Infection and cancer suppress pDC derived IFN-I

Abstract

Plasmacytoid dendritic cells (pDCs) are specialized producers of Type I interferon (IFN-I) that promote anti-viral and anti-tumor immunity. However, chronic infections and cancer inhibit pDC-derived IFN-I. While the mechanisms of this inhibition are multifarious they can be classified broadly into two categories: i) reduction or ablation of pDC IFN-I-production capacity (functional exhaustion) and/or ii) decrease in pDC numbers (altered population dynamics). Recent work has identified many processes that contribute to suppression of pDC-derived IFN-I during chronic infections and cancer, including sustained stimulation through Toll Like Receptors (TLRs), inhibitory microenvironments, inhibitory receptor ligation, and reduced development from bone marrow progenitors and apoptosis. Emerging success leveraging pDCs in treatment of disease through TLR activation illustrates the therapeutic potential of targeting pDCs. Deeper understanding of the systems that limit pDC-derived IFN-I has the potential to improve these emerging therapies as well as help devising new approaches that harness the outstanding IFN-I-production capacity of pDCs.

Introduction

Type I interferons (IFN-I) are potent immune activating, anti-viral, and anti-neoplastic cytokines (reviewed in Refs. [1,2]). They encompass a family of genes including 13 IFNα subtypes in humans (14 in mice), one IFNβ and several other less well-defined gene products (reviewed in Ref. [2]). While most cells can produce some IFN-I, the existence of a highly specialized subset of IFN-I producing cells distinct from T cells, B cells, NK Cells, and Monocytes was established in the 1980s. These interferon producing cells (IPCs), were later shown to be equivalent to a cell type originally identified for its plasmacytoid morphology providing an identity to IPCs as plasmacytoid DCs (pDCs, reviewed in Refs. [3–5]).

pDCs produce IFNsat exceptional levels, including 13 subtypes of IFN-α, IFN-β, 3 subtypes of IFN-λ and IFN-τ (reviewed in Refs. [3–5] and [6]). Indeed, within a few hours of viral activation, 60% of the new pDC transcripts encode IFN-I sequences [6]. In line with their prolific IFN-production capacity, pDCs are known to promote control of multiple types of viral infection through mechanisms that include suppression of viral replication in infected tissues, induction of apoptosis in infected cells, and enhancement of anti-viral NK and CD8 T cell responses (reviewed in Ref. [4]). In contrast, a large body of literature indicates that pDC-derived IFN-I is limited during chronic infections and cancer. This may take place either through reducing pDC numbers by modifying pDC development, survival and/or differentiation (hereafter referred to as changes in pDC population dynamics, Figure 1 top), and/or through promoting a hypo-functional state whereby pDCs produce significantly less IFN-I in response to stimulation (hereafter referred to as pDC exhaustion, Figure 1 bottom). Altered pDC population dynamics and/or pDC exhaustion have been observed in chronic viral infections such as HIV [7–13], HCV [14–21], and HBV [22–24] in humans, as well as in SIV-infected Macaques [25–28], and in chronic LCMV-infected mice [29,30,31••]. Additionally, exhausted pDCs have been observed in many different types of human cancers including head and neck stromal cell carcinoma (HNSCC) [32], breast cancer [33], chronic lymphocytic leukemia [34] and ovarian cancer [35]. Here we will discuss pDC dynamics and exhaustion, including the current understanding of the mechanisms that drive these phenomena, their outcomes, and the potential therapeutic benefit of reversing these processes. It is important to note that while this review focuses on chronic disease settings, altered pDC population dynamics and pDC exhaustion have also been observed, albeit for shorter-term, during acute infection with LCMV, HSV-1, VSV, or MCMV [29,31••,36]. Its hould also be noted that pDCs have been described to play pathogenic roles in the context of certain infections and tumors, including the induction of tolerogenic and/or immunosuppressive responses, which are beyond the scope of this review. For this and for subjects such as steady-state pDC development and function, we direct the readers to other recent reviews [5,37,38].

Figure 1.

Mechanisms inhibiting pDC IFN-I production. Production of IFN-I in pDCs can be inhibited both by reducing available number of functional pDCs (Population Dynamics, top) and the amount of IFN-I produced by pDCs (Functional Exhaustion, bottom). Reduced numbers of functional pDCs may result from reduced developmental capacity from progenitors, apoptosis and/or conversion into cDC-like fates, all of which can be driven by IFN-I signaling, though this IFN-I may not be derived from pDCs. pDC functional exhaustion can be driven by a number of factors including stimulation by TLR ligands, and inhibitory receptor ligation, inhibitory cytokines and hormones. In addition, functionally exhausted pDCs encompass subsets with varying proliferative potential. Factors known to mediate the regulatory processes are in parentheses.

Suppression of pDC IFN-I production in chronic viral infection

pDC functional exhaustion

Decreased IFN-I production by blood pDCs was first observed in HIV patients in the early 2000s [7], and was subsequently confirmed in other HIV studies [8,9,11] as well as other human chronic viral infections [14–16,19,21–24]. The description of this phenomenon by our group and others in LCMV infection of mice [29,30,31••] and SIV infection of macaques [25,26] has permitted a deeper understanding of this phenomenon. We reported that, initially, both acute and chronic LCMV infections induce IFN-I and pDC exhaustion. Importantly, this phenotype is only sustained in the latter persistent infection setting and associated with dramatic reduction of systemic IFN-I (in the face of increasing viral titers in blood and multiple tissues) as well as enhanced susceptibility to unrelated secondary infections [29,31••]. Consistently, systemic IFN-I is attenuated, despite high viremia, early after HIV infection [39], and suppression of IFN-I production capacity by peripheral blood mononuclear cells has been correlated with opportunistic infections in HIV-infected individuals [10,40–42], even independently of CD4 T cell numbers [42]. More recently we showed that, analogous to the role of persistent TCR stimulation in T cell exhaustion [43], pDC stimulation through TLR7 was required for their functional exhaustion during chronic LCMV infection in mice [31••]. The mechanistic underpinnings of this observation are still not fully understood, but our group has established that TLR7 stimulation in the context of infection drives reduction in the levels of the transcription factor E2–2 [31••], which is critical for pDC development and function [44].

A relationship between TLR ligation and pDC function also likely exists in human HIV patients, as pDC exhaustion emerges in patients with high viral load [7,8] rapidly recovers during the administration of successful antiretroviral therapy [11] but subsequently becomes suppressed again upon interruption of treatment [8]. Similar to the LCMV murine infection findings, E2–2 is also reduced in human primary pDCs from HIV-viremic patients [31••] as well as in a human pDC cell line that was stimulated for two days with TLR ligands [44]. Furthermore, in human pDCs, stimulation through TLR has been shown to induce expression of the inhibitory ligand Tim-3 [9], which opposes pDC activation [45]. Coordinately, HIV patients have higher proportions of low IFN-I producing Tim-3⁺ pDCs [9]. While correlative, these data together suggest that persistent TLR stimulation may contribute to IFN-I exhaustion in both murine and human pDCs.

Engagement of pDC inhibitory receptors

Like many immune cells pDCs express a large variety of receptors on their surface that can temper their function such as BDCA2 [46], Siglec-H [47], ILT7 [48], FcεRI [49], DCIR [50], Tim-3 [45], CD28 [51], E-cadherin [52], and Nkp44 [53]. These systems likely represent mechanisms to tune down IFN-I production to avoid immunopathology, although they could also facilitate viral propagation. This is exemplified by the increased numbers of Tim-3⁺ pDCs in HIV-infected patients who exhibit reduced IFN-I production, as discussed above [9]. Additionally, mice deficient in CD28 show enhanced serum IFN-I and increased IFN-I transcript levels in pDCs after either LCMV or MCMV infection [51], and mice deficient in Siglec-H show increased systemic IFN-I after MCMV infection [54]. In contrast, it has been directly shown that the HIV protein Vpu can subvert antiviral pDC function through redistribution of BST2 on infected cells in order to allow viral exit while still engaging the inhibitory receptor ILT7 [55]. Thus, manipulation of pDC inhibitory receptors may be an effective strategy employed by hosts and pathogens to attenuate pDC-derived IFN-I in the context of infections.

Altered pDC population dynamics

Compromised pDC development from progenitors

Another consideration in the long-term suppression of pDC IFN-I production during chronic infection is that pDCs are typically short-lived [56,57], meaning that in the course of a chronic infection pDC populations could be entirely replaced, yet IFN-I levels are still suppressed with respect to peak responses. Our studies on pDC dynamics in chronic LCMV infection lend some insight into this phenomenon. We showed that pDC generation from bone marrow (BM) is decreased in chronically infected mice. This associates with a numerical reduction in progenitors with the potential to generate pDCs, their capacity to generate pDCs and their expression of E2–2 [31••]. In addition, we observed that BM progenitors from LCMV-infected mice develop into hypofunctional, E2–2-low, pDCs when cultured ex-vivo in the presence of Flt3L [31••], raising the possibility that transcriptional regulation inherited from BM progenitors contributes to IFN-I exhaustion in the pDC progeny. A caveat that should be considered in these studies is, however, the presence of virus in the cell culture, which could have caused persistent TLR stimulation to the newly generated pDCs. Remarkably, dysregulation of E2–2 and its target genes in both BM pDC progenitors and persistently stimulated splenic pDCs during chronic infection suggests down-regulation of this transcriptional pathway as a convergent mechanism imposing reductions in IFN-I production at both the levels of pDC development and function. Finally, it should be noted that some of the progenitor populations analyzed in this study are heterogeneous [58•], and so more studies are needed to discern whether the observed reduction in E2–2 in these populations represents a decrease in pDC-committed progenitor subpopulations.

pDC apoptosis

In addition to reduced development it has been observed that pDCs show higher rates of apoptosis in HIV and SIV infection compared to uninfected controls [26,27,59]. The precise mechanisms that drive increased pDC death in these infections is still not fully understood, but IFN-I signaling has been shown to drive pDC death in response to multiple other types of infections in mice [36], and thus it is tempting to speculate that IFN-I driven apoptosis is part of a negative feedback loop limiting pDC IFN-I production after infection. Alternatively, it is established that glucocorticoids (GCs), which are elevated in HIV-infected patients [60], can induce apoptosis in pDCs [61–63]. TLR stimulation, however, opposes GC-induced apoptosis [61–63], and it is therefore unclear whether these elevated levels of GC would induce apoptosis of pDCs in the context of infection.

pDC proliferation

Despite the aforementioned impaired pDC development from progenitors and pDC apoptosis, the numbers of pDCs in peripheral lymphoid tissues are not always dramatically reduced and they are even enhanced at day 10 after chronic LCMV infection [31••]. This was explained by the discovery that exhausted splenic and BM pDCs undergo significant self-renewal early after acute infection and throughout chronic infection with LCMV in mice. Such pDC expansion coincides with increased proportions of CD4-CCR9− and CD4-CCR9⁺ pDC subsets which exhibit the highest proliferative capacity [31••]. This was a striking finding as almost all splenic pDCs are CCR9⁺ and do not proliferate under steady state conditions [56,64,65]. In contrast, CD4-CCR9− and CD4−CCR9⁺ pDCs have been described as minor BM populations with proliferative potential and as the immediate precursors of CD4⁺CCR9⁺ pDCs [56,64,65]. Notably, the gain of pDC proliferation after LCMV infection depends on both IFN-I and TLR signaling, while functional exhaustion only requires the latter [31••]. Similarly, an increase in pDCs expressing the cell cycle marker Ki67 has been reported during chronic SIV infection in macaques [26–28]. Given this, it is tempting to speculate that the suppression of de-novo pDC development together with the gain in self-renewal capacity of exhausted pDCs may perpetuate a pool of hypofunctional cells that compromise IFN-I production throughout chronic infections.

pDC conversion

It is well established that after stimulation with IL-3/CD40L or TLR ligands in vitro pDCs increase their antigen presenting capacity and reduce IFN-I production potential [66–68]. We also demonstrated that viral infections (acute and chronic) and administration of IFN-I primes pDCs to differentiate into CD11b⁺ cDC like cells, exhibiting phenotypic and functional features of bonafide cDCs, both in vitro [69] and in vivo [70]. Similar conversion into CD11b⁺ cDC-like cells was observed by other groups using Ly49Q- pDCs from PolyI:C injected mice [71] or CCR9− pDCs stimulated with intestinal epithelial cell supernatant or GM-CSF [65] or cell transfer under steady-state conditions [64]. The conversion of pDCs into cDC-like cells is consistent with murine and human studies that reported intermediary DCs exhibiting features of both pDCs and cDCs [72–77,78•,79•]. These non-canonical DCs [5] have been defined as Axl⁺ in humans [72,78•], Cx3cr1⁺ CD8⁺ cDC in mice [75,76] and in recent high dimensional mapping and cross-species analysis studies they have been grouped in with the newly described population of transitional (t)DCs [72,79•]. The population defined as tDCs specifically show variable E2–2 expression and have reduced IFN-I production when compared to pDCs. Indeed, as E2–2 is necessary to maintain pDC fate [80] and is downregulated in pDCs and their progenitors during infection [31••], pDC conversion into cDC like cells may be, at least partially, driven via the suppression of E2–2. Further work will be necessary to determine whether the aforementioned intermediary DCs [72–77,78•,79•] are transitional stages along the pDC conversion into CD11b⁺ cDC-like cells that we first reported after LCMV infection [69,70]. Nonetheless, these observations suggest that enhanced pDC conversion into cDC-like cells, which exhibit diminished IFN-I production capacity, may be another mechanism by which infections suppress overall pDC-mediated IFN-I production.

pDC IFN-I exhaustion in cancer

Tumor microenvironments are complex, highly immunosuppressive, and diverse. Thus, disentangling the exact mechanisms by which tumors suppress pDC function is complicated, and it is likely that, as with immunosuppression of other cell types [81] many overlapping mechanisms suppress tumor-associated (TA)pDC function in a given malignancy. Below we provide some examples of mechanisms that have been associated to pDC exhaustion in specific tumor settings.

Immunosuppressive cytokines and hormones

Tumor microenvironments are rife with immunosuppressive cytokines and hormones such as Transforming Growth Factor Beta (TGFβ), Interleukin (IL)-10, and prostaglandin E2 all of which have been shown to inhibit pDC production of IFN-I [35,82–85]. Reducing the levels of PGE2 in HNSCC supernatants slightly relieves pDC IFN-I suppression [83], while blockade of TGFβ partially rescues IFN-I production in pDCs exposed to supernatants from ovarian cancer [35]. The typically pro-inflammatory cytokine Tumor Necrosis Factor Alpha (TNFα), can also inhibit IFN-I production by pDCs [82]. This may represent an autocrine negative feedback loop as pDCs can also produce TNFα upon TLR [5] or FCεRl engagement [49,86]. Indeed, in the above-mentioned ovarian cancer model TNFα also contributed significantly to the inhibition of IFN-I production by pDC in the presence of tumor supernatant [35].

Tumor metabolic environment as a challenge to pDC function

In addition to active evasion mechanisms employed by cancers, the tumor microenvironment can be naturally hostile for immune function as a result of the high metabolic activity of cancerous cells, which puts them in direct competition for nutrients with infiltrating immune populations [87]. As pDC production of IFN is a highly metabolically demanding activity requiring de novo synthesis of a large quantity of mRNA and protein, pDC function has been intrinsically linked to metabolism relying on both glycolysis and OxPhos [88,89]. It is thus plausible that some tumor microenvironments may lack the necessary nutrients to support pDC IFN-I production. Furthermore, many tumors have been established to produce and secrete high levels of lactate [90] which has recently been shown to inhibit pDC function both through nutrient sensing and direct metabolic pathways [91•]. Importantly, if this is the case, typical experiments where pDC are taken ex vivo from tumors and cultured in nutrient rich media for stimulation without lactate may underrepresent the amount of in vivo pDC functional suppression in the tumor microenvironment.

Potential commonalities in pDC exhaustion in chronic infection and cancer

In addition to the above-described work, it is tempting to speculate that, like chronic infection, persistent stimulation of TApDCs through TLRs may also contribute to their exhaustion in cancer. Indeed, nucleic acid release by dying cells, which would be sensed by TApDCs, has been described in tumor microenvironments [92]. In contrast, regarding the dynamics of pDCs in cancer, it is known that pDC can be recruited to tumors that produce Stromal Derived Factor-1 (SDF-1 a.k.a. CXCL12) [93] and that CCL2 is involved in pDC recruitment upon TLR7 ligand administration in tumor settings [94]. However, whether TApDCs are repopulated by BM progenitors or derived from local proliferative pDC populations in the tumor or draining lymph nodes, as observed during chronic infections [26–28,31••], remains unknown. It would also be interesting to explore whether TApDCs or pDCs in tumor draining lymph nodes may differentiate into tDC or cDC-like populations, as observed in the context of infection [69,70]. Answering these questions will be paramount to understanding which populations to target with therapeutics aimed at restoring pDC IFN-I production in the context of cancer.

Regarding pDC inhibitory receptors that have been involved in pDC suppression during chronic infections, Tim-3 is known to suppress pDC function in a tumor microenvironment [45]. In addition, it has been posited that BST2 expression by tumors may suppress pDCs through ILT7, though in multiple myeloma it was found that suppression of pDC function was mediated by E-cadherin and not BST2 [52]. As many cancers express, produce, or present the ligands of pDC inhibitory receptors (e.g. NKp44-L, BST2, phosphatidylserine, Galectin-9, HMGB1) [95–100], and expression of these ligands is a strong predictor of poor outcome or metastasis [96–99], it is conceivable that ligands derived from cancer cells may contribute to the suppression of TApDCs. However, as these ligands have been studied mostly for their impact on other immune cells, which express the same receptors, and their intrinsic impacts on cancer fitness, future work will be needed to elucidate their impact on pDC function in specific tumor microenvironments.

Concluding remarks

Despite the role of pDCs as specialized producers of IFN-I, there are diverse mechanisms, which temper or entirely shut down their IFN-I production. Given the consequences of excessive and/or persistent IFN-I signaling exemplified by both genetic interferonopathies such as Aicardi-Goutieres Syndrome, Chilblain Lupus, Systemic Lupus Erthematosus and others (reviewed by Ref. [101]) as well as IFN-I induced immunopathology that has been observed in some infections (reviewed by Ref. [2]), this may represent an evolved abundance of caution around the power of pDCs to produce such large amounts of IFN-I. Indeed, as mentioned above, altered pDC population dynamics and functional exhaustion still emerges (albeit temporarily) in response to multiple types of acute viral infection, suggesting that pDC suppression may be the result of a general default program that is initiated after pDC activation and is sustained in the context of a chronic insult like a persistent infection or a tumor [31••]. Thus, like PD1 expression in CD8 T cells, which prevents autoimmunity and immunopathology, but favors chronic infections and tumors [81], pDC IFN-I suppression may represent a similar natural mechanism of immune-regulation. Promising studies activating pDCs for treatment of cancer [102–105] suggest the utility of leveraging pDC IFN-I production. However, pDC IFN-I exhaustion and altered pDC population dynamics may be significant stumbling blocks in revealing the full therapeutic potential of pDCs in chronic infections and tumor settings. Further understanding of the mechanisms underlying pDC suppression will not only provide opportunities to restore IFN-I production to fight infections and cancer but may also allow harnessing of these natural braking systems to inhibit pDC function in the context of autoimmunity and other chronic inflammatory diseases.