INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Mycobacterium avium is the most common bacterial infection in patients with AIDS in the United States, leading to substantial morbidity and mortality (6, 21, 24). M. avium is difficult to treat because it is naturally resistant to many antibiotics. It also causes paratuberculosis in ruminants, disseminated lymphadenopathy in swine, and tuberculosis in birds, resulting in significant economic losses. M. avium is closely related to other slowly growing mycobacterial pathogens, including Mycobacterium tuberculosis (the cause of human tuberculosis) and Mycobacterium leprae (the cause of leprosy). The development of effective vaccines against mycobacteria is an important public health goal. An effective M. avium vaccine would be useful in high-risk human and animal populations and might serve as a model for tuberculosis and leprosy vaccines.

DNA vaccination uses plasmids to express antigenic proteins in host cells. DNA vaccines produce both humoral and cell-mediated immune responses and have led to protective immunity in a wide variety of animal models of infectious diseases (10). DNA vaccines yield partial protection in murine models of M. tuberculosis infection (23, 39, 43).

We developed DNA vaccines and tested them in an established murine model of M. avium infection. We generated DNA vaccines which expressed mycobacterial antigens as fusion proteins with enhanced green fluorescent protein (EGFP). EGFP is a mutant form of the jellyfish GFP; it produces more intense fluorescence than the native protein (8). Expression of mycobacterial antigens as EGFP fusion proteins allowed us to monitor the amounts and localization of the expressed proteins in mammalian cells. We developed DNA vaccines expressing the M. avium 65-kDa antigen, the Mycobacterium bovis 85A antigen, and the M. avium 85B antigen as EGFP fusion proteins. Each of these antigens is secreted by pathogenic mycobacteria and generates cellular and humoral immune responses in natural infections. This report describes the development of these DNA vaccines, their expression in mammalian cells, and their protective efficacy in an animal model of M. avium infection.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids used for DNA vaccination. The plasmids used in this study are described in Table 1. pEGFP-N1 (pEGFP) (Clontech Laboratories, Inc., Palo Alto, Calif.) is a mammalian expression vector which contains a cytomegalovirus promoter active in mammalian cells, a multiple cloning site, the EGFP gene, RNA-stabilizing sequences, an Escherichia coli origin of replication, and antibiotic selection markers. Douglas B. Lowrie (National Institute for Medical Research, London, United Kingdom) kindly provided pCMV4.65 and pCMV7.36, which are optimized versions of the plasmids described by Tascon et al. (39).

Construction of plasmids expressing mycobacterial antigen-EGFP fusion proteins. We inserted mycobacterial genes into pEGFP to generate genes expressing fusion proteins with EGFP at the carboxy terminus. We amplified each bacterial gene in its entirety by PCR with primers which modified both ends of the gene. Each forward primer inserted a specific restriction site and replaced the bacterial start codon with a Kozak consensus sequence (GCCACCATGG; positions 6 through +4) to enhance mammalian expression (27). Each reverse primer removed the stop codon and added a second restriction site. Each PCR product was then digested with two restriction enzymes and ligated into the multiple cloning site of pEGFP in frame with the EGFP gene. Each construct was designed to express a fusion protein comprising the complete mycobacterial antigen, a short amino acid spacer, and the complete EGFP. In some cases the second amino acid of the mycobacterial antigen was intentionally changed by the substitution of the Kozak consensus sequence G at nucleotide position 4.

p65K-EGFP. Fouad A. K. El-Zaatari (Baylor College of Medicine, Houston, Tex.) kindly provided pMptb20, which contains the 65-kDa antigen gene from M. avium subsp. paratuberculosis ATCC 43015 (11). We amplified the mycobacterial gene from plasmid DNA by PCR. The forward primer introduced an EcoRI site and the Kozak consensus sequence (5' CGGAATTCGC CACCATGGTG CTAGGTCGGG ACGGTGAGG 3'; the EcoRI site is underlined, and the Kozak consensus sequence is in italics). The reverse primer removed the stop codon and introduced a BamHI site (5' CGGGATCCCA GAAGTCCATGCCGCCCATGC 3'; the BamHI site is underlined). The PCR product was digested and ligated into the EcoRI and BamHI sites of pEGFP. The resultant plasmid, p65K-EGFP, was predicted to express a fusion protein comprising the complete 65-kDa antigen, a 6-amino-acid spacer, and the complete EGFP.

p85A-EGFP. The National Institutes of Health AIDS Research and Reference Reagent Program, Rockville, Md., kindly provided pAg1, which contains the antigen 85A gene from M. bovis BCG (9). We amplified this gene by using the primers 5' TGAGCGAGCT CGCCACCATG GAGCTTGTTG ACAGGGTTCG TG 3' and 5' ACGCGGATCC GCGCCCTGGG GCGCGGGCCC GGTG 3' and subcloned the PCR product into the SacI and BamHI sites of pEGFP.

p85B-EGFP. Naoya Ohara (Nagasaki University, Nagasaki, Japan) kindly provided pAASp56, which contains the antigen 85B gene from M. avium ATCC 15769 (31). We amplified this gene by using the primers 5' TCCGCTCGAG CCACCATGGCAGATCTGAGC GAGAAGGTCC 3' and 5' CGGAATTCCG CCGCCGCCCG GGGACG 3' and subcloned the PCR product into the XhoI and EcoRI sites of pEGFP.

pSodA-EGFP. Vincent Escuyer (Institut National de la Santé et de la Recherche Médicale, Paris, France) kindly provided pSOD2, which contains the sodA gene from M. avium TMC 724 (13). We amplified this gene by using the primers 5' GTCCGCTCGAGCCACCATGG CTGAATACAC CCTGCCCGAC 3' and 5' ACCGGAATTCTGCCGAAGAT CAGGCCTTGG 3' and subcloned the PCR product into the XhoI and EcoRI sites of pEGFP.

Plasmid preparation and verification. Plasmids were transformed by heat shock into competent E. coli INVF' cells (Invitrogen Corp., Carlsbad, Calif.). Plasmid DNA was prepared by using the EndoFree Plasmid Maxi Kit (Qiagen, Inc., Valencia, Calif.) and diluted in phosphate-buffered saline (PBS) to a final concentration of 1 mg/ml for vaccination. Each mycobacterial antigen-EGFP plasmid construct was verified by restriction digestion and by sequencing the complete insert DNA at the Duke University Sequencing Facility.

Cell culture and transfection. Murine 3T3 fibroblasts (2 × 10⁵/well) were grown in 35-mm-diameter tissue culture plates in Dulbecco's medium (Life Technologies, Gaithersburg, Md.) with 10% fetal bovine serum (HyClone Laboratories, Inc., Logan, Utah), penicillin (100 U/ml), and streptomycin (100 µg/ml). Cells in each well were transfected by using 2 µg of plasmid DNA and 5 µg of Lipofectin Reagent (Life Technologies) in 0.1 ml of Opti-MEM Reduced Serum Medium (Life Technologies) with a 12-h incubation at 37°C. Transfected cells were washed and incubated in complete Dulbecco's medium for 48 h prior to analysis. Cells expressing fluorescent proteins were visualized by inverted fluorescence microscopy.

Colocalization experiments. Murine 3T3 fibroblasts expressing fluorescent fusion proteins were fixed with paraformaldehyde, washed with PBS, and stained for 1 h at 37°C with 4',6-diamidino-2-phenylindole (DAPI) or tetramethyl rhodamine isothiocyanate (TRITC)-labelled wheat germ agglutinin (both from Sigma Chemical Co., St. Louis, Mo.). Transferrin was labelled by indirect immunofluorescence with goat antitransferrin serum as the primary antibody and phycoerythrin-conjugated rabbit anti-goat immunoglobulin G as the secondary antibody (both from Sigma). TRAP was labelled by using rabbit anti-TRAP (previously designated SSR [29]) as the primary antibody and phycoerythrin-conjugated goat anti-rabbit immunoglobulin G as the secondary antibody (Sigma). Cells were visualized by fluorescence microscopy.

Vaccination. We vaccinated 4- to 6-week-old female C57BL/6 mice (National Cancer Institute, Frederick, Md.) in groups of 10. Mice were anesthetized with inhaled halothane and injected with 50 µg of plasmid DNA in 50 µl of PBS in each tibialis anterior muscle (100 µg DNA/mouse) by using an insulin syringe with a 28-gauge needle. Mice were vaccinated at week 0, week 3, and week 6 with plasmid DNA. Negative control mice were injected with PBS only. Positive control mice were injected intradermally once at week 0 with 50 µl of Tice BCG vaccine (Organon Teknika Corp., Durham, N.C.), containing 0.5 × 10⁶ to 4 × 10⁶ CFU of live mycobacteria. BCG vaccine is a live attenuated strain of M. bovis and is used in humans to prevent tuberculosis. Previous studies have demonstrated that BCG vaccine protects against M. avium in animal models (22, 33).

Mycobacterial infection. Kevin A. Nash and Clark B. Inderlied (Children's Hospital, Los Angeles, Calif.) kindly provided a spleen homogenate from a mouse infected with M. avium MAC 101. This serovar 1 strain was originally isolated from the blood of a human with AIDS and is widely used in mouse models (16). To maintain virulence, only transparent colonies from primary cultures of murine spleen tissue were used (16). Mycobacteria were grown from spleen homogenates for 10 days at 37°C on Middlebrook 7H11 agar with OADC (oleic acid-albumin-dextrose-catalase) (Difco, Detroit, Mich.) supplementation. Transparent colonies were selected, suspended in PBS, pooled, and stored in aliquots at 70°C at a concentration of 5 × 10⁸ CFU/ml. Three weeks after the final injection of plasmid DNA (week 9), mice were anesthetized with inhaled methoxyflurane and injected intraperitoneally with 5 × 10⁶ CFU of M. avium MAC 101. Vaccination groups of 10 were divided in half; these groups of 5 mice were sacrificed 4 and 8 weeks after infection (weeks 13 and 17).

Quantitative cultures of spleen. Mice were anesthetized with methoxyflurane and euthanized by cervical dislocation. Spleens were removed by aseptic dissection, weighed, homogenized, and serially diluted in Middlebrook 7H9 broth (Difco). Aliquots of the suspension were plated onto Middlebrook 7H11 agar with OADC supplementation. Colonies counts were expressed as log₁₀ CFU per spleen. The logarithmic transformation was used because the CFU counts were not normally distributed. The spleen weight and log₁₀ CFU in each experimental group were compared to those in the negative control group by using a two-tailed t test. A P value of <0.05 was considered statistically significant.

RT-PCR of cytokine mRNA. We amplified mRNAs for -actin, gamma interferon (IFN-), and interleukin-4 (IL-4) from spleen tissues by using reverse transcriptase PCR (RT-PCR). Spleen samples were frozen in liquid nitrogen at the time of euthanasia. Spleen RNA was purified according to the manufacturer's protocol by using Trizol reagent (Life Technologies). cDNA was synthesized in a 20-µl volume containing 1 µg of total RNA, 5 mM random hexamers, 0.5 mM (each) four deoxynucleoside triphosphates, 20 U of RNase inhibitor (Promega, Madison, Wis.), 50 mM Tris-HCl, 75 mM KCl, 3 mM MgCl₂, 10 mM dithiothreitol, and 100 U of Moloney murine leukemia virus RT (Life Technologies). This mixture was incubated for 1 h at 37°C. PCR was conducted in a 50-µl volume containing 1 µl of cDNA, 5 µl of GeneAmp 10× PCR Buffer II (Perkin-Elmer Cetus, Emeryville, Calif.), 1.5 mM MgCl₂, 100 nM each primer, 200 µM (each) four deoxynucleoside triphosphates, and 1.25 U of AmpliTaq Gold DNA polymerase (Perkin-Elmer Cetus). An initial denaturation of 12 min at 95°C was followed by 30 cycles of denaturation at 94°C for 30 s, annealing at 50°C for 1 min, and extension at 72°C for 2 min, followed by a final extension at 72°C for 10 min. Primers were previously published (35). IFN- and IL-4 were amplified by using undiluted cDNA, and -actin was amplified by using a 1:100 dilution. These cDNA quantities were selected empirically such that PCR amplification did not reach a plateau in 30 cycles. Each PCR product was subjected to electrophoresis in 3% agarose, stained with ethidium bromide, and visualized by UV transillumination. The intensity of each fluorescent band was compared to those of DNA standards by using the EDAS 120 digital imaging system (Kodak, Rochester, N.Y.).

Animal care. All animal protocols were approved by the Institutional Animal Care and Use Committees of both the Durham Veterans Affairs Medical Center and Duke University. The method of euthanasia followed the recommendations of the American Veterinary Medical Association. Death was not an experimental end point.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

We constructed plasmids to express mycobacterial antigens as EGFP fusion proteins. We confirmed the expression of the fusion proteins in vitro in murine fibroblasts and in vivo in murine skeletal muscle. We then tested the plasmid DNA vaccines in a murine model of M. avium infection. After completing the vaccination series, mice were injected intraperitoneally with live M. avium bacteria. Protective immunity was determined by spleen weight and spleen CFU at 4 and 8 weeks after infection.

Plasmid sequencing. Plasmid constructs were verified by restriction enzyme digestion and sequencing of the insert (Table 1). Three plasmid constructs had the predicted insert sequences (p65K-EGFP, p85A-EGFP, and p85B-EGFP). However, plasmid pSodA-EGFP had a frameshift error at the site of the forward PCR primer. It was predicted to express a 39-amino-acid polypeptide with only the initial five amino acids corresponding to the M. avium SodA protein. This construct was used in expression and vaccination studies as a negative control plasmid.

Protein expression in murine fibroblasts. We transfected murine 3T3 fibroblasts with pEGFP and each of the four mycobacterial antigen-EGFP constructs. Fluorescent protein was visualized after transfection with pEGFP, p65K-EGFP, p85A-EGFP, and p85B-EGFP but not after transfection with the pSodA-EGFP plasmid containing the frameshift mutation. Fluorescence indicates the production of full-length mycobacterial antigen-EGFP fusion proteins, since the EGFP component is at the carboxy terminus. Three distinct patterns were observed (Fig. 1). EGFP was uniformly distributed in the nucleus and cytoplasm. The 65K-EGFP protein was uniformly distributed in the cytoplasm, with little or no nuclear expression. The 85A-EGFP and the 85B-EGFP fusion proteins were distributed in a speckled cytoplasmic pattern. These patterns of cellular localization were reproducible in at least three independent transfections for each plasmid. The mycobacterial antigen-EGFP fusion proteins were visible in a smaller proportion of cells and fluoresced with a reduced intensity compared to EGFP. This may be due to decreased transfection efficiency (larger plasmid), decreased translation efficiency (larger coding sequence or mycobacterial coding preferences), decreased fluorescence of the fusion proteins, or some combination of these factors.

View larger version (14K): [in this window] [in a new window]   FIG. 1.   Expression and localization of EGFP and mycobacterial antigen-EGFP fusion proteins in vitro. Murine 3T3 fibroblasts were transiently transfected with pEGFP, p65K-EGFP, p85A-EGFP, or p85B-EGFP and visualized 3 days later by fluorescence microscopy. Two photographs are shown for each plasmid. EGFP produced a homogenous pattern in the nucleus and cytoplasm. The 65K-EGFP fusion protein produced a homogenous cytoplasmic pattern with little or no protein in the nucleus. The 85A-EGFP (not shown) and 85B-EGFP fusion proteins produced speckled cytoplasmic patterns.

Expression in mouse muscle. We injected skeletal muscle to confirm in vivo protein expression from the mycobacterial antigen-EGFP constructs. We injected 50 µg of pEGFP, p65K-EGFP, and p85B-EGFP into the tibialis anterior muscles of C57BL/6 mice. Four days later, the mice were euthanized, and frozen muscle sections were examined by fluorescence microscopy (Fig. 2). All three plasmids expressed fluorescent protein throughout the length of individual muscle cells. Both EGFP and 65K-EGFP were distributed in a uniform pattern, but 85B-EGFP had a speckled localization, similar to that seen in cultured fibroblasts. Green fluorescence was not observed in muscle fibers injected with PBS.

View larger version (27K): [in this window] [in a new window]   FIG. 2.   Expression and localization of EGFP and 85B-EGFP fusion protein in vivo. Tibialis anterior muscles of C57BL/6 mice were injected with pEGFP, p65K-EGFP, or p85B-EGFP. Four days later, the mice were euthanized, and frozen sections of the muscle were visualized by fluorescence microscopy. EGFP and 65K-EGFP (not shown) produced homogenous fluorescence, but 85B-EGFP produced a speckled pattern. The clear areas inside the muscle cells are frozen-section artifacts.

Protection against experimental M. avium infection. Mice were vaccinated with plasmid DNA at weeks 0, 3, and 6, challenged by intraperitoneal M. avium injection at week 9, and euthanized at week 13 or 17. Injection of PBS alone served as a negative control. A single intradermal injection of BCG vaccine at week 0 served as a positive control. Two independent experiments (experiments 1 and 2) were conducted with the same methods but different DNA vaccines. Mice in all groups developed disseminated M. avium infection with high bacterial loads in the spleen. All mice developed splenomegaly. A typical spleen weight in uninfected mice of this age is 0.1 g. All mice survived the infection, gained weight, and displayed normal activity.

Vaccination experiment 1. Experiment 1 included two DNA vaccines (p65K-EGFP and pCMV4.65), each of which expressed a mycobacterial 65-kDa antigen (Fig. 3). Vaccination with either BCG or p65K-EGFP yielded a significant reduction in the degree of splenomegaly at 4 weeks after infection compared to the saline control group (Fig. 3A), but there was no difference among the four groups at 8 weeks. Similarly, vaccination with either BCG or p65K-EGFP yielded a reduction in M. avium CFU in the spleen 4 weeks after challenge (Fig. 3B) but had no effect at 8 weeks. p65K-EGFP generated a 0.6-log-unit (fourfold) reduction in CFU at 4 weeks, which was less than that produced by BCG. Mice vaccinated with pCMV4.65 did not differ from negative controls in spleen weight or M. avium CFU at either 4 or 8 weeks after infection.

View larger version (25K): [in this window] [in a new window]   FIG. 3.   Spleen weights (A) and spleen CFU (B) after M. avium infection in vaccination experiment 1. C57BL/6 mice were vaccinated with three intramuscular injections of saline, p65K-EGFP, or pCMV4.65 or with one intradermal injection of BCG vaccine and then challenged with intraperitoneal M. avium, as described in Materials and Methods. All groups contained five mice, and all data represent means ± standard errors. Each experimental group was compared to the saline group by using a two-tailed t test. NS, not significant.

Vaccination experiment 2. Experiment 2 included four DNA vaccines (p85A-EGFP, p85B-EGFP, pSodA-EGFP, and pCMV7.36) (Table 1). The results are shown in Fig. 4. Compared to negative controls, vaccination with either BCG or p85A-EGFP yielded a significant reduction in the degree of splenomegaly at 4 weeks after infection, but there was no significant effect at 8 weeks (Fig. 4A). Vaccination with either BCG or p85A-EGFP yielded a reduction in spleen CFU at both 4 and 8 weeks after challenge (Fig. 4B). p85A-EGFP generated a 0.9-log-unit (eightfold) reduction in CFU at 4 weeks and a 0.6-log-unit (fourfold) reduction at 8 weeks, both of which were similar to the reduction generated by BCG. Mice vaccinated with p85B-EGFP, pSodA-EGFP, or pCMV7.36 did not differ from negative controls in spleen weight or CFU at either 4 or 8 weeks after infection.

View larger version (29K): [in this window] [in a new window]   FIG. 4.   Spleen weights (A) and spleen CFU (B) after M. avium infection in vaccination experiment 2. C57BL/6 mice were vaccinated with three intramuscular injections of saline, p85A-EGFP, p85B-EGFP, pSodA-EGFP, or pCMV7.36 or with one intradermal injection of BCG vaccine and then challenged with intraperitoneal M. avium, as described in Materials and Methods. All groups contained five mice, and all data represent means ± standard errors. Each experimental group was compared to the saline group by using a two-tailed t test. NS, not significant.

Reproducibility. The vaccine model used in these experiments has been used successfully by other researchers. The reproducibility of the vaccine model was demonstrated by the saline and BCG groups, for which the procedure was repeated in the two experiments (Fig. 3 and 4). When the results of experiment 1 and experiment 2 were compared, there were no significant differences in either spleen weight or spleen CFU in either the saline or BCG groups at either 4 or 8 weeks after infection (P > 0.05 for each of the eight comparisons).

Side effects. Some mice vaccinated with intradermal BCG vaccine developed local ulcers as described for humans receiving this vaccine (5). Mice receiving saline or plasmid DNA vaccines had no local signs and had normal gait.

Cytokine mRNAs in spleens after M. avium infection. As an indicator of local Th1/Th2 responses to infection, we amplified cytokine mRNAs from spleen tissue after infection by RT-PCR with primers directed at IL-4, IFN-, and -actin (control). We determined the quantity of each PCR product by using a digital imaging system. IL-4 PCR products were absent or faint at all time points (data not shown). IFN- PCR products increased at 4 weeks after infection and declined toward baseline levels at 8 weeks after infection. However, the quantities of IFN- RT-PCR products in vaccinated mice did not differ from those in the mock-vaccinated controls at either 4 or 8 weeks after infection. This assay had a power of over 80% to detect threefold differences in the amounts of RT-PCR products between groups (alpha = 0.05) (data not shown). Data from vaccination experiment 1 are displayed in Fig. 5. Vaccination experiment 2 yielded similar results.

View larger version (18K): [in this window] [in a new window]   FIG. 5.   Spleen IFN- mRNA after M. avium infection as determined by RT-PCR. Total RNA was extracted from spleen tissue at 4 and 8 weeks after infection. Murine IFN- mRNA was amplified by RT-PCR. PCR products were subjected to electrophoresis on a 3% agarose gel and stained with ethidium bromide. Each lane shows the 243-bp IFN- PCR product from a single mouse. PCR products from mice vaccinated with pCMV4.65 (lanes 1 to 3), p65K-EGFP (lanes 4 to 6), BCG (lanes 7 to 9), and saline (lanes 10-12) are shown. IFN- mRNA was increased at 4 weeks in all groups and declined toward baseline levels at 8 weeks. There were no significant differences among the groups.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Expression of mycobacterial antigens as EGFP fusion proteins. The visualization of EGFP provides rapid confirmation of foreign protein expression in mammalian cells in vitro and in vivo (3). Western blot analysis can also be used for this purpose, but it requires the availability of specific antibodies (23, 39, 43). EGFP fusion proteins can be used to confirm expression of antigens for which specific antibodies are not available. In contrast to many expression markers, EGFP does not require the addition of substrate. EGFP expression in tissue can be quantitated with a fluorometer. Quantitative expression analysis will be important when DNA vaccines are combined with adjuvants, because some adjuvants reduce protein expression by DNA vaccines (40).

Cellular localization of mycobacterial antigen-EGFP fusion proteins. We observed two patterns of localization of mycobacterial antigen-EGFP fusion proteins. The 65K-EGFP protein was distributed uniformly in the cytoplasm, but the 85A-EGFP and 85B-EGFP proteins had a speckled pattern. These patterns were similar in cultured fibroblasts and intact murine muscles. These antigens may contain signal sequences recognized by mammalian cells, although we were unable to identify a specific cellular compartment. Alternatively, the speckled pattern of 85A-EGFP and 85B-EGFP may simply represent intracellular aggregation due to poor solubility of the fusion proteins. Mycobacterial pathogens replicate in vivo inside host macrophages and secrete large amounts of these antigens. Intracellular localization of these antigens in host macrophages may modulate the host immune response in natural infection.

Protection from M. avium challenge. Two plasmid DNA vaccines (p85A-EGFP and p65K-EGFP) generated protective immunity in this animal model of M. avium infection, as demonstrated by reduced spleen weight and reduced spleen CFU (up to a 0.9-log-unit CFU reduction). Several studies have reported similar CFU reductions with DNA vaccines in mycobacterial infection models. Tascon et al. (39) used DNA vaccines expressing the 65-kDa antigen and the 36-kDa antigen and found about a 1.0-log-unit CFU reduction in inbred mice and up to a 2-log-unit CFU reduction in crossbred or outbred mice after intraperitoneal M. tuberculosis challenge. Huygen et al. (23) used DNA vaccines expressing antigen 85A to generate a 0.7- to 1.0-log-unit CFU reduction after aerosol challenge with M. tuberculosis. Zhu et al. (43) used a DNA vaccine expressing the 38-kDa antigen to produce a 0.5-log-unit CFU reduction in spleen after intraperitoneal challenge with M. tuberculosis. Erb et al. (12) demonstrated no protection from intranasal BCG challenge with DNA vaccines expressing the 19-kDa antigen and the AhpC protein.

Protection across mycobacterial species. Antigens from different mycobacteria often have substantial sequence similarity, and monoclonal antibodies directed at these antigens often react with multiple mycobacterial species (42). DNA vaccines expressing these antigens sometimes provide cross-species protection. The DNA vaccine with the highest level of protection against M. avium in our study (p85A-EGFP) actually expressed the M. bovis BCG 85A antigen, which has 79% amino acid identity with its M. avium homologue (32). It is also paradoxical that p85A-EGFP was protective, while p85B-EGFP was not, since M. avium strains express antigen 85A at lower levels than antigen 85B (32). Tascon et al. used a DNA vaccine expressing the M. leprae 65-kDa antigen (pCMV3.65) to protect against M. tuberculosis challenge (39). We used a modified version of this DNA vaccine (pCMV4.65) but achieved no protection against M. avium. The M. leprae and M. avium 65-kDa antigens have 87% amino acid identity. We did observe modest protection with p65K-EGFP, which expresses the M. avium 65-kDa antigen. However, p65K-EGFP also differs from pCMV4.65 in the plasmid backbone and the presence of the EGFP marker.

Limitations of protective immunity results. Our results should be interpreted with several limitations in mind. First, the reductions in spleen weight and CFU which we observed were highly statistically significant, but they represent only modest levels of protection. Second, we do not know the duration of protective immunity, since all mice were challenged 3 weeks after the vaccination series. Most studies of DNA vaccines against mycobacteria have also used short intervals (1 to 4 weeks) between the completion of the vaccination series and the challenge (12, 28, 39, 43). However, Huygen et al. (23) demonstrated similar levels of protective immunity when mice were challenged at either 4 weeks or 10 weeks after completion of the vaccination series. Third, the level of protective immunity in our experiments decreased between 4 and 8 weeks after challenge for both BCG and DNA vaccines. Similar results were obtained with a mouse model of aerosol infection with M. tuberculosis (36). BCG-vaccinated mice had lower numbers of CFU from 3 to 7 weeks after infection, but there was no difference from 8 to 15 weeks. Fourth, each of the protective DNA vaccines was tested only once in a single animal model. Further experiments are needed to determine the reproducibility and generalizability of our results.

Subsets of T-cell responses. Two subsets of CD4⁺ T cells (Th1 and Th2) are identified by their patterns of cytokine production. A variety of lines of evidence support the protective role of a Th1 response in mycobacterial infection. Mice lacking the IFN- gene and humans with mutations in the IFN- receptor are highly susceptible to mycobacterial infection (7, 30). Exogenous administration of Th1 cytokines in humans and mice enhances resistance to mycobacterial infection (20, 26, 30). Th1 cytokine production at the site of infection is associated with control of mycobacterial infection. Yamamura et al. (41) demonstrated that leprosy patients with low M. leprae burdens express Th1 cytokines in their skin lesions, whereas patients with high M. leprae burdens express Th2 cytokines.

Specific versus nonspecific vaccine effects. The host response to mycobacterial infections includes macrophages and natural killer cells, components of the innate immune system (1, 4, 14, 19). Bacterial DNA (including plasmid DNA) has prominent immunostimulatory effects, including Th1 cytokine secretion, macrophage activation, and enhancement of natural killer cell activity (34). Further experiments are needed to determine whether nonspecific immune stimulation contributes to the protection generated by DNA vaccines. These experiments could include trials with a longer delay between vaccination and challenge, cytokine measurement after vaccination, and determination of macrophage and natural killer cell activation after vaccination.

Relevance to human M. avium infections. The M. avium complex is an important cause of disseminated disease in AIDS patients. An effective DNA vaccine would provide a clinical benefit for these patients. Although pulmonary disease is caused by both M. avium and M. intracellulare, disseminated disease in AIDS patients is nearly always caused by M. avium (6, 21, 38). M. avium strains isolated from AIDS patients belong to several serovars based on surface glycopeptidolipids, but they share highly conserved protein antigens (38).