HIV-1 p55Gag Encoded in the Lysosome-associated Membrane Protein-1 as a DNA Plasmid Vaccine Chimera Is Highly Expressed, Traffics to the Major Histocompatibility Class II Compartment, and Elicits Enhanced Immune Responses*

Several genetic vaccines encoding antigen chimeras containing the lysosome-associated membrane protein (LAMP) translocon, transmembrane, and cytoplasmic domain sequences have elicited strong mouse antigen-specific immune responses. The increased immune response is attributed to trafficking of the antigen chimera to the major histocompatibility class II (MHC II) compartment where LAMP is colocalized with MHC II. In this report, we describe a new form of an HIV-1 p55gag DNA vaccine, with the gag sequence incorporated into the complete LAMP cDNA sequence. Gag encoded with the translocon, transmembrane and cytoplasmic lysosomal membrane targeting sequences of LAMP, without the luminal domain, was poorly expressed, did not traffic to lysosomes or MHC II compartments of transfected cells, and elicited a limited immune response from DNA immunized mice. In contrast, addition of the LAMP luminal domain sequence to the construct resulted in a high level of expression of the LAMP/Gag protein chimera in transfected cells that was further increased by including the inverted terminal repeat sequences of the adeno-associated virus to the plasmid vector. This LAMP/Gag chimera with the complete LAMP protein colocalized with endogenous MHC II of transfected cells and elicited strong cellular and humoral immune responses of immunized mice as compared with the response to DNA-encoding native Gag, with a 10-fold increase in CD4+ responses, a 4- to 5-fold increase in CD8+ T-cell responses, and antibody titers of >100,000. These results reveal novel roles of the LAMP luminal domain as a determinant of Gag protein expression, lysosomal trafficking, and possibly of the immune response to Gag.


MHC 1 II-directed antigen activation of CD4 ϩ T-cells is vital to the function of genetic vaccines as demonstrated in studies
with MHC II knockout (1) or CD4 ϩ -depleted mice (2). This includes CD8 ϩ responses, which require CD4 ϩ T-helper cells for secondary expansion and the development of memory (3). In general, antibody class switching, clonal expansion of antigenspecific B cells, T cell expansion and memory cell formation, and many other functions of the immune response require the costimulatory signals and cytokines released by antigen-activated CD4 ϩ T cells (reviewed in Ref. 4). Efficient priming of CD4 ϩ T-cells by DNA vaccination requires sufficient levels of antigen expression in transfected cells and delivery of the protein antigen to the MHC II antigen-processing pathway. This may be facilitated by direct transfection of dendritic cells, a key event in the function of DNA vaccines (5-9, reviewed in Ref. 10). However, although there is evidence for activation of CD4 ϩ T cells by DNA-encoded proteins (9,11,12, reviewed in Ref. 13), epitopes of endogenous proteins expressed in the cytoplasm of transfected cells are commonly presented by MHC class I molecules to cytotoxic CD8 ϩ T-cells. Antigens presented by MHC II molecules to helper CD4 ϩ T-cells are derived mainly from extracellular "foreign" proteins taken into dendritic cells by endocytic receptors (reviewed in Refs. 13 and 14).
With a goal of enhancing endogenous antigen trafficking to the cellular MHC II compartment of antigen-presenting cells, we employ the use of genetic vaccines encoding the antigens as chimeras containing the lysosomal targeting sequences of the lysosome-associated membrane protein (LAMP). Our rationale is that the delivery of antigen to the cellular site of MHC II processing and binding of antigen epitopes could result in an enhanced immune response through greater antigen-specific activation of CD4 ϩ T-cells. LAMP molecules have steady-state localization in the outer membrane of lysosomes (15)(16)(17) with a trafficking pathway from the Golgi complex (18,19) mediated by adaptor protein binding (20) to the carboxyl-terminal YXXØ (where the Ø symbol represents any hydropholic amino acid) recognition sequence of an 11-amino acid cytoplasmic tail (21)(22)(23). This LAMP trafficking pathway was found to coincide with that of MHC II in specialized multilaminar vesicular compartments of immature APCs, termed MIIC, sites associated with the formation of antigenic peptide-MHC II complexes (24 -28). The role, if any, of LAMP in MHC II-related antigen processing and presentation is unknown. Nevertheless, LAMP and MHC II are closely linked in their colocalization in the MIIC, the presence of a dendritic cell-specific LAMP (29), and the activation of lysosomal function during dendritic cell maturation (30). Moreover, a genetic association is suggested by the CHS1 gene defect, the Chediak-Higashi syndrome, which includes abnormalities of lysosome biogenesis and the accumulation of LAMP molecules, and of MHC II trafficking and immune functions (31,32). The application of LAMP to genetic vaccines has been tested with several recombinant protein antigens constructed as chimeras containing the LAMP cytoplasmic domain YXXØ targeting sequence. With a variety of plasmids, antigens, and gene delivery systems, LAMP/antigen chimeras were targeted to the lysosomal membrane and found to elicit enhanced immune responses as compared with vaccines encoding unmodified, native antigens. Antigen/LAMP chimeras encoded in vaccinia virus vectors included HIV-1 gp160/LAMP (33,34), human papilloma virus E7/LAMP (35,36), and cytomegalovirus pp65/LAMP (37). Naked DNA plasmid antigen/LAMP chimera vaccines include human papilloma virus E7/LAMP (38) and dengue virus 2 premembrane/envelope/LAMP (39,40). Enhanced T-cell responses to dendritic cells transfected with RNA encoded antigen/LAMP chimeras of carcinoma antigen (41) and telomerase reverse transcriptase (42) have also been described.
We report here the application of LAMP-trafficking to a DNA vaccine encoding HIV-1 p55 Gag. Gag is attractive as an HIV-1 vaccine, because it is relatively conserved among diverse HIV strains and subtypes, and broad cross-clade anti-Gag cytotoxic T lymphocytes (CTLs) responses have been demonstrated in HIV-infected patients (43)(44)(45)(46). Moreover, several studies relate both Gag-specific CD4 ϩ and CD8 ϩ responses to the control of viremia following infection (47)(48)(49)(50)(51)(52)(53)(54)(55). Application of native gag as a DNA vaccine is limited by a strong dependence on binding of the viral Rev protein to Rev-responsive elements (RRE) for the nuclear export and stability of Gag mRNA (56,57). These inhibitory sequences (INS) have been identified, and their removal by silent-site mutations or humanization of codon usage has resulted in increased expression of Gag protein by DNA vectors in the absence of Rev (58 -63). DNA vaccines encoding these modified Rev/RRE-independent forms of Gag have been reported to elicit Gag-specific immune responses of mice, both antibody and CTL (59,61,64).
In our development of a HIV-1 gag DNA vaccine, we have analyzed two forms of LAMP/gag chimeras as DNA vaccines. The initial construct corresponded to the LAMP/antigen chimera containing the LAMP translocon, transmembrane, and cytoplasmic domains, but lacking the luminal domain (Gag/ lamp), that was previously successfully used with several DNAencoded antigens. It was found, however, that the majority of the Gag/lamp chimera did not traffic to the lysosome or MHC II compartments of transfected cells, despite the presence of the LAMP cytoplasmic YQTI targeting sequence. Also, the Gag/ lamp chimera did not elicit an appropriately enhanced immune response when injected into mice. These problems were overcome by including the LAMP luminal domain in the construct, thus placing Gag within the lumen of the complete LAMP molecule, proximal to the transmembrane domain (LAMP/ Gag). In this case, the LAMP/Gag chimera protein trafficked to the MHC II compartment and application of this construct as a DNA vaccine resulted in enhanced immune responses by immunized mice, as shown herein, and in ongoing studies with non-human primates (data not shown). Additionally, there was a novel finding that this DNA elicited a high level expression of unmodified Gag protein by transfected cells.

EXPERIMENTAL PROCEDURES
Plasmids-Eukaryotic expression plasmids were constructed using nucleotides 1-1503 of the HIV-1 HXB2 p55gag gene (GenBank TM accession number K03455) inserted in two vectors, pcDNA3.1 (see Fig. 1A, plasmids 1-5) (Invitrogen, San Diego, CA) and pITR (65), a vector containing the adeno-associated virus inverted terminal repeat (AAV-ITR) sequences flanking similar expression elements (cytomegalovirus promoter and bovine growth hormone polyadenylation signal) (see Fig.  1A, plasmids 6 -8). The plasmids shown in this study and as labeled in Fig. 1 (see below) are: plasmid 1, pcDNA3.1 gag N and plasmid 6, pITR gag N , which are the native gag sequence inserted into NheI and KpnI sites of pcDNA3.1 and pITR vectors; plasmid 2, SS/gag N /lamp: this construct has an open reading frame of 1695 bp. The 5Ј 72 bp encode the endoplasmic reticulum translocation signal from the mouse LAMP (SS), and an XhoI site at the 3Ј-end links the gag N gene adjoined by an EcoRI site to 108 nucleotides encoding the mouse LAMP transmembrane and cytoplasmic domains (lamp) (GenBank TM J03881); plasmid 3, gag ⌬INS : this pcDNA3.1 vector encodes a mutated HXB2 p55 Gag where the inhibitory sequences were deleted by use of several sitespecific silent mutations (61). This plasmid was a gift from Dr. G. N. Pavlakis from the National Institutes of Health; plasmid 4, pcDNA3.1 SS/gag ⌬INS /lamp and plasmid 7, pITR SS/gag ⌬INS /lamp: these constructs encode the SS at the 5Ј-end and a XhoI site at the 3Ј-end that links the gag ⌬INS gene adjoined by an EcoRI site to 108 nucleotides encoding the LAMP transmembrane and cytoplasmic domains (lamp); plasmid 5, pcDNA3.1 LAMP/gag N and plasmid 8, pITR LAMP/gag N : these plasmids have an open reading frame of 2736 bp, inserted into NheI and KpnI sites. They contain 1113 bp encoding the complete LAMP luminal domain, including the SS, linked to gag N at the 3Ј-end by XhoI and gag N adjoined to LAMP transmembrane and cytoplasmic domains by EcoRI. The plasmids used for vaccination were produced in Escherichia coli and made endotoxin-free (Qiagen, Valencia, CA).
Evaluation of LAMP Targeting by Confocal Microscopy-MHC II (I-E k )-expressing cells (DCEK.ICAM.Hi7, gift from Dr. Susan Swain, The Trudeau Institute, Saranac Lake, NY) (67) were plated onto polylysine-coated coverslips in 6-well plates (2 ϫ 10 6 cells/well) and incubated overnight. The cells were transfected with the DNA plasmids using LipofectAMINE transfection reagent (Invitrogen), and 24 -48 h later they were stained to evaluate cellular localization. Coverslips were fixed in 2% paraformaldehyde in PBS for 5 min and washed with PBS then blocked with PBS containing 4% normal goat serum and 0.1% saponin. Endogenous LAMP was detected with rat anti-mouse LAMP-1 (1D4B) and rat anti-mouse LAMP-2 (ABL93) (15,68,69) supernatant medium at a dilution of 1:50. After washing three times with 0.1% saponin in PBS, the cells were incubated for 1 h with Texas Red-labeled goat anti-rat IgG (BD Pharmingen, San Diego, CA) at a 1:500 dilution. For detection of Gag expression, the cells were incubated for 1 h with mouse anti-Gag monoclonal antibody at a 1:50 dilution. After washing three times with 0.1% saponin in PBS, the cells were incubated for 1 h with Texas Red-labeled goat anti-mouse IgG (BD Pharmingen) at a 1:500 dilution. Colocalization of LAMP-1, LAMP-2, the different forms of Gag, and the LAMP/Gag N chimera with MHC II of transfected DCEK cells was performed by double immunostaining, first for the LAMP or LAMP/Gag N chimera proteins as described above and then with anti-MHC II, by incubating the cells for 1 h with FITC-labeled goat antimouse I-E k (14-4-4S) (BD Pharmingen) at a 1:75 dilution. The cells were then washed three times with PBS, and the coverslips were mounted onto glass slides using ProLong Antifade reagent (Molecular Probes, Eugene, OR). Confocal microscopy was performed using a Wallac confocal laser scanning microscope. Colocalization of the proteins with MHC II was thus determined by merging fluorophore images individually captured and digitally colored by use of Photoshop 5.0 (Adobe).
Antibody Responses of Vaccinated Mice-BALB/c mice (Charles Rivers, Wilmington, MA), 6 -8 weeks old, were immunized intramuscularly with 50 g of the specified plasmid. The mice were immunized four times at 3-week intervals with the same protocol. Blood was collected by tail bleeding 9 days after the second immunization and at day 94 after four immunizations and centrifuged to remove cells, and the serum was collected and stored at 4°C for immediate use. HIV IIIB lysate (ABI, Rockville, MD) was diluted in 0.1 M sodium carbonate-bicarbonate buffer, pH 9.4 (Pierce) at a concentration of 5 g/ml, and 50 l of the solution was added to each well of a 96-well plate (Nunc, Roskilde, Denmark). After overnight incubation at 4°C, the solution was removed and the plates were washed six times with PBS containing 0.05% Tween 20 wash buffer. The plates were incubated for 2 h at 37°C with 200 l of blocking buffer (PBS with 0.05% Tween 20 and 5% fetal bovine serum) and then washed three times. Serum samples were prepared in eight (3-fold) serial dilutions in blocking buffer starting at 1:100. 100 l of each dilution was added to the blocked plate, and the plates were incubated overnight at 4°C. For IgG detection, the reaction was developed using goat anti-mouse anti-IgG-horseradish peroxidase (Cappel, Durham, NC) at a 1:5000 dilution. For IgG 1 and IgG 2a , biotinylated anti-mouse anti-IgG 1 and IgG 2a antibody (BD Pharmingen) diluted 1:500 was conjugated with avidin-horseradish peroxidase (BD Pharmingen) diluted 1000-fold, and 100 l was added to each well for 2 h at 37°C. Turbo tetramethylbenzone substrate solution (BD Pharmingen) was added to each well and incubated for 15 min at room temperature. The reaction was stopped by adding 100 l of 1 M sulfuric acid, and absorbance at 450 nm was measured in a Bio-Rad model 3550 microplate reader.
Preparation of Splenocytes for T-cell-mediated Immune Responses-Single cell suspensions, depleted of red blood cells (RBC), were prepared from freshly isolated splenocytes in culture medium (RPMI 1640 medium supplemented with 5% v/v fetal bovine serum, 100 units/ml penicillin/streptomycin, 2 mM L-glutamine, 50 M 2-mercaptoethanol, and 1 M HEPES buffer). Splenocytes were counted and suspended at a concentration of 10 ϫ 10 6 cells/ml of the culture medium for T cell-mediated assays. When indicated, splenocytes were preincubated with anti-CD4 or anti-CD8 (Southern Biotechnologies, Birmingham, AL, 20 g/ml) for 30 min. CD8 ϩ or CD4 ϩ cells were also depleted by use of magnetic beads (Dynal, Oslo, Norway).
Real-time Quantitative RT-PCR Detection of Cytokine Up-regulation-Splenocytes of immunized mice were cocultured with feeder cells (splenocytes from naive mice irradiated with 3500 rad) in the presence of 10 g/ml HIV SF2 p55 Gag recombinant protein (Baculovirus-produced, National Institutes of Health AIDS Research and Reference Reagent Program, Rockville, MD), 5 g/ml concanavalin A (Sigma, St. Louis, MO) or in culture medium alone, in duplicates, in a 48-well plate cultured overnight for cytokine mRNA detection. After overnight culture, the cells were washed with 1ϫ PBS supplemented with 2% fetal bovine serum and 100 units/ml penicillin/streptomycin and resuspended in 100 l of RNAlater (Ambion, Austin, TX) and frozen until further processing. Total RNA was then extracted using RNeasy kits (Qiagen, Valencia, CA) according to the manufacturer's protocol. The cDNA was made using the Ready-To-Go TM , T-primed first strand kit (Amersham Biosciences) according to manufacturer's protocol starting from 5 g of the extracted RNA from each sample. The primers used for the detection of HPRT, IL-2, and IL-4, and the cycling conditions are described in Reiner et al. (70). For standards, the pPQRS plasmid (containing HPRT, IFN-␥, IL-2, IL-4, IL-5, IL-10, IL-12, tumor necrosis factor-␣, transforming growth factor-␤, and inducible nitric-oxide synthase inserts) was diluted from the original stock to 10,000, 1,000, 100, 10, 1, and 0.1 fg/l dilutions in sterile distilled water. The cDNA of the samples were diluted at 1/20 in sterile distilled water. The PCR amplification was performed on an iCycler iQ TM Multi-color real-time PCR detection system (Bio-Rad) allowing automatic collection of the florescence emission data. The reaction mixture contained 2ϫ solution of SYBR Green PCR Master Mix (PE Biosystems, Foster City, CA), 10ϫ dual (forward and reverse) primers mixture (0.25 M of each primer), and standards or samples (10 l/well) in duplicate, in a final volume of 50 l. Cycling conditions consisted of an activation step at 95°C for 2 min, followed by 45 cycles of 40 s of denaturing at 95°C, 20 s of annealing at 60°C, 40 s of extension at 72°C, and a final extension for 10 min at 72°C. The cytokine expression levels of each sample, run in duplicate, were normalized by hypoxanthine-guanine phosphoribosyltransferase (HPRT) mRNA. The results were expressed as 100 ϫ cytokine-mRNA/HPRT mRNA.
ELISA Analysis of Splenocyte IFN-␥ Production-Splenocytes of immunized mice were cultured as described above in triplicate in a 96-well plate, and the culture supernatants were harvested after 72 h for the detection of secreted IFN-␥ as measured by OPTEIA ELISA kits (BD Pharmingen).
CD8 ϩ -mediated T-cell Responses of Immunized Mice-Gag-specific CD8 ϩ -mediated responses were measured after a single DNA immunization followed in 3 weeks by in vivo expansion of Gag-activated T cells through inoculation with recombinant vaccinia-gag-pol (rVVgag-pol, 10 7 plaque-forming units, intraperitoneal). Five days later, the mice were injected with an immunodominant H-2K d -restricted Gag peptide epitope (position 65-73, AMQMLKETI) at a concentration of 10 g intravenously, and 2 h later they were sacrificed and splenocytes were collected. 51 Cr Release Assay-P815 target cells (2 ϫ 10 6 /ml) were pulsed overnight with 10 M p55Gag peptide AMQMLKETI in culture medium. Target cells were labeled with 50 Ci/ml [ 51 Cr]Na 2 CrO 4 (Amersham Biosciences) for 2 h. The cells were then washed three times, resuspended in culture medium, and diluted to 2 ϫ 10 5 cells/ml. The target cells (1 ϫ 10 4 ) were seeded onto V-bottom 96-well plates with equal volumes of 2% Triton (maximum release), media alone (minimum release), or a suspension of splenocytes from mice immunized for Gagspecific CD8 ϩ -mediated responses (effector cells) at different effector: target ratios, starting at 100:1, all in triplicates. After 4-h incubation at 37°C, 50 l of supernatant was transferred to a LumaPlate TM (Packard, Meriden, CT), dried, and counted in a TOPCOUNT NXT scintillation counter (Packard). The percentage of specific lysis (percent killing) was calculated as: Tetramer Staining-Splenocytes from mice immunized for Gag-specific CD8 ϩ -mediated responses were washed in FACScan buffer at 4°C and blocked with anti-mouse CD16/CD32, Fc␥III/II receptor antibody (BD Pharmingen) at a concentration of 10 g/ml for 10 min at 4°C. The splenocytes (1 ϫ 10 6 cells/well) were stained in duplicate with phycoerythrin-conjugated tetramer (H-2K d /AMQMLKETI, NIAID, National Institutes of Health, Atlanta, GA) and the immunoglobulin isotype control rat IgG1 antibody (BD Pharmingen) at a dilution of 1:100 for 1 h at 4°C. The cells were washed twice with FACScan buffer and incubated with the secondary antibody FITC-conjugated rat anti-mouse CD8 (BD Pharmingen) at a dilution of 1:100 for 30 min at 4°C. Finally, the cells were washed twice with FACScan buffer. Analysis was done on a BD Biosciences FACScan with CellQuest software. A minimum of 200,000 events was analyzed, and results were expressed as a percentage of tetramer binding cells on total CD8 ϩ cells.
Intracellular IFN-␥ Staining-Intracellular IFN-␥ staining for Gagspecific CD8 ϩ -mediated response was performed using the Cytofix/ Cytoperm Plus TM (with GolgiStop TM ) kit (BD Pharmingen). Briefly, GolgiStop TM containing monensin was added to the splenocytes harvested from the immunized mice for 2 h at 37°C. After incubation, cells were washed twice with FACScan buffer and nonspecific binding was blocked by incubating cells with anti-Fc␥R antibody (BD Pharmingen) at a concentration of 10 g/ml for 10 min at 4°C. The splenocytes (1 ϫ 10 6 cells/well) were stained in duplicate with FITC-conjugated rat antimouse CD8 antibody (BD Pharmingen) at a dilution of 1:100 for 30 min at 4°C. The cells were washed twice with FACScan buffer and resuspended in 200 l of Cytofix/Cytoperm solution at 4°C for 20 min. Cells were then washed twice with Perm/Wash solution and stained with phycoerythrin-conjugated rat anti-mouse IFN-␥ antibody or the immunoglobulin isotype control rat IgG1 antibody (BD Pharmingen) diluted 1:100. Analysis was done on a BD Biosciences FACScan with CellQuest software. A minimum of 200,000 events was analyzed, and the results were expressed as a percentage of IFN-␥ ϩ cells on total CD8 ϩ cells.
Statistical Analyses-Unpaired t test analyses and all graphs were made using StatView 5.0 (SAS Institute Inc., Cary, NC).

LAMP Luminal Domain Enhances Expression of Native
HIV-1 p55 Gag as a LAMP/Gag N Protein Chimera-Several DNA plasmids encoding HIV-1 p55Gag or as LAMP protein chimeras were synthesized in both pcDNA3.1 (Invitrogen) and AAV-ITR-modified (pITR) (65) vector backbones for studies of Gag protein expression and cellular trafficking in transfected cells (Fig. 1A). In a typical experiment to examine protein expression (Fig. 1B), COS cells transfected with the pcDNA3.1 or pITR plasmid constructs were analyzed by Western blotting and immunostaining with Gag-specific monoclonal antibodies. Because of cis-acting inhibitory sequences (56,57) The presence of the LAMP/Gag N chimera in protein bands of Ͼ220 kDa (lane 8) is attributed to multimerization of Gag at high concentration, and the Gag-specific protein bands below 100 kDa are attributed to proteolysis. A major Gag-specific protein fragment, with a molecular mass approximate to that expected for Gag adjoined to the LAMP transmembrane and cytoplasmic domains, suggests specific proteolytic cleavage of the LAMP luminal domain from the LAMP/Gag N chimera.  (67), was examined by immunofluorescence microscopy, comparing the localization of Gag N , endogenous LAMP-1 and LAMP-2, and MHC II labeled with monoclonal antibodies. In these cells, the endogenous LAMP-1 ( Fig. 2A, panels a-c) and LAMP-2 ( Fig. 2A, panels d-f) were extensively colocalized with MHC II. In general, the majority of vesicles containing LAMP-1 and LAMP-2 also contained MHC II; in contrast, a significant number of MHC II-containing vesicles lacked the LAMP molecules. Cells transfected with pITR plasmids encoding native Gag (Gag N ) without any specific trafficking signal showed Gag present throughout the cell in a diffuse distribution typical for cytoplasmic proteins in fibroblast cells (71) (Fig. 2B, panel a). A critical finding was that the majority of the Gag/lamp chimera molecules containing the translocon, transmembrane, and cytoplasmic sequences (SS/Gag ⌬INS /lamp), despite the presence of the YQTI lysosomal membrane targeting sequence at the carboxyl terminus of the cytoplasmic domain, were not colocalized with endogenous LAMP or MHC II of DCEK cells but were instead found as a vesicular/tubular component distributed in a manner similar to that seen in labeling of Golgi stacks (71) (Fig. 2B,  panel b). Trafficking of Gag as a LAMP/Gag N chimera to lysosomes was found to require the presence of the LAMP luminal domain in the construct (Fig. 2B, panel c). Confocal immunofluorescence microscopy confirmed these findings of colocalization of Gag with MHC II of the DCEK cells as a LAMP/Gag N chimera (Fig. 2C, panels d-f), but not of the SS/Gag ⌬INS /lamp (Fig. 2C,  panels a-c). Additional analysis of DCEK cells transfected with the LAMP/Gag N chimera encoding the complete LAMP, with 10 serial sections of the confocal Z-plane, showed that Gag as the LAMP/Gag N chimera (Fig. 2D, panels a1 and a2) and MHC II  2D, panels b1 and b2) were colocalized on most of the vesicles that strongly stained for the two proteins (Fig. 2D, panels c1 and c2). A significant amount of MHC II was also visualized in compartments not containing LAMP/Gag. These confocal images are representative of the majority of the cell staining.
CD4 ϩ -mediated Responses of Immunized Mice-DNA vaccines encoding LAMP chimeras of HIV gp160 and human papilloma virus E7 have been shown to provide robust antigenspecific CD4 ϩ -mediated responses as measured by assays of T cell proliferation (34,35,38). In the present study, we have measured CD4 ϩ -mediated responses by assays of mouse spleen cells for Gag-specific IL-2 and IL-4 mRNA expression and IFN-␥ protein production (Fig. 3). Mice, in groups of five animals for each plasmid construct, were injected twice with 50 g of DNA on days 1 and 21 with pITR plasmids encoding: 1) the control LAMP without gag sequences, 2) gag N , 3) SS/gag ⌬INS / lamp chimera lacking the LAMP luminal domain, and 4) LAMP/gag N , with Gag incorporated into the complete LAMP-1 immediately proximal to the LAMP transmembrane domain.
Animals were sacrificed on day 31, and splenocytes were prepared for assay of CD4 ϩ T-cell responses as described under "Experimental Procedures." The mutated SS/Gag ⌬INS /lamp, with the LAMP targeting signals and increased protein expression, elicited a greater response than did the native Gag, but in repeated experiments it was severalfold less effective than LAMP/Gag N (Fig. 3A). Experiments that included CD4 ϩ and CD8 ϩ T-cell blocking antibodies showed that IFN-␥ protein production was specific to CD4 ϩ cells, being completely blocked by prior incubation of splenocytes with an anti-CD4 ϩ but not an anti-CD8 ϩ monoclonal antibody (Fig. 3B). Statistical analysis of the CD4 ϩ IFN-␥ protein production following immunization with these vaccine constructs was analyzed with pooled data taken from all experiments performed with comparable protocols in the schedule, doses, and immunization routes (Fig. 3C). Each experiment involved spleens from two or three mice, with "n" experiments is derived from this vaccine construct used as a positive control in a large number of experiments testing other hypotheses or systems not included in this report. In all experiments there was an ϳ10-fold greater CD4 ϩ IFN-␥ response to the complete LAMP/Gag N chimera antigen as compared with the wild type Gag (p Ͻ 0.01). In addition, there were greater IL-2 and IL-4 cytokine mRNA responses of Gag-stimulated spleen cells from mice immunization with the LAMP/Gag N as compared with the response of cells from mice injected with pITR-DNA encoding the native p55Gag or SS/Gag ⌬INS /lamp (Fig. 3, D and E).
CD8 ϩ -mediated Responses of Immunized Mice-DNA vaccines are able to generate CD8 ϩ T-cell responses through MHC class I presentation of antigenic epitopes by transfected cells, but secondary expansion and memory in CD8 ϩ cells requires stimulus from CD4 ϩ T-helper cells (3). Gag-specific CD8 ϩmediated responses were measured after a single immunization with 50 g of the plasmid DNA as described above for CD4 ϩ -mediated responses, followed in 3 weeks by in vivo expansion of Gag-activated T cells through inoculation with recombinant vaccinia-gag-pol (rVVgag-pol). Five days later, the mice were injected with an immunodominant H-2K d -restricted Gag peptide epitope; after 2 h they were sacrificed and spleen cells were prepared. Gag-specific CTL was assayed for CD8 ϩspecific tetramer staining and intracellular IFN-␥ staining and by chromium release from P815 cells pulsed with the Gag peptide, all performed ex vivo (Fig. 4).
In general, the CD8 ϩ responses of mice injected with the unmodified gag were stronger than the corresponding responses of CD4 ϩ cells, as is commonly found with DNA vaccines. Nevertheless mice immunized with the complete LAMP/ gag N DNA uniformly showed greater CD8 ϩ responses in all three assays (Gag tetramer binding, intracellular IFN-␥ staining, and Gag-specific cell killing) than did those injected with the wild-type gag or SS/gag ⌬INS /lamp DNA preparations (Fig.  4, A-C). Pooled data from all experiments (Fig. 4, D-F) confirmed these results, with the average epitope-specific CD8 ϩ tetramer binding and IFN-␥ responses severalfold higher than the responses to native Gag or SS/Gag ⌬INS /lamp (p Ͻ 0.02). Under this experimental protocol of a single immunization with the DNA vaccine followed by in vivo expansion, 5-10% of the total CD8 ϩ cells became Gag antigen-specific. Removal of CD8 ϩ cells from the effector population abolished the Gagspecific cell killing, but removal of CD4 ϩ did not (not shown).
Combined Immunization with DNA Plasmids Encoding Native Gag and LAMP/Gag N -The response of mice to immunization with a mixture of the two plasmids encoding native Gag and the LAMP/Gag N chimera was studied to investigate the possibility that trafficking of the LAMP/Gag N chimera to lysosomes could result in a deficiency in the amount of protein available to the MHC I processing and presentation pathway of transfected cells. Experiments with cultured cells have shown that the great majority of endogenous LAMP, Ͼ95%, has a steady-state localization in the lysosomal membrane (15)(16)(17)(18)(19). Because all Gag in the transfected cells of mice injected with the DNA is expressed as a LAMP/Gag N chimera targeted to FIG. 4. The LAMP-targeted LAMP/Gag N chimera induces more efficient CD8 priming. Groups of mice (two to three/group) were immunized with 50 g of the indicated pITR plasmids. Three weeks after immunization, they were boosted intraperitoneally with 10 7 plaqueforming units of rVV-Gag-Pol. Five days later, 10 g of the H-2K d -binding HIV-1 Gag peptide was injected intravenously, and splenocytes were harvested after 2 h and analyzed for ex vivo activity. A, flow cytometry quantification of tetramer binding in the CD8 ϩ splenic population. B, flow cytometry quantification of Gag-specific CD8 ϩ T-lymphocytes producing IFN-␥. C, cytolytic activity in a 4-h 51 Cr release assay using P815 target cells pulsed with the H-2K d -binding HIV-1 Gag peptide. The effector cells from the different immunized mice were as indicated. Nonspecific lysis (using unpulsed P815 target cells) was Ͻ5% for all groups (not shown). D-F, statistical analyses of the CD8 ϩ response. The results of immune assays obtained from several different experiments performed in a similar manner were combined and grouped according to each vaccine plasmid. The "n" is the number of experiments performed with a particular plasmid, and each experiment represents the average response of two or three mice. No results were excluded from the analyses. The error bars indicate Ϯ one S.E. D, H-2K d tetramer-binding CD8 ϩ T-lymphocytes. E, Gag-specific IFN-␥-producing CD8 ϩ T-lymphocytes. F, cytolytic activity in a 4-h 51 Cr release assay at the indicated effector:target ratios. lysosomes, it was thought that the amount of protein available for MHC I presentation to CD8 ϩ T-cells might be limiting and that this could be corrected by coimmunization with DNA encoding native Gag. Mice were injected with equal amounts of pITR plasmids encoding gag N and LAMP/gag N , 25 g of DNA each. In this case, Gag expression in cells transfected with the Gag N plasmid would be as a cytoplasmic protein and presumably processed for MHC I presentation. Nevertheless, the combined administration of plasmids encoding Gag N and LAMP/ Gag N did not show a significant effect on either the CD8 ϩ or CD4 ϩ immune responses (Fig. 5). The data thus suggest that expression of Gag as a LAMP/Gag N chimera targeted to the endosomal/lysosomal vesicular pathway is not limiting to the mouse CD8 ϩ T-cell response.
Antibody Responses to HIV Gag DNA Vaccines-An exceptional finding was the strong antibody response to Gag resulting from immunization with the LAMP/Gag N chimera protein. Several immunizations have been performed, all showing 50-to 100-fold higher antibody titers in mice immunized with the LAMP/gag N constructs as compared with mice immunized with constructs encoding cytoplasmic wild type Gag or SS/Gag ⌬INS / lamp (Fig. 6A). As with the T-cell responses, there was no increase in the antibody response of mice injected with a combination of plasmids encoding Gag N and LAMP/Gag N . Additional experiments compared the antibody responses to the Gag DNA vaccine constructs under various protocols and two to four immunizations (Fig. 6B). The end-point antibody titers produced by Gag N and SS/Gag ⌬INS /lamp remained low, less than 10 3 , whereas the Gag-specific IgG end-point titer response after four immunizations with the LAMP/Gag N chimera protein was 5-to 10-fold greater, with a titer of almost 10 6 . The serum of mice immunized with LAMP/gag N was able to detect p55Gag on transfected cells (not shown).
IgG 1 and IgG 2a isotype titers of pooled sera after four immunizations with the LAMP/gag N DNA were 660,000 for IgG 1 and 24,000 for IgG 2a . In contrast, the gag N DNA elicited detectable humoral responses only after four immunizations, with equivalent IgG 1 and IgG 2a titers of 8,100. Thus the LAMP/Gag N protein chimera increased the antibody titers for both immunoglobulin isotypes, but more strongly for IgG 1 . These results are consistent with an increase in IFN-␥ production and IL-4 mRNA up-regulation, which was not detected in mice immunized with DNA encoding the native Gag.
Dose Response to Immunization with pITR/LAMP/gag N -This new p55gag DNA construct, with the high Gag protein expression, was maximally active in mice at the 10-to 50-g range of DNA for both antibody and cellular responses (Fig. 7,  A-D). Mice injected twice intramuscularly with 0.1-50 g of LAMP/gag N DNA showed maximum CD4 ϩ and CD8 ϩ IFN-␥ responses to 10 g of DNA, whereas the antibody response was dependent on a greater dose of DNA.

DISCUSSION
These studies document a novel HIV-1 p55gag DNA vaccine constructed as a chimera with the Gag sequence inserted into the luminal domain of the LAMP lysosomal membrane protein. This construct was developed as a means to provide additional necessary protein elements required for Gag trafficking to the vesicular lysosomal sites of antigen-presenting cells that contain MHC II and act in the presentation of antigen peptide epitopes to CD4 ϩ T helper cells. Gag encoded with the LAMP translocon, transmembrane, and cytoplasmic sequences con- FIG. 5. The immune response to LAMP/Gag N is not increased by combined immunization with DNA plasmids encoding native Gag (Gag N ) and LAMP/Gag N . A and B, mice, in groups of five animals for each plasmid construct, were injected on days 1 and 21 with 50 g of the pITR plasmid DNA constructs encoding the control LAMP without Gag sequences (LAMP), the native p55Gag (Gag N ), the LAMP/Gag N chimera, and with a mixture of DNA plasmids, 25 g each, encoding Gag N and LAMP/Gag N . Animals were sacrificed on day 31, and splenocytes were prepared for assay of CD4 ϩ T-cell responses as described under "Experimental Procedures." Results from the medium control and Gag-specific assays of a representative experiment are shown as mean Ϯ S.D. A, Gag-specific induction of IFN-␥ production. B, Gag-specific up-regulation of IL-2 mRNA expression. C-E, groups of mice (two to three/group) were immunized with 50 g of the indicated pITR plasmids (25 g each of the Gag N and LAMP/Gag N combination). Three weeks after immunization they were boosted intraperitoneally with 10 7 plaque-forming units of rVV-Gag-Pol. Five days later, 10 g of the H-2K d -binding HIV-1 Gag peptide was injected intravenously, and splenocytes were harvested after 2 h and analyzed ex vivo. C, flow cytometry quantification of tetramer binding in the CD8 ϩ splenic population. D, flow cytometry quantification of Gag-specific CD8 ϩ T-lymphocytes. E, cytolytic activity in a 4-h 51 Cr release assay at the indicated effector:target ratios.
taining the YXXØ lysosomal membrane targeting signal, without the LAMP luminal domain, did not colocalize with endogenous LAMP or MHC II molecules of transfected cells. This was in contrast to similar LAMP chimeras of membrane proteins, the CD44 hyaluronate receptor (21), HIV gp160 envelope (34), and the premembrane-envelope of dengue 2 virus, which do traffic to lysosomes as chimeras containing only the lamp transmembrane and cytoplasmic domains (39,40). Unlike these membrane proteins that normally enter the cellular vesicular system, Gag is a cytoplasmic protein, and we speculate that the lack of lysosomal membrane trafficking resulted from the absence of sequences recognized by chaperones or other molecules (72), or required vesicular trafficking elements such as appropriate protein folding or glycosylation. Our rationale for synthesizing gag as a chimera with the complete LAMP molecule was that the LAMP luminal domain might provide additional necessary structural and perhaps functional elements that facilitate the vesicular trafficking of the associated Gag protein. This proved to be the case, as demonstrated by confocal immunofluorescence microscopy of transfected DCE-K.ICAM.Hi7 cells, which express MHC II in a vesicular compartment colocalized with the endogenous LAMP-1 and LAMP-2. The extent of the LAMP/Gag N chimera localization with MHC II of these cells was striking, with Gag present in most of large MHC II-containing vesicles as found by serial confocal sections of individual cells. It should be noted that these results with the Gag protein may not apply to all cytoplasmic proteins; for example, a LAMP chimera of the human papilloma virus E7 lacking the LAMP luminal domain was shown to elicit an enhanced immune response of mice (35,36,38).
The stimulation of immune responses of mice injected with DNA encoding the LAMP/Gag N chimera, including antibody, CD4 ϩ T-cell IL-2 and IL-4 mRNA, and IFN-␥ protein production, and CD8 ϩ T-cell tetramer binding, intracellular IFN-␥ Gag-specific IgG and CD4 ϩ responses. Mice were immunized twice intramuscularly with the indicated dose of the LAMP/gag N construct at 3-to 4-week intervals. The LAMP vector (50 g/mouse) was the negative control. A, titration of end-point titers of total IgGs in pooled serum (two/group) from mice 2 weeks after the second immunization. B, IFN-␥ production by splenocytes harvested 2 weeks after the second immunization, as analyzed by ELISA after in vitro stimulation of splenocyte with the p55 Gag protein. Results are means Ϯ S.D. C and D, flow cytometry quantification of Gag-specific CD8 ϩ responses. Mice (two/ group) were immunized with the indicated dose of the plasmid, with the LAMP vector alone (50 g/mouse) as the control. Three weeks after immunization, mice were boosted intraperitoneally with 10 7 plaqueforming units of rVV-Gag-Pol. After 5 days, 10 g of the H-2K d -binding HIV-1 Gag peptide was injected intravenously, and splenocytes were harvested 2 h later for flow cytometry. C, H-2K d tetramer-binding CD8 ϩ T-lymphocytes. D, Gag-specific IFN-␥-producing CD8 ϩ T lymphocytes. Results are means Ϯ S.D. staining, and Gag-specific chromium release from target cells, adds to the now substantial evidence that trafficking of DNAencoded antigens to the LAMP endosomal/lysosomal vesicular compartments can be an important factor in the function of genetic vaccines. Additionally, in other ongoing studies, monkeys immunized with a comparable human LAMP/gag N construct have shown strong immune responses by each of the five animals after three immunizations (data not shown). A leading hypothesis is that the enhanced immune response results from trafficking of the LAMP/antigen chimera to the MHC II compartment with a resulting increased efficiency of antigen processing and epitope binding to MHC II. However, a variety of alternative mechanisms to explain the LAMP effect may be entertained. For example, the LAMP luminal domain may have some direct role in MHC II antigen processing and presentation independent of the colocalization with MHC II, or in the general activation of lysosomal function during dendritic cell maturation (30). It is also likely that antigen processing and the specific epitopes of the LAMP chimera antigens will differ from those derived from native Gag. LAMP targeting of antigen is not a normal function of APCs, and the products of antigen processing may differ from those associated with conventional endocytic antigen uptake to the MHC II compartment by dendritic cell receptors or other possible trafficking pathways. Trafficking of the LAMP/antigen chimera may provide access to different endosomal/lysosomal antigen-processing compartments where the specificities of the proteases may differ from those present in the natural pathway of the antigen taken into dendritic cells by the normal endocytic route. Moreover, the proteolytic processing might also be conditioned by the structure of the LAMP/antigen chimera or its relationship to other proteins involved in the lysosomal membrane vesicular trafficking. The heavily glycosylated, disulfide cross-linked luminal domain of LAMP is highly protease-resistant (73), as expected for sequences that reside in the lumen of lysosomes, and may partially protect certain Gag sequences from proteolysis. Whatever the mechanism, there is evidence that LAMP chimeras modify the repertoire of epitopes eliciting cellular and humoral immune responses. For example, a CD4 ϩ -mediated Tcell response of a minigene peptide epitope was abolished when constructed as a chimera with the lysosomal targeting signal of the cytoplasmic domain from the lysosomal integral membrane protein-II (LIMP-II) (74). It also has been reported that LAMP targeting of antigen increases the number of immunogenic peptide epitopes that activate CD4 ϩ T cells, thus inducing a qualitatively broadened immune response as compared with untargeted antigen (75). In our ongoing studies it is found that both the antibody and T-cell responses to Gag peptides of the LAMP/Gag N chimera also have significant differences from the responses to native Gag. A further question is the mechanism of the enhanced CD8 ϩ response. While this can be attributed to the stronger CD4 ϩ responses, the nature of MHC I peptide processing and presentation remain to be elucidated. Presumably, in the case of the LAMP-targeted Gag, the majority, if not all, of the Gag protein is degraded by the hydrolytic enzymes of the endosomal/lysosomal compartments of transfected cells rather than by the cytoplasmic proteasome complex. Nevertheless, immunization with a mixture of DNA plasmids encoding native Gag as well as the LAMP/Gag N chimera did not enhance the CD8 ϩ T-cell responses despite the expression of Gag localized to the cytoplasm and available for MHC I processing and presentation. Thus, the strong CTL response to the lysosomal targeting of the LAMP/Gag N chimera protein suggests that the LAMP/Gag N -processing pathway provided sufficient epitope cross-priming for MHC I presentation, or that the greater CD4 ϩ response compensated for a decrease in MHC I epitope presentation by transfected APCs. There is also the question of how trafficking of the LAMP/Gag N chimera enhances antibody responses to Gag. Commonly, DNA vaccines encoding intracellularly expressed proteins do not elicit strong B cell responses (9). Evidently, antibody epitopes of LAMP/antigen chimeras persist and are efficiently presented to B cells. It is known that while LAMP is predominantly localized in lysosomes, a small fraction is present at the cell surface. Our initial identification of LAMP was made by monoclonal antibody binding at the cell surface (68), and cycling of LAMP through the plasma membrane (76) and lysosomal fusion with the plasma membrane (77) have been described. Additionally, some populations of cells transfected in vivo may have lysosomal trafficking patterns that differ from those deduced from previous in vitro studies. For example, mouse peritoneal macrophages are shown to contain a post-translational isoform of LAMP-1 (M150) that is uniquely present on the surface of macrophages and acts as a costimulatory molecule driving the differentiation of naive CD4ϩ T cells into the Th1 subset (78). Thus, although we suggest that the principal element in the enhanced immune response is the improved delivery of Gag antigen to the MHC II compartment, there may be additional effects of LAMP trafficking and the LAMP luminal domain on the immune response to the DNA-encoded LAMP/Gag N protein chimera.
Another novel finding of this study is the enhanced protein expression of HIV-1 p55 Gag as a LAMP/Gag N chimera. Gag expression is minimal in cells transfected with DNA plasmid vectors containing only the native Gag gene. This low expression is attributed to mRNA cis-acting inhibitory or instability region sequence elements, which can be eliminated by gene mutation or humanization of codon usage (56,57,59,60,62,63). Additionally, the nuclear export of Gag mRNA has been promoted by the use of DNA chimeras encoding the constitutive transport RNA element of type D retroviruses that binds to a nuclear element and functions as export factors for the constitutive transport RNA element-containing RNAs (79). The effect of the LAMP luminal sequences 5Ј to the Gag sequences to increase Gag protein expression has not previously been reported and indicates the presence of additional mechanisms acting on the regulation of HIV-1 Gag mRNA and/or protein expression and stability. Ongoing studies suggest that the amino-terminal sequences of LAMP are critical to the increased protein expression, as indicated by deletions, substitutions, and modifications of the amino-terminal proximal sequence of the LAMP luminal domain (data not shown). It has previously been shown that the enhanced gag gene expression of the codonoptimized gag-pol was not inhibited when native gag sequences were inserted downstream, whereas native gag gene sequences upstream of the codon-optimized gag-pol led to diminished RNA levels (60). Expression of the LAMP/Gag N chimera was also increased by the addition of ITR sequences to the plasmid. This effect on transgene protein expression has previously been attributed to the increased stability of the plasmid DNA; the ITR domains can form trans-concatamers or palindromes in a cis configuration and have been shown to induce high level, long term gene expression (65). However, these effects of ITR sequences do not apply to all genes, and we have found no apparent increase in protein expression in several other systems.