Osteoblasts-derived Exosomes as Novel Communicators in Particle Induced Periprosthetic Osteolysis

The inammatory response to titanium implant-derived wear particles is considered as the hallmark of periprosthetic osteolysis, an event that cause pain, reduce patient motility, ultimately leading to the need of a revision surgery. Although macrophages are major cell players, other cell types such as bone cells can indirectly contribute to periprosthetic osteolysis, however the mechanisms are not fully understood. Exosomes (Exos) has been related with several bone pathologies, with growing body of literature recognizing them as actively shuttle molecules through with their cargo being completely dependent of external stimuli (e.g. Till the moment, the role of wear debris on osteoblasts exosomes biogenesis is absent and the possible contribution of Exos to osteoimmune communication and periprosthetic osteolysis is still in its infancy. Taking that in consideration, in this work we investigate the effect of wear debris on Exo biogenesis, where two bone cell models were exposed to titanium dioxide nanoparticles (TiO 2 NPs) similar in size and composition to wear debris associated with prosthetic implants. The contribution of Exos to periprosthetic osteolysis was evaluated performing functional tests stimulating primary human macrophages with bone-derived Exos.

start by exploring the effect of titanium dioxide nanoparticles (TiO 2 NPs) mimicking wear debris released from prosthetic devices on bone derived Exos biogenesis and their involvement in in ammatory responses that are considered as the hallmark of osteolysis. As a proof-of-concept, we investigate the effect of TiO 2 NPs on osteoblasts-derived Exos secretion using two human osteoblastic-like cell lines with different degrees of maturation widely used in bone research. Exos were the focus of this study since they are widely known to be a way for cells to get rid of unneeded or unwanted materials (such cellular contents and TiO 2 NPs), they contribute to the promotion of both innate and adaptative immunity but also there are some evidences suggesting Exos as mediators of chemical toxicity [16][17][18]. For the rst time, we reported that TiO 2 NPs internalization alters osteoblasts Exos biogenesis and cargo.
Interestingly, exosomal cargo stimulated human macrophages towards a mixed phenotype with consequent secretion of in ammatory cytokines that are described to contribute to periprosthetic in ammation and consequent osteolysis. We believe that studies revealing the precise composition of  [19][20]. All the suspensions were also characterized by high-resolution transmission electron microscopy (HRTEM, JEOL 2100 F operating at 200 kV equipped with an X-ray Detector (EDX -Energy-dispersive X-ray spectroscopy)).

MG63 and SAOS-2 cell viability, proliferation, internalization upon nanoparticle exposure
The cells (MG63 and SAOS-2) were exposed to TiO 2 NPs (100 µg/mL) for 72h. After viability, it was measured using 2 x 10 5 cell/mL using the Annexin V Dead Cell Apoptosis Kit (Life and Dead Kit, Life Technologies) in a ow cytometer (FACSAria III, BD Biosciences). The procedure is identical to the one previously reported. The analysis was repeated in three independent experiments. The internalization of NPs in MG63 and SAOS-2; all samples were xed with modi ed Karnovsky for 2h at room temperature and washed with 0.1 M cacodylate buffer with 1% uranyl acetate (diluted in water) (uranyl acetate, Sigma-Aldrich) overnight. The samples were dehydrated in a series of ethanol (Ethanol, VETEC) (30-100%) and nally included in Epon 812 resin (EMS). Ultrathin sections (70 nm) were cut in ultramicrotome and examined under transmission electron microscopy (TEM, Tecnai Spirit G2, FEI) and a three-dimensional reconstruction was performed using the focused ion beam (FIB) technique. At least ten cells from each group (control and 100 µg/mL) were analyzed. For proliferation assay, cells were plated (3 x 10 4 cell/cm 2 ) in 24-well plates (Corning) and after NPs exposure, cells were washed 3 times with PBS 0.01 M and then incubated in 0.1% Triton X-100 (TX-100, Sigma-Aldrich) for 10 min. The DNA was quanti ed by the PicoGreen® dsDNA quanti cation assay kit (PicoGreen®, Invitrogen) according to the manufacturer's instructions. Fluorescence of the plate was read on a microplate reader (Synergy HT, Bioek) using excitation at 480 nm and emission at 520 nm.
Classical TiO 2 NPs nanotoxicity test with primary human macrophages: Human monocytes were isolated from buffy coats from healthy blood donors, as previously described [21]. Brie y, buffy coats were centrifuged for 30 min, at 1200 g, without brake. The whitish layer containing peripheral blood mononuclear cells was collected and incubated with the RosetteSep Human Monocyte Enrichment Cocktail (StemCell Technologies), during 20 min and under rotation, following the manufacturer's instructions. This mixture was diluted (1:1) in phosphate-buffered saline (PBS 0.01 M) supplemented with 2% FBS (Biowest), added over Histopaque-1077 (Sigma-Aldrich) and centrifuged as previously described. The intermediate layer, enriched in monocytes, was collected and washed three times in PBS 0.01 M and centrifuged at 1300 r.p.m. for 6 min. On average, 90% of isolated monocytes were found to be CD14-positive. For monocyte-macrophage differentiation, 1 x 10 6 cell/cm 2 monocytes were seeded on circular glass coverslips with 30 mm diameter (6-well plates, Marienfeld), in complete RPMI medium (10% FBS and 1% Penicillin/Streptomycin (Gibco)), supplemented with 50 ng/mL rhM-CSF (Immunotools), for 7 days. Then, the medium was renewed without rhM-CSF supplementation for additional 3 days. On day 10, the macrophages were treated with TiO 2 NPs (100 µg/mL) or with different concentrations of Exos (4 x 10 3 , 4 x 10 5 and 4 x 10 7 Exos/mL (obtained from MG63 with and without contact with TiO 2 NPs)) and incubated at 5% CO 2 , at 37°C for additional 72h. Human monocytes were isolated from buffy coats from healthy blood donors, obtained at Centro Hospitalar Universitário São João (CHUSJ). All studies using these human samples were approved by CHUSJ Ethics Committee for Health (References 259 and 260/11), in agreement with the Helsinki declaration. Informed consent was obtained from all subjects.

Macrophage cytokine pro le activation
Following treatment with TiO 2 NPs and MG63 Exos (as described above), the human macrophage culture supernatant media was collected. Media were collected and centrifuged at 1200 rpm for 5 min to remove cell debris. The concentration of IL-6, IL-10, IL-1b, TNF-a, IFN-g and TGF-b was determined by ELISA (Biolegend), according to the manufacturer's instructions. As a control were also evaluated in media without addition of NPs.

Macrophage viability
Flow cytometry was used to measure cell viability. 5 x 10 5 cell/mL were analyzed using the Annexin V Dead Cell Apoptosis Kit (Life and Dead Kit, Life Technologies), according to manufacturer's instructions (through the combination of annexin V and propidium iodide (PI), it is possible to distinguish the % viable cells (Annexin V / PI ), apoptotic cells (Annexin V / PI and PI ) and, necrotic cells (Annexin / PI )). The analysis was performed with macrophages from at least 4 different donors. Cells were acquired on a FACSCanto ow cytometer (BD Biosciences) and analyzed with FlowJo software (v10.6.1).All analyzes were performed on a ow cytometer (FACSAria III, BD Biosciences).

Macrophage morphology and cytoskeleton evaluation
Upon NPs or Exos' treatments, macrophages were xed with 4% paraformaldehyde, quenched with 50 mM NH 4 Cl for 10 min and, after PBS 0.01 M washes, permeabilized during 5 min with 0.1% Triton X-100.
Macrophage surface markers: For the ow cytometry analysis of cell surface receptor expression, upon NPs or Exos treatments, macrophages were detached by incubation with Accutase (Grisp) at 37°C for 30 min and harvested by gentle scraping. The cells were washed and resuspended in FACS buffer (PBS 0.01 M, 2% FBS, 0.01% sodium azide) and stained with speci c conjugated uorophore-antibodies, in the dark, for 45 min at 4°C. Macrophages were incubated with the following antibodies: anti-human CD14-APC (clone MEM-18; Immunotools), CD163-PE (clone GHI/61; R&D Systems) and CCR7-PerCPCy5.5 (clone G043H7; Biolegend). Isotype-matched antibodies were used as negative controls. Cells were acquired on a FACSCanto ow cytometer (BD Biosciences) and analyzed with FlowJo software (v10.6.1). The median uorescence intensity (MFI) was calculated by subtracting the respective isotype control intensity. Data was obtained with macrophages from at least 3 different donors.

Reactive oxygen species quanti cation
The identi cation of reactive oxygen species (ROS) was performed using the 2,7-dichloro uorescein diacetate probe (H2D-CFDA). Cells were incubated (in samples interacted with TiO 2 NPs) with 2,7dichloro uorescein at an ambient concentration of 10 µM in the dark (10 min) using protocols provided by the manufacturer (Molecular Probes). Cultures were photographed on an inverted phase contrast microscope (Nikon TMS). They were quanti ed by measuring the uorescence intensity measured at the wavelengths of 488 nm and 530 nm in a microplate reader (Synergy HT, BioTek). The analysis was repeated in 3 independent experiments.

Macrophage lysosome staining
Lysosomes (in samples interacted with TiO 2 NPs) were stained with 500 nM lysotracker (Life) (green) in culture medium for 15 min at 37°C (following kit recommendations) and were incubated with DAPI (Sigma-aldrich) for core labeling. Cultures were photographed on an inverted phase contrast microscope (Nikon TMS).

Macrophage TiO 2 NPs internalization
The internalization of TiO 2 NPs in macrophage were evaluated as previously described.
Isolation and characterization of bone derived-exosomes: The minimal essential medium (α-MEM, Gibco) supplemented with 10% fetal bovine serum (V/V) (FBS, Gibco) was used for the culture of osteoblasts: MG63 (human pre-osteoblasts) and SAOS-2 (osteoblasts) mature humans). Exos-free fetal bovine serum was used in the experiments to obtain Exos [22]. Human osteoblast cell lines were supplied by the Rio de Janeiro Cell Bank (BCRJ), where they were packed in freezer vials and kept in liquid nitrogen. After thawing, the cells were expanded into 25 and/or 75 cm 2 cell culture asks (Corning). The cells used in the experiments were between 2° and 3° and were kept in a humidi ed incubator (5% CO 2 , 37°C).
Contamination of cells with bacteria, fungi and mycoplasma was analyzed.

Isolation of exosomes from MG63 and SAOS-2
To obtain Exos, 5 x 10 5 cells/cm 2 (MG63 and SAOS-2) were cultured in culture medium supplemented with 10% FBS (Exos-free) for 72h. Cells were washed 3 times with PBS 0.01 M and the medium was collected and centrifuged at 4000 x g, 4 for 10 min (Sigma). The supernatant was collected and ltered with 0.2 µm lters (Corning) and transferred to pollinator (Beckman) ultracentrifugation tubes (Beckman Optima L80-XP) using the SW41 rotor (Beckman) for 16h at 100,000 g at 4°C (Beckman Optima L80-XP) using the SW41 rotor (Beckman) for 6h at 100,000 g. Exos were then washed, where the pellet was resuspended in 1 mL of PBS 0.01 M and another ultracentrifugation was performed (Beckman Optima L80-XP). The supernatant was removed completely, and the Exos pellet was resuspended in 100 µL of PBS 0.01 M. The same procedure with the same number of cells (5 x 10 5 cell/cm 2 ) were performed after the interaction of the cells (MG63 and SAOS-2) with the TiO 2 NPs (5, 10, 25, 50 and 100 µg/mL) for 72h.
To remove the high number of NPs in the supernatants, the cells were washed 3 times with PBS 0.01 M and a fresh culture medium was added with 10% FBS (Exos-free). After 24h, the medium was collected and centrifuged as previously described. As controls, cell supernatants were used without interaction with NPs. To eliminate any trace of NPs, a sucrose gradient was performed on the Exos isolates from both cell models.
Puri cation of exosomes: Sucrose Gradient: Puri cation of Exos was performed with a continuous sucrose gradient following the protocol already described [22]. Brie y, the Exos were resuspended in 2 mL of 2.5 M sucrose, 20 mM HEPES/NaOH, pH 7.2. A linear gradient of sucrose (0.25M-2.0M sucrose, 20 mM HEPES/NaOH) was layered on top of the Exos suspension in an SW41 tube (Beckman). Gradients were centrifuged for 16h at 210,000 × g and 1 mL fractions were collected from the top of the tube. The densities were evaluated using a refractometer. Exos have been found to oat at densities ranging from 1.15 to 1.19 g/mL on continuous sucrose gradients [22].
Nanoparticle Trace Analysis (NTA): Malvern NTA (NanoSight NS300) was used to measure the size distribution and concentration of intact Exos isolated from cultures of MG63 and SAOS-2. The isolated Exos were resuspended in PBS 0.01 M (diluted 3:1000) and analyzed. The protocol applied was previously described [22].

MS):
The presence of Ti (titanium) in solution of puri ed Exos and/or not by sucrose gradient (Exos obtained after treatment of 50 µg/mL and 100 µg/mL) was investigated. The positive control was the stock solution of TiO 2 NPs (2 mg/mL) and the negative control was PBS 0.01 M (without contact with NPs). Titanium (48Ti) analysis was performed using an iCAP™ Q instrument (Thermo Fisher Scienti c, Bremen, Germany), equipped with a Meinhard® TQ + high-sensitivity nebulizer, a disconcerted (Peltiercooled) cyclonic spray chamber, a Standard quartz torch and a two-cone design (nickel sample and skimmer cones). High purity argon (99.9997%) (Gasin, Portugal) was used as a nebulizer and as a source of plasma gas. The operating parameters of the ICP-MS instrument were as follows: RF power (1550 W); plasma gas ow (14 L/min); auxiliary gas ow (0.8 L/min); nebulizer ow rate (1.01 L/min). Elemental isotope scandium (45Sc) was used as the internal standard. Samples were diluted 1:20 and their concentrations were derived from linear regression equations representing the relationship between the analytical signal (ICPS) and the elemental concentration of the corresponding standard solution.

Transmission Electron Microscopy (TEM) of exosomes
The Exos isolation and puri cation by gradient sucrose were applied as described before [22]. To remove aggregates of proteins and improve the morphological analysis of Exos, a wash (PBS 0.01 M) was performed. Basically, after isolation and puri cation, Exos were resuspended in PBS 0.01 M and transferred to polyallomer tubes (Beckman) and placed in the ultracentrifuge (Beckman Optima L80-XP) using the SW41 rotor (Beckman), and centrifuged for 16h at 100,000 g at 4°C. After, Exos were resuspended in 50 uL of PBS 0.01 M, and a drop of the suspension was placed onto a holey coated copper grid. Samples were contrasted in 1% uranyl acetate and visualized in the transmission electron microscope (TEM, Tecnai Spirit G2, FEI).

Protein Analysis by SDS-PAGE and Western Blots
For identi cation of the CD63, Exos protein, 20 exos/uL solution (from MG63 and SAOS-2) without interaction with NPs were added to 10 µL of 1x buffer (0.0625 M Tris-HCL, 2.5% SDS, 5% Glycerol) and frozen to -20°C. Samples were boiled 10 min at 100°C and run on 7.5% bis-polyacrylamide gel, 120 V were used to separate the proteins by molecular weight. Therefore, the gel was stained using 1% colloidal Coomassie blue (BIO-RAD). The bis-polyacrylamide gel was transferred to nitrocellulose membrane (Millipore) at 120 V for 120 min. Membranes were blocked in PBS/0.1% Tween-20 (PBS/T) with 5% skim milk, incubated with CD63 primary antibody diluted in PBS/T (Mouse -Novusbio), incubated overnight.
The membranes were washed four times in blocking solution, and nally incubated with conjugated secondary antibody (anti-CD63, Novusbio) followed by washing in PBS/T. Blots were developed using the ECL Plus (GE Healthcare) Western Blotting Detection System accordance with the manufacturer's instructions.

Identi cation of proteins present in exosomes by mass spectrometry
The proteins present in Exos isolated from human osteosarcomas (MG63 and SAOS-2) were analyzed. To this end, we investigated the proteins present in Exos samples obtained from control without contact with TiO 2 NPs of both cell models and in samples obtained after interaction with TiO 2 NPs for 72h. Samples were puri ed by sucrose gradient and washed to remove contaminating proteins. The concentration of Exos was normalized for all points (about 90 µg/mL protein at each point analyzed). An enzymatic digestion was then performed using 0.2 µg trypsin (Promega) diluted in ammonium bicarbonate (50 Mm, 30 min in overnight ice at 37°C). The extracted peptide mixture was lyophilized in 1% formic acid and transferred to StageTip (C18). A drying process was applied. Then 25 µL of methane acid (1%) (Sigma) was added. Samples were analyzed on a mass spectrometer (MS) (EDT-enabled Orbitrap Velos) (Thermo-Fisher scienti c) coupled to an EASY-nLC (Proxeon Biosystem) system using a Proxeon nanoelectrosplay source. The peptides were separated on a gradient of 2-90% acetonitrile in 1% methane acid in a PicoFrit analytical column (20 cm x ID75 µm, 5 µm particle size), with a ow of 300 nL/min for 27 min. The nanoelectrosplay voltage and temperature were adjusted to 2.2 kV and 275°C, respectively. The method con gured for LTQ Orbitrap Velos was data dependent analysis (ADD). SM scanning spectra (m/z 300-1600) were acquired on the Orbitrap analyzer after accumulation to a target value of 1 and 6 and the Orbitrap resolution was adjusted to r = 60,000. Thus, the 20 most intense peptide ions with charge states ≥ 2 were sequentially isolated to a target value of 5,000 and fragmented into the linear low energy CID ion trap (35% normalized collision energy). The signal threshold for triggering an SM/SM event has been set to 1,000 counts. Dynamic deletion was enabled with a size list of 500 and the deletion duration was 60 sec. The activation Q value was 0.25 and the activation time was 10 ms. Data were obtained using the Xcalibur software package and the samples were analyzed in three biological replicates. Peak lists (msf) were generated from les containing raw data using Considering the control versus treatment (p0.01) we obtained the proteins that increased and decreased after the treatment of TiO 2 NPs in the studied cells. Results are the mean ± standard deviation of triplicate independent experiments.

Statistical analysis
Data were presented as mean ± standard deviation (± SD). Gaussian distribution of samples was tested, and the statistical signi cance of the data was evaluated using One-way ANOVA and unpaired t-test was applied to obtain statistical signi cance of means. The P value is shown in gures and statistical signi cance was considered when p < 0.05. Each experiment was performed in three independent experiments with triplicates.

Results And Discussion
Osteoblast-derived exosomes exposed to nano-wear debris alter the cytokine pro le secreted by human macrophages The gradual in ammatory response to titanium implant-derived wear particles is the hallmark of periprosthetic osteolysis that is illustrated by an innate immune response [16][17]23]. This response is characterized by macrophage reactivity to implant wear debris, resulting in pro-in ammatory cytokines, chemokines and pro-osteoclastic factors release that leads to an increased osteoclastogenesis with subsequent bone resorption around the implant (periprosthetic osteolysis) [17,[23][24]. Besides macrophage activation in order to eliminate implant nano-debris that is driven by the chemical and the physical features of the particles, several others cells types including osteoblasts, monocytes, broblasts, osteoclasts, and mesenchymal stem cells present at the interface protheses-bone are affected and contribute to osteolysis [23][24].
We have previously reported that TiO 2 NPs, similar in size to wear debris associated with prosthetic implants, were internalized by primary human osteoblasts inducing the release of pro-in ammatory cytokines. Results that support the direct involvement of osteoblastic cells in in ammatory processes in response to wear debris [19][20]. However, at the implant-bone interface there is a series of differentiation events occurring during bone healing, where debris enter in direct contact with mature and immature osteoblasts among other cells. To understand how osteoblasts with different degrees of differentiation behave in face of nano-debris exposure, MG63 and SAOS-2 cell lines were used and stimulated with 100 µg/mL of TiO 2 NPs during 72h. TiO 2 NPs revealed a round-shape like morphology with a primary size of 20 nm (Fig. 1A). As it can be seen, in medium culture, TiO 2 NPs agglomerate in structures of about 140 nm with speci c proteins and ions adsorbed on them (Fig. 1B), these being the main events occurring at the nano-biointerface, as already reported [20]. The main physicochemical characteristics of TiO 2 NPs are presented in Fig. 1C. After TiO 2 NPs characterization, osteoblasts with different degrees of maturation were exposed to TiO 2 NPs and Exos isolated, following the work ow represented in Fig. 1D.
Our data demonstrated that TiO 2 NPs did not compromise osteoblasts viability and proliferation, con rming that bone cells with diverse degrees of differentiation react in a similar way to TiO 2 NPs ( Fig. 2A-D). Interestingly, in both cell lines, TiO 2 NPs were preferentially located inside multivesicular bodies (MVBs), the nascent of Exos (Fig. 2E, F and G). This nding indicates that NPs in endosomes maturation can proceed in two independent pathways: i) fusion of TiO 2 NPs-containing MVBs with lysosomes, where acid compartments ensure degradation of their content (although TiO 2 NPs are not degraded due to its physicochemical characteristics); ii) MVBs can also fuse with the plasma membrane, releasing their contents as Exos in the extracellular space [13][14][15][25][26]. High resolution images and 3D reconstruction obtained by focusing ion beam, reveal TiO 2 NPs isolated in multivesicular bodies (MVBs) in direct contact with Exos (Fig. 2G).
A detailed description of the cellular and the molecular mechanisms underlying the biological response to implant debris is widely described in literature [1-2, 4, 9, 11, 23]. Accumulating evidence has revealed the roles of osteoblast-derived Exos in mediating intercellular crosstalk in bone remodeling, however the possible contribution of Exos cargo to the systemic diffusion of toxic signals and activation of innate immune activation is lacking [27]. Taking that in consideration, we next questioned if TiO 2 NPs could affect osteoblast-derived Exos biogenesis and affect osteoimmune crosstalk. To assess the in uence of  Fig. 3A). Exos were isolated using established ultracentrifugation methods [22] and puri ed with sucrose gradient to eliminate any trace of TiO 2 NPs. Pro-in ammatory (interleukin 6 (IL-6), tumor necrosis factor α (TNF-α), interleukin 1 beta (IL-1 β)) and anti-in ammatory (interleukin 10 (IL-10), transforming growth factor beta (TGF-β)) cytokines, macrophage viability, cytoskeleton organization and polarization surface markers were evaluated. Excitingly, we observed that the higher concentration of osteoblast-derived Exos was clearly observed, with a higher density of fusiform macrophages, typical of an M1-like phenotype, as previously described [21]. Despite this, Exos did not alter the viability of macrophages (Fig. 3D).
No signi cant differences were observed on IL-1β, TNF-α or TGF-β secretion levels (see Figure in additional les 1B) neither on the expression of macrophage phenotype-associated surfaces markers ( Fig. 4 and Figure in additional les 3), which could be most likely explained by the existence of a mixed M1-M2 macrophage phenotype (as it happens at the implant site), and also to the inherent variability between blood donors. Macrophage polarization in joint tissues remains controversial. Although it has been reported that particle-induced response is prone to drive macrophages towards M1 phenotype, in some studies, the density of metallic particles on aseptic interface tissues did not lead to local preferential M1 macrophage polarization [12]. Clinical studies reveal that in the end-stage osteolysis is characterized by an unconventional macrophage activation pathway within periprosthetic tissues that is distinguished by generation of a speci c cytokine and chemokine milieu, but not by elevated levels of proin ammatory cytokines [33]. It is possible that the pro-in ammatory responses may be transient in nature, with M2 responses being supplanted in the later stages of disease progression [33][34].
Comparing the effect of osteoblast-derived Exos with direct contact of TiO 2 NPs on the behavior of macrophages, it was possible to observe comparable results with exception of macrophage viability ( Fig. 5A), which was directly compromised by ROS (Fig. 5B), overproduction and increased lysosomal activity (Fig. 5C) induced by TiO 2 NPs internalization (Fig. 5D). The decline in macrophage viability after treatment with NPs is probably due to their phagocytic role, causing high levels of internalized NPs, which in turn enter the lysosomal pathway, dramatically increasing ROS levels [12,[34][35]. This induction generates oxidative stress, which will lead to DNA/chromosomal damage and induce death and apoptosis [36].
In sum, our data indicates that Exos derived from pre-osteoblast pre-incubated with TiO 2 NPs modulate macrophages towards a mixed in ammatory pro le that has been reported in particle-induced osteolysis and metallosis cases [3].

Nano wear debris alter osteoblasts derived exosomes biogenesis and cargo
We demonstrated that the identi ed macrophage activation was a result of Exos. We have used sucrose gradient to isolated Exos from ExosMG63/ExosSAOS-2 and ExosMG63 + NPs/ ExosSAOS-2 + NPs. The isolated Exos express the CD63 marker, as shown by western blot analysis (Fig. 6A), they have the expected Exos diameter (75 ± 5 nm and 80 ± 4 nm) measured by nanoSight® nanoparticle tracking analysis and TEM micrographs, respectively (Fig. 6B-G). It is important to refer that the isolation and characterization of Exos followed the Minimal Information for Studies of Extracellular Vesicles ("MISEV") guidelines [37]. The role of osteoblast-derived Exos to maintain bone microenvironment was already described. They are known to facilitate a diverse of intracellular and intercellular signaling cascades that regulates both osteoclastogenesis and osteogenesis [25,38]. An increase of Exos secretion was observed upon TiO 2 NPs stimulation in both cell lines as observed in Fig. 6B and D (see Figure in additional les 4A and B). Exos are important mediators of cell-cell communication, and their biogenesis and fate can be altered by chemicals [14]. The application of Exos research in toxicology is still in its infancy, however emerging studies already reveal that Exos provide a mechanistic link between inhalation exposures and airway in ammation, and systemic effects [14,39]. Here we demonstrate for the rst time that nanoparticles can also impact Exos biogenesis. Our data clearly demonstrates that the observed activation of macrophages is exclusive of Exos and not a dual effect of Exos and NPs. No signi cant differences were observed on Exos diameter when osteoblasts cells were pre-incubated with TiO 2 NPs as evidenced by nanoparticle size analysis results (Fig. 6C and E). Transmission electron micrographs reveal that Exos populations presented a cup-shaped morphology (Fig. 6F and G) and that TiO 2 NPs were not able to enter in Exos. Inductively coupled plasma spectroscopy analysis (ICP-MS) of isolated Exos ( Fig. 6H and I) con rmed that the use of a sucrose gradient was an e cient method to obtain Exos free of metallic components. les 5B and C). At the proteome level, the signi cantly altered cellular components were cytoplasm, cytosol, extracellular region and cellular membrane. Exos shed by pre-osteoblasts and mature osteoblasts were also enriched in cytoskeleton proteins that are among the most abundant components that make up extracellular vesicles, proposing their contribution to Exos formation and/or the regulatory signaling functions. It is possible that as Exos fuse with target cells, the presence of cytoskeletal components on Exos could enhance cellular uptake and cellular responses like direct cell movement [38]. In both cell models, bronectin (FN) was also detected in Exos cargo. There are multiple evidences suggesting that FN facilitate Exos cell binding, uptake and motility [38 40]. Besides that, FN is also contributing to the induction of the mixed-in ammatory microenvironment observed, since human derived exosomal bronectin was reported to induce pro-in ammatory cytokines production [21]. Furthermore, bronectin and cytoskeleton proteins, heparan sulfate (HSPG) were detected on Exos derived from differentiated osteoblasts [38,41]. HSPG reside on the surface of Exos and are considered as key players in Exos biogenesis and uptake, having a great potential as biomarkers for liquid biopsy of glioblastoma and pancreatic cancer diagnostics [42].
Interestingly, treatment with TiO 2 NPs resulted in the enriched and under-represent of speci c exosomal proteins ( Fig. 8A and B). In Exos derived from MG63 (Fig. 8A) pre-incubated with TiO 2 NPs there was an enriched of urokinase-type plasminogen activator (uPA) and 60s ribosomal protein, plexin-A1, immunoglobulin kappa constant and immunoglobulin heavy constant gamma were under-represented.
Urokinase plasminogen activator is an enzyme that catalyzes the conversion of plasminogen to plasmin. From literature it was already reported by immunohistochemistry the localization of uPA in macrophages of periprosthetic tissue that phagocytosed metal, polyethylene, cement particles or accompanying pieces of necrotic bone [43][44]. The expression of the plasminogen activation system by macrophages containing phagocytosed particles suggests undegradable debris as a possible initiator stimulus for a proteolytic activation cascade, which can contribute to loosening of the prosthesis [45]. Interestingly uPA up-regulated was already identi ed in the pseudo-capsular and interface tissue around implants of patients with loosening of total hip prosthesis [46][47] and in osteoarthritic patients who undergone total joint replacement surgery [44]. Aberrant production of uPA and cytokines (IL-1ß and IL-6) has already been described in patients with osteoarthritis and rheumatoid arthritis, demonstrating the active role of uPA on disease progression [48][49]. Interestingly, uPA, C-telopeptide of type I collagen, IL-6, together with toll-like receptor, lipopolysaccharide-binding protein and myeloid-related protein-14, have been also suggested as serological biomarkers for accurate prosthetic joint osteolysis and infection diagnosis [50].
Moreover, recent in vitro and in vivo ndings indicate that the uPA/uPAR system is an active participant in the majority of infection and in ammatory diseases and might serve as a modulator of immunological responses [44,[51][52]. It has been implicated in macrophage adhesion, motility, proliferation, differentiation, invasion and matrix breakdown [43,52]. Previous studies have shown that uPA plays essential roles in immune cell response (macrophages chemotaxis), which aids in the elimination of infectious organisms, tissue in ammation, and regeneration [43]. As uPA is also considered as in ammatory macrophages activator [52], it is possibly contributing to the mixed-in ammatory microenvironment observed upon MG63 Exos macrophage stimulus.
Most of the identi ed enriched proteins were involved on immune response (Fig. 8B). Although no functional tests with macrophages were performed with SAOS-2 derived Exos, it was interesting to observe that the proteins comparing both models were dissimilar. Clearly the degree of cell differentiation affects Exos secretion and cargo upon NPs stimulation.
In conclusion, this study provides, for the rst time, insights into the capacity of the osteoblast derived Exos exposed by wear debris to stimulate macrophages towards a periprosthetic in ammatory pro le.
Besides activation of different cell types migration to the site of in ammation, exosomal cargo enriched in uPA is possibly contributing to the generation of pro-and anti-in ammatory signals. Although there is still the need of analyzing clinical samples, these earlier ndings are considered as the rst steps towards the proposal of exosomal cargo as a new generation of a non-invasive early diagnostic marker for nanoparticle induced osteolysis (Fig. 9). These results are clinically relevant since they reveal, for the rst time, that TiO 2 nano-debris with its low solubility can accumulate in the multivesicular bodies of osteoblast cells affecting Exos biogenesis. Moreover, there is no need of the direct contact of TiO 2 nanoparticles to stimulate macrophage-dominant in ammatory responses, since Exos with its speci c cargo are inducing a mixed in ammatory pro le. Based on our results and the observed cells at the taper interface, it can be suggested that monocytes recruited to the implant site by in ammation or resident macrophages in the surrounding tissue can be activated by osteoblasts derived Exos that were in contact with wear debris. It is important to refer that our data is based on proteomic landscape and efforts are under to focus on the genomic Exos content.     without exposure to NPs (ExoMG63_1:5). Three concentrations of Exos derived from MG63 not exposed (ExoMG63_1:3) and exposed to TiO2 NPs (ExoMG63+NPs_1:3) were used to stimulate human macrophages: 4 x 10³ Exos/mL (ExosMG63_1 and ExoMG63+NPs_1), 4 x 105 Exos/mL (ExosMG63_2 and ExoMG63+NPs_2) and 4 x 107 Exos/mL (ExosMG63_3 and ExoMG63+NPs_3). As control of all the experiments, we analyzed macrophages not exposed (control) and exposed to 100 µg/mL of TiO2 NPs during 72h (NPs 100 µg/mL). (B) Human macrophages from 4 healthy donors were exposed to different concentrations of MG63 derived Exos upon which pro-(IL-6) and anti-in ammatory (IL-10) cytokines were software. The results represent the mean ± standard deviation of at least three independent experiments performed in triplicate. *statistical differences between the groups marked.   comparisons. The results represent the mean ± standard deviation of at least 5 independent experiments carried out in triplicate. *Statistical differences between the groups marked.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.