CD28 co-signaling in the adaptive immune response

Abstract

T-cell proliferation and function depends on signals from the antigen-receptor complex (TCR/CD3) and by various co-receptors such as CD28 and CTLA-4. The balance of positive and negative signals determines the outcome of the T-cell response to foreign and self-antigen. CD28 is a prominent co-receptor in naïve and memory T-cell responses. Its blockade has been exploited clinically to dampen T-cell responses to self-antigen. Current evidence shows that CD28 both potentiates TCR signaling and engages a unique array of mediators (PI3K, Grb2, FLNa) in the regulation of aspects of T-cell signaling including the transcription factor NFkB. In this mini-review, we provide an up-to-date overview of our understanding of the signaling mechanisms that underlie CD28 function and its potential application to the modulation of reactivity to autoimmunity.

Introduction

T-cell activation depends on the presentation of peptide antigen to the antigen receptor complex (TCR) by antigen-presenting cells (APCs) together with additional signals from co-receptors. The ‘two-signal hypothesis’ as first outlined by Bretscher and Cohn¹ proposed regulatory control of the adaptive immune response by two independent receptor mediated signals. From its inception, this notion introduced the possibility of therapeutic intervention at the level of the second signal without interference with an antigen-receptor (TCR; TCR/CD3) signal. One could dampen the co-receptor signal without needing to know the exact nature (i.e., affinity) of the antigen involved in the ‘step 1’ of T-cell activation cascade.

This two signal model was further developed in the 1970s by Lafferty² and Jenkins and Schwartz.³ They showed that the ligation of the TCR complex alone was generally insufficient to induce a T-cell response, and, in many instances, resulted in non-responsiveness or T-cell anergy. The response of most naïve T cells to MHC-peptide fails to occur without an additional co-signal. This underscores the importance of the antigen presenting cell (APC) in T-cell activation with their presentation of ligands such as CD80 and CD86. Dendritic cells (DCs), macrophages and B-cells express co-receptor ligands that are absent on stromal and epithelial cells. DCs express the highest levels of CD80/86 amongst presenting cells. Antigen presentation is context dependent with different ligands being expressed in niches of the peripheral immune system.

Co-stimuli are potent modulators of protein synthesis, metabolism, cell cycle progression, apoptosis and differentiation in T cells. Conversely, inhibitory co-stimuli can prevent the onset or downregulate immune reactions (reviewed in ref. 4–7). The co-stimulatory co-receptors belong to several protein families: immunoglobulin family [CD28 and inducible co-stimulator (ICOS)], tumor necrosis factor receptor family [for example CD27, death receptor 3 (DR3), OX40, 4-1BB], integrins (LFA-1), cytokine receptors (for example IL-2R or IL-4R) and G-protein coupled receptors for chemokines (CCR1, CCR7, CXCR3 and CXCR4) (Fig. 1).

Figure 1.

Co-stimulatory receptors. Activation of T cells depends on two signals. The first is provided by the T-cell receptor upon recognition of a specific peptide in complex with MHC class II molecule on an antigen presenting cell. The second signal is delivered by additional counter-receptors expressed on mature APCs or by soluble factors like cytokines and chemokines. These second signals, called co-stimuli, are per se not able to trigger proliferation and differentiation of T cells. Instead, these co-stimuli are potent modulators of protein synthesis, metabolism, cell cycle progression, apoptosis and differentiation. The co-stimulatory co-receptors belong to several protein families: immunoglobulin family (CD28 and ICOS), tumor necrosis factor receptor family (like CD27, DR3, OX40 and 4-1BB), integrins (LFA-1), cytokine receptors (like IL-2R or IL-4R (not depicted)) and chemokine receptors (CCR1, CCR7, CXCR3 (not depicted) and CXCR4). APC, antigen presenting cell; CCL19, EBI-1-ligand chemokine (ELC), macrophage inflammatory protein 3β (MIP-3β), exodus-3; CCL21, chemokine with 6 cysteines (6Ckine), secondary lymphoid tissue chemokine (SLC), exodus-2; CCL5, regulated upon activation normal T-cell expressed sequence (RANTES); CXCL12, stromal cell-derived factor 1α (SDF-1α); DR3, death receptor 3; ICAM-1, intercellular adhesion molecule 1; ICOS, inducible T-cell co-stimulator; ICOSL, ICOS ligand; IL-2, interleukin 2; IL-2R, interleukin 2 receptor; LFA-1, lymphocyte function-associated antigen 1; TNF, tumor necrosis factor.

The CD28 Co-Receptor

The best established co-stimulatory pairs are CD28 and its binding partners CD80 and CD86. CD28 is constitutively expressed on naïve and activated CD4 and CD8 positive T cells,^8,9 while CD80 and CD86 are induced on DCs with their activation.^10–13 CD28 was first identified in the 1980s as a co-receptor that enhanced TCR-induced proliferation and promoted the differentiation of naïve CD4⁺ T cells.^14,15 It encodes a 44 kDa type I transmembrane glycoprotein that homodimerises due to disulphide bonds between cysteines juxta-positioned in the transmembrane region.¹⁶ Cd28 gene has been mapped to human chromosome 2q33 and mouse chromosome 1.^17,18

The in vivo relevance of CD28 was evident with the generation of CD28 deficient (i.e., Cd28^-/-) mice.¹⁹ These mice are immune compromised, showing reduced T-cell responses to antigen, defective germinal center formation and T-cell differentiation. Anti-CD3 responses are reduced by 60–70 percent. CD28 also preferentially promotes T_H2 differentiation,^20,21 providing help for B-cells with germinal center formation and isotype switching.^19,22,23 In addition, the co-receptor prevents anergy by modulating cell cycle progression and reduces cell death or apoptosis due to the increased expression of anti-apoptotic proteins such as Bcl-2 and Bcl-X_L.^24,25 CD8 T-cell cytolytic responses to viral infection are also reduced due to impaired T-cell help, although this dependency can be over-ridden with a prolonged presence of virus during infection (i.e., repeated antigenic stimulation).^19,26

Overall, CD28 has been implicated in a myriad of functions from altered metabolism to cytokine production and the activation of transcription factors such as NFκB (nuclear factor kappa-light-chain-enhancer of activated B cells) (Fig. 2). Both quantitative and qualitative effects have been noted. In this manner, CD28 progresses the propagation of the adaptive immune response beyond the initial contact of naïve T cells with the immunogenic peptide.

Figure 2.

The major roles of CD28 co-stimulation. CD28 stimulation modulates proliferation, differentiation and effecter functions in T cells by regulation of gene transcription, cell cycle progression, survival, secretion of cytokines and chemokines, metabolism and motility.

CD28 Ligands CD80 and CD86

The extracellular portion of CD28 contains a single Ig-V-like domain incorporating a MYPPPY motif.²⁷ This motif provides the structural specificity for interactions with CD80 and CD86 (B7-1 and B7-2, respectively) with affinities of 4 µM and 20 µM, respectively.^28,29 CD80 and CD86 have overlapping functions, although CD86 is expressed later than CD80 upon DC activation.^11,13,30 The higher affinity of related CTLA-4 for CD80/CD86 has in turn allowed the use of a CTLA-4-Ig fusion protein to out-compete CD28-CD80/86 binding in the treatment of autoimmune disorders.^31–33 CTLA-4-Ig with the commercial name ‘Abatacept’ has been approved for the treatment of rheumatoid arthritis,³⁴ while a modified version termed ‘belatacept’, which binds to CD86 with preferred affinity, may be more effective in the treatment of transplant rejection.³⁵

It is also noteworthy that Freeman and co-workers have identified an additional interaction between CD80 and another co-receptor, programmed death-1 ligand 1 (PD-L1).³⁶ Indeed, human CD80 interacts with human PD-L1 with greater affinity than with CD28. PD-L1 may therefore out-compete CD28 for CD80 on APCs and indirectly influence co-stimulation. PD-L1 itself regulates tolerance, chronic infection and tumor immunity, while its binding partner PD-1 is highly expressed on “exhausted” virus-specific T cells, which have suppressed cytokine and proliferative responses to their antigen.³⁷

Regulation of CD28 Surface Expression

Unlike other immunoglobulin family members, whose synthesis and surface expression is regulated by cell activation, CD28 is constitutively expressed on naïve as well as activated CD4⁺ T cells.^8,9 Understanding the processes, which determine levels of CD28 expression, is important since the mechanisms ultimately affect co-stimulation. Expression is influenced by the rate of protein synthesis, longevity on the cell surface as well the mechanism that removes receptor from the cell surface. Cefai et al. first showed that CD28 endocytosis is regulated by the binding of phosphatidylinositol 3-kinase (PI3K) to a cytoplasmic YMNM motif.³⁸ PI3K-interacting CD28 is preferentially internalized in a clathrin-dependent manner³⁸ (Fig. 3). Mutation of the Y residue prevented CD28 endocytosis, while other mutations with a lesser effect on PI3K binding (i.e., YXXM) had a correspondingly lesser effect on internalization. The Siminovitch lab then extended this finding by showing that the p85 subunit of PI3K binds to phagocyte oxidase homology (PX) domain of sorting nexin 9 (SNX9).³⁹ SNX9 in turn interacts with Wiskott-Aldrich syndrome protein (WASP) and as such associates with clathrin coated vesicles and microtubule cytoskeleton.³⁹ Overexpression of wild-type SNX9 enhances CD28 internalization and mutated SNX9 with impaired ability to interact with p85 subunit reduces CD28 internalization.³⁹ Overall, this work has identified a series of events involving p56^lck activation and its phosphorylation of the YMNM motif leading to PI3K binding followed by its association of SNX9 and WASP in the endocytosis of the co-receptor (Fig. 3).

Figure 3.

Regulation of CD28 internalization. The internalization of CD28 is regulated by a discrete series of events in T cells. (1) CD28 is constitutively expressed on both naïve and activated CD4⁺ T cells. (2) Within seconds after CD28 cross-linking, src kinases p56^lck or p59^fyn (not depicted) phosphorylates the YMNM motif within the intracellular portion of CD28. (3) Phosphorylated YMNM motif binds the SH2-domain of p85 regulatory subunit of PI3K. (4) p85 subunit of PI3K interacts with PX domain of SNX9, which in turn (5) interacts with clathrin associated WASP resulting in the clathrin coated vesicles mediated endocytosis. Lck, lymphocyte-specific protein tyrosine kinase; p110, catalytic subunit of PI3K; p85, regulatory subunit of PI3K; PI3K, phosphatidylinositol 3-kinase; PX, phagocyte oxidase homology domain; SH2, src homology 2 domain; SNX9, sorting nexin 9; WASP, Wiskott-Aldrich syndrome protein.

CD28 Co-Signaling Pathways

The CD28 co-receptor is a discrete signaling unit in keeping with the ‘two-signal model’. However, current data also reveals that the co-receptor does not generate a single signal, but rather is linked to multiple pathways in T cells. The ‘two-signal model’ is actually a ‘multi-signal model’. CD28 binds to multiple signaling proteins that interact with multiple downstream pathways. In one example, CD28 binding and activation of PI3K activate numerous signaling proteins that carry pleckstrin homology (PH) domains (i.e., 60+ proteins). The co-receptor also amplifies TCR signaling, possibly by enhancing adhesion due to CD80/86 binding (i.e., nM range of K_d), or due to the generation of signals that add to or complement TCR signaling. In this vein, it is important to note that CD28 binds with constant affinity to its ligands CD80 and CD86. This contrasts with the TCR complex that binds to a range of different peptides with varying affinity. CD28 therefore provides a constant signal in the context of TCR signals of differing strength. Overall, the co-receptor acts to lower the threshold for TCR activation of naïve CD4⁺ T cells allowing a response to low-affinity peptides. This response would not normally occur in the absence of co-stimulation. By contrast, high-affinity peptides or chronic antigen stimulation trigger activation without CD28 co-stimulation.^40–42 CD28 skewing of the immune response to encompass lower affinity peptide interactions modifies the limits placed on immune reactivity via thymic selection.⁴³

CD28-YMNM and PI3K

The CD28 intracellular domain comprises of 41 amino acid residues that lack intrinsic catalytic activity. Instead, it possesses various binding sites for interactions with intracellular signaling proteins (Fig. 4). Key to this recruitment are tyrosine and proline-based motifs that bind to src homology 2 and 3 (SH2 and SH3) domains, respectively. The first prominent motif is the YMNM sequence that when tyrosine phosphorylated binds to the SH2 domain of PI3K^44–47 and growth factor receptor-bound protein 2 (Grb2)^48,49 (Fig. 4).

Figure 4.

Structure of the cytoplasmic tail of CD28. The 41 amino acid residues of the intracellular tail of CD28 lack intrinsic catalytic activity; however, they create binding sites for interactions with other signaling proteins. Tyrosine phosphorylated YMNM motif allows binding of SH2-domains of PI3K and Grb2. The membrane proximal proline motif (PRRPGP) serves as docking site for SH3-domain of Itk. SH3-domains of FLNa, Grb2 and Lck interact with the distal proline motif (PYAPP). FLNa, filamin-A; Grb2, growth-factor receptor-bound protein; Itk, IL-2-inducible T-cell kinase; Lck, lymphocyte-specific protein tyrosine kinase; PI3K, phosphatidylinositol 3-kinase; SH2, src homology 2 domain; SH3, src homology 3 domain.

PI 3 kinases can be divided into four classes (IA, IB, II and III) based on their structural characteristics and substrate specificity. Class IA PI3K is composed of a heterodimer between a p110 catalytic subunit and a p85 regulatory or adaptor subunit. There are five forms of the p85 regulatory subunit, designated p85α, p55α, p50α, p85β or p55γ and three variants of the p110 catalytic subunit designated p110α, β or δ. The α and β p110 isoforms are expressed in all cells, whilst p110δ is expressed primarily in immune cells. Class IB PI3K deficiency affects T-cell development and function.^50,51 The SH2-domain of the p85 regulatory subunit of PI3K binds to the CD28 pYMNM motif.^44–47 This facilitates localization of the p110 catalytic subunit in the plasma membrane, and allows for receptor mediated activation. The kinase phosphorylates phosphatidylinositol to phosphatidylinositol (3,4)-biphosphate (PIP₂) and phosphatidylinositol (3,4,5)-trisphosphate (PIP₃). These so called D-3 lipids associate with the inner leaflet of the plasma membrane and allow recruitment of proteins containing PH-domains. In this manner, PI3K has a pleiotropic effect in activating multiple signaling pathways. The basis of p85 recruitment by CD28 is the same (i.e., same binding affinity) as binding to growth factor receptors such as the platelet derived growth factor receptor (PDGF-R).⁴⁴ The YMNM motif is phosphorylated by src kinases p56^lck and p59^fyn within seconds after CD28 cross-linking.⁴⁸ The reduced anti-CD3 mediated proliferation in p110γ^-/- T cells can be partially restored by CD28 co-ligation.⁵⁰ This is compatible with CD28 engagement and amplification of signaling by other p110-p85 complexes. A selective p110δ inhibitor IC87114 inhibits cytokine production and reduces hypersensitivity responses in mice and reduces cytokine production by memory T-cells from inflammatory arthritis patients.⁵² Inhibition of PI3Kδ also attenuates allergic airway inflammation and hyperresponsiveness in a murine asthma model.⁵³

One well established downstream pathway of PI3K involves the activation of phosphoinositide-dependent protein kinase 1 (PDK1), which in turn activates protein kinase B (PKB/AKT) (Fig. 5). PDK1 and PKB can themselves phosphorylate and regulate multiple pathways. PKB phosphorylation by PDK1 at threonine 308 residue is involved in the regulation of protein synthesis, cellular metabolism and cell survival as as activated PKB further phosphorylates BAD, caspase-9, transcription factors CREB1 (cAMP responsive element binding protein 1) and forkhead, mTOR (mammalian target of rapamycin) and glycogen synthase kinases-3α and β (GSK3α and GSK3β, respectively).

Figure 5.

CD28 signaling events. CD28 associates with several intracellular signaling proteins (Fig. 4). PI3K generates D-3 lipids, which consequently recruit PDK1. PDK1 can activate PKB/AKT, PKC-θ and S6K by phosphorylation, while PKB phosphorylates numerous proteins, like BAD, mTOR and GSK3. Further, PI3K interacts with SNX9, which in turn interacts with WASP and thus can regulate actin cytoskeleton (not depicted). Similarly, another CD28 binding partner, Grb2 can bind SOS, an activator of GTPase Ras, and another guanine nucleotide exchange factor Vav1. Once activated by tyrosine phosphorylation, Vav1 activates Rac and cdc42 leading to an activation of JNK, p38 (not depicted) and actin cytoskeleton remodelling. Grb-2 binding to the YMNM motif and Vav1 is important in the activation of the NFκB pathway. Similarly, FLNa binding to the PPAYP motif is needed for activation of the NFκB pathway and can regulate actin cytoskeleton. The role of CD28 interaction with Lck and Itk and its consequences remain to be elucidated in more detail (not depicted). BAD, Bcl-2-antagonist of cell death; cdc42, cell-division cycle 42; FLNa, filamin-A; FOXO, forkhead transcription factor O; Grb2, growth-factor receptor-bound protein; GSK, glycogen synthase kinase 3; IKK, IκB kinase; Lck, lymphocyte-specific protein tyrosine kinase; mTOR, mammalian target of rapamycin; PDK1, phosphoinositide-dependent kinase 1; PI3K, phosphatidylinositol 3-kinase; PKB, protein kinase B; PKC-θ, protein kinase C u; Rac1, Ras-related C3 botulinum toxin substrate 1; Ras, p21/Rat sarcoma; S6K, S6 kinase.

Despite this, surprisingly, the role for CD28 bound PI3K is still unclear. This may in part be related to the fact that TCR ligation itself can increase PIP₃ levels,^54,55 and by the unclear role of aspects of PI3K in T-cell development and function. For example, disruption of the gene that encodes p85a has only weak effect on T cells, while B-cell development and proliferation is strongly affected. Surprisingly, T-cells lacking p85β actually proliferaes more, since CD4⁺ and CD8⁺ T-cells completed more cell divisions.^56,57 Both the p85α and β chains bind to the CD28 cytoplasmic domain. Further, as an adaptor, p85 potentiates T-cell activation independently of associated p110.⁵⁸

Mutation of the CD28 YMNM motif residues Y (i.e., disruption of Grb2 and PI3K binding) and M residues (i.e., loss of PI3K binding alone) interferes with the induction of IL-2 in T-cell hybridomas,^46,59–61 while in vivo responses showed either no⁶² or partial dependency on CD28-PI3K.^63,64 Part of this confusion may be due to the generation of sufficient D-3 lipids by high affinity peptide agonist binding to the TCR, and by the fact that CD28 can enhance PIP₃ production at the immunological synapse independently of its YMNM PI3K-recruitment motif,⁵¹ perhaps by increasing the efficiency of conjugate formation, and indirectly TCR signaling. The use of high affinity peptide ligands could obviate the need for supplemental CD28 induction of PIP₃. Although it is possible that cytokine production and proliferation per se are not linked to CD28-PI3K, the requirement for the complex in the context of low affinity peptide has yet to be measured.

Instead, CD28-PI3K was expected to protect against cell death or apoptosis. CD28 has long been known to rescue cells from TCR-driven antigen-induced cell death (AICD),^{62,63,65–67} and earlier work on the nerve growth factor receptor (NGF-R) by the Cooper lab first showed that PI3K generates pro-survival signals. Loss of the PI3K binding site in the NGF-R resulted in the induction of cell death.⁶⁸ Apoptosis can be induced by two pathways, either via death receptors [FasL (CD95L)/Fas (CD95)] or mitochondria associated proteins (i.e., the Bcl family). In the context of Fas/FasL, CD28 utilizes at least three different mechanisms: decreasing FasL expression, increasing expression of FLICE-inhibitory protein (c-FLIP), or by interfering with the formation of the death-inducing signaling complex (DISC).⁶⁹ In the mitochondria associated pathway, CD28 can activate PKB/ATK,^70,71 and increase expression of the pro-survival factors Bcl-2 and Bcl-X_L.²⁴ CD28 ligation can also induce BAD phosphorylation.⁷² However, despite this expected connection, there is also a reported discrepancy in whether CD28-PI3K can prevent apoptosis in in vivo systems. While Okkenhaug and co-workers initially showed that CD28 deficient mice reconstituted with a disrupted YMNM motif for PI3K binding are more prone to cell death (i.e., unable to upregulate Bcl-X_L),⁶² a recent report for the Green lab on CD28-YMNM transgenic knock-in mice found no apparent difference in susceptibility to apoptosis.⁷³ Overall, this discrepancy is surprising and may be due to variations in the response of T cells from different mouse strains and to different antigens.

Another connection to CD28-PI3K binding involves cell metabolism. CD28-PI3K has been reported to uniquely provide signals for increased cell metabolism.⁷⁴ Anti-CD3 ligation alone does increase metabolism, and instead, CD28 uniquely increased glucose transporter 1 (Glut1) expression, glucose uptake and glycolysis in a PI3K/PKB dependent manner. Interpretation of this study was complicated by the use of an anti-CD3 antibody (i.e., OKT3) that is sub-optimal in inducing metabolic changes. Others subsequently found that other anti-CD3 antibodies alone increase metabolic changes.⁷⁵ Nevertheless, the studies support the notion that CD28-PI3K binding has downstream effects on aspects of T-cell function by virtue of PKB/AKT activation. Overall, given the established role of PI3K-PDK1-PKB/AKT pathway in regulating numerous aspects of cell function such as cell cycle and protein synthesis, it is likely that CD28-PI3K will eventually be found to regulate multiple aspects of T-cell function.

CD28-YMNM and Grb2

In 1995, Schneider et al. showed that the CD28 YMNM motif can also bind to the adaptor protein Grb2, by virtue of the YXNX sequence,^49,61,76 where X denotes any amino acid residue (Figs. 4 and 5). The asparagine (N) residue in the motif defines the specificity of Grb2 SH2-domain binding. Grb2 consists of one SH2 domain that is flanked by two SH3 domains, which react with the PYAP motif of CD28 as well.^61,76 Grb2 in turn constitutively binds a guanosine nucleotide exchange factor (GEF) son of sevenless 1 (SOS1), which activates p21^ras and the MAPK/ERK pathway. The loss of Grb2 binding by mutation of the asparagine (N) residue leads to a loss of CD28 mediated phosphorylation of the guanine nucleotide exchange factor Vav1 and consequent activation of the serine/threonine kinase c-Jun kinase (JNK).⁶¹ Confirmation of the importance of the interaction for regulation of IL-2 secretion was seen in an elegant experiment, where the N residue was placed in the YXXM motif of ICOS, a co-receptor that normally fails to upregulate secretion of IL-2.⁷⁷ The introduction of the N residue converted ICOS to a co-receptor that can induce IL-2 production.⁷⁸ This clearly showed that Grb2 binding contributes to the ability of CD28 to potentiate IL-2 production.

In terms of signaling, CD28-Grb2 is needed to phosphorylate and activate the Dbl-homology (DH) domain of Vav1.^61,79 CD28 co-precipitates Vav1.⁸⁰ and the CD28 YXN-AX mutant shows defective JNK activation.⁶¹ Furthermore, Vav1 mutants lacking GTP-GDP-exchange activity block CD28-mediated activation of JNK.^71,81,82 Vav1 phosphorylation activates GDP-GTP-exchange (GEF) activity of the DH domain for Ras-related C3 botulinum toxin substrate 1 (Rac1) and cell-division cycle 42 (cdc42).⁸³ Activated Rac1 then activates JNK which in turn phoshorylates and activates the transcription factor c-Jun that combines with c-Fos to form the AP-1 transcription complex. AP-1 cooperates with NFAT in activation of promoter of the IL-2 gene. Consistent with this connection, CD28-Grb2 cooperates with Vav1 to activate NFAT/AP-1-dependent transcription,⁸⁰ while Grb2-like adaptors such as Gads and Grap fail to substitute for Grb2.⁸⁰ Vav1 also facilitates clustering of the TCRζ/CD3 complex,^84–86 and Rac1 can remodel the actin cytoskeleton, an event needed for TCR clustering.^87,88 In this context, co-expression of Vav1 and its binding partner SLP-76 allowed CD28 to induce NFAT translocation to the nucleus.^88,89 Overall, the collective data clearly point to CD28-Grb-2 and Vav1 cooperation in the generation of co-signals and possibly in amplifying TCR signaling.^4,89–91

NFκB Activation

Current data have also demonstrated a clear link with the activation of transcriptional factor NFκB. NFκB plays a key role in regulating the immune response to infection. Dysregulation of NFκB has been linked to cancer, inflammation, autoimmune diseases and viral infection. Stimulation of T-cells with APCs or cross-linking of TCR/CD3 plus CD28 by antibodies activates the NFκB pathway, where a TCR signal alone is insufficient.^92,93 There are several possible theoretical connections between CD28 and NFκB (Fig. 6). PI3K is required for PKB/AKT activation and activated PKB phosphorylates and activates IκB kinase (IKK). IKK in turn phosphorylates inhibitory IκB, promoting its degradation, and thereby facilitating NFκB translocation into the nucleus. Within the pathway, PDK1 can also influence PKC-θ function. Other potential pathways also focus on PKC-θ as a central regulator node, and include IL-2-inducible T-cell kinase (ITK)-PLC-γ1 activation, Grb2-Vav1 and FLNa (Fig. 6). Downstream of PKC-θ is MAGUK (membrane-associated guanylate kinase) family member CARMA1 (CARD-MAGUK 1, caspase-recruitment domain-membrane-associated guanylate kinase 1) complex formation with BIMP1 (Bcl10-interacting MAGUK protein), Bcl10 and MALT1 (Mucosa-associated lymphoid tissue lymphoma translocation protein 1).

Figure 6.

An outline of the theoretical and established pathways of CD28 regulation of NFκB. In principal, CD28 could regulate NFκB activation by interacting with PI3K, the production of D-3 lipids and recruitment of PDK1 that phosphorylates and activates PKB and PKC-θ. PKB can activate IKK by phosphorylation. Activated IKK phosphorylates IκB, which disrupts its binding to NFκB and targets it to degradation in proteasome. Free NFκB can then translocate to nucleus to initiate transcription. PKC-θ could also facilitate assembly of Bcl10-CARMA1-MALT1 complex and consequent activation of IKK. Despite this, the only established pathways of CD28 regulation of NFκB (solid lines) involve Grb2 binding to the YMNM motif and FLNa binding to the PPYAP motif. CD28-Grb2 regulation further involves Vav1, while both CD28-Grb2 and CD28-FLNa impinge on IKK via their connection to PKC-θ.

Of these options, CD28-Grb2 binding is one clear pathway to NFκB activation. Annibaldi et al. revealed that the mutation of Y in the YMNM motif (i.e., abolished PI3K and Grb2 binding) significantly reduced NFκB activation in comparison to wild-type CD28.⁹² Further, they showed that the inhibition of PI3K by LY294002 or presence of a kinase-dead mutant of PKB/AKT reduced NFκB translocation as well.⁹² Takeda and co-workers further reported that CD28-mediated NFκB activation was dependent on PKC-θ and CARMA1 and required Grb2, but not the PI3K interaction with CD28.⁹³ Mutation of the M in the motif for PI3K binding had no effect on activation, whereas loss of the Grb2 binding (i.e., N) had a marked inhibitory effect (Fig. 6). These data strongly support CD28 binding to Grb2 in NFκB activation, in particular RelA. At the same time, a distal proline-rich P¹⁸⁷YAP region has been implicated. This proline-rich site has been reported to bind to the SH3-domain of src kinase p56^lck, Grb2 and filamin A (FLNa)^61,76,94,95 (Fig. 4). Interaction of the Grb2 SH3 domain with this motif reinforces Grb2 SH2 domain binding to the YMNM motif.⁶¹ Hayashi and co-workers showed that FLNa is required for PKC-θ activity,⁹⁶ and that the FLNa binding site is needed for CD28 induction of lipid raft expression.⁹⁵ FLNa knockdown inhibited CD28-mediated raft expression.⁹⁵ Further, P¹⁸⁷YAP motif is needed for CD28 and PKC-θ localization to the cSMAC⁹⁷ at the immunological synapse.⁹⁸ These observations fit a scenario where the recruitment of PKC-θ to cSMAC requires the binding of FLNa to the P¹⁸⁷YAPP motif in CD28^95,97 (Figs. 4 and 5). In this context, we have reported that P¹⁸⁷YAPP motif increases PI3K binding to CD28 and its rate of endocytosis by two fold.³⁸ The manner by which FLNa affects NFκB is unclear but can induce signals via sphingosine kinase 1 and p21-activated kinase, or provide a platform of PKC-θ movement to the cSMAC via actin re-modeling.⁹⁹ This pathway may work in conjunction with CD28-GRb2 and Vav1 where Vav1 GEF activity is needed for PKC-θ translocation to the membrane.^99–101

CD28 and Negative Regulation

In addition to positive regulation, CD28-related factors have been reported that negatively influence or dampen CD28 co-stimulation. The CD28 P¹⁷⁵RRPGP motif that has been reported to interact with the SH3 domain of Tec family kinases, ITK and Tec.¹⁰² CD28 co-ligation enhances Itk-dependent phosphorylation of phospholipase C γ1 (PLC-γ1).^103,104 PLC-γ1 in turn generates inositol 1,4,5-trisphosphate (IP3) and diacylglycerol that regulate intracellular calcium concentration and activate protein kinase C, respectively. However, surprisingly, T cells deficient in the ITK have been reported to be hyper-responsive to CD28 co-stimulation suggesting a negative role for the kinase.¹⁰⁴ The physiological role of this P¹⁷⁵RRPGP docking site remains to be better understood.

Further, the Cbl-b^-/- T cells are hyper-responsive to peptide ligands, with a stimulation index comparable to cells activated by the combination of anti-CD3 and CD28.^116,117 Cbl-b is an E3 ligase that facilitates the degradation of various proteins such as PI3K and Vav1. Similarly, the SH2-domain-containing protein tyrosine phosphatase 1 (SHP-1) can reduce costimulation as observed in SHP-1-deficient CD4⁺ thymocytes.^106,107 In keeping with this, SHP-1 associates in a complex with Vav1, Grb-2 and Sos1.¹⁰⁵ The suppressor cytokine IL-10 has also been reported to activate SHP-1 and in the process, suppress PI3K binding to CD28.^106,108 Phosphatases such as SHIP¹⁰⁹ and SHP-2,¹¹⁰ have also been implicated in CD28 co-signaling, although the molecular basis remains to be elucidated.

Unique or Distinct Signals

A major question in CD28 co-stimulation is the degree to which it acts as unique signaling receptor, and the degree to which it simply amplifies TCR signaling. Much has been written on this subject matter with surprisingly conclusive summaries. The amplifying effect on TCR signaling was initially observed in gene array studies.^111,112 The vast majority of the genes induced by CD28 were also induced by TCR alone. The limitation of this work involved the use of potent super-agonist anti-CD3 and CD28 antibodies. Nevertheless, this work has been useful in underscoring the role of CD28 in potentiating pathways induced by TCR ligation. Future work will need to examine effects of CD28 ligation on responses to peptide agonists in antigen presentation. This should detect more subtle effects on sub-optimal CD3 signaling by low-moderate affinity peptide. The basis for the CD28 potentiating effects on TCR signals are not clear but may involve CD28 induction of lipid rafts or GEMs (glycolipid enriched microdomains) expression,^113,114 an effect on adhesion and the generation of intracellular signals that intersect with TCR signals. Grb-2-Vav1 may again be significant since Vav-1 regulates Rac1 and the actin cytoskeleton that are important in TCR signaling.^84,90,115

As outlined, besides potentiating TCR signaling, CD28 interacts with a unique combination of signaling proteins that are not associated with the TCR complex. It is well established in many receptor signaling systems that the stoichiometry of different signaling molecules is key in determining the ultimate activation of downstream pathways. It would be therefore remarkable if the unique recruitment of PI3K, Grb2 and FLNa did not manifest itself in some manner in a unique outcome, or at least in a skewing of signaling pathways. Several labs including our own have also reported that CD28 ligation in the absence of TCR engagement can activate NFAT and NFκB activity as well as protein arginine methylation.^{88,89,92,118,119} CD28 has long been found to preferentially activate JNK and NFκB, an event that may be accompanied by a unique outcome such as TH2 differentiation. Similarly, P3K activation of numerous PH-domain carrying proteins would cooperate with aspects of TCR signaling.¹²⁰ As mentioned, this will most likely be relevant in sub-optimal activation. This synergistic signal could differentially activate multiple transcription factors such as NFκB, NFAT and activator protein-1 (AP1).¹⁰⁰ Further work is still needed to establish more clearly the full range of functions controlled by the unique array of mediators coupled to CD28.

CD28 and Autoimmunity

The pro-proliferative effects of CD28 alters the threshold of TCR signaling, and as such, would be expected to impinge on the development of autoimmunity. For example, T-cell anergy is induced in CD4⁺ T-cells by recognition of an antigen in the absence of CD28 co-stimulation,^121,122 and conversely, autoimmunity is reduced in Cd28^-/- mice.¹²³ In one example, CD28 plays a crucial role in induction of autoimmune heart disease.¹²³ In this autoimmune setting, CD28 expression is required for T_H2 and IgG1 responses as well as regulatory T-cell (Treg) function.¹²⁴ Little information is available on the specific signaling pathways linked to autoimmune states except in NOD mice, where CD28 ligation reverses the reduction in Rac1/p38 MAPK signaling and IL-2 and IL-4 gene transcription.¹²⁵ In this context, CTLA-4-Ig has been widely used to treat a wide range of autoimmune diseases and transplant rejection.^31–33 Given the higher affinity for CTLA-4 for CD80 and CD86, the reagent is thought to elicit its effect mostly by outcompeting CD28 binding. Naïve T-cell responses are generally more affected than effecter or memory responses.^126,127 In studies to date, the incidence of re-activation of chronic viral infections such as herpes zoster has been low.¹²⁸ The degree to which CTLA-4-Ig will be useful in treating diseases such as rheumatoid arthritis is not fully clear, although naïve T-cell re-activation may be important and the reagent appears effective.¹²⁹ Overall, the targeting of CD28 linked pathways such as Grb2 and FLNa should provide for an alternate means to modulate co-stimulation in the context of autoimmune disorders and transplantation.

Summary

The CD28 co-receptor provides many second signals that modify that activation threshold and function of T cells. Part of this function is due to enhanced TCR signaling, and part is linked to the association of CD28 with an unique set of mediators such as PI3K, Grb2 and FLNa. A unique complement of factors PI3K, SNX9 and WASP influences the longevity of CD28 expression on the surface of cells. CD28-Grb2 binding plays specific role in the upregulation of NFκB activation and IL-2 production, while PI3K likely plays cooperative roles in the regulation of cell death and cell metabolism. The connection of the different CD28 signaling pathways to specific T-cell functions and in the susceptibility to autoimmune disorders remains to be defined.

Abbreviations

Bcl10

B-cell lymphoma/leukemia 10

CARMA1

CARD-MAGUK 1 (caspase-recruitment domain, membrane-associated guanylate kinase 1)

Grb2

growth-factor receptor-bound protein

IKK

IκB kinase

Itk

IL-2-inducible T-cell kinase

IκB

inhibitor of nuclear factor-κB

MALT1

mucosa-associated lymphoid tissue lymphoma translocation protein 1

NFκB

nuclear factor-γB

PDK1

phosphoinositide-dependent kinase 1

PI3K

phosphatidylinositol 3-kinase

PKB

protein kinase B

PKC-θ

protein kinase C θ

PLC-γ

phospholipase C γ