NK T (NKT) cells are an important component of the innate immune system and recognize the MHC class I-like CD1d molecule. NKT cells possess significant immunoregulatory activity due to their rapid secretion of large quantities of pro- and anti-inflammatory cytokines following CD1d-dependent stimulation. Because the innate immune system is programmed to respond to a multitude of diverse stimuli and must be able to quickly differentiate between pathogenic and endogenous signals, we hypothesized that, apart from stimulation via the TCR (e.g., CD1d-dependent activation), there must be multiple activation pathways that can be triggered through other cell surface receptors on NKT cells. Therefore, we analyzed the ability of CD44, a structurally diverse cell surface receptor expressed on most cells, to stimulate murine NKT cells, compared with conventional T cells. Stimulation of CD44 through Ab cross-linking or binding to its natural ligands hyaluronan and osteopontin induced NKT cells to secrete cytokines, up-regulate activation markers, undergo morphological changes, and resist activation-induced cell death, whereas conventional T cells only exhibited changes in morphology and protection from activation-induced cell death. This CD44-specific stimulation of NKT cells correlated with their ability to bind hyaluronan. Thus, fundamental differences in CD44 function between these lymphocyte subsets suggest an important biological role for CD44 in the innate immune response.
Natural killer T cells (NKT)3 are a component of the innate immune system with significant immunoregulatory capacity as characterized by their rapid secretion of large quantities of cytokines following stimulation. These cells were first described as a class of lymphocytes that exhibit characteristics of both NK and conventional T cells (1) and therefore, are likely important mediators between the innate and adaptive immune response. NKT cell subsets make up at least four populations (2), the most well-characterized being invariant NKT cells (iNKT). The TCR of these cells contain a Vα14-Jα18 TCRα chain rearrangement in mice (Vα24-Jα18 in humans) and recognize lipid Ags presented by the MHC class-I-like CD1d molecule (3). Classical CD1d-mediated activation of NKT cells induces the release of IFN-γ, IL-4, and GM-CSF, which regulates downstream immune function, and is quickly followed by either down-regulation of extracellular markers (4, 5) or activation-induced cell death (AICD) (6, 7) of the NKT cell itself as an apparent autoregulatory mechanism. Conversely, conventional T cells recognize protein-derived peptides presented in the context of MHC class I and II molecules and exhibit an extensive array of TCR combinations allowing for the extreme diversity and specificity that characterizes the adaptive immune response. Although NKT and classical T cells use similar TCR-mediated activation pathways, they maintain independent roles in the immune response and likely have developed additional alternative mechanisms of activation. These alternative activation pathways not only create diversity in the adaptive and innate immune response following stimulation, but also, if activated inappropriately, could cause immune dysfunction and contribute to disease.
CD44 is a structurally diverse cell surface receptor exhibiting a broad range of biological functions exemplified by its expression in most mammalian cells and tissues (8). The extracellular domain of CD44 binds to multiple glycosaminoglycans (GAGs), including hyaluronan (HA), collagen, laminin, and fibronectin, as well as the immunoregulatory cytokine osteopontin (OPN), resulting in the activation of intracellular signaling pathways. Although CD44 is expressed at high levels on many cell types, its function is highly regulated, because not all cells that express the receptor are capable of inducing downstream activation events. For this reason, three activation states of CD44 (inactive, inducible, and active) have been described based on the ability to bind FITC-labeled HA (FITC-HA) (9). The biological function of CD44 has been linked to lymphocyte migration and activation (10, 11, 12, 13), inflammation (14), apoptosis (15, 16) and differentiation (17), and CD44 expression is a marker of metastatic potential in certain tumors (18). How these apparently distinct functions arise from a single receptor molecule is complex and may be attributed to multiple mechanistic and structural variations including diversity in ligand binding, signaling pathways, glycosylation patterns, and the expression of variant exons in the extracellular portion of the gene that arise through alternative splicing. Interestingly, whereas it is the extracellular domain of CD44 that binds directly to its ligands, intracellular components such as kinases, proteases, lipids, and cytoskeletal proteins, also can influence ligand affinity and function (19, 20). Despite these findings and an active field of research, an understanding of the role that CD44 plays in regulating the immune response remains limited.
Because of the link between CD44 and lymphocyte activation and its relatively high level of cell surface expression on NKT cells (21), we attempted to determine whether CD44-mediated activation of NKT cells was distinct from that of conventional T cells. In this study, we show that stimulation of mouse CD44, through Ab cross-linking and natural ligand binding, induces NKT cells to secrete cytokines, up-regulate activation markers, undergo morphological changes, and resist AICD, whereas conventional T cells only exhibit changes in morphology and protection from AICD. This CD44-specific stimulation of NKT cells correlates with the ability to bind HA. Through the use of primary isolates and hybridoma-based NKT and conventional T cells, we describe fundamental differences in CD44 function between these lymphocyte subsets that define its biological significance in the innate immune response.
Materials and Methods
Female wild-type C57BL/6 and B6129SF2/J background CD44-deficient (CD44KO) and wild-type mice were purchased from The Jackson Laboratory and used at 6–8 wk of age for analyses of liver mononuclear cell (LMNC) stimulation as described below. All animal procedures were approved by the Indiana University School of Medicine’s Animal Care and Use Committee.
Reagents and Abs
Purified pan-CD44 Abs used for blocking studies, activation assays and indirect FACS staining were clones IM7 (BD Biosciences), KM201 (BioSource International), KM703 (American Type Culture Collection), and IRAWB14 (a gift from K. Hathcock) (22). The epitope specificity and effect on HA binding by these CD44-specific mAb have been described previously (23). FITC-conjugated rabbit anti-rat Ig secondary Ab (DakoCytomation) was used for indirect FACS staining. The CD3ε-specific mAb (145-2C11) was purchased from BD Biosciences. Abs used for FACS staining were anti-NK1.1-PE (PK136), anti-TCRβ-FITC, and anti-TCRβ-allophycocyanin (H57-597), anti-CD4-FITC (GK1.5), anti-CD8a (53-6.7), anti-CD44-FITC (IM7), anti-IFN-γ-allophycocyanin (XMG1.2), anti-IL-4-allophycocyanin (11B11), anti-CD69-PE (H1.2F3), and anti-CD71-PE (C2), all obtained from BD Biosciences; and anti-GM-CSF-FITC (MP1–22E9; Biolegend). Abs against individual CD44 variants were specific for the following: v4 (10D1), v6 (9A4), both obtained from Bender MedSystems), v7 (AB2083; Chemicon International), and v10 (LYK-1; ATCC). The rat IgG2b-κ isotype control (A95-1) was purchased from BD Biosciences. Recombinant mouse OPN (R&D Systems) and human umbilical cord-derived HA (Sigma-Aldrich) were used as natural CD44 ligands. FITC-HA (Sigma-Aldrich) was used to assess HA binding potential to CD44.
Cells, morphology, and activation assays
Mouse NKT cell hybridoma cell lines N37-1A12, N38-3C3, N38-2C12, N38-2H4, and N37-2H4 (24), all provided by K. Hayakawa (Fox Chase Cancer Center, Philadelphia, PA); DN32.D3 and 431.A11 (25) from A. Bendelac (University of Chicago, Chicago, IL), murine T cell hybridoma lines, DO11.10 (26) from J. Blum (Indiana University School of Medicine, Indianapolis, IN), and 3A9 (27) from G. Bishop (University of Iowa, Iowa City, IA) were used in CD44 stimulation assays. To assess stimulation and changes in morphology, NKT and T cell hybridomas (1 × 105 cells per well) and primary LMNCs (1 × 106 cells per well) were added in a total volume of 200 μl of IMDM supplemented with 5% FBS and penicillin/streptomycin (I+) to 96-well MaxiSorp plates (Nalge Nunc International) that had been previously coated with CD44-specific, CD3ε-specific, or isotype control Abs (100 ng per well) or the CD44 ligands HA (25 μg per well) and OPN (50 ng per well) or BSA (100 μg per well) as a negative control. Cells were allowed to culture for 3–24 h at 37°C. Morphological changes were assessed by viewing the cells by light microscopy using an inverted M200 tissue culture scope (Zeiss) and capturing images using an AxioCam (Zeiss). Activation parameters were assessed by using cell culture supernatants (cytokine ELISAs) or by direct cell surface analysis (e.g., activation markers CD69 and CD71) by flow cytometry.
Mouse LMNCs were harvested from wild-type (B6129SF2/J) and CD44KO mice as described previously (7) and plated in wells of a 96-well flat-bottom plate (1 × 106 cells per well) precoated with anti-CD3 (145-2C11) and the -pan-CD44-specific IM7 (or isotype control) mAb (100 ng per well) in a total volume of 200 μl of IMDM supplemented with 10% FBS and antibiotics for 48 h. Supernatants were harvested and tested for IL-4 production by ELISA.
Secretion of cytokines from hybridomas was measured by ELISA, as described previously (28), and used as an indicator of cellular activation. Cytokine-specific Ab pairs (purified and biotinylated) for IL-2, GM-CSF, IL-4, IFN-γ, IL-12, and IL-10 were purchased from BD Biosciences, and those for IL-13 were obtained from R&D Systems.
Isolation of primary LMNCs
Isolation of LMNCs was performed as described previously (7). Briefly, livers were removed from mice and placed in I+ medium on ice. They were then minced and sequentially pressed through stainless steel and nylon mesh screens. and the resulting homogenate was resuspended in PBS. Following centrifugation at 200 × g for 10 min at 4°C, cell pellets were resuspended in 30 ml of RPMI 1640 medium at room temperature and layered onto 50 ml of a 30% Percoll solution in a 250-ml conical centrifuge tube (Corning) and centrifuged at 2200 rpm for 15 min at room temperature with the rotor brake off. The resulting cell pellet containing the LMNCs was treated with a RBC lysis solution for 10 min at room temperature, washed three times in PBS, and resuspended in IMDM. Cell number and viability was determined by trypan blue dye exclusion.
CD1d–Ig dimer staining of iNKT cells
To detect iNKT cells (Vα14Jα18 TCR positive), recombinant soluble dimeric mouse CD1d–Ig fusion proteins (DimerXI; BD Biosciences) were used. The dimers were loaded overnight with a 40 M excess of α-galactosylceramide (α-GalCer) and labeled with a Zenon R-PE mouse IgG1 labeling kit (Molecular Probes) as described previously (29). LMNCs were stained with the α-GalCer-loaded CD1d1 dimer according to the manufacturer’s instructions for the detection of iNKT cells. Empty CD1d1 dimers were used as a negative control.
Intracellular cytokine staining and flow cytometry
Intracellular cytokine staining procedures were performed using the BD Cytofix/Cytoperm kit (BD Biosciences) according to the manufacturer’s specifications. Primary hepatic mononuclear cells (1 × 106 cells per well) were stimulated with plate-bound Abs specific for mouse CD44 (IM7 clone), CD3ε, or control Ab for 1–12 h in I+ medium supplemented with brefeldin A (1 μg/ml) to inhibit secretion of cytokines. Following stimulation, cells were removed from the plate and washed two times in PBS, stained for cell surface markers (anti-NK1.1 and anti-TCRβ), fixed and permeabilized, and stained again for intracellular cytokines (anti-IFN-γ, anti-IL-4 and anti-GM-CSF). Cells were washed three times and resuspended in HBSS/BSA before analysis by flow cytometry.
FITC-HA binding was performed as described previously (22). Briefly, 1 × 106 NKT and T cell hybridomas or primary LMNCs were incubated for 30 min at 4°C in the presence of 100 μl of RPMI 1640 medium containing 1.0 μg/ml FITC-HA (Sigma-Aldrich). Inhibition of FITC-HA binding was assessed on each cell type by pretreating with 100 μl of RPMI 1640 medium containing 10 μg/ml unlabeled HA (Sigma-Aldrich). Enhancement of HA binding was achieved by pretreating cells with IRAWB14 hybridoma cell culture supernatant (100 μl for 30 min at 4°C). All cells were washed three times in PBS supplemented with 10% bovine growth serum, fixed in 100 μl of 0.05% paraformaldehyde for 10 min at room temperature, and resuspended in HBSS with 5% BSA before analysis by flow cytometry using a FACSCalibur (BD Bioscience) with CellQuest (BD Biosciences) and WinMDI (J. Trotter, University of California, San Diego, CA) software.
RT-PCR analysis of CD44 variant expression
Mouse CD44 isotype-specific RT-PCR analysis has been described previously (30). Total RNA was isolated from NKT and T cell hybridomas using TRI-Reagent (Molecular Research Center) according to the manufacturer’s instructions. cDNA synthesis and RT-PCR followed the two-step RNA PCR protocols outlined in the GeneAmp Gold RNA PCR reagent kit (PE Biosystems). Briefly, for each sample hybridoma, 5 μg of total RNA was reverse transcribed with 30 U MultiScribe reverse transcriptase using oligo(dT)16 primers (1.25 μM) in a 40-μl reaction volume for 10 min at 25°C and 12 min at 42°C. Second-step PCR was performed using 3 μl of the cDNA mixture in 50-μl reactions containing 1.5 U AmpliTaq Gold DNA polymerase. Eleven PCR were set up on each hybridoma using a forward primer for exon 5 in all reactions and reverse primers specific for either individual variant exons (v1–v10) or exon 16 (standard isoform) to assess CD44 isoform expression. PCR cycling was performed on a GeneAmp thermocycler (PE Biosystems) with the following cycling parameters: 95°C for 10 min, 35 cycles at 94°C for 20 s, 58°C for 30 s, and 72°C for 1 min, followed by an 8-min 72°C extension step. Following amplification, 10 μl of the PCR products was analyzed by 1% agarose gel electrophoresis.
Conventional T and NKT cell hybridomas were lysed in 1% NP40 containing 10% glycerol, 50 mM Tris (pH 7.4), 150 mM NaCl, and 1 mM EDTA (pH 8), including complete protease inhibitor tablets (Roche). Cell lyates (200 μg of total protein) were resolved on a 12% SDS-PAGE gel and transferred to a polyvinylidene difluoride membrane (Millipore). The blot was processed using mAb specific for pan-CD44 (IM7) or the CD44 variant exons v4, v6, v7, or v10, and developed using chemiluminescence before exposure on film.
Hybridoma cell lines were plated (1 × 105 cells per well) in microwells coated with either anti-CD3ε plus rat IgG2b-κ isotype control (CD3ε plus isotype) or anti-CD3ε plus anti-CD44 (IM7) (CD3ε plus CD44), each at 100 ng per well, and lactate dehydrogenase (LDH) release into the culture supernatant was monitored after 18 h. The Cytotox96 kit (Promega) was used to quantitate LDH levels and determine the percentage of hybridoma cell lysis.
Data are expressed as mean values ± SD. All analyses for statistically significant differences were performed with Student’s paired t test. Values of p < 0.05 were considered significant.
CD44 is expressed on NKT and CD4+ T cell hybridomas
The use of lymphocyte-derived hybridoma cell lines and clones have long been used to identify and characterize functional roles for different cell populations of the immune response. For this reason, as well as the fact that primary NKT cells are difficult to isolate, composed of diverse populations, and multiple well-characterized NKT and T cell hybridoma clones are readily available, we set out to initially study the role of CD44 function by using NKT and T cell hybridomas. In preliminary experiments, we assessed multiple NKT and conventional T cell hybridomas, all generated by fusing purified primary mouse lymphocytes with the murine T cell lymphoma line BW5147 (Fig. 1), for their cell surface expression of CD44 by flow cytometry using four pan-anti-CD44 mAbs (IM7, KM201, KM703, and IRAWB14). These expression profiles demonstrate that all the hybridoma cell lines tested expressed substantial levels of cell surface CD44 receptor (Fig. 1) independent of the detection Ab used. These results are consistent with previous studies of CD44 expression using primary lymphocytes (31) or NKT (24) and T cell hybridomas (32).
Certain CD44-specific Abs induce NKT and CD4+ T cell hybridomas to undergo morphological changes
Because CD44 plays a role in adhesion and migration of lymphocytes, we first attempted to assess whether stimulation of CD44 on our NKT and T cell hybridomas could induce changes in cellular morphology. To determine this, each hybridoma was cultured for various periods of time in Ab-coated tissue culture wells and visualized by light microscopy to assess gross morphological changes. Indeed, most of the hybridomas appeared to adhere to and spread in wells coated with the anti-CD44 Ab IM7 between 3 and 24 h (Fig. 2, i, j, l– n, and p), whereas there were no observed alterations in cellular morphology in the wells coated with an isotype control Ab during the same duration (Fig. 2, a–h). Although all hybridomas were tested in these studies, only a few representatives are presented. The only exception to this was the NKT hybridoma line N38-3C3, which did not exhibit any changes in morphology (Fig. 2, k and o), although it expresses significant levels of CD44 on its cell surface (Fig. 1). However, somewhat to our surprise, the morphological changes were not observed when other pan-anti-CD44 Abs were used (data not shown), including one, KM703, which maps to the same region as IM7. This discrepancy may represent an enhanced specificity and/or novel recognition and function of the IM7 Ab. Interestingly, following a short 3-h culture on wells coated with both anti-CD3ε and anti-CD44 (IM7), N38-3C3 cells underwent similar morphological changes (Fig. 2,w), suggesting that this phenotype could be modulated upon stimulation of the TCR complex. The same effect was true for the other NKT and T cell hybridomas where an enhanced adhesion and spreading phenotype was observed in the presence of anti-CD3ε and anti-CD44 (Fig. 2, u, v, and x). In this case, the morphological changes could not be linked to anti-CD3ε alone, because samples run in tandem with anti-CD3ε and an isotype control in place of the anti-CD44 failed to induce the same effect (Fig. 2, q–t). In summary, most of the NKT and T cell hybridomas tested in this study underwent morphological alterations following CD44 cross-linking with the anti-CD44 IM7 Ab, and the effect was enhanced with simultaneous TCR-mediated stimulation.
Ab-specific cross-linking of CD44 activates NKT, but not CD4+, T cell hybridomas
Because alterations in cellular morphology are frequently associated with changes in cellular function (33, 34), we next wanted to explore whether CD44 stimulation resulted in the functional activation of the hybridomas. As with the morphology experiments, we used plate-bound CD44-specific (IM7) and isotype control Abs to assess cytokine release from the hybridomas as a primary measure of their activation state. Depending on the cytokine tested, CD44 cross-linking resulted in increased activation of most of the NKT cell hybridomas, whereas the conventional T cell hybridomas showed no activation over isotype control treatment (Fig. 3,A). Other cytokines, such as IL-4, IL-12, IFN-γ, and IL-10, also were tested but were not produced from these cells (data not shown). A time-course analysis of this CD44-mediated stimulation, as measured by IL-2 secretion, suggests that it begins between 3 and 6 h following CD44 cross-linking and reaches a peak between 12 and 24 h (Fig. 3,B). Similar time-course profiles were observed with IL-13 and GM-CSF (data not shown). However, it should be noted that cytokine levels observed following CD44 cross-linking in these cells were typically 5- to 10-fold lower than that with stimulation of the TCR complex using Abs directed against TCRβ or CD3ε (data not shown). This would suggest that these cells were able to temper the degree of response depending upon the stimulus, which would be an important characteristic for a cell involved in the innate immune response. Accompanying this increased cytokine secretion observed following stimulation of the CD44 receptor in the NKT cell hybridomas was a modest up-regulation of the cell surface activation markers CD69 and CD71 (data not shown). However, much like the changes in morphology observed with the pan-CD44 Ab clone IM7, these activation effects were not observed when we used other pan-CD44 Abs, suggesting a specialized recognition by the IM7 Ab clone in NKT cell activation (Fig. 3 C).
Natural CD44 ligands induce activation of NKT cell hybridomas
Next, it was necessary to rule out the possibility that the effects we observed were merely a phenomenon associated with the IM7 Ab clone. To test this, we used two of the natural ligands for CD44, HA and OPN, and again measured cytokine release from the hybridomas, in this case using purified BSA as a negative control. Indeed, stimulation of CD44 with its natural ligands resulted in activation of the NKT cell hybridomas in a similar manner as when the pan-CD44 Ab IM7 was used, whereas no effect was observed with BSA treatment (Fig. 4), suggesting that the CD44-specific effect is physiologically relevant. In addition, much like with the IM7 clone, the conventional T cell hybridomas failed to become activated following treatment with HA or OPN, suggesting again that the observed CD44-mediated activation is NKT cell specific.
CD44 stimulation of primary NKT cells
Because the hybridoma cells used in this study are immortalized derivatives of primary cells and therefore may not react in the same manner as primary cellular isolates, we compared the CD44-mediated stimulation of primary hepatic NKT cells (NK1.1+TCRβ+) to conventional T cells (NK1.1−TCRβ+). To do this, we used intracellular cytokine staining of NKT and T cell-gated populations from total LMNCs as a primary measure of activation by comparing populations stimulated with the CD44-specific Ab IM7 to those treated with an isotype control Ab for different time periods. To verify the expression of CD44 on the surface of these primary isolates, we first performed triple-color staining of unstimulated primary LMNCs and detected CD44 expression on gated NKT and T cells. Consistent with the results using the hybridomas (Fig. 1), it was found that primary hepatic isolates of both NKT and conventional T cells expressed CD44 (Fig. 5,A). As reported earlier by several groups (35, 36, 37, 38), a homogenous CD44high population was clearly evident in the NK1.1+TCRαβint (Fig. 5,A) and α-GalCer loaded CD1d dimer+ (Fig. 5,B) populations, whereas mainstream T cells (NK1.1−TCRβ+ or TCRβ+ α-GalCer-loaded CD1d dimer negative) expressed heterogeneous levels of CD44 on their cell surface. Furthermore, when these primary lymphocytes were stimulated with plate-bound anti-CD44 IM7, only the NKT cell population accumulated significant levels of intracellular IFN-γ, IL-4, and GM-CSF, compared with the same population upon isotype control Ab stimulation (Fig. 5,C). Conversely, only a very small percentage (<3% above control) of the conventional T cell population produced IFN-γ, IL-4, and GM-CSF at the 1-, 6-, and 12-h time points, depending on the cytokine. The expression of CD44 on the surface of CD4+ and CD8+ mainstream T cells is shown in Fig. 6. The NKT cell cytokine response was found to peak 3 h after stimulation and slowly decrease over the course of 12 h, suggesting that CD44-mediated stimulation produces a quick NKT cell response consistent with previously observed CD1d-mediated NKT cell activation (39). Ultimately, by demonstrating similar effects in primary cells, these results represent a validation of the NKT cell-specific CD44-stimulation observations that were made using the hybridoma cell lines. The variability in the production of specific cytokines is consistent with differences in cytokine production by CD4+ and CD4−CD8− NKT cells (40, 41).
To further demonstrate a role for CD44 in enhanced cytokine secretion when this cell surface marker was cross-linked with anti-CD3, LMNCs (in which the predominant subpopulation of T cells is NKT cells) from wild-type and CD44KO mice were added to wells precoated with anti-CD3 with or without anti-CD44. CD44 engagement enhances IL-4 (Fig. 7,A) and GM-CSF (data not shown) production in wild-type (but not CD44KO) mice. Fig. 7 B shows that CD44KO mice have normal levels of NKT cells as reported previously (42), suggesting that, although CD44 is not necessary for NKT cell development, this molecule can modulate NKT cell function upon its engagement.
CD44-specific activation of NKT cell hybridomas correlates with HA binding ability
Because HA is a natural ligand for CD44 and we had shown that it could activate some of the NKT cell hybridomas (Fig. 4), specifically those that underwent activation following cross-linking with the IM7 Ab (Fig. 3), we next wanted to determine whether CD44-mediated stimulation of these cells was associated with the ability to bind HA. To accomplish this, we stained the hybridomas with FITC-HA in the presence or absence of pretreatment with unlabeled HA to demonstrate specificity and quantitated FITC-HA binding by flow cytometry. From these measurements, it was found that only some of the hybridomas retained the capacity to bind HA, namely N37-1A12, 431.A11, DN32.D3, N38-2C12, N38-2H4, and to a lesser extent, N37-2D5, whereas N38-3C3, DO11.10, and 3A9 could not (Fig. 8). Furthermore, this was shown to be a specific interaction because it could be partially or completely blocked with an unlabeled HA pretreatment (Fig. 8). Similar HA binding results were observed when mouse primary LMNCs were analyzed by triple-color (NK1.1-PE, TCRβ-allophycocyanin, and FITC-HA) flow cytometry. In this analysis, both NKT (NK1.1+TCRβInt) and NK (NK1.1+TCRβ−) cells retained the capacity to bind FITC-HA, whereas conventional T cells (NK1.1−TCRβ+ and NK1.1+TCRβhigh) and all other NK1.1−TCRβ− cells lacked this ability (Fig. 9). Interestingly, although all of the hybridomas in this study expressed cell surface CD44 (Fig. 1), only those that had the capacity to bind HA underwent CD44-mediated activation, suggesting variation in a regulatory component of these receptors. The CD44-specific HA binding also could be blocked by pretreatment with any of the pan-CD44 Abs (IM7, KM201, or KM703) (only KM201 is shown; Fig. 10,A), which, in turn, blocked HA-mediated activation when either KM201 or KM703 was used (only KM201 shown; Fig. 10,A). Blocking of HA-mediated activation was not attempted with the IM7 clone because it alone was capable of inducing activation. Conversely, when cells were pretreated with the pan-CD44 Ab IRAWB14, which enhances HA binding (22), we observed a modest up-regulation of HA binding among the hybridomas (Fig. 10,B). Importantly, this Ab-mediated enhancement of HA binding also was capable of inducing HA-mediated activation of NKT (N38-3C3) and T cell (DO11.10) hybridomas that were previously unresponsive to HA treatment (Fig. 10,B). In addition, NKT cell hybridomas that had been shown previously to secrete cytokines in response to HA treatment (Fig. 4) now demonstrated an increase in cytokine production following IRAWB14 pretreatment (Fig. 10 B). Most of the hybridomas used in these assays express differing levels of CD1d1 on their cell surface and can undergo autoactivation through CD1d to varying extents (43). However, plate-bound CD44 stimulation of these hybridomas in the presence of the anti-CD1d mAb, 1H6 (28), did not alter the cytokine secretion profile (data not shown). These findings, in combination with those described earlier suggest that, indeed, the HA binding potential of lymphocytes is linked to activation potential.
NKT and CD4+ T cell hybridomas express CD44 variants
Due to the phenotypic differences in CD44-mediated activation potential observed among hybridoma cells, we next attempted to characterize variations in CD44 gene expression that potentially contributed to these differences. The CD44 gene (Fig. 11,A) is composed of 20 exons, 10 of which are expressed in all isotypes and corresponds to the standard form of the receptor; the remaining 10 exons are termed variants due to their variable presence in any combination in different isotypes of the receptor (44). Expression of variant exons in the CD44 receptor map to the extracellular domain immediately proximal to the transmembrane domain resulting in a lengthening of the extracellular stem structure. CD44 variant gene expression has been linked with differential functional effects in many cell types (44) and, for this reason, we first characterized variant exon expression in the hybridomas by RT-PCR in an attempt to associate a specific variant expression pattern with the observed NKT cell-specific activation. RT-PCR was performed with each hybridoma to qualitatively assess gene expression of each variant and the standard form of the receptor. By using previously published methodology (30) with a forward primer specific for the standard exon 5 in all reactions and a reverse primer with specificity for each variant exon or standard exon 16 in individual reactions (Fig. 11,A) we would be able to determine the presence of each variant within the cell. This technique was useful only in identifying whether a specific CD44 variant exon was present; it was unable to tell us whether a particular variant was being expressed together with others in a specific transcript. Results of representative RT-PCR are presented in Fig. 11,B. Consistent with the CD44 cell surface expression data using pan-CD44 Abs (Fig. 1), all of the hybridomas expressed the CD44 standard isoform (Fig. 11,B, lane S and column S). Interestingly, most of the cells also expressed multiple CD44 variant isoforms (Fig. 11,B, lanes 1–10 and columns v1–v10), but with varying patterns. Not excluding the possibility that each cell contained intrinsic differences in CD44 variant expression patterns, that some variant combinations might be stimulatory with others being inhibitory, or other fundamental alterations in the host cell, we set out to identify variants that could be linked to the CD44-mediated activation observed earlier. It seemed apparent from these comparisons that the expression of a few variants, v4 and v10 in particular, were potentially involved in the activation effects. Because only four mouse CD44 variant-specific Abs (v4, v6, v7, and v10) were commercially available, we were limited in the scope of our analysis. We first attempted to confirm the expression of these specific variants on the hybridomas by flow cytometry. Although this technique is not as sensitive as the RT-PCR method, it could tell us whether the actual CD44 variant receptors were being expressed on the cell surface. From this we found that, depending on the hybridoma tested, low levels of V4, V6, and V7 (but no detectable V10), were expressed as detected by flow cytometry (data not shown), in patterns consistent with the RT-PCR data. We also assessed the expression of variable CD44 exon usage biochemically. Different patterns in CD44 variable exon expression were observed in Western blot analyses (Fig. 12). This is a further demonstration of the distinct differences among conventional T and NKT cells and is consistent with the RT-PCR data in Fig. 11 as well as the functional differences observed in the experiments shown above. Interestingly, it also was found through cross-linking experiments that some of these variant-specific Abs were capable of inducing activation of the hybridomas, although to a lesser extent than that observed with the IM7 Ab or the CD44 ligands HA and OPN (Fig. 13). This finding again suggests a role for CD44 variant expression in NKT cell activation.
CD44-specific activation protects hybridomas from anti-CD3-induced cell death
Although we have demonstrated that NKT and T cells differ in their CD44-mediated activation potential, not all CD44 functions are different between the two cell types. In addition to the fact that both NKT and conventional T cells undergo CD44-mediated morphological changes (Fig. 2), cross-linking CD44 receptors on the NKT and T cell hybridomas was sufficient to protect the cells from undergoing anti-CD3ε-stimulated AICD (Fig. 14). This finding has been reported previously for mainstream T cells (45), but not NKT cells, and when combined with our findings of NKT cell-specific CD44 activation, suggests another potential mechanism for innate immune function.
Many factors encompass the ability of the innate and adaptive immune response to act as a cohesive partnership to handle pathogenic insults. Although there is functional overlap between the two arms of the immune system (46), each component serves a specific role in overall outcome and therefore has developed distinct pathways to eliminate the pathogen. Improper regulation of these responses may result in immune dysfunction and disease. In this study, we investigated the ability of CD44 to differentially stimulate mouse NKT and conventional T cells by demonstrating that Ab cross-linking and natural ligand binding induces specific activation of (and cytokine secretion from) NKT cells, but failed to activate conventional T cells. This activation was related to the HA binding potential of the CD44 receptor, because it could be inhibited by Abs to CD44 that block HA binding, and enhanced following treatment with an Ab that stimulates HA binding. Interestingly, upon Ab-induced HA binding, T cells that were previously unresponsive to HA treatment (i.e., N38-3C3 and DO11.10) produced cytokines in response to HA, confirming a link between HA binding potential and lymphocyte activation. Although this result is not completely consistent with our view of preferential NKT cell activation following CD44 stimulation, because the DO11.10 T cell line also became activated, it is consistent with previous studies demonstrating that conventional T cells undergo CD44-mediated activation when the cells are already in an active state (11, 13, 47). Ultimately, this underscores the fact that, as we have shown here, conventional T cells retain the capacity to undergo CD44-mediated effects. However, we also have shown that hepatic NKT cells bind to HA and become activated, whereas mainstream T cells in the liver do not, suggesting that prior activation is only required for CD44 stimulation in conventional T cells, possibly because NKT cells are always in an activated state (48). Mechanistically, it seems likely that the HA binding potential of NKT cells is at least partially associated with CD44 variant exon expression and possibly the glycosylation state of the receptor, as observed in other systems (49). Identification of these pathways will be a focus of future studies. The challenges associated with identifying a specific mechanistic function for CD44 is endemic in the field, due in large part to its diverse involvement in many cellular processes and highly heterogeneous composition.
Previous in vitro data suggest that CD44 receptor engagement costimulates mainstream human and mouse T lymphocyte activity in conjunction with conventional T cell activation stimuli, such as cytokines and phorbol esters, as well as CD2 and TCR complex signaling (11, 13, 47, 50). However, direct CD44-mediated activation of resting lymphocytes has not been reported previously. Although the prior studies demonstrating CD44 costimulatory capacity also tested and failed to exhibit direct activation by CD44, most used either conventional T cell clones, which we have shown do not undergo direct CD44-mediated activation, or primary cellular isolates from sources low in NKT cell numbers. This, along with the fact that not all CD44-specific Abs are capable of inducing NKT cell activation, may have resulted in an inability to detect a response in those analyses. Additionally, many of the previous reports only measured a few parameters involved in lymphocyte activation and may have overlooked the differential activation, as we have noted that some CD44 phenotypes, such as morphological changes and resistance to AICD, are present in both NKT and mainstream T cells. For these reasons, we performed our studies with NKT and T cell hybridomas and LMNCs, which have a large population of both NKT and mainstream T cells (51), and used multiple CD44-specific Abs and natural ligands while using numerous measures of lymphocyte activation, as well as biochemical and molecular analyses.
As we have noted, CD44-mediated activation of NKT cells is low, compared with stimulation via the TCR complex. This may, in fact, be an inherent mechanism by which NKT cells avoid inappropriate stimulation until they or other immune cells, which may be recruited to the area as a result of this initial activation, can confirm the need to react further through additional stimuli. Because innate effector cells such as NK and NKT cells are designed to react very quickly to a danger signal, they may have developed multiple layers of stimulation states to allow for differing response levels and to avoid overreacting to inappropriate or endogenous stimuli. Because CD44 stimulation of NKT cells likely occurs as a precursor to the full activation and cytokine release following CD1d-dependent TCR activation it may be necessary for CD44 engagement to enhance the TCR-mediated effect. This point also has relevance in studies showing that cross-linking of NK1.1 also can stimulate NKT cells (52) and that IL-12 signaling can synergize with TCR-mediated recognition of endogenous Ags (53).
Ultimately, the question pertaining to the significance of CD44-mediated activation of NKT cells in health and disease remains to be answered. There are multiple potential implications for this activity in immune function and pathogenesis: 1) NKT cell surveying. NKT cells may constitutively express an active form of the CD44 receptor, which allows them to constantly survey the environment for an infection, for example, and, if necessary, trigger the appropriate immune response. If this hypothesis is correct, NKT cell recognition of HA exposed at the site of infection or tissue damage would initially lead to mild NKT cell activation through the release of chemotactic and inflammatory cytokines and up-regulation of activation molecules. If a large-scale immunological alarm is appropriate, this could be further accomplished through conventional CD1d-dependent, TCR-mediated stimulation of the NKT cell in the affected tissue, resulting in large-scale cytokine release. In the event that an infection failed to trigger an NKT cell response via TCR–CD1d interactions, chemotactic factors from the initial CD44-mediated stimulation could recruit professional APCs, which retain the capacity to recognize multiple pathogenic signals, into the region. 2) NKT cell chemotaxis. Because NKT cells may not be the first cells to arrive at the site of an alarm, other mechanisms may be required to draw them there. One potential is the secretion of OPN, which has significant chemoattractant capabilities for macrophages and T cells via binding to CD44 and/or α-integrin receptors (54, 55, 56). Following release from activated macrophages and lymphocytes, OPN recruits additional immune cells and induces Th1 cytokine expression while inhibiting Th2 cytokines (57). In this case, OPN may bind to CD44 on NKT cells and induce activation, which has been described recently in the context of Con A-induced hepatitis (55). Although our data suggest that CD44-mediated stimulation by OPN and HA is similar (Fig. 4), whether or not an active CD44 receptor phenotype, which typically describes the ability to bind to HA, also applies to the ability to bind OPN, is unknown. 3) CD44-mediated enhancement of cell survival. Although prior studies have attributed seemingly opposite apoptotic effects of CD44 depending on the cell type tested (15, 16), we (Fig. 14), and others (45), have demonstrated that CD44 engagement can send an anti-apoptotic signal to TCR-stimulated lymphocytes. When we view this effect in conjunction with the mild activation that occurs in NKT cells following CD44 stimulation, it is reasonable to surmise that this could impair the proposed autoregulatory AICD pathway that NKT cells are believed to undergo following stimulation (6, 7). If CD44-stimulated NKT cells are protected from the TCR-mediated AICD that typically follows lymphocyte activation, persistent NKT cell stimulation and immune dysfunction could thus follow. To contrast with this idea, a nice study by Nagarkatti and colleagues (58) showed in a model of ConA-induced hepatitis that expression of CD44 correlated with less severe disease. Furthermore, CD44 expression in that report correlated with increased AICD. Because this model has been shown by others to be NKT cell dependent (59), alterations in CD44 exon expression on NKT cells may be critical in the regulation of immune-mediated disease, such as in that model. Additional support for this hypothesis also comes from recent work from the Nagarkatti laboratory, in which it was clearly demonstrated that the induction of CD44v7 expression by staphylococcal enterotoxin B is protective against liver injury caused by this toxin (42).
CD44 and its ligands have been associated frequently with the natural and experimental induction of autoimmune diseases (60). Indeed, many autoimmune diseases are characterized by the existence of elevated levels of circulating CD44 ligands (61, 62). Regardless of the etiology, this increase in CD44 ligands could result in constant low-level stimulation of NKT cells and lead to immune dysfunction. Overexpression of OPN also is a risk factor for the development of autoimmunity and lymphoproliferative disorders (63). Furthermore, experimental administration of GAGs, including HA, chondroitin sulfates (A, B, and C), and heparin, all natural ligands of CD44, induce autoimmune connective tissue disease in normal mice (55, 64). Interestingly, a majority of the pathogenic inflammatory cells in this study are CD4+, GAG-binding T cells, consistent with an NKT cell population. In addition, a classical animal model of arthritis uses collagen, another CD44 ligand, to induce disease (65). Consistent with these observations, depending on the stimulus, NKT cell activity has been linked to both the induction of, and in some cases, protection from, autoimmune disorders (66, 67). Taken together with our findings, these observations seem to point to a role for CD44-associated NKT cell activity in the induction of certain autoimmune diseases. This is interesting in light of the fact that intra-articular injections of HA (Hylan GF-20, Synvisc; Genzyme) and Hyalgan (Sanofi-Aventis) are currently used to treat osteoarthritis of the knee and multiple reports of inflammatory complications have surfaced (68, 69, 70, 71, 72, 73, 74).
In conclusion, our observations have important implications for understanding fundamental variations between the innate and adaptive immune response and may lead to a greater understanding of autoimmune pathogenesis. How these two arms of the immune system interact and differentially survey the extracellular environment with regard to polysaccharides, such as conventional CD44 ligands, is not well understood at present. A more thorough understanding of how these major components of the extracellular matrix are involved in inducing the immune response and potentially contribute to the induction of autoimmune disorders may lead to novel therapeutic approaches for modulating the immune response.
We thank A. Bendelac, K. Hayakawa, J. Blum, and G. Bishop for supplying NKT and T cell hybridomas, and K. Hathcock for the IRAWB14 hybridoma cell line. We also thank J. He for providing the use of his microscope and digital imaging equipment.
The authors have no financial conflict of interest.
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.
This study was supported by National Institutes of Health Grants R01 AI46455 and R01 CA89026 (to R.R.B.), National Science Foundation Grant NSF CHE-0194682 (to J.G.-H.), and National Institutes of Health Grant T32 HL07910 (to J.L.). R.R.B. is a Scholar of the Leukemia and Lymphoma Society.
Abbreviations used in this paper: NKT, NK T cell; iNKT, invariant NKT; AICD, activation-induced cell death; GAG, glycosaminoglycan; HA, hyaluronan; OPN, osteopontin; FITC-HA, FITC-labeled HA; LMNC, liver mononuclear cell; α-GalCer, α-galactosylceramide; LDH, lactate dehydrogenase.