UV-induced Immunosuppression in the Balance


  • This paper is dedicated to Professor Margaret L. Kripke on the occasion of her retirement from the University of Texas MD Anderson Cancer Center.

*email: f.r.de_gruijl@lumc.nl (Frank de Gruijl)


Around 1980, experiments with hairless mice showed us that UV-induced actinic keratoses (AK) and ensuing skin carcinomas did not arise independently: the rate of occurrence in one skin area was increased considerably if AKs had already been induced separately in another distant skin area, i.e. a systemic effect. The ground laying work of Margaret Kripke in the 1970s provided a fitting explanation: UV-induced immunosuppression and tolerance toward the UV-induced tumors. From Kripke’s work a new discipline arose: “Photoimmunology.” Enormous strides were made in exploring and expanding the effects from UV carcinogenesis to infectious diseases, and in elucidating the mechanisms involved. Stemming from concerns about a depletion of the ozone layer and the general impact of ambient UV radiation, the groups I worked in and closely collaborated with explored the anticipated adverse effects of UV-induced immunosuppression on healthy individuals. An important turning point was brought about in 1992 when the group of Kevin Cooper reported that immunosuppression could be induced by UV exposure in virtually all human subjects tested, suggesting that this is a normal and sound physiological reaction to UV exposure. This reaction could actually protect us from illicit immune responses against our UV-exposed skin, such as observed in idiopathic polymorphic light eruption. This premise has fruitfully rekindled the research on this common “sun allergy,” affecting to widely varying degrees about one in five Europeans with indoor professions.


Sun exposure has long been known to be a risk factor for skin carcinomas (1,2), and early animal experiments showed the UV radiation to be carcinogenic (3). Some skin rashes, e.g. dubbed “eczema solaris,” were also associated with sun exposures and reported on already around 1800 (4). In parallel, the therapeutic effects of UV radiation were empirically explored in “heliotherapy” with natural sunlight or in “actinotherapy” with artificial UV sources.

Niels Finsen used focused “chemical rays” from carbon arc lamps to cure skin tuberculosis (5) (filtering out wavelengths below 340 nm through glass lenses [6]). He attributed the cure to a bacteriocidal effect. For his work on phototherapy Finsen was awarded the Nobel prize in 1903, the only one in the area of Dermatology/Photomedicine. He and others propagated using heliotherapy also against surgical tuberculosis, and people were cured as they tanned (7,8). Heliotherapy was later on discouraged for lung tuberculosis because of reports on activation of the disease, even with fatal results (9). Heliotherapy at sunburn-inducing dosages was also found to be effective against psoriasis (10). In 1981 the wavelength dependence of this therapeutic effect was published, with maximum efficacy around 300 nm and an increase in the ratio of therapeutic over erythemal efficacy at larger wavelengths (11). This result led to the introduction of the narrow band (311–312 nm) TL-01 lamp (12). Therapeutic success of systemic immune suppressive agents, and specific “biologics” (13) confirm that psoriasis is an immunologically mediated disease, and that the phototherapeutic effect is probably immunologic in nature. In fact, systemic immunologic effects from heliotherapy may have played a role in curing various forms of surgical tuberculosis as well as in aggravating lung tuberculosis.

The work of Margaret Kripke on UV-induced immunogenic skin tumors (14,15) revealed profound immunologic effects of UV radiation, and brought forth the discipline of “Photoimmunology.” This novel discipline shifted paradigms and rejuvenated basic dermatologic research, and provided common ground to the formerly disjointed areas of research on UV carcinogenesis and on phototherapy and photodermatoses. The early work of Kripke’s group and the group of Raymond Daynes identified suppressor (T) lymphocytes as pivotal to the UV-induced tumor tolerance (16–18), and to UV-induced hapten-specific tolerance (19). Surprisingly, the physiologically relevant immunology of UV-induced skin cancers never appeared to have established itself as a prominent model in the field of tumor immunology. Although obvious to photoimmunologists working with mouse models, the concept of suppressor T cells was even outlawed in the mainstream immunology, only to be rehabilitated (20) and re-introduced as “regulatory” T cell (“T-reg”), a less apt name. Breaking the tolerance mediated by the T-regs is now considered quintessential to therapeutic tumor vaccination (21).

The threat of an ozone layer depletion sparked photobiologic research in the 1970s and 1980s to ascertain the effects of elevated ambient UV radiation on the biosphere. Initially, the foremost adverse health effect projected was an increase in skin cancer. But concerns about the ramifications of UV immunosuppression for infectious diseases arose and were confirmed in animal experiments, e.g. for noncutaneous tubercle bacilli infections (22). As a leading expert, Margaret Kripke joined the United Nations Environmental Program panel that assessed the health effects of a depletion of stratospheric ozone (23).

In the group of Jan van der Leun we employed a hairless mouse model to extract dose–response relationships and an action spectrum for UV carcinogenesis (24–26), and fitted the mouse model to epidemiologic data to arrive at estimates of the increases in skin cancer that would result from an ozone depletion (27–30). In our experiments on UV carcinogenesis, systemic and other UV effects surfaced that could be attributable to immune responses. This caught our interest, more so because of the anticipated far-reaching consequences. Therefore, we joined forces with the experimental immuno-toxicology group of Henk van Loveren to investigate UV-induced immunosuppression and its impact on infectious diseases in mice and rats. Here, I present the “meandering” of our research through the “Photoimmunologic landscape” disclosed by Margaret Kripke, leading us from UV carcinogenesis, to infectious diseases, and finally into our dermatologic clinic to investigate a group of patients with “sun allergies.”

UV-induced systemic enhancement of primary skin tumors

Skin carcinomas can occur successively or even simultaneously in sun-exposed skin areas of a patient. This poses the question of whether these tumors arise as truly independent random events (originating from mutations in separate cells in highly exposed skin) or whether there exists an interdependency between the lesions, which conditions a person to develop multiple lesions. Around 1980 we addressed this rather abstract question in a series of experiments in which hairless mice were subjected to sequences of partial UV exposures. Daily UV exposure resulted in skin tumors which in early stages with diameters <2mm were mostly diagnosed as actinic keratoses (AKs), and with diameters >3mm as squamous cell carcinomas (SCCs) (24,25), similar to what is observed in human beings, where AKs were surmised to be precursors of SCCs. From partial shielding of the animals it appeared that the first AKs occurred independently, i.e. the latency time on the right flank was not noticeably affected by whether or not the left flank was simultaneously exposed to develop AKs (31). However, the subsequent latency and rate of occurrence of AKs on the formerly shielded left flank was significantly enhanced if the right flank had priorly been exposed for 13 weeks and had already developed AKs. This latter effect was even more dramatic with larger skin areas, if the mice were preexposed on the abdomen for 15 weeks till some first AKs developed, and were subsequently irradiated on the dorsum to induce AKs and SCCs. Strikingly, the impact of the preexposures in these experiments appeared to increase with increasing tumor size, suggesting that a systemic effect became more pronounced as a tumor grew. However, if exposure of the abdomen was ceased earlier, after 4 weeks, long before AKs appeared (compare to median time till “tumor take” in next section), the impact of the preirradiation was much less and became smaller with increasing tumor size and time (32), suggesting that the systemic effect was not yet fully developed after this short preirradiation. Evidently, this systemic effect could seamlessly be attributed to the UV-induced immunosuppression and tumor tolerance discovered by Margaret Kripke (15,16). The increase in impact with tumor size after prior induction of AKs would suggest that the likelihood of immunogenic elimination of the first tumors rises in proportion to their sizes and times of residence in the skin. These results indicate that the occurrences of multiple AKs could indeed signify that an individual has acquired a state of enhanced susceptibility to UV carcinogenesis, i.e. a high-risk group is formed.

In experiments in which UV exposures were discontinued before the occurrences of AKs we observed that AKs still occurred albeit with much delay and in far lower numbers (25). Also, a substantial portion of the AKs that appeared after discontinued irradiation regressed. This regression diminished with increasing tumor size, which might be attributable to a selection of tumors that “out ran” a partially weakened immunosurveillance, or to a selection of nonantigenic (“progresser”) tumors that escaped immunosurveillance entirely.

UV exposure and tolerance toward skin tumors

To ascertain more precisely at what point in time the hairless mice would acquire the putative tumor-tolerant state in relation to AK development, we wanted to challenge the mice at various time points with a cell line of a UV-induced tumor from a syngeneic mouse. An intracutaneous inoculation with these tumor cells in an unirradiated syngeneic host should lead to rejection of the cells, whereas after sufficient UV exposure, the host should no longer reject the cells which then grow out to a full-blown tumor, i.e. a “tumor take.” Such a “regresser” cell line (T51) from a UV-induced tumor in HRA/SKH mice was kindly given to us by Vivienne Reeve, and we were thus able to determine the percentage of mice susceptible to “tumor take” in the course of UV exposure regimens with either 1 or 0.5 minimal edemal dose (MED) per day. We found that this percentage rose from 0% to 100% in less than 8 weeks with the high daily dose, whereas this took 16 weeks with the low daily dose. The median times till tumor take were about 4 and 8 weeks, respectively (33), substantially shorter than the median latency times of 1 mm AKs: 10.5 and 14.5 weeks, respectively (24,33). The percentage of mice with AKs increased much more steeply than the percentage with tumor take. As found earlier by Kripke’s group in a different mouse model (34), there appeared to be reciprocity between the daily dose and time till tumor take (a two-fold increase in daily dose shortened the time two-fold, i.e. the total UV dose determined tumor tolerance), whereas such a reciprocity did not hold for AKs (a two-fold increase in daily dose shortened the latency time by about 30%). By extrapolation this discrepancy would imply that at low daily dosages (around 0.05 MED day−1) the induction of AKs may actually precede the onset of a tumor-tolerant state as determined by inoculation with T51 cells. Thus, the induction of AKs does not appear to parallel the acceptance of implants of UV tumor cells, indicating that there is no strict interdependence between the two. It could be envisaged that the injection itself of 4 million T51 cells—a sizeable tumor volume—contributes to the induction of tumor tolerance, which may not yet be attained when the first AKs with considerable smaller volumes appear under chronic UV exposure.

Also, there appeared to be no relationship between foci of cells with mutant p53 (microscopic precursors of AKs) and outgrowth of implants of T51 cells (35). Although the steepness of the increase in the number of these foci with time was similar to that of the percentage tumor take (Weibull powers of time equal to 3.7 ± 0.7 and 2.8 ± 0.7, respectively), the dependence on daily dosages did not appear to match: the foci showed no reciprocity between daily dose and induction time, but a relationship similar to that of AKs (36).

Tumor tolerance as ascertained by “tumor take” has now been functionally linked to the induction of regulatory (CD4+CD25+) (37) and natural killer T cells (38). However, the early development of primary tumors in relation to induction of tumor tolerance is not entirely clear, and some UV-induced tumors (“progresser” UV tumors) may develop completely independently of the tumor tolerance (39). Like others in earlier attempts (40), before tracking CD4+CD25+ cells, we could not detect any clear-cut quantitative phenotypic alterations in leukocyte subsets (MHC class II+, NLDC-145+, CD11b+, Thy-1/CD3+, L3T4/CD4+ and Lyt-2/CD8+ cells) in lymph nodes, spleen or skin that could signify the onset of the tumor-tolerant state (41). However, similar to what Kripke’s group found after exposing C3H mice for a week (42), we detected a delayed dramatic decrease in Vγ3+ CD3+,CD4-CD8-dendritic epidermal T cells (DETCs) in the hairless mice after 2 weeks of 1 MED day−1. After 10 weeks large numbers of Vγ3- CD3+CD4-CD8- T cells appeared in the epidermis, around the time the first AKs appeared. DETCs display non-MHC-restricted killing of various skin-derived tumor cells, but no killing of normal keratinocytes (43–45), suggesting they form a first line of defense against tumor cells. And the DETCs appear to be important in the immunization for CD8+ T cell-mediated cytotoxicity against UV-induced tumors (46). As DETCs are not present in humans, their cell-killing function is probably completely absent in human epidermis, and the MHC-I/CD8+ immunization (as in contact hypersensitivity [47]) is probably entirely dependent on antigen presentation by Langerhans cells (LCs).

UV-induced DNA damage and immunosuppression

Immunologic reactions commonly evolve through an intricate network of signaling and cell activation. It should therefore be no surprise that UV immunosuppression appears to have several starting points in various photoreceptors, among which DNA is a very prominent one. There is strong evidence that UV-induced DNA damage, such as cyclobutane pyrimidine dimers (CPDs), plays a role: enhanced removal of CPDs abrogates UV-induced suppression of contact hypersensitivity (CHS) (48) whereas a persistence of the DNA damage due to a repair deficiency enhances the susceptibility to UV immunosuppression (49). Interestingly, we found that removal of CPDs from the entire skin by photorepair in genetically engineered mice reversed systemic suppression of CHS, whereas removal solely from (K14+) basal epidermal cells did not, but the latter did result in a major reduction in UV carcinogenesis (50). Photorepair of 6-4 photoproducts (the other major DNA damage induced by UVB radiation) had no effect on UV-induced immunosuppression, and little or no effect on UV carcinogenesis.

LCs and DETCs were found to decrease in the epidermis upon UV exposure, and this was associated with immunosuppression. The question was whether this loss of LCs and DETCs was caused by cell death, loss of surface markers or migration to lymph nodes. Following a suggestion by Margaret Kripke, we used a then novel H3 antibody against CPDs (generated by Len Roza) to investigate whether the DNA-damaged LCs and DECTs from a UV-exposed epidermis would be detectable in draining lymph nodes. We indeed found CPD-containing cells in the lymph nodes as of 1 h after exposure of the skin and increasing up to 24 h. And by double staining all of these cells proved to be MHC-II+ and some were also NLDC145+, i.e. LCs. No Vγ3+ DETCs were detected (51). DNA-damaged LCs were found to be able to mediate antigen-specific immunosuppression and tolerance in mice (52), and removal of the DNA damage from the LCs made them immunogenic instead of suppressive and tolerogenic (53). Later on, we found that in humans, too, CD1a+ LCs migrated out of epidermis after UV exposure, and could be caught in the fluid of a suction blister we raised. But in most of these migrating LCs the CPDs were repaired and no longer detectable (54). It is therefore unlikely that migration of DNA-damaged LCs mediates UV-induced immunosuppression in humans, where antigen presentation is more likely to be primarily skewed by cytokines, such as IL-10.

Exploiting different types of deficiencies in nucleotide excision repair (NER) we investigated the differential effects of transcription coupled repair (TCR, highly efficient repair of transcribed DNA strands) and global genome repair (GGR, acting on nontranscribed DNA strands) on local and systemic immunosuppression (in systemic immunosuppression UV exposure of a skin area affects immunization through a distant unexposed skin area, whereas in local immunosuppression UV exposure and subsequent immunization are done on the same skin area). A lack of TCR (in CsB and XpA mice) was found to greatly increase the susceptibilities to UV-induced LC migration and local immunosuppression of CHS, whereas a deficiency solely in GGR did not increase these susceptibilities when compared with NER proficient littermates of the mice (55). Likewise, the sunburn sensitivity was only enhanced when TCR was deficient (as in CsB and XpA mice). Remarkably, the susceptibility to systemic immunosuppression was only increased if TCR as well as GGR were deficient (as in XpA mice) (56). This would suggest that after disruption of active genes, owing to a deficiency in TCR, systemic suppression would require a subsequent dysfunction of genes that are “switched on” later (with DNA damage still present owing to deficient GGR), implying that a combination of disrupted initial and delayed signaling is required to enhance systemic immunosuppression.

In confirmation of reports by Vivienne Reeve on a reversal of UVB- (and cis-urocanic acid [UCA]-) induced suppression of CHS by UVA exposure (57), we found that UVA and long-wave UVA1 exposures counteracted the UVB- (and cis-UCA-) induced systemic immunosuppression of delayed hypersensitivity (DTH) to the bacterium Listeria monocytogenes (58). Interestingly, Reeve and Tyrrell (59) and Shen et al. (60) found this action of UVA irradiation to be linked to induction of heme oxygenase, γ-interferon (IFN-γ) and IL-12, which combines interestingly with the work of Schwarz et al. who found IL-12 to enhance the GGR arm of NER and thus break the UVB-induced immunosuppression (61) associated with IL-10. This possible route of UVA-IL-12-DNA repair might shed a new light on mystifying earlier experiments, as we have performed ourselves (62), on “photorecovery” in mammalian and human skin, despite the absence of the required enzyme (photolyase) (63). Evidently, counterbalancing cytokine profiles induced by UVB and UVA radiations may have important practical implications for exposures to full sunlight (with varying ratios of UVB/UVA) and should be studied further in clinical photodermatology.

UV exposure and infections in animal models

From the suppression of skin tumor rejection, CHS and DTH, the inference followed that UV irradiation might disrupt immune reactions against infectious agents. After initial reports on UV effects on skin infections (inflammatory reactions to Leismaniasis suppressed [64] and to herpes simplex activated [65]), Margaret Kripke took on a leading role in exploring UV modulation of various systemic microbial infections in mouse models, e.g. involving bacteria (BGC [22] and Borrelia burdorferi [66]), parasites (Schistosoma mansoni [67]) and fungi (Candida albicans [68]). Exploiting the experimental animal models available at the Dutch National Institute of Public Health and the Environment, we embarked on a program to assess the impact of UV exposure on various forms of infection (69,70), including viral infections (71). And by comparing associated suppression in lymphocytic responses in the animals (mice/rats) and humans, we aimed to ascertain at which levels of sun exposures humans would become more vulnerable to infection owing to UV-induced immunosuppression (72). Thus, we estimated that for 5% of the people a sun exposure of about 90 min around noon at 40° north in July would be sufficient for a two-fold, or higher, reduction in the specific T-cell responses to L. monocytogenes (73). Although not established for man, such suppression was associated with a substantial increase in microbial loads in rats. A 5% decrease in the thickness of the ozone layer was estimated to shorten this exposure time by approximately 2.5%.

As UV exposure was mostly found to suppress Th1-mediated cellular immune responses (e.g. driving CHS and DTH), it was quite remarkable that UV exposure was also found to aggravate infections in which humoral Th2-mediated responses appeared to be important, i.e. IgE against larvae of the worm Trichinella spiralis in muscle tissue. In separate experiments we investigated the effects of UV irradiation on archetypal Th1- and Th2-mediated immune responses, viz. CHS against oxazolone or picrylchloride, and serum IgE against ovalbumin (74). Both responses were decreased by preexposure to UV radiation. UV suppression of the Th1 response was associated with decreased levels of IFN-γ, IL-4 and IL-12 in the spleen of Balb/c mice, whereas suppression of the Th2 response was associated with increases in IFN-γ, IL-4 and IL-10 in the spleen. Although IL-10 is produced by Th2 cells, its role is context dependent, and may inhibit Th2 responses (75).

Experiments with a hepatitis B (HepB) vaccine in mice (Balb/c and C57Bl/6; HepB is not infectious in mice) showed that UV exposure suppressed both cellular and humoral responses, although the latter was restricted to lower levels of IgG-2a, a Th1-associated antibody (76). It was furthermore found that after chronic low-dose exposure (1/3 MED/day for 5 weeks) Balb/c mice lost their susceptibility to UV-induced suppression (after 3/4 MED/day for 5 days) of DTH against HepB vaccine but not of CHS against picrylchloride, i.e. UV adaptation for DTH but not for CHS (77). Evidently, this might be an important factor in experiments on humans in comparison with experiments on mice: in contrast to the laboratory animals, any volunteer in clinical experiments will inevitably have been UV exposed before.

UV exposure and infections in humans

Aside from practical difficulties, it is obviously unethical to perform experiments in which humans are deliberately infected with and without prior UV exposures. The closest surrogate seems to be vaccination with and without prior UV exposures. Although HepB vaccination—like most preventive vaccinations—is Th2 skewed (AlO adjuvant), it appeared worthwhile to investigate whether prior UV exposure would have antigen-specific effects on immunity, both because of an apparent practical relevancy (should people avoid UV exposure before vaccination?) and because of basic scientific importance. The study ran in two consecutive winters on volunteers who had not been sun bathing for at least a month. The subjects were exposed to 1 MED on the first day and to incrementally higher dosages (+10% day−1) on four subsequent days. Three days after the last exposure they received the first vaccination, and a second and third after 1 and 6 months, respectively. Although the preirradiation proved adequate for suppression of CHS and NK activity, no HepB antigen-specific effects were found in the course of time on lymphocyte stimulation or on total antibody titers or isotype titers (78). Interestingly, the levels of cis-UCA leached from the skin after the series of UV exposures showed no correlation with mitogenic lymphocyte stimulation, stimulation by recall antigens or HepB-specific antibody levels, but it did show a negative correlation with HepB antigen-specific stimulation of lymphocytes (79). Furthermore, we found that a certain polymorphism in the IL1β gene (C3953T, increasing IL1β expression) was significantly linked to increases in HepB antigen-specific antibody titers and in HepB antigen-specific stimulation of lymphocytes (the latter only in homozygotes) (80). In the individuals who were homozygous for this minor IL1β variant (allele frequency 0.26), the UV preexposure resulted in a significant suppression of antigen-specific antibody titers, but not in any significant effect on antigen-specific lymphocyte activation (comparing four vs five homozygotes from test and control groups, respectively) (81). Polymorphims like these, either or not in combination with UV exposure, might in part explain the small fraction (about 5%) demonstrating inadequate vaccination responses.

Termorshuizen investigated hospital records on HepB vaccination of paramedical students, and found a slightly retarded antibody response when the vaccination was started in the “sunny season” (April–September). But the antibody titers were no longer different at the end of the vaccination procedure (82). No seasonal effects were detected in vaccination responses of children against rubella and measles. In analyzing the records of renal transplant patients he found significant increases in herpes simplex infections in spring and herpes zoster infections (“shingles”) in summer. Also, skin infections with yeast/fungi were significantly increased in summer. The rate of bacterial infections went up with the accumulated number of reported episodes of sunburn (83). Tapping into an ongoing study on asthma and allergy in a birth cohort, he was able to monitor a cohort of 1-year-old children on their sun exposures over spring and summer by 6 weeks retrospective questionnaires for the parents. Thus, he found that slight sunburn was significantly associated with earache/runny ear, but that, unexpectedly, very low sun exposure increased the risk of upper respiratory tract infections (coughing and runny nose) (84). Hence, he found indications of short-term adverse effects of sunburns on immunity, but also a beneficial effect of regular moderate sun exposure, which—not surprisingly—is attributed by some to maintaining adequate levels of vitamin D3 (85).

UV immunosuppression and “sun allergy”

After some unsuccessful attempts at finding evidence for increased antigenicity in UV-exposed skin of patients with polymorphic light eruption (PLE), as had been reported for actinic purigo/PLE (86), we reconsidered the results from Kevin Cooper’s group which showed that virtually all human volunteers were “UV immunosuppressible” given sufficiently high exposures (87) which in turn indicated that this is probably a sound physiological reaction (88). Pursuing this line of thought, it follows that presumably some illicit pathologic immune reaction would develop in the skin without such an immunosuppressive countermeasure. Hence, it would seem that the immune system in the skin is performing a “balancing act” between elimination of infections and malignant cells, and suppressing illicit reactions against UV-exposed skin cells that harbor photochemically altered “nonself” molecules. An inadequate immunosuppressive response upon UV exposure could thus cause an allergic skin reaction, such as observed in the idiopathic photodermatosis PLE. The balance seems to waver often: in a recent large-scale European study in which several geographically spread academic dermatologic centers collaborated, we found that one in five indoor workers had experienced PLE-like skin reactions (M. Bock, A.S. Janssens, T.C. Ling, L. Anastasopoulou, C. Antoniou, F. Aubin, T. Bruckner, B. Faivre, N.K. Gibbs, C. Jansen, S. Pavel, F.R. de Gruijl, L.E. Rhodes and T.L. Diepgen, submitted).

Cooper’s group found that the immunosuppression in humans by high level UVB exposure was associated with a depletion of LCs in the skin and an influx and proliferation of CD11b+ monocytes/macrophages (87). These latter cells were found to be the main producers of Th1-suppressive IL-10 and their presence was crucial for the development of antigen-specific tolerance upon sensitization after UVB exposure (89,90). We therefore decided to investigate whether PLE patients would deviate in their responses to high level UVB exposure. After establishing that 6 MEDs to the buttock skin resulted in robust and reproducible responses in healthy subjects, we performed a comparative study on healthy subjects and PLE patients who came to the clinic because of recent complaints and who were found to react pathologically to artificial UVB exposure in the clinic. We indeed observed a reduced influx of CD11b+ cells in PLE patients after UVB overexposure, but we found a more dramatic persistence of LCs (91). The large majority of the influx of CD11b+ cells was negative for the macrophage marker CD68. In a follow-up study most of these cells proved to be neutrophils which carried IL-4, IL-10 and tumor necrosis factor-α (TNF-α) (92), and the number of these neutrophils was lower in PLE patients than in controls after UV exposure. Overall, the number of cells with IL1β and TNF-α was lower in PLE patients. These shifts appear to correspond largely to less UV-induced Th2 skewing in PLE patients. The reduced influx of neutrophils in PLE patients was also found at lower levels of UV exposure (3 MEDs) (93), but no abnormalities were found in chemotactic activity of the neutrophils from PLE patients and in the endothelial expression of Intercellular Adhesion Molecule 1 (ICAM1) and E-selectin. The efflux of LCs, caught in the fluid of suction blisters, proved indeed to be lower in PLE patients than in controls after exposure to 6 MEDs (94). The number of dermal HLA-DR+ and activated/matured LCs was higher in PLE patients. Hence, it was inferred that next to a reduced influx of neutrophils, the hampered migration of LCs with a dermal accumulation of activated/mutared LCs might contribute to the pathogenesis of PLE.

A group of people comes to the clinic every spring for preemptive “UV hardening” to avoid PLE by sun exposure; the UV hardening is attained by a course of mild UV exposures (12 exposures in 5 weeks rising from about 0.1 to 1.4 MED). In this group of PLE patients we investigated the migratory responses of LCs and neutrophils upon a 6 MED challenge before and after the UV hardening (after hardening the MED was re-measured). In these people, who did not come to the clinic because of recent complaints, we first checked whether PLE could be provoked by a series of daily UVB and/or UVA exposures. Those subjects who reacted pathologically to UVB exposures in these tests showed significantly reduced migratory responses of LCs and neutrophils to UV overexposure, which was normalized after UV hardening (95). The reduction in migration of LCs and neutrophils in these people with PLE was much less than in the earlier study, which is most likely attributable to the selection of patients (either with or without recent complaints of a sun-induced skin eruption, and a difference in UVB challenges, e.g. 5 or 3 days of repeated exposures). Investigation of cytokines in blister fluid harvested after UVB overexposure yielded no indication of low chemotactic activity but more of a skewing toward inflammatory responses in PLE patients (A.S. Jannssens, S. Pavel, M. Teunissen, J.J. Out-Luiting, R. Willemze and F.R. de Gruijl, submitted). One would thus be led to surmise that a mild increase in MED due to UV hardening may normalize the initial skewing toward inflammatory reactions. And with an increase in MED, a 6 MED overexposure after UV hardening will further increase the immunosuppressive responses, reflected in an increased migration of LCs and neutrophils (these migratory responses were indeed found to increase further in untreated PLE patients with exposures >6 MED; W. Kölgen, personal communication). Of course, it needs to be verified whether the skewing in cytokines indeed normalizes after UV hardening. But such a result would imply that it is not the immunosuppressive arm of the UV response but the inflammatory arm which needs adjustment in PLE. This imbalance in UV responses is apparently easily corrected by a series of mild UV exposures, resulting in a slight increase in MED. And this is, of course, common experience in the vast majority of people with PLE: the problem is only temporary and resolves after a couple of days of (mild) sun exposure.

The work dealt with above is mainly concerned with cellular and cytokine responses related to UV immunosuppression, but gives no direct evidence that PLE patients really demonstrate diminished UV-induced suppression of contact allergic reactions of the skin. Such crucial functional studies were performed by two British groups who provided the experimental evidence that CHS reactions are indeed suppressed less in PLE patients than in controls in the UV dose range around 1 MED (96,97).


Photoimmunology demonstrates how interesting, complex and far-reaching UV-induced skin reactions can be: we wandered from UV carcinogenesis into tumor immunology and from there on to how infections may be affected by UV exposure. And finally this journey took us from the laboratory into the clinic, to look at actual patients with an “idiopathic” photodermatosis for whom we had the audacity of dreaming up a pathogenic mechanism from what seemed to be a “bad” reaction to UV radiation in healthy individuals.

In each of the research areas we passed through many important questions that remain to be answered: e.g.

  • • which cells at what stage of tumor development provide the antigen(s) to initiate the tumor tolerance, and what is the nature of the antigen(s)? what is the mechanism of AK regression?
  • • does UVA-driven IFN-γ and IL-12 release have an impact on UVB effects in humans and on photodermatoses? and is it a possible mechanism for Finsen’s “therapeutic” chemical rays?
  • • what is the scale of the impact of ambient UV radiation on infectious diseases, and what is the balance between beneficial and adverse effects?
  • • and can we identify (have we identified?) the pathogenic mechanism of PLE, and can we exploit this knowledge for preventive intervention and therapy?

UV immunosuppression appeared to imply nothing but trouble, but in the end, it may turn out to be the unappreciated silent but resolute friend bailing us out of a predicament.


Acknowledgements— I am grateful to my mentor Jan van der Leun who has ushered me into the research on photomedicine, UV carcinogenesis in particular, and stratospheric ozone depletion. Don Forbes and the late Fred Urbach are thanked for showing a young physicist the ropes in experiments with hairless mice. Henk van Loveren and Johan Garssen are thanked for their pleasant collaboration, and enabling me to participate in the research on UV effects on various infectious diseases. Len Roza was pivotal in starting me up in research on DNA damage and repair. Bert van der Horst, Harry Versteeg and Errol Friedberg opened up the possibility to dissect the effects of DNA repair with their various strains of genetically modified mice. The collaboration with Huib van Weelden and Stan Pavel was important to the clinical experiments. The recent collaboration with Lesley Rhodes and Neil Gibbs in an EC project on PLE is greatly appreciated. Importantly, besides all these seniors, I want to thank all the PhD students and postdocs (first authors) in our groups who did most of the actual work. And finally, Margaret Kripke for being a trailblazer.