-human Fc (10 �g�ml, 30min), and ICAM-1�Fc (5 �g�ml, 30 min) at 4°C (Fig. 2 Avi).

Preparation of Cells. OT-II (The Jackson Laboratory) and 5C.C7(Taconic Farms) CD4� T cell blasts were prepared by stimula-tion of splenocytes from transgenic mice with 100 �g�ml ovalbu-min (for OT-II) or 1 �M moth cytochrome c peptide (aminoacids 88–103; for 5C.C7). Cells were maintained in complete

RPMI medium (RPMI medium 1640�10% FCS�2 mM L-glutamine�50 �M 2-mercaptoethanol�penicillin�streptomycin)and used on days 5–7.

Time-Lapse Microscopy. Time-lapse fluorescence microscopy wasperformed on a Zeiss Axiovert 200 epifluorescence microscopeequipped with a heated stage (maintaining 37°C and 5% CO2).T cells were loaded with the intracellular Ca2� indicator fura-2AM (Molecular Probes) to permit simultaneous morphologyand intracellular Ca2� imaging (30). For each experiment,fura-loaded T cell blasts were seeded onto patterned synapsearrays, and time-lapse microscopy was immediately initiated (seeSupporting Text).

Immunostaining. T cells were seeded on immunological synapsesurfaces and incubated for 20 min at 37°C and 5% CO2. Cellswere then fixed, permeabilized, and stained as described else-where (42) (see Supporting Text).

ELISA. The functional outcome of T cell interactions with pat-terned surfaces was assayed by measuring IL-2 and IFN-�secretion. T cells (105) were seeded on patterned substrates in300 �l of RPMI media and incubated at 37°C, 5% CO2 for 48 h;100 �l of supernatant was then collected from each sample forELISA analysis of cytokine concentrations.

This work was supported by the DuPont-MIT Alliance and the Arnoldand Mabel Beckman Foundation.

1. Folch, A. & Toner, M. (2000) Annu. Rev. Biomed. Eng. 2, 227–256.2. Whitesides, G. M., Ostuni, E., Takayama, S., Jiang, X. Y. & Ingber, D. E. (2001)

Annu. Rev. Biomed. Eng. 3, 335–373.3. Chen, C. S., Mrksich, M., Huang, S., Whitesides, G. M. & Ingber, D. E. (1997)

Science 276, 1425–1428.4. Koo, L. Y., Irvine, D. J., Mayes, A. M., Lauffenburger, D. A. & Griffith, L. G.

(2002) J. Cell Sci. 115, 1423–1433.5. Jiang, X. Y., Bruzewicz, D. A., Wong, A. P., Piel, M. & Whitesides, G. M. (2005)

Proc. Natl. Acad. Sci. USA 102, 975–978.6. Wu, M., Holowka, D., Craighead, H. G. & Baird, B. (2004) Proc. Natl. Acad.

Sci. USA 101, 13798–13803.7. McBeath, R., Pirone, D. M., Nelson, C. M., Bhadriraju, K. & Chen, C. S. (2004)

Dev. Cell 6, 483–495.8. Grakoui, A., Bromley, S. K., Sumen, C., Davis, M. M., Shaw, A. S., Allen, P. M.

& Dustin, M. L. (1999) Science 285, 221–227.9. Dustin, M. L. & Colman, D. R. (2002) Science 298, 785–789.

10. Giancotti, F. G. & Ruoslahti, E. (1999) Science 285, 1028–1032.11. Davis, D. M., Chiu, I., Fassett, M., Cohen, G. B., Mandelboim, O. &

Strominger, J. L. (1999) Proc. Natl. Acad. Sci. USA 96, 15062–15067.12. Monks, C. R. F., Freiberg, B. A., Kupfer, H., Sciaky, N. & Kupfer, A. (1998)

Nature 395, 82–86.13. Lee, K. H., Holdorf, A. D., Dustin, M. L., Chan, A. C., Allen, P. M. & Shaw,

A. S. (2002) Science 295, 1539–1542.14. Purtic, B., Pitcher, L. A., van Oers, N. S. C. & Wulfing, C. (2005) Proc. Natl.

Acad. Sci. USA 102, 2904–2909.15. Purbhoo, M. A., Irvine, D. J., Huppa, J. B. & Davis, M. M. (2004) Nat.

Immunol. 5, 524–530.16. Faroudi, M., Utzny, C., Salio, M., Cerundolo, V., Guiraud, M., Muller, S. &

Valitutti, S. (2003) Proc. Natl. Acad. Sci. USA 100, 14145–14150.17. Richie, L. I., Ebert, P. J. R., Wu, L. C., Krummel, M. F., Owen, J. J. T. & Davis,

M. M. (2002) Immunity 16, 595–606.18. Hallman, E., Burack, W. R., Shaw, A. S., Dustin, M. L. & Allen, P. M. (2002)

Immunity 16, 839–848.19. Brossard, C., Feuillet, V., Schmitt, A., Randriamampita, C., Romao, M.,

Raposo, G. & Trautmann, A. (2005) Eur. J. Immunol. 35, 1741–1753.20. Friedl, P., den Boer, A. T. & Gunzer, M. (2005) Nat. Rev. Immunol. 5, 532–545.21. Trautmann, A. & Valitutti, S. (2003) Curr. Opin. Immunol. 15, 249–254.22. Mossman, K. D., Campi, G., Groves, J. T. & Dustin, M. L. (2005) Science 310,

1191–1193.

23. Miller, M. J., Wei, S. H., Parker, I. & Cahalan, M. D. (2002) Science 296,1869–1873.

24. Mempel, T. R., Henrickson, S. E. & von Andrian, U. H. (2004) Nature 427,154–159.

25. Leo, O., Foo, M., Sachs, D. H., Samelson, L. E. & Bluestone, J. A. (1987) Proc.Natl. Acad. Sci. USA 84, 1374–1378.

26. Bruniquel, D. & Schwartz, R. H. (2003) Nat. Immunol. 4, 235–240.27. Lee, K. M., Chuang, E., Griffin, M., Khattri, R., Hong, D. K., Zhang, W. G.,

Straus, D., Samelson, L. E., Thompson, C. B. & Bluestone, J. A. (1998) Science282, 2263–2266.

28. Doh, J. & Irvine, D. J. (2004) J. Am. Chem. Soc. 126, 9170–9171.29. Dustin, M. L., Bromley, S. K., Kan, Z. Y., Peterson, D. A. & Unanue, E. R.

(1997) Proc. Natl. Acad. Sci. USA 94, 3909–3913.30. Wulfing, C., Sjaastad, M. D. & Davis, M. M. (1998) Proc. Natl. Acad. Sci. USA

95, 6302–6307.31. Lewis, R. S. & Cahalan, M. D. (1995) Annu. Rev. Immunol. 13, 623–653.32. Donnadieu, E., Bismuth, C. & Trautmann, A. (1994) Curr. Biol. 4, 584–595.33. Vanseventer, G. A., Newman, W., Shimizu, Y., Nutman, T. B., Tanaka, Y.,

Horgan, K. J., Gopal, T. V., Ennis, E., Osullivan, D., Grey, H. & Shaw, S. (1991)J. Exp. Med. 174, 901–913.

34. Kuhlman, P., Moy, V. T., Lollo, B. A. & Brian, A. A. (1991) J. Immunol. 146,1773–1782.

35. Krummel, M. F., Sjaastad, M. D., Wulfing, C. & Davis, M. M. (2000) Science289, 1349–1352.

36. Monks, C. R. F., Kupfer, H., Tamir, I., Barlow, A. & Kupfer, A. (1997) Nature385, 83–86.

37. Depoil, D., Zaru, R., Guiraud, M., Chauveau, A., Harriague, J., Bismuth, G.,Utzny, C., Muller, S. & Valitutti, S. (2005) Immunity 22, 185–194.

38. Balamuth, F., Leitenberg, D., Unternaehrer, J., Mellman, I. & Bottomly, K.(2001) Immunity 15, 729–738.

39. Seder, R. A., Paul, W. E., Davis, M. M. & Destgroth, B. F. (1992) J. Exp. Med.176, 1091–1098.

40. Jonas, U., del Campo, A., Kruger, C., Glasser, G. & Boos, D. (2002) Proc. Natl.Acad. Sci. USA 99, 5034–5039.

41. Sandberg, S. A., Barrett, R. W., Pirrung, M., Lu, A. L., Kiangsoontra, B. &Holmes, C. P. (1995) J. Am. Chem. Soc. 117, 12050–12057.

42. O’Keefe, J. P., Blaine, K., Alegre, M. L. & Gajewski, T. F. (2004) Proc. Natl.Acad. Sci. USA 101, 9351–9356.

Doh and Irvine PNAS � April 11, 2006 � vol. 103 � no. 15 � 5705

APP

LIED

BIO

LOG

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

May

12,

202

0