All tested antibodies exhibited positive immunoreactivity
All tested nmda receptor antagonist exhibited positive immunoreactivity in NSCLC tumors, with N’-epitope being the less expressed (both as overall positivity and number of positive cases, or intensity combined with number of positive cells, as provided by H-Score. This result is in accordance with previous studies , ,  and suggests that full-length ERα might be minimally expressed in lung epithelial cells (both normal and tumor) and further corroborated by the decreased expression of the microarray probe 211627_x_at, targeting ESR1 exon 1 in normal tissues. Lower expression of ESR1 (probe 211627_x_at) has been associated with better response to adjuvant chemotherapy and a positive effect on survival . The lack of immunoreactivity for 6F11 (or 1D5) in bronchial and alveolar epithelial cells in paired non-tumor samples of the same patients further suggests the absence of whole length ERα66 from respiratory epithelial cells. In contrast, the tumor microenvironment (vascular endothelium and infiltrating inflammatory cells) stain heavily for ERα66. The early transcriptional signatures we have identified in treated A549 and H520 cells support this hypothesis. Interestingly, E2 and E2-BSA (membrane acting exclusively) do not trigger the same effects in the two cell lines we have tested. This result could reflect different transcriptional effects mediated exclusively by membrane receptors (E2-BSA) compared to free estradiol effects that also involve different intracellular receptors (located or not at the plasma membrane, as found in our IHC analysis of tumors). The latter might also include cytoplasmic and nuclear ERα variants, but also ERβ (mainly nuclear) receptors. This result is compatible with what we have previously described regarding distinct gene signatures and phenotypic effects of intracellular and membrane bound ERs in breast cancer cell lines , , , , , . The common responses to estrogenic stimuli in A549 and H520 could be indicative of commonly shared ER-mediated pathways in the two cell lines (or types of carcinoma) despite their inherent differences. To further address the identity and localization of receptors that mediate the membranous and intracellular effects, as well as the high heterogeneity of NSCLC tumors, further investigation is required, including more NSCLC lines and binding-competition assays. We also suggest, that the histotype is important, as there are differences between the ADC and SCC cell line, observed in the probesets meta-analysis of NSCLC cell lines. We have identified some pathways, closely related to ER activation including stemness, EMT and metabolic processes. Stemness becomes more evident when tamoxifen is combined with estradiol or estradiol-BSA. This finding is in accordance with our previous report on the early (3h) transcriptional induction of stemness in breast cancer cells following treatment with tamoxifen . The increase of ER transcription in smokers, observed in the probeset metanalysis, could be related to tissue-injury and stemness identified in the functional gene analysis. The conservation of the ligand binding domain and the AF-2 part of the ERα molecule (in the majority of the examined cases) suggests the possible application of these findings in NSCLC treatment. As reported here, addition of tamoxifen in two NSCLC cell lines altered mainly estradiol affected genes and to a lesser extent estradiol-BSA related ones. Hence, for several genes, tamoxifen displayed a similar profile with membrane acting estradiol-BSA. This could suggest that binding of tamoxifen to membrane ERs could trigger partial agonistic effects, whilst the major antagonist effect is observed in intracellular (cytoplasmic and nuclear receptors). Indeed, a previous report suggests that lack of AF-2 part of the receptor leads to transformation of ER antagonists, like fulvestrant, to receptor agonists . Immunodetection of ERα36 provides further insight on estrogenic relevance in the human lung. Non-genomic rapid effects have been described, mediated also by this isoform, in the lung , , . This receptor has been described in murine models, , but also in human NSCLC specimens . As we evidenced ERα36 in a variety of cell types (epithelial, stromal and inflammatory), it is plausible to consider that estrogenic effects could orchestrate diverse autocrine and paracrine functions within lung epithelial and mesenchymal elements, probably mediated though extranuclear signaling. EMT was one of the major functions affected by estrogenic signaling in our transcriptome analysis and could be in line with this hypothesis. Validation of this hypothesis requires additional functional assays in vitro and in vivo. In a previous study in NSCLC specimens, ERα36 was exclusively detected in tumor cells . However, although the same antiserum has been used, the blocking method and primary antibody incubation was different than in our study and could explain (partially) this discrepancy. In previous investigations of our group, we have reported high ERα36 expression by stromal cells and inflammatory infiltrate , ,  associated with anti-inflammatory effects of ERα36. The present study is the first describing ERα36 expression in human lung epithelial cells (both tumor and non-tumor) with membrane/cytoplasmic and nuclear distribution. Recently, we have demonstrated the nuclear translocation of ERα36 in normal human macrophages . If further verified, our data might suggest that in a context of low expression of the WT receptor, ERα36 (and eventually other variants or other estrogen receptors, like GPR30/GPER) lead estrogen signaling .