Dysregulation of the endothelin pathway in lymphangioleiomyomatosis with no direct effect on cell proliferation and migration

LAM is a rare low-grade metastasizing lung neoplasm. Inhibitors of mTOR improve clinical outcome of LAM patients by preventing loss of lung function. Nevertheless, other cell targets may be of interest for drug development. Therefore, we explored the potential role of EDN1 (endothelin) in LAM. We report an increased endothelin blood level in LAM patients as well as EDN1 overexpression and EDN1 receptor downregulation in LAM-derived primary cells and in TSC2NEG cells mutated in TSC2. We evidenced EDN pathway dysregulation based on EDN1, EDNRA, EDNRB and ARRB1 mRNA expression in LAM-derived primary cells. We showed overexpression of EDN1 and ARRB1 mRNAs in TSC2NEG cells; these cells lost their ability to respond to stimulation by endothelin. We analyzed the effects of endothelin receptor antagonists alone or in combination with rapamycin, an mTOR inhibitor, on proliferation and migration of LAM cells. Rapamycin treatment of TSC2NEG cells significantly reduced cell proliferation or migration, while none of the tested inhibitors of EDN receptors impaired these functions. We showed that TSC2NEG cells have acquired a transformed phenotype as showed by their ability to grow as spheroids in semi-solid medium and that unlike endothelin receptors antagonists, rapamycin reduced anchorage-independent cell growth and prevented expansion of TSC2NEG spheroids.

Lymphangioleiomyomatosis (LAM) is a rare pulmonary disease mainly affecting young women 1 . LAM can occur as an isolated disorder, defined as sporadic LAM or in patients with tuberous sclerosis complex, a genetic disease characterized by mutations of the TSC1 and TSC2 (Tuberous Sclerosis Complex) tumor suppressor genes respectively encoding hamartin and tuberin 2 . LAM cells carry mutations of the TSC2 gene 3 , inducing constitutive activation of the PI3K/Akt/mTOR pathway and LAM cell proliferation. LAM causes cystic destruction of the lungs and development of benign renal tumors or angiomyolipomas 1 . Two cell populations are present in LAM lesions: the myofibroblastic-like cells that express markers of smooth muscle cells and fibroblasts, such as α-smooth muscle actin (α-SMA), vimentin and desmin 4,5 and the epithelioid-like cells that express melanocytic markers such as MLANA (Melan A) and proteins evidenced with HMB45 and PNL2 antibodies 5,6 . In LAM patients, circulating VEGF-D (Vascular Endothelial Growth Factor D) is increased in the blood and is associated with lymphangiogenesis, a major pathogenic mechanism in LAM progression 7,8 .
LAM is considered as a low-grade, destructive, metastasizing neoplasm 9 . Circulating LAM cells have been found in the blood, urine and chylous effusions 10,11 of LAM patients. LAM cells invade organs through degradation of the extracellular matrix by metalloproteinases, similarly to metastatic cancer cells 12,13 .
Although mTOR inhibitors (everolimus, sirolimus) have been shown to improve clinical outcomes in preventing loss of lung function 14,15 and have been approved to treat LAM, other pathways must be explored to improve patient treatment. In human cancer cells, high expression levels of EDN1 (Endothelin 1) and of endothelin receptors A and B (EDNRA and EDNRB) are associated with the increase of circulating VEGF and of microvessel density [16][17][18][19] . The EDN1/EDNR/ARRB1 (β Arrestin 1) pathway is implicated in cell proliferation, migration, invasion, survival and angiogenesis in several diseases, among them lung, ovary, prostate and breast cancers 20,21 . The development of endothelin receptor antagonists (ERAs) such as bosentan, a dual EDNRA and EDNRB receptor antagonist, or BQ123 targeting EDNRA, provided targeted treatments for pulmonary arterial hypertension and cancer 22-26 . In this study, we explored the role of EDN1 and of its receptors in LAM-derived primary cells and in angiomyolipoma-derived cells lines. We report an increased blood level of endothelin in LAM patients as compared to controls, and the overexpression of EDN1 and downregulation of its receptors in LAM-derived primary cells as well as in TSC2 NEG cell lines. We analyzed the effects of ERAs, alone or in combination with rapamycin, on LAM cell proliferation and migration.

Materials and Methods
Cell lines. The 621-101 TSC2 NEG and 621-103 TSC2 POS cell lines (respectively named "TSC2 NEG " and "TSC2 POS " along our study) were generously provided by Pr E.P. Henske (Boston, United States) 27 . The TSC2 NEG cell line was derived from a renal angiomyolipoma of a LAM patient. They carry a missense mutation in exon 16 of the TSC2 gene (G1832A) leading to a loss of heterozygosity. The TSC2 POS cell line has been developed by re-expression of normal TSC2 gene in the 621-101 TSC2 NEG cells. These cell lines were cultured in DMEM medium (Sigma) supplemented with 10% inactivated fetal calf serum (Gibco), 100 U/mL penicillin (Sigma), 100 μg/mL streptomycin and with 50 μg/ml zeocin (Thermo Fisher) for the TSC2 POS cells to maintain the selective pressure for TSC2 expression.
Human primary PASMC (Pulmonary Artery Smooth Muscle Cells) (Lonza) were used as controls and maintained for a short time in culture as recommended.
Circulating EDN1 and VEGF-D in LAM patients. Sera from 28 LAM and 12 control patients with unrelated disorders (6 multiple sclerosis, 6 myocardial infarction) were obtained from the Cardiobiotec biobank. EDN1 and VEGF-D were measured in the serum of LAM patients and controls using "solid-phase ELISA Quantikine Endothelin-1" and "VEGF-D immunoassay" (R&D systems) as recommended.
Cell migration assay. Forty-eight hours serum-deprived TSC2 NEG and TSC2 POS cells were plated in the top chamber of 8 μm pore diameter Fluoroblock cell culture inserts (Corning) in serum-free medium, while the bottom chamber was filled with DMEM medium supplemented with 10% fetal calf serum. Cells were cultivated with 10 nM rapamycin, 10 µM BQ123 or 10 µM bosentan. Fifteen hours after plating, cells were stained Expression of EDN1, EDNRA, EDNRB and ARRB1 mRNAs. Total cellular RNAs were extracted with the "PureLink RNA mini kit" (Ambion), treated with the "Turbo DNA-free kit" (Ambion), and reverse transcribed with the "iScript cDNA synthesis" kit (Biorad). The mRNA expression was quantified in triplicate with the "PCR KAPA SYBR FAST qPCR" kit (Clinisciences) using 20 ng cDNA. Primer efficiency was determined by amplification of 10-fold dilutions of cDNAs. PPIA (Peptidylprolyl Isomerase A), YWHAZ (Tyrosine 3 monooxygenase) and RPL32 (Ribosomal Protein L32) ( Table 1) have been selected as reference genes based on their stability in LAM samples evaluated with the geNorm software. Normalized relative expression was calculated with the 2 −ΔΔCq method 28 .

Statistical analysis.
Normalized gene expression was done with REST 2009 software (Qiagen). Statistical analysis was done with SPSS Statistics (Version 21, International Business Machines, USA) and PRISM software (GraphPad). Quantitative variables were compared with Mann-Whitney tests. All tests were done with a significance threshold of α = 0.05.

Results
Elevated levels of circulating EDN1 and VEGF-D in LAM patients. EDN1 and VEGF-D levels were measured in sera obtained from 28 LAM patients and 12 female controls with unrelated disease. The circulating level of VEGF-D was significantly higher in LAM patients compared to controls (Fig. 1) with a VEGF-D level above 800 pg/ml in 82% (23/28) of LAM patients (and over 4000 pg/ml VEGF-D for four patients). The level of circulating EDN1 was significantly higher in LAM patients when compared to controls, suggesting a stimulation of the EDN1 production in LAM patients.

Gene
Approved name 5′-3′ sequences (Forward Reverse) Characterization of LAM-derived cells. We generated lung-derived primary cells from five LAM patients. We observed mixed cell populations in the cultures, with mostly spindle-shaped cells and rare epithelioid-like cells ( Fig. 2A), their respective proportion being patient-dependent. Primary LAM cells expressed α-SMA, vimentin and rarely desmin ( Fig. 2A) and they were hardly positive for the melanocytic markers HMB45 and PNL2 (not shown). LAM cells expressed PMEL (pre-melanosome protein) but not MLANA (Melan-A) mRNAs (Fig. 2B). All but the cells derived from patient #2634 expressed TYRP1 (Tyrosine related protein 1)   (Fig. 2B). Consistent with the protein expression, ACTA2 (Actin alpha 2) mRNA was expressed in all cells. To summarize, the primary cells derived from LAM patients were consistent with myofibroblast-like cells expressing some of the melanocytic markers.
We morphologically characterized the TSC2 NEG and TSC2 POS cell lines; while the TSC2 NEG cells were spindle-shaped, the TSC2 POS cells were polygonal ( Fig. 2A). TSC2 NEG and TSC2 POS cells expressed α-SMA, vimentin and barely desmin ( Fig. 2A), but did not express melanocytic proteins as demonstrated by the absence of staining with HMB45 and PNL2 antibodies (not shown). However, both cell lines expressed mRNAs from melanocytic (PMEL and TYRP1) and myofibroblastic (ACTA2) genes (Fig. 2B). TSC2 NEG and TSC2 POS cell lines were therefore phenotypically compatible with LAM cells.
Because LAM is often associated with TSC mutations, pulmonary LAM cells were screened for mutations in DNA regions corresponding to exons 4041 and intron 5-exon 6 splicing region. We found two silent mutations (as compared to TSC2 reference gene # NC_000016.10) in LAM patient #2634 with an A/C (nt 46399) transversion in the intronic region and a T/C (nt 46450) transversion in exon 41 (Fig. 2C).

Expression of EDN1, EDNRA, EDNRB and β-arrestin (ARRB1) is dysregulated in LAM-derived cells.
We analyzed mRNA expression of EDN1, of its receptors (EDNRA and EDNRB), and of ARRB1 in lung-derived LAM cells and in TSC2 NEG (Fig. 3A). EDN1, EDNRA, EDNRB and ARRB1 mRNAs were differentially expressed in the primary LAM cells derived from the five patients of our study. LAM cells derived from patient #2634 expressed high levels of EDN1, EDNRA, EDNRB and ARRB1 mRNAs when compared to primary PASMC (Pulmonary Artery Smooth Muscle Cells) (Fig. 3A). Overexpression of EDN1 mRNAs and of at least one of two EDNRs was found in three (#1444, #2634, #2749) out of the five lung-derived LAM cultures (Fig. 3A). Despite some heterogeneity of EDN1, EDNRA, EDNRB and ARRB1 mRNA levels, we evidenced a transcriptional dysregulation of the EDN1 pathway in primary lung-derived LAM cells. To investigate TSC2 involvement in EDN pathway dysregulation, we compared mRNA expression of EDN1, EDNR and ARRB1 genes in TSC2 NEG and TSC POS cell lines (Fig. 3A). The mRNA expression of EDN1 and ARRB1 was significantly higher (respectively 3.5 and 6.5-fold) while EDNRB mRNA expression was significantly lower (10 fold) in TSC NEG as compared to TSC POS cells (Fig. 3A); ENDRB protein expression was lower in TSC NEG cells as compared to TSC POS cells (Fig. 3A). Moreover, while EDNRA mRNAs were not differentially expressed in the two cell lines (Fig. 3A), higher levels of EDNRA proteins were detectable in TSC2 NEG cells as compared to TSC2 POS cells (Fig. 3B). Conversely, EDNRB proteins were expressed at a weaker level in TSC NEG compared to TSC POS cells (Fig. 3B).
Next, we tested the effect of endothelin stimulation by incubating both cell lines with 100 nM EDN1 for 24 hours. Following endothelin stimulation, expression of EDN1 and EDNRB mRNAs significantly increased in TSC2 POS cells but stayed unchanged in TSC2 NEG cells (Fig. 3B) suggesting that TSC2 NEG cells lost their ability to respond to endothelin stimulation.

Proliferation of TSC2 NEG cells upon EDNR and mTOR inhibitor treatment.
We first showed that the TSC2 NEG cells had a significant proliferative advantage over the TSC2 POS cells (Fig. 4A). To investigate EDN pathway involvement in LAM, we tested the effect of ERA such as bosentan (EDNRA and EDNRB inhibitors) and BQ123 (EDNRA inhibitor), with or without rapamycin (mTOR inhibitor) on cell proliferation. Twenty-four hours of treatment with 10 µM bosentan, 10 µM BQ123 or 10 nM rapamycin alone or in combination did not significantly reduce TSC2 POS or TSC2 NEG cell proliferation (Fig. 4B). However, at 72 h post treatment, rapamycin was able to reduce the proliferation of both cell line, in a significant way for TSC2 NEG cells (Fig. 4B). Cell proliferation was also reduced when both cell lines were treated with a combination of BQ123 and rapamycin (Fig. 4B). Surprisingly, bosentan alone significantly increased TSC2 POS and TSC2 NEG proliferation (Fig. 4B). These results demonstrated the efficient inhibition of cell proliferation upon rapamycin treatment and suggested that inhibitor of EDNRA alone such as BQ123 do not alter proliferative rate of TSC2 NEG cells, while bosentan (targeting both EDNRA and EDNRB) activates cell proliferation.

Migration of TSC2 NEG cells treated with inhibitors of EDNR or mTOR.
We first established that TSC2 NEG cells had a significant migration advantage over TSC2 POS cells (Fig. 5A). Using fluorescent labeling to follow migration, we showed that while TSC2 POS cells poorly migrated from the upper to the lower compartment of the insert, TSC2 NEG cells efficiently migrated as measured by the level of fluorescence signal (Fig. 5A). We then analyzed EDN involvement in migration of TSC2 POS and TSC2 NEG cells treated with endothelin receptor antagonists, alone or in combination with rapamycin. The addition of 10 μM bosentan or 10 μM BQ123 did not affect cellular migration of TSC2 NEG or TSC2 POS cells (Fig. 5B), suggesting that the EDN1 pathway was not implicated in migration. However, rapamycin alone or in combination with BQ123 significantly reduced migration of TSC2 NEG cells but not of TSC2 POS cells (Fig. 5B). These results suggest that unlike the EDN pathway, the mTOR pathway is implicated in LAM cell migration in vitro.
TSC2 NEG cell growth is anchorage-independent and is reduced by rapamycin in combination to ERA. When maintained in semi-solid medium, TSC2 NEG cells grew as spheroids, while TSC2 POS cells were unable to proliferate in 3D conditions (Fig. 6A), suggesting that TSC2 NEG cells had acquired a transformed or neoplastic phenotype. Interestingly, treatment with rapamycin alone or in combination with BQ123 or bosentan significantly reduced spheroids size when compared to untreated TSC NEG cells (Fig. 6B). Endothelin receptor inhibitors alone (BQ123 or bosentan) were unable to decrease spheroid size (Fig. 6B) while bosentan alone had a reverse effect, favoring the growth of TSC2 NEG spheroids. In conclusion, while ERAs failed to alter anchorage-independent cell growth, rapamycin prevented expansion of TSC2 NEG spheroids.

Discussion
LAM cells are characterized by increased proliferation associated to TSC1 and TSC2 mutations, leading to the activation of the mTOR pathway 30 . We hypothesized that dysregulation of the endothelin pathway could be implicated in LAM. The endothelin/β-arrestin/β-catenin pathway is activated in many cancers such as lung, ovary, prostate or breast cancers. It activates proliferation, migration, invasion and survival of cancer cells and stimulates angiogenesis 23 . To explore our hypothesis, we have successfully derived and characterized primary LAM cells that were compatible with the myofibroblast-like subpopulation of LAM, expressing α-smooth actin, desmin and vimentin. Lung-derived LAM primary cells hardly expressed melanocytic markers recognized by HMB45 and PNL2 antibodies but expressed PMEL and TYRP1 mRNAs. The derivation of pulmonary LAM cells revealed to be challenging, as these cells proliferated slowly and were not easily maintained in culture. Despite their limited lifespan, they may be of interest as an ex vivo model of LAM. We studied the expression of key genes of the endothelin pathway as an insight of what occurs in LAM patients. We have analyzed the expression of EDN1 and of its receptors EDNRA and EDNRB in primary LAM cells as well as in a TSC2 mutated cell line derived from an angiomyolipoma. Interestingly, mRNA expression of EDN1, EDNRA and EDNRB genes was dysregulated. The TSC2 NEG cell line and most primary LAM cells expressed high levels of EDN1 and ARRB1 mRNAs and a low level of EDNRB mRNAs when compared to control cells. Unlike TSC2 POS cells, endothelin stimulation of TSC2 NEG cells did not affect EDN1, EDNRA or EDNRB mRNA expression, suggesting the loss of positive feedback, possibly due to an independent activation of the EDN1 pathway in these cells.
To investigate the role of these dysregulations in LAM, we studied the effect of EDN receptor inhibitors, combined or not with rapamycin in the LAM cell lines. Proliferation, migration and anchorage-independent growth of TSC2 NEG were not modified in presence of BQ123, an inhibitor of EDNRA; surprisingly bosentan, which targets EDNRA and EDNRB, activated cell proliferation. However, we showed that rapamycin was an inhibitor of TSC2 NEG cell migration and anchorage-independent growth.
Despite the expression of EDN in LAM cells, we showed that the EDN1 pathway was not implicated in cellular proliferation or migration or anchorage-independent growth, as demonstrated by the lack of inhibitory effects of ERAs such as BQ123. These results are in contrast with the beneficial effects of ERAs with anti-proliferative action on arterial smooth muscle cells in other diseases such as pulmonary arterial hypertension, where EDN1 is expressed in the pulmonary vasculature [31][32][33] . We have shown that TSC2 NEG were highly proliferating and migrating cells. It has been reported that TSC2 controls cell migration through its N-terminus and that RhoA activity is necessary 34 . TSC2 POS control cells used in this study have been generated by re-establishment of normal TSC2 gene expression in the TSC2 NEG cell line via a viral vector. The TSC2 POS cells that have regained TSC2 expression are characterized by lower proliferation rate, no migration and no anchorage-independent growth; the restoration of TSC2 expression in patients carrying mutations could open to new therapeutic approaches.
Randomized controlled trials have shown that mTOR inhibitors such as sirolimus and everolimus stabilize lung function and reduce angiomyolipoma size in LAM patients 15,35 . Our study is the first report on the inhibitory effect of rapamycin on cellular migration of TSC2-deficient cells. Apart from its effect on cell proliferation, mTOR inhibition may control cell migration in LAM patients; this could participate to the beneficial effect of mTOR inhibitors in LAM patients.
To summarize our results, we have established that the EDN1 pathway is dysregulated in TSC2-deficient cell lines as well as lung-derived primary LAM cells associated with high levels of EDN and ARRB1 mRNA. While targeting EDN receptors with existing antagonists did not prevent proliferation, migration or anchorage-independent growth of LAM cells, β-arrestin 1 could be considered as an alternative target in LAM patients. β-arrestin 1 is a scaffold protein interacting with G protein-coupled receptors and is a major effector of the EDN1 pathway. The binding of EDN1 to EDNRA allows the activation and recruitment of β-arrestin 1 and the formation of β-arrestin 1 and β-catenin complexes 36 that translocate to the nucleus and activate EDN1 transcription 21,[37][38][39] . Barbadin, a β-arrestin inhibitor, has been recently reported 40 and could help to decipher the role of ARRB1 in LAM as a potential target for new therapeutic tools.
Interestingly, we have shown that rapamycin has an inhibitory effect on LAM cell migration and anchorage-independent growth; this may be of importance in understanding the beneficial effect of mTOR inhibitors in LAM patients. Figure 6. Anchorage-independent growth of TSC2 NEG cells. TSC2 POS and TSC2 NEG cells were cultured in semi-solid medium for 6 days and monitored for the formation of spheroids. (A) Microscopic observation of cells growing in semi-solid medium. TSC2 NEG formed multiple spheroids while TSC2 POS were not able to grow without anchorage. (B) Cells were treated for 24 hours with 10 nM rapamycin, 10 μM BQ123, 10 μM bosentan alone or in dual combination and transferred in semi-solid medium. Spheroids formation was monitored at day 6. Spheroid size has been determined and compared between cells maintained in presence of inhibitors or in non-treated semi-solid medium (BQ123, rapamycin or rapamycin + BQ123) or semi-solid medium containing DMSO (bosentan or rapamycin + bosentan). Statistical analyses were performed using the Mann Whitney test (***p < 0.001).