|Home | About | Journals | Submit | Contact Us | Français|
Natural immune tolerance is a formidable barrier to successful immunotherapy to treat established cancers and chronic infections. Conversely, creating robust immune tolerance via immunotherapy is the major goal in treating autoimmune and allergic diseases, and enhancing survival of transplanted organs and tissues. In this review, we focus on a natural mechanism that creates local T-cell tolerance in many clinically relevant settings of chronic inflammation involving expression of the cytosolic enzyme indoleamine 2,3-dioxygenase (IDO) by specialized subsets of dendritic cells. IDO-expressing dendritic cells suppress antigen-specific T-cell responses directly, and induce bystander suppression by activating regulatory T cells. Thus, manipulating IDO is a promising strategy to treat a range of chronic inflammatory diseases.
Prophylactic vaccination to induce long-term protective immunity against certain pathogens has led to effective control and, in the case of smallpox, eradication of infections that formerly caused major epidemics. By contrast, use of similar vaccine strategies to treat established clinical conditions, such as cancer and chronic infections, have achieved little success despite much effort and considerable resources devoted to this goal. What basic immunologic principles explain this dichotomy of responses to prophylactic and therapeutic vaccination? In established clinical syndromes that are immunotherapy resistant, therapeutic vaccines fail to overcome changes brought about in the preclinical period of disease progression. These changes somehow block natural and vaccine-induced immunity to these antigenic insults where it matters, in local tissues where tumors and infections thrive and develop. Conventional approaches to vaccination are predicated, in large part on the premise that efficient delivery of target antigens, with concomitant vaccine adjuvant-induced triggering of innate immune responses, is essential to stimulate effective adaptive immune responses that lead to robust, long-lasting memory. Accordingly, the paradigm that rapid and complex local responses to external insults – known collectively as tissue inflammation – stimulate potent adaptive immune responses to antigens delivered simultaneously has been the dominant principle guiding vaccine design and vaccination procedures.
However, it is increasingly evident that inflammation also creates local tolerance in certain physiologic settings of clinical significance, including tumor progression and chronic infections. Immune suppression induced by local inflammation may be an integral component of disease progression before clinical presentation in such settings. Indeed local tolerance, or immune privilege, established in the preclinical period may explain why tumors and certain chronic infections are typically very resistant to natural and vaccine-induced immunity, and to other immunotherapies designed to enhance immunity to tumors and infected cells . This novel paradigm implies that immunotherapeutic interventions must achieve two key objectives to drive successful clinical outcomes. First, established local tolerance that protects tumors and chronic infections must be destroyed, at least transiently. Second, vaccines, adjuvants and immunostimulatory reagents must be introduced to exploit the window of opportunity created by breaking tolerance to stimulate effective immunity before local tolerance is re-established. In this review, we focus exclusively on a natural immunoregulatory mechanism – involving expression of indoleamine 2,3-dioxygenase (IDO) – that creates potent and dominant T-cell suppression at sites of chronic inflammation. As an immunosuppressive enzyme, IDO has been implicated as a major barrier to successful immunotherapy to treat cancer and certain infections; however, inducing IDO locally may serve as a treatment for autoimmune disorders and transplant recipients. IDO expression is induced by interferons (IFNs) in several cell types including dendritic cells (DCs) in some settings of inflammation. IDO may have immune-regulatory effects when expressed by various cell types. However, in this review, we focus exclusively on IDO expression in DCs because DCs are specialized to acquire, process and present antigens to stimulate naive T cells, and because IDO expression in DCs leads to potent and dominant T-cell suppression in a range of clinically relevant models of chronic inflammatory syndromes. Our goal in this article is to describe the immunologic significance of IDO expression in DCs, the role of IDO in disease pathogenesis, and to review how this knowledge may be applied to promote more effective immunotherapy to treat a range of chronic inflammatory diseases. As previous reviews focus on closely related topics [2–5], we first describe the IDO-dependent regulatory mechanism mediated by specialized DCs, concentrating on recent developments, before summarizing prospects for manipulating IDO for therapeutic benefit in the clinic.
As depicted in Figure 1, IDO is a cytosolic heme-containing enzyme that catalyzes the first, and rate-limiting step of oxidative tryptophan catabolism . This biochemical pathway is not active in all cells, particularly under homeostatic conditions and in gnotobiotic mice, because few cells express IDO unless induced to do so by IFNs released in response to inflammation. Two closely linked, homologous genes (IDO1 and IDO2) located in syntenic regions of chromosome 8 in humans and mice encode IDO proteins [7,8], and six introns in IDO-related genes are conserved from humans to mollusks, implying conservation for 600 million years . All mammalian IDO genes studied to date possess one or more IFN response elements (ISRE and GAS) in their promoter regions, and IFNs produced at local sites of inflammation are potent inducers of IDO in several cell types, such as some DCs, macrophages, eosinophils, epithelial and endothelial cells [10–16].
The IDO enzyme activity catabolizes compounds containing indole rings, including the essential amino acid tryptophan, consumes superoxide radicals, and produces tryptophan metabolites, known as kynurenines (Figure 1). IDO activity is measured by assessing kynurenine levels in serum, or in tissues by high-performance liquid chromatography analysis. Functional IDO activity depends on binding of IDO to heme, substrate supply, redox potentials, the absence of heme toxins such as nitric oxide, as well as other poorly defined post-translational modifications [17,18]. Indeed, reduced IDO activity due to deficient supply of superoxide substrate caused lung hyperinflammation in a murine model of chronic granulomatous disease (CGD) . Hence, it is important to measure enzyme activity, or IDO-mediated effects on T-cell responses, as IDO gene and protein expression, although necessary, may not be sufficient to drive IDO enzyme activity.
IDO-expressing plasmacytoid DCs (IDO+ pDCs) can create local immune privilege. In some physiologic settings, immune privilege is beneficial. For example, IDO inhibits potentially harmful maternal T-cell responses to paternally inherited fetal alloantigens during mammalian pregnancy . In other settings, IDO-induced immune privilege is detrimental: developing malignancies and chronic infections escape attack due to inhibited T-cell immunity, despite the presence of T-cell antigens. IDO regulates local T-cell responses in several syndromes and animal models of clinical significance, as discussed later.
IDO+ pDCs are present in tumor lesions and in tumor-draining lymph nodes (TDLNs) at sites of tumor growth in mice and humans [5,20,21]. Abnormal accumulations of IDO+ cells with DC markers in human sentinel lymph nodes (LNs) draining sites of melanoma growth correlated with significantly worse clinical outcomes [20,22]. In addition, many different types of tumor cells express IDO, although the proportion of tumors that express IDO is variable . Furthermore, inhibiting IDO either pharmacologically or via small interfering RNA (siRNA) inhibited the growth of transplantable melanomas . These data suggested that local IDO activity facilitates tumor growth and survival. Consistent with this notion, the pharmacologic IDO inhibitor 1-methyl-tryptophan (1MT) augmented the antitumor effects of some chemotherapeutic drugs in an autochthanous murine model of breast carcinoma and transplantable melanoma [25,26]. These findings support the hypothesis that IDO protects developing tumors from immune surveillance that could otherwise eliminate premalignant cells, and that IDO impedes the ability of antitumor drugs to destroy tumor cells directly. In a recent study, we reported that IDO+ pDCs created potent local suppression in inflamed LNs draining skin exposed to the tumor-promoting chemical phorbol myristate acetate (PMA, also known as TPA) . Moreover, IDO1-deficient mice exhibited a robust tumor-resistant phenotype in the standard two-stage model of inflammation-driven tumorigenesis after a single exposure to a carcinogen (7,12-dimethylbenz(a)anthracene [DMBA]), and prolonged exposure to PMA thereafter. These data suggested that inflammation induced by chemical (PMA) exposure promoted tumor formation by creating local immunosuppression due to the induction of IDO in pDCs. The key implications to emerge from these preclinical studies are that immunotherapy to block IDO activity is likely to be effective in reducing the risk of developing cancer, as well as enhancing immune responses to established tumors. The d-stereoisomer of 1MT, which was more efficacious in reversing T-cell suppression in preclinical studies than the l-stereoisomer , is currently being tested in Phase I oncology trials as a potential adjuvant to promote more effective antitumor immunity.
IDO-mediated tryptophan catabolism slows growth of intracellular pathogens that require exogenous tryptophan for survival, thus playing a role in host defense for organisms such as Toxoplasma gondii, Chlamydia spp., herpes simplex virus, measles virus and cytomegalovirus [28–32]. Although these studies on intracellular pathogens examined the role of IDO in host defense in vitro, IDO inhibition in vivo has been shown to increase susceptibility to Candida albicans, with a concomitant increase in fungal burden and decreased survival . However, there are conflicting roles for IDO in infectious disease, as some chronic infections, in a manner similar to developing malignancies, exploit IDO to create immune privilege that facilitates pathogen persistence. For example, IDO inhibitors significantly reduced levels of Streptococcus pneumoniae in lungs post-influenza infection, and IDO activity induced by HIV infection may contribute to HIV persistence [34,35]. This paradoxical role for IDO in infectious diseases may arise because IDO inhibits pathogen growth, but does not clear intracellular pathogen infections, while IDO inhibits adaptive immunity, which is required to clear infections. Thus, a pathogen that evolves strategies to evade the antimicrobial effects of IDO may benefit from IDO-mediated T-cell regulatory effects that protect pathogen-infected cells from adaptive immunity. Hepatitis C may be a pertinent example as IDO mRNA levels fell in primates with acute hepatitis C that was subsequently cleared, but IDO mRNA levels remained high in those with chronic disease . Hepatic IDO activity was also higher in patients with chronic hepatitis C that was refractive to IFN therapy. These data are consistent with the hypothesis that IDO is upregulated to create immune privilege that suppresses an adaptive immune response to chronic hepatitis C infection. In this case, IFN therapy may actually harm the patient by inducing higher levels of IDO enzyme that protects pathogen-infected cells from immune attack. However, more studies are needed to verify that IDO has a functional role in preventing clearance of hepatitis C infections.
The evidence for HIV using the immunosuppressive capability of IDO to elicit a tolerogenic state is somewhat stronger . There is a direct correlation between IDO mRNA levels in peripheral blood mononuclear cells (PBMCs) and HIV viral load . IDO inhibition in vivo resulted in a significantly increased cytotoxic T lymphocyte (CTL) response in the brain, followed by clearance of HIV-infected macrophages in a mouse model of HIV encephalitis . IDO inhibition in vitro led to significantly increased PBMC-induced CD4+ T-cell proliferation in HIV patients, and the cell subset responsible for increased IDO expression in HIV-infected PBMCs is the pDC . These data suggest that IDO inhibition in pDCs may promote an adaptive immune response in HIV patients that reduced viral loads or even led to complete immune clearance. This notion also raises the possibility that antiretroviral therapy may elicit some of its effects through partial IDO inhibition, as IDO mRNA levels were comparable with healthy controls in HIV patients treated with antiretrovirals . However, there may be a threshold of T cells necessary for IDO inhibition to elicit an immune response, as proliferation increase after PBMC stimulation directly correlated with the CD4+ count . These results imply that if given early, IDO inhibition (via 1MT) may improve the disease course in HIV-infected patients. Tryptophan catabolism has also been implicated in progression of AIDS-related neurologic disorders, as quinolinic acid, a tryptophan metabolite, is a N-methyl-d-aspartate receptor agonist . Recently, the relationship between neurologic disorders and tryptophan catabolism has been extended to implicate IDO activity and tryptophan metabolites as potential mediators of neurologic depression syndromes in mice with inflammatory syndromes [40,41]. The opportunistic pathogen Mycobacterium avium typically infects only AIDS patients, and can synthesize tryptophan; however, it was susceptible to IDO-mediated growth inhibition in macrophages following oligonucleotide treatment . These studies show that IDO can mediate antimicrobial effects, and protect infected cells from T-cell-mediated attack, identifying levels of complexity that complicate attempts to treat chronic infections using immunotherapeutic intervention.
IDO is a component of granulatomous lesions established in patients by Listeria monocytogenes and Bartonella henselae, which cause persistent infections . Many granulomas are found in patients with CGD, who cannot generate sufficient reactive oxygen species to destroy ingested pathogens due to a lack of NADPH oxidase activity. IDO functionality was recently reported as lacking in a murine model of CGD, probably because IDO utilizes superoxide as a substrate/cofactor for cleaving the indole ring of tryptophan, and lack of superoxide caused functional IDO deficiency . The innate defect in IDO led to increased IL-17 production, and expansion of proinflammatory γδ T cells, which increased susceptibility to Aspergillus fumigatus infection. This defect was corrected by treating defective mice with l-kynurenine and IFN-γ, demonstrating a link between the genetic defect in NADPH oxidase activity and a biochemical defect in IDO function in CGD mice.
IDO retards autoimmune disease progression in several models of autoimmune disorders and allergic diseases. Treatment with 1MT accelerated disease progression in murine models of experimental autoimmune encephalomyelitis (EAE), Type 1 diabetes (T1D) and rheumatoid arthritis [44–46]. Grohmann and colleagues also reported that splenic DCs from prediabetic, T1D-prone nonobese diabetic (NOD) female mice exhibited a transient defect in IDO-mediated T-cell suppressor functions due to excessive production of peroxynitrite by DCs, which was reversed with cytotoxic T-lymphocyte antigen (CTLA)4-immunoglobulin (CTLA4-Ig) treatment to induce IDO [47,48]. This suggested that impaired IDO expression by DCs may help explain why female NOD mice are prone to spontaneous T1D onset, although whether defective IDO is a primary cause or symptomatic of other upstream defects is not clear. Depleting pDCs in NOD mice accelerated T1D onset, and virtually eliminated IDO expression in the pancreas in a recent study, while restoring pDCs reduced the severity of T1D . Combined with the data that 1MT accelerates insulitis progression, these outcomes implicated IDO-expressing pDCs as a critical regulatory DC subset that inhibited T1D progression. Further, daily human chorionic gonadotropin injections significantly inhibited T1D onset in NOD female mice in an IDO-dependent fashion . These findings suggested that immunotherapy to induce IDO may help prevent T1D or slow T1D progression, and further suggested a rationale for reports that autoimmune diseases regress during pregnancy. Consistent with this notion, functional IDO was induced in human islets in response to IFN-γ , although alternative reagents to induce IDO must be identified because of the rare but serious side effects of IFN-γ therapy.
Some immunotherapies to treat hyper-immune syndromes also depended on induced IDO to promote therapeutic outcomes. Thus, 1MT abrogated stem cell therapy to treat EAE , and 1MT also blocked the immunotherapeutic effects of 4-1BB antibody in a model of arthritis . Thus, natural and induced IDO expression ameliorated autoimmune disease severity, and slowed disease progression. T cells may acquire resistance to IDO-mediated inhibition in rheumatoid arthritis patients, potentially due to increased expression of tryptophanyl-tRNA synthetase in T cells, which may overcome the regulatory effects of IDO . Increased T-cell resistance to tolerogenic mechanisms has also been reported for T1D, as T cells from T1D patients became more resistant to regulatory T cell (Treg)-mediated suppression . Determining whether T cells become resistant to IDO-mediated suppression in T1D and other autoimmune disorders requires further study; if verified, inducing IDO at higher levels may overcome resistance to naturally occurring IDO, and may improve disease outcomes.
The role of IDO in allergic disease syndromes is less clear. While Hayashi and colleagues reported that 1MT exacerbated experimental asthma , genetic ablation of IDO reduced the severity of allergic airway disease . This dichotomy is reminiscent of reports that 1MT induced allogeneic fetal rejection [56,57], while allogeneic pregnancies proceeded normally when both parents were IDO1 deficient . However, 1MT did not induce fetal rejection in pregnant IDO1-deficient mice, confirming that IDO (encoded by the IDO1 gene) was the relevant target of 1MT, and ruling out potential off-target effects of 1MT in IDO-sufficient pregnancies. Thus, while IDO is the dominant default mechanism that protects fetal tissues in IDO-sufficient mice, and redundant regulatory mechanisms compensated for genetic loss of IDO1 function; it remains to be seen if IDO2 [7,8] compensates for loss of IDO1 function during pregnancy and other physiologic syndromes where IDO is normally active.
Consistent with a role for IDO in protecting developing fetal tissues during pregnancy, inducing IDO improved graft acceptance rates in a number of murine models. Recently, IDO-competent DCs were shown to regulate spontaneous murine renal allograft acceptance without global immunosuppression , extending earlier observations that IDO mediated spontaneous murine liver allograft acceptance . Guillonneau and colleagues reported that IDO induction in epithelial cells following soluble CD40 (CD40-Ig) treatment protected cardiac allografts . Moreover, IDO induction in DCs via histone deacetylase (HDAC) inhibitors prevented allogeneic bone marrow graft rejection . Adenoviral and transposon vectors encoding IDO protected rat lung allografts [61,62], and IDO-adenoviral vectors also protected rat and murine heart allografts from rejection without additional immunosuppression [59,63]. Thus, IDO induction is an effective strategy to prolong allograft survival and to reduce dependence on use of globally immunosuppressive drugs.
Following the discovery that IDO promotes immune tolerance, several studies have focused on the potential prognostic value of measuring IDO expression and activity. As mentioned above, increased IDO+ cells in sentinel TDLNs of melanoma patients correlated with worse clinical outcomes [20,22]. Patients with high IDO expression in endometrial cancer also had significantly worse overall and progression-free survival; indeed, elevated IDO expression and higher Federation Internationale de Gynecolgie et d’Obstetrique (FIGO) clinical stage were found to be independent prognostic factors for impaired progression-free survival . Similarly, colon cancer patients with high IDO expression in tumor cells had significantly increased liver metastases and worse survival over 45 months . Although IDO expression in ovarian serous adenocarcinoma was a prognostic marker for impaired overall survival, there was no correlation between survival and IDO expression in patients with other ovarian adenocarcinomas, such as clear cell and endometrioid [66,67]. Finally, patients with high IDO expression in hepatocellular carcinoma had significantly lower 5-year survival rates, and IDO expression was an independent predictor of poor prognosis in hepatocellular carcinoma patients .
Other studies examined IDO enzyme activity as a diagnostic marker. Patients with acute kidney rejection episodes had higher serum kynurenine:tryptophan (K:T) ratios as soon as 1 day after surgery, which preceded serum creatinine elevation, typically used to diagnose such events . In addition, patients with primary Sjögren’s syndrome and systemic lupus erythematosus (SLE) had higher serum K:T ratios [70,71], and a higher baseline K:T ratio in SLE predicted activation during the summer in northern climates . Elevated K:T ratios prior to acute kidney rejection and worsened SLE symptoms imply that IDO is activated as a prelude to increased inflammatory activity, but increased pathology implies that naturally induced IDO is insufficient to overcome hyperimmunity, or that resistance to IDO-mediated suppression develops. Thus, in marked contrast to developing malignancies, IDO may inhibit autoimmune disease progression and transplant rejection, but may be overcome, or circumvented, during disease etiology.
IDO expression by DCs is of considerable interest because DCs are professional antigen-presenting cells capable of delivering stimulatory signals to naive and memory T cells. Among cultured human monocytes, the IDO-competent subset is confined to pDCs expressing CD123 and CCR6 . In mouse spleen and TDLNs, IDO-competent pDCs expressed B220, CD8α, the B-cell marker CD19, and high levels of both CD11c and the CC chemokine receptor CCR6, but did not express the conventional pDC marker mPDCA1 [74–78]. IDO-competent pDCs are specialized to express IDO in response to certain inflammatory cues such as IFNs [3,79] and TGF-β , but do not express functional IDO under homeostatic conditions. IDO-competent splenic pDCs exhibited a constitutive mature phenotype (MHCIIhighCD80/86high), and stimulated T cells ex vivo unless induced to express IDO. In our experience, CD19 is the most definitive marker for IDO-competent splenic pDCs and TDLN pDCs in mice [20,76,77]. CD19+ pDCs are rare, accounting for less than 10% of total splenic DCs (~30% of total TDLN DCs). The majority of DCs do not express IDO in response to IDO inducers, probably because negative regulation by signaling adaptor molecules such as DAP12 and suppressor of cytokine signaling 3 (SOCS3) blocks IDO upregulation in response to IFNs, although these restraints may be relaxed in some settings or by appropriate manipulations [81–83]. SOCS3 negatively regulates IDO activity in some DCs by binding to IDO protein and directing the complex for ubiquitination and proteasomal degradation . Conversely, the IFN regulatory factor IRF-8 positively regulated IDO expression by mediating downregulation of DAP12 .
Splenic CD19+ pDCs readily produced large quantities of IFN-α in response to CD80/86 (B7) and Toll-like receptor (TLR)9 ligation, and IFN-α was the obligate upstream inducer of cell-autonomous IDO following these treatments [76–78]. IFN-α receptor signaling induced signal transducer and activator of transcription (STAT)-1 activation selectively in IDO+ DCs , and most, if not all, IDO inducers stimulated IDO upregulation by inducing IFN production. Unexpectedly, we found that an upstream amplification loop via IFN-α controlled IDO upregulation in CD19+ pDCs in some settings, since IFN-α production did not occur under IDO-deficient conditions after B7 ligation, but still occurred after TLR9 ligation . Furthermore, IDO-induced IFN-α production was blocked by excess tryptophan, suggesting that IDO-mediated tryptophan depletion triggered downstream IFN-αproduction after B7 ligation. Consistent with this notion, an intact general control nonderepressible (GCN)2-dependent stress response pathway was essential in pDCs for IFN-α production after B7 ligation. GCN2 encodes a kinase that senses increased binding of uncharged tRNA molecules to ribosomal initiation complexes, and triggers downstream stress responses that shut down translation of most, but not all, proteins in cells experiencing amino acid withdrawal [86,87].
IDO+ pDCs are found in tissues with extensive mucosal surfaces such as lungs, eye, gastrointestinal tract, and placenta in pregnant females. Constant exposure to largely innocuous external insults induces homeostatic inflammation during normal functioning of such tissues. IDO+ pDCs may help mediate tolerance at such mucosal sites; if so, such functions may be redundant because IDO-deficient mice do not exhibit overt signs of excessive mucosal hyperimmunity or hyperinflammation. Some physiologic settings of induced inflammation in mice stimulate IDO-competent pDCs to express IDO constitutively. As discussed previously, a key example is local tumor growth, which creates local inflammation in tumor lesions and TDLNs. Considerable evidence now supports the hypothesis that IDO+ pDCs create immune privilege in TDLNs that facilitates tumor resistance to natural and tumor vaccine-induced immunity . Collectively, local tolerance induced by IDO+ pDCs in TDLNs coupled with tolerogenic mechanisms in tumor lesions themselves, which may be IDO dependent or independent, represent a formidable barrier to successful tumor immunotherapy [21,88]. As discussed above, IDO+ pDCs created potent and dominant suppression in inflamed skin LNs draining sites of topical application of a tumor-promoting chemical . This response is remarkably similar to the response induced by cutaneous melanomas, even though the inflammatory responses induced by these disparate insults are distinctive in multiple respects. Based on this observation, we hypothesized that suppressive local inflammation associated with developing malignancies protects malignant cells from innate and adaptive immune surveillance mechanisms from the very earliest stages in tumor development. Consistent with this hypothesis, IDO1-deficient mice exhibited a robust tumor resistant phenotype in the standard two-stage model of inflammation-driven skin papilloma formation after a single exposure to a carcinogen (DMBA), and prolonged PMA treatment thereafter to maintain inflammation that promotes tumor formation . These findings support the hypothesis that IDO+ pDCs may be critical factors in creating local immune-suppressive microenvironments in response to certain inflammatory insults that create immune privilege. Figure 2 depicts potential checkpoints where DCs expressing IDO in inflamed tissues, or in LNs draining such tissues, may control T-cell responses to self and exogenous antigens encountered at such sites to create immune privilege by suppressing T-cell clonal expansion in LNs, and blocking T-cell effector functions in target tissues.
IDO enzyme activity consumes tryptophan and produces tryptophan metabolites (kynurenines), which vary between cell types. These biochemical changes cause cell cycle arrest, apoptosis and anergy in T cells that activate in response to antigens presented directly to T cells by IDO+ pDCs, or by other DCs in the local microenvironment. IDO activity blocks clonal expansion of naive CD8+ and CD4+ T cells, and generation of CTLs and Th1 cells, while having less impact on Th2 cells . Conversely, IDO activity in pDCs promotes de novo Treg differentiation from naive CD4+ precursors ; the same result occurred when naive CD4+ precursors were cultured with kynurenine/low tryptophan, directly implicating tryptophan catabolism in Treg generation . Similarly, IDO was required for human DCs to promote Treg proliferation . Thus, IDO activity promotes regulatory outcomes by blocking effector T-cell generation and promoting Treg generation. It remains to be seen if IDO affects the generation of effector Th17 cells, which have been detected in infectious and autoimmune disease settings . Collectively, studies on the effect of IDO on T cells and Tregs support the model we depict in Figure 3. We hypothesize that IDO-competent pDCs prevent effector T-cell responses and promote Treg differentiation and activation, but only when local conditions (or treatments) induce pDCs to express IDO. Other cell types expressing IDO may also contribute to tolerogenic outcomes by suppressing T cells and promoting Tregs at the other checkpoints depicted in Figure 2, but the key point is that rare IDO+ pDCs may collaborate with rare Tregs to create potent and dominant T-cell suppressor activity in LNs that drain certain sites of inflammation.
How do IDO+ pDCs block T-cell responses? Some T-cell suppressive effects occur because IDO activity induces T cells to undergo apoptosis. IFN-γ induces T cells to undergo IDO-mediated apoptosis , and tryptophan deprivation causes enhanced Fas-dependent apoptosis and growth arrest of activated T cells . T-cell apoptosis may be triggered by tryptophan withdrawal, or by tryptophan metabolites, some of which induce apoptosis of murine thymocytes and Th1 cells . Indirect (bystander) suppressive effects mediated by IDO-activated Tregs may supplement these direct effects on T cells. In addition to an upstream requirement for intact GCN2 for IDO-dependent IFN-α production by CD19+ pDCs after B7 ligation, signaling via the GCN2-dependent stress response pathway was essential in T cells for susceptibility to suppression by IDO+ pDCs . T cells from GCN2-deficient mice proliferated normally in response to antigenic stimulation in the presence of IDO+ pDCs, revealing that T cells were unaffected by tryptophan starvation per se, since GCN2-deficient T cells experienced the same tryptophan ‘starvation’ conditions that blocked proliferation of normal T cells. Rather, T cells may respond to tryptophan withdrawal by inducing stress responses via the GCN2 pathway.
Intact GCN2 stress response signals were also essential for IDO-mediated activation of resting Tregs . Functionally quiescent Tregs acquired potent suppressor activity when cultured with IDO+ pDCs. Such responses depended critically on cognate MHC + peptide/T-cell receptor (TCR) and IDO/GCN2 signaling from pDCs to Tregs. Moreover, as yet undefined signals from activated T cells were essential to maintain optimal IDO activity, and generation of IDO-mediated Tregs in such cultures. Blockade of CTLA4/B7 interactions had a significant negative impact on IDO enzymatic activity, and Treg activation, providing further support for the notion that CTLA4+ Tregs ligate B7 on pDCs to maintain IDO activity in pDCs [75,90]. Thus, Tregs and pDCs exchange signals in both directions to promote and amplify IDO-mediated suppression. IDO-activated Tregs mediated suppression via a unique and distinctive mechanism dependent on intact programmed death (PD)-1/PDL-1 signaling, which was not dependent on IL-2, IL-10 or TGF-β . IDO-activated Tregs were found in inflamed TDLNs of melanoma-bearing mice. Remarkably, Tregs sorted from inflamed TDLNs of PMA-treated mice also exhibited potent PD-1 dependent suppressor activity ex vivo [Madhav Sharma, Phillip Chandler & Andrew Mellor, Unpublished observations]. Thus, IDO+ pDCs and Tregs, although rare cell types in normal tissues, accumulate and cooperate to create potent and dominant local suppression in LNs draining sites of tumor growth and topical inflammation caused by exposure to the proinflammatory chemical PMA (Figure 2). The key point to emerge from these studies is that local inflammation is not always inherently immunostimulatory, and may create potent immune suppression in certain settings relevant to clinical disease progression. Upstream requirements to induce IDO in LN pDCs in settings of chronic inflammation are still under investigation, but pDCs isolated from skin TDLNs of PMA-treated mice did not mediate suppression ex vivo when prepared from mice with defective genes for IFN type I and II receptors, GCN2, or the signaling adaptor molecule MyD88 , which promotes tumor development [98,99]. These findings reveal that complex upstream signals stimulate pDCs to express IDO in settings of intense inflammation that promote tumor formation. Requirements for IDO-mediated suppression may depend on the nature of the stimulus, the site of induction, and the cell being induced, because IFN type I, but not type II, receptor signaling was required for functional T-cell suppression by splenic pDCs after B7 and TLR9 ligation [76,77], whereas IFN type I receptor signaling was not required for functional IDO induction in lung epithelial cells after TLR9 ligation .
The hypothesis that IDO-competent pDCs help control the balance between antigen-specific effector and suppressor T cells elicited in inflamed LNs (Figure 2 & Figure 3) suggests that diametrically opposite strategies to inhibit or induce IDO will improve immunotherapy in chronic inflammatory disease syndromes by enhancing or blocking (respectively) production of proinflammatory cytokines and T-cell responses (Figure 3). In the settings of cancer and chronic infections where T-cell hyporesponsiveness contributes to disease etiology, strategies to block IDO activity are likely to improve clinical disease outcomes based on experimental outcomes in murine models. Conversely, in the settings of autoimmunity, allergy and transplantation where T-cell hyper-reactivity contributes to disease etiology, IDO induction is likely to enhance T-cell suppression, thereby augmenting therapeutic regimens. Alternatively, patients susceptible to infections where the pathogen requires host tryptophan for survival may benefit from therapy with IDO inducers to slow IDO pathogen growth and allow other immune mechanisms to clear the disease.
1MT is a competitive inhibitor of IDO . 1MT has been used in many studies to detect IDO-mediated effects in vivo and in vitro, but the relative efficacy of d- and l-stereoisomers of 1MT has raised controversy. While the l-1MT isomer was a more effective inhibitor of IDO enzyme activity in cell-free assays and HeLa cell-based assays, unexpectedly d-1MT was more effective in reversing IDO-mediated suppression of effector T-cell responses mediated by physiologic DCs and cultured pDCs from humans and mice [26,53,101]. Importantly, d-1MT was more effective than l-1MT in slowing growth of transplantable melanomas and autochthanous breast carcinomas in mice via T-cell-dependent mechanisms, and d-1MT effectively inhibited IDO activity and reversed IDO-mediated T-cell suppression in other studies with cultured human DCs and DCs isolated from TDLNs or synovial joints of patients with rheumatoid arthritis [26,53]. However, Lob and colleagues reported that d-1MT failed to block IDO activity in human DCs or HeLa cell lines, prompting the suggestion that any immunotherapeutic effects of d-1MT to emerge from ongoing oncology clinical trials cannot be ascribed to inhibitory effects on IDO [102,103]. These disparities in experimental outcomes using d-1MT may arise for technical reasons, such as use of recombinant IFN-γ or allospecific T cells to stimulate DC activation, which may affect the sensitivity of cells to the effects of d-1MT .
Hence, the nature of the stimulus to induce IDO may have distinct effects on IDO enzymology and susceptibility to 1MT stereoisomers in DCs and other cell types. To determine the optimal isomer for immunotherapy application in humans the relevant assays are those using physiologic readouts, such as mixed lymphocyte reactions, kynurenine assays using physiologic cells expressing IDO, tumor growth experiments and other in vivo models. Definitive resolution of the controversy about the effects of 1MT stereoisomers in humans awaits the outcome of ongoing and future NCI and corporate sponsored clinical trials to test the safety and efficacy of d-1MT as a potential tumor vaccine adjuvant, alone or in combination with other therapeutic modalities . These points notwithstanding, the key issues to emerge from studies in rodent models are that 1MT administered as a racemic (d,l) mixture or d-1MT was an effective immunostimulant and antitumor therapy, especially when combined with chemotherapy or tumor vaccines [25,27,105]
The discovery [7,8] of a second IDO-related gene (IDO2) in mice and humans closely linked to the original gene locus (IDO1) has prompted debate on which IDO gene is the relevant immunologic target for immunotherapy. Although Metz and colleagues found that IDO2 was the preferred target for d-1MT , d-1MT was not specific for IDO2, and these studies were performed using transfected cell lines, not physiologic cells. Moreover, the inhibitory effects of d-1MT on growth of transplantable melanomas were completely abrogated by ablating IDO1 gene expression, indicating that IDO activity encoded by IDO1, but not IDO2, protected melanomas from T-cell-mediated antitumor immunity . Moreover, the ability of d-1MT to enhance antitumor immunity in this system was clearly dependent on inhibition of IDO1, but not IDO2, activity. IDO1 gene ablation also led to a robust tumor-resistant phenotype in the DMBA/PMA model of inflammation-driven tumorigenesis, and skin TDLN pDCs from PMA-treated IDO1-deficient mice, or from PMA-treated mice given oral d-1MT, failed to mediate T-cell suppression ex vivo . Moreover, selective IDO1 knockdown in human DCs using siRNA (the DCs still expressed IDO2) resulted in loss of tryptophan catabolism . Collectively, these outcomes suggested that intact IDO1 expressed in pDCs was essential to facilitate tumor formation at sites of chronic inflammation, and that d-1MT blocked this function in mice. It is not clear if IDO2 is important in IDO-mediated suppression in other inflammatory disease syndromes, or if IDO2 genes mediate suppression in cell types other than DCs.
Metz and colleagues also reported that induced IDO activity in transfected cell lines blocked translation of a stimulatory isoform of nuclear factor (NF)-IL-6 (LAP), a transcription factor essential for expression of the proinflammatory cytokine IL-6 . Moreover, IDO activity promoted translation of an inhibitory isoform of NF-IL-6 (LIP), suggesting that IDO may control IL-6 expression at sites of inflammation. If verified in physiologic settings relevant to clinical disease syndromes this finding could have important implications for understanding how IDO+ pDCs create local suppression and affect disease progression. Agaugue and colleagues have also suggested that 1MT may modulate human DC functions independently of inhibitory effects on IDO . Nevertheless, experimental approaches using genetic ablation (or gene silencing) suggest that IDO1 is the relevant target for immunotherapy, obviating the need to invoke ‘off target’ effects to explain why d-1MT enhances antitumor immunity, at least in mice. However, it remains to be seen if the cell type and gene specificity, efficacy and mode of action of 1MT stereoisomers differ substantially in rodents and humans.
Besides using 1MT in vivo to enhance antitumor immunity (with chemotherapy or tumor vaccines), and reverse IDO-mediated T-cell suppression, other reagents can block IDO directly or indirectly. Natural brassinins, found in cruciferous vegetables, such as Chinese cabbage, inhibit IDO activity, and synthetic 5-bromo-brassinin inhibits transplantable melanoma growth in an IDO-dependent manner ; this outcome was somewhat unexpected as 1MT treatment alone had little effect on tumor growth . Hence, IDO may be a major target of 5-bromo-brassinin, but other mechanisms may contribute to the antitumor effects of this reagent. Similarly, the cyclooxygenase (COX)-2 inhibitor celecoxib exhibited antitumor and antimetastatic effects in a murine model of breast carcinoma that correlated with IDO downregulation, and increased effector T-cell responses, suggesting that IDO inhibition may contribute to the antitumor effects of COX-2 inhibitors . However, lipopolysaccharide (LPS)-induced COX-2 expression was attenuated by IDO activity in transfected cell lines leading to profound changes in downstream gene expression and cell-adhesion characteristics . Clearly, interactions between IDO and other biochemical pathways will provide further insights into potential targets to regulate immune outcomes by manipulating the fundamental biochemical processes affected by tryptophan catabolism in cells expressing IDO.
Paradoxically, some immunostimulatory reagents induce pDCs to express IDO under certain conditions in mouse models, potentially because they induce IFN release. Thus, LPS and immunostimulatory oligonucleotides with unmethylated CpG dimers that bind to TLR4 and TLR9, respectively, induced murine IDO-competent pDCs to express IDO [10,77,92,110]. Importantly, these observations imply that the full immunostimulatory potential of certain reagents may not be realized unless induced IDO activity is blocked concomitantly. Indeed, a recent study confirmed that IDO attenuated the adjuvant effects of the otherwise potent natural killer (NK) T-cell activator α-galactosylceramide, a CD1d ligand, in a mouse model of influenza vaccination, possibly because IFN-γ produced by activated NKT cells induced local IDO expression at sites of immunization . Thus, combination therapies with immunostimulants plus 1MT may be more effective in stimulating effective immune responses than immunostimulants alone. In addition to pharmacologic inhibition and genetic ablation of IDO, selective IDO knockdown by delivering siRNA via liposomes has also been effective in inhibiting IDO-mediated resistance to antitumor therapy . Another useful experimental approach has been to eliminate IDO-competent cell subsets in mice by targeting expression of the human diphtheria toxin receptor. Thus, under control of the DC-specific CD11c promoter, Saxena and colleagues elegantly showed that pDCs expressing IDO inhibited T1D progression in NOD mice while other (myeloid) DC subsets exacerbated T1D progression when pDCs were depleted .
IDO-inducing strategies and reagents have been shown to be effective in treating a range of hyperimmune and hyperinflammatory disease syndromes in rodent models. Thus, as stated above, genetic approaches to induce IDO over-expression led to remarkable increases in lung transplant survival with drastic reduction in pathologic states associated with graft rejection [61,62]. In addition, protective effects of administering CD40-Ig on cardiac allograft survival was dependent on increased IDO expression by epithelial cells . Transduction with recombinant lentiviral vectors, which show considerable promise as effective genetic transducers, also induced human DCs to express IDO, albeit at lower levels than achieved with adenoviral vectors . These outcomes raise the prospect that enhancing IDO can be used to reduce current needs to treat transplant patients with global immunosuppressants in perpetuity, thereby potentially lowering the incidence of serious side effects, and the cost of patient care post-transplantation. However, key questions about the safety of recombinant viral vectors to drive prolonged IDO expression in transplanted tissues have yet to be resolved, and nonviral transposons may be an attractive alternative to enhance IDO provided that such vectors can be generated reliably to clinical standards.
IDO-inducing reagents, many of which ameliorate hyperinflammatory and hyperimmune syndromes in mice, include soluble CTLA4 and CD40, lipopolysaccharide (TLR4 ligands), resiquimod (TLR7/8 ligands), CpG oligodeoxynucleotides (TLR9 ligands), soluble CD200 (CD200-Ig), anti-4-1BB, anti-FcεRI, dexamethasone (an inducer of glucocorticoid-induced tumor necrosis factor receptor ligand [GITR-L]), prostaglandin E2 and HDAC inhibitors [52,60,76,77,92,113–119]. As discussed above, the therapeutic effects of these reagents are, at least partially, IDO dependent in many murine models of clinical disease syndromes, such as autoimmune and allergic diseases and transplantation, because their therapeutic effects are abrogated by IDO inhibitors or in IDO-deficient mice.
CTLA4-Ig is a good example of an IDO inducer that was developed originally as an immunomodulatory reagent with a completely distinct mode of action via CD28 costimulatory blockade by binding to B7 (CD80/86) molecules on APCs. CTLA4-Ig (abatacept, marketed as Orencia™) was approved for use as a immunotherapeutic reagent to treat patients with rheumatoid arthritis several years ago . However, not all rheumatoid arthritis patients respond to CTLA4-Ig treatment, and side effects due to immune perturbation, and an increasing number of patients refractory to treatment represent significant barriers to widespread clinical application, which suggests that there may be room for improvement . In 2002, Grohmann and colleagues reported that CTLA4-Ig induced splenic DCs in mice to express IDO and acquire suppressive properties as a consequence . Subsequently, we identified murine CD19+ pDCs as the rare subset of IDO-competent pDCs that responded to B7 ligation by CTLA4-Ig [75,76,115], and showed that human DCs also expressed IDO following CTLA4-Ig treatment . However, optimal IDO inducing functions are species, and CTLA4-Ig isotype specific – particularly with regard to Fc domain structure – such that all isotypes with human Fc domains (hCTLA4-Ig) failed to induce IDO in mice [Babak Baban, Phillip Chandler & Andrew Mellor Unpublished Observations]. Hence studies showing that hCTLA4-Ig slowed murine T1D progression focused exclusively on the costimulatory blockade properties of hCTLA4-Ig, not the potential effects of enhanced IDO . Since IDO inducers have potent suppressive effects on immune responses in a range of murine models of chronic inflammatory diseases syndromes, efforts to optimize the IDO-inducing properties of CTLA4-Ig, and other IDO-inducing reagents are likely to be rewarded by significant improvements in therapeutic potential and benefits in the clinic. The discovery that IDO activity mediates T-cell suppression has generated interest in other drugs that inhibit or enhance tryptophan metabolism. Several reports document the immunosuppressive effects of natural tryptophan metabolites, or synthetic analogs in mouse models of autoimmune and allergic diseases, suggesting that at least some of the suppressive effects of IDO are mediated by metabolites [19,118,124,125]. Hayashi and colleagues also identified a potential mechanism of action of the tryptophan metabolite 3-hydroxyanthranilic acid involving inhibition of 3-phosphoinositide-dependent protein kinase (PDK)1 signaling in T cells, which resulted in inhibition of experimental asthma . HDAC inhibitors are promising new anticancer drugs, which also induced DCs to express IDO. The anti-inflammatory effects of HDAC inhibitors in a mouse model of graft-versus-host disease were dependent on IDO induction in DCs, which significantly reduced levels of proinflammatory cytokines produced postengraftment . Thus, HDAC inhibitors may suppress inflammation in part by inducing IDO, and the anticancer effects of HDAC inhibitors may be enhanced by combining them with 1MT to enhance antitumor immunity, as discussed above.
Another potential IDO inducer is IFN-β, currently used to treat multiple sclerosis. Little is known about the relationship between IFN-β and IDO; however, patients receiving IFN-β (IFN-β1a) for the first time to treat multiple sclerosis had higher K:T ratios than normal controls, suggesting that IFN-β mediated IDO upregulation . Of note, those receiving acute IFN-β (IFN-β1a) had higher K:T ratios than those receiving long-term IFN-β (IFN-β1b); both differ significantly in structure, route of administration and dose, which may account for the differential effects. Nevertheless, these data suggest that as discussed earlier for the related disease EAE, inducing IDO may have protective effects in certain neurological conditions.
IDO activity is a factor in many inflammatory syndromes of clinical significance. IDO slows hyperimmune disease progression, such as autoimmunity or allergy, but exacerbates hypoimmune syndromes, such as cancer and chronic infection, leading to the notion that therapies to enhance, or inhibit innate IDO, respectively, will be therapeutic in these syndromes. IDO-competent pDCs respond to inflammatory cues, and can create potent and dominant suppression in local tissues that overrides natural immunity, and the stimulatory effects of immunotherapeutic interventions. Thus, manipulating IDO-mediated suppression presents a novel approach to designing effective immunotherapies to treat a range of clinical syndromes.
The use of IDO inducers to treat hyperimmune disease syndromes, and IDO inhibitors to treat hypoimmune syndromes is already providing new opportunities to improve immunotherapies designed to treat a range of chronic inflammatory diseases affecting millions of people every year. A key recent development in the IDO field is the initiation of oncology Phase I clinical trials to examine the safety of administering d-1MT to cancer patients. Based on results showing synergistic antitumor effects of chemotherapy plus 1MT in mouse models, we anticipate that evidence of enhanced antitumor efficacy may emerge only when d-1MT is used in combination with other therapies in proposed future Phase II trials, contingent on the outcome of Phase I trials. Additional developments relevant to therapy are likely to emerge from increased knowledge of the effects of IDO activity on fundamental immune processes such as inhibition of T-cell responses to antigenic stimulation, activation and maintenance of Treg suppressor activity, modulation of APC functions, particularly DCs, and inhibition of proinflammatory cytokine production by cells of the innate and adaptive immune systems. Continuing research in these areas is likely to elucidate novel targets for immunotherapy upstream or downstream of IDO that will allow improved and more subtle control of the balance between immune stimulation and immune suppression at sites of local inflammation relevant to a wide range of inflammatory disease syndromes of clinical importance.
The authors wish to thank David Munn and Phil Chandler for insightful discussions and collaboration with research presented in this review, and Doris McCool, Anita Singh and Diane Addis for technical support related to research in this review. The authors also thank Phyllis McKie and Tracy West for administrative assistance.
Financial & competing interests disclosure
Research in the laboratory of Andrew L Mellor cited in this review was supported by grants from the NIH (HD41187, AI063402), and from the Carlos and Marguerite Mason Trust. Andrew L Mellor is a member of the Scientific Advisory Board of NewLink Genetics Inc., and receives compensation for this service. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in the production of this manuscript.
Papers of special note have been highlighted as
of considerable interest