The ubiquitous nanocluster: A molecular scale organizing principle that governs cellular information flow

Abstract

The language of biology at the scale of the cell is constituted of alphabets represented by biomolecules. These are stitched together in a variety of ways to create meaning. We argue that the phrases of this language are nanoscale molecular assemblies or nano-hubs for the purpose of information flow. At the cell surface information is sensed and processed via membrane receptors, often configured as multimers. These nano-assemblies serve as receiver nano-hubs, which are flexibly configured with additional nano-hubs that we term modifiers and transducers. This framework serves to process information that is transmitted for execution inside the cell. Here, we explore some examples about how nano-hubs are built and how they may contribute to cellular information flow.

Introduction

Research over the last two decades has provided compelling evidence that the physical principle of molecular spatiotemporal organization orchestrates information flow at the cell surface. In particular, clustering has emerged as an ubiquitous organizational principle across a myriad of different receptors that bind to distinct ligands, expressed in many cell types and performing different functions [1–4]. What are the fundamental principles that govern receptor nanoclustering? How are nanoclusters assembled (and dissembled)? What is their role? How are they regulated with such a high fidelity and universality? Here, we provide a vision, based on current literature and our own findings, that addresses these questions.

Nanoclusters can take several forms, typically governed by the mechanism of their construction (see Box 1) and location. Here, we focus on nanoclusters that engage in information flow in the form of chemical and mechanical signals received at the cell surface. These signals are sensed and processed at the plasma membrane where information is then conveyed to the intracellular environment via different signaling cascades for appropriate cell responses.

Box 1. Mechanisms of nanocluster formation and rearrangement.

• i)Chemical complexation, i.e., thermodynamically driven interactions: Specific molecular interactions leading to the formation of molecular complexes; protein-protein, protein–lipid, lipid-lipid, protein–DNA, RNA, and DNA, RNA–DNA, RNA. Homomeric and heteromeric complexes are often the nature of these assemblies.
• ii)Ligand-driven rearrangement: An essential feature of chemical communication is that of ligand-induced signaling at the cell surface by membrane receptors. Here, information in the form of extracellular proteins such as growth factors (e.g., epidermal growth factor (EGF)), hormones (e.g., insulin) or small molecules (e.g., serotonin) bind to their cognate receptors (e.g., EGF-receptor, insulin receptor, or the G-Protein coupled receptor, serotonin receptor, respectively) and lead to changes in nanoscale organization of the membrane receptors. Cross-linking of the receptor (B-cell receptor) is a common feature of multi-valent ligands that create nano-assemblies.
• iii)Self-assembled, thermodynamically driven nano-structures: Virus particles, clathrin pits, membrane coat-proteins, biomolecular condensates) or nanoscale assemblies that are shaped by curvature (e.g., BAR-domain containing proteins).
• iv)Self-organized, energy consuming nano-assemblies: Small GTPase driven structures (e.g., Rab domains) or dynamic actin-templated structures such as GPI-anchored protein nanoclusters whose assembly would not occur in the absence of the energy consuming state.
• v)Meso-scale assemblies: Hierarchically from nanoscale structures such as active emulsions in membranes, or meso-scale assemblies that form from barriers imposed by the cortical actin cytoskeleton, or phase separation of weakly interacting molecules resulting in membraneless biomolecular condensates. Some of these meso-scale assemblies formed by different nanoclusters often not coalesce but remain in close spatial proximity to each other.

Nanoclusters as units of functional activity on the cell membrane

Most receptors at the cell membrane are organized as nanoclusters [2] and as we discuss here, many of their downstream signaling components are also organized as pre-assembled nanoclusters. We define these nanoclusters as nano-hubs with elements of the signal receiving and processing system that may be classified as receivers, modifiers and transducers, each with a specific executive function (see Figure 1 and legend). To ensure an effective response while providing versatility and diversifying function, the cell has evolved multiple strategies to exploit the modularity this level of organization offers. This takes place by the spatiotemporal regulation of interactions between different functional nano-hubs leading to their selective concatenation, depending on the input (Figure 1). Such traits have been observed in different receptor systems and are discussed below in their specific context. This general organizational principle provides an explanation why the same receptor may elicit different cellular responses depending on the context, by associating with (or segregating from) a specific nano-hub and for a determined dwell-time. This view differs from the classical picture of the signaling receptor where upon ligand binding, the receptor triggers a set of sequential chemical reactions and consequent molecular assembly steps that results in a signaling output. In this new picture, nano-hubs serve as modular units whose concatenation in a specific spatiotemporal sequence leads to a distinct signaling output (Figure 1).

Figure 1. Nano-hubs as logic gates.

At the cell surface, receptors (as receivers) may be pre-assembled nanoclusters or those that are assembled at the cell surface by a variety of means (see Box 1). Co-receptors and adaptor proteins (modifiers) and different signaling components (transducers) also organize as nano-hubs. Receptor nanoclusters interact with diverse inputs such as small molecules or soluble protein-ligands (a), ligands presented at cell surfaces (b), or virus particles (c), in the presence (a, c) or absence (b) of independent mechanical and chemical inputs from the extracellular matrix. We define nano-hubs as the nano-clustered organization of the various elements of the logic gates of this information processing system. Nano-hubs: i) they concentrate molecules in a hub; ii) they have unique identity (they normally contain only one type of molecule); iii) they have a specific function (like logic gates). These features resemble functional circuits where nano-hubs may be “interconnected” to diversify and perform different functions, or trigger different signaling pathways. The flexible concatenation of the receiver nano-hubs with modifier (co-receptors or adaptors) and transducer (signaling assemblies) nano-hubs lead to a variety of outputs based on the nature of the concatenation. Concatenation between different nano-hubs can be achieved by the cells using different strategies such as the dynamics of the actin cytoskeleton, generation of lipid ordered domains, phase separation, and many others.

Hartwell and colleagues [5] argued that “Modular structures may facilitate evolutionary change, and embedding particular functions in discrete modules allows the core function of a module to be robust to change, but allows for changes in the properties and functions of a cell (its phenotype) by altering the connections between different modules.” We propose that at the level of a signaling receptor, the nano-hub is a unit of such modularity, and that connections between different nano-hubs together encode a specific functional outcome.

Modularity coupled to flexible concatenation enable design principles familiar to engineers. These include: i) positive and negative feedback loops, ii) coincidence detection to prevent triggering of the system by random fluctuations in input, iii) logic gates, iii) amplification of small differences in input signals generated by small asymmetries or stochastic fluctuation of signals. Different from principles familiar to engineers and the static nature of the unit devices, nano-hubs at the cell surface are flexibly made up of diverse biomolecules, which may be further refined to give unique nano-hubs. These units encode a specific and reliable function that is wired into the specific composition and structures of its constituents. On their own, stochastic combinations of individual constituents would otherwise generate noisy, probabilistic intermediates. The flexible composition of the nano-hub also serves a central paradigm of a biological system, that of evolvability, where a particular solution is a result of an elaborate historical process of biological selection, and may be subject to change.

This framework raises questions on how specific nano-clusters may arise in different contexts to convey appropriate information in space and time. Nanoclusters at the single unit level might be pre-complexed inside the cell and emerge at the cell membrane as a “ready-to-use” functional nano-hub. Alternatively, nanoclusters may be assembled at the cell surface via one or more of the mechanisms mentioned in Box 1. Literature in the field also indicates that different nanoclusters do not merge together upon stimulation (such as ligand binding), but reorganize in close proximity to each other [3,6,7]. Regulation of physical proximity or segregation between stimulatory and inhibitory modifiers of receptor nano-hubs is another feature of such a framework. This allows different cell types to use the same receptor and signaling pathways or transducers to achieve unique outcomes (Figure 1). A new understanding of the cell membrane [8], also offers many possibilities for realization of the framework of nano-hubs, their modifiers and transducers. Barriers defined by the cortical actin cytoskeleton such as picket fences [9] as well as the juxtamembrane dynamics of the actin cortex serve to maintain nano-hubs segregation. Phase separation and/or formation of biomolecular condensates [10] or active emulsions [11] might tune the aggregation, segregation or concatenation of specific nano-hubs, ultimately regulating the composition in the vicinity of the receptor nano-hub and thereby its signaling outcome.

Specific nano-hubs, their supramolecular assemblies and functional readouts

In the context of a typical tyrosine kinase receptor involved in ligand-gated signaling, the Epidermal Growth Factor receptor (EGFR) serves as a good example of the nano-hub framework for soluble ligands (Figure 1a). EGFR itself is present mainly as a monomer in dynamic equilibrium with a multimer and depending on the concentration and nature of the ligand it can exhibit multiple oligomeric states [12,13], different signaling kinetics [14], or adopt different conformations of its liganded receptor [15]. A recent single molecule study further showed that the partitioning of EGFR molecules in the proximity of tetraspanin nanodomains (i.e., modifier) regulate ligand binding and subsequent receptor signaling both in terms of its magnitude and sensitivity [16]. Thus, spatial and temporal diversity of the receiver nano-hub allows association with different downstream nano-hubs (transducers) resulting in distinct differentiation or proliferation responses (Figure 1a).

Immune receptors, and their modifiers and transducers

Spatial organization and modularity are highly relevant to the functioning of many immunoreceptors [20]. For instance, the IFN-γ receptor (IFNγR) plays a key role in innate and adaptive immunity for host defense against intracellular infections and tumor control. Using fluorescence correlation spectroscopy, it was shown that partitioning of the receptor either in lipid nano-domains or in nanoclusters formed by the actin cytoskeleton control cytokine signal transduction through the JAK/STAT signaling pathway; lipid nanodomains inhibit JAK/STAT signaling whereas actin-mediated nanoclusters trigger robust signaling [17]. This process seems to be regulated by the binding of galectins which shift the receptor into one or another type of nanoclusters. Thus, a similar ligand and similar receptor exhibit different signaling response according to whether it associates with one or another type of modifier nano-hub.

Another archetypical receptor studied in the context of nanoclustering and signal regulation is the T cell receptor (TCR) which is expressed in Tcells and engages with antigen presenting surfaces, often along with the engagement of multiple co-receptor inputs such as integrins, co-stimulatory proteins and others (Figure 1b). Nanoclustering of not only the TCR, but also of its main co-receptors such as CD4 and major downstream signaling molecules, including Lck, Zap70 and LAT have been invoked to play a major role in the discrimination of different stimulation strengths and signal integration to regulate T cell activation [18–20]. Different forms of super-resolution microscopy have shown that many of these molecules form distinct and spatially segregated nanoclusters prior to TCR-ligation [6,21,22]. Interestingly, TCR engagement to its ligand does not lead to the coalescence of TCR and LAT nanoclusters, but rather to their spatial concatenation while maintaining discrete boundaries between them [6]. A similar form of concatenation has been observed between activated TCR, its co-receptor CD4 and active Lck nanoclusters [23]. The signal modifier/transducer, the Ras-MAPK cascade is also arranged in a modular fashion (see below) and phase transition of the LAT (linker for the activation of TCR)-Grb2-SOS which generates condensates, regulates the activation of the Ras-MAPK signaling pathway downstream of both TCR and EGFR signaling [24]. These observations re-enforce the notion that receiver nano-hubs maintain their specific function but interact in a modular fashion (in space and in time) with other nano-hubs to selectively tune T cell activation, converting analog inputs into nearly digital responses, i.e., amplifying the response when needed or by suppressing noise, in the case of weak stimuli. This nano-hub framework has also been implicated in B cell receptor (BCR) organization and signal propagation [7,25]. Co-crosslinking of the BCR receptor with molecules that modify the membrane composition in the vicinity of the receptor greatly modulate the efficiency of BCR activation [25].

This framework is also observed for receptors involved in host-pathogen interactions, including the Fc receptor, Dectin-1 and their coupling to different toll-like receptors (TLR) expressed on macrophages. These receptors reside in separate but adjacent nanoclusters in the plasma membrane, and their cooperativity with TLR signaling is accompanied by their concatenation, but not the coalescence of individual nanoclusters [26,27]. DC-SIGN, a C-type lectin expressed on dendritic cells (DCs) also exhibits nanoclustering, which is crucial for the capture of HIV [28,29]. Multi-color single molecule imaging together with super-resolution microscopy showed that nanoclusters of CD44 and of Gal-9 contribute to clathrin-dependent DC-SIGN endocytosis [30]. Similar to other immune receptors, engagement of DC-SIGN with the virus induces the concatenation of the tripartite protein nanoclusters (DC-SIGN, CD44 and Gal-9) without coalescence of the individual nanoclusters [31](see also Figure 1c). It should be noted that high-density single molecule imaging approaches using multiple colors are providing valuable insights into the simultaneous spatiotemporal dynamics of the interplay between nano-hubs and between the nano-hubs and their environment [31,32].

Integrins

Recent studies have shown a highly complex organization of integrins and their adaptor proteins inside focal adhesions (FA) [33,34]. Aside from their axial nanoscale segregation [35], integrins and their main adaptor proteins including vinculin, paxillin, zyxin, CAS, FAK form mixed nanoclusters that spatially segregate from each other inside FAs [34]. β₁ integrins also segregate into active or inactive nanoclusters with only a sub-population of the former being present inside FAs [36]. This trait appears common to integrins, since similar nanoscale functional segregation has been documented for the α_Lβ₂ integrin on immune cells [37,38]. Since integrin activation is highly dynamic and transient [39], “ready-to-use” integrin nano-hubs could provide a fast and efficient mechanism to increase the engagement lifetime to the ligand as well as its flexibility to sample the extracellular milieu. Integrin receptors are highly responsive to the activation of signals that modulate the actin cytoskeleton, and the strength of their interactions with the extra cellular matrix is profoundly influenced by inside-out signaling [40,41].

Modifiers nano-hubs

The actin cytoskeleton

The mechanical (stiffness and texture) and chemical environment (extra-cellular matrix) of the cell can function as general inputs to reconfigure modifiers which in turn modulate the information processing via nano-hubs. This is because cells respond to mechanical inputs by generating intracellular mechanical responses. This often takes place via the activation of the actomyosin machinery that is relayed to the plasma membrane to activate mechano-sensitive receptors [25,42].

External forces can also induce actin cytoskeleton remodeling, modulating the binding of different proteins to actin [43,44] thereby influencing the spatio-temporal organization of receptors at the cell membrane. For instance, in the presence of shear-flow, endothelial cells re-arrange their actin cytoskeleton promoting the formation of ICAM-1 nanoclusters, adhesion receptors that do not directly interact with actin. Such nanoclustering strengthens ICAM-1 interaction with their integrin counter-receptors on leukocytes thereby regulating leukocyte adhesion and migration [45]. This reflects the cross-talk that exists between the extracellular milieu of a cell and its ability to influence signaling responses. We argue that this cross-talk is mediated by modifier nano-hubs that are flexibly configured by these inputs.

Siglec-1, the main receptor for HIV on mature DCs, forms nanoclusters prior to virus engagement and the actin cytoskeleton is both responsible for nanocluster formation and their spatial segregation. HIV binding leads to actin remodeling, enhancing Siglec-1 nanoclustering and facilitating virus accumulation in a saclike compartment [46]. Similar actin-mediated processes have been reported for the Fcγ phagocytic receptor [47]. The dynamics of actomyosin also plays an important role in tuning the physical segregation and nanoscale assembly of the cell adhesion protein, E-cadherin, both at celle–cell junctions as well as at extra-junctional membrane [48]. This serves to tune the strength of cell-cell adhesion, making it responsive to elements that control acto-myosin driven cortical cytoskeleton. Thus, the actin cytoskeleton serves as major modifier of many signaling responses due to its dynamic nature, organization, and its capacity to be reconfigured at the nanoscale in response to the mechanical and chemical environment of the cell. In many instances the precise mechanism by which the actin cytoskeleton modifies the nano-hubs, remains to be determined.

Lipid anchored proteins as modifiers of signaling responses

While the cortical actin cytoskeleton plays an important modifier role, it’s influence is exerted over a large spatial scale since it is juxtaposed to almost the entire membrane, whereas lipid-tethered proteins, in particular are capable of modifying the local (lipid) environment in the vicinity of receptor nano-hubs. The inner leaflet lipid-anchored RasGTPase isoforms self-assemble into nanoclusters by deploying combinations of multiple protein interfaces to form oligomers of diverse sizes, topologies and internal structures [49]. These protein-protein interactions interfaces located in the GTPase domain of the protein intersect with the lipid-binding capacity of the hypervariable membrane proximal domain of the Ras isoforms, guiding associations with specific lipid species [50]. This contingent structural assembly process involving both protein-protein and proteine–lipid interactions and generates nanoscale lipidic environments around specific Ras isoforms and the battery of its oncogenic mutants. The availability of different protein interfaces in the different nucleotide states of the Ras isoforms underlies the mechanism behind lipid-mediated spatial segregation of nanoclustering of RasGTP from RasGDP [51]. Importantly, Ras nanoclusters are modular assemblies that require other modifier components such as lipids and membrane-associated proteins such as galectins [52], as well as the actin cytoskeleton [52,53] for their formation. This facilitates diverse Ras-signaling platforms, exemplifying the paradigm of modularity discussed above. These function in specific oncogenic contexts, subtly tuning the output needed for signaling propagation [54].

The rules that have been determined for RasGTPases may likely apply to other lipid-tethered small molecules GTPases. Recently Rac1, a RhoGTPase has been shown to form nanoclusters, especially at sites of its activation [55]. Nanoclustering of Rac1 was associated with the recruitment of specific phosphoinositides and actin regulators such as WAVE indicating that this system also follows the same logic of a locally assembled nanocluster recruiting other protein modules to facilitate function, which in this case is Arp2/3-driven actin polymerization.

Membrane recruited RabGTPases also specify an information cascade about the changing nature of membrane compartments that they associate with. This is exemplified in the context of the maturation of the Rab5-marked early endosomes to the Rab7-marked late endosome [56] or the conversion of Ypt1p and Ypt32p, on the Golgi in the exocytic pathway. They demarcate distinct membrane domains marked by different Rabs that eventually sort membrane and vesicular contents from the same membrane compartment [57]. For these functions they utilize a very similar modular motif involving the concatenation of nano-hubs, each constituted by the construction of a specific Rab domain [58]. The Rab-domain is built by a set of proteins that are specified by the activated Rab and its effectors wherein the cooperativity and the self-assembly properties of the individual components serve to stabilize the Rab domain. This occurs via a positive feedback loop where the localization of one component depends on the recruitment of the other. The activated GTP-bound Rab protein domain then specifies the function of the membrane compartment or the patch of membrane by the recruitment of a diverse set of effectors including GEFs and GAPs for Rabs, SNARE proteins, motor proteins. It is likely that the effectors also exist as nanohubs, therefore permitting more efficient functional outcomes of membrane sorting, compartment maturation or fusion and vesicle motility.

Outer leaflet lipid-anchored molecules such as GPI-anchored proteins and glycolipids form nanoclusters driven by engaging in cholesterol-dependent transbilayer linkages with corresponding long acyl chain lipids of inner-leaflet phosphatidylserine lipids [59,60] and the activity of actomyosin at the inner leaflet [61,62]. These assemble into mesoscale domains with multiple functions in sorting and signaling. Most recently they have been shown to provide cells with the capacity to spread and migrate [63], a function likely mediated by the ability of the liquid ordered (lo)-domains to stabilize activated Rac1 [64]. In membranes maintained far away from phase segregation thresholds, the activity of acto-myosin applying active stresses at the inner leaflet drives the formation of local lo-like nanodomains which in turn assemble into resolvable mesoscale lo-domains called active emulsions by the inherent lateral interactions that exist between molecules of the lipid bilayer [11]. These features may be used to specify the composition of the immediate neighborhood of receptor nano-hubs, thereby modifying the output of the membrane receptor hub as observed in the case of integrin activation [38,63].

Perspectives and conclusion

Cell surface receptors appear to be largely organized as modular nano-hubs flexibly assembled to receive input and subsequently concatenated with other hubs to enable transmission of diverse outputs. Here, we have proposed a modular framework outlining the roles of receptor and modifier nano-hubs. However, due to its complexity and nascent understanding of its organization inside the cell, we have not addressed the functional form of the transducer nano-hubs that must inevitably couple with this system. Our proposal has a number of implications for information flow at this scale. While serving as logic gates, or coincidence detectors, the flexible and dynamic nature (assembly and disassembly) of the nano-hubs lends itself to high sensitivity and fidelity since proof-reading is inbuilt into such a system [65]. For example, the stabilization of SOS at the cell surface occurs via its ability to be linked to the LAT nano-hub (biomolecular condensate constructed of phosphorylated LAT and Grb2) [24]. This stabilized form of SOS efficiently activates downstream Ras, leading to an elegantly encoded proof-reading mechanism in the triggering of the MAPK activity utilizing the principle of nano-hub modularity.

Nano-hubs are influenced by the recruitment of modifiers of the membrane environment (e.g., by the generation of active emulsions) or local biochemistry (via biomolecular condensates). Using this “infrastructure,” they provide a way of modulating information flow from the outside to the inside. How this influence is exerted will require deeper investigation. This will require the development of new technologies that provide molecular and nanometer spatial resolution simultaneously with increased temporal resolution across multiple spatiotemporal scales. Implications of such a modular system on information processing in biological systems will also require deeper engagement with concepts in information and control theory. Multicolor strategies, deep-learning algorithms and modeling will be crucial to resolve and quantify the dynamic nature of nano-hubs interactions while interrogating cellular output. A key ingredient of this system is the inbuilt evolvability by being able to flexibly concatenate nano-hubs and create new outcomes which may be fit(ter) for purpose. Identification of this type of nano-scale modular architecture in the functioning of the myriad information processing systems in the cell is likely to be a subject of future research.

Data availability

No data was used for the research described in the article.