A Triad of Costimulatory Molecules Synergize to Amplify T‐‐Cell Activation in Both Vector‐‐Based and Vector‐‐Infected Dendritic Cell Vaccines

Abstract

The activation of a T cell has been shown to require two signals via molecules present on professional antigen presenting cells: signal 1, via a peptide//MHC complex, and signal 2, via a costimulatory molecule. Here, the role of three costimulatory molecules in the activation of T cells was examined. Poxvirus ((vaccinia and avipox)) vectors were employed because of their ability to efficiently express multiple genes. Murine cells provided with signal 1 and infected with either recombinant vaccinia or avipox vectors containing a TRIad of COstimulatory Molecules ((B7‐‐1//ICAM‐‐1//LFA‐‐3, designated TRICOM)) induced the activation of T cells to a far greater extent than cells infected with vectors expressing any one or two costimulatory molecules. Despite this T‐‐cell “hyperstimulation” using TRICOM vectors, no evidence of apoptosis above that seen using the B7‐‐1 vector was observed. Results employing the TRICOM vectors were most dramatic under conditions of either low levels of first signal or low stimulator cell to T‐‐cell ratios. Experiments employing a four‐‐gene construct also showed that TRICOM recombinants could enhance antigen‐‐specific T‐‐cell responses in vivo. These studies thus demonstrate the ability of vectors to introduce three costimulatory molecules into cells, thereby activating both CD4⁺⁺ and CD8⁺⁺ T‐‐cell populations to levels greater than those achieved with the use of only one or two costimulatory molecules. This new threshold of T‐‐cell activation has broad implications in vaccine design and development.Dendritic cells infected with TRICOM vectors were found to greatly enhance naïve T‐‐cell activation, and peptide‐‐specific T‐‐cell stimulation. In vivo, peptide‐‐pulsed DCs infected with TRICOM vectors induced cytotoxic T lymphocyte activity markedly and significantly greater than peptide‐‐pulsed DCs.

• Vaccines
• Cancer vaccine
• Cancer
• Costimulation
• T cells

Introduction

The extent of the primary response of T cells, which involves their activation, expansion and differentiation, is paramount to a successful immune response to an antigen. The initiation of an immune response requires at least two signals for the activation of naive T cells by antigen presenting cells ((APC)) ((Damle et al., [1992]; Guinan et al., [1994]; Hellstrom et al., [1996])). The first signal is antigen specific, delivered through the T‐‐cell receptor ((TCR)) via the peptide//MHC, and causes the T cell to enter the cell cycle. The second, or “costimulatory,” signal is required for cytokine production and proliferation. At least three distinct molecules normally found on the surface of professional APC have been reported as capable of providing the second signal critical for T‐‐cell activation: B7‐‐1 ((CD80)), Intercellular adhesion molecule‐‐1 ((ICAM‐‐1; CD54)), and Leukocyte function‐‐associated antigen‐‐3 ((LFA‐‐3; human CD58; murine CD48)) ((Cavallo et al., [1995]; Damle et al., [1992]; Dubey et al., [1995]; Guinan et al., [1994]; Hellstrom et al., [1993]; Parra et al., [1993], [1997]; Sperling et al., [1996]; Wingren et al., [1995])). The T‐‐cell ligands for these costimulatory molecules are distinct: B7‐‐1 interacts with the CD28 and CTLA‐‐4 molecules, ICAM‐‐1 interacts with the CD11a//CD18 ((LFA‐‐1//β2 integrin)) complex, and LFA‐‐3 interacts with the CD2 ((LFA‐‐2)) molecules. These molecules have been individually shown to costimulate T‐‐cell proliferation in vitro ((Wingren et al., [1995])). However, because they may be expressed simultaneously on APC, it has been difficult to examine relative potencies of individual costimulatory molecules during the induction of T‐‐cell proliferation ((Damle et al., [1992])).

Because it has been proposed that both antigen and costimulatory molecules must be expressed in the same cell to properly engage the TCR and costimulatory receptor, respectively, an admixture of several recombinant vectors could be utilized to explore the potential cooperation of costimulatory molecules ((Hellstrom et al., [1993])). The disadvantage of this approach, however, is that the admixture of three or more vectors or viruses has a statistically diminished probability of co‐‐infecting the same cell. Thus, a multi‐‐gene construct would be preferable for expression of multiple costimulatory molecule genes in the same cell. We report here a review of the use of constructs containing and expressing a TRIad of COstimulatory Molecules ((B7‐‐1, ICAM‐‐1, and LFA‐‐3, designated TRICOM)) ((Hodge et al., [1999])). The synergistic effect of these costimulatory molecules on the enhanced activation of T cells is demonstrated. More specifically, these three costimulatory molecule genes have been inserted into two vectors: vaccinia, which is replication competent in mammalian cells, and avipox ((fowlpox)), which is replication defective in mammalian cells. In each case, the degree of T‐‐cell activation using vectors containing three costimulatory molecules was far greater than the sum of the constructs, each containing one costimulatory molecule.

DCs are highly specialized APCs that function as the principal activators of quiescent T cells and, thus, cellular immune responses in vivo ((Specht et al., [1997])). Consequently, the unparalleled capacity of DCs to induce antigen‐‐specific T‐‐cell responses has focused the attention of many investigators on the potential effectiveness of these cells in immunoprevention and immunotherapy. In experimental models, DCs pulsed with peptides from tumor‐‐associated antigens have been shown to induce antigen‐‐specific antitumor responses in vivo, and fusion of DCs with tumor cells has also been shown to enhance antitumor immunity. In addition, peptide‐‐pulsed DCs are being explored for the prevention and treatment of infectious diseases such as HIV, tuberculosis, chlamydia, and Epstein–Barr virus.

As DCs express high levels of histocompatibility molecules and most known costimulatory molecules, including B7‐‐1, ICAM‐‐1 and LFA‐‐3, it is generally believed that the addition of a vector to express even higher levels of these costimulatory molecules would be of little, if any, advantage. On the other hand, one could theorize that hundreds of thousands of years of evolution would have placed the capacity of DCs to activate T cells into a “median” state between 1)) the ability to activate T cells to specific antigens such as those of microbial pathogens and 2)) the induction of autoimmunity to self antigens of the host. Thus, we theorize that the natural expression levels of specific costimulatory molecules in either immature or mature DCs would accommodate this “median” state of efficacy of DC populations. Poxvirus vectors were used to hyperexpress the triad of costimulatory molecules because 1)) they can accommodate and express multiple transgenes, and 2)) they possess a high efficiency of infection of most cell types. Both replication‐‐defective fowlpox ((rF)) and replication‐‐competent vaccinia ((rV)) recombinant vectors were employed to hyperexpress the recombinant costimulatory molecules. This report provides a review of studies ((Hodge et al., [1999], [2000])) to determine if the use of vectors that can facilitate hyperexpression of multiple costimulatory molecules can enhance the capacity of DCs to stimulate various T‐‐cell responses.

Results

Expression of Recombinant Costimulatory Molecules

To confirm that each of the recombinant vectors could express the appropriate costimulatory molecule transgene((s)), the murine adenocarcinoma cell line MC38 was infected with the various recombinant vaccinia or fowlpox constructs, and cell‐‐surface expression of the transgene((s)) was demonstrated by flow cytometry. Uninfected cells and cells infected with wild‐‐type vaccinia virus ((V‐‐WT)) failed to express any of the three costimulatory molecules. This observation was confirmed by RT‐‐PCR analysis. In contrast, cells infected with rV‐‐B7‐‐1, rV‐‐ICAM‐‐1, or rV‐‐LFA‐‐3 became positive for their respective transgenes. Similar analysis of a construct containing two costimulatory molecules ((rV‐‐B7‐‐1//ICAM‐‐1)) showed enhanced expression of B7‐‐1 and ICAM‐‐1. Moreover, cells infected with the vaccinia multiple‐‐gene construct rV‐‐TRICOM co‐‐expressed all three costimulatory molecules. Coexpression of all three costimulatory molecules on greater than 79%% of cells was confirmed by three‐‐color cytometric analysis. To determine if the recombinant fowlpox viruses expressed their recombinant proteins, MC38 cells were infected with the fowlpox constructs in a similar manner. Again, cells infected with WT‐‐FP failed to express any costimulatory molecule. Cells infected with rF‐‐B7‐‐1 became positive for B7‐‐1 protein, and cells infected with rF‐‐ICAM‐‐1 became positive for ICAM‐‐1 protein. An rF‐‐LFA‐‐3 vector was not constructed. However, cells infected with the fowlpox multiple‐‐gene construct rF‐‐human carcinoembryonic antigen ((CEA))//TRICOM co‐‐expressed all three costimulatory molecules.

B7‐‐1, ICAM‐‐1 and LFA‐‐3 Synergize to Enhance T‐‐Cell Proliferation

The B7‐‐1, ICAM‐‐1 and LFA‐‐3 molecules have been shown individually to costimulate T‐‐cell proliferation. However, because they may be expressed simultaneously on APC, it has been difficult to examine relative roles of individual costimulatory molecules during the induction of T‐‐cell proliferation ((Damle et al., [1992])). To analyze the contribution of B7‐‐1, ICAM‐‐1 and//or LFA‐‐3 molecules to the induction of naive T‐‐cell proliferation, an in vitro model ((Parra et al., [1997])) was employed where the first signal for T‐‐cell activation was delivered via a pharmacological reagent ((Concanavalin A [[Con A]])). A panel of stimulator cells that differed only in costimulatory molecules was created using the MC38 cell line infected with various recombinant vaccinia ((Figure 1A)) or fowlpox ((Figure 1B)) viruses engineered to express costimulatory molecules. The second, or costimulatory, signal was delivered to the T cell via one or more costimulatory molecules expressed on the surface of these stimulator MC38 cells. As shown in Figure 1A, both uninfected MC38 cells and MC38//V‐‐WT induced marginal proliferation of T cells at all concentrations of Con A examined. MC38//LFA‐‐3 induced a small ((2.1‐‐fold)) but significant ((p < 0.05)) increase in T‐‐cell proliferation. Delivery of signal 2 via MC38//ICAM‐‐1 induced a 3.5‐‐fold increase in T‐‐cell proliferation at 2.5 μg//ml Con A. MC38//B7‐‐1 induced a 7.8‐‐fold and 16‐‐fold increase in proliferation at 2.5 and 1.25 μg//ml Con A, respectively. However, MC38//TRICOM ((MC38 cells co‐‐expressing all three costimulatory molecules)) induced a 17.5‐‐fold increase in T‐‐cell proliferation at 2.5 μg//ml Con A, and a 34‐‐fold increase at 1.25 μg//ml Con A. Moreover, at low Con A levels ((0.625 μg//ml)), expression of ICAM‐‐1 and LFA‐‐3 did not induce T‐‐cell proliferation. While B7‐‐1 induced measurable proliferation ((20,000 CPM)) at 0.625 μg//ml Con A, the coexpression of all three costimulatory molecules induced an even greater level of proliferation ((100,000 CPM)) ((Figure 1A)). These experiments were repeated four times with similar results.

Figure 1. Effect of multiple costimulatory molecules on T cell proliferation. Naive murine T cells, in the presence of varying concentrations of Con A to provide the first signal, were co‐‐cultured with MC38 stimulator cells infected with either recombinant vaccinia ((A)) or recombinant fowlpox ((B)) vectors. Recombinant vectors were wild‐‐type ((i.e., V‐‐WT or WT‐‐FP [[open squares]])), rV‐‐LFA‐‐3 ((closed triangle)), rV‐‐ICAM‐‐1 or rF‐‐ICAM‐‐1 ((closed circles)), rV‐‐B7‐‐1 or rF‐‐B7‐‐1 ((closed diamonds)), and rV‐‐TRICOM or rF‐‐CEA//TRICOM ((closed squares)). Uninfected MC38 cells are open circles. Bars, SD. These data are as presented in Hodge et al. (([1999])).

MC38 stimulator cells were also prepared by infection with recombinant fowlpox vectors ((Figure 1B)). Again, uninfected MC38, MC38//WT‐‐FP, or MC38//rF‐‐CEA induced marginal proliferation of T cells at all levels of Con A examined. MC38//rF‐‐ICAM‐‐1 supported a 2‐‐fold increase, MC38//rF‐‐B7‐‐1 supported a 3.2‐‐fold increase, and MC38//rF‐‐CEA//TRICOM supported a 6‐‐fold increase in T‐‐cell proliferation at 2.5 μg//ml Con A. Similar results were obtained when this experiment was repeated two additional times and when the first signal was delivered via immobilized anti‐‐CD3.

To further confirm the specificity of the proliferative contribution of B7‐‐1, ICAM‐‐1, or LFA‐‐3, MC38 stimulator cells were prepared by infection with V‐‐WT, rV‐‐B7‐‐1, rV‐‐ICAM‐‐1, or rV‐‐LFA‐‐3 and co‐‐cultured with naive murine T cells and Con A in the presence or absence of MAb specific for the given costimulatory molecule. MC38//B7‐‐1 enhanced T‐‐cell proliferation 4.5‐‐fold more than that of MC38//V‐‐WT. This increased proliferation was inhibited 83%% by the addition of a blocking MAb for murine B7‐‐1. Similarly, MC38//ICAM‐‐1 increased proliferation 2.25‐‐fold, which was then reduced by 88%% in the presence of anti‐‐murine ICAM‐‐1 MAb. Finally, MC38//LFA‐‐3 increased proliferation 2.1‐‐fold, which was then reduced by 98%% in the presence of anti‐‐murine CD48 MAb. For each group, incubation with the appropriate isotype control antibody failed to block the noted proliferation. This experiment was repeated two additional times with similar results.

Determination of Costimulatory Molecule Capacity

Modification of the in vitro costimulation assay allowed a quantitative estimation of the relative capacity of B7‐‐1, ICAM‐‐1 and//or LFA‐‐3 to deliver the second signal for T‐‐cell proliferation. To that end, stimulator cells ((MC38 cells infected with the various recombinant vaccinia viruses)) were titered by dilution with varying amounts of MC38 cells infected with V‐‐WT and co‐‐cultured with a constant number of T cells in the presence of 2.5 μg//ml Con A. The total MC38 to T‐‐cell ratio in these experiments remained constant at 1:10. MC38//LFA‐‐3 enhanced proliferation of T cells over that of MC38//V‐‐WT out to a dilution of 40%% ((i.e., of the stimulator cells in the well, 40%% were infected with rV‐‐LFA‐‐3 and the remaining 60%% were infected with V‐‐WT)). MC38//ICAM‐‐1 or MC38//B7‐‐1 supported increased T‐‐cell proliferation out to dilutions of 13%% and 6%%, respectively. In contrast, MC38//TRICOM enhanced proliferation when less than 3%% of stimulator cells contained the TRICOM vector ((extrapolated to less than 1%% via linear least squares analysis)). Given the titration curves of these individual costimulatory molecules, it appeared that the extent of T‐‐cell proliferation mediated by ICAM‐‐1 and B7‐‐1 is 3‐‐fold and 6‐‐fold, respectively, more potent than that mediated by LFA‐‐3 alone. Clearly, the strongest proliferation, however, is mediated by TRICOM. It should be noted that at relatively low stimulator cell concentrations ((i.e., when 3%%–6%% of the MC38 cells are acting as stimulator cells)), expression of LFA‐‐3, ICAM‐‐1, and even B7‐‐1 alone, does not enhance T‐‐cell activation, while the TRICOM‐‐expressing stimulator cells substantially enhance T‐‐cell activation. The MC38 cells infected with the two‐‐gene construct ((rV‐‐B7‐‐1//ICAM‐‐1)) induced little, if any, proliferation of T cells under these conditions, while MC38//TRICOM increased proliferation substantially ((p < 0.0001)).

Costimulation of CD4⁺⁺ and CD8⁺⁺ T Cells

To further characterize the T‐‐cell response to costimulatory molecules expressed singly or in combination, the ability of B7‐‐1, ICAM‐‐1 and LFA‐‐3 to costimulate purified CD4⁺⁺ and CD8⁺⁺ T cells was tested. Figure 2A and B, respectively, show the proliferation of purified CD4⁺⁺ and CD8⁺⁺ cells activated with suboptimal concentrations of Con A. The stratification of stimulator cell effects on proliferation was similar for both CD4⁺⁺ and CD8⁺⁺ cells: MC38//LFA‐‐3 stimulated the weakest proliferation, followed by MC38//ICAM‐‐1 and MC38//B7‐‐1. MC38//TRICOM were the most potent stimulator cells for both CD4⁺⁺ and CD8⁺⁺ T cells. These experiments were repeated three additional times with similar results. It should be noted that at very low concentrations of Con A ((0.625 μg//ml, Figure 2C and D)), there was no significant enhancement in activation of CD4⁺⁺ or CD8⁺⁺ T cells when ICAM‐‐1, LFA‐‐3, B7‐‐1, or the B7‐‐1//ICAM‐‐1 dual recombinant was used to provide the second signal. However, substantial activation of both T‐‐cell subsets was observed when the TRICOM vector was employed. Similar results were noted when the first signal was delivered via immobilized anti‐‐CD3.

Figure 2. Effect of costimulation on specific T‐‐cell populations. Murine CD4⁺⁺ ((A)) or CD8⁺⁺ T cells ((B)) were co‐‐cultured with uninfected MC38 cells ((open circle)), or cells infected with V‐‐WT ((open squares)), rV‐‐LFA‐‐3 ((closed triangles)), rV‐‐ICAM‐‐1 ((closed circles)), rV‐‐B7‐‐1 ((closed diamonds)) or rV‐‐TRICOM ((closed squares)) at a 10:1 ratio for 48 h in the presence of various concentrations of Con A. C and D show the proliferative responses of purified CD4⁺⁺ and CD8⁺⁺ cells, respectively, when co‐‐cultured in the presence of vector‐‐infected MC38 stimulator cells at a low Con A concentration ((0.625 μg//ml)). Bars, SD. These data are as presented in Hodge et al. (([1999])).

Two additional models ((peptide‐‐specific and allospecific)) were also used to demonstrate the efficacy of rV‐‐TRICOM in enhancing T‐‐cell proliferation. Naïve splenocytes from C57BL//6 mice, depleted of T cells, were used as APC either uninfected or infected with V‐‐WT, rV‐‐B7‐‐1, or rV‐‐TRICOM, followed by irradiation. Allogeneic ((BALB//c)) or syngeneic splenic T cells were used as responder cells in mixed lymphocyte reactions. C57BL//6 splenocytes infected with rV‐‐TRICOM enhanced allospecific T‐‐cell proliferation to far greater levels than uninfected splenocytes or splenocytes infected with V‐‐WT or rV‐‐B7‐‐1. All stimulator cell populations, even those infected with TRICOM, that were incubated with syngeneic T cells resulted in no proliferation ((< 1,000 CPM)).

Peptide‐‐specific proliferation of established effector T cells revealed similar results. When MC38 stimulator cells were pulsed with OVA peptide and co‐‐cultured with an OVA‐‐specific T‐‐cell line, MC38//B7‐‐1 induced a 9‐‐fold increase in peptide‐‐specific T‐‐cell proliferation over that of MC38//V‐‐WT, while MC38//TRICOM induced a 20‐‐fold increase in peptide‐‐specific proliferation. Stimulator cells pulsed with control peptide VSVN did not support proliferation of the OVA‐‐specific T‐‐cell line ((< 1,000 CPM)).

Cytokine Studies

It has been reported that B7‐‐1 costimulation prolongs IL‐‐2 mRNA half‐‐life and upregulates of IL‐‐2 transcription, resulting in production of considerable amounts of secreted IL‐‐2 ((Guinan et al., [1994]; Sperling et al., [1996])). Additionally, T‐‐cell costimulation with LFA‐‐3 has been reported to have an effect on a variety of cytokines, notably IL‐‐2 and IFN‐‐γ ((Wingren et al., [1995])). To determine qualitative and quantitative effects of costimulation by single or multiple costimulatory molecules on cytokine production, purified CD4⁺⁺ and CD8⁺⁺ T cells were co‐‐cultured with various stimulator cells expressing either B7‐‐1, ICAM‐‐1 or LFA‐‐3, or expressing all three molecules ((TRICOM)) in the presence of 2.5 μg//ml Con A. Supernatant fluids were analyzed for IL‐‐2, IFN‐‐γ, TNF‐‐α, granulocyte‐‐macrophage colony‐‐stimulating factor ((GM‐‐CSF)), and IL‐‐4 after 24 h. Uninfected MC38 and MC38//V‐‐WT induced a marginal quantity of IL‐‐2 from CD4⁺⁺ cells, while MC38//B7‐‐1 induced 3,979 pg//ml. However, T‐‐cell stimulation with MC38//TRICOM induced a 10‐‐fold greater amount of IL‐‐2. Similarly, MC38//B7‐‐1 induced a marginal quantity of IL‐‐2 from CD8⁺⁺ cells, while MC38//TRICOM induced a 20‐‐fold greater amount ((6,182 pg//ml)). IFN‐‐γ production by stimulated T cells was also examined. MC38//B7‐‐1 and MC38//LFA‐‐3 induced only moderate amounts of IFN‐‐γ from CD4⁺⁺ cells. In contrast, stimulation of CD4⁺⁺ cells with MC38//TRICOM induced 4‐‐fold more IFN‐‐γ than stimulation with any other construct. Stimulation of CD8⁺⁺ cells with MC38//TRICOM induced the greatest amount of IFN‐‐γ, greater than 6‐‐fold more than CD8⁺⁺ cells stimulated with any of the other constructs. Stimulation of either cell type with any construct failed to mediate significant changes ((p > 0.05)) in the levels of secreted TNF‐‐α, GM‐‐CSF, or IL‐‐4. It appears that the predominant culmination of stimulation via the TRICOM construct was IL‐‐2 secretion from CD4⁺⁺ cells and IFN‐‐γ secretion from CD8⁺⁺ T cells. These experiments were repeated three additional times with similar results. Studies were also carried out comparing stimulator cells infected with the two‐‐gene construct ((rV‐‐B7‐‐1//ICAM‐‐1)) with the triad construct ((rV‐‐TRICOM)) for their ability to enhance cytokine production by T cells. Only small differences were observed between the two in IFN‐‐γ production by either CD4⁺⁺ or CD8⁺⁺ cells, or in IL‐‐2 production by CD8⁺⁺ cells. But a substantial difference was seen in the stimulation of IL‐‐2 production by CD4⁺⁺ cells ((5,000 pg//ml employing MC38//B7‐‐1//ICAM‐‐1 vs. 39,600 pg//ml employing MC38//TRICOM)).

Cytokine expression in CD4⁺⁺ and CD8⁺⁺ T cells stimulated with single or multiple costimulatory molecules was also analyzed at the RNA level utilizing the multiprobe RNase protection assay. A radiographic profile and quantitative analysis are depicted ((Figure 3)). These experiments were repeated twice with similar results. Levels of IL‐‐4, IL‐‐5, IL‐‐10, IL‐‐15, and IL‐‐6 were similar in CD4⁺⁺ T cells stimulated with MC38//V‐‐WT, MC38//B7‐‐1, MC38//ICAM‐‐1, MC38//LFA‐‐3, or MC38//TRICOM ((Figure 3A and B, histogram)). IL‐‐2 and IFN‐‐γ expression levels were highest in CD4⁺⁺ cells stimulated with MC38//TRICOM when compared with CD4⁺⁺ cells stimulated with MC38 cells expressing any single costimulatory molecule ((Figure 3A and B)). Slightly higher levels of IL‐‐13, IL‐‐9, and IL‐‐6 were also noted in CD4⁺⁺ cells stimulated with MC38//TRICOM. Expression of cytokine genes was also analyzed in stimulated CD8⁺⁺ T cells. Of the cytokine RNAs analyzed, IL‐‐2 and, in particular, IFN‐‐γ levels were significantly higher when these cells were stimulated with MC38//TRICOM, compared with T cells stimulated with MC38 cells expressing any single costimulatory molecule ((Figure 3A and C)). Thus, the predominant synergistic effect of the triad of costimulatory molecules in cytokine production was IL‐‐2 in CD4⁺⁺ cells and IFN‐‐γ in CD8⁺⁺ T cells.

Figure 3. Effect of costimulation on cytokine RNA expression. A, murine CD4⁺⁺ or CD8⁺⁺ T cells were co‐‐cultured with MC38 stimulator cells infected with V‐‐WT ((lane A)), rV‐‐B7‐‐1 ((lane B)), rV‐‐ICAM‐‐1 ((lane C)), rV‐‐LFA‐‐3 ((lane D)), or rV‐‐TRICOM ((lane E)) at a T‐‐cell:stimulator‐‐cell ratio of 10:1 for 24 h in the presence of 2.5 μg//ml Con A. After culture, T‐‐cell RNA was analyzed by multiprobe Rnase protection assay. The quantitative representation of results from the autoradiograph is normalized for expression of the housekeeping gene L32 in B ((CD4⁺⁺ cells)) and C ((CD8⁺⁺ cells)). Order of histogram columns ((from left to right)): MC38//V‐‐WT, MC38//B7‐‐1, MC38//ICAM‐‐1, MC38//LFA‐‐3, and MC38//TRICOM. These data are as presented in Hodge et al. (([1999])).

Apoptosis Studies

To determine if stimulation of T cells with signal 1 and rV‐‐TRICOM would lead to cell survival or programmed cell death ((PCD)), T cells were activated with Con A for signal 1, cultured with V‐‐WT, rV‐‐B7‐‐1 or rV‐‐TRICOM‐‐infected MC38 cells, and replated for 24 h in medium to measure apoptosis. T cells activated by the combination of MC38 and Con A or MC38//V‐‐WT and Con A in the absence of costimulatory signals exhibited high levels of spontaneous apoptosis ((82.9±3.8 and 78.9±1, respectively [[Figure 4A and B]])). T cells activated by Con A and MC38//B7‐‐1 or Con A and MC38//TRICOM exhibited substantially less spontaneous apoptosis ((31.3±3.8 and 30.7±1, respectively [[Figure 4C and D]])).

Figure 4. Apoptosis of CD8⁺⁺ cells activated by Con A and either MC38 ((A)), MC38//V‐‐WT ((B)), MC38//B7‐‐1 ((C)), or MC38//TRICOM ((D)). Each panel depicts the level of apoptotic cells ((above line)) in each group as measured by the terminal deoxynucleotidyl transferase‐‐mediated nick end labeling assay. FSC, forward scatter. These data are as presented in Hodge et al. (([1999])).

In Vivo Studies

Studies were also conducted to determine if an antigen‐‐specific immune response could be enhanced using a TRICOM vector. A four‐‐gene vaccinia recombinant was constructed that contained the human CEA gene and the B7‐‐1, ICAM‐‐1 and LFA‐‐3 genes, designated rV‐‐CEA//TRICOM. Mice were vaccinated one time with 10⁷ pfu rV‐‐CEA ((Hodge et al., [1995])), rV‐‐CEA//B7‐‐1 ((Kalus et al., [1999])) or rV‐‐CEA//TRICOM, and spleens were harvested 22 days later. As seen in Figure 5, inset, splenic T cells of mice vaccinated with rV‐‐TRICOM showed higher levels of CEA‐‐specific stimulation compared with T cells obtained from mice vaccinated with rV‐‐CEA; ovalbumin and Con A were used as controls. An experiment was then conducted to determine if rV‐‐CEA//TRICOM could induce long‐‐term immunity. Mice ((five//group)) were vaccinated one time with V‐‐WT, rV‐‐CEA, or rV‐‐CEA//TRICOM. One hundred days later, mice were challenged with a high dose ((1 × 10⁶)) of MC38 colon carcinoma cells expresing CEA ((Hodge et al., [1995])). All mice receiving V‐‐WT and rV‐‐CEA succumbed to tumors, while all mice vaccinated with rV‐‐CEA//TRICOM were alive 50 days post‐‐challenge ((Figure 5)).

Figure 5. C57BL//6 mice ((five//group)) were administered HBSS ((closed squares)) or vaccinated with 10⁷ pfu rV‐‐CEA ((closed triangles)) or rV‐‐CEA//TRICOM ((closed circle)). One hundred days later, mice were inoculated with 1 × 10⁶ MC38 carcinoma cells expressing CEA, and survival was monitored. All mice other than the rV‐‐CEA//TRICOM group developed tumors and were sacrificed when tumors exceeded 20 mm in length or width, or when the mice were moribund. Insert: In a second experiment, C57BL//6 mice ((five//group)) were vaccinated with 10⁷ pfu rV‐‐CEA, rV‐‐CEA//B7‐‐1, rV‐‐CEA//TRICOM or HBSS buffer. Lymphoproliferative responses from pooled splenic T cells were analyzed 22 days after vaccination. Values represent the stimulation index of the mean cpm of triplicate samples versus media. Standard deviation never exceeded 10%%. Antigens used were Con A ((5 μg//ml)), CEA ((100 μg//ml)), and ovalbumin ((100 μg//ml)). These data are as presented in Hodge et al. (([1999])).

CEA‐‐transgenic mice ((Kass et al., [1999]; Thompson et al., [1991])) in which the human CEA gene is expressed in normal adult gastrointestinal tissue, and whose serum is CEA‐‐positive, were employed to determine if the rV‐‐CEA//TRICOM vector could enhance T‐‐cell responses to a self‐‐antigen. CEA‐‐transgenic mice were separated into five mice//group ((due to limited availability)). Two mice were vaccinated once with 10⁷ pfu rV‐‐CEA, rV‐‐CEA//B7‐‐1, rV‐‐CEA//TRICOM or buffer and were euthanized on day 30 to analyze CEA‐‐specific T‐‐cell responses. T‐‐cell responses obtained after vaccination with rV‐‐CEA//TRICOM were substantially greater than those obtained with rV‐‐CEA. Responses to ovalbumin and Con A were used as controls. The remaining three CEA‐‐transgenic mice in each group were used to determine if anti‐‐tumor responses to a CEA‐‐expressing tumor could be enhanced employing a TRICOM vector. These mice were first inoculated s.c. with 4 × 10⁵ MC38 carcinoma cells expressing the CEA gene ((Hodge et al., [1995])). Four days later, mice were vaccinated one time at a distal site with 10⁷ pfu viral recombinant or buffer. No tumors grew in mice vaccinated with rV‐‐CEA//TRICOM, whereas tumors continued to grow in mice vaccinated with buffer, rV‐‐CEA and rV‐‐CEA//B7‐‐1. Though preliminary, these results support the in vivo activity of TRICOM vectors.

Increased Expression of Costimulatory Molecules on DCs

To determine the efficiency of poxvirus infection of DCs, the cells were infected with either an rV virus encoding B7‐‐1, ICAM‐‐1 and LFA‐‐3, or an rF virus encoding B7‐‐1, ICAM‐‐1, LFA‐‐3 and CEA. In the latter case, CEA was used as a reporter gene because the majority of fowlpox structural proteins are under the control of late fowlpox promoters and, thus, are not expressed in infected cells. After 18 h, cells were analyzed for the expression of cell‐‐surface markers associated with the particular viral infection ((Figure 6A–D)). Uninfected control DCs expressed CD11b ((97%%–98%%)) and were equivalent to background levels ((1%%)) for the expression of CEA ((Figure 6A)) or vaccinia proteins ((Figure 6C)). After infection with rF‐‐CEA//TRICOM, 87%% of DCs co‐‐expressed both CD11b and CEA ((Figure 6B)). These DCs failed to express fowlpox proteins as detected by polyclonal rabbit anti‐‐fowlpox sera, which is in agreement with previous reports stating that fowlpox does not replicate in mammalian cells. Ninety‐‐four percent of DCs infected with rV‐‐TRICOM co‐‐expressed both CD11b and vaccinia proteins ((Figure 6D)). These data demonstrate that DCs are efficiently infected by both rF‐‐and rV‐‐TRICOM vectors.

Figure 6. Efficiency of poxviral infection of dendritic cells. Uninfected dendritic cells are depicted in panels A and C. Dendritic cells were infected with either a multiplicity of infection ((MOI)) of 50 plaque‐‐forming units per cell of rF‐‐CEA//TRICOM ((panel B)) or 25 MOI of rV‐‐TRICOM ((panel D)) for 5 h. After 18 h, cells were analyzed for surface co‐‐expression of CD11b and either carcinoembryonic antigen ((CEA)), a marker gene for rF‐‐CEA//TRICOM ((panels A and B)), or vaccinia proteins via polyclonal rabbit anti‐‐vaccinia serum ((panels C and D)). For panels A–D, the y‐‐axis represents the fluorescent intensity of cells expressing CD11b. For panels A and B, the x‐‐axis represents the fluorescent intensity of cells expressing CEA proteins. For panels C and D, the x‐‐axis represents the fluorescent intensity of cells expressing vaccinia proteins. The insert boxes in each panel denote the percent positive cells in each quadrant. Dendritic cells infected with TRICOM vectors exhibit enhanced capacity to stimulate naive T cells ((panels E and F)). All dendritic cell populations were co‐‐cultured for 48 h with T cells at a ratio of 10:1 in the presence of different concentrations of cancanavalin A ((Con A)) to provide signal 1. ³H‐‐thymidine was added during the final 18 h. Panel E: Uninfected dendritic cells ((open squares)), mock‐‐infected dendritic cells ((open diamonds)), or dendritic cells infected with wild‐‐type fowlpox ((FP‐‐WT)) ((open inverse triangles)), rF‐‐B7‐‐1 ((closed triangles)) or rF‐‐TRICOM ((closed circles)). Panel F: Dendritic cells ((open squares)), mock‐‐infected dendritic cells ((open diamonds)), or dendritic cells infected with wild‐‐type vaccinia ((V‐‐WT)) ((open inverse triangles)), rV‐‐B7‐‐1 ((closed triangles)) or rF‐‐TRICOM ((closed circles)). Error bars represent 95%% confidence intervals. In some cases, the error bars are obscured by the symbols. These data are as presented in Hodge et al. (([2000])).

The cardinal characteristics of DCs are high expression levels of both histocompatibility antigens and costimulatory molecules. To further characterize the phenotype of DCs prior to and after virus infection, cells were infected with FP‐‐WT, rF‐‐B7‐‐1, rF‐‐TRICOM, V‐‐WT, rV‐‐B7‐‐1, or rV‐‐TRICOM, and analyzed for the expression of 13 different cell‐‐surface markers ((Table 1)). As expected, uninfected and mock‐‐infected DCs expressed high levels of MHC Class I and II, CD11b, B7‐‐2 and CD40 molecules, as well as high levels of B7‐‐1, ICAM‐‐1, and LFA‐‐3. Dendritic cells infected with FP‐‐WT had a similar phenotypic profile to that of uninfected DCs, while DCs infected with rF‐‐B7‐‐1 expressed approximately 5‐‐fold more B7‐‐1 than uninfected DCs [[mean fluorescent intensity ((MFI)) from 329 to 1559]]. Infection of DCs with rF‐‐TRICOM resulted in approximately a 6‐‐fold increase of B7‐‐1, a 3‐‐fold increase in ICAM‐‐1, and a 4‐‐fold increase in LFA‐‐3 when compared with control DCs. Thus, infection with rF‐‐TRICOM increased the expression of the three costimulatory molecule transgenes and altered no other phenotypic marker of the DCs ((Table 1)). The control vector did not alter the expression levels of any of the 13 phenotypic markers. Dendritic cells infected with V‐‐WT expressed lower cell‐‐surface densities ((as determined by MFI)) of some molecules, while DCs infected with rV‐‐B7‐‐1 expressed 5‐‐fold more B7‐‐1 than uninfected DCs ((MFI from 329 to 1689)). Infection of DCs with rV‐‐TRICOM substantially increased MFI and the percentage of cells positive for B7‐‐1, ICAM‐‐1, and LFA‐‐3 ((Table 1)). All DC populations remained negative for T‐‐cell ((CD3)), B‐‐cell ((CD19)), monocyte//neutrophil ((CD14)), and natural killer ((NK))‐‐cell ((Pan NK)) markers both before and after infection with rF or rV vectors ((Table 1)). These experiments were repeated five times with similar results.

Table 1. Infection of dendritic cells ((DCs)) with rV‐‐TRICOM or rF‐‐TRICOM: increase in the expression level of B7‐‐1, ICAM‐‐1, and LFA‐‐3.^a

Infection	DC panel, %% positive cells ((MFI))									Non‐‐DC markers %% positive cells ((MFI))
	I‐‐A^b	H‐‐2K^b//D^b	CD11b	CD11c	B7‐‐2	CD40	B7‐‐1	ICAM‐‐1	LFA‐‐3	CD‐‐14	CD‐‐19	CD3	Pan NK
None	88 ((1124))	89 ((125))	93 ((935))	20 ((74))	68 ((490))	68 ((82))	91 ((329))	96 ((595))	88 ((153))	2 ((20))	2 ((26))	0.3 ((33))	2 ((56))
Mock	87 ((989))	88 ((125))	90 ((1129))	25 ((49))	77 ((432))	71 ((99))	85 ((330))	97 ((519))	86 ((189))	1 ((42))	4 ((30))	0.6 ((37))	2 ((29))
FP‐‐WT	91 ((985))	98 ((126))	89 ((889))	24 ((69))	65 ((487))	71 ((90))	94 ((382))	97 ((464))	91 ((130))	2 ((18))	2 ((10))	1 ((60))	2 ((31))
rF‐‐B7‐‐1	88 ((987))	99 ((114))	86 ((793))	28 ((72))	74 ((404))	68 ((90))	95 ((1559))	98 ((437))	90 ((180))	2 ((30))	2 ((29))	1 ((45))	2 ((54))
rF‐‐TRICOM	90 ((900))	99 ((115))	84 ((789))	27 ((66))	76 ((499))	73 ((93))	99 ((1824))	98 ((1697))	95 ((530))	3 ((22))	3 ((13))	1 ((50))	3 ((33))
V‐‐WT	87 ((890))	86 ((83))	86 ((588))	29 ((54))	61 ((274))	79 ((98))	90 ((197))	95 ((241))	70 ((196))	3 ((20))	3 ((45))	1 ((59))	5 ((50))
rV‐‐B7‐‐1	85 ((856))	85 ((104))	81 ((693))	25 ((62))	73 ((304))	73((103))	94 ((1689))	97 ((364))	72 ((131))	3 ((42))	3 ((40))	0.8 ((67))	4 ((37))
rV‐‐TRICOM	87 ((901))	78 ((103))	77 ((558))	22 ((54))	66 ((298))	71 ((81))	96 ((1442))	94 ((1528))	92 ((304))	3 ((32))	4 ((29))	0.9 ((70))	4 ((31))

^aDendritic cells ((DC)) were uninfected, mock infected, or infected with a multiplicity of infection ((MOI)) of 50 plaque‐‐forming units per cell of either fowlpox wild‐‐type ((FP‐‐WT)), recombinant fowlpox B7‐‐1 ((rF‐‐B7‐‐1)), or rF‐‐TRICOM or 25 MOI of either vaccinia wild‐‐type ((V‐‐WT)), recombinant vaccinia B7‐‐1 ((rV‐‐B7‐‐1)), or rV‐‐TRICOM for 5 h. After an 18‐‐h incubation, cells were phenotyped by three‐‐color flow cytometry analysis. Bold numbers indicate a > 100%% change in cell‐‐surface expression ((mean fluorescent intensity; MFI)). Other abbreviations used: TRICOM, triad of costimulatory molecules ((B7‐‐1, ICAM‐‐1 [[intercellular adhesion molecule‐‐1]], and LFA‐‐3 [[i.e., leukocyte function‐‐associated antigen‐‐3]])); I‐‐A^b == major histocompatibility complex class II allele; H‐‐2K^b//D^b == major histocompatibility complex class I allelles; NK == natural killer cell.

These data are as presented in Hodge et al. (([2000])).

Dendritic Cells Infected with TRICOM Vectors Exhibit Enhanced Capacity to Stimulate Naïve T Cells

An in vitro model was used to analyze how increased levels of B7‐‐1, ICAM‐‐1, and LFA‐‐3 expression affect the induction of naïve T‐‐cell stimulation. In this model, the first signal for T‐‐cell activation was delivered via a pharmacological reagent ((Con A)) and the additional, or costimulatory, signal((s)) were delivered to the T cell via DCs, or DCs expressing higher levels of the three costimulatory molecules as a consequence of recombinant TRICOM vector infection. In these and all subsequent studies reported here, FP‐‐WT and V‐‐WT were also used to rule out effects due to the vector alone. As shown in Figure 6E and F, both uninfected and mock‐‐infected DCs induced proliferation of T cells, which was dependent upon the concentration of signal 1 ((Con A)). Dendritic cells infected with FP‐‐WT ((designated DC//FP‐‐WT)) induced T‐‐cell proliferation similar to that of uninfected DCs. Higher levels of B7‐‐1 expression in DCs infected with rF‐‐B7‐‐1 ((designated DC//rF‐‐B7‐‐1)) significantly increased proliferation of T cells compared with uninfected DCs ((p == 0.002 at highest dose of Con A)). Moreover, DCs infected with rF‐‐TRICOM ((designated DC//rF‐‐TRICOM)) induced even further increases in T‐‐cell proliferation at all concentrations of Con A ((p < 0.001)). In addition, when T cells were stimulated with DC//rF‐‐TRICOM, 17‐‐fold less signal 1 ((Con A)) was needed to induce proliferation to levels comparable to that of uninfected DCs. These experiments were then repeated using wild‐‐type and recombinant vaccinia vectors. Dendritic cells infected with rV‐‐TRICOM ((designated DC//rV‐‐TRICOM)) induced increases in T‐‐cell proliferation at all Con A concentrations compared with uninfected control DCs, or DCs infected with rV‐‐B7‐‐1 or V‐‐WT. As with rF‐‐TRICOM, when T cells were stimulated with DC//rF‐‐TRICOM, 23‐‐fold less Con A was needed to induce proliferation to levels comparable to that of uninfected DCs ((Figure 6F)). It should be noted that DCs infected with V‐‐WT actually showed a decrease in their ability to act as APCs compared with uninfected DCs. This is most probably due to the slight decrease in the expression levels of MHC Class I and II as a result of V‐‐WT infection ((Table 1)). However, as shown in Table 1 and Figure 6F, the increases in expression levels of B7‐‐1, ICAM‐‐1 and LFA‐‐3 as a result of rV‐‐TRICOM infection more that compensated for this vaccinia effect and made these DCs more potent APCs, compared with control uninfected DCs, at all concentrations of signal 1. These experiments were repeated five times with similar results.

Enhanced Allostimulatory Activity by DCs Infected with TRICOM Vectors

The effects of rF‐‐TRICOM or rV‐‐TRICOM infection on DC stimulatory capacity was assessed by allospecific mixed lymphocyte reaction ((MLR)). Both uninfected DCs and mock‐‐infected DC populations ((H‐‐2^b, C57BL//6)) induced a strong proliferation ((78,000 CPM)) of naive allogeneic ((H‐‐2^d, Balb//c)) T cells. The stimulatory capacity of DCs was increased slightly after infection with rF‐‐B7‐‐1. Infection of DCs with rF‐‐TRICOM significantly increased the stimulatory capacity over DCs and DC//rF‐‐B7‐‐1 at all dendritic cell to responder T cell ratios ((p < 0.001)). Importantly, DC populations infected with either FP‐‐WT or rF‐‐TRICOM vectors failed to stimulate syngeneic T cells. When these experiments were repeated using vaccinia vectors, DC//rV‐‐TRICOM induced marked increases in allogeneic T‐‐cell proliferation when compared with uninfected DCs and DCs infected with rV‐‐B7‐‐1 ((designated DC//rV‐‐B7‐‐1)), at all dendritic cell to responder T cell ratios ((p < 0.001)). DC//rV‐‐TRICOM, DC//rV‐‐B7‐‐1, or DCs infected with V‐‐WT ((designated DC//V‐‐WT)), however, failed to stimulate syngeneic ((H‐‐2^b, C57BL//6)) T cells. These experiments were repeated four times with similar results.

Presentation of Peptides to Effector T Cells

Studies were undertaken to determine if the T‐‐cell stimulatory capacity of peptide‐‐pulsed DCs could be enhanced by infecting DCs with rF‐‐B7‐‐1 and rF‐‐TRICOM. To that end, the H‐‐2K^b‐‐restricted OVA peptide and an OVA‐‐specific CD8⁺⁺ effector T‐‐cell line were used. Dendritic cells were exposed to different concentrations of OVA peptide in vitro and incubated in the presence of the OVA T‐‐cell line ((Figure 7)). The conventional ((i.e., uninfected)) DCs induced a strong proliferation of OVA‐‐specific T cells when incubated with the OVA peptide ((Figure 7A, open squares)). These DCs did not induce proliferation of OVA‐‐specific T cells when incubated with the control peptide VSVN ((Figure 7A, open diamonds)). DC//rF‐‐B7‐‐1 increased the overall peptide‐‐specific proliferation of these cells 2‐‐fold at 1 μm peptide. In addition, DC//rF‐‐B7‐‐1 induced similar proliferation to that of uninfected or mock‐‐infected DCs in the presence of 4‐‐fold less peptide ((0.25 μm)). DC//rF‐‐TRICOM, however, markedly and significantly ((p < 0.001)) increased the overall proliferation of these T cells and, in the presence of 13‐‐fold less OVA peptide ((0.077 μm; Figure 7A, closed circles)), induced proliferation comparable to that of uninfected DCs ((Figure 7A)). To further evaluate the capacity of fowlpox‐‐infected DCs to present peptide, DCs were pulsed with a single concentration of OVA peptide ((1 μm)) and incubated in the presence of several ratios of T cells ((Figure 7B)). On a per‐‐cell basis, 2‐‐fold less DC//rF‐‐B7‐‐1 was required to induce proliferation levels comparable to that of DCs ((Figure 7B, closed triangles versus open squares)). The greatest stimulatory effect was that of DC//rF‐‐TRICOM, which induced T‐‐cell proliferation levels comparable to those of uninfected DCs with 8‐‐fold fewer cells ((i.e., T cell to DC ratios of 10:1 for uninfected DCs versus 80:1 for DC//rF‐‐TRICOM)) ((Figure 7B, closed circles versus open squares)). Experiments were also carried out using DCs infected with vaccinia vectors ((Figure 7C and D)). As before, these DCs did not induce proliferation of OVA‐‐specific T cells when incubated with the comtrol peptide VSVN ((Figure 7C, open diamonds)). DC//rV‐‐B7‐‐1 increased the overall peptide‐‐specific proliferation of these cells 2‐‐fold ((Figure 7C)) and induced similar proliferation to that of uninfected or mock‐‐infected DCs in the presence of 2.5‐‐fold less peptide ((1 μm versus 0.4 μm)). However, DC//rV‐‐TRICOM significantly ((p < 0.001)) increased the overall proliferation of these T cells, as compared with uninfected DCs, and induced proliferation comparable to that of uninfected DCs in the presence of 33‐‐fold less OVA peptide ((0.03 μm versus 1 μm)) ((Figure 7C)). When these DC populations were pulsed with a single concentration of OVA peptide and incubated in the presence of several ratios of T cells ((Figure 7D)), 4‐‐fold fewer DC//rF‐‐B7‐‐1 were required to induce proliferation levels comparable to that of DCs ((closed triangles versus open squares)). Among these vectors, the greatest stimulatory effect was that of DC//rV‐‐TRICOM, which induced proliferation levels comparable to that of DCs with 32‐‐fold fewer cells ((10:1 versus 320:1)) ((closed circles versus open squares)). These experiments were repeated five additional times with the same results.

Figure 7. Effect of TRICOM vectors on the stimulation of effector T cells. Panels A, B: Effect of fowlpox virus infection of dendritic cells on OVA ((i.e., ovalbumin_257–264)) peptide‐‐specific T‐‐cell proliferation. Uninfected dendritic cells ((open squares)) or dendritic cells infected with wild‐‐type fowlpox ((FP‐‐WT)) ((open inverted triangles)), rF‐‐B7‐‐1 ((closed triangles)), or rF‐‐TRICOM ((closed circles)) were co‐‐cultured with OVA peptide‐‐specific T cells. Experimental conditions included a fixed effector to stimulator cell ratio of 10:1 in the presence of various concentrations of OVA peptide or negative control peptide VSVN ((i.e., vesicular stomatitis virus N_52–59)) ((open diamonds)) ((panel A)) or a fixed peptide concentration of 1 μM in the presence of various effector to stimulator cell ratios ((panel B)). Panels C and D: Effect of vaccinia virus infection of dendritic cells on peptide‐‐specific T‐‐cell proliferation. Uninfected dendritic cells ((open squares)) or dendritic cells infected with wild‐‐type vaccinia ((V‐‐WT, open inverted triangles)), rV‐‐B7‐‐1 ((closed triangles)), or rV‐‐TRICOM ((closed circles)) were co‐‐cultured with OVA peptide‐‐specific T cells. Experimental conditions included a fixed effector to stimulator cell ratio of 10:1 in the presence of various concentrations of OVA peptide or negative control peptide VSVN ((open diamonds)) ((panel C)), or a fixed peptide concentration of 1 μM in the presence of various effector to stimulator cell ratios ((open diamonds)) ((panel D)). Error bars represent 95%% confidence intervals. In some cases, the error bars are obscured by the symbols. These data are as presented in Hodge et al. (([2000])).

Cytokine and Apoptosis Studies

Cytokine expression in OVA‐‐specific CD8⁺⁺ effector T cells stimulated for 48 h with OVA peptide‐‐pulsed DCs or OVA peptide‐‐pulsed DCs infected with rF‐‐B7‐‐1 or rF‐‐TRICOM was analyzed at the RNA level for expression of IL‐‐2, IFN‐‐γ, IL‐‐10, and IL‐‐4 message utilizing multiplex polymerase chain reaction ((MP‐‐PCR)). Incubation of OVA‐‐specific CD8⁺⁺ T cells with OVA peptide‐‐pulsed DCs or peptide‐‐pulsed DCs infected with FP‐‐WT induced IL‐‐2 message, while pulsed DC//rF‐‐B7‐‐1 induced 32%% more IL‐‐2 message than DCs. However, T cells stimulated with peptide‐‐pulsed DC//rF‐‐TRICOM induced 60%% more IL‐‐2 message than conventional pulsed DCs. Incubation of OVA‐‐specific CD8⁺⁺ T cells with either DCs alone or control peptide VSVN‐‐pulsed DCs induced no detectable IL‐‐2 message ((not shown)). When IFN‐‐γ production by stimulated T cells was examined, IFN‐‐γ message was noted from T cells stimulated with peptide‐‐pulsed DCs or peptide‐‐pulsed DC//FP‐‐WT, while pulsed DC//rF‐‐B7‐‐1 and pulsed DC//rF‐‐TRICOM induced 18%% and 66%% more IFN‐‐γ, respectively, than peptide‐‐pulsed DCs alone. Stimulation of T cells with DCs infected with any viral vector failed to induce detectable levels of IL‐‐4 or IL‐‐10 ((not shown)). These findings were also noted by cytokine ELISA analyzing supernatant fluids from naïve CD4⁺⁺ and CD8⁺⁺ T cells stimulated with the dendritic cell populations noted above for 48 h in the presence of Con A as signal 1. T‐‐cell activation can result in either cell proliferation or cell death. It has been reported that CD28//B7‐‐1 costimulation enhances the survival of activated T cells via enhanced production of IL‐‐2. To determine if stimulation of OVA‐‐specific T cells with peptide‐‐pulsed DCs infected with rF‐‐B7‐‐1 or rF‐‐TRICOM would lead to increased cell survival or programmed cell death, the mRNA levels for selected apoptosis‐‐related proteins were analyzed after 48 h. The proteins bcl‐‐2 and bcl‐‐xL, which have been reported to suppress apoptosis, and Bax and bcl‐‐xS, which reportedly increase cell susceptibility to apoptosis, were examined. T cells activated by peptide‐‐pulsed DCs expressed moderate quantities of bcl‐‐2 message, while T cells stimulated with peptide‐‐pulsed DCs infected with rF‐‐B7‐‐1 or rF‐‐TRICOM expressed 31%% and 75%% higher levels of bcl‐‐2 message, respectively, than those stimulated with peptide‐‐pulsed control DCs. This observation is in agreement with that of Boussiotis et al. (([1997])), who observed that bcl‐‐2 was induced following CD28 costimulation. The expression level of bcl‐‐2 in T cells was decreased by 65%% following stimulation with DC//FP‐‐WT. This was most likely due to the interaction of the T cell with a DC expressing decreased costimulatory molecules due to wild‐‐type virus infection > 60 h prior to message analysis. T cells stimulated with peptide‐‐pulsed DCs infected with any construct failed to modulate expression levels of Bax or of the caspase family LICE. Although it has been reported that CD28 costimulation augmented the expression of bcl‐‐xL in activated T cells ((Boise et al., [1995])), stimulation of these OVA‐‐specific CD8⁺⁺ T cells with peptide‐‐pulsed DCs infected with any construct failed to induce detectable levels of bcl‐‐xL or bcl‐‐xS. These findings were confirmed using TUNEL analysis of naïve CD4⁺⁺ and CD8⁺⁺ T cells stimulated with the dendritic populations noted above for 48 h in the presence of Con A as signal 1.

Effect of rV‐‐TRICOM Infection on DCs Matured with TNF‐‐α, LPS, or CD40

The functional maturation of DCs has previously been shown to correlate with the upregulation of certain T‐‐cell costimulatory molecules. Bone marrow‐‐derived DCs generated in the presence of GM‐‐CSF and IL‐‐4 exhibit an intermediate maturation stage with respect to phenotype and in vitro antigen presenting capacity ((Labeur et al., [1999])). Experiments were conducted to examine the effect of rF‐‐TRICOM or rV‐‐TRICOM infection on DCs that had been further “matured” by co‐‐culture with TNF‐‐α, LPS, or CD40‐‐specific mAb. Treatment of DCs with TNF‐‐α or CD40‐‐specific mAb during the final 24 h of culture resulted in some upregulation of MHC‐‐II, B7‐‐2, and ICAM‐‐1 as determined by flow cytometric analysis; the level of upregulation of these markers is in agreement with previous observations ((Labeur et al., [1999]; Rescigno et al., [1997])). Treatment with LPS resulted in upregulation of MHC‐‐II, B7‐‐2, ICAM‐‐1, and CD40; this too is in agreement with previously reported observations ((Albert et al., [1998]; Labeur et al., [1999]; Rescigno et al., [1997])). None of the other phenotypic markers were altered by TNF‐‐α, LPS, or CD40‐‐specific mAb. Functionally, treatment of DCs with TNF‐‐α, LPS, or CD40‐‐specific mAb culminated in an 18%%–20%% increase in peptide‐‐specific proliferation over that of unmanipulated DCs ((Figure 8A and C)) at the highest level of peptide. These maturation‐‐mediated increases in stimulatory capacity are also similar to those reported by other investigators ((Labeur et al., [1999]; Morse et al., [1998])). Infection of DCs of intermediate maturity ((grown with GM‐‐CSF and IL‐‐4)) with rF‐‐TRICOM ((Figure 8B)) or rV‐‐TRICOM ((Figure 8D)) resulted in their ability to induce a substantial increase in T‐‐cell proliferation. The use of more mature DCs ((matured using TNF‐‐α, LPS, or CD40‐‐specific mAb)) conferred only a slight increase in stimulatory capacity ((Figure 8A and C)). These more mature DCs, however, were markedly enhanced in their T‐‐cell stimulatory capacity when infected with rF‐‐TRICOM ((Figure 8B)) or rV‐‐TRICOM ((Figure 8D)). These experiments were repeated four additional times with similar results.

Figure 8. Enhancing effect of rF‐‐TRICOM or rV‐‐TRICOM infection on the T‐‐cell stimulatory capacity of dendritic cells of intermediate maturity ((granulocyte‐‐macrophage colony‐‐stimulating factor and interleukin 4, and noted as dendritic cells)) and further matured dendritic cells using tumor necrosis factor ((TNF))‐‐α, lipopolysaccharide ((LPS)), or CD40‐‐specific monoclonal antibody ((mAb)). Dendritic cells ((closed squares)) or dendritic cells matured with either 100 ng//mL TNF‐‐α ((open triangles)), 0.1 μg//mL LPS ((open diamonds)), or 5 μg//mL CD40‐‐specific mAb ((open circles)) for the final 24 h of culture were used to stimulate OVA ((ovalbumin_257–264))‐‐specific effector T cells in the presence of several concentrations of OVA peptide ((panel A)). Aliquots of all the above dendritic cell populations were then infected with a multiplicity of infection ((MOI)) of 50 plaque‐‐forming units per cell of rF‐‐TRICOM and used to stimulate OVA‐‐specific T cells under similar conditions ((panel B)). In a separate set of experiments, dendritic cells cultured with TNF‐‐α ((open triangles)), LPS ((open diamonds)), or CD40‐‐specific mAb ((open circles)) as above were used to stimulate OVA‐‐specific effector T cells in the presence of several concentrations of OVA peptide ((panel C)). Aliquots of all dendritic cell populations were then infected with 25 MOI of rV‐‐TRICOM and used to stimulate OVA‐‐specific T cells under similar conditions ((panel D)). For all panels, the T‐‐cell to dendritic cell ratio was 10:1, and the OVA peptide concentration was 1 μg//mL. Closed circles denote proliferation of OVA T cells stimulated with all dendritic cell populations in the presence of 1 μg//mL VSVN ((i.e., vesicular stomatitis virus N_52–59)) peptide. Error bars represent 95%% confidence intervals. In some cases, the error bars are obscured by the symbols. These data are as presented in Hodge et al. (([2000])).

Dendritic Cells Infected with rF‐‐TRICOM or rV‐‐TRICOM Are More Efficient at Priming CTL Responses In Vivo

Experiments were conducted to determine if the enhanced T‐‐cell stimulatory capacity of DCs infected with TRICOM vectors noted in vitro using Con A ((Figure 6, E‐‐F)), MLRs, and peptide‐‐specific effector T‐‐cells ((Figure 7)) would translate to enhanced efficacy in priming naïve T‐‐cell responses in vivo. To that end, DCs, DC//FP‐‐WT, DC//rF‐‐B7‐‐1, and rF‐‐TRICOM were pulsed with 10 μM OVA peptide and administered intravenously to C57BL//6 mice. Mice were also vaccinated with OVA peptide in Ribi//Detox adjuvant subcutaneously. Splenocytes were harvested 14 days following vaccination, restimulated in vitro for 6 days with irradiated splenocytes as APCs and 10 μg//mL OVA peptide, and assessed for their peptide‐‐specific lytic ability against OVA‐‐pulsed EL‐‐4 cells. EL‐‐4 cells pulsed with VSVN peptide were used as control target cells. As seen in Figure 9A, cytotoxic T lymphocyte ((CTL)) generated from mice vaccinated with peptide//adjuvant exhibited modest levels of CTL activity. Mice vaccinated with peptide‐‐pulsed, uninfected DCs exhibited a greater peptide‐‐specific CTL response ((Figure 9B)) than peptide in adjuvant; infection of peptide‐‐pulsed DCs with FP‐‐WT vector yielded results similar to those seen with peptide‐‐pulsed, uninfected DCs ((Figure 9C versus 9B)). Mice vaccinated with peptide‐‐pulsed DC//rF‐‐B7‐‐1 ((Figure 9D)) exhibited a CTL response that was significantly stronger than that of peptide‐‐pulsed DCs ((Figure 9B)) ((LU == 16.1 and 3.2, respectively; p == 0.001)). Mice vaccinated with peptide‐‐pulsed DCs that had been infected with rF‐‐TRICOM exhibited T cells with even more potent lytic capacity ((> 20 LU)) than peptide‐‐pulsed DCs ((p == 0.001)) or peptide‐‐pulsed DCs infected with rF‐‐B7‐‐1 ((p < 0.01)). Again, there was no in vivo effect of the vector alone. Similar experiments were then conducted using the recombinant vaccinia vectors ((Figure 9, F–J)). The induced CTL response was somewhat blunted in mice vaccinated with peptide‐‐pulsed DC//V‐‐WT ((Figure 9H)) versus peptide‐‐pulsed DCs ((Figure 9G)). In contrast, mice vaccinated with peptide‐‐pulsed DC//rV‐‐TRICOM ((Figure 9J)) exhibited a CTL response that was significantly stronger than that of uninfected, or vector control‐‐infected, peptide‐‐pulsed DCs ((LU == 14.3, p == 0.001)).

Figure 9. Effect of poxvirus infection of dendritic cells on induction of cytotoxic T lymphocyte ((CTL)) activity ((panels A–J)). Groups of mice were vaccinated subcutaneously with 100 μg peptide in Ribi//Detox adjuvant ((peptide//adjuvant)) ((panels A and F)) for comparison with dendritic cell vaccination. Uninfected dendritic cells ((panels B and G)), dendritic cells infected with wild‐‐type fowlpox ((FP‐‐WT)) or wild‐‐type vaccinia ((V‐‐WT)) ((panels C and H)), dendritic cells infected with recombinant fowlpox or recombinant vaccinia expressing B7‐‐1 ((DC//B7‐‐1)) ((panels D and I)), dendritic cells infected with recombinant fowlpox or recombinant vaccinia expressing TRICOM ((DC//TRICOM)) ((panels E and J)) were pulsed with 10 μM OVA ((i.e., ovalbumin _257–264)) peptide for 2 h and administered intravenously to mice ((1 × 10⁵ cells//mouse)). Fourteen days later, the spleens were harvested, after which spleen cell suspensions were restimulated for 6 days with the corresponding peptide and assessed for lytic ability against EL‐‐4 cells pulsed with either OVA ((closed squares)) or VSVN ((i.e., vesicular stomatitis virus N_52–59)) peptide ((open squares)). Numbers in panels depict CTL activity as expressed in lytic units ((LU)), as calculated using the following formula: LU_18%% == [[((1 × 10⁶))//((5,000 × number of effector cells to reach 18%% lysis))]]. Error bars represent 95%% confidence intervals. In some cases, the error bars are obscured by the symbols. These data are as presented in Hodge et al. (([2000])).

Discussion

When a naive T cell encounters antigen, several distinct outcomes are possible including proliferation, cytokine secretion, and differentiation into effector cells, inactivation, death, and unresponsiveness ((anergy)). The predominant outcome under physiologic conditions may be determined by whether appropriate costimulatory signals are delivered to the responding T cell ((Dubey et al., [1995])). At least three distinct molecules normally found on the surface of professional APC ((B7‐‐1, ICAM‐‐1 and LFA‐‐3)) have been shown to be capable of providing the signals critical for T‐‐cell activation. Here, the role of costimulatory molecules in naive T‐‐cell activation was examined by utilizing vectors engineered to express B7‐‐1, ICAM‐‐1, LFA‐‐3, or a combination of all three molecules ((designated TRICOM)).

Several groups have investigated the cooperation of two of these molecules in T‐‐cell costimulation. Dubey et al. (([1995])) have reported that costimulation by both B7‐‐1 and ICAM‐‐1 is a prerequisite for naive T‐‐cell activation, while Cavallo et al. (([1995])) determined that B7‐‐1 and ICAM‐‐1 must by coexpressed by tumor cells to established an anti‐‐tumor memory response. In addition, costimulation by B7‐‐1 and LFA‐‐3 has been shown to act additively both upon T‐‐cell proliferation and cytokine production. These previous studies were carried out using two costimulatory molecules and retroviral vectors. One gene was transduced into the target cell line and drug selected. These cells were then transduced again with a second recombinant retroviral construct followed by selection with a different agent. This process often requires weeks or months. Utilizing recombinant poxvirus vectors, one is able to achieve the coexpression of three costimulatory molecules 5 h post‐‐infection. In vitro MC38 cells infected with either recombinant vaccinia or avipox TRICOM vectors were shown to enhance proliferation of T cells to a much greater extent than MC38 cells infected with vectors containing the gene for any single costimulatory molecule ((Figure 2)). In addition, the relative strength of the second signal delivered to the T cell by the combination of costimulatory molecules appeared to be far greater than that delivered by MC38 cells expressing any single costimulatory molecule. Dubey et al. (([1995])) have demonstrated that at low stimulator to T cell ratios, synergy was noted with B7‐‐1 and ICAM‐‐1. Our studies confirm these findings. However, at extremely low stimulator to T cell ratios or weak signal 1 [[0.625 μg//ml Con A ((Figure 2C and D))]], the two‐‐gene construct ((rV‐‐B7‐‐1//ICAM‐‐1)) had little, if any, effect on proliferation. In contrast, stimulation via the TRICOM construct had a substantial and statistically significant effect on proliferation. The predominant effect of stimulation via the TRICOM construct was IL‐‐2 production from CD4⁺⁺ cells and IFN‐‐γ production from CD8⁺⁺ T cells, while few, if any, type 2 cytokines were produced. Cytokine expression analysis by RNase protection provided a profile compatible with the in vitro cytokine assay, manifested by substantially higher expression of IL‐‐2 and IFN‐‐γ in both CD4⁺⁺ and CD8⁺⁺ T cells stimulated with all three costimulatory molecules, compared with stimulation by any single costimulatory molecule ((Figure 3)). These data are in accordance with previous studies, which demonstrated that in the context of low CD28 costimulation, T cells produced low levels of IL‐‐1, whereas strong CD28 costimulation supported production of IL‐‐2, IFN‐‐γ, and IL‐‐13 ((Delespesse et al., [1998])). Taken together, the studies reported here indicate that optimal naive T‐‐cell responses require a higher level of costimulation than was previously thought, and that this could be provided by the combined action of three costimulatory molecules.

One question that is immediately raised is the potential of overstimulated T cells to undergo programmed cell death ((PCD)). The results shown in Figure 4A and B, clearly demonstrate apoptosis in T cells stimulated with MC38 cells in the presence of Con A with or without V‐‐WT infection ((i.e., in the absence of signal 2)). While Con A plus MC38//TRICOM clearly stimulated CD8⁺⁺ cells to far greater levels than Con A plus MC38//B7‐‐1 ((Figure 2)) and resulted in the production of higher levels of IFN‐‐γ and IL‐‐2 ((Figure 3)), this did not result in any greater degree of apoptosis ((Figure 4D)). Our results are in agreement with those of previous studies, which found that costimulation through the CD28 receptor appears to play an important role in enhancing the resistance of activated T cells to undergo PCD in culture ((Boise et al., [1995])). This could be attributed to augmentation of cytokine production by these cells and potential upregulation of survival genes. Further studies are currently under way to analyze the detailed mechanism of survival in these cells.

Perhaps the most studied T‐‐cell costimulatory molecule is B7‐‐1. This molecule's ability to enhance T‐‐cell activation using retroviral vectors, anti‐‐CTLA‐‐4 antibodies and poxvirus vectors is well established. The studies reported here rank the order of T‐‐cell stimulation by a single costimulatory molecule as B7‐‐1 > ICAM‐‐1 > LFA‐‐3. However, the employment of three costimulatory molecules was far superior to B7‐‐1 alone in both proliferation and cytokine production for both CD4⁺⁺ and CD8⁺⁺ T cells. The studies reported here demonstrate the power of the multicostimulatory molecule effect to enhance T‐‐cell proliferation in three very different systems and to employ both naïve and effector T‐‐cell populations as responder cells. These three model systems are: 1)) the activation of naïve T cells using Con A as signal 1 ((Figure 2)); 2)) the use of OVA peptide as signal 1 and the activation of OVA‐‐specific “effector” T cells from an established cell line; and 3)) enhanced allospecific reactivity in a mixed lymphocyte reaction.

Initial in vivo experiments reported here indicated that rV‐‐CEA//TRICOM is more efficient than rV‐‐CEA//B7‐‐1 in the induction of CEA‐‐specific T‐‐cell responses in both intact C57BL//6 mice and in CEA‐‐transgenic C57BL//6 mice ((Figure 5)). Induction of long‐‐term immunity to tumor challenge is shown for rV‐‐CEA//TRICOM in C57BL//6 mice, and anti‐‐tumor activity for rV‐‐CEA//TRICOM vs. rV‐‐CEA//B7‐‐1 or rV‐‐CEA is also shown in CEA transgenic mice. However, only limited amounts of the CEA transgenic mice were available and employed in these studies. Additional comprehensive studies with more CEA transgenic mice and different doses and dose schedules of rV‐‐CEA//TRICOM, rV‐‐CEA//B7‐‐1 and rV‐‐CEA, along with different tumor burdens, are clearly indicated. These mice will also have to be carefully evaluated for evidence of autoimmunity. Due to the scarcity of CEA‐‐transgenic mice, these will be long‐‐term experiments.

There are several possible mechanisms for the observed synergy here between B7‐‐1, ICAM‐‐1 and LFA‐‐3. The ICAM‐‐1//LFA‐‐1 interaction reportedly costimulates the TCR‐‐mediated activation of T cells by sustaining the increase in the same intracellular second messengers as generated by TCR engagement. The ICAM‐‐1//LFA‐‐1 interaction is necessary to upregulate expression of the IL‐‐2R‐‐α chain and CD28 on T cells, which is required to render them competent to respond to IL‐‐2 and B7‐‐1 costimulation. The B7‐‐1//CD28 interaction delivers a TCR‐‐independent costimulatory signal that increases both transcriptionally and post‐‐transcriptionally the expression of IL‐‐2 and other immunoregulatory lymphokines. The LFA‐‐3//CD2 interaction induces tyrosine phosphorylation of several intracellular second messengers, Ca^{2 ++} mobilization, and cAMP production, resulting in elaboration of a variety of cytokines, notably IL‐‐2 and IFN‐‐γ ((Wingren et al., [1995])). Thus, it appears that the three costimulatory molecules could be cooperating by enhancing the antigen‐‐dependent activation of T cells, as well as their production of and response to autocrine and paracrine growth factors. Future studies should involve an analysis of the signal transduction pathways induced by T‐‐cell costimulation via stimulator cell populations expressing B7‐‐1, ICAM‐‐1, and LFA‐‐3, alone or in combination.

These studies demonstrate the ability of vectors to introduce three costimulatory molecules into a cell, and to rapidly and efficiently activate both CD4⁺⁺ and CD8⁺⁺ T‐‐cell populations to levels far greater than those achieved when any one or two of these costimulatory molecules are used. This new threshold of T‐‐cell activation, with the caveat of autoimmunity mentioned above, has broad implications in vaccine design and development for a range of diseases.

The role of the DC in the induction of antigen‐‐specific T‐‐cell responses has focused the attention of many investigators on the potential efficacy of these cells in the immunotherapy of cancer and infectious agents. A potential limitation of the in vivo use of dendritic cells in vaccine strategies has been attributed to both the quality of the dendritic cells and the availability of adequate numbers of dendritic cells for dose escalation. Because of these limitations, methods for improving the efficacy of DCs to present immunogens are being explored. Current strategies to improve dendritic cell effectiveness include: 1)) modulation of the cytokine milieu ((GM‐‐CSF//IL‐‐4)); 2)) maturation of DCs via CD40L, TNF‐‐α or LPS; and 3)) expansion of DC populations in vivo by the administration of Flt3L.

The poxviral vectors examined in these studies were shown to infect DCs with high efficiency ((Figure 6)). Phenotypically, substantial differences were noted in the levels of B7‐‐1, ICAM‐‐1 and LFA‐‐3 expression between DCs, DC//rF‐‐B7‐‐1, and DC//rF‐‐TRICOM or DC//rV‐‐TRICOM ((Table 1)). The expression levels of certain surface markers ((CD11b, B7‐‐2)) decreased after infection with any of the vaccinia constructs and could be attributed to virus‐‐mediated downregulation of host‐‐cell protein‐‐synthesis machinery. The decrease in these markers was not noted following infection with fowlpox vectors. Infection of DCs with rF‐‐B7‐‐1 induced a substantial increase in the surface expression of B7‐‐1, while infection with rF‐‐TRICOM substantially increased B7‐‐1, ICAM‐‐1, and LFA‐‐3 expression. Similar results were noted after infection of DCs with vaccinia vectors. When these DC populations were assessed functionally by measuring either stimulation of naïve T cells in the presence of Con A as signal 1 ((Figure 6E and F)), stimulation in a primary allogeneic culture ((MLR)), or stimulation of a peptide‐‐specific effector T‐‐cell line ((Figure 7)), stimulation with DC//rF‐‐B7‐‐1 or DC//rV‐‐B7‐‐1 was superior to that of DCs alone or DCs infected with control vectors, while DC//rF‐‐TRICOM and DC//rV‐‐TRICOM clearly induced proliferation superior to that of DC//rF‐‐B7‐‐1, DC//rV‐‐B7‐‐1, or conventional uninfected DCs. On a per‐‐cell basis, DC//rV‐‐TRICOM induced levels of proliferation comparable to that of DCs with 12.5‐‐ to 32‐‐fold fewer cells ((Figure 7C and D)). These data taken together suggest that a much lower dose of DC//rF‐‐TRICOM or DC//rV‐‐TRICOM is required to stimulate T cells, which is an important consideration for clinical applications.

Infection with TRICOM constructs led to increased levels of IL‐‐2 and IFN‐‐γ production from CD8⁺⁺ effector T cells, while no detectable levels of type 2 cytokines ((IL‐‐10 and IL‐‐4)) were produced. The question again raised concerns about the potential of overstimulated T cells resulting in apoptosis. The expression of bcl‐‐2 has been shown to block apoptosis in many experimental systems ((Reed, [1994])). The results demonstrate expression of bcl‐‐2 in T cells stimulated with DCs. However, the expression level of bcl‐‐2 in T cells was increased 31%% and 76%% following stimulation with DC//rF‐‐B7‐‐1 or DC//rF‐‐TRICOM, respectively. These results are in agreement with those of previous studies, which found that costimulation through the CD28 receptor appears to play an important role in enhancing the resistance of activated T cells to undergo apoptosis in culture. This could be attributed to augmentation of cytokine production by these cells and potential upregulation of survival genes. There were no substantial changes in any T‐‐cell group in the expression levels of Bax, which has been reported to increase cell susceptibility to apoptosis, or LICE, a member of the caspase gene family responsible for the cascade of catalytic steps toward apoptosis.

The maturation stage of murine DCs is usually defined by phenotypic profile and, at times, by functional characteristics; however, there is considerable variability in these definitions in the literature ((Labeur et al., [1999]; Rescigno et al., [1997]; Winzler et al., [1997])). We have described the DCs used in these studies ((cultured with GM‐‐CSF and IL‐‐4)) as DCs of intermediate maturity; however, based on the phenotypic definitions by some investigators, these DCs would be termed mature. The addition of TNF‐‐α, LPS, or CD40‐‐specific mAb to the DCs used in these studies for “maturation” resulted in slight increases in MHC‐‐II, B7‐‐2, and ICAM‐‐1 molecules on the cell surface, and a modest increase in dendritic cell stimulatory capacity ((Figure 8)). These findings confirm those of Morse et al. (([1998])), who reported that CD40L treatment of DCs resulted in a 17%% improvement in alloreactive proliferation over that of non‐‐treated DCs. In addition, Labeur et al. (([1999])), using a T‐‐cell line specific for OVA_323–339, showed that CD40 or LPS treatment of DCs resulted in an 8.5%% enhancement in T‐‐cell proliferation, while TNF‐‐α treatment of DCs resulted in no enhancement. Treatment of DCs with TRICOM vectors, however, resulted in a DC with far greater stimulatory capacity than that treated with LPS, CD40‐‐specific mAb or TNF‐‐α ((Figure 8)). Dendritic cells pretreated with TNF‐‐α, LPS, or CD40 displayed a substantial increase in stimulatory capacity when infected with rV‐‐TRICOM ((Figure 8B and D)). Thus, the enhanced stimulatory capacity induced by infection of DCs with TRICOM vectors was noted with both mature DCs and DCs of intermediate maturity.

As the in vitro findings demonstrated that infection with TRICOM vectors endowed DCs with an enhanced ability to stimulate T cells, these reagents were also examined in vivo. The model peptide OVA was chosen to demonstrate the broad application of DCs infected with rF‐‐TRICOM or rV‐‐TRICOM. As shown in Figure 9, OVA‐‐pulsed DCs were able to induce OVA‐‐specific CTL activity ((3.2–5.2 LU)) following a single vaccination of 1 × 10⁵ cells. This level of CTL induction was similar to that seen by Porgador et al. (([1996])) when using this system. Vaccination with the same dose of OVA‐‐pulsed DC//rF‐‐TRICOM induced markedly and statistically greater levels of CTL activity ((> 20 LU)) than OVA peptide‐‐pulsed DCs ((Figure 9)). Similar results were demonstrated when peptide‐‐pulsed DC populations were infected with rV‐‐TRICOM ((Figure 9)).

Several groups have been exploring the introduction of exogenous model genes into DCs by retrovirus, adenovirus, or poxvirus vectors containing a single transgene. The advantage of using recombinant poxviruses to introduce genes into DCs is the efficient level of infection ((Figure 6, Table 1)) and the ability of the poxvirus to accommodate large amounts of foreign DNA and express multiple transgenes. To date, as many as seven genes have been inserted into the vaccinia‐‐virus genome ((Ockenhouse et al., [1998])).

The avipox viruses represent potentially attractive vectors for use in DC vaccines, as avipox viruses such as fowlpox and ALVAC ((canarypox)) can be administered numerous times to enhance immunogenicity ((Hodge et al., [1997])). Since they are replication‐‐defective, the consequences of any host immune responses should be minimal. Avipox viruses are also distinct from vaccinia in that the inserted transgene is expressed in infected cells for 14–21 days prior to cell death. In a vaccinia‐‐infected cell, the transgene is expressed for 1 to 2 days until cell lysis. A potential concern inherent in using a dendritic cell immunogen containing a vaccinia vector is that vaccinia immunity might inhibit the effectiveness of subsequent vaccinations. Indeed, a single dose of dendritic cell populations infected with any of the recombinant vaccinia viruses reported here induced anti‐‐vaccinia antibody titers ranging from 1:4,000–1:9,000. However, these antibody titers had no effect on the efficacy of the subsequent dendritic cell vaccinations in the generation of CTL. One potential explanation for this is that vaccinia virus does not replicate in DCs ((Engelmayer et al., [1999])) and, thus, is protected from any potentially neutralizing antibodies due to its intracellular location. Again, there would be less concern about inducing neutralizing antibodies if non‐‐replicating poxviral vector systems such as fowlpox, avipox, or Modified Vaccinia Ankara ((MVA)) were used.

Dendritic cell immunogens, peptide‐‐loaded, apoptotic‐‐body loaded, or transfected with tumor RNA, among other methodologies, are currently being evaluated for clinical use. Infection of these DC immunogens with rF‐‐TRICOM or rV‐‐TRICOM could well improve the efficacy of these reagents in priming specific immune responses. Thus, the use of orthopox TRICOM vectors to infect APC ex‐‐vivo, or used in combination with a TAA transgene in a 4‐‐gene construct ((i.e., rF‐‐CEA‐‐TRICOM)), has implications in vaccine strategies for the immunotherapy of a range of human cancers and infectious diseases.

Acknowledgments

We thank Marjorie Duberstein, Dr. Patricia Greenhalgh, Diane Poole and Marion Taylor for help in conducting these studies, and Nicole Ryder and Debra Jacobs for their editorial assistance. We also thank Dr. Dennis Panicali and Dr. Gail Mazzara of Therion Biologics Corp. for virus production and valuable discussions.