CD147 Promotes Tumorigenesis via Exosome-Mediated Signaling in Rhabdomyosarcoma

Rhabdomyosarcoma (RMS) is an aggressive childhood soft-tissue tumor, with propensity for local invasion and distant metastasis. Exosomes are secreted vesicles that mediate paracrine signaling by delivering functional proteins and miRNA to recipient cells. The transmembrane protein CD147, also known as Basigin or EMMPRIN, is enriched in various tumor cells, as well as in tumor-derived exosomes, and has been correlated with poor prognosis in several types of cancer, but has not been previously investigated in RMS. We investigated the effects of CD147 on RMS cell biology and paracrine signaling, specifically its contribution to invasion and metastatic phenotype. CD147 downregulation diminishes RMS cell invasion and inhibits anchorage-independent growth in vitro. While treatment of normal fibroblasts with RMS-derived exosomes results in a significant increase in proliferation, migration, and invasion, these effects are reversed when using exosomes from CD147-downregulated RMS cells. In human RMS tissue, CD147 was expressed exclusively in metastatic tumors. Altogether, our results demonstrate that CD147 contributes to RMS tumor cell aggressiveness, and is involved in modulating the microenvironment through RMS-secreted exosomes. Targeted inhibition of CD147 reduces its expression levels within the isolated exosomes and reduces the capacity of these exosomes to enhance cellular invasive properties.


Introduction
Rhabdomyosarcoma (RMS) is the most common soft-tissue sarcoma in children and adolescents; it is considered rare in adults [1]. It consists of rhabdomyoblasts that fail to complete differentiation into mature skeletal muscle cells but nonetheless express myogenic transcription factors that control skeletal muscle differentiation [2]. RMS is classified, based on histology, into several subtypes. The most common are the embryonal (ERMS) and alveolar (ARMS) subtypes [3]. In pediatric and adolescent patients, ERMS constitutes

Exosome Isolation
Exosomes were isolated as previously described [15]. Briefly, exosome-free (exo-free) medium was prepared by ultracentrifugation of 40% FBS/DMEM at 100,000× g overnight at 4 • C. The resulting supernatant was filtered with 0.22 µm filter (Millipore, Darmstadt, Germany) and then diluted in a ratio 1/4 to obtain exo-free medium with 10% FBS. JR1 and Rh41 cells were cultured in exo-free medium for 72 h at 37 • C in 5% CO 2 in 15 cm plates. The culture medium was collected and centrifuged 3 times at increasing speeds (300× g for 10 min, 2000× g for 20 min, and 10,000× g for 30 min). The supernatant was collected in ultracentrifuge tubes, mixed with Exoquick solution (SBI, Mountain View, CA, USA) for exosome precipitation, and stored overnight at 4 • C. The solution was ultracentrifuged at 100,000× g for 70 min at 4 • C to remove contaminating elements. The pellet was collected, washed with 700 µL PBS, then centrifuged at 100,000× g for 70 min. The final pellet was resuspended in 300µL PBS for exosome use in functional assays, in lysis buffer for protein extraction, or in Trizol ® for RNA extraction.

Reverse Transcription Real-Time Polymerase Chain Reaction (RTq-PCR)
Total RNA was extracted using Trizol ® reagent (Life Technologies, Carlsbad, CA, USA) according to the manufacturer's instructions and treated with DNase I (Qiagen, Hilden, Germany). cDNA was synthesized using a RevertAid first-strand cDNA synthesis kit (ThermoScientific, Vilnius, Lithuania). Real-time PCR was done with the iQ SYBR green supermix kit in a CFX96 system (Bio-Rad Laboratories, Hercules, CA, USA). Amplification was performed using the following primers: GAPDH sense, AGCCAAAAGGGTCAT-CATCT; antisense, GGGGCCATCCACAGTCTTCT; CD147 primer pair: sense, GGCTGT-GAGTCGTCAGAACAC; antisense, ACCTGCTCTCTCGGAGCCGTTCA. PCR conditions included denaturation at 95 • C for 15 min, 40 cycles of 95 • C for 15 s, 72 • C for 1 min, and then annealing at 55 • C. GAPDH was used as an endogenous control. Experiments were done in triplicate using a CFX96 real-time PCR detection system (Bio-Rad Laboratories, Hercules, CA, USA), and data analysis was performed using the ∆∆CT method.

Cell Viability, Colony Formation, and Scratch Assays
For the MTT cell viability assay, 15,000 cells were seeded onto 96-well plates and cultured as above. The following day, medium was replaced by exo-free medium with exosomes. Control cells were incubated in exo-free medium. MTT cell viability assay (Roche Life Sciences, Penzberg, Germany) was performed according to the manufacturer's instructions. Results were computed as the mean percent absorbance of exosome-treated condition relative to control.
For colony formation assay, cells transduced with shCD147 or shScr control were plated at a density of 15,000 cells/plate onto 6-well plates coated with 1.5 mL of 0.8% agar. Cells were mixed with 1 mL complete medium containing 0.48% agar and added as a top layer. 1 mL complete medium was added to each well every 2 days to prevent drying, and incubated at 21% O 2 , 5% CO 2 . After 10 days, colonies were photographed and counted. To quantify the size of the colonies, the area of each individual colony (in pixels) was measured by ImageJ ® software (Version 1.53m, NIH, Bethesda, MD, USA).
For in vitro scratch assay, cells were seeded in a 24-well plate and incubated at 37 • C until they reached 80-90% confluence. A scratch/wound was created vertically at the center of the well using a 200 µL pipette tip, and the dead cells were washed with phosphatebuffered saline (PBS). Serum-free DMEM was added. Images were taken in randomly selected fields at 0, 8, and 24 h using an inverted light microscope at 10× magnification. To obtain the same field during the image acquisition, markings were used as reference points close to the scratch. For each image, distance between one side of scratch and the other (width) was measured in µm at indicated time using ImageJ ® software (Version 1.53m, NIH, Bethesda, MD, USA).

Cell Proliferation, Transwell Migration, and Transwell Invasion Assays
Human BJ fibroblasts were seeded onto a 24-well plate with 50,000 cells per well in 1 mL medium and incubated at 37 • C for 4 h. The medium was aspirated, and exosomes were added in 1 mL exo-free medium and incubated for 72 h. Then, cells were washed with PBS, trypsinized collected, and counted using a hemocytometer. For the transwell Cells 2022, 11, 2267 5 of 17 migration assay, D Falcon™ Cell Culture Inserts with 8µm pore size were used, and the same inserts coated with 10% growth factor-reduced matrigel were used for the invasion assay (BD Biosciences, Bedford, MA, USA). Human BJ fibroblasts were seeded onto the top chamber (50,000 cells per insert) in 300 µL exo-free medium and 500 µL of serum-free medium was added into the bottom chamber in a 24-well plate. Exosomes were added 4 h later onto the top chamber. Control cells were incubated in exo-free medium. The inserts were fixed after 24 or 72 h, stained with hematoxylin and eosin, mounted, cover slipped, air-dried and photographed, and the migrating/invading cells counted using ImageJ software.

Uptake of CFSE-Labeled RMS-Derived Exosomes by Fibroblasts
Exosomes from each condition were fluorescently labeled by a 20 µM concentration of carboxyfluoresceine diacetate succinimidyl-ester (CFSE) dye by incubating for 30 min at room temperature and in darkness. To stop labelling, approximately 5-fold volume of cell culture medium was added to the solution and ultracentrifuged at 100,000× g for 70 min at 4 • C to eliminate the unincorporated dye. The pellet was resuspended in filtered PBS, and exosomes were quantified as protein concentration by means of a Bradford assay. BJ fibroblast cells were seeded at a density of 50,000 cells/well in a 24-well plate 24 h prior to exosome addition with normal growth medium at 37 • C in a 5% CO 2 incubator. The next day, cells were treated with equal amounts (2 µg) of CFSE-labeled exosomes of each condition, in exo-free medium, incubated for 24 h, washed twice with PBS, then stained with Hoechst dye (20 µM). Pictures were taken on a Microscope Zeiss Axio (Carl-Zeiss, Dresden, Germany) at 20× magnification. Fluorescence intensity was measured by ImageJ ® software, using the pixel count tool, to quantify the amount of exosomes taken up by adherent cells.

Immunohistochemical Staining of Human Tumor Samples
All human studies were approved by the Institutional Review Board (IRB) at the American University of Beirut Medical Center (AUBMC). A total of 46 archived formalinfixed paraffin-embedded rhabdomyosarcoma tumor samples were identified for patients younger than 30 years of age, collected at the AUBMC over the period 2002-2018. Of those, 28 samples had enough material for sectioning and immunostaining, and were therefore included in the study. In addition, 5 samples were identified for patients younger than 30 years of age diagnosed between the period 2016-2020 at the Institut National de Pathologie in Beirut, Lebanon, bringing the total to 33 eligible samples. Clinical characteristics were linked to the tumor samples, with 26 being localized and 7 metastatic. Paraffin-embedded tumors were sectioned at 4 µm. Antigen retrieval was performed in a steamer using citrate antigen retrieval buffer (pH 6.0). Staining was performed using anti-CD147 (Santa Cruz Biotechnology, Heidelberg, Germany), using an ABC Elite Kit (Vector Labs, Burlingame, CA, USA) according to the manufacturer's protocol for detection by biotinylated secondary antibody and streptavidin conjugated to horseradish peroxidase followed by DAB substrate (DAKO, Glostrup, Denmark). Tissues were viewed under a light microscope at 40× magnification. Intensity of CD147 expression was scored as negative (0), weak (1), moderate (2), and high (3), by an experienced pathologist, while blinded to patient clinical characteristics or metastatic stage. Positive tumors were then given a second score based on percent of positive cells, with tumors that had >75% of cells positive receiving a score of 4, 50-75% a score of 3, 25-50% a score of 2, and <25% a score of 1. The product of the first and second scores constituted the CD147 positivity score.

Statistical Analysis and Imaging
All in vitro experiments were performed in biological and technical triplicate unless mentioned otherwise. Comparisons between experimental groups were performed using Mann-Whitney U test, except for categorical variables (positive versus negative staining of human tumor samples), where Fisher's exact test was used. For all tests, a p-value of Cells 2022, 11, 2267 6 of 17 less than or equal to 0.05 was considered statistically significant. A box plot was created using Microsoft Excel ®(version 2019) where each box represents the interquartile range (25th to 75th percentiles), the central horizontal line represents the median value, and the whiskers represent the range. All data is presented as mean ± standard deviation. Digital photomicrographs were obtained using a Zeiss Axio Observer Z1 microscope. Composite images were constructed using Adobe Photoshop CS6 ®software (Version 13.0 ×64, Adobe Systems, San Jose, CA, USA).

CD147 Is Expressed in RMS Cells, and Its Knockdown Diminishes RMS Cell Invasive Properties
Using the RMS cell lines JR1 and Rh41, derived from an ERMS and an ARMS tumor, respectively [35,36], and using four different shRNA targeting CD147 (shCD147), and scrambled construct (shScr) as negative control, we found that the two constructs shRNA-11 (sh11) and shRNA-12 (sh12) exhibited the highest knockdown of CD147 by both protein expression ( Figure 1A,B, with quantitation in Figure 1C,D), and RTq-PCR ( Figure 1E,F), and therefore these two shRNA constructs were subsequently used throughout the remainder of this study.  To investigate the effect of CD147 knockdown on RMS cell migration, we performed an in vitro scratch assay [37]. Knockdown of CD147 in JR1 and Rh41 cells significantly decreased migration compared with cells transduced with shambled construct (Figure 2A-D). To investigate effects on invasion, we performed anchorage-independent growth assays by culturing transduced cells in soft agar. After one week, the colony numbers of CD147 knockdown cells were significantly lower than the numbers of control transduced cells ( Figure 2E-F). In addition to a lower number of colonies, the size of the individual colonies was also significantly smaller in JR1 cells ( Figure 2G), while for Rh41 cells the size difference tended to be smaller, reaching statistical significance with one of the shRNA constructs ( Figure 2H).
To determine the effect of CD147 downregulation on growth and survival pathways in RMS cells, we evaluated the expression levels of proteins that have been previously demonstrated to be impacted by CD147 in other cell types [38,39]. As shown in Figure 3, there was a significant reduction in pro-Caspase 3 and BCL2 upon downregulation of CD147 by one of the two utilized shRNA in JR1 cells, but only a downward trend by the second shRNA, precluding a definitive conclusion but suggesting an effect on antiapoptotic pathways, while there was no discernible effect in Rh41 cells. Similarly, ERK phosphorylation tended to increase in response to CD147 knockdown in JR1 cells, but reached statistical significance for only one of the two utilized shRNA constructs, while there was no effect seen in Rh41 cells. VEGF levels were not reproducibly altered in either cell line ( Figure 3B). Thus, the interrogated pathways do not seem to explain the effect of CD147 on enhancing RMS cell proliferation, migration or invasion. To determine the effect of CD147 downregulation on growth and survival pathways in RMS cells, we evaluated the expression levels of proteins that have been previously demonstrated to be impacted by CD147 in other cell types [38,39]. As shown in Figure 3, there was a significant reduction in pro-Caspase 3 and BCL2 upon downregulation of CD147 by one of the two utilized shRNA in JR1 cells, but only a downward trend by the second shRNA, precluding a definitive conclusion but suggesting an effect on anti-apoptotic pathways, while there was no discernible effect in Rh41 cells. Similarly, ERK phosphorylation tended to increase in response to CD147 knockdown in JR1 cells, but reached statistical significance for only one of the two utilized shRNA constructs, while there was no effect seen in Rh41 cells. VEGF levels were not reproducibly altered in either cell line ( Figure 3B). Thus, the interrogated pathways do not seem to explain the effect of CD147 on enhancing RMS cell proliferation, migration or invasion.

Knockdown of CD147 Modulates the Effect of RMS-Derived Exosomes on Recipient Fibroblasts
Since CD147 is secreted in RMS-derived exosomes, we sought to examine the effect of CD147 downregulation on paracrine signaling of RMS cells. Examination of the protein content in exosomes derived from sh11and sh12-transduced JR1 and Rh41 cells showed that CD147 within exosomes decreased upon its suppression in the parent cells ( Figure 4A,B).    As expected, treatment of normal human BJ fibroblasts with either the control (scrambled shRNA transduced) JR1-or Rh41-derived exosomes resulted in a significant increase in fibroblast proliferation compared to exosome-free control-treated cells ( Figure 4C,D). This increase was abolished upon treatment with exosomes derived instead from cells with CD147 knockdown (Figure 4C,D). Similarly, while there was a significant increase in fibroblast migration when cells were treated with either JR1-or Rh41-derived control exosomes ( Figure 4E,F), this was again abolished when fibroblasts were treated with exosomes derived from CD147 knockdown cells ( Figure 4E,F). The same effects were seen when investigating fibroblast invasion through matrigel when treated with JR1-derived exosomes ( Figure 4G), whereas for cells treated with Rh41-derived exosomes, the number of invasive cells tended to be lower but did not reach statistical significance ( Figure 4H).
We considered whether CD147 knockdown affected the efficiency of exosome uptake by recipient fibroblasts, as one mechanism for a decreased paracrine effect could be through decreased uptake of exosomes. Indeed, when staining exosomes with the CFSE fluorescence marker and treating recipient fibroblasts with equal amounts of exosomes, we found that uptake of exosomes was decreased by around 50% when they were derived from CD147 knockdown RMS cells ( Figure 5A,B).
We considered whether CD147 knockdown affected the efficiency of exosome uptake by recipient fibroblasts, as one mechanism for a decreased paracrine effect could be through decreased uptake of exosomes. Indeed, when staining exosomes with the CFSE fluorescence marker and treating recipient fibroblasts with equal amounts of exosomes, we found that uptake of exosomes was decreased by around 50% when they were derived from CD147 knockdown RMS cells ( Figure 5A,B).  All settings of image processing were kept constant, and fluorescence intensities were calculated using ImageJ ® software. Bars represent standard deviation among triplicates. Asterisks (*) denote a statistically significant difference (p-value ≤ 0.05).

Expression of CD147 in RMS Tumors Correlates with the Metastatic Stage
To explore whether CD147 expression correlates with the aggressiveness of RMS tumors, we analyzed the expression of CD147 by immunohistochemistry in 33 human RMS tumor tissues (26 non-metastatic and 7 metastatic). Intensity of CD147 expression was scored as negative (0), weak (1), moderate (2), or high (3), as shown in Figure 6A. Results are shown in Figure 6B, stratified by metastatic group. Interestingly, while four of the seven metastatic tumors scored positive, none of the 26 localized tumors expressed CD147. Notably, immunoreactivity against CD147 was observed predominantly in the cytoplasm and/or stroma of tumor tissue with some membranous staining, unlike the typically observed membranous stain in normal cells ( Figure 6A). CD147 positivity showed a statistically significant association with metastatic stage, whether by intensity score ( Figure 6C, left panel), positivity score (computed as a composite of intensity and extent of positive cells within the tissue) ( Figure 6C, right panel), or by simple designation as positive or negative stain ( Figure 6D). mors, we analyzed the expression of CD147 by immunohistochemistry in 33 human RMS tumor tissues (26 non-metastatic and 7 metastatic). Intensity of CD147 expression was scored as negative (0), weak (1), moderate (2), or high (3), as shown in Figure 6A. Results are shown in Figure 6B, stratified by metastatic group. Interestingly, while four of the seven metastatic tumors scored positive, none of the 26 localized tumors expressed CD147. Notably, immunoreactivity against CD147 was observed predominantly in the cytoplasm and/or stroma of tumor tissue with some membranous staining, unlike the typically observed membranous stain in normal cells ( Figure 6A). CD147 positivity showed a statistically significant association with metastatic stage, whether by intensity score (Figure 6C, left panel), positivity score (computed as a composite of intensity and extent of positive cells within the tissue) ( Figure 6C, right panel), or by simple designation as positive or negative stain ( Figure 6D).

Discussion
Aggressive disease and distant metastasis are current challenges that hinder treatment strategies in rhabdomyosarcoma patients where overall survival outcomes remain unsatisfactory [40]. This highlights the need for identifying novel therapeutic targets for effective treatment, especially of locally invasive and metastatic disease.
CD147, a member of the immunoglobulin superfamily [27], is highly expressed on the surface of various types of cancer cells, including esophageal cancer, where it has been associated with worse survival outcomes and poor prognosis [41]; bladder cancer, where it correlates with cancer cell proliferation [42]; and breast cancer, where CD147 expression is an indication of increased invasive capacities, presence of metastasis and disease recurrence [43]. We identified CD147 to be expressed in RMS-derived exosomes across a panel of cell lines, which suggested that it may have a role in RMS paracrine signaling, potentially contributing to the known invasiveness and metastatic propensity of this tumor. Exosomes rich in CD147 have been identified in epithelial cancers such as ovarian and colorectal cancer, and found to be associated with higher stages of disease [34], and may promote angiogenesis in endothelial cells [44].
Our work shows that CD147 expression in RMS cancer cells contributes to cellular invasive and migration abilities, and its knockdown was also able to reduce the formation of clones in vitro, suggesting a role in promoting cancer cell stemness. This is consistent with the effect of CD147 knockdown in epithelial cancer types, where it was shown to reduce migration and clonogenic growth [45,46].
In our investigation, we could not demonstrate a reproducibly significant decrease in Bcl2 and pro-caspase-3 upon CD147 suppression, nor a reproducibly significant change in ERK phosphorylation, though the trends correlated with prior reports. Kulyar et al. had previously shown that chondrocytes exhibit enhanced survival when CD147 expression is elevated, which was in part associated with an upregulation of Bcl2 and downregulation of caspase-3 in their system [47]. In hepatoma cells, CD147 was also shown to protect cells from apoptosis by upregulating Bcl2 levels and promoting ERK signaling [48]. Also, our work showed that in RMS cell lines, there was no significant change in VEGF expression upon CD147 downregulation, unlike results observed in other cell lines, such as lung adenocarcinoma where CD147 can upregulate VEGF at both the mRNA and protein levels [28]. CD147 inhibition has also been associated with a decrease in VEGF in breast cancer [49], melanoma [50] and acute myeloid leukemia [51]. Further experiments are therefore needed to identify the specific downstream effectors of CD147 in RMS cells, preferably through an unbiased evaluation of downstream signaling pathways through RNA-sequencing and proteomic profiling.
The crosstalk between tumor and neighboring cells, such as fibroblasts and endothelial cells, plays a crucial role in the development and progression of tumors [52,53]. In this regard, CD147 expressed on cancer cells has been proven to stimulate the adjacent stromal cells to produce several MMPs altering the stromal microenvironment by modifying extracellular matrix composition which aids in tumor growth and invasion. CD147 is released from cancer cells into the tumor microenvironment either in a soluble form or associated to extracellular vesicles [54,55]. Previous studies have shown that shRNA-mediated knockdown of CD147 in malignant melanoma cells, followed by treatment of fibroblasts with the corresponding microvesicles and exosomes, decreased MMPs' enzymatic activity in recipient fibroblasts, suggesting that extravesicular CD147 downregulation is associated with decreased extracellular matrix remodeling and metastasis [56]. In our study, we provide evidence that exosomes shed by RMS cells stimulated invasive properties of normal human fibroblasts, and that this is at least partially mediated by CD147 and its downstream signaling. As such, CD147 may promote tumor invasiveness by acting either directly or indirectly on neighboring stromal cells, promoting pathways that enhance cell growth and motility. At least some of these effects are mediated through improving the efficiency of RMS-derived exosome uptake by recipient cells, though the exact mechanism by which exosome uptake is enhanced remains unclear. This is especially relevant given that we found that CD147 was exclusively expressed (by immunohistochemical staining) in primary RMS tumor samples of patients with metastatic disease. Interestingly, CD147 staining was observed mainly in the tumor stroma, which supports its role in the tumor microenvironment and the tumor/stroma cells' crosstalk. In other types of cancer CD147 is also correlated with disease stage, such as in gastric cancer, where its expression is elevated compared to adjacent normal tissues and is associated with the metastasis and TNM stages [57]. In pulmonary adenocarcinoma, CD147 expression also correlates with lymph node metastasis and can act as a prognostic biomarker of the advanced stage [58].
In our study, we used two representative RMS cell lines, one of ERMS (and fusion gene negative) and one of ARMS (and fusion gene positive) histology. The reproducibility of the findings across the two histologically distinct RMS cell lines, as well as the compelling data from the primary human tumor samples, suggest that CD147 is indeed a modulator of invasive properties in RMS. Based on our findings in the primary human tumor samples, and since the RMS cell lines were initially derived from metastatic tumors [36], we expect that CD147 expression is likely to be relevant primarily to metastatic tumors. However, verification will require evaluation of CD147 expression and the effect of its down-or upregulation, in a panel of primary cultures from metastatic versus localized RMS tumors, availability of which is currently limited by the relative rarity of the disease [59].
In conclusion, we have shown that CD147 enhances RMS tumor cell survival and growth and that RMS-derived exosomes harbor CD147 and stimulate fibroblast cell proliferation and invasion, which supports the hypothesis that paracrine signaling modulated by CD147 plays a role in tumor progression and metastasis. Further studies will focus on identifying the specific downstream effects of CD147 in both RMS cells as well as the RMS tumor microenvironment, for its potential investigation as a therapeutic target for aggressive and/or metastatic disease.  Informed Consent Statement: Patient consent was waived due to minimal harm study using archived tissue without identifiers, as determined by IRB.

Data Availability Statement:
The data presented in this study are available within the article and upon request.