Tumor Vaccination by Salmonella typhimurium After ...

BASIC STUDY

Tumor Vaccination by Salmonella typhimurium AfterTransformation with a Eukaryotic Expression Vector

in MiceImpact of a Salmonella typhimurium Gene Interfering

with MHC Class I Presentation

Susanne Hummel,* Ron N. Apte,† Udi Qimron,† Mario Vitacolonna,* Angel Porgador,†

and Margot Zoller*‡

Summary: Transformed attenuated Salmonella typhimurium (ST)

have been suggested as an efficient means of tumor vaccination.

However, ST themselves might be immunosuppressive, and the

question has arisen as to whether this impedes vaccination efficacy

even if ST are transformed with a eukaryotic expression vector such

that ‘‘tumor antigen’’ will be transcribed by the host. The question

was evaluated using a mutant SL7207, where the yej operon, which

interferes with MHC I-mediated presentation, had been inactivated

(SL7207DyejE). Mice were vaccinated with SL7207 or SL7207DyejE

transformed with a eukaryotic expression vector carrying the lacZ or

the gp100 gene and later received lacZ-transfected RENCA or YC8

or gp100-expressing B16F1 tumor cells. In vaccinated mice, tumor

growth started with a delay and some animals remained tumor-free;

however, the tumor growth rate remained unaltered. No significant

difference was seen between SL7207DyejE versus SL7207 vacci-

nated mice. The latter finding contrasted with ex vivo analyses where

vaccination with SL7207DyejE, compared with SL7207, induced a

significantly stronger response, including nonadaptive defense mech-

anisms. The failure to detect a superior vaccination efficacy of

SL7207DyejE in vivo could be attributed to a stronger effect of the yej

operon on MHC-mediated antigen presentation when driven by a

prokaryotic promoter. Also, additional Salmonella genes apparently

interfere with maintenance of a sustained immune response. Thus, the

immunosuppressive yej operon affects innate and adaptive immunity.

However, when ST are carriers for eukaryotic-expressed tumor anti-

gens, yej does not severely hamper induction of an immune response.

Key Words: tumor vaccination, Salmonella typhimurium, immuno-

suppression

(J Immunother 2005;28:467–479)

The molecular identification of tumor antigens has led toa revival of active immunization as a mode of cancer

therapy.1 So far a wide range of different vaccination scheduleshave been explored, including DNA, RNA, proteins, and pep-tides, as well as carrier-mediated and carrier-free systems.2–4

Every system has advantages but also disadvantages. DNA,RNA, and protein vaccination have the advantages of al-lowing the presentation of multiple epitopes and a system-inherent presentation by the individual patient’s matchingMHC. However, whole proteins are frequently difficult tocreate in large quantities. Carrier-free systems of DNA andRNA application may be burdened by insufficient transcriptionand translation.5 Thus, the transfer of ex vivo loaded ortransfected/infected dendritic cells as the most potent antigen-presenting cells is generally agreed to be one of the methods ofchoice.6 However, the ex vivo generation of dendritic cells isvery time-consuming and cost-intensive and also depends atleast in part on the health of the patient, and there is an urgentneed for easier-to-handle vaccines. Vaccination with trans-formed attenuated Salmonella typhimurium or Listeria mono-cytogenes could well represent such an alternative.7–10 Salmonellatyphimurium is a facultative intracellular pathogen whosecapacity to cause systemic disease depends upon its ability tosurvive and replicate within macrophages.11–14 The bacteriacan cross the epithelial barrier of the gut and will be taken upby macrophages (Mf) and other mucosa-associated phago-cytes.12–15 In fact, the generation of several attenuated forms ofthese bacteria, which still pass the epithelial barrier and aretaken up by Mf but survive for only a limited number of cellcycles, has opened new and promising pathways of vaccina-tion.7,16–19 This accounts for vaccination against the pathogenitself 7,16,17 as well as for using the attenuated bacteria asa DNA vehicle.7,16–21 Depending on whether the DNA isinserted into a prokaryotic or a eukaryotic expression vector,the DNAwill be transcribed by the Salmonella or by the host’sMf after digestion of the attenuated bacteria.8,16,20,22–24 Vac-cination of tumor-bearing mice with attenuated SL, which hadbeen transformed with a prokaryotic or a eukaryotic

Received for publication February 22, 2005; accepted April 14, 2005.From the *Department of Tumor Progression and Tumor Defense, German

Cancer Research Center, Heidelberg, Germany; †Department of Micro-biology and Immunology, Ben Gurion University, Beer Sheva, Israel; and‡Department of Applied Genetics, University of Karlsruhe, Karlsruhe,Germany.

U.Q. is now at Biological Chemistry and Molecular Pharmacology, HarvardMedical School, Boston, MA.

This work was supported by the Israel-German Research Foundation (A.P.,R.N.A., M.Z.), the Mildred Scheel Stiftung fur Krebsforschung (M.Z.),and the Carreras Leukemia Foundation (M.Z.).

Reprints: Margot Zoller, Department of Tumor Progression and ImmuneDefense, German Cancer Research Center, Im Neuenheimer Feld 280,D-69120 Heidelberg, Germany (e-mail [email protected]).

Copyright � 2005 by Lippincott Williams & Wilkins

J Immunother � Volume 28, Number 5, September/October 2005 467

expression vector, has been explored in several animal modelsand has provided promising results.25–31

Mf are equipped with efficient natural killing mecha-nisms.32 Thus, intraphagosomal bacteria can be expected toevolve mechanisms withstanding or suppressing the killingactivities of Mf.7,33,34 While immunosuppressive mechanismsexploited by viruses are well known,35 much less is knownabout immune escape mechanisms and immunosuppressivefeatures of intracellular bacteria. Thus, some Salmonellatyphimurium (ST) strains have evolved mechanisms to limitbacteriostatic and bactericidal effects of reactive nitrogen in-termediates and reactive oxygen species.36 The Salmonellaprotein SpiC is exported into the host cell cytosol and inter-feres with intracellular trafficking and lysosomal fusion withthe Salmonella-containing vacuoles.37,38 Considering bacterialgenes that interfere with Mf-mediated antigen presentationrequired for the induction of a T-cell-mediated immune re-sponse, it has been shown that MHC class II presentation isenhanced in Mf infected with PhoP-mutant Salmonella.39

Also, Chlamydia secretes a factor that suppresses MHC class Iand class II antigen expression.40 Salmonella can induce alter-native splicing of HLA-B27 pre-mRNA such that due to mis-folding of the molecule peptide presentation is impaired.41

In addition, we recently performed a systemic search for STgenes that interfere with the MHC class I presentation pathway.42

Besides the already known gene PhoP, the yej operon in SThas been shown to interfere with peptide presentation on MHCclass I molecules. A mutant in yejE was used as vaccine vectorin mice using ovalbumin 257-265 as model antigen. Thoughinvasiveness was comparable to nonmutated ST, the mutantdid influence presentation of ovalbumin, but only when thebacteria had been transformed with the ovalbumin gene. Thedata were interpreted in the sense that the yejE gene acts spe-cifically in the phagosome.42

We have described before that vaccination with attenuatedST transformed with a eukaryotic expression vector carryingfull-length cDNA of an (artificial) tumor-associated antigenlike the lacZ gene or the murine gp100 cDNA suffices forinduction of a T-cell response, particularly of CD8+ CTL.28,43

Thus it became of interest whether a mutant Salmonella strainenhancing MHC class I-mediated presentation would havea positive impact on antigen presentation even if Salmonellaare transformed with a eukaryotic expression vector, whichimplies that the ‘‘tumor-antigen’’ DNA is transcribed by thehost Mf. As a model system we used the naturally occurringgp100 tumor antigen and b-galactosidase as an artificial tumorantigen. Both cDNAs, inserted in a eukaryotic expressionvector, were used for transforming SL7207 and SL7207DyejE.Although ex vivo analysis confirmed a superiority ofSL7207DyejE compared with SL7207 in immune responseinduction, this effect did not become obvious in the tumor-bearing host.

METHODS

Mice and TumorsC57BL6 and BALB/c mice were obtained from WIGA

(Sulzfeld, Germany). Animals were kept under specific pathogen-free conditions. They were fed sterilized food and water ad

libitum. Animals were used for experiments at the age of 8 to10 weeks. The C57BL6-derived melanoma line B16F144 wasused for cloning of the murine gp100 cDNA as described.28

The BALB/c-derived renal cell carcinoma line RENCA45 andthe BALB/c-derived leukemia line YC846 were co-transfectedwith the lacZ cDNA containing plasmid pVAX (Invitrogen)and the pcDNA3 vector containing the neomycin resistancegene (RENCA-ßGal, YC8-ßGal) by electroporation (1050 mF,260 V). Transfected cells were selected by growth in G418-containing medium (500 mg/mL). Surviving cells were clonedby limiting dilution. Expression was tested by X-Gal staining.43

For in vivo experiments B16F1, RENCA-ßGal, and YC8-ßGalwere used throughout. For in vitro experiments B16F1,RENCA, RENCA-ßGal, YC8, YC8-ßGal, and the NK targetYAC were used. All lines were cultured in RPMI 1640 sup-plemented with 10% FCS. When confluent, adherent growinglines (B16F1, RENCA, RENCA-ßGal) were trypsinized andcultures were split 1:3.

Monoclonal Antibodies and Flow CytometryThe following mAbs were used: YTA 3.1.2 (anti-CD4),

YTS169.4.2.1 (anti-CD8), YBM 6.1.10 (anti-Mac-1), and33D1 (anti-DC). The hybridomas have been obtained from theEuropean Collection of Animal Cell Cultures (Salisbury,England). Anti-CD25, anti-CD28, anti-CD16/CD32, biotinylatedanti-IL2, anti-IL4, anti-IL6, anti-IL10, anti-IL12, anti-IFNg, anti-TNFa and secondary, dye-labeled (FITC or PE) antibodies wereobtained commercially (Pharmingen, Hamburg, Germany).

For flow cytometry 3 to 5 3 105 cells were stainedaccording to standard protocols. When assaying cytokine-expressing cells, lymphocytes were fixed with formaldehydeand permeabilized with PBS/1% Tween-20. Fluorescence wasdetermined using a FACStar (Becton Dickinson, Heidelberg,Germany).

Lymphocyte PreparationLymph nodes and spleen were prepared aseptically and

collected in PBS. Peritoneal exudate cells were collected byflushing the peritoneal cavity with 3 mL PBS containing 20 Uheparin. Spleen and lymph nodes were meshed through finegauze, spleen cells and lymph node cells were washed twice inPBS and were resuspended in RPMI 1640 containingantibiotics, 10% fetal calf serum, glutamine, 10 mM HEPESbuffer, and 5 3 1025 M 2-mercaptoethanol (RPMI-s).

Transformation of Salmonella typhimurium aroA(SL7207 and SL7207DyejE)

The attenuated Salmonella typhimurium aroA strainsSL7207 (SL) and SL7207-yejE (SLDyejE) have been de-scribed previously.47 The yejE gene has been described tointerfere with MHC class I presentation. The gene has beendeleted in SL-yejE by a one-step gene inactivation method.48

SL and SLDyejE were grown in LB medium. Competentbacteriawere transformed with the eukaryotic expression plasmidpVAX1 (SL-mock, SLDyejE-mock) or with pVAXmgp100(SL-gp100, SLDyejE-gp100) or with pVAX containing thelacZ gene cDNA (Invitrogen) (SL-lacZ, SLDyejE-lacZ) byelectroporation (2,500 V, 25 mF). Transformation was con-trolled by selection with kanamycin. After 1 hour at 37�C in

468 q 2005 Lippincott Williams & Wilkins

Hummel et al J Immunother � Volume 28, Number 5, September/October 2005

LB medium, cells were cultured overnight on agar platescontaining 50 mg/mL kanamycin.

In Vitro AssaysProliferative activity of lymph node cells was evaluated

by culturing titrated numbers of cells (2 3 105 to 2.5 3104/well) in the absence or presence of 100 mg/mL gp100peptide mix (H2-Kb motif: GSQSYVPL, TWGKYWQV,VLYRYGSF, QRRSFLPV, H2-Db motif: EGSRNQDWL,GLGENSPLL, LRVINDGPTL, HFLRNHPLI) or 100 mg/mLb-galactosidase (ßGal) or with lysates of C57BL6 or BALB/Clymph node cells or lysates of B16F1 cells, RENCA cells,RENCA-ßGal cells, YC8 cells, YC8-ßGal cells (lysate of 1 3106 cells/mL) in 96-well round-bottom plates. Cells werecultured for 3 days, with 3H-thymidine (10 mCi/mL) addedduring the last 16 hours of culture. Plates were harvested usingan automatic harvester and 3H-thymidine incorporation wasevaluated in a b-counter.

For the activation of lymphokine-activated killer cells,spleen cells were cultured for 48 hours in the presence of50 U/ml IL-2. Macrophage activity (Mf) was determined afterseeding peritoneal exudate cells in 96-well flat-bottom platesand washing out nonadherent cells after 2 hours of incubationat 37�C, 5% CO2 in air. Cytotoxic T cell (CTL) activity wasdetermined after culturing lymph node cells for 10 days in thepresence of 10 U/mL IL-2 plus either the gp100 peptide mix orßGal or cell lysates (as described above).

Cytotoxicity of lymphokine-activated killer cells, Mf,and CTLs was performed according to the protocol describedby Matzinger.49 In brief, target cells were labeled overnightwith 3H-thymidine (10mCi/mL) and were seeded (104 cells/well)after washing on titrated numbers (10 3 105 to 6 3 104) ofeffector cells in 96-well microtiter plates. After incubation for4 to 6 hours (Mf, 8 hours) at 37�C, plates were harvested andthe remaining radioactivity was determined in a b-counter.Cytotoxicity is presented as % cytotoxicity = 1003 (counts incontrol wells 2 counts in test wells):(counts in control wells).The spontaneous release of B16F1, YC8, YC8-ßGal, RENCA,RENCA-ßGal, and YAC cells was in the range of 6% to 12%;the spontaneous release of lymphoblasts (lymph node cellscultured for 48 hours in the presence of 1 mg/mL ConA) wasin the range of 12% to 20%. SDs of triplicate cultures were inthe range of 3% to 5%.

Vaccination and Tumor GrowthFor oral vaccination, transformed SL and SLDyejE were

cultured overnight in LB medium containing 2% NaCl and50 mg/mL kanamycin at 37�C. After determination of OD600

(OD600 1 roughly corresponds to 13 108 SL or SLDyejE/mL),bacteria were centrifuged at 6,000 rpm and washed twice withPBS. Bacteria were resuspended in PBS (containing 10%bicarbonate, pH 9.6), pH 8.3 to 8.5, to neutralize the acid pH ofthe stomach. Mice received 1 to 23 108 bacteria in 100 ml viaprobing. Vaccination was performed in 3-week time intervals.Mice were vaccinated twice in advance of tumor cell appli-cation and received the third vaccination at the time of tumorcell application. The vaccination schedule was maintained aftertumor cell application.

C57BL6 mice received a subcutaneous (s.c.) or in-traperitoneal (i.p.) injection of 1 3 104 B16F1 cells. BALB/Cmice received an s.c. or i.p. injection of 53 103 RENCA-ßGalcells or an intravenous (i.v.) or i.p. injection of 1 3 103 YC8-ßGal cells. Subcutaneous tumor growth was controlled bymeasuring the mean diameter twice per week. Intraperitonealtumor growth was controlled by palpation of the abdomen andestimation of the intraperitoneal tumor mass twice per week.Mice were killed when the i.p. tumor mass reached a meandiameter of 2.5 cm or when animals became anemic, cachetic,or short of breath after i.v. tumor cell application. These timepoints were defined as survival time.

Statistical AnalysisSignificance of differences was calculated according to

the Student t test (in vitro assays) or the Wilcoxon rank sumtest (in vivo assays). P , 0.01 was considered significant.

RESULTS

Monocyte and T-Cell Activation AfterVaccination With SL Versus SLDyejE

Salmonella typhimurium transverse the intestinal epi-thelium and invade macrophages as their natural host.7,13 Thus,attenuated strains with a limited survival time appear wellsuited as a vaccine carrier. Because bacteria, similar to viruses,can develop immune escape mechanisms,32,33 knowledge ofimmune evasive genes is important to improve the vaccinationefficacy. We explored whether a mutant of the yejE operon,which somehow appears to protect Salmonella in the phagosomefrom degradation by the host cell,42 can increase the efficacy oftumor vaccination.

We first tested the impact of SL vaccination on immunecell activation in the tumor-free mouse to see whether deletionof the yejE operon results in improved responsiveness. Micereceived two oral applications of either SL or SLDyejE, whichhad been transformed with the empty vector or with the samevector containing the lacZ gene or the murine gp100 gene.Spleen cells, lymph node cells, and peritoneal exudate cellswere collected 1 week after the second vaccination to evaluateexpression of activation markers and inflammatory cytokinesas well as proliferative and cytotoxic activity.

Expression of the T-cell activation markers CD25 andCD28 was upregulated in spleen cells, lymph node cells, andperitoneal exudate cells irrespective of whether the mice receivedSL or SLDyejE and whether the vector contained a ‘‘tumorantigen’’ cDNA. Thus, SL by themselves induce T-cell activation.However, mainly in the spleen, activation marker expressionwas strikingly upregulated when mice received SLDyejE com-pared with SL (Fig. 1A).

Further evidence for immune activation by SL vaccinationwas obtained by evaluating intracellular cytokine expression inperitoneal exudate cells. The analysis of IL-12, a proinflam-matory cytokine mainly secreted by Mf, revealed upregula-tion by vaccination with SL/SLDyejE, which was furtherincreased by vaccination with lacZ or gp100 cDNA carryingSLDyejE. A similar expression profile was seen for the inflam-matory cytokine IFN-g, secreted by Mf as well as T cells.

q 2005 Lippincott Williams & Wilkins 469

J Immunother � Volume 28, Number 5, September/October 2005 Salmonella typhimurium Strains

There was a tendency toward higher expression in micevaccinated with SL/SLDyejE carrying tumor antigen cDNA.Expression of the inhibitory cytokine IL-10 also appeared to beof interest. IL-10 expression was increased in vaccinated mice;however, the presence of tumor antigen cDNA obviouslyexerted no major effect (see Fig. 1B).

With the exception of a relative increase in the percentageof Mf in the peritoneal cavity, the distribution between CD4+

and CD8+ T cells, B cells, NK cells, andMfwas unchanged inlymphoid organs (data not shown).

These findings provided a first indication for T-cellactivation by oral SL/SLDyejE application. The higher levelof CD25+/CD28+ spleen cells in SL/SLDyejE-lacZ or -gp100vaccinated mice could be indicative for induction of atumor-antigen-specific response; the higher level in SLDyejEvaccinated mice might indicate SLDyejE to be more immuno-genic/less immunosuppressive than SL.

We next assessed the functional activity of the activatedT cells (i.e., for proliferative and cytotoxic activity). The oralapplication of SL or SLDyejE exerted no significant effect onthe overall proliferative activity of spleen cells (data not shown)and lymph node cells, as evaluated by 3H-thymidine in-corporation after stimulation with autologous lymphocytes.

Also, empty vector-transformed SL had a weak or no mea-surable impact on the proliferative response toward ßGal orgp100 peptides. However, the proliferative response towardßGal and gp100 peptides was significantly increased whenmice had been vaccinated with lacZ or gp100 cDNA trans-formed SL/SLDyejE. Deletion of the yejE gene exerted noeffect (Fig. 2A).

This first evidence toward an efficient ‘‘antigen-specific’’vaccination by SL and SLDyejE was confirmed when evalu-ating CTL activity after 1 week of in vitro (re)stimulation. Toexclude a major contribution of NK cells, all CTL assays wereperformed in the presence of a five-fold excess of unlabeledYAC cells as cold target NK inhibitors. Target cells have beenßGal expressing YC8 and RENCA cells; mock-transfectedYC8 and RENCA cells served as controls. For evaluating thegp100 response, gp100 expressing B16F1 cells served astarget and C57BL6 lymphoblasts as control. Three observa-tions should be particularly noted. First, nonspecific bystandercytotoxic activity was increased in SLDyejE compared withPBS or SL vaccinated mice. Second, YC8-ßGal, RENCA-ßGal, and B16F1 cells were more efficiently lysed com-pared with control cells; in other words, vaccination withtransformed SL and SLDyejE apparently sufficed for the

FIGURE 1. T-cell activation and cytokine expression by vaccination with Salmonella typhimurium: comparison between SL7207 andSL7207DyejE. BALB/c mice received two oral applications at a 3-week interval of either PBS or 2 3 108 SL or SLDyejE or SL-lacZ orSLDyejE-lacZ; C57BL6 mice received either PBS or 23 108 SL or SLDyejE or SL-gp100 or SLDyejE-gp100 (as there were no significantdifferences between BALB/c and C57BL6 mice, only the latter two groups of the C57BL6 mice are shown). Mice were killed 10 daysafter the second vaccination. The percentage of (A) CD25+ and CD28+ spleen cells, lymph node cells, and peritoneal exudate cellsand (B) IL-10+, IL-12+, and IFN-g+-expressing peritoneal exudate cells is shown. Significant differences (P, 0.01) in the mean6 SDof 3 mice/group are indicated by s (vaccinated vs. nonvaccinated) or an asterisk (SLDyejE vs. SL).



activation of antigen-specific CTLs. Third, vaccination withtransformed SLDyejE compared with SL apparently wasslightly more efficient in activating antigen-specific CTLs(see Fig. 2B).

Thus, a tumor antigen introduced via SL carryinga eukaryotic expression vector induced an efficient proliferativeand cytotoxic T-cell response with only a minor advantage ofthe yejE operon deletion.

As Mf are the prime host of SL, it was also importantto explore the impact of SL vaccination on Mf activity. Mffrom vaccinated mice exerted a higher cytotoxic activitythan Mf from nonvaccinated mice. Irrespective of whetherSL were transformed with an empty vector or a tumorantigen cDNA carrying vector, vaccination with SLDyejE

induced a significantly higher cytolytic Mf response thanvaccination with SL (Fig. 3A). The latter effect, althoughless pronounced, accounted also for NK activity, whichwas evaluated after overnight incubation of spleen cells inIL-2-containing medium using YAC as target cells (seeFig. 3B).

These findings indicated that SLDyejE induced T-cell,NK-cell, and Mf activation more efficiently than SL in mostinstances. However, when the antigen cDNA had been insertedin a eukaryotic expression vector, differences in ‘‘antigen-specific’’ responses after vaccination with SL versus SLDyejEwere nonexistent or minor. This implies that presentation ofthe eukaryotically transcribed antigen by Mf is not significantlyinfluenced by the yejE operon deletion.

FIGURE 2. Induction of a T-cell response by vaccination with Salmonella typhimurium: comparison between SL7207 andSL7207DyejE. A, Proliferative activity of lymph node cells from control and vaccinated mice (as described in Fig. 1) in response toautologous irradiated lymph node cells, ßGal (100 mg/mL), or a mixture of gp100 peptides (100 mg/mL) is shown. Mean 6 SD oftriplicates. B, Lymph node cells of vaccinated mice (as described in Fig. 1) were cultured in the presence of 10 UIL-2/mL plus either ßGal (100 mg/mL) or a mixture of gp100 peptides (100 mg/mL). Cytotoxic activity was evaluated in a6-hour assay with the indicated 3H-thymidine-labeled target cells and a 5-fold excess of YAC as cold target cells. Mean 6 SD oftriplicates at an E:T ratio of 50:1. A, B, Significant differences (P , 0.01) are indicated by s (vaccinated vs. nonvaccinated) or anasterisk (SLDyejE vs. SL) or a filled asterisk (SL/SLDyejE vs. SL-lacZ/gp100/SLDyejE-lacZ/gp100). All experiments were repeated threetimes. Representative examples are shown.



Efficacy of Vaccination with Transformed SLand SLDyejE on Tumor Growth in thePeritoneal Cavity, at a Subcutaneous Site, andin the Lung

To see whether there are major differences betweena solid tumor and a leukemia, the efficacy of SL vaccinationwas tested using lacZ cDNA transfected YC8 and RENCAcells. Because bGal is a xenoantigen, which we used as

a surrogate tumor antigen, we included as a third model theB16F1 melanoma line that expresses gp100, a low immuno-genic tumor-associated antigen. Mice received two vaccina-tions before tumor cell application. Tumors were appliedintraperitoneally to control for local response induction, andparticularly for the efficacy of the innate immune system,taking into account the abundant presence of Mf at this site. Ina parallel setting the tumor cells were deposited s.c. or, in caseof the leukemia, i.v. to mimic more closely the clinicalsituation of vaccination in patients with cancer and to controlwhether SL vaccination suffices for inducing a systemic re-sponse. In all settings the potential benefit of a deletion of theyejE operon was evaluated (Fig. 4, Table 1).

First, vaccination with transformed SL or SLDyejE wasof benefit for tumor-challenged mice. This accounted for therenal cell carcinoma, the malignant melanoma, and the leukemia.The rate of persistently tumor-free mice was highest after theapplication of RENCA-ßGal; the most pronounced prolongationof the survival time was seen in YC8-ßGal-bearing mice.Second, the therapeutic effect was at least partially ‘‘tumorantigen-specific.’’ Vaccination with SL/SLDyejE transformedwith an empty vector exerted no effect on the survival of micebearing RENCA or B16F1, and only a minor effect on thesurvival of mice bearing YC8-bGal. Notably, in this settingSLDyejE were more efficient than SL. Third, SL or SLDyejEvaccination obviously induced a systemic response; that is, thesurvival rate was improved when the tumors were given i.p.,i.v., or s.c. However, the increase in the survival time was lesspronounced in mice with s.c. tumors or lung metastases.Fourth, deletion of the DyejE operon had no major impact onsurvival time and survival rate when SL were transformed witha eukaryotic expression vector carrying a ‘‘tumor antigen’’cDNA. Fifth, in the vaccinated mice that died with the tumor,tumor growth started with delay, but later the tumor growthrate did not differ from the one in mock-vaccinated mice.

We interpret the data as follows. Deletion of the yejEoperon does not create a special vaccination advantage whenSL are transformed with a eukaryotic expression vector. SLvaccination suffices for induction of a systemic response,which, however, requires repeated and long-lasting vaccina-tion schedules. Natural defense mechanisms such as NK cellsand Mf may be more effective against leukemic than solidtumors, and the lytic machinery of NK and Mf possiblyprofits from yejE deletion. All three tumors obviously exertactive counterregulatory effects. This is deduced from theobservation that tumor growth, although starting with delay invaccinated animals, progressed rather unimpaired thereafter.To support these hypotheses, we evaluated the immuneresponse profile in mice bearing small tumors and in latetumor-bearing mice.

Repeated Vaccination with Transformed SLand SLDyejE Suffices for the Induction of aSystemic T-Cell Response

In early tumor-bearing mice (ie, before visible tumorgrowth), vaccination with transformed SL was accompaniedby an increase in activated T cells as shown for CD25 ex-pression (Fig. 5A) and a strong increase in IL-12, IFN-g andalso in IL-10 cytokine expressing lymph node cells, as shown

FIGURE 3. Mf and NK activation by vaccination withSalmonella typhimurium: comparison between SL7207 andSL7207DyejE. A, Spleen cells of vaccinated (as described in Fig. 1)BALB/c and C57BL6 mice were cultured overnight in thepresence of 50 U/mL IL-2. Cytotoxic activity was evaluated ina 4-hour assay with 3H-thymidine-labeled YAC target cells.Mean 6 SD of triplicates at a NK:T ratio of 100:1. B, Peritonealexudate cells of vaccinated (as described in Fig. 1) BALB/c andC57BL6 mice were seeded in flat-bottomed 96-well plates.After 2 hours of adherence, the nonadherent cells were washedout. Cytotoxic activity was evaluated in an 8-hour assay with3H-thymidine-labeled YC8-ßGAL or B16F1 target cells. Mean6SD of triplicates at Mf:T ratio of 50:1. A, B, Significantdifferences (P , 0.01) in the mean 6 SD of 3 mice/group areindicated by s (vaccinated vs. nonvaccinated) or an asterisk(SLDyejE vs. SL). All experiments were repeated three times.Representative examples are shown.



FIGURE 4. Survival time, survival rate, and tumor growth after vaccination with SL7207 and SL7207DyejE. BALB/c (A, B, C, D, G)and C57BL6 (E, F, H) mice received three times 23 108 SL or SLDyejE (A, C, E) or SL-lacZ or SLDyejE-lacZ (A, B, C, D, G) or SL-gp100or SLDyejE-gp100 (E, F, H) or PBS (A–H) in 3-week intervals and, concomitantly with the third vaccination, 13 103 YC8-ßGal, i.p. ori.v. (A, B) or 53 103 RENCA-ßGal, i.p. or s.c. (C, D, G) or 13 104 B16F1, i.p. or s.c. (E, F, H). Survival time and rate are shown (A–F).The tumor growth rate is shown for 3 mice per group after vaccination with SL-lacZ or SLDyejE-lacZ or PBS and s.c. application ofRENCA-bGal (G) and after vaccination with SL-gp100 or SLDyejE-gp100 or PBS and s.c. application of B16F1 (H). The 3 mice pergroup were selected according to their survival time being closest to the mean survival time of the respective group.



for YC8-ßGal and B16F1 tumor-bearing mice (see Fig. 5B).There was no consistent difference between vaccination withSL versus SLDyejE. Also, lymph node cell activation wasobserved whether the tumor was applied at the site of vac-cination (i.p.) (data not shown) or at a distant site.

Similar to vaccinated, tumor-free mice, vaccination withtransformed SL or SLDyejE had a minor effect on the pro-liferative activity of spleen cells (data not shown) and lymphnode cells toward tumor cells not expressing the antigen usedfor vaccination. On the contrary, a strong proliferative responsewas observed when lymphocytes were restimulated with tumorcells carrying the nominal antigen. The nominal antigen, in theabsence of tumor cells, induced the most profound pro-liferative activity (Fig. 6A). This was shown for the pro-liferative response toward YC8-ßGal and RENCA-ßGal cellsversus ßGal and for B16F1 cells versus gp100 peptides.

We also noted a significant increase in ‘‘antigen-specific’’ CTL activity in the early YC8-ßGal, RENCA-ßGal,and B16F1 tumor-bearing hosts (see Fig. 6B). There wasa minor increase in ‘‘bystander’’ cytolytic activity after vac-cination with SLDyejE-lacZ; that is, YC8 cells were more

efficiently lysed by lymph node cells from SLDyejE-lacZvaccinated compared with SL vaccinated mice. Also, YC8-ßGal cells were more efficiently lysed after vaccination withSLDyejE-lacZ compared with SL-lacZ. No significant differ-ence was seen between lymph node cells from SLDyejE-lacZ/gp100 and SL-lacZ/gp100 vaccinated mice with respectto the lysis of RENCA-bGal and B16F1. Finally, CTLactivation was independent of whether the tumor cells wereapplied i.p. (data not shown), s.c., or i.v.

Vaccination of tumor-free mice had provided evidencethat SLDyejE may be more effective in the activation of Mfand NK cells than SL. This suggestion could be strengthenedby the analysis of NK and Mf cytotoxicity in early tumor-bearing mice. In particular, YC8-bGal cells were more efficientlylysed by Mf and NK cells from SLDyejE than from SLvaccinated mice (see Fig. 6C, D). In addition, RENCA-bGaland B16F1 cells were also more efficiently lysed by Mf andNK cells from SL/SLDyejE vaccinated as compared withnonvaccinated mice. This accounted for i.v. or s.c. tumor cellapplication. We had pointed out repeatedly that vaccinationwith SL and SLDyejE induced a systemic T-cell response and

TABLE 1. Comparative Evaluation of SL Vaccination-Induced Prolongation of Survival Time

Survival Time After Application of Tumor Cells at the Vaccination Site

Tumor* Vaccination Mean Survival Time†P Value (tumor +

SL/SLDyejE vs. tumor)P Value (SL/SLDyejEtransformed vs. mock)

P Value(SLDyejE vs. SL)

RENCA-ßGal None 69.3 6 11.2

SL 75.1 6 10.5 NS

SL-lacZ 99.3 6 19.8 0.002 0.009

SLDyejE 78.1 6 17.9 NS

SLDyejE-lacZ 104.3 6 25.7 0.003 0.034 NS

B16F1 None 38.8 6 5.0

SL 41.8 6 5.5 NS

SL-gp100 73.9 6 29.4 0.005 0.009

SLDyejE 40.4 6 6.3 NS

SLDyejE-gp100 74.1 6 30.2 0.008 0.008 NS

YC8-ßGal None 18.6 6 2.5

SL 22.9 6 4.7 0.041

SL-lacZ 59.5 6 38.6 0.010 0.018

SLDyejE 26.8 6 6.9 0.007

SLDyejE-lacZ 61.8 6 37.2 0.006 0.020 NS

Survival Time After Application of Tumor Cells Distant From the Vaccination Site

Tumor* Application Route VaccinationMean SurvivalTime (days)†

P Value (tumor +SL/SLDyejE vs. tumor)

P Value(SLDyejE vs. SL)

RENCA-ßGal s.c. None 72.5 6 15.7

s.c. SL-lacZ 98.5 6 20.2 0.012

s.c. SLDyejE-lacZ 97.5 6 23.7 0.026 NS

B16F1 s.c. None 41.3 6 8.1

s.c. SL-gp100 70.5 6 25.2 0.007

s.c. SLDyejE-gp100 68.1 6 26.7 0.016 NS

YC8-ßGal i.v. None 32.9 6 4.8

i.v. SL-lacZ 72.0 6 42.4 0.021

i.v. SLDyejE-lacZ 73.8 6 40.4 0.013 NS

*BALB/c or C57BL6 mice were vaccinated as described in text. BALB/c mice received either 13 103 YC8-ßGal or 53 103 RENCA-ßGal, C57BL6 mice received 13 104 B16F1cells, i.p. (first part of table) or s.c. or i.v. (second part of table).

†Mean survival time is shown for 8 mice/group.



that no differences could be seen in effector functions betweenmice carrying the tumor i.p., i.v., or s.c. Surprisingly, this wasdifferent for NK and Mf activity. After i.p. tumor cell ap-plication, NK cells and Mf of SL or SLDyejE vaccinated miceexerted no increased cytolytic activity except for a minorincrease in the lysis of YC8-bGal cells (see Fig. 6C, D).

The ex vivo analysis confirmed that both SL strainssufficed to induce a systemic T-cell response against nominaltumor antigens with a considerable contribution of NK cellsand Mf. The latter were more efficiently activated by SLDyejEthan by SL. Surprisingly, NK and Mf activity was stronglymitigated after i.p. tumor cell application. This locallyrestricted mitigation could be indicative for immunosuppres-sive features of the growing tumor. However, the obviousbreakdown of the vaccination efficacy in 50% to 80% ofvaccinated mice did not depend on the tumor localization.Thus, an additional contribution to immunosuppression by thevaccine itself became likely. To control for the hypothesis,immune reactivity was evaluated in late tumor-bearing mice,which at that time had received four or five SL applications.

Nonresponsiveness of Immune Cells of LateTumor-Bearing Mice

Mice were killed when tumors reached a mean diameterof about 1 cm (depending on the tumor, between 4 and 8 weeksafter tumor cell inoculation). At this stage, mice were notcachectic and, with the exception of RENCA-ßGal-bearing mice,increased numbers of spleen cells (data not shown) and lymph

node cells were recovered from vaccinated mice (Fig. 7A).However, there was a tendency toward a reduction in expres-sion of T-cell activation markers (CD25 and CD28+) (data notshown). In nonvaccinated mice, IL-10, IL-12, and IFN-gcytokine expression was in the same range as in early tumor-bearing mice. On the contrary, cytokine expression was strikinglyreduced in lymph node cells from late versus early tumor-bearing, vaccinated mice. Deletion of the yejE operon did notprevent cytokine downregulation (see Fig. 7B).

The most striking changes were seen with respect toproliferating and cytotoxic T cells. The proliferative activity oflymph node cells from SL and SLDyejE vaccinated mice wasmostly below or in the range of lymph node cells fromnonvaccinated mice cultured in the absence of a nominalantigen (data not shown). In B16F1 and RENCA-bGal-bearingmice, T-cell cytotoxicity was largely abolished. It was alsoreduced in YC8-ßGal-bearing mice. In addition, bystanderkilling of YC8 and RENCA cells as well as of C57BL6lymphoblasts was hardly detectable (see Fig. 7C). Comparedwith early tumor-bearing mice, NK and Mf cytotoxic activitywas also significantly reduced after repeated vaccination withSL or SLDyejE. The only exceptions were NK cells fromSLDyejE-lacZ vaccinated mice, whose activity toward YC8-bGal remained elevated (see Fig. 7D).

Taken together, both transformed SL7207 andSL7202DyejE induced a ‘‘tumor-antigen’’ specific response.Although SL were transformed with a eukaryotic expressionvector, deletion of yejE supported the cytolytic activity of Mf,

FIGURE 5. T-cell activation in tumor-bearing mice after vaccination withSL7207 and SL7207DyejE. Lymphnode cells from vaccinated tumor-bearing mice (as described in Fig. 4)were collected and the percentageof (A) CD25+ and (B) IL-10, IL-12,and IFN-g-expressing lymph nodecells was evaluated. Significant dif-ferences (P , 0.01) in the mean 6SD of 3 mice/ group are indicated bys (vaccinated vs. nonvaccinated).



but also of NK cells and to a minor degree of CTL. However,although some vaccinated mice did not develop tumors, otherssuddenly died with rapidly growing tumors. The latter featuremay be partly due to immunosuppressive features of the tumoritself, but it is obviously supported by the vaccine. This isdeduced from the observation that at a late stage of tumorgrowth, immunoreactivity of vaccinated mice was reduced tothe level or even below that of nonvaccinated mice.

DISCUSSIONSalmonella typhimurium are gram-negative bacteria that

cause systemic disease by their capacity to survive and rep-licate within macrophages.7,12 As macrophages are equippedwith potent bactericidal activities,32 SL have adapted to escapetheir host’s defense.32,33 Although knowledge on escape mech-anisms is still scarce,8,50 vaccination protocols using live at-

tenuated SL have been developed that induce a potent immuneresponse and can protect from virulent bacteria.51–53 Becausetransformed attenuated Salmonella typhimurium provide aneasily accessible and applicable drug, SL-based vaccines arealso considered in cancer therapy.9,19,30,31 In addition, it wasthought that transforming SL with a eukaryotic expressionvector might circumvent potential defects in tumor antigenpresentation caused by the bacteria’s escape mechanisms.30 Inthis report, we intended to experimentally support this hypothesis.We have recently identified the yej operon to interfere withMHC class I-mediated presentation by SL-infected Mf.42

Deletion of yejE was accompanied by a significantly increasedresponse toward a prokaryotically expressed nominal antigen(OVA) that had been introduced via transformed SLDyejE.42

Here we controlled whether yejE deletion could also be ad-vantageous when SL were transformed with a eukaryotic ex-pression vector. Our ex vivo data provided evidence for induction

FIGURE 6. T cell, NK, and MF effector functions in tumor-bearing mice after vaccination with SL7207 and SL7207DyejE. A,Proliferation of lymph node cells from tumor-bearingmice in response to a syngeneic stimulus (YC8, RENCA, C57BL6 lymphocytes)or the nominal antigen presented by syngeneic tumor cells (YCA-bGal, RENCA-bGal, B16F1) or antigen-loaded dendritic cells(bGal, gp100). B, Lymph node cell-mediated cytotoxicity after in vitro restimulation with the nominal antigen presented bysyngeneic tumor cells. Cytotoxicity was evaluated using syngeneic targets (YC8, RENCA, C57BL6 lymphoblasts) or syngeneic,‘‘tumor-antigen’’ expressing targets (YCA-bGal, RENCA-bGal, B16F1). C, NK activity (after overnight culture of spleen cells in IL-2)was evaluated in a 6-hour assay using ‘‘antigen’’-expressing tumor cells as targets (x-axis). D, Mf activity was evaluated usingadherent peritoneal exudate cells and ‘‘antigen’’-expressing tumor cells as target (x-axis). Cytotoxicity was evaluated after 8 hoursof incubation. A-D, Significant differences (P, 0.01) are indicated by s (vaccinated vs. nonvaccinated) or an asterisk (SLDyejE vs. SL)or a filled asterisk (SL/SLDyejE vs. SL-lacZ/gp100/SLDyejE-lacZ/gp100). All experiments were repeated three times. Representativeexamples are shown.



of a stronger immune response by vaccination with SLDyejE,mainly with respect to Mf and NK cytotoxicity. This accountedfor three tumor lines of different histology and was mostpronounced for the YC8 lymphoma line. On the contrary, wedid not observe significant differences in survival time andsurvival rate depending on whether mice were vaccinated withtransformed SL or SLDyejE.

Because SL are immunogenic by themselves and SLdegradation products are potent antigens,21,54–56 we firstevaluated whether the immune response toward the SL andthe empty vector might not override the response against

a ‘‘tumor antigen,’’ which in the present study have been ßGaland gp100. Although we did not evaluate responsivenessagainst defined SL antigens, vaccination with SL trans-formed with an empty vector obviously induced an immuneresponse. There was a significant increase in the percentageof CD28+ and CD25+ cells, which in lymphoid organs ofvaccinated mice mostly were freshly activated rather thanregulatory T cells (ie, they were CD1522) (data not shown).The percentage of cytokine-expressing cells was elevatedand the proliferative activity of lymph node cells wasstrengthened.

FIGURE 7. Lymphocyte responsive-ness in late tumor-bearing mice aftervaccination with SL7207 andSL7207DyejE. Vaccinated mice, whichreceived a tumor cell inoculum eitheri.v. or s.c. as described in Figure 4,were killed 5 weeks (YC8-ßGal) and7 weeks (RENCA-ßGal, B16F1) after tu-mor cell application. A, Impact of tumorgrowth on the recovery of lymphnode cells. The mean number 6 SDof lymph node cells of three untreatedmice is shown as control (open bar). B,The ratio (mean 6 SD of 3 mice/group) of IL-10, IL-12, and IFN-g-expressing lymph node cells in latevs. early tumor-bearing mice. Cyto-kine expression remained unaltered innonvaccinated late tumor-bearingmice, whereas the vaccine-inducedupregulation was lost. C, Lymph nodecell-mediated cytotoxicity afterin vitro restimulation with the nominalantigen presented by syngeneic tu-mor cells. Cytotoxicity was evaluatedusing syngeneic targets (YC8, RENCA,C57BL6 lymphoblasts) or syngeneic,‘‘tumor-antigen’’ expressing targets(YCA-bGal, RENCA-bGal, B16F1).Mean cytotoxicity values of lymphnode cells of early tumor-bearing micevaccinated with SL-lacZ/gp100 (openbar) or SLDyejE-lacZ/gp100 (light graybar) are shown as control D, NKactivity (after overnight culture ofspleen cells in IL-2) was evaluated ina 6-hour assay and Mf activity (ad-herent peritoneal exudate cells) in an8-hour assay using ‘‘antigen’’-express-ing tumor cells (YCA-bGal, RENCA-bGal, B16F1) as targets. Mean cyto-toxicity values of NK and Mf of earlytumor-bearing mice vaccinated withSL-lacZ/gp100 (open bar) or SLDyejE-lacZ/gp100 (light gray bar) are shownas control. A–D, Significant differences(P , 0.01) are indicated by s (vacci-nated vs. nonvaccinated) or an aster-isk (SLDyejE vs. SL) or a filled asterisk (SL/SLDyejE vs. SL-lacZ/gp100/SLDyejE-lacZ/gp100). All experiments were repeated threetimes. Representative examples are shown.



Comparing SL with SL7207DyejE vaccinated mice, wenoted in SL7207DyejE vaccinated mice a significant increasein CTL bystander killing, a further increase in NK and Mfcytotoxicity, and elevated numbers of IL-10, IL-12, and IFN-gexpressing peritoneal exudate cells, although the latter did notreach statistical significance in all instances. These responsefeatures are in line with the recent notion that the yejE geneproduct and the yej operon interfere with MHC class I-mediatedpresentation by infected Mf42 and thus cause a reduced pep-tide presentation for CTLs.57 In addition, our in vitro findingsindicate that the yejE gene product also has a bearing on theactivation of Mf, NK, and T-helper cells. The mechanismwhereby yejE influences activation of the latter cells remains tobe explored. Yet the theory that Salmonella is confusing theimmune system by secreting different peptides that affectvarious immune cells comes to mind, taking into account thatthe yej operon encodes for the components of an ABC-typedipeptide/oligopeptide transport system.42

Given the increased responsiveness after vaccinationwith SLDyejE, it became of particular interest to see whetherdeletion of this gene would have also an impact on the responsetoward a tumor antigen, if the latter is transcribed by the hostrather than by the SL. Neither in the tumor-free nor in the earlytumor-bearing host did the proliferative response toward ßGaland gp100 differ significantly based on whether mice werevaccinated with SL or SLDyejE. With the exception of thelymphoid tumor YC8, this also accounted for the CTL response.Thus, although being advantageous with respect to nonadaptiveimmune and bystander T-cell activation, yejE operon deletiondid not provide an advantage in peptide presentation, when thegene was transcribed by the host.

To our surprise, the overall increase in immune respon-siveness of SL or SLDyejE vaccinated mice had vanished in earlytumor-bearing animals when the tumor cells were injected i.p.Because there is no reason for the efficacy of vaccination toparticularly vanish in the peritoneal cavity, we interpreted thelatter finding as a consequence of tumor-mediated immuno-suppression, which might more rapidly become effective at thelocal (i.p.) versus the more inaccessible (s.c.) site. The growthprofile of tumors in vaccinated mice strengthened the sus-picion of immunosuppression. While some mice remainedtumor-free, the majority of vaccinated mice developed tumors,although with delay. Notably, the tumor growth rate did notdiffer from the one in nonvaccinated animals. Although immu-nosuppressive features have been described for RENCA,58

B16F1,28,59 and YC860 cells, the unaltered growth rate in re-peatedly vaccinated mice argued against immunosuppressiononly by the tumor. To control whether the vaccination regimenmay contribute to the breakdown of the immune system, theimmunoreactivity of late tumor-bearing mice, which had re-ceived the vaccine four or five times, was compared with thatof early tumor-bearing mice. Although the late tumor-bearinganimals were not cachectic, immune responsiveness was stronglyimpaired. There remained no T-helper response, and evenbystander proliferative activity was most strongly impaired (datanot shown). The vaccination-induced increase in cytokineexpression had completely vanished, while cytokine expres-sion remained unaltered in nonvaccinated, tumor-bearing mice.The same accounted for Mf and NK cytotoxicity. Solely in

YC8-ßGal-bearing mice, a low level of CTL and NK activityabove that of nonvaccinated mice was observed. This obvi-ously vaccine-induced immunosuppression was independentof whether SL were transformed with the empty vector or atumor antigen cDNA carrying vector. Also, deletion of the yejEgene apparently did not prevent SL-mediated immunosup-pression. It remains to be explored whether SL-mediated immu-nosuppression is due to additional immunosuppressive operonsor to exhaustion of the immune system by the repeated activation.

An alternative explanation that might account for thedifferences between the ex vivo and the in vivo data is theelimination of the vector by neutralizing antibodies followingthe first vaccine application. It is well known in the field ofvaccination that neutralizing antibodies, especially of the IgAtype, and other specific T-cell responses evolve after the firstvector encounter and eliminate the efficacy of the followingvaccine applications.61,62 The elimination of the SLDyejE vectormay be faster than its parental control due to its enhancedimmunogenicity. Thus, even though the in vivo data showsimilar results, the overall effectiveness of the SLDyejE mightbe superior, taking into account the rate of elimination of bothvectors.

Taken together, vaccination with attenuated transformedSL could well provide a means for cancer immunotherapy. Asfar as the tumor antigen cDNA is inserted into a eukaryoticexpression vector, deletion of the yejE gene is of some ad-vantage during induction of an immune response, which exceedsthat induced by yejE-competent SL. However, irrespective ofthe mode of SL transformation and irrespective of the yejEgene deletion, repeated SL application is accompanied bya reduction in responsiveness. Thus, further studies on theimmunoregulatory mechanisms of SL are required for definingmutants that provide an optimized vaccine.

ACKNOWLEDGMENTThe authors thank Dr. B. A. Stocker, Stanford, CA, for

providing attenuated Salmonella typhimurium.

REFERENCES1. Jager E, Jager D, Knuth A. Antigen-specific immunotherapy and cancer

vaccines. Int J Cancer. 2003;106:817–820.2. Kutzler MA, Weiner DB. Developing DNAvaccines that call to dendritic

cells. J Clin Invest. 2004;114:1241–1244.3. Gilboa E, Vieweg J. Cancer immunotherapy with mRNA-transfected

dendritic cells. Immunol Rev. 2004;199:251–263.4. Lewis JJ. Therapeutic cancer vaccines: using unique antigens. Proc Natl

Acad Sci USA. 2004;101:14653–14656.5. Belardelli F, Ferrantini M, Parmiani G, et al. International meeting on

cancer vaccines: how can we enhance efficacy of therapeutic vaccines?Cancer Res. 2004;64:6827–6830.

6. Whiteside TL, Odoux C. Dendritic cell biology and cancer therapy.Cancer Immunol Immunother. 2004;53:240–248.

7. Raupach B, Kaufmann SH. Bacterial virulence, proinflammatory cytokinesand host immunity: how to choose the appropriate Salmonella vaccinestrain? Microbes Infect. 2001;3:1261–1269.

8. Pasetti MF, Levine MM, Sztein MB. Animal models paving the way forclinical trials of attenuated Salmonella enterica serovar Typhi live oralvaccines and live vectors. Vaccine. 2003;21:401–418.

9. Chabalgoity JA, Dougan G, Mastroeni P, et al. Live bacteria as the basisfor immunotherapies against cancer. Expert Rev Vaccines. 2002;1:495–505.

10. Loessner H, Weiss S. Bacteria-mediated DNA transfer in gene therapy andvaccination. Expert Opin Biol Ther. 2004;4:157–168.



11. Foster JW, Spector MP. How Salmonella survive against the odds. AnnuRev Microbiol. 1995;49:145–174.

12. Sansonetti PJ, Phalipon A. M cells as ports of entry for enteroinvasivepathogens: mechanisms of interaction, consequences for the disease process.Semin Immunol. 1999;11:193–203.

13. Finlay BB, Brumell JH. Salmonella interactions with host cells: in vitro toin vivo. Philos Trans R Soc Lond B Biol Sci. 2000;355:623–631.

14. Hueffer K, Galan JE. Salmonella-induced macrophage death: multiplemechanisms, different outcomes. Cell Microbiol. 2004;6:1019–1025.

15. Jepson MA, Clark MA. The role of M cells in Salmonella infection.Microbes Infect. 2001;3:1183–1190.

16. Levine MM, Galen J, Barry E, et al. Attenuated Salmonella typhi andShigella as live oral vaccines and as live vectors. Behring Inst Mitt. 1997;98:120–123.

17. Sirard JC, Niedergang F, Kraehenbuhl JP. Live attenuated Salmonella:a paradigm of mucosal vaccines. Immunol Rev. 1999;171:5–26.

18. Thole JE, van Dalen PJ, Havenith CE, et al. Live bacterial delivery systemsfor development of mucosal vaccines. Curr Opin Mol Ther. 2000;2:94–99.

19. Dietrich G, Spreng S, Favre D, et al. Live attenuated bacteria as vectors todeliver plasmid DNA vaccines. Curr Opin Mol Ther. 2003;5:10–19.

20. Bumann D, Hueck C, Aebischer T, et al. Recombinant live Salmonellaspp. for human vaccination against heterologous pathogens. FEMS ImmunolMed Microbiol. 2000;27:357–364.

21. Garmory HS, Leary SE, Griffin KF, et al. The use of live attenuated bac-teria as a delivery system for heterologous antigens. J Drug Target. 2003;11:471–479.

22. Bachtiar EW, Sheng KC, Fifis T, et al. Delivery of a heterologous antigenby a registered Salmonella vaccine (STM1). FEMS Microbiol Lett. 2003;227:211–217.

23. Mollenkopf H, Dietrich G, Kaufmann SH. Intracellular bacteria as targetsand carriers for vaccination. Biol Chem. 2001;382:521–532.

24. Pasetti MF, Anderson RJ, Noriega FR, et al. Attenuated deltaguaBASalmonella typhi vaccine strain CVD 915 as a live vector utilizingprokaryotic or eukaryotic expression systems to deliver foreign antigensand elicit immune responses. Clin Immunol. 1999;92:76–89.

25. Darji A, Guzman CA, Gerstel B, et al. Oral somatic transgene vaccinationusing attenuated S. typhimurium. Cell. 1997;91:765–775.

26. Paglia P, Medina E, Arioli I, et al. Gene transfer in dendritic cells, inducedby oral DNA vaccination with Salmonella typhimurium, results in protectiveimmunity against a murine fibrosarcoma. Blood. 1998;92:3172–3176.

27. Revaz V, Benyacoub J, Kast WM, et al. Mucosal vaccination witha recombinant Salmonella typhimurium expressing human papillomavirustype 16 (HPV16) L1 virus-like particles (VLPs) or HPV16 VLPs purifiedfrom insect cells inhibits the growth of HPV16-expressing tumor cells inmice. Virology. 2001;279:354–360.

28. Weth R, Christ O, Stevanovic S, et al. Gene delivery by attenuatedSalmonella typhimurium: comparing the efficacy of helper versus cyto-toxic T cell priming in tumor vaccination. Cancer Gene Ther. 2001;8:599–611.

29. Pan ZK, Weiskirch LM, Paterson Y. Regression of established B16F10melanoma with a recombinant Listeria monocytogenes vaccine. CancerRes. 1999;59:5264–5269.

30. Medina E, Guzman CA, Staendner LH, et al. Salmonella vaccine carrierstrains: effective delivery system to trigger anti-tumor immunity by oralroute. Eur J Immunol. 1999;29:693–699.

31. Pawelek JM, Low KB, Bermudes D. Tumor-targeted Salmonella asa novel anticancer vector. Cancer Res. 1997;57:4537–4544.

32. Kaisho T, Akira S. Toll-like receptors and their signaling mechanism ininnate immunity. Acta Odontol Scand. 2001;59:124–130.

33. Phalipon A, Sansonetti PJ. Shigellosis: innate mechanisms of inflammatorydestruction of the intestinal epithelium, adaptive immune response, andvaccine development. Crit Rev Immunol. 2003;23:371–401.

34. Mastroeni P. Immunity to systemic Salmonella infections. Curr Mol Med.2002;2:393–406.

35. van Kooyk Y, Geijtenbeek TB. DC-SIGN: escape mechanism forpathogens. Nat Rev Immunol. 2003;3:697–709.

36. Vazquez-Torres A, Jones-Carson J, Mastroeni P, et al. Antimicrobialactions of the NADPH phagocyte oxidase and inducible nitric oxide synthasein experimental salmonellosis. I. Effects on microbial killing by activatedperitoneal macrophages in vitro. J Exp Med. 2000;192:227–236.

37. Strauss E. Anti-immune trick unveiled in Salmonella. Science. 1999;285:307–308.

38. Shotland Y, Kramer H, Groisman EA. The Salmonella SpiC proteintargets the mammalian Hook3 protein function to alter cellular trafficking.Mol Microbiol. 2003;49:1565–1576.

39. Wick MJ, Ljunggren HG. Processing of bacterial antigens for peptidepresentation on MHC class I molecules. Immunol Rev. 1999;172:153–162.

40. al-Ramadi BK, Chen YW, Meissler JJ Jr, et al. Immunosuppressioninduced by attenuated Salmonella. Reversal by IL-4. J Immunol. 1991;147:1954–1961.

41. Huang F, Yamaguchi A, Tsuchiya N, et al. Induction of alternative splicingof HLA-B27 by bacterial invasion. Arthritis Rheum. 1997;40:694–703.

42. Qimron U, Madar N, Mittrucker HW, et al. Identification of Salmonellatyphimurium genes responsible for interference with peptide presentationon MHC class I molecules: Deltayej Salmonella mutants induce superiorCD8+ T-cell responses. Cell Microbiol. 2004;6:1057–1070.

43. Zoller M, Christ O. Prophylactic tumor vaccination: comparison ofeffector mechanisms initiated by protein versus DNA vaccination.J Immunol. 2001;166:3440–3450.

44. Fidler IJ. Biological behavior of malignant melanoma cells correlated totheir survival in vivo. Cancer Res. 1975;35:218–224.

45. Murphy GP, Hrushesky WJ. A murine renal cell carcinoma. J Natl CancerInst. 1973;50:1013–1025.

46. Barel M, Morel E. On the mechanism of glycoprotein biosynthesis in Yc8lymphoma bearing Balb/c mice. Biochimie. 1977;59:651–653.

47. Hoiseth SK, Stocker BA. Aromatic-dependent Salmonella typhimuriumare non-virulent and effective as live vaccines. Nature. 1981;291:238–239.

48. Qimron U, Madar N, Ascarelli-Goell R, et al. Reliable determination oftransposon insertion site in prokaryotes by direct sequencing. J MicrobiolMethods. 2003;54:137–140.

49. Matzinger P. The JAM test. A simple assay for DNA fragmentation andcell death. J Immunol Methods. 1991;145:185–192.

50. Yrlid U, Svensson M, Johansson C, et al. Salmonella infection of bonemarrow-derived macrophages and dendritic cells: influence on antigenpresentation and initiating an immune response. FEMS Immunol MedMicrobiol. 2000;27:313–320.

51. Igwe EI, Geginat G, Russmann H. Concomitant cytosolic delivery of twoimmunodominant listerial antigens by Salmonella enterica serovartyphimurium confers superior protection against murine listeriosis. InfectImmun. 2002;70:7114–7119.

52. Russmann H, Shams H, Poblete F, et al. Delivery of epitopes by theSalmonella type III secretion system for vaccine development. Science.1998;281:565–568.

53. Shams H, Poblete F, Russmann H, et al. Induction of specific CD8+memory T cells and long-lasting protection following immunization withSalmonella typhimurium expressing a lymphocytic choriomeningitisMHC class I-restricted epitope. Vaccine. 2001;20:577–585.

54. Shen H, Miller JF, Fan X, et al. Compartmentalization of bacterialantigens: differential effects on priming of CD8 T cells and protectiveimmunity. Cell. 1998;92:535–545.

55. Mittrucker HW, Kaufmann SH. Immune response to infection withSalmonella typhimurium in mice. J Leukoc Biol. 2000;67:457–463.

56. Lo WF, Ong H, Metcalf ES, et al. T cell responses to Gram-negativeintracellular bacterial pathogens: a role for CD8+ T cells in immunity toSalmonella infection and the involvement of MHC class Ib molecules.J Immunol. 1999;162:5398–5406.

57. Houde M, Bertholet S, Gagnon E, et al. Phagosomes are competentorganelles for antigen cross-presentation. Nature. 2003;425:402–406.

58. Gregorian SK, Battisto JR. Immunosuppression in murine renal cellcarcinoma. I. Characterization of extent, severity and sources. CancerImmunol Immunother. 1990;31:325–334.

59. Gerard CM, Bruyns C, Delvaux A, et al. Loss of tumorigenicity andincreased immunogenicity induced by interleukin-10 gene transfer in B16melanoma cells. Hum Gene Ther. 1996;7:23–31.

60. Sensi M, Bergomi M, Formelli F, et al. Eradication of a disseminatedmouse lymphoma by 1,3-bis(2-chloroethyl)-1-nitrosourea is immunolog-ically mediated and accompanied by de novo generation of anti-tumorcytotoxicity. Int J Cancer. 1990;46:1088–1094.

61. Irvine KR, Chamberlain RS, Shulman EP, et al. Enhancing efficacy ofrecombinant anticancer vaccines with prime/boost regimens that use twodifferent vectors. J Natl Cancer Inst. 1997;89:1595–1601.

62. Nasz I, Adam E, Lengyel A. Alternate adenovirus type-pairs for a pos-sible circumvention of host immune response to recombinant ade-novirus vectors. Acta Microbiol Immunol Hung. 2001;48:143–146.



Tumor Vaccination by Salmonella typhimurium After ...

Documents

Transcript of Tumor Vaccination by Salmonella typhimurium After ...