Duality at the gate: Skin dendritic cells as mediators of vaccine immunity and tolerance

Abstract

Since Edward Jenner's discovery that intentional exposure to cowpox could provide lifelong protection from smallpox, vaccinations have been a major focus of medical research. However, while the protective benefits of many vaccines have been successfully translated into the clinic, the cellular and molecular mechanisms that differentiate effective vaccines from sub-optimal ones are not well understood. Dendritic cells (DCs) are the gatekeepers of the immune system, and are ultimately responsible for the generation of adaptive immunity and lifelong protective memory through interactions with T cells. In addition to lymph node and spleen resident DCs, a number of tissue resident DC populations have been identified at barrier tissues, such as the skin, which migrate to the local lymph node (migDC). These populations have unique characteristics, and play a key role in the function of cutaneous vaccinations by shuttling antigen from the vaccination site to the draining lymph node, rapidly capturing freely draining antigens in the lymph node, and providing key stimuli to T cells. However, while migDCs are responsible for the generation of immunity following exposure to certain pathogens and vaccines, recent work has identified a tolerogenic role for migDCs in the steady state as well as during protein immunization. Here, we examine the roles and functions of skin DC populations in the generation of protective immunity, as well as their role as regulators of the immune system.

Keywords:

• dendritic cells
• DC targeted vaccines
• immune tolerance
• migratory DCs

Abbreviations

DCs	=	Dendritic Cells
APCs	=	Antigen Presenting Cells
LCs	=	Langerhans Cells
migDC	=	migratory DCs
FLT-3	=	Fms-like Tyrosine Kinase 3
STAT3	=	Signal Transducer and Activator of Transcription 3
LN	=	Lymph Node.

Introduction

The immune system is responsible for defending its host against a wide variety of pathogens on a daily basis. Two key characteristics of this protection are the generation of specific, sterilizing immunity, and the formation of long lived immunological memory, both of which require an adaptive immune response. Clinically, the protective benefits of the immune system have been successfully harnessed through the development of vaccines. However, the cellular mechanisms underlying long-term protective immunity remain incompletely characterized.

The Dendritic Cell (DC), was discovered in the 1970s by Ralph Steinman and Zanvil Cohn.¹ As the major antigen presenting cells (APC) responsible for driving the generation of the adaptive immune response, DCs are central to establishing long-term protective immunity. Classically, DCs are dependent on the transcription factors ID2² and PU.1,^3,4 as well as FMS-like Tyrosine Kinase 3 (Flt3)^5,6 and Signal Transducer and Activator of Transcription 3 (STAT3)⁷ signaling. In vivo, DCs generate potent, cytotoxic CD8⁺ T cells,⁸ as well as regulate the polarization of helper CD4⁺ T cells toward the appropriate subtype (T_H1, T_H2, T_H17, etc).⁹ Two specific populations of DC, which will not be discussed further but are mentioned here briefly, are plasmacytoid DCs and monocyte-derived DCs. Specialized for innate anti-viral immunity, plasmacytoid DCs express TLR7 and TLR9, and are the primary producers of type I IFNs in response to viral infections.^10,11 DCs may also arise from monocytes in the setting of inflammation, taking on antigen presentation function to both CD4⁺ and CD8⁺ T cells,¹² as well as an anti-microbial role through the production of TNFα and iNOS.¹³

In addition to plasmacytoid and monocyte derived DCs, there are several unique populations of DCs,¹⁴ which can be broadly grouped into 2 categories based on their anatomical location in the steady-state: lymphoid resident DCs ( also know as classical DC), or migratory DCs (migDC). Lymphoid resident DCs are found in the spleen and LNs of both mice and humans, originate from blood precursors, and include CD8α⁺ and CD11b⁺ subsets.¹⁵ In contrast, migDCs are comprised of several different populations (as discussed in detail¹⁶), which constitutively traffic to the local draining LN from a non-lymphoid organ, such as the skin, lungs, or gut, while being notably absent from the spleen.¹⁷ MigDC in the skin are perhaps the best characterized of these tissue DC populations, though their role in the generation of protective immunity is not fully understood.

The skin is a unique immune environment, hosting both a diverse network of immune cells,¹⁸ as well as a significant microbiome made up of both fungal¹⁹ and bacterial²⁰ populations. Furthermore, when directly compared to other routes of vaccine administration, such as intraperitoneal or intramuscular, cutaneous immunization offers superior cellular and humoral immune responses to both influenza²¹ and vaccinia,²² suggesting that the skin might be a specialized site for immune priming. As one such example, intradermal influenza is required at just one fifth of the dose needed via intramuscular injection, to achieve comparable vaccine titers and serum conversion.²¹ However the cellular mechanisms underlying enhanced immunization via the skin are incompletely understood and afford a key opportunity to improve rational vaccine design.

In addition to hosting numerous commensal species, the skin is a site of high cellular turnover, between hair follicles and regular keratinocyte replacement. These features contribute to the unique nature of the skin, and play a significant role in the regulation of the immune system, as it must discriminate between normal cellular turnover and infection induced cell death. Hair follicles are considered to be an immune privileged site,²³ yet when mechanically stimulated, produce CCL2 and CCL20, recruiting Langerhans Cell precursors and allowing the entry of immune cells into the epidermis.²⁴ However, despite high cellular turnover, healthy skin is not constantly inflamed, as the cell death of keratinocytes is closely regulated. Indeed, in mice with keratinocytes deficient in either FADD²⁵ or caspase-8,²⁶ skin lesions and constitutive inflammation develop. The skin must maintain tolerance to self, as well as to commensal bacteria, while responding appropriately to dangerous pathogens. This delicate balance is maintained through a specialized network of dendritic cells.

In the skin of both humans and mice, there are a number of individual DC populations (4 or 5 respectively) as defined by both extracellular markers²⁷ as well as dependence upon specific transcription factors.^14,28-31 Langerhans Cells (LCs) are the only DC located in the epidermis, where they maintain surveillance of tight junctions,^32-34 while binding broadly with keratinocytes through E-cadherin binding.³⁵ In the dermis, an interstitial network of DCs is comprised of at least 3 individual populations, in addition to LC migrating through the dermis en route to the skin draining LN.^31,36-39 These populations include CD103⁺ DC (BDCA3⁺ in humans), CD11b⁺ DC (BDCA1⁺ in humans), and CD11b⁻ Langerin⁻ DC (no human counterpart identified as of yet, although a BDCA3⁺BDCA1⁺ human DC subset exists which has yet to be correlated to a mouse counterpart). Together with LCs, the dermal skin DCs form a complex system responsible for maintaining the equilibrium between barrier immunity and self-tolerance.

Unique Properties of Skin DC

Skin DCs survey the skin for pathogens and are capable of migrating to skin draining LNs bearing antigens to present to T cells (Fig. 1). This migration is CCR7 dependent, as mice genetically deficient for CCR7 mice do not have any migDCs in their cutaneous LNs.¹⁷ Unexpectedly, skin DC constitutively migrate in both germ free and MyD88/TRIF knockout mice,⁴⁰ but fail to migrate when NF-κB signaling is inhibited in DCs,⁴¹ suggesting that the migratory program occurs independently of TLR signaling, but requires NF-κB. However, irritant induced inflammation,⁴² TLR ligation,⁴³ and infection⁴⁴ result in increased trafficking, demonstrating that skin DC migration is responsive to immune stimuli. In addition, recent studies have demonstrated that skin DC will also increase their migration in response to low molecular weight hyaluronan through a TLR4 dependent mechanism, suggesting that skin DC may respond to a wide variety of stimuli.⁴⁵

Figure 1. Graphical Schematic of Skin DC Migration to the Cutaneous Lymph Nodes. Epidermal (Langerhans Cells) and dermal skin DC colonize their respective layers of the skin. In both the inflamed and steady state setting, LC and dermal DC enter the lymphatic vessels, before migrating in a CCR7 dependent manner to the local cutaneous draining lymph node bearing either pathogen derived or self antigens. Upon arrival at the lymph node, skin migratory DC enter the paracortical area and present skin derived antigens to T cells. Additionally, post migration, migratory DC can capture and present antigens that have drained freely, such as DEC205 targeted vaccinations. Images were provided by Matthew Woodruff (Wikimedia commons).

We and others have observed that skin DCs (in the steady state) are transcriptionally far more closely related to each other than to their LN resident counterparts.^28,46,47 This finding was initially surprising, as some skin and LN populations share both common precursors, as well as a capacity for functionally similar properties, such as cross-presentation. Furthermore, even LCs are more closely related to the other skin DC populations,²⁸ despite arising from a unique developmental pathway, more akin to monocytes and microglia than the remaining skin resident DCs.^48,49 Moreover, skin DC in the steady-state and during Flt3L treatment maintain a transcriptional signature that is both mature as well as tolerogenic,^28,46 while lymphoid resident DC tend to selectively express these markers during maturation, highlighting broad programming differences between these 2 groups.

Phenotypically, skin (and other tissue-derived) migDCs can be distinguished from LN resident DCs based on the expression level of classical DC identifiers; MHC II (or HLA-DR in human) and CD11c. While migDCs are identified as MHCII^hiCD11c^int, LN resident DCs can be distinguished as MCHII^intCD11c^hi cells in the steady state. However, during an immune response to an infection, this distinction blurs- LN resident DCs upregulate MHC II and therefore resemble migDCs. Interestingly, Flt3L treatment can expand the numbers of both LN resident DCs, as well as migDCs in the skin and skin draining LN, with the notable exception of Langerhans cells.^6,39 However, in contrast to LN resident DCs, migDCs constitutively express high levels of the classic DC activation markers CD80 and CD86, despite having an immune dampening effect in some models, as will be discussed later. These findings, among others, have stimulated a discussion concerning the definition of a “mature” or even “semi mature” DC based on phenotypic markers, instead opting for functional studies about the licensing state of the DCs in question.⁵⁰ Nevertheless, skin migDCs are a unique network of several DC populations with distinct functional and phenotypic characteristics.

Approaches to Studying Skin DC

A number of different approaches have been utilized to study the function of skin DC. Among these, preclinical models allowing for the restriction or ablation of individual DC populations have been indispensable for understanding how these populations function both individually and as a network.⁵¹ These models drive the human diphtheria toxin receptor under a promoter of specific marker genes, such as Langerin,^52,53 CD11c,⁸ or Zbtb46.⁵⁴ Each of these models results in the ablation of a unique set of cells, sometimes with overlapping results (Table 1). For example, use of the CD11c-DTR model results in ablation of classical DCs preferentially but also ablates some macrophages, monocytes, and pDC,⁸ while the Zbtb46 DTR model ablates only classical Flt3 dependent DC in both lymphoid and non-lymphoid organs.⁵⁴ The use of a Langerin-DTR can deplete both LCs and a dermal DC population that expresses both Langerin and CD103 (but is quite functionally and developmentally distinct from LC).^52,53 However, when bone marrow from Langerin-DTR mice is used to reconstitute a wildtype animal, only the dermal Langerin⁺ CD103⁺ DC population is sensitive to the administration of diphtheria toxin, as LCs are radioresistant, and host LCs (no DTR expression) are left intact from the chimerization.⁵⁵ Another method to selectively delete LCs is the use of Langerin-DTA model, where diphtheria toxin is driven under the human Langerin promoter, resulting in constitutive deletion of LCs.⁵⁶ However, only the LCs, both in the epidermis and migrating through the dermis, are affected in this model, leaving the other dermal resident Langerin⁺ CD103⁺ DCs intact.⁵⁵

Table 1. Approaches to study skin DC

Approach	Experimental Model	Summary	References
Acute Ablation	CD11c-DTR	Deletion of classical DC, PDC, and some monocytes, macrophages	^8
	Zbtb46-DTR	Deletion of classical DC, demonstrated loss of CD103^+, CD11b^+, and CD11b^−, unclear for LC	^54
	Langerin-DTR	Deletion of LC and dermal CD103^+ DC. When used in bone marrow chimera models, only CD103^+ dermal DC are deleted.	^52,53
Constitutive Deficit	Langerin-DTA	Diptheria toxin expressed under the human langerin promoter, resulting in constitutive deletion of Langerhans Cells	^55
	Batf3 −/−	Selective loss of CD103^+ dermal DC and CD8α^+ LN resident DC	^59
	CD11c-Cre IRF4 ^fl/fl	Restriction of CD11b^+ dermal DC to the skin	^62
	CCR7 −/−	Restriction of migDC to the skin, additional defects in Treg, T central memory, and B cell localization	^17
Topical Application	FITC Painting	Dibutyl phthalate acetone sensitization of the skin along with application of FITC	^63
	Dinitrochlorobenzene Sensitization	Regular application of sensitizing agent models contact hypersensitivity	^64
	Imiquimod/Resiquimod	Regular application of TLR7/8 agonist models human psoriasis	^65
	Tape Stripping	Mechanical removal of the stratum corneum, stimulates LC migration	^66
Skin Trophic Infections	Vaccinia Virus	Virally derived antigens are presented to T cells by CD103^+ dermal DC	^74,75
	Herpes Simplex Virus	CD103^+ dermal DC are involved in transporting antigen to the draining LN	^71,72
	Leishmania	Leishmania derived antigens are presented by LN resident DC, while Langerhans Cells dampen the response	^77–79
	Candida Albicans	Langerhans Cells are required for the generation of a TH17 response, while CD103^+ dermal DCs generate a protective TH1 response.	^67-70
	Staphlycoccus Aureus Derived Exfoliative Toxin	Langerhans Cells are required for the production of IgG1 to neutralize exofoliative toxin	^34
Imaging	CD11c-mCherry	Expression of mCherry under the CD11c promoter	^80,81
	Langerin-EGFP	Expression of EGFP under the langerin promoter	^52
	CD11c-EYFP	Expression of YFP under the CD11c (ITGAX) promoter	^141
Isolation Protocols	Skin and Skin Draining LN Digestion	Digestion of skin using a combination of dispase, DNAse, and collagenase to isolate single cell suspensions	^29,42,83,84
	Skin Crawlout	Culture of whole skin explant, with or without the addition of external chemokines, allowing DC to leave the skin	^39,82
In vivo DC Targeting	DEC205 Targeting	Vaccine fusion protein targeted to both migDC and LN resident DC. CD8α^+ LN resident DC stimulate T cell immunity, while migDC inhibit it	^46,86,90,91,93–95,142
	DEC207 (Langerin) Targeting	Vaccine fusion protein targeted to CD103^+ Dermal DC, Langerhans Cells, and CD8α^+ LN resident DC	^87
	DEC209	Vaccine fusion protein targeted to several DC like populations, including monocytes, LN resident DC, and plasmacytoid DC	^88
	DNGR-1 (Clec9a)	Vaccine fusion protein targeted to several DC like populations, including CD8α LN resident DC and CD103^+ dermal DC	^89

While several models of inducible ablation have been established, there is no known model that can specifically ablate all skin migDCs while leaving cDCs intact. One model that has been used to study the effects of migDCs as a whole is the CCR7 knockout mouse. However, while migration of migDCs to the skin draining LN is blocked in these mice (thereby removing them from LN during immune priming), CCR7 also plays a critical role in central memory T cells, T regulatory cells, and B cells, and its loss can alter LN architecture, while naïve and effector T cell are left largely intact. This model may be used in selective contexts (such as excluding a role for CCR7 dependence when immune responses are not lost), but cannot pinpoint a role for migDC exclusive of other CCR7-dependent populations. Thus it can be very useful for broad definitions, but is suboptimal for more detailed study of migDC function.^57,58

In addition to diphtheria toxin driven ablation strategies, several models have been established where DC subsets fail to develop due to the loss of requisite transcription factors. Batf3 knockout mice are deficient in LN resident CD8α⁺ DCs as well as CD103⁺ dermal DC (discussed in detail below), while leaving other cell populations intact.⁵⁹ In Batf3 deficient mice, exogenous or inflammation derived IL-12 results in the recovery of the CD8α⁺ DC subset, but not the CD103⁺ DCs.^60,61 Another model, CD11c-Cre IRF4 ^fl/fl, has been found to prevent the accumulation of CD11b⁺ dermal DC in the skin draining LN, despite their continued presence in the skin.⁶²

The application of topical agents may help define the role of skin DC in inflammatory contexts. “FITC painting,” involves the application of the fluorescent molecule FITC along with an irritant, such as dibutyl phthalate acetone.⁶³ After painting, migDC rapidly transport FITC from the skin to the skin draining LN. Similarly, application of DNCB results in contact hypersensitivity, though unlike the FITC model, inflammation from this model seems to by driven by T_H1 cells.⁶⁴ Repeated application of TLR 7/8 agonists (Imiquimod or Resiqiumod) can model human psoriasis in mice, including increased keratinocyte proliferation, recruitment of multiple immune subsets, and dependence on increased IL-17 signaling.⁶⁵ Mild injury and wounding is modeled through “tape stripping,” which removes part of the stratum corneum while leaving the hair follicle intact.⁶⁶ This mechanical disruption of the epidermis induces inflammation and increases migration of LCs to the skin draining LN, while increasing recruitment of new LC precursors to the affected area.⁶⁶ This model has often been used in conjunction with the topical administration of a target antigen,³⁴ such as ovalbumin, to study T cell responses to topical treatments. In addition to these topical treatments, a number of skin-tropic infections have been studied in the context of skin DC, including Candida Albicans,^67-70 HSV,^71-73 vaccinia virus,^74-76 Staphlycoccus,³⁴ and Leishmania.^77-79 As these pathogens are predisposed to infect the skin, skin DCs have a unique role in the generation of the immune response to them, as will be discussed later on.

Imaging studies of the skin have shed light on the anatomical structure and location of skin DCs, highlighting their location both in the skin as well as in the LN (Fig. 2A). In addition to conventional immunofluorescence, imaging studies may utilize mouse models where fluorescent reporters are expressed under DC specific promoters, such as CD11c⁸⁰ and Langerin,⁵² allowing for either live, in vivo imaging or immunoflourescent staining. Reporter mice have been invaluable in demonstrating the increased motion of LCs following tape stripping,⁵² localization of DCs at the tumor border,⁸¹ or visualizing the interactions between T cells and DCs during the immune response.⁸⁰ In addition to visualizing skin DC, specialized isolation protocols have been developed to isolate skin DC, either by directly digesting the skin²⁹ or skin draining LNs,⁴² or by allowing DC to “crawlout” in culture.^39,82 Furthermore, enzymatic digestion allows for a physical separation of epidermis from dermis, in both murine⁸³ and human⁸⁴ skin, thereby isolating anatomically distinct cell populations prior to further processing.

Figure 2. Migratory DC in the Skin and Cutaneous Draining Lymph Node. (A) Murine skin was stained for CD11c (purple), Langerin (green), CD11b (red), and DAPI (blue) in the steady-state. Images were taken at 20 × and are courtesy of Sze-Wah Tse. (B) Proportions of individual migratory DC populations among the total migratory DC subset in the cutaneous draining lymph node in the steady state. Data are adapted from Henri et al.¹⁰⁶ and Mollah et al.³⁹

In addition to in vitro studies, a number of methods have been explored for targeting DCs in vivo, specifically utilizing antigens fused to DC specific antibodies,⁸⁵ such as DEC205 (CD205),⁸⁶ Langerin (CD207),⁸⁷ DC-SIGN (CD209)⁸⁸ or Clec9a (DNGR).⁸⁹ Of these potential DC targets, DEC205 targeting is the most widely used and first established strategy. By fusing antigens to DEC205, the vaccine is targeted to DEC205⁺ DCs and specifically delivered into the cross-presentation pathway of antigen processing- the major pathway by which viral and tumor antigens are presented. This approach facilitates increased activation of CD4⁺ and CD8⁺ T cells compared to untargeted protein vaccines,^86,90,91 and increased anti-tumor immunity in the preclinical setting of metastatic melanoma.⁹² Recently, we demonstrated that combining DEC205-targeted antigen with adjuvant and Flt3L treatment (to expand DC) can further amplify the resulting immune response.⁴⁶ However, in the absence of an adjuvant to mature or license DC, DEC205 targeting generates immune tolerance. In a murine model of multiple sclerosis (Experimental Autoimmune Encephalitis), DEC205 fused to Myelin Oligodendrocyte Gylcoprotein (MOG) was able to prevent and ameliorate the disease phenotype.^93-95

Clinically, DEC205 targeting has been found to be safe and effective in generating cellular and humoral responses in healthy volunteers.⁹⁶ Additionally, the results of a phase I trial using a DEC205 targeted NY-ESO-1 vaccine (given in combination with resiquimod or Hiltonol) were recently reported.⁹⁷ In this study, patients with NY-ESO-1 expressing tumors saw an increase in both cellular and humoral immunity to the target antigen following vaccination, despite the immune tolerance generated by the tumor. Intriguingly, 75% of patients receiving this vaccine who subsequently enrolled in checkpoint blockade studies had objective tumor regression, suggesting the remarkable potential to combine this vaccine with combinatorial immune checkpoint blockade therapies in the future.⁹⁸

DEC205 targeting has been largely successful in the generation of adaptive immunity in both preclinical as well as clinical studies. However, we and others have observed high DEC205 expression on CD8α⁺ lymphoid resident and all subsets of migDCs.³⁹ As such, DEC205-targeted antigens are directed to both populations.⁴⁶ The identification of a series of individual DC subsets with unique functional properties offers an opportunity to fine-tune this vaccination strategy by directing targeted vaccination to a specific DC population based on the desired cellular or humoral outcome.

Skin Dendritic Cells in Immune Activation

Skin based responses to cytolytic viral infection are highly dependent on cross-presentation of vaccine antigens, suggesting that cell death of locally infected cells may be a licensing cue for optimal skin DC subset activation. In a series of elegant murine experiments, Shen et. al. added the human CMV gene product US11 to the vaccinia virus genome, generating a virus that prevents direct presentation by interfering with MHCI:peptide expression.⁹⁹ In this model, CD8⁺ T cell responses were severely impaired when vaccinia was administered by intravenous, intraperitonial, or intramuscular injection, indicating a role for direct presentation in these vaccination routes. In contrast, when vaccinia was administered by either intradermal injection or skin scarification, the resultant CD8⁺ T cell response was unaffected by the loss of direct presentation, highlighting the unique importance of cross-presentation for these routes of vaccination. In agreement with these findings, the loss of DNGR-1 (a DC receptor that recognizes dead cells and mediates cross-presentation of scavenged antigens) results in significantly decreased CD8⁺ T cell immunity to vaccinia virus (when given by skin scarification).⁷⁵ This requirement for cross-presentation may also translate into superior immunity, as shown by Liu and Kupper et al., who found that skin based vaccinations with vaccinia were superior to other routes in terms of generating prophylactic protective immunity to either viral or tumor challenge through the generation of tissue resident memory T cells.²² However, within skin based vaccination routes, skin scarification is superior to intradermal vaccination in the generation of protection against either viral rechallenge²² or tumor growth,⁷⁴ though the molecular and cellular mechanisms defining these differences in vaccine efficacy are not well understood. Yet, the presence of a unique combination of DC populations (Fig. 2B) suggests that the interaction between the skin and its resident DC network are of major significance for the generation of protective immunity.

Langerhans Cells

In both the murine and human epidermis, Langerhans Cells (LCs) are the only migDC population.¹⁴ Although long considered the prototype DC, LCs have unique characteristics that align them more closely with monocytes than other DC populations, including other skin DC. For example, during neonatal development, LC precursor cells are originally recruited to the skin in a CCR2 and CCR6 dependent manner²⁴ from the yolk sac, but are then replaced by precursors from the fetal liver,⁸⁴ while the other skin DC populations are derived from the bone marrow. LCs require TGFβ, but not Flt3L signaling for their development, while Langerin⁺ CD103⁺ dermal DC are independent of TGFβ, but, like other classical DC, are Flt3L dependent.^39,100 Additionally, LCs demonstrate radio-resistance when compared to dermal skin DC, which are sensitive to radiation.⁵⁵ LCs are largely reconstituted from skin resident, rather than blood derived progenitors in the steady-state,¹⁴ relying on IL-34 signaling through the CSF-1 receptor.¹⁰¹ However, during inflammatory situations, both skin resident precursors¹⁰² as well as blood derived monocytes contribute to the repopulation of LCs in skin.^103,104 Indeed, following infection with Herpes Simplex Virus, LCs are rapidly depleted from the infected area. Once the infection has been cleared, the subsequent LC repopulation of the previously infected skin is derived about half from monocyte derived precursors and half from long lived skin resident progenitor cells.⁷³

One of the major functions of LCs is to monitor the integrity of the skin as a barrier tissue. In order to thoroughly survey this unique environment, LCs extend dendrites through keratinocyte tight junctions to monitor the external barrier of the skin,^33,105 and respond to mechanical injury to the epidermis (in the form of tape stripping) by increasing motility and projection of dendrites.^24,52 By extending dendrites through the tight junctions without compromising the epithelial barrier, LCs are able to generate immunity to antigens that have not breached the barrier yet, thereby generating protection without true infection. This was demonstrated in a model where patch immunization against the Staphylococcus aureus derived exfoliative toxin (ET) protected against blistering during toxin rechallenge, despite the ET not penetrating the epithelial barrier.³⁴ This is advantageous to the host because it allows the immune system to generate adaptive immunity prior to a true pathogenic challenge, similar to the ideology behind prophylactic vaccination.

In the LN, LCs can be distinguished from dermal migDCs by their co-expression of Langerin and EpCam,⁸³ while lacking expression of CD103.¹⁰⁶ In the steady state, LCs in the LN express high levels of CD80, CD86, CD40, PDL1, and PDL2, though this expression can be further upregulated during exposure to a contact allergen, 2,4,6-trinitro-1-chlorobenzene (TNCB).⁴² In addition to these markers, LCs express high levels of IL-15¹⁰⁷ as well as the IL-15 receptor, IL-15R-α, allowing them to present IL-15 to responding T cells. This co-stimulatory cytokine results in increased expansion of the T cell response through STAT5 signaling, even when presenting a normally tolerizing antigen, such as the tumor antigen Wilm's Tumor-1.¹⁰⁸ LCs may also be uniquely prone to tailoring the humoral response to antigens toward an IgG1 response, as the loss of LCs resulted specifically in the decrease of this antibody isotype in a gene gun vaccination model.⁸³

In the setting of infectious disease, LCs have a demonstrated role in adaptive immunity to C. Albicans. In response to skin infection with this pathogen, LCs increase production of IL-1β and IL-6 while presenting pathogen derived antigens to CD4⁺ T cells, resulting in T_H17 differentiation.^69,70 Interestingly, this production of IL-6 is dependent upon the engagement of a specific C type lectin receptor, Dectin-1, whose ligand is only expressed by the yeast form of C. Albicans, when it is located on the epidermis.⁶⁷ Surprisingly, MyD88 signaling is not involved in LC migration in response to C. Albicans, and the generation of both the CD4⁺ T_H1 and CD8⁺ T cell response is instead dependent on a separate, dermal resident DC population, CD103⁺ migDCs, which will be discussed later. However, the T_H17 cells generated by LC are important in generating protective immune memory in the skin, while T_H1 cells provide systemic protection⁶⁷ in this model, thereby highlighting the role of LCs in maintaining and protecting the skin as a barrier tissue.

Langerin⁻CD11b⁻ DCs

In addition to the epidermal resident LC, several populations of migDCs coexist in the dermis of both mice and humans. The first of these populations was uncharacterized until very recently, and is identified as lacking either Langerin or CD11b expression.^39,106 This population shares a common progenitor with other classical DC,¹⁰⁶ is dependent upon FLT3 signaling, and requires the classical DC transcription factor ZBTB46 for development.³⁹ We demonstrated that these cells are skin resident and actively transport FITC from the skin to the draining LN like other migratory DC subsets, present antigen with equal efficiency to other skin and LN DC subsets and, by hierarchy transcriptome analysis, we observed that Langerin⁻ CD11b⁻ DCs likely derive from a precursor (pre-DC) that also gives rise to Flt3L dependent CD11b⁺ migDC, to which they are most closely related.³⁹ These cells are equally capable of presenting antigens with an efficiency that parallels both other migDC populations and lymphoid resident DCs ex vivo.³⁹ Quite recently this population has gained increasing recognition and was found to depend on KLF4 (Kruppel-like factor 4 transcription factor) by an independent group (though KLF4 loss impacted both this subset and CD11b⁺ migratory DC). This study demonstrated a shared role for these populations in response to house dust mite allergen and Schistosoma mansoni infection.¹⁰⁹ However, it is currently unknown if this population has a human counterpart, if phenotypic labeling of BDCA3⁺ dermal DC in humans incorporates its counterpart, or if this population is related to in BDCA1⁺BDCA3⁺ intermediary cells. Further research will explore the in vivo relevance of this population in vaccination and infection, and determine whether these cells have a unique function when compared to other dermal DCs, or functional redundancy with other skin DC subsets. This may also represent a subset that can be further specialized, differentiated, or licensed in an appropriate infectious context.

Langerin⁻CD11b⁺ DCs

In addition to the CD11b⁻ DCs, a separate, Langerin⁻CD11b⁺ population of DCs was identified in mice. The human counterpart of this subset is identified based on expression of BDCA-1.^110,111 Up regulation of CCR7 and the subsequent migration of Langerin⁻CD11b⁺ DCs depends on IRF4 signaling, as tested by modeling contact hypersensitivity in mice lacking IRF4.¹¹² In terms of antigen presentation, in vivo Langerin⁻CD11b⁺ DCs were observed to be uniquely capable of presenting soluble antigens, such as house dust mite antigen,¹¹³ ovalbumin,^114-116 or part of the class II MHC Eα chain,¹¹⁷ to CD4⁺ T cells. In the context of intramuscular vaccination with alum, CD11b⁺ migDC were capable of cross-presentation of soluble OVA to CD8⁺ T cells, though in these studies, CD64⁺ monocyte derived DCs, were equally capable of presenting antigen.¹¹⁶

CD11b⁺ migDC also have a role in CD4⁺ T cell priming during infection. Langerin⁻CD11b⁺ DCs present antigen derived from Herpes Simplex Virus to CD4⁺ T cells⁷¹ while also maintaining a role as the major cytokine producing population of skin DCs. In response to influenza vaccination, Langerin⁻CD11b⁺ DCs produced higher levels of the chemokines MIP1α, MIP1β, and RANTES than CD103⁺ DCs, although the CD103⁺ DCs (to be discussed next) were identified as key presenters of influenza derived antigens.¹¹⁸ In accordance with their role as both cytokine producers and antigen presenters, CD11b⁺ migDC can also skew CD4⁺ T cells toward a T_H2 phenotype, either in the context of a T_H2 inducing adjuvant (papain) or a T_H2 driving infection (Nippostrongylus brasiliensis).⁶² Additionally, CD11b⁺ migDC are required for the generation of a protective T_H17 response to Aspergillus fumigatus challenge,¹¹⁹ demonstrating their highly versatile capacity to shape the resulting T cell response.

Langerin⁺CD103⁺ DCs

This dermal migDC population is associated with a cross-presentation function and identified by co-expression of Langerin and the integrin CD103. These cells depend on the transcription factor BATF3,¹²⁰ as well as the transcriptional regulator IRF8.¹²¹ Although CD103⁺ DC, like other DC, depend on Flt3L and are significantly reduced in numbers in the Flt3L knockout mouse, upon Flt3L treatment they do not expand as significantly as CD11b⁺ and CD11b⁻ dermal DC, suggesting that once they have differentiated from pre-DC in the tissue, they either lose Flt3L dependence or become dependent on or susceptible to pathways opposing Flt signaling.³⁹ CD103⁺ DCs have been ascribed a unique capacity for cross-presentation, as facilitated by high expression of the cell death receptor, DNGR-1 (CLEC9A).⁷⁵ Indeed, CD103⁺ DCs play a key role in cross-presentation of antigens derived from self,^106,122 as well as those derived from Herpes Simplex Virus-1¹²², influenza,^123,124 vaccinia virus,⁷⁴ and C. Albicans.⁷⁰ Concordant with this specialized function, CD103⁺ DCs have decreased endosomal acidification when compared to Langerin⁻CD11b⁺ DCs, allowing CD103⁺ DCs to preserve the integrity of acquired antigens for a longer period of time.^81,123 In the case of influenza, 2 studies have independently demonstrated that CD103⁺ DCs are the primary APC responsible for the generation of the adaptive immune responses. However, in one study, CD103⁺ DCs were not productively infected by the virus, and thus were utilizing cross-presentation,¹²³ while in the other study, CD103⁺ DC were found to be infected with the virus, and thereby directly presenting antigens.¹²⁵ In both studies, type I IFN signaling was capable of preventing further viral infection, though further studies will be necessary to determine which presentation pathway (direct or cross) is the primary route of antigen presentation in this infection. Additionally, a recent study has also identified the potential of CD103⁺ DC to skew the T cell response to Leishmania, despite being dispensable for antigen presentation.¹²⁶ In humans, a similar population of DCs has been identified, based on expression of BDCA-3 or CD141.²⁹ In agreement with the murine studies, human BDCA-3⁺ DCs are uniquely proficient in cross-presentation of antigens.²⁹

In the setting of cancer, CD103⁺ DCs were just reported to infiltrate tumors in 2, separate murine models; ectopically implanted B16 melanoma, as well as spontaneously arising mammary tumors driven by expression of the mouse mammary tumor virus promoter.⁸¹ In agreement with these findings, CD103⁺ DC also infiltrated human metastatic melanoma. In this important study, the authors generated a signature associated with either CD103⁺ tumor-infiltrating DCs or CD11b⁺ tumor-infiltrating DCs, and found that a high CD103 to CD11b signature ratio correlated with a better prognosis in a clinical retrospective analysis of breast cancer, head and neck squamous cell carcinoma, or lung adenocarcinoma.⁸¹ However, it was not clear whether this stratification was driven more by the CD103⁺ DC or by the CD11b⁺ side of this ratio, as in human tumor analysis, PDL1⁺ tolerizing interstitial myeloid cells (represented here within the denominator) correlate with response to immunotherapy.^127,128 Also, while a mechanism for DC recruitment was not demonstrated in this study, tumors secrete high levels of CCL19 and CCL21 (the ligands for CCR7), which could potentially attract migratory DCs as well as other CCR7 dependent populations such as regulatory T cells (as discussed below).¹²⁹ Also it is unclear as to whether intra-tumoral DCs were originally skin-derived and infiltrated the tumor, or were recruited from the bone marrow derived pre-DC circulating through blood, which then entered the tumor and there differentiated into a tissue-resident DC. Many questions remain concerning the interactions between these DCs and tumors, but it is clear that CD103⁺ DCs are critical for the generation of cytolytic CD8⁺ T cell responses to a variety of pathogens, and their unique cross-presentation capability offers distinct potential in the presentation of tumor-derived antigens.

Together, skin migDC form a network of sentinel cells, with each population playing a unique role in shaping the response to a variety of pathogens. Understanding the unique functional properties associated with individual DC subsets will allow for further fine tuning of the immune response being generated, either by specifically targeting one population using antibody therapies as discussed previously or by understanding key pathways utilized by these individual DC populations during their licensing. However, in order to fully utilize the potential of the skin DC network, future studies are necessary to understand the coordination of these subsets, since, in vivo, each population plays a distinct role in shaping pathogen specific responses, as seen by the coordination of CD103⁺ migDC and LC during infection with C. Albicans, or the compensatory roles of CD103⁺ migDC and CD11b⁺ migDC in response to influenza. While individual DC populations are capable of generating an adaptive immune response, the interactions between the members of the skin DC network tailors the response specifically toward an individual pathogen, and thereby shapes the resultant immunological memory. The power to fine tune the immune response by harnessing these DC interactions will be indispensable for future vaccines.

Skin Resident Dendritic Cells in Immune Regulation

As early as the late 1990s, it had been hypothesized that DCs played an active role in the generation and maintenance of self tolerance, and that migratory DC in particular may be especially well suited for this function.¹³⁰ Indeed, DCs are capable of engulfing both apoptotic and necrotic cells, and generating MHC:peptide complexes from them.¹³¹ However, only exposure to necrotic cells results in increased expression of classical DC maturation markers, such as MHC II, CD40, and CD86, as well as triggers DCs potential to activate naïve T cells.^132,133 MigDC are regularly exposed to normal cellular turnover (apoptotic), as mentioned above, and constitutively migrate to the cutaneous lymph node, making them an ideal candidate for DC induced tolerance to self antigens. Indeed, recent data as demonstrated presentation of murine skin-derived antigens in the non-inflamed state is restricted to migDCs, including both LCs and dermal DCs, and this presentation enforces deletional tolerance of self reactive T cells.¹³⁴ In the case of Leishmania infection, deletion of LCs results in a greater production of IFNγ by CD4⁺ T cells, as well as decreased parasite loads at the site of infection,⁷⁷ though it is difficult to discern if this is due to LCs playing a tolerizing role, or due to the fact that Leishmania infects directly propogates in LCs.⁷⁷ Additionally, murine LCs,⁹⁵ CD103⁺ DCs,⁹⁵ and Langerin⁻CD11b⁺ DCs^117,135 are all capable of generating CD4⁺ FoxP3⁺ Tregs, and a similar phenotype has been identified in human LCs.⁸⁴ Furthermore, in some settings, murine CD4⁺ T cells activated by LCs do not fully differentiate into cytokine producing effector cells or persist in vivo, despite initially dividing upon antigen encounter.¹³⁶ In accordance with a tolerizing role in immunity, human BDCA-3⁺ DCs, the counterpart to murine CD103⁺ DCs, have been found to produce large amounts of IL-10, present self-antigens, and induce regulatory T cells.³⁰

Recently, we found that murine migratory DC collectively and individually play a tolerogenic role during the setting of DEC205 protein vaccination.³⁹ Surprisingly, this effect occurred despite the inclusion of an adjuvant targeting TLR4 in our vaccine design, (GLA).¹³⁷ In these studies, LN resident DCs were responsible for the generation of adaptive immunity in response to vaccination, either by subcutaneous or intradermal administration. Despite enhanced antigen capture by migDC that had reached the LN (as compared to LN resident DC), immune responses were enhanced in CCR7 knockout animals, in which migDC could not traffic to the LN, demonstrating that migDCs were not required for productive immunity.

By using a DTR model where LC and CD103⁺ DCs (or just CD103⁺ alone in bone marrow chimera) could be transiently ablated (Table 1), we observed deletion of individual migratory DC subsets also enhanced immunity. Loss of either LC and CD103⁺ migDC or CD103⁺ migDCs alone resulted in increased IFNγ production by effector CD4⁺ T cells, revealing that CD103⁺ DCs can inhibit the immune response to a DEC205 targeted vaccination, despite the inclusion of adjuvant. In our model, an increased CD4⁺ T cell response also translated into increased vaccine specific IgG titers, further extending the immune regulatory function of CD103⁺ DCs to the humoral arm of the adaptive immune system. Furthermore, upon examination of the transcriptional signature of migDCs as compared to LN resident DCs, we found that migDCs expressed high levels of a variety of transcripts associated with immune tolerance. Parallel findings were found when comparing murine lung migDC to LN resident DCs.²⁸ Of translational importance, this tolerance signature was also present when we compared human peripheral blood DCs to skin resident DCs, suggesting that this program was conserved between mice and humans. We identified NF-κB signaling by ingenuity pathway analysis as central to this program. Another recent study has also identified a number of these tolerance associated genes when comparing migratory DC in the skin to those in the lymph node, suggesting that at least a large part of the program we identified may be amplified following migration to the cutaneous lymph node.⁴¹ Collectively, overlapping features of the tolerance signature and migration signature are programmed through NF-κB signaling in DCs. DC specific loss of NF-κB signaling in mice resulted in loss of tolerance to self antigens in the steady-state, resulting in a lupus-like autoimmune phenotype marked by significantly increased lymph node size, as well as increased germinal centers and anti-nuclear antibody deposits in the kidney.⁴¹ In addition to these studies, one study using gene gun based vaccination also noted higher numbers of CD4⁺ and CD8⁺ IFNγ producing cells upon ablation of dermal langerin⁺ DC.⁸³ However, despite having a regulatory role in the T cell response, CD103⁺ dermal DC and LC had a coordinated stimulatory role in the development of the antibody response in this study. The antigen amount and delivery methods may underscore the differences between these 2 studies. By subcutaneous injection we observe antigen rapidly accumulating and captured in LN within 3 hours by both resident and already migrated DC, while a gene gun based approach results in only very small doses of antigen over a longer period of time, which presumably must be actively transported by migratory DC for priming to occur.

Some recent studies have also reported that migDC are not needed to generate adaptive immunity to viral vaccines or particulate antigens. In one study, using UV-inactivated influenza as a model vaccine, UV inactivated influenza was found to drain directly to the LN following footpad or intradermal ear injection.¹³⁸ Additionally, removing the vaccinated ear 30 minutes after vaccination (thereby preventing new skin DC migration) had no affect on the generation of protective immunity against viral rechallenge. Another study used particulate antigen and found that a LN resident DC population (enriched in LN resident CD11b⁺ DC) was positioned to filter particulate antigens that drain directly from the vaccine site. Again, in this model, migDC were unnecessary for generating a T cell response.¹³⁹ In both these studies, as well as in our study, vaccine antigens were capable of draining freely to the local LN, without active transport from migDC. Therefore, in order to effectively target migDC, it may be important to prevent this type of antigen draining. These observations are important to note for future vaccine design as they suggest that targeting vaccinations to migDCs may not be ideal in all contexts, and that a unique set of licensing requirements exist for migDCs.

Recent work has also begun to address the role of migDC in the generation of anti-tumor immunity, as discussed earlier. In a murine melanoma model, B16, it has been demonstrated that expression of CCL21 increases tumor growth, while a tumor line where CCL21 was genetically ablated resulted in decreased tumor growth in a T cell dependent manner.¹²⁹ However, while the authors attributed this to increased infiltration of the tumor by regulatory T cells, skin migDC also express high levels of the receptor for CCL21, namely CCR7, which could be driving the increase in Treg formation. Indeed, when CCR7 signaling was decreased either through the use of a knockdown tumor cell line, or CCR7 knockout host mice, significantly less CD11c⁺ cells were found within the tumor, despite increased tumor specific CD8⁺ T cell infiltration. Furthermore, both CD103⁺ and CD11b⁺ migrate into skin tumors in mice, and are capable of transporting antigen to the draining LN,⁸¹ though their status (stimulatory vs. tolerizing) following interaction with a neoplasm is just beginning to be addressed.⁸¹ How skin cancer affects migDC programming and how in turn these populations contribute to cancer surveillance and therapy is an important area of future research, and could be a contributor to immune priming vs. tolerance to tumors by the immune system.

Conclusions

The skin is a unique microenvironment that interacts with pathogens and self-antigens regularly. In order to maintain the appropriate balance between activation to pathogens and tolerance to self, a network of migDCs surveys the skin and mediates the interactions between the immune system and foreign invaders. The individual characteristics of these populations make them attractive for targeted therapy, and have the potential to further amplify or tune a vaccine to the response desired depending upon the utilized DC populations. For example, targeting a protein vaccine to CD103⁺ DCs could be more effective in generating a cytolytic T cells response when compared to broader DC targeting, while targeting Langerin⁻CD11b⁺ could lead to enhanced production of neutralizing antibodies. In addition, understanding how individual migDC populations coordinate responses to pathogens will allow for greater fine tuning of the response to vaccinations. However, as seen in the setting of DEC205 vaccinations, it will be important to fully understand the licensing cues that drive each of these subsets to activate adaptive immunity, so as to stimulate an optimal immune response.

Despite the need to perform more human skin based immunology, additional pre-clinical models will be invaluable to add mechanistic insight toward understanding the specific roles of individual migDC populations, as well as the function of migDCs as a whole. As a vaccination route, the skin is unique, depending predominantly on cross-presentation of antigens from infected and dying cells, instead of on direct presentation though infection of DCs. When utilized correctly, skin based vaccinations are capable of generating sterilizing and protective immunity, as seen by the overwhelming success of the smallpox vaccine. However, work in our lab and others has recently revealed that the DCs involved in generating this protection are also responsible for tightly regulating the immune system, likely to prevent horror-autotoxicus,¹⁴⁰ and may be hampering the immune response in certain settings.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.