We compared HLA class II expression in a human melanoma line (a nonprofessional APC), induced by IFN-γ or by stable transfection with CIITA, with constitutive class II expression in an EBV-transformed B lymphoblastoid cell line (a professional APC) from the same donor. IFN-γ-induced and CIITA-transfected melanoma cells expressed DR, DP, and DQ at levels similar to those expressed by the professional APC; however, DP and DQ proteins and DM-dependent DR epitopes were delayed in appearing on the cell surface when induced by IFN-γ. The delay in cell surface expression of some IFN-γ-induced class II epitopes was observed even though Northern blots demonstrated class II and DM genes to be coordinately transcribed and their mRNA levels to be equivalent to that in B lymphoblastoid cells. Confocal microscopy suggests that discoordinate cell surface expression of class II results from different intracellular trafficking for IFN-γ-induced class II proteins in the melanoma line compared with that in professional APCs. Specifically, although DR and DM proteins were present 2 days after IFN-γ induction, colocalization of DR and DM proteins intracellularly was not apparent in cells at any time after induction. Failure of DR and DM proteins to colocalize suggests that IFN-γ-induced cells lack an intracellular MIIC-like compartment. The absence of a compartment containing DR and DM to facilitate interaction between the two proteins may account for the delayed surface expression of class II epitopes whose formation requires both class II and DM.
An immune response is initiated when a TCR engages an antigenic peptide bound to an HLA cell surface protein. HLA class II proteins, responsible for stimulating CD4+ T cells, are expressed constitutively on only a few cell types, known as professional APCs, which include B lymphocytes, macrophages, and dendritic cells. Class II proteins also are inducible in other cell types by IFN-γ (1); these latter cells are known as nonprofessional APCs. Regulating the expression of HLA class II proteins as well as the protein DM and invariant chain, which are required for class II presentation of Ag peptides, is fundamental to the control of an immune response (2, 3, 4, 5, 6, 7, 8). Invariant chain associates with class II in the endoplasmic reticulum, preventing the premature binding of peptide, and later is proteolytically cleaved to a fragment termed CLIP3 (9, 10). The HLA encoded protein DM facilitates removal of CLIP, allowing for the loading of antigenic peptide within an intracellular compartment, designated MIIC (11, 12). Invariant chain and DM have constitutive and inducible expression similar to that of class II genes.
The abundance of class II mRNA is a major determinant of the level of class II protein expression on the cell surface (13). Several factors have been identified that regulate class II gene transcription (14). One of these, CIITA, is a non-DNA-binding protein required for both constitutive and inducible expression of HLA class II genes (15, 16, 17, 18). CIITA transcription is induced by IFN-γ via a JAK1 signal transduction pathway (17, 19). CIITA is considered to be the specific IFN-γ-inducible factor required for class II expression (17). In addition to regulating transcription of class II genes, CIITA transfected into nonprofessional APCs results in the expression of DM and invariant chain (20). CIITA is thus a key regulatory transcription factor for the coordinate expression of genes required for HLA class II-mediated immune responses.
Generally, IFN-γ induction is associated with a coordinate increase in abundance of class II, DM, and invariant chain mRNAs. Yet, exceptions to coordinate expression of class II in nonprofessional APCs have been reported (21, 22, 23, 24, 25, 26, 27). Assuming that CIITA must be present for any of the class II genes to be transcribed in response to IFN-γ, lack of coordinate class II expression in nonprofessional APCs may be regulated at a level other than that of CIITA-mediated transcription.
The posttranscriptional events that govern the expression of class II proteins on the cell surface of professional APCs include trafficking through an intracellular compartment, designated MIIC (28, 29, 30). The MIIC is a multivesicular and/or multilamellar compartment defined by its contents, which include lysosomal proteins such as lysosomal-associated membrane protein-1 (lamp-1) and CD63, class II, and DM (15, 21). Studies with endocytosed proteins indicate that MIICs also contain late endosomal components, but not markers of early endosomes, such as transferrin receptor, or markers for trans-Golgi reticulum, such as mannose 6-phosphate receptor (31). Within the MIIC, CLIP is removed from class II molecules via the action of DM, and peptides, proteolytically derived from exogenous proteins, bind to class II molecules before the molecules are transported to the cell surface (32).
Less information is available on intracellular trafficking of IFN-γ-induced class II molecules in nonprofessional APCs. Experiments in which the components of the MIIC were transfected into nonprofessional APCs indicate that class II, DM, and invariant chain are required to generate a functional class II peptide loading compartment (33). Evidence of the physical existence of MIICs in cells without the class II Ag-processing proteins is conflicting. An initial report suggested that class II expression was required to generate multilamellar MIICs in nonprofessional APCs (34), but others have been unable to reproduce this observation. Copier et al. found the morphology of untransfected HeLa cells to be similar to that of cells transfected with class II, DM, and invariant chain. Colocalization of DM and class II was observed in transfected cells, but within an intracellular compartment that was also present by ultrastructure in untransfected cells (35).
In the present study, a human melanoma line, ThM, that does not constitutively express HLA class II proteins was induced to express class II either by IFN-γ or by transfection with CIITA. ThM stably transfected with CIITA expressed all class II isotypes (DR, DP, and DQ), DM, and invariant chain. DM-independent DR epitopes were detectable on the surface of ThM within 2 days after IFN-γ induction, but the expression of DP and DQ proteins and that of DM-dependent DR epitopes were markedly delayed. Confocal microscopy suggests that the delayed surface expression may be due to the lack of an intracellular MIIC-like compartment needed for efficient generation of surface class II antigenic determinants. Based upon these observations, a model for the regulation of class II genes in nonprofessional APCs is described. The implication of these findings is that the expression of class II in nonprofessional APCs has several control points that potentially could be manipulated to interrupt Ag presentation even after transcription of class II genes is initiated.
Materials and Methods
Melanoma and EBV-transformed B cell lines were established from the same donor and designated ThM and ThB, respectively (36). The HLA type of these lines includes DR2, DR6 (DR52a), DP4, A1, A2, B35, and Bw52. A ThM line that expressed CIITA, class II, and DM was achieved by stable transfection with pCIITA-8, received from Jeremy Boss. Transfected cells were selected with hygromycin at a concentration of 60 μg/ml. A DR-expressing clone was isolated from a FACS sort using the Ab, VI-15. A clone that was hygromycin resistant, but VI-15 negative, was also isolated to serve as a control for the effect of hygromycin. This clone was assumed to contain the hygromycin B phosphotransferase element in pCIITA-8, but not CIITA. Clones that expressed DR also expressed the other class II isotypes and DM. Stably transfected clones were maintained in culture with 60 μg/ml hygromycin. ThM and ThB were grown in RPMI medium containing 12% bovine calf serum, 100 U/ml penicillin, and 50 μg/ml streptomycin. ThM was grown as a monolayer, and ThB was grown in suspension.
Wn, a primary fibroblast culture established from lung parenchyma of a donor of HLA type DR3, DR6, DR52, DQ1, DQ2, A28, A29, B18, B60, was grown using the media formulated for ThM and ThB.
IFN-γ (Actimmune, Genentech, Inc., South San Francisco, CA) was added to cell cultures at a final concentration of 600 U/ml for induction of class II genes. This concentration produced maximal induction of surface DR protein after 2 days. Cells maintained in IFN-γ for prolonged periods were given fresh IFN-γ every 3 to 5 days when cell density dictated that the culture be split. After 2 wk of culture in IFN-γ, cell viability and growth rate were diminished.
mAbs with the following specificities and from the indicated sources were used: L243 (DR monomorphic) (37), VI-15 (DR monomorphic) (38), UK8.1 (DR-3, -5,-6 specific) (39), 16.23 (DR3 and DR52a specific with a conformation that requires expression of DMA and DMB, at least for the DR3-derived 16.23 epitope) (40), B7/21 (DP monomorphic) (41), SPVL3 (DQ monomorphic) (42), 1a3 (DQ monomorphic, Biodesign International), cer-CLIP (generated to residues 81–104 of invariant chain) (43), anti-lamp-1 (44), Pin.1 (specific for the amino-terminal cytoplasmic tail of invariant chain) (45), anti-transferrin receptor (46), goat anti-mouse FITC (Sigma F-8646, Sigma Chemical Co., St. Louis, MO), goat anti-rabbit FITC (Sigma F-1262), goat anti-mouse TRITC (Sigma T-5393), and goat anti-rabbit TRITC (Sigma T-6778). Abs to a peptide of the DMB cytoplasmic tail were generated in a rabbit, and serum was affinity purified using the DMB peptide immobilized on Sepharose. Purified anti-DMB sera did not stain the B cell line 9.5.3, which has a single nucleotide change in the DMB gene leading to a null mutation. The progenitor of 9.5.3 containing a normal DMB gene stained positively with the purified DMB antisera. In addition, the cell line T2, which fails to express DMB due to a deletion of the DMB gene, was not stained with the purified DMB antiserum. L243, VI-15, 16.23, and SPVL3 were conjugated with FITC for direct immunofluorescence. Goat Abs conjugated with a fluorophore were used to detect cell surface binding of the other Abs. For triple labeling, L243 was derivitized with digoxigenin (DIG; Boehringer Mannheim, Indianapolis, IN) using the manufacturer’s suggested method. L243-DIG was visualized using mouse monoclonal antidigoxin conjugated to Cy5 (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA).
Viable cells (0.5 × 106) were labeled with Ab for 30 min at 4°C in 50 μl of RPMI 1640 containing 25 mM HEPES, 5% FBS, and 0.02% sodium azide at pH 8.0 using a concentration of Ab sufficient to saturate binding sites on an EBV-transformed B lymphoblastoid line. Cells were washed twice with PBS at 4°C, suspended in 0.5 ml of PBS containing 5 μg/ml propidium iodide, and analyzed immediately. Cells staining with propidium iodide (nonviable) were excluded from gates. A second Ab binding reaction using goat anti-mouse FITC was used to detect the presence of the Abs that were not covalently conjugated with FITC before addition of propidium iodide. Surface immunofluorescence was determined using a FACScan System (Becton Dickinson), and data were analyzed with LYSIS II software.
Total cellular RNA was isolated, and Northern blots were prepared by standard techniques (47, 48). Probes used for hybridization were labeled with [32P]dCTP using random primers (Boehringer Mannheim). PCR was used to generate probes for class II and DM genes: DMA, DMB, DRA, and DQA were full-length cDNAs (49). The DQA cDNA has been demonstrated to detect multiple DQA alleles that generate different length mRNAs (K. A. Muczynski, unpublished observations). A 659-bp PCR product designed to detect all known DPB1 alleles was generated from the second exon (amino acid 10) to the fifth exon (amino acid 229) of DPB1 (50). A 516-bp PCR product designed to detect all known DQB1 alleles was generated from amino acids 39 to 211 (51). Actin was used in linearized plasmids (52). Ku70 was excised from a plasmid (53) and was used as a probe.
Cells plated on Lab-Tek chambered coverglass slides (Nalge Nunc International, Naperville, IL) were washed in cold RPMI, fixed for 20 min at 4°C in 4% paraformaldehyde and 30 mM sucrose in 100 mM phosphate buffer, pH 7.4, and washed in PBS containing 1% BSA (PBS/BSA). Cells were permeabilized for 5 min at room temperature with 0.2% saponin and 30 mM sucrose in PBS/BSA. After washing, cells were incubated for 60 min at room temperature with primary Abs diluted in PBS/BSA, washed, blocked in 5% normal goat serum in PBS/BSA for 15 min, then incubated in goat anti-mouse TRITC and goat anti-rabbit FITC secondary Abs for 30 min at room temperature. Triple-labeled cells were incubated in primary Abs rabbit anti-DMB, mAb SPVL3-FITC, and mAb L243-DIG; washed and blocked as described above, then incubated in goat anti-rabbit TRITC and mouse antidigoxin-Cy5. For triple-labeled cells, elimination of primary Abs in various combinations was used as a negative control. Cells were mounted in Vectashield antifade medium (Vector Laboratories, Inc., Burlingame, CA) and examined with a Bio-Rad MRC 600 or MRC 1024 confocal microscope (Bio-Rad, Hercules, CA).
Reverse transcriptase PCR
The Pharmacia Quick Prep micro-mRNA purification kit (catalog no. 27-9255-01) was used to isolate mRNA from 106 to 107 cells. Single-strand cDNA, made with Life Technologies superscript preamplification system (catalog no. 18089-011, Life Technologies, Gaithersburg, MD), was amplified using Taq polymerase for 35 cycles of 94°C for 30 s (denaturation), 45°C for 2 min (annealing), and 72°C for 1 min (extension).
Two sets of primers were used to amplify CIITA: 1) 5′-ATG-CGC-TAC-TTT-GAG-AGC-TCA-3′ and 5′GTT-GTC-ATA-GGG-CCT-CTT-CTT-3′, which generated a 514-bp product; and 2) 5′-CCT-GAT-GCA-CAT-GTA-CTG-GGC-3′ and 5′ ACG-TCC-ATC-ACC-CGG-AGG-GAC-3′, which generated a 711-bp product. Glyceraldehyde-3-phosphodehydrogenase (GAPDH) was amplified as an internal control for mRNA integrity and RT activity using primers purchased from Stratagene (catalog no. 302047, La Jolla, CA).
Autoradiographic analysis was conducted by the PhosphorImager Facility of the Markey Molecular Medicine Center at the University of Washington (Seattle, WA).
Delayed induction of DP-, DQ-, and DM-dependent DR epitopes in IFN-γ-induced ThM
Class II expression was induced in ThM, a melanoma line that normally does not constitutively express HLA class II mRNA or protein, either by stable transfection with CIITA or by addition of IFN-γ (Fig. 1). ThM stably transfected with CIITA expressed high levels of DR, DP, and DQ on the cell surface, exceeding those of ThB, an EBV-transformed B lymphoblastoid line derived from the same donor. A DR epitope that requires both DMA and DMB gene products, defined by Ab 16.23 binding (11, 37), also was expressed at high levels on the surface of CIITA-transfected Thayer (Fig. 1).
In contrast, after 2 days of IFN-γ treatment, only DR was detected in abundance on the surface of ThM using the antibodies L243 (Fig. 1), VI-15, and UK8.1. The DM-dependent DR epitope detected by 16.23 as well as DP and DQ only appeared on the surface of ThM after 1 to 2 wk of culture in IFN-γ. During this period, the level of DR on the cell surface continued to increase. The epitopes recognized by the DP and DQ antibodies used in this study do not require DMA or DMB, since they are detected on mutant B lymphoblastoid cells that lack DM (K. A. Muczynski, unpublished observations). Additionally, the invariant chain fragment CLIP, which remains complexed with class II molecules on the surface of DM-defective cells (43), was present only in low levels on CIITA-transfected or IFN-γ-induced cells (Fig. 1). Therefore, the delayed expression of DP, DQ, and the 16.23 epitope is unlikely to result from a lag in DM expression or aberrant DM function. Failure of class II and invariant chain to associate intracellularly or the lack of protease activity required to generate CLIP from invariant chain are alternate explanations for the low surface CLIP levels in class II-expressing ThM.
We considered several alternative explanations for the delayed surface expression of the 16.23 epitope and DP and DQ proteins. 1) Although CIITA alone is sufficient to induce high levels of class II expression in ThM, a delay in the IFN-γ induction of CIITA or quantitatively less of this transcription factor might account for a subsequent delay in IFN-γ-induced expression of selected class II isotypes. 2) Since a major regulatory step for class II expression is at the level of transcription, a delay in the IFN-γ induction of mRNA for DM, DP, and DQ compared with DR could account for the respective delayed surface expression of these proteins. 3) If class II genes are transcribed and translated coordinately after IFN-γ induction, delayed surface expression might reflect intracellular trafficking of class II protein, which differs from that in professional APCs. The following studies were designed to investigate these possibilities.
ThB and induced ThM contain similar amounts of CIITA mRNA
The presence of CIITA was estimated by RT-PCR. CIITA is a low abundance mRNA and was undetectable on Northern blots. Two sets of primers were used to amplify CIITA (see Materials and Methods); both produced the same results in terms of identifying the presence or the absence of CIITA. CIITA was not detectable in uninduced ThM or in transfected hygromycin-resistant ThM that did not express class II (Fig. 2). However, CIITA was present in transfected, hygromycin-resistant cells expressing class II and in cells treated with IFN-γ for 2 days. GAPDH was amplified from all cells, indicating that mRNA was intact and that RT was active.
The amounts of CIITA mRNA in stable CIITA transfectants and IFN-γ-induced ThM were compared by analyzing radiolabeled GAPDH and CIITA RT-PCR products serially removed at progressive PCR cycles (Fig. 2). The template for each PCR reaction was the amount of cDNA reverse transcribed from poly(A) RNA isolated from 1 × 105 cells. Similar levels of CIITA mRNA were obtained from stable CIITA transfectants and IFN-γ-induced cells. The cycle number at which PCR products initially were detected, the rate of increase in the PCR product with increasing cycles, and the maximum product after 35 cycles were similar in CIITA transfectants and IFN-γ-induced cells (Fig. 2). Normalization for the starting amount of mRNA was approximated using the same amount of cDNA for GAPDH amplification. GAPDH PCR profiles indicate that slightly more template was present in stable CIITA transfectants than in IFN-γ-induced cells. Assuming that GAPDH is representative of constitutively expressed housekeeping genes, this suggests that 2 days after IFN-γ induction, ThM contained at least as much if not more CIITA mRNA relative to GAPDH mRNA as cells stably transfected with CIITA. Therefore, the delayed cell surface expression of certain class II proteins after IFN-γ treatment appears not to result from the limited abundance of CIITA mRNA. Furthermore, the CIITA levels in IFN-γ-induced ThM and in stable CIITA transfectants were similar to that in the B cell line ThB (Fig. 2), which constitutively expresses CIITA and all class II proteins.
Class II Ag-processing genes are transcribed coordinately in IFN-γ-induced ThM
Total RNA was isolated from ThM at various times after IFN-γ induction and was analyzed by Northern blots for the abundance of class II, DM, and invariant chain mRNAs. Within 2 days of IFN-γ addition, DRA, DPB, DQA, DMA, DMB, and invariant chain mRNAs were abundant (Fig. 3), yet only DR protein was detected on the cell surface (Fig. 1). A DR epitope requiring DM was not expressed despite the presence of DMA and DMB mRNAs. The levels of class II, DM, and invariant chain mRNAs increased roughly coordinately after IFN-γ induction, allowing for the differences in mRNA abundance and the varying specific activities of the probes (Fig. 3). A lag in the induced transcription of selected class II and DM genes, therefore, does not appear to account for the observed delay in surface expression of the DM-dependent DR epitope and DP and DQ proteins.
The abundance of class II and DM mRNAs in IFN-γ-induced and CIITA-transfected ThM was also compared using Northern blots. CIITA-transfected cells contained more DRA, DQA, DQB, DMA, and DMB mRNA than IFN-γ-induced cells (Fig. 4). This might explain the low level of surface expression of DM-dependent DR epitope and DQ protein 2 days after IFN-γ treatment; however, ThB, which constitutively expresses these antigenic determinants at high levels (Fig. 1), had steady state levels of DRA, DMA, and DMB mRNA similar to those of IFN-γ-induced ThM. DQA mRNA in ThB was more abundant than that in ThM, while DQB was less abundant. The facts that class II and DM mRNAs could be readily detected in ThM 2 days after IFN-γ induction and that IFN-γ-induced ThM and ThB contained similar amounts of mRNA for these genes by Northern blotting make it unlikely that the delayed class II expression results from a low abundance of these mRNAs.
Class II and DM proteins fail to colocalize in IFN-γ-induced ThM
We used immunofluorescence with confocal microscopy to evaluate intracellular and surface expression of class II Ag-processing proteins in induced ThM. Two days after IFN-γ induction, both DM and DR were expressed in ThM (Fig. 5,b). DM was located intracellularly. DR, as detected by L243, was only visible on the cell surface (Fig. 5, b–d), unlike in a professional APC (a B lymphoblastoid cell) in which L243 was present both intracellularly and on the cell surface (Fig. 7,f). The absence of L243 binding intracellularly in IFN-γ-induced ThM is not readily explained on the basis of reduced affinity of L243 for DR bound with invariant chain or the inability of L243 to penetrate cells, since the professional APC labeled using the same procedure contained abundant intracellular L243. Furthermore, the DR Abs UK8.1 and HB10A also detected class II on the cell surface, but not within IFN-γ-induced ThM, similar to confocal localization of L243 (data not shown). Intracellular DM and cell surface DR were more abundant in ThM 2 wk after IFN-γ induction, and these proteins were in the same locations and of similar abundance in CIITA-transfected ThM (Fig. 5, c and d). The L243 epitope was not detectable intracellularly and did not colocalize with DM in ThM, possibly because of its rapid transport to the cell surface after translation. DM was present in a lysosome-like compartment in both IFN-γ-induced and CIITA-transfected ThM cells, as indicated by its colocalization with the lysosomal membrane protein lamp-1 (Fig. 6, b–d). DM did not colocalize with transferrin receptor, a marker of cell surface and early endosomes (Fig. 6, g–i). Invariant chain was present in many, but not all, of the intracellular compartments that contained DM (Fig. 6, l–n). Colocalization of invariant chain and DM may denote compartments of Ag processing in ThM.
DM proteins are required to generate the 16.23 epitope (54). Although both DR and DM were expressed 2 days after IFN-γ (Fig. 5, b and g), the 16.23 epitope was not easily visible on the cell surface until 10 to 14 days after IFN-γ induction (Fig. 1). Unlike the L243 epitope of DR, the 16.23 epitope was found intracellularly as well as on the surface of IFN-γ-induced cells (Fig. 5, g and h). Intracellular 16.23 binding was detected 2 days after IFN-γ induction. However, it was not colocalized with DM in the lysosomal compartments (Fig. 5,g). The failure of DR and DM to colocalize in ThM suggests that the intracellular generation of 16.23 in ThM occurs in a compartment distinct from MIIC, the compartment in professional APCs in which DM and class II proteins extensively colocalize (Fig. 7, d–g).
The MIIC appear as multilamellar and multivesicular intracellular compartments in professional APCs (Fig. 7,b). The ultrastructure of IFN-γ-induced ThM revealed abundant intracellular vesicular compartments, but none with the multilamellar or multivesicular structure (Fig. 7,a). Colocalization of class II and DM proteins was present in professional APCs (Fig. 7, d–g), presumably within MIICs, but class II and DM proteins were not colocalized within IFN-γ-induced ThM (Fig. 7 c).
In addition to the delay in surface expression of the 16.23 epitope in IFN-γ-induced ThM, the surface expression of DQ and DP was delayed (Figs. 1 and 8). SPVL3, an Ab that recognizes a monomorphic determinant of DQ, was localized to the cell surface of IFN-γ-induced ThM (Fig. 8,c). Intracellular deposition of SPLV3 was not apparent in the cells. As in the case of L243, SPVL3 labeled intracellular proteins in a professional APC (see Fig. 7,e), indicating the Ab’s ability to penetrate a cell membrane and recognize an intracellular DQ epitope. 1a3, another Ab that recognizes a monomorphic determinant of DQ, produced results identical with those for SPVL3, suggesting that the results for SPVL3 are representative of DQ. DP expression, using monomorphic Ab B7/21, was also delayed after IFN-γ induction of ThM (Figs. 1 and 8, g and h). Confocal data were similar to those for DQ. DP was detected on the cell surface but not intracellularly, even after 2 wk of IFN-γ induction (Fig. 8).
Induced expression of the 16.23 epitope and DQ is delayed in other nonprofessional APCs
The observations made for noncoordinate surface expression of class II epitopes in ThM were not unique to this melanoma line. A primary human fibroblast line derived from lung parenchyma of an individual with the appropriate DR type to generate a 16.23 epitope was established and designated Wn. Wn did not express class II proteins unless it was IFN-γ induced. As in the case of ThM, the L243 DR epitope of Wn was abundant 2 days after induction, but epitopes recognized by 16.23 (DM-dependent DR) and SPVL3 (DQ) were delayed (Fig. 9). Similarly, a delay in the IFN-γ-induced expression of DP and DQ proteins compared with that of DR was observed in primary cultures of kidney proximal tubular cells and in a transformed kidney cell line, HK2 (data not shown).
Regulations of constitutive and inducible class II expressions have been postulated to occur by similar mechanisms, since both require CIITA for transcription of class II genes. In this study of class II regulation in nonprofessional APCs, IFN-γ induced coordinate transcription of class II, DM, and invariant chain (Fig. 3) in the presence of CIITA (Fig. 2) in ThM cells, but cell surface expression of class II proteins was discoordinate (Fig. 1). The levels of CIITA, class II, DM, and invariant chain mRNAs in IFN-γ-induced ThM were similar to those in an HLA-identical professional APC line designated ThB; therefore, an altered pattern of mRNA abundance is unlikely to account for discoordinate class II cell surface expression. Our results suggest that regulation of class II in nonprofessional APCs may occur at levels other than transcription.
Professional APCs derived from invariant chain knockout mice have reduced numbers of class II molecules on the cell surface (5, 7). Diminished class II expression in these cells has been associated with abnormal posttranslational modification of class II proteins involving aberrant terminal glycosylation (5, 7, 55) and with abnormal intracellular trafficking characterized by an accumulation of class II in an endoplasmic reticulum compartment (5, 7). In our study, the observed delay in cell surface expression of some class II proteins in IFN-γ-induced ThM cannot be attributed to the absence of an invariant chain, since cells contained both invariant chain mRNA and protein 2 days after IFN-γ induction (Figs. 3 and 6, l and m).
Professional APCs contain an Ag-processing compartment, designated MIIC, that is characterized by its contents of class II proteins, DM, and lysosomal and late endosomal markers. MIICs in a B lymphoblastoid cell are depicted clearly in Figure 7; DR, DQ, and DM colocalize extensively in discrete intracellular compartments. In contrast, class II and DM proteins were not found to be colocalized in either IFN-γ-induced or CIITA-transfected cells by confocal microscopy (Figs. 5 and 8), suggesting that the MIICs of professional APCs are not present in ThM. MIICs also were not identified in electron micrographs of IFN-γ-induced ThM (Fig. 7,a). Intracellular DM was detected in IFN-γ-induced and CIITA-transfected ThM and was localized with the lysosomal marker, lamp-1 (Fig. 6, b–d). The presence of DM within intracellular compartments other than MIICs has been described. Intracellular lysosomal compartments containing DM, but not class II, were reported in professional APCs by Pierre et al. (56).
We suggest that the appearance on the cell surface of class II epitopes such as 16.23, which require interaction of class II with DM for their formation, may be delayed in IFN-γ-induced ThM because these cells lack a specialized MIIC compartment within which the class II and DM protein interaction is facilitated by close proximity. The observations 1) that surface binding of 16.23 is not present until 2 wk after IFN-γ treatment (Fig. 1) despite the presence of DR and DM protein components 2 days after IFN-γ induction (Fig. 5 b), and 2) that DR and DM do not colocalize in class II expressing ThM are consistent with this hypothesis. Moreover, the finding that 16.23 has both intracellular and cell surface locations indicates that DM-dependent class II epitopes can form intracellularly and then traffic to the cell surface without an MIIC compartment, albeit more slowly or less directly than if trafficking was via an MIIC. Since the generation of the 16.23 epitope requires interaction of DR and DM, one might speculate that these proteins contact one another when separate compartments containing each protein come together on a chance basis and fuse, and that this event is below the detection threshold of the fluorescence confocal microscopy used in this study.
Confocal microscopy with DR Abs L243 and 16.23 (Fig. 5) suggests that there may be two pathways for DR cell surface expression after DR gene transcription. The L243 DR epitope appears on the cell surface within 2 days of IFN-γ induction (Figs. 1 and 5) and is not detected appreciably inside of cells, suggesting that the DR protein follows a direct path to the cell surface after translation. A similar pattern of immunofluorescence was seen with two other DR Abs, UK8.1 and HB10A (data not shown). This pathway to the cell surface is consistent with the constitutive or default secretory pathway that is operational in all cells. A second pathway for DR cell surface expression is suggested by the cell-staining pattern of 16.23. The DM-dependent DR epitope 16.23 appears intracellularly 2 days after IFN-γ induction, but is not present on the cell surface until 10 days after induction (Figs. 1 and 5). This suggests the existence of a DR transport pathway in which DR is retained intracellularly, during which time interaction with DM occurs before transport to the cell surface.
The apparent lack of MIICs in IFN-γ-induced ThM may account for the delayed appearance of 16.23 binding at the cell surface, since it has been shown that this DR epitope requires DM activity (54). However, it does not readily explain delayed surface expression of DQ and DP epitopes, which are not dependent on DM function. Unlike the 16.23 epitope of DR, the SPVL3 and B7/21 epitopes of DQ and DP, respectively, were not detected within intracellular compartments of IFN-γ-induced or CIITA-transfected ThM (Fig. 8). Explanations for delayed cell surface expression and lack of intracellular expression of DQ and DP must also take into account the presence of mRNA for DQ and DP (Figs. 3 and 4). Possible explanations to account for delayed DQ and DP cell surface expression include the following: 1) delayed translation of DQ and DP with subsequent rapid transit to the cell membrane; 2) posttranslational modification of DQ and DP required to generate SPVL3 and B7/21 epitopes that occurs after DQ and DP reach the cell membrane; 3) the requirement of an unknown protein to interact with DQ and DP to generate SPVL3 and B7/21 epitopes, analogous to the situation for the 16.23 epitope; and 4) intracellular degradation of DQ and DP proteins that are not loaded efficiently with peptide within a MIIC. Whatever the mechanism for delayed DQ and DP expression, CIITA alone is sufficient to induce expression; other IFN-γ-inducible genes are not required.
The observed delayed cell surface expression of selected class II proteins in IFN-γ-induced ThM was also true for normal human fibroblasts and human kidney proximal tubular cells. Further studies exploring the basis for delayed class II surface expression in these nonprofessional APCs are in progress. That three unrelated cell lines all had delayed expression of selected class II proteins suggests that the observations for ThM might be characteristic of nonprofessional APCs in general. In contrast, THP1 and U937 cells, two monocyte/histiocyte lines, were examined as examples of professional APCs. Both cell lines had basal levels of class II protein on the cell surface detected by FACS. Addition of IFN-γ to THP1 and U937 cells resulted in a coordinate increase in cell surface expression of DR, DP, and DQ protein, unlike the situation in nonprofessional APCs.
The function of inducible class II expression in nonprofessional APCs is unknown. It has been speculated that induced class II may function in Ag presentation (57, 58); however, there are reports of nonprofessional APCs expressing class II that do not activate T cells (59, 60). The failure of induced cells with class II to initiate a T cell response has been associated with the lack of costimulatory factors such as B7 (60, 61), which are not induced with class II. In this paper we present data on another difference between professional and nonprofessional APCs that may affect Ag presentation and subsequent T cell stimulation; namely, some nonprofessional APCs may lack the MIIC compartment present in professional APCs. The functional significance of this remains to be determined.
We thank Tom Cotner for critical review of the manuscript and Dan Hill for preparation of the manuscript.
This work was supported by a Physician-Scientist Award (to K.A.M.) and National Institutes of Health Grants K1AI01150, R37AI16689 (to D.P.), and R01AI30527 (to D.P.).
Abbreviations used in this paper: CLIP, class II-associated invariant chain peptide; lamp-1, lysosomal associated membrane protein-1; TRITC, tetraethylrhodamine isothiocyanate; DIG, digoxigenin; GAPDH, glyceraldehyde-3-phosphodehydrogenase.