Immunologists have long tried to exploit the immune system to control human disease, in many cases with great success. The prevention of viral diseases by immunizations that stimulate durable antibody responses is a cardinal example. Cancer and certain other infections have proven more elusive, however. Among the myriad reasons for this are that most tumor antigens (Ags) are self or differentiation Ags to which the immune system is not responsive, the tumor microenvironment itself may be inhibitory, and tumor cells lack the other surface molecules required for immunogenicity.
The stimulation of an immune response also requires more than just tumor Ags and lymphocytes. T lymphocytes in particular will only respond to Ag bound to special cell surface molecules that comprise the major histocompatibility complex (MHC) and are abundantly expressed by specialized Agpresenting cells (APCs). Moreover, the sensitization of resting or naïve T cells to Ag requires multiple additional costimulatory and adhesive ligands expressed on the APC surface together with the Ag and MHC. The dendritic cell (DC) is a specialized type of leukocyte, rare in the steady state, which provides a means of orchestrating all of these various requirements for initiating cellular immunity.
Tumors express antigens, and DCs possess everything else needed to stimulate T cell immunity. The challenge now is how best to combine the two, given that investigators can now generate sufficient numbers and purity of DCs for large-scale experimental and clinical applications. Various methods for loading tumor Ag(s) onto DCs are under evaluation, especially those that ensure durable expression and stimulation of T cells, which in turn target tumors expressing the same Ags. This article discusses the various genetic approaches used to transfer Ag into highly immunostimulatory DCs.
II. THE DENDRITIC CELL SYSTEM
A. DCs in Vivo
Dendritic cells are widely distributed throughout the body, notably at peripheral sites of Ag exposure. The physiologic process of antigen capture, followed by maturation and migration to secondary lymphoid organs where DCs select and stimulate Ag-specific, resting T cells, parallels an inducible increase in accessory molecule expression and stimulatory function. This segregates antigen uptake from antigen presentation, both physically and functionally, which are the two components of the afferent or sensitization arm of the immune response. T cells activated in secondary lymphoid organs then exit into the periphery to perform the effector functions of the efferent arm of the immune response.
DCs reside in epidermal and mucosal surfaces as Langerhans cells (LCs) and in the dermis and interstitia as dermal or interstitial DCs (DDC-IDCs). DCs at various degrees of maturation are also found migrating in afferent lymph and in draining secondary lymphoid organs, but DCs are never found in efferent lymph. DCs are not identifiable in fresh blood, e.g., in peripheral blood smears, without immunocytochemical detection of surface markers, which in the aggregate distinguish DCs from other leukocytes. Freshly isolated DCs do not exhibit their unique, stellate, veiled cytoplasmic extensions unless first activated in some way. Nonlymphoid, mononuclear leukocytes, or monocytes are now recognized as precursors of at least one type of DC; these are readily found in peripheral blood.
B. Generation of DCs ex Vivo
Prior methods for isolating these specialized APCs from circulating blood were laborious, lengthy, technically demanding, and of low yield. Methods for the generation of DCs ex vivo using defined precursors and cytokine conditions have fostered a much greater understanding of the heterogeneity of DCs. The complexity of the DC system is proving useful, given the resultant variety of important T-cell responses that can be stimulated.
Like all leukocytes, all DCs develop initially from CD34+ as well as more primitive hematopoietic progenitor cells (HPCs). Hematopoiesis is inherently a stochastic process, but cytokines can skew this process, especially in vitro. With increasing differentiation down a particular pathway, there is decreasing opportunity for differentiation into alternative progeny. While it is now recognized that myeloid (or nonlymphoid) DCs, as well as so-called lymphoid or plasmacytoid DCs, constitute the two broad types of DCs, this article focuses on myeloid DCs because their immunogenic capacities are better established (Fig. 1).
FIGURE 1 Hematopoietic development of human myeloid dendritic cells. Human DCs are all originally derived from CD34+ hematopoietic progenitor cells (HPCs), but follow alternative differentiation pathways depending on cytokine exposure. GM-CSF is the pivotal cytokine for all DC growth and differentiation. IL- 4 is useful for the suppression of macrophage differentiation. TGF-β supports the differentiation of Langerhans cells. Stem cell factor (c-kit-ligand) and FLT-3L are useful for progenitor expansion but do not affect differentiation. Inflammatory cytokines (e.g., TNF-α, IL-1, IL-6, PGE-2, and type I IFNs) and/or CD40L has proven useful for terminal maturation or activation. All terminally matured DCs are MHC bright, CD14 negative, and CD83+. Myeloid DCs are always CD11c+. Langerhans cells are CD11b negative, whereas the other myeloid DCs are CD11b+.
1. Monocyte-Derived DCs
CD14+ peripheral blood monocytes are the most accessible precursors for generating DCs. These precursors are postmitogenic and capable only of further differentiation, which bears directly on the type of gene transfer that can target these cells. Despite their more restricted differentiation potential, blood monocytes yield more highly purified populations of DCs with fewer enrichment steps than needed for generating DCs from multipotent CD34+ HPCs.
The granulocyte-macrophage colony-stimulating factor (GM-CSF) has consistently been the critical cytokine for supporting DC growth, differentiation, and survival from all precursors, including blood monocytes. Interleukin (IL)-4 has also proven essential for suppression of the alternative monocyte differentiation pathway into macrophages. By themselves, GM-CSF and IL-4 will only generate immature monocyte-derived DCs (moDCs). Upon removal of cytokines, these may revert to a precursor form or even differentiate into macrophages. Immature moDCs can also have the effect of stimulating a regulatory or suppressor type of T-cell response rather than a preferred immunogenic response. Exposure of immature DCs to a mixture of inflammatory cytokines and/or CD40 ligand ensures that these DCs achieve irreversible maturation and/or activation before exposure to T cells.
2. CD34+ HPC-Derived DCs
CD34+ HPCs can give rise to two types of myeloid, or nonlymphoid, DCs. Langerhans cells (LCs) develop in the presence of GM-CSF and transforming growth factor (TGF)-β. In the absence of TGF-β, dermal-interstitial DCs (DDC-IDCs) develop. Investigators now avoid fetal calf serum (FCS) supplementation of culture media, even for preclinical work in the laboratory, but especially for cultures generating progeny for administration to humans. The absence of FCS compromises the total expansion and yield, but this is an acceptable concession for avoiding the introduction of xenogeneic antigens into a highly immunogenic system.
Cytokines like FLT-3L and/or stem cell factor (ckit- ligand) are useful for recruiting progenitors into the cell cycle and for expanding these clonogenic progenitors. These precursor populations then remain sensitive to the differentiating effects of cytokines like GM-CSF, tumor necrosis factor (TNF)-α, and IL- 4 in the case of DDC-IDCs or to GM-CSF, TNF-α, and TGF-β in the case of LCs. Several lines of evidence support differentiation of DDC-IDCs, but not LCs, through a CD14+ intermediate. DDC-IDCs and moDCs are very similar, but simultaneous evaluations have not established whether these DCs are exactly homologous, albeit derived from different starting populations.
3. Phenotypic and Functional Characterization of DCs
There are consistent and reliable characteristics that distinguish DCs from other leukocytes, and specifically from other APCs. Most of these features are useful both in situ and in vitro, and they characterize all the different DC types.
The unique, circumferential, cytoplasmic, dendritic extensions, which are eponymous for DCs, are readily identified in situ, in epidermal sheets for example. These morphologic characteristics develop in vitro as well, becoming most pronounced with maturation and activation.
Mature myeloid DCs lack significant phenotypic expression of any epitopes specific to lymphocytes or macrophages, e.g., CD3, CD14, CD16, CD19, and CD20. Mature DCs increase expression of MHC, especially class II, and CD83. CD83 is unique to DCs among myeloid cells, can be detected intracellularly before terminal maturation and surface expression, and indicates commitment to the DC differentiation pathway. Its surface expression is the best available marker of maturation, and its function is under study.
DC maturation also increases the expression of all costimulatory and adhesive molecules (e.g., CD40, CD50, CD54, CD58, CD80, and CD86, among others) for interacting with and stimulating T cells. These epitopes are not restricted to DCs, however, so they cannot be used alone for DC identification. Differential expression of chemokines and chemokine receptors are being increasingly recognized as important components of the migratory and maturational stages of DCs.
The most straightforward and reliable assay of DC function in vitro, compared with that of other candidate APCs, is the allogeneic mixed leukocyte reaction (alloMLR). Terminally matured DCs are consistently the most potent stimulators of allogeneic T cells, including naïve or resting populations, by one to two logs compared with any other candidate, physiologic APCs like B lymphocytes or macrophages.
Dose-response titrations of DCs versus other APCs are especially revealing on this point. More sensitive assays (e.g., ELISpot, tetramer staining) in vitro are now available for monitoring responses stimulated in vivo by the clinical administration of DCs. Careful dose titrations still support the superior potency of DCs as stimulators of cellular immunity.
III. GENE TRANSFER IN DENDRITIC CELLS AND THEIR PRECURSORS
A. Rationale for a Genetic Approach to Express Antigens in DCs
The origin of DCs and their requirements for differentiation and activation are critical determinants of their biological activity and hence their therapeutic efficacy. Another essential consideration is how and when to provide relevant antigens to DCs or their precursors. Gene transfer in DCs provides a means of enhancing presentation and immunogenicity of tumor antigens, as well as optimizing DC maturation and function.
Following generation ex vivo, DCs can be activated and loaded with antigen in a controlled manner. This is essential to assessing the function of well-defined cellular reagents and establishing proof of principle of their therapeutic efficacy. Loading DCs with antigen is classically performed using protein extracts, purified or recombinant proteins, or synthetic peptides. Pulsing DCs with necrotic or apoptotic cells, or peptides bound to chaperones like heat shock proteins, are alternative ways of providing tumor antigens. Genetic approaches to antigen loading are based on the transduction of DCs or their precursors with cDNA or mRNA encoding the antigen.
In principle, gene transfer may offer several advantages over pulsing with peptide or cellular extracts.
a. A set of selected, pure (recombinant) proteins can be expressed, either alone or in defined combinations.
b. The level of antigen expression can be controlled at the transcriptional level.
c. Antigens can be processed into HLA-restricted peptides, irrespective of the HLA type of each individual, allowing an individual’s own cells to process those peptides best suited for presentation on its own MHC molecules.
d. Antigen presentation may be more sustained if the antigen is expressed endogenously rather than pulsed onto the cell surface.
e. Genetic alterations (point mutations, gene fusions) can be easily introduced in the cDNA encoding the antigen in order to augment antigenicity or facilitate processing.
f. The generation of DCs from their genetically modified precursors could greatly facilitate the production of effective DCs ex vivo or, perhaps, in vivo.
g. Genetic approaches offer attractive means to codeliver antigens and signals designed to enhance immune stimulation, e.g., CD40 ligand, cytokines, or chemokines.
B. Vectors Used for DC Transduction
Various strategies to generate and transduce DCs, using an expanding repertoire of available vector systems, are currently under investigation. Most use viral vectors (see Table I). Nonviral methods using RNA or DNA transfection are also actively investigated.
TABLE I Viral Vectors Used to Transduce DCs or Their PrecursorsVector type Prototypic virus Packaging size Vector immunogenicity Required target cell division Lytic activity Example Replication defective (RD) adenovirus Human adenovirus 8 kba +++a No No Dietz (1999) RD lentivirus Human immunodeficiency virus-1 8-9 kb + No No Dyall (2001) RD oncoretrovirus Murine leukemia virus 8 kb + Yes No Szabolcs (1997) Vaccinia Vaccinia (poxvirus) 30 kb +++ No Yes Chaux (1999)
1. Stable Transduction Systems
Strategies for modifying HPCs genetically require stable gene delivery systems. Stable gene transfer can be achieved with viral vectors that integrate into host cell chromosomes. Oncoretroviral vectors stably transduce CD34+ HPCs from bone marrow, cord blood, and cytokine-elicited peripheral blood. As oncoretroviral vectors require division of the infected cell to integrate (Table I), their efficacy is restricted to progenitor cells that proliferate before differentiation into DCs. This approach is thus not applicable to moDCs.
Lentiviruses, however, are able to integrate stably into certain nondividing cells. Wild-type human immunodeficiency virus-1 (HIV-1) in particular is capable of infecting moDCs in the absence of cell proliferation. Recombinant vectors derived from HIV-1 have been used successfully for transduction in vivo or in vitro of terminally differentiated cells like neurons, hepatocytes, retinocytes, and macrophages. To improve the safety of lentiviralmediated gene transfer, one or more of the four accessory genes vpr, vpu, vif, and nef, which contribute to viral pathogenicity, have been removed in the second generation of lentiviral vectors. Such multiply attenuated vector systems efficiently transduce moDCs.
2. Transient Expression Systems
Because DCs can prime T cells in a few hours (or days), transgene integration is not necessary to obtain an immune response. Other gene transfer approaches are then possible using nonintegrating viral vectors, plasmid DNA, or mRNA.
a. Nonviral Approaches Plasmid DNA and mRNA transfection can be performed in DCs by electroporation, lipofection, or passive pulsing. On the one hand, transfection with plasmid DNA, especially in nondividing cells, has proven extremely difficult and is further complicated by the high immunogenicity of plasmid backbones. On the other hand, mRNA can be easily transfected and has been shown to be very efficient at priming T cells.
A principal advantage of using mRNA is that the antigen does not need to be known. T cells can then be primed against several tumor antigens at the same time, avoiding the risk of escape that could occur with the use of a single protein or peptide. mRNA is also relatively easy to obtain, even from a very limited number of tumor cells. A promoter can be added to the primers used to generate tumor-derived libraries without recombinant DNA intermediates. Finally, the half-life of the mRNA is a few hours, and no insertional mutagenesis can occur, rendering the procedure very safe.
Nevertheless, the main concern is that the use of total mRNA obtained from tumor cells could elicit a deleterious autoimmune response. mRNA could, in principle, be selected by substraction techniques to avoid such a problem, but the logistics would be complicated. Two other relative disadvantages of this approach are that access to tumor tissue is required for every patient and it may not be possible to immunize against antigens encoded by rare mRNA molecules.
b. Viral Vectors Several viral vector systems have been used to transduce moDCs, including those based on adenovirus and poxvirus (Table I). Expression from these vectors is relatively short-lived, ranging from a few days in rapidly dividing cell populations to several weeks in nonreplicating cells. This is not an obstacle to expressing antigens in DCs that have a limited lifespan, however. Of greater concern is the expression of viral proteins encoded by these vectors, which may perturb DC differentiation and activation or alter the immunogenicity of the transduced DCs.
Expression of viral antigens may sometimes prove to be beneficial, e.g., by providing helper epitopes, or detrimental, e.g., by providing dominant epitopes that blunt the response to tumor antigens. The presence of highly immunogenic viral antigens may also prove to be problematic upon repeated administration of transduced DCs.
DCs are currently viewed as ideal adjuvants for immunizing against tumor antigens. Advances in the definition of DC subtypes and their requirements for activation hold the promise of generating effective DCs for clinical investigation. There are nonetheless a number of remaining questions about how best to use DCs for cancer immunotherapy.
In principle, genetic strategies offer a number of advantages in terms of introducing known or unknown antigens into DCs. Tumor-derived or genetically engineered recombinant antigens can be expressed using a range of technologies based on electroporation, chemical transfection, or viral vector-mediated transduction. A number of vector systems appear promising in terms of transduction efficiency in either moDCs or their hematopoietic progenitors.
Viral vectors differ in terms of their molecular requirements for successful transduction, duration of antigen expression, inherent immunogenicity, and safety. The availability of multiple vector systems for transducing moDCs and CD34+ HPC-derived DCs will allow for systematic investigation of the biological activity and therapeutic efficacy of different DC subsets. Comparisons between different antigen combinations and vector types will most certainly yield valuable information on the induction of potent immune responses against tumor antigens.
James W. Young
Memorial Sloan-Kettering Cancer Center
ADENO-ASSOCIATED VIRUS ; ANTI-IDIOTYPIC ANTIBODY VACCINES ; CANCER VACCINES: PEPTIDE- AND PROTEINBASED VACCINES ; CARBOHYDRATE-BASED VACCINES ; CYTOKINES ; DNA-BASED CANCER VACCINES ; RETROVIRAL VECTORS ; TARGETED TOXINS ; TUMOR ANTIGENS
antigen presentation The display of peptide fragments bound to MHC molecules on a cell surface in a manner required by T lymphocytes for recognition and response.
chemokine Small cytokine that influences migration and activation of cells, especially phagocytes and lymphocytes, but also dendritic cells.
cytokine Protein made by one cell that affects the development or function of another cell by interacting with a specific cell surface signaling receptor.
major histocompatibility complex A highly polymorphic gene cluster on human chromosome 6 (mouse chromosome 17) composed of codominant alleles that encode specialized membrane proteins for binding and presenting peptide antigens to CD8+ (class I MHC) and CD4+ (class II MHC) T lymphocytes.
transduction Gene transfer in mammalian cells mediated either by nonviral vectors or by viral vectors. Vector entry is an infectious process in the latter case, albeit not followed by viral replication when using a replication-defective viral vector.
transfection In mammalian cells, refers to gene transfer mediated by physical or chemical means, e.g., electroporation, calcium-phosphate precipitation, and lipofection.
viral vector (adenoviral, retroviral, etc.) Usually, replication- defective viral vectors are minimal recombinant viral genomes that retain cis-acting sequences required for permitting efficient transduction.
Banchereau, J., Schuler-Thurner, B., Palucka, A. K., and Schuler, G. (2001). Dendritic cells as vectors for therapy. Cell 106, 271-274.
Banchereau, J., and Steinman, R. M. (1998). Dendritic cells and the control of immunity. Nature 392, 245-252.
Bell, D., Young, J. W., and Banchereau, J. (1999). Dendritic cells. Adv. Immunol. 72, 255-322.
Gilboa, E., Nair, S. K., and Lyerly, H. K. (1998). Immunotherapy of cancer with dendritic cell-based vaccines. Cancer Immunol. Immunother. 46, 82-87.
Hart, D. N. J. (1997). Dendritic cells: Unique leukocyte populations which control the primary immune response. Blood 90, 3245-3287.
Kay, M. A., Glorioso, J. C., and Naldini, L. (2001). Viral vectors for gene therapy: The art of turning infectious agents into vehicles of therapeutics. Nature Med. 7, 33-40.
Young, J. W. (1999). Dendritic cells: Expansion and differentiation with hematopoietic growth factors. Curr. Opin. Hematol. 6, 135-144.