Tissue-Based Class Control: the Other Side of Tolerance

PErsPEcTIvEs

the model did not offer any clues as to how ESSAY one effector class might be chosen over another, as it was designed to cover only the Tissue-based class control: the other immune system’s first decision (whether to respond or not). It proposed that per- side of tolerance turbed tissues initiate immune responses by sending alarm signals that activate local antigen-presenting cells (APCs), whereas Polly Matzinger and Tirumalai Kamala healthy tissues display their own antigens Abstract | In this Essay, we offer a new perspective on how immune responses are or allow ‘resting’ APCs to display those antigens to induce peripheral tolerance. In regulated. We do not cover how they are turned on and off, but focus instead on effect, this model suggested that turning the second major aspect of an immune response: the control of effector class. immune responses on or off was the pre- Although it is generally thought that the class of an immune response is tailored to rogative of the tissues. It takes only a small fit the invading pathogen, we suggest here that it is primarily tailored to fit the step to suggest that tissues may also control tissue in which the response occurs. To this end, we cover such topics as the nature the effector class, such that the class of an immune response is tailored to the tissue in of T helper (T ) cell subsets (current and yet to be discovered), the nature of H which it occurs, rather than to the invading privileged sites, the difference between oral tolerance and oral vaccination, why pathogen. The basics of this idea were out- 12,14 the route of immunization matters, whether the TH1‑type response is really the lined in two earlier articles . In this Essay, immune system’s primary defense, and whether there might be a different role for we describe the idea more fully, suggesting some regulatory T cells. mechanisms by which tissues could carry out this function, describing some well- known immunological phenomena in light When confronted with a potential threat, organisms; and the ‘innate’ receptors, such of this view, and pointing out the possibility the immune system faces two decisions: as the Toll-like receptors (TLRs) and NOD- that a complete definition of the immune first, whether to respond or not, and second, like receptors (NLRs), are so promiscuous system should perhaps include every tissue what kind of response to make. The first that they don’t distinguish between ligands in the body. decision has fascinated immunologists for from different phyla, or between pathogen- decades, sparking several theories and much derived and self-derived alarm signals6–9. Class and T helper cell subsets experimental work, most of which rests on Although current data suggest that TLR5 Let us start by defining what we mean by the assumption that the immune system and NOD2 (nucleotide-binding oligomeriza- immune effector class. Although the term responds to ‘foreign’ antigens, and that tion domain protein 2) may be fairly specific, “class” has historically been used to define maintaining self tolerance is a matter of con- it would be difficult to use even these recep- different antibody isotypes (such as IgM trolling autoreactive T and B cells. But deal- tors to make decisions about effector class. and IgG), we prefer a definition that also ing with autoreactive lymphocytes is only NOD2 does not distinguish between intra- includes the participating cells. Thus, each half of the problem. Even in the absence of cellular and extracellular bacteria, as it binds effector class combines a particular set of any autoreactivity, the wrong immune effec- muramyl dipeptide, a component of almost helper cells and the antibodies and effector class can completely destroy a tissue1–5. all bacterial cell walls10. Similarly, TLR5 tor cells that they promote. Currently, three The control of effector class, however, has binds the flagella of both intracellular patho- main subclasses are generally accepted. had little theoretical input. Students gener- gens (such as Listeria monocytogenes) and TH1-type responses consist of T cells that ally learn that the immune system matches extracellular pathogens (such as Escherichia produce interleukin-2 (IL-2), interferon-γ the effector class to the pathogen that it is coli and Pseudomonas aeruginosa)11, whereas (IFNγ) and tumour necrosis factor (TNF), as fighting (for example, making IgE against it does not bind the flagella of several well as B cells that make complement-fixing worms, and cytotoxic T lymphocytes (CTLs) other important pathogens (including IgG antibodies, CTLs, activated natural killer against viruses and intracellular bacteria). Helicobacter pylori, Staphylococcus aureus (NK) cells and macrophages that produce 11 But it is not easy to see how the immune and Campylobacter jejuni) . free oxygen radicals. TH2-type responses system could discriminate between worms, So, what controls the effector class of an consist of T cells that produce IL-4, IL-5, viruses or intracellular versus extracellular immune response? The idea that it might be IL-13 and IL-10, B cells making IgE and bacteria, as T cell receptors bind peptide– the tissues, rather than the immune system, IgG1, macrophages that express arginase,

MHC complexes; B cell receptors bind has grown slowly over the 13 years that we and the influx of eosinophils. TH17-type small epitopes on proteins, carbohydrates have been studying immunity from the per- responses consist of T cells that produce and lipids that are present in most living spective of the danger model12,13. Initially, IL-17 and the influx of neutrophils.

However, there have long been clues response, it has limited our ability to recognize There is, however, a world of differ- that this is an oversimplification. Human the potentially enormous diversity of immune ence between data and interpretations. The T cells, for example, often show non-classical responses. If we were to stop consigning interpretation that immunity cannot occur cytokine expression patterns, and it took TH cells to a small group of numbered subsets, in privileged sites makes little evolutionary a long time to persuade researchers work- but instead name them by what they produce sense. These tissues are wet, warm and full of ing with human cells to adopt the TH1- and (as is done for TH17 cells) or by the responses nutrients. Without protection by the immune 15 TH2-type classifications. In mice, Kelso they promote (as is done for follicular helper system, wouldn’t they promptly be exploited found that micromanipulated single T cells T cells), we would uncover the possibility that by pathogens? Luckily, other interpretations do not stably make TH1- or TH2-type cytokine there are a large number of differentiation exist. Streilein’s work on the eye reveals a

‘cassettes’. She suggested that the TH1- and paths that TH cells can take. We would suggest much more interesting picture.

TH2-type patterns are only the extremes of a that each particular effector cell (such as each The eye is a complex tissue, containing multidimensional continuum; that individual type of B cell, CTL, NK cell, macrophage, eosi- delicate specialized cells that cannot sur-

T cells normally make only a small number nophil, neutrophil and basophil) is controlled vive a full-blown TH1-type or delayed-type of stochastically produced cytokines; and by a particular set of secreted and membrane- hypersensitivity (DTH) response. It protects that populations of T cells can diversify to bound signals (from TH cells and from other itself by a process Streilein called “anterior produce any number of different cytokine sources) and can be combined with any chamber-associated immune deviation”27, combinations. More recently, O’Shea and other effector cell to make a wide variety of in which cells lining the anterior chamber Paul16 proposed a somewhat similar sce- carefully tailored immune responses. secrete cytokines — transforming growth nario, but these views are far from universal. Given the existence of such a variety of factor-β (TGFβ), vasoactive intestinal pep-

For example, Pulendran’s and Oppenheim’s effectors, and the TH cells that facilitate them, tide (vIP) and α-melanocyte stimulating groups found that dendritic cells (DCs) what determines the ultimate effector combi- hormone (αMSH) — that suppress TH1-type stimulated by Porphyromonas gingivalis- nation that arises in any particular immune and DTH responses and increase the activity 17 28 (Table 1) derived lipopolysaccharide (LPS) or by response? We were drawn to the possibility of regulatory T (TReg) cells . 18 eosinophil-derived neurotoxin induce TH that this is the responsibility of the tissues Although this appears to be immune sup- cells that secrete IL-5 but not IL-4 (the signa- by two old immunological phenomena: pression, a closer look suggests something ture TH2-type cytokine). In addition, Prussin immune-privileged sites and oral tolerance. different. TGFβ, vIP and αMSH all promote and colleagues19 found two different subsets IgA production29–32, but IgA cannot reject a Immune-privileged sites of TH cells in atopic patients, some of which transplant. So if we measure ocular immu- produce IL-5 but not IL-4. Nevertheless, Immune-privileged sites are organs in which nity only by transplant rejection, and ignore rather than postulate the existence of a TH cell allogeneic transplants are not rejected. the perfectly functional IgA response, we call that did not fit into the standard categories, all Neonatal hearts, for example, are rejected it “tolerance”, “deviation”, “suppression” or of these authors called these IL-5-producing when transplanted under the skin or kidney “regulation”. But this is none of those things. 20 cells “TH2 cells”. capsules of adult recipients , but survive It is simply a class of response that protects We believe it is time to follow Kelso’s lead indefinitely if placed into the anterior chamber the eye without destroying it. and stop forcing various kinds of immune of the eye21, the brain21, the testes22 or the The cells of the eye also express FAS lig- responses into a few common categories. hamster cheek pouch23. These observations and (also known as CD95L), which can trig-

Although the TH1/TH2 paradigm has been are generally interpreted as evidence that ger T cells through surface FAS (also known useful in establishing the concept that different ‘privileged’ sites exclude24, disable25 or as CD95) to die by apoptosis25. Although 26 sorts of TH cells promote different classes of suppress immune cells. initially interpreted as evidence that lymphocytes entering the eye are eliminated,

newer evidence that TH1 cells express more Table 1 | Cytokines that tailor immune effector class in the eye and gut 33 FAS than TH2 cells suggests that the eye ‘chooses’ what kinds of T cells it allows or Cytokine DTH or TH1-type TReg cell induction IgA production H response excludes. We predict that the other privileged In the eye sites also exert similar control over the local class of immune response. TGFβ ↓140 ↑141 ↑29,30 vIP ↓142 ↑143 ↑31,32 Oral tolerance and oral vaccination Oral tolerance has been extensively studied in αMsH ↓144 ↑145 ? experimental autoimmune encephalomyelitis. In the gut* This is an experimental system in which TGFβ ↓140 ↑141 ↑29,30 an animal that is immunized with a strong vIP ↓142 ↑143 ↑31,32 adjuvant plus a brain-derived protein, such as myelin basic protein (MBP), acquires an TsLP ↓ ↑146 ↑147‡ autoimmune disease mediated by TH1-type, 148,149 119,150,151 57 34 vitamin A (retinoic acid) ↓ ↑ ↑ TH17-type and DTH responses that some- DTH, delayed‑type hypersensitivity; αMsH, α‑melanocyte stimulating hormone; TGFβ, transforming what resembles multiple sclerosis in humans.

growth factor‑β; Treg, regulatory T; TsLP, thymic stromal lymphopoietin; vIP, vasoactive intestinal peptide. Feeding the animal MBP prior to MBP *There are other molecules secreted by gut epithelial cells (such as APrIL (a proliferation‑inducing ligand; immunization prevents disease and reduces also known as TNFsF13) and BAFF (B cell‑activating factor; also known as TNFsF13B) that are not as well 35,36 studied but which are likely also to influence the class of the immune response. ‡TsLP promotes IgA T cell responses . The T cells no longer indirectly by promoting APrIL and interleukin‑10 production. proliferate in response to MBP or make IFNγ

or TNF and, when transferred into another Box 1 | Class control by organs, tissues or regions? mouse, these T cells suppress the recipient’s autoimmune response. The interpretation of Is it organs or tissues that control the immune response? On the one hand, one could argue that these results, that orally administered antigen the function of each organ dictates that it should encourage particular types of response and discourage others. But are organs homogeneous in their needs? For example, the skin has a barrier generates tolerance and T cells, has spurred Reg function that may require certain types of immune functions, but the dermis and epidermis are not a novel treatment for allergy known as sub- the same. They have different populations of antigen-presenting cells (APCs) and might promote 37 lingual immunotherapy , in which patients different immune response classes. The jejunum and ileum, although comprising parts of the small apply small amounts of allergen sublingually intestine, have their own subsets of microflora and may need somewhat different types of immune each day. protection. Even within these intestinal regions, the epithelial microenvironments are different. In But what about oral vaccination, which the villus crypts, where most epithelial cell division occurs and where bacterial infection would be can elicit protective immunity against polio- particularly hazardous, Paneth cells produce large amounts of antimicrobial peptides. Further up virus in humans38 and against rabies virus the villus, the epithelial cells take on more of their own protection, expressing Toll-like receptors in raccoons and coyotes39? What is the dif- and producing different types of bactericidal molecules. ference between oral vaccination and oral One could argue, then, that each cell type has the ability to produce immune protective and immune-modulating signals, and that this implies that control of the immune response lies at the tolerance? In many cases, very little. Oral level of the cell. However, the same cell type might behave differently in different organs. For administration of antigen in mice elicits at example, the vascular endothelium is unlikely to be the same in the lungs, liver, heart, skin and least three different kinds of response. When kidneys. Will it communicate differently in those different sites with the cells of the immune system? given in large doses, it can induce systemic At the moment we don’t have answers to these questions. We don’t know what comprises a deletional tolerance40,41 (presumably because minimum tissue ‘unit’ that communicates with the immune system. For that reason, we use the enough antigen leaks into the circulation word “tissue” to define a local mixture of tissue cells that communicate with each other and with to be presented by resting DCs), as well as the bone marrow-derived cells that constitute the rest of the immune system. In some cases, a a local IgA response42. The addition of vari- local tissue might also communicate systemically with other tissues to help define the initiation, ous adjuvants converts this into a systemic longevity and effector class of an immune response. response that includes CTLs and IgG43. Lower doses of antigen can lead to the acti- 59 vation of TH3 cells that produce IL-4, IL-5, neutrophils; CTLs and NK cells kill target DTH responses and enhance TH2-type IL-10 and TGFβ and promote the production cells directly or through antibody-dependent responses60. In the gut, TGFβ acts as a 42,44 49,50 of IgA . Because the TH3 cells can suppress cell-mediated cytotoxicity ; macrophages switch factor that induces B cells to produce 51 61 inflammatory mechanisms through their release oxygen radicals ; and complement IgA , and vitamin D helps to recruit TH2 52 secretion of IL-10 and TGFβ, and because the drills holes in cell membranes . These are cells rather than TH1 cells by promoting the presence of IgA is rarely assessed, this type devastating weapons! production of CC-chemokine ligand 22 of response is also often labelled as tolerance. The eye and gut are not alone in being (CCL22)62, which recruits CC-chemokine + 63 But this is not tolerance. It is simply a switch susceptible to damage by these powerful receptor 4 (CCR4) TH2 cells . In the skin, to an immune response that is appropriate responses. Each organ is made of an intri- locally produced vitamin D can suppress for the intestinal environment. cate combination of tissues (bOX 1), precisely local DTH responses64. How does the gut promote the IgA tuned to perform particular functions that Second, many tissues have resident T cells response? Intestinal epithelial cells express can easily be compromised by destruc- that respond to stress-induced self molecules TLRs45 and secrete cytokines46 (Table 1). tive effector mechanisms. For example, rather than foreign antigens. The gut has 65 66 Some of these cytokines (including TGFβ strong TH1-type responses can destroy the intraepithelial αβ T cells and γδ T cells , and vIP) are similar to those produced in placenta1,53, pancreatic islets3, skin1, eye4, as well as mucosa-associated invariant T the eye, others (such as thymic stromal lym- brain54 and small intestine2. So most tissues (MAIT) cells67. These cells respond to the phopoietin (TSLP)) are unique to the gut, are likely to have mechanisms to avoid such stress-induced self molecules RAE1 (retinoic and some factors (such as vitamins A and D) destruction while promoting appropriate acid early-inducible protein 1) and MR1 are acquired in the diet and modified for local immunity. (MHC class I related) in mice, and to use. All of these factors shape the immune How might tissues communicate their MICA (MHC class I polypeptide-related response (Table 1). For example, TGFβ, TSLP, preferences to local and circulating cells of sequence A), MICB and MR1 in humans68. vitamin A and vitamin D have been shown to the immune system? First, the local tissue Similarly, up to 40% of liver-resident T cells suppress TH1-type and DTH responses and can produce (or modify) cytokines, chemo- are NKT cells that recognize several lipid promote the production of IgA. Thus, the kines and other communicating molecules molecules presented by the stress-induced gut seems to promote the antibody subclass (for example, antimicrobial peptides such as antigen-presenting glycoprotein CD1d69. (IgA) that is locally most useful, while simul- LL37 (Ref. 55); neuroactive molecules such Mouse and cattle skin contains dendritic epi- 56 taneously preventing destructive TH1-type as vIP and noradrenaline ; or vitamins dermal γδ T cells (DETCs), which respond and DTH responses. such as vitamins A57 and D58). These factors to RAE1 (Ref. 70) and help to heal the skin by can influence tissue-resident APCs to pro- making keratinocyte growth factor71, IL-2 Tissue-appropriate immunity mote a certain effector class (or classes) while and IFNγ72. The purpose of these tissue- Why would a tissue suppress DTH and discouraging others. They can also affect resident cells seems to be to survey the

TH1-type responses? Why not make many the entry and exit of innate and adaptive tissues they reside in for signs of stress and different effector classes to ensure pathogen immune cells, and govern what these cells to maintain the health of these tissues. The clearance? The reason is that these mecha- do in the local environment. For example, cells may achieve this by making molecules nisms are terribly destructive. TNF and fluid from the eye’s anterior chamber can that are important for healing and/or IFNγ induce cell death47,48; IL-17 recruits induce peritoneal macrophages to suppress by producing cytokines (and probably

chemokines, endogenous danger signals and Nevertheless, such destruction is a tre- is probably necessary. This fits with the com- neuropeptides) that promote the class of mendous price to pay for good health and mon findings that T cells stimulated by DCs immune response appropriate for that tissue should only be summoned when necessary. generated without tissue-derived signals at that time. How does a tissue determine when to (such as DCs generated in plastic dishes) or Third, neuronal signals may contribute to switch from one class of immune response by CD3-specific antibody tend to produce tissue-specific class control. Neuropeptides to another? TH0- or TH1-type responses, whereas T cells can influence effector class73, and many stimulated by DCs from mesenteric lymph leukocytes express neuropeptide receptors Controlling the switch nodes or Peyer’s patches produce IL-4, IL-10 that were previously thought to be restricted When considering what might control the and TGFβ88,92. Thus, tissue-derived signals to the nervous system74–78. Overall, tissues switch between the phases of an immune serve as checkpoints to prevent the destruc- seem to have multiple ways of communicat- response, we begin with three assumptions. tive default backup response. ing with the immune system and promoting The first is that tissues ‘educate’ resident Third, we assume that APCs carrying appropriate local immune functions. We APCs (and incoming T and B cells) to pro- tissue instructions to draining lymph nodes assume that each tissue has a particular set mote certain types of immune response survive for a while and are then replaced by of effectors that it prefers. IgA, for example, and suppress others81–86. Although it has new APCs. Their experimentally deduced is appropriate for the eye79, the gut and other been suggested that there are many subsets functional lifespans range from 3 days93 to mucosal surfaces, but may not be the right of DCs because of the need to distinguish 3 weeks94. effector for the brain, which would have its many different pathogens16, these subsets Given these assumptions, there are at least own preferred response class. might also correlate with the different kinds two non-exclusive possibilities for how tis- of tissues that need protection (for exam- sues might manage the switch from locally The three phases of an immune response ple, Langerhans cells can be distinguished tailored immune responses to destructive

If TH1-type, TH17-type and DTH responses from dermal DCs in unperturbed skin). backup responses: one based on time, the are so harmful, why have they evolved? One Furthermore, DCs remember their origins. other on signal strength. possibility is that tissues that regenerate easily DCs that migrate from the gut to mesenteric (such as the skin and liver) can tolerate the lymph nodes induce T cells to express gut- Time-dependent class switch. When acti- damage, whereas others (including the eye, homing receptors (such as CCR9)87 and vated by exogenous or endogenous6,95,96 alarm brain and pancreatic islets) might not. Thus, produce cytokines (including IL-4, IL-10 signals, tissue-resident APCs migrate to a DTH response in a non-regenerative tis- and TGFβ) that suppress TH1-type responses the draining lymph nodes and are replaced sue might be a case of the right action in the and promote IgA production41,61,81,88, whereas by new APC precursors. If the immune wrong place. DCs that leave damaged skin induce T cells Another possibility is that DTH is the to express skin-homing receptors (namely immunological counterpart of frostbite. On α4β1 integrin, CCR4 and CCR10)89,90. Figure 1 | A model for tissue-based class exposure to cold, capillaries dilate to keep We predict that a thorough analysis will control of immune responses. a | resting tis‑ extremities warm, but with intense cold or show that each tissue imparts specific instruc- sues educate local antigen‑presenting cells (APcs). b | Following an insult (such as an injury or long exposure the capillaries constrict, caus- tions that result in different populations of infection), the APcs leave the tissue to stimulate ing loss of extremities while preserving core TH cells promoting different combinations of naive T cells to make tissue‑educated responses. temperature. Perhaps the immune system’s effectors, each appropriate for that particular c | If the innate immune response clears the infec‑ response is similarly biphasic. It first uses injured or infected tissue. For example, three tion (or injured tissue), the tissue heals and edu‑ less destructive immune mechanisms to deal different TH cell subsets protect three differ- cates newly arriving APcs. An adaptive immune with adversity in a way that maintains the ent organs: CD62L+ T cells protect islets from response is not needed and ceases. d | If the innate health of all tissues. But if that doesn’t work, diabetes, CD25+ T cells protect the stomach immune response does not stop the infection, it switches to a second phase of destruc- from gastritis, and CD45RBlow T cells protect then tissue‑educated adaptive immune responses tive responses, sacrificing some tissues to the gut from colitis91. are initiated. If these clear the pathogen, then the preserve life. The instructions that tissues pass to tissue heals. e | If the tissue‑educated adaptive immune response cannot resolve the infection, In fact, there could be three phases. In T cells might be communicated through then a second wave of newly entering APcs will phase one, the tissue summons innate cells tissue-derived molecules, which could influ- be activated in a local tissue environment that for clean up and repair. If this isn’t sufficient, ence local or newly entering T cells directly now contains more extensive damage. The new it recruits the adaptive immune system, or which could be transported to the T cells APcs may be properly educated or they may not which attempts to clear the problem with by APCs. Alternatively, the signals might be be (because the high level of damage would result a locally tailored effector class. And if that produced by APCs as a consequence of the in fewer signals from the tissue). If not, they will doesn’t work, the tissue brings in the highly education that they received from their local leave the tissue and stimulate the emergency backup response. f | If the backup response clears destructive TH1-type, TH17-type and DTH tissue. If antigen recognition delivers ‘signal responses. DTH responses might destroy one’ and co-stimulation delivers ‘signal two’, the pathogen, then the tissue heals, but with some the eye while clearing an ocular infection80, then tissue-influenced signals governing the scarring or fibrosis occurring. g | If the initial insult is severe, the local APcs leave the tissue without but they ensure that the individual survives. ensuing effector class could be called ‘signal receiving a complete education. This could be In the gut, DTH responses may cause tem- t h r e e’. because the severely damaged tissues cannot porary flattening of gut villi and result in The second assumption we make is that provide the right signals or because the tissue diarrhoea, but villi regenerate quickly. And, when signal three is absent, the default action provides signals that override the original educa‑ although a TH1-type response may cause a is a TH1-type or DTH response. If a tissue tion. These APcs launch the immediate backup fetus to abort while clearing an infection, it becomes so damaged that it cannot deliver response. DTH, delayed‑type hypersensitivity; saves the mother’s life. APC-educating signals, the backup response Tcr, T cell receptor; TH1, T helper 1.

E6KUUWGJGCNGF 6JGVKUUWGKUJGCNGFD[GKVJGT KPPCVGQTCFCRVKXGKOOWPG TGURQPUGU0GY#2%UTGRNCEG +HKPUWNVKUUVQRRGFD[ VJGQPGUVJCVNGȎCPFDGEQOG GFWECVGF KPPCVGKOOWPGTGURQPUG 'FWECVKPI UKIPCN

D(KTUVFGEKUKQPKPKVKCVGCVKUUWGGFWECVGFTGURQPUG +HKPUWNVKUUVQRRGFD[ CPFDGIKPVJGTGRCKTRTQEGUU CFCRVKXGKOOWPGTGURQPUG +PLWTGFQTKPHGEVGF VKUUWGEGNN F5GEQPFFGEKUKQPVKUUWGGFWECVGFCFCRVKXG KOOWPGTGURQPUGDGIKPUVQENGCTVJGKPHGEVKQP +PVGTOGFKCVG #PVKIGP #PVKDQF[ 0GWVTQRJKKPUWNVN #EVKXCVKPI CPF GFWECVKPI! UKIPCN #2% 6KUUWGGFWECVGF GȭGEVQT6EGNN +HKPUWNV 0GY /CETQRJCIG EQPVKPWGU 5KIPCN #2% 0CKXG #EVKXCVKPI /KPKOCNQT 'FWECVGF /*% 6%4 6EGNN UKIPCN KPVGTOGFKCVG #2% KPUWNV EQOOQP 7PGFWECVGF &TCKPKPI #2% 5KIPCN N[ORJPQFG +PEQOKPIGFWECVGFGȭGEVQT6EGNNUCPFCPVKDQFKGUWUG +PPCVGKOOWPGEGNNUCTTKXGVQUVCTVVJGTGRCKTRTQEGUU VKUUWGGFWECVGFTGURQPUGUVQDGIKPENGCTKPIVJGKPHGEVKQP C4GUVKPIVKUUWG 'FWECVGF#2%UOKITCVGVQFTCKPKPIN[ORJPQFGUFGNKXGT 0GY#2%UOKITCVGKPVQVJGKPLWTGFVKUUWGCPFCTG EQUVKOWNCVQT[UKIPCNU UKIPCN CPFGFWECVKPIUKIPCNU 6KUUWGEGNN CEVKXCVGFVQNGCXGYKVJQWVIGVVKPICEQORNGVGGFWECVKQP UKIPCN VJCVKPFWEGPCKXG6EGNNUVQOCMGVKUUWG GFWECVGFGȭGEVQT6EGNNUUQOGQHYJKEJOKITCVGVQ +HKPUWNV KPLWTGFQTKPHGEVGFVKUUWGUQVJGTUVQ$EGNNHQNNKENGU 4GURQPUGUKPXQNXKPI EQPVKPWGU VKUUWGGFWECVGF#2%U #2% 4GURQPUGUKPXQNXKPI 'FWECVKPI WPGFWECVGF#2%U UKIPCN I+PKVKCVGVJGKOOGFKCVGDCEMWRTGURQPUG G6JKTFFGEKUKQPKPKVKCVGVJGGOGTIGPE[DCEMWRTGURQPUG 4GUVKPIVKUUWGURTQFWEGUKIPCNU #PVKIGP #PVKDQF[ VJCVGFWECVGTGUKFGPV#2%U

5GXGTG #EVKXCVKPI CPF #EVKXCVKPI KPUWNV TCTG QXGTTKFKPI! UKIPCN UKIPCN

6*EGNN 5KIPCN 6*EGNN 5KIPCN 7PGFWECVGF #2% 7PGFWECVGF #2%

&TCKPKPI 0QUKIPCN N[ORJPQFG &TCKPKPI 0QUKIPCN N[ORJPQFG #2%UNGCXGVJGKPLWTGFVKUUWGYKVJQWVTGEGKXKPIC EQORNGVGGFWECVKQP QTJCXKPIJCFVJGKTGFWECVKQP #PGYYCXGQH#2%UNGCXGUVJGKPLWTGFVKUUWGYKVJQWVTGEGKXKPIC QXGTTKFFGP! CPFUVKOWNCVGVJGGOGTIGPE[6*V[RGQT EQORNGVGGFWECVKQPCPFVJGUGEGNNUUVKOWNCVGVJGGOGTIGPE[6* &6*DCEMWRTGURQPUG'ȭGEVQT6EGNNUCPFEQORNGOGPV V[RGQT&6*DCEMWRTGURQPUG'ȭGEVQT6EGNNUCPFEQORNGOGPV ȮZKPICPVKDQFKGUIQVQVJGVKUUWGCPFENGCTVJGKPHGEVKQP ȮZKPICPVKDQFKGUIQVQVJGVKUUWGCPFENGCTVJGKPHGEVKQP

+PUWNVKUUVQRRGF +PUWNVKUUVQRRGFD[ D[DCEMWRTGURQPUG DCEMWRTGURQPUG

H6KUUWGJGCNGFYKVJUECTTKPI +PEQOKPIDCEMWRGȭGEVQTUCPF CPVKDQFKGUJCXGENGCTGFVJG 5ECT KPHGEVKQPDWVCNUQETGCVGFOQTG FCOCIG6JGVKUUWGJGCNUYKVJ 'FWECVKPI ȮDTQUKUQTUECTTKPI0GY#2%UIGV UKIPCN GFWECVKPIUKIPCNUHTQOVJG TGOCKPKPIVKUUWGEGNNU

response elicited by the original wave of the textbook consensus that the immune therefore, if the IgE responses reported tissue-resident APCs clears the infection, system tailors its responses to the pathogen against worms might be due to instruc- the new APC precursors that migrate to the (by initiating CTL responses to viruses tions from the worm, rather than decisions healing or healed tissue will become the new and intracellular bacteria, and TH2-type by the immune system. resident tissue-educated APCs. If, however, responses to extracellular bacteria and When studying immunity to a pathogen, the infection persists, this second wave worms)16,102. However, the evidence is not it is not always clear whose agenda we are of APCs may be activated before they are as clear as the textbooks would like to us studying. viruses are known to have evolved educated. When these uneducated APCs to believe. all sorts of mechanisms to subvert or modu- migrate to the draining lymphoid tissue, they First, the view that the immune system late immune responses115, and we should not will elicit the default backup TH1-type or mobilizes CTLs to clear viruses and IgE to expect less from pathogens with larger and DTH response. In this scenario, the backup eliminate worms is rather oversimplified. more complex genomes116. Their survival mechanism occurs only because the initial Most viruses elicit antibody responses, and depends on it and they have had evolution- tissue-oriented response has failed (fIG. 1). the isotypes vary with the infection site. ary time to devise the mechanisms. bOX 2 Measles and rubella viruses, for example, gives a few examples of the strategies used by Damage- or stimulus-dependent class inhabit the skin103 and elicit mostly IgG1, parasites and worms to modulate immune switch. under conditions of severe infec- IgG3 and IgG4 (Ref. 104), but rotavirus and responses in their favour. tion97, cell stress or destruction8, or in the influenza virus, which inhabit the gut and presence of noxious adjuvants28, the local the lungs, respectively, induce strong IgA The influence of history tissue may need to promote a strong backup responses105 (which can be more important If tissues and their resident immune cells response right from the start or may simply than CTLs for protection from reinfec- tend to promote immune response classes 106 have difficulty sending educating signals. tion ). unfortunately these local tissue other than the TH1-type or DTH response, APCs migrating from the tissue without hav- responses are often missed, either because why have immunologists thought of the ing had the proper final set of instructions they simply aren’t measured (for example, TH1-type response as the ‘normal’ response will, therefore, induce the backup response. TGFβ and secretory IgA are often left out for so long? This may be a historical acci- Later, when most of the infection has been of standard tests because they are difficult dent. Before we had in vitro culture sys- cleared and the healing tissue is once again to measure) or because the relevant B and tems and other laboratory tools, there were able to send educating signals, the response T cells home to distinctly local sites using two main ways of measuring cellular (as will revert to the tissue-oriented response98. specific chemokine receptors and adhesion opposed to humoral) immunity: the tuber- This might account for the occasional late molecules107,108, and discharge their antibod- culosis skin DTH test and graft rejection. appearance of less noxious responses99 ies and cytokines into local secretions rather Later, as we developed in vitro correlates of (fIG. 1). than serum109. such cellular immunity, we looked for com- These are not mutually exclusive sce- ponents that were part of these responses. narios. There might be situations in which For example, we generated assays for memory T cells influence APCs after they Tissues are not simply CTLs and NK cells, for macrophages that arrive at the draining lymph node and passive recipients of immune produce oxygen radicals, for T cells that counteract the tissue-derived instruc- protection produce TNF and IFNγ, and for antibod- tions that the APCs arrived with. For ies that fix complement. In other words, example, T cells induced by oral immu- we measured things that kill! If anyone nization can enter popliteal lymph nodes What about the idea that TH2-type wondered why immune responses should and re-educate DCs to promote TH2- or responses are the best mechanism to be so destructive, they probably assumed

TH3-type responses rather than TH1-type clear helminths? In mice infected with that this is the only way to fight pathogens responses83,92. There might be times when schistosomes, treatment with antibody and that the collateral damage is simply the the signals that APCs receive through their specific for IgE actually causes a reduc- price we pay. innate receptors are so strong that they tion in worm burden110, and worm num- Because what we think influences what overcome the tissue-educating signals. bers do not differ between IL-4-deficient we do, we also geared our model systems to There might also be positive or negative and wild-type mice111. Furthermore, generate these responses. We found culture feedback loops. For example, DCs that IL-4-deficient mice have lower burdens conditions to promote them in vitro and promote TH1-type or DTH responses make of Onchocerca microfilariae and greater adjuvants that elicit them in vivo (is it any IL-12, which stimulates NK cells. In turn, resistance to reinfection than wild-type surprise that our ‘best’ adjuvant for cellular activated NK cells can rapidly kill activated mice112. In human studies, people living in immunity, complete Freund’s adjuvant, 100 DCs , ensuring that a TH1-type or DTH Schistosoma mansoni endemic areas who contains mycobacteria?). We named the + response doesn’t last long. And when the remain clinically uninfected make strong CD4 T cells that enhance them TH1 cells, damage gets too great, the resulting hypoxia IFNγ responses to worm antigens, whereas and when we discovered that IgE produc- can downregulate the destructive immune those who maintain chronic low worm tion and allergy seem to be driven by a dif- response and switch it to another class101. burdens make IL-4 (Refs 113,114). This ferent kind of helper cell, we named them

suggests that an IFNγ-associated response TH2 cells and called their suppressive effect What about the pathogens? clears the worms and prevents reinfection, on TH1 cell responses ‘immune deviation’.

One could argue that the idea that tissues whereas TH2 cell functions instead allow When we discovered that orally admin- promote their own particular types of the worms to establish low-level coloniza- istered antigen elicits TH3 cell responses, response, irrespective of the infecting agent, tion and produce more worms without which involve TH cells making TGFβ and goes against a wealth of data supporting doing serious damage. One could ask, IL-10 and promoting IgA production, we

Box 2 | Class control by pathogens We are all guilty of this. Out of all of the ways to evaluate immune responses, each Viruses are known to have a plethora of mechanisms to influence and/or avoid the immune laboratory tends to use only a small subset. system115. Other organisms have not been as extensively studied, but some data are beginning to emerge. Some pathogens might choose to reside in tissues that not only offer good shelter and So when we do something (such as feed nutrition, but that also promote an effector class that is unable to effectively clear that parasite. antigen to animals, or add cytokines or Pathogens can also generate their own immune-influencing signals to exploit host defense antibodies to in vitro cultures) and notice strategies to their advantage. Leishmania parasites are an example. Sandflies taking a that the immune response we are measuring blood meal on mammalian skin induce a typical wound healing response that summons decreases, we call it tolerance. Then when we ‘alternatively activated’ macrophages that express high levels of arginase130. This is precisely transfer cells from that animal to another, or what Leishmania parasites need: readily available macrophages that they can infect to start the add them to other cells in vitro, and find that next stage of their life cycle. How does the macrophage-associated arginase help in this process? the recipient also fails to respond in our lim- Catabolism of arginine involves two enzymes that compete with each other. Inducible nitric ited set of assays, we call it suppression. But oxide synthase (iNOS) generates nitric oxide, which effectively kills intracellular Leishmania, we don’t notice that, unobserved, another whereas arginase catabolizes arginine to ornithine, a precursor of essential nutrients for Leishmania131. Leishmania parasites influence events even more in their favour by inducing the type of response is increasing. If, for exam- sandfly to secrete promastigote secretory gel (PSG), a potent inducer of arginase-expressing ple, we measured TH2 cells only by their abil- 132 macrophages . The more parasites harboured by the sandfly, the more PSG in its midgut and the ity to suppress TH1-type responses, we would more it deposits in the skin when it bites. Thus, the process by which Leishmania are deposited miss their capacity to promote IgE and IgG1 into host skin during a natural infection induces precisely the responses that are most likely to production and their role in allergy and maximize Leishmania survival and propagation, namely the migration to the site of infection of asthma. This is not really suppression, but a macrophages that synthesize the nutritional compounds that Leishmania require. switch of effector class. Perhaps, if we were Mycobacteria also have immune-subverting effects. Recent studies133 on early events in the to begin measuring the helper functions of establishment of mycobacterial infections have reversed long-standing assumptions about the T cells, we might find a new range of tools purpose of the granuloma, one of the most ancient host defense strategies by which multicellular Reg to help us modulate immunity. organisms wall off infectious agents and prevent their spread through the body134. Surprisingly, granulomas form rapidly during infection with virulent mycobacteria and have greater levels of We suggest that it is time to discard macrophage recruitment, motility and apoptosis than those that form during infection with the idea that the destructive TH1-type or non-virulent mycobacteria (which are poorly formed and result in attenuated infections). The DTH response is the immune system’s core result of the accelerated granuloma formation with virulent mycobacteria is early dissemination mechanism. When someone tells us that the

of the infection, through the release of infected macrophages from the primary granulomas and TH1-type response is the “natural” response production of secondary granulomas at distal sites133. Thus, the virulent mycobacterium converts in their experimental system, we need to an evolutionarily ancient form of host defense into a convenient and pliant tool that enables its ask, “what adjuvant are you using, what survival and more efficient propagation. dose of antigen are you using, where are you 135 136–138 9 Bordetella bronchiseptica , Schistosoma mansoni and Fasciola hepatica express molecules injecting it and what other responses are that promote T helper 2 cell responses, whereas Lewis antigen-expressing Helicobacter pylori139 you missing?” specifically block T helper 1 cell differentiation, presumably because such strategies work to their advantage. These strategies are likely to be only the tip of the iceberg, and we will find many more as we study the relationships between pathogens, commensals, symbionts and their hosts. A few final thoughts Although the body must clear some pathogens, or at least keep them in their place, mostly ignored them, or focused instead will have been stimulated mostly by anti- our commensal microflora must also be on their ability to suppress the TH1-type gens entering through the nose, mouth and maintained and our tissues kept healthy in inflammatory response and called them intestine. To help to promote appropriate the process. To accomplish this, the immune 117–120 TR1 cells or TReg cells . responses for these mucosal tissues they system does not use a set of rigidly defined

Although TReg cells were originally may also need to suppress other classes of TH1, TH2 or TH17 cells, but a wide variety of thought to be autoreactive T cells that response. It would not be surprising, there- TH cells that respond to signals from their are educated in the thymus121 to become fore, to discover that many CD25+ memory environment to mount a carefully balanced 122,123 suppressors , there are now known to TH cells from a normal mouse can suppress response to adversity. Some of these cells be several subsets. There are natural and IFNγ or TNF production, graft rejection or are tissue resident and some circulate. Some induced TReg cells, thymic and peripheral CTL activity. have genetically defined invariant T cell

TReg cells and self-reactive and non-self- One might ask why our postulated receptors, whereas others are somatically reactive TReg cells. There are TReg cells helper activity of tissue-specific TReg cells generated in each individual. Each of these induced by culture in the absence of co- has not been observed. The main reason is TH cells associates with various groups of stimulatory molecules and those induced that TReg cells are not often tested for helper B cells, macrophages, cytotoxic cells, neu- in the presence of co-inhibitory molecules functions. More than 99% of the work on trophils, eosinophils, basophils and tissue or other molecules such as TGFβ, vita- TReg cells uses assays that measure only their cells to tailor the response to the local milieu 125 min D, retinoic acid or antibodies specific ability to suppress TH1-type functions . and the pathogen. for CD40 ligand. There are TReg cells that If we don’t measure the mucosal memory In fact, we should perhaps redefine the make TGFβ or IL-10 or neither. And there response, we will mistakenly call this toler- immune system to include every tissue in have been endless discussions about which ance, suppression or regulation. In fact, in the body. Tissues are not simply passive 124 ones are the ‘real’ TReg cells . We would two studies in which intestinal antibody recipients of immune protection, but are suggest that many of these TReg cells are production was measured, forkhead box active participants in their own defense. actually misunderstood memory helper P3 (FOxP3)+ cells were shown to become They express TLRs128,129. They produce anti- T cells. In an unimmunized animal, they helper cells for the production of IgA126,127. microbial peptides and antiviral cytokines,

such as type I IFNs. They produce ‘eat me’ 18. Yang, D. et al. Eosinophil-derived neurotoxin acts as 42. Mestecky, J., Russell, M. W. & Elson, C. O. an alarmin to activate the TLR2–MyD88 signal Perspectives on mucosal vaccines: is mucosal signals to bring in scavenger cells, alarm sig- pathway in dendritic cells and enhances Th2 immune tolerance a barrier? J. Immunol. 179, 5633–5638 nals to activate local APCs, class-influencing responses. J. Exp. Med. 205, 79–90 (2008). (2007). 19. Prussin, C., Lee, J. & Foster, B. Eosinophilic 43. Maloy, K. J., Donachie, A. M., O’Hagan, D. T. & signals to modulate local immune responses gastrointestinal disease and peanut allergy are Mowat, A. M. Induction of mucosal and systemic alternatively associated with IL-5+ and IL-5– T 2 immune responses by immunization with ovalbumin and chemokines to recruit cells for repair, H responses. J. Allergy Clin. Immunol. 124, 1326–1332 entrapped in poly(lactide-co-glycolide) microparticles. remodelling and immunity. Finally, they (2009). Immunology 81, 661–667 (1994). may potentially also transmit ‘health’ signals 20. Hume, D. M. & Egdahl, R. H. Progressive destruction 44. Weiner, H. L. Induction and mechanism of action of renal homografts isolated from the regional of transforming growth factor-β-secreting Th3 to send away all of these cells when they lymphatics of the host. Surgery 38, 194–214 (1955). regulatory cells. Immunol. Rev. 182, 207–214 are no longer needed. To fully understand 21. Medawar, P. B. Immunity to homologous grafted skin; (2001). the fate of skin homografts transplanted to the brain, 45. Abreu, M. T. Toll-like receptor signalling in the these complex interactions we will need to to subcutaneous tissue, and to the anterior chamber intestinal epithelium: how bacterial recognition step back, have another look, start using of the eye. Br. J. Exp. Pathol. 29, 58–69 (1948). shapes intestinal function. Nature Rev. Immunol. 22. Barker, C. F. & Billingham, R. E. Immunologically 10, 131–144 (2010). assays that measure a wider array of immune privileged sites. Adv. Immunol. 25, 1–54 (1977). 46. Kaplan, H. J. & Niederkorn, J. Y. Regional immunity functions, and embrace the complexity that 23. Barker, C. F. & Billingham, R. E. The lymphatic status and immune privilege. Chem. Immunol. Allergy 92, of hamster cheek pouch tissue in relation to its 11–26 (2007). we find. properties as a graft and as a graft site. J. Exp. Med. 47. Boehm, U., Klamp, T., Groot, M. & Howard, J. C. 133 Annu. Rev. Polly Matzinger and Tirumalai Kamala are at the , 620–639 (1971). Cellular responses to interferon-γ. 24. Fijak, M. & Meinhardt, A. The testis in immune Immunol. 15, 749–795 (1997). Ghost Lab, Laboratory of Cellular and Molecular privilege. Immunol. Rev. 213, 66–81 (2006). 48. Tracey, K. J. & Cerami, A. Tumor necrosis factor: Immunology, T‑Cell Tolerance and Memory Section, 25. Ferguson, T. A. & Griffith, T. S. A vision of cell death: a pleiotropic cytokine and therapeutic target. National Institute of Allergy and Infectious Diseases, Fas ligand and immune privilege 10 years later. Annu. Rev. Med. 45, 491–503 (1994). National Institutes of Health, 9000 Rockville Pike, Immunol. Rev. 213, 228–238 (2006). 49. Kagi, D., Ledermann, B., Burki, K., Zinkernagel, R. M. 26. Caspi, R. R. Ocular autoimmunity: the price of & Hengartner, H. Molecular mechanisms of Bethesda, Maryland 20892, USA. privilege? Immunol. Rev. 213, 23–35 (2006). lymphocyte-mediated cytotoxicity and their role in Correspondence to P.M. 27. Streilein, J. W. Ocular immune privilege: therapeutic immunological protection and pathogenesis in vivo. e‑mail: [email protected] opportunities from an experiment of nature. Annu. Rev. Immunol. 14, 207–232 (1996). Nature Rev. Immunol. 3, 879–889 (2003). 50. Mukaida, N. Antibody-dependent cell-mediated doi:10.1038/nri2940 28. Niederkorn, J. Y. & Stein-Streilein, J. History and cytotoxicity (ADCC). Nippon Rinsho 57 (Suppl.), physiology of immune privilege. Ocul. Immunol. 571–573 (1999). 1. Streilein, J. W. Pathologic lesions of GVH disease in Inflamm. 18, 19–23 (2010). 51. MacMicking, J., Xie, Q. W. & Nathan, C. Nitric oxide hamsters: antigenic target versus ‘innocent 29. Coffman, R. L., Lebman, D. A. & Shrader, B. and macrophage function. Annu. Rev. Immunol. 15, bystander’. Prog. Exp. Tumor Res. 16, 396–408 Transforming growth factor β specifically enhances 323–350 (1997). (1972). IgA production by lipopolysaccharide-stimulated 52. Carroll, M. C. The role of complement and 2. Elson, C. O., Reilly, R. W. & Rosenberg, I. H. murine B lymphocytes. J. Exp. Med. 170, complement receptors in induction and regulation of Small intestinal injury in the graft versus host 1039–1044 (1989). immunity. Annu. Rev. Immunol. 16, 545–568 (1998).

reaction: an innocent bystander phenomenon. 30. Sonoda, E. et al. Transforming growth factor β induces 53. Chaouat, G. et al. TH1/TH2 paradigm in pregnancy: Gastroenterology 72, 886–889 (1977). IgA production and acts additively with interleukin 5 paradigm lost? Int. Arch. Allergy Immunol. 134, 3. Krook, H. et al. A distinct Th1 immune response for IgA production. J. Exp. Med. 170, 1415–1420 93–119 (2004). precedes the described Th2 response in islet (1989). 54. Linthicum, D. S., Mackay, I. R. & Carnegie, P. R. xenograft rejection. Diabetes 51, 79–86 (2002). 31. Boirivant, M. et al. Vasoactive intestinal polypeptide Measurement of cell-mediated inflammation in 4. Knisely, T. L., Luckenbach, M. W., Fischer, B. J. & modulates the in vitro immunoglobulin A production experimental murine autoimmune encephalomyelitis Niederkorn, J. Y. Destructive and nondestructive by intestinal lamina propria lymphocytes. by radioisotopic labeling. J. Immunol. 123, patterns of immune rejection of syngeneic intraocular Gastroenterology 106, 576–582 (1994). 1799–1805 (1979). tumors. J. Immunol. 138, 4515–4523 (1987). 32. Kimata, H. & Fujimoto, M. Vasoactive intestinal 55. Gilliet, M. & Lande, R. Antimicrobial peptides and 5. Abram, M. et al. Effects of pregnancy-associated peptide specifically induces human IgA1 and IgA2 self-DNA in autoimmune skin inflammation. Listeria monocytogenes infection: necrotizing production. Eur. J. Immunol. 24, 2262–2265 Curr. Opin. Immunol. 20, 401–407 (2008). hepatitis due to impaired maternal immune response (1994). 56. Nijhuis, L. E., Olivier, B. J. & de Jonge, W. J. and significantly increased abortion rate. 33. Fang, Y., Yu, S., Ellis, J. S., Sharav, T. & Braley-Mullen, H. Neurogenic regulation of dendritic cells in the Virchows Arch. 441, 368–379 (2002). Comparison of sensitivity of Th1, Th2, and Th17 cells intestine. Biochem. Pharmacol. 80, 2002–2008 6. Seong, S. Y. & Matzinger, P. Hydrophobicity: an to Fas-mediated apoptosis. J. Leukoc. Biol. 87, (2010). ancient damage-associated molecular pattern that 1019–1028 (2010). 57. Tokuyama, H. & Tokuyama, Y. Retinoids enhance IgA initiates innate immune responses. Nature Rev. 34. Shinohara, M. L., Kim, J. H., Garcia, V. A. & Cantor, H. production by lipopolysaccharide-stimulated murine Immunol. 4, 469–478 (2004). Engagement of the type I interferon receptor on spleen cells. Cell. Immunol. 150, 353–363 (1993). 7. Kawai, T. & Akira, S. The role of pattern-recognition dendritic cells inhibits T helper 17 cell development: 58. Baeke, F., Takiishi, T., Korf, H., Gysemans, C. & receptors in innate immunity: update on Toll-like role of intracellular osteopontin. Immunity 29, 68–78 Mathieu, C. Vitamin D: modulator of the immune receptors. Nature Immunol. 11, 373–384 (2010). (2008). system. Curr. Opin. Pharmacol. 10, 482–496 8. Zhang, Q. et al. Circulating mitochondrial DAMPs 35. Khoury, S. J., Hancock, W. W. & Weiner, H. L. (2010). cause inflammatory responses to injury. Nature 464, Oral tolerance to myelin basic protein and natural 59. Wilbanks, G. A., Mammolenti, M. & Streilein, J. W. 104–107 (2010). recovery from experimental autoimmune Studies on the induction of anterior chamber- 9. Dowling, D. J. et al. Major secretory antigens of the encephalomyelitis are associated with downregulation associated immune deviation (ACAID) III. Induction of helminth Fasciola hepatica activate a suppressive of inflammatory cytokines and differential ACAID depends upon intraocular transforming growth dendritic cell phenotype that attenuates Th17 cells upregulation of transforming growth factor β, factor-β. Eur. J. Immunol. 22, 165–173 (1992). but fails to activate Th2 immune responses. Infect. interleukin 4, and prostaglandin E expression in the 60. Kosiewicz, M. M., Alard, P. & Streilein, J. W. Immun. 78, 793–801 (2010). brain. J. Exp. Med. 176, 1355–1364 (1992). Alterations in cytokine production following 10. Koch, A. L. Bacterial wall as target for attack: past, 36. Miller, A., Lider, O., Roberts, A. B., Sporn, M. B. & intraocular injection of soluble protein antigen: present, and future research. Clin. Microbiol. Rev. 16, Weiner, H. L. Suppressor T cells generated by oral impairment in IFN-γ and induction of TGF-β and IL-4 673–687 (2003). tolerization to myelin basic protein suppress both production. J. Immunol. 161, 5382–5390 (1998). 11. Andersen-Nissen, E. et al. Evasion of Toll-like in vitro and in vivo immune responses by the release 61. Stavnezer, J. & Kang, J. The surprising discovery that receptor 5 by flagellated bacteria. Proc. Natl Acad. of transforming growth factor β after antigen-specific TGFβ specifically induces the IgA class switch. Sci. USA 102, 9247–9252 (2005). triggering. Proc. Natl Acad. Sci. USA 89, 421–425 J. Immunol. 182, 5–7 (2009). 12. Matzinger, P. Tolerance, danger, and the extended (1992). 62. Penna, G. et al. 1,25-Dihydroxyvitamin D3 selectively family. Annu. Rev. Immunol. 12, 991–1045 (1994). 37. Scadding, G. & Durham, S. Mechanisms of sublingual modulates tolerogenic properties in myeloid but not 13. Matzinger, P. The danger model: a renewed sense of immunotherapy. J. Asthma 46, 322–334 (2009). plasmacytoid dendritic cells. J. Immunol. 178, self. Science 296, 301–305 (2002). 38. Salk, D. & Salk, J. Vaccinology of poliomyelitis. 145–153 (2007). 14. Matzinger, P. Friendly and dangerous signals: is the Vaccine 2, 59–74 (1984). 63. Imai, T. et al. Selective recruitment of CCR4-bearing tissue in control? Nature Immunol. 8, 11–13 39. Sterner, R. T., Meltzer, M. I., Shwiff, S. A. & Slate, D. Th2 cells toward antigen-presenting cells by the (2007). Tactics and economics of wildlife oral rabies CC chemokines thymus and activation-regulated 15. Kelso, A. Th1 and Th2 subsets: paradigms lost? vaccination, Canada and the United States. chemokine and macrophage-derived chemokine. Immunol. Today 16, 374–379 (1995). Emerg. Infect. Dis. 15, 1176–1184 (2009). Int. Immunol. 11, 81–88 (1999). 16. O’Shea, J. J. & Paul, W. E. Mechanisms underlying 40. Chen, Y. et al. Peripheral deletion of antigen-reactive 64. Lemire, J. M. & Adams, J. S. 1,25-Dihydroxyvitamin lineage commitment and plasticity of helper CD4+ T cells in oral tolerance. Nature 376, 177–180 D3 inhibits the passive transfer of cellular immunity T cells. Science 327, 1098–1102 (2010). (1995). by a myelin basic protein-specific T cell clone. 17. Pulendran, B. et al. Lipopolysaccharides from distinct 41. Alpan, O., Rudomen, G. & Matzinger, P. The role of J. Bone Miner. Res. 7, 171–177 (1992). pathogens induce different classes of immune dendritic cells, B cells, and M cells in gut-oriented 65. Cheroutre, H. & Lambolez, F. The thymus chapter in responses in vivo. J. Immunol. 167, 5067–5076 immune responses. J. Immunol. 166, 4843–4852 the life of gut-specific intra epithelial lymphocytes. (2001). (2001). Curr. Opin. Immunol. 20, 185–191 (2008).

66. Groh, V., Steinle, A., Bauer, S. & Spies, T. Recognition 89. Mora, J. R. et al. Reciprocal and dynamic control of 113. Brito, C. F., Caldas, I. R., Coura Filho, P., of stress-induced MHC molecules by intestinal CD8 T cell homing by dendritic cells from skin- and Correa-Oliveira, R. & Oliveira, S. C. CD4+ T cells of epithelial γδ T cells. Science 279, 1737–1740 gut-associated lymphoid tissues. J. Exp. Med. 201, schistosomiasis naturally resistant individuals living in (1998). 303–316 (2005). an endemic area produce interferon-γ and tumour 67. Treiner, E. et al. Selection of evolutionarily conserved 90. Calzascia, T. et al. Homing phenotypes of tumor- necrosis factor-α in response to the recombinant 14 mucosal-associated invariant T cells by MR1. Nature specific CD8 T cells are predetermined at the tumor kDa Schistosoma mansoni fatty acid-binding protein. 422, 164–169 (2003). site by crosspresenting APCs. Immunity 22, 175–184 Scand. J. Immunol. 51, 595–601 (2000). 68. Hue, S. et al. A direct role for NKG2D/MICA (2005). 114. Viana, I. R. et al. Interferon-γ production by peripheral interaction in villous atrophy during celiac disease. 91. Alyanakian, M. A. et al. Diversity of regulatory CD4+ blood mononuclear cells from residents of an area Immunity 21, 367–377 (2004). T cells controlling distinct organ-specific autoimmune endemic for Schistosoma mansoni. Trans. R. Soc. Trop. 69. Bendelac, A., Savage, P. B. & Teyton, L. The biology diseases. Proc. Natl Acad. Sci. USA 100, Med. Hyg. 88, 466–470 (1994). of NKT cells. Annu. Rev. Immunol. 25, 297–336 15806–15811 (2003). 115. Tortorella, D., Gewurz, B. E., Furman, M. H., (2007). 92. Alpan, O., Bachelder, E., Isil, E., Arnheiter, H. & Schust, D. J. & Ploegh, H. L. Viral subversion of the 70. Jameson, J. & Havran, W. L. Skin γδ T-cell functions in Matzinger, P. ‘Educated’ dendritic cells act as immune system. Annu. Rev. Immunol. 18, 861–926 homeostasis and wound healing. Immunol. Rev. 215, messengers from memory to naive T helper cells. (2000). 114–122 (2007). Nature Immunol. 5, 615–622 (2004). 116. Terrazas, C. A., Terrazas, L. I. & Gomez-Garcia, L. 71. Boismenu, R. & Havran, W. L. Modulation of epithelial 93. De Smedt, T. et al. Antigen-specific T lymphocytes Modulation of dendritic cell responses by parasites: cell growth by intraepithelial γδ T cells. Science 266, regulate lipopolysaccharide-induced apoptosis of a common strategy to survive. J. Biomed. Biotechnol. 1253–1255 (1994). dendritic cells in vivo. J. Immunol. 161, 4476–4479 2010, 357106 (2010). 72. Boismenu, R., Hobbs, M. V., Boullier, S. & Havran, W. L. (1998). 117. Thorstenson, K. M. & Khoruts, A. Generation of Molecular and cellular biology of dendritic epidermal 94. Helft, J. et al. Antigen-specific T-T interactions anergic and potentially immunoregulatory CD25+CD4 T cells. Semin. Immunol. 8, 323–331 (1996). regulate CD4 T-cell expansion. Blood 112, T cells in vivo after induction of peripheral tolerance 73. Braun, A., Wiebe, P., Pfeufer, A., Gessner, R. & Renz, H. 1249–1258 (2008). with intravenous or oral antigen. J. Immunol. 167, Differential modulation of human immunoglobulin 95. Martinon, F., Mayor, A. & Tschopp, J. The 188–195 (2001). isotype production by the neuropeptides substance P, inflammasomes: guardians of the body. Annu. Rev. 118. Zhang, X., Izikson, L., Liu, L. & Weiner, H. L. Activation NKA and NKB. J. Neuroimmunol. 97, 43–50 (1999). Immunol. 27, 229–265 (2009). of CD25+CD4+ regulatory T cells by oral antigen 74. Dunzendorfer, S. & Wiedermann, C. J. Neuropeptides 96. Yu, L., Wang, L. & Chen, S. Endogenous Toll-like administration. J. Immunol. 167, 4245–4253 (2001). and the immune system: focus on dendritic cells. receptor ligands and their biological significance. 119. Coombes, J. L. et al. A functionally specialized Crit. Rev. Immunol. 21, 523–557 (2001). J. Cell. Mol. Med. 14, 2592–2603 (2010). population of mucosal CD103+ DCs induces Foxp3+ 75. Ho, W. Z., Lai, J. P., Zhu, X. H., Uvaydova, M. & 97. Eisenbarth, S. C. et al. Lipopolysaccharide-enhanced, regulatory T cells via a TGF-β and retinoic acid- Douglas, S. D. Human monocytes and macrophages Toll-like receptor 4-dependent T helper cell type 2 dependent mechanism. J. Exp. Med. 204, express substance P and neurokinin-1 receptor. responses to inhaled antigen. J. Exp. Med. 196, 1757–1764 (2007). J. Immunol. 159, 5654–5660 (1997). 1645–1651 (2002). 120. Sun, J. B., Raghavan, S., Sjoling, A., Lundin, S. & 76. Lai, J. P., Douglas, S. D. & Ho, W. Z. Human 98. Belkaid, Y. The role of CD4+CD25+ regulatory T cells Holmgren, J. Oral tolerance induction with antigen lymphocytes express substance P and its receptor. in Leishmania infection. Expert Opin. Biol. Ther. 3, conjugated to cholera toxin B subunit generates both J. Neuroimmunol. 86, 80–86 (1998). 875–885 (2003). Foxp3+CD25+ and Foxp3–CD25– CD4+ regulatory 77. Qian, B. F., Zhou, G. Q., Hammarstrom, M. L. & 99. Jankovic, D. et al. Conventional T-bet+Foxp3– Th1 T cells. J. Immunol. 177, 7634–7644 (2006). Danielsson, A. Both substance P and its receptor are cells are the major source of host-protective regulatory 121. Fontenot, J. D., Gavin, M. A. & Rudensky, A. Y. expressed in mouse intestinal T lymphocytes. IL-10 during intracellular protozoan infection. Foxp3 programs the development and function of Neuroendocrinology 73, 358–368 (2001). J. Exp. Med. 204, 273–283 (2007). CD4+CD25+ regulatory T cells. Nature Immunol. 78. Shibata, M., Hisajima, T., Nakano, M., Goris, R. C. & 100. Shah, P. D., Gilbertson, S. M. & Rowley, D. A. 4, 330–336 (2003). Funakoshi, K. Morphological relationships between Dendritic cells that have interacted with antigen are 122. Sakaguchi, S., Yamaguchi, T., Nomura, T. & Ono, M. peptidergic nerve fibers and immunoglobulin A- targets for natural killer cells. J. Exp. Med. 162, Regulatory T cells and immune tolerance. Cell 133, producing lymphocytes in the mouse intestine. 625–636 (1985). 775–787 (2008). Brain Behav. Immun. 22, 158–166 (2008). 101. Sitkovsky, M. & Lukashev, D. Regulation of immune 123. Shevach, E. M. Mechanisms of Foxp3+ T regulatory 79. Meek, B., Speijer, D., de Jong, P. T., de Smet, M. D. & cells by local-tissue oxygen tension: HIF1α and cell-mediated suppression. Immunity 30, 636–645 Peek, R. The ocular humoral immune response in adenosine receptors. Nature Rev. Immunol. 5, (2009). health and disease. Prog. Retin. Eye Res. 22, 712–721 (2005). 124. Shevach, E. M. From vanilla to 28 flavors: multiple 391–415 (2003). 102. Pulendran, B., Palucka, K. & Banchereau, J. varieties of T regulatory cells. Immunity 25, 195–201 80. Lausch, R. N., Monteiro, C., Kleinschrodt, W. R. & Sensing pathogens and tuning immune responses. (2006). Oakes, J. E. Superiority of antibody versus delayed Science 293, 253–256 (2001). 125. Mempel, T. R. et al. Regulatory T cells reversibly hypersensitivity in clearance of HSV-1 from eye. 103. Takahashi, H. et al. Detection and comparison of viral suppress cytotoxic T cell function independent of Invest. Ophthalmol. Vis. Sci. 28, 565–570 (1987). antigens in measles and rubella rashes. Clin. Infect. effector differentiation. Immunity 25, 129–141 81. Everson, M. P. et al. Dendritic cells from different Dis. 22, 36–39 (1996). (2006). tissues induce production of different T cell cytokine 104. Isa, M. B. et al. Comparison of immunoglobulin G 126. Tsuji, M. et al. Preferential generation of follicular B profiles. J. Leukoc. Biol. 59, 494–498 (1996). subclass profiles induced by measles virus in helper T cells from Foxp3+ T cells in gut Peyer’s 82. Bode, U. et al. Dendritic cell subsets in lymph nodes vaccinated and naturally infected individuals. patches. Science 323, 1488–1492 (2009). are characterized by the specific draining area and Clin. Diagn. Lab. Immunol. 9, 693–697 (2002). 127. Cong, Y., Feng, T., Fujihashi, K., Schoeb, T. R. & influence the phenotype and fate of primed T cells. 105. Gonzalez, R., Franco, M., Sarmiento, L., Romero, M. & Elson, C. O. A dominant, coordinated T regulatory Immunology 123, 480–490 (2008). Schael, I. P. Serum IgA levels induced by rotavirus cell–IgA response to the intestinal microbiota. Proc. 83. Iliev, I. D., Mileti, E., Matteoli, G., Chieppa, M. & natural infection, but not following immunization Natl Acad. Sci. USA 106, 19256–19261 (2009). Rescigno, M. Intestinal epithelial cells promote colitis- with the RRV-TV vaccine (Rotashield), correlate with 128. Zarember, K. A. & Godowski, P. J. Tissue expression of protective regulatory T-cell differentiation through protection. J. Med. Virol. 76, 608–612 (2005). human Toll-like receptors and differential regulation dendritic cell conditioning. Mucosal Immunol. 2, 106. Liew, F. Y., Russell, S. M., Appleyard, G., Brand, C. M. of Toll-like receptor mRNAs in leukocytes in response 340–350 (2009). & Beale, J. Cross-protection in mice infected with to microbes, their products, and cytokines. J. Immunol. 84. Rimoldi, M. et al. Intestinal immune homeostasis is influenza A virus by the respiratory route is correlated 168, 554–561 (2002). regulated by the crosstalk between epithelial cells with local IgA antibody rather than serum antibody 129. Hammad, H. et al. House dust mite allergen induces and dendritic cells. Nature Immunol. 6, 507–514 or cytotoxic T cell reactivity. Eur. J. Immunol. 14, asthma via Toll-like receptor 4 triggering of airway (2005). 350–356 (1984). structural cells. Nature Med. 15, 410–416 (2009). 85. Xia, S. et al. Hepatic microenvironment programs 107. Mora, J. R. et al. Generation of gut-homing IgA- 130. Shearer, J. D., Richards, J. R., Mills, C. D. & hematopoietic progenitor differentiation into secreting B cells by intestinal dendritic cells. Caldwell, M. D. Differential regulation of macrophage regulatory dendritic cells, maintaining liver tolerance. Science 314, 1157–1160 (2006). arginine metabolism: a proposed role in wound Blood 112, 3175–3185 (2008). 108. Fagarasan, S. & Honjo, T. Intestinal IgA synthesis: healing. Am. J. Physiol. 272, E181–E190 (1997). 86. Zeuthen, L. H., Fink, L. N. & Frokiaer, H. Epithelial regulation of front-line body defences. Nature Rev. 131. Kropf, P. et al. Arginase and polyamine synthesis are key cells prime the immune response to an array of gut- Immunol. 3, 63–72 (2003). factors in the regulation of experimental leishmaniasis derived commensals towards a tolerogenic phenotype 109. Brandtzaeg, P. Induction of secretory immunity in vivo. FASEB J. 19, 1000–1002 (2005). through distinct actions of thymic stromal and memory at mucosal surfaces. Vaccine 25, 132. Rogers, M. et al. Proteophosophoglycans regurgitated lymphopoietin and transforming growth factor-β. 5467–5484 (2007). by Leishmania-infected sand flies target the L-arginine Immunology 123, 197–208 (2008). 110. Amiri, P. et al. Anti-immunoglobulin E treatment metabolism of host macrophages to promote parasite 87. Kunkel, E. J. et al. Lymphocyte CC chemokine decreases worm burden and egg production in survival. PLoS Pathog. 5, e1000555 (2009). receptor 9 and epithelial thymus-expressed Schistosoma mansoni-infected normal and 133. Davis, J. M. & Ramakrishnan, L. The role of the chemokine (TECK) expression distinguish the small interferon γ knockout mice. J. Exp. Med. 180, granuloma in expansion and dissemination of early intestinal immune compartment: epithelial expression 43–51 (1994). tuberculous infection. Cell 136, 37–49 (2009). of tissue-specific chemokines as an organizing 111. King, C. L., Malhotra, I. & Jia, X. Schistosoma 134. Ulrichs, T. & Kaufmann, S. H. New insights into the principle in regional immunity. J. Exp. Med. 192, mansoni: protective immunity in IL-4-deficient mice. function of granulomas in human tuberculosis. 761–768 (2000). Exp. Parasitol. 84, 245–252 (1996). J. Pathol. 208, 261–269 (2006). 88. Iwasaki, A. & Kelsall, B. L. Freshly isolated Peyer’s 112. Hogarth, P. J., Folkard, S. G., Taylor, M. J. & 135. Siciliano, N. A., Skinner, J. A. & Yuk, M. H. patch, but not spleen, dendritic cells produce Bianco, A. E. Accelerated clearance of Onchocerca Bordetella bronchiseptica modulates macrophage interleukin 10 and induce the differentiation of microfilariae and resistance to reinfection in phenotype leading to the inhibition of CD4+ T cell T helper type 2 cells. J. Exp. Med. 190, 229–239 interleukin-4 gene knockout mice. Parasite Immunol. proliferation and the initiation of a Th17 immune (1999). 17, 653–657 (1995). response. J. Immunol. 177, 7131–7138 (2006).

136. Everts, B. et al. Omega-1, a glycoprotein secreted by receptor for vasoactive intestinal peptide. Proc. Natl conducive for Th1 cell development. Eur. J. Immunol. Schistosoma mansoni eggs, drives Th2 responses. Acad. Sci. USA 98, 13854–13859 (2001). 25, 1673–1679 (1995). J. Exp. Med. 206, 1673–1680 (2009). 143. Delgado, M., Chorny, A., Gonzalez-Rey, E. & Ganea, D. 150. Kang, S. G., Lim, H. W., Andrisani, O. M., Broxmeyer, 137. Steinfelder, S. et al. The major component in Vasoactive intestinal peptide generates CD4+CD25+ H. E. & Kim, C. H. Vitamin A metabolites induce gut- schistosome eggs responsible for conditioning regulatory T cells in vivo. J. Leukoc. Biol. 78, homing FoxP3+ regulatory T cells. J. Immunol. 179, dendritic cells for Th2 polarization is a T2 ribonuclease 1327–1338 (2005). 3724–3733 (2007). (omega-1). J. Exp. Med. 206, 1681–1690 (2009). 144. Taylor, A. W., Yee, D. G., Nishida, T. & Namba, K. 151. Sun, C. M. et al. Small intestine lamina propria 138. van der Kleij, D. et al. A novel host-parasite lipid Neuropeptide regulation of immunity. The dendritic cells promote de novo generation of Foxp3 cross-talk. Schistosomal lyso-phosphatidylserine immunosuppressive activity of alpha-melanocyte- T reg cells via retinoic acid. J. Exp. Med. 204, activates Toll-like receptor 2 and affects immune stimulating hormone (α-MSH). Ann. NY Acad. Sci. 917, 1775–1785 (2007). polarization. J. Biol. Chem. 277, 48122–48129 239–247 (2000). 145. Taylor, A. W. Modulation of regulatory T cell immunity (2002). Acknowledgements 139. Bergman, M. P. et al. Helicobacter pylori modulates by the neuropeptide alpha-melanocyte stimulating We thank A. Bendelac, T. Honjo, B. Jabri, Y. Rosenberg, F. Di the T helper cell 1/T helper cell 2 balance through hormone. Cell. Mol. Biol. 49, 143–149 (2003). Rosa, R. Schwartz, N. Singh, the ‘ghosts’ (K. Abdi, A. Perez- phase-variable interaction between 146. Mazzucchelli, R. et al. Development of regulatory Diez, A. Morgun and N. Shulzhenko) and especially P. Chappert lipopolysaccharide and DC-SIGN. J. Exp. Med. 200, T cells requires IL-7R stimulation by IL-7 or TSLP. α for commenting on the manuscript. T.K. would like to express 979–990 (2004). Blood 112, 3283–3292 (2008). special thanks to D. Usharauli for his encouragement and sup- 140. Meade, R. et al. Transforming growth factor-beta 1 147. He, B. et al. Intestinal bacteria trigger T cell- port during the preparation of this Essay. We apologize to the inhibits murine immediate and delayed type independent immunoglobulin A class switching by 2 authors whose work we didn’t cite owing to lack of space. This hypersensitivity. J. Immunol. 149, 521–528 (1992). inducing epithelial-cell secretion of the cytokine work was supported by the intramural program of the National 141. Bommireddy, R. & Doetschman, T. TGF 1 and APRIL. Immunity 26, 812–826 (2007). β Institute of Allergy and Infectious Diseases, National Institutes T cells: alliance for tolerance. Trends Mol. Med. 13, 148. Wiedermann, U. et al. Vitamin A deficiency increases reg of Health, Bethesda, Maryland, USA. 492–501 (2007). inflammatory responses. Scand. J. Immunol. 44, 142. Goetzl, E. J. et al. Enhanced delayed-type 578–584 (1996). hypersensitivity and diminished immediate-type 149. Cantorna, M. T., Nashold, F. E. & Hayes, C. E. Competing interests statement The authors declare no competing financial interests. hypersensitivity in mice lacking the inducible VPAC2 Vitamin A deficiency results in a priming environment