The JI
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     
 

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Chattopadhyay, K.
Right arrow Articles by Nathenson, S. G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Chattopadhyay, K.
Right arrow Articles by Nathenson, S. G.
The Journal of Immunology, 2006, 177: 3920-3929.
Copyright © 2006 by The American Association of Immunologists, Inc.

Structural Basis of Inducible Costimulator Ligand Costimulatory Function: Determination of the Cell Surface Oligomeric State and Functional Mapping of the Receptor Binding Site of the Protein1

Kausik Chattopadhyay*, Sumeena Bhatia*, Andras Fiser{dagger},{ddagger}, Steven C. Almo2,{dagger},§ and Stanley G. Nathenson2,*,¶

* Department of Microbiology and Immunology, {dagger} Department of Biochemistry, {ddagger} Seaver Center for Bioinformatics, § Department of Physiology and Biophysics, and Department of Cell Biology, Albert Einstein College of Medicine, Bronx, NY 10461


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 Disclosures
 References
 
Inducible costimulator (ICOS) ligand (ICOSL), a B7-related transmembrane glycoprotein with extracellular IgV and IgC domains, binds to ICOS on activated T cells and delivers a positive costimulatory signal for optimal T cell function. Toward determining the structural features of ICOSL crucial for its costimulatory function, the present study shows that ICOSL displays a marked oligomerization potential, resembling more like B7-1 than B7-2. Use of ICOSL constructs lacking either the IgC or IgV domain demonstrates that receptor binding is mediated solely by the IgV domain but requires the IgC domain for maintaining the structural integrity of the protein. To map further the receptor recognition surface on ICOSL, a homology-based protein structure model of the ICOS:ICOSL complex was constructed. Based on predictions from the model, a series of mutations were generated targeting the potential receptor binding surface on ICOSL, and the mutants were tested for their biological function in terms of ICOS binding and T cell costimulation ability. The results provide experimental validation of the model and show that the receptor binding site on ICOSL is constituted chiefly by aromatic/hydrophobic residues. Critical ICOSL residues essential for ICOS binding map to the GFCC'C'' beta-sheet face of the IgV domain and approximately overlap with the B7-1/B7-2 motif(s) that recognize CD28/CTLA-4. Altogether, similar structural features of ICOSL and B7 isoforms suggest a close evolutionary relationship between these costimulatory ligands, yet differences at the same time explain their unique specificity for the cognate binding partners, ICOS and CD28/CTLA-4, respectively.


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 Disclosures
 References
 
The process of optimal T cell activation and differentiation requires two distinct but synergistic signals. The first signal, which is Ag specific, is delivered by the TCR upon recognition of specific peptide-MHC complexes displayed on APCs. The second signal, which is Ag independent, is provided by the costimulatory receptors on T cells through binding to their cognate ligands on APCs. In the absence of this costimulatory signal, T cells enter a state of nonresponsiveness called T cell anergy (1).

The best characterized costimulatory receptor-ligand interaction is that involving CD28/CTLA-4 and B7-1/B7-2. CD28 is constitutively expressed on naive T cells and, upon binding to B7-1/B7-2 on APCs, delivers a positive costimulatory signal that augments and sustains the T cell activation process. CTLA-4, which is induced on activated T cells, also binds to B7-1/B7-2, but with much higher affinity and down-regulates the T cell activation process. The balance of CD28-mediated positive and CTLA-4-mediated negative costimulatory signals is crucial for maximizing protective immune responses while maintaining a state of immunological tolerance (1, 2, 3).

Another member of the CD28/CTLA-4 family is the inducible costimulator (ICOS),3 which is induced upon T cell activation and binds to a B7-related molecule called ICOS ligand (ICOSL; also known as B7h, GL50, B7RP-1, and B7-H2) (3, 4, 5). ICOSL is expressed on B cells, macrophages, and dendritic cells, as well as on multiple nonlymphoid tissues. Ligation of ICOS enhances T cell proliferation and the production of various effector cytokines such as IFN-{gamma}, IL-4, and IL-10. However, unlike CD28, ICOS only minimally augments IL-2 production, suggesting that the ICOS:ICOSL interaction triggers a positive costimulatory signaling cascade distinct from that associated with the CD28:B7 interaction. Consistent with a positive role in costimulation, ICOS-deficient mice exhibit severely affected T cell activation-proliferation along with a marked reduction of T cell-dependent B cell responses, a profound deficit in Ig class switching, and impaired germinal center formation (6, 7, 8). Moreover, studies with ICOSL-deficient mice have shown that ICOSL is required for proper Th cell activation, differentiation, and effector cytokine production (9).

ICOS and ICOSL are both transmembrane glycoproteins belonging to the Ig superfamily. ICOS is a disulfide-linked homodimer of extracellular IgV domains with 24 and 17% aa identity to CD28 and CTLA-4, respectively (4). Three-dimensional modeling of the extracellular IgV domain of ICOS and subsequent mutagenesis studies have demonstrated that a "FDPPPF" motif of ICOS is crucial for its interaction with ICOSL as deletion/substitution mutations in FDPPPF motif resulted in loss of binding of ICOS to ICOSL (10). Notably, this FDPPPF motif is analogous to the strictly conserved "MYPPPY" sequence motif in the IgV domains of CD28 and CTLA-4, a region critical for their binding to the B7-1 and B7-2 ligands (11). Thus, involvement of overlapping regions with different patterns of conserved and nonconserved residues in ICOS and CD28/CTLA-4 suggests an explanation for different ligand binding specificities of these receptors.

ICOSL is a two-domain protein composed of a membrane-distal IgV and a membrane-proximal IgC domain and exhibits ~20% sequence homology with B7-1 and B7-2 (5, 12). Both human and mouse ICOSL contain a free cysteine in the extracellular IgC domain, yet ICOSL does not form a disulfide-linked dimer on the cell membrane (12, 13). B7-1 and B7-2, which lack this cysteine residue, appear to exist as a noncovalent dimer and monomer on the cell surface, respectively (14). It is currently not known whether ICOSL is present as a monomer or a noncovalent oligomer on the cell membrane. Furthermore, there is no information available on the ICOSL motif that specifically recognizes its receptor ICOS. Also, no data are available on the structural, molecular, and signaling properties of ICOS and ICOSL, such as structural similarities and differences with other costimulatory molecules of the CD28 and B7 family, organization at the immunological synapse, and most importantly the mechanism by which the ICOS:ICOSL interaction activates the downstream signaling pathways.

We have characterized ICOSL with particular emphasis on the two critical structural features of the protein relevant for its costimulatory function: 1) the cell surface organization and 2) the receptor binding properties. First, using biochemical and biophysical studies, we demonstrate that the protein has a propensity to oligomerize, both in solution as well as on the cell surface. In the second part of our study, we show that ICOS binding is solely mediated by the IgV domain of ICOSL. Furthermore, with the aid of 1) a structure-based sequence alignment between ICOSL and B7-1/B7-2 and 2) a homology-based three-dimensional model of ICOS:ICOSL complex, we have identified critical amino acid residues on ICOSL responsible for its receptor binding activity. Our data suggest that the ICOS recognition surface is chiefly composed of aromatic/hydrophobic residues located on the GFCC'C'' strands in the IgV domain of ICOSL. Consistent with the loss of binding, mutations at the receptor binding interface of ICOSL also result in a reduced T cell costimulatory response, suggesting functional involvement of the respective residues in ICOS:ICOSL-mediated T cell costimulatory responses.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 Disclosures
 References
 
Soluble forms of ICOS and ICOSL

cDNAs for human ICOS and ICOSL were obtained as gifts from G. J. Freeman (Dana-Farber Cancer Institute, Boston, MA). A DNA fragment encoding the ICOS extracellular domain (residues 21–139) with N-terminal human beta2-microglobulin signal peptide (residues 1–20) was fused in-frame upstream of DNA sequence encoding the C-terminal 232 residues (Fc region) of mouse IgG2a. For extracellular domains of ICOSL, DNA segments encoding either residues 1–252 (for ICOSL-VC) or residues 1–135 (for ICOSL-V) were fused in frame with the mouse IgG2a-Fc part. The Ig-fusion constructs were introduced into the pIRES2-EGFP vector (BD Biosciences Clontech), and the resultant plasmids were transfected into HEK293T cells using LipofectAMINE 2000 (Invitrogen Life Technologies) as per manufacturer’s protocol. Seventy-two hours posttransfection, culture supernatants were harvested and used as a source of Ig-fusion proteins. Ig-fusion protein production in the culture supernatant was tested and estimated by Western blotting and ELISA, respectively, using anti-mouse IgG2a (Santa Cruz Biotechnology and BD Pharmingen).

For large-scale production of soluble ICOSL-VC, stable clone producing ICOSL-VC-Ig was established. The Ig-fusion protein was purified from the culture supernatant by affinity chromatography on Ni-NTA Agarose (Qiagen), using the tag of six histidine residues at the C-terminal end of the Ig-Fc region. The protein was eluted from the Ni-NTA Agarose column with 250 mM imidazole. The Fc region was cleaved off by proteolytic processing at an engineered preScission protease (Amersham Biosciences) site between the ICOSL-VC and the Fc region and was removed by passing the protein solution through an Immunopure Immobilized Protein A (Pierce) column. Homogeneity of the protein preparation was confirmed by SDS-PAGE and Coomassie Blue R-250 staining. The extinction coefficient of ICOSL-VC was determined by amino acid analysis to be 0.75 (for 1 mg/ml).

Cloning and expression of full-length proteins tagged with CFP and YFP

Full-length proteins (ICOS and ICOSL) were PCR-amplified using specific 5'- and 3'-end primers and subcloned into ECFP-N1 or EYFP-N1 (BD Clontech) expression vectors (14). All the mutants of ICOSL were generated by PCR-based mutagenesis strategies. All the constructs were verified by DNA sequencing. HEK293T or Chinese hamster ovary (CHO) cells were transfected with the recombinant vectors coding for CFP- or YFP-tagged proteins, using either LipofectAMINE 2000 or Fugene6 (Roche Biochemicals). Twenty-four hours posttransfection, cells were analyzed for protein expression using flow cytometry and Western blotting. CHO cell lines with stable expressions of YFP-tagged ICOSL variants were established by selection with 1 mg/ml G418 sulfate (Mediatech).

Western blot analysis of ICOSL expressed on the cell membrane

Intact HEK293T or CHO cells expressing YFP/CFP-tagged variants of ICOSL were washed twice with PBS (pH 7.4). For cross-linking experiments, intact cells were incubated with 5 mM BS3 (Pierce) in PBS (pH 8.0) for 30 min at 37°C, followed by quenching of the reaction with 0.5 M Tris-HCl (pH 7.5) for 5 min. Membrane protein fractions were extracted using the Mem-PER Eukaryotic Membrane Protein Extraction Reagent Kit (Pierce) as per the manufacturer’s protocol and subjected to immunoblotting using Abs either against ICOSL or GFP (Santa Cruz Biotechnology).

Flow cytometry

Cell surface expression of ICOS and ICOSL was determined by flow cytometry using biotin-conjugated primary Ab (anti-human ICOS mAb, clone ISA-3, and anti-human ICOSL mAb, clone MIH12, respectively; eBioscience) and the streptavidin-allophycocyanin (SAv-APC; BD Pharmingen) conjugate as the second-step reagent. Binding to Ig-fusion proteins was measured by treating the cells with appropriate concentrations of Ig-fusion proteins, followed by secondary staining with Cy5-coupled anti-mouse IgG (Jackson ImmunoResearch Laboratories). The stained cells were analyzed using a FACSCalibur (BD Biosciences) equipped with CellQuest software.

Photobleaching-based fluorescence resonance energy transfer (FRET) experiment

FRET experiments were done following the method described earlier (14). Briefly, CHO cells were cotransfected with the CFP- and YFP-tagged constructs. Twenty-four hours posttransfection, fixed cells were examined with a Leica TCS SP II AOBS laser-scanning confocal microscope. CFP was excited at 405 nm, and emission was monitored over the range of 416–492 nm; YFP was excited at 514 nm, and emission was recorded over the range of 525–600 nm. Photobleaching of YFP was achieved by using a 514-nm line in full power for 1–2 min. FRET efficiency (%FRET) was calculated using the relationship: percentage of FRET = ((Dpost – Dpre)/Dpost) x 100, where Dpre and Dpost are the CFP fluorescence intensity before and after photobleaching, respectively.

In vitro T cell costimulation assay

This study was performed under a protocol approved by the Albert Einstein College of Medicine. Purified human T cells (>95%) were isolated by negative selection of fresh PBMC using Ab-labeled magnetic beads, as per the manufacturer’s protocol (Dynal Beads), and stimulated with plate-bound anti-CD3 (3 µg/ml) and soluble anti-CD28 (1 µg/ml) for 48 h. At 48 h, the activated cells were harvested, and the dead cells were removed by Ficoll-Hypaque density gradient centrifugation. The activated cells were rested for an additional 24 h. A small aliquot of the cells was used to confirm the expression of ICOSL by flow cytometry (data not shown). Activated T cells (105 cells) were cultured in 96-well flat-bottom plates with various concentrations of coated anti-CD3 in the presence of 5 x 104 mitomycin C-treated CHO cells expressing ICOSL variants. At 24 h, culture supernatants were harvested and analyzed for IL-10 and IFN-{gamma} by ELISA, according to the manufacturer’s protocol (BD Pharmingen).

Structure-based sequence alignment

A sequence alignment of human ICOSL with B7-1/B7-2 was constructed using the program ClustalW (15) and was manually adjusted and was rendered with human B7-1 structure using the program Espript (16).

Molecular modeling of the three-dimensional structure of the ICOS:ICOSL complex

The ICOS-ICOSL protein complex was constructed using comparative structure modeling using Modeller (17). The most suitable template found in the Protein Data Bank (18) by Psi-Blast (19) was the crystal structure of the CTLA-4:B7-1 costimulatory complex (Protein Data Bank code 1i8l). The sequence alignment between CTLA-4:B7-1 (template) and ICOS:ICOSL (target) was generated by the Multiple Mapping Method (20), which optimally splices together alignments from a variety of alternative inputs. The overall sequence identity between the template and target was 25%. For graphical representations of the models PyMOL program (PyMOL Molecular Graphics System found online (<www.pymol.org>)) was used. Solvent accessible surface in the bound as well as in the free states of the proteins was calculated by the DSSP (21) program and was used to determine the buried surfaces at the protein interfaces.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 Disclosures
 References
 
ICOSL has the tendency to form an oligomer in solution

The soluble extracellular domain of ICOSL (ICOSL-VC) was expressed in HEK293T cells. The purified protein migrated as a broad band of ~60 kDa under both reducing and nonreducing condition in SDS-PAGE (Fig. 1A), suggesting that ICOSL-VC did not form a disulfide-linked dimer in solution. A relative molecular mass of 60 kDa for ICOSL-VC was consistent with heavy glycosylation of the protein at up to six potential sites (12). When subjected to gel-filtration chromatography on a calibrated Superdex G-200 column (30 x 1 cm) at 4°C, the purified protein (at a concentration of 1 mg/ml) was distributed into two symmetric peaks: 1) one corresponding to the ICOSL-VC monomer of ~60 kDa; and 2) a second peak corresponding to an oligomeric assembly of the protein (Fig. 1B). The existence of a significant amount of noncovalent oligomer was also observed by native PAGE (Fig. 1B, inset). The identity of the monomeric and oligomeric forms of the protein was confirmed by Western blotting using anti-ICOSL Ab (data not shown). Both the oligomeric and monomeric forms of ICOSL-VC bound to soluble ICOS-Ig as demonstrated by the ligand-blotting experiment (Fig. 1C), indicating that both species were functionally active. Additional information regarding the stoichiometry of the ICOSL oligomer was obtained through a BS3-mediated cross-linking experiment, which showed presence of only dimer (~120 kDa) and no other oligomeric species of higher order assembly (Fig. 1D). The distribution between monomer and oligomer did not readily reappear when gel-filtration chromatography was performed on fractions corresponding to either monomer or oligomer, suggesting the absence of a dynamic equilibrium between the two species. However, when stored at 4°C, the monomer converted slowly but irreversibly into the oligomeric form in a concentration-dependent manner (data not shown). The conversion to oligomer was enhanced significantly by incubating at 37°C for 24 h. Such a concentration- and temperature-dependent conversion of the monomer to the oligomer could explain generation of ICOSL-VC oligomer during extraction and purification of the protein from the mammalian protein expression system. Taken together, these results demonstrated that this soluble form of ICOSL has a marked potential to form oligomer, predominantly dimer, in solution.


Figure 1
View larger version (22K):
[in this window]
[in a new window]
  		FIGURE 1. Formation of noncovalent oligomer of soluble ICOSL-VC. A, SDS-PAGE and Coomassie staining of purified ICOSL-VC under reducing (lane 1) and nonreducing (lane 2) conditions demonstrate that ICOSL-VC does not form disulfide-linked dimers. Molecular mass markers are shown (lane M). B, Elution profile of purified ICOSL-VC from Superdex G-200 gel filtration chromatography column. The two peaks at ~13.6 and ~11.6 ml correspond to the monomeric and noncovalent oligomeric form of the protein, respectively. Inset, Native PAGE and Coomassie staining of purified ICOSL-VC. C, Ability of the ICOSL-VC monomer and oligomer to bind to ICOS-Ig is shown by ligand blotting. Purified ICOSL-VC preparation containing both monomer and oligomer were subjected to native PAGE, transferred electrophoretically onto the polyvinylidene difluoride membrane, treated with ICOS-Ig fusion protein (10 µg/ml) for 16 h at 25°C, and finally probed with HRP-conjugated anti-mouse IgG2a. D, Cross-linking of ICOSL-VC with BS3 (5 mM) shows presence of dimer as the only oligomeric species (lane 2). Protein bands were visualized by SDS-PAGE and Coomassie staining.

 
ICOSL has the propensity to exist as a noncovalent dimer on the cell membrane

To characterize the organization of ICOSL on the cell membrane, recombinant ICOSL fusion proteins tagged with either yellow fluorescent protein (YFP) or cyan fluorescent protein (CFP) at the C terminus were expressed in cultured mammalian cell lines (CHO or HEK293T cells). The presence of YFP- or CFP-tag at the C terminus of ICOSL did not affect its cell surface expression or binding to soluble ICOS-Ig (Fig. 2A; data not shown for ICOSL-CFP). Immunoblot analysis of membrane fractions prepared from CHO cells stably transfected with ICOSL-YFP showed presence of an ~90-kDa band (Fig. 2B), which corresponds to the molecular mass range expected for a glycosylated monomer of ICOSL-YFP fusion protein. The presence of multiple bands is the consequence of different glycosylation states of the protein, as treatment with PNGaseF generated a single band of ~70 kDa (data not shown). Moreover, a similar ~90-kDa band for ICOSL-YFP was observed under reducing as well as nonreducing conditions, supporting the earlier observation that ICOSL does not form any disulfide-linked dimer on cell membrane (13) (Fig. 2B). Similar results were also obtained for ICOSL-CFP (data not shown).


Figure 2
View larger version (36K):
[in this window]
[in a new window]
  		FIGURE 2. Noncovalent association of ICOSL on cell surface. A, Stable expression of ICOSL-YFP on CHO cell surface (black curve, left panel) and its binding to soluble ICOS-Ig (5 µg/ml) (black curve, right panel) were detected by flow cytometry. Filled gray curves represent immunostaining control of untransfected CHO cells. B, Stable expression of ICOSL-YFP on CHO cell membrane was detected by Western blotting using anti-ICOSL Ab under reducing (lane 1) and nonreducing (lane 2) SDS-PAGE (4–20% Tris-HCl gel). C, ICOSL-YFP expressed on HEK293T cell membrane after transient transfection was cross-linked with BS3 and immunoblotted with anti-GFP Ab under reducing SDS-PAGE (3–8% Tris-acetate gel). Cross-linking shows presence of ICOSL oligomers of higher stoichiometry (lane 4). B7-2-YFP was used as control (lane 5). D, Percentage of FRET for ICOSL increases with increasing YFP intensity at each YFP:CFP ratio; 1:1 (blue), 2:1 (red), and 3:1 (green). E, Percentage of FRET for ICOSL shows moderate linear dependence on YFP:CFP ratio. F, Average percentage of FRET for ICOSL ({blacksquare}) increases with increasing YFP:CFP ratio. B7-2 ( Figure 2) is shown for control. The error bars represent the 0.01% confidence limits. G, ICOSL-Cys-YFP expressed on HEK293T cell membrane after transient transfection migrates almost exclusively as dimers under nonreducing SDS-PAGE (lane 6) as detected by immunoblotting with anti-GFP Ab. ICOSL-Ctrl-YFP is incorporated for control. H, Comparison of the average percentage of FRET values for ICOSL-Ctrl-YFP ({blacksquare}) and ICOSL-Cys-YFP (Figure 2) at 1:1 and 2:1 YFP:CFP ratios.

 
To further investigate whether ICOSL can form noncovalently associated dimeric or higher order assemblies on the cell membrane, a noncleavable membrane impermeable cross-linking agent, BS3, was used. HEK293T cells were transiently transfected with ICOSL-YFP and pretreated with 5 mM BS3 before cell lysis. Immunoblot analysis of the membrane fraction showed that a small fraction of the protein could potentially exist as dimers as well as higher order oligomers (Fig. 2C). For comparison, B7-2-YFP was used, which existed almost exclusively as monomer after BS3-mediated cross-linking (Fig. 2C, lane 5).

To further demonstrate noncovalent association of ICOSL on the cell membrane, the interaction between ICOSL-CFP and ICOSL-YFP molecules was studied using photobleaching-based FRET (pbFRET) experiments (14, 22) as described in Materials and Methods. Briefly, FRET occurs due to nonradiative energy transfer from an excited donor fluorophore to an acceptor fluorophore when they are in close proximity (10–100 Å). In pbFRET, the FRET efficiency is measured as an increase in the fluorescence intensity of donor fluorophore after photobleaching of the acceptor fluorophore. Using theoretical models for the distribution of molecules on the cell membrane (23, 24), the FRET efficiency was analyzed for its dependence on the YFP intensity and the ratio of YFP:CFP intensities (Fig. 2, D–F). CHO cells cotransfected with ICOSL-CFP and ICOSL-YFP exhibited an average FRET efficiency of, ~9, ~10, and ~14% for YFP:CFP ratio 1:1, 2:1, and 3:1, respectively (Fig. 2F). The FRET efficiency displayed considerable dependence on increasing YFP intensity, as well as with increasing YFP:CFP ratio (Fig. 2, D–F). This behavior is consistent with a considerable degree of self-association between YFP- and CFP-tagged molecules (22, 24), thereby suggesting that ICOSL shows a marked potential to form noncovalently associated oligomers on the cell surface. For comparison, CFP- and YFP-tagged B7-2 was used, which exhibited an average FRET efficiency value of 6–7% over the same YFP:CFP ratios described for ICOSL (Fig. 2F). Since FRET efficiency depends, among all other factors, on the proximity between donor and acceptor fluorophores, our data suggest that ICOSL shows a considerable degree of concentration-dependent homo-association on the cell surface in comparison to B7-2 that has recently been shown to exist as monomer on the cell surface (14). Moreover, the average FRET efficiency between CFP- and YFP-tagged ICOSL displayed a linear dependence on the YFP:CFP ratio (slope of the plot in Fig. 2E). Such a linear dependence of FRET efficiency on acceptor:donor ratio has been shown to indicate a monomer-dimer equilibrium and absence of higher order oligomers (14, 25, 26). Our data, therefore, suggest that ICOSL is present on the cell surface as a mixture of monomers and oligomers, predominantly dimers.

To further explore the propensity of ICOSL to form noncovalently associated homodimers on cell membranes, we generated a "cysteine-trap" mutant, described for signaling molecules like CD45, and B7-1/B7-2 (27, 14). Specifically, a short linker sequence of 8 aa containing a cysteine residue (Cys: GAGAGCGA) was introduced into the extracellular stalk region adjacent to the transmembrane domain of ICOSL. This would allow trapping of any potential homodimers of ICOSL through formation of an intermolecular disulfide bond. The constructs with a linker lacking the cysteine (Ctrl: GAGAGAGA) served as control. Flow cytometry-based analysis revealed that both the ICOSL-Cys-YFP and ICOSL-Ctrl-YFP expressed on HEK293T cell surface and bound to soluble ICOS-Ig equally efficiently compared with the wild-type protein (data not shown). As revealed by Western blot analysis of the membrane fraction isolated from transiently transfected HEK293T cells (Fig. 2G), ICOSL-YFP containing a cysteine linker (ICOSL-Cys-YFP) migrated almost exclusively as a dimer under nonreducing conditions in SDS-PAGE. The presence of cysteine, however, did not change the average FRET efficiency values of ICOSL (Fig. 2H), indicating that there was no overall alteration in the extent of intermolecular association between ICOSL molecules after a cysteine was introduced. This behavior is consistent with a preformed oligomeric assembly of the protein on the cell surface. Altogether, these data suggest a model in which ICOSL oligomerizes on the cell membrane, with a marked tendency to form noncovalent dimers.

The IgV domain of ICOSL is responsible for receptor binding but requires the IgC domain for structural integrity

To map the receptor binding domain of ICOSL, we generated ICOSL fusion proteins composed of either both IgV and IgC domains (ICOSL-VC) or only IgV domain (ICOSL-V) fused in frame to mouse IgG2a Fc fragment (Fig. 3A). The ability of these ICOSL-Ig fusion proteins to bind to the ICOS receptor expressed on the surface of HEK293T cells was assessed by flow cytometry (Fig. 3B). Both the recombinant proteins, ICOSL-VC-Ig and ICOSL-V-Ig, exhibited similar equilibrium binding to ICOS, indicating that the IgV domain of ICOSL alone could bind to ICOS. However, when tested over a broad concentration range, ICOSL-V-Ig was found to bind at least 8- to 10-fold less well to ICOS compared with ICOSL-VC-Ig (Fig. 3C). Similar results were obtained when the binding activity of ICOS-Ig was analyzed with HEK293T cells expressing either the full-length ICOSL (containing both IgV and IgC domains; ICOSL-YFP) or a truncated version of the protein containing only IgV domain (ICOSL-{Delta}C-YFP) fused in frame with YFP (Fig. 4A). Notably, the truncated version of ICOSL lacking the IgV domain and containing only the IgC domain (ICOSL-{Delta}V-YFP) was expressed on HEK293T cell surface but did not show any detectable binding to soluble ICOS-Ig (data not shown). Remarkably, the replacement of the ICOSL IgC domain with the IgC domain of B7-2 was capable of restoring at least wild-type binding to ICOS-Ig (our unpublished observation). The results described above suggest that the IgV domain of ICOSL alone serves as the receptor-binding unit. However, the presence of the IgC domain of ICOSL augments the efficacy of ICOS:ICOSL interaction, likely by maintaining the structural integrity/stability of ICOSL rather than by directly contributing contact residues for ICOS binding.


Figure 3
View larger version (22K):
[in this window]
[in a new window]
  		FIGURE 3. Receptor binding activity of ICOSL-VC-Ig and ICOSL-V-Ig. A, Culture supernatants of HEK293T cells transiently transfected to express ICOSL-VC-Ig (lane 1) or ICOSL-V-Ig (lane 2) were immunoprecipitated with protein A, and the precipitated proteins were separated on SDS-PAGE under reducing condition and visualized by Coomassie staining. B, Binding of soluble ICOSL-VC-Ig (black curve, right panel) and ICOSL-V-Ig (dark gray curve, right panel) at a concentration of 1.6 µg/ml to ICOS-YFP expressed on HEK293T cells after transient transfection, as detected by flow cytometry. ICOS-YFP expression is shown by the black curve in the left panel. Filled gray curve represents immunostaining control of the untransfected cells. C, Binding of ICOSL-VC-Ig ({square}) and ICOSL-V-Ig ({circ}) to ICOS-YFP was detected by flow cytometry at different concentrations. Binding data shown are the mean fluorescence intensities after subtracting the values for control supernatant from mock-transfected HEK293T cells. The result shown is a representative of at least three independent experiments.

 

Figure 4
View larger version (16K):
[in this window]
[in a new window]
  		FIGURE 4. Role of ICOSL IgC domain in receptor binding and cell surface organization. A, Binding of soluble ICOS-Ig to ICOSL-YFP ({triangleup}) and ICOSL-{Delta}C-YFP ({circ}) transiently expressed on HEK293T cell surface was measured by flow cytometry. The binding data were calculated as the mean fluorescence (MF) normalized for the cell surface expression of ICOSL variants using the formula: (MFmutant – MFmock)Binding x KExpression, where KExpression = ((MFWt – MFmock)/(MFmutant – MFmock))Expression when anti-ICOSL Ab was used to measure cell surface expression. Data were presented as the percentage of binding obtained with the wild-type protein. The result shown is a representative of three independent experiments. B, Percentage of FRET vs YFP intensity plot for ICOSL-{Delta}C at different YFP:CFP ratios; 1:1 (blue), 2:1 (red), and 3:1 (green). C, Increased average percentage of FRET of ICOSL-{Delta}C (Figure 4) compared with that of wild-type ICOSL ( Figure 4) suggests a higher degree of self-association for the truncated protein on cell surface. D, ICOSL-{Delta}C-Cys-YFP transiently expressed on HEK293T cell membrane migrates as a diffuse oligomeric species and not as a precise dimer (lane 2) as detected by immunoblotting with anti-GFP Ab under nonreducing SDS-PAGE. ICOSL-{Delta}C-YFP without Cys-linker serves as control (lane 1).

 
To assess the role of the IgC domain in maintaining the structural integrity and proper molecular orientation of ICOSL on the cell membrane, cell surface organization was analyzed for ICOSL variants either containing both the IgV and the IgC domains or only the IgV domain alone. pbFRET was used to address the issue. FRET efficiency was compared for CFP- and YFP-tagged ICOSL or ICOSL-{Delta}C proteins transiently expressed on the surface of CHO cells. As shown in Fig. 2F, the average FRET efficiency for ICOSL was found to be 9 and 10% at 1:1 and 2:1 ratio of YFP:CFP, respectively. A significantly higher FRET efficiency, ~17% was obtained at 1:1, as well as 2:1 ratio of YFP:CFP for ICOSL-{Delta}C (Fig. 4, B and C), suggesting that ICOSL-{Delta}C displayed higher degree of homo-association on cell membrane.

To further characterize the observed increased homo-association of ICOSL-{Delta}C on the cell membrane, a cysteine-trap mutation was introduced in the stalk region of ICOSL-{Delta}C-YFP. As shown by the Western blot profile in Fig. 4D, lane 2, ICOSL-{Delta}C-Cys-YFP migrated as a diffuse oligomeric species, and not as a precise dimer as observed for ICOSL-Cys-YFP, in SDS-PAGE under nonreducing conditions. Altogether, these data suggest that, in the absence of an IgC domain, ICOSL has a tendency to form nonspecific aggregates on the cell membrane. This was further supported by the fact that the soluble form of ICOSL containing only the IgV domain readily formed insoluble aggregates (gel filtration chromatography and PAGE under native condition; data not shown), whereas the other soluble form of ICOSL containing the entire extracellular domain (both the IgV and IgC domain) existed as a homogeneous population in solution (as displayed by elution profile from gel filtration chromatography column). Since removal of IgC domain results in considerable decrease in receptor binding ability of the protein (Figs. 3C and 4A), we conclude that formation of nonspecific aggregates by ICOSL-{Delta}C is deleterious for the ligand-receptor interaction supporting the hypothesis that the IgC domain stabilizes the structural integrity of ICOSL protein necessary for optimal ICOS:ICOSL binding.

Identification of amino acid residues in ICOSL crucial for ICOS binding

To further define the binding determinants in ICOSL, 1) a structure-based sequence alignment of ICOSL with the B7 isoforms (Fig. 5A) and 2) a homology-based protein structure model of the ICOS:ICOSL complex (Fig. 5, B and C) based on the CTLA-4:B7-1 structure (28) was constructed. The ICOS:ICOSL model contains several features: 1) The "FDPPPF" loop in ICOS, previously predicted to be crucial for ligand binding, makes substantial contacts with ICOSL. This is similar to the binding mode (MYPPPY loop) present in the CTLA-4:B7-1/B7-2 complexes (11, 28, 29). 2) ICOSL residues that are predicted from the model to make contacts with ICOS are well aligned with the B7-1/B7-2 residues that interact with CTLA-4 (Fig. 5A). 3) None of the potential glycosylation sites in ICOS/ICOSL appear to interfere with the proposed ICOS:ICOSL interface.


Figure 5
View larger version (55K):
[in this window]
[in a new window]
  		FIGURE 5. Mapping of receptor binding surface on ICOSL by structure-based sequence alignment and homology modeling. A, Structure-based sequence alignment of ICOSL with B7 isoforms. IgV- and IgC-set beta-strands are marked with black and blue letters, respectively. B7 residues critical for receptor binding are marked with asterisks. As predicted from the model (C), ICOSL residues forming putative receptor binding surface are indicated with red triangles. Potential N-glycosylation sites in ICOSL are marked green. B, Homology-based protein structure model of the costimulatory complex formed by the extracellular domains of ICOS (red) and ICOSL (green). C, ICOSL residues buried at the ICOS:ICOSL model interface are listed and colored according to the buried surface area (cyan, 10–25 Å2; blue, 25–50 Å2; and orange, 50–75 Å2). The 114FDPPPF119 loop (pink, indicated with the arrowhead) of ICOS is also buried at the interface (F114, 108 Å2; D115, 16 Å2; P117, 75 Å2; P118, 58 Å2; and F119, 95 Å2).

 
In the ICOS:ICOSL model, residues were defined as contributing to the binding interface if the solvent accessible surface area decreased by at least 10 Å2 upon forming the complex (calculated by the DSSP program). On this basis, 13 residues in the ICOSL IgV domain were predicted to contribute to the binding interface (Fig. 5C) and were targeted for mutagenesis (Table I). Notably, mutational studies on B7-1/B7-2 show involvement of additional residues in receptor binding. For example, alterations of the residues in the IgC domains of B7-1 affect its interactions with CD28/CTLA-4 (30); however, these effects must have been indirect as CTLA-4 only contacts residues of the B7-1 IgV domains (28). Therefore, in addition to the homology-based protein structure modeling, an amino acid sequence alignment of ICOSL with B7-1/B7-2 was also used to identify additional, if there is any, ICOSL residues potentially critical for receptor binding. From the alignment, ICOSL residues were selected for alterations at the sites where the corresponding B7 residues were found to be important for receptor binding (Fig. 5A and Table I) (30, 31).


View this table:
[in this window]
[in a new window]
  		Table I. Binding activities of ICOSL mutants

 
Combining the predictions from the three-dimensional model building and the amino acid sequence alignment strategies, mutagenesis was performed for the respective residues from ICOSL sequence at 26 positions (Table I). Each of these residues was replaced by alanine, and the full-length mutant YFP fusion proteins expressed on the cell surface were assessed for their abilities to bind to soluble ICOS-Ig fusion protein by flow cytometry. As detected by YFP fluorescence and Western blotting using anti-GFP Ab, all the mutants were expressed in HEK293T cells after transient transfection (data not shown). Based on flow cytometry assay using the anti-ICOSL mAb (clone MIH12; eBioscience), all mutants exhibited wild-type cell surface expression (Table I), suggesting that the overall structural integrity of these mutants remained unaltered after alanine substitutions.

The ICOSL mutants expressed on the cell surface were tested for their ability to bind soluble ICOS-Ig using flow cytometric analysis (Table I and Fig. 6A). Nine of the mutations (Y51A, Y53A, Q55A, V62A, H66A, Y80A, L114A, L116A, and F122A) exhibited none or significantly reduced binding to ICOS-Ig. In particular, Y51A, Y53A, Y80, L116A, and F122A mutations completely abrogated binding, whereas Q55A, V62A, and L114A mutations caused ~70–80% reduction in binding and H66A caused ~50% loss of binding. The double mutant of Q55A-V62A completely abolished ICOS binding. When more conservative substitutions with phenylalanine were examined at positions 51 and 53, Y51F supported ~80–90% of the wild-type binding activity while the Y53F mutation resulted in the loss of all binding activity. The relative locations of these critical residues involved in ICOS recognition could be mapped on the ICOSL extracellular IgV domain as depicted in the Fig. 6B. Notably, in the model, most of these residues (Y51, Y53, Q55, H66, L114, L116, and F122) were in close proximity on the front beta-sheet surface (strands GFCC'C'') of the ICOSL IgV domain directly facing the ICOS FDPPPF motif (Figs. 5C and 6B).


Figure 6
View larger version (31K):
[in this window]
[in a new window]
  		FIGURE 6. Characterization of ICOSL mutants to map the residues involved in ICOS binding. A, Binding of ICOS-Ig (30 µg/ml) to HEK293T cells transiently expressing ICOSL variants tagged with YFP was determined by flow cytometry. The binding data (representative of at least three independent experiments) are calculated as the mean fluorescence normalized for the cell surface expression of ICOSL variants using the formula in Fig. 4A and are presented as the percentage of binding obtained with the wild-type protein. B, ICOSL residues whose mutation affects receptor binding are mapped onto the IgV-like domain of ICOSL in the ICOS:ICOSL complex model and are colored according to the percent reduction in binding (orange, 100%; blue, 70–80%; and cyan, 50%). With the exception of V62 and Y80, all the residues are predicted to be exposed on the domain surface in close proximity of the FDPPPF loop of ICOS. Potential N-linked glycosylation sites in the IgV domain are shown in yellow.

 
Glycosylation contributes significant mass (~30 kDa) to ICOSL (13) as is common in many other Ig superfamily members. The importance of the six potential N-linked glycosylation sites (N70, N102, N137, N173, N186, and N225) to ICOSL structure/function (Fig. 5A and Table I) was investigated by individually introducing N-to-Q mutations at each of these sites. As shown in Table I, five of the six glycosylation mutants expressed on the cell surface and bound to ICOS-Ig almost as well as the wild-type protein, indicating that glycosylation at these sites do not play any critical role in ICOS recognition. This behavior is consistent with the placement of these putative N-glycosylations distant from the modeled ICOS:ICOSL interface. Locations of the two IgV domain N-glycosylations sites, N70 and N102, are shown in Fig. 6B; the other four are located on the IgC domain of the protein. N173Q mutant was expressed in HEK293T cells based on YFP fluorescence (data not shown) but could not be detected on cell surface (Table I), presumably due to the fact that this mutation interfered with transport of ICOSL to the cell surface.

Mutations at Y51, Y53 and V62 that disrupt ICOS:ICOSL interaction inhibit ICOS-mediated T cell costimulation

ICOSL mutants with altered ICOS binding properties were further compared with their corresponding biological functions. Stable transfectants of CHO cells expressing either the wild-type or various mutants of ICOSL (Y51A, Y53A, and V62A) were generated and tested for their ability to provide ICOS-mediated costimulation to purified human T cells. The wild-type and the mutants of ICOSL showed equivalent levels of expression on the cell surface (Fig. 7A, upper panels). When tested for their ability to bind soluble ICOS-Ig, CHO cells stably expressing either ICOSL-Y51A-YFP or ICOSL-Y53A-YFP showed no detectable binding, whereas CHO cells expressing ICOSL-V62A-YFP displayed significantly reduced binding as compared with the wild-type ICOSL-YFP (Fig. 7A, lower panels). These data agree well with the results obtained from the transient transfection experiments described above.


Figure 7
View larger version (42K):
[in this window]
[in a new window]
  		FIGURE 7. Disruption of ICOS:ICOSL interaction due to mutations of ICOSL residues blocks T cell costimulation. A, Stable expression of ICOSL variants on CHO cell surface (upper panel) and binding to ICOS-Ig (30 µg/ml) (lower panel) were determined by flow cytometry. Filled gray histograms, immunostaining control with untransfected cells; black curves, wild-type and mutants of ICOSL (Y51A, Y53A, V62A). Numbers shown are the mean fluorescence intensities. B, These CHO cells were used for in vitro T cell costimulation assay. Mutations affecting ICOS binding caused reduced IL-10 (left panel) and IFN-{gamma} production (right panel) in the T cell costimulation assay; ({blacksquare}) wild-type ICOSL, ({circ}) ICOSL-Y51A, ({triangleup}) ICOSL-Y53A, ({blacktriangleup}) ICOSL-V62A; and ({square}) untransfected CHO cells. Data shown are the mean (±SEM) of triplicate cultures and are representative of three independent assays.

 
To assess the effect of Y51A, Y53A, and V62A mutations in ICOSL on T cell costimulation, human peripheral blood T cells, preactivated by anti-CD3 and anti-CD28, were restimulated for 24 h with varying concentrations of anti-CD3 in the presence of CHO cells expressing either the wild-type or various mutants of ICOSL. The T cell response was evaluated by measuring IL-10 and IFN-{gamma} in the cell culture supernatant. IL-10 production by T cells, which is hallmark cytokine for the ICOS:ICOSL costimulation, was critically dependent on an intact ICOS:ICOSL interaction (Fig. 7B, left panel). When CHO cells expressing Y51A or Y53A mutants of ICOSL were used, there was complete abrogation of IL-10 production, which is consistent with the loss of ICOS binding observed for these mutants. Furthermore, the V62A mutation, which resulted in partial loss of binding to ICOS, resulted in partial loss of IL-10 production. The Y51A, Y53A, and V62A mutations caused similar, albeit much more modest, reductions in IFN-{gamma} production (Fig. 7B, right panel). Collectively, these data suggest that mutations disrupting the ICOS:ICOSL interaction are also deleterious in triggering ICOS on activated T cells, thereby confirming the functional significance of the corresponding amino acid residues in ICOS:ICOSL-mediated T cell costimulatory response.


   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
Disclosures
 References
 
The present study identifies structural features of ICOSL important for costimulatory function. Homology-based protein structure modeling combined with binding studies of cell surface expressed mutants allows for the experimental mapping of critical amino acid residues that form the ICOS recognition surface on ICOSL. As suggested by the model in Figs. 5C and 6B, the ICOS:ICOSL binding interface is formed by the GFCC'C'' face of the IgV domain of each molecule, with an ~90° angle between the two interacting beta-sheets. Such an orthogonal mode of binding between the GFCC'C'' face of the IgV domains is also observed in the crystal structure of CTLA-4:B7 complexes (28, 29). The model also suggests that ICOSL makes substantial contacts with the FDPPPF sequence in the FG loop of ICOS through a surface with considerable hydrophobic character (Y51, Y53, Q55, S59, T61, H66, L114, L116, L120, G121, F122, Q123, and E124). Our subsequent mutagenesis data provide evidence for the important role of at least nine residues located on the ICOSL IgV domain in ICOS binding (Y51, Y53, Q55 and V62, H66, Y80, L114, L116, and F122). Alanine substitutions of these residues lead to decreased receptor binding and importantly equivalent decreased costimulatory response. As visualized in the model of the ICOS:ICOSL complex (Fig. 6B), seven of these residues become buried upon formation of the ICOS:ICOSL interface and are positioned in close spatial proximity to the FDPPPF motif of ICOS. Two exceptions are the V62 and Y80 that do not make direct contacts with the residues from ICOS but remain buried in the ICOSL core. It is possible that V62 and Y80, which are partially buried within ICOSL IgV domain, elicit an indirect effect on binding through the stabilization of the C strand in a conformation that is optimal for receptor binding. Among the seven residues (Y51, Y53, Q55, H66 L114, L116, and F122) contributing to the potential receptor binding interface in ICOSL, major impact on ICOS binding is observed with the mutations of Y51, Y53, L116, and F122. Moreover, Y53 appears to be more critically involved in receptor recognition since semiconservative replacement with phenylalanine at this position could not restore the binding (Fig. 6A). Amino acid sequence alignment indicates that this tyrosine residue is highly conserved in ICOSL and B7-1 molecules (mouse and human variants; data not shown) and is replaced by another aromatic residue phenylalanine in case of B7-2 (Fig. 5A). The importance of this conserved tyrosine has also been implicated in the CTLA-4:B7 interaction through structural and mutational studies (28, 29, 30, 31). Examination of the CTLA-4:B7-1 complex structure reveals that Y65 (equivalent of Y53 in ICOSL) of B7-1 participates in a stacking interaction with the second proline residue of the CTLA-4 MYPPPY loop, thereby forming the core of the CTLA-4:B7-1 interface (28). Altogether, the present data on the ICOS binding site on ICOSL and the previous study on the ICOSL binding site (the FDPPPF motif) on ICOS (10) support an aromatic/hydrophobic interface between the two proteins. This is very consistent with the binding interface formed between CTLA-4/CD28 and the B7 isoforms as shown by the previous structural and mutagenesis studies. The overall aromatic/hydrophobic nature of these receptor:ligand interfaces, consisting of a similar yet distinct set of overlapping amino acid residues, suggests a close evolutionary relationship between the CD28/CTLA-4:B7 and ICOS:ICOSL costimulatory interactions, while at the same time providing a structural basis for the specificity for cognate binding partners in these costimulatory pathways.

The role of the IgC domain of ICOSL in receptor binding has also been addressed. Deletion of the IgC domain led to the conclusion that this domain is not critically involved in receptor binding. Nonetheless, the presence of the IgC domain is implicated in maintaining the structural integrity of the protein. Characterization of recombinant ICOSL lacking the IgC domain reveals that the protein, in its transmembrane form as well as in its soluble form, shows marked tendency to form nonspecific aggregates with compromised binding ability for ICOS, suggesting that the presence of the IgC-like domain somehow blocks this nonspecific aggregation. Interestingly, replacement of ICOSL IgC domain with the B7-2 IgC domain could restore at least the wild-type binding capacity to ICOS (our unpublished observation). In conclusion, our data indicate that, although the membrane-distal IgV domain of ICOSL alone serves as the sole receptor binding unit, the presence of an additional membrane-proximal IgC domain is critical in maintaining the structural integrity and aggregation state of the protein required for its functional interaction with ICOS.

ICOSL has a tendency to form noncovalent oligomers, predominantly dimers, both in solution as well as on the cell surface. The specific oligomeric states of costimulatory molecules have been shown to be an important parameter in determining their function. For example, B7-1 and B7-2 adopt different oligomeric states in solution as well as on the cell surface, B7-1 being a dimer and B7-2 being a monomer; these differences in oligomeric states have been suggested to have functional implications in terms of generating T cell costimulatory responses through their interactions with CD28/CTLA-4 (14). Our data demonstrate that ICOSL behaves more like B7-1 by displaying considerably higher tendency to form noncovalently associated homodimers compared with B7-2 (32, 33). The marked dimerization/oligomerization potential of ICOSL in solution as well as on the cell surface suggests a working model in which the preformed homodimers of ICOSL on an APC could potentially interact with the obligate disulfide-linked dimers of ICOS on the apposing T cell. The detailed organization of the ICOSL will determine whether it is monovalent, bivalent, or can support an extended receptor:ligand complex as suggested for CTLA-4:B7 (28, 29). A more elaborate structural characterization of the ICOS:ICOSL complex will be required to determine the overall organization of this assembly and its potential constraints on signaling at the immunological synapse.

In summary, we have characterized the oligomerization property and the receptor recognition surface of ICOSL. Being a prominent member in the B7 family of costimulatory ligands, the molecular description of the structural features of ICOSL demonstrates significant similarities and differences with the other B7-homologs and suggests the structural basis of specific costimulatory interaction with its cognate receptor, ICOS.


   Acknowledgments
 
We thank Drs. T. DiLorenzo and N. Deb for help with the T cell costimulation experiment; Drs. T. DiLorenzo, F. Macian, D. Brims, and E. Lazar-Molnar for insightful discussions; E. Cao and Q. Yan for useful comments; and staff at the Analytical Imaging Facility and W. King at the FACS facility at AECOM for assistance.


   Disclosures
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Disclosures
References
 
The authors have no financial conflict of interest.


   Footnotes
 
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1 This work was supported by National Institute of Health Grants 5 R01 AI07289 (to S.G.N.) and 5 R01 DK065247 (to S.C.A.), Center for AIDS Research Grant P30AI051519, and a postdoctoral fellowship (to K.C.) from Cancer Research Institute Back

2 Address correspondence and reprint requests to Dr. Stanley G. Nathenson or Dr. Steven C. Almo, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY 10461. E-mail address: nathenso{at}aecom.yu.edu or almo{at}aecom.yu.edu Back

3 Abbreviations used in this paper: ICOS, inducible costimulator; ICOSL, ICOS ligand; FRET, fluorescence resonance energy transfer; pbFRET, photobleaching-based FRET; CFP, cyan fluorescent protein; YFP, yellow fluorescent protein; CHO, Chinese hamster ovary. Back

Received for publication April 18, 2006. Accepted for publication June 28, 2006.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Disclosures
 References
 

  1. Sharpe, A. H., G. J. Freeman. 2002. The B7-CD28 superfamily. Nat. Rev. Immunol. 2: 116-126. [Medline]
  2. Alegre, M. L., K. A. Frauwirth, C. B. Thompson. 2001. T cell regulation by CD28 and CTLA-4. Nat. Rev. Immunol. 1: 220-228. [Medline]
  3. Greenwald, R. J., G. J. Freeman, A. H. Sharpe. 2005. The B7-family revisited. Annu. Rev. Immunol. 23: 515-548. [Medline]
  4. Hutloff, A., A. M. Dittrich, K. C. Beier, B. Eljaschewitsch, R. Kraft, I. Anagnostopoulos, R. A. Kroczek. 1999. ICOS is an inducible T cell co-stimulator structurally and functionally related to CD28. Nature 397: 263-266. [Medline]
  5. Yoshinaga, S. K., J. S. Whoriskey, S. D. Khare, U. Sarmiento, J. Guo, T. Horan, G. Shih, M. Zhang, M. A. Coccia, T. Kohno, et al 1999. T cell co-stimulation through B7RP-1 and ICOS. Nature 402: 827-832. [Medline]
  6. Dong, C., A. E. Juedes, U. A. Temann, S. Shresta, J. P. Allison, N. H. Ruddle, R. A. Flavell. 2001. ICOS co-stimulatory receptor is essential for T cell activation and function. Nature 409: 97-101. [Medline]
  7. Tafuri, A., A. Shahinian, F. Bladt, S. K. Yoshinaga, M. Jordana, A. Wakeham, L. M. Boucher, D. Bouchard, V. S. F. Chan, G. Duncan, et al 2001. ICOS is essential for effective T helper cell responses. Nature 409: 105-109. [Medline]
  8. McAdam, A. J., R. J. Greenwald, M. A. Levin, T. Chernova, N. Malenkovich, V. Ling, G. J. Freeman, A. H. Sharpe. 2001. ICOS is critical for CD40-mediated antibody class switching. Nature 409: 102-105. [Medline]
  9. Nurieva, R. I., X. M. Mai, K. Forbush, M. J. Bevan, C. Dong. 2003. B7h is required for T cell activation, differentiation, and effector function. Proc. Natl. Acad. Sci. USA 100: 14163-14168. [Abstract/Free Full Text]
  10. Wang, S., G. Zhu, K. Tamada, L. Chen, J. Bajorath. 2002. Ligand binding sites of inducible costimulator and high avidity mutants with improved function. J. Exp. Med. 195: 1033-1041. [Abstract/Free Full Text]
  11. Peach, R. J., J. Bajorath, W. Brady, G. Leytze, J. Greene, J. Naemura, P. S. Linsley. 1994. Complementarity determining region 1 (CDR1)- and CDR3-analogous regions in CTLA-4 and CD28 determine the binding to B7-1. J. Exp. Med. 180: 2049-2058. [Abstract/Free Full Text]
  12. Wang, S., G. Zhu, A. I. Chapoval, H. Dong, K. Tamada, J. Ni, L. Chen. 2000. Costimulation of T cells by B7-H2, a B7-like molecule that binds ICOS. Blood 96: 2808-2813. [Abstract/Free Full Text]
  13. Khayyamian, S., A. Hutloff, K. Buchner, M. Grafe, V. Henn, R. A. Kroczek, H. W. Mages. 2002. ICOS-ligand, expressed on human endothelial cells, costimulates Th1 and Th2 cytokine secretion by memory CD4+ T cells. Proc. Natl. Acad. Sci. USA 99: 6198-6203. [Abstract/Free Full Text]
  14. Bhatia, S., M. Edidin, S. C. Almo, S. G. Nathenson. 2005. Different cell surface oligomeric states of B7-1 and B7-2: Implications for signaling. Proc. Natl. Acad. Sci. USA 102: 15569-15574. [Abstract/Free Full Text]
  15. Thompson, J. D., D. G. Higgins, T. J. Gibson. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22: 4673-4680. [Abstract/Free Full Text]
  16. Gouet, P., X. Robert, E. Courcelle. 2003. ESPript/ENDscript: extracting and rendering sequence and 3D information from atomic structures of proteins. Nucleic Acids Res. 31: 3320-3323. [Abstract/Free Full Text]
  17. Sali, A., T. L. Blundell. 1993. Comparative protein modelling by satisfaction of spatial restraints. J. Mol. Biol. 234: 779-815. [Medline]
  18. Bourne, P.E., K. J. Addess, W. F. Bluhm, L. Chen, N. Deshpande, Z. Feng, W. Fleri, R. Green, J. C. Merino-Ott, W. Townsend-Merino, et al 2004. The distribution and query systems of the RCSB Protein Data Bank. Nucleic Acids Res. 32: (database issue):D223-D225. [Abstract/Free Full Text]
  19. Altschul, S. F., T. L. Madden, A. A. Schaffer, J. Zhang, Z. Zhang, W. Miller, D. J. Lipman. 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25: 3389-3402. [Abstract/Free Full Text]
  20. Rai, B. K., A. Fiser. 2006. Multiple Mapping Method: a novel approach to the sequence-to-structure alignment problem in comparative protein structure modeling. Proteins 63: 644-661. [Medline]
  21. Kabsch, W., C. Sander. 1983. Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 22: 2577-2637. [Medline]
  22. Edidin, M.. 2002. Fluorescence resonance energy transfer: techniques for measuring molecular conformation and molecular proximity. J. E. Coligan, and A. M. Cruisbeek, and D. H. Margulies, and E. M. Shevach, and W. Strober, eds. In Current Protocols in Immunology 18.10.1-18.10.18. Wiley, Hoboken.
  23. Kenworthy, A. K., M. Edidin. 1998. Distributions of a GPI-anchored protein at the apical surface of MDCK cells examined at a resolution of <100Å using imaging fluorescence resonance energy transfer. J. Cell Biol. 142: 69-84. [Abstract/Free Full Text]
  24. Pentcheva, T., M. Edidin. 2001. Clustering of peptide-loaded MHC class I molecules for ER export imaged by fluorescence resonance energy transfer. J. Immunol. 166: 6625-6632. [Abstract/Free Full Text]
  25. Adair, B. D., D. M. Engelman. 1994. Glycophorin A helical transmembrane domains dimerize in phospholipid bilayers: a resonance energy transfer study. Biochemistry 33: 5539-5544. [Medline]
  26. Li, E., M. You, K. Hristova. 2005. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and forster resonance energy transfer suggest weak interactions between fibroblast growth factor receptor 3 (FGFR3) transmembrane domains in the absence of extracellular domains and ligands. Biochemistry 44: 352-360. [Medline]
  27. Majeti, R., Z. Xu, T. G. Parslow, J. L. Olson, D. I. Daikh, N. Killeen, A. Weiss. 2000. An inactivating point mutation in the inhibitory wedge of CD45 causes lymphoproliferation and autoimmunity. Cell 103: 1059-1070. [Medline]
  28. Stamper, C. C., Y. Zhang, J. F. Tobin, D. V. Erbe, S. Ikemizu, S. J. Davis, M. L. Stahl, J. Seehra, W. S. Somers, L. Mosyak. 2001. Crystal structure of the B7-1/CTLA-4 complex that inhibits human immune responses. Nature 410: 608-611. [Medline]
  29. Schwartz, J. C., X. Zhang, A. A. Fedorov, S. G. Nathenson, S. C. Almo. 2001. Structural basis for co-stimulation by the human CTLA-4/B7-2 complex. Nature 410: 604-608. [Medline]
  30. Peach, R. J., J. Bajorath, J. Naemura, G. Leytze, J. Greene, A. Aruffo, P. S. Linsley. 1995. Both extracellular immunoglobulin-like domains of CD80 contain residues critical for binding T cell surface receptors CTLA-4 and CD28. J. Biol. Chem. 270: 21181-21187. [Abstract/Free Full Text]
  31. Fargeas, C. A., A. Truneh, M. Reddy, M. Hurle, R. Sweet, R. P. Sekaly. 1995. Identification of residues in the V domain of CD80 (B7-1) implicated in functional interactions with CD28 and CTLA4. J. Exp. Med. 182: 667-675. [Abstract/Free Full Text]
  32. Ikemizu, S., R. J. C. Gilbert, J. A. Fennelly, A. V. Collins, K. Harlos, E. Y. Jones, D. I. Stuart, S. J. Davis. 2000. Structure and dimerization of a soluble form of B7-1. Immunity 12: 51-60. [Medline]
  33. Zhang, X., J. C. Schwartz, S. C. Almo, S. G. Nathenson. 2003. Crystal structure of the receptor binding domain of human B7-2: insights into organization and signaling. Proc. Natl. Acad. Sci. USA 100: 2586-2591. [Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 Proc. Natl. Acad. Sci. USAHome page
K. Chattopadhyay, U. A. Ramagopal, M. Brenowitz, S. G. Nathenson, and S. C. Almo
Evolution of GITRL immune function: Murine GITRL exhibits unique structural and biochemical properties within the TNF superfamily
PNAS, January 15, 2008; 105(2): 635 - 640.
[Abstract] [Full Text] [PDF]

Home page
 Protein Sci.Home page
E. Feyfant, A. Sali, and A. Fiser
Modeling mutations in protein structures
Protein Sci., September 1, 2007; 16(9): 2030 - 2041.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Chattopadhyay, K.
Right arrow Articles by Nathenson, S. G.