Proteomic Profiles of Human Arterioles Isolated from Fresh Adipose Tissue or Following Overnight Storage

Abstract

Arterioles are key determinants of the total peripheral vascular resistance, which, in turn, is a key determinant of arterial blood pressure (BP). However, the amount of protein available from one isolated human arteriole may be less than 5μg, making proteomic analysis challenging. In addition, obtaining human arterioles requires manual dissection of unfrozen clinical specimens. This limits its feasibility, especially for powerful multi-center clinical studies in which clinical specimens need to be shipped overnight to a research lab for arteriole isolation. We performed a study to address low input, test overnight tissue storage, and develop a reference human arteriolar proteomic profile. In tandem mass tag proteomics, use of a booster channel consisting of human induced pluripotent stem cells-derived endothelial and vascular smooth muscle cells (1:5 ratio) increased the number of proteins detected in a human arteriole segment with FDR < 0.01 from 1,051 to more than 3,000. The correlation coefficient of proteomic profile was similar between replicate arterioles isolated freshly, following cold storage, or before and after the cold storage (one-way ANOVA, p=0.60). We built a human arteriolar proteomic profile consisting of 3,832 proteins based on the analysis of 12 arteriole samples from three subjects. Of 1,945 BP-relevant proteins that we curated, 476 (12.5%) were detected in the arteriolar proteome, which was a significant overrepresentation (Chi-squared, p<0.05). These findings demonstrate that proteomic analysis is feasible with arterioles isolated from human adipose tissue following cold overnight storage and provide a reference human arteriolar proteome profile highly valuable for studies of arteriole-related traits.

Introduction

Microvascular dysfunction results in tissue damage and organ dysfunction, which can lead to cardiovascular disease, renal failure, stroke, blindness, hypertension, and dementia^1–4. The resistance arteries, or arterioles, have particular importance as the major determinant of total peripheral vascular resistance, which, in turn, is a major determinant of arterial blood pressure. It is essential to recognize the patterns and dynamic features of arteriolar protein networks that constitute the molecular signature of healthy and diseased arterioles in humans to improve early disease detection and interrupt the disease process before clinical consequences occur^5,6.

Very few studies have reported proteomic profiles of human arterioles. The few that did used more than 25 μg of input protein^7,8. However, the amount of protein available from one isolated human arteriole may be less than 5 μg. The tandem mass tag (TMT) approach allows multiplexed proteomic analysis of samples of limited amounts^9–11. TMT proteomics involves the labeling of peptides with tags unique to each sample. All samples in a mass spectrometry run are combined for the identification of each peptide, and the sample origin and the quantity of a peptide in a sample can be ascertained using the tag. Using “boosting” samples in the multiplexed analysis enables higher sensitivity when the input sample amounts are low¹². Booster samples are biological samples similar to study samples but available in large quantities. Similar to combining several tagged samples, the inclusion of booster samples increases the amount of each peptide in a mass spectrometry run and enhances the sensitivity of detecting and identifying the peptide.

Clinical proteomic research depends on the availability of clinical samples. In the ideal condition, the proteomic analysis will be performed in fresh or fresh-frozen tissue¹³. However, many informative tissues, including arterioles, require manual dissection of unfrozen specimens. Isolation and collection of arterioles from fresh tissue is limited in its feasibility. Technical staff with expertise in arteriole isolation may not be available at clinical specimen collection sites, especially for otherwise powerful multi-center studies. In addition, clinical specimens may become available outside of regular work hours. Therefore, storing clinical specimens for a short duration, up to 24 hours, is an attractive alternative that may substantially improve the feasibility of obtaining human arterioles. MACS tissue storage solution (Miltenyi Biotec, 130-100-008) is designed for the storage of fresh organs and tissue samples for 24 to 48 hours. The MACS tissue solution has been tested on various human tissues including tumors, skin, heart, brain, and skeletal muscle. However, a comprehensive study of the impact of unfrozen storage suitable for subsequent dissection on global proteome stability for human tissues has not been reported.

Building on our experience with proteomic analysis, human arteriole isolation and iPSC derived EC and VSMC differentiation^3,4,14–19, we performed a study with three technical and biological goals. The first was to develop a sample for the booster channel and test whether it would improve the identification of proteins from a human arteriole sample. The second was to examine the effect of cold storage of unfrozen human adipose specimens in MACS tissue solution on the proteomic profiles of isolated arterioles. The third was to identify the characteristics of the human arteriolar proteome.

Experimental procedures

Human specimen processing and arteriole Isolation

All procedures were approved by the Institutional Review Boards at the Medical College of Wisconsin with patient consent.

Experimental Design and Statistical Rationale

The workflow is illustrated in Figure 1. The proteome of the samples was studied by Data Dependent Acquisition (DDA-MS) using Booster Channel and TMT based labeling. The reproducibility of DDA-MS was measured using 2 technical replicates for test TMTzero sample and three replicate for TMTPro 16 plex samples. Pearson’s correlation and Spearman’s correlation were used to evaluate the repeatability of protein quantification. The analysis of DDA-MS is based on false discovery rate (FDR) value < 0.01 as the significant threshold. Protein annotations were analyzed using Swissport human with isoforms, 2019-05-01, MaxQuant Contaminants. Proteomic ruler was applied to determine protein copy number per cell²⁰ using default parameters. Gene ontology analysis was performed using Panther Classification system (v17.0) (http://pantherdb.org/).

Figure 1 –

Study schematic for human arteriolar proteome analysis.

Arterioles isolation

Arterioles, approximately 100–150 μm in diameter, were isolated from subcutaneous adipose tissues discarded from surgical procedures as described³. Upon arrival in our laboratory, adipose tissue samples were divided into two aliquots: one was used for the same day arteriole isolation, and the other was stored in MACS tissue storage buffer at 4°C for 24 hours prior to arteriole isolation. Perivascular fat was removed during vessel isolation. Following isolation, vessels were stored in liquid nitrogen.

Differentiation of human induced pluripotent stem cells (iPSC)

The human iPSC line 039B was reprogrammed from urine cells obtained from a 35-year-old male Caucasian using Sendai virus following the protocol described previously²¹. iPSCs were differentiated into endothelial cells (EC) and vascular smooth muscle cells (VSMC) following previously published protocols²². In brief, iPSCs were grown in Matrigel coated 6-cm dishes with mTeSR^™ plus (STEMCELL Technologies) and routinely passaged at a dilution of 1:3 to 1:5. iPSCs were dissociated using Accutase (STEMCELL Technologies) and plated on 6-well Matrigel plates at a density of 47,000 cells/cm² in mTeSR^™ plus with 10 μM Rock inhibitor Y-27632 (STEMCELL Technologies). After 24 hours, cells were treated with N2B27 medium (a 1:1 mixture of DMEM:F12 with Glutamax and Neurobasal media supplemented with N2 supplement and B27 supplement minus vitamin A; all Life Technologies) plus 8 μM CHIR99021 (Selleck Chemicals) and 25 ng/ml BMP4 (PeproTech) for 3 days to generate mesoderm cells. After induction for 2 days with StemPro-34 SFM medium (STEMCELL Technologies) with 200 ng/ml VEGF (PeproTech) and 2 μM forskolin (Abcam), ECs were purified with CD144 magnetic beads (Miltenyi Biotec). CD144-positive ECs were cultured in StemPro-34 SFM medium supplemented with 50 ng/ml VEGF for 5 days before cell harvest. For VSMC induction, mesoderm cells were treated with N2B27 medium supplemented with 10 ng/ml PDGF-BB (PeproTech) and 2 ng/ml Activin A (PeproTech) for 2 days. Induction of contractile VSMCs were then achieved with N2B27 media supplemented with 2 ng/ml Activin A and 2 μg/ml Heparin (STEMCELL Technologies). CD144 magnetic beads were used to remove CD144+ cells from VSMCs. For mass spectrometry analysis (MS), iPSC derived EC and VSMC were mixed 1:5 to create a sample for the booster channel for the analysis of arteriolar proteomes as an arteriole consists of one layer of endothelial cells and 1–3 layers of smooth muscle cells and one smooth muscle layer may contain more proteins than the endothelial layer.

Sample digestion for mass spectrometry

The samples were transferred to 1.5 mL microcentrifuge tubes in two washes of 100 mM ammonium bicarbonate, 40% Invitrosol, and 20% acetonitrile. Using an ice bath, samples were sonicated for 20 cycles of 10 seconds on, 30 seconds off. We reduced cysteine in 5 mM TCEP for 30 minutes at 37°C and alkylated it with 10 mM iodoacetamide for 30 minutes at 37°C, then digested it with 10 μg of trypsin overnight at 37°C. After cleanup using the PreOmics Phoenix kit according to the manufacturer’s instructions, dried samples were redissolved in 100 mM TEAB, their concentrations measured using the Pierce Quantitative Fluorometric Peptide Assay and 4 μg of each (20 μg of the booster channel sample) were taken for labeling with Thermo TMTzero and TMTPro 16 plex reagents (ThermoFisher Scientific). The labeled samples were dried and fractionated using a Pierce High pH Reversed-Phase Peptide Fractionation Kit (ThermoFisher Scientific), using one water and one 10% acetonitrile wash, then eluted in 10, 12.5, 15, 17.5, 20, 22.5, 25, and 50% acetonitrile/TEA steps. The fractions were concatenated by combining fractions 1+5, 2+6, 3+7, and 4+8 before drying. The samples were dissolved in 135 μl of 2% acetonitrile 0.1% formic acid and, after test runs on diluted aliquots, were further diluted 2.5-fold for replicate 20 μl full-loop HPLC injections.

Mass Spectrometry Analysis

The samples were analyzed on a Thermo Scientific Orbitrap Fusion Lumos MS using technical replicate injections. For the test TMTzero sample, injections F1 and F2 represented 1+5 pooled fractions, injections F3 and F4 for 2+6 pooled fractions, injections F5 and F6 for 3+7 pooled fractions and injections F7 and F8 for 4+8 pooled fractions. For the TMTPro 16 plex samples, injections F2 and F7 represented 1+5 pooled fractions, injections F4, F6 and F11 for 2+6 pooled fractions, injections F5, F8 and F12 for 3+7 pooled fractions, and injections F3, F9 and F13 for 4+8 pooled fractions. The settings of the data-dependent acquisition (DDA) HCD MS2 instrument method were shown in Supplementary table S1a. Briefly, the parameters were set as 2.5–9% solvent B (0.1% formic acid in 80% acetonitrile) for 3 min, 9–37% for 76 min, 37–60% for 10min, and increased to 99% for 4 min all at a constant flow rate of 300 nl/min. Electrospray voltage was set at 2.1 kV, and peptides were detected with MS range 350–1500 m/z using an Orbitrap with 120,000 @ 200 m/z resolution. The MS/MS fixed first mass was set at 100 m/z with 50,000 @ 200 m/z resolutions.

Data analysis

Proteome Discoverer 2.4 (Thermo Scientific) platform was used to analyze MS data, the settings of which were shown in Supplementary table S1b. Briefly, the secondary mass spectrum data were searched against a FASTA file containing recombinant sequences and Swissprot human with isoforms, 2019-05-01, MaxQuant Contaminants. The search parameters were configured as follows: trypsin (semi) enzyme digestion with a maximum of two missed cleavages allowed and minimum peptide length of 6 amino acids; Carbamidomethyl (C) and TMTPro (K and peptide N-term) static modifications; Oxidation (M), and Acetylation (protein N-terminus) dynamic modifications. The peptide precursor mass tolerance was set to 10 ppm, with the fragment mass tolerance set at 0.02 Da. The TMTpro-16plex method was used for quantification using reporter ions quantifier with most confident centroid 20 ppm tolerance. Protein, peptide, and PSM identification were controlled at 1% false discovery rate (FDR). A matrix containing peptide abundances was created out of the raw LC-MS datasets for quantitative analysis. Normalization was performed using none and total peptide amount. For hypothesis testing, Anova-individual protein was used. The identifications of proteins were filtered to include only those identified by one or more unique peptides.

The Reporter Ions Quantifier node and the Precursor Ions Quantifier node classify which peptide groups are used for protein quantification. We applied Unique + Razor method to identify protein abundances. Razor peptides are shared among multiple protein groups or proteins. Unique + Razor uses both unique peptide groups and peptide groups containing razor peptides for the best associated master protein. The best master protein is the protein with the largest value in the “Protein Unique Peptides column” on the Proteins page and with the smallest value in the “Coverage column (the longest protein)”. Therefore, abundance score was assigned to the master proteins only. The calculated protein abundance was a simple summation of its associated and used peptide group abundances.

Proteomic ruler was applied to determine protein copy number per cell²⁰ using default parameters. The proteomic ruler algorithm scales intensity for a protein using its sequence length and molecular weight to estimate copy number per cell. Scaling copy number was done using histone proteomic ruler where cumulative histone amount was considered proportional to the expected DNA amount per cell and histones were used as reference to estimate copy numbers of individual proteins per cell. Alternatively, total protein amount per cell was used where total intensity will be considered proportional to the total protein amount per cell (default 200pg).

Panther Classification system (v17.0) was used to annotate identified proteins into various protein classes.

Statistical analyses were performed using GraphPad Prism 9. Specific statistical tests were described in the Results section.

Results

Human arteriole proteomes are not affected by overnight cold storage of adipose tissue.

We reasoned that, if cold storage (4°C for 24 hours) of human adipose tissue affected the arteriole proteome, the consistency of proteomic profiles between arterioles isolated from fresh adipose tissue and following the cold storage would be lower than that between multiple arterioles isolated from fresh adipose tissue. To test this, adipose tissue was obtained from three human subjects following surgery as described in the study schematic (Figure 1). The adipose tissue from each subject was divided into two aliquots. Two arterioles were isolated from one aliquot of adipose tissue from each subject on the same day of surgery. These 6 arterioles were labeled as Fresh1A, Fresh1B, Fresh2A, Fresh2B, Fresh3A and Fresh3B. The numbers referred to the subjects, and A and B referred to the two arterioles from each subject. The other aliquot of adipose tissue was stored in the MACS buffer at 4°C. Two arterioles were isolated 24 hours later from the stored adipose tissue from each subject and labeled as Stored1A, Stored1B, Stored2A, Stored2B, Stored3A and Stored3B). Total protein was extracted from each of the 12 arterioles using an ice bath sonication, yielding between 4.3 μg – 8.2 μg of protein (Table 1).

Table 1 -.

Total isolated proteins and number of detected proteins in the arteriole proteome from each sample.

Ion Channel	Nomenclature used in Manuscript	Total isolated proteins (μg)	Number of detected Protein having abundance score
130N	Stored1A	6.6	3416
132C	Stored1B	8.0	3430
127N	Fresh1A	6.0	3078
129C	Fresh1B	8.2	3361
130C	Stored2A	4.7	3404
132N	Stored2B	4.6	3427
127C	Fresh2A	4.3	3360
129N	Fresh2B	7.6	3434
131N	Stored3A	7.1	3340
131C	Stored3B	8.1	3411
128N	Fresh3A	6.6	3326
128C	Fresh3B	7.2	3331
126	Pool		3370
134N	Booster	345.8	3436

The TMT label binding efficiency and technical reproducibility of mass spectrometry (MS) were tested using a test sample. A 4.0 μg total protein isolated from a small arteriole was labeled with TMTzero and high pH reverse-phase peptide fractionation was performed as described in the methods. The test sample was run twice, and 1051 proteins were detected by MS with FDR ≤ 0.01 (Supplementary Table S2). The Pearson correlations of log-transformed protein abundance were high (r ≥ 0.98) between the two replicate runs, indicating high technical reproducibility of the MS analysis (Supplementary Figure S1a–f).

We used a mixture of iPSC derived EC and VSMC in a 1:5 ratio as a booster sample. In conjunction with a booster channel, the multiplexed Thermo TMTPro 16 plex MS analysis of the 12 arteriole samples detected 3836 proteins with an FDR ≤ 0.01 and a distinct peptide threshold of 1, whereas 3012 proteins would be detected if minimum detected peptide threshold was increased to 2 peptides (Supplementary Table S3). Among the 3836 proteins, 3819 proteins were unique, while 17 proteins were isoforms. A total of 3436 proteins out of 3836 could be assigned an abundance score. The number of proteins with an abundance score detected in the 12 arteriole samples ranged from 3078 to 3434 (Table 1 and Supplementary Figure S2a). No significant difference was observed for the total isolated protein amount (unpaired t test, p=0.11) and the number of detected proteins based on abundance score (unpaired t test, p=0.88) between same-day collected samples and next-day collected samples.

Next, we examined the correlation of log normalized protein abundance between replicate arterioles isolated freshly from the same subject (Fresh1A vs Fresh1B, Fresh2A vs Fresh2B, Fresh3A vs Fresh3B) or isolated after 24 hours of cold storage (Stored1A vs Stored1B, Stored2A vs Stored2B, Stored3A vs Stored3B), or between arterioles isolated from the same subject before and after the cold storage (Fresh1A vs Stored1A, Fresh1B vs Stored1B, Fresh2A vs Stored2A and so on). The Pearson correlation (Figure 2a and Supplementary Figure S2b–m) was not significantly different between the three sets of comparisons (one-way ANOVA, p=0.60). For example, the correlation was 0.85 for Fresh3A and Fresh3B, 0.82 for Stored3A and Stored3B, 0.81 for Fresh3A and Stored3A, and 0.86 for Fresh3B and Stored3B. In addition, the correlation between replicate arterioles from the same subject was similar to that between arterioles from different subjects (Figure 2b). One of the freshly collected samples, Fresh1A, appeared to be an outlier compared with all other arterioles analyzed (Figure 2b). A similar correlation pattern was observed for Spearman correlation (Supplementary Figure 3a–b).

Figure 2 -. Pearson correlation among 12 samples.

(a) illustrates the Pearson correlation between proteins abundances in the arterioles isolated form human subcutaneous adipose tissues; same-day (Fresh) isolated arterioles, next-day (24-hour storage at 4°C, Stored) isolated arterioles, and between same day and next day isolated arterioles. A and B are two isolated arterioles at the same time and in the same condition. One-way anova test shows nonsignificant difference among same day, next day, and between same day and next day isolated arterioles (p=0.5984) (b) show Pearson correlation matrix among all 12 samples.

Characteristics of the human arteriolar proteome

A PANTHER term analysis revealed that most of the 3819 unique proteins detected in human arterioles were classified as metabolite interconversion enzymes (553 proteins), protein modifying enzymes (371), RNA metabolism enzymes (351), translational proteins (225), cytoskeleton proteins (220), protein binding activity modulators (180), scaffold proteins (156) and membrane traffic proteins (150) (Figure 3a). Other proteins with potential regulatory functions were also detected, examples of which included chromatin binding or regulatory proteins (94), transcriptional regulators (79), cell adhesion molecules (34), calcium-binding proteins (31), transmembrane signal receptors (22), and intercellular signal molecules (11). In addition, 886 proteins were not classified into a specific group (Figure 3a). Several hemoglobin subunit proteins were detected, which were removed from the subsequent analysis of the arteriolar proteome as hemoglobin came from residual blood rather than the arteriolar proteome.

Figure 3 -. Characteristics of the human arteriole proteome.

(a) Panther based protein classification resulted in characterization of 3794 out of 3819 detected unique arteriole’s protein in 12 samples (b) Arteriolar proteome copy number percentage and protein copy number for top 10 detected proteins (constitute 7.62% of total detected arteriolar proteome) (c) protein copy number and proteome copy number percentage for selected endothelial cell and smooth muscle cell markers. The copy number calculation was performed using a method reported by Wisniewski et al 2014²⁰.

Next, we applied ‘proteomic ruler’ algorithm to the dataset to determine protein copy numbers per cell. Since sample Fresh1A appeared to be an outlier in the correlation analysis, it was not included in the protein copy number analysis. As only 3436 out of 3836 detected proteins were assigned abundance scores, and four hemoglobin proteins were removed, protein copy number was calculated for 3432 proteins. Two methods, histone proteomic ruler and total protein amount per cell, resulted in similar protein ranking although variation was detected in copy number estimation (Supplementary Table S4a–b). The average protein had a mean protein copy number per cell of 2.76×10⁶ and a median value of 1.36×10⁶ protein copies per cell based on histone proteomic ruler. The top 10, 20, 50 and 100 proteins make up 7.62%, 11.16%, 18.30% and 26.08%, respectively, of the total protein abundance based on copy number per cell (Supplementary Table S4c). Smooth muscle marker gene Transgelin (TAGLN, 1.36% of the detected proteome), trypsin-1 precursor (PRSS1, 1.34%), Cysteine-rich protein 1, a intracellular zinc transport protein (CRIP1, 0.84%), fat cell development protein Adipogenesis regulatory factor (ADIRF, 0.71%), Histone H4 (HIST1H4A, 0.70%), motility proteins, Protein S100-A4 (S100A4, 0.69%) and Actin cytoplasmic 1 (ACTB, 0.65%) appear among the top 10 proteins (Figure 3b). Further, PANTHER based protein classification for the top 100 proteins revealed that most of the protein belong to metabolite interconversion enzymes (19), cytoskeletal proteins (17), transfer/carrier proteins (6) and calcium binding and chromatin binding/regulator classes (6) (Supplementary Figure S4).

Next, we looked for several marker genes for endothelial cells including CD34, Cadherin 5 (CDH5), Von Willebrand Factor (vWF), Platelet and Endothelial Cell Adhesion Molecule 1 (PECAM1), Intercellular Adhesion Molecule 1 (ICAM1), Tyrosine Kinase With Immunoglobulin Like and EGF Like Domains 1 (TIE1), Endothelin 1 (EDN1), Kinase Insert Domain Receptor (KDR), Lymphatic Vessel Endothelial Hyaluronan Receptor 1 (LYVE1) and Angiopoietin 2 (ANGPT2), and marker genes for smooth muscle cells including Actin Alpha 2, Smooth Muscle (ACTA2), Myosin Heavy Chain 11 (MYH11), Transgelin (TAGLN), Calponin 1 (CNN1), Smoothelin (SMTN), Melanoma Cell Adhesion Molecule (MCAM), Fibulin 5 (FBLN5), Periostin (POSTN), Elastin (ELN) and Caldesmon 1 (CALD1) in the arteriolar proteome. Smooth muscle cell markers TAGLN, CNN1, ACTA2, and FBLN5 had copy numbers greater than the mean protein copy number per cell in the arteriole (Figure 3c, Supplementary Table S3 and S4). PECAM1, vWF, CD34, KDR, MYH11, ELN, MCAM, and POSTN were detected at lower copy numbers. The other endothelial or smooth muscle cell markers that we looked for were not detected.

Blood pressure-relevant proteins are overrepresented in the human arteriole proteome.

Arterioles play a central role in determining total peripheral vascular resistance, which, in turn, is a major determinant of blood pressure (BP). We recently reported the curation of 251 BP physiology genes based on Gene Ontology, 136 genes with nonsynonymous single nucleotide polymorphisms (SNPs) associated with BP based on genome-wide association studies (GWAS), and 3801 genes that were expression quantitative trait locus (eQTL) genes for BP-associated SNPs²³. Excluding genes without corresponding human proteins, these genes corresponded to 220, 132, and 1708 proteins with unique Uniprot IDs, respectively. Some of these proteins overlapped, resulting in a total of 1945 proteins.

We examined the representation of these BP-relevant proteins in the human arteriole proteome. Of the 1945 proteins encoded by BP-relevant genes, 476 were detected in the arteriole proteome (Figure 4a). In other words, 12.5% of the detected arteriolar proteome were BP-relevant proteins. As the 1945 proteins represented 9.5% of all human proteins with unique Uniprot IDs, BP-relevant proteins were significantly overrepresented in the arteriole proteome (p<0.05, Chi-squared test) (Supplementary table 5a). The 476 BP-relevant proteins detected in the arteriole proteome included 38 encoded by BP physiology genes, 36 by genes with nonsynonymous SNPs associated with BP, and 433 by eQTL genes for SNPs associated with BP (Figure 4b–d). The proteins encoded by genes with nonsynonymous SNPs associated with BP or eQTL genes for SNPs associated with BP were significantly overrepresented in the arteriole proteome (p<0.05, Chi-squared test).

Figure 4 -. Blood pressure (BP)-relevant proteins detected in arteriolar proteome.

(a) total BP-relevant proteins (b) BP-physiology proteins (c) nonsynonymous SNPs associated with BP (d) SNPs with BP-associated eQTLs (e) An analysis of the percentage copy number distribution of BP-relevant proteins in the total detected arterioles proteome and the panther-based classification of those proteins. The copy number calculation was performed using a method reported by Wisniewski et al 2014²⁰.

Lastly, we examined the characteristics of the 476 proteins encoded by BP-relevant genes and detected in the arteriole proteome. We were able to estimate copy numbers for 425 of these proteins, which, combined, contributed 12.54% of the arteriolar proteome, with 243 having copy numbers greater than 1×10⁶ per cell (Supplementary Table S5b). Panther based protein class distribution analysis showed that the majority of the 476 proteins belonged to metabolite conversion enzymes (74), protein modifying enzymes (56), cytoskeleton proteins (34), RNA metabolism (30) and protein binding activity modulators (25) (Figure 4e). Other proteins with potential regulatory functions were also included, such as transcriptional regulators, cell adhesion molecules, signaling molecules and receptors, and calcium-binding proteins.

Discussion

The results of this study show that broadly available human adipose tissues may be stored for 24 hours at cold storage before arterioles are isolated for proteomics analysis. The findings greatly increase the feasibility of obtaining human arterioles for proteomic analysis. Adipose tissues from surgeries performed outside of normal business hours can be stored for next-day isolation of arterioles. Additionally, Adipose tissues obtained from multiple clinical centers can be shipped overnight to one location for arteriole isolation.

The study shows that more than 3,000 proteins can be identified and quantified from 4 ug of proteins in one short segment of arteriole obtained from human adipose tissues. Our number of detected proteins is comparable with recent studies where >30 μg protein per human arteriole sample was used as input^7,8. We achieved this using multiplexed TMT proteomics that incorporated a booster channel sample consisting of iPSC-derived EC and VSMC in 1:5 ratio. Just over 1000 proteins were detected in a test sample without using the booster channel. It is possible that the TMT label in conjunction with booster channel can increase the ionization efficiency of peptides and thereby increase their chance of detection^12,24,25.

The proteomic ruler makes several assumptions that allow the omission of any spike-in standards and eliminate several experimental steps such as cell counting and absolute protein concentration determination, which are themselves prone to errors²⁰. Using the proteomic ruler, we determined that the arteriolar proteome consists of proteins with 10⁴–10⁸ protein copies per cell. Several known markers of endothelial and smooth muscle cells are present in the detected arteriole proteome, although some of them have low copy numbers per cell. Many of the most abundant proteins in the human arteriole are involved in cellular metabolism and structure.

The detected arteriolar proteome is significantly enriched for BP-relevant proteins. This is consistent with the key role of the arteriole in the regulation of BP. Human GWAS has suggested the potential importance of hundreds of genes in the BP regulation, many of which have not been previously shown to regulate BP²³. These genes may contain nonsynonymous SNPs associated with BP or are eQTL genes for BP-associated SNPs. The significant enrichment of proteins encoded by these genes in the human arteriolar proteome further supports the potential relevance of these genes to the BP regulation. In addition, these findings provide candidate proteins that are expressed in human arterioles and may mediate the effect of BP-associated SNPs on BP.

The current study has generated one of the most comprehensive proteomic profiles for human arterioles, which may serve as a powerful reference dataset for numerous studies of the arteriole. However, the detected proteome likely represents a fraction of the entire arteriolar proteome that includes more abundant proteins. Future studies should aim for establishing a more complete human arteriolar proteome and gaining new insights into the physiological and disease relevance of the arteriole by examining changes of the human arteriolar proteome in response to stimuli or disease conditions. Likewise, our conclusion that overnight cold storage of human adipose tissues does not alter the proteomic profile of human arterioles was based on the detected fraction of the arteriolar proteome. It remains to be investigated whether the storage affects the remaining proteome.

Nonstandard Abbreviations and Acronyms

BP

Blood Pressure

MS

Mass spectrometry

TMT

Tandem mass tag

iPSC

human induced pluripotent stem cells

EC

Endothelial cells

VSMC

Vascular smooth muscle cells

SNPs

Nonsynonymous single nucleotide polymorphisms

eQTL

Expression quantitative trait loci

Data availability

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE²⁶ partner repository with the dataset identifier PXD044454 and 10.6019/PXD044454”. Reviewer account details to access dataset are: Username: reviewer_pxd044454@ebi.ac.uk; Password: keGagbBN