low neurotrophin receptor CD271 regulates phenotype switching in melanoma

Abstract

Cutaneous melanoma represents the most fatal skin cancer due to its high metastatic capacity. According to the "phenotype switching" model, the aggressive nature of melanoma cells results from their intrinsic potential to dynamically switch from a high-proliferative/low-invasive to a low-proliferative/high-invasive state. Here we identify the low affinity neurotrophin receptor CD271 as a key effector of phenotype switching in melanoma. CD271 plays a dual role in this process by decreasing proliferation, while simultaneously promoting invasiveness. Dynamic modification of CD271 expression allows tumor cells to grow at low levels of CD271, to reduce growth and invade when CD271 expression is high, and to re-expand at a distant site upon decrease of CD271 expression. Mechanistically, the cleaved intracellular domain of CD271 controls proliferation, while the interaction of CD271 with the neurotrophin receptor Trk-A modulates cell adhesiveness through dynamic regulation of a set of cholesterol synthesis genes relevant for patient survival.

Fig. 1

CD271 correlates with an invasive signature and decreased proliferation. a Scheme illustrating the experimental approach using human organotypic skin substitutes as the basis for a humanized in vivo melanoma model to study melanoma invasion. Human melanoma cells (M070413) stably expressing GFP are mixed with human keratinocytes (HKC) and seeded onto high-density type-I collagen hydrogels containing human dermal fibroblasts (HFB). This human dermo–epidermal skin equivalent is then transplanted onto immunocompromised rats. b Immunofluorescent staining of GFP and CD271 on sections of transplanted human organotypic skin substitutes (M Melanoma, D Dermis). Scale bar 75 µm. c After excision, cells were isolated from the skin equivalent and sorted for GFP and CD271 high and low populations. d Gene expression analysis of the CD271 high and low populations (n = 2, Log2 ratio < −0.6, > 0.6; FDR < 0.5). e Pathway analysis of the CD271 high vs. low population with MetaCore^TM (http://thomsonreuters.com/en.html). f Selected genes with highest significance from the MetaCore^TM analysis representing enrichment of the EMT, Cell Adhesion, ECM Remodeling, Cell Cycle, Upregulation of MITF in melanoma and other pathways. g–p Correlation of RNA seq. data from TCGA for CD271 with MITF, Tyrosinase (TYR), Melan-A (MLANA), PMEL, E-cadherin (CDH1), ZEB1, TWIST1, WNT5A, c-JUN, and AXL genes

Fig. 2

Endogenous levels of CD271 influence tumor growth. a, b Melanoma cell lines with different levels of CD271 (CD271 low = M010817; CD271 high = M050829 were stably transduced with lentiviral backbones expressing infrared fluorescent protein (iRFP) and injected in to Nude mice. The growth of the xenografts was assessed by IVIS a and macroscopically after euthanasia b. c H&E staining and CD271 immunofluorescence of tumors formed from low and high CD271-expressing melanoma cells. Scale bars upper panel 2 mm, scale bars lower panel 50 µm. d, e qRT-PCR and western blot of the CD271 low melanoma cell line (M010817) treated with TGFβ1 ligand (5 ng/ml) at different time points. (FN: Fibronectin; CDH2: N-cadherin; β-act.: beta actin). N = 3, P values CD271 > 0.05 and ≤0.001, P value FN ≤ 0.001, P value N-cadh. ≤0.001. Error bars indicate S.D

Fig. 3

CD271 overexpression impairs proliferation and adhesion. a Schematic representation of the lentiviral vector used to stably overexpress CD271 in melanoma cell lines. The viral backbone carries the Red Fluorescent Protein (RFP) and infrared fluorescent protein (iRFP) reporters under the SV40 and PGK promoters respectively, as well as the CD271 overexpression cassette (CD271 OE) or an empty vector (EV) under the CMV promoter. b FACS analysis for EdU incorporation in cells 72 h after infection with EV or CD271 OE vectors (n = 3; P value ≤ 0.05). c Upper panel: in vivo imaging of Nude mice injected with cells carrying the empty vector (EV) or the CD271 overexpressing construct (CD271 OE). The signal is from the infrared protein present in the lentiviral backbone. Lower panel: Immunofluorescence for CD271 protein in xenografts obtained by injection of melanoma cells containing the EV or the CD271 OE vector. Scale bars 100 µm. d Quantification of the signal from iRFP by IVIS at different time points (four mice for a total of eight injections were analyzed for both control and CD271 overexpression; P value ≤ 0.001). Error bars indicate S.E.M. e Brightfield and fluorescent micrographs of melanoma cells in culture 72 h after infection with the EV or CD271 OE vector. Scale bars 50 µm. f Crystal violet staining of melanoma cells 72 h after infection with the EV or CD271 OE vector after washing (upper panel) and 2 h after suspension cells reattached on FN plates (lower panel visualized as inverted, black and white micrographs). g Quantification of the suspension cells re-attached on fibronectin-coated plates. (n = 3, P value ≤ 0.001) h qRT-PCR for CD271 levels in melanoma cells infected with EV or CD271 OE in the adherent and in the suspension fractions (n = 3, P value ≤ 0.001). Error bars for g and h indicate S.D. i Western blot for CD271 levels in melanoma cells infected with EV or CD271 OE in the adherent and in the suspension fractions. j Crystal violet staining of suspension cells re-attached on fibronectin-coated plates 10 days after seeding. All experiments done with cell line M010817

Fig. 4

CD271 transient overexpression in vitro reveals a reversible phenotype. a Schematic representation of the inducible vector (TetON) to overexpress CD271 in melanoma cell lines. The viral backbone carries the green fluorescent protein (GFP) and infrared fluorescent protein (iRFP) reporters under the SV40 and PGK promoters respectively, as well as the CD271 overexpression cassette (CMVTOCD271) or an empty vector (CMVTOEV) under the inducible CMVTetOperon (CMVTO) promoter. The overexpression of the gene is controlled by doxycycline and is based on the TRex^TM system (pLenti CMV TetR Blast, Addgene no. 17492). b Crystal violet staining of adherent cells (upper panels) and suspension cells forced to reattach on fibronectin-coated plates (lower panels). Not treated (day 0), treated with doxycycline for 24 h (day 1), and subsequently released from doxycycline for 48 h (day 4). c Quantification of cell numbers of b (n = 3, P values > 0.05, ≤ 0.01). Error bars indicate S.D. All experiments done with cell line M010817

Fig. 5

Transient CD271 overexpression in vivo increases metastasis. a Time course for the transient CD271 expression experiment in vivo. At day 0 (d0) the mice were injected with melanoma cells carrying CMVTOEV or CMVTOCD271 vectors. 1 week after injection (d7), the mice were treated with 200 ng/ml of doxycycline for 1 week until day 14 (d14). The mice were then released from doxycycline and analyzed 2 weeks later (d28). Different measurements (m1, m2, m3, m4) were done by IVIS to check tumor growth at different time points. b In vivo imaging of Nude mice injected with cells carrying CMVTOEV or CMVTOCD271 vectors. m1 = start of doxycycline treatment; m2 = during doxycycline treatment; m3 = end of doxycycline treatment; m4 = 2 weeks after release of doxycycline. c Quantification of the IVIS signal for infrared protein (iRFP) (4 mice for a total of 8 injections were analyzed for both conditions; P value_m1 > 0.05, P value_m2 ≤ 0.05, P value_m3 ≤ 0.05, P value_m4 > 0.05). Error bars indicate S.E.M. d qRT-PCR for GFP transcripts in lung lysates of Nude mice treated according to a (n = 6 for each condition. P value ≤ 0.05). Error bars indicate S.E.M. e Representative pictures of lungs at the end of the experiments to detect GFP. To enhance the visibility of the GFP spots, the pictures were set to grayscale and inverted. f Time course for the control experiment, in which CD271 expression was not switched. The experiment was designed as in a with the exception that between day 7 (d7) and day 28 (d28) the mice were not released from doxycycline but continuously treated. g In vivo imaging of Nude mice injected with cells carrying CMVTOEV or CMVTOCD271 vectors at different time points of the experiment (m1 = Start of doxycycline treatment; m2 = During doxycycline treatment; m3 = During doxycycline treatment). h Quantification of the IVIS signal for iRFP (4 mice for a total of 8 injections were analyzed for both conditions; P value_m1 > 0.05, P value_m2 ≤ 0.05, P value_m3 ≤ 0.001). Error bars indicate S.E.M. i qRT-PCR for GFP transcripts on lung lysates of Nude mice treated according to f. (n = 4 for each condition P value > 0.05). Error bars indicate S.E.M. All experiments done with cell line M010817

Fig. 6

CD271 regulates proliferation and adhesion via separate signaling axes. a Representation of the lentiviral vector for stable overexpression of the Intra Cellular Domain (ICD) of CD271. The viral backbone contains Green Fluorescent Protein (GFP) and infrared fluorescent protein (iRFP) reporters under the SV40 and PGK promoters respectively, as well as the ICD overexpression cassette (ICD OE) or an empty vector (EV) under the CMV promoter. b FACS analysis for EdU incorporation in cells overexpressing ICD (ICD OE) compared to control cells (EV) (n = 3; P value ≤ 0.05). c In vivo imaging of Nude mice injected with cells carrying the empty vector (EV) or the ICD overexpressing vector (ICD OE). The signal is from the iRFP present in the lentiviral backbone. d Quantification of the signal obtained by IVIS for iRFP (three mice for a total of six injections were analyzed for each conditions; P value_m1 ≤ 0.05, P value_m2 ≤ 0.05). Error bars indicate S.E.M. e Brightfield (upper panel) and fluorescent micrographs (lower panel) of melanoma cells in culture 72 h after infection with the EV or ICD OE vector. Scale bars 100 µm. f Crystal violet staining of adherent cells (upper panel) and suspension cells, after re-attachment on fibronectin-coated plates (lower panel), carrying either the control vector (EV) or the ICD overexpressing vector (ICD OE). g Quantification of adherent and suspension cells after infection with EV or ICD OE constructs (n = 3, P value_adh.cells ≤ 0.05, P value_susp.cells > 0.05). Error bars indicate S.D. h FACS analysis for EdU incorporation in cells infected with CMVTOEV or CMVTOCD271 constructs and transfected with sicontrol or siTrk-A. One day after siRNA transfection the cells were treated with doxycycline (1 μg/ml) for 24 h. Cells were then pulsed with EdU for 30 min and collected for FACS analysis (n = 3; P value ≤ 0.05). i Crystal violet staining of suspension cells (forced to reattach on fibronectin-coated plates) infected with CMVTOEV or CMVTOCD271 and transfected with siControl (upper panel) or siTrk-A (lower panel). The cells were treated, as described in h. j Quantification of suspension cells (CMVTOEV or CMVTOCD271 infected) after siRNA transfection and doxycycline administration (as described in h) (n = 3, P value ≤ 0.05, P value ≤ 0.001). Error bars indicate S.D. All experiments done with cell line M010817

Fig. 7

Transcriptome analysis upon transient CD271 overexpression. a RNAseq of cells before, during and after CD271 transient overexpression was performed and analyzed using edgeR. Genes were filtered for Log2 ≥ +0.27 and ≤−0.27, p-value < 0.05, FDR < 0.05, which resulted in 160 genes significantly upregulated upon CD271 overexpression (comparing CMVTOCD271 + DOX 24 h over CMVTOCD271 –DOX24 h) with subsequent significant downregulation (comparing CMVTOCD271 + DOX 24 h/−DOX 48 h over CMVTOCD271 + DOX 24 h) and vice versa in 303 genes significantly downregulated upon CD271 overexpression with subsequent significant upregulation. Heatmap represents row z-scores of normalized counts from RNAseq. b Top 20 gene ontology process enrichments from MetaCore^TM. Numbers in brackets indicate the number of differentially expressed genes above the total number of genes within an annotated pathway. c Selected “switching” genes from a involved in cholesterol and lipid biosynthesis pathways. Again, heatmap represents row z-scores of normalized counts from RNA Seq. d Survival curves based on TCGA data for expression of cholesterol biosynthesis genes in human melanoma samples. Data from 114 patients per condition were analyzed, which represent 25% of highest and lowest expressing patients. e Selected “switching” genes from a involved in cell cycle regulation. Again, heatmap represents row z-scores of normalized counts from RNA Seq. f qRT-PCR for cell cycle genes in ICD overexpressing and EV melanoma cells. g qRT-PCR for cholesterol and lipid biosynthesis genes in ICD overexpressing and EV melanoma cells. For f and g, n = 3 for each condition. P values > 0.05 = n.s., P values ≤ 0.05 = *, P values ≤ 0.01 = **, P values ≤ 0.001 = ***. Error bars for f and g indicate S.D. All experiments done with cell line M010817

Fig. 8

CD271-mediated upregulation of cholesterol synthesis genes is TrkA dependent and regulates cell adhesion. a–i qRT-PCR for cholesterol synthesis genes in CMVTOEV and CMVTOCD271 cells transfected with siCtrl or siTrkA and 24 h later treated with doxycycline (1 µg/ml) for 24 h. N = 3 for each condition. P values > 0.05 = n.s., P values ≤ 0.05 = *, P values ≤ 0.01 = **, P values ≤ 0.001 = ***. j Crystal violet staining of cells in suspension reattached on FN plates after 24 h of combinatorial treatment with doxycycline and vehicle (DMSO), Lovastatin (2.5 µM) or MF-438 (15 µM). k Quantification of amount of cells in adherent and suspension fractions after combinatorial cell treatment with doxycycline and vehicle (DMSO), Lovastatin or MF-438. N = 3 for each condition. P values > 0.05 = n.s., P values ≤ 0.05 = *. Error bars for a–i and k indicate S.D. All experiments done with cell line M010817