Aberrant gene expression with deficient apoptotic keratinocyte clearance may predispose to polymorphic light eruption

Dear Editor, Polymorphic light eruption (PLE) is the most common photodermatosis, demonstrating intermittently appearing nonscarring pruritic, erythematous papules, vesicles or plaques on patient skin within hours of ultraviolet radiation (UVR) exposure. It depends on genetic susceptibility, as well as environmental UVR exposure, and is apparently a failure of normal UVR‐induced immunosuppression in the presence of simultaneously produced skin photoneoantigens,1 2 with a resulting increased immune response. Reduced tumour necrosis factor‐α, interleukin (IL)‐4 and IL‐10 expression, a lack of neutrophils, reduced T helper 1/T helper 2 skewing and reduced Langerhans cell migration have also been demonstrated in ultraviolet B‐irradiated PLE skin.3 Bead‐array immunoassay studies have further revealed significantly higher PLE baseline IL‐1β and CCL11 (eotaxin) spring plasma levels,4 while decreased transforming growth factor‐β1, IL‐10 and receptor activator of nuclear factor kappa B ligand epidermal and dermal expression, and reduced regulatory T‐cells (Tregs) occur in patients irradiated with ultraviolet A1,5 as well as decreased Treg function in all patients with PLE, irradiated or not.6 However, the genetic basis for the impaired UVR‐induced immunosuppression and the nature of any photoinduced PLE neoantigens have yet to be identified. Investigating PLE candidate genes to try and define its genetic basis has generally been unrewarding, although segregation analysis suggests 72% of U.K. patients carry a low‐penetrance PLE susceptibility allele,7 perhaps either functionally important with a polymorphism resulting in only slightly altered function, or else relatively unimportant with a mutation abolishing all function. To investigate this further, we used genome‐wide expression analysis to try and identify possible candidate genes. After meeting the requirements of the Declaration of Helsinki, keratinocyte cultures from punch biopsies of sun‐protected (buttock) skin of eight white females with PLE and eight age‐matched female controls (Table 1) were exposed either to a moderate dose (970 mJ cm−2) of solar‐simulated radiation (SSR) from a 1‐Kw xenon arc source (Oriel, Stratford, CT, U.S.A.) or else to no radiation.8 This produced four sets of samples (32 samples in total): PLE with SSR; PLE without SSR; controls with SSR; and controls without SSR. Six h later, keratinocyte RNA was extracted from the samples and the expression of > 22 000 genes analysed and selectively confirmed by real‐time polymerase chain reaction (RT‐PCR) studies. Genespring® v.7.3.1 (Silicon Genetics, Redwood City, CA, U.S.A.) was used to examine gene expression levels, with genes being filtered by so‐called ‘flags’ (i.e. the marginal or full presence in at least six samples) and also by expression levels (the raw data needed to have a value of at least 40 in at least 16 samples). All arrays passed standard quality‐control requirements and 11 405 genes passed both filters. Expression levels were thereafter analysed by parametric two‐way anova. Results were corrected for multiple testing by the Benjamini and Hochberg method, or that of Bonferroni. P‐values < 0·05 were considered significant, while filtering on fold change indicated the most up‐ and downregulated genes. Characteristics of the two groups PLE, polymorphic light eruption; MED, minimal erythema dose; UVB, ultraviolet B; PLESI, PLE Severity Index; NA, not applicable. Characteristics of the two groups PLE, polymorphic light eruption; MED, minimal erythema dose; UVB, ultraviolet B; PLESI, PLE Severity Index; NA, not applicable. By anova, 1979 genes were SSR‐regulated: 194 upregulated more than twofold and 446 downregulated more than twofold. The most upregulated gene was COX2 (up 21‐fold), as also noted elsewhere.9 Only 16 genes were differentially expressed between patients with PLE and controls when P‐values were derived following multiple test correction by the Benjamini and Hochberg method (Table 2), while with the more stringent Bonferroni method only the first four genes in Table 2 were found to be significantly different. Genes differentially expressed in patients with polymorphic light eruption (PLE) vs. controls and their regulation by solar‐stimulated radiation (SSR) Genes are listed in order of the difference between patients with PLE and controls, with genes with the most statistically significant differences listed first. In the gene array data, the P‐values were obtained by correcting for multiple comparisons with the Benjamini and Hochberg method; when correction was made with the more stringent Bonferroni method, only the first four genes were significantly different. The interaction between a diagnosis of PLE and the effect of SSR was not significant for any gene. RT‐PCR, real‐time polymerase chain reaction; NS, nonsignificant; NT, not tested; NF‐κB, nuclear factor kappa B. Genes differentially expressed in patients with polymorphic light eruption (PLE) vs. controls and their regulation by solar‐stimulated radiation (SSR) Genes are listed in order of the difference between patients with PLE and controls, with genes with the most statistically significant differences listed first. In the gene array data, the P‐values were obtained by correcting for multiple comparisons with the Benjamini and Hochberg method; when correction was made with the more stringent Bonferroni method, only the first four genes were significantly different. The interaction between a diagnosis of PLE and the effect of SSR was not significant for any gene. RT‐PCR, real‐time polymerase chain reaction; NS, nonsignificant; NT, not tested; NF‐κB, nuclear factor kappa B. Of the 16 differentially expressed genes 14 showed lower expression in PLE, and two higher. Of particular note was that the interaction between a diagnosis of PLE and the effect of SSR was not significant for any gene; in other words, we did not detect a differential effect of SSR in PLE. To check the validity of the array result, eight of the 16 genes were further investigated by RT‐PCR, with statistically significant differences being confirmed for six, but not for CASP1 and PON2, although they showed similar trends. We found that 1979 genes were regulated by SSR; this complexity of UVR‐induced keratinocyte transcriptional responses has also been noted in previous gene‐expression array studies.10 Such studies have also shown variations in responses to UVR. These variations are likely to reflect differences in cells, radiation sources, time courses, arrays, statistical methods and other such factors. We used just one 6‐h time point after SSR because of cost constraints, but this was considered the time frame likely to provide the most reliable overview of SSR‐induced growth factor and cytokine effects, and also that best associated with PLE rash development after irradiation. However, given the potential for methodological error and possible discrepancy between in vitro and in vivo responses, it is reassuring that COX2 was the gene most affected by SSR exposure.9 Of the 14 genes with lower expression in patients with PLE than in controls (Table 2), two are closely associated with apoptotic cell clearance: namely C1S and SCARB1. Four others may also be associated with such clearance: FN1, IGSF3, CASP1 and PON2. During such clearance, protein accumulation develops on cell‐surface blebs, where it is subject to proteolysis and other modification, leading to potential autoantigen formation.11 Therefore, the reduced expression in our patients of six genes related to apoptotic cell clearance suggests a possible autoantigen source promoting PLE outbreaks during UVR‐induced apoptosis. Furthermore, as normal apoptosis has been shown to induce immunosuppression,12 the impaired PLE apoptotic response may well inhibit this, thereby permitting enhanced antigen recognition. In support of these concepts, autoantigen accumulation also occurs after irradiation in another photosensitive condition, lupus erythematosus,13 14 Kuhn et al. irradiated 20 patients with lupus erythematosus and seven controls, as well as three patients with PLE, and found that the patients with lupus erythematosus and those with PLE showed increased apoptotic cell accumulation vs. controls.15 From these considerations, therefore, we postulate that a genetic deficiency in PLE apoptotic keratinocyte clearance leads to a partial failure of normal UVR‐induced immunosuppression along with the production of persistent, potentially photoautogenic debris, with both these factors arguably permitting PLE outbreaks after UVR exposure.