Innate immune receptor clustering and its role in immune regulation

ABSTRACT

The discovery of receptor clustering in the activation of adaptive immune cells has revolutionized our understanding of the physical basis of immune signal transduction. In contrast to the extensive studies of adaptive immune cells, particularly T cells, there is a lesser, but emerging, recognition that the formation of receptor clusters is also a key regulatory mechanism in host–pathogen interactions. Many kinds of innate immune receptors have been found to assemble into nano- or micro-sized domains on the surfaces of cells. The clusters formed between diverse categories of innate immune receptors function as a multi-component apparatus for pathogen detection and immune response regulation. Here, we highlight these pioneering efforts and the outstanding questions that remain to be answered regarding this largely under-explored research topic. We provide a critical analysis of the current literature on the clustering of innate immune receptors. Our emphasis is on studies that draw connections between the phenomenon of receptor clustering and its functional role in innate immune regulation.

Summary: This Review discusses the formation and mechanisms of innate immune receptor clustering and how nanotechnologies have contributed to understanding the regulatory role of receptor clustering in innate immune cells.

Introduction

Since the discovery of the large-scale segregation of proteins at the intercellular junction between T cells (Grakoui et al., 1999; Monks et al., 1998) and B cells (Batista et al., 2001), extensive studies have shown that the phenomenon of receptor clustering is shared by several types of immune cells (Carroll-Portillo et al., 2010; Davis et al., 1999; Goodridge et al., 2011) and is likely a general theme in biology (Cebecauer et al., 2010). It is now recognized that the activation of those immune cells, such as T cells and B cells, relies on the collective interactions between receptors, from oligomers to the microscale (Dustin and Groves, 2012; Fooksman et al., 2010; Kuokkanen et al., 2015; van der Merwe and Dushek, 2011). This recognition has transformed our understanding of the physical basis of immune signal transduction. In contrast to the extensive studies on adaptive immune cells, the existence of receptor clusters and their biological significance in innate immune functions remains largely unknown.

A key function of the innate immune cells, including macrophages, neutrophils and mast cells, is to detect and eliminate non-self particulates, from invading microbes to apoptotic cells. Unlike adaptive immune cells (T and B cells) that express a massive repertoire of antigen-specific receptors, the innate immune system relies upon a limited number of germline-encoded receptors that recognize either structural shapes conserved among microbial species or endogenous danger molecules released from damaged or dying cells (Takeuchi and Akira, 2010). Those receptors are called pattern recognition receptors (PRRs). For example, Toll-like receptors (TLRs), a type of PRR, can recognize a diverse range of ligands that share similar structural characteristics (Moresco et al., 2011). Because of this broad specificity of PRRs, the innate immune system can detect a diverse range of pathogens. However, it is also capable of discriminating between ligands from different bacteria. For example, activation of TLR4 by lipopolysaccharides (LPS) from Escherichia coli leads to potent inflammation and activation of adaptive immunity (Pulendran et al., 2001; Rallabhandi et al., 2008). However, signaling of the same TLR is negatively regulated by LPS derived from the oral bacterium, Porphyromonas gingivalis (Coats et al., 2005). How do the innate immune cells generate these highly specific responses using the seemingly non-specific PRRs?

Studies have shown that innate immune cells combine the function of two or more different types of receptors to increase their detection selectivity and accuracy, while still maintaining the broad specificity of individual type of PRRs (Ferwerda et al., 2008; Gantner et al., 2003; Inoue et al., 2011; Mukhopadhyay et al., 2004). The co-receptors act as context determinants; signals from combinatorial receptors are integrated to synergize or antagonize the immune responses through a process known as receptor crosstalk (Hill, 1998). This is analogous to the ‘two-signal'model of T cell activation, in which the primary signal from T cell receptors (TCRs) after antigen binding must be primed and augmented by the co-stimulatory signal from co-receptor CD28 to achieve a full activation (Acuto and Michel, 2003). In general, a plausible mechanistic explanation for the crosstalk between immune receptors is that the receptors assemble into signaling complexes on various length scales (Hartman and Groves, 2011). On the small multi-molecular scale, two or more receptors can oligomerize into homodimers and multimers, or associate with other proteins to form heterodimers and higher-order oligomers. This directly affects the specificity, selectivity and efficiency of receptor–ligand recognition (Inoue and Shinohara, 2014). Many innate immune receptors, especially TLRs, are known to form oligomers during signaling (Gay et al., 2014; Inoue and Shinohara, 2014). On the large length scale, typically from tens of nanometers to sub-micron lengths, a large number of receptor–ligand pairs can be sorted into distinct clusters or other spatial patterns that extend to microns in size in the cell plasma membrane. These receptor clusters provide a physical platform where the proximity, interactions and reactions between receptors and signaling proteins can be tightly regulated, like in B cells (Kuokkanen et al., 2015) and T cells (Fooksman et al., 2010; Goyette et al., 2019). This allows signals from individual receptors to be integrated and amplified. A few types of PRRs have been found to form large-scale receptor clusters (Aaron et al., 2012; Goodridge et al., 2011), but this is a phenomenon that is much less understood than the receptor oligomerization in innate immune cells. Furthermore, the functional significance of receptor clustering in innate immune functions and its underlying physical mechanisms remains poorly understood.

Membrane clusters are difficult to study for several reasons. The first challenge is to directly visualize the clustering of membrane receptors in living cells. The membrane clusters can be as small as a few nanometers in size, and their assembly is often transient and dynamic (Lenne et al., 2006; Sieber et al., 2007). Our understanding of receptor clusters has largely been propelled by the advancement of imaging microscopy techniques, such as super-resolution localization microscopy (Betzig et al., 2006; Dietz and Heilemann, 2019; Hess et al., 2006; Rust et al., 2006; Wen et al., 2020). The second challenge, after confirming the existence of receptor clusters, is to identify their roles in immune cell signaling. In this Review, we summarize the available literature on receptors that are known to form large-scale clusters in the plasma membrane of innate immune cells, including Fc receptors (FcRs), C-type lecithin-like receptors (CLRs) and TLRs. We discuss the mechanisms that may regulate the formation and spatial organization of such receptor complexes. Additionally, we review several up-and-coming methodologies that, when combined with advanced imaging techniques, are promising tools for revealing the inner workings of receptor clusters and the functional regulation of innate immune systems.

Representative innate immune receptors that form clusters

Fc receptors

FcRs recognize the Fc fragment of antibodies that are often found on invading pathogens, and when active, elicit antibody-dependent innate immunity (Bournazos et al., 2017; Halstead et al., 2010). Fc receptors are categorized based on the types of antibodies they bind to. One prominent phenomenon shared by different types of FcRs is that the receptors bind strongly to multivalent ligands, such as those on the surface of opsonized pathogens, which generate significantly stronger activation responses than that caused by soluble monomeric ligands (Bakema and van Egmond, 2011; Hansen et al., 2017; Kanamaru et al., 2007). The multivalent ligands crosslink receptors at the host cell–pathogen interface, leading to large-scale receptor clusters that can be visualized even with low-resolution microscopy (Menon et al., 1986b). This has been confirmed for Fc alpha receptor I (FcαRI; also known as FCAR) (Otten and van Egmond, 2004; Wines et al., 2001), Fc epsilon receptor I (FcεRI) (Carroll-Portillo et al., 2010), Fc gamma receptors I (FcγRI), and FcγRII (Kwiatkowska and Sobota, 2001).

FcεRI is a multimeric immune receptor that has high affinity for IgE and plays a critical role in IgE-mediated allergic reactions (Galli et al., 2008). Earlier studies have shown that adding IgE to rat basophilic leukemia cells and mast cells leads to reduced mobility of IgE receptors (Menon et al., 1986a). The IgE–FcεRI complexes can be further crosslinked by soluble multivalent antigen to form puncta on cell plasma membranes (Menon et al., 1986b). These studies provided early evidence that ligation of FcεRI has significant effects on its large-scale spatial distribution and diffusion behavior. Indeed, FcεRI was later shown to be randomly distributed on the plasma membrane of resting mast cells, but to reorganize into nanoclusters after multivalent antigen crosslinking, together with its downstream signaling proteins, including Lck/Yes-related novel tyrosine kinase (Lyn), spleen tyrosine kinase (Syk) and linker of activated T cells (LAT) (Veatch et al., 2012). Each nanocluster contains ∼100 FcεRIs and has an average radius of 70 nm, as revealed in a super-resolution imaging study (Shelby et al., 2013). Interestingly, the morphology of large-scale FcεRI clustering appears to depend on the ligand mobility. While immobile multivalent IgE leads to the formation of FcεRI nanoclusters in mast cells, mobile monovalent IgE tethered on a fluidic lipid bilayer triggers the coalescence of nanoclusters into large micron-sized clusters that accumulate at the central area of the contacting membrane (Carroll-Portillo et al., 2010). The morphology of the large-scale receptor clustering is similar to that of the immune synapse in T cells (Grakoui et al., 1999). The degranulation in mast cells, an indicator of cell activation, is stronger with immobile multivalent ligands than with mobile monovalent ones. Clearly, the spatial organization and signaling of FcεRI depend on the way its ligands are presented. However, the underlying mechanisms remain unclear.

FcγRs are receptors for the Fc region of IgG. FcγRs include FcγRIa (FCGR1A), FcγRIb (FCGR1B) and FcγRIc (FCGR1C), which have high affinity for monomeric IgG, and the low-affinity receptors FcγRIIa (FCGR2A), FcγRIIb (FCGR2B) and FcγRIIc (FCGR2CP) and FcγRIIIa (FCGR3A) and FcγRIIIb (FCGR3B) (Nimmerjahn and Ravetch, 2006). FcγRII was presumed to be evenly distributed as monomers in the membrane of resting cells, based on early results from immunoblotting and low-resolution immunofluorescence microscopy studies (Greenberg et al., 1993; Kwiatkowska and Sobota, 1999, 2001) and a single-molecule tracking analysis (Jaumouille et al., 2014). However, a super-resolution stochastic optical reconstruction microscopy (STORM) study recently revealed that FcγRI, FcγRII and the inhibitory receptor signal regulatory protein α (SIRPα; also known as SIRPA) all reside in separate nanoclusters of less than 100 nm in resting macrophage membrane (Lopes et al., 2017). After stimulation by IgG, the FcγRI and FcγRII nanoclusters separately reorganize into micron-sized concentric ring patterns. Another important observation from the study by Lopes et al. (2017) is that the signaling integration between FcγRI (positive) and SIRPα (negative) changes the spatial distribution of the receptor nanoclusters. FcγRI nanoclusters are in close proximity with SIRPα clusters in resting cells and remain unchanged after ligation of both FcγRI and SIRPα, but they become separated after stimulation of FcγRI alone (Lopes et al., 2017). The physical association or segregation between the stimulatory and inhibitory receptor nanoclusters appears to be an effective mechanism to regulate immune signaling. This was separately validated by Freeman et al. (2016). They showed that activated FcγR nanoclusters are spatially segregated from membrane-associated tyrosine phosphatase CD45 (also known as PTPRC) by an actin diffusion barrier (Fig. 1A). The receptor signaling is sustained as a result of this segregation. This study went further to elucidate the role of actin cytoskeleton in correlating the receptor clustering with their signaling consequences (Freeman et al., 2016).

Fig. 1.

Representative innate immune receptors that form clusters. (A) Fc γ receptors (FcγRs), after binding to immunoglobulin G (IgG), cluster with downstream signaling proteins. Inhibitory molecules, such as the tyrosine phosphatases CD45 and SIRPα, are excluded from the activated receptor clusters by integrin barriers. (B) Dectin-1 binds to β-glucans and forms a ‘phagocytic synapse’ at the host-cell–pathogen contact area. Inhibitory molecules, including CD45 and CD148, that can dephosphorylate the immunoreceptor tyrosine-based activation motif (ITAM) of Dectin-1, are excluded from activated Dectin-1 receptor clusters at the synapse. (C) TLR4 forms a heterodimer complex with MD2 in response to LPS. The heterodimer can further associate with CD14 and LBP, resulting in the formation of a TLR4–MD2–CD14–LBP complex.

C-type lectin receptors

CLRs are a type of PRR responsible for detecting the diverse range of carbohydrate structures on the cell wall of fungi (Chiffoleau, 2018; Dambuza and Brown, 2015). Dectin-1 (also known as CLEC7A) is one such CLR expressed mainly in dendritic cells and macrophages. Dectin-1 recognizes β-glucans on the fungal cell wall and transduces signals through its immunoreceptor tyrosine-based activation motif (ITAM)-like motif in the cytoplasmic domain (Schorey and Lawrence, 2008). During interaction with microbes, Dectin-1, like FcRs, concentrates in the host cell membrane at the contact site with microbes. Underhill and coworkers discovered in 2011 that Dectin-1 and other signaling molecules in macrophage cells spatially reorganize into large-scale patterns at the contact interface with yeast, forming a so-called ‘phagocytic synapse’ (Goodridge et al., 2011). Similar to the protein organization in T cell immune synapse (Grakoui et al., 1999), Dectin-1 accumulates in the center of the phagocytic synapse in macrophage cell membranes. Meanwhile, tyrosine phosphatases, including CD45 and CD148 (also known as PTPRJ), were excluded (Fig. 1B). Because CD45 and CD148 can dephosphorylate the activating ITAM motif of Dectin-1, the spatial segregation of Dectin-1 and phosphatases was proposed as a regulatory mechanism to sustain and amplify the phagocytic signaling. This study revealed for the first time that the large-scale reorganization of proteins at the host cell-pathogen interface is critical for innate immune responses (Goodridge et al., 2011).

In terms of interactions of Dectin-1 with other PRRs, Dectin-1 has been shown to cooperate with TLRs, particularly TLR2, in antimicrobial immune functions, as genetic knockout of either receptor leads to impaired immune responses (Gantner et al., 2003). This signaling synergy was long believed to be facilitated by the physical colocalization of Dectin-1 and TLR2 (Ferwerda et al., 2008; Inoue et al., 2011; Xu et al., 2009). However, by using high-resolution fluorescence imaging, we recently demonstrated that Dectin-1 and TLR2 actually reside in separate but adjacent nanoclusters in the plasma membrane of macrophage cells (Li et al., 2019). By physically separating the receptor nanoclusters on phagosome membranes, we showed that the cooperativity between Dectin-1 and TLR2 signaling requires the proximity, but not the colocalization, of the respective receptor clusters (Li et al., 2019). It therefore appears that the interactions between this receptor pair are confined to the overlapping interface between their nanoclusters. This is surprisingly similar to the observation that TCRs, co-receptor CD4 and the lymphocyte-specific protein tyrosine kinase Lck, reside in spatially separated nanoclusters even in activated T cells (Roh et al., 2015). In contrast to the perception that receptor nanoclusters colocalize to synergize, it is possible that receptors in separate nanoclusters from their signaling partners can better modulate their interaction with freely diffusing signaling molecules in the membrane and suppress nonspecific activation (Cebecauer et al., 2010; Roh et al., 2015).

Toll-like receptors

The TLR family is an essential group of PRRs expressed by professional phagocytes, such as monocytes and neutrophils (Hayashi et al., 2003; Hornung et al., 2002), as well as mammalian non-phagocyte cells (Cario et al., 2000). Ten different types of TLRs have been identified in humans. TLR1, TLR2, TLR4, TLR5, TLR6 and TLR10 are plasma-membrane receptors that mainly recognize pathogen-associated molecular patterns (PAMPs) from extracellular bacterial products, whereas the other four TLRs, TLR3, TLR7, TLR8 and TLR9, are mostly located on intracellular endosomes for recognizing internalized PAMPs, particularly DNAs and RNAs from pathogens (Kawasaki and Kawai, 2014). A few TLRs, such as TLR5, can detect ligands directly. However, most of them require heterodimerization with other TLRs or PAMP-binding co-receptors to achieve productive receptor–ligand recognition and downstream signaling.

Among all TLRs, TLR4 is the only one that has been shown to form both small-scale dimerization and large-scale clusters (Fig. 1C). TLR4 forms a heterodimer with co-receptor myeloid differentiation factor 2 (MD2; also known as LY96) to recognize LPS from Gram-negative bacteria (Brubaker et al., 2015; Nagai et al., 2002; Viriyakosol et al., 2001; Visintin et al., 2001). Upon LPS binding, the TLR4–MD2 heterodimer further dimerizes to bring their cytoplasmic Toll/interleukin-1 receptor (TIR) domains together to trigger pro-inflammatory responses (Kim et al., 2007; Ohto et al., 2007; Park et al., 2009). The TLR4–MD2 receptor complex can further assemble with LPS-binding protein (LBP) and CD14, leading to enhanced sensitivity of macrophages to LPS by 100- to 1000-fold (Kusunoki et al., 1995; Wright et al., 1990). CD14 is thought to also facilitate the complex formation by recruiting TLR4 and its co-receptors into lipid rafts (Plociennikowska et al., 2015).

Beyond the heterodimer complexes, TLR4 also assembles into nanoclusters, as revealed by a few super-resolution imaging studies. However, the reported size of nanoclusters and effect of receptor activation varies greatly. One study reported that TLR4 forms nanoclusters of ∼376 nm in the plasma membrane of resting murine macrophages and the nanocluster size increases to ∼515 nm after LPS activation (Aaron et al., 2012). In contrast, two other studies reported have TLR4 nanoclusters of 40–50 nm in the membrane of human glioblastoma cells (Zeuner et al., 2016) and primary human macrophages (Neumann et al., 2019). Both studies found that LPS stimulation has no effect on the nanocluster size. The discrepancy possibly stems from the intrinsic differences between murine and human cells. It may also reflect the different compositions of those nanoclusters in the three studies. Presumably, each TLR4 nanocluster consists of many TLR4–MD2 and TLR4–MD2–CD14–LBP complexes, but the exact composition may vary depending on the stimulation condition. More in-depth studies will be necessary to reveal the details and function of the TLR4 nanoclusters.

TLR2, unlike TLR4, requires the action of other TLRs for their activation by microbial ligands. TLR2 forms heterodimers with TLR1 or TLR6 to recognize a diverse range of microbial cell wall ligands, particularly lipopeptides and lipoteichoic acids (Oliveira-Nascimento et al., 2012). For instance, the TLR2–TLR6 heterodimer detects peptidoglycan on Gram-positive bacteria (Ozinsky et al., 2000), whereas the TLR2–TLR1 heterodimer recognizes mycobacterial lipoproteins and lipopeptides (Alexopoulou et al., 2002; Morr et al., 2002; Takeuchi et al., 2001, 2002). Despite the different ligands they bind to, both the TLR2–TLR6 and TLR2–TLR1 heterodimers signal through the same myeloid differentiation primary response 88 (MyD88)-mediated pathway and lead to similar pro-inflammatory responses (Farhat et al., 2008). This finding supports the notion that innate immune cells use the oligomerization and possibly clustering of receptors to diversify their pathogen recognitions, while achieving the same immune responses. Once formed, TLR2-associated heterodimers can further interact with non-TLR receptors, such as CD14, the scavenger receptor CD36 and chemokine receptor 4 (CXCR4) for synergistic immune responses (Inoue and Shinohara, 2014; Kawai and Akira, 2011; van Bergenhenegouwen et al., 2013). It has been reported that TLR2 exhibits confined diffusion after stimulation, providing indirect evidence that TLR2 and its co-receptors may be clustered into lipid rafts (Triantafilou et al., 2004). However, there is currently no direct evidence validating the large-scale clustering of TLR2 heterodimers.

Proposed mechanisms of clustering of innate immune receptors

Actin-mediated clustering of innate immune receptors

Actin is known to have a critical role in signaling of immune cells (Lagrue et al., 2013; Yu et al., 2013). In general, the actin cytoskeleton forms a dynamic mesh network that provides forces for cell polarization and motility, and facilitates the transports of intracellular organelles and molecules (Blanchoin et al., 2014). The actin cortex supports and scaffolds the shape of the cell plasma membrane in functions such as endocytosis and regulates the compartmentalization and molecular interactions in the membrane (Head et al., 2014; Koster and Mayor, 2016; Viola and Gupta, 2007). In innate immune cells, actin plays indispensable but distinct roles in the various stages of host cell–pathogen interactions, from initial receptor–ligand recognition (Kubelkova and Macela, 2019) and phagocytosis (Freeman and Grinstein, 2014; May and Machesky, 2001) to antigen presentation (Al-Alwan et al., 2003, 2001; Menager and Littman, 2016). We highlight here studies on actin in receptor clustering, but note that its multi-faceted roles are often interconnected.

Actin regulates the clustering and signaling of innate immune receptors through multiple possible mechanisms. One predominant mechanism is by altering receptor mobility (Fig. 2A). In this mechanism, the plasma membrane of resting immune cells can be described by the ‘picket-fence model’ (Kusumi et al., 2005). Transmembrane proteins that are anchored to the actin cortex act as rows of ‘pickets’ and effectively restrict the lateral diffusion of mobile proteins in the membrane. This view is supported by findings that the mobility of FcγRIIa in resting macrophages was restricted by the actin-anchored transmembrane protein CD44 (Freeman et al., 2018). In another study, single FcγR receptors (FcγRIIa, FcγRIIb and FcγRIII) in macrophages were shown to display different mobilities, some freely diffusive while others confined (Jaumouille et al., 2014). Actin disruption leads to more diffusive receptors and increased mobilities of the receptors, which is consistent with the ‘picket-fence” model. After initial receptor activation, actin remodels and debranches possibly through the synergistic action of coronins (F-actin-debranching proteins) (Kueh et al., 2008; Yan et al., 2005), as well as gelsolin and cofilins (F-actin severing proteins) (Bamburg and Bernstein, 2010; Nag et al., 2013). The actin debranching removes the ‘pickets’ to promote lateral movement of receptors (Freeman and Grinstein, 2014). The increased mobility of receptors presumably allows more frequent collisions between them, resulting in clustering and greater signaling. This may act as a signal amplification mechanism, which is particularly important for immune cell activation in response to low densities of extracellular stimuli.

Fig. 2.

Models of innate immune receptor clustering. (A) Actin-mediated receptor clustering during phagocytosis. In resting cells, the actin cortex confines the diffusion (orange arrows) of receptors (such as FcγRs) and actin-anchored transmembrane proteins. Upon the detection of pathogens, the actin cortex is debranched through the synergistic action of coronins (F-actin-debranching proteins), and gelsolin and cofilins (F-actin-severing proteins), leading to increased diffusion (orange arrows) and clustering of receptors. As more FcγR clusters are formed at the cell-pathogen interface, new actin filaments are assembled by the actin-nucleator Arp2/3 complex. Integrins form diffusion barriers that confine FcγRs clusters, while excluding phosphatase such as CD45 from the site of phagocytosis. (B) Lipid-raft-mediated TLR4 clustering. In resting cells, CD14 and CD36 reside in lipid-rafts (shown on the left). Upon ligand recognition, CD14 and CD36 bind to LPS and subsequently transfer LPS to TLR4. This leads to the coalescence of lipid rafts (shown in the middle) and clustering of CD14 and CD36 with TLR4 within the lipid rafts into an activated state (shown on the right).

For innate immune receptors, such as FcRs, that trigger phagocytosis, the role of actin extends beyond the initial clustering of receptors. As more phagocytic receptor clusters are formed at the host cell–pathogen interface, new actin filaments are assembled again, driven by the actin-nucleating activity of the Arp2/3 complex (Pollard, 2007). This allows the extension of pseudopodia that engulf the target pathogen. Indeed, the debranching and re-assembly of the actin cortex during FcεRI clustering has been observed in live mast cells using super-resolution microscopy (Colin-York et al., 2019). With the F-actin filaments also here acting as anchored membrane ‘pickets’, the receptor clusters become confined and eventually immobile to form the ‘phagocytic synapse’. Furthermore, integrins can also serve as an actin-anchored barrier that excludes membrane-associated tyrosine phosphatase CD45 from the host cell–pathogen contact site where activated FcγRs and Src kinases accumulate (Freeman et al., 2016). Such spatial segregation between activating and inhibitory proteins ensures that phagocytosis is successful.

Beyond altering receptor mobilities, actin likely plays additional roles in other aspects of receptor clustering. First of all, actin serves as a scaffold to maintain the interactions of receptors with their ligands and other signaling proteins, supported by the observation that FcγRs failed to bind to IgG-opsonized particles in macrophages, dendritic and microglial cells in the presence of pharmacological drugs that alter actin assembly or disassembly (Flannagan et al., 2010). Second, actin is required for forming and maintaining the ‘close-contact zone’ between a host cell and its target pathogen, where the innate immune receptors are concentrated and clustered (Freeman et al., 2016; Goodridge et al., 2011). In addition, actin may also modulate receptor clustering and signaling by affecting the mechanical properties of the cell membrane (Delanoe-Ayari et al., 2004; Faure et al., 2004; Le Roux et al., 2019; Liu et al., 2012).

The lipid raft hypothesis

The term ‘lipid raft’ initially referred to cholesterol-rich lipid microdomains in which lipid acyl chains are tightly packed (Simons and Ikonen, 1997). However, owing to the failed attempts to visualize lipid microdomains in cell membranes, the definition of ‘lipid raft’ has evolved to encompass nanoscale membrane domains that are enriched with cholesterol and saturated sphingolipids, are highly dynamic and involve both protein and lipid interactions (Kusumi et al., 2004; Pike, 2006). The existence of lipid rafts has been under intense debate due to conflicting evidence (Frisz et al., 2013; Gaus et al., 2003; Lai, 2003; Lee et al., 2015; Lingwood and Simons, 2010; Munro, 2003; Sevcsik et al., 2015; Shaw, 2006). Nevertheless, the lipid raft hypothesis is frequently cited as an explanation for receptor clustering in immune cell signaling (Koberlin et al., 2016; Ruysschaert and Lonez, 2015; Triantafilou et al., 2011).

In the lipid raft hypothesis, the detergent-insoluble lipid domains have different compositions than the rest of the plasma membrane. Such localized membrane compartments provide a favorable environment to recruit certain signaling proteins, such as GPI-anchored receptors and some Src family tyrosine kinases, while they simultaneously exclude phosphatases, such as CD45, that are generally detergent soluble (Foster et al., 2003; Harder et al., 1998; Simons and Toomre, 2000). This hypothesis therefore provides a convenient explanation for how positive and negative signaling proteins are spatially segregated in the cell plasma membrane to regulate their signaling outcome. IgE-bound FcεRI was first shown to reside in detergent-insoluble membrane domains, leading to the conclusion that it is recruited into lipid rafts during signaling (Field et al., 1995). Since then, a few more innate immune receptors, including FcαRI (Lang et al., 2002, 1999), FcγRIIa (Bournazos et al., 2009), Dectin-1 (Huang et al., 2015; Xu et al., 2009), C-type lectin-like receptor 2 (CLEC-2) (Pollitt et al., 2010), as well as TLR2 and TLR4 have all been shown to reside in lipid rafts, and validated based on their detergent resistance or association with cholesterol. Lipid rafts were considered to serve as docking sites for the clustering and signaling of the innate immune receptors. For TLRs, their recruitment into lipid rafts was thought to be facilitated by the GPI-anchored co-receptor CD14 and the scavenger receptor CD36, both of which are believed to partition in lipid rafts (Dorahy et al., 1996; Haziot et al., 1988; Pugin et al., 1998) (Fig. 2B). For example, CD14 has been shown to associate with TLR2 or TLR4 in lipid rafts during ligand binding, and may further form complexes with other receptors, such as with CD36, FcγRIII (CD16), CD81, heat-shock protein (Hsp) 70 and 90 family proteins, CXCR4 and growth differentiation factor 5 (GDF5) (Pfeiffer et al., 2001; Triantafilou et al., 2001, 2002). However, the association of CD14 with TLR2-TLR1 heterodimers has been called into question, as immunoprecipitation analysis showed no physical association between them (Nakata et al., 2006). This discrepancy highlights the controversial nature of methods used to confirm the existence of lipid rafts.

Nearly all studies mentioned above used one or more of the three traditional methods to confirm the existence of lipid rafts – the isolation of detergent-resistance membrane fractions (DRMs), cholesterol depletion by drugs such as methyl-β-cyclodextrin (Shogomori and Brown, 2003; Zidovetzki and Levitan, 2007) and/or direct visualization of lipid rafts using raft markers such as cholera toxin B (Kenworthy et al., 2000; Klymchenko and Kreder, 2014; Komura et al., 2016). All three methods are widely used but have serious caveats (Lichtenberg et al., 2005; Shogomori and Brown, 2003; Zidovetzki and Levitan, 2007). The solubility of proteins in detergents has been shown to vary depending on many experimental conditions, ranging from temperature and incubation time to detergent type (Schuck et al., 2003). The use of cholesterol depletion drugs, such as methyl-β-cyclodextrin, can also be problematic, because they not only extract cholesterol and lipids from cell membranes but also cause cell cytotoxicity that is unrelated to cholesterol depletion (Biswas et al., 2019; Kiss et al., 2010). Many lipid-raft markers such as cholera toxin B have been found to alter membrane organization (Day and Kenworthy, 2015). Recent studies using super-resolution microscopy have provided evidence supporting the existence of lipid-raft-like membrane domains (Eggeling et al., 2009; Owen et al., 2012a,b), but results have been conflicting (Frisz et al., 2013; Lee et al., 2015; Sevcsik et al., 2015). The super-resolution microscopic techniques have their unique complications (Huang et al., 2009; MacDonald et al., 2015; Yu et al., 2019). A more detailed discussion on the lipid raft hypothesis can be found elsewhere (Levental et al., 2020; Levental and Veatch, 2016; Sevcsik and Schutz, 2016; Sezgin et al., 2017).

Nanotechnology-based methods for investigating innate receptor clusters

Receptor clusters in plasma membranes of a cell are challenging to study. The first challenge is to actually prove the existence of receptor nanoclusters. For this, high-resolution microscopic techniques have been developed and applied to directly visualize receptor clusters and nanoscale membrane organization. This research area is blossoming and has been reviewed extensively (Aaron et al., 2012; Dietz and Heilemann, 2019; Sherman et al., 2013; Stone et al., 2017; van Zanten et al., 2010). The second, and possibly bigger challenge, is to demonstrate the role of receptor clustering in cell functions. Despite the importance of this research topic, it is much less explored than the imaging of receptor clusters.

To pinpoint the role of clustering and the spatial organization of receptors in the function of immune cells and other types of cells, a straightforward approach is to manipulate the receptor clusters and observe the corresponding changes in cell function, for instance, by stimulating cells with ordered patterns of ligands. The idea behind this approach is that the patterns of ligands reorganize the distribution of their receptors, providing a way to directly control the size or spacing of the receptor clusters. This approach has been used to reveal the role of receptor clustering in cancer cells (Biswas et al., 2018; Chen et al., 2018), T cells (Cai et al., 2018; Doh and Irvine, 2006; Matic et al., 2013; Mossman and Groves, 2007; Shen et al., 2008) and mast cells (Torres et al., 2011; Wakefield et al., 2017). These approaches have also been adopted for studying innate immune cells. For instance, micropatterned arrays of IgG (a ligand for FcγRs) have been used to study the mechanism of exclusion of phosphatase CD45 during FcγR-mediated phagocytosis (Fig. 3A) (Freeman et al., 2016). Here, the spatial organization relationship between individual receptor microclusters and the downstream signaling molecules could be directly observed because FcγRs were re-organized into well-ordered and distinct activation foci on the micropatterned ligand arrays. This study also revealed that integrins help sustain FcγR signaling during phagocytosis through two mechanisms; integrins form diffusional barriers that segregate phosphatase CD45 (inhibitory signals) from activated FcγRIIa and signaling proteins at the phagocytic synapse, and they help bridge neighboring phagocytic sites and engage more immobile FcγRs (Freeman et al., 2016). Planar micropatterned substrates have also been employed to study neutrophil migration (Henry et al., 2016). Here, adhesion molecules were presented in micropatterned islands of high ligand density, surrounded by regions that were largely devoid of ligands. Using this set up, it was found that motile neutrophils not only respond to local adhesive cues on the length scale of receptor clusters, but also integrate adhesion signaling across the entire contact interface (Henry et al., 2016).

Fig.

3. Nanotechnology-based methods for investigating innate receptor clusters. (A) Planar substrates with micropatterned IgG provide a platform to study the role of receptor spatial segregation at phagocytic synapse during FcγR-mediated phagocytosis. Integrins form diffusional barriers segregating CD45 from activated FcγR microclusters and, simultaneously, help bridge neighboring phagocytic sites to engage more activated FcγRs. (B) Nanopatterned arrays of Pam₃CSK₄ (a ligand for TLR2–TLR1) can be employed to identify the effect of TLR2–TLR1 nanocluster proximity on macrophage activation. (i) Illustration of the experimental set up, with a pseudo-colored scanning electron microscopy (SEM) image of a macrophage cell on the nanopatterned array (top right) and a fluorescence image (top left) showing the distribution of immunostained MyD88; this indicates activated TLR2-TLR1 nanoclusters in the macrophage cell on the nanopattern array. Scale bars: 5 μm. Images were taken by M.L. (ii) A decrease in the proximity of TLR2–TLR1 nanoclusters results in an increase in pro-inflammatory immune responses. Maximum immune responses are reached when TLR2–TLR1 nanoclusters become as close as possible based on their intrinsic spacing. (C) Particles displaying patterned ligands can be used to re-arrange the organization of Dectin-1 and TLR2–TLR1 on the phagosome membranes of macrophage cells. Particles with spatially mixed ligands (uniform particles) lead to a synergistic activation of Dectin-1 and TLR2 with enhanced immune responses. Particles with spatially segregated ligands (‘Janus particles’) no longer allow synergistic receptor signaling and result in a reduced immune response.

Compared to micropatterned substrates, nanoscale patterning of ligands is technically more challenging. Recently, we created nanopatterned arrays of Pam₃CSK₄ (a ligand for the TLR1–TLR2 heterodimer) to investigate how the spacing of TLR2–TLR1 receptor nanoclusters regulates macrophage activation (Li et al., 2020a) (Fig. 3B). We showed that macrophage pro-inflammatory responses, including secretion of tumor necrosis factor (TNF) and activation of nuclear factor κB (NF-κB), increase with decreased proximity of activated TLR2–TLR1 nanoclusters. There is, however, an intrinsic limit. When the TLR2–TLR1 clusters are adjacent to one another, the immune responses reach the maximum. The nanoclusters do not coalesce into larger ones, even with excess density of ligands (Li et al., 2020a). This study thus provides quantitative and direct evidence that the proximity of TLR2–TLR1 nanoclusters modulates immune responses in macrophages. It is possible to further decrease the size of the ligand ‘islands’ such that only a single or few ligands are presented on each gold nanoparticle on the patterned array. Such gold nanoparticle arrays have been used to study the effect of T cell receptor spacing on cell activation (Cai et al., 2018; Deeg et al., 2013; Guasch et al., 2018; Matic et al., 2013), but it has not been used yet for studying innate immune receptors.

While planar patterned substrates are useful for studying the clustering of receptors on the cell surface, they only allow the analysis of so-called ‘frustrated phagocytosis’, meaning phagocytic cells are spread on the ligand-coated planar substrate and attempt to engulf it without ultimate closure of the phagocytic cup (Takemura et al., 1986). It is desirable to engineer ligand patterns on particles that can serve as actual phagocytic targets, but creating patterns of proteins on micron- and nano-sized particles can be technically challenging. By developing a geometric fabrication strategy that patterns two types of ligands onto spherical microparticles, either homogeneously mixed or spatially segregated onto opposite sides of the particle, we were able to study the synergistic signaling between Dectin-1 and TLR2 clusters (Fig. 3C) (Li et al., 2019). To that end, we compared the proinflammatory responses, such as TNF secretion, activation of NFκB and reactive oxygen species (ROS) production, of macrophages stimulated by particles with either a uniform ligand distribution on their surface or those displaying spatially segregated ligands (termed ‘Janus’ particles). On the macrophage phagosomes encapsulating these engineered particles, the Dectin-1 and TLR2 nanoclusters in the phagosome membranes follow the patterns of their ligands on the particles; they are either spatially mixed on phagosomes encapsulating particles of uniformly mixed ligands, or segregated to the opposite sides of a phagosome containing a ‘Janus’ particle. This consequently changes the macrophage immune responses. When Dectin-1 and TLR2 are spatially mixed on the membrane of phagosomes, these two receptors synergistically trigger maximal anti-fungal immune responses. However, when they are spatially separated to be farther apart than 500 nm, their synergistic effect diminishes, leading to significantly reduced cell responses (Li et al., 2019). This approach thus allows the spatial manipulation of receptors within intracellular organelles in living systems and can also be used to study the clustering of other innate immune receptors, including those inside the cell.

Synthetic and biological polymers are also promising material platforms to manipulate receptor clustering, even though they have not been used thus far for studying innate immune receptors. One such example is synthetic stimuli-responsive polymers. These polymers change conformation in response to external environment, such as light. An optothermal-responsive polymer actuator has been developed to manipulate the clustering of TCRs (Liu et al., 2016). In this system, gold nanorods immobilized on a surface are coated with a thermo-responsive polymer. The other end of the polymer is functionalized with ligands for TCRs. When the gold nanorods are illuminated with near-infrared light, the induced localized heating causes the polymers to collapse and this conformational change generates forces in the piconewton range that pull on and cluster the ligand-bound TCRs. This approach provides a way to temporally control the clustering of immune receptors and the subsequent cell activation (Liu et al., 2016). Another example is DNA origami nanostructures (Rothemund, 2006), which can be assembled into almost any geometric shape (Hong et al., 2017). More importantly, ligands for receptors of interest can be conjugated over the DNA nanostructures at specific locations (Wamhoff et al., 2019). In an example of a DNA origami-based ‘nanocaliper’, ligands for the receptor EphA2 were positioned with precisely controlled spacing (Shaw et al., 2014). Using such an approach, the transcriptional responses of human glioblastoma cells were found to be affected by the spacing of ligands with maximal receptor activation at a ligand spacing of 40 nm (Verheyen et al., 2020). In another recent study, DNA nanostructures were used to study how nanoscale organization of antigens affects B cell receptor activation (Veneziano et al., 2020). Here, as few as five antigens, spaced ∼25–30 nm apart on the DNA origami, were found to be sufficient to trigger maximal B cell activation (Veneziano et al., 2020). The high precision in controlling the presentation of single ligands on DNA nanostructures renders them a highly promising approach for studying innate immune receptors.

Conclusions and perspectives

In this Review, we have summarized studies addressing the clustering and spatial organization of innate immune receptors. There is compelling evidence that some of these receptors form clusters during the recognition of pathogens, but such evidence is unclear for some others, particularly the TLRs. Moreover, the functional roles of those receptor clusters in regulating innate immune responses remain poorly understood and there are a number of important questions that warrant future study. First, the role of receptor clustering in the ability of innate immune cells to detect and respond to a diverse range of pathogens remains unclear. Another question relates to the mechanisms by which innate immune cells transduce information from spatially organized extracellular stimuli into the production of activation events at the cellular level. It is known that many of the cell wall components of microbes, which are ligands for innate immune receptors, are not uniformly distributed (Dague et al., 2008; Dufrene, 2014; Li et al., 2020b; Wheeler et al., 2011). Therefore, it is reasonable to speculate that the chemical and structural heterogeneity of the surface of a microbe directly impacts the spatial organization of innate immune receptors and subsequently the responses of the host cell. Future studies testing this hypothesis will be important for understanding the pathogenesis of microbial infections. Recent work has also demonstrated that innate immune receptors are promising drug targets for improving tumor immunotherapy (Feng et al., 2019; Wang et al., 2018). For example, co-activation of TLRs and FcRs can enhance the efficacy of antibody-based cancer immunotherapy by modulating the tumor microenvironment (Sharma et al., 2019). These emerging immunotherapy strategies highlight the importance of developing a comprehensive understanding of innate immune receptors and their collective interactions and functions in the regulation of signaling.