INTRODUCTION

CD69 is a member of the NK gene complex family of type II oligomeric signal transmitting receptor proteins that contain C-type lectin-binding domains and is expressed on a variety of hematopoietically derived cells, including bone marrow cells, monocytes, platelets, T and B lymphocytes, and natural killer cells (1-6). In all cell types examined, CD69 cross-linking transduces intracellular signals that generate a variety of cellular responses, suggesting that CD69 is a pleiotropic immune regulator important in the biology of many different hematopoietic cell types (2, 7-9). A specific ligand for CD69 has not been identified but has been postulated to involve carbohydrate moieties (10, 11).

CD69 is encoded by a single nonpolymorphic gene and is expressed in both the mouse and human as disulfide-linked homodimers and heterodimers of differentially glycosylated CD69 polypeptides or glycoforms (12-15). For mouse CD69, three putative N-glycan addition sites (Asn-X-Ser/Thr) exist within the extracellular region, accounting for the capacity to synthesize multiple CD69 glycoforms (3). Human CD69 synthesis is somewhat enigmatic in that only one N-glycan addition sequence (Asn-Val-Thr) has been identified within the extracellular domain (1, 13, 14). The molecular basis for human CD69 heterogeneity is unknown but has been proposed to result from qualititative heterogeneity in chain glycosylation, indicative of immature and mature oligosaccharides on CD69 species (2, 7). Recent studies by Hamann et al. (14) argue against this idea, however, as CD69 polypeptides synthesized in vitro in the presence of microsomes (in which immature N-linked glycans predominate) existed as two glycoforms, indistinguishable from those made in intact cells. Thus, the cellular mechanisms responsible for the generation of human CD69 glycoforms remain to be established (7).

The tripeptide sequence Asn-X-Ser/Thr is widely recognized as the consensus motif for attachment of N-linked oligosaccharides to polypeptides (16-20). However, early studies by Bause and Legler (21) showed that oligosaccharyl transferase can also transfer N-linked glycans to peptides having the atypical sequence Asn-X-Cys. To date, only three proteins have been demonstrated to contain N-linked glycans attached via atypical Asn-X-Cys motifs: human plasma protein C, bovine plasma protein C, and human von Willebrand factor (22-24) (see Table I). Interestingly, an atypical glycosylation motif exists within the extracellular domain of human CD69 that is similar to those previously identified (see Table I). In the current study we examined the hypothesis that two N-glycan addition sites exist on human CD69 molecules by determining the importance of Asn residues contained within typical and atypical glycosylation sites in CD69 heterogeneity. These data establish that multiple dimeric forms of human CD69 result from the differential post-translational modification of CD69 polypeptides with 1-2 N-linked glycan chains at typical (Asn-X-Ser/Thr and atypical (Asn-X-Cys) glycosylation motifs.

Table I. Atypical Asn-X-Cys glycosylation sites on polypeptides Sequence Protein Reference No. Asn-Ser-Cys-Ala-Pro-Ala Human von Willebrand factor 24 Asn-Ser-Cys-Val-His-Ala Bovine plasma protein C 22 Asn-Glu-Cys-Ser-Glu-Val Human plasma protein C 23 Asn-Ala-Cys-Ser-Glu-His Human CD69 This study

EXPERIMENTAL PROCEDURES

The human T cell line Jurkat was used throughout these studies and was maintained by weekly passage in RPMI containing 5% fetal calf serum in 5% CO₂ atmosphere (25); similar results were obtained using peripheral blood lymphocytes freshly isolated from normal human donors (data not shown). For activation, T cells were cultured for 3-4 h with phorbol myristate acetate (PMA),¹ (Sigma) at a final concentration of 5-10 ng/ml. BW5147 thymoma cells were maintained by weekly passage in RPMI containing 5% fetal calf serum in 5% CO₂ atmosphere (26). COS-7 cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum in 5% CO₂ atmosphere and were passaged by trypsinization every 5 days (27).

Metabolic labeling and surface-biotin labeling were performed as described previously (28). Cells were lysed by solubilization in 1% Nonidet P-40 lysis buffer (20 mM Tris, 300 mM NaCl, 10 mM iodoacetamide, plus protease inhibitors) for 20 min at 4 °C. Insoluble material was pelleted by centrifugation, and supernatants were removed and transferred to new tubes. For immunoprecipitation, lysates were mixed with anti-human CD69 mAb (FN50) (Pharmingen, San Diego, CA) preabsorbed to protein A-Sepharose beads for 3-4 h at 4 °C. Precipitates were washed 3 times in 0.2% Nonidet P-40 wash buffer and 1 time in phosphate-buffered saline.

Digestion of N-linked oligosaccharides by endoglycosidase H (Endo H) (Genzyme, Cambridge, MA) was performed as previously reported (29). N-glycosidase F digestion was performed using a N-glycosidase F digestion kit (Glyko Inc., Novato, CA) according to manufacturer instructions. One-dimensional and two-dimensional nonreducing × reducing gel electophoresis was carried out according to previously published methods (29), except that a 10% acrylamide gel was used in the first dimension and a 13% acrylamide gel was used in the second dimension. For detection of surface biotin-labeled proteins, samples were transferred to nitrocellulose (Immobilon P; Millipore, Bedford, MA) and visualized by chemiluminescence performed according to manufacturer instructions (Renaissance chemiluminescence reagent; NEN Life Science Products).

Cell surface staining of CD69 proteins was performed by incubating cells with primary anti-human CD69 mAb (FN50; Pharmingen) and, after washing, with secondary goat anti-mouse Ig conjugated to fluorescein isothiocyanate (Pharmingen). Cells were analyzed on FACSSCAN and FACSSTAR flow cytometers (Becton Dickinson, Mountain View, CA). All fluorescence data were collected using logarithmic amplification on 50,000 viable cells as determined by forward light scatter intensity or propidium iodide exclusion.

cDNA encoding human CD69 was cloned by reverse transcription-polymerase chain reaction from interleukin 2-activated peripheral blood lymphocytes using the 5 oligonucleotide primer GCTAAGCTTAAGATGAGCTCTGAAAATTGT and the 3 primer GCTGGATCCTTATTATTTGTAAGGTTTGTTACATATCCA. (These primers added a HindIII and BamHI enzyme restriction site to the 5 and 3 ends of the CD69 cDNA, respectively.) The polymerase chain reaction product was then cloned directly into pCR^TMII with an Invitrogen (San Diego, CA) TA cloning kit. The insert encoding CD69 was removed by enzymatic digestion with HindIII and BamHI enzymes and subcloned into the HindIII/BamHI sites of pBluescript II KS (Stratagene, La Jolla, CA). Site-directed mutagenesis was done using the Chameleon double-stranded site-directed mutagenesis kit (Stratagene) with the selection primer CGACCTCGAGGGGGGGCCCAGATCTCAGCTTTTGTTCCC (conversion of a KpnI site to a BglII site) and the following mutagenic primers: Asn_A  Lys(N_A  K), GGACTTCAGCCCAAAAGGCTTGTTCTGAACATGG; Asn_T  Ile (N_T  I), GGCAAAGAATTTAACAACTGGTTCATCGTTACAGGGTCTGAC; Asn_A  Lys + Asn_T  Ile, both of the mutant primers given above; Asn_Control  Lys, GACAAGTGTGTTTTTCTGAAAAAGACAGAGGTCAGCAGC; Cys_A  Ser (C_A  S), GGACTTCAGCCCAAAATGCTAGTTCTGAACATGGTGC. CD69 mutations were verified by sequencing BglII-sensitive mutant plasmids using the dideoxy chain termination procedure performed according to manufacturer instructions (Perkin-Elmer Corp.).

CD69 transcription was done by linearization of CD69 plasmids with BamHI restriction enzyme; CD69 RNA was transcribed from each of the plasmids using the T3 RiboMax^TM large scale production system (Promega, Madison, WI) followed by two rounds of extraction, first with Tris-EDTA-saturated phenol/chloroform/isoamyl alcohol and then with chloroform/isoamyl alcohol. Extracted RNA was precipitated with sodium acetate and isopropanol and resuspended in nuclease-free water. Translation of CD69 RNA was done with the Flexi^TM rabbit reticulocyte lysate system supplemented with canine pancreatic microsomal membranes (Promega) in the presence of [³⁵S]methionine for 60 min at 30 °C. Translation reactions were terminated by the addition of stop buffer (0.75 M KCl, 20 mM Tris-HCl, pH 7.6, 10 mM EDTA) and layered onto 0.5 M sucrose containing 20 mM Tris-HCl, pH 7.6, and 10 mM EDTA. Samples were centrifuged at 100,000 × g for 20 min, the supernatants were removed, and translation products were resuspended in 3 × SDS-PAGE sample buffer or digested with Endo H as previously detailed (29).

Approximately 1 × 10⁷ COS-7 cells were transfected with 15 µg of CD69 cDNA by electroporation at 200 V and 960 microfarads using a Biotechologies and Experimental Research Transfector 300 (Biotechologies and Experimental Research, San Diego, CA). 24 h after transfection, cells were harvested and radiolabeled with [³⁵S]methionine for 30 min at 37 °C. Nonidet P-40 lysates were immunoprecipitated with anti-CD69 mAb and analyzed on 13% SDS-PAGE gels under reducing conditions.

RESULTS

Initially, surface expression of CD69 was examined on Jurkat T cells cultured in the presence of PMA. In agreement with previous reports, surface density of CD69 increased rapidly within hours of PMA addition (Fig. 1A), which was verified by biochemical studies on surface-labeled proteins (Fig. 1B). As demonstrated, two CD69 species existed on PMA-stimulated Jurkat cells: a 28-kDa chain and a 32-kDa chain (Fig. 1B) previously shown to represent differentially glycosylated CD69 polypeptides (12).

CD69 glycoforms have been proposed to result from differential processing of CD69 glycoproteins by Golgi maturation enzymes, resulting in the presence of immature and mature oligosaccharides on 28- and 32-kDa species, respectively (2, 7); however, the maturation status of surface CD69 glycoforms has not been rigorously evaluated (12, 30). Glycosidase digestion studies were performed on CD69 precipitates of surface-labeled PMA-stimulated T cells using Endo H, which is specific for immature N-glycan chains, and peptide N-glycosidase F, which cleaves both immature and mature N-linked glycans (31). As shown in Fig. 2, both forms of surface CD69 were succeptible to N-glycosidase F treatment but were completely resistant to Endo H digestion (Fig. 2, left side). The effectiveness of Endo H in these studies was verified by parallel digests of T cell receptor alpha-glycoproteins from murine BW thymoma lysates (Fig. 2, right side), which are localized in the endoplasmic reticulum (ER) of this cell type (29). Thus, in agreement with previous studies (12, 30), these results demonstrate that 28- and 32-kDa CD69 species represent CD69 proteins that are differentially modified by N-linked glycan chains. Importantly, these data show that surface CD69 glycoforms expressed on PMA-stimulated Jurkat T cells contain exclusively mature, Endo H-resistant N-linked glycan chains.

Next, metabolic labeling studies were performed to examine the biosynthesis of CD69 proteins at time intervals when most CD69 proteins should exist within the ER and contain only immature glycan chains. As shown in Fig. 3A, CD69 proteins synthesized during a short 5-min pulse period existed in two glycoforms (CD69 A and B), both of which contained immature glycans as indicated by their sensitivity to Endo H digestion (Fig. 3A; CD69 Aimmature and B immature). In agreement with our findings in Fig. 2A on surface-labeled CD69, metabolically labeled CD69 glycoforms were both processed to mature, Endo H-resistant species by the conclusion of a 60-min chase period (Fig. 3A). To confirm that CD69 heterogeneity does not involve maturation of N-linked oligosaccharides in the Golgi complex, CD69 proteins were synthesized in the presence of the Golgi mannosidase inhibitor deoxymannojirimycin (32). As is evident, CD69 glycoforms persisted in deoxymannojirimycin-treated groups under conditions where conversion of immature, high mannose glycans to mature, complex type oligosaccharides was significantly inhibited (Fig. 3B). Taken together, these data show that CD69 glycoforms exist coincident with the synthesis and translocation of CD69 polypeptides into the ER and occur independently of processing by Golgi maturation enzymes. These results strongly suggest that CD69 heterogeneity results from quantitative differences in the number of N-linked glycans added to the polypeptide backbone and not from qualitative differences in oligosaccharide processing as previously proposed (2, 7, 14). Note that CD69 glycoforms do not reflect differential removal of glucose residues from nascent glycan chains by ER glucosidase enzymes as the mobility of both CD69 species was retarded following treatment with the glucosidase inhibitor castanospermine (32) (data not shown).

To determine the efficiency of assembly of newly synthesized CD69 glycoproteins into disulfide-linked dimers, immunoprecipitates of metabolically labeled lysates were analyzed on two-dimensional nonreducing × reducing gels. As shown in Fig. 4, most CD69 proteins synthesized during a 10-min pulse period existed in the low molecular weight B glycoform (approximately 73%), with the remaining CD69 proteins present in the A glycoform (Fig. 4, left lane); approximately 54% of newly synthesized CD69 was assembled into BB homodimers, 40% into AB heterodimers, and 6% into AA homodimers (Fig. 4). These results are consistent with the random assembly of CD69 glycoforms into disulfide-linked homodimers and heterodimers and are similar to the proportion of CD69 dimers expressed on the cell surface (Ref. 12 and data not shown). In addition, these data show that CD69 rapidly disulfide-links to itself in the ER as few, if any, newly synthesized CD69 proteins migrated as monomeric proteins on the diagonal (Fig. 4).

Careful examination of the amino acid sequence of human CD69 (1, 3, 13) shows that two potential N-linked glycan addition sequences exist within the extracellular domain: the previously identified consensus glycosylation motif Asn-Val-Thr (NVT) at positions 166-168 and the atypical Cys-containing glycosylation motif Asn-Ala-Cys (NAC) at positions 111-113 (Fig. 5). As noted, glycosylation of Asn-X-Cys sequences has been described for at least three other proteins (Table I). To determine the importance of specific amino acid residues in CD69 heterogeneity, site-directed mutagenesis studies were performed to create mutant CD69 proteins lacking Asn residues within the atypical site (Asn_A  Lys), the typical site (Asn_T  Ile), or both (Asn_A  Lys + Asn_T  Ile) (Fig. 5). In agreement with previous results, WT CD69 proteins translated in vitro in the presence of microsomes showed similar heterogeneity as CD69 synthesized in intact cells (Ref. 14 and this study) with both CD69 A and B glycoforms present (Fig. 6A). Synthesis of the higher molecular weight A glycoform was completely abolished by mutation of Asn residues within the atypical motif (Asn_A Lys) (Fig. 6A), demonstrating that Asn residues within this site do indeed contribute to CD69 heterogeneity. Consistent with the idea that two sites for N-linked oligosaccharide addition exist within the CD69 extracellular domain, glycosylated CD69 products (B glycoform) were present in Asn_T  Ile groups lacking Asn residues within the typical motif (Fig. 6A). As expected, mutant CD69 proteins in which both Asn residues were deleted (Asn_A  Lys + Asn_T  Ile) showed no evidence of glycosylation and migrated coincident with glycosidase-digested CD69 proteins (Fig. 6A). Interestingly, all CD69 proteins synthesized in WT groups were modified by carbohydrate addition (Figs. 6, A and B), whereas nonglycosylated CD69 proteins were present in Asn_A  Lys and especially Asn_T  Ile groups (Fig. 6A) (see below). These results were specific in that synthesis of CD69 glycoforms was unaffected by mutation of Asn residues localized within an unrelated tripeptide motif, Asn-Thr-Gln (Asn_Control  Lys, Fig. 5) (Fig. 6B). Taken together, these data show that Asn residues within both typical and atypical glycosylation motifs contribute to the synthesis of CD69 glycoforms, providing a molecular basis for human CD69 heterogeneity. These data are consistent with the modification of CD69 polypeptides with 1-2 oligosaccharide chains in the ER attached to Asn residues localized with typical (NVT) and atypical (NAC) glycosylation motifs.

We reasoned that variable usage of atypical glycosylation motifs on CD69 proteins might result from inefficient recognition of Cys residues by ER glycosylation machinery compared with Ser/Thr amino acids or, alternatively, result from the inaccessibility of this region to N-glycan attachment because of protein folding constraints. To investigate these possibilities, mutant CD69 proteins were created in which the atypical Asn-X-Cys motif was replaced by a typical Asn-X-Ser motif generated by a cysteine-to-serine conversion (Fig. 5, Cys_A  Ser). As demonstrated, such CD69 proteins were exclusively modified to the A glycoform containing two N-glycan chains (Fig. 6B, right side). These results indicate that glycosylation of CD69 polypeptides is primarily regulated by amino acid identity two positions distal to Asn, with Ser/Thr-containing sequences more efficiently modified than those containing Cys residues.

Finally, to verify that synthesis of CD69 glycoforms occurred similarly in intact cells, CD69 cDNAs were transfected into COS cells, and biosynthesis of CD69 glycoforms was examined by metabolic labeling and immunoprecipitation analysis. As shown in Fig. 7, wild-type CD69 proteins were synthesized as A and B glycoforms in COS cells (Fig. 7), data which are in agreement with previous findings that post-translational modification of CD69 is not cell-type specific (7, 14). Importantly, these data show that mutation of Asn residues within atypical (Asn_A  Lys) or typical (Asn_T  Ile) CD69 glycosylation motifs completely extinguished synthesis of CD69 A glycoforms in COS cells (Fig. 7), corroborating our results on CD69 synthesis in cell-free translation systems (Figs. 6, A and B). Similarly, CD69 variants lacking Asn residues at both sites (Asn_A  Lys + Asn_T  Ile) were not modified by carbohydrate addition in COS cells (Fig. 7), and synthesis of CD69 glycoforms was similar in WT and control (Asn_Control Lys) groups (Fig. 7). As previously observed in our in vitro studies, glycosylation of CD69 Asn_A  Lys proteins was more efficient than in CD69 Asn_T  Ile proteins, with significant amounts of nonglycosylated proteins present in CD69 Asn_T  Ile groups (Fig. 7). Interestingly, however, unlike in vitro translated Asn_A  Lys proteins, the vast majority of Asn_A  Lys proteins made in COS cells were modified by carbohydrate addition (Fig. 7). Although the significance of this latter difference is unknown, it is likely that these results reflect more efficient cotranslational modification of CD69 Asn_A  Lys proteins in COS cells relative to cell-free systems using microsomes. In conclusion, these studies show that two sites for N-glycan attachment exist within the extracellular domain of human CD69, both of which may be utilized for N-glycan addition and demonstrate that variable modification of typical and atypical glycosylation motifs represents the molecular basis for CD69 glycoform biosynthesis.

DISCUSSION

In summary, the current study provides a molecular basis for human CD69 heterogeneity by documenting that human CD69 polypeptides contain two sites for attachment of N-linked glycans within their extracellular domain that are variably utilized during CD69 biosynthesis. To our knowledge, CD69 represents only the fourth protein known to contain N-glycans attached to atypical Asn-X-Cys sites. Interestingly, unlike human CD69, which contains a single atypical glycosylation motif, the human von Willebrand factor protein contains eight occurrences of the Asn-X-Cys sequence, only one of which is modified by N-glycan addition (24). Multiple factors within the ER environment have been suggested to influence the usage of glycosylation motifs on polypeptides, including protein folding constraints, translation rate, and the amount of time a protein spends within the ER (23, 33). Our studies on CD69 biosynthesis indicate that modification of CD69 proteins at atypical glycosylation motifs is regulated primarily at the level of amino acid identity, although the contribution of Cys versus Ser residues to polypeptide folding was not directly addressed in these experiments. Determining the circumstances that promote or disfavor the modification of atypical Cys-containing glycosylation sequences on CD69 polypeptides should provide valuable information regarding the quality control mechanisms that regulate their utilization during polypeptide synthesis.

As described in the current report, variable glycosylation of CD69 polypeptides at atypical and typical motifs represents the molecular mechanism by which multiple dimeric forms of human CD69 molecules are generated. Indeed, several forms of CD69 dimers may exist, including CD69 AA homodimers, AB heterodimers, and BB homodimers (Fig. 8). Moreover, because oligosaccharide chains on monoglycosylated CD69 proteins can be attached to either typical or atypical glycosylation sequences, further heterogeneity among CD69 B glycoforms is possible (Fig. 8). Our data indicate that most monoglycosylated CD69 proteins (B glycoform) utilize the typical site for N-glycan addition, as nonglycosylated proteins were consistently synthesized in CD69 Asn_T  Ile groups lacking functional typical motifs. Importantly, the data in the current study provide a corrected model for the molecular basis of human CD69 heterogeneity by showing that CD69 glycoforms do not represent qualitative heterogeneity due to differential processing of a single glycosylation site as previously proposed (2, 7, 11) but rather result from quantitative differences in the number of N-glycans that are added to CD69 polypeptides during their synthesis (Fig. 8). Indeed, the results in this study document that two sites for N-glycan attachment exist within the human CD69 extracellular domain and identify the second, cryptic site as the atypical glycosylation sequence, Asn-Ala-Cys.

Concerning the potential role of glycosylation in CD69 structure and function, it is interesting that multiple glycosylated CD69 species are observed in both murine and human species (3, 15). Similar to mouse CD69, N-glycan addition sequences of human CD69 are localized within the lectin-binding portion of the extracellular domain (3). Conceivably, N-linked oligosaccharides could influence the recognition of as yet undetermined ligands by CD69 and the delivery of activation signals to the interior of the cell. Interestingly, two other members of the NK gene complex family of proteins, NKR-P1 and NKG2, contain similarly situated N-linked glycans within their carbohydrate-binding domains (2, 34), and NKG2 has atypical Asn-X-Cys glycosylation sequences localized within this region (34), although the extent of their usage is unknown. Finally, surface expression of human CD69 appears to be decreased following treatment with the glycosylation inhibitor tunicamycin (30), suggesting that N-glycans are important for the effective assembly and intracellular transport of CD69 molecules. However, because tunicamycin treatment would affect glycosylation of both typical and atypical sites on CD69 as well as glycosylation of numerous other cellular proteins, further studies are needed to fully define the role of N-linked glycosylation in CD69 assembly and expression. Experiments designed to address these issues are currently ongoing in the laboratory.