Solar UVR is highly mutagenic but is only partially absorbed by the outer stratum corneum of the epidermis. UVR can penetrate into the deeper layers of the epidermis, depending on melanin content, where it induces DNA damage and apoptosis in epidermal cells, including those in the germinative basal layer. The cellular decision to initiate either cellular repair or undergo apoptosis has evolved to balance the acute need to maintain skin barrier function with the long-term risk of retaining precancerous cells. Langerhans cells (LCs) are positioned suprabasally, where they may sense UV damage directly, or indirectly through recognition of apoptotic vesicles and soluble mediators derived from surrounding keratinocytes. Apoptotic vesicles will contain UV-induced altered proteins that may be presented to the immune system as foreign. The observation that UVR induces immune tolerance to skin-associated antigens suggests that this photodamage response has evolved to preserve the skin barrier by protecting it from autoimmune attack. LC involvement in this process is not clear and controversial. We will highlight some basic concepts of photobiology and review recent advances pertaining to UV-induced DNA damage, apoptosis regulation, novel immunomodulatory mechanisms and the role of LCs in generating antigen-specific regulatory T cells.
Solar UVR makes up just 5% of the electromagnetic spectrum that reaches the earth’s surface. Three spectral regions have been designated based on their biologic effects. Terrestrial UVR consists of 3–6% UVB (290–320 nm) and 94–97% UVA (320–400 nm). Negligible amounts of UVC (200–290 nm) reach the earth’s surface due to the filtering capacity of the ozone layer (1). UVR is a potent environmental carcinogen and is largely responsible for the development of the most common cancer worldwide, skin cancer. In the United States, one million new cases of basal and squamous cell carcinomas (SCC) as well as 60 000 cases of melanoma are expected in 2007 (2). The steady increases in melanoma and nonmelanoma skin cancer cases contrast with the recent downward incidence for all other cancers (excluding lung cancer in women) (2). The increases in skin cancer are largely attributed to recreational sun exposure (including tanning beds) practiced by the population (3–6). Concern that further increases in skin cancer incidence may result because of ozone depletion may be tempered by positive global efforts to reduce ozone-depleting substances in the atmosphere (7). In order to ameliorate the worldwide burden of UVR-related pathologies such as sunburn, aging, autoimmunity, immune suppression and cancer, it is imperative that we gain a better understanding of the mechanisms of UVA- and UVB-induced photodamage and how they relate to the molecular and immunologic nature of photodamage responses.
The genetic mechanisms by which UVR transforms and promotes various skin cancers have been under intense investigation for decades, and much progress has been made in identifying genes that contribute to the oncogenic process in the development of melanoma, SSC and basal cell carcinoma (8,9). However, in addition to generating genetic mutations, UVR actively suppresses the normal processes of immune surveillance responsible for eliminating mutant cells, which permits tumor growth. This was first shown in a series of experiments that launched the field of photoimmunology over 30 years ago. Kripke and colleagues demonstrated that UV-induced tumors were highly immunogenic. These tumors were not recognized by immune-deficient mouse strains or in parabiotic immune-competent mice sharing the same hematopoietic system as UV-treated mice (10). The effectors of UV immunosuppression could be transferred by injecting T cells from UV-treated mice into naive recipients (11). Such treatment permitted the outgrowth of many transplanted UV-induced tumors but not tumors generated by different means, such as chemically induced tumors. Collectively those experiments established that UV-induced suppression generated a subset of T suppressor cells that were UV antigen specific. Indeed, UV-induced suppression can develop against many types of antigens that appear in conjunction with UVR in both murine and human systems. This was first shown for contact hypersensitivity (CHS) responses to hapten in 1963 (12) and then, with revived interest in this field, was shown to be mediated by CD4+ T cells (13,14). In addition to effecting CHS responses to hapten, UVR-induced immune suppression can affect delayed-type hypersensitivity responses (15,16), susceptibility to infectious diseases (17), vaccine immunogenicity (18) and recall responses (19). It is now appreciated that multiple mechanisms are invoked to orchestrate UVR-induced immunosuppression. Differences in UV wavelength (UVA versus UVB), UV dose (low versus high), duration of UV exposure (acute versus chronic) and genetic factors (20) have revealed that the chief mediators of suppression may differ according to the protocol that is used (reviewed in Leitenberger et al. ). Many biochemical and molecular factors have been identified as important mediators of photodamage-related immune suppression (reviewed in Ullrich ). In this review, we focus on recent advances made toward understanding the molecular and cellular triggers of this process. In addition, we will re-examine some key experiments that cutaneous photobiology has built upon, and present new information regarding specific DNA photoproducts generated by different UV spectrums, molecular mechanisms involved in cellular decision-making, apoptosis regulation and discuss how those processes may program Langerhans cells (LCs) to directly or indirectly activate antigen-specific regulatory T (Treg) cells.
The biologic receptors of UV damage are molecules that directly absorb the energy of UVR and undergo a conformational or structural change. The altered molecules are recognized as an alarm signal that initiates cellular photodamage responses that are detected as erythema, sunburn, and immune suppression. Two well-characterized photodamage signaling molecules are trans-urocanic acid (trans-UCA) and DNA.
trans-UCA is present at high concentrations in the stratum corneum (SC) and strongly absorbs short-wave UVR, which can attenuate the penetration of UVC and UVB radiation into skin tissue. UVR isomerizes trans-UCA to cis-UCA and this isomer has potent immunosuppressive qualities. Recently the serotonin [5-hydroxytryptamine (5-HT)2a] receptor was shown to bind cis-UCA and mediate suppression. UV- and cis-UCA-induced immune suppression was blocked by antiserotonin antibodies or with 5-HT2A receptor antagonists. Human dendritic cells (DC), monocytes, mast cells, activated murine T cells and keratinocytes have been shown to express the 5-HT2A receptor or its gene (23–26). cis-UCA is thought to promote systemic suppressive effects that occur with high doses of UVR. As a direct effect on LC function has not been shown, it has little effect on local tolerance mechanisms (27,28).
DNA is another important photodamage signaling molecule. DNA photoproducts are altered DNA structures that activate a cascade of responses, beginning with the initiation of cell cycle arrest and DNA repair mechanisms. A variety of photodamage responses are detected shortly after UV exposure and include edema/erythema, sunburn cell formation (apoptotic cells) and immunosuppression. The severity of these responses correlates well with the levels of DNA damage sustained in skin tissue (29,30–33). The biologically harmful effects associated with sun exposure are largely the result of errors in DNA repair, which can lead to oncogenic mutations. The relationship of DNA repair mechanisms to UV pathogenesis was first observed in patients with the autosomal recessive disease xeroderma pigmentosum (XP). This syndrome is the result of a genetic defect in at least one of three critical enzymes involved in nucleotide excision repair, and consequently results in the accumulation of UV-induced DNA mutations and the development of skin cancer within the first decade of life (34). Therefore, it is not surprising that UV-induced DNA damage acts as an important trigger of photodamage responses, and is necessary (but not sufficient) for tolerance induction.
DNA Damage is a Key Trigger for UV-induced Tolerance
The hypothesis that DNA damage serves as a trigger for immune suppression was supported in a series of experiments using topical liposomal delivery of enzymes that either damaged DNA or facilitated repair of UV photoproducts. Liposomes containing the bacterial restriction enzyme, HindIII, induced the expression of suppressive cytokines (e.g. interleukin [IL]-10, tumor necrosis factor-α [TNF-α]) from epidermal cells and inhibited CHS responses (35,36), while application of liposomes with DNA repair enzymes (T4 endonuclease V or photolyase, which efficiently repairs pyrimidine dimers upon photo-reactivation) inhibited local immunosuppression (37) and blocked induction of Treg cells by UV-treated cutaneous antigen-presenting cells (APCs) in vitro (37,38). Interestingly, experiments using transgenic mice that selectively express photolyase in the basal cell layer demonstrated that the trigger for immune suppression does not require basal cell DNA damage (39). These data suggest that DNA damage sustained by suprabasal epidermal cells and/or LCs are important triggers, and this will be discussed further in a later section. It will be interesting to see what the impact of LC-targeted photolyase expression will be on UV tolerance induction.
There is a growing interest in studying Treg cell biology, because this subset of cells plays a prominent role in immune pathologies and cancer development and has potential for therapeutic applications. Therefore, a greater number of immunologists are entering the field of photoimmunology to study Tregs in UV-induced immune suppression models. In addition, more tools and reagents are now available to study LC biology, and UVC is being used as a specific ablator of these cells in mice (40,41). Therefore, it is helpful to review some landmark studies that guide research in cutaneous photobiology today.
In the 1980s, a study performed by L. Berrens and his colleagues established the transmission efficiency of UVR through human skin. They measured the capacity of selected UV bandwidths to penetrate through samples of full thickness human skin in an ex vivo system (42). Figure 1 illustrates that each spectra has a characteristic limit of efficiency in penetrating the epidermal and dermal layers of human and murine skin. Figure 1b presents the tabular data from this study in graph form and illustrates how efficiently the shorter, more mutagenic wavelengths, are absorbed by the SC and cellular layers of human skin. Similar values for UVB transmission have been calculated for mouse skin (43). UVC exposure has become an established technique for eliminating LCs in mouse models (40,41,44,45). This approach is feasible due to the mouse’s thin epidermis. Assuming that the physical properties of energy transmission for any incident wavelength are similar for human and murine skin, approximately 3% of UVC energy may reach basal cells (Fig. 1a). Consequently, UVC irradiation requires long exposure times (>30 min) to effectively deplete LCs from the epidermis (44).
The most widely recognized UV-induced mutations arise from errors in repairing cyclobutane pyrimidine dimers (CPDs). These are prominently formed by DNA absorption of UVB, and to a lesser extent by absorption of UVA radiation. Approximately 10% of the damaged DNA is in the form of pyrimidine (6-4) pyrimidone photoproducts (6-4PPs), which are efficiently removed in both human and rodent cells (46). Enhanced repair of these photoproducts in transgenic mice expressing a plant enzyme, 6-4PP photolyase, does not protect against UV-induced tumor formation. In contrast, expression of an enzyme that enhances CPD repair, the marsupial CPD photolyase, does protect (47). The DNA absorption maximum (260 nm) lies in the UVC spectrum, and therefore UVC is most harmful to genetic integrity. Fortunately, the stratospheric ozone layer protects us from these potent mutagenic rays. However, with the looming concern regarding future ozone depletion, investigators have examined the entire UV spectra, including UVC, to characterize the physical behavior and biologic effects on skin tissues and systemic responses in both humans and animal models (e.g. mice).
DNA Lesions Induced by UVC, UVB or UVA Spectra
UVR-induced CPDs that form at cytosine (C) molecules are highly mutagenic because of error-prone repair that results in thymine (T) substitutions. Up to 60% of the CPDs formed occur at cytosine sequences (48). These photolesions are responsible for the high proportion of p53 mutations present in SCC (49). T-T lesions are proposed to play a role in generating N-ras mutations found in rodent UV tumors (50) as well as in human melanomas (51).
UVA-induced photodamage effects have principally been attributed to the generation of reactive oxygen species (ROS) and the secondary consequences of lipid peroxidation, protein oxidation and DNA oxidation, particularly of guanosine nucleotides to form the “UVA signature” adduct, 8-oxo-7,8-dihydro-2-deoxyguanosine (8-oxo-dG) (52,53). However, it is well established that UVA also generates “UVB signature” CPDs in cultured cells (54,55) and human skin in vivo (56,57). Figure 1c presents data from Sutherland’s group, who were the first to quantify the absolute frequency of CPDs formed in skin by different UV bandwidths (56). The cutaneous DNA damage action spectrum they defined indicated that the peak damage bandwidth occurred not in the UVC spectrum, but in the UVB spectrum at 300 nm. Further, the UVA spectrum also generated significant levels of CPDs but at efficiencies reduced by 2 to 4 orders of magnitude. Indeed, the shape of the CPD action spectrum defined by Sutherland and colleagues describes the frequency of CPDs detected by others (58) as well as the action spectrum for erythema responses by skin Type I/II individuals (29). Furthermore, recent work from Douki and colleagues quantified all possible DNA photoproduct species formed by UVB and UVA irradiation using sensitive mass-spec techniques, and found the yield of CPDs generated by UVA irradiation was substantially greater than the ROS-associated photoproduct 8-oxo-dG, indicating that nuclei are relatively protected from oxidative damage (55,57). Interestingly, T-T dimers formed the majority of UVA-induced CPD derivatives (85%) compared to CPD formed by UVB (40%). While T-T lesions tend to be less mutagenic than T-C and C-T dimers, they may have importance in the development of melanoma (51). Considering that UVA makes up ∼95% of terrestrial UV and can penetrate skin more deeply than UVB, UVA may be most relevant in targeting epidermal and melanocyte stem cells that reside in hair follicle structures (59,60). In addition, UVA-induced repair mechanisms may be less efficient, compared to UVB-induced repair (57). Therefore, UVA radiation may be responsible for generating more oncogenic DNA lesions than previously appreciated (61,62).
The Cellular Photodamage Response
The photodamage response has evolved to protect an organism against genotoxic events that can lead to cancer and immune reactivity to “altered-self” proteins, while promoting recovery and growth of healthy cells for tissue repair. However, protection afforded by UV-induced immunosuppression can lead to increased tumor formation (63). And, while apoptosis is the principal mechanism for suppressing cutaneous tumor formation, global elimination of all UV-damaged cells would endanger the critical barrier functions of the skin. Therefore, a cellular decision-making process permits repairing cells with low levels of damage to maintain skin barrier function at the cost of retaining some cells with a low level of mutations (49,64). The molecular mechanisms involved in the cellular decision-making process are complex and to some extent, cell type specific (65). Indeed, substantial differences are seen between keratinocytes and fibroblasts with respect to their susceptibility to apoptosis and the DNA repair processes (reviewed in D’errico et al. ).
As “guardian of the genome” the tumor suppressor protein p53 plays a central role in the photodamage response by judging the cellular health “status” and determining the cell’s fate. These signals are perceived by p53 through site-specific posttranslational modifications (phosphorylation and acetylation) that stabilize and activate its transcription factor function. The chief prosurvival activity identified for p53 is activating cell cycle arrest at the G1/S checkpoint. This arrest provides time to repair damaged DNA before progressing through the cell cycle. The enzyme that specifically senses CPD formation, ataxia–telangiectasia, mutated (ATM) RAD3-related kinase (ATR), directly activates p53 by phosphorylation of serine-15 and downstream ATR activated kinase, checkpoint kinase-1 (Chk1), phosphorylates p53 on serine-20 (Fig. 2). This promotes p53-dependent transcription of cell cycle arrest proteins (p21WAF1/CIP1, a cyclin-dependent kinase inhibitor) and DNA repair enzymes (XP-A, -B and -C) and antiapoptotic genes (survivin, serpins) (66,67). However, if too much DNA damage is sustained then apoptosis-promoting genes are targeted by p53 and their encoded products activated to induce apoptosis (reviewed in Helton and Chen ).
In addition to the well-known transcription-dependent functions of p53, extranuclear, transcription-independent functions have been identified. A fraction of p53 can accumulate in mitochondria within cells during apoptosis, but not during p53-independent apoptosis or during cell cycle arrest (69). The translocation of p53 to mitochondria was shown to induce apoptosis in a transfection model system. A transcription-deficient mutant of the p53 gene was fused to a mitochondrial targeting sequence, and its expression readily induced apoptosis in p53 negative cells, without external damage signals (70). Other studies have demonstrated that p53 interacts with Bcl-2 family member proteins. It directly binds to and inhibits antiapoptotic Bcl-2 and Bcl-XL (71), and can trigger oligomerization of proapoptotic Bax and Bak that is required to disrupt the mitochondrial membrane (72). Thus, p53 behaves similar to the BH3-only subset of proteins (such as Bad, Bik or Noxa) termed “enablers” (73) or antideath protein “inactivators” (74). Enabler proteins release BH3-only “activators” (such as Bid, Bim or Puma) from sequestration by apoptosis inhibitors (such as Bcl-2, Bcl-XL or Mcl-1) to disrupt mitochondrial membranes directly (tBid-cardiolipin interaction) or indirectly through Bak or Bax activation. p53’s BH3-like activity maps to its N-terminal proline-rich domain as well as its DNA-binding domain, where numerous mutations related to oncogenesis have been identified (75). Thus, p53 prosurvival functions are nuclear and transcription dependent, while propapoptotic functions can be nuclear (energy-dependent transcription) or cytosolic (energy independent). A proposed schema presenting these concepts is shown in Fig. 2. How p53 is regulated during cellular recovery, with respect to transcriptional program choices or cellular location and apoptotic activity is of great interest, and under intense investigation (76,77).
Death Receptor Activation by UVR
When the amount of DNA damage is more than the cell can efficiently repair, then p53-dependent mechanisms of apoptosis will be initiated, or at higher UV doses, p53-independent mechanisms of apoptosis will be activated. Under such conditions, high levels of cellular stress will activate apoptotic programs. Death receptors, such as Fas or TNF-RI, have been shown to aggregate in a ligand-independent manner, because of UV-induced membrane alterations and protein cross-linking by ROS-dependent mechanisms (78–80). Furthermore, it is well established that FasL, TNF-α and their receptors are upregulated in response to UVR, and that mice deficient in these genes or their receptors (Fas or TNFR p55) display resistance to sunburn cell formation, as well as UV-induced immune suppression (81, reviewed in Guzman et al. ). In addition, a role for TRAIL in UV-induced apoptosis has been characterized in cultures of human keratinocytes (83). These TNF family ligands and receptors are regulated at many levels, and alterations in these mechanisms have been identified in basal and SCC as well as melanomas (82).
BH3-only Bcl-2 Family Members Required for UV-induced Apoptosis
The Bcl-2 family members regulate the intrinsic mitochondrial pathway of apoptosis and are divided into antiapoptotic and proapoptotic proteins. They dynamically pair up by interacting with Bcl-2 homology domains that are highly conserved and can be divided further into four sequence motifs. Molecules containing just the third motif are called the “BH3-only” proteins and they are all proapoptotic (reviewed in Ranger et al.  and Danial and Korsmeyer ). It was initially proposed that each BH3-only protein selectively responded to different intrinsic and extrinsic death signals, such as cytokine deprivation (Bim), death receptor activation (Bid) or DNA damage (Puma or Noxa); however, it is evident that selectivity is not stringent and is cell specific. For example, Bim, Bmf and Bid have also been reported to respond to DNA damage inducing agents (86–88). A more recent understanding regarding the hierarchy of interaction affinities between Bcl-2 family members has provided an additional layer of complexity regarding their death-related functions and has established a new classification system (reviewed in Galonek and Hardwick ). This will be highlighted in a later section.
The proapoptotic BH3-only protein Bid is present at high levels in human epithelial cells (89) and is targeted by p53 for transcriptional upregulation in UV-exposed skin (67). Bid is potently activated through cleavage at specific sites by caspase-8 (which is activated through death receptor aggregation) forming truncated (t)Bid, revealing a myristolation modification site, that when modified, increases the efficiency of mitochondrial targeting 300-fold (reviewed in Yin  and diagrammed in Fig. 2). In addition, Bid can be cleaved at a unique site by a JNK-dependent protease generating the Bid isoform jBid (91). Further, jBid appears to permit selective release of mitochondrial Smac and Diablo (but not cytochrome c) and this activity was required for TNF-α-induced caspase 8 activation and apoptosis in NFκB-deficient cells (NFκB activation inhibits TNF-R-mediated death signals). As UV-induced keratinocyte sunburn cell formation depends on TNF-α or TNFR-I expression, it is tempting to speculate that the jBid pathway may be operating, but further experiments are needed to establish this.
A role for Bid in the photodamage response has been observed in recent experiments performed in our laboratory. We have demonstrated that Bid-deficient LCs are highly resistant to apoptosis induced by T cells (92) and UVB exposure (93). We observe that Bid-deficient mice exhibit profound inhibition of UVB-induced sunburn cell formation in vivo (94) and in cultured primary (untransformed) fibroblasts in vitro (H. K. Kim, C. Martin, S. Katiyar, C. Elmets, M. Athar and L. Timares, unpublished). Further, Bid-deficient mice are highly resistant to UV-induced local and systemic immune suppression (, S. Pradhan, H. K. Kim, C. Thrash, S. Mantena, S. Katiyar, C. Elmets and L. Timares, unpublished). Collectively, our results support the notion that Bid-dependent processes have an important impact on photodamage responses by epidermal and dermal cells. Our data are in line with those of Kamer et al. (87) showing that Bid-deficient murine embryonic fibroblasts were resistant to UVC (20 J m−2)-induced apoptosis. Interestingly, a novel prosurvival role has been characterized for a nuclear isoform of Bid, which was found to mediate S or S/G2 phase cell cycle arrest in response to some forms of DNA damage (Fig. 2) (88). However, these findings are controversial, as studies by Zinkel et al. (88) and more recently by Kaufmann et al. (95) do not support a role for Bid in photodamage responses. Clearly, more work must be done to clarify the issues regarding these discrepancies.
A recent report provides evidence that another BH3-only protein, namely Noxa, is required to activate UV-induced apoptosis. These studies show that E1a/ras-transformed fibroblasts or epidermal tissue derived from Noxa-deficient mice were partially resistant to UVC-induced apoptosis, but this level of protection was not observed in cells or tissues from Puma- or Bim-deficient mice (96). In the same study, the authors observed that Bcl-2 overexpression provided complete protection in Noxa- or p53-deficient, but only partial protection in Puma-deficient transformed fibroblasts against UV-induced apoptosis (up to 50 J m−2 UVB). These findings suggest that other downstream proapoptotic activators contribute to the apoptotic mechanism. This is consistent with Noxa’s classification as an “enabler” BH-3-only protein, which functions to release BH3-only “activator” proteins, Bid, Bim or Puma, to directly trigger the mitochondrial apoptotic process (Fig. 2). Another “enabler” BH-3 only protein, Bmf, has been shown to translocate from sequestering scaffold proteins to associate with Bcl-2 in response to UVR, and may play a role in certain settings (86). As Puma, Bim and Bid comprise all the known “activator” BH3-only proteins to date, these data suggest that Bid may be an important activator of UV-induced apoptosis, as our data support, or that additional BH3-only activator proteins have yet to be discovered.
Interestingly, Naik et al. observed no role for death receptor activation by their fibroblast lines in response to UVR, as transfection of Noxa- or p53-deficient transformed fibroblasts with genes encoding inhibitors of death receptor activation did not inhibit apoptosis. However, as previously discussed, it is well established that FasL and TNF-α are upregulated in response to UVR, and that mice deficient in these genes or death receptors (Fas or TNFR p55) display resistance to sunburn cell formation (81). Therefore, the relative contribution of death receptor activation to UV-induced apoptosis can differ substantially between in vivo and in vitro systems.
JNK-dependent UV-induced Apoptosis
The c-Jun-NH2-terminal kinases (JNKs) are a subgroup of mitogen-activated protein kinases. The JNK signaling pathway is activated by environmental stresses, including UVR damage associated with ROS formation. JNK activation has been shown to be critical for UV-induced apoptosis of murine fibroblasts in a number of studies. Tournier et al. performed a systematic evaluation of the contribution of individual members of the Jnk genes. Murine embryonic fibroblasts deficient in both Jnk1 and Jnk2 genes exhibited compete resistance against apoptosis induced by UVC (60 J m−2) exposure, while single gene deficiencies provided partial protection (97). The authors demonstrated that UV-induced Bid cleavage, mitochondrial cytochrome c release and caspase 3 activation were abrogated in Jnk-deficient cells. Further, the pan-caspase inhibitor zVAD, but not the caspase 8-specific inhibitor zIETD, inhibited UV-induced apoptosis, but not UV-induced Bid cleavage in fibroblast cells, suggesting that a JNK-dependent process induced Bid activation. Indeed, JNK-dependent cleavage of Bid has been observed at a novel site, generating ‘‘jBid’’ with properties that induced selective release of Smac and Diablo, but not cytochrome c from mitochondria, to initiate TNF-α-induced apoptosis in susceptible cell lines (91). The nature of Bid cleavage product that is generated by UVR, particularly in keratinocytes, will need further characterization to confirm a role for jBid. JNK has been shown to phosphorylate other Bcl-2 family members in vitro; however the relevance of this activity to UV-induced apoptosis in vivo needs further investigation (reviewed in Weston and Davis ).
JNK-dependent apoptosis is activated in response to UV-induced CPDs that form in transcribed genes and is therefore associated with transcription-coupled repair (TCR) (99). This was established in experiments employing primary human fibroblasts from XP and Cockayne syndrome patients that are deficient in TCR and global genome repair (GGR), respectively. They observed that protracted activation of JNK did not occur in cells deficient in GGR, and that apoptosis in TCR-deficient cells occurred at much lower UV doses and was dependent on JNK activation but not on p53. Taken together, these studies suggest that in noncycling differentiated cells, like suprabasal epidermal keratinocytes or LCs (100), the JNK pathway may play a prominent role in the photodamage response.
Langerhans Cell Function—A Current Mystery
“Planned programmed cell death” (101) is an essential part of keratinocyte differentiation and formation of the SC. It takes place in the outermost layer of nondividing cells, away from epidermal APCs (i.e. LCs). In contrast, UVR induces damage and apoptosis in cells throughout the epidermis, including the germinative basal layer (102). Resting LCs reside just above the germinative layer and are well positioned to act as sentinels of skin damage. Whether they can distinguish between transient formations of neo-self antigens from stable expression of neo-antigens that mark an oncogenic process is not known. Depending on the level of damage sustained by UVR exposure, LCs are likely to be important contributors in defining the immune response to skin-associated antigens. It is assumed that, because LCs are depleted from the epidermis after UVR treatment, they undergo apoptosis in situ (103). However, detecting apoptotic LC in vivo has been elusive. It has also been reported that LC migratory capacity is impaired and they accumulate in the dermis (104). Interestingly, we have observed that LCs still remaining in the epidermis 5–7 days after UV treatment display apoptotic morphology ( and L. Timares, unpublished). In addition, Langerin+ CPD+ LCs can be detected in dLN 3 days after chronic (4-day) UVB exposure, which is consistent with what others have reported ([105,106] and L. Timares, unpublished). The ability to detect CPD-positive LCs in draining LNs indicates that at least a portion of LCs successfully reach their destination. And the delay in LC apoptosis would provide the time necessary to reach the dLN. Kissenpfennig et al. established that LCs require 3–5 days, while dermal DC (dDC) require only 2 days to reach the dLN after hapten painting (107). An interesting observation made by these authors was that LCs and dDCs colonized distinct areas of dLN. dDC colonized the outer paracortex near the B cell zone, while LCs trickled into the deeper inner cortex in the T cell zone. These distinct migratory behaviors suggest that LCs and dDCs are dedicated to very different specialized functions that have yet to be understood. In fact, the established, well-characterized functions of LCs have recently been challenged (reviewed in Refs. [108–110]).
LCs are considered to be the primary APC of the skin. They have been studied extensively and their behavior has defined the paradigm that describes how naive CD4 and CD8 T cells are instructed to initiate immune responses to peripheral antigens (111,112). However, the recent findings from three independent groups, employing similar transgenic mouse models that selectively ablate LCs, have called into question (albeit, not unanimously) how well the paradigm applies to LCs’ true function (113–115). When LCs are ablated from birth, augmented CHS responses result and this suggests that LCs normally downregulate immune responses which may be important for maintaining tolerance. However, when LCs are ablated in adult animals just prior to hapten sensitization this either has no effect (114) or results in the partial inhibition of CHS responses (115). Additionally, the ability of LCs to directly cross-prime and activate CD8 T cell responses has also been challenged. Results favor a specialized role for lymph node (LN) resident CD8α+ DCs in the direct activation of CD8 T cells for tolerogenic (116) or immunogenic responses (117). But studies by others come to opposite conclusions and indicate that LCs are not tolerogenic in the steady state (118), and they can cross-present protein antigens (119) and virally encoded products (120) to CD8 T cells. It is likely that LCs may perform these functions or not, depending on the antigen and the context of environmental signals. An intriguing view is that the phenotypic markers that define a particular DC subset may change depending on environmental cues (reviewed in Larrengina and Falo ). For example, the definitive LC marker Langerin (CD207) also identifies a minor subset of LN-resident CD8α DC when intracellular staining is used (121), and LCs can upregulate CD8α expression under certain conditions (122).
DC subsets participate in a division of labor with respect to the set of foreign or self-antigens they can recognize, and how they will be presented to the immune system. This is indicated by the unique constellation of receptors that identify each subset in their immature state, including antigen capture receptors (C-type lectin receptors: mannose-R CD206, Langerin CD207, DC-SIGN CD209 and dectin molecules) (123), β-integrins (CD11b, CD11c) (124), toll-like receptors (125) and CD1 molecules (CD1a on LCs, CD1b on dDCs and CD1d on dDCs and MØs) (reviewed in Larrengina and Falo ). At this time, a consensus view of LCs’ biologically relevant function has yet to come. With new transgenic animal models, GFP-tagged LCs can be followed in vivo, and the impact of their activity tested in conditional knock out experiments. Thus, for the first time, the unique behavior of LCs may be sorted out from other cutaneous DC subsets. There are differences in the three existing models, and each provides different answers regarding the importance of LC function. It will be of great interest to use these models to study UV-induced immune suppression and to assess the role of LCs in immune surveillance against tumor development.
UV-induced RANK Signaling and LC-dependent Treg Cell Generation
Recent experiments provide new evidence that LCs are critical for the generation of Treg cells. The effectors that mediate UV-induced suppression of CHS responses were identified by Elmets et al. (13) as T cells which were able to transfer hapten-specific tolerance to naive mice. A series of investigations have shown that they display classic Treg cell markers CD4+ CD25+ (FoxP3 expression has not yet been determined). Further, they secrete high levels of IL-10 and express a ligand for dectin-2 receptors on LCs, suggesting that they represent a specialized subset of skin-specific Tr1-like regulatory cells (, reviewed in Beissert et al. ). Recently, Loser et al. (128) demonstrated that the capacity to induce Treg cells was dependent on receptor activator of NF-κB (RANK) activation on epidermal LCs by interaction with its ligand (RANKL; also known as CD254, OPGL and TRANCE) on keratinocytes. Interestingly, RANKL expression could be induced after UVR or was associated with cutaneous diseases such as psoriasis (128). In other previous studies, this receptor–ligand interaction was shown to be critical for generating Treg cells that can suppress diabetes (129). A well-characterized consequence of RANK activation on DCs is an increase in their survival due to the upregulation of Bcl-XL expression (130–132). This mechanism may be operating in UV-damaged LCs to promote their recovery and delay their own demise until after they reach the dLNs. Because RANK-activated LCs have demonstrated potent Treg cell activation in vitro (128), it is possible that UV-damaged RANK-activated LC may similarly activate Treg cells directly, once they reach the dLN (Fig. 3, step 7). Alternatively, they may indirectly promote Treg generation by undergoing apoptosis to distribute peripheral “altered self” antigens to resident CD8α DC which have a known role in tolerance induction (Fig. 3, step 8) (116,133). In the report by Loser et al., RANKL expression was upregulated in murine epidermis after UV exposure. This may be due to UV-generated vitamin D, an activator of human RANKL expression, known to be an important regulator of myeloid osteoclast activity in bone remodeling (134,135). Vitamin D may also act directly on keratinocytes and LCs to promote recovery and tolerogenic activities (136–138). The cutaneous RANK-RANKL system may be analogous to the mucosal system, where RANKL was shown to specifically upregulate IL-10 mRNA expression in mucosal but not splenic DCs expressing similar levels of surface RANK (139). These new findings support the hypothesis that LCs are specialized in their response to UV-induced RANK activation and play an important role in generating antigen-specific Treg cells that mediate UV-induced tolerance. The potential to block immune suppression or to reverse established suppression with RANK-RANKL inhibitors will make this an exciting area of research for developing new approaches to treat cutaneous diseases and cancer.
Regulators of Apoptosis and their Impact on Immune Suppression
Multiple events are activated in response to photodamage, yet the key triggers and mediators of immunosuppression are not completely understood. A schematic of events that have been shown to impact UV-induced tolerance and LCs’ proposed role in this process is depicted in Fig. 3. Many of the key elements that contribute to tolerance induction have been identified in studies using mouse model systems that ablate or overexpress particular genes to modulate the response. It is remarkable that the loss of one component in this complex response can significantly impact the development of suppression. This indicates that many of these cellular and molecular mechanisms are nonredundant and need to act in concert to achieve suppression. Many of these mediators and mechanisms have been well characterized and extensively discussed in recent reviews (22,58,140–142).
It is well established that apoptotic bodies provide strong tolerogenic signals to phagocytic immature DCs (, reviewed in Refs. [144,145]). The engulfment of apoptotic bodies by LCs, dDCs and macrophages (MØs) depend on overlapping as well as unique sets of “self”-recognizing receptors involved in scavenging, and likely transmit tolerogenic signals when engaged. Apoptotic vesicles that derive from basal or suprabasal layers will contain common, as well as differentiation-specific components, and these differences may define their tolerogenic capacity based on the phagocytic receptors they engage. The complement receptor 3 (CR3; CD11b-CD18) is an example of one such phagocytic receptor present on DC that signals tolerogenic responses (146). Many other phagocytic receptors are involved in apoptotic body engulfment, but how ligation of these receptors signal changes in the cellular program is still not known and under intense investigation.
Table 1 provides a list of mouse model systems that have been useful in revealing a relationship between apoptosis regulators and cytokines with immune suppression. In general, when data from genetically altered mice are available for all three manifestations of photodamage (apoptosis, immune suppression and susceptibility to skin cancer) a strong correlation is observed, supporting the hypothesis that DNA damage induces apoptosis which triggers immune suppression that permits tumor outgrowth. However, the epidermal cell subsets that are key to each step in this process have not been fully defined.
Table 1. Mouse models of altered apoptosis related genes and their impact on UV-induced apoptosis, immunosuppression and cancer.
TNF = tumor necrosis factor; LC = Langerhans cell; NKT = natural killer T cell; UCA = urocanic acid; CPD = cyclobutane pyrimidine dimer. –, Designates not reported. *Gene altered by targeted disruption or tissue specific overexpression by the indicated promoter (pr). †Mode of altered gene expression in the mouse model: KO, genetic knockout; Tg, transgenic, neutralizing antibody. ‡Note that mouse strains differ in susceptibility to UV-induced suppression (167,168). §Bold font designates discordance with other parameters. ||Chemical carcinogenesis protocol. ¶Parentheses designate a predicted outcome based on related reports (see comments).
First paper to correlate inhibition of apoptosis to increased cancer
One report listed in Table 1 has provided data that initially appear discordant with the relationship between apoptosis, tumor development and immune suppression. In this transgenic animal model the K14 promoter is linked to the photolyase gene (39). Thus, CPDs are selectively removed (upon photoreactivation) and enhance DNA repair only in basal layer cells. Not surprisingly, apoptosis in the basal layer and skin cancer development was sharply inhibited in these mice. Unexpectedly however, the induction of immunosuppression was still observed. These data are consistent with the following scenarios: (1) Photodamage sustained by suprabasal cells provides all the signals and mediators necessary to induce tolerance while enhanced DNA repair of basal cells inhibits tumor initiation or (2) LCs are altered by direct UV-damage and this triggers their migration to dLN to initiate tolerogenic responses. The first of these scenarios seems quite plausible, but additional observations reported in this study suggest that typical markers of photodamage responses, such as hyperplasia with infiltrating dermal leukocytes, were not observed after chronic UV treatment. Mediators that may act independently of cellular photodamage, such as trans-UCA, may play a role. Alternatively, directly damaged LCs may be important in this process. The independent findings from Loser et al. and Schwarz et al. support the latter, as they have also shown a direct correlation with the presence of cutaneous LCs (that are CPD+) either to the level of Treg cells or immunosuppression.
There have been significant advances in photobiology and photoimmunology that have provided new insight into many areas that affect cutaneous photodamage responses. An increased respect for the mutagenic properties of UVA will certainly impact the priority given to its future research. The elucidation of mechanisms that govern the integration of cellular signals that determine cell fate will continue to be a growing challenge. Recent investigations indicate that an ever-growing number of molecules will become a part of this process, and may involve complex dynamic interactions. And it is increasingly apparent that dual functions can be ascribed to these regulators so they can participate in either cell survival or apoptosis mechanisms. This may be an efficient way for the cell to quickly respond to cell status signals. The nature of how a cell undergoes apoptosis is communicated to the immune system. In addition to the environmental cues, the remnants of apoptotic cells provide innate signals that are interpreted by phagocytic subsets of APCs in specific ways such that autoimmunity or tolerance occurs. The status of the LC has undergone a tremendous paradigm shift, and is currently changing from the prototypic immunogenic DC to the prototypic tolerogenic DC. The sophistication of new animal models can now specifically address the function of LCs in many immunologic scenarios. Advances in understanding suppressor T cell biology have been especially productive over the years due to the well-established 30-year-old model of UV-induced immune tolerance. It is apparent that the insight gleaned from studying mechanisms of cutaneous photodamage responses that lead to immune suppression and carcinogenesis carries over to many disciplines. Therefore, it is hoped that the development of new therapeutic approaches, based on advances in understanding photobiology and cutaneous immunology, can be applied to treat an even broader variety of diseases.
Acknowledgements— The authors thank Dr. Mohammad Athar for his helpful discussions and critical reading of this manuscript. This work has been supported by grants from the American Cancer Society, Dermatology Foundation, Charlotte Geyer Foundation, NIH grants RO1-AI50150, RO1-CA86172, P30-AR050948, USAMRAA grant W81XWH-0510296 and a VA Merit Award.