Expression profiles of proton-sensing G-protein coupled receptors in common skin tumors

The proton-sensing GPCRs (pH-GPCRs) GPR4 (GPR19), TDAG8 (GPR65, T-cell death associated gene 8), OGR1 (GPR68, ovarian cancer GPCR1), and G2A (GPR132, G2 accumulation protein) are involved in sensing and transducing changes in extracellular pH (pHe). Extracellular acidification is a central hallmark of solid cancer. pH-GPCR function has been associated with cancer cell proliferation, adhesion, migration and metastasis, as well as with modulation of the immune system. Little is known about the expression levels and role of pH-GPCRs in skin cancer. To better understand the functions of pH-GPCRs in skin cancer in vivo, we examined the expression-profiles of GPR4, TDAG8, OGR1 and G2A in four common skin tumors, i.e. squamous cell carcinoma (SCC), malignant melanoma (MM), compound nevus cell nevi (NCN), basal cell carcinoma (BCC). We performed immunohistochemistry and immunofluorescence staining on paraffin-embedded tissue samples acquired from patients suffering from SCC, MM, NCN or BCC. We show the expression of pH-GPCRs in four common skin cancers. Different expression patterns in the investigated skin cancer types indicate that the different pH-GPCRs may have distinct functions in tumor progression and serve as novel therapeutic targets.


Results
We summarised our findings of immunohistochemistry data for individual SCC, MM, NCN and BCC tumors from each 5-6 patients. In order to present more data, we performed additional immunohistochemistry on TMA-format including 24-27 samples per tumor type. The tissue microarray analysis (TMA). TMA revealed that while only 4% of the SCC samples were strong positive for GPR4, 92% showed a weak GPR4 expression and 4% were GPR4 negative (Fig. 6a, Fig. S18). MM revealed strong positive results, especially in the epidermal areas of the tissue (94.7%) compared to the dermal section (52.6%) (Fig. 6d). NCN epidermal and dermal parts appeared both weak positive (epidermal: 61.1%, dermal: 87.5%) (Fig. 6c). In the dermal portions of MM and NCN especially the giant tumor cells appeared to be strong positive. 47.8% of BCC tissue samples strongly expressed GPR4 (Fig. 6b). Other TMA BCC tissue exhibited a weak positive expression (47.8%) or were negative for GPR4 (4.4%). For most instances, the TMA supported the trend of results described for the combined IHC and IF data, except of the BCC where one case not expressing GPR4 was observed in the TMA. TDAG8 (GPR65). IHC  www.nature.com/scientificreports/ TMA. In the TMA, 87% of the SCC samples showed a weak positive expression, supporting the other immunostaining results (Fig. 6a, Fig. S18). The majority of MM strongly expressed TDAG8, where the epidermal section was strong positive in 78.3% and the dermal part in 69.6% (Fig. 6d). In contrast, NCN tumors appeared strong positive (57.9%) in their dermal section, while the epidermal portion was predominantly only weak positive (64.7%) (Fig. 6c). Large NCN tumor cells were strong positive for TDAG8 in both, dermal and epidermal tissue. 40.9% of the BCC cells were strong positive and 59.1% were either partial or weak positive (Fig. 6b).
Overall, the TMA corresponded well with the previous results, although the TMA did not show any negative results for TDAG8 in BCC.

OGR1 (GPR68). IHC and IF.
The evaluation of the OGR1 (GPR68) expression-profiles based on IHC showed that 20% of SCC tissue samples were weak positive for OGR1, while the other 80% showed no expression of OGR1 (Figs. 1c,g,k, 5a, and Supplementary Figs. S1-S4 third column). 40% of MM epidermal tissue samples strongly expressed OGR1, 40% were weak positive and the other 20% were negative for OGR1. In contrast, 80% of the MM dermal sections showed a partial or weak positive expression and 20% showed no expression ( Fig. 2c,g,k, 5d and Supplementary Figs. S5-S8 third column). 60% of the NCN epidermal tissue samples were weak positive and the other 40% strongly expressed OGR1. Dermal areas in NCN tissue samples revealed strong expression of OGR1 in 20% of the samples, partial expression in 60% of the cases and no expression in 20% of the tested samples (Figs. 3c,g,k, 5c and Supplementary Figs. S9-S12 third column). 33.3% of the BCC samples (Figs. 4c,g,k, 5b and Supplementary Figs. S13-S17 third column) were weak positive and the other 66.6% were negative for OGR1. www.nature.com/scientificreports/ TMA. 13% of the SCC cells were strong positive and 43.5% were weak positive, while 43.5% did not express OGR1 at all (Fig. 6a, Fig. S18). The epidermal part of MM was mostly positive for OGR1 (95.5%), whereas the MM dermal section appeared predominantly weak positive (63.6%) (Fig. 6d). Dermal MM cells appear to be partially strong positive. 64.3% of the NCN epidermal portion was OGR1 negative, while the NCN dermal part reached 100% weak positive results (Fig. 6c). About 75% of the BCC cells expressed OGR1 positive, while 25% did not express this GPCR (Fig. 6b). Taken together, the TMA correlated with the results of the IHC/IF except of the BCC. According to the TMA data, BCC were more likely positive for OGR1.  TMA. 83.4% of the SCC cells showed a weak G2A expression, 8.3% of the samples revealed strong positive expression and the other 8.3% were negative for G2A (Fig. 6a, Fig. S18). 52.4% of the MM epidermal section expressed G2A strongly and 85.7% of the dermal portion of MM had a weak positive expression (Fig. 6d). Giant MM tumor cells appeared to be strong positive in the epidermal and dermal parts. The NCN dermal part appeared to be mostly weak positive (88.2%) in contrast to the NCN epidermal zone, which revealed more negative results (33.3%) (Fig. 6c). 60% of the BCC cells expressed G2A weakly, whereas 40% of BCC showed strong expression (Fig. 6b). The TMA results fully confirmed the IHC/IF results.

Discussion
In   www.nature.com/scientificreports/ except of two tissue samples. GPR4 was found to be overexpressed in several human cancers 31 . GPR4-deficient mice showed a significantly reduced angiogenic response to VEGF, which accordingly led to a reduction in tumor growth in orthotopic models 29 . Acidification-activated GPR4 in endothelial cells increased the expression of a number of inflammatory genes and promoted angiogenesis in head and neck cancer, likely via secretion of angiogenic factors 32 . Regarding cancer cells themselves, it was shown that ectopic expression of GPR4 in murine 3T3 cells induced malignant transformation 33 . In contrast, GPR4 overexpression in B16F10 melanoma cells inhibited their acidic pH-induced migration, invasion and metastasis formation 22 . Taking this knowledge and our results into account, GPR4 might be an indicator of dysplasia of dermal melanocytes similar to HMB45. The latter is also found to often remain positive in deep dermal portions of dysplastic nevi or melanoma while expression levels decrease with increasing depth in normal nevi.

TDAG8 (GPR65).
In all four tumor types investigated in this study expression of TDAG8 was high, except for dermal portions of MM, epidermal portions of NCN and BCC, where TDAG8 occurrence was often only moderate or missing. TDAG8 is predominantly expressed in lymphoid cells and tissues, including peripheral blood leukocytes, spleen, lymph nodes, and thymus and has also been detected in some selected cancers 31,34 .
Overexpression of TDAG8 in lung carcinoma cells was associated with enhanced tumor development and cancer cell survival under acidic conditions 33 . Ectopic TDAG8 expression malignantly transformed a normal mammary epithelial cell line and led to ligand-independent activation of SRE and CRE promoter-driven gene transcription in HEK293 cells 31 . On the other hand, ectopic overexpression of TDAG8-GFP fusion protein enhanced apoptosis and sensitivity to dexamethasone-induced apoptosis in lymphoma cells 35 . TDAG8-deficiency in different KO www.nature.com/scientificreports/ mouse models was associated with an exacerbation of inflammation in selected pathologies [36][37][38] . In summary, it seems that TDAG8 attenuates immune-mediated inflammation, while the overall effect on non-blood-cell tumor cell behaviour remains less clear. Based on this knowledge, we hypothesize that the high expression of TDAG8 in the investigated skin tumors might be crucial for tumor growth and/or tumor cell survival. However, another possible mechanism could be that TDAG8 acts as a tumor suppressive receptor to control tumor growth under acidic conditions. These questions have to be addressed with cell culture experiments.

OGR1 (GPR68).
Regarding the tumor tissue analysed in this study, OGR1 is not (IHC) or only moderately (TMA) expressed in SCC and BCC, but particularly present in MM and NCN. Overexpression of OGR1 in human prostate and ovarian cancer cells mediated an inhibitory effect on cell migration and metastasis 27,28,39 .
In addition, OGR1 overexpression in ovarian cancer cells also inhibited cell proliferation, while increasing cellmatrix adhesion 27 , suggesting a tumor-supressing effect of ORG1. In contrast, when the host cells of ORG1 knock-out mice were depleted of ORG1 the tumorigenesis of injected melanoma cells and prostate cancer cells was decreased 40,41 , indicating tumor-promoting function of ORG1 in the host organism. In other cell types ORG1 expression and acid stimulation was associated with the expression of inflammatory and immune modulatory factors [42][43][44][45] . These findings state that the high expression of ORG1 in the mesenchymal tumors NCN and www.nature.com/scientificreports/ MM clearly differs from the expression in the epithelial tumors SCC and BCC. Further cell culture experiments are needed in order to study the exact effect of OGR1 in the skin.

G2A (GPR132).
In skin G2A (GPR132) is proposedly expressed in keratinocytes, fibroblasts, epidermal cells and melanocytes 14 . The incidence of G2A in the investigated tumors was high in both, epidermal and dermal parts of MM, NCN and BCC, but less frequent in SCC. G2A is predominantly expressed by different immune cells 46,47 . G2A was identified as a stress-inducible gene, activated by genetic recombination processes in immature B lymphocytes and developing thymocytes, or by exposure to DNA-damaging stress, like UV, X-ray, etoposide or doxorubicin 48 . G2A expression led to cell cycle arrest and attenuated transformation potential of oncogenes 48 . The important role of G2A in controlling immune cell homeostasis was supported by the finding that G2A-deficiency caused autoimmune syndrome in ageing G2A-depleted mice 49 . However, expression of G2A in other cell types appears to have oncogenic potential, as high-level expression of G2A in NIH3T3 cells induced malignant transformation 50 . In human epidermal keratinocytes, G2A mediated the secretion of cytokines, and induced cell cycle arrest 46 . UVB radiation and H 2 O 2 enhanced G2A expression in HaCaT cells, indicating that G2A might function as sensor for DNA damage and oxidative stress in keratinocytes 51 . With its high expression in skin tumors MM, NCN, SCC and BCC, G2A might play a pivotal role as an immune checkpoint of the tumor. Regarding the expression of pH-GPCRs in different skin tumors, TMA results (Fig. 6) reveal that the overall expression of all four GPCRs increases in MM compared to NCN. Especially the incidence of strong positive expression of the pH-GPCRs is increased in both, dermal and, even more pronounced, in epidermal portions. Thus results suggest that an increase in pH-GPCR expression in MM could be a marker for increased malignancy, which requires, however, further investigation.
The prevailing hypothesis, that influencing factors such as type of cancer, the micro-and macro environment as well as the variation between every human individual influence individual pH-GPCR expression, can be supported with the results of this study, containing a large data set for the four most common skin cancers.
In summary, the current evidence on the expression of pH-GPCRs in tumors is still only a first step towards understanding the role of pH-GPCRs and their function as transmembrane messengers of extracellular pH in cancer development or control. Further functional studies are undoubtedly required to fully understand the individual role of each pH-GPCR in the development and progression of different skin cancers.
The cancer type-specific differential expression of individual pH-GPCRs underpins their potential value in the field of cancer therapy. Our investigations may lead to more specific cell culture studies of the pH-GPCRs in different skin tumor cell lines and their use as a potential therapeutic target.

Materials and methods
Tissue samples. For all experiments, we used tissue samples older than 10 years from the department of Dermatology at the University Medical Center Regensburg (IHC/IF: n = 5, exception: BCC n = 6; TMA: n = 24-27). Routine paraffin-embedded skin biopsies obtained from affected areas of patients with localized skin tumors were used anonymized. The diagnosis of localized tumors had been previously confirmed histologically by a dermatopathologist. Handling of human skin tumor biopsies older than 10 years was approved by the ethical committee of the University of Regensburg. Under German law the tumor tissue left after surgery after the final diagnosis can be discarded after 10 years or are free to use.
General data regarding the tissue sample origin is given in Supplementary Table S1. The following tissue types served as positive controls: primary human tonsil tissue and lung tissue for G2A and TDAG8, endosomal membrane of the testis and lung for OGR1 and pancreas as well as the endosomal membrane of the lung for GPR4. As negative control tissue we used liver for GPR4, heart muscle for GPR65, pancreas for GPR68 and ovary for GPR132. Respective images from IHC staining on control tissue including secondary antibody controls and isotype antibody controls are shown in supplementary figure S19. Additional antibody controls on dermal tumor tissue are shown in supplementary figure S20.

Immunohistochemistry (IHC).
Hematoxylin and eosin (HE)-stained, paraffin-embedded and fixed tissues as well as positive and negative controls were freshly cut into 2 μm thin pieces and superimposed on slides. Tissue sections were incubated for 30 min at 72 °C before they were rehydrated by washing with alcohol solutions at descending concentrations as follows: 2 × xylol for 5 min, 2 × 100% ethanol for 5 min, 2 × 96% Ethanol for 5 min, 2 × 70% ethanol for 5 min. Endogenous peroxidase was neutralised by incubation with 3% H 2 O 2 (Fisher Scientific GmbH, Schwerte, Germany) for 10 min. Afterwards, the slides were washed in distilled water and were submersed in precooked citrate buffer (boiled for 30 min) (Zytomed Systems GmbH, Berlin, Germany) for 20 min. After cooling the sections on ice, they were incubated in phosphate-buffered saline (PBS) (Sigma-Aldrich, St. Louis, United States of America) for 10 min at RT. Subsequently, sections were clamped to Shandon coverplate immunostaining chambers Fisher Scientific GmbH, Schwerte, Germany) and transferred to PBS. Samples were incubated for 10 min at RT in a blocking solution (ZytoChem Plus HRP Kit/Rabbit, Zytomed Systems GmbH, Berlin, Germany) in order to minimise unspecific binding of antibodies.
In the following, tissue sections were treated with polyclonal primary antibody (rabbit anti-human GPR4 www.nature.com/scientificreports/ three washes with PBS, samples were incubated with streptavidin-HRP-conjugate for 20 min at RT, followed by another washing step with PBS. Finally, chromogen solution AEC plus (Dako, Glostrup, Denmark), was added. The reaction was stopped by several washes with distilled water as soon as the positive controls showed distinct staining. Mayer's Haemalm (Carl Roth GmbH & Co., Karlsruhe, Germany) was used to counterstain the tissue. Samples were embedded with Aquetex mounting medium (Merck KGaA, Darmstadt, Germany). Specimen were inspected with a Leitz Wild Biomed microscope (Leica Microsystems GmbH, Wetzlar, Germany, Type: 020-507.010) and afterwards scanned with the PreciPoint M8. Digital images were edited using the analysis software ViewPoint online (PreciPoint, Freising, Germany). Images were evaluated via visual inspection. Scores were assigned for ++: strong positive/positive reaction; +: weak positive/partial positive reaction; −: negative reaction.
Tissue microarray (TMA). The immunohistochemical multiple-labelling tissue microarray (TMA) allows for simultaneous IHC staining of multiple tissue samples. Representative tumor material from 24-27 tissue samples per tumor type was assembled into a paraffin matrix (5 × 6) with 1 mm diameter spots. Samples on the TMA tissue slide were subjected to IHC staining following the protocol above.
Immunofluorescence (IF). Samples were incubated in the heating cabinet for 20 min at 70 °C and rehydrated with the descending order of the alcohol concentration as described above. The slides were washed with PBS and subsequently incubated in citrate-tris-EDTA-buffer (Zytomed Systems GmbH, Berlin, Germany) for 25 min. The sections were cooled on ice for 25 min and then submersed in PBS for 10 min. The samples were incubated for 15 min in tris-glycine-buffer (Trishydroxymethylaminomethan: Merck KGaA, Darmstadt, Germany; Glycine: Merck KGaA, Darmstadt, Germany) to reduce autofluorescence. After three rinses with PBS, samples were incubated with 5% BSA (Sigma-Aldrich, St. Louis, United States of America) in tris-buffered saline with 0.1% Tween-20 (TBST) (Tween-20: Carl Roth GmbH & Co., Karlsruhe, Germany) for 60 min in order to block unspecific binding of antibodies. Tissue sections were incubated with primary antibody (GPR4 1:100, GPR65 1:300; GPR68 1:250, GPR132 1:60) in phosphate-buffered saline 1% Tween-20 (PBST) at 4 °C overnight. Afterwards, samples were washed three times for 15 min. Alexa-594-conjugated goat anti-rabbit specific second antibody (Life Technologies, Carlsbad, United States of America, A11037) was diluted with 1% BSA in PBST (1:1,000) and added to the slides for 30 min. Afterwards, tissue sections were rinsed once with PBS for 15 min, and finally stained for cell nuclei with 4,5-diamindino-2-phenylindole (DAPI). Fluorescence was exposed with a Zeiss Axio Imager. Ethics. All experiments were done in accordance with the declaration of Helsinki. No identifying data of tissue donors were used during experiments or in the paper. All tissue samples were from patients older than 18 years.