Differential Clustering of CD4 and CD3ζ During T Cell Recognition

See allHide authors and affiliations

Science  25 Aug 2000:
Vol. 289, Issue 5483, pp. 1349-1352
DOI: 10.1126/science.289.5483.1349


Whereas T helper cells recognize peptide–major histocompatibility complex (MHC) class II complexes through their T cell receptors (TCRs), CD4 binds to an antigen-independent region of the MHC. Using green fluorescent protein–tagged chimeras and three-dimensional video microscopy, we show that CD4 and TCR-associated CD3ζ cluster in the interface coincident with increases in intracellular calcium. Signaling-, costimulation-, and cytoskeleton-dependent processes then stabilize CD3ζ in a single cluster at the center of the interface, while CD4 moves to the periphery. Thus, the CD4 coreceptor may serve primarily to “boost” recognition of ligand by the TCR and may not be required once activation has been initiated.

In higher vertebrates, T lymphocytes systematically scan cells and tissues for non-self antigens embedded in molecules of the MHC. Recognition of these peptide-MHC complexes is achieved through the αβ TCR, and signal transduction occurs through phosphorylation of the tightly associated CD3 γ, δ, ɛ, and ζ polypeptides (1). Several receptors on T cells can potentiate the response of the T cell, and their engagement is often required for successful activation in vivo. One of these is CD4 on T helper cells, which binds to relatively invariant sites on MHC class II molecules outside the peptide-binding groove (2, 3) and accumulates at the T cell–antigen-presenting cell (APC) interface (4). During activation, CD4 increases T cell sensitivity to antigen by 10- to 100-fold (5–7). A cysteine motif present in the CD4 tail binds src-family kinase p56lck (8), and Lck appears to be responsible for phosphorylating CD3ζ, the earliest known event in T cell signaling (9).

Despite CD4's importance in generating signals, its role in potentiating TCR recognition remains unclear. TCR binding affinity varies considerably (10), and it has been suggested that CD4 binds MHC class II simultaneously with the TCR (11) or perhaps shortly after TCR–peptide-MHC engagement (6,7) and serves to stabilize the complex. Recently it has been shown that some of the signaling molecules involved in T cell recognition do not form a homogeneous “cap” but rather segregate into distinct areas of the T cell–APC interface, forming a “supramolecular activation cluster” (SMAC) (12) and contributing to what has been called the “immunological synapse” (13). By antibody staining of fixed cells (12), or by individually labeling molecules in a lipid bilayer in place of the APC (13), it has been shown that after just a few minutes of T cell coupling, TCRs on T cells and MHC molecules on APCs segregate to the central core of the contact zone surrounded by lymphocyte function–associated antigen-1 (LFA-1) and intracellular adhesion molecule–1 (ICAM-1) accumulations, respectively.

In order to analyze the role of CD4 and CD3ζ during synapse formation, we made COOH-terminal fusions of these molecules with green fluorescent protein (GFP) as described for ICAM-1 (14) and separately transfected them into the T cell line D10 (15, 16). The clones analyzed had moderate levels of surface GFP expression and were indistinguishable from the parental line in their responses to antigens (Web fig. 1A) (17). Additionally, CD3ζ-GFP and CD4-GFP associate with ZAP-70 and Lck, respectively, and CD3ζ-GFP was phosphorylated in response to signaling (Web fig. 1B) (17). The cells were analyzed with an instrument that coordinated a high-speed piezoelectric objective positioning device to rapidly change the focal plane while “streaming” the acquired data from a fast-cooled charge-coupled device camera. Twenty optical sections through a T cell–APC couple were captured within a 3-s window, and calcium and bright-field images were collected within an additional 2 s.

In Fig. 1, the response of D10 CD3ζ-GFP cells to peptide-loaded CH27 (18) after correcting for photobleaching and background (19) shows CD3ζ clustering in the contact area coincident with the rise in intracellular calcium. Such clusters were typically very dynamic, forming, breaking apart, and re-forming over time, but ultimately coalesced into a single, large central cluster.

Figure 1

Clustering of CD3ζ during the onset of signaling, followed by mature synapse formation. CD3ζ-GFP T cells and strong agonist (10 μM)–pulsed CH27 cells were mixed on coverslips and imaged at 15-s intervals. Each time point shows the DIC image overlayed with the calcium ratio obtained with FURA-2 (340/380), a mid-cell z section of GFP, and projection of the contact-region view obtained by a Max Intensity method. The Max Intensity projection method was applied to a region encompassing the interface [assessed by the DIC image (20)]. Times relative to the onset of the calcium signal are indicated. Pixel values within the resulting central-core clusters were typically at least twice the intensity of the average pixel located along the cell surface. Concentrations of CD3ζ-GFP at the trailing edge, opposite the site where the interface most often forms, are often observed before the first calcium flux, and these typically shifted forward in the first minutes after activation. Within minutes of forming, these larger clusters often diminish in intensity, probably as a result of receptor internalization (33). Occasionally, a small region of internalized GFP was briefly (≤30 s) observed just under the interface, suggesting that internalized receptors and/or GFP are quickly proteolysed or inactivated. We have observed these same phenomena with tagged CD3δ and TCRα constructs (34). Time-lapse movies of these and other CD3ζ-GFP accumulations are available in Web fig. 2 (17).

CD4-GFP accumulation was initially very similar to that of CD3ζ, with punctate accumulations forming coincident with calcium mobilization (Fig. 2). However, starting at about 2 min after calcium signaling, CD4-GFP was frequently diminished in the central interface. This appears as a ring when the interface is viewed en face, and similar structures were observed within 15 min of activation in 69% (n = 29) of T cells responding to 10 μM CA 134-147. Both wide exclusions (reaching the perimeter of the contact zone) and very narrow exclusions (∼2 μM in diameter) were observed and in most cases decreased in diameter over time. The distinct pattern of CD3ζ versus CD4 localization seen with GFP fusions was also confirmed by antibody staining of fixed T-B cell conjugates (Web fig. 4) (17).

Figure 2

Clustering of CD4 during the onset of signaling followed by mature synapse formation. Clustering patterns of CD4-GFP indicate an early recruitment and subsequent exclusion at the T cell–APC interface. D10 CD4-GFP clones were imaged while interacting with strong agonist (10 μM)–pulsed CH27 cells. Typical accumulations in the interface were only twofold higher than average cell membrane intensities. CD4 exclu- sions occasionally collapsed and re-formed during the minutes after activation. This pattern was also evident at similar frequencies when lower peptide concentrations were used, and where no clear exclusion was observed, CD4-GFP accumulations appeared as actively moving patches. A time-lapse movie of this and other CD4-GFP accumulations is available in Web fig. 3 (17).

The appearance of stable CD3ζ and CD4 accumulations was highly coincident with the onset of calcium signaling, and the magnitude of the maximum intensity (20) correlated with the concentration of antigenic peptide (Web fig. 5A) (17). When challenged with weak agonists or with no peptide at all, T cells interfaces typically did not show appreciable, stable accumulation of CD3ζ-GFP or CD4-GFP and frequently had an excluded appearance (Web fig. 5B) (17). Strong-agonist–mediated clusters slowly decrease in number over time as they coalesce into the center of the interface, while the brightest cluster increases in intensity, indicating that small clusters are merging into larger ones (Fig. 3). Weak-agonist peptide stimulation gave clusters that generally lasted only one frame or less (∼15 s).

Figure 3

Coalescence of multiple clusters into a central core cluster. (A) The number of discrete clusters within the interface at various times after the calcium signal onset mediated by APC pulsed with strong agonist (CA 134-147) (•) or weak agonist (E8T) (□) at 10 μM. At 15-s intervals relative to the calcium signaling onset (FURA ratio >150% of nonreacting level), the number of individual clusters, defined as contiguous regions within the interface with GFP intensities of >100% above the cell surface average, were counted. Data show the mean ± SEM for 15 couples. (B) The brightest single-pixel intensity at the front of the cell was compared to the average cell-surface pixel intensity to obtain a ratio. This corresponds to the degree of aggregation of the brightest cluster in the contact zone. Data show the mean ± SEM from scoring 15 couples for each condition in which a calcium ratio change was observed. This allows a consistent and activation-relevant reference point to be used, and those E8T-mediated couples that did not induce calcium signals were similarly unable to mediate stable clustering or coalescence of clusters.

The formation of a stable synapse was modulated by peptide concentration and was severely compromised by blocking either the calcium increase (BAPTA/Ni2+) or cytoskeletal rearrangement (Cytochalasin D or LAT-A), although neither prevented the formation of transient CD3ζ clusters (Web table 1). Blocking B7 costimulation allowed transient clustering and calcium increase, but the frequency of stable CD3ζ accumulations was reduced by about fivefold. Similar cytoskeletal and costimulation dependence is observed for the clustering of labeled MHC molecules on APCs during T cell recognition (21). Previous work has shown that costimulation triggers a cytoskeleton-dependent movement of membrane patches (rafts) (22, 23) into the interface, and it seems likely that this helps to create and maintain the central accumulation, particularly because CD3ζ associates with the cytoskeleton in activated T cells (24, 25).

To examine the coordination of these molecules, we first treated T cells with antibody Fab fragments before adding APCs. Treatment with antibody to CD4 (anti-CD4) prevented CD4-GFP accumulation, slightly reduced CD3ζ GFP accumulation, and significantly inhibited calcium mobilization (Web fig. 6) (17). Anti-TCR blockade prevented CD3ζ-GFP accumulation and calcium increase and abolished the high-density accumulation of CD4-GFP. Thus, CD4 accumulation is dependent on TCR binding and/or signaling even though its ligand (MHC class II) is present on B cells, suggesting that some feature of early TCR–peptide-MHC recognition serves to recruit CD4 to the interface and initiate signaling. This is not a function of the cytoplasmic tail because a deletion mutant behaved in the same way as the intact molecule (Web fig. 7) (17).

In this work we have shown two phenomena relevant to T cell recognition: (i) the appearance of small, dispersed clusters of CD3ζ and CD4 that appear at the same time as the initial calcium response and that, if a sufficient amount of agonist peptide is present, adopt a mature synapse structure and (ii) the differential accumulation of CD3ζ versus CD4.

The initiation of activation in the T cell did not correlate with the formation of the SMAC/synapse structure described by Monks et al.(12) and Grakoui et al. (13), but rather with the formation of much smaller, very unstable clusters of CD3ζ and CD4. These clusters were also seen with weak agonists, although in those situations they rarely coalesced into large, central groupings. Such clusters may represent the initiators of activation, or perhaps even smaller TCR aggregates are the primary initiators (26, 27).

In the case of histamine receptor signaling, threshold levels of calcium can be reached by a combination of frequency, amplitude, and the integration of signals from multiple dispersed receptor clusters (28). We suggest that a similar process occurs with T cell signaling and is responsible for the initiation of T cell activation. This early “fluid” phase might be driven by specific TCR-p/MHC oligomerization, as observed in solution studies (29) and independent of the cytoskeleton. Soon after signaling is initiated, however, CD3ζ and other molecules would move into the center of the interface by costimulation-induced, cytoskeleton-mediated membrane raft movement (22, 23). This stable accumulation is likely to be required for sustained signaling and subsequent progression to full activation and cytokine secretion (30).

With respect to the differential localization of CD4 versus CD3ζ, these findings do not support a model in which CD4 acts to stabilize the TCR-pMHC complexes. No evidence for such stabilization has been seen in solution studies, and direct MHC class II–CD4 binding was not detected (31). These results and the data presented here instead suggest that the function of CD4 may be to “boost” or “trigger” the early phase of activation. Once that process is under way, CD4 seems to be excluded from the central core of the synapse, perhaps owing to the formation of some lattice-like structure by the remaining molecules.

  • * Present address: Center for Immunology, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390–9093, USA.

  • To whom correspondence should be addressed: E-mail: mdavis{at}


Stay Connected to Science

Navigate This Article