Ultraviolet Radiation Signaling through TLR4/MyD88 Constrains DNA Repair and Plays a Role in Cutaneous Immunosuppression

Human skin is ubiquitously exposed to UV radiation (UVR). The longer wavelengths of UVA are able to penetrate down through the epidermis to the dermis and associated connective tissue, whereas the shorter UVB wavelengths are absorbed by epidermal cells, primarily keratinocytes (KCs) (1). UVR has both positive and negative effects on human health. It is responsible for the biosynthesis of vitamin D₃, can stimulate the production of photoprotective melanin (2–5), and is used therapeutically to treat inflammatory skin diseases, such as psoriasis, vitiligo, localized scleroderma, and atopic dermatitis (6–8). At the same time, UVR has many negative effects, most notably that it is directly absorbed by cellular DNA, resulting in cyclobutane pyrimidine dimer (CBPD) mutations that can lead to carcinogenesis of resident epithelial cells. Contributing to the carcinogenic effects of UVR is the fact that it is also a known potent immunosuppressant, rendering the recipient less able to stimulate an immune response against cancerous cells that may otherwise be destroyed (9, 10).

CBPDs are the main type of DNA damage associated mainly with UVB irradiation. These dimers require the nucleotide excision repair (NER) complex to be excised and corrected; once this process occurs, the integrity of the affected DNA is maintained (11). Another type of DNA damage–causing radiation, gamma irradiation, results in double-strand breaks that cannot be repaired using a complementary template; it is associated with mutagenesis and chromosome abnormalities that often lead to cell death. Repair of both UV and gamma irradiation damage was shown to be altered in the presence of endosomal TLR agonists (12, 13), implying that both the NER and double-strand breaks repair complex may be affected by the status of TLR activation. In this study, we focus on DNA damage caused by UVR in the form of CBPDs, which requires NER machinery to be excised and repaired.

The NER pathway is responsible for excising and repairing CBPD mutations after UVR, and it is a necessary mechanism that is used to repair a wide range of DNA lesions (14). NER can be categorized into two subpathways: global genome repair (GGR) and transcription-coupled repair (15, 16). The removal of UV-induced CBPDs falls into the category of GGR, whereas transcription-coupled repair ensures that an intact and accurate genome is maintained and passed on during cell division (17). GGR begins when a DNA damage–sensing molecule, such as PARP or DDB, recognizes a DNA lesion. Xeroderma pigmentosum (XP) family proteins are then recruited to the damaged bp. These XP proteins begin to repair the lesion by unwinding the surrounding DNA, stabilizing the now ssDNA, and excising the mutated bp. A DNA polymerase and DNA ligase are recruited to fill in the necessary bp that are complementary to the intact template strand (1). The importance of an intact NER pathway is highlighted by XP patients (18); they are deficient in one or more of the XP family proteins and present with multiple basal cell carcinomas and other skin malignancies at an extremely young age (19, 20). Often, XP patients have to be completely protected from exposure to sunlight and other sources of radiation to avoid skin cancers.

Interestingly, it was found that mice deficient in the innate immune signaling pathway TLR4 are resistant to cutaneous UV immunosuppression (21). Expression of TLR2 and TLR4 also were shown to be increased in epidermal cells following UVR, and an observed increase in immune signaling molecules, such as MAPK and NF-κB, is dependent on this TLR expression (22). Previously, we found that the TLR4/MyD88 signaling pathway is necessary for UV-induced apoptosis and that, without this intact pathway, UV-induced cell death is skewed to an inflammatory, necroptotic cell death (23). We hypothesized that, in addition to TLR4-deficient mice, MyD88-deficient mice are resistant to UV-induced immunosuppression and that this phenotype is a result of increased resolution of UV-induced DNA damage to skin-resident APCs.

Mouse experiments were approved by the University of Maryland Institutional Animal Care and Use Committee. Fourteen- to sixteen-week-old female mice were used in all experiments. Wild-type (WT) C57BL/6 mice (The Jackson Laboratory; strain C57BL/6J; stock no. 000664), MyD88^−/− mice on a C57BL/6 background (The Jackson Laboratory; strain B6.129P2(SJL)-Myd88^tm1.1Defr/J; stock no. 009088), and TLR4^−/− mice on a C57BL/6 background (The Jackson Laboratory; strain B6.B10ScN-Tlr4^lps-del/JthJ; stock no. 007227) were housed in the Baltimore Veterans Affairs Medical Center animal facility and used in the experiments described. Where indicated, hair was removed from the abdomen of mice using electric clippers. When ears were exposed to UVR, mice were anesthetized by i.p. injection of a ketamine (80 mg/kg)/xylazine (10 mg/kg) mixture, and ears were taped down to a surface so only the dorsal side was exposed to UVR. All in vivo experiments were repeated three times to assure reproducibility.

Groups of four mice at a time or cell culture plates with the media removed were irradiated with a Panasol II Two-Foot Broad Band UVB light source, equipped with eight UVB broad band lamps (National Biologic). We calibrated the UVB dosing with National Biologic measuring devices—the UVB 500C radiometer (measures output in the 290–320 nm range) and the UVA 400C radiometer (measures output in the 320–400 nm range)—to determine the time necessary to deliver the desired doses of UVB. Cell monolayers or animal skin were subjected to UVR at a distance of 12 in from the light source.

The dinitrofluorobenzene (DNFB) contact sensitization model was carried out as previously described (24). Briefly, hair was removed from the abdomen of WT or MyD88^−/− mice with electric clippers, and they were exposed to an immunosuppressive dose of UVR, 70 mJ/cm², for four consecutive days (days −3 through 0). On the last day of UVR, day-0 and day-1 mice were sensitized with 20 μl 0.5% DNFB on the abdomen. On day 5, baseline ear thickness measurements were taken, and immune reaction was elicited by painting 20 μl 2% DNFB onto the ear. Twenty-four hours after elicitation, day-6 ear thickness was measured, and the amount of swelling was determined by subtracting the baseline thickness.

A Mouse IFN-γ ELISA kit was purchased from BioLegend (cat. no. 575309), and samples were run according to the manufacturer’s protocol for cell culture supernatants. DNFB-specific IgG2a serum ELISA was run as previously described (25). Briefly, serum was collected from contact hypersensitivity (CHS) mice 21 d after initial sensitization. ELISA plates (Corning Glass Works) were coated with 10 μg trinitrophenyl-OVA or DNP-OVA (Sigma-Aldrich, St Louis, MO), and the plates were blocked with 0.25% BSA in Tween/PBS. Sera were plated in triplicate to allow for Ab binding. Unbound Ab was washed off, and biotinylated anti-mouse IgG2a (2 μg/ml; BD Pharmingen, San Diego, CA) was added. O-phenylenediamine dihydrochloride (Sigma-Aldrich) was added, according to the manufacturer’s instructions, for color development. All ELISA plates were read on a Bio-Rad benchmark microplate reader (Bio-Rad, Hercules, CA), and OD was read at 450 nm.

Genomic DNA was extracted and purified from mouse peritoneal macrophages (PMs) and human KCs using a DNeasy Kit (QIAGEN, Valencia, CA). CBPDs were measured in duplicate and in a random order using an OxiSelect Cellular UV-Induced DNA Damage ELISA Kit (CPD) (Cell Biolabs), according to the manufacturer’s instructions. Briefly, the extracted DNA was heat denatured and coated onto microtiter plates provided by the manufacturer of the ELISA. UV-irradiated calf thymus DNA was provided by the manufacturer, and it served as a set of internal standards for each assay. The plates were probed for DNA damage using a mAb, followed by a biotinylated secondary reagent and subsequent color development. OD was measured at 450 nm.

PARP activity was measured in histone-coated strip wells using the High Throughput (HT) Colorimetric PARP/Apoptosis ELISA Kit (Trevigen), following the manufacturer’s procedures. This ELISA measures the incorporation of biotinylated poly(ADP-ribose) (PAR) into histone proteins after samples are incubated with anti-PAR Ab and then HRP-conjugated secondary Ab (anti-mouse IgG-HRP). Absorbance was read at 450 nm. PARP activity for each sample was calculated from a standard curve run in duplicates within each experiment.

Whole-genomic DNA was extracted using a DNeasy Blood and Tissue Kit (QIAGEN), following the manufacturer’s protocol. For DNA dot blots, 500 ng heat-denatured DNA was spotted onto a nitrocellulose membrane and allowed to dry at room temperature for 1 h and then baked at 80°C for 20 min. The membrane was probed with an anti-CBPD Ab (Cosmo Bio; cat. no. NMDND001) and developed using secondary Ab and reagents from the WesternBreeze Kit (Invitrogen; cat. no. WB7104), according to the manufacturer’s protocol. Image density quantification was performed using Image-Pro Plus software version 4.5.1.29 (Media Cybernetics). We normalized DNA to the amount spotted onto membrane (500 ng, as measured by NanoDrop ND1000 spectrophotometer) versus the OD, as measured by our image-analysis software.

Formalin-fixed WT and MyD88^−/− mouse ear biopsies were embedded in paraffin, sectioned, and stained with H&E. Ear sections were visualized using a Nikon Eclipse E600 microscope, and the images were documented using the SPOT imaging system (Diagnostic Instruments, Sterling Heights, MI). To harvest mouse epidermal sheets, ears were obtained 24 h after irradiation of one side of the ear with 100 mJ/cm² UVR. The irradiated side was split from the control side and incubated with Dispase (BD Biosciences; cat. no. 354235) at 4°C for 2 h. The epidermis was separated and fixed in ice-cold methanol for 20 min. The epidermal sheets were stained with FITC-conjugated anti-mouse I-A/I-E clone M5/114.15.2 (BioLegend; cat. no. 107605) and mounted onto slides with DAPI-containing mounting medium for fluorescent visualization. Epidermal sheets were visualized using a Zeiss LSM700 microscope.

PMs obtained by thioglycollate administration in mouse abdomen were purified by short plastic adherence in culture media (DMEM supplemented with 20% FCS and 100 μg/ml Pen-Strep), washed three times, and cultured, as described, for experimental use. Primary human KCs were derived, as previously described, from neonatal foreskins (26). Use of human samples was approved by the University of Maryland Medical School Institutional Review Board. KCs were cultured in Epi-Life growth medium supplemented with epidermal growth factor and pituitary extracts (Cascade Biologics, Portland, OR). Lymph node single-cell suspensions were made by passing lymph node tissue through a 70-μm nylon cell strainer; cells were washed twice with PBS and plated at a concentration of 1 million cells/ml RPMI 1640 supplemented with 10% FCS and 100 μg/ml Pen-Strep. For some experiments, PJ-34 (Sigma-Aldrich; cat. no. P4365) or Dynabeads Mouse T-Activator CD3/CD28 (Life Technologies; cat. no. 11456D) was added to cell culture medium at the indicated concentration. All in vitro experiments were repeated three times to assure reproducibility.

Expression of MyD88 was knocked down in a pool of primary KCs from three donors using small interfering RNA (siRNA; QIAGEN; cat. no. SI00300909) transfected into the cells with an Amaxa Human Keratinocyte Nucleofector Kit, following the manufacturer’s instructions. Scrambled siRNA was used as a control (QIAGEN; cat. no. 1022076). RNA extraction, cDNA synthesis kit, and real-time PCR were run according to previously published methods (27, 28). Relative levels of MyD88 mRNA were normalized to 18s mRNA. Quantitative real-time PCR was performed using QPCR SYBR Green Supermix (QIAGEN). The following primers were used in the reactions: MyD88 (RefSeq accession no. NM_002468.4) (reference position: 979); 18s (RefSeq accession no. X03205.1) (reference position: 1447); IFN-γ (RefSeq accession no. NM_008337.3); IL-12p40 (RefSeq accession no. NM_008352.2); IL-23p19 (RefSeq accession no. NM_031252.2); Foxp3 (RefSeq accession no. NM_054039.1); IL-4 (RefSeq accession no. NM_021283.1); IL-10 (RefSeq accession no. NM_010548.1); 18s (RefSeq accession no. X03205); XPA (RefSeq accession no. NM_011728); XPC (RefSeq accession no. NM_009531.2); Xab2 (RefSeq accession no. NM_026653.1); Rad23b (RefSeq accession no. NM_009011.4); Ddb1 (RefSeq accession no. NM_015735); and Ddb2 (RefSeq accession no. NM_028119.4). All of these primers were purchased from SABiosciences (QIAGEN).

Protein concentrations in cell lysates made using RIPA buffer were measured using a DC Protein Assay (Bio-Rad). For Western blots, 10 μg protein/lane was loaded onto a NuPAGE 10% Bis-Tris gel (Life Technologies cat. no. NP0301), electrophoresed, and transferred to nitrocellulose membranes using the Novex mini-gel system (Invitrogen). Membranes were probed with one of the primary Abs—anti–β-actin polyclonal Ab (Cell Signaling; cat. no. 4967), anti-MyD88 mAb (Cell Signaling; cat. no. D80F5), or anti-PARP mAb (Cell Signaling; cat. no. 46D11)—and developed using the WesternBreeze kit (Invitrogen; cat. no. WB7104 or WB7106), according to the manufacturer’s protocol.

Quantitative data were analyzed for statistically significant differences, between groups of n replicates, as described in the figure legends, using GraphPad InStat software (GraphPad, La Jolla, CA). ANOVA analysis was performed for the quantitative data and p < 0.05 was considered significant. Data are mean ± SD from three independent experiments.

To determine how MyD88 induces UV-induced immunosuppression, we studied MyD88^−/− mice using the DNFB CHS model. Twenty-four hours after elicitation, both WT and MyD88^−/− mice that were sensitized with DNFB exhibited a robust ear swelling reaction, whereas mice elicited with DNFB only showed a low level of baseline ear swelling. WT mice that were subjected to UVR before DNFB sensitization showed a significant (p < 0.0001) reduction in ear swelling compared with WT mice that were not exposed to UVR. However, ear swelling in MyD88^−/− mice that were exposed to the same UV doses was not significantly diminished and was significantly (p = 0.0096) greater than the paired UVR WT group of mice (Fig. 1A). H&E-stained ear sections from unirradiated and UV-irradiated MyD88^−/− mice showed similar histologic morphology, with significant edema and cellular infiltrates. Although the ear sections from unirradiated WT mice showed similar histologic signs of an intact immune response, the ear sections from UV-irradiated WT mice showed limited edema and decreased histologic signs of inflammation (Fig. 1B). Thus, after acute exposures to four consecutive doses of UVR, MyD88^−/− mice are resistant to immune suppression compared with WT mice. Further studies will be necessary to determine the level of resistance of MyD88^−/− mice to immune suppression in the context of chronic UVR exposures.

It is known that UVR depletes epidermal Langerhans cells (LCs) by two main mechanisms: by causing cell death and migration to the local lymph node (LLN) (29). To address the possibility that UVR is causing emigration of APCs to LLNs at different rates in WT and MyD88^−/− animals, epidermal sheets were stained for MHC class II, which stains epidermal LCs (30). Twenty-four hours after 100 mJ/cm² UVR, similar levels of APC emigration from the epidermis were observed in WT and MyD88^−/− mice (Fig. 2A, 2B). Examination of LC densities at 96 h after UVR exposures revealed similar levels of depletion in WT and MyD88^−/− mice (data not shown). Single-cell suspensions made from LLNs of the same irradiated mice showed that MyD88^−/− mice are capable of producing significantly more (p = 0.0094) IFN-γ when stimulated with a mitogen compared with WT mice (Fig. 2C). Although WT and MyD88^−/− mice have the same amount of APC emigration, the IFN-γ production capability of LLN cells from MyD88^−/− mice is higher.

To further support this conclusion, serum was obtained from each mouse on day 21 after sensitization, and DNFB-specific IgG2a, an Ab response that is dependent on Th1 lymphocyte–derived IFN-γ (21), was measured. There was significantly (p = 0.0127) less DNFB-specific IgG2a in the serum of UV-irradiated WT mice than in the serum from UV-irradiated MyD88^−/− animals (Fig. 2D). DNFB is known to elicit primarily a Th1 response, which results in the production of IgG2a; as expected, a significant DNFB-specific IgG1 response was not observed (data not shown). However, there was a significant difference with regard to DNFB-specific IgG2a. We could not detect IFN-γ in the serum of WT or MyD88^−/− mice in the steady-state or following sensitization (S) and elicitation (E) or UV+S+E (data not shown).

A fundamental concept of UVR-induced cutaneous immune suppression is the arrival of APCs with DNA damage in the LLN, which is thought to play a role in the generation of regulatory T cells (31). Therefore, we examined whether UVR, skin-derived MyD88^−/− APCs arrived in the LLN with less DNA damage compared with UVR, skin-derived WT APCs. Our initial flow cytometry analyses of APC populations (MHC-II⁺, CD11c⁺), before or 24 h after UVR, were not significantly different in the LLN of WT or MyD88^−/− mice (data not shown). Twenty-four and forty-eight hours after a single exposure to UVR, a significantly lower (p < 0.01) amount of DNA damage was detected in the LLNs of MyD88^−/− mice compared with WT mice, as measured with a quantitative DNA damage ELISA (Fig. 3A). Additionally, the expression of a family of genes associated with sensing DNA damage (32) was elevated in the LLNs of UVR-irradiated WT mice compared with MyD88^−/− mice (Fig. 3B). As a result of fewer DNA-damaged APCs arriving in the LLN after UVB exposure in MyD88^−/− mice, there was a relatively higher level of expression of genes associated with immune competence (IFN-γ, IL-12, and IL-23) and a lower level of genes associated with immune suppression compared with WT mice (Fig. 3C). By demonstrating that there is a functional Th1 skewing environment in the LLN of MyD88^−/− mice, these data provide mechanistic insights as to why CHS is intact in MyD88^−/− mice compared with WT mice.

The reduction in DNA damage in the LLNs of MyD88^−/− mice led us to wonder whether this phenotype also can be observed as increased resolution of CBPDs in UVR skin. To test this hypothesis, serial skin biopsy specimens were taken from the abdomen of UV-irradiated WT and MyD88^−/− mice. Immediately after being exposed to 70 mJ/cm² UVR, both strains of mice showed similar levels of CBPDs in epidermal DNA. However, 24 h after UVR, levels of CBPDs in epidermal DNA from MyD88^−/− mice were significantly (p < 0.05) lower than in WT epidermal DNA (Fig. 4A). These data suggest that the skin of mice without the MyD88 signaling pathway is able to resolve UVR-induced CBPDs more efficiently than is WT mouse skin. A similar phenotype of increased resolution of CBPDs was observed in TLR4^−/− mice (data not shown).

To confirm that this phenotype also can be observed in human skin cells, siRNA knockdown of MyD88 in human primary KCs derived from surgical specimens from three healthy donors was performed. A >80% knockdown of MyD88 was achieved at both the mRNA and protein levels, as measured by RT-PCR at 24 h after transfection and as measured by Western blot at 48 h after transfection, respectively (Fig. 4B). KCs were subjected to UVR 48 h after transfection, DNA was extracted 24 h after UVR, and a DNA damage ELISA was run to evaluate levels of CBPD after increasing doses of UVR. KCs with MyD88 knocked down exhibited significantly less CBPD 24 h after 25 mJ/cm² UVR compared with control transfected KCs (Fig. 4C). MyD88 siRNA-treated human KCs, similar to MyD88^−/− mice, are able to repair UV-induced DNA damage at a faster rate than are paired control-treated KCs. These data suggest that loss of the MyD88-dependent signaling pathway has similar effects on DNA repair of UVR-induced damage in KCs and APCs.

DNA damage recognition is crucial for effective repair and resolution of lesions. Damage-recognition molecules, such as PARP, are cleaved by caspases during apoptosis, a condition when DNA repair is no longer necessary because the cell is dying (31). Consistent with our previous finding that cells of MyD88^−/− and TLR4^−/− mice do not undergo efficient apoptosis after UVR compared with WT mice (23), we found that PARP is cleaved in a time-dependent manner after UVR in WT PMs; however, this cleavage is significantly decreased in MyD88^−/− PMs (Fig. 5A). To determine whether PARP activity is maintained in MyD88^−/− PMs, we used a functional PARP assay (ELISA, see 2Materials and Methods) to compare PARP activity in PMs after increasing doses of UVR. There was maintenance of PARP activity in MyD88^−/− PMs after increasing doses of UVR (Fig. 5B), in contrast to WT PMs, which exhibited a dose-dependent loss of PARP activity after UVR exposure. To investigate whether the DNA damage–recognition molecule PARP is necessary for the increased resolution of dimers observed in MyD88-deficient models, the water-soluble PARP inhibitor, PJ-34, was used in PM culture experiments. In the presence of PJ-34, MyD88^−/− PMs and epidermal cells lost their ability to resolve CBPDs at a higher rate, whereas PJ-34 treatment had no effect on the resolution of CBPDs in WT PMs or epidermal cells (Fig. 5C, 5D). Consistent with the DNA repair data, PJ-34 also interfered with the enhanced survival that was observed in MyD88^−/− PMs, but it had no effect on the survival of WT PMs (data not shown). These data show that PARP is necessary for increased DNA repair in vitro.

In addition to damage recognition by PARP, we investigated whether the observed enhanced resolution of CBPDs is dependent on NER machinery, the predominant repair pathway for CBPDs (33). It was shown that bone marrow–derived cells accumulate CBPDs after UVR similarly to epithelial cells that are routinely physiologically exposed to sunlight (29). Because bone marrow–derived cells respond similarly, we were able to use a lymphoblast cell line from an XP patient who is deficient in XPA, one of the critical NER molecules, in our experiments. A >75% knockdown of MyD88 was achieved, by siRNA transfection, at both the mRNA and protein levels in XP and WT lymphoblasts compared with control transfected lymphoblasts (data not shown). Forty-eight hours after transfection, the cells were irradiated with 50 mJ/cm² UVR, and CBPDs were detected by DNA dot blot immediately following and 24 h after irradiation (Fig. 6A). As predicted, the WT cells with MyD88 knocked down showed an increased resolution of CBPDs 24 h post-UVR; however, the XPA-deficient cells did not have this phenotype of increased repair. These DNA dot blot data were confirmed by a CBPD ELISA (Fig. 6B). These data describe a pathway for DNA repair that is dependent on CBPD recognition by PARP and repair by NER machinery. This repair is activated by UVR in the absence of the apoptosis-initiating TLR4/MyD88 signaling pathway.

Apoptosis initiated by UVR has long been thought of as a protective mechanism that enables the skin to get rid of damaged cells in a way that does not cause potentially damaging inflammation (34). Apoptotic cell fragments cause immune-suppressive reactions in APCs (35, 36), but their role in UVR-induced immune suppression has not been investigated thoroughly. We demonstrated previously that cell death in the skin in vivo, as well as in cultured APCs in vitro, is nonapoptotic; instead, it proceeds by a necroptotic pathway (23). The immunosuppressive effects of UVR have far-reaching effects on the immune system; by allowing cells to die without inflammation, apoptosis caused by signaling through MyD88 may substantially contribute to this impaired immune surveillance. To substantiate this hypothesis we carried out CHS experiments in WT and MyD88^−/− mice and, as hypothesized, found that MyD88^−/− mice are resistant to UV-induced immunosuppression (Fig. 1). These data are similar to the study in which TLR4^−/− mice were demonstrated to be relatively resistant to the immune-suppressive effects of UVB (21), suggesting that UVB immune suppression via TLR4 signaling is mediated by a MyD88-dependent signaling pathway, because the phenotype of MyD88^−/− mice (resistance to UVR immune suppression) is identical to that of TLR4^−/− mice.

Work by Margret Kripke and colleagues in the 1970s found that this UV-induced immunosuppression was a critical step in allowing for the growth of cancerous tumors on the skin (37–39). UVR causes the proliferation of tumor-specific regulatory T cells (31), and it damages the DNA of skin-resident APCs and causes them to migrate out of the skin to the LLN (40). In our model, we found that skin-resident APCs migrated out of the epidermis at a similar rate in WT and MyD88^−/− mice (Fig. 2A); however, when probed for DNA damage, the LLN of MyD88^−/− mice presented with fewer CBPDs (Fig. 3A). Because the CBPDs were a result of direct UVR exposure, our data support the idea that migratory epidermal APCs from WT mice arrive in the LLN with DNA damage, whereas APCs from MyD88^−/− mice arrive in the LLN without DNA damage, because they repair such damage more efficiently. This idea is further supported by our observation that DNA damage–sensing genes are upregulated in the LLN of WT mice but not MyD88^−/− mice (Fig. 3B). The arrival of immune-competent APCs in the LLN of MyD88^−/− mice results in a microenvironment that supports Th1 development, because both IL-12 and IL-23 can further promote the repair of UVR-induced DNA damage (41, 42).

Maintenance of Th1 responses by APCs that reach the LLN with less DNA damage supports the integrity of CHS responses (Fig. 1), which indicates intact skin immune surveillance after an acute UVR exposure. Additionally, when stimulated with anti-CD3/anti-CD28, LLN cells from MyD88^−/− mice were capable of producing significantly more IFN-γ (Fig. 2C). Although IFN-γ production is suppressed in WT mice by UVR, it is augmented slightly in MyD88^−/− mice, which could be a consequence of the previously described inflammatory necroptotic cell death. These data suggest that there is increased DNA repair in MyD88^−/− cells that may allow for the observed resistance to UV immunosuppression. We demonstrated previously that UVR suppresses the systemic immune response after hapten sensitization, in that the IgG2a (IFN-γ–driven) anti-hapten (in this case DNP) response, is suppressed (43). We demonstrate that a lack of MyD88 signaling results in a preserved systemic anti-hapten (DNP) response, because the sera from UVR WT mice exhibited a significantly decreased anti-hapten Ab response, whereas this was relatively intact in MyD88^−/− mice (Fig. 2C).

As discussed previously, TLR4^−/− and MyD88^−/− cells do not undergo a typical apoptotic cell death after UVR; instead, they die by necroptosis (44). This observation supports our findings that MyD88^−/− mice are resistant to UV immunosuppression because the apoptotic fragments released during classical UV-induced cell death are known to contribute to UV-induced systemic immunosuppression. Not only is the irradiated organism systemically affected during UV-induced apoptosis, but at the cellular level, many proteins are routinely cleaved by caspase-3, including the DNA damage–recognition molecule PARP (45), which is considered a biomarker for apoptosis.

In our experimental systems, we found that PARP was not cleaved (i.e., inactivated) after UVR in MyD88^−/− PMs (Fig. 5A). PARP plays a role in the recognition of DNA damage, and it works synergistically with NER enzymes (46–48). Lysates derived from MyD88^−/− (but not WT) PMs maintained functional PARP activity (ADP-ribosylate histones in vitro) (Fig. 5B).

We also found that increased resolution of CBPDs is observed after UVR in both mouse and human cells without a functional MyD88 pathway (Fig. 4). The DNA damage–recognition/repair molecule PARP is cleaved and inactivated during apoptosis, and the TLR4^−/−/MyD88^−/− cells die by necroptosis instead of apoptosis. Therefore, we hypothesized that the decrease in UV-induced DNA damage is due to increased DNA repair. PARP is able to recognize CBPDs and function as a recruiter for the NER machinery, which can come in and repair the DNA lesion (49). It is not surprising that we observed increased resolution of CBPDs in MyD88-deficient cells, which maintain full-length/functional PARP after UVR. If this increased resolution of CBPDs is dependent on full-length (active) PARP, when PARP is inhibited, more CBPDs should be detected in UV-irradiated MyD88-deficient cells. Using PJ-34, a water-soluble PARP inhibitor, we found that, in vitro (Fig. 5C), CBPDs occur in MyD88^−/− cells at levels similar to those observed in WT PMs after UVR. This supports the idea that preservation of PARP activity in UVR cells that lack an intact MyD88 signaling pathway is responsible for maintenance of DNA repair, as well as enhanced cell survival (23).

Using MyD88 knockdown in cells from an XP patient, we find that increased repair of CBPDs in MyD88-deficient cells is also dependent on the presence of functional NER machinery (Fig. 6). This suggests that enhanced resolution of CBPDs after UVR exposure is dependent on intact NER machinery, and the resolution of CBPDs is not an artifact of nonapoptotic cell death. It also implies that intact PARP-1, in the setting of nonapoptotic death, synergizes with DNA repair mechanisms.

This study outlines a pathway that describes a novel property of MyD88, above and beyond inflammation and PAMP recognition. We hypothesize that the death domain of MyD88, with or under UV irradiation, is able to cascade with another death domain–containing molecule, such as FADD or TRADD, which traditionally leads to extrinsic apoptotic cell death. If this is correct, then the UV-activated, TLR4 noncanonical extrinsic apoptotic pathway diverges from the classical TLR4 signaling pathway after MyD88. This would be a critical finding, because it may allow for alteration of the cell death pathway without perturbing canonical TLR4 signaling. Therapeutics that can alter UV-induced cell death without affecting PAMP/DAMP recognition by TLR4 will allow for specific treatment to obtain the desired effect. It is important for the future application of the findings presented in these studies to clearly define the signaling molecules involved in this novel MyD88-dependent pathway, which is the subject of some of our ongoing studies.

These data provide a previously unappreciated, direct link between TLR signaling and DNA repair after UVR. This may begin to change the way that we think about classical TLR signaling, along with the possible conditions that activate and the consequences of activating them. Manipulation of cell death responses by pharmacologic agents may be able to maintain immune integrity by enhancing DNA repair and allowing for the propagation of fewer cancerous mutations.

The authors have no financial conflicts of interest.