Dendritic cells · Tumor vaccines · Peptide vaccines · Vaccine adjuvants. ABSTRACT ..... together, these findings support that the effect of DC-peptide antitumor ...
Concise Review
Bone Marrow-Derived Dendritic Cells Serve as Potent Adjuvants for Peptide-Based Antitumor Vaccines JOSE I. MAYORDOMO, TATIANA ZORINA, WALTER J. STORKUS, LAURENCE ZITVOGEL, MARIA D. GARCIA-PRATS, ALBERT B. DELEO, MICHAEL T. LOTZE University of Pittsburgh Cancer Institute, Pittsburgh, Pennsylvania, USA Key Words. Dendritic cells · Tumor vaccines · Peptide vaccines · Vaccine adjuvants
A BSTRACT Dendritic cells (DCs) are considered the most effective antigen-presenting cells (APCs) for primary immune responses. Since presentation of antigens to the immune system by appropriate professional APCs is critical to elicit a strong immune reaction and DCs seem to be quantitatively and functionally defective in the tumor host, DCs hold great promise to improve cancer vaccines. Even though they are found in lymphoid organs, skin and mucosa, the difficulty of generating large numbers of DCs has been a major limitation for their use in vaccine studies. A simple method for obtaining DCs from mouse bone marrow cells cultured in the presence of GM-CSF + interleukin 4 is now available. In four different tumor models, mice injected with DCs grown in GM-CSF plus
interleukin 4 and prepulsed with a cytotoxic T lymphocyte-recognized tumor peptide epitope developed a specific cytotoxic T lymphocyte response and were protected against a subsequent tumor challenge with tumor cells expressing the relevant tumor antigen. Moreover, treatment of day 5-14 tumors with peptidepulsed DCs resulted in sustained tumor regression in five different tumor models. These results suggest that presentation of tumor antigens to the immune system by professional APCs is a promising method to circumvent tumor-mediated immunosuppression and is the basis for ongoing clinical trials of cancer immunotherapy with tumor peptide-pulsed DCs. Stem Cells1997;15:94-103
INTRODUCTION Dendritic cells (DCs) were first described by Steinmann and Cohn in 1973 [1]. Cells with “veiled” morphology (that is, round cells with extensive cytoplasmic processes) on light microscopy and electron microscopy were first observed in the spleen and peripheral lymph nodes of mice. Soon, similar cells were found in several organs including skin (where they had been previously termed Langerhans cells) [2], Peyer’s patches [3] and liver [4]. A long search for surface markers present on these cells has been only partially successful [5]. A pattern of markers characteristic of DCs was soon defined: CD1a+CD3–CD4–CD8–CD14–CD20–CD40+CD80+CD86+ for human DCs and CD3–CD4–CD8–B220–CD40+CD80+CD86+ for mouse DCs. Still, no single marker is specific for DCs. Several groups have subsequently attempted to isolate monoclonal antibodies directed against surface markers expressed exclusively by DCs and not by other cell types. The most promising antibodies include 33D1, the anti-interdigitating cell antigen NLDC 145 and the anti-CD11c integrin N418.
It is to be remembered, though, that while most DCs stain at least with one of these monoclonal antibodies, many are organ-specific: i.e., 33D1 stains splenic DCs but not DCs present in other organs, and most stain either mouse or rat DCs but not human DCs. There is an urgent need for monoclonal antibodies directed to human DCs. At the moment, DCs are phenotypically defined with multiple-marker panels, based mostly on absence of staining with markers specific for other cell lineages. This situation is not ideal and certainly identification of surface markers specific for DCs is critical not only to improve identification of DCs, but to better differentiate subsets of DCs with important functional differences. FUNCTION OF DCS Soon after the first description of DCs as a separate cell type it was noticed that these cells are the most potent stimulators of the allogeneic mixed leukocyte reaction [6]. When used as stimulators, numbers of splenic DCs as low as 103 were able to induce [3H]thymidine uptake levels comparable
Correspondence: Dr. Jose I. Mayordomo, Servicio de Oncologia Medica, Hospital Clinico Universitario, Avenida Gomez Laguna s/n, 50009, Zaragoza, Spain. Accepted for publication June 12, 1996. ©AlphaMed Press 1066-5099/97/$5.00/0
STEM CELLS 1997;15:94-103
95
to those obtained with 106 unpurified spleen cells. The allogeneic mixed leukocyte reaction is still a useful simple test to measure function of DCs. This observation suggested that DCs are antigen-presenting cells (APCs). In fact, DCs were subsequently shown to present antigens to B and T lymphocytes [7]. DCs are now considered to be potent professional APCs which are essential for priming of B and T lymphocyte-mediated immune responses. DCs are capable of active pinocytosis of exogenous antigens. Intracellularly, antigens are processed to short peptides that enter the assembly chain of the class I and class II molecules of the major histocompatibility complex (MHC). Peptides capable of being recognized by T lymphocytes (both MHC class I-restricted cytotoxic T lymphocytes and MHC class II-restricted helper T lymphocytes) are presented on MHC molecules on the cell surface of DCs. Moreover, costimulatory molecules, which are essential for priming of naive lymphocytes, are expressed at high levels on DCs. The list of costimulatory molecules expressed on DCs includes CD80 (B7-1), CD86 (B7-2) and CD40 [8, 9]. Lymphocytes are preferentially primed by antigens presented to the immune system in the presence of costimulatory molecules. This provides a protection against autoimmunity, since autoantigens presented by nonprofessional APCs expressing low levels of costimulatory molecules will not generate a strong immune response. The distinction between professional and nonprofessional APCs is only a part of the picture. In fact, the immune response against a given antigen may be preferentially a cell-mediated response (what is also termed a Th1 response) or a preferentially antibody-mediated response (what is also termed a Th2 response). Which APC presents the antigen to the immune system is crucial to drive the immune response towards a Th1 or Th2 pattern. For most pathogens, one of these patterns (i.e., Th1 for Listeria monocytogenes) results in protective immunity while the other results in progressive lethal infection. DCs are found in multiple organs. It seems that DCs residing in the skin (Langerhans cells) or mucosae are able to process exogenous antigens. These DCs have been termed “immature” [10]. After antigen uptake in the context of aggression (i.e., infected wound), possibly under the influence of tumor necrosis factor-α (TNF-α), DCs undergo changes (“maturation”) including downregulation of protein antigen processing capacity [11], and migrate to the regional lymph nodes where they present the antigen to T lymphocytes. “Mature” and “immature” are operative terms to describe the fact that DCs in a given moment and site exhibit preferentially either the ability to process antigens or the ability to migrate and prime T cells.
Dendritic Cells for Cancer Vaccines PURIFICATION OF DCS In the years subsequent to the description of DCs by Steinmann et al. [1] methods to isolate DCs from the spleen [1], skin [2] and liver [4] were developed. These methods involve depletion of cell types other than DCs with cocktails of lineage-specific antibodies and also exploit the ability of DCs and macrophages to adhere to plastic plates. Separation between DCs and macrophages is based on the fact that DCs detach from plastic within 24 h while macrophages remain attached longer. The yields of these methods are low (typically, in the order of 105 DCs per animal). Methods to isolate human DCs from peripheral blood mononuclear cells have also been described [12]. They are time-consuming and the yield of DCs is low (0.08% to 0.4% of the starting mononuclear cell population or 2 to 32 × 106 DCs per patient) [13]. CULTURE OF DCS The finding by Koch et al. [14] that the addition of GM-CSF plus TNF-α to a culture of human CD34+ hematopoietic progenitors purified from the peripheral blood induced the growth of cells with phenotype and functional features typical of DCs was a major breakthrough in the knowledge of the ontogeny and maturation of DCs, confirming the derivation of DCs from hematopoietic stem cells, and facilitating the use of DCs as immunogens. Cultured cells were found to be as active as DCs purified from other sources in stimulating the allogeneic mixed leukocyte reaction. Several groups have subsequently described different methods to culture large numbers of human, mouse or rat DCs derived from normal tissues in cytokine combinations. Inaba et al. [15] described a method to culture mouse DCs from bone marrow (BM) by immunodepleting lymphocytes, macrophages and granulocytes and culturing the remaining cells in GM-CSF for seven days. The resulting cultured cells included large numbers of newly generated granulocytes that had to be eliminated by repeated washes and finally by sedimentation to obtain purified DCs. Since Steinmann et al. [1] had previously found no DCs in mouse bone marrow both by morphology and by allogeneic mixed leukocyte reaction, the results obtained by Inaba et al. [15] support that DCs derive from hematopoietic stem cells. Sallusto and Lanzavecchia [10] described the culture of human DCs from peripheral blood mononuclear cells in the presence of either GM-CSF, GM-CSF + TNF-α or GM-CSF + interleukin 4 (IL-4). The phenotype and function of the cultured cells varied dramatically depending on the culture conditions. Cells cultured in GM-CSF + IL-4 were found to be the most potent APCs as measured by stimulation of the
Mayordomo, Zorina, Storkus et al. allogeneic mixed leukocyte reaction and priming of naive B lymphocytes against tetanus toxoid. The basic role of TNF-α in DC cultures appears to be the inhibition of granulocyte maturation. However, cells cultured in GM-CSF + TNF-α still include a significant proportion of macrophages. In addition, TNF-α promotes “maturation” of DCs to a state where they lose ability to process antigens but are very active in stimulating naive lymphocytes. IL-4 inhibits both granulocyte and macrophage development, and seems to keep DCs in an “immature” state, thus more capable of processing exogenous antigens. IL-13 has similar properties to IL-4. Romani et al. [16] described the culture of large numbers of human DCs from peripheral blood mononuclear cells obtained during recovery from cancer chemotherapy-induced myelosuppression. Mononuclear cells in these conditions contain a proportion of hematopoietic progenitors much higher than mononuclear cells obtained in steady state. When these mononuclear cells were cultured in GM-CSF alone, few DC colonies developed. However, culture in GM-CSF + IL-4 resulted in the generation of large numbers of phenotypically and functionally typical DCs. The fact that the yield of DCs is much higher using mononuclear cells obtained during recovery from myelosuppression than in steady state supports the theory that most DCs derive from hematopoietic progenitors. Cells indistinguishable from DCs have also been shown to derive from CD14+ monocytes cultured in GM-CSF + IL-4. A recent report by Caux et al. [17] resolved this apparent controversy (that is, do DCs generate directly from early hematopoietic progenitors or do they derive from blood monocytes?). They demonstrated the existence of two independent pathways of DC development. When CD34 + hematopoietic progenitors from human cord blood are cultured in the presence of GM-CSF + TNF-α, two sets of DC precursors emerge by day 5-7. The first subset, identified by the expression of CD1a and the absence of CD14, will mature into DCs with features of epidermal Langerhans cells (expression of Birbeck granules, Lag antigen and E-cadherin). The second subset, which is CD14+CD1a–, will mature into CD1a+ DCs with the same typical morphology, phenotype and ability to stimulate the allogeneic mixed leukocyte reaction, but lacking Birbeck granules, Lag antigen and E-cadherin. Interestingly, the CD14+ precursors but not the CD1a+ precursors represent bipotent cells that can be induced to differentiate, in response to macrophage-colony stimulating factor, into macrophage-like cells. Most groups studying DCs as APCs now use culture in GM-CSF + IL-4 to generate human DCs. However, culture of mouse BM-derived DCs in GM-CSF + TNF-α or GM-CSF + IL-4 was only recently described by Zorina et al. [18] (Table 1). Mouse BM cells are harvested from the femur and
96
tibia, cultured in the presence of GM-CSF (103 U/ml) plus either IL-4 (103 U/ml) or TNF-α (102 U/ml). By day 8, large numbers of cell aggregates with the characteristic DC “veiled” morphology have proliferated. The yield of cultured cells at day 8 from each mouse is approximately 10 × 106 (culture with GM-CSF + IL-4), 6 × 106 (GM-CSF + TNF-α) and 5 × 106 cells (GM-CSF alone). Antigens expressed on these DCs as assessed by flow cytometry include CD45, CD44, CD11b (Mac-1), CD18, CD40, CD80(B7-1), CD86(B7-2), the interdigitating cell antigen NLDC 145 and class I and class II MHC antigens. The levels of MHC class II and the costimulatory molecules B7-1 and B7-2 were found to be higher in cells cultured with GM-CSF + IL-4 than in those cultured with GM-CSF + TNF-α or GM-CSF alone (Table 1). No evidence of B220, Thy 1.2, CD4 and CD8 expression was detected. The phenotype of BM-derived DC is similar in many aspects to that seen in splenic and epidermal DC, with the exception of a high level of the CD11b integrin, in agreement with the findings of Inaba et al. [14]. DCs grown in GM-CSF + IL-4 were more potent stimulators of the allogeneic mixed leukocyte reaction than cells grown in GM-CSF + TNF-α or GM-CSF alone. These results are in agreement with those reported by Sallusto and Lanzavecchia [10] in human DC cultured from peripheral blood mononuclear cells. Groups interested in DC maturation are now investigating the effect of other cytokines on DC generation. As for the effects of adding additional cytokines to cultures in GM-CSF + IL-4, stem cell factor has been shown to increase the size of DC colonies and the yield of DCs with no apparent functional modification [19]. In humans, addition of stem cell factor (kit ligand) or flt-3 ligand to cultures of peripheral blood CD34+ hematopoietic progenitors supplemented with GM-CSF + TNF-α has been shown to increase the yield of DCs [20]. As
Table 1. Comparison of phenotype and function (stimulation of the allogeneic mixed leukocyte reaction) of BM-derived DC cultured in GM-CSF, GM-CSF + TNF-α and GM-CSF + IL-4 Surface Ag.1
GM-CSF
GM-CSF + TNF-α
GM-CSF + IL-4
CD45
282
277
283
Ia (cl. II MHC)
168
139
201
CD80 (B7-1)
40
37
49
CD86 (B7-2)
46
40
124
38 ± 3
59 ± 4
97 ± 4
10,000
3,000
1,000
MLR stim.2 3
DC 50% stim. 1
Expression of surface antigens was measured as mean fluorescence channel. Stimulation of the allogeneic mixed lymphocyte reaction is measured by the 2 maximum [3H]thymidine uptake (counts per million) ± standard deviation. 3 Number of DCs needed to obtain 50% of the maximum [ 3H]thymidine uptake. 2
97
for the effect of cytokines administered in vivo to mice before BM harvest for DC culture, flt-3 ligand has been recently found to markedly increase the yield of DCs [21]. WHY USE DCS AS VACCINE ADJUVANTS? The concept of vaccine adjuvant has been developed empirically. The biologic basis of the effect of adjuvants has been studied only very recently. In the late nineteenth and early twentieth centuries, attempts to protect humans against infectious diseases by injection of inactivated or attenuated forms of the causative agent had only been successful in some cases. Injection of the vaccinia virus with no adjuvants protected humans against smallpox. However, for many other infectious diseases, injection of the pathogenic organism by itself induced no protection. It was tested whether emulsion of the organism in a dense substance such as alum would induce prolonged release of the antigens of the germ to the dermis thus enabling the immune system to mount a stronger immune response. When serological tests were developed to detect a specific antibody-mediated immune response, it was indeed found that immunization with antigens emulsified in alum or other adjuvants induced antibody titers which were higher than those induced with antigen alone. Route of immunization was also found to be critical for antibody titers: s.c. or i.m. injections resulted in antibody titers much higher than the i.p. or i.v. routes. When a better understanding of the role of the cellmediated arm of the immune system was achieved, it was found that some adjuvants induced preferentially an antibody-mediated response (termed also Th2 pattern), others a cell-mediated (Th1) response and still others induced a mixed pattern. Aichele et al. [22] recently studied the importance of route and dose of antigen for vaccines consisting of short peptides (capable of being recognized by T lymphocytes) emulsified in adjuvants. The model peptide was GP (33-41), derived from the lymphocytic choriomeningitis virus, and the adjuvant was incomplete Freund’s adjuvant. They found that for a dose of peptide in the range of 50-500 µg, the s.c. route induced a specific T lymphocyte-mediated immune reaction, while the i.p. route resulted in antigen-specific tolerance. However, such phenomenon was more complex when the dose of antigen was titrated for each route; tolerance could also be induced by the s.c. route if sufficiently high doses of antigen were injected. The mechanism postulated by Aichele et al. to explain these results is that low doses of antigen injected by the s.c. route are uptaken preferentially by dermal DCs (Langerhans cells), which are professional APCs expressing high levels of costimulatory molecules, and will prime T lymphocytes,
Dendritic Cells for Cancer Vaccines thus inducing a strong specific cell-mediated immune reaction. In contrast, when the antigen is injected in the peritoneal cavity, where the numbers of professional APCs are very low, nonprofessional APCs lacking adequate levels of costimulatory molecules are the presenters to the T lymphocytes. Presentation of antigens in the absence of adequate levels of costimulatory molecules induces antigen-specific tolerance. When extremely high doses of antigen are injected s.c., the ability of Langerhans cells to uptake them is overwhelmed and most of the antigen reaches nonprofessional APCs instead, thus inducing tolerance. An obvious implication of this hypothesis is that immunization with DCs prepulsed ex vivo with tumor antigens would represent an optimal means to elicit a strong immune response. When dealing with cancer vaccines it has to be assumed that the cancer host has been contacting whatever foreign antigens may be expressed by the tumor for a long time. The obvious question is what is the reason that a successful immune response has not been mounted. How can we present tumor antigens with more efficacy than the tumor cells themselves? Porgador and Gilboa [23] compared different strategies to immunize with peptides capable of being recognized by cytotoxic T lymphocytes. They found that immunization with DCs pulsed with peptide is superior to injection of peptide emulsified in adjuvant (incomplete Freund’s adjuvant) as measured by specific lysis in a standard cell-mediated cytotoxicity assay. Studies on tumor-mediated immunosuppression have started to characterize the reasons why the host fails to mount an adequate immune response to the tumor. Defects in the ζ chain of the T cell receptor have been found [24]. In animal models, BM-derived professional APCs (a concept roughly equivalent to DCs) were found to be critical for induction of an effective immune response in the tumor host by Huang et al. [25]. Their results suggest that tumor cells themselves, being nonprofessional APCs, are not adequate to prime the immune system against tumor antigens. Ideally, these antigens should be presented by DCs. These studies have resuscitated the interest to quantify the number of DCs found within tumors. A number of early studies [26-28] reported a decreased number of DCs infiltrating tumors when compared to normal tissues. This fact has been reported for lung cancer [26], laryngeal carcinoma [27] and nasopharyngeal carcinoma [28]. A recent study found that DCs isolated from tumors were also functionally impaired when compared to DCs from normal donors as measured by stimulation of the allogeneic mixed leukocyte reaction [29]. IL-10 may be critical for tumor-mediated suppression of DC function in vivo [30]. DCs from cancer patients obtained by culture of mononuclear cells obtained
Mayordomo, Zorina, Storkus et al.
98
from cancer patients are functionally comparable to those cultured from normal donors. The findings of the previous studies can be summarized as follows: A) DCs are critical to elicit a strong immune reaction against tumor antigens and B) DCs in the cancer host are both functionally impaired and decreased in number within the tumor. In the light of these findings, immunization with autologous DCs cultured ex vivo (and thus functionally intact) prepulsed with tumor antigens is a promising strategy to elicit a strong immune response in the cancer host [31].
Table 2. Tumor antigens recognized by class I MHC-restricted cytotoxic T lymphocytes in human and murine tumors Antigen
Tumor
Tissue expression
Human tumors Mage-1 Mage-3 MART-1/Melan A Tyrosinase gp100 gp75 Her-2/neu
Melanoma, lung, etc. Melanoma, lung, etc. Melanoma Melanoma Melanoma Melanoma Ovarian CA
Testicle Testicle Melanocytes Melanocytes Melanocytes Melanocytes Multiple
Mouse tumors P1A Mut Connexin 37 Mutated p53
P815 Mastocytoma Lewis Lung CA Meth A sarcoma
Testicle Lung None
DCS AS TUMOR VACCINES Tumor antigens recognized by T lymphocytes have been identified in human and mouse tumors [32, 33] (Table 2). The antigens identified in mouse tumors provide a useful model to design effective methods of preventive and therapeutic immunization to be subsequently tested in human clinical trials. We directly assessed the ability of vaccines consisting of DCs carrying tumor peptides to elicit protective antitumor immune responses in murine models. Day 8 BM-derived DCs were pulsed with a synthetic tumor-associated peptide (Table 3),
Table 3. Synthetic peptides used in DC-based tumor peptide vaccination studies Peptide
H-2 Restriction
Expressed by tumor
E7(49-57)
Db
C3 Sarcoma
OVA (257-264)
Kb
M05 Melanoma
b
Lewis Lung CA
*
MUT1 (Connexin37 )
K
234CM (mut. p53*)
Kd
*
Meth A Sarcoma
The MUT1 and 234CM peptides derive from mutated proteins expressed in a tumor cell line.
Table 4. Immunization with tumor peptides pulsed onto DCs protect mice from a subsequent tumor challenge Vaccine
C3
M05
Lewis
Meth A
DC + no peptide
155
99
166
214
DC + irrel. peptide
148
93
152
197
4
0
0
7
DC + tumor peptide
Naive C57/BL6 or balb/c mice were immunized on days 0 and 7 with syngeneic BMderived DCs pulsed with synthetic tumor-associated peptide, with irrelevant peptide or with no peptide. Vaccinated animals were then challenged with either C3, M05, Lewis lung carcinoma or Meth A tumors on day 14. Results represent tumor size (mm2) on day 30 post-tumor inoculation.
washed extensively to remove unbound peptide and injected i.v. in the tail vein of syngeneic mice on days 0 and 7. Animals were then challenged on day 14 with relevant or irrelevant tumors injected s.c. and tumor progression was monitored [34]. As shown in Table 4, animals vaccinated in this manner were resistant to challenge with tumor cells expressing the relevant tumor antigen in each of four different tumor models, including a transfected antigen of viral origin (E7 (49-57), derived from the type 16 human papilloma virus, expressed on the C3 cell line) [35]), a transfected xenogeneic antigen (OVA (257-264), from chicken ovalbumin, expressed on the M05 melanoma) [36] and two mutated antigens naturally expressed on tumor cells (MUT1, expressed by the Lewis lung carcinoma [37] and the 234CM peptide, derived from a mutated form of p53, expressed by the Meth A sarcoma [38, 39]). In comparative experiments, we found that DCs cultured in GM-CSF + IL-4 were significantly better immunogens in vivo than DCs grown in GM-CSF + TNF-α, DCs grown in GM-CSF alone or skin-derived DCs (Langerhans cells) [34]. Of note, the observed efficacy of DC-peptide vaccines was directly correlated to the ability of the vaccine to prime in vivo splenic antitumor cytotoxic T lymphocytes [34, 40] (Table 5). Previous reports on antitumor DC-based vaccines in murine models [41-44] include data on immunization with semipurified DCs pulsed with tumor extracts or with DCs purified from the lymph nodes draining the tumor (presumably “pulsed in vivo” with tumor antigens). Protection against a subsequent tumor challenge was seen in these studies. Flamand et al. [45] reported successful protection against a B cell lymphoma by prior immunization with splenic DCs pulsed with tumor-specific idiotype protein (the monoclonal surface immunoglobulin expressed by each B cell lymphoma). The availability of simple methods to culture large numbers of DCs in mice and humans, and the identification of short peptides as the antigens recognized by T lymphocytes has allowed us to extend these earlier reports and to dissect the immune response induced by DC-based cancer vaccines.
Dendritic Cells for Cancer Vaccines
99
Table 5. In vivo generation of specific antitumor cytotoxic T lymphocytes by immunization with DCs pulsed with tumor-associated peptides Vaccine
% Specific lysis (E:T = 33:1)
% Specific lysis (E:T = 100:1)
Saline serum
2
1
Langerhans cells + OVA(257-264)
0
0
24
41
Dendritic cells + OVA (257-264)
Mice were immunized twice with DCs pulsed with the OVA(257-264) peptide, Langerhans cells pulsed with the same peptide or saline serum as described in Table 4. Spleens were harvested on day 14 and restimulated in vitro with the ovalbumin-transfected E.G7 thymoma. Data show percentage of specific lysis of the E.G7 thymoma in a standard cytotoxicity assay. The nonovalbumin expressing EL-4 thymoma was not lysed by splenocytes from mice immunized with DCs pulsed with OVA(257-264).
DCS AS TUMOR THERAPIES A critical test of the efficacy of antitumor DC-mediated immunizations is whether they can only prevent the growth of tumor cells injected after a set of DC-mediated immunizations, or if they can also suppress the growth of tumors present in the host before DC-mediated immunization, a situation more similar to the treatment of cancer patients. In order to test the efficacy of DC-mediated vaccines on the growth of established tumors, mice were inoculated s.c. with twice the minimum tumorigenic dose of a given tumor, and the tumors were allowed to grow for 7-14 days. Day 8 BM-derived DCs cultured in GM-CSF + IL-4 were pulsed with tumor peptides as described for protection experiments [34] and injected i.v. in the tail vein of tumor-bearing animals every four to seven days. This therapy cured 80%-100% of C57/BL6 mice bearing day 14 C3 tumors or day 7 Lewis lung carcinomas (Table 6). Tumor stasis and subsequent regression was typically seen 7-10 days after the first immunization. This Table 6. Therapy of established tumors with DCs pulsed with tumor peptides % animals with palpable tumor on day 30 Therapy
C3
Lewis Lung CA
DC + no peptide
100
100
DC + irrelevant peptide
100
100
10
20
DC + tumor peptide
C57/BL6 mice were injected with tumor cells and tumors allowed to progress for 7 (Lewis lung carcinoma) or 14 (C3) days. Tumor-bearing animals were then treated with syngeneic DCs pulsed with either no peptide, irrelevant tumor peptide, or relevant tumor peptide every four days. Results are reported as the percentage of animals with palpable tumor on day 30 post-tumor challenge.
observation suggests that the first one or two DC-peptide vaccines are critical for the observed antitumor effect. This is in agreement with the critical role of DCs for generation of primary immune responses. All disease-free animals were subsequently shown to specifically reject rechallenge with relevant tumor but to display progressive outgrowth of irrelevant tumors. The effectiveness of DC-peptide therapy was also evaluated for later-stage tumors in the C3 model. Sustained tumor regression was observed in 60% of the animals in a day 21-established C3 model and in 20% of cases in a day 28 C3 model (data not shown). The potential impact of this strategy on human cancer therapy when appropriate targets such as mutated oncogenes are identified is exemplified by the efficacy of vaccines consisting on DCs pulsed with peptide epitopes derived from p53 on established murine sarcomas [39]. Synthetic short peptides capable of being recognized by cytotoxic T lymphocytes are only one of the immunogens that can be presented onto DCs. Our group [46] and others [47] have successfully identified peptides recognized by cytotoxic T lymphocytes from peptide mixtures eluted from the surface of tumor cells. The method developed by Storkus et al. [48] involves only a short incubation of tumor cells in pH 3.0-3.3 phosphate-acetate buffer to denature the surface MHC class I-peptide complexes. When pulsed onto appropriate target cells, the peptide mixture can be recognized by tumor-specific T lymphocytes. We set out to determine whether immunization with DCs pulsed with tumor-derived acid eluted peptide mixtures was able to generate an antitumor immune response capable of suppressing the growth of established tumors. Since few antigens recognized by T lymphocytes have been identified for human tumors other than melanoma, this strategy was designed as a model of vaccine which may be applied to tumor histologies where tumor antigens and peptide epitopes have not yet been identified. Peptides were extracted by elution in pH 3.3 acid buffer of single-cell suspensions derived from tumors grown in vivo in syngeneic animals, concentrated and pulsed onto day 8 BM-derived DCs cultured in GM-CSF + IL-4. Repeated i.v. injection of these DCs in mice bearing either the C3 tumor, the MCA 205 sarcoma, the CL8.1 melanoma or the TS/A mammary carcinoma induced transient tumor stasis, but no cures [49] (Table 7). The exception was the day 14 C3 tumor model in which all mice were cured. Therapeutic efficacy of these vaccines was tumor antigen-specific since vaccines composed of DCs pulsed with irrelevant tumor peptides were ineffective. Inoculated DCs rapidly localized to the spleen, liver and inguinal lymph nodes of the treated animals (data not shown). Elevated production of interferon-γ was observed in the spleen and lymph nodes of the animals receiving DC pulsed with the relevant peptide but not in those receiving DC alone or DCs pulsed with irrelevant
Mayordomo, Zorina, Storkus et al.
100
Table 7. Therapy with DCs pulsed with unfractionated tumor-derived peptides inhibits tumor progression in tumor-bearing mice Mean Tumor Size (mm2) Therapy
C3
MCA205
CL8-1
TS/A
DC alone
99
248
183
152
DC + spleen peptide
102
242
178
155
DC + tumor peptide
7
45
39
42
Animals were inoculated with twice the minimum tumorigenic dose of the C3 tumor, the MCA 205 sarcoma, the CL8-1 melanoma or the TS/A mammary adenocarcinoma. After five to seven days (MCA205, CL8-1, TS/A) or 14 days (C3), syngeneic DCs pulsed with no peptide, normal splenic peptides or relevant tumor peptides were injected i.v. in the tail vein. Immunization was repeated every four days. Results are reported in mean tumor size 30 days post-tumor inoculation.
peptides. A significant tumor infiltration by CD4 and CD8 lymphocytes and by S100+ cells (presumably DCs) was seen only in animals receiving DCs pulsed with relevant peptides. The therapeutic effect of the vaccines is critically dependent on interferon-γ, TNF-α, IL-12 and B7-mediated costimulation as demonstrated in blocking experiments [49]. Taken together, these findings support that the effect of DC-peptide antitumor vaccines is mediated through the induction of a strong Th1-associated cell-mediated immune reaction. An alternative approach to immunize mice against a tumor for which no defined tumor antigens have been identified was recently described by Boczkowski et al. [50]. Mice were vaccinated with DCs pulsed with RNA from an ovalbumin-expressing tumor. Specific cytotoxic T lymphocytes were generated and mice were protected against a subsequent tumor challenge. DC-BASED TUMOR IMMUNOTHERAPY IN HUMANS The therapeutic efficacy of DC-peptide immunization on established tumors in murine models is impressive. Two other recent preclinical studies complete the picture. One reports successful protection against a poorly immunogenic tumor by immunization with peptide-pulsed RMA-S cells, but only when these cells had been transfected with the CD80 (B7-1) costimulatory molecule [51]. The other demonstrates regression of established murine tumors when the gene encoding the tumor antigen is transfected onto Listeria monocytogenes, a very immunogenic bacteria capable of attracting professional APCs in vivo [52]. Taken together, these studies support that presentation of tumor antigens by professional APCs is a critical step to circumvent tumor-induced immunosuppression. At the moment, DCs cultured from peripheral blood in GM-CSF + IL-4 are certainly the most readily available professional APCs to test this hypothesis in humans. However, it must be emphasized that in preclinical models, very advanced tumors cannot be successfully treated with
DC-based immunotherapy by itself, so the immune defect present in late-stage cancer patients must certainly be more complex than inadequate presentation of tumor antigens, probably involving defects in T lymphocytes themselves [24]. Beyond these arguments, the impact of immunotherapy with DCs pulsed with tumor antigens in human cancer therapy can only be determined by adequate clinical trials. Supported by the encouraging mouse experiments shown here, many groups are currently designing trials for cancer patients involving DCs. Hsu et al. at Stanford pioneered DC-based immunotherapy of cancer in humans [13]. Autologous DCs were purified from the peripheral blood of four patients with follicular lowgrade B cell lymphoma previously treated with chemotherapy by leukapheresis and density gradient centrifugation. The tumor antigen pulsed onto DCs was the monoclonal surface immunoglobulin (idiotype protein) present in each patient’s lymphoma. Idiotype-pulsed DCs and DCs pulsed with KLH, an immunogenic protein recognized by T lymphocytes, were injected i.v. on days 1, 29, 57 and 150. Subcutaneous injections of idiotype protein and KLH were given 14 days after each DC injection to boost the primary response induced by the DC infusion. The number of DCs infused ranged from 2 to 32 million DCs per treatment (median 5 million). All patients developed measurable antitumor cellular immune responses as determined by peripheral blood mononuclear cell proliferative responses against tumor idiotype protein. These responses, which were absent prior to immunization, appeared after one to two vaccinations and were specific for the autologous tumor idiotype. Meaningful clinical responses have been seen with one partial response, one minor response and one disease stabilization in three patients with progressive measurable disease, and one complete response in a fourth patient with minimal disease detectable only by polymerase chain reaction. All patients have remained progression-free for up to two years. These results have been recently updated [53]. It is remarkable that in previous clinical trials at the same institution [54], immunization with idiotype protein alone or emulsified in adjuvants had consistently failed to induce regression of lymphoma, even though a detectable specific immune response was generated. The task of identifying the idiotype protein for each patient with B cell lymphoma is enormous. However, easier targets for DC-based immunotherapy are already available. Tumor antigens shared by tumors of the same histology have been identified for melanoma (Table 2) [32, 33]. A prerequisite for in vivo clinical trials is to check whether lymphocytes from melanoma patients, immunosuppressed by the tumor, are still able to generate specific antitumor cytotoxic T lymphocytes when stimulated in vitro with the specific peptide presented by autologous DCs. Our group [55] and others [56] have
Dendritic Cells for Cancer Vaccines
101
recently documented the ability to elicit human antimelanoma HLA-A2 cytotoxic T lymphocytes in vitro from healthy donors by coculture of autologous peripheral blood lymphocytes with autologous DCs pulsed with melanoma peptides. Our group has also succeeded in generating cytotoxic T lymphocytes from melanoma patients [55]. A list of peptides for which specific cytotoxic T lymphocytes could be generated in vitro for a majority of melanoma patients and are thus candidates for clinical trials of in vivo immunization is shown in Table 8. The culture techniques previously described can be used to generate large numbers of DCs from cancer patients from peripheral blood mononuclear cells obtained in steady state or during stem cell mobilization with chemotherapy and/or cytokines. Siena et al. [20] have consistently generated 4 × 1010 DCs from mononuclear cells obtained by leukapheresis from cancer patients after treatment with high-dose cyclophosphamide followed by colony-stimulating factors. This number is 10,000-fold higher than the amount obtained by Hsu et al. [13] by density gradient centrifugation. All major cancer centers are already routinely performing leukapheresis to harvest stem cells to support high-dose chemotherapy. This technology can now be used to generate DCs for cancer immunotherapy. The challenge now is to test whether immunization with autologous DCs pulsed with already identified tumor antigens, such as those identified in melanoma, has a discernible antitumor effect in cancer patients. Minimal residual disease may be more responsive than macrometastatic disease. Clinical trials aimed at addressing this question are currently underway at the University of Pittsburgh and in other American and European cancer centers. FUTURE DEVELOPMENTS IN DC-BASED VACCINES The exquisite sensitivity of DCs to the culture conditions, with small changes resulting in major phenotypic and functional differences, leads us to suggest that the best possible conditions to culture DCs for cancer immunotherapy may not have been defined yet, and this is now an area of intense research. In addition to GM-CSF, IL-4 and TNF-α, other cytokines found to increase yield of human DCs include IL-12 (Storkus et al., manuscript in preparation), flt-3 ligand and kit ligand [20]. Multiple cytokines added to human DC cultures during the last 24 hours result in altered phenotype and/or enhanced stimulation of the allogeneic mixed leukocyte reaction. Increasing interest is being focused on IL-12, a cytokine which is secreted by DCs upon interaction with CD4+ T helper lymphocytes through ligation of the costimulatory molecule CD40 [57, 58]. IL-12 is crucial for the development of Th1-type cell-mediated immune responses.
Table 8. Successful generation of HLA-A2 restricted antimelanoma cytotoxic T lymphocytes in vitro derived from peripheral blood lymphocytes from melanoma patients stimulated with peptide-pulsed autologous DCs Melanoma peptide
Sequence
Number of patients
MART-1 (27-35)
AAGIGILTV
9/10
MART-1 (32-40)
ILTVILGVL
9/10
gp 100 (280-288)
YLEPGPVTA
8/10
Tyrosinase (368-76)
YMDGTMSQV
6/10
Results indicate the number of melanoma patients for which a specific response was generated/patients tested.
An alternative way of modifying DC maturation is to insert genes encoding cytokines or other substances onto DCs. Successful expression of transfected genes in DCs has been achieved using either retroviral gene transfer, gene gun, liposomal transfection or simply coupling of DNA to latex beads which are then phagocytosed by DCs [59]. Profound changes in the phenotype and function of DCs can be found after transfection of cytokine genes. Transfection of genes encoding tumor antigens has also been successful. Immunization with mouse DCs transfected with genes encoding mouse melanoma tumor antigens induces tumor-specific cytotoxic T lymphocytes [59]. This is a promising approach to avoid the problem of MHC restriction. Whereas any given peptide antigen within a tumor antigen is recognized only by patients expressing the appropriate MHC allele, DCs transfected with the gene encoding the whole tumor antigen can process it and express on their surface peptides capable of being recognized by cytotoxic T lymphocytes expressing whatever MHC alleles, provided that both DCs and cytotoxic T lymphocytes are autologous. It may even be possible that injection of the appropriate cytokines combined with tumor antigens in vivo will attract autologous DCs to the immunization site where they would uptake the injected antigens. This would obviate the need for ex vivo culture of DCs. Issues related to tumor-mediated immune suppression, however, caution that ex vivo generated DCs may be more active than endogenous DCs, unless ways to circumvent tumor-mediated immunosuppression of DCs are identified. In any case, immunization with tumor antigens vehiculized by ex vivo cultured autologous DCs provides important insights into the function of these fascinating APCs and their role in the generation of immune responses against tumors.
Mayordomo, Zorina, Storkus et al.
102
R EFERENCES 1 Steinmann RM, Cohn ZA. Identification of a novel cell type in peripheral lymphoid organs of mice. I. Morphology, quantitation, tissue distribution. J Exp Med 1973;137:1142-1162.
16 Romani N, Gruner S, Brang D et al. Proliferating dendritic cell progenitors in human blood. J Exp Med 1994;180:83-93.
2 Romani N, Schuler G. The immunologic properties of epidermal Langerhans cells as a part of the dendritic cell system. Semin Immunopathol 1992;13:265-279.
17 Caux C, Vanbervliet B, Massacrier C et al. CD34+ hematopoietic progenitors from human cord blood differentiate along two independent dendritic cell pathways in response to GM-CSF + TNFα. J Exp Med 1996;184:695-706.
3 Kersall BL, Strober W. Distinct populations of dendritic cells are present in the subepithelial dome and T cell regions of the murine Peyer’s patch. J Exp Med 1996;183:237-247.
18 Zorina T, Mayordomo JI, Watkins S et al. Culture of dendritic cells from murine bone marrow supplemented with GM-CSF and TNF-alpha. J Immunother 1994;16:247.
4 Lu L, Woo J, Rao AS et al. Propagation of dendritic cell progenitors from normal mouse liver using granulocyte/macrophage colony-stimulating factor and their maturational development in the presence of type-1 collagen. J Exp Med 1994;179:1823-1834.
19 Santiago-Schwarz F, Rappa DA, Laky K et al. Stem cell factor augments tumor necrosis factor-granulocyte-macrophage colony-stimulating factor-mediated dendritic cell hematopoiesis. STEM CELLS 1995;13:186-197.
5 Aager R, Crowley MT, Witmer-Pack MD. The surface of dendritic cells in the mouse as studied with monoclonal antibodies. Int Rev Immunol 1990;6:89-101.
20 Siena S, Di Nicola M, Mortarini R et al. Efficient ex vivo generation of functional dendritic cells (DCs) utilizable for tumor vaccination from blood cell transplants (BCT) in cancer patients. Proc Am Soc Clin Oncol 1996;15:554.
6 Steinman RM, Inaba K. Stimulation of the primary mixed leukocyte reaction. CRC Crit Rev Immunol 1985;5:331-338. 7 Steinman RM. The dendritic cell system and its role in immunogenicity. Annu Rev Immunol 1991;9:271-296. 8 Caux C, Vanbervliet B, Massacrier C et al. B70/B7-2 is identical to CD86 and is the major functional ligand for CD28 expressed on human dendritic cells. J Exp Med 1994;180:1841-1847. 9 Inaba K, Witmer-Pack M, Inaba M et al. The tissue distribution of the B7-2 costimulator in mice: abundant expression on dendritic cells in situ and during maturation in vitro. J Exp Med 1994;180:1849-1860. 10 Sallusto F, Lanzavecchia A. Efficient presentation of soluble antigen by cultured human dendritic cells is maintained by granulocyte/macrophage colony-stimulating factor plus interleukin-4 and downregulated by tumor necrosis factor alpha. J Exp Med 1994;179:1109-1118. 11 Koch F, Trockenbacher B, Schuler G et al. Antigen processing capacity of dendritic cells from mice of different MHC backgrounds: down-regulation upon culture and evidence for heterogeneity of dendritic cell populations. Adv Exp Med 1995;378:203-206.
21 Shurin MR, Pandharipande PP, Sikora SS et al. Characterization of dendritic cells obtained from mice treated with flt3 ligand and IL-12. Fourth International Symposium on Dendritic Cells in Fundamental and Clinical Immunology. Venice (Italy), October 5-10, 1996. 22 Aichele P, Brduscha-Riem K, Zinkernagel RM et al. T cell priming versus T cell tolerance induced by synthetic peptides. J Exp Med 1995;182:261-266. 23 Porgador A, Gilboa E. Bone marrow-generated dendritic cells pulsed with a class I-restricted peptide are potent inducers of cytotoxic T lymphocytes. J Exp Med 1995;182:255-260. 24 Mizoguchi H, O’Shea JJ, Longo DL et al. Alterations in signal transduction molecules in T lymphocytes from tumor-bearing mice. Science 1992;258:1795-1798. 25 Huang AYC, Golumbeck P, Ahmadzadeh M et al. The role of bone marrow-derived cells in presenting class I-restricted tumor antigens. Science 1994;264:961-965. 26 Zeid NA, Multer HK. S100 positive dendritic cells in human lung tumors associated with cell differentiation and enhanced survival. Pathology 1993;25:338-343.
12 Fagnoni FF, Takamizawa M, Godfrey WR et al. Role of B70/B7-2 in CD4+ T immune responses induced by dendritic cells. Immunology 1995;85:467-474.
27 Gallo O, Libonati GA, Gallina E et al. Langerhans cells related to prognosis in patients with laryngeal carcinoma. Arch Otorhinolaryngol 1991;117:1007-1010.
13 Hsu FJ, Benike C, Fagnoni F et al. Vaccination of patients with B-cell lymphoma using autologous antigen-pulsed dendritic cells. Nat Med 1996;2:52-58.
28 Giannini A, Bianchi S, Messerini L et al. Prognostic significance of accessory cells and lymphocytes in nasopharyngeal carcinoma. Pathol Res Pract 1991;187:496-502.
14 Koch F, Heufler C, Kampgen E et al. Tumor necrosis factor α maintains the viability of murine epidermal Langerhans cells in culture, but in contrast to granulocyte/macrophage colony-stimulating factor, without inducing their functional maturation. J Exp Med 1990;171:159-171. 15 Inaba K, Inaba M, Romani N et al. Generation of large numbers of dendritic cells from mouse bone marrow cultures supplemented with granulocyte/macrophage colony-stimulating factor. J Exp Med 1992;176:1693-1702.
29 Becker Y. Dendritic cell activity against primary tumors: an overview. In Vivo 1993;7:187-191. 30 Qin Z, Noffz G, Mohaupt M et al. Interleukin 10 prevents dendritic cell infiltration and vaccination with granulocytemacrophage-colony-stimulating-factor gene modified tumor cells. J Mol Med 1996;74:B9. 31 Grabbe S, Beissert S, Schwartz T et al. Dendritic cells as initiators of tumor immune responses: a possible strategy for tumor immunotherapy? Immunol Today 1995;16:116-120.
103
32 Pardoll D. Tumour antigens: a new look for the 1990s. Nature 1994;369:357-358. 33 Storkus WJ, Lotze MT. Biology of tumor antigens: tumor antigens recognized by immune cells. In: DeVita VT, Hellman S, Rosenberg SA, eds. Biologic Therapy of Cancer, 2nd edition. Philadelphia: J.B. Lippincott Company, 1995:64-77. 34 Mayordomo JI, Zorina T, Storkus WJ et al. Bone marrowderived dendritic cells pulsed with synthetic tumour peptides elicit protective and therapeutic antitumour immunity. Nature Medicine 1995;1:1297-1302.
Dendritic Cells for Cancer Vaccines 47 Cox AL, Skipper J, Chen Y et al. Identification of a peptide recognized by five melanoma-specific human cytotoxic T cell lines. Science 1994;264:716-719. 48 Storkus WJ, Zeh HJ III, Salter RD et al. Identification of T cell epitopes: rapid isolation of class I-presented peptides from viable cells by mild acid elution. J Immunother 1993;14:94-103. 49 Zitvogel L, Mayordomo JI, Tjandrawan T et al. Therapy of murine tumors with tumor peptide-pulsed dendritic cells: dependence on T cells, B7 costimulation and T helper cell 1-associated cytokines. J Exp Med 1996;183:87-97.
35 Feltkamp MCW, Smits HL, Vierboom MP et al. Vaccination with cytotoxic T lymphocyte epitope-containing peptide protects against a tumor induced by human papilloma virus type 16-transformed cells. Eur J Immunol 1993;23:2242-2249.
50 Boczkowski D, Nair SK, Snyder D et al. Dendritic cells pulsed with RNA are potent antigen-presenting cells in vitro and in vivo. J Exp Med 1996;184:465-472.
36 Falo LD, Kovacsovics-Bankowski M, Thompson K et al. Targeting antigen into the phagocytic pathway in vivo induces protective tumour immunity. Nature Medicine 1995;1:649-653.
51 Bellone M, Iezzi G, Manfredi AA et al. In vitro priming of cytotoxic T lymphocytes against poorly immunogenic epitopes by engineered antigen-presenting cells. Eur J Immunol 1994;24:2691-2698.
37 Mandelboim O, Berke G, Fridkin M et al. CTL induction by a tumour-associated antigen octapeptide derived from a murine lung carcinoma. Nature 1994;369:67-71. 38 Noguchi Y, Chen YT, Old LJ. A mouse mutant p53 product recognized by CD4+ and CD8+ T cells. Proc Natl Acad Sci USA 1994;91:3171-3175. 39 Mayordomo JI, Loftus DJ, Sakamoto H et al. Therapy of murine tumors with p53 wild-type and mutant sequence peptide-based vaccines. J Exp Med 1996;183:1357-1365.
52 Pan Z-K, Ikonomidis G, Lazenby A et al. A recombinant Listeria monocytogenes vaccine expressing a model tumour antigen protects mice against lethal tumour cell challenge and causes regression of established tumours. Nature Medicine 1995;1:471-477. 53 Hsu FJ, Benike C, Fagnoni F et al. A clinical trial of antigenpulsed dendritic cells in the treatment of patients with B-cell lymphoma. Proc Am Soc Clin Oncol 1996;15:419.
40 Celluzzi C, Mayordomo JI, Storkus WJ et al. Peptide-pulsed dendritic cells induce antigen-specific CTL-mediated protective tumor immunity. J Exp Med 1996;183:283-287.
54 Kwak LW, Campbell MJ, Czerwinski DK et al. Induction of immune responses in patients with B-cell lymphoma against the surface-immunoglobulin idiotype expressed by their tumors. N Engl J Med 1992;327:1209-1215.
41 Gyure LA, Barfoot R, Denham S et al. Immunity to a syngeneic sarcoma induced in rats by syngeneic lymph cells exposed to the tumour either in vivo or in vitro. Br J Cancer 1987;55:17-20.
55 Storkus, WJ, Mayordomo JI, DeLeo A et al. Dendritic cells pulsed with tumor epitopes elicit potent antitumor CTL in vitro and in vivo. Meeting of the International Association of Immunology, 1995:3493a.
42 Knight SC, Hunt R, Dore C et al. Influence of dendritic cells on tumor growth. Proc Natl Acad Sci USA 1985;82:4495-4497.
56 Bakker ABH, Marland G, de Boer AJ et al. Generation of antimelanoma cytotoxic T lymphocytes from healthy donors after presentation of melanoma-associated antigen-derived epitopes by dendritic cells in vitro. Cancer Res 1995;55:5330-5334.
43 Caux C, Liu Y-L, Banchereau J. Recent advances in the study of dendritic cells and follicular dendritic cells. Immunol Today 1995;16:2-4. 44 Grabbe S, Bruvers S, Gallo RL et al. Tumor antigen presentation by murine epidermal cells. J Immunol 1991;146:3656-3661. 45 Flamand V, Sornasse T, Thielemans K et al. Murine dendritic cells pulsed in vitro with tumor antigen induce tumor resistance in vivo. Eur J Immunol 1994;24:605-610. 46 Castelli C, Storkus WJ, Maeurer MJ et al. Mass spectrometric identification of a naturally processed melanoma peptide recognized by CD8+ cytotoxic T lymphocytes. J Exp Med 1995;181:363-368.
57 Koch F, Stanzl U, Jennewein P et al. High level IL-12 production by murine dendritic cells: upregulation via MHC class II and CD40 molecules and downregulation by IL-4 and IL-10. J Exp Med 1996;184:741-746. 58 Cella M, Scheidegger D, Palmer-Lehmann K et al. Ligation of CD40 on dendritic cells triggers production of high levels of interleukin-12 and enhances T cell stimulatory capacity: T-T help via APC activation. J Exp Med 1996;184:747-752. 59 Storkus WJ, Mayordomo JI. Gene-modified dendritic cells as biologic adjuvants for cancer immunotherapy. J Mol Med 1996;74:B11.
