Control of adaptive immunity by pattern recognition receptors

Abstract

One of the most significant conceptual advances in immunology in recent history is the recognition that signals from the innate immune system are required for induction of adaptive immune responses. Two breakthroughs were critical in establishing this paradigm: the identification of dendritic cells (DCs) as the cellular link between innate and adaptive immunity and the discovery of pattern recognition receptors (PRRs) as a molecular link that controls innate immune activation as well as DC function. Here, we recount the key events leading to these discoveries and discuss our current understanding of how PRRs shape adaptive immune responses, both indirectly through control of DC function and directly through control of lymphocyte function. In this context, we provide a conceptual framework for how variation in the signals generated by PRR activation, in DCs or other cell types, can influence T cell differentiation and shape the ensuing adaptive immune response.

eTOC

Signals from the innate immune system initiate and guide adaptive immune responses. Barton and colleagues recount key discoveries leading to the current understanding of this communication, and discuss how variation in the signals generated from pattern recognition receptors in dendritic cells or other cell types can influence T cell differentiation and the ensuing adaptive immune response.

Introduction

In 1908, Ilya Mechnikov and Paul Ehrlich shared the Nobel Prize in Physiology or Medicine for their work on what we now conceptualize as the two branches of mammalian immune systems. Mechnikov’s observation that microbes can be engulfed and destroyed by immune cells laid the foundation for our current understanding of innate immunity.¹ Similarly, Ehrlich’s studies of the immune responses against toxins and his proposed “side chain theory”, in which immune cells are induced to release antibodies specific to the stimulating antigen,² established much of the conceptual framework of our current understanding of adaptive immunity. At the time, it was unclear to what extent the two different branches of immunity represented by Mechnikov’s and Ehrlich’s works were linked, and most research over the following decades focused on studying either innate or adaptive immunity in isolation. However, two transformative discoveries in the late 20^th century established the cellular and molecular basis by which the innate immune system controls and instructs activation of the adaptive immune system: the identification of dendritic cells (DCs) and the conceptualization and discovery of pattern recognition receptors (PRRs).³

The links between innate and adaptive immunity are now so well established that younger immunologists may struggle to imagine life before DCs and PRRs, but this integrated understanding of the immune system represents a relatively new paradigm. In fact, many of the key discoveries that have shaped our current understanding occurred within the past 3 decades. Therefore, on the occasion of the 30 year anniversary of Immunity, we take this opportunity to reflect on how the discoveries of DCs and PRRs have transformed the field of immunology. In addition, we review how PRR signaling on DCs, and in some cases lymphocytes themselves, influences T cell and B cell activation and differentiation. Our intention is not to provide a comprehensive description of the state of the field but rather to highlight key discoveries and concepts that illustrate how the innate immune system, largely through the activity of PRRs expressed by DCs, controls and shapes activation of the adaptive immune system.

The cellular link: dendritic cells

Dendritic cells (DCs) were identified by Ralph Steinman and Zanvil Cohn as they searched for “accessory cells” necessary for lymphocyte responses to antigens in vitro. The fact that pure populations of lymphocytes could not respond to antigen on their own had been established by earlier groups.^4,5 Many believed that macrophages were the relevant accessory cells⁶; however, macrophages had several characteristics that were not ideal for accessory cells, including rapid degradation of phagocytosed cargo.⁷ In 1973, Steinman and Cohn described a cell type in mouse spleen and lymph nodes with a morphology distinct from macrophages.⁸ They proposed the term “dendritic cell” for these new cells because of their branch-like appearance. Experiments over subsequent years established definitively that DCs are vastly superior accessory cells compared to macrophages as well as potent activators of T cells.^9,10,11

We now know that DCs possess several key characteristics that enable them to function as the cellular bridge between the innate and adaptive immune systems, especially in the context of activation of naive T cells (Figure 1). DCs populate all tissues throughout the body, and, within tissues, developmentally mature but functionally immature DCs exist in a highly endocytic and phagocytic state which enables sampling of antigens in their local environment.¹² Antigen acquisition (coupled with PRR activation, as discussed in the following section) is linked to a series of coordinated events that are integral to potent stimulation of T cells.^12,13 This functional transition has historically been referred to as DC “maturation”, but we will refer to it as DC “activation” for consistency with more current terminology.¹⁴

Figure 1: Dendritic cells (DCs) serve as the cellular link between the innate and adaptive immune systems.

In tissues, DCs exist in a developmentally mature but functionally immature state with low levels of surface MHC, costimulatory molecules (e.g., CD80 and CD86), and CCR7 (a chemokine receptor involved in migration the lymph nodes). These immature DCs sample antigens in their environment through a variety of mechanisms including endocytosis, phagocytosis, and pinocytosis. Microbial recognition by PRRs on DCs triggers a process of DC activation, which results in several changes that make DCs potent activators of naïve T cells. These changes include upregulation of costimulatory molecules and CCR7, and increased lysosomal acidification that promotes antigen breakdown and the generation of peptide-MHC complexes. Activated DCs migrate from tissues to the draining lymph node where they drive the activation and differentiation of naive T cells.

First, upon activation, DCs upregulate the chemokine receptors CCR7, which enables them to migrate from tissues to secondary lymphoid organs where they interact with naive T cells.^{15,16,17,18,19,20} The ability to migrate distinguishes DCs from most tissue-resident macrophages. Second, DC activation is linked to degradation of antigens into peptides that are loaded onto the major histocompatibility complex (MHC), and displayed on the cell surface where they can be recognized by the T cell receptor (TCR) on T cells.²¹ This binding of peptide-MHC complexes to the TCR constitutes “Signal 1” for T cell activation. The endosomes and lysosomes of DCs are less degradative than those of macrophages, which favors MHC presentation over degradation, and levels of surface MHC are higher on activated DCs.^22,13 Third, activated DCs express costimulatory molecules, such as CD80 and CD86, which engage CD28 on T cells.^23,24 CD28 signaling provides an essential "Signal 2” to T cells.²⁵ Interaction of a T cell with antigen in the absence of costimulation leads to inactivation or anergy.²⁶ Finally, activated DCs secrete key cytokines that provide “Signal 3” for T cell activation and influence T cell polarization (discussed in greater depth below).^27,28

Several early studies established that DCs are heterogeneous in both peripheral and lymphoid tissues.^29,30,31 These studies typically defined populations of DCs by flow cytometry based on their expression of particular surface molecules, but different markers were often used to define the DC populations in various tissues. The advent of technologies for broadly measuring gene expression showed that DCs across different tissues shared a similar core transcriptional program.^32,33 Based on these shared programs, it was proposed that conventional dendritic cells (cDCs) in both mice and humans be separated into two subpopulations: cDC1s and cDC2s.³⁴ Note that the term “conventional” is used to distinguish cDCs from plasmacytoid DCs (pDCs) and monocyte-derived DCs. cDCs arise from a common precursor in the bone marrow called pre-cDCs and express the transcription factor Zbtb46.^35,36,37 cDC1s and cDC2s express distinct molecules on their surface, arise from separate precursors downstream of the pre-cDC (pre-cDC1s and pre-cDC2s), and depend on different transcription factors for their development (For a more extensive discussion of the differences between cDC1s and cDC2s, we refer the reader to the comprehensive review by Reis e Sousa and colleagues Ref. 14).^14,38,39 Recently, more in-depth characterization of cDC2s has shown them to be a heterogeneous group of cells, and several subdivisions of the cDC2 lineage have been described.^40,41,42,43 A discussion of these subtypes, and their specific role in T cell activation is beyond the scope of this review.

The molecular link: pattern recognition receptors

While the discovery of DCs linked antigen acquisition by the innate immune system to activation of naive T cells and induction of adaptive immunity, there remained a conceptual gap in understanding how the process was controlled. The splenic and lymph node DCs characterized by Steinman et al were potent activators of T cells due to their expression of MHC and costimulatory molecules, but, as discussed above, subsequent studies showed that DCs in tissues had lower surface expression of MHC and costimulatory molecules and were less stimulatory.^12,13,44 What triggered DCs to transition from one state to the other?

In a prescient address at the 1989 Cold Spring Harbor Symposium on Quantitative Biology⁴⁵, Charles Janeway proposed that this job must be carried out by yet-to-be-discovered receptors, which he called pattern recognition receptors (PRRs), that are broadly expressed on cells of the innate immune system. He hypothesized that PRRs evolved to recognize conserved, largely invariant features of microbes, so-called pathogen associated molecular patterns (PAMPs). Janeway further reasoned that because PAMPs are absent from the host, PRR activation would induce signals that initiate the immune response, including upregulation of costimulatory molecules on DCs. Due to the random nature of V(D)J recombination, each cell of the adaptive immune system is ignorant as to the degree of threat posed by the antigen recognized by its receptor and must therefore rely on the signals from the innate immune system to provide that context. In this way, PRRs serve as the gatekeepers of adaptive immunity, ensuring that an immune system favors responses against infectious microorganisms rather than self. Janeway argued that this requirement explains why the use of adjuvants (i.e., mixing of antigen with dead microbes which he called the Immunologists’ dirty little secret) is required to elicit adaptive immune responses to antigens of non-microbial origin.

The first PRRs to be described were the Toll-like receptors (TLRs). Jules Hoffman, Bruno Lemaitre, and colleagues, working with flies, demonstrated the importance of the Drosophila Toll pathway for immunity to fungal pathogens via activation of NF-kB and induction of antimicrobial peptides.⁴⁶ Janeway and Ruslan Medzhitov identified the first human Toll (later designated TLR4) and showed that its activation induced NF-kB as well as the expression of the costimulatory molecule CD80.⁴⁷ While both discoveries represented huge conceptual advances, they did not establish that TLRs recognize PAMPs; in flies, the ligand for Toll is an endogenous protein Spaetzle, and Medzhitov and Janeway’s functional studies of TLR4 employed a clever trick to bypass ligand binding. Three papers published in 1998 and 1999 established this link. First, Paul Godowski’s group showed that cells ectopically expressing TLR2 induced NF-kB in response to lipopolysaccharide (LPS), the abundant component in the outer membranes of gram-negative bacteria.⁴⁸ Of course, TLR4, not TLR2, is the receptor for LPS, so this paper is incorrect in terms of the specificity of TLR2. This error was likely due to the crude nature of LPS preparations used in those days which led to contamination with bacterial ligands for TLR2. Sadly, this paper is often overlooked despite clearly establishing the link between TLR recognition of microbial products and activation of NF-kB. A few months later, the groups of Bruce Beutler and Danielle Malo published the culmination of long efforts mapping Tlr4 as the gene responsible for lack of responsiveness to LPS in two inbred mouse strains: C3H/HeJ and C57BL/10ScCr.^49,50 Shortly thereafter, Shizuo Akira’s group demonstrated that mice with engineered deficiencies in Tlr2 and Tlr4 genes no longer responded to bacterial lipopeptides and LPS, respectively.⁵¹

Collectively, the discoveries summarized above launched an explosion of research focused on understanding the molecular basis of innate immune recognition. We have refrained from a detailed discussion of these diverse PRR families here, especially because each is the focus of a separate article in this issue; however, a description of the major PRR families is useful when discussing how they function to shape adaptive immunity. Within a few years of the identification of TLR4, the mammalian TLR family quickly grew to 13 members whose ligands represent highly conserved features of microbes as Janeway had predicted.⁵² All TLRs induce signaling pathways that culminate in the activation of NF-kB, IRF, and AP-1 families of transcription factors. There can be differences in the specific family members activated by individual TLRs which can result in differential gene induction in some contexts (Akira and colleagues, this issue). C-type lectin-like receptors (CLRs) localize to the plasma membrane and/or endosomes, similar to TLRs, and recognize PAMPs in the extracellular space or PAMPs taken up by the cell within endosomes or phagosomes.⁵³ Signaling induced by CLRs is distinct from that induced by TLRs (Reis e Sousa, this issue). Several PRR families sense products of pathogens that gain access to the cytosol. For example, RIG-I like receptors (RLRs) and cGAS recognize cytosolic RNA and DNA, respectively.^54,55,56 Activation of these PRRs induces production of type I interferon (IFN). Members of a diverse family of PRRs called NLRs (nucleotide binding domain and leucine-rich repeat containing proteins) recognize a variety of PAMPs, including peptidoglycan fragments and bacterial flagellin.⁵⁷ In some cases, activation of these NLRs leads to transcriptional responses that are conceptually analogous to the responses induced by TLR, RLR, and cGAS activation. However, other NLRs induce formation of multiprotein complexes called inflammasomes which serve as platforms for activation of inflammatory caspases such as caspase-1 (Kanneganti and colleagues, this issue).⁵⁸ The outcome of inflammasome activation is largely post-transcriptional; caspase-1 activation leads to pore formation in the plasma membrane, release of inflammatory cytokines such as IL-1b and IL-18, and eventual death of the host cells by a mechanism called pyroptosis.

In addition to PAMPs, certain endogenous molecules can also promote innate immune activation. Some endogenous ligands, such as IL-1a and IL-33, are released upon necrotic death of host cells and activate receptors on innate immune cells.^59,60 In fact, the cytosolic domain of the receptors for IL-1 family of cytokines share homology with TLRs and induces similar downstream signaling.⁶⁰ Necrotic cell death also exposes the normally hidden Filamentous actin (F-actin) and ATP. F-actin is recognized by Clec9a, a CLR expressed by certain DCs,^61,62,63 and ATP binds to purinergic receptors resulting in assembly and activation of the NLRP3 inflammasome.⁶⁴ These instances illustrate the immune system’s ability to recognize cellular damage or “danger”, a concept developed by Polly Matzinger.⁶⁵ In some cases, endogenous ligands driving immune activation are molecularly identical to PAMPs, as in the case of self-nucleic acids that can activate TLRs, RLRs, or cGAS. Innate immune activation induced by such endogenous molecules is associated with chronic inflammation or autoimmunity.

PRR control of DC function and T cell activation

Microbial recognition by PRRs induces a series of functional changes in DCs that enhance their ability to induce the activation and differentiation of microbe-specific naïve T cells (Figure 1).^66,67 TLR signaling increases presentation of microbe-derived peptides (Signal 1 for T cells) through upregulation of surface MHC and the preferential loading of peptides in phagosomes from which TLR signaling occurs.^13,68,69 TLR and CLR signaling can also favor the presentation of exogenously acquired antigens on MHCI, which will be discussed in the subsequent section on cross presentation. In addition to these mechanisms that promote Signal 1 generation, PRR signaling on DCs induces expression of costimulatory molecules, such as CD80 and CD86, which constitute Signal 2 for T cells.^23,24,25 The requirement for costimulation limits the likelihood that T cells with high affinity for self-peptides will become activated. While such self-reactive cells are often deleted via central tolerance, the process is not perfect, so induction of peripheral T cell tolerance through presentation of self-peptides by DCs lacking expression of costimulatory molecules serves as an important additional mechanism. The linkage of PRR activation on DCs to generation of Signals 1 and 2 that are required for T cell activation focuses immune responses toward microbes. In addition to this gatekeeping function, PRR activation can provide critical contextual information to T cells about the nature of infection in the form of cytokines, referred to as Signal 3. In the remainder of this section, we discuss our understanding of how PRR activation is linked to generation of Signal 3 cytokines and how these cytokines influence T cell differentiation and function.

The cytokines secreted by DCs can determine the functional properties that activated T cells acquire as they proliferate and differentiate (Figure 2). During an infection, CD4⁺ T cells can adopt a number of distinct functional states, including Th1, Th2, Th17, and T follicular helper (Tfh) cells.⁷⁰ Each Th type directs the ensuing immune response to favor particular effector mechanisms, ideally those optimized to address the nature of the threat at hand. For example, Th1 cells produce IFNγ⁷¹, which induces macrophage microbicidal mechanisms.⁷² As such, Th1 cells play an important role during intracellular bacterial and viral infections, as well as against tumors. Th1 differentiation is promoted by DC-derived IL-12.^73,74 Th17 cells produce IL-17 and IL-22 and are most relevant for infections of extracellular bacteria and fungi.^75,76,77 The cytokines IL-23, TGFβ, IL-6, IL-21, and IL-1b promote the differentiation of Th17 cells.^{75,78,79,80,81} Tfh cells promote B cell activation and differentiation by providing key survival and proliferation signals through expression of CD40L and ICOS and secretion of IL-21 and IL-4.⁸² Finally, Th2 cells produce IL-4, IL-5, and IL-13 and are important during infections with large parasites, such as worms, but are also implicated during allergic responses.⁸³ IL-4⁸⁴ and thymic stromal lymphopoietin (TSLP)^85,86 have been implicated as critical Signal 3 cytokines for Th2 responses, but the source and/or precise mechanism of action of these cytokines during DC driven Th2 cell priming remains somewhat unclear. Regulation of Th2 responses may not fit under the same conceptual framework as other Th types, in that a role for PRRs on DCs has not been established.

Figure 2: DCs shape the activation and differentiation of T cells.

During an infection, CD4⁺ T cells can differentiate into a variety of distinct effector phenotypes, including Th1, Th2, Th17, and T follicular helper (Tfh) cells. These effector Th cells engage different immune effector mechanisms through their interaction with other immune cells and non-immune cells. The specific combination of cytokines secreted by DCs during initial T cell priming influences CD4⁺ T cell polarization towards a particular Th type. DCs are also responsible for promoting the activation of CD8+ T cells, and certain cytokines have been shown to promote robust CD8+ T cell responses. Several factors affect which cytokines are secreted by a DC including intrinsic differences between different subsets of DCs (e.g., cDC1 versus cDC2), and which PRRs are activated on the DC. The DC subsets, PRRs, and cytokines responsible for driving CD8 and CD4 T cell activation and differentiation towards particular fates are depicted. Note that the PRRs depicted in this figure for each scenario are not meant to be exhaustive; rather, we have depicted the relevant PRRs linked to the differentiation of each Th type. Also, in some cases, these PRRs may not be expressed by cDCs themselves, but by bystander cells that produce cytokines that influence T cell differentiation.

Much work has gone into understanding what determines the type of cytokines secreted by DCs (Figure 3). The specific combination of PRRs engaged on a DC certainly influences the types of cytokines produced; however, not all DCs respond to activation of PRRs in the same way, due to intrinsic differences between DC subsets in terms of the PRRs they express and how those PRRs signal. How these two factors – the type of DC presenting antigen and the specific PRRs engaged on the DC – combine to influence T cell differentiation is an area of active investigation. In the remainder of this section, we discuss how PRR activation on distinct DC subsets influences CD4 T cell differentiation. We note that several additional cell populations are often discussed as members of the DC family (e.g., plasmacytoid DCs (PDCs), monocyte-derived DCs, and transitional DCs). These cells, as well as other accessory cells such as innate lymphoid cells (ILCs) and non-immune cells, can be important sources of Signal 3 cytokines, especially after initial priming as T cells continue to differentiate. While we discuss some of these specific instances briefly, for the most part we have focused on conventional DCs and their role in priming naive T cells.

Figure 3: Determinants of differential Signal 3 cytokine production.

DCs provide at least three critical signals to naïve T cells during activation: peptide-MHC complexes to stimulate the T cell receptor (Signal 1), costimulation (Signal 2), and cytokines (Signal 3). Signal 3 cytokines secreted by DCs or other cell types drive CD4⁺ T cell differentiation towards particular Th types, as outlined in Figure 2. The figure illustrates the DC-intrinsic factors that can influence which Signal 3 cytokines are secreted by DCs following their interaction with microbes: 1) the specific type of DC that is interacting with the pathogen and 2) selective PRR activation, either due to distinct expression or selective engagement based on the nature of the microbe. Other cell types can also serve as sources for Signal 3 cytokines. Both (3) immune cells, such as innate lymphoid cells (ILC) or plasmacytoid DCs (pDCs), and (4) non-immune cells, such as epithelial cells, can produce cytokines that influence T cell differentiation.

cDC1s and cDC2s each preferentially induce the differentiation of different types of effector CD4 T cell subsets. While the dichotomy is not absolute, in general, cDC1s induce the differentiation of Th1 cells, whereas cDC2s have been linked to Th2, Th17, and Tfh cell differentiation.⁸⁷ cDC1s also play a critical role in the cross presentation of exogenous antigens to CD8+ T cells on MHC class I, as discussed in greater detail in the next section. This division of labor is due to a combination of factors including the location of the two cell types, how they handle endocytosed antigens, the expression of distinct PRR repertoires, and the nature of the cytokines they produce. cDC1s and cDC2s localize to different regions of lymph nodes.⁸⁸ cDC1s are found deep in the T cell zone in the areas of the lymph node where CD8 T cells and Th1 cells are activated and differentiate. In contrast, cDC2s localize to the outer regions of the cortex and the T-B border. cDC1s also produce more IL-12 than cDC2s which explains their ability to drive Th1 responses.^{89,90,91,92,93} cDC2s are major drivers of Tfh and Th17 cell differentiation, both of which depend on IL-6.^79,94 Whether and how cDC2s produce more IL-6 following recognition of microbes that elicit Tfh or Th17 responses is not well understood. Finally, there are differences in the expression of PRRs between cDC1s and cDC2s (summarized in Table 1); the specific implications of these differences for T cell priming are not yet completely understood.

Table 1: Expression of PRRs in cDC1s, cDC2s, and B Cells.

Wherever possible, the expression data summarized here are based on protein expression as measured by western blot, flow cytometry, or mass spectrometry, or functional readouts (e.g., cytokine production) downstream of PRR activation. For DCs, we have focused on data from primary cDC1s and cDC2s, not bone marrow-derived DCs or monocyte-derived DCs. ImmGen and the Human Cell Atlas were used to fill in expression in cases where protein level data were unavailable (cases based on RNA data only are italicized). In cases where it is known that expression patterns vary between human and mouse, this is indicated. The focus on cDC1s and cDC2s is because of the primary role for these cells in activating naïve T cells. See the main text for a discussion of PRR expression in T cells.

Family	PRR	Expressed on cDC1s?	Expressed on cDC2s?	Expressed on B Cells?
TLRs	TLR1	No in mice178 Yes in humans179	Yes178,179	Yes158,136,152,180
	TLR2	Yes181,178,179	Yes178,181	Yes 136,152,180,182,183
	TLR3	Yes120,184,179	Low or no in mice181,184 Yes in humans179	Expressed based on RNA on MZ B cells in mice but they do not respond to ligand133,136 No in humans except on plasma cells152,185
	TLR4	Yes in mice181 No in humans179	Yes181,179	Yes in mice136,180 Not expressed on most B cells in humans except plasma cells; expression can be increased by stimulation186, 185
	TLR5	No181,179	Yes181,179	May be expressed in plasma cells but not other B cell subsets180, 136,183,187
	TLR6	Low in mice178 Yes in humans179	Yes181,179	Yes158,152,180
	TLR7	Varies by tissue in mice (No in spleen181; Yes in intestines188) No in human179	Yes181,179	Yes158,136,180,183
	TLR8 (only functional in humans)	Yes179	Yes179	No 158,180
	TLR9	Yes in mice181,179 No in humans181,179	Yes in mice181,179 No in humans181,179	Yes158, 189, 180
	TLR10 (pseudogene in mice)	Yes 179	Yes 179	Yes158,190
	TLR11 (Only in mice)	Yes191	No191	No 187,192
	TLR12 (Only in mice)	Yes184	No184	No 192
	TLR13 (only in mice)	Yes184	No184	Low or no 187,193
NLRs	NOD1	Low or no in mice184 Yes in humans194	Yes184,194	Yes195,196
	NOD2	Yes in mice197, 187 Some human DCs seem to be responsive but unclear if all human DCs express it194,198	Yes in mice197, 187 Some human DCs seem to be responsive but unclear if all human DCs express it194,198	Only in certain B cells in mice (GC, PB, PC)187 Yes in humans195
	NAIP (7 in mice but one in humans)	Yes199,200,187,201	Yes199,187,201	Yes 187,195
	NLRP1	Yes187,194,202 Ref. 202 shows expression in human DCs but does not distinguish between cDC1s and cDC2s	Yes187,194,202 Ref. 202 shows expression in human DCs but does not distinguish between cDC1s and cDC2s	Yes187,202
	NLRP3	Yes102,103,202 Ref. 202 shows expression in human DCs but does not distinguish between cDC1s and cDC2s	Yes102,103,202 Ref. 202 shows expression in human DCs but does not distinguish between cDC1s and cDC2s	No in mice187 Unclear in humans (clearly upregulated upon stimulation but mixed results about expression at steady state)202–204
	NLRP11 (not expressed in mice; primate specific)	Low or no 205	Low or no 205	Low205
Miscellaneous Inflammasome Forming Proteins	Pyrin (MEFV)	Yes in mice 199 No in humans 199	Yes in mice 199 No in humans 199	No 199
	AIM2	Yes206, 187	Yes206,187	Yes 187,194
Cytosolic Sensors	cGAS	Yes187,194,207	Yes187,194,207	Very low in mice 187 Yes in humans156
	RIG-I	No in mice184 No or low in humans194	Yes184,194	Yes 187,194
	MDA5	No in mice184 Yes in humans194	Yes184,194	Very low or no in mice 187 Yes in humans 194
	DAI (ZBP1)	No184,194	Yes184,194	Yes 187,194
CLRs	DEC205 (Ly75)	Yes208,209	No208,209	Yes194,210
	DC-SIGN (CD209)	Low 187,194,211	Yes211	No194,211
	Dectin1 (Clec7A)	No in mice178 Yes in humans212	Yes178,212	No in mice213 Yes in humans214
	Clec9a (DNGR-1)	Yes116,117,61 but may vary by tissue209	No116,117,61	No in mice116 No or low in humans116,117

Ultimately the abilities of cDC1s and cDC2s to induce distinct T cell responses are a combination of the cell-intrinsic features described above and PRR-induced changes. PRR activation induces a series of changes in DCs, including the production of Signal 3 cytokines (Figure 1). DCs stimulated with different classes of microbes, such as Escherichia coli and influenza, have unique transcriptional programs.^95,96 What leads to these differences in gene expression? In some cases, intrinsic differences between DC subsets leads to differences in PRR-induced gene expression. For example, TLR activation leads to activation of NF-kB and production of IL-12 in cDC1s whereas the same stimuli leads to much less IL-12 in cDC2s.⁸⁹ In other cases, the activation of unique combinations of PRRs will induce differential gene expression that skews downstream T cell differentiation.⁹⁷ Such differential activation of PRRs may be particularly relevant for cDC2s because these cells can induce multiple TH types (Figure 2). Some PRRs, such as certain TLR family members, will be activated by most, if not all, pathogens, and differences in signaling downstream of individual receptors can contribute to differential gene expression. Other PRRs, especially cytosolic sensors, are only activated by certain pathogens that express toxins or virulence factors or gain access to the cytosol for replication. The activation of these PRRs can lead to differential production of cytokines that shape T cell differentiation.

Inflammasome activation can synergize with TLR signaling to enhance production of cytokines that impact T cell activation and differentiation. The expression of two key cytokines, IL-1β and IL-18, is induced by TLR signaling, but the release of active cytokine typically requires inflammasome assembly and activation of inflammatory caspases.^98,99 In general, IL-1β promotes Th17 responses whereas IL-18 promotes Th1 responses.^100,80 Both cytokines are translated as inactive proteins without secretion signal sequences, and they only gain activity when cleaved by inflammatory caspases, such as caspase-1. Inflammasome activation also induces Gasdermin-D to assemble into pores in the plasma membrane, which enables extracellular release of cleaved IL-1b and IL-18. Gasdermin-D activation is also often associated with an inflammatory type of cell death known as pyroptosis (Kanneganti and colleagues, this issue). One might imagine that pyroptosis of DCs would be detrimental to the induction of adaptive immune responses because dead cells can no longer migrate to the lymph node and present antigens to naive T cells. One potential solution to this paradox is that inflammasomes function differently in DCs or perhaps do not function at all. Indeed, DCs derived from bone marrow precursors (BMDC) suppress certain downstream sequelae of inflammasome activation, enabling them to survive while still secreting IL-1β to facilitate T cell priming – a state referred to as “hyperactive.”¹⁰¹ Whether cDCs resident in tissues and secondary lymphoid organs adopt such a state requires further investigation. The lineage defining transcription factors IRF8 and IRF4 can inhibit expression of a variety of genes associated with inflammasome activation in cDC1s and cDC2s, respectively,¹⁰² suggesting that cDCs lack inflammasome function altogether.¹⁰³ In this case, the IL-1 and IL-18 must come from another cellular source or their production must be induced by other PRR pathways.¹⁰⁴ One possibility is that cDCs resident in tissues and secondary lymphoid organs have evolved to suppress inflammasome genes so that they can focus on initiating T cell priming, while recruited monocyte-derived DCs undergo inflammasome activation and adopt the “hyperactive” state to provide critical signals necessary for clonal expansion and T cell differentiation. It seems likely that multiple mechanisms are involved in determining the distinct outcomes of inflammasome activation in DCs versus macrophages, and identifying such mechanisms remains an area of active investigation.

Type I IFNs can also act as Signal 3 cytokines for T cell activation and clonal expansion. In cDCs, TLRs induce type I IFN expression via TRIF-dependent signaling and to a lesser extent via MyD88-dependent signaling. PDCs are not classical antigen presenting cells but specialize in producing large quantities of type I IFNs in response to TLR7 and TLR9 ligation.^105,106 In addition, cytosolic PRRs, such as cGAS, RIG-I, and MDA5, are potent drivers of type I IFN production in many cell types including cDCs. CD8 T cells in particular depend on IFNAR signaling for differentiation and function.¹⁰⁷ For CD4+ T cells, type I interferon is an important driver of clonal expansion during viral infections, but it may be dispensable during bacterial infections.¹⁰⁸ In addition, the presence of type I IFN shapes the polarization of CD4+ T cells, although how it does so remains somewhat unclear. Some data suggest that type I IFN drives a Tfh phenotype^109,110; however, other data suggest that type I IFN inhibits Tfh while promoting Th1 differentiation.¹¹¹ It is likely that the specific source of IFN (infected cells vs PDC), as well as the context of the infection (bacterial vs viral) influences the outcome of CD4 T cell differentiation. A recent study also suggested that the timing of when DCs are exposed to type I IFN may influence CD4+ T cell differentiation. Early exposure to type I IFN, as occurred during vesicular stromatitis virus, drove Tfh polarization whereas late exposure to type I IFN, as occurred during LCMV, drove Th1 polarization.¹¹²

Altogether, we propose that there are four determinants that collectively influence CD4+ T cell differentiation during interactions with DCs (Figure 3). Two of these determinants are DC intrinsic: — the type of DC (#1) and the specific combination of PRRs (#2) activated on the DC. Both will impact the transcriptional changes that occur upon DC activation, including which Signal 3 cytokines are produced. Signal 3 cytokines can also be produced by immune (#3) or non-immune cells (#4) other than the primary DCs presenting antigen in lymphoid tissues. While we have focused our discussion here primarily on Signal 3 cytokines produced by DCs, these secondary sources can affect T cell differentiation, both during initial activation in lymphoid organs and by reinforcing differentiation signals in tissues.

Cross Presentation to CD8+ T Cells

Proteins acquired exogenously by antigen presenting cells are typically presented on MHC Class II molecules while cytosolic proteins are presented on MHC Class I molecules; however, certain DCs have the ability to bypass this strict bifurcation through a process called cross presentation. cDC1s are particularly adept at presenting exogenously acquired antigens on MHC class I, and their expression of certain PRRs can facilitate this process (Figure 2). Cross presentation is critical for CD8 T cell responses against pathogens that do not infect DCs, pathogens that inhibit antigen presentation, and tumors. Mice lacking cDC1s, or lacking critical components of the antigen processing and presentation pathway in cDC1s, lack robust CD8+ T cell responses against certain viral pathogens and tumors.^113,114,115

The CLR Clec9a (also known as DNGR-1) is highly expressed by cDC1s and plays an important role in promoting the cross presentation of cell-associated antigens following necrotic death.^116,117,61 Intriguingly, Clec9a is dispensable for sensing and internalizing dead cells, but is required for efficient cross presentation.⁶¹ Mechanistically, the activation of Clec9a by filamentous actin on necrotic cells promotes the cytosolic pathway of cross presentation, in which antigen is degraded in the cytosol, by triggering phagosomal rupture.^61,118,119 This phagosomal rupture leads to the release of endocytosed antigen into the cytosol for processing by the proteasome and loading into the endogenous MHC class I pathway.

TLR activation can enhance cross-presentation in certain contexts. TLR3-mediated recognition of viral dsRNA present in infected cells enhances the activation of CD8 T cells by cDC1s after phagocytosis of infected cells.¹²⁰ TLR ligands can also promote cross presentation by altering the trafficking of critical molecules required for antigen processing and presentation^121,122 as well as by improving the ability of cells to internalize antigens and reducing antigen degradation.¹²³ However, in some contexts only certain TLR ligands increase cross presentation, while others hinder it.^124,125,126 Systemic administration of CpG, a TLR9 agonist, blocked cross presentation in vivo.¹²⁷ Differences between these studies could be due to differences in dose or purity of TLR ligand, timing (whether the TLR ligand was administered prior to antigen or simultaneously with antigen), differences in the type of DC (monocyte-derived DCs vs cDCs), and differences in the activation state of the DCs. In some cases, it is difficult to distinguish whether the TLR ligand is truly promoting cross presentation, or whether it is simply activating DCs and therefore making them better at priming T cells.

Type I IFN, which is linked to sensing by cytosolic PRRs, can also promote cross presentation, and is required for tumor rejection in vivo.^128,129 In vitro, the production of type I interferon by DCs in response to tumor cells is dependent on sensing of tumor-derived DNA through the cGAS-STING pathway.^128,130 How these ligands gain access to the cytoplasm is an open question. One attractive possibility is that the sensing of necrotic tumor cells by Clec9a leads to phagosomal rupturing which allows for the release of PAMPs into the cytoplasm leading to activation of cytosolic sensors.¹³¹ cGAMP, the ligand for STING, can be generated within tumor cells and then transferred to non-tumor cells where it activates STING and promotes type 1 interferon production.¹³²

Direct control of lymphocyte function by PRRs

Activation of T and B lymphocytes is primarily driven by engagement of their antigen receptors (TCR and BCR); however, there is a large body of evidence that activation of PRRs, especially TLRs, can directly influence responses by lymphocytes. The best evidence for direct activation by PRRs comes from studies in B cells. B cells express several TLR family members (Table 1). Conventional B cell activation relies on dual engagement of the BCR by antigens and CD40 through T cell help, via engagement of CD40L expressed by Tfh cells (Figure 4). BCR and CD40 thus provide first and second signals to drive activation and differentiation of B cells into antibody producing plasma cells. Both mouse and human B cells express multiple TLRs, and, depending on the developmental origin of the B cells, TLR stimulation can either lead to direct activation of B cells as in the case of marginal zone or B1 B cells, or serve as second or third signals to shape antibody responses by follicular B cells.

Figure 4: PRRs control the activation and differentiation of B cells through cell-extrinsic and cell-intrinsic mechanisms.

In the case of B cell extrinsic PRR activation, PRRs drive DC activation which in turn controls activation and differentiation of Tfh cells. The precise mechanisms by which PRR engagement in DCs leads to Tfh differentiation remain unclear and is an active area of investigation. Following ligation of BCR by specific antigen which acts as Signal 1 for B cell activation, Tfh cells provide a critical second signal to B cells via CD40L-CD40 interactions when they recognize peptide-MHC complexes on B cells. In many instances, direct activation of PRRs on B cells acts as a third signal to boost antibody responses as well as induce class switching to specific isotypes. In the case of B cell intrinsic PRR activation, engagement of the BCR and TLRs on the same B cell induces antibody production. In this scenario, BCR ligation still provides Signal 1, and TLR signaling can replace Signal 2 that is typically provided by Tfh cells. This obviates the need for any third signal since TLR signaling induces both B cell proliferation and plasma cell differentiation. The most well characterized example of this B cell intrinsic PRR activation is seen in lupus where B cells specific for nuclear antigens are activated by BCR and TLR7/9 ligation to produce pathogenic antibodies.

In certain B cell subsets, particularly marginal zone B cells, B1 B cells, and transitional B cells, TLR activation alone can induce B cell proliferation and differentiation into short lived plasma cells. ^{133,134,135,136} B cells activated this way produce antibodies irrespective of their specificity and do not differentiate into memory B cells. In both human and mouse studies, TLR activation of these specific B cell subsets can also lead to class switching to certain isotypes.^137,134 In contrast to B1 B cells and marginal zone B cells, follicular B cells are less responsive to direct stimulation by TLR ligands.^137,136 However, follicular B cells have the capacity to integrate signals initiated by dual ligation of BCR and TLRs. The consequences of dual engagement of BCR and TLRs were revealed by a seminal study that examined the signals needed for activation of B cells that express antibodies against rheumatoid factor (RF).¹³⁸ Chromatin-IgG complexes that engage both the RF-reactive BCR as well as TLR9 drive production of auto-antibodies responsible for pathogenic lupus.¹³⁹ Similarly, RNA-associated autoantigens that engage both BCR and TLR7 activate autoantigen specific B cells.¹⁴⁰ These studies revealed the critical role of B cell intrinsic TLR signaling in lupus and suggest that direct targeting of TLR signaling in B cells could be a useful therapeutic strategy. More importantly, when BCR is engaged by immune complexes, TLR signaling can serve as a critical second signal bypassing the need for help from T cells (Figure 4). While detrimental outcomes of TLR signaling in B cells when BCR and TLRs are engaged simultaneously are seen in autoimmune disease, the sequential engagement of BCR and TLRs likely evolved for beneficial responses against microbial antigens when T cell help is unavailable.

In some contexts, TLR activation on B cells appears to provide a critical third signal following BCR activation and CD40 ligation by TH cells. This third signal helps boost antibody titers and can promote class switching, but it does not replace the need for a second signal from CD40. Absence of the TLR signaling adapter MyD88 specifically in B cells hampers optimal antibody responses in response to a variety of immunizations and infections. Mice lacking MyD88 or TLR4 only in B cells mounted defective germinal center reactions and produced reduced titers of IgG2c antibodies in response to protein antigens mixed with LPS,¹⁴¹ papilloma virus like particles,¹⁴² virus like particles containing CpG,¹⁴³ and synthetic nanoparticles containing antigen plus TLR ligands.¹⁴⁴ TLR or MyD88 deficiency restricted solely to B cells also results in reduced antibody responses against several viruses (Respiratory Syncytial virus, gamma herpes virus, influenza virus, retroviruses) and impairs plasma cell differentiation, robustness of antibody titers, and plasma cell survival.^{145,146,147,148,149} Defects in antibody responses are apparent even when TH cell priming is unaffected.^147,141 In contrast, there are no significant defects in antibody responses when mice lacking MyD88 and TRIF are immunized with proteins conjugated to the hapten, trinitrophenol (TNP)¹⁵⁰, suggesting that haptenation bypasses the need for TLR engagement on B cells.¹⁵¹ Finally, TLR stimulation acts as a third signal during activation of human B cells,¹⁵² and, more importantly, human memory B cells depend on TLR activation to maintain long term serological memory.¹⁵³

B cells also express cytosolic PRRs to varying levels, but there is no strong evidence for any cell-intrinsic role for these receptors in either B cell activation or antibody production.^154,155 In fact human B cells appear to be non-responsive to cytosolic DNA making them ideal for harboring viruses such as EBV.¹⁵⁶ Overall, there is a dearth of compelling studies examining the function of cytosolic PRR pathways in B cells, so this area is ripe for further investigation.

Murine and human T cells express TLRs, as measured by mRNA expression, and both CD4 and CD8 T cells respond to specific TLR activation.^157,158 TLR2 acts as a costimulatory receptor on both CD4 and CD8 T cells promoting effector function, cytokine production and memory generation.^{159,160,161,162,163} Regulatory T (Treg) cells express several TLRs including TLRs 4,5 7 and 8; however, the role of TLR signaling in Treg cells is debated.¹⁶⁴ In mice, TLR4 engagement on Treg cells enhances their suppressive capacity.¹⁶⁴ Engagement of TLR8 on human Treg cells reverses suppressive function¹⁶⁵ and aids anti-tumor immunity¹⁶⁶ by inhibiting glucose uptake. However, engagement of TLR5 on human T cells enhances Treg cell suppressive capacity.¹⁶⁷ Thus, there is no clear consensus on the role of TLRs on Treg cells. In all in vitro studies that examine the function of TLRs on either conventional T cells or Treg cells, there is a concern of DC contamination that could confound their interpretation. A seminal study by Jonathan Sprent,¹⁶⁸ well before the discovery of TLR9, demonstrated that all the effects of immunostimulatory CpG on T cells was because of the presence of antigen presenting cells, and highly purified T cells were unresponsive to TLR9 ligands. Importantly, there are no compelling studies where TLRs have been deleted specifically in T cells to examine their importance in T cell function in vivo. Conceptually, the presence of functional TLRs on T cells is hard to reconcile with our understanding of how microbial sensing by DCs is linked to the acquisition of features on DCs that promotes T cell activation and differentiation. Direct sensing of TLR ligands by T cells would seem to bypass this strict control of T cell activation and might lead to undesirable outcomes of nonspecific T cell activation as seen in the case of superantigens. In any event, this topic of T cell-intrinsic TLR function, as well as any role for other PRRs, most certainly deserves further investigation and would benefit from the use of definitive in vivo experiments that utilize genetic tools.

Evaluation of cytosolic PRR expression in T cells provides conflicting data on whether cGAS and other cytosolic nucleic acid sensors are expressed in human and mouse T cells.^169,170 Even in studies that do suggest cGAS is expressed by T cells, there is no compelling evidence that cGAS can be activated in T cells by transfected DNA.¹⁷⁰ STING, on the other hand, is expressed by human T cells.¹⁷⁰ STING agonists, such as cGAMP, can induce type I IFN production by human T cells, but this activation seems to negatively impact their canonical function. T cells exhibit decreased proliferation and increased cell death following exposure to STING ligands.¹⁷⁰ Similar outcomes are seen in mouse studies where T cell-intrinsic STING activation during Listeria monocytogenes infection leads to T cell apoptosis and impaired T cell memory.¹⁷¹ Curiously, HIV infection fails to induce activation of the cGAS-STING pathway in CD4+ T cells¹⁷² suggesting that the virus successfully suppresses recognition or hides the ligands from sensing by cGAS or that cGAS is simply nonfunctional or not expressed in T cells. While HIV infection fails to activate the cGAS-STING pathway, it has been reported to induce NLRP3 inflammasome activation resulting in pyroptosis that contributes to CD4 T cell loss seen in HIV patients.^173,174 Other studies also implicate the CARD8 inflammasome in pyroptosis of HIV infected T cells.^175,176,177 Thus, cytosolic PRR activation appears to be generally detrimental to T cell survival and function.

Concluding Remarks

As potent activators of naïve T cells, DCs function as the bridge between innate and adaptive immunity. Engagement of PRRs on DCs improves their ability to activate and influence the differentiation of T cells by inducing generation and display of peptide-MHC complexes on the cell surface, expression of costimulatory molecules, and secretion of Signal 3 cytokines. PRR signaling also induces migration of DCs from peripheral tissues to secondary lymphoid organs where they present antigens they acquired in the tissues to naive T cells. Because PRRs have evolved to recognize the highly conserved microbial features that are PAMPs, their activation signals that the antigen presented by a DC is infectious in origin. Therefore, PRR activation on DCs leading to T cell activation is an essential aspect of any immune response, whether against a natural infection or elicited for therapeutic purposes. The impact that this paradigm has had on therapeutic efforts to induce (e.g., vaccines) or manipulate (e.g., cancer immunotherapy) immune responses cannot be overstated.

A major challenge going forward is to extend this paradigm to include the complexity of DC heterogeneity, multiple PRR families, and differences in PRR signaling that we have summarized here. Each of these components can result in distinct signals that influence T cell differentiation, but in many instances we lack a complete understanding of the precise mechanisms by which activation of specific PRRs in different DC subtypes, perhaps in specific tissues, leads to induction of a particular TH type and the ensuing adaptive immune response that is most appropriate for that challenge. The same can be said of how innate immunity controls B cell responses, which can be shaped by PRRs indirectly via Tfh cells as well as directly via cell-intrinsic PRR signaling. The ultimate manifestation of the DC/PRR paradigm will be an ability to elicit tailored immune responses against an antigen of interest via selective activation of specific PRRs and/or selective delivery to a particular DC type. While such precision may be difficult to imagine today, the remarkable progress made over the past 30 years can only leave one optimistic at the prospect.