Toll Bridge to Immunity
Immune molecules hold promise for adjuvant discovery
By Mary Lee MacKichan, PhD*
Vaccine researchers have long depended on ill-defined additives called adjuvants to potentiate immune responses to immunogens. Injections of even large amounts of foreign protein from a pathogen are rarely enough to register on the radar of an animal's immune system. But add a pinch of an ingredient to the vaccine that bears no relation to the pathogen's protein—say mycobacterial cell wall extracts or alum—and the immune system springs to life. The late, great immunologist Charles Janeway famously termed the reliance on adjuvants to switch on the immune system the vaccinologist's "dirty little secret."
The pursuit of this mystery led to the discovery that some adjuvants work their magic on the immune system by activating Toll-like receptors (TLRs), now a red hot topic in immunology and vaccinology. They are the sentinels of the innate immune system, immediately sounding the alarm when a microbial invasion is detected. Crucially, TLRs, notably those on dendritic cells (DCs), also serve as a bridge to the adaptive immune system and the antigen-specific responses a vaccine elicits. The innate immune system and TLRs are now appreciated as principal actors in the drama between pathogen and host and their study will likely provide important lessons for vaccine design, including ones for the prevention of AIDS. "The beauty is that different TLRs seem to be specialized in the quality of the immune response they generate" says Bali Pulendran at Emory Vaccine Center. "This is great from a vaccinologist's point of view because it gives you new ways to manipulate the immune response."
TLRs are an evolutionarily ancient protein family. Toll, the founding member, was originally identified in the fruit flyDrosophila as an essential embryonic patterning molecule which is also required for innate defense against fungal infection. Human TLR genes were soon identified based on their sequence similarity to Toll, but a function in mammalian immunity was not demonstrated until Bruce Beutler from the Scripps Research Institute showed that in mice normal inflammatory responses to bacterial lipopolysaccharide (LPS) required TLR4. From there the linking of TLR family members with their microbial ligands proceeded quickly. To date 11 mouse and 10 human TLRs have been identified and shown to recognize a growing list of diverse molecules (Table 1; see Nat. Rev. Immunol. 4, 499, 2004 for a more exhaustive list).
TLRs are expressed by nearly all mammalian cells and are now recognized as the first switch that needs to be thrown to initiate protective immune responses against pathogens. TLRs serve their role as immunological watchdogs mainly by recognizing highly conserved molecules comprising critical structures of bacteria, fungi, or viruses. Many of these structures, such as cell wall constituents, are essential to the microbe's survival or infectivity, making it difficult for the microbe to go undetected by a host immune system or for the microbe to evolve effective replacements. Other molecules recognized by TLRs, notably single-stranded RNAs, are not unique to pathogens and instead the distinction between self and non-self depends on their subcellular localization. TLRs recognizing such molecules are situated so that they come into contact with their ligands only when the source is a pathogen rather than the host.
Within minutes of activation, TLRs contribute to the innate immune response by triggering the release of inflammatory cytokines and inducing the antimicrobial activities of macrophages, including migration, phagocytosis, and direct killing. Through their effects on DCs in blood and tissues, TLRs also lead to the induction of adaptive immunity, acting on both the myeloid DC (mDC) and plasmacytoid DC (pDC) subsets. When DCs encounter TLR ligands in tissues, they are stimulated to increase antigen uptake, migrate to lymph nodes and mature, up-regulating expression of MHC and co-activator molecules. Such mature DCs are required as antigen-presenting cells (APCs) that activate naive CD4+ T lymphocytes to become helper cells and differentiate into memory cells. Another APC type, B lymphocytes, is affected both indirectly and directly by TLR signaling: cytokines from CD4+ T cells influence antibody affinity and class switching, and B cells undergo polyclonal activation in response to some TLR ligands, including LPS (see Research Briefs).
Anatomy of TLR
TLRs are membrane spanning proteins, many of which function on the surfaces of cells, while others work in internal cell compartments. The ligand recognition domain of TLRs is comprised of leucine-rich repeats (LRRs) and serves as the antenna which detects microbial molecules. Though not yet conclusively demonstrated, data suggest these repeats bind ligand directly. Additional co-receptors can be involved in ligand recognition, as is the case for TLR2 (CD36) and TLR4 (CD14 and MD2), while as yet unidentified co-receptors may interact with other TLRs.
The signaling domain of a TLR transmits news of microbial invasion to cells through a classical signal transduction pathway, comprised of a complex set of molecular relays. This domain consists of a Toll–IL1 receptor (TIR) element, so called because it is also present in the receptor for the pro-inflammatory cytokine interleukin-1. Revealing the mechanism of how TLR activation is translated into immune function is a rapidly moving area of research. What's clear is that the signaling is mediated in large part by the cytosolic adaptor protein MyD88, a protein that has its own TIR domain that directly engages its twin on TLRs. Through a separate domain, MyD88 then recruits signaling molecules, including IRAK family kinases and TRAF6, to turn on downstream pathways leading to activation of MAP kinases and the transcription factors NF-κB and interferon response factors (IRFs). TLR3 signaling is uniquely independent of MyD88 and instead relies on another adaptor, TRIF, to link it to downstream pathways. Activation of these pathways ultimately leads, depending on the TLR and cell type, to expression of cytokine and co-stimulatory molecules, cellular proliferation or increased survival, and changes in actin and cell motility (Figure 1).
Microarray studies of TLR-induced gene expression confirm that different TLRs modulate expression of common target genes while other genes are specifically responsive to a given TLR. Differences in adaptor protein recruitment can account for some of the specificity of responses to different TLRs. But other subtle differences in the signaling cascades downstream of TLRs can also translate into large differences in gene transcription and ensuing adaptive immunity. For example, initial signaling responses following activation of TLR2 and TLR4 overlap significantly. But Pulendran's laboratory has demonstrated that DC activation following engagement of TLR2 results in Th2 CD4+ helper T cell responses, while TLR4 results in Th1 responses. Further investigation revealed that the basis of this difference lies in the quantitative, rather than qualitative, aspects of the initial response: the TLR2 ligand induced more prolonged and intense activation of the signaling molecule ERK MAP kinase than did the molecule binding TLR4. The enhanced ERK signaling downstream of TLR2 stabilized the transcription factor c-Fos which, in turn, repressed IL-12 expression by DCs, resulting in a Th2 response (J. Immunol. 171, 4984, 2003).
While the detailed roadmap of the links between TLRs, downstream pathway activation, and immune responses is still being drawn, the importance of TLR 3, 7, 8, and 9 for antiviral immunity is emerging rapidly (reviewed in J. Virol. 78 7869, 2004). All four receptors recognize nucleic acids and endosomal acidification is required for their activation. It is believed that viral ligands come into contact with TLRs in APCs through receptor-mediated uptake of virus or viral fusion with endosomal membranes (see Proc. Natl Acad. Sci. USA 101, 6835, 2004 for discussion). For viruses that either do not infect or cannot replicate in APCs, CD8+ T cells can be activated by cross-priming from DCs that engulf and then present antigen from apoptotic cells infected with virus. TLR3 in APCs appears to be involved in mediating such cross-priming by recognizing double-stranded RNAs found in infected apoptotic cells. A recent report in Nature (433, 887, 2005) showed that the robust cytotoxic T lymphocyte (CTL) response produced by immunization with virally-infected cells required this receptor and phagocytosis.
Interferon-a (IFN-a) is a hallmark of the response to many viral infections and pDCs—which express TLRs 7 and 9—are the major source of this cytokine (see J. Virol. 79, 17, 2005 for review of pDCs). IFN-a stimulates multiple innate protective pathways, including intracellular RNAse activity, that inhibit viral replication and can lead to viral clearance. Human pDCs have been shown to be activated via TLRs to make IFN-a in response to multiple enveloped viruses, including influenza virus and HIV.
TLRs and HIV
The role of TLRs in the HIV-specific immune response is slowly being unraveled. Multiple studies support the idea that HIV nucleic acids, and possibly envelope protein, are involved in activating TLRs during infection. A ssRNA oligonucleotide derived from HIV-1 U5 was shown to act through TLR8, and possibly TLR7, to induce production of IFN-a and other cytokines (Science 303, 1526, 2004). In addition to evidence that synthetic HIV sequences can be detected by TLRs, human pDCs are activated and mature following exposure to intact HIV in vitro (J. Virol. 78, 5223, 2004). This pDC response to HIV is unaffected by chemical inactivation of the virus, suggesting that membrane fusion or replication is not required for detection by TLRs. Nina Bhardwaj's laboratory at New York University recently demonstrated that viral nucleic acid is required for HIV stimulation of pDC through TLRs and identified TLR7 as the likely target for recognition of HIV RNA. However, HIV uptake, mediated by envelope protein and cellular CD4 interactions, as well as subsequent endosomal acidification, are also required for viral RNA to reach TLR7 (J. Clin. Invest. 115, 3265, 2005).
While HIV clearly can activate pDCs via TLRs in vitro, during natural infection this interaction may fail to occur or the virus may interfere with downstream responses allowing establishment of a chronic infection. Precedent for such interference has been uncovered in other viral infections, including vaccinia and hepatitis C virus (HCV). Vaccinia encodes a TIR-containing decoy protein that blocks TLR signaling (J. Exp. Med. 201, 1007, 2005), whilein vitro evidence suggests HCV NS3/4 protease can cleave TRIF to suppress TLR3-mediated production of IFNs and other cytokines (Proc. Natl Acad. Sci. USA 102, 2992, 2005). If it turns out that HIV latency or chronicity involves interference with TLR pathways, any such mechanism could offer a novel target for intervention.
TLRs and vaccines
The goal for vaccinologists in tapping TLR responses is to strike a balance between immune activation and the potential damage that can result from inflammation. The importance of TLR-mediated responses in achieving protective immunity is being reinforced by research from the Pulendran laboratory where he and his colleagues have begun analyzing the workings of the highly efficacious yellow fever vaccine, YF-17D. Their results suggest YF-17D activates multiple TLRs, including TLR 2, 7, 8, and 9 on multiple subsets of DCs. "It's almost too good to be true. The vaccine induces a remarkably broad spectrum of responses including CTL, broad neutralizing antibodies, Th1, and Th2 responses. It is clear that this diversity is achieved by activating multiple TLRs," says Pulendran. "I think YF-17D might be teaching us vaccinologists a lesson, providing a clear scientific rationale for incorporating more than one (TLR) ligand into a vaccine formulation." However, he notes that the need to license multiple TLR ligands could be a hurdle to commercialization of this approach.
As with any attempt to tinker with natural responses, manipulating the immune system via TLRs may have a dark side. Because TLRs recognize nucleic acids some experts have proposed a role for TLRs in the generation of lupus-like autoimmune syndromes characterized by the production of anti-DNA and -RNA antibodies. TLR activation has also been suggested to be involved in the genesis of asthma and allergy. In addition, a polymorphism identified in human TLR4 not only correlates with increased susceptibility to certain bacterial infections but also to a lower risk of atherosclerosis and acute cardiac events, suggesting that activation of other TLR4 alleles could promote atherosclerosis (see Nat. Immunol. 5, 975, 2004 for a review of TLRs in disease.) "The key for vaccines will be to harness immunostimulatory parts of the pathways and limit the toxic parts of the pathways," says Richard Ulevitch of Scripps Research Institute. Robert Seder at the US National Institutes of Health believes some "side-effects" will be unavoidably associated with any TLR agonist. "These are inflammatory responses," he says. "People are going to get feverish." Pulendran suggests correlates of immunity versus toxicity may be defined by thresholds rather than distinct pathways. "It may be that just more of the same thing is bad," he says. He emphasizes the need for quantitative experiments with titrated doses of adjuvants to separate out beneficial effects from toxicity.
One complication to development of novel TLR ligands for use in human vaccines is the species-specificity of TLR expression, ligand recognition, and DC subsets. Mice can make a robust immune response that includes CTL in response to vaccination with TLR ligands such as CpG oligonucleotides and protein antigen, but the CD8+ DC subset responsible may not be present in primates. Because of the limitations of mouse models to answer all questions surrounding TLRs, Susan Barnett of Chiron says her group uses multiple animal models as well as in vitro screens to test TLR-related vaccine adjuvants. But Barnett says that TLR ligands, such as CpG oligonucleotides, might be tested clinically as part of an AIDS vaccine in the next few years, allowing direct assessment of their efficacy in humans.
Using what is already known about TLRs, vaccinologists have begun shifting from a largely empirical approach to the more rational design of adjuvants to take advantage of these mechanistic insights. The investigation of microbial and synthetic TLR ligands already underway has intensified, with the majority of attention focusing on CpG oligonucleotides (for TLR9) and small molecules in the imidazoquinoline family. CpG oligonucleotides have received much attention for their potential to activate multiple cell types and promote Th1 responses and are currently undergoing testing in the clinic for treatment of cancer and viral infection. Imiquimod, a small molecule imidazoquinoline ligand for TLR7 and 8, is already marketed as a topical antiviral to treat human papillomavirus and basal cell carcinoma. The drug’s activity was discovered empirically and later the mode of action was shown to involve TLR-dependent activation of DCs to secrete IFNs and other cytokines. With the TLRs in hand the search for new synthetic TLR ligands is now booming (see Nat. Rev. Drug Discov. 4, 879, 2005 for a detailed summary of TLR-targeted therapeutics). While most of these efforts are focused on therapeutic applications, "in principle it should be possible to set up screens for TLRs that would identify novel ligands that could act as adjuvants" says Ulevitch.
Using drug-like chemicals as adjuvants may also have the advantage of allowing the TLR stimulus and antigen to be delivered together, possibly as conjugated molecules. "Synchronous delivery is going to be critical. If protein arrives after the TLR stimulus is gone it's a bad thing because DCs won't take it up," says Seder. "These (conjugated TLR ligand and protein antigen) are terrific for antibody and Th1 responses," he says, "and I think it is possible to generate CD8+ T cell responses with proteins under the right conditions. I don't think they will be as good as replication-defective viral vaccines, but that doesn't mean they can't be used in combination with virus." While citing evidence from his own work (see Research Briefs) and others suggesting conjugation dramatically improves T cell responses, he cautions that conjugation could sometimes produce suboptimal antibody responses.
The full potential of harnessing TLR signaling pathways will only be clear once they are fully elucidated. And that work is, at best, half done. Based on saturation mutagenesis studies, Beutler calculates that approximately fifty proteins may have non-redundant function in TLR signaling to induce TNF, and quite a few more may be involved in IFN production. More than half of these proteins remain to be identified. Even so, he thinks that TLRs may not offer the best way to stimulate every aspect of immunity. "While TLRs give mainly a CD4+ T cell response, other TIR-independent pathways induced by apoptotic cells are better inducers of CD8+ responses." Pulendran cites other open questions in the field. "Understanding how TLRs interact with other receptors in the context of a real infection will be important. We don't yet know how to get a good B cell response, whether we need to engage TLR on the B cell itself in addition to APC activation. There is scope for a lot of research."
Looking ahead, Pulendran envisions a day when designer adjuvants will combine TLR ligands and small molecules to other targets to push the immune system in precisely the right direction for a given vaccine. "Although it sounds like science fiction, in 15 years or sooner adjuvants may be nanoparticles that contain not only specific TLR ligands but also specific inhibitors of intracellular signalers like ERK or c-Fos," he says.
*Mary Lee MacKichan, PhD, is a freelance science writer based in San Francisco.