|Home | About | Journals | Submit | Contact Us | Français|
An assay proposed to quantify endothelial progenitor cell (EPC) colonies in humans was investigated to determine the phenotype of recovered cells and their relevance to in vivo endothelial function.
Twelve sedentary subjects participating in a worksite wellness program underwent endothelial flow-mediated dilation (FMD) testing of the brachial artery and blood sampling for EPC colony assay. Microarray-based genotypic characterization of colonies showed surface markers consistent with T lymphocyte phenotype, but not with an EPC (CD34, CD133, VEGFR-2) or endothelial (CD146) phenotype. Gene expression patterns more closely matched T lymphocytes (r=0.87) than endothelial cells (r=0.66) in our microarray database. Flow cytometry of colonies confirmed large populations of CD3+CD45+ T cells (>75%) and few CD146+CD45− endothelial cells (<1%). Further, there was no correlation between colony number and the magnitude of FMD (r= −0.1512, P=0.6389). After exercise training, subjects improved FMD, from 6.7±2.0 to 8.7±1.9% (P=0.0043). Colonies also increased (P=0.0210), but without relation to FMD (r=0.1074, P=0.7396). T lymphocyte phenotype persisted after exercise (r=0.87).
Cells in a commonly used EPC colony assay have a gene expression and cell surface marker profile consistent with a predominance of T lymphocytes and have an unclear relevance to endothelial function, either before or after exercise training.
Endothelial progenitor cells (EPCs) were first described in 1997 as a novel lineage of bone marrow–derived hematopoietic progenitor cells, initially defined by CD34 and subsequently by VEGFR-2 and CD133 cell surface markers, which showed endothelial cell characteristics when expanded in culture on matrix protein.1–4 Subsequent studies in animal models showed homing of cells to arterial injury and tissue ischemia, resulting in endothelial repair or neovascularization. 5–8 However, the exact phenotype(s) of EPCs remains a matter of dispute. In 2003, Hill et al9 used an assay with peripheral blood mononuclear cells cultured in fibronectin-coated wells to quantify EPCs in men with a spectrum of cardiovascular risk and endothelial function. The number of colonies, designated endothelial progenitor cell-colony forming units (EPC-CFUs), in this study was inversely related to cardiovascular risk and positively associated with endothelial function, as measured by brachial-artery flow-mediated dilation (FMD), which they proposed supported an important role of EPCs quantified in this assay in maintaining normal endothelial function.
Phenotypic characterization of cells in EPC-CFU assays has commonly been accomplished by microscopy, including spindle-shaped appearance of cells, fluorescence staining (eg, DiI-acetylated low-density lipoprotein [LDL] and ulexlectin), and immunostaining (eg, CD31; platelet-endothelial cell adhesion molecule).1–4 Using flow cytometry or immunostaining, however, other investigators have reported the presence of cell surface markers for monocytes, macrophages, and lymphocytes in their culture assays, with variable evidence for endothelial outgrowth.10–15 Markers for cell components and surface antigens, however, are shared by many cells and may not define unique population phenotypes.16,17
Given the uncertainty in defining phenotype using cellular markers, we performed gene expression profiling on EPC-CFUs cultured using a commercial assay based on that used by Hill et al.9 We then repeated gene expression profiling of EPC-CFUs after exercise—which has been reported to independently increase EPCs and to improve endothelial function18– 22—for 3 months in a worksite wellness program. Our purpose was to determine whether phenotypic characterization of these CFUs and changes in their gene expression patterns would provide insight into endothelial function and EPC differentiation capacity, before and after exercise training.
This study was conducted at the Clinical Center of the National Institutes of Health in employees who reported no previous exercise training and provided written informed consent for participation in this protocol, approved by the institutional review board of the National, Heart, Lung, and Blood Institute.22 Brachial artery flow-mediated dilation (FMD) testing was performed by an experienced technician who also performed the testing in Hill et al.9 Peripheral blood mononuclear cells were cultured for the EPC-CFU assay (Stem Cell Technologies EndoCult Liquid Medium Kit) according to the manufacturer’s instructions. All testing was performed 48 hours after exercise to avoid acute effects of exercise on EPC mobilization and endothelial function.
RNA was extracted from EPC-CFUs with an RNAqueous isolation kit (Ambion) according to the manufacturer’s directions. T7-based RNA amplification was performed on 10 ng of the isolated total RNA with the Riboamp OA 2-round amplification kit (Arcturus) and subjected to biotin labeling with Affymetrix’s IVT labeling kit according to the manufacturer’s directions (Affymetrix). Fragmented cRNA was hybridized to Affymetrix Human Genome (HG) U133 Plus 2.0 chips, washed, stained on an Affymetrix fluidics station, and scanned with Affymetrix GeneChip scanner.
Affymetrix GeneChip operating software version 1.4 was used to calculate signal intensity and the percent present calls on the hybridized chip. The signal-intensity values obtained for probe sets were transformed with an adaptive variance-stabilizing, quantile-normalizing transformation.23 JMP statistical software package 7.0 (www.jmp.com, SAS Institute) was used in microarray data analysis. Gene ontology analysis and functional annotation clustering was performed with DAVID Bioinformatic Resources 2008 (http://david.abcc.ncifcrf.gov/). Microarray data are available at http://www.ncbi.nlm.nih.gov/geo/, accession number GSE12155.
Flow sorted CD3+CD45+CD146− resting T lymphocytes (n=3), flow sorted CD3+CD45+CD146+ activated T lymphocytes (n=5), cultured human umbilical vein endothelial cells (HUVECs, n=5), and cultured human microvascular endothelial cells (HMVECs, n=5) were selected from our microarray database as representative cell types for comparison because of similarities in RNA isolation and amplification.24 A set of T lymphocyte specific genes identified by Palmer et al was used for comparative analysis.25 Gene expression profiles for these 388 probe sets was compared between the amplified transcriptomes.
EPC-CFUs from 4 healthy subjects were analyzed by flow cytometry using the Cyan flow cytometer (Dako) and FCSExpress (De Novo Software). T lymphocytes were identified by monoclonal antibodies to CD45 and CD3; endothelial cells were identified as CD146+CD45−.
Clinical data are reported as mean value ± SD. Two-tailed paired t tests and Wilcoxon signed-rank tests were used in computing probability values for clinical data. All methods are described in greater detail in the supplemental materials (available online at http://atvb.ahajournals.org).
Twelve subjects (Table) completed the program and had RNA of sufficient quantity and quality after amplification for microarray analysis. Seven of the subjects were female, and the average age was 50 years (range 32 to 62 years). Seven subjects were on drug treatment for hypertension (4) or hypercholesterolemia (3) for at least 2 months before the study, with normal test values at baseline and no change in treatment during the exercise program. Exercise at the work-site averaged 89±37 minutes per week.
To identify the phenotype of cells in the EPC-CFU assay, we examined transcript data for cell identification markers. Several highly expressed transcripts corresponded to markers consistent with lymphocytes, including CD2, CD3, CD20, CD25, CD38, CD45, CD69, and CD71 (Figure 1A). The traditional endothelial marker platelet-endothelial cell adhesion molecule (CD31) was highly expressed, yet other endothelial markers (CD105, CD144, CD146, and von Willebrand factor [vWF]) were not expressed above median values. EPC markers CD34, CD133, and VEGFR-2, as well as the monocyte/macrophage markers CD14 and CD115, were expressed below median values. Inspection of highly expressed transcripts of cell surface markers after 3 months of exercise training showed near-identical patterns compared with baseline transcripts (Figure 1B).
To confirm the phenotype of cells in the EPC-CFU assay we performed comparative analysis on transcriptional patterns from EPC-CFUs at baseline with other cells of known phenotype in our database. Preliminary gene expression and principal component clustering analysis suggested a lymphocyte phenotype for EPC-CFUs. We performed regression analysis using a list of 159 T lymphocyte specific genes25 (388 probe sets) on Affymetrix GeneChip HG-U133 Plus 2.0 arrays to compare gene expression patterns of cells from the EPC-CFUs to those of T lymphocytes and endothelial cells. From this analysis, cells from the EPC-CFU assay display a high degree of similarity with CD3+CD45+CD146− resting T lymphocytes (Figure 2A) but a low degree of similarity with HUVECs (Figure 2B). We also performed analysis against CD3+CD45+CD146− activated T lymphocytes (r=0.87) and against HMVECs (r=0.62), with similar results.
Regression analysis of gene expression patterns from EPC-CFUs after exercise training showed persistence of a strong relation with resting T lymphocytes (Figure 2C) and a weak relation with endothelial cells (Figure 2D). These data were virtually identical to transcript data at baseline, suggesting that cells in culture after repetitive exercise were unchanged in phenotype.
To determine whether cultured cells were a mixture of T lymphocytes and endothelial cells, flow cytometric analysis of EPC-CFUs from 4 healthy subjects was performed using markers for T lymphocytes (CD3+CD45+) and endothelial cells (CD146+CD45−). In all 4 samples, large populations (>75%) of CD3+CD45+ T lymphocytes were present, with few CD146+CD45− endothelial cells (<1%) apparent in samples from any subject (Figure 3).
FMD at baseline averaged 6.7±2.0% and was not associated with EPC-CFU number (r=−0.1512, P=0.6385). Exercise training for 3 months had no effect on weight, vital signs, or cholesterol levels (Table). Treadmill exercise duration (standard Bruce protocol) increased from 531±71 seconds to 601±95 seconds (P=0.0005). Brachial artery FMD improved to 8.7±1.9% (P=0.0043). The number of EPC-CFUs also increased with exercise, from 44±37 to 60±52 colonies per well, minimum 4 wells counted (P=0.0210), although there was no correlation between increase in colonies and improvement in FMD (r=0.1074, P=0.7396).
To identify genes with potential relevance to changes in colony number and endothelial function after exercise training, we selected probe sets either upregulated or downregulated in EPC-CFUs after 3 months of exercise (P<0.001, fold change >1.2). The cohort of 12 participants exhibited a significant change in 29 probesets representing 25 unique genes. Examination of differentially regulated genes from this analysis revealed 5 upregulated genes and 20 downregulated genes (supplemental Table I). Gene ontology analysis of these modulated genes identified them to be involved in functions such as signal transduction (C14orf100, CREB1, GNAI3, RAP2C, RIPK3, TBC1D14), metabolic processes (ADIPOR1, ADSL, CRLS1, NDUFA10), immune response (IGKC, IGL, PRELID1), translation (EIF4E2, MRPL11, MRPL42), gene expression and transcription (CREB1, E2F6, PRRX1), and transport (GNAI3, NUP160, TMEM1).
Colonies of cells grown in fibronectin culture assays from peripheral blood mononuclear cells— believed to represent bone marrow–derived EPCs—have been associated with endothelial function in men,9 reduced in patients with coronary artery disease,21 and inversely associated with cardiovascular outcomes.26 However, markers of endothelial phenotype used by several groups to validate these assays— including CD31, CD105, CD144, lectin binding, and DiIacetyl LDL uptake—are shared by many subsets of mononuclear cells.11–17,27–29 Because of the uncertainty involved in phenotypic characterization using cellular markers, we considered whether genotypic characterization could serve as a more exact determinant of cellular phenotype. Our data suggest that cells derived from a commercially available EPC-CFU assay, based on the assay used by Hill et al,9 are primarily T lymphocytes by cell surface marker expression, by flow cytometry, and by transcriptome regression analysis with T lymphocytes. Although gene expression patterns correlated weakly with those of endothelial cells in our database, few endothelial cells (CD146+CD45−) were observed by flow cytometry.
Marginal expression of the transcript for CD146, a surface marker traditionally present on endothelial cells, was present in our microarray analysis. Flow cytometry of EPC-CFUs showed a small number of CD146+ cells, but the majority coexpressed CD3, suggesting that these cells are T lymphocytes and not endothelial cells. Coexpression of CD146 on lymphocytes is representative of an activated lymphocyte phenotype with increased capacity to bind to endothelial cells and extravasate to sites of inflammation.24,30,31 Blood samples from healthy subjects in this study showed that approximately 2% of CD3+ cells in circulation were also CD146+. Accordingly, cells from our assay likely represent an expansion of this unique population.
Our findings are consistent with reports from Hur et al13 and Rohde et al,15 who reported that colonies from mononuclear cells plated on fibronectin are predominately lymphocytes. Rohde et al15 used the same commercial assay as was used in our study and demonstrated predominance of T cells and monocytes in their colonies by flow cytometric determination of cell surface markers. Of interest, inclusion of both CD2+ lymphocytes and CD14+ monocytes in their starting population of blood mononuclear cells in culture was necessary for colony formation to be observed after five days of culture. Our data extends these findings by using a microarray approach to determining the phenotype of cells in this assay, demonstrating expression of genes encoding surface markers consistent with T lymphocyte phenotype, but not with an EPC (CD34, CD133, VEGFR-2) or endothelial (CD146) phenotype. Gene expression patterns closely matched T lymphocytes in our microarray database. Flow cytometric analysis of cells from colonies was also consistent with T lymphocytes, with few endothelial cells present after 5 days of plating on fibronectin-coated plates. We also investigated whether colonies in this assay may have relevance in subjects participating in a worksite wellness program, the majority of whom had risk factors for impaired endothelial function. Contrary to our original hypothesis and to the report of Hill et al,9 we found no correlations between EPC-CFUs and FMD at baseline. Several groups have suggested that cells in colonies, despite limited potential of endothelial differentiation, may have paracrine effects on endothelium via secretion of cytokines and growth factors. In this regard, our findings are consistent in part with data of Hur et al,13 who also reported a predominance of CD3+CD31+ cells in their colony assay. They claimed that these cells represent a unique population of T lymphocytes capable of secreting angiogenic cytokines and facilitating endothelial differentiation.
Exercise has been reported to mobilize bone marrow–derived EPCs into the circulation,18–22 with the potential of attachment to arteries in the circulation and replacement of dysfunctional endothelium. Although we saw an increase in EPC-CFUs in our culture assay after exercise, the cells remained predominately T lymphocytes by microarray analysis of cell surface markers and by transcriptome regression analysis with T lymphocytes. Gene expression profiling before and after exercise training revealed 25 differentially regulated genes, in a magnitude of change that has been considered relevant for human microarray studies.32 Gene ontology analysis of these genes revealed an enrichment of transcripts involved in signal transduction, metabolism, immune response, translation, transcription, and transport. Yet the majority of these genes were downregulated after exercise, thus the relevance to the observed increase in colonies is obscure. Despite absence of correlations with endothelial testing to our study, it remains possible that the observed T lymphocyte population recovered from the EPC-CFU assay has a functional relationship to the endothelium or EPC capacity. Alternatively, the observed improvement in endothelial function by FMD may be independent of these T lymphocytes and may be attributable to more conventional responses. For example, repetitive shear stress attributable to exercise has been shown to increase expression and activity of endothelial nitric oxide synthase (eNOS), the product of which (NO) improves endothelial function, and could account for the enhanced dilator response of the brachial artery to shear stress in our study participants.33,34
Although other assays than the one used in our study have been proposed for demonstrating endothelial differentiation potential of blood-derived progenitor cells,12 none to date has been shown to be relevant to endothelial function in humans. Further, the role of bone marrow–derived cells in postnatal angiogenesis or arterial repair has been questioned by some groups.35–39 Animal models of bone marrow transplant and parabiosis, with donor animals genetically engineered to permit cells tracking, have shown few-to-no cells incorporated into vasculature of recipient animals, even with stimulated mobilization and homing by VEGF administration or tumor induction.39 Accordingly, improvement in endothelial function with exercise training may depend more on upregulation of eNOS within endothelium or reparative effects of resident progenitor cells within vasculature rather than bone marrow–derived progenitor cells in the circulation.40–43
We appreciate the valuable contributions of Leigh Samsel and Ann Williams of the Flow Cytometery Core Facility for technical support.
Sources of Funding
This work was supported by the Intramural Research Program of the National Heart, Lung, and Blood Institute.
The online version of this article, along with updated information and services, is located on the World Wide Web at: http://atvb.ahajournals.org/cgi/content/full/29/1/121