In Vitro Analysis of Photosensitizer Accumulation for Assessment of Applicability of Fluorescence Diagnosis of Squamous Cell Carcinoma of Epidermolysis Bullosa Patients

Epidermolysis bullosa (EB) is a group of inherited skin disorders characterized by blistering following mechanical trauma. Chronic wounds of EB patients often lead to tumors such as squamous cell carcinoma (SCC). Early diagnosis may prevent its invasive growth—frequently the reason of premature mortality of EB-patients. Early detection of tumors is achieved by fluorescence diagnosis (FD), where photosensitizers localize selectively in tumors and fluoresce upon illumination. Excessive accumulation of photosensitizers in inflamed areas, as occasionally found at chronic wounds and tumors due to inflammatory processes, leads to false-positive results in FD. This study analyzed accumulation kinetics of the photosensitizers hypericin and endogenous protoporphyrin IX (PpIX) in different skin cell lines including the three EB subtypes under normal and proinflammatory conditions (stimulated with TNF-alpha). The aim was to assess the applicability of FD of SCC in EB. All cell lines accumulate hypericin or PpIX mostly increasing with incubation time, but with different kinetics. SCC cells of recessive dystrophic EB (RDEB) accumulate less hypericin or PpIX than nonmalignant RDEB cells. Nevertheless, tumor selectivity in vivo might be existent. Non-EB cell lines are more active concerning photosensitizer enrichment. Proinflammatory conditions of skin cell lines seem to have no major influence on photosensitizer accumulation.


Introduction
Epidermolysis bullosa (EB) is a group of skin disorders which are genetically determined. ey are characterized by blistering of the skin and mucosa following mechanical trauma [1][2][3]. EB can be divided into three classes.
EB simplex (EBS) is the most common form of EB. Its inheritance is normally autosomal dominant but in some cases an autosomal recessive trait can be found. e blister formation begins intraepidermally with a subnuclear disruption of the basal keratinocytes. e reason for this is mutations in speci�c genes encoding for keratin 5 and keratin 14 (KRT5 and KRT14) [4,5] and for plectin (PLEC1) [6].
EB junctionalis (EBJ) is a group of autosomal recessive disorders. ere are two main categories within this group of EB, the Herlitz (lethal) and non-Herlitz (nonlethal) form.
e tissue separation of these forms is through the lamina lucida of the basement-membrane zone beneath the plasma membrane of epidermal basal cells. Nonscarring blistering is the result of this separation. Mutations in genes encoding for laminin 5 subunits (LAMA3, LAMC2, and LAMB3) and collagen, type XVII, alpha 1 (COL17A1) are causative for this form of EB [7].
EB dystrophica (EBD) has an autosomal recessive or dominant inheritance. e blistering level of this type of EB lies below the lamina densa of the epidermal basement membrane. Mutations are occurring in COL7A1, the gene encoding for collagen, type VII, alpha 1 [8].
All these forms of EB are resulting in the pain of blistering, in�ammation, and in some cases scarring and cancer because of loss of the skin's barrier function [9]. e chronic wounds of EB patients are accompanied by in�ammatory processes, which may promote induction and growth of skin tumors such as squamous cell carcinoma (SCC), especially when the in�ammation lasts for a long period or is derailed [10]. Early diagnosis of SCC is important, since early stages of SCC can be treated more easily than invasively growing SCC, which oen is the main reason of premature mortality of the EB patients. To this purpose, a new, effective, and noninvasive technique for early detection of SCC would be offered by �uorescence diagnosis (FD) using a photosensitizer. e latter localizes selectively in tumor tissue and is able to �uoresce upon irradiation with visible light of a wavelength matching the absorption spectrum of the substance. is modality can be applied for tumor diagnosis, even in early stages, and it is especially helpful in �uorescence-guided resection [11].
Beyond diagnosis, the tumor-localizing photosensitizer is able to kill the target cells when light activated. In the presence of oxygen, most photosensitizers generate either superoxide radicals, that might form peroxides and hydroxyl radicals in a type I reaction, or singlet oxygen molecules ( 1 O 2 ) in a type II reaction. e tumor destruction occurs �nally due to reactive oxygen species (ROS) [12] or reactive nitrogen species [13]. is treatment is called "photodynamic therapy (PDT)" and was already used for basal cell carcinoma treatment of an RDEB-patient [14].
Chronic wounds, especially a problem for EB patients, as well as tumors are oen accompanied by in�ammatory processes, which may lead to false-positive results in FD, decreasing the speci�city. e reason for this is unclear, but some clinical studies supposed local immune cells such as macrophages, which invade in�amed areas, as source for an excessive accumulation of the photosensitizer [15][16][17][18]. Nevertheless, it cannot be excluded that nonimmune cells accumulate the photosensitizer at a higher rate under in�ammatory conditions, and that proin�ammatory cytokines could play a role in this process.
Proin�ammatory cytokines control in�ammation and modulate neovascularisation, cell proliferation, and migration [19]. In�ammation is an essential part of wound healing, but it can turn to a problem, when this controlled process is switching to an uncontrolled or excessive one. is is oen seen in diseases like chronic wounds, tumor metastasis, psoriasis, and arthritis [20]. Most of all deregulated wound healing is caused by an increase of interleukin 1 (IL-1alpha and IL-1beta) and tumor necrosis factor-alpha (TNF-alpha) levels [21][22][23][24]. Interleukin 6 (IL-6) also seems to play an important role in the pathogenesis of in�ammation [25]. On the other hand, secretion of these cytokines was found to be upregulated by PDT [26][27][28].
As indicated before, detection of early stages of SCC in EB-patients via �uorescence diagnosis would be a new approach to prevent invasive SCC growth by early intervention. erefore, the aim of the present study should be the analysis of the �uorescence kinetics of photosensitizers in EB cell lines to assess the applicability of FD on SCCs of EBpatients. e uptake of an externally applicable photosensitizer such as hypericin and accumulation of the endogenously formed photosensitizer protoporphyrin IX (PpIX) should be analysed in the malignant EB-cell line SCCRDEB4 and compared to nonmalignant EB cell lines and a malignant non-EB cell line. To induce endogenous PpIX formation, its precursor ALA (5-aminolevulinic acid) in the heme biosynthesis will be applied. PpIX is currently successfully used in tumor diagnosis [11]. Hypericin is a plant constituent from St. John�s wort with excellent �uorescent properties that can modulate several signaling pathways [29,30].
To prove the hypothesis that in�ammation of tissues causes excessive accumulation of photosensitizers oen leading to false-positive results in FD, the effect of proin�ammatory conditions on uptake or accumulation of hypericin or PpIX, respectively, in normal and malignant EB-cell lines and their respective reference cell lines should also be analysed. e question here is whether FD is in�uenced by the proin�ammatory state of EB cells.
In order to address these issues, we performed analysis of uptake kinetics on seven different skin cell lines (GABEB, EBS-MD, RDEB-CL, SCCRDEB4 as EB cell lines, representative of the three main subtypes, and HaCat, Skin, A431 as keratinocyte, �broblast, and SCC cancer cell lines as control and reference). For establishment of a proin�ammatory milieu, we stimulated the cells with TNF-alpha (tumor necrosis factor-alpha) to activate proin�ammatory pathways [31]. e SCCRDEB4 cell line was routinely grown in Keratinocyte-SFM (Life Technologies, Vienna, Austria) and RDEB-CL and GABEB cell lines in Keratinocyte-SFM with 100 U/mL penicillin and 0.1 mg/mL streptomycin. Keratinocyte-SFM was always supplemented with bovine pituitary extract and recombinant epidermal growth factor.

Photosensitizers.
In preliminary experiments, cell proliferation characteristics were analysed by the MTT assay in order to determine the respective cell number for each cell line, which should be used to yield comparable results due to comparable cell mass and monolayer density. Twentyfour hours aer seeding the cells to con�uency of about 80% cell culture medium was replaced with medium containing hypericin or the PpIX precursor ALA.
Hypericin is a secondary metabolite predominantly extracted from the Hypericum perforatum (St. John's wort). It was purchased from Planta Natural Products (Vienna, Austria) and added to the serum-free medium of the cell cultures in �nal concentrations of 3 and 5 M. ese concentrations were chosen according to previous work [33]: irradiation of A431, HaCat, and SCCRDEB4 cells with a diagnostic protocol using 3 M hypericin was sublethal, but using 5 M hypericin was phototoxic with 40% to 70% survival having therapeutical impact.
Endogenous PpIX is the last molecule in the heme biosynthesis prior to heme and depends on its precursor 5aminolevulinic acid (ALA). If ALA is given in excess, PpIX is accumulated in cells and can be used as a very effective photosensitizer for many hours. ALA was purchased from Sigma-Aldrich (Vienna, Austria). It was applied to the serumfree medium of the cell cultures in �nal concentrations of 0.5 and 1 mM. Also, these concentrations were chosen according to a previous study [34], in which 0.5 mM and 1 mM ALA was used to induce PpIX efficiently in a linear relationship for �uorescence detection of mononuclear and circulating tumor cells.
Handling with photosensitizers was performed under subdued light conditions.

2.�. �n�uction of Proin�ammatory �ilieu.
Fibroblast and keratinocyte cell lines were incubated with lipopolysaccharide (LPS) or TNF-alpha (both Sigma-Aldrich, Vienna, Austria) in preliminary tests to induce in�ammation. Based on these tests TNF-alpha was then chosen as most applicable inductor. e culture medium was replaced by corresponding serumfree medium containing TNF-alpha (5 ng/mL) 4 h before photosensitizer treatment was started.
IL-6 and IL-1beta Ready-Set-Go! ELISAs (eBioscience, Vienna, Austria) were performed according to the manufacturer's protocol to identify the best procedure for inducing a proin�ammatory milieu, which was repeatedly checked throughout the study.

Uptake Experiments.
Aer incubation with hypericin or ALA, respectively, for 1 h up to 8 h, cells were washed twice with 100 L DPBS and lyzed for 10 min by addition of 50 L 1% Triton X-100. Subsequently, the �uorescence intensity of the photosensitizers hypericin and PpIX in the cells was measured in 96-well plates (Greiner, Kremsmuenster, Austria) using a microplate reader (In�nite M200pro, Tecan, Groedig, Austria). Hypericin �uorescence was detected at lambda(ex) = 340 nm and lambda(em) = 604 nm and PpIX �uorescence at lambda(ex) = 435 nm and lambda(em) = 635 nm. e �uorescence signals were related to the protein content of each sample (BCA assay� Fisher Scienti�c, Vienna, Austria) to correct for variations in the cell mass. Measurements were performed in triplicates and series were repeated independently for at least two more times.

Data Analysis, Statistics.
Comparisons between data points were statistically evaluated by the Student's t-test for independent samples. At least three independent series were included in the analysis.

Results
e results show the analysis of the �uorescence kinetics of EB and non-EB cell lines under normal and proin�ammatory conditions. In order to check for the proin�ammatory state of the selected cell lines aer TNF-alpha induction, the IL-6 level of cells was checked at random (data not shown).
All cell lines take up hypericin or generate PpIX, respectively, at the selected concentrations.
SCCRDEB4 cells generate PpIX from 0.5 mM ALA with linear kinetics, equally under normal and proin�ammatory conditions (Figure 1(a)), and from 1 mM ALA with linear kinetics and moderately higher in the proin�ammatory state ( Figure 1(b)). However, these variations are not signi�cant. is differs from normal state, which shows a �nal �uorescence intensity similar to the lower concentration but with a curve shape forming a plateau.
SCCRDEB4 cells take up hypericin at a concentration of 3 M with linear kinetics, equally under normal and proin-�ammatory conditions (Figure 2(a)), and at a concentration of 5 M almost equally under normal and proin�ammatory conditions with curve shapes showing the onset of a plateau. e �nal �uorescence level at the proin�ammatory state is marginally higher than the �uorescence aer 3 M hypericin (Figure 2(b)).
PpIX is generated by RDEB-CL cells from 0.5 (Figure 3(a)) and 1 mM ALA (Figure 3(b)) equally under normal and proin�ammatory conditions, with more or less linear kinetics for 0.5 mM and with a beginning �attening to a plateau for 1 mM ALA. e �uorescence intensity is slightly lower at the higher concentration.
RDEB-CL cells take up hypericin at a concentration of 3 M in a moderately �attening curve course, statistically nonsigni�cantly higher under proin�ammatory than normal conditions (Figure 4(a)). Uptake of 5 M hypericin occurs in a curve forming a plateau also more or less equally under normal and proin�ammatory conditions (Figure 4(b)). Under normal conditions, the �uorescence increases with the hypericin concentration, but under TNF-alpha pretreatment, the �nal �uorescence values at 8 h incubation time with GABEB cells generate PpIX from 0.5 mM ALA in a linear correlation with the incubation times, with a delay of 2 h. Under proin�ammatory conditions, the �uorescence between 5�7 h is signi�cantly increased ( , Figure 5(a)), in contrast to 1 mM ALA, which induces about equal PpIX �uorescence in a linear relation with a delay of 1 h (Figure 5(b)). Fluorescence intensity increases with ALA concentration and is about double without TNF-alpha induction.
Hypericin at a concentration of 3 M is taken up by GABEB cells in a linear curve course, equal under proin-�ammatory and normal conditions (Figure 6(a)). While the uptake of 5 M hypericin occurs also more or less equally under normal and proin�ammatory conditions, the curve shape shows here a plateau (Figure 6(b)). Under both conditions, �uorescence increases with hypericin concentration. EBS-MD cells generate PpIX from 0.5 mM ALA with linear kinetics up to 8 h (Figure 7(a)). e course is equal under proin�ammatory and normal conditions, for 0.5 mM, and more or less also for 1 mM ALA (Figure 7(b)). However, PpIX formation aer application of 1 mM ALA leads to a curve shape with a plateau. Noteworthy is the fact that the �uorescence decreases with increasing ALA concentration.
EBS-MD cells take up hypericin at a concentration of 3 M in a moderately �attening curve course, equal under proin�ammatory and normal conditions (Figure 8(a)). While the uptake of 5 M hypericin occurs also more or less equally under normal and proin�ammatory conditions, the curve shape shows a distinct plateau and a very rapid increase already within the �rst two hours (Figure 8(b)). Under both conditions, �uorescence increases with hypericin concentration. PpIX is generated by A431 cells from 0.5 mM ALA with linear kinetics up to 8 h under nonin�ammatory conditions (Figure 9(a)). Under proin�ammatory conditions, the �nal �uorescence intensity is signi�cantly ( ) reduced, and the curve reaches a plateau. PpIX formation aer application of 1 mM ALA shows a linear curve under proin�ammatory conditions, which is �attened under normal conditions with lower �uorescence endpoints. A signi�cant difference can be found for 2 and 3 h ( ) (Figure 9(b)). Fluorescence increases with increasing ALA concentration.
A431 cells take up hypericin at a concentration of 3 M in a moderately �attening curve course, equal under proin�ammatory and normal conditions until 5 h. Between 5 and 8 h, the �uorescence increase is reduced under proin�ammatory conditions with statistical signi�cance at � h ( ) (Figure 10(a)). Uptake of 5 M hypericin shows a similar curve course under proin�ammatory and normal conditions, which is linear until � h. Aer � h, a reduced �uorescence increase aer TNF-alpha application leads to a moderate difference in both conditions (Figure 10(b)). However, under both conditions, �uorescence increases more than doubles with increasing hypericin concentration.  HaCat cells generate PpIX from 0.5 (Figure 11(a)) and 1 mM ALA (Figure 11(b)) equally under normal and proin-�ammatory conditions with more or less linear kinetics. �e only difference is that the �uorescence of PpIX induced by 1 mM ALA shows no further increase from � to 8 h. �e �uorescence intensity shows almost double the amount with the higher concentration.
Hypericin at a concentration of 3 M is taken up by HaCat cells in a moderately �attening curve shape, equal under proin�ammatory and normal conditions (Figure 12(a)). While the uptake of 5 M hypericin occurs also more or less equally under normal and proin�ammatory conditions, the curve shows a distinct plateau (Figure 12(b)) as well as a steep increase within the �rst hour. �nder both conditions, �uorescence increases with hypericin concentration. PpIX formation in Skin cells is about linear aer incubation with 0.5 mM ALA (Figure 13(a)) and almost linear aer incubation with 1 mM ALA (Figure 13(b)). Differences in �uorescence between in�ammatory and normal conditions are not signi�cant. Fluorescence decreases with increasing ALA concentration.
Skin cells take up hypericin at concentrations of 3 M (Figure 14(a)) and 5 M (Figure 14(b)) in curve courses with plateaus under proin�ammatory as well as normal conditions. �nder proin�ammatory conditions, the �uorescence of hypericin, 3 or 5 M, respectively, is lowered, highly signi�cant for 6 and � h incubation with 3 M hypericin

Discussion
�e aim of this study was the analysis of �uorescence kinetics of photosensitizers in EB cell lines representing the three types of EB, to assess the suitability of FD for squamous cell carcinoma of EB patients. Since proin�ammatory conditions due to the chronic wounds [10] are oen present in EBpatients, and excessive accumulation of photosensitizers in in�amed areas had been occasionally observed in the clinical situation, the �uorescence kinetics were also measured in all cell lines aer TNF-alpha induction. For this purpose, the photosensitizer hypericin and the precursor ALA were externally applied; the latter one to induce the endogenous generation of the photosensitizer PpIX in the cells.
All cell lines produce PpIX and take up hypericin, as demonstrated by the �uorescence measurements, but with different accumulation kinetics. Since all cell lines represent �broblasts or keratinocytes, this is in line with other studies [35][36][37].  Based either on the �uorescence kinetic curves or their endpoints at 8 h (Figures 15(a)-15(d)), the following comparisons can be made with special emphasis on the applicability of FD for SCC of EB-patients.
�.�. C�r�e ����e �n� ��e�i�� A�����l�ti�n �� ���� ���t�sen� sitizer in All Cell Lines. A moderate-curve �attening until to a distinct plateau is found in all cell lines aer 5 M hypericin incubation and in most cases also aer 3 M hypericin. In cell lines where it is found aer ALA incubation, it is limited to 1 mM.
Curves with a plateau normally indicate saturation with the photosensitizer but could in some cases be based on aggregation of the molecules, which �uenches the �uorescence [38]. Especially the application of 5 M hypericin seems to be at the upper limit, and incubation with a higher concentration cannot be recommended for the used cell lines. A retarded �uorescence increase was observed aer ALA incubation in GABEB cells and a rapid initial increase aer 5 M hypericin incubation of EBS-MD, HaCat, and Skin cells. e latter phenomenon points to a rapid initial uptake of hypericin mainly to �broblast cell lines, which are known to carry out increased receptor-mediated endocytosis of lowdensity lipoprotein (LDL) in contrast to keratinocytes [39]. Lipophilic photosensitizers such as hypericin bind reportedly to LDL [40,41]. However, uptake kinetics soon reach a distinct plateau in most cell lines, including Skin �broblasts.

Effect of Concentration on the Photosensitizer Accumulation.
In most cell lines, higher photosensitizer or precursor concentrations induce higher or-in the case of saturation-at least equal �uorescence (Figures 15(a)-15(d)). e observed seeming decrease in a few cell lines is not statistically signi�cant. RDEB-CL). �CCRDEB4 cells show lower �uorescence levels than RDEB-CL cells, independent of TNF-alpha induction. is negative selectivity is the highest (more than double) aer 0.5 mM ALA incubation, about double aer hypericin treatment, with both concentrations, and reduced aer 1 mM ALA (Figures 15(a)-15(d)). However, a statistically signi�cant �uorescence reduction ( ) was only detected with 5 M hypericin. In spite of its negative in vitro selectivity, hypericin tumor selectivity in vivo might well be existent [42]. RDEB-CL cell �uorescence is equal to or higher than that of the other EB cell lines.

EB-S�eci�c Photosensitizer Accumulation in �ali�nant
Cells (SCCRDEB4 versus A431). e comparison of squamous cell carcinoma cells of an RDEB patient with those of a non-EB patient shows a 2-3 times higher �uorescence of A431 cells aer PpIX formation and aer 3 M hypericin accumulation, and an up to 8 times higher �uorescence aer 5 M hypericin, for both pretreatment conditions (with and without TNF-alpha induction) (Figures 15(a)-15(d)). All di�erences are highly signi�cant ( ) with the only exception of a signi�cance for 0.5 mM ALA in the proin�ammatory state. �e EB background of the malignant cells leads to drastically reduced �uorescence by decreased PpIX formation or hypericin uptake. Due to constant de�ciency compensation processes of the recessive dystrophic EB cell line [43], other physiological activities could be limited.

Tumor Selectivity of Non-EB Cell Lines (A431 versus
HaCat and Skin Cells). �e �uorescence of PpIX is signi�cantly higher in A431 cells than in HaCat and Skin cells under nonin�ammatory conditions when incubated with 0.5 mM ALA (both: ) and signi�cantly higher than in Skin cells under both conditions when incubated with 1 mM ALA (both: ). All three cell lines take up 3 M hypericin in a similar rate. When the cells are incubated with 5 M hypericin, the dye is accumulated to a much higher degree in A341 cells than in HaCat and Skin cells (highly signi-�cant with ), independent of TNF-alpha pretreatment, (Figures 15(a)-15(d)). Except for 3 M hypericin and restricted for 1 mM ALA, the protocols are suitable to generate tumor selectivity in non-EB cell lines, at least under nonin�ammatory conditions.

4.�. EB-S�eci�c ��otosensiti�er Accumulation in Non-Malignant Cells (GABEB versus HaCat and EBS-MD versus
Skin). Cells of junctional EB patients (GABEB), compared to keratinocytes (HaCat) as well as cells of an EB could also play a role-at least with regard to the decreased hypericin uptake.

�.�. E�ect of TN�-Alpha-Induced Proin�ammatory State on
Photosensitizer Accumulation. In general, the proin�ammatory state shows no major in�uence on photosensitizer accumulation as only a few nonsigni�cant differences to the nonin�ammatory state were found. Signi�cantly increased �uorescence aer TNF-alpha induction is restricted to the presence of PpIX in GABEB cells (5-7 h with 0.5 mM ALA) and A431 cells (2-3 h with 1 mM ALA). Reduced �uorescence was measured in A431 cells (incubated with 0.5 mM ALA and 3 M hypericin) and in Skin cells aer hypericin application.
Concomitant measurements for checking the proin�ammatory state of the cells showed that GABEB cells generate high levels of IL-6 and Skin cells even double the amount whereas A431 and HaCat cells present a low IL-6 level. Based on these data, it is hardly possible to correlate an in�uence of TNF-alpha induction on the �uorescence of the photosensitizers in the studied cell lines with the amount of released IL-6.
However, with our data, we support the hypothesis that the increased �uorescence found in in�amed tissue during FD of tumors is not due to higher accumulation of photosensitizers in nonimmune cells with a proin�ammatory status but might rather be due to additional photosensitizer accumulation in the extracellular matrix and�or in in�ltrating immune cells such as neutrophils, mast cells, monocytes, and macrophages [45,46].

Conclusions
Following conclusions can be drawn from the results above.
(1) All cell lines take up hypericin or generate PpIX mostly increasing with the incubation time, but with different kinetics.
(2) SCCRDEB4 cells take up less hypericin and generate less PpIX than the nonmalignant RDEB-CL cells. is is in contrast to the non-EB cell lines, which show tumor selectivity.
From the here found in vitro results, we cannot conclude whether �uorescence diagnosis of SCC in EB patients will be feasible. Even though the applied photosensitizers exhibit no tumor selectivity in vitro, their tumor selectivity in vivo might well be existent.
(3) EB cell lines are less active than non-EB cell lines concerning uptake of hypericin or formation of PpIX.
(4) Since uptake of hypericin or formation of PpIX is hardly modi�ed under proin�ammatory conditions, the proin�ammatory state of the cells seems to have no in�uence on the �uorescence detection of the photosensitizers. erefore, the higher �uorescence of in�amed areas in tissue might rather be due to photosensitizer accumulation in in�ltrating immune cells.