Differential availability/processing of decorin precursor in arterial and venous smooth muscle cells

Abstract

The existence of specific differentiation markers for arterial smooth muscle (SM) cells is still a matter of debate. A clone named MM1 was isolated from a library of monoclonal antibodies to adult porcine aorta, which in vivo binds to arterial but not venous SM cells, except for the pulmonary vein. MM1 immunoreactivity in Western blotting involved bands in the range of M_r 33–226 kDa, in both arterial and venous SM tissues. However, immunoprecipitation experiments revealed that MM1 bound to a 100-kDa polypeptide that was present only in the arterial SM extract. By mass spectrometry analysis of tryptic digests from MM1-positive 130- and 120-kDa polypeptides of aorta SM extract, the antigen recognized by the antibody was identified as a decorin precursor. Using a crude decorin preparation from this tissue MM1 reacted strongly with the 33-kDa polypeptide and this pattern did not change after chondroitinase ABC treatment. In vitro, decorin immunoreactivity was found in secreted grainy material produced by confluent arterial SM cells, although lesser amounts were also seen in venous SM cells. Western blotting of extracts from these cultures showed the presence of the 33-kDa band but not of the high-molecular-weight components, except for the 100-kDa monomer. The 100/33-kDa combination was more abundant in arterial SM cells than in the venous counterpart. In the early phase of neointima formation, induced by endothelial injury of the carotid artery or vein-to-artery transposition, the decorin precursor was not expressed, but it was up-regulated in the SM cells of the media underlying the neointima in both models. Collectively, these data suggest a different processing/utilization of the 100-kDa monomer of proteoglycan decorin in arterial and venous SM cells, which is abolished after vein injury.

Introduction

Arteries and veins display marked structural differences, which account for the unique functional tasks carried out by the two types of vessels in the blood circulation, namely the amount and direction of regional blood flow and pressure (Guyton & Hall, 1996). Arteries and veins are required to be separated by a sharp interconnecting region, established during early development (Conway et al. 2001; Lawson & Weinstein, 2002; Adams, 2003), similar to the spatial boundaries existing between the myocardium and the large vessels proximal to the heart (Ausoni & Sartore, 2001). Recent studies in mouse, Xenopus and zebrafish have pointed out that arterial–venous demarcation involves tissue-specific molecules, such as ephrins and their Eph cognate receptors, that act as guidance cues on mobile cells of arterial and venous endothelial cell lineages (Wang et al. 1998; Helbling et al. 2000; Lawson & Weinstein, 2002). Structural identity of either endothelia occurs independently from the establishment of the proper haemodynamic conditions, under the control of related but distinct gene programs (Adams et al. 1999; Yancopoulos et al. 2000). Formation of arterial–venous endothelial junctions depends upon the functional integrity of receptors, such as type I receptor for the TGF-β superfamily of growth factors (Urness et al. 2000). The venous network development relies on distinct signal transduction pathways, such as the angiopoietin-1 and the orphan receptor TIE1 (Loughna & Sato, 2001). Stabilization of vascular identity seems to be attained late during development, as shown in the expression of the markers neuropilin-1 (receptor for members of the semaphorin/collapsin family) and TIE2 in artery and vein, respectively (Shin et al. 2001).

Moreover, ephrin-B2 may also play a role in conferring a unique identity to the whole arterial vessel. In fact, in an embryo this marker is specifically expressed in the surrounding mesenchymal cells destined to become arterial smooth muscle (SM) cells or pericytes, and this selective distribution is maintained in adulthood (Gale et al. 2001; Shin et al. 2001).

As the endothelium strongly influences the phenotypic profile of the newly incorporated parietal mesenchymal cells driving their differentiation towards SM cells (Hungerford & Little, 1999), one might expect arterial and venous SM cells to be structurally and perhaps functionally distinct. Some studies have shown a difference when arterial and venous SM cells are compared for their in vitro growth response to low-density lipoproteins (Ulrich-Merzenich et al. 2002), serum or BB platelet-derived growth factor (PDGF-BB) (Yang et al. 1998), or antiproliferative drugs (Kim et al. 2004). Other studies have identified that an α-actinin isoform (Ratajska et al. 2001), smoothelin (Van der Loop et al. 1997) and tenascin-C (Wallner et al. 1999) are expressed in the artery but not in the vein (or at least to a lesser extent, or at a later developmental stage). A particular venous-specific myosin heavy chain (MHC), named MHC3, has been found in the porcine inferior vena cava (Okai-Matsuo et al. 1991). As far as the extracellular matrix in the blood vessels is concerned, in vivo structural differences between artery and vein were described for the chondroitin sulfate proteoglycan NG2 (Murfee et al. 2005) and the glycosaminoglycan hyluronan (Hellström et al. 2003), and more recently in vitro for the small proteoglycan decorin (Wong et al. 2005).

cDNA array analysis has indeed demonstrated that a number of genes are differentially expressed in macaque aorta vs. vena cava (Adams et al. 2000). Among these are the regulator of G-protein signalling 5 (RGS5), elastin, the aortic preferentially expressed gene 1 (APEG-1; Hsieh et al. 1996) and the B-type MHC (Adams et al. 2000). More interestingly, Li et al. (1996) found that the promoter for SM22α is selectively activated in the arterial but not in the venous system in transgene mice, thus indicating that distinct regulatory mechanisms control the expression of this contractile protein in arterial vs. venous SM cells.

In an effort to identify specific markers for arterial and venous SM cells, to be used in monitoring the phenotypic changes that the venous SM cells undergo when vein segments are grafted in the arterial position, we screened a library of hybridomas produced by immunizing mice with porcine aorta SM tissue. The MM1 antibody was found to be arterial specific when tested on vascular tissues and is able to react with an antigen whose expression is developmentally regulated and activated in the venous medial SM cells when a vein segment is transposed in the arterial position.

Materials and methods

Collection of tissue specimens and preparation of tissue and cell extracts

Female adult (weighing ∼120 kg), fetal (days 30 and 90) and newborn farm pigs were studied. Vascular and non-vascular tissue specimens were examined (see Table 1).

Table 1.

Distribution of AgMM1 in vascular and non-vascular tissues as determined by immunocytochemistry

Type of tissue	Immunoreactivity
Vascular SM tissues
Thoracic aorta	++
Abdominal aorta	++
Coronary arteries	+++
Pulmonary a.
extrapulmonary branch	+++
intrapulmonary branch	+
Carotid a.	+++
Renal a.	+++
Femoral a.	+++
Arterial vasa vasorum	+++
Umbilical a.	+++
Inferior vs. cava	−
Coronary v.	−
Pulmonary v.
extrapulmonary branch	+
intrapulmonary branch	+
Jugular v.	−
Renal v.	−
Femoral v.	−
Umbilical v.	−
Venous vasa vasorum	−
Non-vascular SM tissues
Lung	−
Trachea	+ (cartilage)
Oesophagus	−
Stomach	−
Small and large intestine	−
Bladder	−
Lymphatic vessels	−
Kidney (Bowman's capsule)	++
Other tissues
Brain	−
Spinal cord	−
Lymph node	+ (interstitial tissue)
Spleen	+ (interstitial tissue)
Liver	−
Mammary gland	−
(myoepithelial cells)	+
Ovary gland	+ (interstitial tissue)
Tendons	−
Cartilage (hyaline and elastic)	+++
Blood cells	−

⁺⁺⁺the majority

⁺⁺the minority

⁺rare

⁻no reactive cells with MM1 antibody.

Crude extracts were obtained from tissues and cells. Minced tissues were powdered in a mortar in the presence of liquid nitrogen. After sonication in phosphate-buffered saline (PBS; three times, 5 s each round, in an ice bath, at a 1–7 w/v ratio) in the presence (thoracic aorta, inferior vena cava for dot-blot immunoassay) or absence (thoracic aorta for immunization of mice) of the Protease Inhibitors Cocktail (Sigma, Milan, Italy), the slurry was spun down at 9500 g at 4 °C in a Eppendorf microfuge. Vascular SM cells (coronary artery and jugular vein) were obtained from primary or secondary cultures at confluency. Cultures were, in sequence, rinsed in cold PBS containing the protease inhibitors, removed by a rubber policeman and finally centrifuged. Pellets of tissues and cells were treated with Laemmli's Sample Solution containing type I DNAase (30 µg mL⁻¹), boiled for 3 min and spun down. The same treatment was also applied to some tissues (neck lymph node, cartilage from external ear, coronary artery, thoracic aorta and inferior vena cava) used as crude extracts in Western blotting experiments. The total protein content of each sample was determined by the Bradford dye-binding assay (Sigma).

Tissue cultures

Primary and secondary cultures from coronary artery and jugular vein from newborns were obtained from tissue explants. Small fragments (about 1–2 mm³) of the vessel walls, devoid of adventitia and endothelium, were rinsed in cold Hanks's balanced salt solution and cultured on gelatin-coated (10 µg mL⁻¹) Petri dishes in the presence of Dulbecco's modified Eagle's medium (DMEM; Sigma) with 10% fetal calf serum (FCS; Sigma). After 2 weeks, cells migrated from the explants were detached with a trypsin solution (0.25% in Hank's solution), seeded on new coated Petri dishes (10⁴ cells cm⁻² as initial seeding per each preparation) and grown until confluency. Some confluent cultures were shifted to 2% FCS for 3 days. Ten and 2% FCS confluent cultures were fixed in acetone or 2% p-formaldehyde in PBS, pH 7.4, for subsequent immunofluorescence processing. In the latter case, the fixed cultures were permeabilized with 0.1% Triton X-100 in PBS for a few seconds before applying the primary antibody.

Production of hybridomas

The supernatant of sonicated aorta extract was used as an immunogen. Eight- to 12-week-old female BALB/c mice were injected i.p. three times with the PBS-soluble extract mixed with Freund's adjuvant, with an interval of 14 days between immunizations (Borrione et al. 1989). Blood samples from immunized mice were collected 7 days after the third immunization and the sera tested by dot-blot assay and indirect immunoperoxidase labelling using arterial and venous extracts. Three days after the i.v. boost, spleen cells from a selected mouse were fused with NS-0 cells. Arterial vs. venous screening of hybridoma clones was performed using indirect immunoperoxidase labelling and dot-blot immunoassay with porcine aorta/inferior vena cava cryosections or tissue extracts, respectively. The MM1 hybridoma secreting an arterial-specific immunoglobulin was cloned by limiting dilution. Immunoglobulin isotype was determined using subclass-specific goat anti-mouse peroxidase-conjugated antibody (Amersham Biosciences, Milan, Italy).

Immunocytochemistry

Freshly obtained cryosections of 8 µm from porcine vascular and nonvascular tissues (see Table 1) were reacted with MM1 in comparison with a panel of primary antibodies according to the procedure described in Chiavegato et al. (1999). The following antibodies were used: unconjugated and FITC-conjugated anti-SM α-actin (Sigma), SM-E7 anti-SM myosin (Borrione et al. 1989), 1B8 anti-SM22 (Chiavegato et al. 1999), NM-F6 anti-platelet myosin heavy chain (MyHC)-A_pla1 (Sartore et al. 1999) and IST-9 anti-EIIIA fibronectin (Abcam, Cambridge, UK). Bound primary antibodies were revealed by a swine IgG anti-mouse IgG (Dako, Dakopatts, Denmark) coupled with horseradish peroxidase (for immunoperoxidase experiments). Bound IgG was detected by incubation in amino-ethyl-carbazole solution and counterstaining with Harris haematoxylin. Non-immune IgG instead of the primary antibody and incubation of the secondary antibody alone were used as controls. For indirect immunofluorescence experiments bound primary antibody was detected by the secondary antibody rabbit IgG to mouse IgG conjugated with tetramethylrhodamine isothiocyanate (Dako). In double immunofluorescent experiments, MM1 binding was shown by rabbit IgG to mouse IgG conjugated with tetramethylrhodamine isothiocyanate, whereas the other primary antibody (anti-SM α-actin) was directly labelled with fluorescein isothiocynate. Controls were performed as reported in Chiavegato et al. (1999). Some tissue cryosections were treated with chondroitinase ABC (Sigma; 100 mU/5 µg decorin) for 30 min at 37 °C in 50 mm Tris-HCl buffer containing 60 mm sodium acetate, pH 8.0, prior to MM1 incubation to assess whether glycosaminoglycan removal might interfere with the MM1 binding pattern (Sorrell et al. 1999).

Processed cryosections were observed with a Leica DMR photomicroscope (Wetzlar, Germany) or a Zeiss Axioplan (Oberkochen, Germany) fluorescence microscope. Digitized images collected by a Leica DC 300 camera were elaborated and mounted using Adobe Photoshop 7.0 software.

Confocal microscope examination of cultured SM cells reacted with MM1 was performed using a Bio-Rad MRC 1024 laser scanning microscope (Bio-Rad, Milan, Italy) equipped with a Kr/Ar laser. To reduce bleed-through, double-labelled confocal images (both xy and xz sections) were acquired sequentially. Noise reduction was achieved by Kalman filtering during acquisition.

Dot-blot immunoassay

PBS-sonicated tissue extracts from aorta and inferior vena cava prepared as described above were spotted on Protran BA nitrocellulose paper (Schleicher & Schuell, Brentford, UK) using the Bio-Rad dot-blot apparatus. A volume of 3 µL per well was used for each extract and the amount of protein was in the range 0.25–15 µg per well. The filter was dried at room temperature, saturated in 5% defatted milk in TBS (20 mm Tris-HCl, pH 7.4, 137 mm NaCl) and then incubated with the hybridoma supernatant diluted in TBS plus 3% bovine serum albumin and 0.1% Triton X-100 for 2 h at 25 °C. After three rinses with TBS plus 0.1% Triton X-100 the paper was reacted with swine IgG to mouse IgG₁ coupled with horseradish peroxidase (The Binding Site, Birmingham, UK). Bound secondary antibody was revealed by the chemiluminescence SuperSignal kit (Pierce, Celbio, Milan, Italy) in this case, following the blocking procedure suggested by the manufacturer. Monitoring was as described for the immunofluorescence.

Electrophoresis and Western blotting

One-dimensional SDS-PAGE was performed as previously described (Chiavegato et al. 1999), using 5% and 7.5% SDS-slab gels and vascular and non-vascular SDS extracts or a decorin crude extract from thoracic aorta obtained according to the method of Wheatley et al. (2004). After electrophoretic separation, the proteins were stained with Coomassie blue or subjected to immunoblotting. The separated polypeptides were transferred to a nitrocellulose paper (Schleicher & Schuell) for 16 h at 250 mA in the presence or absence of 10% methanol. The following steps were the same for the dot-blot assay. Controls were the same as described for the immunofluorescence. Blotted decorin was enzymatically deglycosylated with chondroitinase ABC (Sigma; 100 mU/5 µg decorin) for 2 h at 37 °C in 50 mm Tris-HCl buffer containing 60 mm sodium acetate, pH 8.0 (Johnstone et al. 1993). The paper sheets were then rinsed with 5% bovine serum albumin in 10 mm Tris-HCl, pH 7.4, and containing 150 mm NaCl and 0.05% Tween-20 before performing incubation with the primary antibody.

Protein degradation

Aorta SDS extracts were kept at 25 °C in the presence of a cocktail of protease inhibitors for times ranging from 5 to 30 min before electrophoresis. The resulting pattern was studied with MM1 in comparison with the F8 anti-COMP antibody (a gift of Dr Jack Lowler), which also labelled the apparent M_r 130-kDa band.

Immunoprecipitation

Frozen fragments of porcine thoracic aorta and inferior vena cava, devoid of endothelium and adventitia, were powdered in a mortar in the presence of liquid nitrogen. Cold RIPA buffer (PBS, 1% Nonidet P40, 0.1% SDS, 0.5% sodium deoxycholate, pH 7.2) containing the Sigma anti-proteases cocktail was then added to the powdered tissue (10 µL/2 mg). The mixture was collected and sonicated three times (2 s each) in a ice bath and kept in this condition for 30 min. The slurry was then centrifuged in an Eppendorf minifuge at 9500 g for 10 min. Mouse non-immune IgG (0.25 mg; Sigma) and Protein A/G PLUS-Agarose (20 µL; Santa Cruz) were added to the supernatant (1 mL) and incubated for 90 min in a rocker platform at 4 °C. The pellet obtained after centrifugation (1000g for 4 min) was discarded whereas to the supernatant was added MM1 (1 : 200 v/v). The antigen–antibody incubation was performed at 4 °C for 3 h in a rotating device. Protein A/G PLUS-Agarose was then added and the mixture incubated at 4 °C overnight. The agarose beads were first spun down at 600 g, then the pellet was rinsed three times with the RIPA buffer and finally boiled in Laemmli's Sample Solution for 3 min.

Antigen identification

Identification of polypeptide antigens recognized by MM1 in Western blots was performed by mass spectrometry. Gel plugs of electrophoresed thoracic aorta extract containing the proteins of interest were excised by hand and sent for LC/MS/MS analyses (Eastern Quebec Proteomics Centre, Centre Hospitalier de l'Université Laval, Quebec, Canada). Gel plugs were placed in 96-well plates and then washed with water. Tryptic digestions were performed on a MassPrep liquid handling robot (Waters Ltd, Mississauga, Ontario, Canada) according to the manufacturer's specifications and using sequencing-grade modified trypsin (Promega, Fisher Scientific, Ontario, Canada). After extraction from the gel into 50% acetonitrile/water, peptides were lyophilized in a speed vacuum and resuspended in 10 µL of 0.1% formic acid solution. Peptide tandem mass spectra (MS/MS) were obtained by capillary liquid chromatography coupled to an LCQ DecaXP (ThermoFinnigan, San Jose, CA, USA) quadrupole ion trap mass spectrometer with a nanospray interface. Each sample was loaded onto a reversed-phase column (PicoFrit 15-µm tip, BioBasic C18, 10 cm × 75 µm, New Objective, Woburn, MA, USA). Peptides were eluted from the column with a linear gradient of water–acetonitrile in 0.1% formic acid at a flow rate of approximately 200 nL min⁻¹. Mass spectra were acquired using a data-dependent acquisition mode in which each full scan mass spectrum was followed by collision-induced dissociation of the three most intense ions. The dynamic exclusion function was enabled, and the relative collisional fragmentation energy was set to 35%. Resulting peptide MS/MS spectra were interpreted using MASCOT software (Matrix Science Inc., Boston, MA, USA) and searched against mammals in the NCBI non-redundant protein database. Carbamidomethylation of cysteine and partial oxidation of methionine, two missed cleavages, and an error tolerance of 2.0 Da for peptides and 0.5 Da for fragments were considered in the search. Each peptide identification was confirmed by manual inspection of the spectrum.

Experimental models

The experimental protocol and the surgical procedures used to set up the endothelial injury, vein-to-artery transposition, were approved by the Italian Ministry of Public Health and the Institutional Animal Care and Use Committee of the University of Padua. All animals received human care according to guidelines issued by the National Institutes of Health and contained in the Guide for the Care and Use of Laboratory Animals. Animals were anaesthetized by an i.m. injection of ketamine (20 mg kg⁻¹) and xylazine (4 mg kg⁻¹). After endotracheal intubation, pigs were ventilated with halothan (0.75%) and oxygen to maintain anaesthesia. Before surgery the animals received an i.v. bolus of heparin (150 units kg⁻¹). All experimental models were set up using 30-kg female pigs from a local supplier. All pigs were subjected to an antibiotic prophylaxis and analgesic treatment after surgery. Endothelial lesion was produced at the level of the common carotid artery using a 4F Fogarty catheter (Groves et al. 1995). Balloon positioning was carried out under fluoroscopic control. After inflation with 2 mL of saline the catheter was pulled back and forth three times with constant rotation. The balloon catheter was removed and the wound closed. Pigs were killed after 7 (n = 2) and 14 (n = 2) days from surgery when intimal thickening was already detectable (Groves et al. 1995). The animals were killed with an i.v. injection of Tanax.

Vein-to-artery transposition was performed using the internal jugular vein grafted at the level of the common carotid artery. About 6 cm of vein was harvested, flushed and gently distended using a saline solution containing heparin (5 units mL⁻¹) and papaverine hydrochloride (0.7 mg mL⁻¹). One centimetre of vein was kept to control the ungrafted venous wall. About 3 cm of vein was anastomosed by a continuous suture with the arterial ends. The incision was closed and the animals were allowed to recover. Pigs were killed after 7 (n = 2) and 14 (n = 2) days from surgery using an i.v. injection of Tanax.

Results

Binding properties of MM1 antibody

In total, 1100 hybridomas produced by immunizing mice with a porcine aorta extract were screened by dot-blot and immunocytochemistry assays using non-denatured arterial and venous extracts or cryosections, respectively. Of these, 247 were found to be equally cross-reacting with either artery or vein, and only one, named MM1, was specific for an arterial antigen (Fig. 1). IgG₁ secreted by this hybridoma heterogeneously stained the SM cells of the aorta but were negative with the corresponding venous SM cells (Fig. 1). Pre-absorption of MM1 antibody with aorta extract before applying to arterial cryosections (in immunocytochemistry) or spotted extracts (in dot-blot immunoassay) eliminated the immunostaining. In addition, the use of several fixative solutions (acetone, ethanol or 1–4% p-formaldehyde in PBS, pH 7.4, for 10 min at room temperature) prior to MM1 in immunocytochemical experiments did not change the pattern of immunoreactivity, suggesting that the observed antigenic distribution was not the result of artefacts. The extensive staining performed with other vascular tissues consistently showed that this antibody was able to label all the arterial SM tissues examined (see Table 1). Conversely, MM1 was not reactive with all venous tissues, the only exception being with the pulmonary tree (supplementary Fig. S1). Confocal microscopy analysis (Fig. 2) revealed that the antigen recognized by MM1, provisionally named AgMM1, is a secreted molecule particularly abundant in the extracellular matrix (ECM) that surrounded arterial SMC and the subendothelial region of this type of vessel. MM1 antibody did not cross-react with other mammalian species (human, rat, rabbit) in immunocytochemistry and Western blotting experiments (data not shown).

Fig 1.

Screening of MM1 hybridoma. (A) Dot-blot analysis of non-denatured arterial and vein extract with MM1 hybridoma supernatant or 1B8 anti-SM22 antibody (used as control of SM cells). Lanes 1, 1′: aorta extract; lanes 2, 2′: inferior vena cava extract. (B) Immunoperoxidase staining of cryosections from adult porcine thoracic aorta (a,c) and inferior vena cava (b,d) with MM1 hybridoma or 1B8 antibody. Note that AgMM1 is heterogeneously distributed in the medial SM cells whereas SM22 and smooth muscle tissue (asterisk) in the vein is negative. Scale bar, 90 µm.

Fig 2.

Confocal microscopy analysis of AgMM1 distribution (in red) in comparison with SM α-actin (in green). Small (a,c,e) and large (b,d,f) branches of the left descending coronary artery are shown. AgMM1 appeared to be present in the extracellular matrix, particularly abundant in the subendothelial space in the small branches of coronary arteries (e) and in the space among SM bundles in the large branches of this vessel (f). The thin white line in (e) identifies the endothelium (ec). sm, Smooth muscle. Scale bar, 50 µm.

Electrophoresis in denaturing conditions and Western blotting were used to characterize AgMM1. Surprisingly, MM1 reacted equally in Western blotting with SDS extracts from both the artery and the vein, i.e. giving six immunoreactive bands of apparent M_r of 226, 220, 217, 130, 120 and 33 kDa (Fig. 3A). The only exception was the 100-kDa band, which was present in the artery exclusively and was the most reactive band. This pattern did not change if arterial and vein extracts were performed in the presence of guanidinium chloride, or varying the percentage of SDS and 2-mercaptoethanol in the extraction solution or the time of extraction (data not shown).

Fig 3.

Western blotting analysis of MM1 binding in artery and vein extracts (A), MM1-immunoprecipitated arterial and venous antigens (B) and crude decorin from thoracic aorta (C) examined on 7.5% SDS gels. (A) Extracts from artery in lanes 1,1′ and from veins in lanes 2,2′. The electrophoretic pattern is shown in lanes 1,2 and the Western blotting profile in lanes 1′,2′. (B) Immunoprecipitate of antigens reacted with MM1 from arterial (lane 1) and venous (lane 2) extracts. H and L, heavy and light chains of IgGs. Arrows indicate standard molecular weights. Equal amounts of proteins (5 µg) were loaded in the gels. Note the 100-kDa band reacting with MM1 in arterial but not venous extract. (C) Electrophoretic profile of crude decorin (lane 1) and the corresponding Western blotting pattern (lane 2). Note the marked reaction of MM1 with the 33-kDa band.

The immunoprecipitation pattern obtained with arterial and venous SM extracts obtained in non-denaturing conditions treated with MM1 (see Fig. 3B) confirmed that the major difference between the two tissues can be imputed to the 100-kDa component that is selectively identified in the arterial extract.

To rule out the possibility that some of the bands recognized by MM1 were the result of a degradation process, despite the presence of the anti-proteolytic cocktail in the preparation of the muscle extracts, we performed an autolytic experiment on the aorta extract (supplementary Fig. S2). At least for the 130–120-kDa polypeptides, the MM1 binding pattern cannot be ascribed to incomplete activity of anti-protease cocktail used in the preparation of tissue extracts.

We approached the problem of AgMM1 identification by cutting out the bands corresponding to polypeptides of 130 and 120 kDa in the Coomassie-stained gel and performing a mass spectrometry study of the respective tryptic digests (Table 2). LC/MS/MS results (see Table 2) suggest that the antigen recognized by MM1 is a decorin precursor, i.e. an ECM molecule that undergoes numerous post-translational modifications to attain its final form.

Table 2.

Result of polypeptide identity search as determined by mass spectrometry analysis

Mr (kDa)	# Access	pI/nominal mass	Source	Probability based Mowse score	Protein name
130	Q9XSD9	8.85/40216	Pig	84	Decorin precursor
120	Q9XSD9	8.85/40216	Pig	94	Decorin precursor

To confirm such data, MM1 was reacted in Western blotting with a crude decorin preparation from porcine aorta, and the corresponding binding profile is shown in Fig. 3(C). Here, the major reactive band is the 33-kDa component whereas the other bands are weaker (130–120 kDa) or not reactive (226–217 kDa; compare blots in Fig. 3A, lane 1′, and Fig. 3C, lane 1′). Pretreatment of blotted decorin with chondroitinase ABC prior to incubation with MM1 induced an even greater reactivity of 33-kDa band without any effect on the other bands (data not shown).

In SM cells grown in vitro in high-concentration serum and examined by confocal microscopy, bound MM1 was visible only after 5 days in confluent cultures showing a grainy or cluster-shaped pattern (Fig. 4A) in both cell types, although in arterial SM cells AgMM1 was generally more abundant than in venous SM cells (data not shown). Interestingly, when SM cells were cultivated in low-concentration serum MM1 staining disappeared (data not shown). Apparently, there was no obvious spatial relationship between the intracellular distribution of AgMM1 with the microfilament (Fig. 4A), microtubular or intermediate (data not shown) filament system. Cultured SM cells were also examined for AgMM1 expression in Western blotting (Fig. 4B). A major band of M_r 33 kDa was seen in the arterial and, to a lesser extent, in venous SM cell extract. No immunostained band in the range of M_r 226–217 or 130–120 kDa was detectable in the extracts, although a 100-kDa band was frequently seen in both arterial and venous cell extracts.

Fig 4.

Confocal microscopy images of double immunofluorescence assay on cultured SM cells from coronary arteries (A) and Western blotting analysis of extracts from arterial and venous SM cells grown in vitro (B). (A) MM1 in red and anti-SM α-actin in green. Arrowheads indicate clusters of different size, apparently not related to the microfilament system. Asterisks indicate MM1-positive secreted material. Scale bar, 25 µm. (B) Western blotting analysis of MM1 binding to arterial (lanes 1,1′) and venous (lanes 2,2′) SM cell antigens examined in 5% (upper panel) and 7.5% (lower panel) SDS gels. Lanes 1,1′ are representative of the electrophoretic pattern and lanes 2,2′ show the corresponding Western blotting profile. Note that the 33- and 100-kDa bands are reactive with MM1 in both extracts. However, in venous SM cells the two bands stain less intensely even if equal amounts of proteins (3 µg) were loaded in the gels. High-molecular-weight MM1-reactive components are lacking in culture.

AgMM1 was not expressed solely in the SM cells of vascular tissues but in a few non-muscle tissues as well (Table 1; supplementary Figs S3 and S4). In lymphoid tissue, such as the lymph node, AgMM1 was localized to the interstitial tissue that characterizes the germinal centre (Fig. S3A). Elastic (Fig. S3C) and hyaline (Fig. S4a) cartilage contained a large amount of this antigen. AgMM1 was also found in close apposition to the Bowman's capsule of the renal corpuscle (Fig. S4b). Western blotting of the lymph node extract with MM1 shown in Fig. S3(B) shared some features with the corresponding pattern obtained with the arterial tissue, such as the 130-, 120-, 100- and 33-kDa bands. The 226–217-kDa antigens expressed in the vascular tissue extract, however, were not detectable here. Interestingly, the binding pattern of MM1 in Western blotting found with the elastic cartilage (Fig. S3D) is distinct from the lymphoid (Fig. 4B) tissue showing reactivity for the high-molecular-weight components but not for the 100- and 33-kDa components.

AgMM1 expression during development and in experimental models

We investigated the time course of AgMM1 during development to identify when the arteriovenous difference appears. The earliest evidence for this pattern in the coronary vessels could be demonstrated from about day 30 of pregnancy (Fig. 5a), when in elastic vessels such as the aorta it was barely discernible (Fig. 5b). However, even among elastic vessels, there were differences in the time when AgMM1 appeared: in a 90-day-old fetus, pulmonary artery was still negative for this antigen whereas it was already present in aorta (data not shown). Umbilical veins as with the other fetal veins were not reactive with MM1 (supplementary Fig. S5). Spatial distribution of AgMM1 during arterial maturation was also interesting inasmuch as in the aorta this antigen displayed a ‘centrifugal propagation’ (Fig. 5d) from the first layers of differentiating SM cells towards the forming adventitia. This pattern of expression was unrelated to those of the markers known to identify the SM cell lineage and differentiation profile, such as SM α-actin/SM22 and SM myosin. In fact, even in the AgMM1-negative cells SM α-actin, SM22 and SM myosin were already present (data not shown). Noticeably, pretreatment of arterial and venous cryosections with chondroitinase ABC prior to incubation with MM1 did not change the distribution pattern of MM1 immunoreactivity but the intensity of immunostaining was markedly increased in the arterial SM cells (data not shown).

Fig 5.

Distribution of AgMM1 in developing coronary artery (ca) and vein (cv) and aorta (ao). Note that differences in decorin expression in the coronary vessels are evident at embryonic day 30 (E30, panel a) whereas at the same developmental time only very rare foci of MM1-positive cells are seen in the aortic media (panel b, arrow). At E60, AgMM1 is particularly expressed in the region surrounding the aortic endothelium (arrowhead in d) while in the coronary artery, besides SM cells, traces of decorin labelling can also be observed in the interstitial tissue at the interface with the myocardial cells (double arrowheads). Scale bar, 100 µm.

Because AgMM1 expression seems to be related to differentiating SM cells, we tested whether under naturally occurring or experimentally induced conditions where SM cells are known to be partially differentiated (or de-differentiated/phenotypically modulated, see Sartore et al. 1999) the antigen recognized by MM1 is expressed in a peculiar manner. To this end, we tested the ‘intimal cushions’ and the neointima formed in response to the endothelial cell injury of the arterial wall. ‘Intimal cushions’ are characterized by the presence of partially differentiated (or de-differentiated) SM cells, i.e. cells showing immunoreactivity for SM myosin (Fig. 6b) but also for platelet myosin (Fig. 6c) and fetal fibronectin (Fig. 6d) (see Sartore et al. 1999). The ‘immature’ SM cells of ‘intimal cushions’ did not express AgMM1 (Fig. 6a). In intact carotid artery wall, decorin is heterogeneously distributed (Fig. 6e) with respect to medial SM myosin-positive SM cells (Fig. 6f), but in the endothelial injured arterial wall neointimal decorin-positive SM cells appeared mainly in few cells near the internal elastic lamina (Fig. 6g, inset). Interestingly, AgMM1-positive SM cells were markedly more numerous in the innermost region of the media (Fig. 6g).

Fig 6.

AgMM1 distribution in the intimal cushions of abdominal aorta (a–d) and in intact (e,f) and endothelial-injured (g,h) carotid artery in comparison with SM cell markers. Note that the decorin is almost absent in the intimal cushions (ic; panel a) whereas it is heterogeneously expressed in the underlying media (m). SM-E7, anti-SM myosin (b); NM-F6, anti-MyHC-A_pla1 (c); IST-9, anti-EIIIA fibronectin (d) Reactivities for platelet MyHC-A_pla1, EIIIA fibronectin and SM myosin indicate that intimal cushion tissues contain SM cells of ‘immature’ type (Sartore et al. 1999). In (e), decorin is scattered in the media whereas anti-SM myosin homogeneously stained this wall layer (f). After 14 days from balloon injury, an intimal thickening (it) is formed in the carotid artery, which is positive for SM myosin but almost negative for MM1 (panels g,h). Some immunoreactivity in the SM cells near the internal elastic lamina (iel; large arrowhead in the inset, panel g) can, however, be seen. Almost all the cells of the innermost region of the media now express AgMM1 (asterisk) whereas the outermost layers of this vessel are now AgMM1-negative: av, arterial vasa vasorum. Scale bars: (a)–(d), 70 µm; (e)–(h), 100 µm; insets, 40 µm.

We also investigated whether AgMM1, which was not detectable in the veins in vivo, could be ‘rescued’ when the vein is grafted in the arterial position and can form a neointima. When the jugular vein, whose SM cells were positive for SM myosin (Fig. 7b) but negative for MM1 (Fig. 7a), was grafted in the carotid position in autotransplantation experiments, venous SM cells gave with MM1 the pattern shown in Fig. 7(c). SM myosin-positive SM cells (Fig. 7d) accumulated in the intimal thickening following transplantation were almost completely devoid of decorin immunoreactivity (Fig. 7c) whereas medial SM cells underlying the venous neointima were stained for MM1. On the whole, these data suggest that, first, the arterial environment can modulate the expression of AgMM1 in venous medial SM cells, and, secondly, ‘immature’ SM cells in the arterial or venous neointima are very similar with regard to AgMM1 expression, i.e. weak or absent.

Fig 7.

AgMM1 distribution in intact (a,b) and arterially grafted (c,d) internal jugular vein in comparison with SM-E7 anti-SM myosin antibody (b,d). AgMM1 (a,c) is absent from the SM cell layer (sm) in intact vessel (a) but it is expressed in the media of venous SM segment transplanted in the arterial position (c). In the intimal thickening (it), which is formed after vein-to-artery graft and contains SM myosin-positive SM cells, AgMM1 immunostaining is barely detectable (d). a, adventitia; vv, vasa vasorum; t, thrombus. Scale bar, 110 µm.

Discussion

The aim of this study was to test the hypothesis that endothelial-based arteriovenous demarcation, which relies on tissue-specific genetic and haemodynamic factors (Adams, 2003), also involves the establishment of structural differences in the underlying SM tissues (Gale et al. 2001). We attempted to identify arterial-specific proteins via expression profiling of vascular SM antigens using a crude porcine arterial extract and monoclonal antibody technology. Contrary to expectation, the results of the hybridoma screening indicate that only one clone (MM1) was able to identify an arterial-specific antigen expressed in the porcine vascular SM cells. The antigen recognized by this antibody is heterogeneously expressed in both small and large muscular and elastic arteries but not in veins, except for the extrapulmonary tract of the pulmonary vein. Interestingly, AgMM1 is expressed in the arterial system from the late fetal phase of development, though at different times between muscular vs. elastic arteries and within the elastic arteries themselves. In the developing elastic arteries the spatial propagation of AgMM1 antigenicity is of the ‘centrifugal type’, closely resembling that of elastin, a component of extracellular matrix, in maturating vessel (Gale et al. 2001). It is also present in a few non-vascular tissues such as the hyaline and elastic cartilage, lymphatic tissues and in the kidney at the level of cells of Bowman's capsule.

In Western blots, the binding pattern of MM1 in crude extracts of artery and vein is peculiar. In fact, it gives a multiband profile of 226, 220, 217, 130, 120 and 33 kDa. Noticeably, the 100-kDa band is expressed in the artery and not in the vein, as also confirmed by the selective immunoprecipitation of this polypeptide with MM1 in arterial extract. Interestingly, when the venous SM cells are grown in vitro, AgMM1 expression appears (Fig. 4) and this event correlates with the appearance of the 100-kDa band and, to a lesser extent, of the 33-kDa band.

Based on the immunocytochemical (Figs 2 and 4) and Western blotting (Fig. 3) data, one could hypothesize that AgMM1 is a secreted protein whose biosynthetic pathway is that of a multimeric protein complex or the result of a tissue-specific processing of a large preprotein. In both cases the final destination of the functional molecule seems to be the ECM. Indeed, LC/MS/MS mass spectrometry analysis suggests that AgMM1 is actually the precursor of the ECM proteoglycan decorin. This is also confirmed by the marked immunoreactivity of MM1 with the 33-kDa component (and to a lesser extent with the 130–100-kDa bands) present in a crude decorin preparation of aorta SM (Fig. 3C). Decorin, a chondroitin/dermatan sulfate proteoglycan, is synthesized in chondrocytes and nonchondrocyte cells, such as SM cells, via a complicated sequence of events involving numerous post-translational steps of protein and carbohydrate processing: addition of Asn-linked high mannose oligosaccharides (1–3 chains) and O-linked oligosaccharides to the core protein, cleavage of the precursor peptide, addition and elongation of glycosaminoglycan (single chain), secretion and subsequent organization to form dimers and oligomers (Järveläinen et al. 1991; Sawhney et al. 1991; Scott et al. 2003). Importantly, the biosynthetic process of decorin as determined by electrophoretic and Western blotting analysis is in concordance with the pattern of MM1 immunoreactivity shown in Fig. 3. The 33-kDa band corresponds to the core protein whereas the 100-kDa band identifies the monomer (Järveläinen et al. 1991; Sawhney et al. 1991; Scott et al. 2003). The 120–130-kDa bands might represent further processing of the monomer, and the higher-molecular-weight components (217–226 kDa) could be attributable to dimers of decorin or cross-reactivity with biglycan (another proteoglycan with which decorin shares a marked homology; Day et al. 1987; Järveläinen et al. 1991; Sawhney et al. 1991; Scott et al. 2003, 2004). In non-vascular tissue, such as lymph node (supplementary Fig. S3), the decorin monomer is also expressed but the high-molecular-weight components are lacking, suggesting that oligomers/aggregates of decorin are assembled in a tissue-specific manner. This property of decorin is further supported by data obtained with the elastic cartilage extract (supplementary Fig. S3), in which the 33- and 100-kDa bands are undetectable but the 226–120-kDa bands are clearly visible.

The antigenic epitope of MM1 is likely to be localized to the 33-kDa core protein as suggested by the deglycosylation experiments performed on cryosections of arterial/venous SM tissues or blotted decorin from aorta SM tissue. Here, immunocytochemical and immunochemical patterns of MM1 binding do not change except for the intensity of immunostaining, which is even greater with chondroitinase ABC-treated samples than with untreated samples. Available monoclonal antibodies to decorin belong to two classes: those specific for the proteoglycan (Sawada et al. 2002) and those directed against the core protein (Scott et al. 1993). The peculiar MM1 binding pattern could be due to the modality used for preparing the antigen to immunize the animals, i.e. the ultrasonic treatment. It is well known that large oligomeric molecules in aqueous solution can undergo an ultrasonic depolymerization and, hence, exposure of previously hidden antigenic sequences (Grönroos et al. 2004). The fact that the 33-kDa core protein cannot be identified in Western blots of cartilage extract (supplementary Fig. S3D) suggests that most of the core protein in this tissue is engaged in dimers and oligomers, which limits its biological availability.

In the vascular system, the selective in vivo binding of MM1 to arterial SM cells, despite the presence of polypeptides shared with the vein in the range 226–217, 130, 120 and 33 kDa is likely to be due to two favourable conditions: (1) the biological availability in the arterial wall of the 100-kDa monomer in such amounts to be detected in immunocytochemical and dot-blot assays by MM1, and (2) a different, tissue-specific processing/assembling of decorin. It seems plausible that in venous SM cells, high-molecular-weight decorin intermediates and aggregates do not necessarily go through 100-kDa monomer (again this property seems tgo be shared with the elastic cartilage; see supplementary Fig. S3D). This conclusion does not imply that venous SM cells are incapable of producing the 100-kDa monomer. In fact, when these cells are grown in vitro enough 100 kDa (and 33 kDa) becomes detectable by MM1. Interestingly, the biosynthetic progression in both types of SM cells seems to be arrested in vitro as the high-molecular-weight intermediates/aggregates are lacking, probably because a different microenvironment is needed to complete the biosynthetic pathway.

Recent data further support the conclusion that decorin is differently synthesized in cultured arterial and venous SM cells (Wong et al. 2005). Conditioned media from rabbit arterial SM cells display a much higher content of the 100-kDa decorin monomer than the corresponding media from venous SM cells. The level of decorin is also paralleled by changes in: (1) PDGF and TGF-β1 expression, key regulatory factors for decorin synthesis (Fischer et al. 2001; Nili et al. 2003); (2) the pattern of SM cell differentiation and proliferation; and (3) the collagen synthesis profile (Wong et al. 2005). These authors and others (Fischer et al. 2001; Nili et al. 2003) have suggested that decorin can have an inherently modulatory (essentially inhibitory) role on SM cell proliferation in vivo, such that in vein-to-artery grafts venous SM cells are indeed more prone to develop atherosclerotic lesions, characterized by changes in the basic medial SM cell properties such as differentiation, proliferation and migration.

Among the putative factors that can regulate the unique molecular processing of decorin, certainly blood pressure is not involved as in the low-pressure environment of the (extra)pulmonary vein or arterial vasa vasorum SM cells express decorin (see Table 1). Similarly, the oxygen blood content does not seem to be involved in modulating the decorin expression in veins (supplementary Fig. S5).

Mechanical (endothelial injury) injury or haemodynamic (vein-to-artery transposition) changes to the vascular wall can indeed modulate the expression of the 100-kDa monomer. In particular, the carotid artery undergoes a marked rearrangement in the spatial distribution of MM1 immunoreactivity after lesioning (see Fig. 6) with a ‘shift’ of labelling to the innermost layers of the media underlying the intimal thickening. This pattern is reminiscent of that observed in vitro (Fig. 4) and during development (Fig. 5), and is possibly caused by the interaction of the 100-kDa monomer with cytokines and growth factors (such as PDGF, TGF-β1 and bFGF) released from damaged cells and ECM (Ross, 1993), according to a gradient emanating from the endothelium towards the adventitia. Similarly, the medial SM cells of a vein segment autotransplanted in arterial position are activated by locally released cytokines and growth factors (Newby & George, 1996), which facilitate detection of the 100-kDa monomer. This could be made possible because in these conditions more antigen is ex novo synthesized. Weakened interactions of decorin with other ECM proteins such as type I collagen, fibronectin, tropoelastin and fibrillin-1 (Järveläinen et al. 1991, 2004; Crippes Trask et al. 2000; Fischer et al. 2001; Reinboth et al. 2002; Nili et al. 2003), whose binding may indirectly influence decorin antigenicity, cannot be excluded. Importantly, profound structural and functional changes occur in SM cells during this post-injury phase, particularly in ECM (Sartore et al. 1999).

In concordance with other reports (Zhang et al. 1999; Yamakawa et al. 2000), neointimal SM cells in both experimental models do not show 100-kDa monomer expression. This could be related to the relatively late expression of this antigen, in comparison with other SM differentiation markers. It is well known that SM cells in the neointima are considered to be poorly differentiated, particularly in the early post-injury phase. These data suggest that the 100-kDa monomer of decorin is not strictly necessary for the initial acquisition of proliferation-migration potential by SM cells in the neointima. Instead, 100-kDa decorin could be more important for the subsequent phase of lesion expansion, as suggested by experiments in rat carotid artery where the cell-mediated transfer of bovine core protein does not influence SM cell proliferation, but it can alter the composition and organization of ECM (Fischer et al. 2000).

Supplementary material

The following supplementary material is available for this article online at

http://www.blackwellsynergy.com

Fig S1

Immunoperoxidase staining of pulmonary artery (pa, panel a), extrapulmonary (epv, panel b) and intrapulmonary (ipv, panel c) branches of pulmonary vein with MM1 antibody. Note that the SM cells of pa and epv are labelled, those of ipv are scarcely stained and grouped in small foci (asterisks). sm, smooth muscle layer. Scale bar, 70 µm.

Fig S2

Western blotting with MM1 (lanes 2–5) and F8 anti-COMP antibody (lanes 2′-5′) performed in an autodigestion experiment. Aorta SDS extracts were kept at 25 °C for times ranging from 5 min (lanes 2,2′) to 30 min (lanes 5,5′) before electrophoresis. The resulting pattern was studied with MM1 in comparison with the F8 anti-COMP antibody, which also labelled the apparent M_r 130-kDa band. While the immunoreactivity pattern of the 130- and 110-kDa bands with F8 changes with the time of autodigestion, the 130/120-kDa bands identified by MM1 remain substantially unaltered. These data indicate that these two polypeptides are not formed as a result of a proteolytic process. Lane 1: the electrophoretic profile of arterial extract at time 0.

Fig S3

MM1 immunoperoxidase staining of a neck lymph node (A) and elastic cartilage (C), and the corresponding Western blotting patterns of lymph node (B) and cartilage (D) extract with MM1. Panel A: note that AgMM1 is present in the periphery of lymph node (l) and in the germinal centre (gc) (inset). (C) All the chondrocytes in the elastic cartilage of the external ear are labelled with some staining localized to the cytoplasm encircling the nucleus (in b, arrowheads) or more diffusely around the chondrocytes as well as peripheral to the isogenous (i) cells (inset in b, arrows). Scale bars: a, 90 µm; b, 40 µm; insets: 30 µm. (B,D) Electrophoretic pattern (lanes 1) and Western blotting profile (lanes 1′). Arrows indicate standard molecular weight. Four micrograms of protein was loaded in the gels. Note the different Western blotting pattern of MM1 in lymph node and cartilage extract.

Fig S4

Immunoperoxidase staining of non-vascular tissues with MM1. (a) Bronchus wall; (b) kidney. Note that the outermost layers of the hyaline cartilage in the bronchus are MM1 positive. Some staining is present in the extracellular matrix (inset in a, arrow). In the renal tissue AgMM1 is localized to the Bowman's capsule of the renal corpuscle (g; inset in c). Scale bar: 80 µm; insets: 30 µm.

Fig S5

Immunoperoxidase staining of MM1 on umbilical veins (uv). Note that MM1 immunostaining (b) is negative with venous SM cells whereas these cells are positive for the SM cell marker SM22 (a). Scale bar, 60 µm.