A major recent development in innate immunity is the realization that microbial “non-self” and “damaged self” are recognized by a shared system of receptors. Acute inflammation is generated to deal with the “danger” inherent in such situations. Surprisingly, as will be discussed later, non-pathogenic commensal microbes that are tolerated by the body are recognized by the same system, suggesting regulatory control that is dictated by the nature of the challenge. Janeway (1
) first fully recognized that innate immune cells must have evolved a system for recognizing conserved “non-self” microbial products through pattern recognition receptors (PRRs). His group later identified a Toll-like receptor (TLR) as a key PRR capable of activating innate immune responses to bacterial lipopolysaccharide (LPS) (2
). Recently, the spectrum of PRRs has been widened to encompass intracellular nucleotide binding oligomerization domain (NOD)-like and retinoic acid inducible gene (Rig)-like receptors (3
) and C-type lectins (4
PRRs were originally proposed as a recognition system for exogenous (microbial) pathogen-associated molecular patterns (PAMPS). However, the same receptor superfamily was later found to recognize endogenous damage-associated molecular patterns (DAMPS), also known as “alarmins” (5
). This suggests that PRRs may have a role in maintaining tissue homeostasis, wound healing and tissue regeneration after damage, for which there is some evidence (7
). It would therefore be surprising if they were not major players in radiation-induced normal tissue damage and repair.
Most of what we know about PRRs today concerns TLRs, and we will focus on these as a model for how PRRs interface with our microbial world and with tissue damage to maintain homeostasis. TLRs are evolutionary ancient sensors that lie at the heart of our innate immune system. They are members of a superfamily of receptors that has homology to the drosophila Toll protein, but also to IL-1RI; all members share a TIR (Toll-IL-1 receptor) domain. There are currently about 12 members of the TLR subfamily, with some species variation (8
). The ligands for some TLRs have yet to be identified, but TLR4/MD2 dimers are particularly important in the response to LPS, a process that also involves CD14 in the formation of an activation cluster and sends a stronger signal than other TLRs. Of the other TLRs, TLR2 can form heterodimers with TLR1 or TLR6 and responds to lipopeptide components of gram-positive and -negative bacteria, and TLR5 recognizes bacterial flagellins. In contrast to these cell surface dimers, TLR3, TLR7 and TLR9 are intracellular receptors that sense mainly microbial RNA and DNA, as do NOD- and RIG-like receptors () [reviewed in ref. (7
FIG. 1 The Toll-like receptor system. DAMP and PAMP ligands activate dTLRs in the plasma membrane (TLR4/2/1/6/5) or in lysosomal vesicles (TLR3/7/9) through MyD88 or TRIF adapter proteins. NF-κB, AP1 or IRF3/7 transcription factors result in production (more ...)
DAMPS that we know are recognized by TLRs include the high-mobility-group box 1 (HMGB1) proteins. These are abundant chromatin-binding proteins that bind within the minor DNA groove and are released from damaged and activated cells. They share with LPS the ability to activate TLR4/MD2 but may also activate TLR2 (9
). Other DAMPS include heat-shock proteins, degradation products of extracellular matrix (surfactant protein A, fibronectin extradomain A and hyaluronan fragments), and other damage-associated proteins, such as beta defensin, uric acid, S100, minimally modified LDL and possibly proteins damaged by reactive oxygen species (ROS) (5
). As is the case with microbial nucleotides, endogenous DNA and DNA-activated autoantigens activate cells through TLR9, and the role of TLRs in various human autoimmune diseases is therefore an area of intense research (10
). The exact requirements for a molecule to act as a DAMP is not clear, but primarily only TLR2 and TLR4/MD2 seem to act as receptors. Presumably this restricts the response that can be made, as may the fact that a high concentration of DAMP is needed for TLR activation. Such control mechanisms must exist to ensure that the need for the response outweighs the damage that inflammation might cause.
In the real world, multiple TLRs will respond simultaneously to a challenge and the cellular distribution of the TLRs, their co-receptors and accessory proteins, and their downstream adaptor molecules form a mosaic signal that orchestrates the response. In general, immune cells express varying TLR profiles and were thought for a while to be the only players. But recently, epithelial, fibroblast and other cells also have been found to express TLRs, and although they may have a more restricted profile, they appear to be functionally important (see below). In spite of their complexity, all TLRs essentially signal through the adapter proteins, MyD88 and/or TRIF (11
), to activate primarily the transcription factors NF-κB and AP-1 and interferon regulatory factor (IRF) 3 or 7 (). Although this is an oversimplification, it is certainly true that the pathways are restricted and the target genes for NF-κB and AP-1 activation are pro-inflammatory cytokines, while IRFs signal type I interferon production. In fact, a spectrum of cytokines is produced in keeping with the need to eliminate viruses and intracellular bacteria on the one hand and deal with extracellular microbes on the other. How these polarized responses are orchestrated is not clear, but understanding the mechanism(s) will be important if we are to manipulate such responses for therapeutic benefit. In particular, while we know that radiation generates pro-inflammatory cytokines, the MyD88/TRIF dependence of the profile in different tissues has not yet been defined.
A compelling aspect of PRR signaling is that PAMPS and DAMPS can be classified as “danger” signals (12
) that link inflammation to antigen-specific immunity (5
). The signals that are generated can license immature dendritic cells (DCs), which normally maintain peripheral immune tolerance, to mature into potent antigen presenting cells that initiate antigen-specific immunity (14
). For DAMPS, this must be carefully controlled since autoimmunity is the flip side of this coin.
PRRs in Radiation and Immunity
The importance of PRR signaling in radiation responses has yet to be fully explored, but there are compelling hints as to its relevance. In retrospect, the older findings that LPS and IL-1 protect mice against lethal whole-body irradiation (16
) might now be seen as implicating the Toll-IL-1R superfamily. In fact, radiation has been shown to affect expression of TLR-related molecules. Shan et al.
) reported that 5 cGy to 2 Gy increased TLR4/MD2 and CD14 expression on mouse macrophages as well as elevating intracellular levels of MyD88, and this was thought to be responsible for their radiation-enhanced secretion of IL-12 and IL-18. More recently, a homolog of TLR4, but lacking the TIR domain, called radioprotective 105 (RP105) has been discovered in mouse B cells (18
). Its co-receptor, MD1, is a homolog of MD2. An antibody to RP105 caused B cells to proliferate and protected them against radiation-induced apoptosis. RP105 is also expressed in myeloid cells and, at least in some systems, it serves as a negative regulator of LPS/TLR4 signaling and cytokine production (19
). Thus, in the lung, while TLR4–/–
mice are more sensitive to bleomycin-induced epithelial injury and have decreased survival, the opposite seems to be true for RP105–/–
). Further work is needed to elucidate the situations in which TLR4 or RP105 is dominant, but clearly these mutually antagonistic signaling pathways may dictate pro-inflammatory cytokine production in response to radiation and TLRs may serve as useful targets for radiotherapeutic intervention.
The most compelling emerging concept is that radiation-induced DAMP signaling through TLR4 and TLR2 might affect the outcome of cancer treatment. Apetoh et al.
have shown that radiation releases HMGB1 from dying tumor cells and that HMGB1 is mandatory for host DCs to become licensed to present tumor antigens and generate tumor-specific immunity (9
). Intriguingly, patients with breast cancer who carry a TLR4 loss-of-function allele, which prevents HMGB1 binding, relapse more quickly after radiotherapy and chemotherapy than those carrying the normal TLR4 allele (9
The relationship between infection and cancer regression has a long history and has prompted many attempts to use microbial products for cancer immunotherapy, as with Coley's toxins in the early 20th century. Pathologists have frequently shown inflammation to correlate with the outcome of cancer treatment, for example in colorectal cancer (20
), and this response may in fact be a little-recognized factor in conventional treatment success (21
). It seems likely that the discovery of TLRs will herald a new era of investigation into how the host balances anti-tumor reactivity and normal tissue damage after radiation therapy and how to rebalance this equation to encourage a favorable outcome. In fact, we have known since the beginning of radiation therapy that it has a pro-inflammatory component, and numerous studies have explored the subsequent dialogue between the immune system and the mesenchymal and epithelial components that is required for successful tissue repair. The recent discovery that PRRs are not the sole property of immune cells but are expressed on other lineages, including epithelial cells (22
), and are key players in this dialogue forces a reassessment of these lines of communication in irradiated tissues.