Abstract

   Ewing sarcoma (EWS) is an aggressive mesenchymal tumor with unmet
   clinical need and significant social impacts on children, adolescents,
   and young adults. CD99, a hallmark surface molecule of EWS,
   participates in crucial biological processes including cell migration,
   differentiation, and death. EWS cells can release CD99 through exosomes
   (EXOs), specialized extracellular vesicles with major cell
   communication roles. Here we show that, as a consequence of CD99
   silencing, EWS cells deliver exosomes with oncosuppressive functions
   that significantly reduce tumor aggressiveness. These CD99-lacking
   microvesicles modulate gene expression of the EWS-recipient cells,
   reduce proliferation and migration, in turn inducing a
   more-differentiated less-malignant phenotype. The most relevant effects
   were detected on the activator protein-1 signaling pathway whose
   regulation was found to be dependent on the specific cargo loaded in
   vesicles after CD99 shutdown. Investigation of the miRNA content of
   CD99-deprived EXOs identified miR-199a-3p as a key driver able to
   reverse EWS malignancy in experimental models as well as in clinical
   specimens. All together, our data provide evidence that the abrogation
   of CD99 in EWS tumor cells leads to produce and release EXOs capable to
   transfer their antineoplastic effects into the nearby tumor cells,
   suggesting a novel atypical role for these microvesicles in reversion
   of malignancy rather than in priming the soil for progression and
   metastatic seeding. This conceptually innovative approach might offer a
   new therapeutic opportunity to treat a tumor still refractory to most
   treatments.

   Subject terms: Paediatric cancer, Sarcoma, miRNAs

Introduction

   Ewing sarcoma (EWS) is a highly aggressive mesenchymal tumor affecting
   children, adolescents, and young adults. Genetically, three major
   sequencing studies have shown that EWS is characterized by a striking
   paucity of somatic mutations^[40]1–[41]3.

   Apart from an oncogenic hybrid transcript that derived from the fusion
   of the EWSR1 gene with a member of the E26 transformation-specific
   (ETS) transcription factor family, the other most common alterations
   are found in STAG2 (15–20% of cases), CDKN2A (12% of cases), and TP53
   (6–10% of cases). The quiet genome of EWS, together with the high
   levels of inter- and intratumoral heterogeneity^[42]4, has prevented
   the identification and prioritization of preclinical candidates that
   might improve patient outcomes. Even the use of immune checkpoint
   inhibitors, which has been reported to be successful in several tumor
   types^[43]5–[44]7, has produced discouraging results in EWS^[45]8.
   Thus, very few advances in the treatment of EWS have been recorded in
   the last decade, and current therapy remains confined to conventional
   chemotherapy. Survival of patients with metastases at diagnosis remains
   dismal^[46]9,[47]10, and very few treatment options can be offered to
   the patients with localized tumors that recur after first-line
   treatments. In addition, even in more favorable situations in which
   patients with localized tumors at diagnosis respond well to therapy
   (survival rate up to 70–75% at 5 years), the patients experience
   significant side effects that limit their quality of life. Therefore,
   there is a strong demand from patients, families, and physicians for
   additional therapies. It is very likely that novel therapeutic
   approaches to combat this very aggressive and elusive tumor may be
   derived only from the development of innovative strategies exploiting
   new biomolecular features of the disease. The latest advances have been
   derived from cutting-edge technological applications. However, it is
   necessary to return to biology to properly contextualize this
   knowledge. In this paper, we focused on exosomes (EXOs), a subset of
   extracellular vesicles recognized to have a major role in cellular
   communications^[48]11. EXOs carry and transfer many types of cargo,
   including proteins, mRNAs and miRNAs, which are major components of the
   mechanisms used by tumor-derived EXOs to carry out their functions.
   Indeed, EXOs can be absorbed by recipient cells, thereby transferring
   both signaling proteins and genetic information and exerting observable
   and function-altering effects. We have recently demonstrated that CD99,
   a cell surface molecule that is a hallmark of EWS^[49]12, can be
   released by EWS cells, mainly through EXOs^[50]13. CD99 is involved in
   crucial biological processes, including cell migration, cell death, and
   differentiation^[51]14. EWS cells harboring the oncogenic EWS-ETS
   fusion gene are inhibited with regards to growth, migration and
   metastasis and are prone to differentiate toward neural lineage if
   deprived of CD99^[52]12. In this study, we thoroughly investigate the
   capability of EXOs lacking CD99 (CD99neg EXOs), released from
   CD99-silenced EWS cells, to interfere with the aggressive profile of
   tumor cells. Exploring the specific cargo loaded in these vesicles, we
   focused on miRNA content to identify players able to affect malignancy
   and to shape the genetic landscape of EWS neoplastic cells.

Results

CD99neg EXOs halt EWS aggressiveness and shape neural differentiation

   In a panel of eight patient-derived EWS cell lines, we evaluated the
   levels of released CD99 to confirm observations obtained in genetically
   modified cells^[53]13. Indeed, all the EWS cells released CD99, and its
   levels in cell supernatants were consistent with the cellular
   expression of the molecule (Fig. [54]1a) and were positively correlated
   with tumor cell aggressiveness (Table [55]1). Released CD99 was
   demonstrated to be mostly associated with EXOs^[56]13. Therefore, we
   used two EWS cell lines stably deprived of CD99, the TC-CD99-shRNA and
   CAR-CD99-shRNA variants^[57]12,[58]13, as donor cells of CD99neg EXOs
   (Supplementary Fig. [59]1a) and verified their impact on EWS
   malignancy. The EXOs were purified with the ExoQuick-TC (EQ) system
   from conditioned media of CD99-lacking cells, and the sizes and amounts
   of EXOs characterized by using NanoSight technology (Supplementary Fig.
   [60]1b). CD99 expression on the EXO surface was determined by flow
   cytometry (Supplementary Fig. [61]1c). CD99neg EXOs were then incubated
   with EWS cell lines. The internalization of the EXOs into acceptor
   cells was verified by staining the cells with the fluorescent fatty
   acid molecule BODIPY® FL C16, which, as it is incorporated into
   membrane phospholipids, made possible to label and visualize the
   EXOs^[62]15. A similar uptake by target cells of CD99neg EXOs or EXOs
   isolated from CD99-expressing parental cells (CD99pos EXOs) was
   confirmed (Supplementary Fig. [63]2a, b). EWS cells exposed to and
   incorporating CD99neg EXOs displayed lower cell growth than cells
   absorbing CD99pos EXOs (Fig. [64]1b). There was a reduction in cell
   proliferation, as shown by Ki-67-labeling index (Fig. [65]2a), but no
   significant induction of cell death (Supplementary Fig. [66]3a, b). In
   addition, EWS cells receiving CD99neg EXOs displayed increased neural
   differentiation, as shown by the expression of β-III Tubulin and the
   number of neurites (Fig. [67]2b). Thus, EXOs derived from CD99-silenced
   EWS cells are able to induce the same phenotype that we have proven to
   be associated with stable CD99 silencing^[68]12,[69]13. Accordingly,
   when CD99-deprived cells, which showed a more differentiated phenotype,
   received CD99pos EXOs, the expression of β-III Tubulin was inhibited
   (Supplementary Fig. [70]4a). In addition, EWS cells exposed to CD99neg
   EXOs exhibited significantly reduced migration (Fig. [71]3a, b),
   whereas CD99-silenced cells, which showed a reduced migratory
   capability with respect to controls, regained their ability when
   exposed to CD99pos EXOs derived from parental EWS cells (Supplementary
   Fig. [72]4b). To exclude aspecific effects that simply reflect a
   cellular response to EXO exposure at high concentration, both CD99pos-
   or CD99neg EXOs were used to treat the respective producer cells
   (Supplementary Fig. [73]5).

Fig. 1. Released CD99 affects cell growth in EWS recipient cells.

   [74]Fig. 1
   [75]Open in a new tab

   a Representative western blotting showing CD99 secreted in the
   supernatants of eight EWS cell lines; CD99 protein expression in the
   corresponding EWS cells is shown for comparison. GAPDH was used as a
   loading control. b Evaluation of cell growth in serum-free condition in
   EWS cells receiving CD99pos- or CD99neg EXOs isolated from the TC-71
   (purple bars) and IOR/CAR (green bars) experimental models. Columns
   show the mean values of at least two independent biological experiments
   with two replicates/each, and the bars represent the SE (*p < 0.05,
   **p < 0.01, Student's t-test, n = 2–4)

Table 1.

   Relationship between released CD99 and EWS malignancy in a panel of EWS
   cell lines
   Cell line OD CD99^a Released CD99 (score) Motility Growth in soft agar
   Bone metastases
   Incidence (%) Latency (days)
   r = 0.92 r = 0.95 r = 0.96 r = −0.46
   P = 0.027 P = 0.013 P = 0.011 P = −0.437
   6647 27,812 High 487 ± 44 1259 ± 99 9/9 (100%) 33 ± 1
   TC-71 20,199 High 148 ± 13 1048 ± 28 23/29 (79%) 44 ± 5
   IOR/CAR 10,304 Medium 22 ± 5 422 ± 70 2/10 (20%) 109 ± 5
   LAP-35 8441 Medium–low 23 ± 3 5 ± 2 0/5 (0%) ND
   SK-N-MC 5898 Low 42 ± 13 256 ± 21 1/5 (20%) 42
   [76]Open in a new tab

   ^a Released CD99 levels were measured by densitometric analysis after
   western blotting. Correlation was performed between CD99 OD and
   parameters of malignancy (migration, growth in anchorage-independent
   conditions, and capability to induce metastases). For each parameter,
   the correlation coefficient r and the significant p value P were
   determined by Pearson’s test (*P<0.05, Pearson’s correlation test)

Fig. 2. CD99neg EXOs reduce the proliferation rate and foster neural
differentiation in EWS cells.

   [77]Fig. 2
   [78]Open in a new tab

   a Ki-67 immunostaining was performed in a panel of EWS cell lines
   receiving or not receiving (CTR) CD99neg EXOs obtained from
   CD99-silenced cells. Ki-67-positive cells were counted, and the
   percentages are shown. Values are from one experiment representative of
   two independent biological experiments, with 6–10 fields/each. Columns
   show the mean values and the bars represent the SE (*p < 0.05,
   **p < 0.01,***p < 0.001, one-way ANOVA, n = 6–10). b Immunofluorescence
   staining of β-III Tubulin (green) in EWS cells receiving or not
   receiving (CTR) CD99neg EXOs isolated from TC-CD99-shRNA or
   CAR-CD99-shRNA cells. The nuclei were labeled by Hoechst 33258 (blue).
   Representative images are displayed. Scale bars: 50 µm. Quantitative
   analysis of neurite outgrowth (%) in EWS cells treated or not treated
   (−) with TC- or CAR-CD99neg EXOs from CD99-silenced models. Values are
   from one experiment representative of two independent biological
   experiments, with 6–10 fields/each. Columns show the mean values and
   the bars represent the SE (**p < 0.01, ***p < 0.001, one-way ANOVA,
   n = 6–10)

Fig. 3. CD99neg EXOs promote reductions in cell migration.

   [79]Fig. 3
   [80]Open in a new tab

   a Wound-healing in EWS cells receiving or not receiving (CTR) CD99neg
   EXOs purified from CD99-silenced TC-71 or IOR/CAR cell lines. The assay
   was performed over 24 h. Scale bars: 100 µm. b Motility assay with
   transwell chambers in EWS cell lines untreated (−) or pretreated for
   30 min with CD99neg EXOs. The data are shown as the mean ± SE of two
   independent biological experiments with three replicates/each
   (***p < 0.001, Student's t-test, n = 6)

   In addition, CD99 was equally expressed in EWS cells receiving CD99neg
   EXOs and in those not receiving EXOs (Supplementary Fig. [81]6a, b),
   suggesting that the modulation of cell behavior is specific and
   independent of the presence of CD99 on the recipient cell surface but
   rather dependent on EXO cargo.

CD99neg EXOs modulate the gene expression profile of recipient EWS cells

   EWS cells exposed to and incorporating CD99neg EXOs showed a different
   gene expression profile than parental cells. Twenty-four hours after
   cell exposure to CD99neg EXOs, 1839 transcript clusters, including
   those from protein coding and non-protein coding genes, were found to
   be differentially expressed compared with the levels in the parental
   TC-71 cell line (fold-change ± 1.5; Benjamini–Hochberg-adjusted p
   value ≤ 0.05) (Fig. [82]4a). We performed a functional annotation of
   the differentially expressed genes using GeneGo MetaCore software
   (Thomson Reuters) and DAVID bioinformatics tool^[83]16. Of note, the
   top-scored pathway based on the enrichment analysis of GeneGo software
   was Development Ligand‑dependent activation of the ESR1/AP‑1 pathway
   (Fig. [84]4b, Supplementary Table [85]1), with significant
   downregulation of FOS/JUN genes (Supplementary Table [86]2).
   Consistently, EWS cells exposed to CD99neg EXOs showed significant
   downregulation of activator protein-1 (AP-1) activity in luciferase
   assays, reduced expression of c-Fos (Fig. [87]4c), and inhibited
   expression of some validated AP-1 target genes, such as CCND1, MMP9,
   and MMP1 (Fig. [88]4d).

Fig. 4. CD99neg EXOs modulate gene expression in recipient EWS cells.

   [89]Fig. 4
   [90]Open in a new tab

   a Cluster analysis of TC-71 cells with or without CD99neg EXOs
   performed using a list of differentially expressed Affymetrix
   transcriptional clusters (adjusted p < 0.05). Genes (columns) and
   samples (rows) were grouped by hierarchical clustering (Pearson’s
   correlation). High expression and low expression were normalized to the
   average expression across all samples. b Top-scored map (the map with
   the the lowest p value) based on MetaCore pathway enrichment analysis
   of 1839 transcriptional clusters (corresponding to 508 annotated
   genes). Changes in gene expression after CD99neg EXOs addition are
   depicted on the map as a thermometer-like figures. Upregulated genes
   have an upward thermometer and a red color, whereas downward (blue)
   thermometers indicate downregulated genes in TC-71+TC-CD99neg EXOs vs
   TC-71 cells. c AP-1 transcriptional activity as assessed by luciferase
   assay and c-Fos expression as assessed by western blotting in TC-71
   cells receiving or not receiving (CTR) CD99neg EXOs. For the luciferase
   assay, the data are represented as RLU and are expressed as the fold
   induction over the control value. The values were normalized to Renilla
   luciferase activity. The data are shown as the mean ± SE of three
   independent biological experiments, with three replicates/each
   (**p < 0.01, Student’s t-test, n = 6). For western blotting, GAPDH was
   used as a loading control. d The mRNA expression levels of CCND1, MMP9,
   and MMP1 were evaluated by qPCR in TC-71 cells treated or not treated
   (−) with CD99neg EXOs. The results were quantified by the 2^−ΔΔCt
   method, and untreated cells were used as calibrators (2^−ΔΔCt = 1).
   GAPDH was used as a housekeeping gene. The data are shown as the
   mean ± SE of three independent experiments, with two replicates/each
   (**p < 0.01, Student’s t-test, n = 6)

   Other major pathways in DAVID analysis of modulated genes based on KEGG
   pathways and Uniprot keywords are related to neural differentiation:
   KEGG hsa04360: Axon guidance, p 0.03; UP_KEYWORD Neurogenesis, p 0.01
   (Supplementary Table [91]3). These results confirm that the phenotypic
   variations observed were at least partially due to gene expression
   alterations.

miRNA cargo of CD99neg EXOs suggests miR-199a-3p as a novel tumor suppressor
of EWS cells

   We evaluated the miRNA content of CD99pos EXOs isolated from the
   parental cell line TC-71 (three samples) and of CD99neg EXOs isolated
   from TC-CD99-shRNA cells (four samples) by microarray analysis. We
   detected 350 miRNAs in the CD99neg EXOs and 398 miRNAs in the CD99pos
   EXOs. In addition, we identified a signature composed of 56 miRNAs (10
   up- and 46 downregulated; Supplementary Table [92]4) that characterized
   the cargo of CD99neg EXOs. The results of a clustering analysis of
   CD99pos- and CD99neg EXOs according to the levels of these 56 miRNAs
   are presented in Fig. [93]5.

Fig. 5. CD99pos- and CD99neg EXOs display a different miRNA cargo.

   [94]Fig. 5
   [95]Open in a new tab

   Cluster analysis of TC-71 EXOs with or without CD99 performed using a
   list of 56 differentially expressed miRNAs (adjusted p < 0.2). Genes
   (rows) and samples (columns) were grouped by hierarchical clustering
   (Pearson’s correlation). High expression and low expression values were
   normalized to the average expression across all samples

   The most enriched miRNA inside the CD99neg EXOs was miR-199a-3p.
   Through quantitative PCR (qPCR), we confirmed the increased expression
   of miR-199a-3p in CD99-deprived cell lines and in CD99neg EXOs (Fig.
   [96]6a, Supplementary Fig. [97]7a). Exposure of TC-71 and IOR/CAR cells
   to miR-199a-3p mimic inhibited cell growth and migration (Fig. [98]6b;
   Supplementary Fig. [99]7b, c), as was observed for CD99neg EXO
   exposure. Notably, in silico analysis with different bioinformatic
   algorithms (DIANA-MicroT and TargetScan) identified c-Fos as one of the
   predicted targets of miR-199a-3p; accordingly, we found inhibition of
   AP-1 activity and reduction in c-Fos levels (Fig. [100]6c,
   Supplementary Fig. [101]7d) together with induction of neural cell
   differentiation (Fig. [102]6e) in mimic-treated cells, replicating the
   observed effects of CD99neg EXOs (Fig. [103]4c). To reinforce our
   observations, we transfected the parental, CD99 positive TC-71 cells
   with the miR-199a-3p mimic in turn isolating EXOs enriched for the
   expression of this miRNA (named CD99pos EXOs [miR-199a-3p]). When TC-71
   cells were exposed to these CD99pos EXOs [miR-199a-3p], effects on AP-1
   activity and c-Fos expression (Fig. [104]6d) were found comparable to
   those obtained by direct miRNA mimic transfection (Fig. [105]6c). In
   addition, we analyzed the consequences on CD99-low EWS cells
   (TC-CD99-shRNA) derived from treatment with CD99pos EXOs enriched or
   not with miR-199a-3p (Supplementary Fig. [106]8). We focused on cell
   growth, migration and AP-1 activity and, in all the cases, the data
   indicated a specific action of miR-199a-3p with respect to control
   (scrambled = SCR).

Fig. 6. miR-199a-3p acts as an oncosuppressor in EWS, reducing tumor
malignancy while inducing neural differentiation.

   [107]Fig. 6
   [108]Open in a new tab

   a Validation of miR-199a-3p expression by qPCR in CD99neg- and CD99pos
   EXOs and their respective producing cells. The relative miRNA levels
   were normalized to the expression of miR-16 (for EXOs) or RNU6b (for
   cells). The data are shown as the mean ± SE of three independent
   experiments, with two replicates/each (**p < 0.01, ***p < 0.001,
   one-way ANOVA, n = 4). b miR-199a-3p mimic (30 nm) reduced cell growth
   (left; trypan blue vital count) and cell migration (right; transwell
   chambers) in TC-71 cells. The data are shown as the mean ± SE of three
   independent biological experiments with two duplicates/each (*p < 0.05,
   one-way ANOVA, n = 6). c miR-199a-3p mimic (30 nm) inhibited AP-1
   transcriptional activity (as evaluated by luciferase assay) and c-Fos
   expression (as evaluated by western blotting) in TC-71 cells. The data
   are shown as the mean ± SE of three independent experiments with three
   replicates/each (**p < 0.01, Student’s t-test, n = 9). SCR, nonspecific
   control miRNAs. GAPDH was used as a loading control. d After
   transfection of miR-199a-3p mimic in TC-71, EXOs enriched for this
   miRNA were harvested (CD99pos EXOs [miR-199a-3p]) and used to treat
   parental EWS cells, TC-71. These enriched vesicles inhibited AP-1
   transcriptional activity (as evaluated by luciferase assay) and c-Fos
   expression (as evaluated by western blotting) in TC-71 cells. The data
   are shown as the mean ± SE of at least three independent experiments
   (***p < 0.001, Student’s t-test). SCR, nonspecific control miRNAs.
   GAPDH was used as a loading control. e miR-199a-3p mimic (30 nm)
   induced neural differentiation. Representative immunofluorescence
   images from one experiment representative of two, showing β-III Tubulin
   expression (green) and H-NF expression (green) in TC-71 cells after
   exposure to miR-199a-3p. The nuclei were labeled with Hoechst 33258
   (blue). Scale bars: 50 μm

   To validate these observations in a clinical setting, we determined the
   expression levels of miR-199a-3p in a series of localized, untreated
   primary tumors (n = 62) in comparison with metastatic lesions (n = 51:
   34 lung metastases and 17 bone metastases) from EWS patients. The
   expression of miR-199a-3p was significantly lower in metastases than in
   localized primary tumors (p < 0.05, Student’s t-test) (Supplementary
   Fig. [109]9), suggesting a relevant contribution of this miRNA to EWS
   aggressiveness.

Discussion

   EXOs, which are specialized extracellular nanovesicles ranging from 50
   to 150 nm in diameter^[110]17, are now viewed as pivotal mediators in
   intercellular communication^[111]18 and have become the focus of
   exponentially growing interest as potential novel diagnostics or
   therapeutics. In cancer, EXOs may influence cancer cell behavior and
   the tumor microenvironment through instructions that are believed to be
   delivered by specific mRNAs, miRNAs and proteins that characterize
   EXOs' cargo^[112]19–[113]21. Most investigations have explored the
   protumorigenic, proinvasive, and prometastatic effects of cancer
   cell-derived EXOs^[114]22. In this paper, we provide one of the few
   lines of evidence that cancer-derived EXOs may also have a detrimental
   effect on tumor cell aggressiveness. These oncosuppressor EXOs were
   obtained after silencing the cell surface molecule CD99 in the EWS
   cells. CD99 is a hallmark of EWS and is involved in several cellular
   processes, including cell adhesion, migration, death, and
   differentiation as well as intracellular protein trafficking,
   endocytosis, and exocytosis^[115]23. All EWS cells express the molecule
   at high levels, and CD99 is required for the maintenance of tumor
   aggressiveness. The molecule indeed prevents cell differentiation by
   communicating with EWS-FLI, the oncogenic driver of EWS, to maintain
   the proliferative and migratory as well as the metastatic capabilities
   of tumor cells^[116]12,[117]13. Here, we confirmed that the reversion
   of malignancy and the induction of neural differentiation induced by
   genetic silencing of CD99 in EWS cells^[118]12,[119]13 are replicated
   by the delivery of CD99neg EXOs to recipient EWS cells. EXOs derived
   from two EWS cell lines deprived of CD99 regulated gene expression in
   recipient cells and were able to inhibit cancer cell growth and
   migration while inducing cell differentiation in a panel of human
   patient-derived cell lines. These processes were oppositely regulated
   when CD99-deprived cells received EXOs derived from parental EWS cells,
   indicating that migration and differentiation in EWS cells are
   specifically regulated by CD99. Whether the presence or absence of this
   cell surface molecule determines the specific shuttling of particular
   cellular constituents into EXOs^[120]24 or whether EXOs packaging is
   just a reflection of the cell of origin^[121]25,[122]26 is still
   unclear. Our data tend to favor the idea that EXOs reflect the
   experience of the cell and become relevant messengers and powerful
   propagators of cell status.

   CD99neg EXOs were able to modulate the expression profile of EWS
   recipient cells, leading to up- or downregulation of genes relevant to
   EWS malignancy and neural differentiation. Among the top upmodulated
   genes (Supplementary Table [123]2) was the gene encoding galectin 3
   binding protein (LGALS3BP), a matricellular protein with a role in
   tumor progression that has been shown to inhibit EWS proliferation,
   invasion, and metastasis^[124]27. Among the top downregulated genes
   were early growth response 1 (EGR1), a transcription factor that has
   been described to be downregulated in the nonmetastatic sarcoma
   cells^[125]28, and Jun/Fos, members of the AP-1 family of transcription
   factors^[126]29. AP-1 members are frequently overexpressed in
   sarcomas^[127]30, and c-Fos has been shown to specifically contribute
   to tumorigenesis and metastasis of bone sarcomas^[128]31–[129]33. AP-1
   lies downstream of many signaling pathways and transduces extracellular
   stimuli such as growth factors, cytokines, and environmental stresses,
   modulating a variety of biological processes including cell growth,
   death, differentiation, and oncogenic transformation^[130]32. The
   activity of the AP-1 complex can be regulated at many levels, including
   by constitutive protein expression changes, dimerization partners,
   phosphorylation, and protein–protein interactions with other
   transcription factors^[131]34. Here, we further investigated also the
   miRNA-specific content of CD99neg EXOs to identify the particular
   regulators that may be responsible for the modified profile of the
   recipient cells. The most differentially increased miRNA in CD99neg
   EXOs compared with CD99pos EXOs was miR-199a-3p; quite interestingly,
   c-Fos is among the predicted targets of this miRNA. Exposure of EWS
   cells to miR-199a-3p mimic as well as to miR-199a-3p-enriched EXOs
   indeed replicated the phenotypic effects of CD99neg EXOs, including
   inhibition of AP-1 activity and the expression of its target genes
   (i.e., MMP9, MMP1, and CCND1). miR-199a-3p has been reported to
   suppress tumor growth, migration, invasion, and metastasis while
   increasing chemosensitivity in several tumors, including
   osteosarcoma^[132]35,[133]36; in addition, it has been described as an
   independent marker of improved survival in bladder cancer^[134]37. In
   this paper, we confirmed that miR-199a-3p is decreased in metachronous
   metastases compared with primary tumors, supporting its clinical
   relevance even in EWS. This indicates that the effects of CD99neg EXOs
   may be mediated, at least in part, by enhanced delivery of this miRNA.
   Nevertheless, other than miR-199a-3p miRNAs might be relevant as well.
   Indeed, we identified a signature composed of 56 differentially
   expressed miRNAs whose impact on EWS cell behavior needs further
   consideration.

   In addition, our previous observations have highlighted an important
   role also for the miR-34a^[135]13. Considering that miRNAs have
   multiple targets and their action is strictly interconnected, it is
   conceivable that even minimal fluctuations in the expression of
   multiple miRNAs will result in substantial changes in the mechanisms of
   tumor progression. A schematic representation of the so far described
   molecular connections that link CD99-deprived EXOs with malignant
   reversion of EWS cells is shown in Fig. [136]7.

Fig. 7. Schematic representation of the main molecular mechanisms that link
CD99-deprived EXOs with malignant reversion of EWS cells.

   [137]Fig. 7
   [138]Open in a new tab

   The figure depicts the role of two oncosuppressor microRNAs, which were
   enriched in CD99neg EXO cargo: miR-199a-3p and miR-34a, the first
   acting through AP-1 activity, the former exerting its action by means
   NOTCH/NF-kB axis. They both contribute to the reversion of malignancy
   resulting in the induction of neural differentiation together with the
   decrease of cell growth and migration

   Taken together, we have shown that modulating the expression of a
   fundamental cell surface molecule in EWS cells leads to the release of
   EXOs delivering a specific group of miRNAs with oncosuppressor
   functions that may be transferred to other tumor cells and
   substantially reduce malignancy. Development of best-practice models
   for the therapeutic use of EXOs is still a challenge^[139]38, but the
   use of EXOs as “natural” nanocarriers of anticancer therapies has
   gained growing interest, and engineering of EXOs is an active field
   offering new therapeutic opportunities to control primary tumors and
   metastatic disease^[140]39. Here, we provide a novel conceptual
   therapeutic approach against a tumor type that remains refractory to
   most treatments except conventional chemotherapy.

Materials and methods

Cell lines and primary cultures

   EWS cell lines were grown as previously described^[141]13,[142]40. The
   lines 6647 and TC-71 were kindly provided by T.J. Triche (Children's
   Hospital, Los Angeles, CA, USA); SK-N-MC, SK-ES-1, and RD-ES were
   provided by American Type Culture Collection, ATCC (Rockville, MD,
   USA); IOR/CAR and LAP-35 were previously established in our laboratory;
   and the A673 sarcoma cell line was provided by Dr. H. Kovar (St. Anna
   Kinderkrebsforschung, Vienna Austria). Stable CD99-silenced cells were
   obtained from the TC-71 and IOR/CAR cell lines as previously
   described^[143]12,[144]13. The cells were cultured in Iscove's modified
   Dulbecco's medium (IMDM; EuroClone, Milan, Italy) enriched with 10%
   fetal bovine serum (FBS) (EuroClone) supplemented with 100 U/ml
   penicillin and 100 mg/ml streptomycin and incubated at 37 °C in a
   humidified atmosphere at 5% CO[2].

   All cell lines were tested for mycoplasma contamination (MycoAlert
   Mycoplasma Detection Kit, Lonza) and authenticated by short tandem
   repeat (STR) polymerase chain reaction (PCR) analysis (last control:
   December 2017) using a PowerPlex ESX Fast System kit (Promega, Madison,
   WI, USA).

Isolation, characterization and labeling of EXOs

   EXOs were isolated from cell culture medium with the EQ (System
   Biosciences, Mountain View, CA, USA) or by ultracentrifugation (UC)
   methods according to the manufacturer’s instructions or standard
   procedures^[145]15.

   In EQ, briefly, cells were cultured for 24 h in IMDM supplemented with
   EXO-depleted FBS. The serum was depleted of bovine EXOs by UC at
   100,000 × g for 6 h, followed by filtration through a 0.2 µm filter
   prior to use. The conditioned medium was centrifuged at 3000 × g for
   15 min to remove cell debris. The supernatant was transferred to a new
   sterile tube and EQ precipitation solution was added for an overnight
   incubation at +4 °C. EXOs were isolated the day after by centrifugation
   at 1500 × g for 30 min at room temperature (RT). The EXO pellets were
   then resuspended in IMDM 0% FBS for in vitro functional studies, in
   Trizol for RNA extraction or in lysis buffer for western blot. The
   protein concentration of the EXOs was determined using a protein assay
   kit (Bio-Rad, Hercules, CA, USA), and in some cases, the number and
   size of the EXOs were directly tracked using a NanoSight NS300 system
   (NanoSight technology, Malvern, UK), configured with a 488 nm laser and
   a high-sensitivity sCMOS camera. Videos were collected and analyzed
   using NanoSight NTA software (version 3.0). For each sample, multiple
   videos 60 s in duration were recorded to generate replicate histograms,
   which were averaged.

   The fatty acid molecule BODIPY® FL C16 (16 μm) was included in the
   culture medium to label EWS cells, as it is incorporated into the cell
   membranes. EXOs (10 µg) were cocultured with 5,000 recipient cells
   grown in eight-well chamber slides (Falcon, NY, USA). After 3 h of
   incubation, the cells were fixed in 4% paraformaldehyde (Sigma-Aldrich,
   St Louis, MO, USA) for 10 min. Cell staining was analyzed with a Nikon
   A1R confocal microscope with a Plan Apo 60 ×/NA 1.4 DIC N2 objective.
   The uptake of fluorescent vesicles in recipient cells was assessed by
   flow cytometric analysis as previously reported^[146]41.

In vitro functional assays

   EXOs in equal amounts (10 µg/5,000 cells), purified from either
   parental or CD99-silenced EWS cell lines, were incubated in serum-free
   conditions with recipient cells at 37 °C before functional studies were
   performed.

Cell viability assessment with trypan blue dye

   A total of 50,000 cells/well were seeded in 24-well plates. CD99neg- or
   CD99pos EXOs were added after 24 h. The cells were exposed to the EXOs
   for up to 48 h before being harvested and counted with trypan blue
   vital cell dye (Sigma-Aldrich). Cells that were not treated were used
   as controls.

Cell proliferation assessment with Ki-67 staining

   The EWS cell lines were exposed to CD99neg EXOs 48 h and then fixed
   with methanol for 7 min at − 20 °C. The slides were incubated with a
   primary antibody against Ki-67 (Mib1; Sigma-Aldrich) overnight at +4 °C
   and then with an anti-mouse fluorescein isothiocyanate (FITC) secondary
   antibody (Thermo Fisher Scientific, Waltham, MA, USA) for 30 min at RT.
   The nuclei were counterstained with Hoechst 33258 (0.5 μg/ml;
   Sigma-Aldrich).

Cell death analysis

   EWS cell lines (TC-71, IOR/CAR, and LAP-35) were treated with CD99neg
   EXOs for 48 h. Detection and quantification of apoptotic cells were
   performed by flow cytometric analysis (FACSCalibur; Becton Dickinson)
   of Annexin-V-FITC-labeled cells, according to the manufacturer's
   instructions (MEBCYTO Apoptosis Kit, Annexin-V FITC Kit, MBL Life
   Science). Propidium iodide (PI) incorporation was evaluated in
   association with the fluorescent signal intensity to allow the
   discrimination between necrotic and apoptotic cells.

Migration assay (transwell chambers)

   Migration ability in the EWS cell lines was assessed using transwell
   chambers (Costar, Cambridge, MA) with an 8 μm pore size and
   polycarbonate filters. A cell suspension (100,000 cells) was pretreated
   with CD99neg- or CD99pos EXOs for 30 min in serum-free IMDM at 37 °C.
   The cells were seeded in the upper compartment, whereas IMDM plus 10%
   FBS was placed in the lower compartment of the chamber as a
   chemoattractant. Incubation was performed overnight at 37 °C in a
   humidified atmosphere. The migrated cells were fixed with methanol,
   stained with Giemsa dye and counted at 10 × magnification. A673 cells
   had a low capability to migrate, so they were seeded in 1% FBS in the
   upper compartment, and IMDM 20% FBS was placed in the lower compartment
   (gradient).

Wound-healing assay

   EWS cells were seeded in 24-well plates coated with fibronectin
   (3 μg/cm^2; Sigma-Aldrich). The cells were allowed to grow until 100%
   confluence was achieved. A pipette tip was used to obtain the cell free
   lane and the medium was then renewed and supplemented with purified
   EXOs from TC-CD99-shRNA or CAR-CD99-shRNA cells or the parental cell
   line TC-71. Images were obtained at time 0 and after 24 h under an
   inverted microscope (Zeiss, Inc., Thornwood, NY).

Neural differentiation

   To assess neural differentiation in EWS cultures, cells were plated at
   low densities (5,000 cells/cm^2) in standard medium. The cells were
   then exposed to purified CD99neg EXOs for 48 h. Immunofluorescence was
   performed on adherent cells grown on coverslips for 72 h, fixed in
   methanol/acetone 3:7 or 4% paraformaldehyde, permeabilized with 0.15%
   Triton X-100 in phosphate-buffered saline (PBS) and incubated with
   anti-heavy neurofilament (H-NF, cat. #2836; Cell Signaling Technology,
   Beverly, MA, USA) and anti-β-III Tubulin (cat. #T5076; Sigma-Aldrich)
   antibodies. The nuclei were counterstained with bisbenzimide (Hoechst
   33258; Sigma-Aldrich). Cell fluorescence was then evaluated using a
   Nikon Eclipse 90i microscope (Nikon Instruments, Florence, Italy)
   equipped with a Plan Apo VC 60 oil, NA 1.4 objective at RT. Images were
   captured under identical conditions using a digital camera (Nikon
   DS5MC) and the software NIS‐Elements AR 3.10 (Nikon). The same software
   was used to merge all images with double labeling. For the neurite
   outgrowth assay, cells were classified as differentiated if they
   exhibited an outgrowth extending from the cell that was at least 1.5
   times the diameter of the cell. At least 100 cells were counted on each
   slide.

Luciferase reporter gene assay

   A luciferase assay was used to evaluate the transcriptional activity of
   AP-1. Cells were seeded in standard medium in 24‐well plates (30,000
   cells/well) previously coated with fibronectin (3 μg/cm^2;
   Sigma-Aldrich). Transfection was performed 24 h later with 250 ng of
   AP-1 signal reporter (Qiagen, Hilden, Germany) using Lipofectamine 2000
   (Thermo Fisher Scientific) according to the manufacturer’s
   instructions. At 24 h after transfection, the same amounts of CD99neg
   EXOs were incubated with recipient cells for 30 min at 37 °C before
   assessment of luciferase activity. The cells were lysed, and luciferase
   activity was measured according to the manufacturer’s protocol using a
   Dual‐Glo Luciferase Assay System (Promega) with a GloMax Luminometer
   (Promega).

   The firefly luciferase luminescence signal was normalized to that
   produced by Renilla luciferase, which was included as an internal
   control. The data are represented as relative light units (RLU).

Immunoblotting analysis

   Western blot experiments were performed according to standard
   protocols. Samples were lysed in radioimmunoprecipitation assay (RIPA)
   buffer (50 mm Tris-HCl, pH 7.4, 150 mm NaCl, 0.1% sodium dodecyl
   sulfate (SDS), 1% Triton X-100, 5 mm ethylenediaminetetraacetic acid
   (EDTA), 1% deoxycholate) supplemented with protease inhibitors.

   Equivalent amounts of lysates were run on SDS gels under denaturing
   conditions and blotted onto nitrocellulose membranes. The membranes
   were incubated overnight with anti-CD99 (12E7; sc-53148; Santa Cruz
   Biotechnology, Santa Cruz, Dallas, USA), anti-GAPDH (sc-25778; Santa
   Cruz Biotechnology), anti-c-Fos (9F96; #2250; Cell Signaling
   Technology) primary antibodies. Anti-rabbit or anti-mouse antibodies
   conjugated to horseradish peroxidase (GE Healthcare, Little Chalfont,
   UK) were used as secondary antibodies. The proteins were visualized
   with an ECL Western Blotting Detection System (GE Healthcare). The
   signal for CD99 protein was quantified by using GS-800 imaging
   densitometer (Bio-Rad, Hercules, CA) and the Quantity One 4.6.9
   software (Bio-Rad). The values are represented as optical Density (OD).

RNA extraction and qPCR

   RNA from EXOs was isolated using miRNeasy columns (Qiagen) according to
   the manufacturer’s protocol; total RNA from cell lines was extracted
   with a TRIzol extraction kit (Life Technologies, Grand Island, NY,
   USA). Nucleic acid quality and quantity were assessed with a NanoDrop
   spectrophotometer (NanoDrop Technologies, ThermoFisher Scientific).
   Total RNA for each sample was reverse transcribed into cDNA using a
   High-Capacity cDNA Reverse Transcription Kit (Life Technologies)
   according to the manufacturer’s protocols. qPCR was performed on a
   ViiA7 system (Life Technologies) using TaqMan PCR Master Mix (Life
   Technologies). Predesigned TaqMan probes (Life Technologies) were used
   for Matrix Metalloproteinase 1 (MMP1, Hs00899658_m1), Matrix
   Metalloproteinase 9 (MMP9, Hs00957562_m1), Cyclin D1 (CCND1,
   Hs00765533_m1), CD99 (Hs00908458) and miR-199a-3p ([147]Hs002304).
   Relative quantification as performed with the ΔΔCT method, and the
   expression levels of the target genes were normalized to those of the
   housekeeping gene GAPDH (Hs99999905_m1), RNU6b ([148]Hs001093) or
   miR-16 ([149]Hs000391).

Microarray profiling and data analysis

   Gene expression patterns were assessed using a GeneChip® Human
   Transcriptome Array 2.0 (Affymetrix, Santa Clara, CA, USA) starting
   from 100 ng of total RNA. Microarray data are available at the GEO
   database (https://www.ncbi.nlm.nih.gov/geo) with the accession number #
   [150]GSE128681. The microarray target sample processing, target
   hybridization, washing, staining, and scanning steps were completed
   according to the manufacturer's instructions (Affymetrix). Raw data
   were analyzed with GeneSpring software v.14.9 (Agilent Technologies,
   Santa Clara, CA, USA). Specifically, .CEL files were imported and
   processed using the ExonRMA16 summarization algorithm. Quantile
   normalization was applied without baseline transformation.
   Differentially expressed genes were defined as having a ≥ 1.5-fold
   expression difference between groups and an adjusted p-value of less
   than 0.05 in a moderated t-test with Benjamini and Hoechberg correction
   for false positive reduction. Hierarchical clustering was performed for
   the TC-71 samples with the GeneSpring clustering tool using Pearson’s
   centered correlation as a measure of similarity. Pathway and network
   analyses of the differentially expressed genes were performed using the
   web-based software MetaCore (GeneGo, Thomson Reuters).

microRNA profiling and data analysis

   miRNA expression was investigated using an Agilent Human miRNA
   Microarray #G4870C and Sanger miRBase v.21 (Agilent Technologies).
   Microarray data are available at the ArrayExpress database
   (https://www.ebi.ac.uk/arrayexpress) with the accession number
   E-MTAB-7585. RNA labeling and hybridization were performed in
   accordance with the manufacturer’s instructions. The microarray results
   were analyzed with GeneSpring software v.14.9 (Agilent Technologies).
   Quantile normalization and log2 transformation were applied.
   Differentially expressed miRNAs were selected that had a 1.5-fold
   difference between CD99pos- and CD99neg EXOs and an adjusted p value
   < 0.20 in a moderated t-test with Benjamini and Hoechberg correction.
   Hierarchical clustering was performed using Pearson’s centered
   correlation as a measure of similarity.

Flow cytometry

   CD99 expression was evaluated by flow cytometry in EXOs isolated with
   EQ or by UC^[151]15. We used a staining protocol involving latex beads
   (aldehyde/sulfate latex #A37306; Invitrogen) that were incubated with
   20 μg of EXOs for 30 min at RT. Bovine serum albumin (BSA, 1%) was
   added to the samples, and the mixture was incubated overnight at 4 °C
   on a test tube rotator. After three washes in PBS with 0.5% BSA, the
   EXOs-bead mixture was incubated with an anti-CD99 antibody (12E7;
   sc-53148; Santa Cruz) for 30 min at 4 °C and with an anti-mouse FITC
   secondary antibody (#31569; Thermo Scientific). The data were acquired
   using a BD FACS LSR III (BD Transduction Laboratories, Lexington, KY,
   USA).

   CD99 levels were analyzed in EWS cells with a FACSCalibur flow
   cytometer (BD Transduction Laboratories). The antibodies used were
   anti-CD99 O13 (Signet Laboratories, Dedham, MA, USA) and an anti-mouse
   FITC secondary antibody (Thermo Scientific).

miR-199a-3p transfection

   Twenty-four hours after cell seeding, cells were transfected with
   pre-miR-199a-3p mimic or with nonspecific control miRNAs (SCR) (30 nm)
   (assays #AM17100 and #AM17110; Thermo Scientific) using a TransIT-X2
   Dynamic Delivery System (Mirus, Madison, WI, USA). The expression level
   of miR-199a-3p was determined by qPCR up to 48 h after transfection.
   Functional studies were performed as previously described for the
   CD99neg- and CD99pos EXO treatments.

   In parallel, from the transfected TC-71 cells, EXOs enriched for
   miR-199a-3p were harvested and used for biological assays in TC-71
   cells and CD99-deprived cells, TC-CD99-shRNA. The experiments performed
   reproduce the setting previously described in this section of material
   and methods.

Clinical samples

   A cohort of 113 patients with confirmed diagnosis of EWS treated at the
   Istituto Ortopedico Rizzoli (Bologna, Italy) was considered. The EWS
   patients underwent homogenous treatments based on local control of the
   disease (surgery and/or radiotherapy, depending on the tumor site) and
   systemic neoadjuvant chemotherapy. All the patients were enrolled in
   clinical studies, previously reported in detail
   approved^[152]42,[153]43. miR-199a-3p expression was evaluated by qPCR
   in localized primary tumors (62 cases) or metachronous metastases that
   developed from 6 months to 10 years after the end of treatments (51
   cases: 17 bone and 34 lung). The ethical committee of the Istituto
   Rizzoli approved the study (0019012/2016), and informed consent was
   obtained. The study was conducted in accordance with the Declaration of
   Helsinki ethical guidelines.

Statistical analyses

   All statistical analyses were performed using Prism version 7.0
   (GraphPad Software, La Jolla, CA). Differences among means were
   evaluated by one-way analysis of variance (ANOVA) with Tukey’s multiple
   comparisons test, whereas two-tailed Student’s t-test was used for
   comparisons between two groups. The data were considered statistically
   significant at p < 0.05. Pearson’s test was used to evaluate
   correlations.

Supplementary information

   [154]Supplementary Figures^ (4.5MB, pdf)
   [155]Supplementary Table 1^ (11.7KB, xlsx)
   [156]Supplementary Table 2^ (248.4KB, xlsx)
   [157]Supplementary Table 3^ (17.8KB, xlsx)
   [158]Supplementary Table 4^ (14.8KB, xlsx)
   [159]original blots^ (665.5KB, pdf)
   [160]Supplementary figure legends^ (37KB, doc)

Acknowledgements