Structural Biology in Immunology: Structure/Function of Novel Molecules of Immunologic Importance

possible.

Chapter 1

Organization of Immunological Synapses and Kinapses

Marco Fritzsche*,†; Michael L. Dustin*,‡    * Kennedy Institute of Rheumatology, Oxford, United Kingdom,

† Weatherall Institute of Molecular Medicine, University of Oxford, Oxford, United Kingdom,

‡ New York University School of Medicine, New York, NY, United States

Abstract

Immunological synapses and kinapses describe two modes through which cells of the immune system exchange information based on specific recognition. This chapter focuses on T-cell immunological synapses, particularly, the structural aspects building from molecules that undergo stepwise assembly into the complex supramolecular activation clusters that captivated biology 20 years ago. Modern imaging methods have provided insights into the linkage between physical processes such as liquid-phase separation in adhesion, membrane, and signaling layers of the interface. These processes then link to the F-actin cortex and dramatic changes in membrane topology emerging from exocytic, endocytic, and extracellular vesicle budding mechanisms to shape and maintain this intricate communication interface.

Keywords

Synapse; Kinapse; Signaling; Activation; Phase separation; Actin; Cortex; Extracellular vesicles; Exosomes; Ectosomes; Exocytosis

Abbreviations

Ag 

antigen

Arp 

actin-related protein

CD(#) 

cluster of differentiation (1,2,3…)

CRTAM 

class-I MHC-restricted T-cell-associated molecule

CSK 

C-terminal Src kinase

CTLA-4 

cytotoxic T lymphocyte antigen-4

DC 

dendritic cell

ICAM(#) 

intercellular adhesion molecule-(1,2,3…)

ICOS(L) 

inducible T-cell costimulator (ligand)

ITAM 

immunotyrosine activation motif

LAT 

linker of activated T cells

LFA-(#) 

lymphocyte function associated-(1,2,3)

(p)MHC 

(peptide) major histocompatibility complex

PD-1 

programmed cell death-1

SH(#) 

Src homology domain (1,2,3)

SIM 

structured illumination microscopy

SLAM 

signaling lymphocyte activation molecule

SLB 

supported lipid bilayer

(c,p,d)SMAC 

(central, peripheral, distal) supramolecular activation cluster

STED 

stimulated excitation depletion

TCR 

T-cell antigen receptor

(e)TIRF(M) 

(extended) total internal reflection fluorescence (microscopy)

VASP 

vasodilator-stimulated phosphoprotein

WASp 

Wiscott Aldrich syndrome protein

ZAP-70 

TCR zeta-associated protein of 70 kili-Daltons

Acknowledgments

We thank our groups for stimulating data and discussions. MLD was supported by Wellcome Trust and Kennedy Trust for Rheumatology Research (PRF 100262Z/12/Z).

1.1 Introduction

Immunological synapses are stable adhesive junctions formed by immune cells for priming of immune responses and effector function.¹ The archetype is based on radially symmetric junctions formed by helper or cytotoxic T cells, with B cells as antigen-presenting cells in which concentric zones referred to as supramolecular activation clusters (SMACs) can be resolved by wide-field microscopy in vitro and in vivo.²,³ The most important function of the immunological synapses is directed secretion of regulatory signals in the form of proteins and vesicles between immune cells, and also killing of pathogens in the context of innate immune recognition (,⁴–⁷ p. 5140). Immunological synapses are formed by T cells, innate lymphoid cells, B cells, mast cells, neutrophils, macrophages (phagocytic synapse), and likely others.⁵,⁷–¹⁰ The key common denominator is that specificity is driven by receptors that couple to nonreceptor tyrosine kinases, often through immunotyrosine activation motifs in the cytoplasmic domain.¹¹ The T-cell antigen receptor (TCR) is the relevant receptor for T cells,¹² which will be the major focus of this chapter.

The formation of the archetypal T-cell immunological synapse can also be reconstituted with supported lipid bilayers (SLBs) containing major histocompatibility complex (MHC) proteins with bound agonist peptides (pAgMHC) and the adhesion molecule intercellular adhesion molecule-1 (ICAM-1, CD54) in which SMACs form in a time-dependent manner.¹⁰ The two major SMACs are the central SMAC (cSMAC), which is enriched in T-cell receptors (TCRs), pAgMHC, and some costimulatory receptors such as CD28, and the peripheral SMAC (pSMAC) that is enriched in the integrin lymphocyte function associated 1 (LFA-1, CD11a) and ICAM-1² (Fig. 1.1A). Subsequently, Freiberg et al. defined the distal SMAC (dSMAC) as a ring outside the pSMAC enriched in CD45 and dynamic filamentous actin (F-actin).¹³,¹⁴ Some aspects of immunological synapse formation can also be reconstituted with glass substrates coated with anti-CD3 antibodies to trigger TCR signals and other surface treatments, such as poly-l-lysine or serum attachment factors (fibronectin, vitronectin, etc.) that provide an adhesion component.¹⁵,¹⁶ In the absence of this adhesion component, anti-CD3 elicits actin-rich protrusions that push the cell away from the anti-CD3-coated surface.¹⁷,¹⁸ We will incorporate data from cell-cell systems and model substrates to generate a current picture and trajectory for future questions.

Fig. 1.1 Supramolecular activation clusters in synapse and kinapse. (A) Schematic view of the Kupferian SMACs as originally defined in immunological synapse and remapped to kinapse. Note that the cSMAC, pSMAC, and dSMAC (defined in text) are related to structures in kinapse by symmetry breaking. Based on Sims et al. ¹³ the activity of PKC-θ favors kinapse whereas the activity of WASp favors synapse. (B) eTIRF-SIM image of F-actin network in Jurkat T cell. The regions of the rosette-like network that correspond to the SMACs are indicated. See text for further discussion of the limitations of these assignments.

Stability of the immunological synapse depends on its symmetry. Breaking of the symmetry of the immunological synapse leads to a migratory junction defined as a kinapse, which appears to be an important mode of T-cell interaction with, for example, dendritic cells (DCs) and germinal center B cells in vivo¹⁰,¹³,¹⁹–²³ (Fig. 1.1B). Kinapse formation may lead to multifocal synapses observed in T-DC interactions in vitro and in vivo.²⁴,²⁵ The immunological synapse and kinapse share signaling elements, referred to as microclusters, which are formed by both the TCR and LFA-1.²⁶–²⁹ While the formation of microclusters likely has some aspects driven by the biophysics of receptor- ligand-driven self-assembly in the interface, most aspects of immunological synapse and kinapse formation and maturation are dependent upon intact F-actin cytoskeleton and later stages of synapse maturation are regulated by microtubules.³⁰–³² The chapter will build a structural model for the immunological synapse beginning with some major classes of receptors and then moving to the underlying cytoskeletal networks and important issues of membrane topology.

1.2 Receptors and Ligands of the Immunological Synapse and Kinapse

1.2.1 T-cell Receptor and pAgMHC

There are two known TCR heterodimers—αβ and γδ—that are expressed on distinct subsets of T cells.¹² The classical adaptive T-cell subsets use the αβ TCR, whereas subsets of tissue-resident T cells use the γδ TCR.³³ Most γδ T cells and a small subset of αβ T cells are innate like, meaning that they use stereotypical rearrangements to generate largely invariant receptors for conserved ligands that are often presented on nonclassical MHC-like proteins. For example, the so-called natural killer T cells use invariant αβ TCR to recognize glycolipids bound to CD1d, which has a structure similar to MHC class I, but possesses a binding groove adapted for the recognition of glycolipids.³⁴ Mucosal-associated invariant T cells recognize microbial metabolites presented on MHC-related protein.³⁵ This chapter focuses on the majority of αβ T cells that use MHC class I or class II proteins as ligands in combinations with short peptides. Each of these αβ T cells expresses one TCR that binds weakly to self-peptide (pself)-MHC based on thymic selection and has the potential to bind a variety of pAgMHC.³⁶ Since each TCR is generated through multiple recombination events involving germline V, (D), and J segments with random addition of base pairs during joining, there is huge diversity of these weakly self-reactive TCR in the naïve repertoire that are selected for expansion and conversion into memory T cells over the life of an individual. The interaction of TCR with pself–MHC is challenging to measure physically using the available methods, but in solution have dissociation constants on the order of 100 μM to > 1 mM,³⁷,³⁸ yet these interactions are essential for thymic selection during development and appear to influence mature T cells continuously.³⁹ The interactions of TCR with pAgMHC typically have Kd in the range of 100 nM–10 μM.³⁸ All individuals seem to harbor some higher affinity pself-MHC-specific T cells that often take on suppressive roles as regulatory T cells (Treg),⁴⁰ but can also be recruited into anti-pathogen responses and contribute to autoimmune diseases.⁴¹,⁴²

The description of the interaction of TCR with pMHC through solution affinity is an imperfect predictor of function,⁴³ as the receptor system exclusively operates in an interface between two biological membranes. This introduces two key features. First, in order for the TCR to interact with pMHC, the two membranes need to be separated by < 15 nm, with measurements from electron microscopy suggesting ~ 13 nm as the average intermembrane distance generated by large numbers of these interactions.⁴⁴ The other aspect is that due to the cytoskeletal dynamics that we will describe, the interactions are subjected to physical forces.¹⁰,¹⁵,⁴⁵,⁴⁶ There are two well-described bond types for force-bearing interactions in biology—slip bonds in which the off-rate increases with force in the pN range, and catch bonds, where the bond off-rate decreases at an optimal force.⁴⁷ It has been shown by two independent methods, biomolecular force probe and laser trap, that TCR-pAgMHC interactions are catch bonds with a force optimum around 10 pN.⁴⁸,⁴⁹ This force requirement is a key feature that distinguishes the immature T cell selecting pself-MHC interactions, which form slip binds, from the mature T-cell-activating pAgMHC, which form catch bonds.⁴⁹ However, recent studies suggest that the pre-TCR, expressed by thymocytes following rearrangement of the TCR β subunit, can also form catch bonds with pself-MHC complexes.⁵⁰ The structural features that distinguish the slip and catch bonds are not known, but laser trap measurements suggest that the catch bond undergoes a conformational transition that extends the complex by 10 nm, which is 66% of the apparent length of the unstrained complex.⁴⁸ This suggests significant unfolding of β-sheet structures as part of the force-dependent transition. There is some evidence that the tangential orientation of force relative to the T-cell surface is more effective than normal orientation.⁵¹,⁵² Possible explanations for this are discussed in the context of functions of TCR microclusters.

1.2.2 Adhesion Molecules

T cells are sensitive to a single pMHC and this can be accomplished in part because of the activity of adhesion molecules.⁵³,⁵⁴ We will define adhesion molecules as surface receptors that are present in high copy number (on the order of 100,000 molecules) on both the T cell and the antigen-presenting cell. This high expression level allows adhesion molecules to contribute to the generation of the close contacts needed for TCR to reach the pMHC on the surface of the APC. There are two general adhesion molecules in human T cells that are extensively utilized in cell-cell interactions: the integrin family member LFA-1 and the signaling lymphocyte activation molecule (SLAM) family member CD2.⁵³ Other integrins and other SLAM family members participate in different settings, but we will focus on these as the best-studied paradigms.

Integrins are large cell surface heterodimers that interact with ligands in a divalent cation and in an energy-dependent manner. The large extracellular domains have an N-terminal globular domain made up of large (α) and small (β) subunits that bind to ligand with high or low affinity depending upon the conformation.⁵⁵ The globular domain is connected to the two transmembrane domains by 15 nm stalks. The small cytoplasmic domains interact with a number of regulatory factors, most importantly the F-actin linker talin and the regulatory protein kindlin 3.⁵⁶ Genetic deficiency of the β subunit of LFA-1 (referred to as β2 or CD18) or kindling 3 leads to leukocyte adhesion deficiency diseases (types I and III, respectively) that particularly impact specific innate cellular functions involved in protection from bacterial infection, and quantitatively impair aspects of adaptive immunity.⁵⁶,⁵⁷ Antibodies against LFA-1 were used in the treatment of autoimmune diseases including psoriasis until adverse events including fatal viral infections forced their removal from the market.⁵⁸,⁵⁹ This experience underscores the great importance of adhesion molecules in many aspects of the immune response, including the homing of cells through interactions with endothelial cells, such that global inhibition of a major adhesion molecule has significant risks. LFA-1 is well characterized structurally with crystal structures of its ligand-binding domain without and with the relevant immunoglobulin (Ig)-like domain from the ligand ICAM-1.⁶⁰ There is also extensive information about LFA-1 conformations from electron microscopy studies.⁶¹ Like other integrins, LFA-1 exists in low affinity form, where the stalks are bent to bring the ligand-binding domain very close to the T-cell membrane. Activation of the molecules by separation of the cytoplasmic domains results in extension so that the ligand-binding site projects to ~ 20 nm from the membrane, and together with the ~ 14 nm length of ICAM-1 (which has an L-shaped molecule with 5 Ig-like domains), the LFA-1-ICAM-1 interaction could span up to 34 nm between membranes.⁶² However, like the TCR, LFA-1 may subjected to mechanical forces that are tangential to the T-cell membrane during synapse or kinapse formation, such that the actual intermembrane distance may be smaller due to the lateral pulling against ICAM-1 that is anchored to the cytoskeleton.⁶³ In addition to ICAM-1, human LFA-1 binding to several other ligands, the most important of which is the 2 Ig-like domain ICAM-2.⁶⁴ It is not clear if interaction with this shorter ligand generates different intermembrane spacing or clustering of LFA-1. Generally, TCR-pMHC interactions segregate laterally from LFA-1-ICAM-1 interaction. The precise mechanism by which LFA-1-ICAM-1 interactions promote TCR-pMHC interactions is not entirely clear, but it may involve generating force-resistant adhesion against which the cell can push TCR-rich projections against the substrate (see the section on scanning interactions).¹⁰,⁶⁵,⁶⁶

CD2 continuously diffuses into the T-cell membrane and binds to CD58 (T11, LFA-2, sheep erythrocyte receptor), a ligand expressed on all cells, even erythrocytes.⁶⁷ CD2 and CD58 are both members of the SLAM subfamily of the Ig-superfamily, which resides in a cluster on chromosome 1 in the mouse and human. CD2 and CD58 bind to each other with a Kd of ~ 1 μM and the interaction dissociates very rapidly with a koff of > 7 s− 1.⁶⁸ A crystal structure exists for human CD2 interacting with CD58, which provides a detailed structural basis of its adhesive function.⁶⁹ In rodents, the major CD2 ligand is CD48, which also exists in humans as a distinct gene from CD58. Mice lack a homologue of CD58. In both mouse and human, CD48 is a ligand for the 2B4 receptor (CD244), an activating receptor highly expressed on natural killer cells, but also activated T cells.⁷⁰ Thus, in rodents CD2 and 2B4 may compete for CD48 and the affinity is mouse CD2 for the mouse CD48 is 10-fold lower than for human CD2 and CD58.⁷¹ Nonetheless, mouse CD2 contributes to T-cell activation in a nonredundant manner.⁵³,⁷² The distinct contributions of CD2 likely arise from its distinct mode of adhesion and from its highly conserved cytoplasmic domain. In contrast to the situation with integrins, adhesion mediated by CD2 interactions with CD58 is neither divalent cation nor energy dependent. Like the TCR-pMHC, it spans an apparent distance between membranes of ~ 15 nm with a measured intermembrane distance of 13 nm.⁷³,⁷⁴ The CD2-CD58 interaction is the first adhesion system for which a 2D affinity, as defined by Bell,⁷⁵ was measured,⁷⁶ partly because its mode of adhesion is highly amenable to this type of analysis. The 2D Kd is ~ 1 μm− 2 indicating a highly efficient assembly process that positions the binding sites in a very small volume, with an axial dimension of ~ 5 nm, and a lateral dimension of micrometers.⁷⁷ An immunological synapse has an interaction area of ~ 50 μm². This means that the ~ 10,000 CD2 and CD58 molecules in the interface when the cells come into contact are at an effective concentration of 66 μM, which exceeds the three-dimensional (3D) Kd for mouse or human CD2 for its ligand.⁷¹ While there are some barriers to forming such close contacts between cells,⁷⁸ once the close contact is nucleated this adhesion system is very efficient, but at the same time very dynamic, such that the CD2-CD58 interaction turns over rapidly and does not impede cell migration. As CD2-CD58 interactions span the same distance at the TCR-pMHC, it is not surprising that the presence of CD2-ligand interactions increases the sensitivity of TCR for pAgMHC.⁷² In fact, increasing the size of CD48 to produce a mismatch with the TCR-pMHC dimensions profoundly inhibits T-cell detection of pMHC on APCs expressing the extended CD48 molecules.⁷⁹ There are autoimmune disease risk alleles associated with CD2 and CD58, and recently, a CD2 signaling signature was associated with increased severity of autoimmune diseases.⁸⁰ CD2 has a long cytoplasmic domain with five polyproline motifs that interact with SH3 domains of Fyn and an adapter protein referred to as CD2-associated protein.⁸¹ Under specific cross-linking conditions CD2 ligation alone can be strongly mitogenic for T cells,⁸² but there is no known physiological signaling mode that fully exploits this strong, antigen-independent activation.⁸³ The CD2 cytoplasmic domain lacks any tyrosine residues and thus may utilize the regulatory mechanisms based on tyrosine phosphatase recruitment.

The SLAM family contains a number of other receptors most of which are homophilic (self-binding) and have cytoplasmic domains that bind to SH2D1A and B, which are single SH2 domain adapters that interact with the Src family kinase Fyn.⁸⁴ SH2D1A mutations result in X-link lymphoproliferative disease 1, which is characterized by defects in the regulation of myeloid cells and hypogammaglobulinemia due to defects in T-cell help for B cells.⁸⁵ In this situation, the tyrosine in the cytoplasmic domain of most SLAM family members (excluding CD2, 2B4, and CD48) recruits the tyrosine phosphatase SHP2 leading to a dominant inhibitory effect.⁸⁶ The normal function of SLAM family members is in adhesion, likely in conjunction with CD2-CD58/CD48 and LFA-1-ICAM-1 interactions. Another nonredundant function of SLAM family members and SAP is in the selection of NKT cells in the thymus.⁸⁷ CD1d is expressed on other thymocytes and positive selection and expansion of NKT cells is mediated by thymocyte-thymocyte interactions that require SLAM family members and the Fyn kinase that is recruited to engage SLAM receptors by SH2D1A. The relevant SLAM family members for T-B and NKT selection include CD84, CD150, CD229, and CD352.⁸⁸ Crystal structures for CD84, CD150, and CD229 reveal homophilic interactions with overall similar to the CD2-CD58 heterophilic interactions that would likely operate with a similar intermembrane spacing to TCR-pMHC. In general, the SLAM family members play an important role in lymphocyte-lymphocyte