The proteome of mouse cerebral arteries

Abstract

The cerebral vasculature ensures proper cerebral function by transporting oxygen, nutrients, and other substances to the brain. Distribution of oxygenated blood throughout the neuroaxis takes place at the level of the circle of Willis (CW). While morphologic and functional alterations in CW arteries and its main branches have been reported in cerebrovascular and neurodegenerative diseases, accompanying changes in protein expression profiles remain largely uncharacterized. In this study, we performed proteomics to compile a novel list of proteins present in mouse CW arteries and its ramifications. Circle of Willis arteries were surgically removed from 6-month-old wild-type mice, proteins extracted and analyzed by two proteomics approaches, gel-free nanoLC-mass spectrometry (MS)/MS and gel-based GelLC-MS/MS, using nanoAcquity UPLC coupled with ESI-LTQ Orbitrap XL. The two approaches helped maximize arterial proteome coverage. Six biologic and two technical replicates were performed. In all, 2,188 proteins with at least 2 unique high-scoring peptides were identified (6,630 proteins total). Proteins were classified according to vasoactivity, blood–brain barrier specificity, tight junction and adhesion molecules, membrane transporters/channels, and extracellular matrix/basal lamina proteins. Furthermore, we compared the identified CW arterial proteome with the published brain microvascular proteome. Our database provides a vital resource for the study of CW cerebral arterial protein expression profiles in health and disease.

Introduction

A healthy cerebral circulation is essential for maintaining brain perfusion and function.¹ In human, oxygenated blood to the brain is delivered by the internal carotid and vertebrobasilar arteries, and is distributed throughout the neuroaxis at the level of the circle of Willis (CW), a ringlike arterial structure located in the subarachnoid space at the base of the brain. The CW is formed by the confluence of the anastomotic branches of the two internal carotid arteries, rostral portion of the vertebrobasilar artery, and the anterior and posterior communicating arteries.² The arterial wreath allows for communications between the anterior and posterior circulations providing blood to the forebrain and the hindbrain, respectively, and insures redundancies in the cerebral circulation.³ The three principal brain arteries arising from the CW are the left and right anterior, middle, and posterior cerebral arteries, cortical branches of which penetrate the brain parenchyma to irrigate the cerebral cortex and deep structures of the brain.² In mice, the CW is similarly located along the ventral aspect of the brain extending from the pons–midbrain junction to the anterior cerebrum and involves the same major arteries.⁴

Several studies have reported morphologic and protein expression changes in the CW and its surface branches in healthy aging⁵ as well as pathologic conditions, such as cerebrovascular and neurodegenerative diseases (e.g., Alzheimer's disease).^{5, 6} Arterial wall thickening, loss of elasticity, and alterations in perivascular innervation are common changes observed in aged cerebral arteries.⁵ Pathology-associated alterations observed in brain arteries/arterioles resemble an admix of accelerated vascular aging and disease-specific hallmarks, such as amyloid deposits in Alzheimer's disease. To date, in comparison with the plethora of structural and functional changes documented in CW arteries of human and animal models, only a limited number of differentially expressed proteins have been identified. The main aim of our study was to generate, and compile a database of the mouse CW arterial proteome that will provide a valuable resource to study differential protein expression patterns in healthy and diseased brain arteries. In this paper, we have developed methodologies for efficient, mass spectrometry (MS)-compatible arterial protein extraction and peptide detection, and compared the CW arterial proteome with a previously published⁷ brain microvascular proteome.

Materials and methods

Mice

Eighteen 6-month-old C57BL/6J wild-type (WT) mice from two different breedings (B1 and B2) of nine mice each were used in this study. Furthermore, two 6-month-old WT mice from a third breeding were used for protein validation. To eliminate potential gender-related differences in brain vasculature only males were used. Mice were housed under a 12-hour light-dark cycle, in a temperature (23°C) and humidity (50%) controlled room, with food and tap water available ad libitum. All experiments were performed in compliance with the Animal Ethics Committee of the Montreal Neurological Institute and the Canadian Council on Animal Care guidelines.

Surgical Extraction of Circle of Willis Cerebral Arteries

Mice were anesthetized (65 mg/kg sodium pentobarbital, intraperitoneal) and transcardially perfused (0.9% saline, 4°C) for 5 minutes to clear the brain vasculature of blood contaminants. After each perfusion, brain was extracted, immersed in a precooled (4°C) plexiglass-dissection bath containing 0.9% saline, and secured (ventral side up first) with two pins to the transparent silicone cushion at the bottom of the bath. The CW, cerebral arteries (anterior, middle, and posterior), and their main ramifications were surgically removed from the pia mater under a dissecting microscope. Thereafter, isolated arteries underwent a second round of attached pia mater removal. Similarly prepared ‘pure' cerebral arteries from three mice were combined (constituted one biologic replicate) and stored at −80°C. A total of six (B1a, B1b, B1c, B2a, B2b, and B2c) biologic replicates were prepared for this study.

Protein Extraction from Circle of Willis Arteries

Proteins were extracted from surgically isolated arteries using Rapigest SF (Waters, Milford, MA, USA), a MS-compatible acid-labile surfactant that enhances solubilization and tryptic digestion efficiencies of proteins in complex biologic mixtures.⁸ To each surgically isolated sample, we added 100 μL of 0.2% (w/v) Rapigest SF in 50 mmol/L ammonium bicarbonate. The mixture was then exposed to three 20-second pulses of sonication (550 Sonic Dismembrator; Fisher Scientific, Ottawa, ON, Canada) to further facilitate protein extraction. The above steps were all performed on ice. After sonication, each preparation was incubated at 95°C for 10 minutes, followed by a 20-minute centrifugation (Eppendorf Centrifuge 5415D; Brinkmann Instruments, Westbury, NY, USA) at 10,000 g. The supernatant (S₀), containing soluble proteins, was transferred to a fresh tube and used as follows: 5 μL for protein concentration measurement (using Bradford assay; Bio-Rad Corporation, Philadelphia, PA, USA), 80 μL for gel-free proteomics (strong cation exchange (SCX)-LC-MS/MS), and the remaining for gel-based (GelLC-MS/MS) proteomics (see below). The pellet, however, was collagenase treated to recover unextracted/insoluble proteins. Collagenase treatment alters the structure of the collagen fiber network within the arterial wall,⁹ loosening the extracellular matrix (ECM) and promoting protein solubility. Each pellet was resuspended in 50 μL of buffered collagenase (0.1 mg/mL crude collagenase, 50 mmol/L Tris, pH 7.5, 2 mmol/L CaCl₂) and incubated for 60 minutes at 37°C. As crude collagenase also contains active proteases, these samples could only be used for gel-free analysis and not gel-based methods (as the latter requires proteins to be intact). Thus, the collagenase-treated sample was mixed back with 80 μL of its saved S₀ fraction, and the entire mixture sonicated three times (30 seconds each), followed by a 10-minute 95°C incubation in a dry bath. Thereafter, the mixture was centrifuged at 10,000 g for 20 minutes to remove debris and the supernatant (S₁) collected for gel-free proteomics using ion exchange separation (see below). A total of six biologic replicates were collected for gel-free analysis.

Proteomics Methods for Analyses of Cerebrovascular Proteins

We used two different approaches, gel-based and gel-free, to maximize CW arterial proteome coverage. Both approaches reduced sample complexity by fractionating either (1) at the level of the protein or (2) at the peptide level, respectively. It should be noted that while all six biologic replicates underwent gel-free proteomics, only three samples (B1a, B1b, and B1c) were additionally subjected to the gel-based method.

Gel-based approach (GelLC-MS/MS)

Each soluble protein extract S₀ was separated on a 12% one-dimensional SDS-PAGE. A gel cutter was used to consistently cut each sample lane into 29 contiguous pieces and allocated to 10 fractions. Each fraction was allotted 3 contiguous gel pieces, except for fractions 1 and 10. Fraction 1 containing the highest molecular weight proteins was assigned the top two contiguous gel pieces, while fraction 10 containing the lowest molecular weight proteins was apportioned the four contiguous pieces at the bottom of the gel. Each fraction was subjected to in-gel tryptic digestion, and the tryptic peptides were analyzed by nanoLC-MS/MS.

Gel-free approach (strong cation exchange-LC-MS/MS)

This approach involved protein deglycosylation and digestion followed by SCX-based separation of peptides. The proteins from soluble and collagenase-digested extracts (i.e., S1) were first denatured (10 minutes at 95°C in 100 mmol/L 2-mercaptoethanol) and then deglycosylated using N-Glycosidase F (Roche; Cat. # 11365185001, Laval, QC, Canada) at 37°C overnight to increase accessibility to trypsin in later step. Each sample was then reduced in 4 mmol/L dithiothreitol (10 minutes at 95°C) followed by alkylation in the dark in 10 mmol/L iodoacetic acid (30 minutes, at room temperature (RT)). Subsequently, each sample was trypsin digested (Trypsin Gold, MS grade, Promega, Madison, WI, USA; Cat. # V5280) by overnight incubation at 37°C. The resulting peptides were fractionated by SCX using a protocol described previously.¹⁰ In brief, an ICAT Cation Exchange Cartridge and Buffer Pack (Applied Biosystems, Carlsbad, CA, USA) were used to elute and collect peptide fractions in a stepwise manner, from low to high salt-containing SCX elution buffer concentration, by slowly injecting (∼1 drop/s) 0.5 mL of each elute buffer. Peptide fractions were collected at 20%, 40%, and 100% elution buffer concentrations.¹⁰ The approach is similar to the MUDPIT methodology used for microvessels,⁷ both being gel-free and label-free methods involving strong SCX separation followed by reverse phase nanoLC-MS/MS. However, unlike MUDPIT, our approach involved off-line SCX separation.

NanoLC-MS/MS Analysis

For each of the six biologic replicates, each SCX and/or in-gel-digest fraction was analyzed twice (technical replicates) by nanoLC-MS/MS. This culminated to a total of 96 nanoLC-MS/MS runs. In detail, each fraction was analyzed using ESI-LTQ-Orbitrap-XL mass spectrometer (Thermo) coupled to a NanoAcquity UPLC system (Waters). Samples were loaded on a NanoAcquity symmetry C18 trap (Waters) and separated on a 10 cm × 100 μm I.D. C18 column (Waters, 1.7 μm BEH130C18) at 250 nL/min using a 68-minutes gradient: 0% to 45% solution B (100% ACN/0.1% formic acid) over 66 minutes and 45% to 95% solution B over 2 minutes. Ten-minute washes in 40% solution B were followed by 30 minutes blank gradients between each sample to minimize carryover effects. The mass spectrometer was set to automated MS/MS analysis with an MS scan at 60 k resolution in the Orbitrap analyzer and data-dependent turbo MS/MS scans on the top three ions in the trap with dynamic exclusion (180 seconds).

Protein Identification and Bioinformatics

The raw data generated from LTQ-Orbitrap were converted to mzXML format using ReAdW program (http://tools.proteomecenter.org). MGF files were then generated from the mzXML file using MzXML2Search from the Trans Proteomics Pipeline project (http://tools.proteomecenter.org) and the resulting MS/MS spectra were searched against Mus musculus Swiss-Prot database using Mascot v2.2.0 (Matrix Science, Boston, MA, USA) search engine for protein identification with the following parameters: enzyme=trypsin; modifications=carbamidomethyl (C, fixed), oxidation (M, variable); peptide tolerance=1.5 Da; fragment tolerance=1.5 Da; 1 missed cleavage allowed). Decoy database searches were used to estimate the false positive identification rate. Search results were subsequently filtered to remove peptides with a delta mass of >15 p.p.m. or a score of <35. Under these conditions, the false positive identification rate based on decoy database searches was 1%. Peptides with a score of >50 had a false positive rate of 0.2%. As an independent statistical measure of peptide identification, Peptide Prophet probabilities (P scores) were also measured. All identified peptides had P scores ⩾0.95. For proteins identified based on only one peptide, additional peptides corresponding to the same proteins within the same gel band with P scores >0.9 were also examined and reported where indicated. To extract quantitative MS data, align all runs, and integrate protein search results, in-house software MatchRx¹⁰ version QnD-2.0 was used. MatchRx defines peptide peaks, extracts peptide abundance values from LC-MS data, allows quantitative comparison of peptide levels in two or more samples, and overlays the quantitative differences of peptides and LC-MS/MS identification results (from Mascot) into MSight images.¹⁰ Peptide intensity validation was performed using MSight software version 1.0 (http://www.expasy.ch/MSight), a visualization tool that allows graphical representation of the LC-MS and LC-MS/MS data.

Characterization of Arterial Protein List

Two protein lists (List 1 and List 2) of varying stringency were generated. All proteins with P scores ⩾0.95 were included in List 1. The second, more stringent List 2 only contained proteins identified with at least two peptide hits and P scores ⩾0.95. Subcellular localizations and function of proteins were determined using a combination of Gene Ontology (www.geneontology.org) and the Universal Protein Resource (UniProt, www.uniprot.org) databases. The PANTHER Classification System (www.pantherdb.org) was used to classify proteins into pathways. Only pathways with >2 proteins were accounted. Proteins were classified as endothelial cell (EC) proteins if present in an in-house EC database compiled through proteomics/literature mining, and as vascular smooth muscle cell (SMC) proteins if present in an in-house vascular SMC database compiled via literature mining of cultured arterial SMC proteomics data sets.

Western Blotting

Western blots were performed to validate List 1 proteins discoidin domain receptor tyrosine kinase 2 (DDR2) and hyperpolarization-activated cyclic nucleotide-gated cation channel 2 (HCN2), both known to be present in peripheral but not in cerebral arteries. DDR2 is a collagen receptor implicated in the regulation of SMC-mediated collagen turnover in atherosclerosis.¹¹ The ion channel HCN2 is suggested to have a role in the control of vascular tone and dysregulation of its expression may contribute to vascular pathology.¹²

Frozen CW arteries from each mouse (n=2) were digested in 35 μL Laemmli buffer.¹³ In brief, each preparation was repeatedly exposed to a pulse of vortexing (20 seconds, RT), followed by incubation (10 minutes, RT) for 2 hours. Thereafter, each preparation was incubated (10 minutes, 100°C), centrifuged (2 minutes, RT), and the supernatant retained. In all, 9 μL of the supernatant was loaded in 5% to 10% Tris/tricine SDS-PAGE gradient gels, separated by electrophoresis, and transferred onto nitrocellulose membranes. Membranes were incubated (1.5 hours, RT) in a blocking buffer containing 5% skim milk and then (overnight, 4°C) with rabbit anti-HCN2 (1:100; Alomone Labs, Jerusalem, Israel) or -DDR2 (1:150; Abcam, Cambridge, MA, USA). Actin was used as the loading control. Membranes were further incubated (1 hour, RT) with horseradish peroxidase-conjugated secondary antibodies (1:1,500 and 1:1,000, respectively) and proteins visualized by chemiluminescence (ECL plus kit; GE Healthcare) using a phosphorImager (Scanner STORM 860; GE Healthcare, Mississauga, ON, Canada).

Comparison with Microvascular Proteome

List 2 arterial proteins were compared with the previously released cerebral microvessel protein data set.⁷ Inclusion criteria used for microvascular proteins were the following: (1) the presence of ⩾2 unique tryptic peptides, (2) the presence of protein in all technical replicates, and in at least one biologic replicate, and (3) an average spectral count of ⩾5. Common and unique proteins to each data set were identified. The PANTHER Classification System was used to classify proteins by pathways. Proteins were listed as (1) blood–brain barrier (BBB)-specific cell type proteins, (2) tight junction (TJ) and adhesion proteins, (3) membrane transporter and channel proteins, and (d) ECM and basal lamina proteins, if they were (1) identified as such in the microvascular proteomics data set,⁷ (2) classified accordingly by Gene Ontology or UniProt databases, and were (3) present in either the mouse¹⁴ or rat¹⁵ BBB transcriptome repositories.

Results

Extraction and Solubilization of Cerebral Arteries

The CW and its ramifications (Figure 1A) were extracted in a reproducible manner using the surgical procedure described (see Materials and methods). Figure 1B outlines the protein solubilization and digest workflow developed in-house for efficient protein extraction from CW cerebral arteries. On average, 87 μg (ranged from 70 to 110 μg) proteins were extracted per biologic replicate. Figure 1C shows the reproducibility of two technical and biologic replicates, where the median relative standard deviation between technical replicates is 6% and between biologic replicates is 18%, values similar to that published for gel-free analysis.¹⁶

Figure 1.

Generation of the circle of Willis (CW) cerebral arterial database. (A) Surgically extracted fresh unfixed CW and its major ramifications from one mouse. Scale bar: 1 mm. (B) Simplified outline of the proteomics workflow used for the study. (C) Reproducibility of two technical (sample 1 versus sample 1) and two biologic (sample 1 versus sample 2) replicates. RSD, relative standard deviation; vs, versus; SCX, strong cation exchange.

Proteomic Profile

A total of 6,630 CW cerebral arterial proteins (List 1 proteins, Supplementary Table 1) were identified at P scores ⩾0.95. Of these proteins, 1,911 (28.8%) were detected using gel-based approach only, 2,875 (43.4%) using gel-free approach only, while 1,844 proteins (27.8%) were detected using both methods (Figure 2A). The latter was considered as ‘common CW cerebral arterial proteins'. Subcellularly, the top five locations List 1 proteins corresponded to membrane (41.3%), cytoplasm (40.1%), nucleus (32.1%), mitochondrion (10.2%), and secreted (8.8%) (Figure 2C). In addition, List 1 proteins DDR2 and HCN2 were validated using western blots (Figure 2D).

Figure 2.

Circle of Willis (CW) cerebral arterial protein database. (A) Percentage of List 1 proteins uniquely and commonly detected using gel-based and gel-free proteomics approaches. (B) Percentage distribution of the stringent List 2 proteins uniquely and commonly detected using gel-based and gel-free proteomics approaches. Approximately, half of the proteins present in List 2 were detected using both methods. (C) Distribution (in percentage) of subcellular localizations of proteins in List 1 and 2 proteins. (D) Detection of List 1 proteins, discoidin domain receptor tyrosine kinase 2 (DDR2) and hyperpolarization-activated cyclic nucleotide-gated cation channel 2 (HCN2), in CW arteries showed using western blots. Actin was used as the loading control. WT, wild type.

A total of 2,188 List 1 proteins (33%) were identified with ⩾2 peptides each and were cataloged as List 2 proteins (Supplementary Table 1). Approximately half (52%) of these were detected using both gel-based and gel-free approaches (Figure 2B). The top five subcellular locations of List 2 proteins were cytoplasm (46.3%), membrane (41.7%), nucleus (33.7%), mitochondrion (12.6%), and secreted (8.6%) (Figure 2C). A total of 466 proteins (21% of List 2 proteins) belonged to curated pathways in the PANTHER database. Ten pathways with >20 proteins were observed. In descending number of proteins they were as follows: integrin signaling (89 proteins), inflammation mediated by chemokine and cytokine signaling (53 proteins), heterotrimeric G-protein signaling (49 proteins), Wnt signaling (41 proteins), angiogenesis (36 proteins), platelet-derived growth factor signaling (36 proteins), cytoskeletal regulation by Rho GTPase (34 proteins), epidermal growth factor receptor signaling (27 proteins), cadherin signaling (23 proteins), and phosphatidylinositol 3-kinase pathway (23 proteins).

Vasoactivity

We detected 78 proteins shown to regulate the cerebral circulation or exert vasomotor effects on cerebral arteries (Table 1; Supplementary Table 2 for references and additional information). A few of these, specifically, neuronal mediators of vasoactivity and their vascular receptors detected in our arterial data set have been highlighted in the subsections below, and are illustrated in Figure 3.

Table 1. Proteins involved in vasomotricity.

Protein ID	Symbol	Protein name
P97718***	ADRA1A	Alpha-1A adrenergic receptor (ADRA1C)
Q8R429***	ATP2A1	Sarcoplasmic/endoplasmic reticulum calcium ATPase 1 (SERCA1)
O55143***	ATP2A2	Sarcoplasmic/endoplasmic reticulum calcium ATPase 2 (SERCA2)
Q64518***	ATP2A3	Sarcoplasmic/endoplasmic reticulum calcium ATPase 3 (SERCA3)
Q9ERZ4***	CHRM2	Muscarinic acetylcholine receptor M2
P97445***	CACNA1A	Voltage-dependent P/Q-type calcium channel subunit alpha-1A
P62204***	CALM1	Calmodulin
P35363***	HTR2A	5-Hydroxytryptamine receptor 2A
P34968***	HTR2C	5-Hydroxytryptamine receptor 1 or 2C
O55222***	ILK	Integrin-linked protein kinase
Q9Z329***	ITPR2	Inositol 1,4,5-trisphosphate receptor type 2 (IP3R2)
P70227***	ITPR3	Inositol 1,4,5-trisphosphate receptor type 3 (IP3R3)
P63143***	KCNAB1	Voltage-gated potassium channel subunit beta-1 ( Kvb1)
P63143***	KCNC2	Potassium voltage-gated channel subfamily C member 2 (Kv3.2)
Q14B80***	KCNH5	Potassium voltage-gated channel subfamily H member 5 (Kv10.2)
P59111***	KCNH8	Potassium voltage-gated channel subfamily H member 8 (Kv12.1)
P97794***	KCNJ8	ATP-sensitive inward rectifier potassium channel 8 (Inward rectifier K(+) channel Kir6.1)
Q08460***	KCNMA1	Calcium-activated potassium channel subunit alpha-1 (BK channel)
O89109***	KCNN4	Intermediate conductance calcium-activated potassium channel protein 4
Q8K3F6***	KCNQ3	Potassium voltage-gated channel subfamily KQT member 3 (Kv7.3)
Q8CFS6***	KCNV2	Potassium voltage-gated channel subfamily V member 2 (Kv8.2)
Q64133***	MAOA	Amine oxidase (flavin-containing) A
Q8BW75***	MAOB	Amine oxidase (flavin-containing) B
Q3THE2***	MYL12B	Myosin regulatory light chain 12B (myosin regulatory light chain 20 kDa)
Q6PDN3***	MYLK	Myosin light-chain kinase, smooth muscle (MLCK)
Q9Z0J4***	NOS1	Neuronal nitric oxide synthase
P29477***	NOS2	Inducible nitric oxide synthase
Q8K4S1***	PLCE1	1-Phosphatidylinositol-4,5-bisphosphate phosphodiesterase epsilon-1 (phospholipase C-epsilon-1)
Q4KWH5***	PLCH1	1-Phosphatidylinositol-4,5-bisphosphate phosphodiesterase eta-1 (phospholipase C-eta-1)
A2AP18***	PLCH2	1-Phosphatidylinositol-4,5-bisphosphate phosphodiesterase eta-2 (phospholipase C-eta-2)
P62137***	PPP1CA	Serine/threonine-protein phosphatase PP1-alpha catalytic subunit
P62141***	PPP1CB	Serine/threonine-protein phosphatase PP1-beta catalytic subunit
Q9DBR7***	PPP1R12A	Protein phosphatase 1 regulatory subunit 12A (myosin phosphatase-targeting subunit 1 or MYPT1)
Q8BG95***	PPP1R12B	Protein phosphatase 1 regulatory subunit 12B (myosin phosphatase-targeting subunit 2 or MYPT2)
Q3UMT1***	PPP1R12C	Protein phosphatase 1 regulatory subunit 12C (protein phosphatase 1 myosin-binding subunit of 85 kDa or MBS85)
Q61056***	TRPC1	Short transient receptor potential channel 1
Q9WVC5***	TRPC7	Short transient receptor potential channel 7
Q7TN37***	TRPM4	Transient receptor potential cation channel subfamily M member 4
O70176**	ADCYAP1	Pituitary adenylate cyclase-activating polypeptide
Q01815**	CACNA1C	Voltage-dependent L-type calcium channel subunit alpha-1C (Cav1.2)
Q8R3Z5**	CACNB1	Voltage-dependent L-type calcium channel subunit beta-1
Q8BIG7**	COMTD1	Catechol O-methyltransferase domain-containing protein 1
P11881**	ITPR1	Inositol 1,4,5-trisphosphate receptor type 1 (IP3R1)
P97382**	KCNAB3	Voltage-gated potassium channel subunit beta-3 ( Kvb3)
Q17ST2**	KCNA7	Potassium voltage-gated channel subfamily A member 7 (Kv1.7)
O88932**	KCNJ15	Inward rectifying potassium channel Kir4.2
Q9Z351**	KCNQ2	Potassium voltage-gated channel subfamily KQT member 2 (Kv7.2)
Q9JK97**	KCNQ4	Potassium voltage-gated channel subfamily KQT member 4 (Kv7.4)
Q9JK45**	KCNQ5	Potassium voltage-gated channel subfamily KQT member 5 (Kv7.5)
P70313**	NOS3	Endothelial nitric oxide synthase
P30549**	TACR2	Neurokinin A receptor
P24529**	TH	Tyrosine 3-monooxygenase
Q9R244**	TRPC2	Short transient receptor potential channel 2
Q9QX29**	TRPC5	Short transient receptor potential channel 5
P97751**	VIPR1	Vasoactive intestinal polypeptide receptor1
Q01337*	ADRA2C	Alpha-2C adrenergic receptor
Q9R0K7*	ATP2B2	Plasma membrane calcium-transporting ATPase 2 (PMCA2)
Q99246*	CACNA1D	Voltage-dependent L-type calcium channel subunit alpha-1D (Cav1.3)
Q9JIS7*	CACNA1F	Voltage-dependent L-type calcium channel subunit alpha-1F (Cav1.4)
O88427*	CACNA1H	Voltage-dependent T-type calcium channel subunit alpha-1H (Cav3.2)
Q03059*	CHAT	Choline O-acetyltransferase (ChAT)
P16390*	KCNA3	Potassium voltage-gated channel subfamily A member 3 (Kv1.3)
Q61762*	KCNA5	Potassium voltage-gated channel subfamily A member 5 (Kv1.5)
Q61923*	KCNA6	Potassium voltage-gated channel subfamily A member 6 (Kv1.6)
P58390*	KCNN2	Small conductance calcium-activated potassium channel protein 2
P97414*	KCNQ1	Potassium voltage-gated channel subfamily KQT member 1 (Kv7.1)
P70180*	NPR3	Atrial natriuretic peptide receptor 3 (NPRC)
Q61212*	NPY6R	Neuropeptide Y receptor type 6 (NPY5R)
Q9Z1M0*	P2RX7	P2X purinoceptor 7
Q9Z1B3*	PLCB1	1-Phosphatidylinositol-4,5-bisphosphate phosphodiesterase beta-1 (phospholipase C-beta-1)
P51432*	PLCB3	1-phosphatidylinositol-4,5-bisphosphate phosphodiesterase beta-3 (phospholipase C-beta-3)
P63087*	PPP1CC	Serine/threonine-protein phosphatase PP1-gamma catalytic subunit
Q91VC7*	PPP1R14A	Protein phosphatase 1 regulatory subunit 14A or 17 kDa PKC-potentiated inhibitory protein of PP1 (CPI17)
P20444*	PRKCA	Protein kinase C alpha type
P63318*	PRKCG	Protein kinase C gamma type
P70414*	SLC8A1	Sodium/calcium exchanger 1 (NCX)
Q9QUQ5*	TRPC4	Short transient receptor potential channel 4
Q61143*	TRPC6	Short transient receptor potential channel 6

Protein ID refers to those assigned by UniProt (www.uniprot.org). Cerebral arterial proteins identified with (1) *** peptides ⩾2, with at least two peptides exhibiting Peptide Prophet scores ⩾0.95, (2) ** peptides ⩾2, with one peptide exhibiting Peptide Prophet scores ⩾0.95, and (3) * 1 peptide with Peptide Prophet scores ⩾0.95.

Figure 3.

Neuronal mediators of vasoactivity. Perivascular nerve fibers originating from various peripheral ganglia release vasoactive mediators that act on specific receptors in the vascular wall, many of which were detected in our arterial data set and have been highlighted with colored circles. Protein IDs in figure refer to those assigned by UniProt (www.uniprot.org). In alphabetical order, detected proteins were as follows: ADCYAP1 or PACAP, pituitary adenylate cyclase-activating polypeptide; ADRA1A and 2C, the α-1A and -2C adrenergic receptors; CACNA1A, alpha1A subunit of the voltage-dependent P/Q-type calcium channel; CHAT (or ChAT), choline o-acetyltransferase; CHRM2 (or mAChR2), muscarinic acetylcholine receptor M2; HTR2A and 2C, serotonin receptors 2A and 2C; NOS1, neuronal nitric oxide synthase; NPR3, natriuretic peptide receptor C (also known to bind VIP); NPY6R, neuropeptide Y receptor 6; VIPR1, vasoactive intestinal polypeptide receptor 1; TACR2, neurokinin A receptor. 5-HT, 5-hydroxytryptamine; ACh, acetylcholine; CGRP, calcitonin gene-related peptide; CNS, central nervous system; NA, noradrenaline; NKA, neurokinin A; NO, nitric oxide; NPY, neuropeptide Y; PNS, peripheral nervous system; SCG, superior cervical ganglion; SP, substance P; SPG/OG, sphenopalatine and otic ganglia; TG, trigeminal ganglion; and VIP, vasoactive intestinal polypeptide. Adapted from Hamel¹ with permission from author (EH).

Neuronal proteins

Ultrastructural investigations of cerebral arteries show a rich neuronal network at the adventitia–media junction.¹⁷ Originating from the peripheral nervous system, these ‘extrinsic' nerve fibers are well-established modulators of vasoactivity.¹ Of the classic vasoactive substances released by these nerves, we detected the pituitary adenylate cyclase-activating polypeptide (ADCYAP1 or PACAP) (Figure 3). We also detected eight other neuronal facilitators of vasomotricity, in particular, neuronal nitric oxide (NO) synthase (NOS1), muscarinic acetylcholine (ACh) receptor M2 (CHRM2), alpha1A subunit of the voltage-dependent P/Q-type calcium channel (CACNA1A), and choline O-acetyltransferase (ChAT) represented in Figure 3, as well as, tyrosine 3-monooxygenase (or tyrosine hydroxylase, TH), monoamine oxidases type A and B (MAOA, B), and catechol O-methyltransferase domain-containing protein 1 (COMTD1).

Arterial proteins

We also detected receptors, known to be present on the vessel wall, that bind vasoactive molecules released by the extrinsic nerve fibers. These included the α-1A and α-2C adrenergic receptors (ADRA1A and ADRA2C), vasoactive intestinal polypeptide receptor 1 (VIPR1 or VPAC1), neurokinin A receptor (TACR2 or NK2R), and the 5-hydroxytryptamine receptors 2A and 2C (HTR2A and HTR2C) (Table 1; Figure 3).

In addition, we detected various smooth muscle and endothelial mediators of vasoactivity (Table 1; Supplementary Table 2 for references and additional information), including endothelial NO synthase (NOS3 or eNOS) that generates NO, a potent vasodilator. A cartoon diagram summarizing how some of Table 1 proteins mediate arterial smooth muscle motricity is presented in Figure 4.

Figure 4.

Protein effectors of arterial smooth muscle contraction and dilation. Simplified representation of proteins and protein pathways involved in vascular smooth muscle contraction. Protein effectors present in our arterial database are indicated in italics and are listed in Table 1. It should be noted that while protein effectors of vascular smooth muscle relaxation (in particular, K⁺ channels, PMCAs, Na⁺/Ca²⁺ exchanger, and SERCAs) are present in this cartoon diagram, they are not in their activated states. ADP, adenosine diphosphate; ATP, adenosine triphosphate; Ca²⁺, calcium; CaM, calmodulin; G, G protein; LC20 (or MYL12B), myosin 20 kDa light chain; LC20.P (or MYL12B.P), phosphorylated myosin 20 kDa light chain; MLCK (or MYLK), myosin light-chain kinase; MLCP, myosin light-chain phosphatase; NO, nitric oxide; PLC, phospholipase C; PIP₂, phosphatidylinositol 4,5-bisphosphate; PKC, protein kinase C; PMCAs, plasma membrane calcium ATPases; ROCCs, receptor-operated calcium channels; SERCAs, smooth endoplasmic reticulum calcium ATPases; SOCCs, store-operated calcium channels; VGCCs, voltage-gated calcium channels. Inspired from Horowitz et al¹⁹ with permission from last author (KGM).

Cerebral Arterial Versus Microvascular Proteomes

A total of 3,411 unique proteins were present in the arterial (List 2, 2,188 proteins total) and microvascular (1,824 proteins total) data sets combined (Figure 5A). Of these, 1,587 (47%) and 1,223 (36%) proteins were unique to the arterial and microvascular data sets, respectively, whereas 601 (18%) were common to both (Figure 5A). These common proteins comprised 28% of the List 2 CW cerebral arterial data set and 33% of the microvascular data set. Analysis of common proteins yielded a total of 62 PANTHER pathways, the top 5 being integrin signaling (35 proteins), heterotrimeric G-protein signaling (26 proteins), inflammation mediated by chemokine and cytokine signaling (23 proteins), cytoskeletal regulation by Rho GTPase (17 proteins), and Wnt signaling (13 proteins). A total of 58 pathways were identified for proteins solely present in either arteries or microvessels. Of these, 2 and 7 pathways were unique to arterial and microvascular proteomes, respectively, whereas 49 were common. Listed are data set-specific pathways with ⩾4 proteins: (1) hypoxia response via hypoxia-inducible factor activation pathway (unique to arterial data set) and (2) 5HT4 type receptor mediated signaling pathway, corticotropin releasing factor receptor signaling pathway, opioid (prodynorphin, proenkephalin, and proopiomelanocortin) pathways, β3 adrenergic receptor signaling pathway, and androgen/estrogen/progesterone biosynthesis pathway (unique to microvascular data set). In the following subsections, we compare protein distributions of (1) smooth muscle and endothelial components and (2) BBB and associated categories in the arterial and microvascular data sets.

Figure 5.

Cerebral arterial versus microvascular proteomes. (A) Distribution of unique and common proteins present in the arterial and microvascular databases. (B) Percent distribution of unique vascular smooth muscle and endothelial proteins present in the two-abovementioned data sets. Proteins were classified as belonging to one of the two categories based on their presence in the in-house compiled endothelial cell (EC) and SM cell (SMC) databases. Distribution of arterial and microvascular proteins present in four different categories, namely, (C) BBB-specific cell types, (D) tight junction and adhesion proteins, (E) membrane transporter and channel proteins, and (F) ECM and basal lamina proteins. BBB, blood–brain barrier; ECM, extracellular matrix; SM, smooth muscle.

Smooth muscle cell and endothelial cell protein distribution

We identified 230 (11%) SMC- and 1,035 (47%) EC-classified proteins uniquely present in the arterial data set containing a total of 2,188 proteins (Figure 5B). In all, 204 (9%) of these proteins were common to both cell types. In comparison, we established that 122 (7%) SMC- and 623 (34%) EC-classified proteins were unique to the microvascular data set containing a total of 1,824 proteins (Figure 5B). In all, 107 (6%) of these proteins were common to both cell types. In addition, we identified 218 SMC- and 515 EC-classified proteins common to both arterial and microvascular data sets (containing a total of 3,411 proteins), with 206 (6%) belonging to both cell types.

Blood–brain barrier-specific proteins

In the arterial and microvascular data sets combined, we detected 86 BBB-specific proteins, 28 (33%) of which were detected in both sample types (Table 2; Figure 5C). While 52 BBB-enriched proteins were microvascular data set specific, we detected only 6 such proteins exclusive to our arterial data set. These included two transporters (ABCB1A and SLC2A1), three transcriptional regulators (EBF1, FOXQ1, and LEF1), and the DNA repair protein RAD54B.

Table 2. Cerebral artery proteome classified into BBB-specific proteins, tight junction and adhesion molecules, membrane transporters and channels, and ECM and basal lamina proteins.

UniProt ID	Symbol	Protein name	M
BBB-specific proteins
P20152***	VIM	Vimentin	y
P04370***	MBP	Myelin basic protein	y
Q91ZX7***	LRP1	Prolow-density lipoprotein receptor-related protein 1	y
P21279***	GNAQ	Guanine nucleotide-binding protein G(q) subunit alpha	y
Q61490***	ALCAM	CD166 antigen	y
Q9CZJ2***	HSPA12B	Heat shock 70 kDa protein 12B	y
P48678***	LMNA	Lamin-A/C	y
Q8BTM8***	FLNA	Filamin-A	y
Q60634***	FLOT2	Flotillin-2	y
P08228***	SOD1	Superoxide dismutase [Cu-Zn]	y
P10833***	RRAS	Ras-related protein R-Ras	y
Q9R0P5***	DSTN	Destrin	y
Q8R2Z5***	VWA1	von Willebrand factor A domain-containing protein 1	y
P49817***	CAV1	Caveolin-1	y
Q3UMF0***	COBLL1	Cordon-bleu protein-like 1	y
P15208***	INSR	Insulin receptor	y
Q8CIZ8***	VWF	von Willebrand factor	y
P27600***	GNA12	Guanine nucleotide-binding protein subunit alpha-12	y
P03995***	GFAP	Glial fibrillary acidic protein	y
Q8R0Y6***	ALDH1L1	10-formyltetrahydrofolate dehydrogenase	y
Q9QUP5***	HAPLN1	Hyaluronan and proteoglycan link protein 1	y
P62737***	ACTA2	Actin, aortic smooth muscle	y
P63094***	GNAS	Guanine nucleotide-binding protein G(s) subunit alpha isoform short	y
Q60675***	LAMA2	Laminin subunit alpha-2	y
Q61830***	MRC1	Macrophage mannose receptor 1	y
Q61696***	HSPA1A	Heat-shock 70-kDa protein 1A	y
Q8VHY0***	CSPG4	Chondroitin sulfate proteoglycan 4	y
P05622***	PDGFRB	Beta-type platelet-derived growth factor receptor	y
P21447***	ABCB1A	Multidrug resistance protein 3	—
Q07802***	EBF1	Transcription factor COE1	—
O70220***	FOXQ1	Forkhead box protein Q1	—
P27782***	LEF1	Lymphoid enhancer-binding factor 1	—
Q6PFE3***	RAD54B	DNA repair and recombination protein RAD54B	—
P17809***	SLC2A1	Solute carrier family 2, facilitated glucose transporter member 1	—
Tight junction and/or adhesion proteins
Q61490***	ALCAM	CD166 antigen	y
P15379***	CD44	CD44 antigen	y
P26231***	CTNNA1	Catenin alpha-1	y
Q61301***	CTNNA2	Catenin alpha-2	y
Q02248***	CTNNB1	Catenin beta-1	y
P30999***	CTNND1	Catenin delta-1	y
O35927***	CTNND2	Catenin delta-2	y
Q62108***	DLG4	Disks large homolog 4	y
Q3V3R4***	ITGA1	Integrin alpha-1	y
Q62470***	ITGA3	Integrin alpha-3	y
Q61738***	ITGA7	Integrin alpha-7	y
P43406***	ITGAV	Integrin alpha-V	y
P09055***	ITGB1	Integrin beta-1	y
O54890***	ITGB3	Integrin beta-3	y
Q02257***	JUP	Junction plakoglobin	y
Q6RHR9***	MAGI1	Membrane-associated guanylate kinase, WW & PDZ domain-containing protein 1	y
Q9EQJ9***	MAGI3	Membrane-associated guanylate kinase, WW & PDZ domain-containing protein 3	y
Q8R2Y2***	MCAM	Cell surface glycoprotein MUC18	y
Q8VBX6***	MPDZ	Multiple PDZ domain protein	y
P70290***	MPP1	55 kDa erythrocyte membrane protein	y
Q810U4***	NRCAM	Neuronal cell adhesion molecule	y
Q08481***	PECAM1	Platelet endothelial cell adhesion molecule	y
Q80 × 82***	SYMPK	Symplekin	y
P39447***	TJP1	Tight junction protein ZO-1	y
Q9Z0U1***	TJP2	Tight junction protein ZO-2	y
P29533***	VCAM1	Vascular cell adhesion protein 1	y
P83741***	WNK1	Serine/threonine-protein kinase WNK1	y
Q8K371***	AMOTL2	Angiomotin-like protein 2	—
Q9WTR5***	CDH13	Cadherin-13	—
Q99PF4***	CDH23	Cadherin-23	—
Q9Z0S3***	CLDN14	Claudin-14	—
Q65CL1***	CTNNA3	Catenin alpha-3	—
Q8BJ42***	DLGAP2	Disks large-associated protein 2	—
Q63ZW7***	INADL	InaD-like protein	—
Q00651***	ITGA4	Integrin alpha-4	—
P11688***	ITGA5	Integrin alpha-5	—
A2ARA8***	ITGA8	Integrin alpha-8	—
P05555***	ITGAM	Integrin alpha-M	—
A2A863***	ITGB4	Integrin beta-4	—
Q9Z0T9***	ITGB6	Integrin beta-6	—
P97350***	PKP1	Plakophilin-1	—
Q9QY23***	PKP3	Plakophilin-3	—
P32507***	PVRL2	Poliovirus receptor-related protein 2	—
A2ALU4***	SHROOM2	Protein Shroom2	—
Q4G0F8***	UBN1	Ubinuclein-1	—
Q3UH66***	WNK2	Serine/threonine-protein kinase WNK2	—
Q80XP9***	WNK3	Serine/threonine-protein kinase WNK3	—
Membrane transporter and channel proteins
P41233***	ABCA1	ATP-binding cassette sub-family A member 1	y
P21447***	ABCB1A	Multidrug resistance protein 3	y
Q61102***	ABCB7	ATP-binding cassette sub-family B member 7	y
P41234***	ABCA2	ATP-binding cassette sub-family A member 2	—
Q8R420***	ABCA3	ATP-binding cassette sub-family A member 3	—
Q8K441***	ABCA6	ATP-binding cassette sub-family A member 6	—
Q91V24***	ABCA7	ATP-binding cassette sub-family A member 7	—
Q5SSE9***	ABCA13	ATP-binding cassette sub-family A member 13	—
P21440***	ABCB4	Multidrug resistance protein 2	—
Q9QY30***	ABCB11	Bile salt export pump	—
Q8VI47***	ABCC2	Canalicular multispecific organic anion transporter 1	—
Q9R1 × 5***	ABCC5	Multidrug resistance-associated protein 5	—
P61222***	ABCE1	ATP-binding cassette subfamily E member 1	—
Q64343***	ABCG1	ATP-binding cassette subfamily G member 1	—
Q8VDN2***	ATP1A1	Sodium/potassium-transporting ATPase subunit alpha-1	y
Q6PIE5***	ATP1A2	Sodium/potassium-transporting ATPase subunit alpha-2	y
Q6PIC6***	ATP1A3	Sodium/potassium-transporting ATPase subunit alpha-3	y
P14094***	ATP1B1	Sodium/potassium-transporting ATPase subunit beta-1	y
P14231***	ATP1B2	Sodium/potassium-transporting ATPase subunit beta-2	y
P97370***	ATP1B3	Sodium/potassium-transporting ATPase subunit beta-3	y
Q8R429***	ATP2A1	Sarcoplasmic/endoplasmic reticulum calcium ATPase 1	y
O55143***	ATP2A2	Sarcoplasmic/endoplasmic reticulum calcium ATPase 2	y
Q64518***	ATP2A3	Sarcoplasmic/endoplasmic reticulum calcium ATPase 3	y
Q64436***	ATP4A	Potassium-transporting ATPase alpha chain 1	y
Q03265***	ATP5A1	ATP synthase subunit alpha, mitochondrial	y
P56480***	ATP5B	ATP synthase subunit beta, mitochondrial	y
Q91VR2***	ATP5C1	ATP synthase subunit gamma, mitochondrial	y
Q9CQQ7***	ATP5F1	ATP synthase subunit b, mitochondrial	y
Q9DCX2***	ATP5H	ATP synthase subunit d, mitochondrial	y
P97450***	ATP5J	ATP synthase-coupling factor 6, mitochondrial	y
P56135***	ATP5J2	ATP synthase subunit f, mitochondrial	y
Q9CPQ8***	ATP5L	ATP synthase subunit g, mitochondrial	y
Q9DB20***	ATP5O	ATP synthase subunit O, mitochondrial	y
P50516***	ATP6V1A	V-type proton ATPase catalytic subunit A	y
P62814***	ATP6V1B2	V-type proton ATPase subunit B, brain isoform	y
Q8BVE3***	ATP6V1H	V-type proton ATPase subunit H	y
P70704***	ATP8A1	Probable phospholipid-transporting ATPase IA	y
Q9Z1W8***	ATP12A	Potassium-transporting ATPase alpha chain 2	y
A7L9Z8***	ATP2C2	Calcium-transporting ATPase type 2C member 2	—
Q06185***	ATP5I	ATP synthase subunit e, mitochondrial	—
Q3UHK1***	SLC2A13	Proton myo-inositol cotransporter	y
Q761V0***	SLC6A5	Sodium- and chloride-dependent glycine transporter 2	y
Q8BLV3***	SLC9A7	Sodium/hydrogen exchanger 7	y
P57787***	SLC16A3	Monocarboxylate transporter 4	y
Q8VEM8***	SLC25A3	Phosphate carrier protein, mitochondrial	y
P48962***	SLC25A4	ADP/ATP translocase 1	y
P51881***	SLC25A5	ADP/ATP translocase 2	y
Q9CR62***	SLC25A11	Mitochondrial 2-oxoglutarate/malate carrier protein	y
Q8BH59***	SLC25A12	Calcium-binding mitochondrial carrier protein Aralar1	y
Q5U680***	SLC25A26	S-adenosylmethionine mitochondrial carrier protein	y
Q8R0Y8***	SLC25A42	Solute carrier family 25 member 42	y
Q80SU6***	SLC34A3	Sodium-dependent phosphate transport protein 2C	y
P43006***	SLC1A2	Excitatory amino-acid transporter 2	—
O35544***	SLC1A6	Excitatory amino-acid transporter 4	—
P17809***	SLC2A1	Solute carrier family 2, facilitated glucose transporter member1	—
P10852***	SLC3A2	4F2 cell-surface antigen heavy chain	—
Q8K0E3***	SLC5A11	Sodium/myo-inositol cotransporter 2	—
O88576***	SLC6A18	Sodium- and chloride-dependent transporter XTRP2	—
Q8BUE1***	SLC9A4	Sodium/hydrogen exchanger 4	—
Q6UJY2***	SLC9A10	Sodium/hydrogen exchanger 10	—
Q91Y63***	SLC13A3	Solute carrier family 13 member 3	—
Q8VC69***	SLC22A6	Solute carrier family 22 member 6	—
Q8BMD8***	SLC25A24	Calcium-binding mitochondrial carrier protein SCaMC-1	—
Q8VC69***	SLC26A1	Solute carrier family 22 member 6	—
P97441***	SLC30A3	Zinc transporter 3	—
Q60825***	SLC34A1	Sodium-dependent phosphate transport protein 2A	—
Q9DBP0***	SLC34A2	Sodium-dependent phosphate transport protein 2B	—
Q8CD26***	SLC35E1	Solute carrier family 35 member E1	—
ECM and basal lamina proteins
A2ASQ1***	AGRN	Agrin	y
P07356***	ANXA2	Annexin A2	y
P02463***	COL4A1	Collagen alpha-1(IV) chain	y
P08122***	COL4A2	Collagen alpha-2(IV) chain	y
P11276***	FN1	Fibronectin	y
Q9QUP5***	HAPLN1	Hyaluronan and proteoglycan link protein 1	y
Q05793***	HSPG2	Basement membrane-specific heparan sulfate proteoglycan core protein	y
P97927***	LAMA4	Laminin subunit alpha-4	y
Q61001***	LAMA5	Laminin subunit alpha-5	y
Q61292***	LAMB2	Laminin subunit beta-2	y
P02468***	LAMC1	Laminin subunit gamma-1	y
Q9R0B6***	LAMC3	Laminin subunit gamma-3	y
P10493***	NID1	Nidogen-1	y
O88322***	NID2	Nidogen-2	y
Q60675***	LAMA2	Laminin subunit alpha-2	y
Q9R001***	ADAMTS5	A disintegrin and metalloproteinase with thrombospondin motifs 5	—
Q811B3***	ADAMTS12	A disintegrin and metalloproteinase with thrombospondin motifs 12	—
P59384***	ADAMTS15	A disintegrin and metalloproteinase with thrombospondin motifs 15	—
Q69Z28***	ADAMTS16	A disintegrin and metalloproteinase with thrombospondin motifs 16	—
P59509***	ADAMTS19	A disintegrin and metalloproteinase with thrombospondin motifs 19	—
P59511***	ADAMTS20	A disintegrin and metalloproteinase with thrombospondin motifs 20	—
Q80T21***	ADAMTSL4	ADAMTS-like protein 4	—
Q99MQ4***	ASPN	Asporin	—
P28653***	BGN	Biglycan	—
Q8R2G6***	CCDC80	Coiled-coil domain-containing protein 80	—
Q61245***	COL11A1	Collagen alpha-1(XI) chain	—
Q64739***	COL11A2	Collagen alpha-2(XI) chain	—
Q60847***	COL12A1	Collagen alpha-1(XII) chain	—
Q80 × 19***	COL14A1	Collagen alpha-1(XIV) chain	—
O35206***	COL15A1	Collagen alpha-1(XV) chain	—
Q8BLX7***	COL16A1	Collagen alpha-1(XVI) chain	—
Q07563***	COL17A1	Collagen alpha-1(XVII) chain	—
P39061***	COL18A1	Collagen alpha-1(XVIII) chain	—
Q0VF58***	COL19A1	Collagen alpha-1(XIX) chain	—
P11087***	COL1A1	Collagen alpha-1(I) chain	—
Q01149***	COL1A2	Collagen alpha-2(I) chain	—
Q30D77***	COL24A1	Collagen alpha-1(XXIV) chain	—
Q5QNQ9***	COL27A1	Collagen alpha-1(XXVII) chain	—
Q2UY11***	COL28A1	Collagen alpha-1(XXVIII) chain	—
P28481***	COL2A1	Collagen alpha-1(II) chain	—
P08121***	COL3A1	Collagen alpha-1(III) chain	—
Q9QZS0***	COL4A3	Collagen alpha-3(IV) chain	—
O88207***	COL5A1	Collagen alpha-1(V) chain	—
Q3U962***	COL5A2	Collagen alpha-2(V) chain	—
Q04857***	COL6A1	Collagen alpha-1(VI) chain	—
Q02788***	COL6A2	Collagen alpha-2(VI) chain	—
A2AX52***	COL6A4	Collagen alpha-4(VI) chain	—
A6H584***	COL6A5	Collagen alpha-5(VI) chain	—
Q8C6K9***	COL6A6	Collagen alpha-6(VI) chain	—
Q63870***	COL7A1	Collagen alpha-1(VII) chain	—
Q00780***	COL8A1	Collagen alpha-1(VIII) chain	—
P25318***	COL8A2	Collagen alpha-2(VIII) chain	—
Q05722***	COL9A1	Collagen alpha-1(IX) chain	—
Q07643***	COL9A2	Collagen alpha-2(IX) chain	—
Q8R555***	CRTAC1	Cartilage acidic protein 1	—
P28654***	DCN	Decorin	—
P54320***	ELN	Elastin	—
Q91VF5***	EMID1	EMI domain-containing protein 1	—
Q99K41***	EMILIN1	EMILIN-1	—
Q08879***	FBLN1	Fibulin-1	—
P37889***	FBLN2	Fibulin-2	—
Q61554***	FBN1	Fibrillin-1	—
Q61555***	FBN2	Fibrillin-2	—
P50608***	FMOD	Fibromodulin	—
O35367***	KERA	Keratocan	—
Q61789***	LAMA3	Laminin subunit alpha-3	—
Q61087***	LAMB3	Laminin subunit beta-3	—
Q61092***	LAMC2	Laminin subunit gamma-2	—
P16045***	LGALS1	Galectin-1	—
Q9D1H9***	MFAP4	Microfibril-associated glycoprotein 4	—
Q9R1A3***	NTN3	Netrin-3	—
Q9JI33***	NTN4	Netrin-4	—
Q62009***	POSTN	Periostin	—
Q9JK53***	PRELP	Prolargin	—
P04202***	TGFBI	Transforming growth factor beta-1	—
Q8R2Z5***	VWA1	von Willebrand factor A domain-containing protein	—
Q8CIZ8***	VWF	von Willebrand factor	—
P70701***	WNT10A	Protein Wnt-10a	—
P22725***	WNT5A	Protein Wnt-5a	—

Cerebral arterial proteins were identified with ⩾2 peptides each and with at least two peptides exhibiting Peptide Prophet scores ⩾0.95 (indicated with ***). Note that proteins present only in the microvascular data set are not listed. In column M (or microvascular), proteins detected in the study by Chun et al⁷ and present in our arterial data set are indicated as ‘y'. y, yes.

Tight junction and adhesion proteins

We identified 83 TJ and/or adhesion proteins. Of these proteins, 19 and 37 were specific to arteries and microvessels, respectively, while 27 proteins were common to both data sets (Figure 5D; Table 2). In descending order, common proteins belonged to the following groups: nine TJ plaque proteins: DLG4, MAGI1, MAGI3, MPDZ, MPP1, SYMPK, TJP1, TJP2, and WNK1; six immunoglobulin superfamily cell adhesion molecules (ALCAM, CD44, MCAM, NRCAM, PECAM1, and VCAM1), six integrins (α1, 3, 7, V and β1, 3), and five catenins (α1 and 2, β1, and δ1, 2) (Supplementary Table 3—for protein function and cellular component details).

Membrane transporter and channel proteins

We detected 138 proteins in the arterial and microvascular data sets involved in transport across the plasma membrane. Sixty-eight of these proteins were present in our arterial data set and include 28 solute carrier proteins (SLCs), 14 adenosine 5′-triphosphate (ATP)-binding cassette proteins (ABCs), and 26 ATPases (Na⁺/K⁺, H⁺/K⁺, and Ca²⁺) (Table 2). Twenty-nine of the above-mentioned proteins were unique to our data set, while thirty-nine were also present in the microvascular data set (Table 2; Figure 5E). In comparison, a larger number (70) of transporters were found exclusively in microvascular data set (Figure 5E).

Extracellular matrix and basal lamina proteins

A total of 94 ECM and/or basal lamina proteins were identified in the two data sets combined (Figure 5F). Of these 64 were present in our arterial data set only (Table 2), while a smaller number (15) were microvascular specific. Fifteen proteins were common to both data sets (Table 2). These included six laminins (subunits α2, α4, α5, β2, γ1, and γ3), two collagens (IVα1 and IVα2), two nidogens (1 and 2), agrin, annexin A2, fibronectin, hyaluronan/proteoglycan link protein 1, and basement membrane-specific heparan sulfate proteoglycan core protein.

Discussion

We have generated the first extensive database of 6,630 proteins expressed in the wall of fresh cerebral arteries. An advantage of using freshly isolated over cultured tissue is that it allows for maximal preservation of the tissue microenvironment experience in vivo in the absence of potential tissue culture-induced artifacts, such as phenotypic drift, that can result in altered protein expression.¹⁸ Of the 6,630 proteins in our database, 33% were detected with ⩾2 peptides at P scores ⩾0.95, and the remaining 67% with at least 1 peptide and the same stringent P scores cutoff. We established that approximately half (56%) of the proteins in our arterial databank are of known vascular origin, being copresent in the EC and/or the vascular SMC protein databases. We further established that almost a third of these proteins are robustly expressed in cerebral arteries and can be detected by both gel-based and gel-free approaches.

Arterial Proteome Coverage

To generate a comprehensive arterial protein database, we developed highly reproducible methodologies that allowed for (1) rapid surgical isolation of fresh cerebral arteries, (2) efficient protein extraction using the MS-compatible surfactant Rapigest, with yields ∼3 × higher compared with SDS-based extraction (unpublished data), and (3) optimal detection and quantification of low-abundant peptides. Concomitant use of gel-based and gel-free approaches contributed to increased detection of peptides (and proteins) as showcased by (1) the presence of EC layer proteins, which are harder to detect since unlike the SMC and adventitia layers, the EC layer is a single layer of cells and (2) good agreement between the breakdown of detected arterial proteins by subcellular localization and the relative sizes of cellular compartments (e.g., membrane>mitochondria).

Cerebral Arteries and Vasoactivity

Large arteries as well as smaller arterioles at the brain surface constitute resistance vessels that control blood flow and influence global perfusion.⁵ Vascular resistance is essential for pushing blood through the cerebral circulation, and is increased through contraction of the vessel muscular wall, and decreased via vasodilation stemming from the relaxation of SMCs.⁵ Interestingly, 3 of the top 10 PANTHER pathways detected in our arterial data set, namely heterotrimeric G-proteins (e.g., Figure 4), integrin signaling (e.g., Figure 4), and cytoskeletal regulation by Rho GTPases, are established modulators of vasomotricity.^{19, 20, 21} In addition to signaling cascade proteins, we also identified several perivascular and vascular mediators of vasoactivity that are discussed in the following two sections.

Neuronally released vasoactive substances and their receptors

Nerve fibers originating in the peripheral nervous system, specifically, the superior cervical ganglion (sympathetic innervation), the sphenopalatine and otic ganglia (parasympathetic innervation), and the trigeminal ganglion exert vasomotor control on cerebral arteries via release of neurotransmitters and other vasoactive substances.¹ We detected vascular receptors for five of these substances in our arterial data set, in particular, the α-1A and α-2C adrenergic receptors (ADRA1A and ADRA2C); the neuropeptide Y receptor, NPY6R; serotonin receptors, HTR2A (or 5-HT2A) and HTR2C (or 5-HT2C); the established vasoactive intestinal polypeptide receptor, VIPR1 (or VPAC1); the lesser known intestinal polypeptide receptor, NPR3 (or NPRC (natriuretic peptide receptor C)); and finally, the neurokinin A receptor TACR2 (or NKR2) (Figure 3). The presence of all above-mentioned receptors, with the exception of NPY6R and HTR2C, has been previously shown in cerebral arteries (for references, see Supplementary Table 2). Moreover, the detected receptors are G protein-coupled and modulate vasoactivity via heterotrimeric G-protein signaling, interestingly, a top 10 PANTHER pathway detected in our study.

We also detected nine neuronal mediators of vasoactivity (Table 1; Supplementary Table 2 for references; and Figure 3). Of these, the pituitary adenylate cyclase activating peptide (ADCYAP1 or PACAP), released by perivascular nerve fibers originating in the trigeminal ganglion, is an established vasodilator.²² The calcium channel CACNA1A facilitates release of the calcitonin gene-related protein,²³ another vasodilator compound released by trigeminal ganglion originating nerves.¹ The presynaptic muscarinic ACh receptor CHRM2 (or mAChR2), located on ACh releasing nitric oxidergic nerves originating in sphenopalatine and otic ganglia, is known to mediate inhibition of neurogenic vasodilation in cerebral arteries via reduction in NOS activity and the production and release of the vasodilator NO.²⁴ Both ChAT and NOS1 (or nNOS) enzymes involved in the synthesis of ACh and NO, respectively, were present in our arterial data set. In addition, we detected four enzymes (TH, MAOA, MAOB, and COMTD1) involved in the formation and degradation of superior cervical ganglion released vasoactive substance, norepinephrine (or noradrenaline).²⁵

Proteins involved in vascular smooth muscle contractility and relaxation

Among the key effectors of smooth muscle contraction we detected calmodulin (CALM1), the 20-kDa myosin light chain (MYL12B or LC20), as well as the myosin light chain-specific kinase (MYLK or MLCK) and phosphatase (PPP1 or MLCP). Both catalytic (PPP1CA, B, and C) and myosin-targeting (PPP1R12A, B, and C) subunits of the phosphatase were detected, with only PPP1R12A previously reported in cerebral arteries (Supplementary Table 2 for references). It is well established that intracellular-free calcium (iCa²⁺) binds with calmodulin, which then binds and activates the myosin light-chain kinase.¹⁹ The activated kinase complex then phosphorylates myosin on the 20-kDa light chains, resulting in smooth muscle contraction.¹⁹ Conversely, dephosphorylation of myosin light chains induces relaxation (Figure 4). Activated protein kinase C (PKC) and integrins are known to inhibit dephosphorylation under physiologic conditions via the PKC-potentiated inhibitory protein of MLCP (PPP1R14A) and the integrin linked kinase, respectively²⁰ (Figure 4). We detected two classic PKCs (PRKCA and PRKCG), PPP1R14A and integrin linked kinase in our arterial data set, with all except for PRKCG previously reported in cerebral arteries (Supplementary Table 2 for references), as well as several integrins (see Table 2).

Increases in iCa²⁺ are a major determinant of arterial smooth muscle contractility.¹⁹ In SMCs, rise in iCa²⁺ is facilitated by (1) Ca²⁺ release from intracellular stores in the smooth sarcoplasmic reticulum and (2) Ca²⁺ entry from the extracellular space via plasma membrane Ca²⁺ channels (Figure 4). Ca²⁺ release from sarcoplasmic reticulum requires hydrolysis of plasma membrane phosphatidylinositol 4,5-bisphosphate by phospholipase C (PLC).²¹ We detected five PLC isozymes belonging to β (B1, 3), ɛ (E1), and η (H1, 2) subtypes, of which, PLC-β, and -ɛ are activated by agonist (including vasoactive substances) binding to plasma membrane G protein-coupled receptors.²¹ All PLC isozymes detected are known generators of second messengers, inositol 1,4,5-trisphosphate (IP3) and 1,2-diacylglycerol.²¹ Binding of IP3 to sarcoplasmic reticulum surface receptors (IP3Rs or ITPRs) activates Ca²⁺ release.²⁶ We detected three ITPRs, namely ITPR1, 2, and 3, all previously showed to be present in rat cerebral artery SMCs with ITPR1 being the most abundant.²⁷ In addition, the second messenger 1,2-diacylglycerol has been shown to activate PKC, which in turn increases smooth muscle contractility via several different mechanisms such as MLCP inhibition and increased actin availability.²⁰

Several pathways of Ca²⁺ entry from the extracellular space have been suggested, and include passive leak Ca²⁺ channels, as well as channels activated by membrane depolarization (voltage gated), agonist binding (receptor operated), depletion of Ca²⁺ from intracellular stores (store operated), or stretch.²⁸ Voltage-gated Ca²⁺ channels enriched in SMCs of major cerebral arteries and arterioles are of L, and to a lesser extent, T type.²⁹ We detected four α-1 subunits belonging either to the high-voltage-activated L-type (CACNA1C, D, and F) or to the low-voltage-activated T-type (CACNA1H) channels. Of these, mRNAs of CACNA1C (or Cav1.2) and CACNA1H (or Cav3.2) have been previously detected in rat cerebral artery SMC and mouse coronary artery, respectively (Supplementary Table 2 for references). We also detected a known β subunit isoform (CACNB1) of the L-type channel. Interestingly, the α1-subunit forms the ion-conducting pore of voltage-gated Ca²⁺ channels, while the β subunit is involved in the modulation of gating.³⁰ Receptor-operated Ca²⁺ channels frequently detected in SMC plasmalemma belong to the P2X purinoreceptor family and activate upon binding of extracellular ATP.^{28, 31} Of these, we detected the P2RX7 receptor, expression of which has been previously shown in cerebral arteries (Supplementary Table 2 for references). In addition, emerging evidence suggests relatedness between receptor- and store-operated Ca²⁺ channel types, both being formed by proteins of the transient receptor potential channel (TRPC) family.³² We identified six proteins (TRPC1, 2, 4, 5, 6, and 7) belonging to this family, all of which can form Ca²⁺ permeable cation channels, and except for TRPC2 have been previously shown to be present in cerebral arteries (Supplementary Table 2 for references). Mechanical stretch-induced Ca²⁺ influx is known to occur via opening of nonspecific cation channels in SMC membranes. We detected one such channel, specifically, the transient receptor potential cation channel TRPM4, an inherently mechanosensitive channel³³ shown to be present in cerebral arterial myocytes.

In contrast to smooth muscle contraction, its relaxation is accompanied by a decrease in iCa²⁺ levels to resting values. Decrease in iCa²⁺ is facilitated by (1) reuptake into intracellular stores by Ca²⁺ ATPase pumps (smooth endoplasmic reticulum calcium ATPases, SERCAs) on the sarcoplasmic reticulum membrane and (2) extrusion from the cell by the plasma membrane Ca²⁺ ATPases (PMCAs) and the Na⁺/Ca²⁺ exchanger (Figure 4). We detected three SERCAs (ATP2A1, A2, and A3), one PMCA (ATP2B2), and one Na⁺/Ca²⁺ exchanger (SLC8A1 or NCX) in our arterial data set. To date, only mRNAs of ATP2B2 and SLC8A1 have been reported in cerebral arteries (Supplementary Table 2 for references). The SMC relaxation is further promoted by the activation of multiple arterial smooth muscle potassium (K⁺) channels. These channels provide an opposing hyperpolarizing influence reducing Ca²⁺ channel activity. We detected multiple K⁺ channels in our arterial data set, in particular, large (KCNMA1 or BKCa), intermediate (KCNN4), and small (KCNN2 or SKCa) conductance Ca²⁺-activated K⁺ channels, as well as ATP-sensitive (KATP channel subunits KCNJ8 and 15), and voltage-activated K⁺ channels (KCNA3, A5-7, C2, H5 and H8, Q1-5, V2, AB1 and AB3). The majority of these K⁺ channels have been previously observed in cerebral arteries (Supplementary Table 2 for references). Particularly, the peptide calcitonin gene-related peptide potently dilates brain cerebral arteries through activation of SMC K⁺ channels.³⁴ Each of the endothelium-derived relaxing factors, namely NO, prostacyclin, and endothelium-derived hyperpolarizing factor, can activate one or several types of SMC K⁺ channels.³⁵ We detected three NOSs in our arterial data set, specifically the previously mentioned NOS1 (or nNOS), as well as NOS2 (or iNOS) and NOS3 (or eNOS).

Insights into the Proteomes of Arteries and Microvessels

The recent release of the cerebral microvessel proteomics study⁷ provided an opportunity to compare and contrast protein-based findings in pial/extracerebral CW arteries (this study) and intraparenchymal brain vessels. Vessel diameters in our sampling of mouse cerebral arteries were ∼100 μm and lower, with the majority being large diameter vessels. The wide range of arterial diameters in our sampling is representative of the varied vessel sizes present at the brain surface, starting with the CW and its major arteries ranging from 75 to 100 μm in diameter.³⁶ Furthermore, it has been shown that peripheral branches of the major arteries range between 50 and 70 μm in diameter, while their ramifications coursing over the brain surface, reach 35 μm or lower in diameter, just before perforating as penetrating intracerebral arterioles.^{36, 37} In comparison with our arterial sample, vessel diameters in the microvascular sampling⁷ ranged between >20 μm and <100 μm. Thus, it can be concluded that while the majority of vessels in the microvascular sample tended toward small diameters, some overlap in vessel diameter size exists between the two studies.

Comparatively, the most striking difference between the arterial and microvascular data sets was the ∼9 × increase in the number of unique BBB-specific cell types in the latter. Disparities in unique protein numbers in ‘TJ and adhesion proteins' and ‘membrane transporter and channel proteins' categories, while lesser (∼2x more in the microvascular data set), were still present. Overall, our findings of protein number inequalities in the above-mentioned categories are in line with the knowledge that at the level of cerebral arteries, the BBB is not as developed, nor specialized, as in intracortical microvessels. Incomplete barrier characteristic of cerebral arteries has been shown in smaller diameter surface vessels using endothelial barrier antigen staining.³⁸ Compared with the uniform staining present in intracortical microvessels, those on the cerebral surface showed heterogeneity in endothelial barrier antigen expression.³⁸ Moreover, two populations of interendothelial TJs have been shown in these surface microvessels, (1) those with the BBB-type fusion of adjacent membranes and (2) those with discernible gap between them.³⁹ Unlike the three above-mentioned protein categories, the arterial data set housed higher numbers (∼5 × ) of ‘ECM and basal lamina proteins' compared with intracortical microvessels, probably as a consequence of an adventitia layer present in the former. This layer, largely composed of ECM proteins, is absent in intracortical microvessels.⁴⁰

In summary, we have compiled a novel resource database for proteins present in the arterial wall of mouse CW and its surface ramifications. To ensure detection and identification of an extensive and comprehensive list of CW cerebral arterial proteins, we used fresh unfixed tissue and employed two parallel, custom proteomic approaches. Our database provides an excellent resource for the study of protein expression profile in healthy cerebral arteries and its perturbations in cerebrovascular diseases. Furthermore, we shed light on the difference and the similarities between brain arteries and microvessels at the level of the proteome.

The authors declare no conflict of interest.