Specific Activation of the CD271 Intracellular Domain in Combination with Chemotherapy or Targeted Therapy Inhibits Melanoma Progression

CD271 (p75NTR, NGFR) is a transmembrane receptor expressed in the nervous system and skin. It interacts with numerous ligands to modulate multiple pathways (1–6). The known ligands are the four neurotrophins. CD271 can bind the neurotrophins alone or in association with another class of neurotrophin receptors: the tropomyosin receptor kinases (TrkA, TrkB, and TrkC). Interestingly, melanoma cells synthesize and secrete all neurotrophins and express both receptor classes, revealing the presence of a complex autocrine pathway regulating proliferation and migration (7).

Several studies showed that CD271 plays a critical function in melanoma (6–10). CD271 downregulation increases cell proliferation, whereas its overexpression is associated with a stem-like quiescent state. In contrast, its role in metastasis remains controversial. CD271 expression was shown to be lost in early progression when melanoma cells invade the dermis (11), whereas other studies showed its association with increased metastatic abilities (8, 10).

Live-cell imaging revealed that exposure of melanoma cells to MAP-kinase inhibitors (MAPKi) results in multiple cell fates including a drug-adapted state associated with CD271 overexpression, reversible upon drug removal (5). Consistently, Rambow and colleagues identified drug-tolerant transcriptional states in BRAF-mutant patient-derived xenografts exposed to MAPKi, showing CD271 to be one of the key markers of adaptive resistance (12).

CD271 expression is upregulated in response to chemotherapeutic agents in association with DNA repair genes, revealing a role in chemoresistance (13, 14). Notably, some studies demonstrated CD271 to be a suppressor of p53 leading to MDM2-mediated p53 proteolysis (15).

CD271 can signal alone or in association with other receptors (e.g., Trks, Nogo, sortilin, DR6), promoting either survival or apoptosis. This flexibility allows CD271 to play a fundamental role in the regulation of cell fate by acting as an on/off switch to modulate opposing pathways (6).

Unlike neurotrophins, which bind CD271 and activate opposite pathways according to the cell tissue context, β-amyloid has been shown to bind CD271 with a high affinity and to selectively activate its apoptotic pathway in the nervous system (16).

Unlike previous approaches that inhibited the receptor or prevented its overexpression, we activated the CD271 pro-apoptotic function with a short β-amyloid derived peptide. Since the minimum sequence able to bind CD271 was shown to be amino acids 25–35, all experiments were performed using this short peptide (Aβ^(25–35); refs. 17–20).

WM115, WM266–4, SKMEL28, WM793B, and 1205Lu (ATCC) were cultured as indicated by the manufacturer. Other cell lines were generated from biopsies at the University Hospital Zurich after written informed consent and approval by the local Institutional Review Board (BASEC-2017–0494) as previously described (21). The cultures p0-p20 were tested for Mycoplasm once a month. Cells were grown in RPMI1640 (Sigma Life Science) supplemented with 10% FBS (Gibco, Life Technologies), 2 mmol/L glutamine (Biochrom), and sodium-pyruvate (Sigma Life Science). Human cells experiments were approved by the ethical review board (KEK No. 2014–0425).

Drugs used: PLX4032 and MEk162 (Selleckchem), JNKi (SP600125, Santa Cruz Biotechnology), MG132, DAPT and TAPI2 (Calbiochem), N-acetyl-L-cysteine (Sigma), cisplatin, dacarbazine, and carmustine (Selleckchem).

Aβ^(25–35) (Bachem; catalog no. H-1192). “Cntr” refers to the inverted sequence^(35–25) (Bachem; catalog no. H-2964). Both peptides were diluted in PBS and incubated at 37°C for 2 hours. Fibril formation was induced by sonication (5 seconds, low intensity). For the treatment with soluble ECD, we used the recombinant protein from LSBio (catalog no. LS-G27179).

Tumor slices were kept in medium with antibiotic and then embedded in 1% agarose on the same day as biopsy. The tumors were cut with a vibratome (300 μm), and the slices were placed on filters in 6-well plates and treated for 3 to 5 days.

A total of 5 × 10³ melanoma cells/well were seeded on agarose-coated plates as described previously (22). Seventy-two hours later, melanoma spheroids were implanted into collagen I as described previously (23). For live/dead evaluation, spheres were stained with calcein 7 μmol/L (Sigma, catalog no. 17783) and ethidium homodimer 10 μmol/L (Sigma, catalog no. E1510) for 1 hour in melanoma medium. Pictures were taken with a confocal microscope (Leica TCS SPE). Analysis of spheroids area was performed as described previously (11). Live/dead staining was quantified with Photoshop.

A total of 5 × 10³ cells/well were seeded in 96-well plates. Cells were incubated with 0.5% MTT for 4 hours at 37°C and then dissolved with 100 μL isopropanol. The plate was read at 560 nm with a reference filter of 650 nm.

Cells were incubated with anti-CD271 antibody (1:100 in PBS; Merck, catalog no. 05–446) and with secondary Alexa Fluor anti-mouse 488 (1:50; Invitrogen, catalog no. A21202) for 20 minutes at 4°C and analyzed by flow cytometry (LSR Fortessa II).

Reactive oxygen species (ROS) measurements: cells were stained with MitoSOX (5 μM; Invitrogen, catalog no. M36008) for 20 minutes at room temperature, resuspended in FACS Buffer (0.2% FBS in PBS) and analyzed. For cell death analysis, cells were stained with PI solution (50 μg/mL propidium iodide, 0.1% sodium citrate, 0.5% Tryton X-100) for 15 minutes. For caspase-3 staining, the “CellEvent Caspase-3/7 green detection Reagent” Kit (Invitrogen, C10723) was used.

Cells were incubated in blocking buffer (DMEM, 10% FBS, 4% human gamma globulin) for 20 minutes on ice. Cells were stained with anti-CD271 antibody (1:100 in PBS, Merck, catalog no. 05–446) at room temperature for 15 minutes, washed and stained with Alexa Fluor 488 anti-mouse antibody (Invitrogen, catalog no. A21202) for 15 minutes at room temperature. Melanoma medium was added to the cells. A FACS Aria III flow cytometer (BD Biosciences) was used and analyzed on FACS Diva software (BD Biosciences).

Reconstructs were generated by seeding human keratinocytes and melanoma cells on dermal equivalents, as described previously (11). After 12 days, skin reconstructs were fixed with formalin for 2 hours at room temperature and embedded in paraffin.

Cells were harvested in lysis buffer pH 7.5 (150 mmol/L NaCl, 15 mmol/L MgCl, 1 mmol/L EGTA, 50 mmol/L Hepes, 10% Glicerol, 1% Triton-X100). Membranes were incubated with CD271 (1:1000, Cell Signaling Technology, catalog no. 8238), CD271_ICD (1:1000; Merck, catalog no. 07–476), APR1 (1:1,000; Thermo Fisher Scientific, catalog no. PA5–67686), NRAGE (1:250; R&D System, catalog no. MAB38352), PARP (1:1,000; Cell Signaling Technology, catalog no. 9542), cleaved-PARP (1:1,000; Cell Signaling Technology, catalog no. 5625), p53 (1:1,000; Cell Signaling Technology, catalog no. 2527), p21 (1:500; Cell Signaling Technology, catalog no. 2946), GSK (1:1,000; Cell Signaling Technology, catalog no. 9315), pGSK (1:1,000; Cell Signaling Technology, catalog no. 9336), JNK (1:1000; Cell Signaling Technology, catalog no. 9252), pJNK (1:1,000; Cell Signaling Technology, catalog no. 4668), GAPDH (1:500; Cell Signaling Technology, catalog no. 3683), Snail (1:1,000; Cell Signaling Technology, catalog no. 3879), Slug (1:1,000; Cell Signaling Technology, catalog no. 9585), N-cadherin (1:1,000; Cell Signaling Technology, catalog no. 13116), E-cadherin (1:1,000; Cell Signaling Technology, catalog no. 3195), and secondary peroxidase-conjugated antibody (1:3,000; Cell Signaling Technology, catalog no. 7074).

The immunoprecipitation kit (Abcam; catalog no. ab206996) was used in pull down experiments. M121224 wt and CD271 KO cells were treated for 48 hours with Aβ^(25–35) 40 μmol/L. Lysates were pulled-down with CD271 Ab (1 μg Ab/300 μg lysate).

Lentiviral vectors [CD271 full-length and extracellular domain (ECD) overexpression] were a gift from Lukas Sommer (8). For CD271 transient induction, M130429 melanoma cells with inducible CMVTO_EV and CMVTO_CD271 constructs, were induced in vitro with doxycycline at a concentration of 1 μg/mL as described previously (8).

CD271-specific sgRNAs (GGTGTCCCTTGGAGGTGCCA) were used. HEK293T, cells were seeded for transfection with the viral plasmids. After 24 hours, the LentiCRISPRv2GFP expression plasmid (Addgene, plasmid no. 82242426, catalog no. 82416) containing CD271 sgRNA was mixed with the packaging plasmid psPAX2 (2 μg; Addgene, plasmid no. 12260), the envelope plasmid pMD2.G (1 μg; Addgene, plasmid no. 12259), and PEImax (21 μg; Polysciences, catalog no. 24765) in 1 mL of serum-free DMEM (Gibco, catalog no. 11960–044) and incubated 15 minutes at room temperature. The DNA/PEImax mixture was add to the HEK293T cells and the medium was collected and sterile filtered 48 hours. Melanoma cells were incubated overnight in virus-containing medium + 8 μg/mL polybrene (H9268, Sigma-Aldrich). GFP-positive cells were sorted by the FACSAria III (BD Biosciences).

Cells plated for 24 hours in antibiotic-free medium were transfected with 100 nmol/L APR1 (OnTargetPlus SMART pool human MAGEH1, catalog no. L-009135-00-0005) or scrambled siRNA (Dharmacon Inc.; Scramble, siGENOME Control pool Non-targeting #1, catalog no. D-001206-13-05) in antibiotic/FBS-free medium for 24, 48, and 72 hours. For transfection we used the INTERFERin Transfection Kit (Polyplus-transfection, catalog no. 101000016) and we followed the manufacturer's guidelines.

Skin reconstructs and melanoma lesions were stained with hematoxylin and eosin (H&E), S100 (1:400; Novocastra, Agilent Technologies, catalog no. NCL-L-S100p), CD271 (1:100 in PBS; Lab Vision Corporation, catalog no. LS-C124234-100), Ki67 (1:200 in PBS; Dako, catalog no. M7240), TUNEL (Sophistolab), Snail/Slug (1:500 in PBS, Abcam, ab85936) and caspase-3 (1:800, Abcam ab2302).

Zebrafish experiments were performed at the University of Padova, Italy, under authorization 407/2015-PR (OPBA). All procedures complied with the European Legislation for the Protection of Animals used for Scientific Purposes (Directive 2010/63/EU). Embryos were obtained from natural spawning of albino adults. Human cells were stained with Vybrant Cell-Labeling Solution (5 μg/mL, Molecular Probes) for 20 minutes, loaded in a glass capillary needle, and microinjected into the yolk as previously described (11). Analysis was performed by a blind investigator. Imaging was performed using a Leica MZFLIII dissecting microscope with a Leica DFC7000T camera.

For in vivo experiments 9-week-old female Hsd:Athymic Nude-Foxn1nu were used.

All animals were housed according to institutional guidelines and experiments were approved by CNIO, the ISCIII Ethical Committee, and the Comunidad Autónoma de Madrid (protocol PROEX225/17). The experiments were performed in accordance to the guidelines for Ethical Conduct in the Care and Use of Animals as stated in The International Guiding Principles for Biomedical Research involving Animals, developed by the Council for International Organizations of Medical Sciences (CIOMS). 5 × 10⁵ M121224 cells were subcutaneously injected in 40 μL of 1:1 DMEM: Matrigel (BD Biosciences) and when the tumor reached 50 mm³, 15 μg of Aβ^(25–35)/BSA in 20 μL were intratumoral injected three times a week. Mice were sacrificed when control group reached 1,000 mm³.

Statistical analysis was performed using GraphPad Prism5.0. P values ≤0.05 were considered significant. Cell line experiments were triplicated and error bars represent the mean ± SD. Student t test or 2-way ANOVA were applied.

The authors declare that the data of this study are available within the paper and its supplementary files or from the authors upon request.

CD271 marks neural crest-like cells that are associated with resistance to targeted therapy (12). We thus evaluated its expression after treatment with carmustine (BCNU), cisplatin (CISP), and dacarbazine (DTIC) in ATCC cell lines (Fig. 1A). All chemotherapies strongly induced CD271 expression. In WM115 cells, CD271 dramatically increased from 32.51% to 85.51% (BCNU), 94.59% (CISP), and 95.22% (DTIC); in WM266–4 cells from 27.53% to 66.76% (BCNU), 56.06% (CISP), and 89.37% (DTIC) and in SKMEL28 cells, from 2.11% to 25.86% (BCNU), 85.50% (CISP), and 54.59% (DTIC).

We next assessed the response to MAPK inhibitors in cells extracted from eight melanoma patients' metastases. At a basal level, CD271 was heterogeneously expressed (Supplementary Fig. S1A). Confirming previous results, cisplatin induced a strong upregulation of CD271 in all cells, independent of the driver mutation (Fig. 1B; Supplementary Fig. S1B). A strong increase was also observed following treatment with vemurafenib (PLX4032, BRAFi), binimetinib (MEK162, MEKi), alone or in combination (Fig. 1B; Supplementary Fig. S1B). We confirmed this on matched tumor biopsies collected before and after the patients received BRAFi+MEKi therapy. We observed that CD271 was significantly higher after targeted therapy (P < 0.01; Fig. 1C and D; Supplementary Fig. S1C).

To study the role of CD271 in drug resistance, we evaluated its expression in three-dimensional (3D) spheroids derived from resistant and sensitive melanoma cells. Interestingly, MAPKi-resistant spheres displayed significantly higher CD271 levels in comparison to sensitive spheres (Fig. 1E).

This result was confirmed by comparing CD271 expression in sensitive and resistant subpopulations derived from the same cells after making them resistant in vitro through the long-term administration of BRAFi/MEKi (Fig. 1F), suggesting that CD271 is associated with acquired resistance. We then induced the stable overexpression of CD271 full length (FL) in 1205Lu cells and we treated the cells. Although no difference was observed with BRAFi, CD271_FL cells showed a significantly reduced sensitivity to MEKi compared with the empty vector (EV; Supplementary Fig. S1D). To better understand the role of CD271 in drug response, we analyzed the correlation between IC₅₀ to BRAFi/MEKi and CD271 levels (Supplementary Fig. S1E). However, no significant correlation was observed. We finally checked the expression of multidrug resistant (MDR) proteins, the so called “ATP-binding cassette” (ABC) proteins, which export drugs out of the cell (Supplementary Fig. S1F). None of these showed a correlation with CD271 suggesting resistance mechanisms independent of MDR proteins.

Given the rapid upregulation of CD271 in response to therapy, we aimed to activate apoptosis in CD271-expressing cells with a short peptide derived from the β-amyloid (Aβ) sequence. All experiments used the short peptide Aβ^(25–35) derived from the total Aβ sequence^{(1–42 aa)}.

Aβ^(25–35) significantly reduced viability in CD271^high cells, whereas no effect was observed in CD271^low cells (Fig. 2A). However, the combination with cisplatin resulted in a strong induction of cell death in CD271^low cells (Fig. 2B; Supplementary Fig. S2A). Combinations with BCNU and DTIC produced the same effect (Supplementary Figs. S2B and S2C).

To evaluate the effects of Aβ^(25–35) on 3D spheroids, melanoma cells were seeded in agarose-coated plates and implanted into a collagen I matrix (Supplementary Fig. S2D). M121224 spheroids, which displayed the strongest invasive capacities, showed a significant reduction in sphere area after treatment, especially in combination with cisplatin at 96 hours (Fig. 2C and D). Moreover, the evaluation of caspase-3/7 revealed the induction of apoptosis (Fig. 2E). Combinations with DTIC were consistent (Supplementary Fig. S2E). Cell-cycle analysis revealed that Aβ^(25–35) significantly induced cell death in spheroids and the addition of chemo- or MAPKi showed a synergistic effect (Fig. 2F).

Despite the ability of 3D spheroids to recapitulate some tumor features, they oversimplify the tumor microenvironment (23, 24). Thus, we employed ex vivo organotypic slice cultures. We collected fresh tumors from three patients with MAPKi-resistant, BRAF-mutated melanoma (Fig. 3A).

The tumors were sectioned and treated for 5 days (Fig. 3B and C; Supplementary Figs. S3A and S3B). In patient 1, Aβ^(25–35) significantly reduced proliferating (Ki67+) cells, whereas the combination with cisplatin/BRAFi was synergistic. In patient 2, the BRAFi+MEKi failed to decrease Ki67 levels, as expected. To evaluate intrapatient variability, two different areas of the tumor were treated with Aβ (labeled as Aβ1 and Aβ2). Strikingly, the percentage of proliferating Ki67+ cells were rare with Aβ^(25–35)-treatment, which was similar in combination with cisplatin. Patient 3 confirmed these observations. Notably, the same area that was Ki67 negative was TUNEL positive, indicating apoptosis. Taken together, these data confirmed that Aβ^(25–35) treatment induces cell death and strongly reduces melanoma proliferation in ex vivo models of highly resistant melanoma.

As CD271 is known to control melanoma progression (8, 11), we tested if Aβ^(25–35) could influence melanoma invasion. Spheroids derived from MAPKi-sensitive (M980513) and resistant (M121224) cells were treated and the percentage of dead cells was assessed by live/dead staining (green/red, respectively; Fig. 4A and B). Although cisplatin had no effect on spheroid viability or invasion, Aβ^(25–35) alone significantly increased cell death in the core of the sphere, while invading cells were alive.

Conversely, the combination of Aβ^(25–35) with chemo- or targeted therapy induced the death of the entire sphere. Interestingly, the treatment with Aβ^(25–35) reduced the expression of N-cadherin, Snail, and Slug and induced an increase in E-cadherin, suggesting the attenuation of the cells' invasive capacity (Supplementary Figs. S4A and S4B).

Consistently, a reduction of the area of the spheres was observed with Aβ^(25–35)+DTIC (Supplementary Fig. S4C). The percent of fragmentation (i.e., percent of invading cells outside the sphere's core) and factor shape (i.e., sphericity), two measurements correlated to increased invasion, were also significantly reduced (Supplementary Fig. S4D).

For in vivo validation, we generated zebrafish xenografts of human melanoma (25). We injected fluorescent human melanoma cells into the zebrafish yolk at 2 days after fertilization (2 dpf). One day after injection (1 dpi), we added chemo- or MAPKi into the water and at 2 dpi, Aβ^(25–35) was applied. The number of fish with metastases was counted by two blinded investigators (Supplementary Fig. S4E). Notably, no in vivo toxicity for Aβ^(25–35) was observed (Supplementary Fig. S4F). Strikingly, 100% of larvae injected with BRAFi-sensitive cells (M000921) were metastasis free, whereas those treated with BRAFi alone showed a higher degree of metastasis (Fig. 4C and D). Similarly, Aβ^(25–35) alone or in combination with MEKi resulted in a significant reduction in the percentage of metastasis in MEKi-sensitive cells (M130425; Fig. 4C and D).

Finally, zebrafish injected with BRAF/NRAS-mutated MAPKi-resistant cells (M121224) showed a significant reduction in metastases with Aβ^(25–35) alone or in combination, whereas MEKi was ineffective (Fig. 4C and D). The same reduction was observed in zebrafish treated with Aβ^(25–35)+DTIC (Supplementary Fig. S4G). Thus, these data show that Aβ^(25–35) reduces metastases regardless of sensitivity or resistance to MAPKi (Supplementary Fig. S4H).

Moreover, Aβ^(25–35) strongly reduces proliferation of melanoma in vivo and significantly reduces the expression of the invasion markers Slug/Snail (Fig. 4E; Supplementary Fig. S4I). Notably, a significant increase in dead cells was detected, according to caspase-3 and TUNEL staining (Fig. 4E; Supplementary Figs. S4J and S4K). Finally, our findings were confirmed in nude mice, revealing a significant reduction in the tumor volume and the induction of caspase-3 after 15 days of Aβ^(25–35) monotherapy (Fig. 4F and G; Supplementary Figs. S5A–S5D).

We next tested the selectivity of Aβ^(25–35) for CD271 (Fig. 5A). Although Aβ^(25–35) induced cell death in M130425 cells, a significant reduction was observed in CD271-KO cells (% sub-G₁: 54.8% vs. 31.6% at 2 days and 93% vs. 47.1% at 6 days in wt vs. KO, respectively; Fig. 5B; Supplementary Fig. S6A). Cell viability was also increased in CD271 KO cells, revealing a lower susceptibility to Aβ^(25–35) (Supplementary Fig. S6B). To evaluate whether cell death was the result of apoptosis, we performed a double staining with PI/AnnexinV (Fig. 5C). Analysis of early and late apoptosis revealed a strong increase of cells undergoing apoptosis, which was significantly reduced after CD271 silencing and was confirmed in additional melanoma cells (Supplementary Figs. S6C–S6E).

We then transiently overexpressed CD271 by using a TetON lentiviral expression vector, which induces CD271 after doxycycline addition. As expected, CD271 levels increased at 48 hours in M130429 transfected cells (Fig. 5D) compared with empty vector (EV) control (i.e., CMVTO_EV).

Although no difference was appreciable after treatment with Aβ^(25–35) alone, a significant increase in cell death was observed with Aβ^(25–35)+doxycycline (16.4% vs. 43.5% in CMVTO_EV vs. CMVTO_CD271, respectively; Supplementary Figs. S7A and S7B; Fig. 5E). Consistently, a significant reduction in viability occurred after stable CD271 overexpression (Supplementary Fig. S7C).

For further validation, M121224 cells were FACS-sorted to isolate CD271⁺ and CD271⁻ cells. The two subpopulations were used to generate 3D spheroids and immediately treated for 144 hours. At 24 hours, Aβ^(25–35) prevented the formation of the sphere, giving rise to a single-cell suspension. After 144 hours of treatment, CD271⁻ cells generated multiple spheres, whereas CD271⁺ cells grew poorly, revealing a greater susceptibility to Aβ^(25–35) (Fig. 5F). Consistently, cell death was significantly higher in CD271⁺ than in CD271⁻ spheres (% sub-G₁: i.e., 24 hours: 88.5% vs. 28.9%; 144 hours: 38.4% vs. 3.49%, respectively; Fig. 5G; Supplementary Fig. S7D). Notably, although Aβ^(25–35) allowed CD271⁻ cells to form multiple spheres, the addition of cisplatin prevented their growth as a consequence of CD271 upregulation (Supplementary Figs. S7E and S7F).

A further confirmation of the affinity between CD271 and Aβ^(25–35) was demonstrated by treating the cells with Aβ^(25–35) and a recombinant CD271 ECD soluble protein, which showed a reduction in Aβ^(25–35) -induced cell death in a concentration dependent-manner, most likely by peptide sequestration (Supplementary Fig. S7G).

Although 100% of the M130425 wt-injected zebrafish were metastasis-free upon treatment, the percentage of metastases in CD271_KO larvae was comparable with controls, demonstrating that Aβ^(25–35) is inefficient in the absence of the receptor also in vivo (Fig. 5H and I). We observed a reduced chance of survival in MEKi-treated fish in the presence of CD271 compared with CD271-KO. However, the addition of Aβ^(25–35) induced a significant increase in survival probability (Fig. 5J).

Finally, 3D skin equivalents were generated from the same cells. Although Ki67+ was high in CD271-KO skin reconstructs after treatment, its expression was strongly reduced in those derived from control cells (Supplementary Fig. S7H).

After ligand binding, CD271 undergoes two proteolytic cleavages, with the first releasing the ECD and a carboxyl-terminal fragment (CTF). The second cleavage of CTF yields a soluble intracellular fragment (ICD). The proteolytic enzymes are the α- and γ-secretases, respectively (26). The activities of these enzymes can be suppressed by two inhibitors: the α-secretase inhibitor TAPI2, and the γ-secretase inhibitor DAPT (Fig. 6A). We asked whether CD271 cleavage was necessary for Aβ^(25–35) induced apoptosis. Aβ^(25–35) induced an initial upregulation between 16 and 24 hours, whereas at 48 hours CD271 levels gradually decreased and finally disappeared (Fig. 6B). Consistently, CD271 was strongly reduced 72 hours after treatment in four different cell lines (Supplementary Fig. S8A). We hypothesized that this reduction was due to the cleavage and the consequent degradation into the proteasome. To test this hypothesis, cells were treated with Aβ^(25–35) with DAPT (Fig. 6C). Although no difference was observed at 16 hours, likely because the receptor activation had not yet occurred, the prolonged treatment for 72 hours with Aβ^(25–35)+DAPT prevented CD271 cleavage and its subsequent degradation (Fig. 6C).

To confirm this, we overexpressed CD271 in 1205Lu cells and we evaluated its cleavage by using an antibody that detects the ICD (Fig. 6D). The treatment of 1205Lu CD271_FL cells with Aβ^(25–35) and the proteasome inhibitor MG132 allowed the detection of both CTF and ICD fragments (Fig. 6D). No cleavage was detectable in the absence of Aβ^(25–35). Notably, the addition of DAPT prevented the formation of the ICD and strongly decreased apoptosis in CD271_FL cells (Fig. 6D and E).

The same finding was confirmed in M121224 cells (Supplementary Fig. S8B). Treatment with TAPI2/DAPT in CD271⁺ cells prevented the cleavage (Fig. 6F and G). Consistently, cell death rescue was observed in CD271⁺ spheres after treatment with Aβ^(25–35)+DAPT/TAPI2, whereas no difference was appreciable in the CD271⁻ spheres (Fig. 6H).

To evaluate if the ICD is necessary for the activation of apoptosis, we overexpressed a truncated form of the receptor containing only the ECD (Fig. 6I and J). Interestingly, the percentage of cell death after treatment was dramatically reduced from 70.8% (FL) to 42.9% (ECD), being comparable with the EV (Fig. 6K). Taken together, these data strongly suggest that Aβ^(25–35) activates CD271 cleavage and that the ICD is necessary for mediating apoptotic cell death.

To investigate the mechanisms involved in Aβ-CD271 apoptosis, we evaluated the expression of known CD271 interactors (Fig. 7A; refs. 27–29). APR1 increased over time, whereas JNK was strongly phosphorylated at 6 hours only in wt but not in CD271_KO cells. Interestingly, the highest APR1 expression occurred at 72 hours, when CD271 was downregulated (Supplementary Fig. S8C). To verify the involvement of APR1, the protein was silenced in M121224 cells. However, APR1 silencing did not significantly decrease death compared with controls (Supplementary Fig. S8D). Thus, we asked whether the phosphorylation of JNK was affected by CD271 activation. We treated cells with Aβ^(25–35)+DAPT observing no difference in JNK phosphorylation, suggesting activation upstream of CD271 cleavage (Fig. 7B). To test this hypothesis, we treated the cells with Aβ^(25–35) in the presence of the pJNK inhibitor (SP600125) and we observed CD271 rescue in comparison to Aβ^(25–35) alone, confirming a delayed cleavage (Fig. 7C). Moreover, a CD271 co-immunoprecipitation assay revealed a direct interaction with pJNK (Fig. 7D). The JNKi rescued only 1205Lu_FL cells, whereas the treatment was ineffective in 1205Lu_ECD cells, demonstrating the role of JNK in inducing apoptosis through the generation of ICD (Fig. 7E). Moreover, the inhibition of pJNK in ex vivo melanoma slices rendered the treatment ineffective (Fig. 7F).

It has been shown that Aβ impairs the mitochondrial redox activity, leading to ROS formation and cell death (30, 31). To verify whether mitochondrial ROS (mROS) were also involved in Aβ^(25–35)-mediated apoptosis, we evaluated their levels in the presence/absence of CD271 (Fig. 7G). Interestingly, mROS strongly increased after CD271 silencing in M121224, whereas its overexpression in 1205Lu cells significantly reduced mROS levels (Fig. 7G). These data revealed an unexpected role of CD271 as a mROS inhibitor. The same result was observed in 1205Lu ECD transfected cells, showing that CD271-mediated ROS inhibition does not require its ICD (Supplementary Fig. S8E). Consistently, the treatment with doxycycline to induce CD271 strongly reduced ROS compared with control (Supplementary Fig. S8E). Despite its function as a ROS scavenger, the treatment with Aβ^(25–35) for 72 hours induced a significant upregulation of mROS in control but not in KO cells, revealing their production to be CD271-dependent (Fig. 7H). Moreover, this induction was reverted in the presence of the scavenger N-acetylcysteine (NAC; Fig. 7H; Supplementary Fig. S8F). These results were confirmed in 1205Lu cells (Supplementary Fig. S8G).

Aβ^(25–S35) treatment with JNKi+NAC significantly reduced cell death, showing a synergistic effect and demonstrating the co-existence of two apoptotic pathways: (i) a pJNK-CD271 cleavage pathway; and (ii) a mROS pathway, both dependent on the presence of CD271 (Fig. 7I).

To better understand the kinetics of ROS production, we treated M121224 cells and analyzed ROS levels over time: before CD271 upregulation (0–16 hours), during its induction (16–24 hours), and after cleavage (48–72 hours; Fig. 7J). Interestingly, mROS significantly increased between 3 and 9 hours. However, at 24 hours, when CD271 upregulation was maximal, mROS returns to the basal levels confirming its inhibitory role. Finally, after its cleavage, the levels begin rising, presumably as a consequence of the lack of CD271 (Fig. 7J). Interestingly, the upregulation of CD271 between 24 and 48 hours was prevented by NAC (Fig. 7K). These results show that Aβ^(25–35) cell death is triggered by the activation of the CD271–JNK–ROS pathway. Moreover, we discovered a feedback loop in which mROS induces CD271 upregulation, which acts as a ROS scavenger (Fig. 7L).

Finally, to evaluate the induction of apoptosis, PARP cleavage and activation of p53 were also detected (Fig. 7A). Surprisingly, p53 and PARP were activated also in CD271_KO cells as well as in the presence of DAPT and in CD271^low cells (Fig. 7A; Supplementary Figs. S8H–S8J), suggesting the presence of a CD271-independent pathway. In line with this assumption, caspase-3/7 activation in CD271_KO cells revealed the existence of an alternative mechanism (Supplementary Fig. S8K). In addition, immunofluorescence showed that Aβ^(25–35) can enter the cells and accumulate preferentially in the cytoplasm of CD271^low cells, whereas it localizes at the membrane in the presence of the receptor (Supplementary Fig. S8L).

Despite data demonstrating a role for CD271 in melanoma, few studies go beyond merely describing it as a biomarker. The pathways modulated by CD271 in melanoma, and the possibility of activating CD271 with specific ligands has been poorly explored.

In this study, we observed a significant induction of CD271 following different treatments, which makes CD271 an attractive therapeutic target. Ngo and coworkers conjugated an anti-CD271 antibody with saporin to kill CD271-expressing cells (32). Although the combination of an anti-CD47 Ab+CD271-saporin Ab in mice significantly reduced the volume of primary tumors, the treatment with CD271-saporin Ab alone was ineffective. However, it strongly reduced the percentage of metastases. These data highlight the necessity of activating CD271 to induce apoptosis.

In contrast, other studies attempted to target CD271 by inhibiting its upstream regulators. Mohammad and colleagues used vemurafenib with cJun/FAK/Scr inhibitors, thus preventing CD271 upregulation (5). Similarly, Rambow and colleagues used a RXRγ antagonist, observing a significant reduction in CD271^high cell accumulation (12).

In contrast to these approaches, we exploited the high CD271 levels induced by therapies to activate ligand-specific receptor cleavage and apoptosis by using a 10-amino-acid Aβ^(25–35) peptide in vitro, ex vivo, and in vivo.

We demonstrated that Aβ^(25–35) significantly induced apoptosis, and reduced metastasis in vivo in a CD271-dependant manner. Aβ^(25–35) induced CD271 cleavage, releasing the ICD death domain, which mediates apoptosis. A γ-secretase inhibitor prevented CD271 cleavage, significantly rescuing peptide-treated cells. Consistently, several studies showed that mutations in the ICD prevent cell death (33).

After recruiting its cytosolic interactors, CD271 mediates JNK phosphorylation, which in turn induces the α-secretase TACE/ADAM17, leading to CD271 cleavage (34, 35). Here, we demonstrate that JNK phosphorylation was necessary for the generation of the ICD and that CD271 and pJNK directly interact. Pharmacologic inhibition of JNK not only delayed CD271 cleavage, but also significantly reduced treatment efficacy.

The role of JNK in β-amyloid–induced death was also previously shown in a neuroblastoma cell line (35). Specifically, β-amyloid induced death by activating p38 and JNK kinases in an ICD-dependent manner. Interestingly, JNK was not phosphorylated in CD271_KO cells, demonstrating the pathway to be receptor-dependent.

β-Amyloid generates pores in the mitochondrial membrane, leading to cytochrome c release (36). Moreover, β-amyloid can directly inhibit ATP production, altering the enzymatic mitochondrial machinery and inducing a strong production of ROS (37).

We showed that CD271 silencing induced mROS production 1,000-fold. Moreover, the pharmacologic inhibition of both pathways (i.e., JNK and ROS) demonstrated a synergistic effect by rescuing the cells from apoptosis and demonstrating the presence of two CD271-apoptotic pathways.

These findings highlight a role for CD271 in ROS inhibition. Indeed, although CD271^low cells treated with a BRAFi upregulate mROS, CD271 overexpression prevents ROS induction (Supplementary Fig. S8M). Despite these results, further studies are needed to understand to what degree this can abrogate the cell killing effect of MAPK inhibitors.

In conclusion, we revealed that ligand-mediated CD271 cleavage induces apoptosis in melanoma by releasing its ICD and preventing metastasis in vivo. Although this demonstrates the potential efficacy of CD271 targeting in vivo, the ability of the ECD to sequester Aβ^(25–35), and thereby attenuate this affect should be studied further for clinical application. However, amyloids have been used in patients successfully. Amyloid-peptides conjugated with a cell penetrating peptide showed a selective killing of cancer cells without toxicity in healthy cells (38). In addition, previous studies suggest amyloids in the formulation of long-acting drugs (GnRH analogs Degarelix and Antagon) for the treatment of prostate cancer and in assisted reproduction, respectively (39). Both drugs are in phase III clinical trials and showed a good safety margin. Thus, understanding the interaction between the Aβ^(25–35) peptide and CD271_ECD will facilitate developing drugs that mimic the effect of amyloid without toxicity.

N. Tiso reports grants from AIRC and Italian Telethon during the conduct of the study and grants from AIRC and Italian Telethon outside the submitted work. R. Dummer reports personal fees from Novartis, Merck Sharp & Dhome, Bristol-Myers Squibb, Roche, Amgen, Takeda, Pierre Fabre, Sun Pharma, Sanofi, Catalym, Second Genome, Regeneron, Alligator, T3 Pharma, MaxiVAX SA, Pfizer, and touchIME outside the submitted work. M.P. Levesque reports project focused research funding from Roche, Novartis, Molecular Partners, and Oncobit AG. No disclosures were reported by the other authors.

A. Saltari: Conceptualization, data curation, formal analysis, funding acquisition, validation, methodology, writing–original draft, project administration, writing–review and editing. A. Dzung: Formal analysis, validation, methodology. M. Quadri: Data curation, validation, investigation, visualization, methodology. N. Tiso: Resources, funding acquisition, investigation, visualization, methodology, writing–review and editing. N. Facchinello: Formal analysis, validation, methodology. A. Hernández-Barranco: Data curation, validation, investigation, visualization. S. Garcia-Silva: Data curation, formal analysis, validation, visualization, methodology. L. Nogués: Validation, methodology. C.I. Stoffel: Validation, visualization, methodology. P.F. Cheng: Software, formal analysis. P. Turko: Software, formal analysis. O.M. Eichhoff: Conceptualization, supervision. F. Truzzi: Conceptualization, data curation, supervision. A. Marconi: Resources, supervision, funding acquisition. C. Pincelli: Conceptualization, funding acquisition, writing–review and editing. H. Peinado: Conceptualization, methodology. R. Dummer: Conceptualization, resources, funding acquisition. M.P. Levesque: Conceptualization, resources, supervision, writing–original draft, writing–review and editing.

The authors thank Prof. Lukas Sommer and Dr. Gaetana Restivo for helping with cloning and Melanie Maudrich and Federica Sella for all their valuable help. This work was supported by the FIRC grant 2015 18145 (to A. Saltari), AIRC grant IG 2017 19928 (to N. Tiso), and the Telethon grant GGP19287 (to N. Facchinello). N. Facchinello is a fellow of the Umberto Veronesi Foundation.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.