|Home | About | Journals | Submit | Contact Us | Français|
Ex vivo production of highly-stimulator mature Dendritic Cells (DCs) for cellular therapy has been used to treat different pathological conditions with the aim of inducing a specific immune response. In the last decade several protocols have been developed to mature monocyte-derived DCs: each one has led to the generation of DCs showing different phenotypes and stimulatory abilities, but it is not yet known which one is the best for inducing effective immune responses. We grouped several different maturation protocols according to the downstream pathways they activated and reviewed the shared features at a transcriptomic level to reveal the potential of DCs matured by each protocol to develop Th polarized immune responses.
Dendritic Cells (DCs) are considered essential for the initiation, programming and regulation of antigen-specific immune responses. They play a fundamental role in supporting the survival and the effector function of primed T cells while coordinating communication among cells of the immune system. The observation that patients with cancer have DCs that exhibit a reduced ability to activate Th1-polarized immune responses[2, 3], makes the understanding of DC maturation a prerequisite for the design of rational immunotherapies based on antigen-specific reactivation of adaptive immune responses. Since DCs were first generated ex vivo from monocytes for immunotherapy and used in clinical trials[5, 6], they have been found to be effective in treating patients with lymphoma, melanoma and renal cell carcinoma[7, 8], although their potential utilization is wider and goes beyond cancer therapy. However, the clinical response rates of DC therapies are only 10–15%[7, 9]. Various reasons for the limited efficacy of monocyte-derived-DC-based vaccines have been hypothesized, with inefficient activation of Th1-polarized responses due to incomplete DC maturation being the most cited[7, 10, 11].
T cell activation by DCs relies on three different signals[7, 8]: antigen presentation, co-stimulatory and polarizing. The first two signals are intrinsic to mature DCs. During maturation, in fact, the antigen processing ability of DCs decreases while their antigen presentation and expression of co-stimulatory molecules increases. For example the expression of the B7 family of co-stimulator molecules is increased in mature DCs. Optimizing the expression of polarizing molecules is complicated: DCs can secrete different cytokines, growth factors and chemokines that can, in turn, attract different immune cells and differentially influence their activation in situ[13–15]. Moreover, the methods that are currently used to produce DCs for adoptive immune therapy may not be optimized: immune responses against tumors, pathogens and autoimmunities need different Th-specific polarization. For these reasons Th-specific immune responses should be obtained by stimulating DCs to produce cells with a specific Th polarizing ability, however, DC induced Th polarization is still not well understood or characterized.
Over the last decade, a large body of information has accumulated concerning the functional properties of DCs generated ex vivo under different conditions. However, the molecular determinants that regulate effective immune responses are still poorly defined. Moreover, the heterogeneity of the clinical trials done so far make it impossible to define the most efficient protocol according to clinical results. DNA microarray technology has often been used to study the maturation of DCs, but only a small number of studies have systematically compared different conditions of maturation[14, 15]. This review summarizes the salient results from these studies following a pathway-associated principle and clustered the different maturation strategies according to conceptually convergent theorems with the aim of suggesting future studies and directions to improve the effectiveness of DC-based immunotherapy.
We first selected publications on Pubmed using the following sequence of terms: “Monocyte” “Dendritic Cells” “Human” “Gene Profiling”. Then we selected only those published between September 2000 and June 2009 that focused on DCs production for immunotherapy and that included a protocol for generating mature DCs. The compounds used for the maturation of DCs were clustered according to the overlapping downstream signaling pathways that they activated. By using this method a less-complex and sharper description of the maturation of DCs resulted. All the pathways used for this analysis were canonical pathways described by Ingenuity Pathways Analysis (Ingenuity® Systems, www.ingenuity.com, Mountain View, CA). To restrict the analyses to more reliable and consistent data, genes were selected as induced only when cited by at least two papers. The only exception was when only one paper described the transcription profiling of a specific maturation protocol . Considering that the experiments used different microarray platforms and designs, we did not include any fold-change or p-value criteria for the selection of the genes, but trusted the methods used by the authors to define a gene as induced.
Eighteen reports fit the criteria for this study (Table 1). All reports were based on studies using monocyte-derived DCs. Generally, peripheral blood mononuclear cells (PBMCs) were collected by apheresis and monocytes were obtained by elutriation, CD14 antibody selection, or selection of adherent cells after overnight culture on plates. Protocols to generate immature DCs from circulating monocytes most frequently used granulocyte macrophage colony-stimulating factor (GM-CSF) and interleukin-4 (IL-4), although the concentrations of these factors and time in culture varied among studies. One study used interferon-α (IFN-α) instead of IL-4 during the differentiation step. Then, because of the low T-cell activation potential of the resulting immature DCs, cells were matured with immune response activation stimuli. Several strategies have been used to produce mature DCs, which are characterized by high immune cell activation potential. Factors used to mature immature DCs included lipopolysaccharide (LPS), CD40 ligand (CD40L), tumor necrosis factor-α (TNF-α), IFN-α and IFN-γ. Cocktails combining several factors, to better recreate the inflammation environment have also been used. Factors used in maturation cocktails include prostaglandin E2 (PGE2), interleukin-1β (IL-1β), IL-6 and polyinosinic: polycytidylic acid (poly (I:C)).
Based on the level of overlap of canonical pathways believed to be primarily activated by each compound (Figure 1), we clustered various maturation protocols into three major categories: LPS, CD40L/TNF-α, and IFN-dependent maturation strategies. In addition, a fourth category was needed for protocols that included cocktails of factors that represented a combination of the other categories. For instance, LPS interacts with Toll-like receptors (TLR)-4 leading to the activation of different transcription factors including nuclear factor -kB (NF-kB), p38 mitogen-activated protein kinase (p38 MAPK), c-Jun N-terminal kinase (JNK), and extra-cellular signal-regulated protein kinase (ERK1/2). On the other hand, although both CD40L and TNF-α activate NF-kB, this activation occurs through different and only partially overlapping signaling cascades. In contrast, IFN-α and IFN-γ signal through the Janus kinase (JNK) and tyrosine kinase (TYK) cascade leading to the activation of signal transducer and activator of transcription proteins (STATs), which leads to down-stream effects where the action of NF-kB is also complemented by the action of several interferon regulatory factors (IRFs). We summarized the effects of the different maturation protocols on the transcriptional profiles of DCs according to these broad categories. It should be kept in mind, that this report is obviously limited by the protocols used for DC activation which generally refer to a specific time point after the maturation process is started.
Regarding the different monocyte purification methods, there are no specific studies aimed at identifying the effects of the purification method on the final product at the transcriptomic level. Dohnal showed that the greatest differences due to the purification method were on the purity and yield of the starting monocytes, but functionally no statistical differences were noticed. In addition, the high degree of similarity of the phenotypic characteristics of the DCs derived from different monocyte selection techniques [14, 20–23], as well as the observation that the monocyte-derived DCs do not show differences at a transcriptomic level due to differences in processing suggests that the purification method has only minor effects on the final product. Thus, it is possible that other information could have been gathered if analyses were performed at different time points as was clearly shown by Piqueras.
Because of its bacterial origin and its predominant role as a pathogen associated pattern (PAP), LPS represents a prototypical model of DC maturation. Gene profiling studies (see Table 1) have shown that several chemokine genes are induced by LPS: CCL5 (RANTES), CCL4 (MIP-1β), CCL20, CCL18 (DC-CK1), CCL19 (MIP-3β), and CCL23 (MPIF-1)[20, 22, 23, 26, 27]. The expression of CCL5, CCL4, CCL20, CCL18 have also been confirmed by RT-PCR, ELISA or protein array[28, 29] [30, 31]. CCL5 primarily mediates the trafficking and homing of classical lymphoid cells such as T cells and monocytes, but also acts on a variety of other chemokine receptor (CCR)-5-expressing cells, including basophils, eosinophils, natural killer (NK) cells, DCs and mast cells. Increased CCL5 expression has been associated with a wide range of inflammatory disorders and pathologies. It is thought to act by promoting leukocyte infiltration to sites of inflammation. CCL4 has both chemotactic and pro-inflammatory effects, attracting, in particular, monocytes to the site of inflammation. CCL20 interacts with CCR-6 and recruits immature DCs, their precursors and NKT cells. Both CCL18 and CCL23 can attract lymphocytes, immature DC and monocytes. It remains unclear whether CCL18 predominantly favors pro-inflammatory (Th1 or Th2) or anti-inflammatory/tolerogenic responses depending upon the microenvironment in which it is secreted. Another transcript that was often observed to be induced by LPS-matured DCs is the BCL-related protein A1 (BCL2A1), which, like BCL-2, has anti-apoptotic function and whose induction has been confirmed by ribonuclease protection assay. It promotes the survival of immune cells and consequently leads to stronger immune responses. Taken together these data suggest that after DCs detect pathogens, they produce factors to enhance the recruitment of different immune system cells, and consequently, activate the primary immune response against pathogens in the area of infection.
Other genes whose expression has often been found to be enhanced in DCs matured with LPS are DUSP-5, DUSP-1 and TNFAIP3. The increased expression of these latter two have been confirmed at protein level[38, 39]. These three genes play an important role in inhibiting MAPK activity and in particular the activity of NF-kB, JNK and ERK2, indicating that the activation signal induced by LPS has a self regulatory loop that dampens its effects over time. Other genes whose expression is increased are super-oxide dismutase 2 (SOD2) and ICAM1 (CD54) and whose induction has been confirmed at a protein level [41–43]. Both genes have a putative role in inflammation. The adhesive molecule ICAM1 is directly implicated in leukocyte adhesion and transmigration during inflammation. SOD2, instead, is primarily involved in the catabolisys of Reactive Oxygen Species usually generated during inflammation.
According to our selection criteria, TNF-α/CD40L induces the expression of a pattern of genes tightly associated with TNF family signaling: CCL17 (TARC), IL-10R, IL-13R, MYO1A and GM-CSFR[14, 18, 23, 46–50]. CCL17, IL-10R and Il-13R have been observed to be induced by TNF-α/CD40L also with other techniques [46, 51]. CCL17 is a chemokine that is able to recruit memory DCs, CD4+ cells, and Th2-T cells. High levels of CCL17 in serum are associated with progression-free survival in advanced melanoma patients in response to dendritic cell-based immunotherapy. Both IL-10R and IL-13R are cytokines receptors involved in Th2 cell differentiation; both receptors can also make DCs more responsive to the Th2-cytokine imbalance that has been observed in several tumor microenvironments, and thus in this way weakening the immunogenic properties of DCs. The transcription profile of DCs matured with TNF-α/CD40L suggests that these DCs polarize T cells toward a Th2 response. This observation perfectly correlates with the impaired ability of TNF-α-matured DCs to induce the Th1 immune response described by Decker. In fact, TNF-α-matured DCs show a lack of secretion of IL-12 and IFN-γ and consequentially, a reduced ability to activate cytotoxic T cell responses in vitro.
Due to the central role played by IFNs in the initiation of innate and adaptive immune responses, the effects on maturation of DCs by type I (IFN-α) and type II (IFN-γ) IFNs have been widely studied[55, 56]. Although the physiological role of IFNs in DC function in vivo is still not clear, it has been reported that in vitro IFN-α induces the differentiation and maturation of DCs and IFN-γ can be used alone or in combination with other compounds to mature DCs, leading to the secretion of large quantities of IL-12[18, 57–59]. Moreover, type I IFNs enhance the maturation of DCs through a positive feedback loop, acting as autocrine mediators either when used alone or in combination with other factors[58, 60, 61]. Gene profiling studies indicate that the early response (6–8 hours following stimulation) of DCs to IFN-stimulation includes the expression of well-described IFN stimulated genes (ISGs), such as the CXCR3 ligand chemokines CXCL-9/Mig, CXCL-10/IP-10, CXCL-11/I-TAC, MXa, and ISG-15. Furthermore, both type I and type II IFNs stimulate, though with different intensities, the STAT-1/IRF-1/IL-15 axis which has recently been shown to play a key role in immune-mediated tissue rejection[63, 64]. In general, it has been concluded that IFN-based maturation of DCs is strongly biased toward a Th1 immune response. This observation is also suggested by a study of Longhi et al., where the induction of Th1 immune response has been shown to require type I IFNs.
Since IFN-dependent signaling is time and cell type dependent, it is important to keep in mind that different methods and/or stimulation of different cell types will yield different results. For instance, type I IFNs induce phosphorylation of STAT1 or STAT4 in immature DCs and mature DCs respectively. These differences are, however, likely to be quantitative in most cases rather than qualitative and the large majority of ISGs induced by either type of IFNs will likely separate the transcriptional patterns of DCs induced by IFNs from those of all other methods of maturation. In fact IFN-α induces a pattern of expression that is similar to the majority of the ISGs in both immature and mature DCs. Obviously, since stimulation of several TLRs (i.e. TLR-7, -8 and -9) is tightly linked to the secretion of IFNs and activation of IRFs, it is likely that maturation protocols involving nucleic acid containing PAPs of damage associated patterns (DAPs) may lead to the activation of similar pathways and produce similar Th1 polarization.
In an attempt to recreate a physiological environment for DC maturation, some investigators advocate the utilization of balanced cocktails of maturation agents that may be the most representative of various inflammatory states. At a cellular level a cocktail of agents leads to the activation of several downstream pathways and the consequent expression profile could differ from the simple sum of profiles of the genes induced by each single agent because of synergistic or antagonistic effects. The first and most frequently used cocktail contains TNF-α, IL-1β, IL-6 and PGE2. The rationale for the use of this cocktail is to enhance the pro-inflammatory effects of TNF-α by adding other pre-inflammatory cytokines in an attempt to mimic the inflammatory environment that physiologically leads to the maturation of DCs. It’s interesting to note that, although LPS can by itself induce the production and secretion of these same cytokines, gene expression profiles of DCs matured with LPS differ markedly from that of DCs matured with the cocktail. Like TNF-α alone, this cocktail leads to the induction of CCL17, CXCR4 and IL13R, but not IL10R and GM-CSFR. Interestingly the expression of IL-6 is enhanced. IL-6 is a potent, pleiotropic, inflammatory cytokine that mediates a plethora of physiological functions, including the developmental differentiation of lymphocytes, cell proliferation, and cell survival. Since IL-6 is biologically active only on cells that express constitutively IL-6R (B cells and hepatocytes), it is likely that it acts through transignaling in which IL-6 dimerizes with the soluble IL-6R shed by activated neutrophils and/or monocytes. However, this needs to be further explored in the context of DC activation. Other genes found to be increased by this cocktail are NF-kB1/2, CXCL-2/GRO-β, TNFAIP3/6, CXCL-8/IL-8, CXCL-16, CD58, IL-2R, IL-7R, CRIP1, ISG20, BCL2A1, IER3, CFLAR, NCK2, and RUNX3. The functions of these genes are associated with inflammatory response and apoptosis. In vitro assays have clearly shown that DCs matured with this cocktail strongly stimulate the proliferation of allogeneic lymphocytes[70, 71].
Although PGE2 is believed to play an important role in DC migration and lymph node homing, it is also considered responsible for some immune inhibitory properties of DCs. In particular, the induction of indoleamine-2,3-deoxygenase (IDO), the increased secretion of IL-10 and the diminished secretion of IL-12 suggest that PGE2 may induce a tolerogenic immune response[15, 68, 72]. Additionally, several clinical studies have shown that plasma from patients with metastatic melanoma or renal cell carcinoma treated with DCs matured with this cocktail contain a mixture of Th1/Th2 cytokines, without resulting in a shifting toward a Th1-cytokines profile[73–75]. For these reasons several groups use cocktails of cytokine that do not contain PGE2.
One of these alternative cocktails contains IL-1β, TNF-α, IFN-α, IFN-γ and poly (I:C). This maturation protocol seems to bias the maturation of DCs toward a Th1 phenotype since IL-12p40, IL-1α/β, IL-2Rα/β/γ, IL-6, CXCL-8/IL-8, IL-15, CXCL-9/Mig, CXCL-10/IP10, and CXCL-11 are expressed at much higher levels compared with the PGE2-containing protocols. Only the protocol containing poly (I:C) upregulates CCL4/MIP-1β, CCL5/RANTES, CCL20/MIP-3α and CXCL-2/GROβ, but CCL17/TARC and CXCL16 are not induced by the poly (I:C) containing protocol. However, this cocktail of cytokines also induces the expression of functionally active IDO and it has been shown that the presence of poly(I:C) could affect the antigen expression ability of DCs. In vitro stimulation of autologous T cells by poly (I:C) matured DCs, in fact, failed to show a clear improvement respect to the stimulation of PGE2 matured DCs.
Another of the PGE2-free maturation protocols contains LPS and IFN-γ. Gene expression analysis of DCs matured under these conditions clearly demonstrate the expression of several chemokines, proinflammatory cytokines and inflammatory-related genes. The expression of CCL2, CCL3, CCL4, CCL5, CCL8, CXCL-9, CXCL-10, CXCL-11, IL-1, IL-6, IL-8, IL-10, IL-12, TNF, DUSP5, and SOD2 were increased after 24 hours and these factors were found in the supernatants of DCs cultured with LPS and IFN-α, indicating a strong Th1 polarization potential. In vitro assays have shown that DCs matured with LPS and IFN-γ induce the stimulation of antigen specific CD4+ and CD8+ T cells much more than PGE2-matured DCs. These data suggest that a cocktail of cytokines induces factors that predominantly could be anticipated from the effects of the individual cytokines with dominance by IFN-dependent activation when either type I or type II IFNs or TLR agonists are added to the mixture, resulting in the induction of Th1 cells and effector T cell expansion.
Although gene expression profiling does not provide conclusive information about cell function, it does indicate predominant signaling pathways induced by a defined state of stimuli. This review, which addressed the transcriptional activation of DCs upon various maturation stimuli, indicates that maturation protocols can be categorized according to target pathways and presumed effects on cell function (Table 2). Furthermore, because of the lack of comprehensive comparison studies on different DCs maturation protocols, this review addresses important concepts concerning the differences of maturation protocols. In fact, by looking at the downstream pathways we have categorized the different protocols and have shown that some lead to similar expression patterns. Then, focusing on the most consistent data, we delineated the potential of DCs matured according to each protocol, showing the in vitro assay results that support the phenotype described by gene profiling. However, the ultimate comparison of protocols involves cellular assays and/or biomarkers associated with clinical outcomes. We, however, can speculate that gene expression profiling could identify biomarkers useful for quality control of DCs during manufacture and for potency evaluation of cellular products.
Although antigen-specific immunotherapy holds promise as an approach to cancer therapy, from a practical standpoint, limited, if any, clinical benefits have been achieved for several possible reasons[9, 78]. None the less, it is likely that the function of DCs as antigen presenting cells may play a central role, at least in the afferent phase, in the induction of adaptive immune responses. DCs, in particular, have attracted much attention because of their primary role in presenting antigens to lymphocytes and simultaneously contributing to the recruitment and activation of various immune cells through cell-to-cell contact or through secretion of soluble factors.
Fifteen years after the first studies of DC generation from monocytes, several different protocols have been described to differentiate DCs from circulating mononuclear cells and to mature them into efficient antigen-presenting units, but only a few have been widely tested in clinical trials. Consequently, it still remains questionable which DC maturation protocol provides the best cellular products for in vitro activation of adaptive immune responses or the adoptive transfer of antigen-loaded DCs.
The first maturation protocols used clinically were based on a single compound and lead to a phenotypical complete maturation of DCs but not to a functional DC. For a long time, the cocktail containing TNF-α, PGE2, IL-6 and IL-1β was considered the gold standard maturation protocol because it increased the ability of DCs to home in lymph nodes and recruit T cells. However, recent studies have shown that DCs matured with this protocol enhance the recruitment and activation of CCR4-expressing T regulatory cells by secreting CCL22/MDC[13, 79], and display a reduced ability to secrete active IL-12, which is thought to be fundamental for the induction of a Th1 immune response. Thus, different groups are testing in clinical trials maturation protocols that contain IFNs since it appears that these cytokines are more directly linked and are likely essential for the induction of an effector immune response. Indeed, gene expression profiling as well as other experimental evidence suggests that the maturation of DCs in the presence of IFNs seems to lead to a more powerful Th1-polarized immune response.
Our analysis came to these same conclusions, but only by considering the most consistent genes induced by each maturation compound we did reveal several differences in DCs matured by different protocols. According to our analysis, TNF-α and CD40L-based maturation protocols shift the polarization of DCs towards a Th2 response; whereas IFNs based protocols seem to be more able in activating Th1 cells. Moreover, our analysis suggested that the use of a cocktail of cytokines could lead to a more complete activation of DCs, better mimicking the one that occurs naturally in vivo where a combination of danger signals mature DCs. Among the cocktails of cytokines used, the potential of LPS in attracting different immune cells could be the best utilized when combined with the Th1 activating power of IFN-γ.
Further studies are necessary to define the best way to differentiate, mature and use DCs in order to obtain an effective Th1 immunogenic response or a Th2 tolerogenic response. Moreover, although some antigen processing and presentation related genes have been observed to be induced by maturation protocols, it is still not clear how specific maturation protocols affect antigen presentation. Antigen cross-presentation has been shown to be affected by maturation stimulus, but a comprehensive comparison of different protocols is lacking.
DC activation of T cells can be used as a marker of the co-presence of an activating environment and antigen presentation. In recent years several lines of evidence have indicated that during in vivo maturation DCs secrete a well-defined cascade of cytokines. These waves of cytokine production synergistically attract different immune cells in a time-dependent fashion triggering a logical and systematic activation of the immune response which evolves from the early induction of innate effector mechanisms to the subsequent expansion of effector adaptive immune responses and, in the end, the induction of regulatory mechanisms aimed at limiting and eventually concluding the host’s reaction to the pathogenic stimulus. While this physiological cascade is probably determined in vivo by a distinct group of maturing factors during the various phases of the immune response, the artificial over-exposure to higher than physiological doses of the same factors in the context of cellular therapy may effectively bias the natural behavior of DCs toward one or the other outcomes of the immune response.
The lack of comprehensive comparisons among protocols and the time-dependent effect of IFNs on DCs discussed before shed light on the importance of defining and comparing protocols currently used for DC maturation in the clinical arena. It is likely that DCs are not only affected by the compounds/cocktail used, but also by the timing of their application in relation to the utilization of the DCs. In fact, several studies described a loss of DCs’ activating power after they gained full maturation which was associated with reduced IL-12 secretion, absence of DCs response to further stimulation[81, 82] and impaired ability in inducing T cells activation. Perhaps, by using the optimal timing and maturation protocol combination it will be possible to obtain mature DCs that mimic physiological maturation and activate an effective immune response.
This work is supported by the Intramural Programs of the National Institutes of Health Clinical Center and National Cancer Institute.