Interaction Between the Mucosal Vaccine Adjuvant CDG and Pulmonary DCs

Abstract

Adjuvants significantly enhance vaccine efficacy, in part, by targeting, modulating and activating key functions of dendritic cells (DCs). A novel mucosal vaccine adjuvant candidate, 3’, 5’ cyclic di-GMP (CDG), is well-known to induce protective immunity against bacterial and viral pathogens. However, the direct effect CDG has on DCs in vivo remains unknown. Chapter I provides evidence that CDG directly targets and activates pulmonary DCs to drive adaptive immune responses and generate protective immunity. Using an intranasal vaccination strategy with CDG, we observe an increase in antigen uptake and processing by pulmonary DCs. CDG selectively activates and mobilizes pinocytosis-efficient DCs to the draining lymph nodes, where these cells initiate adaptive immune responses. Additionally, conditional removal of the CDG receptor, stimulator of interferon genes (STING), in DCs impairs the functions of pulmonary DCs and ultimately reduces the adjuvant activity of CDG in vivo. While we show that CDG can directly activate DCs in vivo, pulmonary DCs are heterogeneous and are equipped with specialized functions.

In chapter II, we specifically identify a subset of pulmonary CD11b+ DCs as the main mediator of the adjuvant activity of CDG, and delineate a novel two-signal requirement for the maturation of CD11b+ DCs in vivo. The IRF4fl/flCD11ccre mouse model is used to demonstrate that CD11b+ DCs drive CDG-induced antibody production and T cells responses, specifically Th2 and Th17 cells. We found that pulmonary CD11b+ DCs are characterized into two functional subsets based on expression of tumor necrosis factor receptor 2 (TNFR2). Use of an adoptive transfer model shows the TNFR2-CD11b+ DCs as the mediators of CDG-induced immunity in vivo, and require the activation of RelB for maturation.

Furthermore, we found that CDG drives the maturation of CD11b+ DC through a two-signal mechanism: the CDG/STING/TBK1 pathway for production of membrane-bound TNF (mTNF)-a and mTNFa/TNFR2/RelB signaling pathway for maturation. Collectively, the data in this thesis shows that CDG is an attractive mucosal vaccine adjuvant due to its ability to target and activate unique functions of pulmonary DCs. Specifically, our novel finding that CDG promotes the maturation of a unique subset of CD11b+ DCs through the TNFR2-RelB signaling axis provides insight into the mechanism of CDG activity in vivo and contributes to the development of novel mucosal vaccine strategies.

Introduction

Vaccines have effectively eliminated many illnesses caused by viral and bacterial pathogens, such as smallpox, polio, and measles, and therefore are important for the prevention of infectious diseases. There are over 25 vaccines available for human use, and are mostly administered either through subcutaneous or intramuscular routes [1]. These vaccines can elicit antigen-specific humoral and cellular immune responses systemically, however they are incapable of generating protective mucosal responses [2]. It is becoming increasingly clear that eliciting local mucosal immune responses through vaccination is important for the prevention of many infectious diseases. Mucosal vaccines induce local immune responses at the site of entry to prevent colonization and invasion by infectious pathogens, and are therefore superior to the current systemic vaccines [2].

A. Mucosal Vaccines

Vaccines are considered to be the greatest medical achievement of the 20th century, reducing the number of diseases caused by infectious agents [3]. However, there are no effective vaccines available for respiratory infections such as: influenza, pneumonia, and tuberculosis, which are responsible for over 5 million deaths annually worldwide [3]. Features of an effective vaccine include: (i) safety; (ii) protection against infection and (iii) long-term protection. The ultimate goal for a successful vaccine is to prevent infectious disease through the development of long-term protective immunity, which entails a strong humoral response and balanced cellular immunity, at the site of infection. Furthermore, a vaccine must be easy to administer, have very few side effects and must be biologically stable [2].

1. Strategies for Mucosal Vaccination

Current vaccines for use against these respiratory bacterial and viral pathogens are administered systematically and fail to elicit protective mucosal immunity in the lung.  Mucosal vaccines can induce antigen-specific humoral and cellular immunity, systemically and mucosally, by inducing long lasting memory immunity  [4-6]. Despite its important role in providing frontline immunity, there are only a dozen of mucosal vaccines approved for human use [2].  This lack of mucosal vaccines is due to the vaccine formulation, which depends on route of administration and immunogenicity. Vaccines are characterized into three main categories: (i) killed, inactivated vaccines; (ii) live, attenuated vaccines; and (iii) subunit vaccines.

a. Killed, Inactivated Cell Vaccines

Inactivated vaccines contain killed bacteria and virus that cannot cause disease, yet still maintain some of their pathogenic integrity. Neutralizing antibodies are produced against a wide range of surface antigens for a particular pathogen [7-9]. This method is successfully used to produce vaccines against polio (IPV), cholera (Dukoral), typhoid fever, pertussis and influenza. Inactivated vaccines are unable to replicate or mutate, and will not spread, thus are considered safe. However, this method induces weak immune responses, thus require an adjuvant and multiple doses to induce a sufficient immune response [10].

The only licensed inactivated vaccines for mucosal use are the cholera vaccines: Dukoral and Shanchol [5]. Both vaccines use the whole killed Vibrio cholera, with Dukoral using the cholera toxin subunit B as an adjuvant. These vaccines are administered orally, and require 2-3 doses to generate IgA antibodies as well as IFN- producing T cells in the gut [11]. However, these inactivated vaccines induce short-term protection, with around three to five years [12].

b. Live, Attenuated Vaccines

Live, attenuated vaccines induce the most potent immune response and can stimulate both humoral and cellular immunity without the need of an adjuvant [13]. This occurs due to the pathogen’s ability to replicate, although for a limited time, without causing infection at the site of immunization. Several examples of successful attenuated vaccines include: the Bacille Calmette Guerin (BCG) tuberculosis vaccine, the oral polio vaccine (OPV), intranasal influenza vaccine, and typhoid vaccine. A major disadvantage for this type of vaccine is the possibility of reversion back to a virulent state and cause infection and disease.

The majority of mucosal vaccines contain live, attenuated pathogens. These vaccines include: the influenza vaccine FluMist, H1N1 influenza vaccine NASOVAC, OPV, and the Salmonella Typhi vaccine Vivotif [5]. FluMist and NASOVAC are administered intranasal while OPV and Vivotif are administered orally. Mucosal IgA and serum IgG antibodies as well as T cell responses are induced by these live, attenuated vaccines [5, 11, 14-16]. Although these vaccines are highly effective at generating long-lived immunity, there are some safety concerns. While FluMist has been shown to be more effective than the inactivated, parental (intramuscular administered) influenza vaccine, wheezing and hospitalization rates were higher in children vaccinated with FluMist [17]. Additionally, the OPV has been shown to revert back to a virulent state, and in some cases cause paralysis similar to wild poliovirus [18, 19]. These drawbacks indicate that although this strategy is effective at stimulating mucosal immunity, there are still concerns over the safety of these vaccines.

c. Subunit Vaccines

While live, attenuated vaccines are effective at stimulating both mucosal and systemic immunity, there are still concerns over the safety and stability of these vaccines [5]. Subunit vaccines are ideal for the development of mucosal vaccines since these vaccines contain specific antigens, protein or polysaccharide derived, of a particular pathogen without including live, virulent components of the pathogen. Vaccines from live, attenuated microorganisms are at risk of reversion and are unsafe in humans. Subunit vaccines do not have the same safety concerns as live, attenuated vaccines and would be ideal for mucosal vaccines [5]. Currently, there are no licensed subunit mucosal vaccine available due to the difficultly to stimulate strong mucosal production of IgA and protection by mucosal administration of antigens. Since these vaccines contain purified antigens, they are less immunogenic and need an adjuvant to improve their immunogenicity [5].

2. Mucosal Adjuvants

An adjuvant by definition is a compound that when administered with an antigen will enhance immunogenicity [20]. The ultimate idea behind an adjuvant is to enhance and elicit the desired humoral and cellular immune responses against an antigen. In the case of subunit vaccines, where the antigen itself is weakly immunogenic, an adjuvant will help in generating a strong immune response towards the antigen and ultimately will help in inducing protective immunity needed for a successful mucosal vaccine. Adjuvants can regulate the type, quality, and magnitude of the adaptive immune response [13]. Since adjuvants can affect the immune responses in different ways, it is important to know the mechanism behind the adjuvant before choosing a candidate adjuvant for a specific mucosal vaccine. An adjuvant can enhance an immune response through different ways:

(i) induction of cytokines and chemokines for the recruitment of innate immune cells to the site of injection;

(ii) enhance antigen uptake and presentation to T and B cells; and

(iii) activation and maturation of DCs [21].

Several studies have shown that adjuvants targeting the TLR pathway can induce the generation of humoral and cellular immunity indirectly by increasing the production of proinflammatory cytokines, which activate B cells to increase the production of IgG antibodies and CD8⁺ T cells [22]. Other studies have shown that adjuvants such as alum, oil-in-water emulsions, and micropatricles can induce humoral and cellular immunity directly by targeting and enhancing antigen uptake by DCs [21, 23-25].

Despite all the advantages associated with potent adjuvants, there are several limitations: (i) safety and toxicity; (ii) limited immune responses and; (iii) chemical properties, such as size, electric charge and hydrophobicity. These disadvantages highlight why there are no mucosal adjuvants approved for human use [5, 15, 26]. The most promising mucosal adjuvants are derived from bacterial toxins and are potent stimulators of humoral and cellular immunity. Cholera toxin and labile enterotoxin are experimental adjuvants able to induce antigen specific antibody and T cell responses at the mucosal and systemic compartments, making them very potent mucosal adjuvants [2, 27-31]. Moreover, these adjuvants can selectively enhance the activation, maturation and migration of DCs in the intestine leading to the activation of T cells following oral administration [32-34].

Although these adjuvants have all the characteristics of a promising mucosal adjuvant, both have severe side effects when administered through the intranasal route and are not suitable for mucosal vaccines [35-37].The only FDA approved vaccine adjuvant in United States, alum, has minimal side effects, however this adjuvant has a limited immune response, capable of generating one type of T cell response. Alum induces a predominantly Th2 response, while failing to generate a Th1 or Th17 response, thus making this adjuvant ill-suited for vaccines against intracellular bacterial pathogens and viruses [38]. The lack of mucosal adjuvants available for human use highlights the need for the development of potent adjuvants that can elicit robust humoral and balanced cellular immune responses that are also safe for human use.

B. Cyclic Dinucleotides as Mucosal Vaccine Adjuvants

Recent interest has grown in a new class of adjuvants known as cyclic dinucleotides (CDNs) due to their immunomodulatory properties. CDNs are small molecule second messengers synthesized by bacteria to regulate important biological processes, but also serve as potent stimulators of the innate immune response. There are three types of naturally occurring, bacterial derived CDNs: bis-(3’,5’)-cyclic dimeric guanosine monophosphate (CDG); bis-(3’5’)-cyclic dimeric adenosine monophosphate (CDA) and; cyclic [G(3’5’)pA(3’5’)p] (3’3’-cGAMP). CDG is mainly produced by gram negative bacteria and serves as a second messenger to regulate virulence, survival, motility and biofilm formation [39]. CDA is found to play an important role in cell wall biosynthesis, potassium transport, and stress by gram positive bacteria [40]. 3’3’-cGAMP is involved in the virulence of Vibrio cholera [41]. A fourth CDN was found to be produced in mammalian cells in response to cytosolic DNA: cyclic [G(2’5)pA(3’5)p] (2’3’-cGAMP) [42, 43]. CDNs all serve as PAMPs to activate the innate immune system in response to invading pathogens.

CDNs, especially those that are bacterial derived, are ideal candidates for mucosal vaccine adjuvants since they can potently stimulate the innate immune response. Additionally, there is great interest in developing CDNs as potential treatments for cancer, infectious diseases, and autoimmunity[44]. Studies have shown that CDG and CDA are capable of producing higher levels of mucosal IgA than CT, the gold standard for mucosal adjuvants, and also generated much better memory T cell responses than CT [45-47]. However, not all CDNs are ideal as mucosal adjuvants. When administered intranasally, 3’3’-cGAMP induces moderate production of antigen specific antibodies and only induces Th1 and Th2 responses [48]. Since 2’3’-cGAMP is a self molecule, generated in response to self-DNA, it is also not an ideal mucosal vaccine adjuvant due to the risk of autoimmune diseases [49].

1. Cyclic di-GMP is an attractive mucosal adjuvant

A promising mucosal vaccine adjuvant must achieve the following: (i) protect against viral and bacterial infections; (ii) produce systemic and mucosal IgG and IgA; (iii) induce a balanced Th1, Th2, and Th17 cell response; (iv) must have stability and; (v) must be safe, with minimal side effects. CDG has been shown to have all the characteristics of an effective mucosal adjuvant, as described below.

a. Cyclic di-GMP induces protective, humoral and cellular immunity

Early studies have shown that CDG was able to generate protection against pathogens. It was demonstrated that CDG treatment significantly reduced the colonization of Staphylococcus aureus, Klebsiella pneumoniae, Bordetella pertusis and Streptococcus pneumoniae [50-53]. Furthermore, CDG did not directly kill Staphylococcus aureus or Streptococcus pneumoniae, but instead it was its immunostimulatory effects, through the recruitment of innate cells and production of proinflammatory cytokines, which protected these mice [51, 53]. In a bacterial pneumonia model, it was shown that pretreatment with CDG, either through intranasal or subcutaneous routes, resulted in a reduction of bacterial CFU in the lung and significantly increased survival [52]. To investigate whether the protective effect of CDG can be translated into a vaccine, Hu et al. immunized mice with Staphylococcus aureus antigens and CDG subcutaneously. Following intravenous challenge with methicillin-resistant Staphylococcus aureus, mice immunized with CDG had increased survival and reduction in bacterial load [54].

The immunomodulatory effects of CDG are even more drastic when administered intranasally, solidifying the idea that CDG is an ideal mucosal adjuvant. Intramuscular vaccines required high doses of CDG, around 145ug, to elicit adjuvant activity, whereas intranasal vaccines only require 5ug [45, 54, 55]. The first study to directly show the mucosal adjuvant activity of CDG intranasally administered CDG with the pneumococcal surface adhesion A (PsaA) [45]. Mice had reduced bacterial burden in the lung when immunized with CDG and generated robust antigen specific IgG in the serum and IgA in the lung. Importantly, mice immunized with CDG produced comparable antigen specific antibodies as those immunized with CT.

Other studies have also shown that intranasal administration of CDG with a vaccine antigen generated antigen-specific responses in the serum and lung, further illustrating the ability of CDG to generate humoral immunity [56]. Additionally, intranasal vaccines adjuvanted with CDG produce more antigen-specific antibody in the serum than alum and CT and this correlated with enhanced protection [45, 51]. In agreement with the serum antibody results, subcutaneous immunization with CDG enhanced the formation of germinal centers with B cells [57].

Vaccination with CDG induces a balanced Th1, Th2, and Th17 cellular immune response, providing a stronger rationale for the use of CDG as a mucosal adjuvant. It was first reported that CDG immunizations have an effect on priming T cells [56]. Mice vaccinated with CDG and -galactosidase induced the proliferation of antigen specific T cells, specifically Th1 and Th2 memory cells. Intranasal vaccination with CDG and b-galactosidase induced the generation of CD8⁺ memory T cells, as well as Th1, Th2, and Th17 memory CD4+ T cells in the spleens [58]. These observations have also been seen in intranasal CDG vaccinations with influenza antigens [59, 60]. Intranasal immunization with CDG and the influenza antigen H5 generated stronger Th1, Th2, and Th17 responses than intramuscular immunization [60]. It also generated multifunctional Th1 cells, which are important for protective immunity.

Taken together, these studies highlight the effectiveness of CDG as a mucosal vaccine adjuvant. These studies convincingly demonstrate that intranasal administration of CDG provides protection against bacterial and viral pathogens by generating antibodies in the lung and serum, while also inducing a balanced Th1, Th2, and Th17 immune response. It is also interesting to note that intranasal administration of CDG also generates antigen specific IgA in other distal mucosal compartments, such as the saliva and intestine [45, 58, 59]. Furthermore, CDG has been shown to be highly stable under various conditions and temperatures [57, 61]. However, the mechanisms of CDG in vivo are poorly understood, hindering its development as a mucosal vaccine adjuvant.

b. The adjuvant activity of CDG requires STING in vivo

Direct recognition of CDNs by the receptor stimulator of interferon genes (STING), alternatively known as MPYS, TMEM173, MITA and ERIS, activates the production of type I interferons [43, 62-65]. Several studies have shown that CDG induces a type I interferon response that was dependent on STING [66, 67]. In vitro studies using bone marrow derived macrophages showed that cytosolic CDG stimulated IFN- production [61, 67]. The induction of type I interferon by CDG completely depends on STING [66, 68]. In addition to sensing CDG, STING plays a crucial role in response to cytosolic nucleic acids, such as double stranded DNA, and has been found to play protective roles against bacterial and viral pathogens [62, 63, 65, 69-72]. STING has been shown to play a role in inducing humoral and cellular immune response to DNA vaccines [71]. In addition, STING is also involved in the control tumor development and attenuation of autoimmunity, making this pathway ideal not only for mucosal vaccines but also for the development of cancer and autoimmunity therapies [44, 73].

CDG directly binds to the dimeric form of STING, which is found on the endoplasmic reticulum, to induce the activation of STING [43, 74]. CDG is believed to simultaneously activate two distinct pathways (Figure 1). Following the recognition of CDG by STING, TANK-binding kinase 1 (TBK1) is recruited and activated, which subsequently phosphorylates the transcription factor IRF3 [71, 75]. The phosphorylation of IRF3 then results in the production of type I interferon. STING also activates the NF-B transcription factor through the phosphorylation of IKK to trigger the expression of type I interferon and other inflammatory cytokines, such as TNF- [55, 76]. These studies were all performed in vitro, so it is not clear whether CDG activates these signaling pathways in vivo for its mucosal adjuvant activity.

Recently, work from our lab found that the mucosal adjuvant activity of CDG was completely dependent on STING in vivo [55]. We showed that intranasal immunization with CDG and ovalbumin (OVA) antigen of STING^-/- mice failed to produce antigen specific IgG and IgA in the lungs, and failed to induce Th1, Th2 and Th17 responses. CDG failed to induce inflammatory cytokines, such as TNF-, in the lungs of immunized STING^-/- mice. Furthermore, CDG was found to activate both the TBK1-IRF3-type I interferon and NF-B-TNF- pathway in bone marrow derived macrophages. It is important to note that these two pathways are STING dependent, but are also distinct and separate. Inhibition of IRF3 by the chemical inhibitor Bx-795 failed to produce type I interferon, however CDG was still able to induce TNF- production in the macrophages. This study provided, for the first time, that STING was required for the mucosal adjuvant activity of CDG in vivo.

c. The mucosal adjuvant activity of CDG depends on STING-dependent TNF- signaling

CDG can activate two STING-dependent pathways: (i) the TBK1-IRF3-type I interferon pathway and (ii) the IKK-NF-B-TNF- pathway. Importantly, stimulating type I interferon production is the signature function of STING, having a role in mediating the protective effects against bacterial and viral pathogens [63, 71, 72]. However, one of our key findings was that although CDG can activate STING-dependent type I interferon production in vitro, this was dispensable in vivo [55]. Intranasal immunization of IFNAR^-/- mice, which lack the type I interferon receptor, with CDG and OVA produced normal levels of antigen specific humoral and cellular immune responses [55]. Furthermore, knocking out TNF- signaling through the use of TNFR1^-/- mice resulted in reduced levels of humoral and cellular immunity in response to CDG immunizations. This highlights the importance of TNF- signaling, and not type I interferon, in mediating the mucosal adjuvant activity of CDG. Another study found that nanoparticle delivery of CDG acts through TNF- and type I interferon, although to different degrees [77]. Specifically, blocking TNF- with anti-TNF- neutralizing antibodies showed a reduction in serum IgG production, while blocking IFNAR1 had no effect on IgG levels. These studies provide evidence as to the importance of TNF- in mediating the mucosal adjuvant activity of CDG.

d. Cyclic di-GMP activates dendritic cells in vitro

Although CDG has been shown to have immunostimulatory effects when used as a vaccine adjuvant, the mechanism behind the adjuvant activity of CDG remains unknown. To date, the cellular mechanism of CDG has been only characterized by in vitro experiments with murine and human cells. Splenic DCs stimulated in vitro with CDG increased expression of activation markers CD86 and CD80 [61]. Furthermore, human immature DCs stimulated with CDG upregulates activation markers CD80 and CD86 as well as the maturation marker CD83 [61].

At a functional level, immature human DCs stimulated with CDG enhanced T cell proliferation in vitro, indicating that CDG has an effect on DC maturation and function. Furthermore, cytosolic CDG triggers the activation and maturation of bone marrow derived DCs in a STING-dependent manner [55]. It is interesting to note that treatment of STING-deficient bone marrow derived DCs with TNF- increased DC activation, since the adjuvant activity of CDG depends on TNF- in vivo. Important unanswered questions include: (i) whether DCs contribute to the mucosal adjuvant activity of CDG in vivo, (ii) how does CDG affect the functions of DCs in vivo; and (iii) whether CDG-induced TNF- affects DCs in vivo. The work within this dissertation will further examine the impact CDG has on pulmonary DCs.

C. Dendritic Cell Subsets and Immune Responses

DCs are a heterogeneous population made up of different subsets that are highly specialized and functionally distinct. In the lung, DCs comprise of conventional (cDCs), the potent antigen presenting cells, and plasmacytoid DCs (pDCs), producers of type I interferon [78]. cDCs express the integrin CD11c and MHCII, and can be further divided into CD103⁺ cDCs, CD11b⁺ cDCs and monocyte derived DCs (moDCs) [79-81]. pDCs are identified as CD11c^lowCD11b^–PDCA1⁺B220⁺ [82], express low levels MHCII and are not potent stimulators of T cells [83-85]. Lung DCs have become attractive therapeutic targets for the development of vaccines due to their ability to recognize foreign antigens and prime an adaptive response. DCs perform their essential role based on their anatomic location, where they are located throughout the epithelium and interstitium, extensive repertoire of pattern recognition receptors (PRRs), which is used to recognize and internalize pathogens, and migratory properties, which help in directly activating T cells [86, 87]. Pulmonary cDCs arise from distinct lineages and are equipped with unique set of receptors that play a role in sensing microbial stimuli, activating specific T cell subsets, and inducing tolerance [88].

DC maturation is not defined by markers, but rather is based on functionality. A fully mature cDC will take up and process antigen, upregulate MHCII and co-stimulatory molecules, have migratory capabilities, and must be able to activate T cells.  All cDCs can take up antigen either through receptor-mediated endocytosis or pinocytosis [89]. Functionally, receptor-mediated endocytosis by cDCs primary stimulates CD8⁺ T cells, while pinocytosis efficiently stimulates CD4⁺ T cells [90]. Mature cDCs will increase expression of C-C chemokine receptor type 7 (CCR7) to help migrate toward T cell zones in the tissue draining lymph nodes [91]. T cells require three signals from cDCs in order to fully develop: (i) antigen presentation on either MHCI or MHCII; (ii) co-stimulation through CD86 and CD80 receptors and; (iii) polarizing cytokines to direct the differentiation of a particular T helper cell subset [92]. It is becoming increasingly clear that different cDCs in the lung have unique functions that allow them to contribute to different aspects of the adaptive immune response, a concept termed “division of labor” [93].

1. Conventional Dendritic Cell Subsets

cDCs are lung resident and can be subdivided into two distinct subsets: CD103⁺ cDCs and CD11b⁺ cDCs. CD103⁺ cDCs are characterized as: CD11c⁺MHCII⁺CD103⁺CD11b^–CD24⁺CD64^–, while CD11b⁺ cDCs are CD11c⁺MHCII⁺CD103⁺CD11b^–CD24⁺CD64^– [94]. cDCs arise from the common DC progenitor in the bone marrow, express the cDC-specific transcription factor Zbtb46, and require the cytokine fms-like tyrosine kinase 3 (FLT3) ligand for their development [95-99]. Interestingly, cDCs do not require the cytokine, granulocyte/macrophage colony-stimulating factor (GM-CSF) for development, which is the key cytokine used to generate DCs in vitro [100, 101]. Once in the lung, the DC progenitor will commit to either the CD103⁺ or CD11b⁺ lineage [99].

a. CD103⁺ dendritic cells

CD103⁺ cDCs uniquely require the transcription factors IFR8, Id2, NFIL3 and Batf3 for their development [99, 102-104]. Early studies using the Batf3^-/- mouse model showed a selective absence of lung CD103⁺ cDCs, and the closely related lymphoid-resident CD8a⁺ DCs, while other DCs remained intact [103]. This was attributed to an intrinsic mechanism, as reconstitution of Batf3^-/- mice with normal hematopoietic cells generated CD103⁺ cDCs [103, 105].  These mice had no defects in other immune populations, did not succumb to DSS-colitis or bacterial infections and had normal CD4⁺ T cell responses [105].  The number of pre-DC progenitors were normal in the bone marrow and blood of Batf3^-/- mice, suggesting that this transcription factor plays a role in the terminal differentiation of CD103⁺ [103].

IRF8 is expressed in the common DC progenitor, and is increased in the pre-DC progenitor in the lung [103, 106]. Furthermore, pre-DCs were reduced in IRF8^-/- mice, suggesting that IRF8 plays a role in the commitment of the progenitors to the CD103⁺ cDC lineage [107]. CD103⁺ DCs require both IRF8 and Batf3 for their development, and absence of either transcription factor results in a complete loss of this subset.

CD103⁺ cDCs are mainly found embedded in the epithelial layer and express the epithelial-interacting integrin CD103 (E-cadherin–binding integrin E7) [108-110]. In addition, CD103⁺ cDCs express high levels of tight junction proteins Claudin-1, Claudin-7, and ZO-2 [109]. This allows the CD103⁺ cDCs to display long protrusions in between epithelial cells, allowing direct contact into the alveolar space for antigen sampling [109, 111, 112]. CD103⁺ cDCs express the chemokine receptor XCR1, CD207 (langerin), and the c-type lectin receptor Clec9a (DNGR-1) [113, 114].

One of the key functions of CD103⁺ cDCs is the enhanced ability to cross present antigens to CD8⁺ T cells.  CD103⁺ cDCs constitutively express higher levels of the MHCI peptide loading machinery and can cross present soluble and viral antigens to CD8⁺ T cells [115-118]. OVA-pulsed CD103⁺ cDCs from the lung preferentially stimulated CD8⁺ T cells, while having little to no effect on CD4⁺ T cells in vitro [116]. Moreover, CD103⁺ cDCs directly present antigens to stimulate the differentiation of effector CD8⁺ T cells, while having no effect on memory CD8⁺ T cells [117, 119, 120]. Batf3^-/- mice infected with influenza virus, Sendai Virus, West Nile Virus had impaired antigen cross presentation ability and failed to induce IFN- CD8⁺ T cells [103, 105, 119].

In addition, pulmonary CD103⁺ cDCs can: (i) take up and cross present self-apoptotic antigens for the maintenance of self-tolerance under homeostatic and inflammatory conditions; (ii) directly transport intact influenza virus to the lymph node; (iii) generate local IgA production through the activation of B cells and ;(iv) mediate anti-tumor immunity through a type I interferon dependent manner [121-125].

The role of pulmonary CD103⁺ cDCs in the activation of CD4⁺ T cells remains controversial. In certain cases, CD103⁺ cDCs have been found to be essential for the induction of Th2 cells to inhaled allergens, while others have found that CD103⁺ cDCs play little to no role in Th2 induction [126-128]. The induction of regulatory CD4⁺ T cells in the lung to provide airway tolerance is also thought to be dependent on CD103⁺ cDCs [129-131].

Several studies have shown that targeting CD103⁺ cDCs in vivo is a promising vaccination strategy. Targeting antigen to the XCR1 and DNGR-1 receptors, which are selectively expressed on CD103⁺ cDCs, was found to generate protective CD8⁺ T cell responses as well as induce a Th1 response [132-134].  In addition, targeting CD103⁺ cDCs for antigen delivery enhanced antibody production and formation of memory B cells [135]. These receptors act as endocytic receptor, allowing for the internalization of the antigen when administered to mediate the responses required for vaccines. It remains to be seen if CD103⁺ cDCs are required for mucosal vaccines, as these studies used intradermal and intramuscular routes.

b. CD11b⁺ dendritic cells

CD11b⁺ cDCs are the dominant cDCs in the lung and depend on the transcription factor IRF4 for development [96, 136]. This subset is functionally heterogeneous and remains less characterized. Conditional depletion of IRF4 in CD11c cells results in the total absence of CD11b⁺ cDCs in the lung, while other DCs remain intact [96, 106, 136]. IRF4 is important for the terminal differentiation of this lineage, as it is hardly expressed in circulating progenitors and IRF4^-/- mice had normal numbers of progenitors [106, 107, 137]. Unlike the closely related lymphoid-resident CD4⁺CD11b⁺ DCs and lamina propria CD103⁺CD11b⁺ cDC, pulmonary CD11b⁺ cDCs are unique in that their development does not depend on Notch2 or RelB [138, 139]. The transcription factor KLF4 plays differential roles in the development and maturation of lung CD11b⁺ cDCs [137]. Early depletion of KLF4 resulted in the loss of progenitors committed to the CD11b⁺ cDC lineage, skewing the progenitors towards the CD103⁺ lineage. In mature CD11b⁺ cDCs, KLF4 is expressed only by a small subset and is needed to induce Th2 responses [137].

CD11b⁺ cDCs reside in the lamina propria of the lung, and are defined based on the expression of signal regulatory protein (SIRP)- and the fractalkine receptor CX3CR1 [109, 140]. Initially thought to be the source of proinflammatory cytokines, these cells are particularly important in the priming and restimulation of CD4⁺ T cells in the lung [116, 127, 141-143]. CD11b⁺ cDCs are required for the development of Th2 immunity in an asthma model [144]. CD11b⁺ cDCs express high levels of the MHCII pathway molecules and lower levels of the MHCI pathway, indicating that this subset does not have the capacity to present to cross present to CD8⁺ T cells [125, 143]. However, several studies have indicated that under infection and inflammatory conditions, pulmonary CD11b⁺ cDCs have the ability to stimulate the generation of CD8⁺ memory T cells, though much later during infection [119, 145].

Selectively depleting the CD11b⁺ cDCs in the lung through the use of the IRF4^fl/flCD11c^cre mouse model have identified the key functions of this subset. IRF4-dependent CD11b⁺ cDCs are necessary for Th2 lung inflammation, but dispensable for Th1 responses to viral infection in the lung [136, 146]. Interestingly, a subset of the KLF4-expressing CD11b⁺ cDC in the lung were found to be specifically important for the induction of Th2 immunity [137]. Furthermore, CD11b⁺ cDCs are also important for the generation of Th17 responses to fungal infection by producing Th17 polarizing cytokines IL-23 and IL-6 [96]. These CD11b⁺ cDCs depend on IRF4, but not KLF4, to induce Th17 immunity [137].

IRF4-dependent DCs were found to be important for the generation of regulatory T cells [143]. Specifically, these DCs expressed tolerogenic markers, such as the enzyme aldehyde dehyrogenase (ALDH), programmed death-ligand (PD-L)- 1 and -2, and had impaired peripheral tolerance. In other organs, CD11b⁺ cDCs induced peripheral regulatory T cells through the production of retinoic acid [144, 147-149]. It remains to be seen whether CD11b⁺ cDCs are required for the generation of regulatory T cells in the lung.

Several studies have implicated CD11b⁺ cDCs in the regulation of humoral immunity. Depletion of cDCs with CD11c-DTR mice completely reduced antibody production against allogenic red blood cells (RBC) [150]. Specifically, IFR4^fl/flCD11c^cre mice failed to produce antibody, while Batf3^-/- mice had normal levels of anti-RBC antibodies. This is in line with several other studies, which suggest that CD11b⁺ cDCs are required for the induction of T follicular helper cells and the formation of germinal centers [151]. In addition, IRF4-dependent CD11b⁺ cDCs were found to be the DC subset in the gut necessary for IgA production [152, 153].  In the lung, CD11b⁺ cDCs produce TGF- to induce the production of IgA by B cells following intranasal administration of CT [123]. Taken together, CD11b⁺ have the unique ability to not only initiate and regulate CD4⁺ T cell responses in the lung, but to also regulate B cell responses.

2. Monocyte-derived dendritic cells

The exact role of moDCs in the lung were not fully identified until recently due to the phenotypic similarities they share with CD11b⁺ cDCs, such as expression of CD11b⁺, SIRP- and CX3CR1 [93]. moDCs however express monocyte defining markers such as Ly6c, CD64 and the high-affinity IgE receptor FcR1 chain (MAR-1) [127, 154, 155]. These moDCs arise from Ly6chi classical monocytes and require the C-C chemokine receptor 2 (CCR2) for migration to the lung [156]. Once in the lung, monocytes will upregulate MHCII, CD11c, and CD11b to become moDCs [157]. Although these cells increase in number during inflammation, moDCs are found to be present in the lung at steady-state [93, 127, 158].

While first identified as TNF-/iNOS producing (tip)-DCs, recent studies have highlighted the ability of moDCs to be prominent stimulators of adaptive immunity in the lung [159, 160]. moDCs play crucial roles in the defense against viral and bacterial infections in the lung, as well as contribute to the induction of asthma, through the modulation of T cells [93, 160]. Pulmonary moDCs have a Th1 bias when stimulating CD4⁺ T cells in vitro, indicating that these cells can directly influence the differentiation of T cells [127]. However, this is not the case in vivo. During viral infection, moDCs were required for the reactivation of Th1 cells and memory CD8⁺ T cells, but were dispensable for the priming and recruitment of Th1 cells during viral infection [161, 162]. Similarly, CCR2^-/- mice, which lack moDCs, were shown to have significantly reduced numbers of viral-specific CD8⁺ T cells in the lung, but not in the draining lymph nodes, during influenza infection [160]. In an asthma mouse model, moDCs were able to induce Th2 immune response [127]. Adoptive transfer of antigen-loaded moDCs showed that these cells are capable of inducing Th2 cells in vivo [127].

moDCs are highly efficient at taking up antigen, however are poorly migratory [127, 163]. These DCs do not express the migration marker CCR7, and are not found in the lung draining lymph nodes. Under low-dose allergen exposure, moDCs fail to migrate to the lymph nodes and induce Th2 cells [127]. However, exposure of very high doses of the allergen house dust mite (HDM) in Flt3^-/- mice, which only have moDCs, induce Th2 cell mediated immunity. This suggests that moDCs primarily interact with T cells in the lung instead of the lymph node.

D. Summary

Previous work in our lab has established the requirement of STING in mediating the mucosal adjuvant activity of CDG [55]. Specifically, this work emphasizes on the requirement of TNF- signaling for the induction of humoral and cellular immunity in vivo. Taken together, we believe that CDG induces STING-dependent TNF- production, which is needed for the protective, humoral and cellular immune responses in vivo (Figure 2). The work within this dissertation extends our previous knowledge by demonstrating that CDG can specifically target and activate unique functions of pulmonary DCs, therefore providing insight into the in vivo mechanism of CDG. Further investigation reveals a role for CD11b⁺ cDCs in mediating the adjuvant activity of CDG and characterization of this population delineates a previously unappreciated functional heterogeneity. Lastly we demonstrate a two-signal requirement for the maturation of pulmonary DCs by CDG, which directly depends on TNF- signaling. Thus, this work emphasizes on the relationship between CDG and pulmonary DCs function.

Conclusions

The overall purpose of this dissertation was to obtain a detailed understanding of the interaction between the mucosal vaccine adjuvant CDG and pulmonary DCs. The experiments herein, specifically examine the direct and indirect effect CDG has on DCs in vivo in order to define the cellular and molecular mechanisms required for the mucosal adjuvant activity of CDG. A particular focus was to characterize the differential roles of pulmonary DC subsets in CDG-induced immune responses. Furthermore, special efforts were aimed at understanding the molecular mechanism of CDG-induced DC function in the different subsets. Work from this dissertation provides insight into the function of pulmonary DCs in regards to a potent mucosal adjuvant, and allows for a better understanding of how to target their specific functions when designing effective mucosal vaccines.

A. Characterization of the in vivo response to CDG

1. CDG directly targets and activates pulmonary DCs in vivo

Lung DCs have become attractive therapeutic targets for the development of mucosal vaccines due to their diverse functions and ability to tailor immune responses.  This study shows, for the first time, a total requirement of pulmonary DCs for the activation of antigen-specific humoral and cellular immunity after immunization with CDG. Mice depleted of STING, the receptor for CDG, in CD11c⁺ DCs were unable to induce Th1, Th2 and Th17 responses as well as produce antigen-specific antibodies. Furthermore, adoptive transfer of CD11b⁺ cDCs into STING^-/- mice rescued humoral immunity, demonstrating that pulmonary DCs are sufficient and necessary for the adjuvant activity of CDG. Our results show that CDG directly targets and activates pulmonary DCs in a STING-dependent manner to elicit adaptive immune responses in vivo.

To this day, STING activation of DCs has been limited to in vitro work, using bone marrow derived DCs transfected with CDNs and other STING agonists [205, 206]. The use of bone marrow derived DCs has its limitations, as these DCs have no similarities to tissue resident DCs [100]. Secondly, transfection of CDG and other CDNs is very artificial as CDG is a charged molecule, thus unlikely to cross the membrane directly. We have demonstrated for the first time that CDG directly targets the STING pathway in pulmonary DCs and that STING plays an intrinsic role in DC activation and maturation.

2. CDG preferentially activates pinocytosis-efficient DCs in vivo

Targeted delivery of antigen to DCs in vivo has promising potential in mucosal vaccine design, and can be enhanced through the use of adjuvants. As previously documented, DCs are professional APC and have the ability to present antigen to T cells, irrespective of whether it is through pinocytosis or receptor-mediated endocytosis in vitro [89, 207]. However, targeting certain endocytic pathways can generate much stronger T cell responses [207-209]. We demonstrate, through intranasal administration of fluorescently labelled antigens, that CDG selectively activates pulmonary DCs to take up antigen through pinocytosis in vivo. In addition, pulmonary CD11b⁺ cDCs and CD103⁺ cDCs efficiently take up CDG, increase activation through production of cytokines, and enhance maturation markers, such as CD86 and CCR7, to induce humoral and cellular responses.

Our findings demonstrate that the route of antigen internalization is important for the adjuvanticity of CDG. These results also highlight that CDG, a charged molecule with two phosphate groups, could also be taken up through pinocytosis. Although from these in vivo studies we could not determine whether pinocytosis was directly required for the adjuvant activity of CDG, these results are consistent with several, primarily in vitro, observations that the route of antigen uptake results in differential functions of DCs [90, 210]. It remains to be seen how CDG gets into the cytosol to interact with STING, and the mechanism by which STING enhances antigen uptake.

It is interesting to note that a population of DCs, which take up antigen through receptor-mediated endocytosis are also activated in response to CDG. This population is small, around ~25% of total DCs, and is primarily made up of CD103⁺ cDCs. In vitro studies have implicated receptor-mediated endocytosis with cross-presentation and activation of CD8⁺ T cells, while pinocytosis is important for the activation of CD4⁺ T cells [90, 211]. It is well documented that CD103⁺ cDC can mediate CD8⁺ T cell and Th1 responses in vivo, thus it is possible CDG differentially enhances antigen uptake in CD103⁺ cDCs. It is also possible that this population of DCs, originally thought to be homogenous due to phenotypic and developmental characterizations, can be functionally heterogeneous, which warrants further investigation.

B. Pulmonary DCs have differential roles in mediating the adjuvant activity of CDG

1. CD11b⁺ cDCs play a central role in mediating the adjuvant activity of CDG

Until recently, CD103⁺ cDCs have been the center of interest for vaccination strategies due to their anatomic location, ability to directly acquire antigen, and for their enhanced ability to cross present antigen to CD8⁺ T cells. However, these characteristics only make this DC subset suitable in the defense against intracellular pathogens. An ideal target would be able to generate robust Th1, Th2 and Th17 responses as well as the production of antigen-specific antibodies. We demonstrate, through DC subset-specific knockout mouse models, that in CDG-adjuvanted vaccines, CD11b⁺ cDCs play a superior role in mediating humoral and cellular responses in the lung. Loss of this subset completely abrogated all immune responses, including Th1 cells which require CD103⁺ cDC.

Our results indicate that priming of Th1 T cells is mediated by CD103⁺ cDCs, but the local differentiation or retention of tissue memory Th1 cells depend on CD11b⁺ cDCs. Furthermore, Th1 responses are also diminished in TNFR2-deficient mice, in which CD103⁺ cDC function remains normal, signifying a requirement for fully mature CD11b⁺ cDCs for mediating Th1 responses. It is entirely possible that CD103⁺ cDCs are required for the differentiation of Th1 cells, but CD11b⁺ cDCs are required for the homing or even the generation of memory T cells in the lung. Notably, human lung-resident CD1c⁺, which are related to murine CD11b⁺ cDCs, induce lung homing markers on CD8⁺ T cells and while mouse CD11b⁺ cDCs are important in generating CD8⁺ memory T cells in response to influenza vaccine antigens [119, 212, 213]. In light of the importance of CD11b⁺ cDCs as the main regulators of CDG-induced mucosal immune responses, it will be important to understand how CD11b⁺ cDCs can promote long-term responses in the lung.

2. CDG specifically targets a subset of CD11b⁺ cDCs for systemic immunity

CD11b⁺ cDCs represent a major subset of cDCs in the lung, however their in vivo functional activities remain to be fully elucidated. One of our novel observations was that CDG uniquely targets TNFR2^–CD11b⁺ cDCs to mediate systemic immunity. In our model, TNFR2^–CD11b⁺ cDCs can directly take up CDG and antigen, increase expression of CD86, MHCII, CCR7, as well as TNFR2, and migrate to the draining lymph nodes to stimulate adaptive immune responses. We favor this hypothesis for several reasons:

(i) TNFR2-expressing cells are the only CD11b⁺ cDCs able to take up antigen;

(ii) the majority of activated CD11b⁺ cDCs express TNFR2;

(iii) TNFR2-deficient DCs cannot induce adaptive immunity to CDG and;

(iv) adoptive transfer of TNFR2^–CD11b⁺ cDCs induces humoral immunity in response to CDG.

These observations are of great importance since targeting select subsets of DCs that are immunogenic and migratory, as opposed to those that are tolerogenic, is ideal for mucosal vaccinations [214, 215]. By identifying this subset, it will be important to further identify molecular and surface markers to better target this subset for the induction of systemic immunity.

We also found that TNFR1^-/- mice had a complete reduction in CDG-induced immunity in the lung but not systemically. This suggests that TNFR1 signaling, and soluble TNF-, is somehow important for CDG-induced mucosal immunity. This is in line with several recent publications which show that TNFR1 is implicated in regulating local immune responses to inhaled antigens, although it is not clear by which mechanism [216, 217]. Together, our results demonstrate a novel finding by which sTNF- and mTNF- differentially regulate mucosal immunity in vivo. We show for the first time that mTNF- is required for the direct maturation of CD11b⁺ cDCs, while the role of sTNF- in mediating CDG-induced mucosal immunity remains unknown. TNFR1 signaling can enhance the expression of maturation markers on human moDCs through soluble TNF-, it is therefore possible that sTNF- could potentially stimulate moDCs in the lung to induce mucosal immunity in response to CDG [218].

3. A role for TNFR2⁺CD11b⁺ cDCs in mediating tolerance

CD11b⁺ cDCs are extremely heterogeneous, and identification these DCs at a functional level has yet to be seen [219].  We demonstrate for the first time that CD11b⁺ cDCs in the lung can be characterized into two functional subsets based on expression of TNFR2. At steady-state, a small population of CD11b⁺ cDCs, ~30%, constitutively express TNFR2, PDL-1 and activate RelB. This is consistent with previous in vivo findings, which suggest that RelB activation in DCs at steady-state induce T regulatory cells for peripheral tolerance, as well as in vitro findings, which suggest that CD11b⁺ cDCs have the highest expression of RelB at steady-state [203, 204]. It is not known whether this DC population is required for maintaining homeostasis through the generation of T regulatory cells, or whether they play a local role in regulating lung immune cells, such as activating moDCs or regulating memory T cells. In light of the importance of maintaining tolerance in the lung during immunization, it will be important to understand how this CD11b⁺ cDC subset functions in response to CDG.

C. CDG differentially activates pulmonary DCs

Transcriptional profiles of human and mouse cDCs describe unique transcriptional signatures that control DC function at steady state and during immunization [88, 219, 220]. One unique observation that we found is that CDG can induce NF-B activation in both CD11b⁺ cDCs and CD103⁺ cDCs differently. In response to CDG, CD11b⁺ cDCs activate the NF-B pathway through RelB, but does not activate RelA. In contrast, CD103⁺ cDCs activate RelA but not activate RelB in response to CDG. This is novel as we provide evidence that RelB is important for CDG-induced maturation in vivo. The activation of RelB depends on TNFR2 signaling in response to CDG. We believe that the function of RelB depends on the nature of the stimuli, as RelB-deficient DCs stimulated with TLR2 agonists still mature and prime T cells, while stimulation of bone marrow derived DCs with LPS enhanced RelB activation and maturation [138, 221, 222]. Our results signifying the importance of understanding how CDG interacts with cDCs in order to design more effective mucosal vaccines.

Significance

The data within this dissertation represents a series of novel findings that further elucidates the in vivo mechanism of the mucosal adjuvant CDG. First, it provides insight into the effect of CDG on pulmonary DC activation in vivo. This topic is relevant as adjuvant modes of action are very different and can directly or indirectly act on DCs. Our data defines a direct effect of CDG on DC function by enhancing antigen uptake, and further showing that the route of antigen entry is important in activating DC functions. Second, we provide evidence that CDG can also indirectly activate a subset of DCs through TNF- signaling. While the current notion is that CDG induces STING-dependent type I interferon production, our data suggests that it is TNF- that is important for mediating CDG activity. This finding is important as it sheds light on the complexity of DC maturation, a key feature of DCs required by adjuvants to elicit the required responses, and defines a signaling pathway that can be targeted when designing vaccines. Third, we demonstrate a unique role for CD11b⁺ cDCs in mediating the adjuvant activity of CDG in vivo. This DC subset expresses distinct receptors and activates unique transcriptional pathways that are not evident in other DC subsets. Taken together, our findings highlight the unique function of pulmonary DCs and their activation requirements in response to the mucosal vaccine adjuvant CDG. Our understanding of CDG activity and its effect on pulmonary DCs at this level in mucosal vaccines could provide also important insight into DC function in other therapeutic models, inducing cancer, allergy, and autoimmunity.