Abstract
Humans express four MHC-like CD1 molecules CD1a, b, c and d that are capable of presenting a wide variety of self or foreign lipid antigens to T cells. Much progress has been made in elucidating the function of CD1d-restricted NKT cells in both innate and adaptive immune responses. However, knowledge of the other CD1 molecules is less well defined in terms of lipid presentation and immune regulation. We have previously shown that immunoglobulin-like transcript 4 (ILT4) binds to CD1d and inhibits its recognition by NKT cells. In this study, we show that CD1c can also interact specifically with ILT4 with a higher affinity than that of CD1d. Furthermore, changes in CD1c expression seem to modulate CD1d function; up-regulation of CD1c enhances NKT recognition of CD1d and down-regulation reduces CD1d recognition. We propose that CD1c can act as a sink for the inhibitory receptor ILT4: when CD1c is up-regulated, ILT4 is recruited to CD1c, thus reducing the inhibitory effect of ILT4 on CD1d recognition. Consequently, CD1c could be a potential target for modulating NKT activity.Keywords: NKT, CD1d, CD1c, ILT4, antigen presentation
Introduction
The human CD1 family has four functional isoforms, namely CD1a, b, c and d, whereas in the mouse, there is only one member and it is most related to human CD1d. They form a group of cell surface expressed glycoproteins that represent a third lineage of antigen-presenting molecules in addition to MHC-I and MHC-II (1). Structurally, CD1 proteins share a number of features with MHC-I molecules in that their heavy chains are non-covalently associated with β2 microglobulin (β2m) (1, 2) and an antigen-binding groove is formed between the α1 and α2 domains of their heavy chains. In contrast to MHC-I and MHC-II, which bind short peptide antigens, the antigen-binding surface of CD1 molecules is extended and hydrophobic, giving CD1 the ability to bind and present a variety of lipid antigens (3–6).
The lipid antigens presented by CD1 molecules include self-lipids, such as the ganglioside GM1 (7), phospholipids (8) and isoglobotrihexosylceramide (9), and exogenous antigens, including mycobacterial glucomonomycolates, Sphingomonas glycosphingolipids (10), Borrelia burgdorferi-derived diacylglycerol glycolipids (11) and the synthetic antigen α-Galactosylceramide (αGalCer). All these lipid molecules share similar structures with two alkyl chains binding to the two major pockets in the CD1 antigen-binding grooves and their sugar heads pointing upwards to be recognized by TCRs.
Among the four CD1 molecules, CD1d has been well characterized and drawn the most attention in recent studies. CD1d-restricted T cells, termed NKT cells, are suggested to have important roles in autoimmunity and host defence, and measures to enhance NKT function have been demonstrated to be beneficial to treating infectious diseases and tumours (12, 13).
We have previously reported that ILT4, an inhibitory receptor expressed by antigen-presenting cells (APCs), was able to interact with and inhibit CD1d antigen presentation (14). ILT4 belongs to the immunoglobulin-like transcript (ILT)/leukocyte immunoglobulin-like receptor family. This family consists of more than 10 genes clustered within the leukocyte receptor complex (LRC) that is situated on the long arm of human chromosome 19 (15, 16). With the exception of ILT6, which lacks a transmembrane anchor and is soluble, the majority of ILT family members are type I transmembrane proteins expressed mainly on macrophages/monocytes and dendritic cells (DCs). They are further classified as activating or inhibitory receptors based on the length of their cytoplasmic regions. The members with short cytoplasmic tails are normally associated with activating transmembrane adaptor protein FcRγ and transduce stimulatory signals. Other members, such as ILT4, have immunoreceptor tyrosine-based inhibitory motif within their long cytoplasmic regions and have been found to be inhibitory (16). ILT4, along with other inhibitory ILT family members, has been found to recognize HLA-A, -B and -C alleles and atypical MHC-I such as HLA-E, -F and -G, thus having potentially important immunomodulatory roles in both the innate and adaptive immune systems (17–20).
Considering the homology among CD1 family members, our previous finding that ILT4 interacts with CD1d prompted us to investigate further its interaction with CD1 family proteins. In these studies, we show that ILT4 additionally interacts with CD1c with greater affinity than CD1d. We propose that this up-regulation provides a ‘sink’ for ILT4, thus relieving its inhibitory effect on CD1d antigen presentation to NKT cells; targeting CD1c may provide an additional approach for modulating NKT activity.
Methods
mAbs and recombinant proteins
CD1c mAb (clone L161, Serotec, UK), CD1d mAb (CD1d42, BD Pharmingen, UK) and anti-FLAG M2 mAb (Sigma, UK) were purchased commercially. ILT4X13 was generated by conventional hybridoma technology by immunizing BALB/c mice with an ILT4-Fc fusion protein. ILT4-Fc, ILT4D1D2, HLA-A*0201 and CD1d have been described previously (14), and CD1c protein and tetramer were generated in a similar way to that of CD1d (21). Briefly, cDNA encoding the extracellular region of human CD1c was linked to that of human β2m through a (GlySer)6 linker, which was cloned into a lentiviral vector to generate stable cell lines that secrete β2m-CD1c-human IgG1Fc fusion protein. β2m-CD1c was subsequently released from the fusion protein by enzymatic digestion and biotinylated to generate tetramers.
Cells
Fresh human PBMCs were isolated from whole blood of healthy donors by Lymphoprep (Nycomed Pharma, Switzerland) density gradient centrifugation. Monocytes were purified with human CD14 MACS MicroBeads (Miltenyi Biotec, UK) following the manufacturer’s protocol. DCs were generated by inducing purified monocytes with IL-4 (100ng/ml) and granulocyte macrophage colony-stimulating factor (100ng/ml) (both from Peprotech EC, UK) for 6 days. NKT cell lines and clones have been described previously (14) and were expanded with irradiated mixed allo-PBMC from three donors with periodic addition of IL-2 (200 IU/ml).
Flow cytometry
All flow cytometry (FACS) staining was conducted in 50 μl FACS washing buffer (PBS with 2% FCS and 0.02% azide). After the addition of antibodies, cell suspensions were incubated on ice for 30min, then washed twice with washing buffer before fixation with 1% paraformaldehyde/PBS. For tetramer staining, cells were incubated with tetramers at a final concentration of 10 μg/ml at 37°C for 30min before staining with other fluorescently labelled mAbs on ice. For blocking experiments, blocking reagents were added at the indicated concentrations to the cell suspension before addition of tetramers. FACS data acquisition and analysis were performed on a FACSCalibur flow cytometer using CellQuest software (BD Biosciences), and data were analysed using Flowjo software (Tree Star, CA, USA).
mAb branching assay
Jurkat cells were incubated with CD1c mAb at 10 μg/ml for 30min at 4°C, then washed with FACS washing buffer, followed by PE-conjugated CD1c tetramer incubation at 10 μg/ml for 30min at 37°C. Cells were then washed and analysed by FACS.
NKT stimulation
Jurkat or THP-1 cells were pulsed with 100ng/ml of αGalCer (AXXORA, Switzerland) for 1h in a 37°C CO2 incubator, then the cells were washed with RPMI1640-10% FCS and seeded into 96-well plates at 5×104 per well. Equal numbers of NKT cells were added and the cells were cultured for a further 16h. The culture supernatant was collected and IFN-γ and IL-4 concentrations were measured by ELISA (R&D Systems, UK).
RNA interference
ON-TRAGETplus SMARTpool siRNA for CD1c and ILT4 were purchased from Thermo Scientific, UK. FlexiTube GeneSolution for CD1c siRNA was purchased from Qiagen, UK. Next, siRNA were transfected into DCs using an Amaxa Nucleofector with the human dendritic cell Nucleofector kit following the manufacturer’s protocol VPA-1004 (Lonza, Switzerland).
Immunoprecipitation
Streptavidin-dynabeads (Dynal, UK) were coated with biotinylated CD1c monomer at room temperature for 30min. The beads were then washed with PBS and incubated with supernatants of HEK293T-ILT2-GFP or HEK293T-ILT4-GFP stable cell lines lysed in Nonidet overnight at 4°C. The beads were then washed extensively before boiling in loading buffer and subjected to SDS-PAGE. Proteins were transferred to Hybond-C membrane (GE Healthcare, UK) in a Trans-Blot SD Semidry Transfer Cell (Bio-Rad, UK) and western blotting was conducted with HRP-labelled anti-hu-IgG-Fc antibody; blots were visualized using an ECL kit (GE Healthcare).
Surface plasmon resonance
Surface plasmon resonance (SPR) was performed on a BIAcore 3000 system (GE Healthcare Bioscience, NJ, USA) at 25°C. Biotinylated HLA-A*0201, CD1c, CD1d or biotin (as a control) was immobilized on a streptavidin sensor chip. ILT4D1D2 was freshly purified and serially diluted with the running buffer (10mM HEPES, pH 7.4, 150mM NaCl, 3.4mM EDTA, 0.005% surfactant P20) before injection into the flow cell at a flow rate of 30 μl/min for 2 min. Data were analysed by using BIAevaluation 4.1 (GE Healthcare Bioscience), and the equilibrium binding constants (KD) were calculated with the steady-state analysis model.
Statistical analysis
Student’s t-test was used to test the significance between experimental groups.
Results
ILT4 is a ligand for CD1c
We have recently reported that a member of the leukocyte receptor family, ILT4, is a ligand for CD1d and that the interaction between CD1d and ILT4 inhibits NKT cell activation (14). Because ILT4 and other members of the family, such as ILT2, have previously been shown to bind other MHC-I molecules including HLA-A2, HLA-B27 and HLA-G (19, 22, 23), we set out to investigate whether such interactions also occurred between other members of CD1 family and ILT family. We generated a CD1c tetramer by a lentivirus-mediated mammalian expression system from HEK293T cells using a single chain construct in which the CD1c heavy chain is covalently linked by a glycine serine linker to β2M. After protein purification, CD1c monomers were biotinylated and tetramerized with fluorescently labelled streptavidin.
Before performing any binding test, we checked the conformation of the expressed single chain β2m-CD1c using an antibody-branching assay. In the assay, CD1c-expressing cells were stained with a CD1c-specific mAb followed by CD1c tetramer staining. The rationale was that a proportion of the bound CD1c mAb would only use one of their two antigen-binding arms to interact with surface CD1c; therefore, their free arms would be available to pick up subsequently added CD1c tetramer if the tetramer had the correct conformation. Jurkat cells, which were shown to be positive for CD1c protein expression, were used in this assay, and CD1c tetramer showed positive staining, confirming that β2m-CD1c protein was correctly folded during expression (Fig. 1A).
ILT4 is a receptor for CD1c protein. (A) CD1c tetramer validation. Jurkat cells were stained with CD1c mAb, followed by PE-conjugated CD1c tetramer (filled histogram) or anti-mouse secondary mAb (open solid line histogram). The Isotype control is shown as a dotted line histogram. (B) PE-conjugated CD1c and CD1d tetramers were used to stain HEK293T cells stably transfected with ILT2-GFP or ILT4-GFP. Streptavidin-PE was used as a control. (C) Blocking of CD1c tetramer binding to surface-expressed ILT4. HEK293T-ILT4-GFP cells were incubated with blocking reagents at the indicated concentrations, then stained with PE-conjugated CD1c tetramer and analysed by FACS. Anti-FLAG M2 mAb (IgG1) was used as the control antibody (left panel), and DR5-Fc protein was used as control protein (right panel). Inhibition was calculated as the percentage of reduction of MFI compared with that of the control staining. Figures show one of the three repetitive experiments. (D) CD1c blocks ILT4 recognition by its mAb. ILT4-specific mAb (ILT4X13) was used to stain HEK293T-ILT4-GFP in the presence of blocking reagents at the indicated concentrations. DR5-Fc was used as a control. (E) CD1c co-immunoprecipitation with ILT4. HEK293T-ILT2/ILT4-GFP cell lysate was immunoprecipitated by CD1d-Dynabeads (lanes 1 and 2) and detected by anti-GFP mAb. Cell lysate was added to lanes 3 and 4 as controls.
These verified CD1c tetramers were then used to stain cells transfected with ILT family expression constructs to investigate whether CD1c could recognize cell surface-expressed ILT proteins. CD1c tetramers behaved similarly to CD1d tetramers, staining only ILT4-, but not ILT2-, transfected HEK293T cells (Fig. 1b). This binding could be blocked by ILT4 mAb and CD1c soluble protein in a dose-dependent manner (Fig. 1c). Similarly, CD1c mAb presence in the solution could also sequester CD1c tetramer from binding the cells (Fig. 1c). In addition, CD1c protein blocked ILT4 mAb staining of ILT4-transfected cells (Fig. 1d), proving that the CD1c/ILT4 interaction was specific. Neither of the control blocking reagents, anti-FLAG mAb (control mAb in Fig. 1c left panel) or DR5-Fc (control in Fig. 1c right panel), an Fc fusion protein produced in a similar system to that of CD1c and CD1d, showed any blocking effect. Interestingly, CD1d also blocked CD1c binding to the ILT4 transfectants (Fig. 1c, right panel), suggesting that the binding sites of CD1c on ILT4 overlapped with that of CD1d. The direct interaction between CD1c and ILT4 was again confirmed by co-immunoprecipitation where soluble CD1c could precipitate ILT4 but not ILT2 (Fig. 1E).
Having established that CD1c bound to ILT4 transfectants, we next wanted to investigate whether CD1c could recognize naturally expressed ILT4 protein. ILT4 is normally expressed on myeloid cells including monocytes, macrophages and DCs (16). We used human PBMCs as a source of monocytes for investigation and found that CD1c tetramer positively stained monocytes and this staining was much stronger than that of CD1d tetramer (Fig. 2A). Both CD1c and ILT4 soluble protein could block this binding does-dependently. An ILT4-specific mAb blocked around 40% of the binding (Fig. 2b), suggesting that there may be additional receptors for CD1c on monocytes.
CD1c binds to naturally expressed ILT4. (a) Human PBMC were stained with FITC-CD14 and PE-CD1c tetramer, PE-CD1d tetramer or PE-Streptavidin. Monocytes were gated on larger, more granular cells comparing with lymphocytes using FSC/SSC. (b) CD1c binds monocytes through ILT4. PBMC were stained with CD1c tetramer in the presence of blocking antibodies (left panel) or proteins (right panel) at the indicated concentrations. Anti-FLAG M2 mAb was used as control antibody (left panel), and DR5-Fc was used an control protein (right panel).
CD1c competes with CD1d for ILT4 binding
We previously showed that ILT4 interacted with CD1d via its two N-terminal Ig domains (14). The fact that CD1d blocked CD1c binding to ILT4 transfectants (Fig. 1C) suggested that it was also likely to be the case for CD1c. To prove this, we transfected HEK293T cells with wild-type or mutated ILT4 constructs in which the N-terminal or C-terminal two Ig domains (D1D2 or D3D4) of ILT4 were replaced with those from ILT2, respectively. When CD1c tetramers were used to stain the transfectants, it was found that the removal of ILT4D1D2 abolished staining, confirming that CD1c also binds to ILT4 via its first two Ig domains (Fig. 3A). Consistent with this, ILT4D1D2 soluble protein was able to block CD1c tetramer binding to ILT4-expressing cells although it was less efficient compared with full-length ILT4 protein (ILT4-Fc) (Fig. 3B).
CD1c competes with CD1d for ILT4 binding. (a) CD1c binds to the first 2 Ig domains of ILT4. HEK293T cells were transiently transfected with GFP fusion constructs for wild-type (wt), D1D2-mutated or D3D4-mutated ILT4. The cells were then stained with PE-conjugated CD1c tetramer. (b) ILT4 soluble proteins block CD1c tetramer staining of ILT4-transfected cells. HEK293T-ILT4-GFP stable cells were stained with PE-conjugated CD1c tetramer in the presence of blocking reagents at the indicated concentrations. (c) Sensorgrams of ILT4D1D2 binding to CD1d and CD1c. Biotinylated CD1d and CD1c were coupled to the surfaces of SA biosensor chips and increasing concentrations (6.875, 13.75, 27.5, 55, 110 and 220 μM) of ILT4D1D2 were sequentially injected over each surface. The zero (0) time point corresponds to the start of the injection of the soluble analyte. Background binding to a biotin-coupled surface was subtracted. Calculated KD values based on a simple monovalent interaction model were determined from steady-state evaluation of global curve fits using BIAevaluation 4.1. (d) Plots of the equilibrium binding responses of ILT4D1D2 versus concentration. Open circle, CD1d; solid circle,CD1c. Insets: scatchard plots of the same data. (e) CD1c competes with CD1d for ILT4 binding. Rat basophilic leukaemia cell line RBL stably expressing ILT4 (RBL-ILT4) (left panel) or monocytes (right panel) were stained with PE-conjugated native CD1d tetramer in the presence of CD1c, CD1d or control protein at the indication concentrations. DR5-Fc was used as a control.
SPR was performed to further characterize the interaction between CD1c and ILT4. Biotinylated CD1c, along with control proteins CD1d and HLA-A*0201, were immobilized through streptavidin on sensor chips. Soluble ILT4D1D2 protein was injected over the sensor surfaces at various concentrations and binding affinity was measured by equilibrium binding analysis. ILT4D1D2 showed fast kinetics with rapid binding and dissociation with both CD1c and CD1d (Fig. 3C). Fitting of conventional and Scatchard plots of ILT4D1D2 binding data confirmed the binding to be monophasic and CD1c had higher affinity than CD1d (KD = 149 μM and 243 μM, respectively) (Fig. 3d). In agreement with this, CD1c protein blocked CD1d tetramer binding to both transfected and naturally expressed ILT4 more efficiently than CD1d itself (Fig. 3e).
CD1c enhances CD1d function through binding ILT4
As a group 1 CD1 family member, which also include CD1a and CD1b, CD1c is mainly expressed on APCs, which are also crucial for NKT cell function. The fact that there was a triangular interaction between CD1c, CD1d and ILT4 raised the interesting question: could the up-regulation of CD1c affect the function of APC-expressed CD1d through its interaction with their common binding partner ILT4?
Jurkat cells express both CD1c and CD1d, but not ILT4, so to create an artificial APC, we stably transduced Jurkat with an ILT4-encoding lentivirus (Columns 1 and 2, Fig. 4a). When we enhanced CD1c expression in these cells by CD1c-encoding lentivirus transduction, we found that the level of ILT4 expression was significantly reduced with the overall mean fluorescence intensity (MFI) (gated on all live cells) dropping from 90.2 to 46.0. In the FACS histogram of ILT4 mAb staining, a negative population appeared in addition to the positive population that had reduced levels of ILT4 expression. Meanwhile, the level of CD1d remained unchanged (Row 2, Fig. 4a). This experiment was repeated thrice and all showed similar results.
CD1c over-expression down-regulates ILT4 and enhances CD1d function. (a) Jurkat cells and ILT4 stably transfected Jurkat cells (Jurkat-ILT4) were transduced with CD1c-expressing lentivirus, then stained with CD1c, CD1d or ILT4 mAbs in parallel with parental cells. Results from one of the three repetitive experiments are shown. (b) THP-1 cells and ILT4-GFP or CD1c-GFP stably expressing THP-1 cells were stained with CD1c, CD1d or ILT4 mAbs. Reduction of MFI was calculated between GFP+ and GFP− populations. Results from one of the four repetitive experiments were shown. (c) Jurkat cells or Jurkat cells stably expressing ILT4, CD1c or ILT4 and CD1c were pulsed with αGalCer (100ng/ml), then were incubated with NKT cells for 16h. IFN-γ in the culture supernatant was measured by ELISA (left panel). THP-1 cells and their derivative cell lines were similarly pulsed and used as APCs to stimulate NKT cells (right panel). The Experiment was repeated with four different NKT cell lines/clones three times; one representative result is shown. *P < 0.5, **P < 0.01 as compared with parental cell line (left panel) or as indicated (right panel). n = 3.
We supposed this down-regulation of ILT4 expression was caused by CD1c’s binding to ILT4 because we had shown that ILT4 mAb binding was inhibited by CD1c (Fig. 1d), but there was a possibility that in this particular experiment CD1c blocked ILT4 synthesis by competitive use of the same promoter that both CD1c and ILT4 shared in their lentiviral constructs. To exclude this possibility, we next used another cell line THP-1, which constitutively express endogenous CD1c, CD1d and ILT4, to do the same experiment. The cells were transduced with CD1c-GFP fusion construct at a virus titre that gave around 50% transduction efficiency, so that two cell populations could be detected when CD1c-GFP expression was measured: the transduced population (GFP positive) and the un-transduced population (GFP negative). The levels of endogenous ILT4 expression on the transduced population had an MFI of 87.2, and the un-transduced population MFI = 162.0, suggesting a 46% reduction (Fig. 4b). Combining the results from both cell lines, we were able to draw the conclusion that over-expression of CD1c could reduce the accessibility of ILT4 by other molecules, such as ILT4 mAb in this case, through their direct interaction.
In both cases, the changes to ILT4 surface expression did not seem to cause any change in CD1d expression. Because we have proved that ILT4 could inhibit CD1d function through direct interaction (14), we wanted to see whether this reduction of ILT4 expression caused by CD1c over-expression would cause any change in CD1d function. The above-mentioned Jurkat cells and THP-1 cells, along with their ILT4/CD1c transfectants, were pulsed with αGalCer and used to stimulate NKT cells. In both cell types, ILT4 over-expression caused significant reduction in IFN-γ secretion by NKT cells (Fig. 4c). The additional introduction of CD1c on ILT4-expressing cells (Jurkat-ILT4-CD1c and THP1-CD1c) reversed this reduction substantially, but CD1c over-expression on non-ILT4 expressing cells (Jurkat-CD1c) caused only a very modest (albeit significant) change to NKT activation (Fig. 4c). These results suggested a model in which CD1c up-regulation could, by recruiting ILT4, reduce ILT4’s inhibitory effect on CD1d function.
Having found that CD1c enhanced CD1d function in immortalized cell lines, we next wanted to test whether this also applied to professional APCs such as DCs. Mature DCs express high levels of CD1c and ILT4 (14, 24). We used RNAi technique to knock down the expression of CD1c or ILT4 on DCs, then pulsed these cells with αGalCer to stimulate NKT cells. As shown in Fig. 5a, ILT4 expression was almost completely knocked down, whereas CD1c expression showed a significant but incomplete reduction by RNAi. Similar to the results we observed in Jurkat and THP-1 cell over-expression experiments (Fig. 4a and b), the levels of CD1c and ILT4 did not show significant effect on CD1d cell surface expression. These modified DCs were then pulsed with αGalCer and used to stimulate NKT cells. ILT4 knock-down as expected enhanced the response of NKT cells, whereas the partial knock down of CD1c expression nevertheless has a substantial inhibitory effect on NKT recognition of αGalCer measured by IFN-γ or IL-4 secretion (Fig. 5b). These results confirmed that CD1c could modulate CD1d function via the intermediary of ILT4.
CD1c enhances CD1d function through ILT4 in DCs. (A) ILT4 and CD1c knock-down by siRNA. DCs were transfected with CD1c or ILT4 siRNA, and protein expression was analysed by mAb staining 36h later. One of the three representative results is shown. (B) Cytokine secretion by CD1c/ILT4 siRNA-treated DCs. DCs transfected with CD1c or ILT4 siRNA were cultured for 36h in the presence or absence of 10ng/ml αGalCer. Cells then were washed and incubated with NKT cells for another 24h, and IFN-γ and IL-4 levels in the culture supernatant were measured by ELISA. *P < 0.5, **P < 0.01. n = 3.
Discussion
Most mammals have a multiple CD1 family proteins and it is suggested that the different CD1 molecules coordinate with each other by surveying different parts of the endocytic systems and present differentially distributed lipid, glycolipid or lipopeptide antigens (25) . CD1d mainly travels to the late endosomes and lysosomes and presents a wide range of endogenous and exogenous antigens that are processed there, whereas CD1c preferentially routes to the early endosomes and presents lipopeptides or mycobacteria-derived glycolipids (26–28). Different members of the CD1 family present antigens to different subsets of T cells. CD1d presents antigens to T cells with invariant chain TCR, and CD1c restricts common αβ T cells or γδ T cells (25). Up to now, there has not been any strong evidence suggesting any direct or indirect interaction between CD1c and CD1d at the protein level although they are frequently found to be co-expressed on the surface of various kinds of APCs (29). Our finding here that they have a common partner ILT4 not only provides an additional link between them but also opens up new potential for modulating NKT functions.
There are clear similarities between CD1c and CD1d’s interaction with ILT4. Both of them specifically recognize the two N-terminal Ig domains of ILT4 but not that of a close relative ILT2. This is in contrast to that of classical HLA class I molecules that bind both ILT2 and ILT4 (22, 30). The binding affinities of CD1c and CD1d with ILT4 are similar; both are much weaker than that of classical HLA class I molecules, as the latter have KD values of 4.8–45 μM (22). When CD1c and CD1d are compared, CD1c has stronger binding than CD1d (KD = 149 versus 243 μM). Although this difference seems marginal in number terms, when the proteins are tetramerized, CD1c tetramer is much more efficient in blocking both CD1c and CD1d binding to ILT4 (Fig. 1c and unpublished data), suggesting a much higher avidity when protein is clustered, which is probably the case on cell surface.
Our unexpected finding that CD1c over-expression down-regulates ILT4 surface expression (Fig. 4a and b) suggests that this may have functional implications. Both molecules interact with ILT4; if the stronger binder can make the common binding partner inaccessible by the other, then it will have an impact on the latter’s function. Subsequent functional assays proved that the down-regulation of ILT4 could affect stimulation of CD1d-restricted NKT cells both in transduced cell lines and in DCs. Therefore, a reasonable scenario seems to be as follows: when CD1c expression is low, ILT4 is more associated with CD1d, therefore APC cells have a low efficiency in stimulating NKT cells; whereas when CD1c is up-regulated, such as when DCs are activated, CD1c, which binds more strongly to ILT4 than CD1d, will compete for ILT4 binding and free CD1d molecules to present lipid antigens to T cells. These observations provide a novel insight into APC function and may in the future open up an additional route to manipulate APC function to regulate T-cell responses and vaccine efficacy.
Another possible explanation to CD1c-mediated enhancement of CD1d functioning could be through intracellular signaling. ILT4 is an inhibitory receptor and has been found to be able to transduce negative signals that could inhibit early signaling events triggered by stimulatory receptors (31). Although the direct link between MHC-I binding to signaling pathways has yet to be fully established, it is entirely possible that CD1d binding to surface ILT4 could result in reducing APC functioning, whereas CD1c is capable of overriding this inhibitory effect. Further elucidation of the relationship between surface events and signaling cascades will provide better understanding of this scenario.
In this study, we proposed that interaction between CD1 and ILT4 occurs in cis. This is in agreement with similar studies in which murine ILT4 homologue PIR-B was shown to bind MHC-I molecules on the same cell surface (32, 33). However, past studies on interactions between ILT4 and HLA class I molecules all suggested that the interactions were more likely to happen in trans, i.e. the receptor and ligand are expressed by different cells (30, 34). At present, we do not have sufficient evidence to exclude one or the other, but the data shown here seem to indicate that cis-interaction does occur. Further investigation may shed light on the exact mechanism of these interactions.
In conclusion, we have identified a novel interaction between CD1c and ILT4; exploiting this could open new avenues for modulating NKT cell function.
FundingUK Medical Research Council; National Natural Science Foundation of China (NSFC) Innovative Research Group (Grant No. 81021003); NSFC Project grant (Grant No. 81172792).
References
