¹The Wistar Institute, 3601 Spruce Street, Philadelphia, PA 19104, USA

²ICRF Molecular Oncology Unit, Hammersmith Hospital, London, UK

Correspondence to: HCJ Ertl

The effect of palindromic CpG sequences on the B cell response to plasmid vectors expressing a highly immunogenic viral glycoprotein was investigated. Methylation of the CpG sequences of bacterial expression vectors abolished their ability to induce an antibody response to the transgene product in mice. The antibody response could be rescued by concomitant injection of oligonucleotides carrying immunostimulatory sequences. The B cell response to two plasmid vectors, both expressing the same viral glycoprotein but containing a different content of the highly stimulatory AACGTT motif, was compared. Comparable B cell responses were induced to the two constructs given at an optimal vaccine dose while the vector containing additional palindromic sequences resulted in higher antibody titers at a suboptimal dose. These data indicate that deletion of CpG motifs or methylation of such sequences in plasmid DNA can abrogate the immune response to the vector encoded antigen and might thus enhance their usefulness as gene therapy vehicles.

gene therapy; expression vector; CpG sequences; methylation; immune responses

Introduction

Palindromic CpG sequences have immunostimulatory activities and are present at a high rate in bacterial DNA but are rare or methylated in the mammalian genome. They cause activation of macrophages and natural killer cells, induce release of cytokines including IL-12, IL-6 and IFN-, and polyclonally stimulate B cells.^1,2,3 The ability of unmethylated, bacterial CpG sequences to trigger innate immune responses is assumed to be instrumental for the immune response to DNA vaccines by providing an activation signal to antigen presenting cells (APCs).⁴

Gene therapy seeks permanently to replace missing or faulty genes by transferring such genes into somatic cells with suitable carriers. Unfortunately, many gene therapy vehicles, such as E1-deleted adenoviral recombinants, induce potent B and T cell-mediated immune responses which cause rapid elimination of transgene expressing cells and prevent efficacious gene transfer upon readministration.^5,6,7 In the case of mis-sense mutations or deletions, the protein that needs to be replaced can also stimulate such immune responses, thus further complicating long-term gene therapy even with carriers that are non-immunogenic. Hemophilia, caused by functional lack of factor IX or VIII activity, is one of the genetic diseases which could potentially be cured by gene therapy. The current therapy for this disease complex consists of intravenous application of clotting factors either derived from plasma or generated as a recombinant protein. In a significant proportion of patients this treatment eventually leads to formation of antibodies to the missing clotting factor which inhibit their coagulation activity.^8,9 A similar reaction can be anticipated in these patients upon gene transfer of clotting factors using most of the currently available gene replacement vehicles.^10,11

Thus, approaches have to be developed that allow gene delivered replacement of a potentially immunogenic protein without inducing an immunological reaction. Here, we tested if the antibody response to a highly immunogenic viral glycoprotein expressed by a plasmid vector could be ablated or at least diminished by reducing the immunostimulatory activity of the carrier. CpG methylation of the plasmid vectors was shown to abolish the B cell response to the vector-encoded protein. This effect was not related to a reduced expression of antigen upon methylation of the promoter. Furthermore, vectors with reduced AACGTT motifs were also shown to result, at least at low vector doses, in a reduced B cell response.

Results

The effect of SssI methylation on the B cell response to the vector encoded antigen

To test if unmethylated CpG sequences are required to provide an inflammatory signal to the innate immune system, which in turn promotes generation of an antigen-specific immune response, groups of C3H/He mice were immunized with 50 g of unmethylated or SssI methylated pVR.amp.rab.gp vector. An additional group was immunized with 50 g of the empty pVR.amp control vector. Mice were bled 2, 4 and 8 weeks later and antibody titers to rabies virus were tested by an ELISA. As shown in Figure 1, unmethylated pVR.amp.rab.gp vector induced a strong B cell response to the rabies virus antigen that could be detected as early as 2 weeks after immunization and reached peak titers within 4 weeks and then remained at plateau levels at 8 weeks after vaccination. The empty control vector failed to induce an immune response and antibody levels were comparable with those of sera from unimmunized C3H/He mice (data for normal mouse sera are not shown). Methylated plasmid DNA expressing the viral antigen only induced a marginal antibody response that was statistically insignificant (for all dilutions compared with the control serum, P values obtained by Student's t test were >0.05) at any of the three time-points tested. These results confirm an earlier study conducted in a different system which showed that unmethylated CpG sequences are required for induction of an immune response to DNA vaccines.¹² The CMV promoter used in the pVR.amp.rab.gp vector and in the vector used in the previous report has a CpG motif that is methylated by the SssI enzyme.¹² Methylation of this site strongly reduces the activity of the CMV promoter.^12,13 To determine if the inhibition of the immune response to the methylated pVR.amp.rab.gp DNA vaccine was merely a reflection of the reduced antigen expression or was indeed linked to methylation of immunostimulatory sequences, we next immunized mice with a methylated or unmethylated pVR.amp.rab.gp vector or an unmethylated empty control vector mixed together with oligonucleotides containing immunostimulatory sequences.¹⁴ Vectors were mixed with an oligonucleotide lacking such sequences as a negative control. Mice were injected with a mixture of 50 g of vector DNA and 1 g of either of the oligonucleotides. Mice were bled 2 and 4 weeks later and serum antibody titers were determined by an ELISA. As shown in Figure 2, unmethylated DNA mixed with the control oligonucleotide induced an antibody response that was readily detectable after 2 weeks and increased further by 4 weeks after vaccination. Addition of the oligonucleotide containing the immunostimulatory motif had no significant effect on the antibody response. Methylated DNA mixed with the control oligonucleotide failed to induce a significant response after 2 or 4 weeks. The absorbance values measured by ELISA were overlapping with those obtained with the control plasmid. Upon addition of the CpG motif expressing oligonucleotide the methylated DNA vaccine induced a B cell response to rabies virus which was detectable after 2 weeks and further increased after 4 weeks. The response was well below that achieved with unmethylated DNA which presumably reflects the decrease in the levels of expressed antigen following methylation of the promoter.

To ensure further that the lack of a B cell response to methylated vector DNA is caused by inhibition of the immunostimulatory activity of CpG sequences upon methylation rather than by the reduction of the activity of the methylation-sensitive CMV promoter, we conducted an additional experiment using the pSG5rab.gp vector which expresses the rabies virus glycoprotein under the control of the early SV40 promoter, which is one of the few regulatory elements that is unaffected by CpG methylation.¹⁵ In a pre-experiment we compared in a transient transfection assay the activity of the SV40 promoter in a SssI methylated or unmethylated pSG5 vector expressing luciferase as the reporter protein. Methylated and unmethylated vector gave comparable luciferase expression levels upon transfection of COS-1 cells confirming that CpG methylation does affect the activity of the SV40 promoter (Table 1). Next, groups of C3H/He mice were immunized with 50 g of methylated or unmethylated pSG5rab.gp DNA using an empty pSG5 vector as the negative control. Mice were bled 2, 4 and 8 weeks later. As shown in Figure 3, the unmethylated pSG5rab.gp vector induced a strong antibody response to rabies virus that could be detected as early as 2 weeks after immunization. The response increased by 4 weeks and further by 8 weeks after vaccination. The empty pSG5 vector caused a slight increase in activity above that seen with a normal mouse serum (not shown) as tested by ELISA which presumably reflects polyclonal activation of B cells. In contrast, the methylated pSG5rab.gp vector was immunologically inert; the ELISA absorbance curves were well below those obtained with the control construct at all three time-points measured.

Taken together, these data clearly show that unmethylated CpG sequences, regardless of antigen expression levels, are required to elicit an antibody response to an antigen encoded by vector DNA.

The effect of additional CpG sequences on the B cell response to an antigen encoded by a plasmid vector

A number of different DNA sequences have immunostimulatory activity. The palindromal AACGTT sequence is one of the most potent immunomodulatory motifs in mice.¹⁶ The original pVR1012.2 vector, which contains the kanamycin resistance gene, lacks such a motif while the ampicillin resistance gene has two AACGTT sequences.¹⁶ The cDNA encoding the rabies virus glycoprotein does not have such a motif. Both versions of the pVR1012 vector have additional immunostimulatory CpG motifs which are listed in comparison with those of the pSG5 vector in Table 2. To investigate if the addition of two AACGTT sequences gained by replacing the kanamycin resistance gene with the ampicillin resistance gene would affect the ability of the DNA vaccine to induce a B cell response to the rabies virus glycoprotein, we next compared the antibody response with the pVR.kana.rab.gp and the pVR.amp.rab.gp vector using a high dose of 50 g and a moderate dose of 10 g of DNA per mouse. In this set of experiments, groups of C3H/He mice were inoculated i.m. with pVR.amp.rab.gp, VR.kana.rab.gp or empty pVR.amp plasmids. Antibody titers to rabies virus were tested at three different time-points after immunization (ie 2, 4 and 8 weeks). As shown in Figure 4a, antibody titers became positive 4 weeks after immunization in the two groups that received the antigen-encoding vectors. At 50 g of vector the pVR.kana.rab.gp construct gave slightly superior antibody titers compared with the pVR.amp.rab.gp vector, nevertheless, the difference was marginal and statistically insignificant. Sera of mice immunized with 10 g of the two different vector constructs showed a more pronounced difference; the pVR.amp.rab.gp vector resulted in slightly superior antibody titers compared with the pVR.kana.rab.gp vector in sera harvested 4 weeks after immunization. This difference was more pronounced after 8 weeks when mice immunized with the pVR.amp.rab.gp vector had still readily detectable antibody titers while sera of mice vaccinated with pVR.kana. rab.gp vector were only borderline positive. At both time-points the difference between antibody titers to the two constructs given at 10 g was statistically highly significant (Student's t test, P < 0.05). These data were confirmed by a neutralization assay using sera harvested 8 weeks after immunization (Table 3); sera of mice immunized with 50 g of either of the rabies virus glycoprotein-expressing vectors had comparable VNA titers of 20 and 21 IU, respectively, while sera of mice immunized with 10 g of DNA had a titer of 6.6 IU upon vaccination with the pVR.amp.rab.gp construct compared with 1.3 IU upon vaccination with the pVR.kana.rab.gp construct. To confirm that at an optimal dose of a DNA vaccine additional CpG sequences are not able to further enhance the B cell response to the transgene product, we repeated the experiment in a genetically different mouse strain, ie in C57Bl/6 mice. Groups of five C57Bl/6 were immunized with 50 g of pVR.amp.rab.gp, pVR.kana.rab.gp or as a control empty pVR.amp vector. Mice were bled 2, 4 and 8 weeks later and antibody titers to rabies virus were determined. As shown in Figure 4b, mice immunized with either of the DNA vaccines mounted a rapid B cell response to rabies virus that was readily detectable after 2 weeks. At this time-point the response to the two vectors was indistinguishable. At 4 weeks after immunization the antibody response in C57Bl/6 mice was slightly higher upon immunization with the pVR.kana.rab.gp vector while at 8 weeks the pVR.amp.rab.gp vector induced marginally higher titers. Overall, the differences observed in the B cell response of C57Bl/6 mice to the two different vector constructs was statistically insignificant.

To characterize further the immune response of mice to plasmids encoding for the same antigen but with a different content of CpG sequence, the isotype profile of rabies virus-specific antibodies harvested from vector immunized C3H/He mice was determined. The response was predominated by antibodies of the IgG2a isotype for both groups of mice immunized with the two different vectors given at either dose indicating that the difference in the number of AACGTT sequences did not influence the type of the immune response (Figure 5).

Discussion

In this report we describe two experimental approaches that both reduce the antibody response to a transgene product transferred into mice by plasmid vectors. Long-term gene therapy has been hampered by immune responses elicited to the gene therapy vehicle as well as to the transgene product. Such immune responses, mainly through the action of cytolytic CD8⁺ T cells, cause rapid elimination of transgene expressing cells.¹⁷ In the case of viral gene therapy vehicles, the concomitantly induced neutralizing antibody response against the carrier reduces efficacious uptake of the gene therapy vehicle upon its reapplication.⁷ Even gene therapy vehicles that do not induce specific immune responses against their own antigens, such as expression vectors, were shown to stimulate B and T cell responses to the transgene product thus inviting their use as vaccines, called DNA vaccines, but lessening their suitability for long-term replacement of missing or faulty genes.^18,19

The immune system is assumed to require three signals for activation of B and T cells: signal 1 is the antigen; signal 2 is a co-stimulatory molecule such as B7.1 or B7.2 expressed on professional antigen presenting cells; signal 3, also called signal 0 or the danger signal, is assumed to be an activation signal for professional antigen presenting cells such as dendritic cells.⁴ Dendritic cells in their resting stage engulf antigen very efficiently but they only express low levels of MHC class II and co-stimulatory molecules and are thus inefficient at presenting antigenic peptides to naive T cells. They are, furthermore, stationary and require additional activation to gain the ability to migrate to lymph nodes where stimulation of T and B cells takes place. In the case of DNA vaccines, the antigen encoded by the vector provides signal 1. Vector DNA inoculated into mice has been shown to transfect directly dendritic cells in situ; furthermore, antigen presented by in vitro transfected muscle cells, the main in vivo target of DNA transfection upon intramuscular inoculation of DNA vaccines, has been shown to be reprocessed and presented by bone marrow-derived antigen presenting cells of the host.^20,21,22 Antigen encoded by DNA vaccines is thus expressed in the context of co-stimulatory molecules providing signal 2 to the immune system. Unmethylated CpG sequences present in the bacterial part of plasmid vectors presumably provide signal 0 to the immune system by causing activation of innate immune responses including polyclonal activation of B cells and induction of cytokine release.

One previous report demonstrated that mice injected with a DNA vaccine methylated by the SssI methyl transferase, which targets CpG sequences, failed to mount an immune response to the vector-encoded antigen while the same vector in its unmethylated form induced a potent response. In this study, the transgene was under the control of the CMV promoter, which has a CpG motif that is sensitive to methylation by the SssI enzyme.¹³ Methylation of this site causes a strong decrease in the activity of the CMV promoter thus reducing the level of antigen expression which also affects the magnitude of the immune response. In an unrelated study the local inflammatory response following intramuscular inoculation of plasmid vectors was investigated.²³ This study showed that unmethylated vectors elicited a strong cellular infiltrate at the site of injection while SssI methylated DNA failed to induce such a reaction; again showing that the vector's CpG sequences induce a local inflammatory reaction. Furthermore, this study demonstrated that cells transfected with unmethylated plasmid vectors were cleared more rapidly compared with those transfected with methylated DNA suggesting that methylation of CpG sequences had compromised the ability of the immune system to mount an efficient response against transgene expressing cells.

In this report, we have extended these studies by comparing the antibody response to SssI methylated and unmethylated plasmid vectors using either constructs with the CpG methylation-sensitive CMV promoter or the CpG methylation-insensitive SV40 early promoter. With both constructs, SssI methylation abrogated the immunogenicity of the DNA vaccine demonstrating that unmethylated CpG sequences were indeed crucial to provide an activation signal to the immune system. Using the methylation-sensitive CMV promoter, addition of oligonucleotides carrying immunostimulatory CpG sequences partially rescued the response. Nevertheless, the response to methylated DNA in the presence of CpG expressing oligonucleotides remained well below that of the unmethylated construct, which most likely reflects the reduced expression of antigen by the CpG methylated CMV promoter.

Gene therapy requires a sufficient level of transgene expression to achieve efficacious levels of the therapeutic protein. The SV40 promoter that fully retains its activity upon SssI methylation is a weak promoter when compared with the methylation-sensitive CMV promoter that achieves in most cell types 40 times higher levels of protein production. Many gene therapy vehicles are based on potent promoters which are in general methylation sensitive. We therefore tested in a second set of experiments if other modifications of the vector backbone aimed at diminishing the content of immunostimulatory CpG sequences might reduce the B cell response to the vector-encoded antigen. For these studies we used the pVR1012.2 vector backbone that in its original form contains the kanamycin resistance gene. This construct completely lacks an AACGTT sequence which is one of the most potent immunostimulatory CpG motifs in mice.¹⁶ By replacing the kanamycin resistance gene with the ampicillin resistance gene we generated a construct that contains two such sequences. The B cell response to an optimal dose of 50 g of these two vectors was indistinguishable in the two mouse strains tested showing that with this fairly high amount of vector DNA the addition of two AACGTT motifs failed to affect the immune response. Upon reducing the dose to 10 g, the antibody response to the ampicillin resistance gene containing vector was slightly reduced compared with the higher dose of the same construct while the antibody response to the kanamycin containing plasmid given at this low dose was barely above control levels confirming that, at least under limiting conditions, this particular motif is crucial for providing an activation signal to the immune system.⁴ The pVR1012.2 vector has several other motifs (three GACGGT, two GTCGTT and five GACGCT) which have been described as providing activation signals to the immune system.²⁴ Nevertheless, previous studies have shown them to be at least in vitro less potent compared with the AACGTT sequence which was confirmed by our results.²⁴ These data suggest that targeted modulation of the vector backbone can profoundly affect the antibody response to the encoded antigen; while addition of immunomodulatory sequences increases the B cell response and might thus serve to optimize DNA vaccines, deletions of such motifs reduce the B cell response which would be useful for gene therapy vehicles.

Materials and methods

Mice

Female C3H/He and C57Bl/6 mice were purchased from Jackson Laboratory (Bar Harbor, ME, USA). Mice were housed in a temperature controlled, light-cycled room in the animal facility of The Wistar Institute. They were used at 8-12 weeks of age.

Cells

Baby hamster kidney (BHK)-21 cells and COS-1 cells were grown in Dulbecco's minimal essential medium (DMEM) supplemented with 10% fetal bovine serum (FBS).

Virus

Rabies virus strain Evelyn Rokitniki Abelseth (ERA) was grown on BHK-21 cells, purified and inactivated with betapropionolactone (BPL) as described.²⁴ The Challenge Virus standard (CVS)-11 strain of rabies virus was propagated and titrated on BHK-21 cells.²⁵

Construction of plasmids

The pVR1012 plasmid containing the kanamycin resistance gene was obtained from Vical (San Diego, CA, USA). The full-length cDNA encoding the rabies virus glycoprotein of the ERA strain, excised from the pSG5rab.gp vector, was cloned into the unique BglII restriction site downstream of the CMV promoter thus generating pVR.kana.rab.gp. The ampicillin gene was obtained from the pUC19 plasmid upon digestion with AflIII and SspI. The kanamycin resistance gene was excised from the pVR.1012.2 vector by digestion with the restriction enzymes DraIII and XmnI. The ampicillin gene was ligated into the pVR1012 plasmid by blunt-end ligation to generate pVR.amp. The full-length cDNA of the rabies virus glycoprotein of the ERA strain was cloned into the BglII site of the pVR.amp vector (pVR.amp.rab.gp). The new constructs were analyzed by restriction enzyme mapping to ensure correct orientation of the inserts. Furthermore, upon transfection of L929 mouse fibroblasts, they were tested for expression of the rabies virus glycoprotein as described previously.²⁶ The pSG5rab.gp vector has been described in detail earlier.²⁷ A pSG5 vector expressing luciferase (pSG5luc) was constructed by excising the full-length luciferase cDNA from the pG2/Basic vector using BglII and BamHI restriction enzymes. The fragment was ligated into the BglII site of the pSG5 vector's multicloning site downstream of the SV40 promoter. The orientation of the insert was determined by restriction enzyme mapping using BamHI and BglII.

Purification of plasmids

Plasmids were amplified in E. coli strain DH5 in LB broth supplemented with 50 g/ml ampicillin or 25 g/ml kanamycin. Large scale purification of plasmids was carried out using the Qiagen Plasmid Mega Kit (Santa Clarita, CA, USA) according to the manufacturer's protocol. The DNA was quantified by spectrophotometry at an optical density of 260 nm or by agarose gel electrophoresis against a known marker.

Methylation of plasmids

The vectors pVR.amp.rab.gp, pSG5rab.gp and pSG5.luc were methylated by SssI methylase overnight according to the manufacturer's specifications. Vectors were subsequently purified by phenol/chloroform extraction followed by precipitation with isopropanol then ethanol. To ensure complete methylation, 1 g of vector was digested with 1 U of HpaII using unmethylated vector DNA for comparison. Lack of activity of this restriction enzyme that is sensitive to methylation of CpG sequences was confirmed by agarose gel electrophoresis which showed multiple bands upon digestion of unmethylated vector and a single band upon digestion of fully methylated DNA.

Luciferase activity

COS-1 cells were plated at 2 ´ 10⁵ cells in six-well Costar plate wells. Cells were transfected the next day with 1 g of vector DNA using lipofectin according to the manufacturer's protocol. Cells were lysed 48 h later with 300 l of buffer containing 5% Triton X; 20 l of the cell lysate was added to 100 l of luciferase assay reagent (20 mm tricine, 1.7 mm MgCO₃, 2.67 nm MgSO₄, 0.1 mm EDTA, 33.3 mm DTT, 279 M Coenzyme A, 530 m ATP, 470 m luciferin) immediately before analysis in a Monolight 2010 Luminometer (PharMingen, San Diego, CA, USA). Sham transfected cells were used as a negative control.

Oligonucleotide

A phosphorothioate oligonucleotide containing an optimized CpG motif (TCCATGACGTTCCTGACGTT) as well as a control phosphorothioate oligonucleotide (TCCAGGACTTCTCTCAGGTT) were most generously provided by Dr A Krieg (University of Iowa, Iowa City, IA, USA).

Immunization of mice

Groups of five mice were injected intramuscularly into both quadriceps with a total of 10 or 50 g of plasmid diluted in 100 l of saline or water. In one set of experiments 1 g of a phosphodiester oligonucleotide was mixed into this solution. Mice were bled 2, 4 and 8 weeks after immunization by retro-orbital puncture using Netelson Blood Heparinized Collecting Tubes (Fisher Scientific, Pittsburgh, PA, USA). Capillaries were spun down at 2500 r.p.m. for 25 min at room temperature, sera were pooled, aliquoted, and stored at -20°C till further use. Sera were heat-inactivated for 30 min at 56°C before testing for rabies virus neutralizing activity.

Enzyme-linked immunoadsorbent assay (ELISA)

ELISAs were conducted in 96-well U-bottom microtiter plates coated with 100 l of 0.2 g per well of BPL-inactivated rabies virus as described.²⁸

Mapping of antibody isotypes

The isotypes of antibodies to rabies virus were determined by ELISA on rabies virus coated plates using the Calbiochem (San Diego, CA, USA) hybridoma subisotyping kit, according to the manufacturer's suggestions, with some minor modifications as described.²⁹ Optical density (OD) was determined at 450 nm.

Neutralization assay

Virus neutralizing antibodies (VNA) titers were determined on BHK-21 cells using infectious CVS-11 virus at 1 plaque-forming unit per cell against a reference serum obtained from the World Health Organization.³⁰ Data are expressed as international units.

Statistical analysis

Data were analyzed by Student's t test. Antibodies that had P values <0.05 compared with the reference serum were considered to show a significant difference.

Acknowledgements

We wish to thank Dr L Otvos for critical review of this manuscript. This work was supported in part by the Cancer Core Support Grant from the National Cancer Institute, No. CA10815, The Wistar Institute.

1 Yakaukas MA et al. Long term anti-nucleoprotein cellular and humoral immunity is induced by intramuscular injection of plasmid DNA containing NP gene. DNA Cell Biol 1993; 12: 771-777, MEDLINE

2 Krieg AM et al. CpG motifs in bacterial DNA trigger direct B-cell activation. Nature 1995; 374: 546-551, MEDLINE

3 Yamamoto S et al. Unique palindromic sequences in synthetic oligonucleotides are required to induce IFN and augment IFN-mediated natural killer cell activity. J Immunol 1992; 148: 4072-4076, MEDLINE

4 Matzinger P. Tolerance, danger and the extended family. Annu Rev Immunol 1994; 12: 991-1045, MEDLINE

5 Yang Y, Li Q, Ertl HCJ, Wilson JM. Lung-directed gene therapy with recombinant adenoviruses is limited by destructive class 1 restricted cytotoxic T cell responses to virus-infected cells. J Virol 1995; 69: 2004-2015, MEDLINE

6 Yang Y, Xiang ZQ, Ertl HCJ, Wilson JM. Upregulation of class I MHC antigens by interferon- is necessary for the T cell-mediated elimination of recombinant adenovirus infected hepatocytes in vivo. Proc Natl Acad Sci USA 1995; 92: 7257-7261, MEDLINE

7 Yang Y et al. Immune responses to viral antigens versus transgene product in the elimination of recombinant adenovirus infected hepatocyte in vivo. Gene Ther 1995; 3: 137-144,

8 Biggs R. Jaundice and antibodies against factor VIII and IX in patients treated for hemophilia or Christmas disease in the United Kingdom. Br J Haematol 1974; 26: 313-329, MEDLINE

9 Ehrenfort S et al. Incidence of development of factor VIII and factor IX inhibitors in hemophiliacs. Lancet 1992; 339: 594-598, MEDLINE

10 Kay MA et al. In vivo hepatic gene therapy: complete albeit transient correction of factor IC+X deficiency in hemophilia B dogs. Proc Natl Acad Sci USA 1994; 91: 2353-2357, MEDLINE

11 Monahan P et al. Direct intramuscular injection with recombinant AAV results in sustained expression in a dog model of hemophilia. Gene Therapy 1998; 5: 40-49, MEDLINE

12 Klinman DM, Yamshchikov G, Ishigatsubo Y. Contribution of CpG motifs to the immunogenicity of DNA vaccines. J Immunol 1997; 158: 3635-3639, MEDLINE

13 Hug M et al. Transcriptional repression by methylation: cooperativity between a CpG cluster in the promoter and remote CpG-rich regions. FEBS Lett 1996; 379: 251-254, MEDLINE

14 Klinman DM et al. CpG motifs present in bacterial DNA rapidly induce lymphocytes to secrete interleukin 6, interleukin 12 and interferon . Proc Natl Acad Sci USA 1996; 93: 2879-2883, Article MEDLINE

15 Fradin A, Manley JL, Prives CL. Methylation of simian virus 40 HpaII site affects late, but not early, viral gene expression. Proc Natl Acad Sci USA 1982; 79: 5142-5146, MEDLINE

16 Sato Y et al. Immunostimulatory DNA sequences necessary for effective intradermal gene immunization. Science 1996; 273: 352-354, MEDLINE

17 Jooss K, Ertl HCJ, Wilson JM. CTL target proteins and their MHC class I restriction in response to adenoviral vectors delivered to mouse liver. J Virol 1998; 72: 2945-2954, MEDLINE

18 Tang D, DeVit M, Johnston SA. Genetic immunization is a simple method for eliciting an immune response. Nature (London) 1992; 356: 152-154, MEDLINE

19 Ulmer B et al. Heterologous protection against influenza by injection of DNA encoding a viral protein. Science 1993; 259: 1745-1749, MEDLINE

20 Condon C. DNA-based immunization by in vivo transfection of dendritic cells. Nature Med 1996; 2: 1122-1128, MEDLINE

21 Porgador A. Predominant role for directly transfected dendritic cells in antigen presentation to CD8+ T cells after gene gun immunization. J Exp Med 1998; 188: 1075-1082, MEDLINE

22 Fu TM et al. Priming of cytotoxic T lymphocytes by DNA vaccines: requirement for professional antigen presenting cells and evidence for antigen transfer from myocytes. Mol Med 1997; 3: 362-371, MEDLINE

23 McMahon JM et al. Inflammatory responses following direct injection of plasmid DNA into skeletal muscle. Gene Therapy 1998; 5: 1283-1290, MEDLINE

24 Ballas ZK, Rasmussen WL, Kreig AM. Induction of NK activity in murine and human cells by CpG motifs in oligonucleotides and bacterial DNA. J Immunol 1996; 157: 1840-1845, MEDLINE

25 Wiktor TJ, Dietzschold B, Leamson N, Koprowski H. Induction and biological properties of defective interfering particles of rabies virus. J Virol 1977; 21: 626-633, MEDLINE

26 Dietzschold BD et al. Mechanism of rabies virus neutralization by glycoprotein specific monoclonal antibodies. Virology 1987; 161: 29-36, MEDLINE

27 Xiang ZQ et al. Vaccination with a plasmid vector carrying the rabies virus glycoprotein gene induces protective immunity against rabies virus. Virology 1994; 199: 132-140, MEDLINE

28 Burger SR et al. Stable expression of rabies virus glycoprotein in Chinese hamster ovary cells. J Gen Virol 1991; 72: 359-367, MEDLINE

29 Xiang ZQ, Ertl HCJ. Transfer of maternal antibodies results in inhibition of specific immune reponses in the offspring. Virus Res 1992; 24: 297-314, MEDLINE

30 Wang Y, Xiang ZQ, Pasquini S, Ertl HCJ. Immune response to neonatal genetic immunization. Virology 1997; 228: 278-284, Article MEDLINE

31 Wiktor TJ. Tissue culture methods. In: Kaplan M, Koprowski H (eds). Laboratory Techniques in Rabies. 2nd edn. Vol 23: World Health Organization Monograph: Geneva, 1973, pp 101-120.

Figure 1 The antibody response to a methlyated expression vector encoding the antigen under the control of the CMV promoter. Groups of five C3H/He mice were inoculated intramuscularly with 50 g of empty pVR vector (X), 50 g of unmethylated pVRamp.rab.gp () or 50 g of methylated pVRamp.rab.gp () vector. Sera were harvested 2, 4 and 8 weeks later and tested for antibodies to rabies virus by an ELISA.

Figure 2 The effect of oligonucleotides with CpG motifs on the B cell response to a methlyated expression vector. Groups of five C3H/He mice were inoculated intramuscularly with 50 g of vector DNA mixed with 1 g of either a CpG sequence containing oligonucleotide (Oligo) or a control oligonucleotide: () unmethylated pVRamp.rab.gp + CpG oligo; () unmethylated pVR.amp.rab.gp + control oligo; () methylated pVR.amp.rab.gp + CpG Oligo; () methylated pVR.amp.rab.gp + control oligo; (+) empty pVR.amp + CpG oligo; () empty pVRamp + control oligo. Mice were bled 2 (a) and 4 (b) weeks later and serum antibody titers were measured by an ELISA.

Figure 3 The antibody response to a methylated expression vector encoding the antigen under the control of the SV40 early promoter. Groups of five C3H/He mice were immunized intramuscularly with 50 g of empty pSG5 (X), 50 g of unmethylated pSG5rab.gp () or 50 g of SssI methylated pSG5rab.gp vector (). Mice were bled 2, 4 and 8 weeks later and antibody titers to rabies virus were measured by an ELISA.

Figure 4 The antibody response to expression vectors with varied contents of the AACGTT motif. Groups of C3H/He (a) or C57Bl/6 (b) were immunized with 10 g (C3H/He mice only) or 50 g of empty pVR.amp (X), pVR.amp.rab.gp () or pVR.kana.rab.gp (). Sera were harvested 2, 4 and 8 weeks later and tested for antibodies to rabies virus in an ELISA.

Figure 5 Antibody isotypes elicited by expression vectors with varied contents of the AACGTT motif. Sera from C3H/He mice immunized 8 weeks previously with 10 g or 4 weeks previously with 50 g of empty pVR.amp (), pVR.amp.rab.gp () or pVR.kana.rab.gp () were tested for the isotype distribution of antibodies to rabies virus by an ELISA.

Table 1 Levels of reporter protein expression in cells transiently transfected with methylated and unmethylated vectors

Table 2 Analysis of the CpG contents of the different DNA constructs

Table 3 Induction of neutralizing antibodies by methylated and unmethylated expression vectors