Melanosomal membrane proteins are frequently recognized by the immune system of patients with melanoma and vitiligo. Melanosomal glycoproteins are transported to melanosomes by a dileucine-based melanosomal transport signal (MTS). To investigate whether this sorting signal could be involved in presentation of melanosome membrane proteins to the immune system, we devised a fusion construct containing the MTS from the mouse brown locus product gp75/tyrosinase-related protein-1 and full-length OVA as a reporter Ag. The fusion protein was expressed as an intracellular membrane protein, sorted to the endocytic pathway, processed, and presented by class II MHC molecules. DNA immunization with this construct elicited CD4+ T cell proliferative responses in vivo. Ag presentation and T cell responses in vitro and in vivo required a functional MTS. Mutations of either the upstream leucine in MTS or elimination of the entire MTS negated in vitro Ag presentation and in vivo T cell responses. In a mouse melanoma model, DNA immunization with MTS constructs protected mice from tumor challenge in a CD4+ T cell-dependent manner, but complete deletion of MTS decreased tumor rejection. Therefore, MTS can target epitopes to the endocytic pathway leading to presentation by class II MHC molecules to helper T cells.
Immune recognition of human cancer has been most intensively investigated in melanoma. Melanosome membrane proteins have been the most prevalent and best characterized melanoma Ags recognized by both Abs and T cells. They form a family of differentiation Ags that are expressed by melanocytes and their malignant counterpart, melanoma (1, 2). Melanosomes are membrane-bound organelles that synthesize and package the pigment melanin. They are localized in the endocytic pathway of melanocytes (3, 4). The melanosome membrane glycoproteins are encoded by genetic loci that determine coat color in mice and thus are implicated in the type of melanin synthesized and how melanin is packaged. Prototypes include tyrosinase (encoded by the albino locus), the tyrosinase-related proteins (TRP)4 gp75/TRP-1 (brown locus) and TRP-2 (slaty locus), and gp100/pMel17 (silver locus) (5).
These melanosome Ags can be recognized by multiple components of the immune system, including cytotoxic and helper T cells and Abs in patients with melanoma and vitiligo as well as healthy individuals (1, 6, 7, 8). This broad recognition, along with the presence of IgG Abs, suggests that T cell help is involved. In fact, the melanosome membrane protein tyrosinase is recognized by CD4+ T cells, indicating that these molecules can be naturally processed through the class II MHC pathway (9, 10, 11).
Melanoma cells constitutively express class II MHC molecules that are up-regulated during the process of malignant transformation (12, 13). Because class II MHC is not normally expressed by cutaneous melanocytes and melanosome membrane proteins are only expressed by melanocytic cells, presentation of melanosome proteins through the class II pathway by melanoma cells could be operationally tumor-specific, i.e., endogenously presented only by cancer cells but not normal cells.
The melanosomal glycoproteins are transported to melanosomes by a dileucine-based sorting and retention motif, the melanosomal transport signal (MTS), which is comprised of a conserved hexapeptide sequence with a neighboring upstream glutamic acid (4, 14). To investigate if MTS can lead to class II MHC presentation, we constructed fusion constructs containing a reporter Ag linked to the transmembrane and cytoplasmic domains of gp75 containing wild-type and mutant MTS. These studies show that MTS can traffic Ags to the endocytic pathway for class II MHC presentation, providing an explanation for frequent recognition of melanosomal glycoproteins in melanoma and a strategy for inducing CD4+ T cell responses by DNA immunization.
Materials and Methods
Mice, cell lines, and Abs
[BALB/c × C57BL/6]F1 (CB6F1) female mice, 6–12 wk old, were acquired from The Jackson Laboratory (Bar Harbor, ME) or the National Cancer Institute (Bethesda, MD). Class II MHC (I-Ad)-restricted T cell hybridomas DO.11.10 and 3DO-54.8 were kindly provided by Dr. Phillipa Marrack (National Jewish Center for Respiratory Medicine, Denver, CO). B16F10LM3 is a derivative from our laboratory of B16F10 murine melanoma, kindly provided by Dr. Isaiah Fidler (15, 16). MO4 is the mouse melanoma cell line B16F10 transfected with full-length OVA (17), which was provided by the laboratory of Pramod Srivastava (University of Connecticut Health Center, Farmington, CT).
Plasmid and vaccinia constructs
A fusion construct containing full-length OVA and the carboxyl terminus of gp75 connected with a 9-aa linker, Ser-Gly-Gly-Ser-Gly-Gly-Ser-Gly-Gly, was constructed by PCR. The OVA cDNA was obtained from Mark Moore (Genentech, South San Francisco, CA) and was originally cloned into the NotI/BamHI site of the WRG7077 plasmid to create the plasmid WRGBEN/ova, which has previously been described (18). The OVA cDNA was released from WRGBEN/ova, which was then cloned into the multiple cloning site of pBK-CMV (Stratagene, La Jolla, CA) to create pBK-ova. The OVA gene coding amino acids 1–386 was amplified from plasmid WRGBEN/ova (18) with primer pairs 5′-CGCCACCAGACATAATAGC-3′ and 5′-GCCTCCTGAACCTCCGGAACCACCAGAAGGGGAAACACATCTGCC-3′. The transmembrane and cytoplasmic domains of gp75, amino acid 488–539, were amplified using primers 5′-TCTGGTGGTTCCGGAGGTTCAGGAGGCATCATTACCATTGCTGTAGTG-3′ and 5′-GGTTGCTTCGGTACCTGCTGCG-3′ from pSVK3-mgp75 (14). These PCR products were purified and subjected to a second round of PCR using primers 5′-CGCCACCAGACATAATAGC-3′ and 5′-GGTTGCTTCGGTACCTGCTGCG-3′.
The product of the second-round PCR was then digested with EcoRI and KpnI, releasing ova/gp75, which was then cloned into the multiple cloning site of pBK-CMV (Stratagene, La Jolla, CA) to create pBK-ova/gp75. The fusion gene had a combined open reading frame of 1365 bp capable of coding a protein of 455 aa, which includes 386 aa from OVA, 9 aa from the linker, and 60 aa from gp75 transmembrane and cytoplasmic tail (Fig. 1). Two MTS mutants of the fusion protein were also constructed by PCR (Fig. 1). The primers used for construction of a mutant fusion protein with a deletion of Glu427 to Asp435 (pBK-Del) were 5′-CTCAGCATAGCGTTGATAGTGATTCTTGGTGCTTCTAGAACG-3′ and 5′-CGTTCTAGAAGCACCAAGAATCACTATCAACGCTATGCTGAG-3′.
A leucine to alanine mutant, Leu432 to Ala432 (pBK-L2A), was constructed with the primer pair 5′-GAGTGCAGGCTGGTTGGCTTC-3′ and 5′-CCTGCACTCACTGATCACTAT3′. All constructs have been sequenced, and their primary structure confirmed.
The pUV1 plasmid (19) was used to generate recombinant vaccinia virus (rVV). Inserts containing ova/gp75, L2A, and Del were released from the pBK-CMV vectors by EcoRI/KpnI digests. The gel-purified insert fragments were cloned into the same sites of pUV1. The rVV were constructed as described previously (20).
Transient transfection and immunofluorescence microscopy
Mouse fibroblast L929 cells (5 × 104) were plated in 8-well chamber slides (Nunc, Naperville, IL) the day before transfection. On the day of experiment, cells were transfected with 0.5–1.0 mg DNA by either calcium phosphate, DEAE, or lipofectamine (Life Technologies, Rockville, MD) using standard methods. After transfection, cells were cultured for 48–72 h for protein expression before immunofluorescence staining or metabolic labeling.
Before staining, cells were washed with cold PBS and fixed with 2% paraformaldehyde, and then with or without 100% methanol for permeabilization. The fixed cells were probed with mAb OVA-14 (BioMaker, Rehovot, Israel) followed by FITC-labeled secondary goat anti-mouse Ab (Dako, Carpinteria, CA). For double staining studies, cells were costained with anti-mouse endocytic membrane glycoprotein LAMP-1 Ab 1D4B from Developmental Studies Hybridoma Bank (Iowa City, IA) followed by Texas red-conjugated anti-rat Ab. Slides were observed under a fluorescence microscope (Optiphot; Nikon, Garden City, NY). The desired images of transfectant cells were photographed with a mounted Nikon camera on the microscope.
Expression and in vitro Ag processing and presentation of OVA
Mouse B cell lymphoma A20 cells (I-Ad) infected with rVVs were used as APCs. To confirm the expression of OVA, A20 cells (5 × 106) were infected with vaccinia at a multiplicity of infection (MOI) of 20 for 16 h. The cell lysates (equivalent to about 2 × 105 infected cells) were run on 8.5% SDS/PAGE and transferred to polyvinylidene difluoride membranes. The blot was then probed by rabbit anti-OVA Ab and developed by enhanced chemiluminescence. To evaluate presentation of OVA epitopes, A20 cells were infected with 5 MOI of rVVs for 24 h and then fixed with 0.5% paraformaldehyde for 5 min. These fixed A20 cells (105) were cocultured with 105 cells/well of class II MHC (I-Ad)-restricted T cell hybridomas DO.11.10 and 3DO-54.8 for 24 h. The production of IL-2 was determined by adding 50-μl aliquots of the incubation supernatant to 5 × 103 CTLL cells for an additional 24 h. Then, 10 μl of MTT (5 mg/ml) was added to each well during the last 4 h of incubation. The absorbance at 570 nm was read on the Bio-Rad EIA Reader 2550 (Bio-Rad, Hercules, CA). Data represent the mean OD reading of MTT incorporation from triplicate culture wells.
Genetic immunization with the OVA/gp75 fusion constructs
CB6F1 mice were immunized with DNA plasmids purified by the Qiagen ion-exchange columns (Qiagen, Hilden, Germany) as previously described (18). Briefly, purified DNA was coated to the gold particles (0.95 μm in diameter) by ethanol precipitation. These DNA-coated beads were then instilled into Tefzel tubing and cut into desired length for delivery using a hand-held helium-driven gene gun (Powderject Pharmaceuticals, Oxford, England). For cutaneous DNA immunizations, mice were anesthetized with Metofane inhalation (Pitman-Moore, Mundelein, IL). Abdominal hair was removed with Nair depilatory cream (Carter-Wallace, New York, NY), so that depilated abdominal skin was exposed for immunization. Animals were immunized by delivering the gold beads in one bullet into each abdominal quadrant, for a total of four injections per immunization. Each injection delivered 1 μg DNA and therefore a total of 4 μg DNA per mouse each immunization. The bullet was delivered to the abdominal skin at a helium pressure of 400 pounds per square inch.
To measure Ab responses, CB6F1 mice were immunized with different plasmid constructs by gene gun once a week for 4 wk and a booster at week 6. Sera samples were collected at weekly intervals for 2 mo. Purified OVA (Sigma, St. Louis, MO) was used as Ag and plated 50 μg each well in a 96-well plate overnight at 4°C. The diluted serum samples were added to the Ag-coated plate and incubated for 1 h at room temperature. After washing, goat anti-mouse IgG conjugated with alkaline phosphatase (Sigma) was added and incubated for 1 h at 37°C. The plates were developed using the Fast p-nitrophenyl phosphate substrate (Sigma), and the reactions were terminated with the addition of 3 N NaOH. The absorbance at 405 nm were obtained by the Bio-Rad EIA Reader 2550 (Bio-Rad).
Cytokine ELISA for IFN-γ and IL-4 were performed using kits from Genzyme (Cambridge, MA). CD4+ cells purified from splenocytes from immunized and control mice were incubated with syngeneic APC splenocytes pulsed with 100, 10, or 1 μg denatured OVA (boiled for 5 min) for ∼18 h. Supernatants were harvested and tested by ELISA. The sensitivity of ELISA was >20 pg/ml.
For determining isotype responses, specific goat anti-mouse Abs (Sigma) against mouse IgG isotypes were incubated with serum samples over the Ag. The plates were then developed with tetramethylbenzidine dihydrochloride (TMB) and stopped by 2 M H2SO4. The absorbance at 450 nm were obtained by the Bio-Rad EIA Reader 2550.
CD4+ T cell proliferative assay
CB6F1 mice were immunized once a week for 2 wk by gene gun, and at day 14 the mice were sacrificed. CD4+ T cells were purified from pooled splenocytes by Cellect · Plus column (Biotex Laboratories, Alberta, Canada). The purified CD4+ T cells (3 × 105) were stimulated by incubation with syngeneic naive splenocytes (1 × 105) pulsed with denatured OVA at different concentrations for 4 days at 37°C. On day 4, 100 μCi of [3H]TdR was added to each well, and cpm were counted after 16–18 h. The proliferation response was expressed as the net cpm minus background.
Tumor challenge and T cell depletion
For tumor protection studies, CB6F1 mice were immunized by gene gun weekly for 2 wk, and at day 14 they were challenged by injecting s.c. 1 × 106 MO4 melanoma cells, a B16 melanoma cell line transfected with the full-length OVA (17). In vivo depletion of CD4+ T cells was accomplished by injection of 0.2 ml ascites i.p. of GK1.5 (American Type Culture Collection, Manassas, VA). The ascites was injected at day −3, 7, and 14 in the immunization schedule and depleted CD4+ T cells by >95% (16). The tumor growth was monitored every other day after tumor cell injection. All experiments were performed according to National Institutes of Health guidelines with approval by the Institutional Animal Care and Utilization Committee of Memorial Sloan-Kettering Cancer Center. Tumor-free survival was evaluated by Kaplan-Meier method and compared using the log-rank test.
Three fusion constructs containing MTS (ova/gp75) and mutant (L2A) or deleted (Del) MTS were generated. The constructs linked the transmembrane and cytoplasmic domains of the mouse brown locus protein gp75 (amino acids 478–539 of gp75) to the carboxyl terminus of the full-length OVA gene (Fig. 1). ova/gp75 contained wild-type MTS, whereas Del and L2A contained either in-frame deletion or leucine to alanine substitution of MTS, respectively.
To confirm cellular location of the fusion proteins and the effects of the MTS mutations on intracellular sorting, mouse fibroblast L929 cells were transiently transfected with different plasmid constructs and stained with mAb OVA-14 against OVA (Fig. 2). The ova/gp75 fusion protein had a distinct punctuate, cytoplasmic localization pattern consistent with an intracellular vesicular distribution (Fig. 2,B). The ova/gp75 protein colocalized with LAMP-1, an endocytic marker, in double-labeling experiment (Fig. 3). Thus ova/gp75 localized to the endocytic pathway.
A single mutation of the upstream leucine (L2A) markedly reduced intracellular retention in vesicles, although weak vesicle staining (Fig. 2,C) was still detected (the photomicrographs of vesicle staining in Fig. 2,C required 26 s exposure vs 3 s exposure for ova/gp75 fusion protein in Fig. 2,B). However, mutation of the upstream leucine led to cell-surface staining (Fig. 2,D), presumably by sorting through the default pathway to the cell-surface due to loss of the MTS intracellular retention signal. Complete deletion of MTS (Del) led to a fine intracellular reticular staining with a perinuclear blush, consistent with staining of the endoplasmic reticulum and Golgi, without any vesicle staining (Fig. 2 E). A weak cell-surface staining was observed (results not shown because surface staining was too weak to convincingly demonstrate in photomicrographs). These results showed that MTS was necessary for efficient intracellular sorting to the endocytic pathway, but the weak vesicle staining of the L2A mutant protein was consistent with inefficient sorting to the endocytic pathway when only a single leucine remained in the dileucine pair.
To compare the effects of functional and mutant MTS on Ag processing and presentation, recombinant vaccinia constructs expressing different versions of the MTS fusion protein were used to express the fusion proteins endogenously in A20 cells. The transduced A20 cells expressed the appropriate fusion proteins, estimated by molecular mass, and expressed approximately equivalent amounts of each fusion protein (Fig. 4). The two mutant proteins, L2A and Del, were expressed at steady-state levels that were at least as high as the ova/gp75 parental protein (Fig. 4). Ag processing and presentation through the class II MHC pathway was assessed using T cell hybridomas specific for the OVA323–339/I-Ad-restricted epitope. The ova/gp75 protein was processed and presented, but L2A mutation was not effectively processed and presented (the Del construct was not assessed in this assay) (Fig. 5). Therefore, a fully functional MTS was required for efficient presentation of the endogenously synthesized Ag through the class II MHC pathway.
To verify that processed and presented ova/gp75 could prime T cells in vivo, we genetically immunized CB6F1 mice with plasmid DNA by particle bombardment into the skin (18). The efficiency of the DNA constructs to prime T cell responses was monitored through proliferative response using purified CD4+ T cells. The ova/gp75 fusion construct primed CD4+ T cells efficiently (Fig. 6 shows results from one of four representative experiments). However, mutations in MTS markedly reduced CD4+ T cell responses, confirming the requirement for a functional MTS to prime CD4+ T cell responses in vivo. Purified CD4+ splenic T cells from mice immunized with ova/gp75 secreted IFN-γ, but little IL-4, after in vitro stimulation with denatured OVA (for 5 days) (Fig. 7). These results show that the T cell response elicited by DNA immunization against OVA using MTS was predominantly of the Th1 type.
Mice were assessed for IgG Ab responses to OVA after DNA immunization with the different MTS constructs. IgG responses were induced after immunization with secreted ova (Fig. 8). MTS mutants L2A and Del induced weak but detectable Ab responses (Fig. 8). DNA immunization with ova/gp75 generated no detectable IgG responses (Fig. 8). Secreted OVA (encoded by ova) readily elicited IgG responses, but we presume that ova/gp75 did not because the protein was sequestered in an intracellular compartment and not available for B cell recognition. The strong Th1 response generated by immunizing with ova/gp75 may also favor a cellular rather than humoral immune response. The weak Ab responses induced by L2A and perhaps Del could be related to expression of these proteins at the cell surface, in contrast to the intracellular retention of ova/gp75. We do not have a good explanation for the ability of L2A or Del to generate IgG responses in the absence of apparent T cell help.
To test the in vivo ability of these constructs to induce immunity against OVA, we used DNA immunization. CB6F1 mice were immunized cutaneously with MTS constructs twice, 7 days apart. Mice were subsequently challenged with MO4 melanoma cells expressing OVA. All mice immunized with control vector and unimmunized mice developed palpable tumors within 12–20 days, whereas the large majority of mice immunized with ova/gp75 did not develop tumors over 60 days (Fig. 9; data shown only through day 48). Mice immunized with Del construct showed only weak tumor protection with almost all mice developing tumors within 35 days (Fig. 9). Mice immunized with ova/gp75 were significantly protected from tumor challenge compared with mice immunized with Del (p < 0.01). The majority of mice immunized with the L2A construct showed significant tumor protection similar to mice immunized with ova/gp75.
To investigate the role of CD4+ T cells in tumor protection experiment, mice were depleted of CD4+ T cells and then challenged with tumor. Depletion of CD4+ T cells abrogated tumor protection induced by immunization with ova/gp75 (Fig. 10 is representative of two experiments). Treatment with vector alone followed by CD4+ T cell depletion showed no protective effect. Therefore, CD4+ T cells were necessary for tumor protection.
Melanosomes are differentiated organelles that are believed to be derived from the endocytic pathway. These results show that molecules sorted to the endocytic compartment by MTS are processed and provide peptides that are available to the MHC class II Ag-loading compartments, presumably during transit through the endocytic pathway. This notion is supported by the finding that the dileucine-based MTS is required for Ag presentation through the class II MHC pathway. Because MHC II molecules are broadly distributed throughout the endocytic pathway, specifically in late endosomes, loading of peptides sorted by MTS could occur potentially throughout the endocytic pathway. However it is not clear whether these results reflect promiscuous processing and loading of MHC II molecules throughout the endocytic pathway. Alternatively, sorting by MTS could localize Ag to more specialized compartments for processing and subsequent loading on MHC II molecules. A corollary to the second scenario, processing and loading in specialized compartments, is that melanosomes would contain machinery for protein degradation and loading MHC II. We are currently addressing these possibilities.
The MTS dileucine-based signal is shared by a set of melanosome membrane proteins, including tyrosinase, gp75/TRP-1, TRP-2, gp100, and the pink locus protein (which is a type II membrane protein where MTS is located at the amino terminus). It is known that mutation of both leucine residues in MTS can ablate endocytic localization (21), and our results suggest that mutation of a single leucine can markedly reduce localization in intracellular vesicles although still permit weak intracellular retention (Fig. 2,C). At the cellular level, effective class II presentation of these endogenous transmembrane proteins requires intact MTS with a complete dileucine signal. We believe this is due to a strict requirement for targeting of endogenous intracellular proteins to endosomes by MTS or other endocytic signals and not related to levels of protein expression (see Fig. 4). The dileucine motif is crucial to the MTS signal, either directing cargo from the trans Golgi to endosomes or triggering internalization from the cell surface into endosomes (see Refs. 4 , 14 , and 21). If proteins in the secretory pathway lack a sorting signal, they can move through a default pathway to the cell surface where they stay. In the case of the L2A mutant, the dileucine signal required for internalization from the cell surface would be altered although not completely disabled (21). This presumably explains the cell-surface expression of the L2A mutant protein, which loses most of its ability to sort or internalize to the endocytic compartment, leading to stable surface expression. In turn, the L2A protein is not presented through the class II pathway. In the case of our Del mutant, the entire MTS domain is deleted. We have previously reported (14) that the region of MTS in the gp75 tail contains not only endocytic sorting signals but also is required for efficient egress from the endoplasmic reticulum. This might be related to a diacidic peptide downstream of MTS, but this is not established. Based on these previous studies and our present results, we presume the Del protein stays within the secretory pathway, moving inefficiently from the endoplasmic reticulum to Golgi. Retention of the Del protein intracellularly without access to the endocytic pathway would explain the lack of class II presentation.
However, at the level of the host, the issue of specific endocytic targeting is more complex. Ags can access the endocytic pathway both endogenously (e.g., using MTS) or through exogenous pathways (e.g., by endocytosis or fluid phase uptake). For instance, secreted OVA induces immunity presumably through an exogenous pathway. Our results show the L2A protein is not endogenously processed and presented but still induces immunity. One possibility is that genetic immunization with L2A in vivo allows MHC class II presentation of the endogenous cellular OVA, which we were unable to detect in our assays, although this is hard to reconcile with the lack of effective Ag presentation in vitro. A different scenario, which we favor, is that genetic immunization with L2A induces qualitatively a different type of immunity because the mutant protein is expressed on the cell surface. Cell-surface OVA or peptides could be shed (e.g., due to a protease sensitive linker) and cross-presented by other APCs. Because the L2A construct induces Abs (but ova/gp75 does not), efficient uptake of OVA by Ag-specific B cells for presentation is possible. Cell-surface membrane-bound OVA could be degraded extracellularly for loading of MHC II molecules. Because the Del protein is largely expressed inside the cell but without endocytic targeting, less immunity in vivo was observed. Because some tumor protection was still detected with Del immunization, it is possible that a low level of cell-surface expression induced immunity or that retention in the endoplasmic reticulum led to efficient presentation through the class I pathway. At this point, we do not understand how immunization with the Del construct leads to Ab responses because the Del product seems to be largely retained intracellularly, although there may be low but sufficient levels of cell-surface expression for inducing Ab responses.
Another dileucine-based sorting signal, from the amino terminus of the invariant chain that traffics the assembled class II MHC complex to the MHC II compartments, has also been used to elicit T cell immunity (22). The invariant chain sorts to MHC II compartments, and presentation would be expected. In addition, the sorting signal of LAMP-1 also can be used to elicit T cell immunity (23). LAMP-1 is distributed throughout the endocytic pathway, including late endosomes and lysosomes. LAMP-1 is expressed within the MHC II compartment and therefore would be predicted to target Ags for class II MHC presentation. However, the MTS signal presumably segregates melanosome membrane proteins into a specialized vesicle subcompartment within late endosomes. Our results are consistent with the notion that molecules sorted to melanosomes are available for class II MHC presentation. It will be interesting to further evaluate how MTS could be used as a cassette linked to genes encoding Ags for DNA immunization to elicit immunity requiring T cell help.
We thank Setulari Vijayasaradhi and Yiqing Xu for helpful discussions, Bernard Moss and Philip Greenberg for assistance in vaccinia vectors, and Philippa Marrack for providing T cell hybridomas.
This work was supported by National Cancer Institute Grants CA58621, CA47179, and CA33049 and Swim Across America, the Dillman Fund, the Ruben Foundation, the Lymphoma Foundation, and the Louis and Anne Abrons Foundation.
Abbreviations used in this paper: TRP, tyrosinase-related protein; MTS, melanosomal transport signal; rVV, recombinant vaccinia virus; MOI, multiplicity of infection.