Possible Effects of Sunlight on Human Lymphocytes

The human population is exposed to both the ultraviolet A (UVA) and B (UVB) regions of the solar spectrum. UVB induces mainly dipyrimidine photoproducts in DNA by a direct photochemical mechanism, whereas UVA is absorbed by other cellular constituents and induces mainly oxidative damage indirectly. The proportions of the different dipyrimidine photoproducts, and the ratio of dipyrimidine to oxidative damage depend on the exact spectral output of a UV source. Irradiation of human epidermal keratinocytes induces release of cytokines, with cyclobutane pyrimidine dimers playing a significant role in the process. These cytokines may then modulate the activity of cells of the immune system. Freshly isolated human lymphocytes are exquisitely sensitive to UVB irradiation, because of their low deoxyribonucleotide pools. They also have a separate defect in removal of cyclobutane pyrimidine dimers from their DNA. We have observed that frequencies of mutations at the hprt locus in human T-lymphocytes and translocations involving the bcl2 locus in B-lymphocytes appear to be associated with sunlight levels over the period before the blood sample was taken. This may be an indirect cytokinemediated effect, and may be relevant to the possible link between non-Hodgkin's lymphoma and sunlight. On the other hand, sunlight can have beneficial effects, and may protect against autoimmune diseases including type I diabetes and multiple sclerosis. J Epidemiol, 1999 ; 9 : S48-S57.


Effects of Sunlight on Human Lymphocytes
Michael H. L. Green 1,2 , Corinne Petit-Frere 1.4,Peter H. Clingen 1, 5, Graham Bentham 3, Jane Cole 1, and Colin F. Arlett 1 The human population is exposed to both the ultraviolet A (UVA) and B (UVB) regions of the solar spectrum.
UVB induces mainly dipyrimidine photoproducts in DNA by a direct photochemical mechanism, whereas UVA is absorbed by other cellular constituents and induces mainly oxidative damage indirectly.The proportions of the different dipyrimidine photoproducts, and the ratio of dipyrimidine to oxidative damage depend on the exact spectral output of a UV source.
Irradiation of human epidermal keratinocytes induces release of cytokines, with cyclobutane pyrimidine dimers playing a significant role in the process.
These cytokines may then modulate the activity of cells of the immune system.Freshly isolated human lymphocytes are exquisitely sensitive to UVB irradiation, because of their low deoxyribonucleotide pools.They also have a separate defect in removal of cyclobutane pyrimidine dimers from their DNA.We have observed that frequencies of mutations at the hprt locus in human T-lymphocytes and translocations involving the bcl2 locus in B-lymphocytes appear to be associated with sunlight levels over the period before the blood sample was taken.This may be an indirect cytokinemediated effect, and may be relevant to the possible link between non-Hodgkin's lymphoma and sunlight.
On the other hand, sunlight can have beneficial effects, and may protect against autoimmune diseases including type I diabetes and multiple sclerosis.J Epidemiol, 1999 ; 9 : S48-S57.
Ultraviolet B; human lymphocytes; hprt (hypoxanthine guanine phosphoribosyl transferase); bcl2; mutation; cytokines ABBREVIATIONS 6-4, pyrimidine pyrimidone photoproduct; 6TG, 6thioguanine; bcl2, anti-apoptotic oncogene, commonly involved in a specific translocation in follicular B-cell lymphoma; CPD, cyclobutane pyrimidine dimer; Dewar, Dewar isomer of the pyrimidine(6-4)pyrimidone photoproduct; hprt, hypoxanthine guanine phosphoribosyl transferase gene; ICAM-1, intercellular adhesion molecule one; IDDM, insulindependent diabetes mellitus; NHL, non-Hodgkin's lymphoma; (UVC, below 280 mn) (Figure 1) l).The shorter the wavelength of UV, the more energy it transmits and the more damage it causes.However, shorter wavelengths are more readily blocked, either by the atmosphere or outer layers of skin.Humans are not normally exposed to UVC, since it is completely blocked by the earth's atmosphere.Exceptions are arc welders (which is why they must have such effective eye and skin protection) and perhaps laboratory scientists, since most research with UV makes use of UVC from germicidal lamps emitting predominantly at 254 nm.It is simple and convenient to use UVC as a research tool, but effects can be significantly different to those of natural sunlight.UVB comprises the shortest wavelengths to reach the surface of the earth from the sun.It is UVB that is attenuated by the ozone layer, so that depletion of the ozone layer may lead to increased UVB exposure.The proportion of UVB in sunlight depends not just on ozone, but on the distance that the light must travel through the atmosphere, so that altitude, latitude and time of day are important.Since UVA is less energetic, it passes readily through the atmosphere and it is the main UV component of sunlight to which we are exposed.
The key difference between UVB and UVA is that UVB has the ability to introduce direct photochemical damage into DNA.The major types of damage formed in this way are between adjacent pyrimidines on the same DNA strand: the cyclobutane pyrimidine dimer (CPD), the pyrimidine(6-4)pyrimidone photoproduct , and its Dewar isomer (Dewar).Not only the overall yields of these photoproducts, but their relative proportions depend on the exact spectral distribution of the UV.The 6-4 is induced at shorter wavelengths, but is isomerised to the Dewar by longer wavelength UVB and even UVA.UVC forms the same direct photoproducts as UVB, but in different proportions, with negligible levels of the Dewar 2).
Because UVA is less energetic than UVB, it does not cause direct photodamage to DNA.It can, however, be absorbed by other molecules (chromophores) within the cell, which absorb

SUNLIGHT AS A CARCINOGEN
There is overwhelming evidence that sunlight is a human carcinogen1,4), and in terms of absolute numbers of cancers, it is one of the most significant carcinogens to which the United Kingdom population is exposed.There are approximately 40,000 new cases of basal cell carcinoma , 10,000 cases of squamous cell carcinoma, and 5,000 cases of malignant melanoma per year in the United Kingdom 5).Fortunately basal cell carcinoma is almost completely curable and squamous cell carcinoma and early malignant melanoma largely curable.Nevertheless there are about 1500 skin cancer deaths per year in the United Kingdom, about 1000 of which are due to malignant melanoma 6).There has been an enormous increase in all forms of skin cancer in the United Kingdom and Europe over the last 40 years 4).This pre-dates depletion of the ozone layer, and reflects a change in lifestyle, with the enormous increase in sunshine package holidays, and the belief that a suntan is attractive and a sign of good health.(I will point out later that sunlight can have good as well as bad effects).
In hairless mice, both UVB and UVA have been shown to induce squamous cell carcinoma, and an action spectrum has been determined7,8,9).Although association of all three forms of human skin cancer with sunlight is unequivocal, in the case of malignant melanoma (and to some extent basal cell carcinoma) the nature of the link is not straightforward4).
The British population are enthusiastic users of sunbeds 10) Although it is plausible to expect that this results in an increased risk of cancer, we do not have the data to predict how much risk, and which cancer.Sunbed manufacturers target their advertising accordingly.Initially they reduced the UVB component and advertised their lamps as providing a lower ratio of harmful UVB to beneficial UVA.A difficulty with this approach is that UVA was relatively ineffective in inducing tanning, but since the lamps provided a higher proportion of UVA than natural sunlight, the overall exposure of the client to UVB, and hence the ability to cause tanning, could be broadly similar.
As dermatologists have started to complain that UVA is also a known animal carcinogen, and that it has other skin-damaging effects, the manufacturers have tended to increase the proportion of UVB again and to emphasise how their lamps have a similar spectrum to pure natural sunlight 11).

SUNLIGHT AFFECTS SKIN CANCER BY MORE THAN ONE MECHANISM
Both the UVB and UVA components of sunlight form types of DNA damage a, 12. 13.14) likely to induce mutations.UVB has been shown to induce chromosomal alterations in human cells 15).Thus sunlight is a classic initiating carcinogen.UVB, however, can also affect the promotion and progression stages of cancer.In mice, the elegant experiments of Kripke and her colleagues 16) have provided overwhelming evidence that UVB acts as a promoter, by inducing immunosuppression.
In humans the evidence is indirect that this immunosuppression leads to cancer promotion.Transplant patients undergoing immunosuppression show greatly increased susceptibility to squamous cell carcinoma 17) (also non-Hodgkin's lymphoma (NHL) 18)).Contact hypersensitivity is more likely to be impaired by UVB in patients with skin tumours 19).Patients with the genetic disease xeroderma pigmentosum show a defect in the repair of DNA damage induced by a wide range of carcinogens, but only show a large increase in susceptibility to skin cancer.Increases in other types of cancer are small a0), implying that an effect of UVB other than mutagenesis must be involved 21).
It has been proposed that induction of apoptosis in response to DNA damage is a defence mechanism against cancer and mutation 22), and a proportion of squamous cell carcinomas show mutations in the p53 gene, which should render them resistant to UVB-induced apoptosis 23).There is therefore a possibility that the ability of UVB to induce apoptosis in skin cells is important in the prevention of skin cancer.We suspect that it is not.
Figure 2 sets out some of the possible interactions between the different components of sunlight (UVB and UVA), the different intermediates through which they might act, and the processes they might affect.In the remainder of this paper, I will summarise our work addressing several of these issues:  immunosuppressive cytokines from skin cells, but the overall balance is in favour of immunosuppression.Induction appears to be linked to UVB, rather than UVA, and to the formation of direct photoproducts.In the case of the pro-inflammatory cytokine interleukin-6, we have found that the wavelength dependency closely follows those for absorption of ultraviolet by DNA and for formation of CPDs 24).By use of liposomes containing a photolyase which specifically reverses the formation of CPDs (Generously provided by Dr D Yarosh) 25), we have been able to show that the CPD makes a major contribution to induction of cytokine release.This is a surprising result for several reasons.We have previously shown that CPDs do not make a significant contribution to cell killing 2) or to inhibition of RNA synthesis 26), For these events, the less abundant 6-4 photoproduct seems to be critical.The approach used in these earlier studies was to measure the proportions of specific photoproducts formed by different defmed UV sources and to compare biological responses to the same sources.The work was dependent on Professor 0 Nikaido, Dr T Mori and Dr L Roza providing monoclonal antibodies to each of the specific photoproducts.
An important distinction between these photoproducts is that the CPD is relatively poorly recognised by excision repair enzymes, whereas the 6-4 and Dewar are rapidly excised 2, 27.28).It is sometimes suggested that although overall removal of CPDs is not important in cell survival, removal of CPDs in actively transcribed genes is critical 29).However, if repair of CPDs in critical regions was essential for survival, we would have expected to see a different result in our experiments comparing responses to UV sources 2).But if cells seem largely able to ignore CPDs without coming to harm, how can they be a trigger for cytokine release?A cell can only respond to DNA damage, if it "knows" that the damage is there.The signal might be an inhibition of RNA synthesis, but again our results do not suggest an effect of CPDs, at least on initial inhibition of RNA synthesis 26).The overall inhibition in these experiments, however, would largely relate to non-mRNA synthesis, and it might be that 6-4s were more effective than CPDs in sequestering the polymerase machinery.
A further problem with understanding the mechanism of signalling involved in cytokine release is that although there is evidence of a role of oxidative events, it has been very difficult to show any role of direct photoproducts in early signalling effects following UV irradiation.Nevertheless direct photoproducts clearly are involved in late events, and some intermediary mechanism must exist.We have encountered yet another complication in assessing the risks associated with sunbeds.We had predicted from our wavelength dependency studies 24) that it would be possible to define the ability of a sunbed to induce release of cytokines from the yield of CPDs that it formed.Instead we have found that the UVA component of radiation from a sunbed completely suppresses release of the cytokines interleukin-6 and tumour necrosis factor-a .Furthermore, UVA prevents cytokine release from UVB-irradiated keratinocytes.Surprisingly, induction of mRNA synthesis is not suppressed.
A second marker of UV immunomodulation, the inhibition of upregulation of intercellular adhesion molecule 1 (ICAM-1), does appear to be mediated by direct photoproducts, and not affected by the UVA component of sunbed irradiation (Clingen, Berneburg, Woollons, in preparation).
Overall, this does not amount to a very clear picture.Photoproducts induced by the UVB component of sunlight play a key role modulating the immune response, but the mechanism is not at all clear.Moreover, some effects appear to be reversed in a paradoxical way by UVA.Thus at the present time we do not have the knowledge to predict how an increase in the UVB component of sunlight, or a particular class of sunbed will modulate the immune response.

SUNLIGHT AND LYMPHOCYTES
There are two mechanisms by which sunlight might affect cells of the immune system.The most obvious is that lymphocytes pass through the skin and will receive some exposure to sunlight during their passage.This will obviously be very much lower than the exposure of keratinocytes and other skin cells.On the other hand, human T-lymphocytes are exceptionally sensitive to UVB irradiation 30), more so than to UVC.The reason for this hypersensitivity is that circulating non-dividing lymphocytes have exceptionally low deoxyribonucleotide pools 31.32).Rapid excision repair of the 6-4 and Dewar photoproducts leads to formation of gaps in the DNA, which require resynthesis.Levels of deoxyribonucleotides become rate-limiting and persistent breaks lead to death of the cell.These breaks also appear to play a role in UVB clastogenicity 15) Addition of deoxyribonucleosides to the medium allows strand rejoining to proceed and enhances survival 32).Lymphocytes appear to have a second, separate defect in their ability to recognise and excise CPDs 33.34,35)The hypersensitivity of lymphocytes to UVB may make it possible to envisage some direct damaging effect of sunlight.On the other hand, we and others have never succeeded in demonstrating UV damage in lymphocytes following UVB irradiation of patients, even using a highly sensitive assay (Clingen unpublished).
The second way in which sunlight could affect lymphocytes is via release of cytokines from cells of the skin.We know that such release occurs, the only question is the extent to which it could modulate the immune system.In this connection, we have had a surprising lead, from a quite separate line of investigation within our unit.For a period of about 15 years, Dr Jane Cole in the MRC Cell Mutation Unit led a major project, screening human populations for evidence of exposure to environmental mutagens 36,37.38).She used a clonal assay, measuring the proportion of Tlymphocytes that were able to grow in the presence of the toxic guanine analogue 6-thioguanine (6TG).Resistance arises thorough inactivation of the X-linked gene hypoxanthine/guanine phosphoribosyl transferase.We have found a proportion of 6TG-resistant T-lymphocytes in blood samples from all adult subjects that we have tested, although the frequency varies greatly.By carefully optimising conditions (activation with phytohaemagglutinin, use of interleukin-2 and a thioguanine-resistant lymphoblastoid feeder layer) 39), it is possible to obtain colonies from virtually every T-lymphocyte in a blood sample.We have established that mutant frequencies tend to increase with age and to be higher in smokers.There is a lower apparent mutant frequency when the cloning efficiency is high 1).Mutant frequencies are higher in xeroderma pigmentosum patients 40).We found no effect of alcoholic beverages 41) occupational radiation exposure 42), or exposure to crude oil following a shipwreck 43).We were unable to confirm an initial indication that there might be an association between domestic radon levels and mutant frequency 45) During these studies, we began to realise that another factor appeared to influence mutant frequency, which was related in some way to the date when the blood sample was obtained.Following a fortuitous meeting between GB and CFA, we began to investigate the possibility that this novel factor might be sunlight.We analysed all the data for which we had information on when and where the blood sample was taken.This gave us information on over 750 assays for hprtmutant frequency on over 440 blood samples.
We found 61) that for blood samples obtained during the summer months, hprt-mutant frequencies tended to be higher following a period of sunny weather.The effect was highly significant (P<0.001) and cumulative, the strongest association being with sunlight records for the three weeks prior to taking the blood sample (Figure 3).We did not see any correlation with variation in winter sunlight levels, but in winter actual amounts of sunlight are very low.We did not carry out a direct comparison of summer versus winter, because of other possible seasonal factors, such as diet or infection.
The literature is full of examples of epidemiological correlations which disappear on further investigation.Is this another example?We have used a standard method of analysis (multivariate ANOVA of log transformed mutant frequencies), but models of distributions as complex as human mutant frequencies will never be exact.Nevertheless, the apparent significance is sufficiently high to give some margin of confidence.Since all the blood samples were originally obtained to test for quite different environmental factors, the timing of taking samples was not optimal to investigate any effect of sunlight.Large numbers of samples were obtained for specific studies over relatively short periods, whereas other periods were underrepresented.Nevertheless, analyses of subsets of the data confirm the apparent sunlight effect.Obviously we would not have set out to design a study in this way, but on the other hand we would never have set out to look for this particular effect.Another criticism could be that we do not know the actual exposure to sunlight of the subjects, but are relying on sunshine records from the nearest meteorological station.We have found, however, that the strength of the association has tended to increase as we have refined our analysis from crude monthly national sunshine records to daily local records.Another factor that tends to increase our confidence is that factors which we had previously identified as affecting mutant frequency are virtually identical between winter and summer blood samples.It is only sunlight that is specific for summer.
If sunlight is associated with hprt-mutation, an obvious question is whether the mutations found show a UV-specific fingerprint.UV-induced mutations tend to occur at dipyrimidine sites, and to include a significant proportion of tandem mutations mutations at adjacent base pairs, where dipyrimidine photoproducts could have formed.A substantial number of hprtmutations in T-lymphocytes have been derived from other studies and sequenced.There is no evidence for a UV fingerprint among this group, although we can have no indication of sun exposure prior to the blood sample.
In the last few years, in association with Dr G Cortopassi and Dr Y Liu, we have extended our monitoring studies to chromosome translocations in B-lymphocytes 3&4B).At (14:18) translocation places the bcl2 pro-oncogene under the control of the immunoglobulin heavy chain promoter.This is the commonest translocation in follicular B-cell lymphoma, which is in turn the commonest form of NHL 47,48).The translocation is not a sufficient condition for cancer, and has also been found at a low frequency in B-lymphocytes from normal individuals 49), including the majority of adult donors in our studies.It arises in the bone marrow as an error in the site-specific recombination process that generates immune diversity in B-cells.The translocation can be detected by a PCR approach, using primers for the two chromosomes (Figure 4).A PCR fragment is only obtained if a DNA molecule is present where the two chromosomes have become joined.The method can detect one translocation in DNA from over 105 cells from the mononuclear fraction.We had previously shown a statistically significant correlation between hprt-mutation and bcl2 translocation frequencies 38) in those blood samples for which we had data for both endpoints.Although our data set for bcl2 translocation is far smaller, we again find a statistically significant association with sunlight over the same three week period before the blood sample was taken.As an inevitable consequence of the smaller sample size (95), this association is not so strong (P<0.002).

ON THE OTHER HAND
If the ultraviolet component of sunlight induces release of cytokines from the skin and modulates our immune system, could this have other effects on our health?The epidemiologies of type I diabetes (IDDM, insulin-dependent diabetes mellitus, juvenile diabetes) and of multiple sclerosis may be relevant.Both show a strong geographical dependency, with high incidence associated with high latitudes.Ireland and Finland show high levels of diabetes, for instance, whereas levels are low in Mediterranean countries 56).A similar pattern is observed for multiple sclerosis 57) Various explanations of these distributions have been offered, including diet, and exposure to triggering infections.It is worth bearing in mind that these are autoimmune diseases, and sunlight is the most widespread and potent immunosuppressive agent to which the population is regularly exposed 57 58).It is important also to remember that sunlight is necessary for vitamin D formation 4.59), and this too may be linked to diabetes 60).
If sunlight does protect against specific diseases, it may be possible to determine the mechanism and develop a more focused strategy for protection.Processes leading to skin cancer and preventing diabetes cannot be identical, because the British love of sunshine holidays has not led to a reduction in diabetes to compensate for the increase in skin cancer.To date, we have only evolved a strategy to achieve the harm, without obtaining any compensating benefit.The British enthusiasm for excessive sunbathing is placing a significant burden on health care resources, but with a better understanding and use of sunlight, we may be able to turn the balance of risk/benefit to our advantage.

Figure 1 .
Figure 1.Properties of different regions of the ultraviolet spectrum.

1 )
Does DNA damage induce the signalling events which are associated with ultraviolet-induced immunosuppression?If so what types of DNA damage are important?2) Could sunlight affect cells of the immune system or cancers other than skin cancer?3) Does sunlight have important health effects other than skin cancer?INDUCTION OF CYTOKINE RELEASE BY UV Sunlight induces the release of both pro-inflammatory and

Figure 2 .
Figure 2. Links of UVB and UVA with possible mechanisms of sunlight-induced skin cancer .Sunlight-induced damage is mutagenic , can induce signalling events, and can induce apoptosis, each of which may affect the induction of skin cancer .

Figure3.
Figure3.Effect of sunlight record before taking

Although the distressing genetic
disease Lesch-Nyhan syndrome 50) is the result of a defective hprt locus, almost all of us have between 1 in 104 and 1 in 106 T-lymphocytes which are mutant at this locus, and come to no harm.Most of us also have a proportion of B-lymphocytes carrying the bcl2 translocation, but this must be considered more seriously in view of the association with non-Hodgkin's lymphoma.Several studies have suggested a possible link between sunlight and NHL.Increases in the incidence of skin cancer and NHL have tended to occur over the same time period, and in the same populations 51) Skin cancer patients show an increased incidence of NHL 52).Within the UK there is a geographical association between NHL and high sunlight 33).On the other hand, no such geographical association has been seen in the USA 54), and the issue is highly controversial 55).An apparent association between sunlight and a chromosomal translocation found in NHL is relevant in this respect.Since bcl2 translocation arises as an error in site-specific recombination during B-lymphocyte differentiation in the bone marrow, any effect of sunlight must be indirect, possibly mediated by cytokines.

Figure 4 .
Figure 4. Mechanism of detection of translocations by PCR.With appropriate primers , a fragment is only amplified if a translocation has occurred in a specific region.