Ectopic calcification and formation of mineralo-organic particles in arteries of diabetic subjects

Vascular calcification occurs in various diseases including atherosclerosis, chronic kidney disease and type 2 diabetes but the mechanism underlying mineral deposition remains incompletely understood. Here we examined lower limb arteries of type 2 diabetes subjects for the presence of ectopic calcification and mineral particles using histology, electron microscopy and spectroscopy analyses. While arteries of healthy controls showed no calcification following von Kossa staining, arteries from 83% of diabetic individuals examined (19/23) revealed microscopic mineral deposits, mainly within the tunica media. Mineralo-organic particles containing calcium phosphate and proteins such as albumin, fetuin-A and apolipoprotein-A1 were detected in calcified arteries. Ectopic calcification and mineralo-organic particles were observed in a majority of diabetic patients and predominantly in arteries showing hyperplasia. While a low number of subjects was examined and information about disease severity and patient characteristics is lacking, these calcifications and mineralo-organic particles may represent signs of tissue dysfunction.

knowledge, these results are the first to indicate that lower limb arteries of individuals with type 2 diabetes harbor mineralo-organic particles similar to the particles described earlier in biological fluids.

Methods
Human tissue samples. The use of human samples was approved by the Institutional Review Board of Chang Gung Memorial Hospital (Linkou, Taiwan). The experiments were performed in accordance with the Declaration of Helsinki and the approved guidelines and regulations. Written informed consent was obtained from the subjects. Artery tissues were obtained from type 2 diabetes mellitus patients (n = 23). Tissues consisted of discarded lower limb arteries (~1 cm in length) removed during surgical amputation. Discarded lower limb arteries of healthy controls (n = 4) who underwent amputation following trauma were also examined. Each artery sample was divided for preparation of paraffin sections and ultrathin sections as described below.
Tissue preparation. Blood vessel specimens were fixed in 10% formaldehyde and mounted on paraffin blocks which were sectioned using a microtome (Cambridge Scientific, HM350) to produce sections with a thickness of 4-5 μm.
Hematoxylin and eosin staining. Histological staining was performed as before 6 . Briefly, for hematoxylin and eosin (H&E) staining, tissue sections were deposited on glass slides and incubated overnight in a laminar flow hood. Specimens were heated at 70 °C for 30 min. Sections were immersed into a fresh solution of xylene three times for 15 min to remove paraffin. Sections were rehydrated with 95%, 80% and 70% ethanol for 5 min each time. Rehydrated specimens were washed with double-distilled water (thereafter referred simply as "water") for 5 min. Cell nuclei were stained with hematoxylin for 8 min, followed by washing in water. Specimens were dipped 10 times in ethanol (95%). Eosin Y was used as counterstain for 1 min. Dehydration was done in 95% and 100% ethanol, followed by treatment with xylene, for 10 min each time. Microscopy observations were done with a BX51 light microscope (Olympus) and a Flex digital camera (Diagnostic Instruments) as done before 6 . von Kossa staining. This procedure was performed as described previously 6 , with minor modifications.
Briefly, deparaffinized and rehydrated blood vessel specimens were incubated with a solution of 5% silver nitrate. Samples were exposed to UV light for approximately 20 min, prior to washing in water. Staining was done with 5% sodium thiosulfate and nuclear fast red (Sigma-Aldrich) for 5 min each time. Dehydrated samples were examined by light microscopy as above. Some microscopy images were combined to form larger images using the PTGui Pro software (New House Internet Services, version 10.0.16).
Scanning electron microscope and energy-dispersive X-ray spectroscopy. Paraffin sections of blood vessels were placed onto copper pieces and dried overnight in a laminar flow hood. Specimens were coated with gold using a sputter coater (E-1010, Hitachi Science Systems). Samples were observed under an S-3000N scanning electron microscope (SEM) (Hitachi Science Systems). Energy-dispersive X-ray spectroscopy (EDX) analysis was performed using an EMAX Energy EX-400 EDX device (Horiba). Samples were irradiated for 30 sec and data acquisition was performed in point mode using the EMAX software (Horiba).
Transmission electron microscopy. Ultrathin sections were prepared as before 6 . Sections were placed onto carbon-coated formvar copper grids. Contrast staining was done with 2% uranyl acetate. Specimens were observed under a JEOL 1230 transmission emission microscope (TEM) operated at 100 kV and equipped with a Gatan Orius SC1000 digital camera (JEOL).

Results
We examined lower limb artery tissues from 23 diabetic subjects for the presence of vascular calcification. For comparison, we also examined lower limb arteries of four healthy controls. Hematoxylin and eosin (H&E) staining of arteries from healthy controls showed normal histological features and no mineral precipitate was observed following von Kossa staining (Table 1 and Supplementary Fig. S1). Arteries from diabetic subjects showed subendothelial hyperplasia or intimal thickening in 75% of cases ( Fig. 1, Table 1, and Supplementary Table S1). Based on von Kossa staining and optical microscopy, we observed that specimens from 19 out of 23 diabetic subjects (83%) contained calcification ( Table 1; representative images of calcified tissues are shown in Fig. 1). Calcification level was classified into four stages ( Table 1): absence of calcification (stage 0); fine mineral granulations (stage 1); mineral deposits with plaques extending in the layers of the media tunica (stage 2); and extensive mineral deposits involving the entire circumference of the artery (stage 3). Representative images for each calcification stage are shown in Fig. 1.
We used SEM to examine the morphology and ultrastructure of the calcified areas in arteries of diabetic individuals (Fig. 2). High magnification observations revealed the presence of mineral granules with sub-micron size in artery specimens corresponding to all stages ( Fig. 2E-H). Low amounts of particles were detected in stages 0 and 1 based on SEM observations (Fig. 2E,H) but these particles did not contain calcium or phosphorus based on EDX analysis (Fig. 2I,J). In contrast, the mineral particles observed in samples from stages 2 and 3 ( Fig. 2G,H) contained calcium and phosphorus (Table 2 and Fig. 2K,L; the EDX spectra in K and L correspond to the particles highlighted in yellow in G and H, respectively). A few NPs were observed in the arteries of healthy controls under SEM but these particles did not contain calcium and phosphorus when analyzed with EDX ( Supplementary  Fig. S1).
TEM observations also revealed the presence of mineral particles within the artery samples of all stages (Fig. 3). Particles in stages 0 and 1 appeared to be delineated by a lipid membrane (Fig. 3E,F, see insets for higher magnification). Particles with a multilayer structure consisting of alternating electron-dense and electron-lucid layers were observed in all stages (Fig. 3, see insets). These multilayer mineral particles are similar to the particles observed earlier in biological fluids 7 and renal tissues 6 .
We used TEM and EDX analysis to examine the vesicles that appeared to be delineated by a lipid membrane (Fig. 4). While the vesicles observed in specimens from stages 0 and 1 showed no calcium or phosphorus (Fig. 4A,B), some mineral deposits found in these samples showed peaks of carbon and phosphorus (Fig. 4C). Vesicles detected in specimens from stages 2 and 3 showed calcium and phosphorus peaks (Fig. 4D-F), consistent with the presence of calcium phosphate within these vesicles.
We compared the morphologies and composition of artery-derived mineral particles with mineralo-organic particles prepared in human serum (HS-Particles) or fetal bovine serum (FBS-Particles), which have been found earlier to mimic the mineral particles detected in body fluids 7,22 . Under TEM, the prepared particles showed a round morphology and a size distribution (~50 nm to 1.5 μm) similar to that of the particles detected in calcified arteries from diabetic individuals (Fig. 5). The round morphology of the mineralo-organic particles was attributed earlier to stabilization of amorphous calcium phosphate phase by organic components such as proteins found in biological fluids 7,8,14 .
In order to confirm the presence of proteins in artery particles, we used immunogold staining coupled with TEM. Our results showed that antibodies that react against HSF, HSA, and apo-A1 bound to artery particles as shown by the presence of gold particles in association with the external or internal layers (Fig. 6). A negative control antibody that reacts against fengycin B (α-FenB), a bacterial protein from Bacillus subtilis, failed to bind to Intimal Calcification ---3 5 Table 1. Calcification level of lower limb arteries from type 2 diabetic subjects. Calcification staging of lower limb arteries from healthy controls (n = 4) and diabetic subjects (n = 23) was based on the detection and abundance of black precipitates in the tunica media following von Kossa staining and observation under optical microscopy as described in the Methods. Calcification level was based on four stages; stage 0: absence of black mineral deposit; stage 1: detection of black granules; stage 2: abundant black mineral plaques with some extending in the layer of the media; stage 3: extensive black mineral deposition including in the tunica media and surrounding the whole artery. Intimal thickening was assessed by observation of H&E-stained tissues. Specimens with intima over 0.2 mm and containing multiple cell layers were classified as positive for intimal thickening.
artery particles (Fig. 6A,F,K). We also used an antibody characterized earlier as binding to various proteins from human serum (α-HS-NPs). As expected, this antibody labeled all artery particles (Fig. 6E,J,O). Similar results were obtained for particles prepared in HS and used as a control for comparison ( Fig. 6P-T). We also performed immunofluorescence staining of artery tissues using antibodies that react against serum proteins. Our results showed that both HSF and HSA were enriched in calcified arteries belonging to stages 2 and 3, whereas minimal fluorescence was noted in specimens from stages 0 and 1 (Fig. 7, arrows in C, D, G, and H show fluorescence staining and calcification based on von Kossa staining). HSF and HSA colocalized and formed small particles in artery tissues examined using confocal fluorescence microscopy ( Fig. 7I-L). While specimens of stage 1 showed minor colocalization of HSA and HSF (Fig. 7I,J), specimens from stage 3 showed more extensive colocalization (Fig. 7K,L). Some particles containing HSF and HSA were also in close proximity with nucleic acids (Fig. 7, see insets).

Discussion
We observed that the majority of artery tissues from diabetic patients contain macroscopic calcification as well as mineralo-organic particles. TEM observations indicated that artery tissues contained lipid membrane vesicles, and the presence of calcium and phosphorus (and thus possibly phosphate) within the particles was associated with a higher propensity for ectopic calcification. A possible interpretation for this observation is that the von Kossa staining of lower limb arteries from diabetic individuals (n = 23) were processed as described in the Methods. Intimal thickening was observed in (B-D), while the intima in A was normal. White rectangles represent the magnified area observed in the images found in the row below. Calcification level was graded into four stages based on the abundance and location of black mineral precipitates; stage 0: absence of black mineral deposit; stage 1: low amount of black precipitates; stage 2: abundant black mineral plaques extending into the layers of the media; stage 3: extensive black mineral deposition, including in the tunica media and surrounding the whole artery. Red arrows in K and L indicate the presence of intimal calcification. Table 1 shows the information about media calcification level, intimal thickening, and intimal calcification noted for the collected artery specimens examined in the present study. TI, tunica intima; TM, tunica media. www.nature.com/scientificreports www.nature.com/scientificreports/ membrane vesicles may represent the nucleating agents that induce the formation of mineral particles in calcified artery tissues. The mineral particles are thus likely to represent precursors of ectopic calcification in this context. Consistent with this possibility, we observed earlier that membrane vesicles derived from serum can nucleate mineralo-organic NPs in biological fluids 20 .
Our findings are consistent with the observations made by other groups. For instance, Price et al. described the formation of fetuin-mineral complexes in the serum of rats treated with the bisphosphonate etidronate 23 or vitamin D 24 . Jahnen-Dechent and colleagues described the formation of calciprotein particles (CPPs) in the ascites of a patient with calcifying peritonitis 25 . Primary CPPs consisted of small clusters of amorphous calcium phosphate and fetuin-A which gradually ripened with time to form more crystalline and larger secondary CPPs 25 . Matsui et al. observed that a precipitate of calciprotein particles formed in the serum of rats treated with adenine to induce kidney failure 26 . Schlieper et al. described mineral particles in iliac arteries of human subjects with chronic kidney disease requiring dialysis 15 . This group identified apoptotic bodies or matrix vesicles as possible nucleating agents for the minerals 15 . Yamada and colleagues identified CPPs in the serum of 10 diabetic patients and observed that their levels increased after meal intake and in subjects with reduced renal functions 27 .
The main limitations of this exploratory study include the low number of subjects and the absence of information regarding disease severity or the gender and age of the patients studied. Given that the specimens were obtained from diabetic subjects who required amputation, the samples probably reflect a late disease stage. While we observed mineralo-organic particles in artery tissues from these diabetic subjects, the clinical significance of vascular deposition and the presence of mineral particles in this context will require further studies. Recent   Table 2. Chemical elements detected using EDX in mineral particles found in lower limb arteries of diabetic patients. Calcification level was assessed as described in the legend of Table 1. Chemical elements in the media tunica of arteries were determined using EDX analysis. "Ca, P" refers to the number of particles in which both calcium and phosphorus were detected, which allowed determination of "Ca/P ratios".  www.nature.com/scientificreports www.nature.com/scientificreports/ studies indicate that vascular calcification is associated with increased mortality risk in patients with chronic kidney disease 28,29 . The presence of mature calciprotein particles was also associated with hypertension or chronic kidney disease 30 . Similarly, artery calcification-especially in the iliac arteries but also in the aorta and carotid arteries-was associated with increased total and cardiovascular mortality in human patients 31 . Medial calcification was also identified as a strong predictor of cardiovascular mortality in type 2 diabetes patients 4 . These  www.nature.com/scientificreports www.nature.com/scientificreports/ results suggest that the presence of mineral NPs in human blood may be used as a marker of calcification status and disease progression. Detection of calciprotein particles in body fluids represent an interesting approach in this context 32,33 .
We observed earlier that various minerals that form in the presence of organic compounds will form an amorphous phase with rounded, cell-like morphologies before transforming into crystalline mineral particles 7,34 . These mineralo-organic NPs activate caspase-1 and NLRP3 inflammasome in macrophages and may therefore induce inflammation 22 . Similarly, the particles activated neutrophils and the release by these cells of