Efficient induction of T-cell responses to carcinoembryonic antigen by a heterologous prime-boost regimen using DNA and adenovirus vectors carrying a codon usage optimized cDNA

Abstract

The immunogenic properties of plasmid DNA and recombinant adenovirus (Ad) encoding the carcinoembryonic antigen (CEA) were examined in mice by measuring both the amplitude and type of immune response, and the immunogenicity of codon usage optimized cDNA encoding CEA (CEAopt) was assessed both in C57Bl/6 and CEA transgenic mice. Vectors were injected into quadriceps muscle either alone or in combination, and plasmid DNA was electroporated to enhance gene expression efficiency and immunogenicity. Injection of plasmid pVIJ/CEA followed by Ad-CEA boost elicited the highest amplitude of both CD4⁺ and CD8⁺ T-cell response to the target antigen, measured by both IFNγ-ELIspot assay and intracellular staining. Vectors carrying cDNA of CEAopt expressed a greater amount of the CEA protein than their wild-type counterparts, and this enhanced expression was associated with greater immunogenicity. Both CD4⁺ and CD8⁺ T-cell epitopes were mapped in the C-terminal portion of the protein. In CEA transgenic mice, only immunization based on repeated injections of pVIJ/CEAopt followed by Ad-CEAopt was able to elicit a CEA-specific CD8⁺ T-cell response, whereas the wild-type vectors did not break tolerance to this target antigen. MC38-CEA tumor cells injected s.c. in CEA transgenic mice vaccinated with CEAopt vectors exhibited delayed growth kinetics. These studies demonstrate that this type of genetic vaccine is highly immunogenic and can break tolerance to CEA tumor antigen in CEA transgenic mice. © 2005 Wiley-Liss, Inc.

Despite improvements in prevention, early detection and treatment, the possibility of curing many cancer patients remains elusive. Thus, cancer continues to be a largely unmet medical need for which more efficient therapeutic strategies must be developed.1

Particular attention has been given to active specific immunotherapy of cancer, whereby patients are immunized against antigens presented by tumor cells. In fact, experimental and clinical evidence have demonstrated the critical role played by the cellular and humoral responses in controlling tumor growth and metastasis.2 Many of these therapies are specifically targeted to tumor-associated antigens among which carcinoembryonic antigen (CEA) is a frequent example due to its ectopic and deregulated expression in a large percentage of adenocarcinomas.3 Human CEA is the prototypic member of the human CEA family, a group of highly glycosylated homotypic/heterotypic cell surface intracellular adhesion molecules, and part of the immunoglobulin gene superfamily. CEA is widely used as human tumor marker, is expressed mostly in the gastrointestinal tract and is overexpressed in many human cancers, including epithelial tumors originating from the gastrointestinal tract, lung, thyroid, breast, prostate, cervix and ovaries.4 CEA has been the focus of extensive preclinical and clinical investigation aimed at developing a CEA-specific vaccine with a therapeutic impact on tumor progression.5

In this context, genetic vaccine strategies are probably the most promising to achieving effective antitumor responses. In fact, gene-based vaccines have distinct potential advances over traditional protein vaccination due to the induction of antigen-specific T helper (Th1) and cytotoxic T-lymphocytes (CTL) responses, the prolonged antigen expression and the long-lived effector activity.6

Plasmid DNA vaccine has emerged as a safe and promising method for genetic vaccination. The intrinsic advantages of DNA vaccines are their ease of production and storage; their ability to mimic the effect of live attenuated vaccines in their capacity to induce MHC class I restricted CD8⁺ T-cell responses; and their ability to elicit antibody responses.7 Indeed, DNA vaccine approaches have been applied to various pathogens,8 but the immune responses induced to date by DNA vaccines have been relatively weak compared to conventional vaccines. One reason for this limited efficacy is that in larger animals the level of protein expression might be insufficient to elicit strong immune responses. In vivo electroporation (EP) of plasmid DNA has resulted in increased DNA uptake, leading to enhanced protein expression in treated muscle9 and in a concomitant increase in the immune responses to the target antigen in a variety of species.10, 11, 12, 13 This enhancement of immunogenicity mediated by EP could, however, also derive from undefined adjuvant effects connected with transfection of other cells, muscle damage and/or release of danger signals.14, 15, 16

Viruses have highly evolved structures that enable them to bind cells and deliver genes into the cells they infect. It is thus reasonable to consider using live virus vaccines to induce CTL for treating a disease in which cellular immunity is an absolute requirement. To this end, a number of different viruses such as vaccinia, rabies virus, canarypox virus, Semiliki Forest virus and adenovirus (Ad) have been used to develop recombinant viral vaccines.17 Among these, the replication-defective recombinant Ad has been shown to be safe and to induce strong humoral and cellular antigen-specific immune responses.18 In addition, combinations of heterologous modalities of immunization have indicated that enhanced immune responses to the target antigen can be elicited by vaccinating with different vectors encoding the same immunogen. A similar immunization regimen has been tested for plasmid DNA coupled with Ad or pox vectors in mice and non-human primates.19, 20, 21, 22, 23 The rationale behind this strategy is that by using different vectors as boosters, it should be possible to bypass the immune response elicited against the primer and also strengthen the immune response against the target antigen.

The potency of current gene delivery methods that include plasmid DNA and viral vectors can also be improved through increasing the expression efficiency of the encoded antigens. Studies performed with viral antigens have demonstrated that elevated AT content in the target gene exerts a negative influence on its expression efficiency.24 Higher percentages of AU in human mRNA have been shown to result in instability, increased turnover and low expression level.25 These findings have prompted modification of the target gene coding sequence by the reduction of the AT content on the assumption that these modifications could improve mRNA stability and increase expression. These changes have been justified by the observation that for highly expressed genes, G or C is generally preferred over A or T. In fact, in a variety of experimental systems, optimizing the codon usage of the target gene has been shown to enhance expression and increase immunogenicity.26, 27, 28, 29, 30, 31, 32

In our study, the immunogenic properties of a genetic vaccine based on plasmid DNA-EP and Ad vector have been examined by monitoring the amplitude and type of immune responses to CEA. In addition, the immunogenicity of a codon usage optimized cDNA encoding CEA was examined both in C57Bl/6 and in CEA transgenic mice. Our results demonstrate that the combined use of plasmid DNA-EP and Ad vector carrying the codon usage optimized cDNA of CEA are able to elicit significant immune responses to CEA and to break tolerance to this target antigen.

Material and methods

Human CEA optimized codon sequence

The entire CEAopt coding sequence was synthesized and assembled by BIONEXUS (Oakland, CA). The CEAopt cDNA carrying an optimized Kozak sequence at its 5′ end was by PCR and inserted into the pCR-Blunt vector from Invitrogen (Frederick, MD) yielding pCR-CEAopt. The integrity of the CEAopt cDNA was determined by sequencing.

Plasmid constructs and adenovirus vectors

pVIJ/CEAopt

CEAopt DNA digested from pCR-CEAopt was cloned into the EcoRI site of plasmid pVIJnsB.33

pVIJ/CEA

CEA cDNA from plasmid pCI/CEA34 was cloned into the EcoRI site of plasmid pVIJnsA.33 The CEA cDNAs were placed under the control of the human cytomegalovirus (CMV)/intron A promoter plus the bovine growth hormone (BGH) termination signal.

Ad-CEAopt

Plasmid pCR-CEAopt was digested with EcoRI and the 2156 bp insert was purified and cloned into the EcoRI of the polyMRK-Ad5 shuttle plasmid.

Ad-CEA

The shuttle plasmid pMRK-CEA for generation of Ad5 vector was obtained by digesting plasmid pDelta1sp1B/CEA with SspI and EcoRV. The 9.52 kb fragment was then ligated with a 1272 bp BglII-BamHI fragment from plasmid polyMRK. A PacI/StuI fragment from pMRK-CEA and pMRK-CEAopt containing the CMV/BGH expression cassette for CEA and E1 flanking Ad5 regions was recombined to ClaI linearized plasmid pAd5 in BJ5183 E. coli cells. The resulting plasmids were pAd5-CEA and pAd5-CEAopt, respectively. Both plasmids were cut with PacI to release the Ad ITRs and transfected in PerC-6 cells. Ad5 vectors amplification was carried out by serial passages. Ad-CEA and Ad-CEAopt were purified through standard CsCl gradient purification and extensively dialyzed against A105 buffer (5 mM Tris-Cl pH 8.0, 1 mM MgCl₂, 75 mM NaCl, 5% sucrose, 0.005 Tween 20).

CEA expression and detection

Expression of CEA by the plasmid and Ad vectors was monitored by ELISA. Plasmids were transfected in HeLa cells or PerC.6 cells with Lipofectamine 2000 (Life Technologies, Gaithersburg, MD). PerC.6 cells were infected by Ad in serum-free medium for 30 min at 37°C, and then fresh medium was added. After 48 hr incubation, whole cell lysates and culture supernatant were harvested. The secreted CEA was detected in the cell supernatants and in peripheral blood of injected mice (3 days postinjection) using the Direct Elisa CEA Kit (DBC-Diagnostics Biochem Canada, Ontario, Canada).

Peptides

Lyophilized CEA peptides were purchased from Bio-Synthesis (Lewisville, TX) and resuspended in DMSO at 40 mg/ml. Pools of peptides 15 aa long overlapping by 11 residues were assembled as previously described.35 Final concentrations were the following: pool A, 1.2 mg/ml; pool B, 0.89 mg/ml; pool C, 0.89 mg/ml; pool D, 0.8 mg/ml. Peptides were stored at −80°C.

Mice immunization

All animal studies described in our study were approved by the IRBM institutional animal care and use committee. Female C57Bl/6 mice (H-2^b) were purchased from Charles River (Lecco, Italy). CEA.tg mice (H-2^b) were provided by J. Primus (Vanderbilt University)36 and kept in standard conditions. Fifty micrograms of plasmid DNA were electroporated in a 50 μl volume in mice quadriceps as previously described.37 Ad injections were carried out in mice quadriceps in 50 μl volume. Humoral and cell-mediated immune response were analyzed at the indicated time. Two weeks after the last injection, mice were challenged s.c. with the cell line MC38-CEA,36 which expresses the tumor antigen CEA, at the dose of 5 × 10⁵ cells/100 μl of PBS. At weekly intervals, mice were examined for tumor growth and tumor volume was measured with a caliper and calculated as previously described.36

Antibody detection and titration

Sera for antibody titration were obtained by retro-orbital bleeding. ELISA plates (Nunc maxisorp) were coated with 100 ng/well of CEA protein (Fitzgerald, Concorde, MA, highly pure CEA), diluted in coating buffer (50 mM NaHCO₃, pH 9.4) and incubated O/N at 4°C. Plates were then blocked with PBS containing 5% BSA for 1 hr at 37°C. Mouse sera were diluted in PBS 5% BSA (dilution 1/50 to evaluate seroconversion rate; dilutions from 1:10 to 1:31,250 to evaluate titer). Pre-immune sera were used as background. Diluted sera were incubated O/N at 4°C. Washes were carried out with PBS 1% BSA, 0.05% Tween 20. Secondary antibody (goat anti-mouse, IgG Peroxidase, Sigma, St. Louis, MO) was diluted 1/2,000 in PBS, 5% BSA and incubated 2–3 hr at RT on a shaker. Plates were developed with 100 μl/well of TMB substrate (Pierce, Rockford, IL) and were read at 450 nm/620 nm. Anti-CEA serum titers were calculated as the reciprocal limiting dilution of serum, producing an absorbance at least 3-fold greater than the absorbance of autologous pre-immune serum at the same dilution.

IFN-γ ELISPOT assay

Ninety-six wells MAIP plates (Millipore, Billerika, MA) were coated with 2.5 μg/ml solution of purified rat anti-mouse IFN-γ (IgG1, clone R4-6A2, BD Pharmingen, Franklin Lakes, NJ).

Splenocytes were obtained by grating mice spleen on a metal grid. Red blood cells were removed by osmotic lysis by adding 1 ml of 0.1× PBS to the cell pellet. Cells were washed with PBS and resuspended in 1 ml R10 medium. Viable cells were counted using Türks staining.

Splenocytes were plated at 2.5 × 10⁵ and 5 × 10⁵ cells/well in duplicate and incubated for 20 hr at 37°C with 1 mgr;g/ml suspension of each peptide. Concanavalin A (ConA) was used as positive internal control for each mouse at 5 μg/ml. After washing with PBS, 0.05% Tween 20, plates were incubated O/N at 4°C with 50 μl/well of biotin-conjugated rat anti-mouse IFNγ (RatIgG1, clone XMG 1.2, Pharmingen) diluted to 1:2,500 in assay buffer. Plates were developed by adding 50 μl/well NBT/B-CIP (Pierce) until spots were clearly visible, then the spots were counted using an automated ELISPOT reader.

Flow cytometry

One to 2 million mouse splenocytes in 1 ml RPMI 10% FCS were incubated with a pool of peptides (5–6 μg/ml final concentration of each peptide) and brefeldin A (1 μg/ml; BD Pharmingen cat. #555028/2300kk) at 37°C and 5% CO₂ for 12–16 hr. Cells were then washed with FACS buffer (PBS 1% FBS, 0.01% NaN₃) and incubated with purified anti-mouse CD16/CD32 Fc block (BD Pharmingen cat. #553142) for 15 min at 4°C. Cells were then washed and stained with surface antibodies: CD4-PE conjugated anti-mouse (BD Pharmingen cat. #553049), PercP CD8 conjugated anti-mouse (BD Pharmingen cat. #553036) and APC conjugated anti-mouse CD3e (BD Pharmingen cat. #553066) for 30 min at room temperature in the dark. After washing, cells were fixed and permeabilized with Cytofix-Cytoperm Solution (BD Pharmingen cat. #555028/2300kk) for 20 min at 4°C in the dark. After washing with PermWash Solution (BD Pharmingen cat. #555028/2300kk), cells were incubated with the IFNγ-FITC antibodies (BD Pharmingen). Cells were fixed with formaldehyde 1% in PBS and analyzed on a FACS-Calibur flow cytometer, using CellQuest software (Becton Dickinson, Franklin Lakes, NJ). At least 15,000 gated events were acquired for the analysis of the immune response.

Statistical analysis

Where indicated, results were analyzed by the log rank, Student's t-test (2-tailed distribution), by Kruskal-Wallis one-way analysis of variance on ranks (ANOVA on ranks) or by Mann-Whitney rank sum test. All these tests were executed using SigmaStat software. A p-value ≤ 0.05 was considered significant.

Results

Construction of human CEA expression vectors

To assess the efficiency of anti-CEA genetic vaccination, the cDNA of human CEA was cloned into plasmid pVIJ.33 The construct was named pVIJ/CEA. In addition, an adenovirus type 5 vector was constructed carrying the CEA cDNA (Ad-CEA).

To verify whether codon usage optimization of the CEA cDNA would enhance immune response, a synthetic gene of human CEA (CEAopt) was designed to incorporate human-preferred (humanized) codons for each amino acid residue. The CEAopt cDNA was modified and maintained 76.8% nucleotide identity to the original clone (Fig. 1). The codon usage optimized cDNA was cloned into the pVIJ vector. The construct was named pVIJ/CEAopt. In addition, an adenovirus type 5 vector was constructed carrying the CEAopt sequence (Ad-CEAopt).

Figure 1. Nucleotide sequence of human CEA cDNA and of the codon usage optimized clone. The deduced amino acid sequence is shown on the top rows. The substituted nucleotides of the synthetic codon optimized cDNA are shown below the CEA cDNA sequence.

Western blot analysis of HeLa cells transfected with plasmids pVIJ/CEA and pVIJ/CEAopt yielded a protein with the predicted molecular mass (180–200 kDa) that was recognized by a CEA-specific antiserum. Similarly, a CEA-specific band was detected in PerC.6 cell lysates that had been infected with Ad-CEA or Ad-CEAopt (data not shown).

Identification and characterization of epitope-containing peptides for direct quantification of CEA-specific T cells

To characterize the immune response against CEA, different immunization modalities were evaluated. In view of recent studies that indicate that high levels of cellular immunity can be induced against viral and bacterial antigens by utilizing plasmid DNA prime-Ad boost modality,38, 39, 40 the same immunization protocol was employed in our study. C57Bl/6 mice were immunized intramuscularly by different regimens: (i) 2 doses of 1 × 10⁷ plaque forming units (pfu) of Ad-CEA (Ad/Ad); (ii) two 50 μg injections of plasmid pVIJ/CEA (DNA/DNA); (iii) plasmid DNA followed by Ad-CEA (DNA/Ad); and (iv) Ad-CEA followed by plasmid DNA (Ad/DNA). As control, a group of mice was immunized with the DNA/Ad modality with a dose of vector pVIJ followed by 1 × 10⁷ pfu of an Ad-empty vector that does not express any transgene (DNA/Ad-empty). Immunizations were 2 weeks apart. The plasmid DNA was routinely electroinjected into mouse skeletal muscle in view of the enhanced transduction and immunogenicity connected with this particular procedure.11, 12

The cellular immunity elicited by the different immunization regimens was measured by Elispot assay 2 weeks after the boost. To compare the immunogenic efficiency of the different vaccination regimens, a pool of 15 mer peptides overlapping by 11 aa and covering aa 497–703 (pool D) were used to stimulate antigen-specific IFNγ secretion from splenocytes. As shown in Figure 2, the most vigorous responses indicated by the higher geometric mean values of the spot forming cells (SFC) were observed in C57Bl/6 mice from the DNA/Ad-injected group (p < 0.05). Thus, this regimen was utilized to further analyze the immune response.

Figure 2. Comparison of different immunization regimens. Groups of C57Bl/6 mice (n = 6) were immunized with different combinations of plasmid pVIJ/CEA (50 μg/dose injected in the quadriceps muscle) and Ad-CEA (1 × 10⁷ pfu/dose). A group of mice was also immunized with pVIJ and Ad-empty vectors in the DNA/Ad modality. Timeline graph is shown on top. The number of IFNγ-secreting T cells in splenocytes in each individual mouse was determined using a pool of peptides covering aa 497–703 (pool D) as described in Material and Methods. Geometric mean values are also shown (empty circles). ANOVA shows that groups immunized with DNA/DNA, DNA/Ad, Ad/Ad are significantly different from the control group with DNA/Ad empty treatment (p < 0.05), whereas Ad/DNA group is not.

To determine whether the immune response was equally distributed across the entire CEA protein, splenocytes from immunized C57Bl/6 mice were stimulated in vitro with each of 4 pools of 15 mer peptides that collectively encompass the entire protein sequence. In addition to pool D, pools A, B and C were used in our study. Pool A covers aa 1–147, pool B aa 137–327 and pool C aa 317–507. As shown in Figure 3, the immune response elicited by the DNA/Ad vaccination regimen in C57BL/6 mice was primarily biased toward the C-terminal region of the protein. Significant SFC values were obtained with peptide pools C and D (geometric mean values: 170 and 244 SFC/10⁶ splenocytes, respectively), whereas pools A and B yielded much lower values (10 and 27 SFC/10⁶ splenocytes, respectively). SFC values observed with peptide pools B, C and D were significantly different from control (p < 0.05). No responses against a pool of unrelated peptides were noted in either groups of mice (data not shown).

Figure 3. Mapping of T-cell responses to selected regions of the CEA protein. Groups of C57Bl/6 mice (n = 9) were immunized with 50 μg of plasmid pVIJ/CEA and boosted 3 weeks later with 1 × 10⁷ pfu of Ad-CEA. Timeline graph is shown on top. The number of IFNγ-secreting T cells in splenocytes of each individual mouse was determined 2 weeks post-boost using a pool of peptides covering the entire protein. This experiment has been performed twice. Geometric mean values are also shown (empty circles). ANOVA shows that groups stimulated with pool B, pool C and pool D of peptides are significantly different from the control group stimulated with DMSO (p < 0.05), whereas the group stimulated with pool A of peptide is not.

To identify the individual peptides present in the peptide pools that elicit the responses, spleens from 4 mice immunized with the DNA/Ad vaccination regimen were analyzed in an IFNγ-ELISPOT assay against each of the individual peptides comprising the pools against which a significant immune response had been observed. Splenocytes from C57Bl/6 mice were tested against peptides 80–173 included in pools C and D (aa 315–702). CEA-specific responses were mapped to 4 pairs of 15 mer peptides that had overlapping sequences (aa 431–435 and 425–439; 529–543 and 533–547; 565–579 and 569–593; 613–627 and 617–631) (Fig. 4).

Figure 4. Identification of immunoresponsive peptides of CEA. Pooled splenocytes from immunized C57Bl/6 mice (n = 4) were assayed for IFNγ secretion against each indicated peptide covering aa 315–702. Timeline graph is shown on top. Elispot assay was carried out as described in Material and Methods. These results are representative of 4 independent experiments.

To define the T-cell specificity of the epitopes contained within the selected peptides, IFNγ intracellular staining assay was carried out on splenocytes from injected mice. The results obtained are shown in Table I. Three peptides (107, 133 and 155) were identified as CD4⁺-specific epitopes and 1 (143) contained a CD8⁺-specific epitope.

Table I. Sequence of Immunoreactive CEA Peptides in C57Bl/6 Mice1

Pool	Peptide	Position	Sequence	CD8^+	CD4^+
1 C57Bl/6 mice (n = 10) were immunized with 50 μg of pVIJ/CEA and 1 × 10^7 pfu of Ad-CEA as described in Material and Methods. IFNγ intracellular staining was carried out with individual peptides using a pool of splenocytes. These results are representative of 3 independent experiments. Data refer to % of IFNγ producing CD8^+ (CD4^+) CD3^+ cells.
C	CEA-107	425–439	TYYRPGVNLSLSCHA	0.02	0.37
D	CEA-133	539–543	NTTYLWWVNGQSLPV	0.01	0.1
D	CEA-143	569–583	YVCGIQNSVSANRSD	0.17	0.01
D	CEA-155	617–631	YVCGIQNSVSANRSD	0.03	0.17
	DMSO			0.01	0.01

To identify the CD8⁺ epitope contained within peptide 143, a set of overlapping 9 mer peptides were used in an intracellular IFNγ staining assay to map the CEA protein sequence recognized by antigen-specific CD8⁺ T cells. As shown in Table II, analysis of these peptides identified as the MHC-I binding motif C₅₇₂GIQNSVSA₅₇₉. The identification of the CD8⁺ T-cell-specific epitope was carried out with 3 immunization regimens (D/D, D/A or A/A). The results obtained were comparable among the different modalities. This observation was further confirmed by analysis of the epitope sequence by the SYFPEITHI algorithm that predicted it to be H2D^b specific.41 Thus, CD8⁺- and CD4⁺-specific epitopes were identified, which can be used to quantify the cellular immune responses in immunized mice.

Table II. Identification of CEA-Specific CD8⁺ T-Cell Epitope1

Peptide	Sequence	CD8^+
1 C57Bl/6 mice (n = 8) were immunized with 2 injections of 1 × 10^8 pfu of Ad-CEAopt as described in Material and Methods. IFNγ intracellular staining of pooled splenocytes was performed with the indicated peptides. These results are representative of 2 independent experiments. Data refer to % of IFNγ producing CD8^+ (CD4^+) CD3^+ cells.
CEA-142	DARAYVCGIQNSVSA	2.19
CEA-143	YVCGIQNSVSANRSD	2.11
143-1	YVCGIQNSV	0.02
143-2	VCGIQNSVS	2.25
143-3	CGIQNSVSA	2.09
143-4	GIQNSVSA	2.4
Pool D		3.79
DMSO		0.01

Codon usage optimized cDNA of CEA significantly increased CEA expression

The use of codon usage optimized cDNAs for genetic vaccination against viral diseases has been shown to elicit a greater immune response due, at least in part, to an increased expression of the target protein.26, 27, 28, 29, 30, 31, 32 To compare the efficiency of expression of the CEAopt to that of CEA, groups of 10 C57Bl/6 mice were injected into the quadriceps with different doses of the Ad-CEAopt vector ranging from 1 x 10⁷ to 1 × 10⁴ pfu. Three days post injection, CEA protein levels in the mice sera were determined and compared to those of control groups that had been injected with the same doses of Ad-CEA. As shown in Figure 5a, a 6-fold increase in the geometric mean values of CEA levels was observed upon injection of 1 × 10⁷ pfu of Ad-CEAopt (48.2 μg/l) relative to the Ad-CEA injected mice (7.8 μg/l), whereas a 10-fold increase in protein level was observed upon injection of 1 × 10⁶ pfu of the same virus (19.1 μg/l vs. 1.9 μg/l). In contrast, injection of lower doses of Ad-CEAopt did not result in a substantial increase in circulating CEA levels compared to Ad-CEA. The enhancement of CEA protein levels was also noted, albeit to a lower extent, upon injection of 25 or 50 μg of plasmid pVIJ/CEAopt relative to pVIJ/CEA (Fig. 5b). Thus, these results indicate that, independently of the gene transfer vehicle utilized, the codon usage optimized cDNA is expressed in vivo with greater efficiency than the corresponding wild-type sequence.

Figure 5. Expression of CEA in injected mice. Groups of C57Bl/6 mice (n = 10) were injected via the quadriceps muscle either with various doses of Ad/CEA or Ad/CEAopt (a) or with 25 or 50 μg of plasmids pVIJ/CEA or pVIJ/CEAopt (b). Blood samples were collected 3 days postinjection and CEA levels were measured. Filled triangles represent CEA measurement of individual mice. Timeline graph is shown on top. This experiment has been performed at least 3 times. Geometric mean values are also shown (empty circle). Ad/CEAopt 1 × 10⁷ pfu vs. Ad/CEA 1 × 10⁷ pfu (p = 0.004), Ad/CEAopt 1 × 10⁶ pfu vs. Ad/CEA 1 × 10⁶ pfu (p = 0.007), Ad/CEAopt 1 × 10⁵ pfu vs. Ad/CEA 1 × 10⁵ pfu (p = 0.151) and Ad/CEAopt 1 × 10⁴ pfu vs. Ad/CEA 1 × 10⁴ pfu (p = 0.82) (Mann-Whitney rank sum test). pVIJ/CEAopt 25 μg vs. pVIJ/CEA 25 μg (p = 0.03) and pVIJ/CEAopt 50 μg vs. pVIJ/CEA 50 μg (p = 0.002) (2-tailed Student's t-test).

Increased immunogenicity of CEAopt

To examine in vivo immune responses induced by the CEAopt expression vectors, C57Bl/6 mice were immunized intramuscularly with different doses of Ad-CEAopt ranging from 1 × 10⁵ to 1 × 10³ pfu. As comparison, groups of 8–10 mice were immunized with Ad-CEA in doses ranging from 1 × 10⁶ to 1 × 10⁴ pfu. Mice were subjected to 2 injections 3 weeks apart, and 2 weeks after the second immunization splenocytes were isolated from each mouse. To quantify the IFNγ secreting CEA-specific CD8⁺ T-cell precursor frequencies generated by the Ad-mediated immunization, the ELISPOT assay was performed using the H-2D^b-restricted T-cell epitope C₅₇₂GIQNSVSA₅₇₉ as stimulator. As shown in Figure 6a, immunization with 1 × 10⁴ pfu elicited a measurable immune response yielding 53 IFNγ spot forming cells (SFC) per 1 × 10⁶ splenocytes specific for the C₅₇₂GIQNSVSA₅₇₉ epitope (geometric mean value), whereas injection of 1 × 10³ pfu elicited negligible SFC values. The SFC increased to 302/1 × 10⁶ splenocytes in the group immunized with 1 × 10⁵ pfu of Ad-CEAopt. By contrast, at least 1 × 10⁵ pfu of Ad-CEA were necessary to elicit significant CD8⁺ T-cell precursor frequencies that increased to 168 SFC in the mouse group immunized with 1 × 10⁶ pfu. No peptide-specific IFNγ SFC was detected in the Ad-empty immunized mice (data not shown).

Figure 6. Codon usage optimization increases immune response to CEA. Groups of C57Bl/6 mice (n = 8) were injected via the quadriceps muscle either with various doses of Ad-CEA and Ad-CEAopt. Virus injections were carried out at 0 and 21 days. (a) At 2 weeks post-boosting injection, the number of CD8⁺ IFNγ-secreting T cells specific for CEA was determined by Elispot assay on splenocytes from individual mice (filled triangles) using peptide 143 that covers aa 569–583 and includes a CD8⁺ epitope. Two different amounts of splenocytes (2.5 × 10⁵ and 5 × 10⁵) and 2 replicas of each tested amount of splenocytes. Average values were calculated, from the background level determined in the absence of peptides (typically <10 SFC/10⁶ total splenocytes) was subtracted, and the result was expressed as the number of SFC/10⁶ total splenocytes. Timeline graph is shown on top. Values from individual mice (filled triangles) and the geometric mean values (empty circles) are shown. Ad-CEAopt (1 × 10³ pfu) vs. Ad-CEA (1 × 10⁴ pfu) p < 0.05 (Mann-Whitney rank sum test). (b) Anti-CEA antibody titers in sera from individual mice (filled triangles) were measured using 10 days post-boost serum samples. Geometric mean titers (empty circles) are also shown. The difference in the median between the 2 groups is statistically significant (p < 0.001) (Mann rank test).

Sera from mice immunized with 1 × 10⁵ pfu of each CEA Ad vector were tested in ELISA using the purified human CEA protein as substrate (Fig. 6b). CEA-specific antibody titer in Ad-CEAopt immunized mice was detected in all immunized mice, and the geometric mean value of the antibody titer was 46,474. By contrast, the Ad-CEA immunized group showed an approximately 100-fold lower geometric mean titer of CEA-specific antibody (454) (p = 0.0016).

Comparative immunization of different mice cohorts with DNA/DNA, DNA/Ad, Ad/Ad and Ad/DNA modalities confirmed that, independently of the intrinsic immunogenic properties of the injected cDNA, the DNA/Ad combination elicited the most significant immune response as measured by intracellular staining assay using peptides containing CD4⁺- and CD8⁺-specific epitopes. In addition, the immune response elicited by the DNA/Ad injection protocol was characterized by a greater CD4⁺ and CD8⁺ T-cell response than that observed in the other 3 groups (Table III). A representative dot blot analysis of IFNγ intracellular staining is shown in Figure 7. Thus, these results demonstrate that the codon usage optimized cDNA of CEA is more efficient than the wild-type cDNA in eliciting cellular and humoral immune responses directed against this tumor antigen. Also, these data further demonstrate that the DNA/Ad vaccination regimen is the most efficient in eliciting a significant CD4⁺ and CD8⁺ T-cell response.

Figure 7. Dot blot analysis of INFγ ICS. Representative dot blot of INFγ ICS of splenocytes obtained from a single mouse vaccinated with pVIJ/CEAopt and Ad-CEAopt as described in Table III. Splenocytes were stimulated with peptides CEA-107, CEA-133, CEA-155 (CD4⁺ peptides) and CEA-143 (CD8⁺ peptide).

Table III. Comparison of Immunogenic Efficiency of Different Immunization Regimens in C57Bl/6 Mice1

	A/A		D/A		D/D		A/D
	CD8^+	CD4^+	CD8^+	CD4^+	CD8^+	CD4^+	CD8^+	CD4^+
1 Mice (n = 6) were immunized with 50 μg of pVIJ CEAopt and 1 × 10^7 pfu of Ad-CEAopt as described in Material and Methods. IFNγ intracellular staining of splenocytes from single mice was performed with peptides CEA-107, CEA-133, CEA-155 (CD4^+ peptides) and CEA-143 (CD8^+peptides). Data refer to % of IFNγ producing CD8^+ (CD4^+) CD3^+ splenocytes.–A/A, Ad-CEAopt regimen; D/A, pVIJ/CEAopt-Ad-CEAopt regimen; D/D, pVIJ/CEAopt-pVIJ/CEAopt regimen; A/D, Ad-CEAopt-pVIJ/CEAopt regimen.
CD4^+peptides	0.01 ± 0.004	0.56 ± 0.45	0.02 ± 0.002	1.85 ± 0.9	0.03 ± 0.002	0.76 ± 0.6	0.01 ± 0.004	0.97 ± 0.4
CD8^+peptide	1.34 ± 0.4	0.04 ± 0.003	10.92 ± 3.5	0.08 ± 0.004	0.99 ± 0.6	0.02 ± 0.002	0.9 ± 0.6	0.04 ± 0.004
DMSO	0.01 ± 0.001	0.01 ± 0.001	0.01 ± 0.001	0.01 ± 0.001	0.01 ± 0.001	0.01 ± 0.001	0.01 ± 0.001	0.01 ± 0.001

cDNAopt breaks tolerance in CEA.tg mice

To determine whether the enhanced immunogenic properties of the codon usage optimized cDNA of CEA would more efficiently break tolerance to human CEA, CEA transgenic mice were immunized with vectors carrying either the wild type or the CEAopt sequences. These transgenic mice carry the entire human CEA gene and flanking sequences and express the CEA protein in the intestine, mainly in the cecum and colon. Thus, this mouse line is a useful model to study the safety and efficacy of immunotherapy strategies directed at this tumor self-antigen.36

Groups of 10 mice were subjected to 4 injections of 50 μg plasmid DNA followed by a final injection of 1 × 10⁸ pfu of Ad. In addition, a high dose (1 × 10⁸ pfu) of Ad-CEAopt was used in view of the unresponsiveness to CEA antigen of the CEA transgenic mice.36 The immune response to CEA was analyzed by IFNγ-ELIspot assay on splenocytes obtained from 4 injected mice. As comparison, groups of C57Bl/6 mice were immunized with a single injection of either pVIJ/CEA or pVIJ/CEAopt followed by immunization with the corresponding Ad vector. As shown in Table IV, the immune response to CEA was detected only with the splenocytes from the CEA transgenic mice immunized with the CEAopt cDNA. The immune response was detected primarily with peptide pool D. Comparison of the immune response elicited by the CEA and CEAopt vectors in C57Bl/6 and CEA transgenic mice clearly showed that the latter are tolerant to CEA and the codon optimized cDNA is required to elicit a measurable immune response to this target antigen. Intracellular staining assay on mouse splenocytes showed the induction of a CD8⁺ T-cell response. Also, the immune response within the peptide pool D was primarily targeted against the CD8⁺ epitope present within peptide CEA-143 (Table V). Interestingly, no obvious CD4⁺ T cell or humoral response to CEA was detected, suggesting that, unlike the wild-type C57Bl/6 mice, induction of a CD4⁺ T-cell response against the self-CEA antigen is more difficult to achieve in this transgenic mouse line.

Table IV. IFN_γ-Elispot Analysis of Immunized Mice1

Mouse	Immunogen	Pool A	Pool B	Pool C	Pool D	DMSO
1 Mice were immunized with plasmid DNA (50 μg) and Ad vectors (1 × 10^8 pfu) carrying the wild-type or codon optimized cDNA of CEA as described in Material and Methods. Data shown one the averages of the values obtained from 6 injected mice. SFC/10^6 splenocytes.
CEA tg	CEAopt	55 (± 10)	16 (± 1)	19 (± 3)	323 (± 80)	6 (± 6)
CEA tg	CEA	0	2 (± 2)	3 (± 2)	3 (± 2)	0
C57B1/6	CEAopt	–	–	–	> 1,562 (± 440)	0
C57B1/6	CEA	–	–	–	500 (± 100)	0

Table V. Comparison of immunogenic efficacy in CEA transgenic mice of plasmid and Ad vectors carrying the wild-type or codon optimized cDNA of CEA1

Peptides	D/A		D/Aopt
	CD8^+	CD4^+	CD8^+	CD4^+
1 CEA transgenic mice were immunized with immunized plasmid DNA (50 μg) and Ad vectors (1 × 10^8 pfu) carrying the wild-type or codon optimized cDNA of CEA as described in Material and Methods. These results are representative of 4 independent experiments. Data refer to % of IFNγ-producing CD8^+ (CD4^+) CD3^+ splenocytes. Data were obtained with pooled splenocytes of 6 injected mice.
Pool D	0.01	0.01	0.92	0.03
CEA-143-3	0.01	0.01	0.24	0.01
DMSO	0.01	0.01	0.01	0

To verify whether immunization with vectors carrying the codon optimized cDNA of CEA would elicit an immune response capable of protecting mice from tumor growth, a group of 10 CEA tg mice were immunized with repeated DNA-EP followed by an injection of 1 × 10⁸ pfu of Ad-CEAopt. Two weeks after the last immunization, mice were challenged with an s.c. injection of 5 × 10⁵ MC-38-CEA cells, a syngenic tumor cell line that expresses CEA.36 Tumor growth in vaccinated mice was monitored for 35 days and compared to the growth rate observed in control group. As shown in Figure 8a, although only 2 of 10 vaccinated mice were completely protected from tumor challenge, the onset of tumor in the treated mice was significantly delayed compared to the control group (p = 0.0209). In addition, the tumor volume in the vaccinated mice measured at day 30 post-challenge was on average significantly smaller than that in control mice (Fig. 8b, geometric mean values 61 and 561 mm³, respectively) (p = 0.05). Interestingly, complete protection from tumor growth was observed upon vaccination of C57Bl/6 mice with a single injection of pVIJ/CEAopt followed by Ad-CEAopt. In contrast, no protection from tumor growth was observed upon immunization with control plasmid DNA and Ad-empty vectors (Fig. 8c). Thus, these results indicate that the codon usage optimized cDNA of CEA is more immunogenic than the wild-type sequence and that, under these experimental conditions, tolerance to CEA can be broken only by using the codon optimized cDNA as immunogen. In addition, a moderate antitumor effect can be exerted upon vaccination with plasmid and Ad vectors carrying the synthetic gene of CEA.

Figure 8. Antitumor effect of vaccination with vectors carrying codon optimized cDNA of CEA. Groups of CEA transgenic mice (n = 10) were immunized with repeated weekly injections of 50 μg of pVIJ/CEAopt followed by a boost with 1 × 10⁸ pfu of Ad-CEAopt. Similarly, C57Bl/6 mice were vaccinated with a single injection of pVIJ/CEA-opt followed by Ad-CEAopt. Two weeks after the last injection, mice were challenged with s.c. inoculation of 5 × 10⁵ MC38-CEA tumor cells. Percentage of tumor-free mice in the vaccinated group was determined at weekly intervals and compared to that of untreated controls. The timeline graph is shown. (a) The CEA transgenic mice vaccinated group (empty circles) is significantly different from control (filled triangle, dashed line) (log rank test p = 0.0209). (b) Tumor volume of each mouse (filled triangles) was measured at day 30 post-challenge. Geometric mean values (empty circles) are shown. The CEA transgenic mice vaccinated group is significantly different from control (Student's t-test p = 0.05). (c) Tumor-free percentage of C57Bl/6 mice vaccinated with pVIJ/CEAopt and Ad/CEAopt (empty circles), control DNA and Ad vectors (empty squares) and untreated mice (filled triangles, dashed line) (n = 10). These results are representative of 2 independent experiments.

Discussion

In our study, we have examined the immunogenic properties of a genetic vaccine for CEA based on the electroporation of plasmid DNA and injection of recombinant adenovirus vector. The results obtained indicate that EP of DNA followed by Ad injection can elicit a significant immune response to the target antigen and that immunogenicity of the genetic vaccine is enhanced by the use of the codon usage optimized cDNA of CEA.

Comparison of different modalities of immunization has demonstrated that DNA EP followed by Ad injection leads to a greater immune response to CEA than DNA EP or Ad alone (Fig. 2, Table III). This observation is in agreement with an emerging body of data suggesting that vaccination protocols based on the use of different vaccine vehicles administered sequentially are capable of eliciting antigen-specific immune responses with greater efficiency than those that are based on a single vector.42, 43, 44 Why DNA-EP followed by Ad is more immunogenic than either vector alone is not clear, but several factors may contribute to the enhanced immunogenic properties of this vaccination regimen. Priming the immune responses with DNA EP probably drives the immune system to recognize exclusively CEA since it is the only protein expressed by the plasmid vector. In addition, although Ad vectors are extremely efficient vehicles for genetic vaccine, the immune response elicited upon repeated injection of this virus may be directed not only against target antigen but also against viral proteins. Injection of recombinant Ad vectors has been shown to elicit in mice T-cell response to viral hexon and DNA binding protein.45 Thus, induction of a virus-specific cell-mediated response may compete, at least in part, with the immune recognition of CEA as target antigen. Finally, it has been reported that injection of Ad vectors is likely to induce neutralizing antibodies to the virus, thereby hampering its immunogenic efficiency.46, 47 Indeed, we have detected high titers of Ad neutralizing antibodies in the transgenic mice after Ad injection that have a detrimental effect on the immunogenicity of subsequent Ad injections. However, unlike Ad, plasmid DNA injection can fully boost the T-cell response elicited by Ad injection (data not shown). Thus, immunization with plasmid DNA will not be affected by neutralizing responses against the viral vector and also drives the immune system to focus primarily on recognition of CEA as target antigen.

In light of these considerations, the efficacy of immunization regimens based on the use of plasmid DNA and baculovirus vectors should also be investigated. The latter has been shown to elicit a CEA-specific immune response in C57Bl/6 mice35 and, since it does not normally infect humans, could work well in combination of plasmid DNA.

Vectors carrying the codon usage optimized cDNA of CEA elicit stronger immune response than their wild-type counterparts both in C57Bl/6 and in CEA transgenic mice (Tables IV and V). The difference in immunogenic efficiency is probably connected, at least partially, to enhanced expression of the target protein (Fig. 6). This observation is in line with data obtained in mice and nonhuman primates with genetic vaccines for HCV, HIV and HPV, whereby vectors carrying codon sequence optimized cDNA are characterized by increased expression and enhanced immunogenicity of the viral polypeptides.26, 27, 28, 29, 30, 31, 32 Although it is reasonable to assume that the higher expression of CEA will lead to enhanced presentation by the MHC of CEA-derived epitopes, an enhanced immune response can also be ascribed, at least in part, to an increase in CpG motifs in the synthetic gene administered. The role of immune modulation of unmethylated CpG in governing the immune response has been described both in rodents and in nonhuman primates.48, 49 Thus, as a number of CpG has been introduced into the synthetic cDNA of CEA, these may also contribute to the enhanced immunogenic properties of the CEA expressing vectors.

Since host T-cell immunity has been correlated with protection from CEA-positive tumor development both in preclinical and clinical studies,50, 51 being able to measure the induction of CD4⁺ and CD8⁺ T cells quantitatively in mouse models is important. Prior to this work, only limited information has been published on the identification of CEA epitopes specific for C57Bl/6 and CEA transgenic mice.52 The latter strain is commonly used as the mouse model of choice in determining the efficacy of a variety of immunotherapy approaches that target CEA.53, 54 In our study, T-cell responses to pools of CEA peptides were induced in C57Bl/6 mice after DNA EP prime Ad boost regimen. The data reported confirm the immune response elicited by this antigen was primarily directed toward the C-terminal region in C57Bl/6 mice as observed upon immunization with a baculovirus vector carrying the CEAopt cDNA.35 Further dissection of the responses to individual peptides revealed CEA epitopes for both CD4⁺ and CD8⁺ T cells in the H-2^b background. The CD8⁺ T-cell epitope is located at the carboxy-terminus aa 572–579 of CEA and stands out as the most immunodominant epitope both in C57Bl/6 and in CEA transgenic mice. However, the induced CTL did not lyse MC38-CEA when used as target cells (data not shown). The data were similar to the observation that an H2D^b-restricted CD8⁺ CTL line specific for the CEA_526-533 peptide efficiently lysed peptide-pulsed target cells but not CEA-expressing tumors.52 Thus these data suggest that these epitope-specific CTLs might be low-avidity T cells that are unable to lyse tumor cells expressing CEA. Whether the CD8⁺ T cells specific for the CEA_572–579 epitope contribute directly to the antitumor effect observed in CEA transgenic mice (see below) or alternatively by production of factors that facilitate the antitumor activity of other components of the immune response is not entirely clear and is currently subject of investigation in our laboratory.

Interestingly, the immunologic profile of the CEA transgenic mice differs significantly from that of the C57Bl/6 mice. The lesser immunogenic vectors carrying the wild-type cDNA of CEA are unable to elicit an antigen-specific immune response in these mice, whereas vaccination with vectors carrying the synthetic cDNA of CEA only partially succeeds in inducing an immune response to the tumor antigen. Specifically, vaccination with pVIJ/CEAopt and Ad-CEAopt results only in the induction of CEA-specific CD8⁺ T-cell response, whereas no CD4⁺ T-cell response or anti-CEA antibodies could be detected in the immunized mice. In view of the observation that CEA transgenic mice are fully immunocompetent and can mount an immune response to exogenous antigens with the same efficiency as the C57Bl/6 mice (data not shown), it is reasonable to conclude that tolerance to CEA curbs the immunoresponsiveness of the transgenic mice to CEA-derived epitopes. Nonetheless, the induction of a CEA-specific CD8⁺ T-cell response suggests that it is possible to break tolerance to this target antigen. It is surprising that the CD4⁺ T cells' tolerance to CEA could not be overcome by the immunization regimen. Immunization of colorectal carcinoma patients with recombinant CEA protein has been shown to elicit proliferative T-cell responses that are both MHC class I and II restricted, suggesting that immunization against this particular tumor antigen can elicit both CD4⁺ and CD8⁺ T cells.55 The lack of induction of CD4⁺ T cells' in the CEA transgenic mice is not limited to vaccination with plasmid DNA and Ad vectors but can also be observed upon injection of dendritic cells (DC) transduced with Ad-CEAopt (data not shown). Thus, it would appear that CD4⁺ T-cell tolerance in the CEA transgenic mice is greater than in humans. This phenomenon may be a peculiarity of transgenic mice, since vaccination of MUC1 transgenic mice with DC-based vaccine results in the preferential induction of a CD8⁺ T-cell response without any detectable CD4⁺ T-cell response specific for MUC I.56 Nonetheless, the lack of detection of CD4⁺ T-cell response can be dictated, at least in part, by the nature of the immunization rather than the animal model.

Vaccination with plasmid DNA followed by Ad boost exerts a protective antitumor effect resulting in slower tumor growth kinetics in immunized mice (Fig. 8). This observation is in line with published data demonstrating that vaccination with CEA expressing pox vectors or plasmid DNA can have a therapeutic effect in mice and protects against tumor development.57, 58 In addition, the extent of protection from tumor challenge observed in CEA transgenic mice appears to be comparable to that observed upon vaccination of the same mouse strain with an oral DNA vaccine followed by challenge with MC38-CEA cells.59, 60 Although protection from tumor challenge was limited, it is entirely possible that further optimizing the intrinsic immunogenic properties of the cDNA of CEA could lead to a far greater antitumor effect. In this regard, fusion to the coding sequence of viral and tumor antigens of immunoadjuvant moieties such as cytokines, helper epitopes and bacterial toxins has resulted in enhanced immunogenicity of the target antigen and in a concomitant increased protection from infection or tumor challenge.61 Interestingly, protection from tumor growth upon vaccination of C57Bl/6 mice was complete, thus indicating that to evaluate the potency of an antitumor vaccine, studies must be carried out in a tolerant animal model. It is likely that evaluation of immunogenic and antitumor potency of a vaccine candidate in wild-type mice will give misleading results since it will not include breakage of tolerance to the target antigen as an additional potency criterion.

These results provide evidence that vaccination strategies based on the use of plasmid DNA followed by Ad boost are likely to elicit significant immune responses to tumor antigens. In addition, they demonstrate for the first time to our knowledge that the cDNAopt of CEA is a valid immunogen that can be incorporated into any genetic vector to increase immune responses to this target antigen. Improvement of the amplitude and type of the CEA-specific immune response will contribute to an antitumor therapeutic effect.

Acknowledgements

We thank Dr. A. Folgori, Dr. A. Vitelli, Dr. R. Savino and Dr. A. Nicosia for helpful suggestions during the course of our study. We also thank Dr. G.J. Prud'homme for plasmid pCR-CEA, Dr. J. Clench for editorial assistance, Ms. M. Emili for graphics and LAR personnel for technical support in conducting animal experiments. We are also grateful to D. Peruzzi for critical review.