Characterisation of resistance mechanisms developed by basal cell carcinoma cells in response to repeated cycles of Photodynamic Therapy

Photodynamic Therapy (PDT) with methyl-aminolevulinate acid (MAL-PDT) is being used for the treatment of Basal cell carcinoma (BCC), but recurrences have been reported. In this work, we have evaluated resistance mechanisms to MAL-PDT developed by three BCC cell lines (ASZ, BSZ and CSZ), derived from mice on a ptch+/− background and with or without p53 expression, subjected to 10 cycles of PDT (10thG). The resistant populations showed mesenchymal-like structure and diminished proliferative capacity and size compared to the parental (P) cells. The resistance was dependent on the production of the endogenous photosensitiser protoporphyrin IX in the CSZ cell line and on its cellular localisation in ASZ and BSZ cells. Moreover, resistant cells expressing the p53 gene presented lower proliferation rate and increased expression levels of N-cadherin and Gsk3β (a component of the Wnt/β-catenin pathway) than P cells. In contrast, 10thG cells lacking the p53 gene showed lower levels of expression of Gsk3β in the cytoplasm and of E-cadherin and β-catenin in the membrane. In addition, resistant cells presented higher tumorigenic ability in immunosuppressed mice. Altogether, these results shed light on resistance mechanisms of BCC to PDT and may help to improve the use of this therapeutic approach.

Proliferation capacity, adhesion efficiency and cell cycle. By using the clonogenic assay, we evaluated both the ability to form colonies (each colony is formed by >50 cells) and the proliferative capacity of each cell population, by quantifying the number and the size of the colonies formed, respectively. ASZ cells formed a higher number of colonies although they were small in size, while BSZ and CSZ gave rise to a lower number of colonies of different sizes. Comparing P and 10 th G populations, no differences in the number of colonies were observed. Regarding the size, BSZ P and CSZ P formed significantly higher number of large colonies than 10 th G. Crystal violet staining absorbance indicated a higher cellular density of ASZ and BSZ P in relation to their 10 th G (Fig. 3A). To conclude, ASZ (P and 10 th G) presented higher adhesion capacity than BSZ and CSZ cells since they formed a higher number of colonies. Conversely, BSZ and CSZ cells formed larger colonies, which indicate that they are more proliferative.
No differences were found in the distribution of the three cell lines among the different phases of the cell cycle, with approximately 50% of cells in G0/G1 phase, 25% in S phase and 20-25% in G2/M. No differences were found between P and 10 th G for any of the cell lines (Fig. 3B). subcellular localisation and production of protoporphyrin IX. To determine PpIX localisation, P and 10 th G cells were observed by fluorescence microscopy after 24 h of incubation with appropriate MAL concentrations (ASZ 0.3 mM, BSZ 0.4 mM and CSZ 0.2 mM) under UVA excitation light (Fig. 4A). PpIX was localised in the plasma membrane in all populations; and a very low fluorescence was also detected in the cytoplasm. Besides, nuclear fluorescence signal was detected in ASZ P and BSZ P and scarcely in ASZ 10 th G. In CSZ 10 th G, the fluorescent signal was higher in the polygonal than in the spindled cells. In addition, a bluish autofluorescence in the cytoplasm and particularly in mitochondria was observed in control cells without MAL ( Fig. 4A; Suppl. Fig. 1A).
Since the amount of intracellular PpIX might affect the phototoxicity, we next examined the PpIX content by flow cytometry (λexc = 625 nm, after incubation with the fixed MAL concentrations for 5 and 24 h) (Fig. 4B). All populations showed higher intracellular PpIX content after 24 h than after 5 h of MAL incubation. There were no significant differences in PpIX production either at 5 or 24 h of incubation between P and 10 th G, except for CSZ at 24 h, in which 10 th G produced less PpIX than P. www.nature.com/scientificreports www.nature.com/scientificreports/ To evaluate the potential relation between PpIX production and cell death we used the AO-EB assay that allowed distinguishing between viable (fluorescing in green) and dead (fluorescing in orange-red) cells after PDT (Suppl. Fig. 1B). After incubation for 5 h with MAL followed by irradiation, P cultures presented a higher percentage of cell death than 10 th G, confirming the MTT results. However, in CSZ, when MAL incubation was for 24 h, time for which P cells presented a high PpIX production, differences between P and 10 th G cell death were higher than at 5 h, confirming that a higher PpIX production is linked to a higher PDT-MAL effectiveness.
In vivo studies: tumorigenic capacity of BCC lines. The tumorigenic capacity of P and 10 th G populations was evaluated in immunosuppressed mice. After subcutaneous injection into mice, all populations generated tumours. Tumours induced by 10 th G were bigger than those caused by P cells (P ≤ 0.05). ASZ and CSZ lines grew quickly, reaching tumour maximum size of 500 mm 3 in 23 days, while BSZ was the slowest, needing to be monitored for 35 days. ASZ presented differences in tumour size between P and 10 th G since day 8 after the inoculation, BSZ since day 21 and CSZ since day 14 (Fig. 5). ASZ generated the smallest tumours; BSZ tumours showed the highest size difference between P and 10 th G; and CSZ (P and 10 th G) generated the largest tumours. Tumours formed by ASZ P and 10 th G cells presented two patterns: the first one was mixed, combining fibroblast-like with fibrosarcoma and osteosarcoma-like cell morphologies, and the second one was composed by fibroblast-like cells only. Tumours induced by BSZ and CSZ P cells exhibited similar characteristics. However, BSZ 10 th G tumours were composed by fibroblast-like cells and some hemangiopericytoma-like areas (with abundant vascularization). Tumours from CSZ 10 th G cells showed the three aforementioned morphologies: fibroblast-like, osteosarcoma-like and hemangiopericytoma-like. In general, the tumour progression from both P and 10 th G cells of the three lines was accompanied by invasion of the muscle and the adipose tissue without inflammation (Suppl. Fig. 2). The areas of calcified osteoid, characteristic of osteogenic differentiation, were stained by alizarin red and were observed only in tumours generated by BSZ P and CSZ (P and 10 th G) (Suppl. Fig. 2). www.nature.com/scientificreports www.nature.com/scientificreports/ In order to characterise the cells composing the tumours, we obtained two new cell populations related to each line (P T and 10 th G T) by explant culture of the tumours, and studied the cell morphology by phase contrast microscopy (Fig. 5). Both ASZ populations showed a more evident fibroblastoid morphology than the original ones. The most relevant change was observed for BSZ cells, whose 10 th G T population showed a clear spiky shape vs. the polygonal morphology of 10 th G cells. No evident morphological changes in CSZ T were observed. Genetic validation. We next evaluated the expression of p53 and of ptch and their protein products by RT-PCR and Western blot (WB), respectively. The results obtained (Suppl. Fig. 3) confirmed some of those reported by So et al. 19 . As expected, no p53 expression was detected for BSZ and CSZ, as both copies of the p53 alleles had been 'floxed' out. Only cells derived from the ASZ cell line (ASZ 10 th G, P T and 10 th G T) expressed the gene p53 as their corresponding P cells did. We have also checked the status of p53 in ASZ at the exons 5 and 8, which correspond to certain 'hot-spots' in the p53 gene where mutations are commonly found 20 . In particular, www.nature.com/scientificreports www.nature.com/scientificreports/ we have found changes in exon 5 at codons 149 (CCA to CTA) and 176 (CAT to CT/AT), but not at codon 137 (ACG) neither at the exon 8, codon 275 (CCT) as previously described 20 . At the protein level, the evaluation by WB validated the expression of p53 in all ASZ populations and no differences were observed between P and 10 th G. However, its expression was significantly higher in 10 th G T than in P T (Suppl. Fig. 3A). The expression of ptch was also studied by RT-PCR, confirming that 10 th G, P T and 10 th G T cell populations had only the mutated form of the gene and they were lacking the wild type one as their corresponding P 19 . The murine fibroblast cell line 3T3 was used as a positive control of the wild type sequence (Suppl. Fig. 3B).

evaluation of proteins implicated in tumour progression.
It is well documented that during tumour progression, cells can lose their epithelial characteristics and acquire a mesenchymal phenotype that confers invasive properties in an EMT process. In this context, we have evaluated the expression and localisation of some markers related with this process, including: E-cadherin, N-cadherin, vimentin and components of the Wnt/β-catenin pathway.
The expression and distribution of E-cadherin assessed by IF was variable (Fig. 6A). In ASZ, membrane localisation was only observed in P T cells, whereas the protein was weakly expressed in the cytoplasm of P, 10 th G and 10 th G T cells. In BSZ, while it was detected in both the cytoplasm and the membrane of P, 10 th G and P T cells, cytoplasmic localisation was predominant in the 10 th G and 10 th G T populations being completely excluded from the membrane in the 10 th G T cells. In CSZ P, P T and the polygonal cells of 10 th G, E-cadherin was observed also in the cell membrane, while the spindled population did not express this protein.
The analysis by WB and RT-PCR (Fig. 6B,C) confirmed the results observed by IF; BSZ and CSZ 10 th G showed lower expression of E-cadherin than their respective P populations; in ASZ cells, only the P T population showed expression of this protein. www.nature.com/scientificreports www.nature.com/scientificreports/ In the case of the EMT marker N-cadherin, in ASZ only 10 th G cells showed expression of the protein in the membrane, while in ASZ P, P T and 10 th G T cells the signal observed was diffuse in the cytoplasm. BSZ populations showed heterogeneous N-cadherin expression; some cells presented the protein in the membrane and in the cytoplasm, while others only in the cytoplasm. In CSZ P and 10 th G N-cadherin was also presented in both, cytoplasm and membrane, but in P T and 10 th G T cells it was only in the cytoplasm; spindled-shaped cells showed higher expression of the protein especially in 10 th G T (Fig. 7A). The WB analysis revealed higher expression in ASZ 10 th G and BSZ 10 th G T cells related to their corresponding P. Conversely BSZ P presented higher expression than BSZ 10 th G. No differences in the expression of N-cadherin were reported between CSZ populations (Fig. 7B).  www.nature.com/scientificreports www.nature.com/scientificreports/ Remaining focused on proteins implicated in EMT, we also evaluated the expression of vimentin. No differences in the expression or localisation of this protein were revealed either by IF or WB in any of the lines (Suppl. Fig. 4).
Finally, we analysed potential variations in the expression of two important proteins of the Wnt/β-catenin pathway: β-catenin and Gsk3β. The IF assay revealed that β-catenin was located mainly in the membrane and diffusely in the cytoplasm, showing a more intense signal in BSZ and CSZ lines. Also, it was observed in the nucleus of some populations (ASZ P and P T, BSZ 10 th G T). Spindled-shaped cells of CSZ 10 th G and 10 th G T populations presented lower β-catenin expression than polygonal cells (Fig. 8A). WB analysis confirmed these results; no changes were noticed for ASZ, while resistant populations of BSZ and CSZ decreased their expression of β-catenin.
In the case of Gsk3β, the protein was diffusely expressed in the cytoplasm of all lines except for BSZ 10 th G, in which it was observed in the nucleus. In addition, the fluorescent signal was more intense in CSZ 10 th G polygonal than in the spindled-shaped cells. WB analysis indicated higher expression in ASZ resistant (10 th G and 10 th G T) in relation to their parental cells. Conversely, in BSZ and CSZ resistant populations was lower than in the corresponding parental cells (Fig. 8B).

Discussion
Several therapies are being used for the treatment of BCCs, including surgery, topical Imiquimod or PDT, and also specific inhibitors of the Hh signalling pathway (which is constitutively activated in most BCCs), such as the SMO inhibitor Vismodegib. However, in some cases, resistance of BCCs to clinical treatments occurs [21][22][23][24][25][26][27] . In the case of PDT for BCC, detection of recurrences after treatment has been as high as up to 30% [28][29][30] . Therefore, understanding the resistant mechanisms to PDT constitutes a very important goal for the management of BCC.
Several intrinsic and extrinsic factors have been described to be implicated in the resistance to PDT, including those related with the treatment -location and production of PS-, cancer cells features -proliferation, adhesion www.nature.com/scientificreports www.nature.com/scientificreports/ and expression of certain genes implicated on the development of the tumour including p53 and components of shh/ptch and Wnt/β-catenin pathways 18,31 . In this study, we found that the resistance of BCC to PDT is associated to p53 expression, components of Wnt/β-catenin pathway and with the EMT process.
Cancer cells from different origin resistant to PDT in vitro have been obtained by using several PSs [15][16][17]32,33 . However, resistance of BCC cells to PDT has not been studied. In this context, we have used BCC cells (ASZ, BSZ and CSZ) isolated from mice with different genetic background 19 . BSZ and CSZ completely lack expression of p53, whereas ASZ expresses it. As previously described, in ASZ we determined mutations within the coding region of the gene at exon 5 20 . In any case the implication of such mutations in its function is unclear. From these cells we have obtained resistant cells to 10 cycles of MAL-PDT in order to stress the mechanisms of resistance.
We have detected changes in cell morphology. The 10 th G populations of ASZ and BSZ appeared more spindled than their P. In CSZ, two populations were identified: one composed of polygonal cells, similar to their P, and the other one showing spindled-shaped cells that grew surrounding their polygonal counterparts. Changes to spindled shapes have also been seen in LM3 (mammary adenocarcinoma cells) 16 and in SCC-13 cells 17 , resistant to PDT with ALA or MAL, respectively; in both cases, the authors related the morphological changes with increased aggressiveness. No differences were reported in the cell cycle distribution between P and 10 th G cells in any of the lines, in agreement with previous studies in SCC-13 17 . However, size and complexity decreased in CSZ 10 th G cells, as well as size in ASZ and BSZ 10 th G. Conversely, it has been described an increase in size in RIF-1 murine fibrosarcoma cells resistant to Photofrin II-PDT 15,34 , and in LM3 resistant to ALA-PDT 16 ; and in cancer cells resistant to chemotherapeutic agents 35,36 . In addition, lung cancer cells smaller in size appeared to be more sensitive to PDT with a benzoporphyrin derivative 37 . We consider that this lack of concordance between our results and those described by other authors, could be due to the different biology of BCC, an indolent and rarely metastatic tumour.
Proliferation analyses showed that ASZ (p53+) presented higher plating efficiency, generating more colonies, but of small size; while colonies of BSZ and CSZ (p53−) were larger, revealing higher proliferation ability. These results highlight the role of p53 in promoting adhesion to substrate and controlling proliferation 38,39 . No differences were seen in the total number of colonies between P and 10 th G for any line, but we observed variable results related with the size in BSZ and CSZ: P cells formed a higher number of larger colonies than their corresponding 10 th G. From the aforementioned results, it can be concluded that 10 th G populations are less proliferative than their corresponding P but, in general, do not change neither the cell cycle dynamics nor the complexity.
Two relevant factors directly implicated in the efficacy of PDT are the production and the subcellular localisation of PpIX 9,16,31,40,41 . We have indicated no differences in PpIX production, between P and 10 th G in ASZ and BSZ populations, supporting the results obtained in SCC-13 cells 17 . However, CSZ 10 th G produced more PpIX than P after 24 h of MAL incubation, although the fluorescence signal was lower in the spindled than in the polygonal population; this difference could confer resistance to PDT in the former population. Supporting these results, primary murine keratinocytes treated with ALA accumulated more PpIX than Pam212 tumour cells 42 . In addition, the intracellular localisation has direct effects on the mode of cell damage 9,43 . Due to limited average life (40 s) and action radius (20 nm) of 1 O 2 , the main ROS generated after MAL-PDT, different cell structures (e.g. mitochondria, lysosomes, plasmatic membrane and nucleus) in which PS is located represent primary targets 9,11,43,44 . Our results indicate that PpIX is localised on the cell membrane and also in the nucleus of ASZ and BSZ P cells, being related with a higher sensibility to PDT 45 (as measured by MTT and AO-EB assay). Previous reports did not describe changes in localisation of PpIX between P and resistant RIF-1 cells -with ALA or Photofrin-46 nor SCC-13, localizing in the membrane 17 . In conclusion, the sensibility of P cells to PDT in ASZ and BSZ is favoured by nuclear PpIX localisation and in CSZ by its higher production.
Our results indicated that both P and 10 th G cells were capable of generating tumours in mice after its inoculation with Matrigel. In addition, the tumours induced by 10 th G were bigger than those induced by P cells. This could point to a larger cancer stem cell component in the resistant populations 47 ; although this should be confirmed. Similar results were obtained with the squamous cell carcinoma SCC-13 17,48 and the mammary carcinoma 4T1 49 . Regarding to the type of tumours developed, no relevant histological differences were observed, except for the presence of osteoid tissue in the tumours generated from BSZ P, CSZ P and CSZ 10 th G. These histological features presented similarities to one uncommon human BCC subtype, the basal cell carcinosarcoma, very rare in patients and with an equivalent metastasis capacity to conventional BCCs. However, we have not found conclusive differences between P and resistant cells 50 .
There are many evidences that connect resistance to anticancer therapies to the EMT process 51,52 . The EMT process is linked to morphological changes from epithelial/polygonal to fibroblastoid/spindled; phenotypical changes -loss of cell-cell adhesion mediated by E-cadherin-and an increase in the expression of mesenchymal markers, including vimentin or N-cadherin. It has been described that, among other factors, EMT can be promoted by the activation of Wnt/β-catenin pathway [53][54][55] . As mentioned before, the morphology of 10 th G population of ASZ and BSZ underwent a slight elongation process and a new spindled-shaped cell population was observed in the cultures of CSZ 10 th G. These changes could indicate an increased aggressiveness in the resistant BCC populations.
The best-studied indicator of EMT is the loss or decrease of E-cadherin expression, which confers a higher invasive capacity 56,57 . This was observed in the BCC resistant populations, except for the ASZ line. Lower expression of E-cadherin has been associated to more aggressive subtypes of BCC [58][59][60] . In the case of ASZ, neither P nor 10 th G presented expression of E-cadherin, but it was expressed on the membrane and cytoplasm of P T cells, indicating a phenotypical change of ASZ P in the in vivo model.
Loss of E-cadherin is, in general, accompanied with expression of mesenchymal markers such as N-cadherin 61 . In this context, we did not find a total correlation between both proteins. The expression of N-cadherin was only increased in ASZ 10 th G and BSZ 10 th G T, and decreased in BSZ 10 th G. Other authors have also related SCC progression with a decrease of E-cadherin expression but not with changes on N-cadherin expression 62 . Regarding (2019) 9:4835 | https://doi.org/10.1038/s41598-019-41313-y www.nature.com/scientificreports www.nature.com/scientificreports/ vimentin, there are evidences indicating a relation between its expression and resistance to therapy depending on the tumours 63,64 . However, we have not found differences in its expression between the different populations, indicating that this protein may not be related with BCC resistance to PDT.
Since the EMT process is promoted by the activation of Wnt/β-catenin pathway, expression levels by WB and subcellular localisation of two proteins involved in this route were studied: β-catenin and Gsk3β kinase [65][66][67] . β-catenin was located in the membrane and cytoplasm of all ASZ populations and in ASZ P T was also detected in the nucleus. In resistant populations of BSZ and CSZ, its expression was lower than in P; and in BSZ 10 th G cells was clearly also located in the nucleus. These data confirmed the lower adhesion of tumour resistant populations (10 th G T) of the three cell lines. Besides, the loss of β-catenin expression in the membrane has been related to a bad prognosis in colorectal cancer 68 . In any case, these observations do not correlate to those described in SCC-13 cells, where no differences in E-cadherin and β-catenin expression patterns were reported 69 . Even so, Casas et al. 70,71 described the deregulation of E-cadherin/β-catenin complex in LM3 resistant cells, but no decrease in protein expression levels was found. Regarding Gsk3β, its expression has been reported to be decreased in cells with constitutive activation of the pathway, as it occurs in most solid tumours 72,73 . In this study, we show that Gsk3β expression increased in resistant populations of ASZ and decreased in those of BSZ and CSZ (particularly in the spindled cells). Moreover, Gsk3β was located in the nucleus of BSZ 10 th G cells, which has been related to apoptosis inhibition 74 and replicative senescence induction 75 . The role of this protein in the resistance processes is not well defined. Whereas an increase in its expression has been linked to gemcitabine and radiation resistance in pancreatic cancer cells 76 , decreased expression has been reported in lung adenocarcinoma cells resistant to cisplatin 77 .
All these observations, together with the evidences that relate the lack of p53 with the increase in expression of components of the Wnt/β-catenin pathway, which favour EMT 38,78-81 , allow us to hypothesize two mechanisms of BCC resistance to MAL-PDT.
BSZ and CSZ cell lines lack expression of p53 whereas ASZ cell line expresses the gene at the RNA and the protein level. Therefore, in the case of ASZ, in which point mutations have been described, it is possible that (1) the gene in not completely inactivated by the mutations and retains some wt function, or (2) the mutations confer advantages in promoting tumorigenesis. In relation to (1), whereas the majority of tumour suppressor genes (RB, APC or BRCA) are inactivated during cancer progression, by mutations, the p53 gene is often found to undergo missense mutations. In relation to (2), there is growing evidence that mutant p53 have lost wt p53 tumour suppressor activity and gained functions that contribute to malignant progression 82 . In addition, in vivo experiments showed that mice expressing mutant p53 display a tumour profile that is more aggressive and metastatic than p53 null or wild-type mice 83 . From the results obtained, we cannot indicate what occurs in ASZ, further studies must be done. In any case, the results indicate that resistant ASZ cells, exhibited differential characteristics to those observed in BSZ and CSZ. p53 expression (ASZ) favours the decreased expression of β-catenin and increased that of Gsk3β, repressing Wnt/β-catenin pathway, especially in resistant populations. Cell adhesion is lower in this line and the EMT process would be associated with an increase in N-cadherin expression in resistant populations. In the absence of expression of p53 (BSZ and CSZ), Gsk3β levels decrease and nuclear β-catenin becomes more evident in resistant cells, which seems to indicate a higher activity of the Wnt/β-catenin pathway. The EMT process, in this case, is associated to decreased adhesion (lower expression of E-cadherin and β-catenin in the membrane), but not with higher expression of mesenchymal proteins such as N-cadherin, except for the BSZ 10 th G, in which this protein is increased in relation to P cells.

Materials and Methods
Cell lines and culture. The study has been performed by using cell lines (kindly provided by Dr. Epstein's laboratory) obtained from BCCs induced in a ptch1 +/− mouse by UV irradiation (ASZ001, ASZ); in a ptch1+/−, K14CreER2/+; p53fl/fl mouse exposed to γ radiation (BSZ2, BSZ) and from a spontaneous tumour developed in a ptch1+/−, K5-CrePR, p53fl/fl mouse (CSZ1, CSZ) 19,20 . All cell lines were cultured in DMEM (Thermo Scientific, Hyclone) supplemented with 10% foetal bovine serum (FBS) (Thermo Scientific, Hyclone) and 1% of Penicillin/Streptomycin (Thermo Scientific, Hyclone). The cells were grown at 37 °C in an atmosphere with 5% of CO 2 . pDt and generation of resistant populations. The procedure followed for obtaining PDT resistant populations was based in those previously published [15][16][17][18] . For PDT, MAL (Sigma) was used as a precursor of the endogenous PS PpIX. Cells seeded in F12.5 were incubated for 5 h with different concentrations of MAL in FBS-free DMEM and thereafter subjected to red light irradiation. The light source employed was a red LEDs lamp (635 nm) (Segainvex, UAM). Treatment conditions that caused survival rates of 5-15% were chosen (tested by MTT assay). The selected initial treatment conditions that induced a survival rate lower than 15% were: 0.3 mM, 0.4 mM or 0.2 mM MAL for ASZ, BSZ or CSZ, respectively and a light dose of 2 J/cm 2 . For the rest of the PDT cycles, the concentration of MAL was maintained and light doses were increased if necessary, to obtain surveillance between 5-15%. After each cycle, surviving cells were allowed to reach 50-60% of confluence before applying a new PDT treatment 15,17 . The final population received a total of 10 cycles of PDT. The initial population, not subjected to PDT, was called parental population (P), and 10 th G refers to the population resistant to 10 cycles of PDT. For the experiments, P and 10 th G cells were used until 7 passages. Cytotoxic assays. The evaluation of the cytotoxic effect of PDT was performed by the MTT assay (Sigma).
Cell viability was tested 24 h after each PDT treatment by incubating the cultures with 500 µg/mL of a MTT solution for 3 h. Formazan crystals were dissolved in DMSO (Panreac) and the optic densitometry was measured in a Spectra Fluor, Tecan reader plate at 542 nm. Cells with neither drug nor light exposition were used as controls and their viability rate was expressed as 100%. From the survival values obtained from a determined irradiation Scientific RepoRts | (2019) 9:4835 | https://doi.org/10.1038/s41598-019-41313-y www.nature.com/scientificreports www.nature.com/scientificreports/ conditions (1.36 or 2.36 J/cm 2 ) fold-change indexes were calculated as the ratio between the % of survival of each 10 th G and P cell type (Fold-change = %10 th G survival/%P survival). The experiments were performed, at least, three times.
Cell death was also evaluated by the acridine orange (AO, Sigma) and ethidium bromide (EB, Sigma) assay that allowed distinguishing between viable and dead cells. After 24 h of MAL-PDT, AO and EB were added to the cultures at a final concentration of 50 µg/ml. Immediately after staining the cells were observed in the fluorescence microscope under blue excitation light (450-490 nm, BP 490). According to the fluorescence colour observed, cells were classified as viable and dead fluorescing in green or orange, respectively. tumorigenic assay and Isolation of BCC keratinocytes. For the tumorigenic assays, 8 week-old Athymic Nude-Foxn1 nu mice (Envigo, France) were used. Mice were classified randomly in 3 groups with 5 mice per group. Each mouse was inoculated in the left flank with 1.5 × 10 6 of P cells in 50 µL of PBS and 50 µL of Matrigel (Corning), and in the right flank with the same number of 10 th G cells. During the subsequent days, the animals were monitored, measuring the progressive increase of tumour size with an automatic calliper. The sum of all the different lobules was considered the tumour volume, and it was calculated with the following formula: When the tumour reached the maximum established volume of 500 mm 3 , mice were sacrificed with CO 2 . Tumours were surgically removed and divided into 2 pieces, one for histopathological evaluation and the second one for tumour keratinocytes isolation. In the first case, pieces were fixed with 3.7% formaldehyde (Panreac) in PBS (Thermo Scientific, Hyclone), washed in PBS and included in paraffin. In the second case, tumour cell cultures were performed by explant. For this end, the tissue was washed with 96% ethanol (Panreac), then in ethanol/ PBS 1:1 and three times with PBS in sterile conditions. After that, the tumours were mechanically disaggregated and seeded on 35 mm Petri plates with DMEM supplemented with 1% P/S, 10% FBS and 1% amphotericin B. When keratinocyte colonies started to form, they were isolated by trypsinization and amplified. Cells obtained from tumours induced by P or 10 th G cells were named P T or 10 th G T, respectively 84 .
All methods were carried out in accordance with relevant guidelines and Spanish regulations. All experimental protocols were approved by the Committee of ethical investigation of Autonomous University (CEI-85-15809) and the Committee of ethic in human and animal experimentation of CSIC (Centro Superior de Investigaciones Científicas) on the 15 th of June 2015 (number 280790000188).
Cell morphology and immunofluorescence. To analyse the cell morphology, cells were cultured over coverslips and observed on a phase contrast microscopy connected to a CCD DP70 camera (Olympus BX-61).
size, complexity and cell cycle. Size and complexity parameters and cell cycle distribution were analysed by flow cytometry (Cytomics FC500, 1 laser, Beckman Coulter). For that, cells were trypsinized and washed with PBS by centrifugation. Size evaluation was made based on forward scatter and the complexity was evaluated by side scatter. For cell cycle analysis, DNAprep kit (Beckman Coulter) was used. The pellet obtained by centrifugation was resuspended in 50 µL of detergent of the kit, and 1 mL of propidium iodide, incubating 30 min at 37 °C. Cells were maintained at 4 °C in dark until performing the measurement.
Cell proliferation. Proliferation rate was estimated using the clonogenic assay. Cells were seeded at 125 cells/mL and they were let to grow for 7 days. Then, cultures were washed with cold PBS (4 °C) and stained with 0.2% crystal violet (Aldrich Company, Inc) in 2% ethanol in distilled water for 20 min in constant shaking at room temperature. Thereafter, the cultures were washed with PBS, plates were let to dry and colonies (each colony is formed by >50 cells) were counted in number and classified in groups by their size as: small (<1.5 mm), medium (>1.5 mm; <2.5 mm) and big (>2.5 mm). Next, cells were lysed and dye was dissolved in 1% SDS (Sodium Dodecyl Sulfate) (Panreac); and optic density was measured with a plate reader at 570 nm (Spectrafluor, Tecan). production and subcellular localisation of ppIX. The production of PpIX was evaluated by flow cytometry (FC500 Cytomics 2 lasers, Beckman Coulter) after incubation with the appropriate concentrations of MAL for 5 or 24 h. After MAL incubation cells were trypsinized, centrifuged 7 min at 481 g and fixed with 1 mL of 3.7% formaldehyde in PBS at room temperature for 10 min. Fixing solution was removed washing the cells with PBS twice by centrifugation, resuspended in clean PBS and kept in the dark at 4 °C until evaluation. PpIX emission measurements were obtained employing the flow cytometer Cytomics FC500 (λexc = 625 nm; λem = 670 nm). Fluorescence intensity was determined for 10 4 cells per each cellular population. PpIX subcellular localisation was determined by fluorescence microscopy using the UVA excitation line. For that, cells were seeded on cover slips and incubated or 24 h at 37 °C with different concentrations of MAL in DMEM supplemented with 1% FBS. Then,