Nano-Hydroxyapatite Accelerates Vascular Calcification by Inhibiting Lysosomal Acidiﬁcation in Smooth Muscle Cells.


 Background: Vascular calcification (VC) is a common characteristic of aging, diabetes, chronic renal failure, and atherosclerosis. The basic component of VC is hydroxyapatite (HAp). Nano-sized HAp (nHAp) has been identified as the initiator of pathological calcification of vasculature. However, whether nHAp can induce calcification in vivo and the mechanism of nHAp in the progression of VC remains unclear.Results: We discovered that nHAp existed both in vascular smooth muscle cells (VSMCs) and their extracellular matrix (ECM) in the calcified arteries from patients. Synthetic nHAp had similar morphological and chemical properties as natural nHAp recovered from calcified artery. nHAp induced rapid progression of VC by stimulating osteogenic differentiation and accelerating mineralization of VSMCs in vitro. Synthetic nHAp could also directly induce VC in vivo. Mechanistically, nHAp was internalized into lysosome, which impaired lysosome vacuolar H+-ATPase for its acidification, therefore blocked autophagic flux in VSMCs. The accumulated autophagosomes and autolysosomes were converted into calcium-containing exosomes which were secreted into ECM and accelerated vascular calcium deposit. Inhibition of exosome release in VSMCs decreased calcium deposition. Conclusions: Our results illustrated a novel mechanism of nHAp-induced vascular calcification. Understanding the role of nHAp in autophagy-lysosome-exosome pathway in SMCs could have great clinical significance in preventing the progression of VC.


Introduction
Vascular calci cation (VC) is a detrimental pathology highly prevalent in aging, diabetes mellitus, atherosclerosis, and chronic renal failure. Development of VC involves aberrant calcium deposition in the vessel wall, apoptosis of vascular smooth muscle cells (VSMCs) and the phenotypic transformation of VSMCs into osteoblast-like cells. [1,2]. During calci cation, cytoplasmic Ca 2+ and phosphate (Pi) incorporate with alkaline phosphatase (ALP) into exosomes that bud off the plasma membrane and associate with extracellular proteins, such as collagen. Crystals in the exosomes nally grow into nanosized hydroxyapatite (nano-HAp or nHAp) [Ca 10 (OH) 2 (PO 4 ) 6 ] [3]. Studies have shown that nHAp stimulated osteogenic transdifferentiation of SMCs in vitro and induced calci cation on extracellular matrix (ECM) [4][5][6], but there was no experimental result in vivo. The above studies suggested that nHAp may lead to a pathological feedback loop causing more in ammation, cell death, phenotypic change, matrix degradation, and calci cation. This may explain why calci cation progresses rapidly once microcalci cation exists. Whether nHAp can induce calci cation in vivo and the underlying mechanism by which nHAp affects the VSMCs remain unclear.
Autophagy is a "housekeeping" subcellular process for lysosome-mediated turnover of damaged cytosolic materials that are engulfed by a double membraned phagophore to form an autophagosome.
The autophagosomes fuse with single membraned acidic lysosomes to form autolysosomes, where the engulfed content is degraded and recycled [7]. Autophagy has been shown to regulate Ca 2+ homeostasis in VSMCs. Studies proved that autophagy dysfunction contributes to VC [8,9]. These studies demonstrated a protective role of autophagy in vascular disorders. Mice with VC had increased expression of the autophagy markers microtubule-associated light chain 3 II (LC3-II) and sequestosome 1 (SQSTM1/ p62), etc. in aortic VSMCs as compared to that in normal mice [10]. Evidence also demonstrated that increased cytosolic concentrations of Ca 2+ induced autophagosome accumulation, which was caused by a decline in autophagic clearance via lysosomes [11,12]. These studies indicated that impaired autophagy may contribute to VC, and lysosomal function may involve in VC. These data suggest that autophagy-lysosomal function plays an important regulatory role in vascular calci cation.
Lysosomes are single membrane-enclosed organelles present in all eukaryotic cells that contain large numbers of hydrolytic enzymes. Lysosomes exhibit their maximal enzymatic activity at a low pH (pH ≤ 5) and digest extracellular material from the endocytic pathway and intracellular material from the autophagic pathway. Disorders of lysosome acidi cation disrupt the clearance of autophagy [13].
Nanoparticles could lead to impairment of both autophagy and lysosome [14]. However, whether and how nHAp affects autophagy and lysosome in VSMCs remain to be elucidated.
Recently, secretary autophagy as a new concept was presented in addition to the traditional degradative autophagy [15]. Autophagic proteins were found on matrix vesicles (MVs) that were secreted out of cells [15] . MV membranes are enriched in phosphatidylethanolamine [16,17] which is also a main constituent of the mature autophagosome membrane [18]. These data suggest that autophagic vesicles can form MVs to be released into matrix outside of cells. In addition, one study proved that the calciumcontaining MVs in osteoblasts were transported via lysosome and secreted by exocytosis, indicating that lysosome plays a central role in mineralization [19]. Studies have shown that MVs are enriched with a concoction of calcifying enzymes and crystals of calcium phosphate HAp mineral [16,20]. Such MVs are released from osteogenic cells and facilitate HAp formation in ECM [21]. In context of atherosclerosis, VSMCs and macrophages are the primary source of these calci ed MVs which are released into the collagen rich matrix in the intima to promote atherosclerotic calci cation, directly leading to the formation of calci ed plaques [22]. It is reasonable to speculate that MVs may be entwined with the network of autophagic vesicles either at the stage of their formation or release during the process of vascular calci cation. Accumulating evidence indicates that matrix vesicles are secreted from multivesicular bodies and are enriched with the exosome markers CD63, CD9, CD81, and MHC I. Secretion of these exosome-like structures is regulated by sphingomyelin phosphodiesterase 3, and inhibition of this enzyme prevents vascular calci cation [23]. Comparative proteomics showed that VSMC derived exosomes were compositionally similar to exosomes from other cell sources but also shared components with osteoblast-derived MVs including calcium-binding and extracellular matrix proteins [24]. Despite the clinical importance of exosomes in triggering vascular calci cation, little is known of the link between exosomes biogenesis and VC induced by nHAp.
In this study, we demonstrated that nHAp induced VC through lysosomal impairment and blockage of autophagy ux in VSMCs and enhanced the release of calcifying-exosomes. We for the rst time illustrated the effect of autophagy-lysosome impairment on VC, and highlighted the link between nHApinduced de ciencies of lysosome activity and calcium deposition of VSMCs.

Methods And Materials
Characterization of Nanoparticles Synthetic nHAp (purity 97%; particle size less than 100 nm) was obtained from Aladdin Biochemical Technology (Shanghai, China). Human-derived HAp crystals were prepared from samples of human carotid arteries removed by endarterectomy. The HAp crystals were dissected as described previously [27]. The hardened, obviously calci ed regions were dissected from the atherosclerotic vessels and digested with collagenase (3 mg/mL) and elastase (1 mg/mL) overnight at 37°C. After digestion, the crystals were washed 3 times in Hanks' Balanced Salt Solution. Samples were then placed on a 70-m sieve (Gibco) and washed further. The remaining crystals ( 70 um) were crushed using a pestle and mortar into Hanks' Balanced Salt Solution. Washing of the crystals at this stage was achieved using ultracentrifugation at 100000 g to capture very small particles. The morphology and size of the crystals were characterized by scanning electron microscope (SU8010, HITACHI, Japan). The phase composition of the prepared nanosized crystals was con rmed by X-ray diffraction (XRD, PANALYTICAL, The Netherlands) with Cu-Kα radiation and Fourier transform infrared (FT-IR) spectroscopy (FITR-8400S, SHIMADZU, Japan).
Cell Culture Vascular SMCs were isolated from aortas of 8-week-old wild-type C57BL/6 mice (Slac Laboratory Animal). Brie y, the adventitia and endothelium were removed and the median layers were digested with 1 mg/ml collagenase type I (Worthington) and 0.714 U/ml elastase (Worthington) for about 1.5 h at 37°C. Cells were maintained in Dulbecco's modi ed Eagle's medium (Hyclone) containing 20% fetal bovine serum (Hyclone) and 100 U/mL penicillin, 100mg/mL streptomycin at 37°C in a humidi ed atmosphere with 5% CO 2 . Cells were used between passage 3-5 in the experiments.
Measurements of activity of Alkaline phosphatase (ALP).
The ATP activity of VSMCs was determined by using a luminescence ATP detection kit in accordance with the manufacturer's instructions (Beyotime Biotechnology, Shanghai, China). After SMCs were cultured with HAp for 7 days, SMCs were washed with cold PBS, scraped into 100 μL of lysis buffer (0.2% NP-40 with l mM MgCl 2 ) and incubated for 30 min on ice. Then the lysis was centrifugated at 5000rpm for 5 min and the supernatant was collected. All samples were stored under -80°C until use for ALP assay which was performed following the manufacturer's instructions (Beyotime Biotechnology, Shanghai, China). By catalyzing p-nitrophenyl phosphate (pNPP) to pNP, ALP activity is referred as the change of absorbance at 405nm, and then normalized to the protein level. Unit De nition: One ALP DEA unit equals hydrolyzation of 1 mole of p-nitrophenyl phosphate per minute at pH 9.8 at 37℃.

Alizarin Red Staining
After cultured at speci ed conditions for 14 days, SMCs on dishes were xed with 10% formalin for 30 min, then washed with double distilled water (ddH 2 O) twice and incubated with Alizarin Red (Solarbio Life Sciences, China) for ve minutes and washed with ddH 2 O twice to remove the excessive dye. After examination and photography under a microscopy, the dye on the cells was extracted with 100ul Hexadecyl Pyridinium chloride (Sigma-Aldrich) and the optical density (OD) at 560 nm was measured using a microplate reader (Spectra MAX 190, Molecular Devices, USA).
von Kossa staining A segment of the abdominal aorta was excised and xed in 10% formalin for 24 h, then dehydrated, embedded in para n, and cut into 5-µm -thick sections. The slides were depara nized, dehydrated and washed in distilled water (ddH 2 O),the samples were incubated with reagent A of von Kossa staining Kit (Solarbio Life Sciences, China) and placed under ultraviolet light for 10 min ,and then placed into reagent B for 2 min followed by staining with hematoxylin for the demonstration of cell nucleus.
Acridine orange (AO) staining VSMCs were exposed to 100 μg/mL sHAP for 24 h. The cells were then stained with 5μg/mL acridine orange (Aladdin, China) for 20 min at 37°C and 5% CO 2 . Then the cells were washed with PBS, followed by the addition of DMEM without serum to the cells. The uorescence images of the cells were immediately acquired on a uorescence microscope (Leica).

Analysis of EGFR degradation
After treated with nHAp for 24 h, VSMCs were incubated with serum-free DMEM containing 50 ng/ml of EGF for 0, 1, 2 and 3 h at 37 °C, and then the cells were either lysed and analyzed by western blot, or were xed in 4% formaldehyde, stained with anti-EGFR antibody and imaged by uorescence microscopy (Leica).

Calcium Quanti cation
To determine the calcium deposition in cells or exosomes, PBS-washed VSMCs on a dish or puri ed exosomes were dissolved in100 ul 0.6 N HCl for 48 h. Calcium content was measured by Calcium Assay Kit (Cayman Chemical, USA). The amount of calcium was normalized to the total amount of protein or the number of cells.

Exosome isolation
Exosomes were harvested by following a modi ed Exosome isolation protocol [52]. Con uent VSMCs were washed twice with PBS and cultured with DMEM medium with sHAp (100 μg/ml) for 3 days, and then the medium was collected and centrifuged at 10,000x g for 1 h to remove cells and apoptotic bodies. Exosomes were then harvested from the supernatant by centrifugation at 110,000x g for 70 min twice at 4℃ in an Ultracentrifuge (Optima L-90K; Beckman Coulter, USA). Exosomes were resuspended with PBS or 0.6N HCl for protein or calcium content measurement, respectively. Exos were observed by TEM (Tecnai G2 spirit, Thermo FEI, USA) which demonstrated typical cup shapes with diameters around 100 nm ( Fig S11A). Dynamic light scattering analysis further con rmed that the average diameter of Exos was approximately 133nm (Fig S11B). Western blot con rmed the expression of the Exos' marker proteins Alix, TSG101 and CD9 in Exos, but no GAPDH expression. While the VSMCs have no expression of CD9. And the above markers all don't exist in the supernatants after ultracentrifugation ( Fig S11C).

mRFP-GFP-LC3 Adenovirus Puncta Assay
For the analysis of autophagic ux, VSMCs were transfected with stubRFP-sensGFP-LC3 Adenovirus (JIKAI, China) following the manufacturer's instructions. After 36 h the transfected cells were starved or treated with 100ug/ml sHAp for 24 h. Then the cells were xed with 4% paraformaldehyde in PBS and then stained with DAPI, and imaged using a uorescence microscope (Leica, Germany).

Murine models of calci cation
Two different murine models of calci cation were created for this study. One is ectopic calci cation on soft tissue inducing via injection of Matrigel mixed with GFP-labeled VSMCs and sHAp. The other is by smearing HAp on the abdominal aorta. Mice (C57BL/6, 3-month old) were anesthetized with iso urane. The mice were shaved and the skin was disinfected with 75% ethanol prior to operation. For the rst one, three longitudinal skin incisions about 1cm in length on the dorsal surface of each mouse were made, and subcutaneous pouches were created by blunt dissection. The implants were 400 μl in volume containing 5mg sHAp or 6 x 10 6 SMCs or both, which was thoroughly mixed in advance, and were placed into pouch as following: left pouch, 200 μl sHAp + 200 μl Matrigel; right pouch, 100 μl sHAp + 100 μl GFPlabeled SMCs+200 μl Matrigel; lower pouch, 200 μl GFP-labeled SMCs+200 μl Matrigel. The incisions were closed with surgical sewing. For the second model, a longitudinal skin incision about 2cm in length on the abdominal surface of a mouse was made, and the abdominal aorta was isolated (between renal artery and the bifurcation of common iliac artery), the vascular adventitia was torn off. The artery was incubated with 0.9% NaCl, 0.5 M/l CaCl 2 , or sHAp (25 mg/ml 0.9% NaCl) for 15min with the aid of Stereo Microscope (SZ61, Olympus, Japan). The incisions were closed with surgical sewing. Calci cation in vivo was evaluated using micro-CT (U-CT-OI, Milab, USA).
Statistical analysis.
Data were expressed as means ± standard deviation (SD) and analyzed by Program Graph-Pad Prism version 6.0. Multiple comparisons were performed by one-way ANOVA followed by Bonferroni correction.
A value of p < 0.05 was considered as statistically signi cant.

Results
Nano-sized hydroxyapatite exists in human calci ed aorta To study the relationship between nHAp and vascular calci cation, human calci ed aorta specimen was collected. Calcium deposition in both media and adventitia of the artery was detected (Fig 1A). Nanosized HAp was observed on the surface of the vascular cells by scanning electron microscope (SEM) (Fig.  1B). Interestingly, we found that nHAp was adhered to the VSMC surface and there was nHAp in the lysosome of the cell (Fig 1C). In addition, abundant α-SMA positive cells expressing osteogenic marker Runx2 were observed on the calci ed artery (Fig 1D). These ndings indicate that nHAp internalization potentially contribute to the rapid progression of VC.

Characterization and cellular internalization of HAp crystals
To analyze the effect of HAp on calci cation in SMCs, we used synthetic nHAp (sHAp) to mimic human naturally crystallized HAp (hHAp) that was derived from atherosclerotic plaques of human aorta. The crystalline structures of synthetic and hHAp crystals were similar in morphology under SEM (Fig 2A), both were homogeneous and rod-like crystals. The size is about 100 nm for sHAp and ~80 nm for hHAp crystals. The X-ray diffraction (XRD) spectrum of synthetic and hHAp crystals showed both had eight diffraction peaks of crystals (Fig 2B), consistent with the standard HAp (JCPDS09-0432) [25]. The samples were further investigated by Fourier transform infrared spectrum (FT-IR) ( Fig 2C). Both crystals had the vibration peaks at 563, 603 and 1029 cm −1 which belonged to the asymmetric stretching vibration peaks of P-O in the PO 4 3− groups. Vibration peak at 3421 cm −1 were attributed to the O-H stretching vibration in HAp. Both XRD and FT-IR spectra data con rmed that synthetic and hHAp are similar and well crystallized in pure-phase.
To study engraftment of the HAp crystals by VSMCs, HAp was dyed bright green by the uorexon ( Fig  S1). As shown in Fig 1D, both hHAp and sHAp were internalized into VSMCs after the HAp crystals were mixed with VSMCs for 24 h. Under EM, irregularly aggregated HAp crystals in different size were visible inside cytoplasm of VSMCs after mixing cells with HAp for 24 h (Fig 2E). The images show that a number of HAp crystals were gathered in the lysosomes, autophagosomes and autolysosomes.

Nano-HAp induced osteogenic differentiation of VSMCs and calci cation deposition
To evaluate the effect of HAp on osteogenic differentiation of VSMCs, osteogenic marker proteins were examined. Expression of ALP, Runx2 and OPN was all increased signi cantly after VSMCs were treated with sHAp ( Fig 3A) or hHAp (Fig S2A) for 5 days as compared with those of the controls. The activity of ALP in VMSCs was also increased signi cantly after treated with HAp ( Fig S2B).
After VSMCs were treated with osteogenic medium (OM) or/and sHAp (VSMC sHAp ) for 14 days, signi cantly more calcium deposition in VSMC sHAp was observed (Fig 3B-D) as compared with VSMCs cultured in OM, whereas VSMCs under normal medium showed no calcium deposition. VSMCs treated with both OM and sHAp showed the maximal calcium deposition. Similar phenomena were also observed for VSMCs cultured with hHAp (Fig S2C-E). These data indicated that both sHAp and hHAp crystals could induce calci cation of VSMCs, and HAp crystals accelerated the development of calci cation compared with the normal OM. Such capacity was notably enhanced with OM. In view of the fact that human samples are hard to obtain, we may use sHAp instead of hHAp to conduct the following experiments on the basis of the above experiments.

Nano-HAp induced ectopic tissue calci cation in vivo
To investigate whether nHAp can directly contribute to mineralization of ECM in vivo, Matrigel mixed with VSMCs or/and sHAp were subcutaneously implanted into the dorsum of C57BL/6 mice ( Fig 4A). Greater calci cation was detected using micro-CT at the region injected with both VSMCs and sHAp as compared to that with sHAp only (Fig 4B-C) after implantation for 1 month. No calci cation was observed at the site injected with VMSCs only. Likewise, the calcium content in the recovered subcutaneous tissues with both VSMCs and sHAp was the highest among 3 groups (Fig 4B-C). Calci c deposits and cells were identi ed in the recovered subcutaneous tissues (Fig S3). Most of the cells in the recovered tissues with VSMCs and sHAp were GFP positive, indicating that they were the originally implanted VSMCs, and they were strong positive for osteogenic marker Runx2; whereas in the tissue with VSMCs only, signi cantly less Runx2 was detected in the GFP + VSMCs ( Fig 4D). As expected, neither GFP + cells nor Runx2 were detected in the recovered tissues with sHAp only. These results con rmed that nHAp possessed the ability to induce osteogenic differentiation of VSMCs and ectopic matrix mineralization in vivo.
To further con rm the effect of nHAp on vessel calci cation, we applied sHAp to the out surface of mouse abdominal aorta whose adventitia was torn off. Calci cation dots on the aorta being applied with either sHAp or CaCl 2 were detected by enhanced micro-CT (Fig 4E) or directly viewed under stereomicroscope ( Fig 4F) 14 days after the surgery, while application of control NaCl to aorta resulted in no calcium deposits, which was further con rmed by the von Kossa staining of the vessels (Fig 4G).
Signi cantly more Runx2 was co-localized with SMCs marker a-SMA on the aorta treated with sHAp or CaCl 2 as compared to that treated with NaCl ( Fig 4H). These results indicate that nHAp could induce osteogenic differentiation of SMCs and vascular calci cation in vivo.

Nano-HAp induced accumulation of autophagosomes in VSMCs
Under TEM, we observed increased number of double-membrane autophagosomes in VSMC HAp (Fig 5A and Fig S4A) as compared to the control cells, indicating that internalization of HAp could cause the accumulation of autophagic organelles. Indeed, almost 4-fold more LC3, a classical marker for autophagy, was detected on the calci ed artery than that on the control (Fig 5B). More puncta formation of endogenous LC3 was observed in VSMC sHAp , while less and evenly dispersed LC3 throughout the cytoplasm was seen in the control cells ( Fig 5C). Furthermore, signi cantly higher LC3-II was detected in VSMC HAp than the untreated controls ( Fig 5D, and Fig S4B), indicating that HAp stimulated the conversion of cytoplasmic LC3-I to membrane-conjugated LC3-II. Taken together, these data demonstrate that both types of HAp crystals induced accumulation of autophagosomes in VSMCs.
The classical pathway of autophagy induction involves inhibition of mTOR, usually induced by energy reduction. We found that sHAp did not alter the phosphorylation of either mTOR or AMPK (Fig S5A), suggesting that sHAp stimulates formation of autophagosomes through a pathway different from starvation-induced mTOR inhibition. Rapamycin, an inhibitor of mTOR, which induces autophagy, additively increased LC3-II level in VSMC sHAp (Fig S5B), further con rming that sHAp-stimulated production of autophagosomes involves a pathway independent of mTOR inhibition.
Nano-HAp inhibited autophagic degradation without effecting fusion of autophagosome with lysosome.
Accumulation of autophagosomes could be a result of autophagy induction or blockage of autophagosome clearance which involves fusion of autophagosomes with lysosomes and thereafter degradation of the resulted autolysosomes [26]. S-HAp did not increase the mRNA of LC3 (Fig S6A), indicating that HAp-induced increase in LC3 was not due to enhanced LC3 expression for more autophagosome formation.
Protein sequestosome 1 (p62) is a selective autophagic receptor that is incorporated into autophagosomes and preferentially degraded along with other substrates by lysosomal hydrolysis. Level of p62 protein was increased after VSMCs were treated with sHAp or hHAp ( Fig 6A, and Fig S4B).
Immunostaining showed more cytoplasmic p62 proteins were assembled into aggregates in VSMC sHAp (Fig S7). Such effect of HAp on accumulation of p62 was not due to more p62 expression as its mRNA level was not different from that in the untreated cells ( Fig S6B). When chloroquine (CQ), an inhibitor of autophagosome-lysosome fusion, was added into the culture, both LC3II and p62 were additively elevated ( Fig 6B). These data indicate that nHAp could block the autophagy ux.
When 3-methyladenine (3-MA), an inhibitor of autophagosome synthesis, was added to the starved cells, LC3-II was signi cantly reduced; however, such 3-MA mediated reduction of LC3-II did not occur in VSMC sHAp (Fig 6C), also suggesting that nHAp blocks autophagysome degradation.
Autophagic ux was further examined by transferring mRFP-GFP-LC3 fused genes using adenoviral vector. Under starvation, fusion of autophagosomes with lysosome lowered the pH that quenches the GFP signal (Fig 6D). When the cells were cultured with sHAp, starvation-induced quenching of GFP uorescence was signi cantly reduced, con rming that sHAp blocks autophagic flux and autophagic degradation.
To investigate the formation of autolysosomes, co-localization of autophagosomal LC3 with lysosomeanchored Ras-related protein 7 (Rab7), a GTPase, was examined to implicate the fusion of autophagosome with lysosome. LC3 was poorly colocalized with Rab7 in control VSMCs, while starvation enhanced the colocalization, demonstrating the formation of autolysosome (Fig 6E). Nano-HAp treatment of VSMCs resulted in a similar colocalization of LC3 and Rab7 as that of starved cells, indicating that the formation of autolysosomes was not affected by nHAp. This was further con rmed by colocalization of LC3 with the lysosomal associated membrane protein 1 (LAMP1) (Fig S8).
Autophagy is a degradation process in cooperation with lysosomes. Since nHAp does not affect the fusion of autophagosome with lysosome, we then examined the function of lysosomes. The expression of early endosomal antigen 1 (EEA1), and LAMP1 both increased in VSMCs after sHAp treatment ( Fig  S10), indicating that sHAp may enter VSMCs through endosome/lysosome pathway. Entry of sHAp into lysosome was observed after mixing sHAp with VSMCs for 24 h (Fig 7A), which could impair the function of lysosomes. We then evaluated the ability of lysosomal degradation using a degradation assay of epidermal growth factor receptor (EGFR). EGFR was localized primarily on cell surface and was internalized and clustered into puncta after binding with its ligand EGF. Clustered EGFR was observed at 1 h post EGF treatment, and then was degraded later in the control VSMCs (Fig 7A). However, such EGFR puncta were retained in VSMC sHAp , indicating weakened lysosomal proteolysis. This was further con rmed by western blot analysis showing that total cellular EGFR was retained in VSMC sHAp while a timely reduction of EGFR was observed in the control cells (Fig 7B). These data indicate that nHAp impairs lysosomal degradation capacity.
Next, we examined the acidi cation ability of lysosomes since acidi cation is essential for the maturation and activation of most lysosomal enzymes. Acridine orange (AO) staining was used to assess the acidi cation of lysosomes. In cytosol, AO exists in nonprotonated monomeric form and emits green uorescence. When it enters acidic lysosomes, the protonated form of AO aggregates and uoresces bright red. As shown in Fig 7C, red uorescence of AO was dramatically reduced in VSMC sHAp as compared with that in control cells, indicating a reduced acidi ed-compartments in VSMC sHAp .
In addition, more pre-matured cathepsin D (CTSD) at 43-kDa, a lysosomal protease, and less matured CTSD were detected in VSMC sHAp in comparison with that in the control cells (Fig 7D), indicating a suppression of the conversion from pre-matured to matured form.
To understand the alteration of acidi cation and subsequent lysosome immaturation in VSMC HAp , the subcellular localization of V-ATPase, an ATP-driven proton pump that imports protons into lysosomal lumen for the acidi cation of the compartment was examined. V1D subunit of V-ATPase was colocalized strongly with lysosome membrane protein LAMP1 in the control cells, while this colocalization was disrupted in VSMC sHAp (Fig 7E), suggesting a failure in the targeting of V-ATPase V1D subunit to lysosomes in VSMC sHAp . These results demonstrated that nHAp impairs lysosomal acidification.
To investigate the relationship between autophagy and calci cation, we utilized autophagy stimulator rapamycin (RAPA) and autophagy inhibitor chloroquine (CQ) Fig S9 . RAPA and CQ themselves did not affect calci cation in VSMCs (Fig 7G). When sHAp was added along with either RAPA or CQ, stronger Alizarin red staining with more calcium deposition in VSMCs was detected (Fig 7G), suggesting that both RAPA and CQ enhanced the calci cation induced by sHAp.
Nano-HAp promoted release of exosomes containing LC3 and LAMP1.
Since exosome budding off plasma membrane has been proposed as a mechanism of calci cation, and the interplay of autophagy and exosomal secretion was reported recently, we postulated that the accumulated autophagosomes and autolysosomes can be converted into exosomes and then secreted out of cells to lead calci cation. To con rm that, exosomes (Exos) were puri ed from the conditioned medium of VSMCs and characterized (Fig S11). More Exos (18.8 ± 0.6 mg/10 7 cells) were harvested from VSMC sHAp than that from VSMCs (7.5 ± 0.5 mg/10 7 cells) (Fig 8A). There were more exosome speci c marker CD9 as well as autolysosome-associated LC3 and LAMP1 proteins in Exos from VSMC sHAp than VSMCs ( Fig 8B). Calcium content in Exos from VSMC sHAp was signi cantly higher (139.7 ± 1.6 ng/10 7 cells) than that from VSMCs (64.4 ± 1.6 ng/10 7 cells) (Fig 8C). When GW4869, an inhibitor of neutral sphingomyelinase (nSMase), was added to the culture to block Exos release, signi cantly less Exos were harvested from VSMC HAp . Exosome speci c protein CD9, and autolysosome-associated LC3 and LAMP1 were correlatively reduced in Exos from VSMC sHAp treated with GW4869 (Fig 8D and E), as well as the calcium content ( Fig 8F). Accordingly, when VSMC were treated with sHAp in the presence of GW4869, calcium deposits were signi cantly reduced as compared with VSMC sHAp .
To con rm that sHAp could be carried out of the cells via Exos after they were internalized into VSMCs, uorexon-labeled sHAp were cultured with VSMCs. Fluorescence signal was detected in the Exos puri ed from the conditional medium of VSMC sHAp . These data suggest that nHAp promoted release of Exos from VSMCs, and these Exos were at least partly originated from the accumulated autophagic vesicles which were resulted from the blockage of autophagy ux by nHAp (Fig 8H).

Discussion
In this study, we demonstrated nHAp existed in calci ed artery, and rst observed internalization of nHAp into VSMC. Nano-HAp stimulated osteogenic differentiation and accelerated mineralization of VSMCs in vitro, and enhanced calcium deposits in artery in vivo. Such effect of nHAp on promoting VC was through altering lysosomal acidi cation and impairing the degradation function of thereafter formed autolysosomes. We rst discovered that the blockage of autophagy ux by nHAp resulted in accumulation of autophagosomes and autolysosomes in VSMCs, which were turned into more calciumcontaining exosomes released into ECM.
Nano-HAp could be both a result and cause of vascular calci cation processes. HAp crystals had been isolated from human carotid atherosclerotic plaques [27] and suggested to contribute to early osteoblastic transdifferentiation of cardiovascular cells [28]. Here we detected nano-crystalline deposits in human calci ed aorta, onset of Runx2 expression, and nHAp existence in the VSMC (Fig 1). And for the rst time demonstrated that synthetic and human-derived HAp crystals were similar in morphology and characteristics. Both can accelerate calcium deposition of VSMCs via blockage of autophagy ux and accumulation of autophagic vacuoles. Studies have shown that nHAp promoted osteogenic differentiation of mesenchymal stem cells (MSCs) and osteoblasts with the increased expression of Runx2, OCN and ALP [29][30][31][32][33]. Such effect of nHAp on mineral deposition has been translated into clinical use of nHAp for bone repair [34]. The effect of HAp on VSMCs had also been studied, and upregulation of BMP-2 and OPN gene expression was found in VSMCs after treated with Pi-induced nanocrystals [35].
Calcifying nanoparticles (CNPs) extracted from human serum was found to induce accumulation of apoptotic bodies and accelerate vascular calci cation [5]. CNPs were detectable in calci c but not healthy human arteries. They could propagate and induce mineral deposition in vitro with su cient Pi [36] , [4]. However, the above studies are lack in clinical evidence, thorough mechanism study and in vivo experiments. Our study con rmed that nano-sized HAp rested in calci ed aorta, attached to the surrounding cells and internalized to the VSMC. Either sHAp or hHAp alone elevated the expression of Runx2 and OPN and the activity of ALP (Fig 2A and FigS3A-B), and expedited calcium deposition in VSMCs (Fig 2B). With additional Pi supplied in the osteogenic medium, HAp-induced calci cation was signi cantly enhanced (Fig 2B), which is supported by the recently published report [6].
To con rm the effect of nHAp on calci cation in vivo, sHAp was implanted underneath the skin or outside of aorta. Implantation of sHAp directly induced calcium deposition (Fig 3A-C). When nHAp was implanted along with VSMCs, more calcium deposition was observed ( Fig 3C) and the implanted VSMCs were differentiated in osteogenic phenotypes with more Runx2 expression (Fig 3D), demonstrating that nHAp can induce VSMCs osteogenic differentiation in soft tissue. When nHAp was applied to the outside of abdominal aorta, signi cant calci cation was observed in the vascular wall after 2 weeks (Fig 3E-G).
According to previous studies, both CaCl 2 and CaPO 4 could induce calci cation in the arterial segments [37]. Here, we employed CaCl 2 as a positive control which indeed showed substantial calci cation in artery ( Fig 3A). This is the rst time to demonstrate that nano-sized HAp induced direct vascular calci cation in vivo.
Autophagy is a key regulator of cellular metabolism and homeostasis. Emerging evidence has demonstrated that autophagy plays an important role in vascular calci cation. It has been documented that atorvastatin protected VSMCs from TGF-β1-stimulated calci cation by inducing autophagy through suppression of the β-catenin pathway [9]. Another study demonstrated that estrogen inhibited osteoblastic differentiation of VSMCs in vitro and prevented arterial calci cation in vivo by promoting autophagy through the ERα signaling pathway [38]. These studies demonstrated a protective role of autophagy in vascular calci cation [10].Recent investigations point to the critical and defensive role of autophagy in the behavioral function of vascular cells against different insults [39], which was mainly focused on the death and phenotypic change of VSMCs without precise mechanism. Here, we found that nHAp was internalized by VSMCs ( Fig 2B) and caused accumulation of autophagosomes (Fig 5A & S4A). It has great clinical signi cance to gure out whether blockage of autophagy ux contributes to nHAp-induced calci cation.
In our study, we found that both sHAp and hHAp crystals increased intracellular autophagosomes (Fig 5A, and Fig S4A) and elevated autophagic cargo proteins, including LC3-II and p62 in the VSMCs (Fig 5C,6A and Fig S4B), and the process is mTOR-independent. This was due to blockage of autophagic ux by nHAp-induced dysfunction of autolysosome degradation. The underlying mechanism was further elucidated that acidi cation of lysosome was inhibited by sHAp ( Fig 7D). However, Wang, et al proved that nHAp modulated osteoblast differentiation by mediating Autophagy via mTOR signaling pathway [40], which is inconsistent with our results. Our data suggest an involvement of an inhibitory effect of nHAp on the late stage of autophagy. This conclusion was supported from 3 lines of evidence. First, nHAp had an additive effect on rapamycin-induced production of LC3-II ( Fig S5B); second, nHAp arrested the degradation of LC3 and p62, the selective autophagy substrates ( Fig 6); third, nHAp prevented the maturation of lysosomes, leading to lysosomal dysfunction (Fig 7). Our results rst demonstrate that nHAp-induced autophagy dysfunction by lysosomal impairment resulted in vascular calci cation. Lysosomal impairment could be a general toxic mechanism of nanoparticles to cells [41,42]. Previous studies testi ed lysosomal disfunction exacerbates vascular calci cation [43,44].This provides evidence to focus on lysosomal function in our study. Mechanism underlining nHAp-induced dysfunction of autolysosome degradation was further studied by analyzing lysosomal acidi cation. We found that nHAp impaired lysosome maturation by inhibiting lysosomal acidi cation without disturbing fusion between autophagosome and lysosome ( Fig 6E). Degradation of internalized EGFR was impaired (Fig 7A & B). Maturation of lysosomal protease CTSD was inhibited ( Fig   8C). Based on the dampened acidity of lysosome with a de ciency in CTSD maturation, we considered V-ATPase be a potential target of damage by nHAp. V-ATPases are membrane-embedded protein complexes that function as ATP hydrolysis-driven proton pumps. V-ATPases are the primary source of organellar acidi cation in all eukaryotes, making them essential for many fundamental cellular processes [45]. In our study, we detected a disruption in lysosomal anchoring of V-ATPase ( Fig 7E).
Although further studies are required, our data rst suggest that nHAp-caused defect of lysosomal acidi cation may involve impairment in V-ATPase transport to lysosomes. This may provide us new idea that cyclic-3′,5′-adenosine monophosphate (cAMP) which restores lysosomal pH may control the progression of vascular calci cation. As nHAp blocks autophagy ux by impairing autolysosome degradation (Fig 6), the accumulated autophagic vacuoles promoted calcium deposits. This was supported by our observation that either autophagy stimulator rapamycin or inhibitor CQ aggravated nHAp-induced calci cation (Fig 7G), and both had an additive effect on the accumulation of LC3 when they were used with nHAp ( Fig 6B and Fig S5B).
Previous study proved that blocking autophagy promotes more rampant calci cation by promoting release of MVs [46], and VSMC vesicle release contributed to calci cation [47]. It has been shown that MVs played a central role in skeletal mineralization as they served as a nidus for initiation of hydroxyapatite crystal formation [48]. MVs were also identi ed in human aortic calci cation, demonstrating that vascular calci cation and bone mineralization occur by similar mechanisms [49]. Proteomic studies showed similarities between exosomes from VSMCs and skeletal osteoblasts, supporting that vascular calci cation recapitulates skeletal mineralization [50].We found that the exosomes released from VSMCs nHAp contained more LC3 and LAMP1 than that from control VSMCs (Fig 8B). It has been reported that more "calcifying" exosomes were secreted from VSMCs after pro-calcifying stimulation, which induced VC procession [46]. Calcium binds with Pi to form hydroxyapatite nodes on the inner and outside of "calcifying" exosomes membranes, which further initializes mineral deposition [46]. Nollet et al. found that autophagosomes from calci ed mouse osteoblasts are packed with calci ed hydroxyapatite [51]. Our study also showed there were HAp crystals in autophagosomes in VSMCs nHAp (Fig 1E). Autophagic proteins were detected in MVs released from valve interstitial cells [20]. These results suggest that exosomes may have a crosstalk with the network of autophagic vesicles either at the stage of their formation or release during the process of VC [7]. Our study showed the accumulation of autophagosomes and autolysosomes, the enhancement of ALP activity and the increased release of exosomes from VSMCs induced by HAp (Fig 8A). Based on these previous studies, we assume that the autophagic vesicles in VSMCs may be secreted out of the cells in the form of exosomes. We found that there were more exosomes released from VSMC HAp as compared with the control (Fig8A, Fig S11A&B), which resulted in more calcium contents released into ECM (Fig 3B, &8C). GW4869 reduced not only the number of exosomes but also the protein content of LC3 and LAMP1, and calcium deposition as well ( Fig   8D-G). All these con rmed our postulation that accumulated autophagosomes and autolysosomes could be converted into calcium-containing exosomes that were released into ECM (Fig 9). This agrees with the report showing that VSMC calci cation is mediated by regulated exosome secretion [24]. However, more studies are needed to further explore the relationship between autophagosomes and exosomes in regards the VC process. Our study unveiled that once nano-crystals were formed, they would facilitate the more rapid progression of calci cation, showing the positive feedback of nHAp in the development of VC. Therapies aim to either prevent initial nHAp formation or nHAp-induced autophagy blockage or exosome releasing may be useful to control VC.

Conclusion
In summary, we report nHAp could directly induce VC in vivo and put new light on the mechanism by which nHAp promotes VC. Nano-HAp internalizes into VSMCs, damages lysosomal V-ATPase and then impairs lysosomal acidi cation, thus causing blockage of autophagy ux and release of Ca 2+ -containing exosomes, eventually resulted in calcium deposition on ECM. Our results suggest that targeting this novel autophagy-lysosome-exosome pathway may help to control vascular calci cation development.