Signaling clusters in the cell membrane

Abstract

Large-scale molecular assemblies, or signaling clusters, at the cell membrane are emerging as important regulators of cell signaling. Here, we review new findings and describe shared characteristics common to signaling clusters from a diverse set of cellular systems. The well-known T cell receptor cluster serves as our paradigmatic model. Specifically, each cluster initiates recruitment of hundreds of molecules to the membrane, interacts with the actin cytoskeleton, and contains a significant fraction of the entire signaling process. Probed by recent advancements in patterning and microscopy techniques, the signaling clusters display functional outcomes that are not readily predictable from the individual components.

Introduction

Living cells interpret and respond to numerous signals from their environment. Receptors in the cell membrane are generally the initial point of interaction for incoming signals. At the most basic level these receptors provide specificity, for example by binding a unique chemical ligand. This information is transduced across the membrane to a cascade of chemical signaling reactions that perform logical operations and, ultimately, make decisions. In recent years, the role of spatial arrangement and assembly of signaling molecules into organized structures on the membrane is emerging as a significant component of signal regulation [1••,2,5,6,7•,8–13]. Advancements in the application of optical, spectroscopic, and materials patterning techniques to questions in cell biology have provided new angles of illumination on membrane signaling processes. Of particular interest is the discovery of submicronscale receptor signaling clusters in several cell systems [1••,2,5,6,7•,8–13]. These clusters can contain hundreds of molecules, are often associated with the cytoskeleton, and appear to encompass much more of the entire signaling cascade than originally thought to occur on the membrane surface. Most importantly, the physical assembly of the signaling system gives rise to emergent functional properties that can transcend the simple approximations of cooperativity, such as Hill coefficients and allosteric effects [14–16]. In this review, we present a study that compiles recurrent themes in membrane signaling that may link disparate systems based on large-scale spatial assembly of signaling molecules. Although current knowledge of many specific details in these systems is still incomplete, the common motifs presented here provide new insight and may help to predict yet undiscovered properties of these and related systems.

T cell receptor microclusters as a paradigm

Whereas small-scale receptor oligomerization has been discussed within the context of membrane signaling for decades, prominent examples of large-scale spatial patterns, such as the immunological synapse (IS) [8–11] have emerged more recently. The IS comprises an intercellular junction between immune cells and their target cells [8–11]. Upon binding, various receptor-ligand pairs become sorted into distinct spatial patterns that extend to microns in size across the interface. Among the various spatial components that can be identified within the IS, it has become clear that the T cell receptor (TCR) microclusters, along with clusters of other downstream signaling molecules, are the active signaling units—sometimes referred to as signalosomes [15] (Figure 1). There have been significant advances in our understanding of TCR microclusters over the past few years, and these new discoveries may reflect a more general theme in biology.

Figure 1. A signaling cluster.

Schematic of a T cell receptor microcluster, also called the TCR signalosome, which is densely packed with a diverse set of signaling molecules. Activation of actin assembly initiates centripetal transport. Further details can be found in the text and in references [1••,2–4].

Upon first engagement with antigen peptide-bound major histocompatibility complexes (pMHC), TCR molecules congregate into large clusters containing a hundred or more individual receptors. Here, other key molecules, such as lck [17,18], Lat [19], SLP76 [20,21] and actin [22,23] are heavily recruited [1••,2,3,13,24]. Much effort to understand TCR microclusters has focused on documenting their assembly and content (for in depth reviews see [1••,3,4]). Some very recent work probes the functionality of the signaling cluster by direct manipulation. Supported membranes have long been used as surrogate antigen presenting cell (APC) surfaces for T cell activation. By patterning structures onto the underlying substrate, it is possible to guide and restrict the assembly of signaling clusters in living T cells. This technique, which we refer to as a spatial mutation [25,26] has been used to demonstrate that T cell triggering thresholds are determined based on the number of agonist-bound ligand in a single microcluster, and not by the total number encountered by the cell (Figure 2). Thus, TCR microclusters appear to function with a high degree of internal cooperativity but little to no cooperativity between clusters [27•].

Figure 2. Titration of signaling clusters.

(a) Brightfield images of cells off and on the indicated grid pattern size (scale bar = 5 mm). (b) Schematic of TCR microcluster within indicated area in (a) with pMHC bound to activating agonists (stars) and non-activating null (circles) peptides. (c) Corresponding heat maps that display calcium flux for a population of cells on and off the grids, with each cell shown as a horizontal line and >100 cells per heat map.Adapted from reference [27•].

Interactions of signaling microclusters with the actin cytoskeleton

A common feature of signaling clusters is association with the actin cytoskeleton. Actin inhibition reveals that transport of TCR microclusters at the periphery is actin-dependent [22–24]. Further insights have been obtained from live cell imaging of the transport process as microclusters traverse maze-like configurations of mobility barriers [28]. Upon encountering a barrier during centripetal transport, the TCR microcluster trajectory is deflected along the barrier until it can continue its course through an opening. No elastic recoil of the cluster towards its original trajectory was observed. Furthermore, the speed of the deflected cluster scales with the cosine of its deflection angle to that of the flow, which is consistent with a dissipative or frictional coupling mechanism. In other studies with Jurkat cells, TCR and actin have been measured to move with different velocities [29•] and actin flow has been observed to slow but not stop over regions of trapped TCR [30]. Thus dynamic associations between TCR and actin allow force transmission without direct coupling. Similar actin coupling has also been observed for cadherin [31,32]. Cadherin clusters at adherens junctions regulate actin growth through the adaptor protein alpha-actinin [31–33]. Recently emerging imaging methods [16], such as image time autocorrelation analysis [34] (Figure 3), provide alternative ways to monitor and quantify interactions between the cytoskeleton and cell surface signaling clusters. For many of these studies, the use of patterned substrates to block or control TCR cluster movement is key.

Figure 3. Measuring TCR–actin interactions.

(a) Schematic of a T cell on a substrate patterned with diffusion barriers. (b) Left, Time average of a stack of fluorescent actin-GFP. Right, Trace and plots illustrate temporal fluctuations along with time correlation, G(τ). (c) Pixel by pixel image of autocorrelation times reveals regions of high actin volatility and stability.

The association of actin with TCR microclusters may do more than drive lateral translocation. Two recent studies, using different but equally ingenious methods, have measured the on-off kinetics of the TCR–pMHC interactions in living cells. Both report these interactions to be more dynamic than previously thought, with much faster k_off rates than was measured in vitro [35••,36••]. It may have been natural to assume that actin would stabilize TCR–pMHC at the molecular level, but Huppa et al. [36••] found that actin polymerization somehow contributes to the more dynamic TCR–ligand interaction. Consequently, actin appears to exert direct influence over the ligand-binding properties of TCR, providing a level of active control that the T cell can, in principle, modulate. This observation exemplifies how the emergent properties of signaling microclusters are not easily predictable from the primary inputs.

Triggering actin assembly

Several reviews discuss in great detail all the molecules thought to be involved in TCR–actin interactions [13,37–41]. One general theme is that molecules upstream of the Arp 2/3 complex become activated through a series of phosporylation-dependent reactions initiated at TCR microclusters [41]. Activation of the Arp 2/3 complex ultimately promotes actin branching and lamellipodia formation [42]. This event brings together signaling molecules, which may already be pre-clustered, to the inner leaflet of the membrane and the cytosolic side of the clusters. Recently, Lebensohn and Kirschner showed that Rac and cdc42, which activate Arp2/3, needed to be membrane-associated in the presence of the lipid PIP3 to fully activate Arp2/3 [43••]. More importantly, the WAVE complex must be activated to initiate Arp2/3 nucleation, which is the rate-limiting step for actin polymerization. This necessitates the simultaneous interaction of the WAVE complex with both Rac-GTP and acidic phospholipids at the membrane, such as PIP3. In accordance with this, some believe that negatively charged lipids, which are products of the TCR signaling cascade, may also promote actin nucleation [43••].

Impact of protein clustering

Crosslinking antibodies are commonly employed as alternative methods to trigger cell surface receptors. For example, antibodies to CD3, which is closely associated with TCR, are widely used to trigger TCR. Immobilized Anti-CD3 antibody on glass can induce Jurkat T cell activation, bypassing the need for TCR ligand binding [44,45]. Anti-CD3 not only crosslinks TCR into dimers, but also ultimately amplifies the TCR cluster assembly process (possibly in unnatural ways). MHC oligomers have also been widely used to activate T cells in solution; the monomer is inactive [46]. Crosslinking of CD28, a T cell costimulatory molecule, with a superagonist antibody without concomitant TCR or CD3 ligation produces similar results [47]. Hünig and Dennehy propose that interaction of the superagonist with bivalent CD28 can lead to supramolecular structures that favorably position receptors and signaling molecules to mediate signal transduction [47].

The effect of clustering on receptor signaling can also be observed when higher order complexes of downstream signaling molecules is induced. For example, the nucleotide switch that activates cdc42 can be bypassed through artificial dimerization of the GTPase with WASP [48]. Moreover, increased activation and affinity for Arp 2/3 have been observed for VCA domain dimers and N-WASP molecules that are incorporated into assemblies [49,50]. Padrick et al. [51] systematically tested three agents—EspFu, SH2 dimers, and PIP₂ containing vesicles—that can cluster WASP proteins. They found that both allosteric activation and oligomerization are necessary for dramatically increasing actin assembly [51].

Protein cluster coupling to actin may be examined using antibody crosslinking experiments. A laser tweezers study measured drag forces on GPI-linked proteins as a function of antibody-crosslinking and revealed evidence of cluster size-based protein coupling to actin [52]. Clustering cell surface receptors can also lead to long-range spatial sorting in the context of actin-driven flows. In the T cell IS, differential clustering of the primary adhesion molecule lymphocyte function associated antigen-1 (LFA-1) leads to quantitative changes in its distribution within the synaptic junction [53•]. Bivalent or tetravalent antibody crosslinking of LFA-1, or its ligand, caused the protein to be sorted progressively closer to the lateral center of the circular junction [29•]. Local changes in receptor clustering are thus translated into global changes in receptor organization over the cell surface.

Interestingly, the idea of flow-assisted sorting has previously been observed for the effective clearance of antibody from the surface of motile trypanosome parasites [54]. Engstler et al. [54] showed that parasite swimming induced hydrodynamic flow, which led to the size-based sorting of antibodies to the posterior of the cell. Larger antibodies (IgM) are transported more rapidly than the smaller antibodies (IgG) to the flagellar pocket for endocytosis and degradation. In the case of the IS, the cell is static and does not undergo extensive hydrodynamic flow. Instead, the actin centripetal flow transports molecules and the cluster size determines the coupling strength and, ultimately, extent of radial transport [53•].

The growing use of antibody-based drugs in the clinical setting underscores the need to better understand receptor clustering and its broader effects on signal transduction processes. For example, an anti-CD20 antibody for cancer therapy induces lysosomal release of the reactive oxygen species for cell killing in an actin-dependent manner [55]. A potential link for this effect is the actin-associated protein Vav, which can control superoxide production [56]. Additionally, Benson et al. [57] observed increased killing of cancer cells by natural killer (NK) cells in multiple myeloma using an antibody against PD-1. Actin also appears to be affected, as the migration and cell morphology of the NK cells are changed with the antibody addition. The effect of these antibodies on actin is not well understood, and we speculate that such secondary effects may result from underlying, actin-involved receptor assembly processes.

Signaling clusters in other systems

Juxtacrine signaling in other cell systems also display features similar to those found in the T cell immunological synapse. One recent example is the EphA2-EphrinA1 system, which regulates cell adhesion, motility, and angiogenesis [58]. EphA2 binding to EphrinA1 leads to the formation of clusters that undergo actin-driven transport at the cell membrane [7•]. Spatial mutation studies, in which EphA2-EphrinA1 cluster transport was blocked, reveal discrete changes in the recruitment of the metalloprotease ADAM10 [7•]. This suggests the possible existence of mechanoregulation in this system as an emergent property of the signaling cluster [59].

In another example, the receptor tyrosine kinase MEK forms oligomers that preferentially sort to actin-dependent dorsal ruffles, where they undergo increased internalization [60]. Similar behavior has also been reported for activated epidermal growth factor receptor (EGFR) [61]. This same study further demonstrates that endocytosis may be independent of the traditional pathways, which are clathrin-pit or caveolae dependent [61]. Actin enrichment near the membrane has also been observed during trans-Notch-Delta binding [62]. To regulate signaling, cis-Delta-Notch complexes are quickly degraded [63], and it would be interesting to see if their internalization were actin-dependent.

Lipid rafts

Clusters of membrane signaling proteins can hardly be discussed without invoking the concept of membrane lipid rafts [64]. Despite immense experimental effort, the nature of lipid rafts and the balance of contributions from lipids or proteins in their formation remain controversial. What is clear is that signaling clusters of the types highlighted here are highly specific with respect to their molecular content, spatial positioning, and function. Lipids and cholesterol may contribute to their assembly and stabilization, but the proteins define the specificity and ultimate functionality.

Concluding remarks

Signaling clusters from a number of different biological systems share a surprisingly large number of physical features, such as actin recruitment and large-scale transport. Currently, much of biology is driven by to the goal of understanding how specific functions arise from underlying molecular mechanisms. In the case of signaling clusters, there is an opportunity to uncover common physical mechanisms that are employed in a diverse range of biological functions.