Nature AOP, published online 22 September 2004; doi:10.1038/nature02916
 
 
A role for the immunological synapse in lineage commitment of CD4 lymphocytes

Activation of the naive T-helper lymphocyte (Thp) directs it down one of two major developmental pathways called Th1 and Th2. Signals transmitted by T cell, co-stimulatory and cytokine receptors control Thp lineage commitment but the mechanism by which these signals are integrated remains a mystery. The interferon- (IFNGR) and interleukin 4 (IL-4R) cytokine receptors, in particular, direct the earliest stages of T-helper commitment. Here we report that on engagement of the T-cell receptor (TCR) on Thp cells, rapid co-polarization of IFNGR with the TCR occurs within the developing immunological synapse. Thp cells from the intrinsically Th1-like C57BL/6 mouse strain have significantly more receptor co-polarization than Th2-prone BALB/c Thp cells. Remarkably, in the presence of IL-4, a cytokine required for Th2 differentiation, IFNGR co-polarization with TCR is prevented. This inhibition depends on Stat6, the transcription factor downstream of IL-4R that is required for Th2 differentiation. This cytokine receptor crossregulation provides an explanation for the effect of IL-4 in inhibiting Th1 differentiation. These observations suggest a scenario in which physical co-polarization of critical receptors directs the fate of the naive Thp, and offer a novel function for the immunological synapse in directing cell differentiation. They further suggest a new mechanism of membrane-bound signalling control by the physical disruption of large receptor-rich domains on signalling through a functionally antagonistic receptor.

The immunological synapse can be described as a superstructure initiated at the contact surface between the T cell and the antigen presenting cell (APC) that optimizes activation by congregating signalling molecules in specialized regions of the membrane of both cell types1-4. Differences in calcium transport, assembly of adaptor protein/signalling molecules and receptor distribution into lipid raft structures have been described in mature Th1 and Th2 cells5, 6 raising the possibility that fundamental differences in immunological synapse formation on the Thp might influence T helper lineage commitment by mechanisms as yet unknown.

TCR and IFNGR co-polarize at the immunological synapse on TCR engagement
Most evidence indicates that T helper differentiation in vivo is dictated by cytokines in concert with TCR signalling7-9. Autocrine or paracrine production of IFN and IL-4 are required for Th1 and Th2 commitment, respectively, in part through the induction by these two key cytokines of the T-bet (Th1) and GATA-3 (Th2) subset specific transcription factors. Mice lacking components of the IFN or IL-4 signalling pathways (including T-bet and GATA-3), lack Th1 or Th2 cells7, 9. We analysed the relative membrane topology of TCR and cytokine receptors, focusing on these two cytokine receptors (IFNGR and IL-4R) on the naive Thp, that are critical for the earliest stages of T-helper commitment. Many previous studies have been performed using polyclonally activated T cells and transformed B cells as APC. To more closely mirror physiological conditions as observed in vivo10, we studied the formation and evolution of the immunological synapse using Thp and mature splenic dendritic cells (DC). Two methodologies were pursued with similar outcomes. First, the cell suspension was sedimented by centrifugation to optimize interaction between the two cell types (almost 100% of T cells form conjugates with dendritic cells), the reaction stopped by fixation with paraformaldehyde (PFA), Thp cell–DC couplets permeabilized, labelled and observed by confocal microscopy. Analysis of receptor distribution on T cells demonstrated that at 0 min the TCR, IFNGR1 (IFNGR subunit and IFNGR2, the  subunit) and IL-4R (data not shown) were randomly distributed on the cell surface with little or no receptor accumulation at the Thp cell–DC interface (Fig. 1a; and Supplementary Fig. 1). In contrast, if cells were incubated 30 min (and 10 min, not shown) before fixation, 60 to 80% of the 50 cells analysed per experiment (n = 4) displayed increased polarization of TCR in the presence of ovalbumin (OVA)-loaded dendritic cells (Fig. 1a) but not with unpulsed dendritic cells (data not shown). Strikingly, IFNGR1 (and IFNGR2, Supplementary Fig. 1) was recruited to the Thp cell–DC interface and colocalized with TCR. As reported, activation of the Thp leads to patching and internalization of the TCR. However, this is a continuous and not completely synchronized process. Given that polarization and subsequent internalization occur in the same geographical location, we chose 30 min, a time when the majority of Thp has achieved polarization (and some receptor internalization) to perform imaging studies. Further studies will be necessary to determine whether TCR and IFNGR receptors are internalized in the same vesicles or segregated after internalization.

 Figure 1 TCR and IFNGR colocalize at the immunological synapse.   Full legend
 
High resolution image and legend (64k)
 

Second, we performed four-dimensional (4D) imaging by mixing prelabelled transgenic DO11.10 Thp cells and OVA-loaded dendritic cells in order to visualize better the dynamics of DC–Thp cell interactions and receptor movement relative to the Ca2+ influx that follows T cell activation (assessed by FURA-2AM radiometric analysis). A representative experiment (out of four) of time-lapse microscopy in Fig. 1b–d (and Supplementary Movies 1 and 2) shows that at 0 s (no cell–cell contact) Ca2+ levels were low and both TCR and IFNGR1 were uniformly distributed in the cell membrane. Immediately after Ca2+ influx, both IFNGR1 and TCR on the naive T cell progressively migrate to the contact interface with the dendritic cell (Fig. 1b, c).

Restricted co-polarization of IFNGR and IL-2R cytokine receptors with the TCR at the immunological synapse
We obtained similar results when Thp activation was induced by crosslinking of TCR at 37 °C with a secondary antibody in the absence of dendritic cells. Figure 2a shows that crosslinking of TCR induced its segregation and co-polarization with the IFNGR1 (and IFNGR2, not shown) but not IL-4R. In order to visualize better the co-polarization of receptors by a more quantitative method, values of their fluorescence intensity were measured (or 'linearized'). The left panels in Fig. 2c show that the distribution of TCR and IFNGR1 molecules at 0 min is random and uniformly represented across the entire cell surface. At 30 min and 10 min (data not shown) after TCR crosslinking both TCR and IFNGR1 molecules were polarized to one region at the periphery of the cell (Fig. 2a) as reflected by the one-peak shape of the almost completely overlapping curves (Fig. 2c, first row, right panel). In this system TCR-induced copatching of cytokine receptors was quite specific to the IFNGR as the IL-4R, IL-6R, IL-7R and IL-10R (Fig. 2b, c) did not colocalize with TCR and their distribution was similar to CD45, a molecule known to be excluded from the immunological synapse3. The single exception was the IL-2R alpha chain, previously noted to be present in membrane rafts11, which displayed a similar (although less pronounced) co-polarization with the TCR after activation. This is interesting because IL-2, the earliest cytokine secreted by the Thp, is an overall growth factor for both T-helper subsets but is especially important for Th2 development12. Because ligand-induced IFNGR polarization to caveolar membrane domains and internalization occurs13, the patching observed could be a consequence of small quantities of IFN secreted by contaminating mature T-helper cells or by early activated Thp. This was not the case, however, as a comparison of TCR crosslinking on WT, IFN-deficient and Stat1-deficient Thp showed an identical pattern of distribution and motility of IFNGR1 and TCR (Fig. 2c, bottom row and Supplementary Fig. 2). Of note, neither the presence of IL-4 nor the absence of IFN (using IFN-deficient Thp as above) induced the co-polarization of TCR and IL-4R (Supplementary Fig. 3). We conclude that activation of the naive Thp results in a rapid and selective co-polarization of TCR and IFNGR and that this co-polarization is IFN-independent.

 Figure 2 IFNGR co-polarization and partial IL-2R co-polarization with the TCR after T cell activation.   Full legend
 
High resolution image and legend (156k)
 

Concordance between the intrinsic Th1-'ness' of a Thp cell and the degree of IFNGR and TCR co-polarization
Genetic background controls the inherent tendency of Thp to differentiate along a Th1 or Th2 pathway, traits that have been mapped to specific chromosomal loci14. No firm explanation, however, has as yet emerged to explain such differences. We explored the possibility that Th1- or Th2-like strains of mice might display differences in their capacity to co-polarize IFNGR and TCR. In Fig. 3a, imaging analysis and linearization of the Thp surface (as in Fig. 2) shows that TCR crosslinking results in the extensive co-recruitment and polarization of IFNGR1 and TCR in Thp from the B6 (Th1-prone) strain. In contrast, in Thp from the BALB/c (Th2-prone) strain, the distribution of IFNGR is less polarized and its superimposition on the TCR only partial. In a more extensive analysis (Fig. 3b), the correlation coefficient ( x,y) was calculated for the normalized data (relative to the highest pixel peak) in order to determine whether the variations in the expression of IFNGR and TCR at different sectors of the Thp membrane are linked. The value of  x,y increased up to fivefold from 0 min to 30 min after Thp stimulation, reflecting the active recruitment of both receptors. The chance of TCR/IFNGR co-polarization occurring at random in the B6 Thp is approximately 1/5000 (average  x,y value of 0.9; P = 0.0002) wheras it is approximately 1/25 (average  x,y value of 0.58; P = 0.04) in the BALB/c Thp, a 200-fold difference. These results indicate a concordance between the tendency of a Thp to differentiate along the Th1 pathway and the degree of IFNGR and TCR co-polarization. This was not directly attributable to IL-4 itself as IL-4-/- BALB/c Thp behaved similarly (Supplementary Fig. 4) suggesting that another intrinsic genetic mechanism was operative in the BALB/c strain. The less robust recruitment of IFNGR to the immunological synapse in the BALB/c Th2-like Thp led us to test directly whether differential recruitment might be important for Th1/Th2 differentiation.

 Figure 3 B6 Thp display increased co-polarization of TCR and IFNGR compared to BALB/c Thp.   Full legend
 
High resolution image and legend (34k)
 

IL-4 impedes the co-polarization of IFNGR with TCR in a Stat6-dependent manner
T helper differentiation is a rather unique system where a relatively neutral signal (TCR engagement) is accompanied by positive polarizing signals (cytokine receptor engagement) that are also mutually inhibitory8. Thus IL-4 inhibits the differentiation of Th1 cells from Thp and tends to be dominant over the Th2-inhibitory properties of the IFN Th1-enhancing cytokine. The mechanism by which such inhibition occurs is unknown. We tested the hypothesis that ligand-induced signalling through the IL-4R might affect the physical distribution of the opposing IFNGR. Thp were cultured under Th2 polarizing conditions and receptor localization was visualized. Remarkably, crosslinking of the TCR in the presence of IL-4 completely inhibited the migration and co-polarization of the IFNGR1 (Fig. 4a, and IFNGR2 not shown) with TCR. Population analysis (10 to 50 cells per experiment, n = 3) as above showed that the  x,y values were significantly different when cultured in the presence of IL-4 (Fig. 4b). This was true when IL-4 was added at times ranging from 0 to 30 min after TCR crosslinking suggesting that IL-4 can actually reverse as well as prevent co-polarization. Other cytokines such as IL-2 and the Th1-antagonistic cytokine IL-10 did not interfere with co-polarization (Fig. 4c). Because IL-4 acts primarily through the Jak/Stat signalling pathway to accomplish Th2 lineage commitment, we tested the ability of IL-4 to inhibit TCR/IFNGR co-polarization in mice lacking the IL-4R signalling molecule, Stat6. Figure 4d shows that the ability of IL-4 to prevent TCR/IFNGR co-polarization is abolished in Stat6-/- Thp cells. Occupancy of IL-4R with subsequent activation of Stat6 therefore translates into events that prevent the IFNGR from migrating into the immunological synapse with TCR. There is precedence for the ability of an inhibitory receptor to prevent the recruitment of an activating receptor into the immunological synapse. Engagement of the natural killer (NK) inhibitory receptors, killer cell immunoglobulin-like receptor (KIR) 2DL1 or CD94/NKG2, rapidly blocks tyrosine phosphorylation of the (CD244) 2B4 activating receptor by preventing its recruitment into lipid rafts15. However, it was not clear from those studies whether NK inhibitory receptors blocked recruitment because synapse formation was deficient, similar to the blockade of phosphorylation and Ca2+ flux by TGF (ref. 16), or whether this was a specific process.

 Figure 4 IL-4 impedes the co-polarization of IFNGR with TCR in a Stat6-dependent manner.   Full legend
 
High resolution image and legend (112k)
 

Discussion
The function of the T cell synapse remains controversial although there is general agreement that sustained signalling, which correlates with immunological synapse formation17, 18, leads to optimal T cell activation characterized by cycles of receptor signalling and degradation19. A recently espoused view that is consistent with most available data envisions the immunological synapse as a physical structure with the required complexity to guide T cell activation under diverse settings over time1. Our data suggest that the IFNGR is a novel member of that architectural platform whose co-polarization with TCR may guide lineage commitment. One function of such polarization might include the concentration of cytokine receptors at the actual sites where cytokines are produced. This hypothesis fits nicely with previous contentions that cytokines act locally rather than at a distance20 and assumes directional cytokine secretion focused towards the presentation pole of the cell, a phenomenon already described for CTLs (cytotoxic T lymphocytes) and NK cells21. On target recognition, these cells polarize the secretion of cytolytic factors with a 'kiss of death'21. Hence, for naive Thp the immunological synapse could provide a platform for 'cytokine presentation' of key activating and polarizing factors such as IL-2, IFN, IL-12, IL-23 or IL-27 from dendritic cells. Such presentation might be reciprocal to include cytokines presented from the T cell to the dendritic cell. Indeed, activated mature TCR transgenic T cells co-polarize TCR and IL-2 secretory vesicles in situ on injection of specific peptide suggesting secretion of IL-2 towards the immunological synapse22. Further, this autoactivating loop might also be initiated by dendritic cells, which secrete IL-2 shortly after maturation, another reason for the T cell to bring IL-2R to that interface23. Thus, polarized IL-2R (as shown above) and IL-2 production will not only maximize the efficiency of IL-2 autoactivation but will also ensure optimum activation of APC presentation capacity.

It is tempting to hypothesize that TCR-induced co-polarization leads to IFNGR signalling to preferentially drive Th1 differentiation. The presence of the IL-2R and IFNGR in the immunological synapse under 'neutral' conditions (the absence of IFN and IL-4) is consistent with previous reports that TCR signalling can induce Stat5 tyrosine24 and Stat1 serine phosphorylation25 and results in transient downregulation of IL-4R signalling26. However, the majority of these experiments involved strong TCR signalling and much evidence emphasizes the importance of cytokine signalling under TCR 'neutral' stimuli8, 9. In the Thp cell, signalling through the IFNGR and the IL-27R and Stat1, rather than the IL-12R (which is only expressed on activated T helper but not naive Thp cells) is probably the primary Th1-polarizing stimulus7. Indeed, we did not observe TCR/IFNGR co-polarization on mature T helper cells (data not shown). Our observations suggest a scenario where IFNGR co-recruitment provides the early link between TCR and cytokine signalling. Strong TCR signalling leads to accentuated IFNGR co-polarization and the assembly of a Th1 signallosome which is further stabilized by the subsequent secretion of IFN (ref. 27) unless an inhibitory signal (IL-4 induced Stat6 activation) is delivered and co-polarization prevented. Parallel engagement of TCR and IL-4R might then lead to assembly of a Th2 signallosome6. Such a scenario fits with the observation that activated Thp produce IL-4 promptly and can become Th2 cells in the absence of IL-12 or IFN (ref. 28). The quantity of IL-4 in the local milieu may ultimately determine Thp cell fate thus implying that the Th1 pathway may be the default response of the Thp. These speculations will need to be tested experimentally but our observations provide intriguing evidence for a new function of the immunological synapse in controlling cell fate decisions. It may be that similar processes occur for other receptors such as chemokine receptors whose residence in membrane microdomains with CD4 influences HIV binding29 and in non-T cells where cytokine antagonism exists. Further, receptor co-polarization in cells of other lineages such as neurons, adipocytes, endocrine tissues and muscle—where lipid rafts have been described to control effector function—may also control cell fate30.

Methods
Cell preparations and Thp activation Naive Thp cells were isolated from the lymph nodes of 5–6-week-old Rag2-/- DO11.10 TCR transgenic mice by MACS (magnetic activated cell sorting) negative selection of CD25highCD11c+CD8+CD11b+CD45RB+DX5+CLII+cells. Dendritic cells were purified by digesting the spleens of BALB/c mice with collagenase/dispase followed by MACS positive selection. CD11c+ sorted cells were cultured for 10 h to induce spontaneous maturation in the presence of 100 µg ml-1 of OVA protein and 1 µM of OVA CLII-restricted peptide. The cell suspension was sedimented by centrifugation to optimize interaction between the two cell types (almost 100% of T cells form conjugates with dendritic cells), the reaction stopped by fixation with paraformaldehyde (PFA), Thp cell–DC couplets permeabilized, labelled and observed by confocal microscopy. In other experiments, crosslinking of the TCR was induced by adding a secondary goat anti-FITC Alexa 488-coupled antibody at 37 °C and the reaction was stopped by fixation of the cells at various times. Fixed cells were permeabilized and stained for different markers. 4D observations were performed by staining the cells with monoclonal antibody followed by culture at 37 °C in 5% CO2 in separate droplets of complete media or three-dimensional (3D) collagen matrices (1 mg ml-1). After 30 min rest (or collagen polymerization), a bridge between the droplets was created allowing Thp cell–DC cell contact. Markers used were as follows: DAPI, a DNA marker; Alexa-594-labelled (A594)- or A488-hamster anti-mouse IFNR (2E2); A488- or A594-hamster anti-mouse TCR (H57); A647-hamster anti-mouse IFNR (MOB-47); A647-rat anti-mouse IL-4R (M1); (phycoerythrin (PE)-labelled) PE-rat anti-mouse IL-2R (PC61); PE-rat anti-mouse IL-6R (D7715A7); PE-rat anti-mouse IL-7R (SB/14); PE-rat anti-mouse IL-10R (1B1.3a); PE-rat anti-mouse CD45RB (16A). Values of Fura-2AM radiometric analysis were obtained as Ratio = Fura emissions at 340 nm/380 nm.

Linearization analysis and calculation of correlation coefficient Linearization analysis allows us to compare the location of the molecules observed in one plane of the cell versus the whole cell. Scans of the cell surface were made by drawing ring-shaped regions at the mid focal plane of every cell that included the membrane and the subjacent cytoplasm. The two arrays of fluorescence intensities per cell (TCR and IFNGR number of pixels) were normalized relative to the highest peak (Y axis) and their correlation coefficient (a statistic that measures the degree to which two variables are related, http://noppa5.pc.helsinki.fi/koe/corr/cor7.html, http://mathworld.wolfram.com/CorrelationCoefficient.html,  x,y) was calculated as x,y=(x - )(y - )/(x - )2(y - )2 This value is a version of the covariance expressed as -1   x,y  + 1 and represents the relationship between two arrays of numbers where in 100 datapoints, a value of the correlation coefficient greater than 0.197 has less than a 5% probability of arising by chance.