Basic Research in Cardiology

, 106:879

Opposing effects of monomeric and pentameric C-reactive protein on endothelial progenitor cells

  • I. Ahrens
  • H. Domeij
  • S. U. Eisenhardt
  • D. Topcic
  • M. Albrecht
  • E. Leitner
  • K. Viitaniemi
  • J. B. Jowett
  • M. Lappas
  • C. Bode
  • I. Haviv
  • K. Peter
Open AccessOriginal Contribution

DOI: 10.1007/s00395-011-0191-y

Cite this article as:
Ahrens, I., Domeij, H., Eisenhardt, S.U. et al. Basic Res Cardiol (2011) 106: 879. doi:10.1007/s00395-011-0191-y

Abstract

C-reactive protein (CRP) has been linked to the pathogenesis of atherosclerosis. The dissociation of native, pentameric (p)CRP to monomeric (m)CRP on the cell membrane of activated platelets has recently been demonstrated. The dissociation of pCRP to mCRP may explain local pro-inflammatory reactions at the site of developing atherosclerotic plaques. As a biomarker, pCRP predicts cardiovascular adverse events and so do reduced levels and function of circulating endothelial progenitor cells (EPCs). We hypothesised that mCRP and pCRP exert a differential effect on EPC function and differentiation. EPCs were treated with mCRP or pCRP for 72 h, respectively. Phenotypical characterisation was done by flow cytometry and immunofluorescence microscopy, while the effect of mCRP and pCRP on gene expression was examined by whole-genome gene expression analysis. The functional capacity of EPCs was determined by colony forming unit (CFU) assay and endothelial tube formation assay. Double staining for acetylated LDL and ulex lectin significantly decreased in cells treated with pCRP. The length of tubuli in a matrigel assay with HUVECs decreased significantly in response to pCRP, but not to mCRP. The number of CFUs increased after pCRP treatment. RNA expression profiling demonstrated that mCRP and pCRP cause highly contradictory gene regulation. Interferon-responsive genes (IFI44L, IFI44, IFI27, IFI 6, MX1, OAS2) were among the highly up-regulated genes after mCRP, but not after pCRP treatment. In conclusion, EPC phenotype, genotype and function were differentially affected by mCRP and pCRP, strongly arguing for differential roles of these two CRP conformations. The up-regulation of interferon-inducible genes in response to mCRP may constitute a mechanism for the local regulation of EPC function.

Keywords

CRPmCRPEPCCardiovascular diseaseGene arrayInterferon-alpha

Introduction

C-reactive protein (CRP) is an acute phase protein that consists of five non-covalently linked subunits forming a disc-shaped pentamer (pCRP) with a molecular weight of 115 kDa. The serum level of pCRP has been established as an independent risk factor for cardiovascular events [23, 51, 71]. We recently demonstrated the dissociation of pCRP to monomeric (m)CRP on the cell membrane of activated platelets and the deposition of mCRP in atherosclerotic plaques, thereby suggesting a causal link between localised inflammation, C-reactive protein and atherosclerosis [15, 17]. In addition to our recent findings, evidence from other studies points towards a differential effect of mCRP and pCRP on the activation of platelets and leukocytes [16, 34, 40], which are known key players involved in the initiation of atherosclerotic plaque development [26].

While elevated serum levels of C-reactive protein are predictive of adverse cardiovascular events [23, 51, 71], decreased numbers of circulating endothelial progenitor cells (EPCs) and impaired function of EPCs have been described [18, 31, 45, 48, 55, 67, 68], thereby generating the hypothesis that C-reactive protein might have a negative effect on EPC number and function. In addition, in vitro experiments have shown that pCRP reduces the number of differentiated EPCs along with a reduction in EPC-mediated endothelial tube formation [64]. Furthermore, a recent study by the same group demonstrated a dose-dependent increase in pCRP-induced formation of reactive oxygen species and apoptosis in EPCs [19].

However, the effects of C-reactive protein are versatile and still not fully understood. It appears that inflammation accompanied by elevated levels of C-reactive protein in general does not necessarily lead to decreased levels of circulating EPCs [20, 47]. In fact, the EPC number may even be elevated [20].

The recent description of the dissociation of pCRP to mCRP on cell membranes thereby generating two biologically distinct conformations of C-reactive protein [15, 28] prompted us to examine whether mCRP and pCRP have a differential effect on EPC differentiation and function.

The term EPC has initially been applied to describe a cell population possessing the ability to regenerate damaged endothelium and contribute to the development of new vessel structures (EPC) [3]. Investigations into the different purification and culturing methods have sparked an ongoing discussion about a more differentiated definition of these progenitor cells with regenerative capacities [13, 24, 50, 53, 58, 61, 62, 69]. The new definition allows a crude separation into early outgrowth EPC or pro-angiogenic cells [13] and endothelial outgrowth cells (EOCs) also described as endothelial colony forming cells (ECFCs) [24, 61, 69]. The cells that were used in the current study were derived from CD34+ selected human umbilical cord blood mononuclear cells. After in vitro expansion, the cells were differentiated into early outgrowth EPCs [1]. CD34 is a hematopoietic progenitor cell (HPC) marker, which has been linked to the term EPC throughout the majority of the available EPC literature [63, 66]. Therefore, for simplification purposes we use the term HPC to describe the CD34+-purified cells and the general term EPC to describe the cells that were used for the experiments throughout our study.

CD34+ cells purified by positive selection from human cord blood samples were expanded in vitro and thereafter differentiated into EPCs in the presence or absence of mCRP or pCRP, respectively. pCRP and mCRP showed differential effects on survival, differentiation and function of EPCs. In addition, whole-genome gene expression analysis revealed a highly differentially and opposing gene expression pattern in response to treatment with the two isoforms of CRP. Interferon-responsive genes (IFI44L, IFI44, IFI27, IFI 6, MX1, OAS2) were among the highly up-regulated genes in response to mCRP, but not pCRP treatment of EPCs.

Materials and methods

Isolation and expansion of CD34+ cells

Mononuclear cells (MNCs) were isolated from human umbilical cord blood (HUCB) obtained from healthy donors following normal full-term deliveries after their written informed consent. Ethics approval was granted by the Human Research Ethics Committee, Mercy Health, Mercy Hospital for Women, Melbourne, Australia (Project number R08/24).

HUCB was collected in 50 ml Falcon tubes (BD Bioscience, NJ, USA) containing 15 ml of the anticoagulant citrate phosphate dextrose. After collection, HUCB was diluted 1:3 in isolation buffer (PBS, 0.1% BSA, 0.6% citrate, pH 7.4). MNCs were isolated from the diluted HUCB by density gradient centrifugation, whereby 20 ml of diluted HUCB were layered onto 15-ml Ficoll-Paque™ (GE Healthcare, Chalfont, UK) and centrifuged for 30 min at 800×g. Thereafter, the interphase containing MNCs was collected and washed with an equal amount of isolation buffer, followed by centrifugation for 30 min at 500×g. The washed MNCs were then subjected to magnetic beads-based selection of CD34+ cells using Dynal® CD34 Progenitor Cells Selection System (Invitrogen, Oslo, Norway) following the manufacturer’s recommendation. The number of positively selected CD34+ cells, the haematopoietic progenitor cells (HPC), was assessed in a Neubauer haemocytometer and the purity (>90%) of the CD34+ cells evaluated by flow cytometry.

The cord blood-derived CD34+ HPCs were cultured at a density of 3–5 × 104 cells/400 μl/1.8 cm2 in a humidified incubator at 37°C with 5% CO2. The cells were cultured for 7 days in serum-free StemSpan® medium (StemCell Technologies, Vancouver, Canada) during the initial expansion period and supplemented with 1% penicillin–streptomycin (Sigma-Aldrich, St. Louis, USA) and recombinant human (rh) Flt-3 ligand (100 ng/ml), rh stem cell factor (100 ng/ml), rh IL-3 (20 ng/ml) and rh IL-6 (20 ng/ml), all purchased from StemCell Technologies. Thereafter, the HPCs were cultured (3 × 105–1 × 106/1.5 ml/9.6 cm2) for an additional 3 days in endothelial cell growth medium-2 (EGM-2) containing FBS (2%), hydrocortisone, hFGF, VEGF, R3-IGF-1, ascorbic acid, hEGF, gentamicin, amphotericin-B and heparin (Lonza, Basel, Siwtzerland).

Treatment of EPCs with monomeric and pentameric CRP

pCRP (Chemicon, Temecula, USA) and mCRP were prepared as described previously [15]. The expanded HPCs described above were collected and transferred to fibronectin (10 μg/ml) (Sigma-Aldrich)-coated plates, cultured and differentiated into EPCs for 3 days in fresh EGM-2 media in the presence or absence of mCRP (1, 5 or 25 μg/ml) or pCRP (1, 5 or 25 μg/ml), respectively. A control group of HPCs treated with EGM-2 media containing PBS (diluted 1:40) was also included.

Cell viability assays

Phosphatidylserine translocation (annexin V binding)

The binding of annexin V to phosphatidylserine was determined using FITC-conjugated annexin V (BD Biosciences, San Jose, CA, USA). In brief, the EPCs were washed in annexin-binding buffer (10 mM HEPES, 140 mM NaCl, 2.5 mM CaCl2, pH 7.4) and thereafter incubated with annexin V-FITC at RT for 15 min. After additional washing in annexin-binding buffer, the cells were analysed for annexin V-FITC binding in a FACSCalibur TM flow cytometer using CellQuest software (Becton and Dickinson, San Jose, CA, USA). Between 10,000 and 20,000 events per test were acquired. A negative control with unstained cells was included for each treatment.

CytoTox-Glo™ cytotoxicity assay

The relative viability of EPCs treated with mCRP (1, 5, or 25 μg/ml), pCRP (1, 5, or 25 μg/ml) or PBS was analysed using the CytoTox-Glo™ cytotoxicity assay (Promega, Madison, WI, USA) following the manufacturer’s recommendations. In brief, EPCs were added to a 96-well luminescence plate at a concentration of 10,000 cells/well in a volume of 100 μl of EGM-2. Thereafter, to detect protease activity that had been released from dead cells, 50 μl of alanyl-alanyl-phenylalanyl-aminoluciferin luminogenic peptide substrate (AAF-Glo™ Substrate) mix was added to the wells and incubation was carried out for 15 min at RT. The plate was measured in a 96-well plate luminometer. Immediately after measurement of dead cells by luminescence, the EPCs were exposed to a lysis buffer containing digotinin for 15 min at RT and the luminescence was measured again. The number of viable cells was determined by subtracting the number of dead cells from the total cell count. A no-cell background control was included in the assay.

Uptake of acetylated LDL and binding of ulex lectin

To investigate the endothelial progenitor characteristics of the EPCs, binding of ulex lectin and uptake of Dil-labelled acetylated LDL (AcLDL) by the cells were analysed by dual staining of the cells at 37°C in PBS.

After expansion, the cells were seeded on a fibronectin-coated 24-well plate with a density of 3 × 105cells/0.5 ml/1.8 cm2 and cultured in EGM-2 for 3 days in the presence or absence of mCRP, pCRP or PBS. Thereafter, cells were washed twice with PBS at 37°C. The cells were then incubated with Dil-AcLDL (6 μg/ml) (Invitrogen, Carlsbad, USA) for 1 h at 37°C in the dark. Thereafter, the cells were washed twice with PBS at 37°C and incubated with FITC-conjugated ulex lectin (10 μg/ml) (Sigma-Aldrich, St. Louis, USA) for 1 h at 37°C in the dark. After two final washing steps with PBS at 37°C, the cells were fixed with 0.3 ml CellFIX solution (BD Biosciences, NJ, USA), and subsequently analysed for uptake of Dil-AcLDL and binding of FITC-ulex lectin using an Olympus 1X81 inverted fluorescence microscope.

The effect of mCRP and pCRP on functional characteristics of EPCs

The EPCs treated with mCRP or pCRP were studied using colony forming assay (CFU-Hill) and Matrigel™ tube formation assay.

Colony forming unit (CFU) assay

Culture of CFU-Hill colonies (StemCell Technologies) was performed according to the manufacturer’s recommendations with the exception that we used ten times less cells (5 × 105 cells/well) and in vitro differentiated cord blood-derived EPCs instead of freshly prepared MNCs. In brief, after 72 h of culture on fibronectin in EGM-2 media in the presence or absence of mCRP or pCRP, the cells were collected and counted. The collected cells were then cultured (5 × 105 cells/1.5 ml/9.6 cm−2) on fibronectin pre-coated 6-well plates (BD Biosciences) in EndoCult® liquid medium containing EndoCult supplements (StemCell Technologies). After 2 days, the non-adherent cells were collected and transferred to fibronectin pre-coated 24-well plates (BD Biosciences) at a density of 3 × 105 cells/ml/1.8 cm2. After a further 3 days of culture, the cells were washed twice with PBS, fixed with methanol, and the colonies were visualised with Giemsa staining (Gibco, Carslbad, USA), following the manufacturer’s recommendations. The number of colonies per well was counted with an inverted microscope (Olympus CKX41).

Endothelial tube formation assay

The capability of the mCRP- and pCRP-treated EPCs to support endothelial tube formation was assessed using Matrigel™ (BD Biosciences) coated 96-well plates and human umbilical vein endothelial cells (HUVECs) in early passages (P3-P6). In brief, wells of a 96-well plate were coated with 50 μl of ice-cold Matrigel™ followed by incubation at 37°C for 1 h. Thereafter, 100 μl of EGM-2 medium containing 25,000 HUVECs and 100 μl of EGM-2 containing 25,000 EPCs were added to the Matrigel™. Incubation was carried out for 16 h in a humidified atmosphere at 37°C in the presence of 5% CO2. Tube formation was assessed with an inverted microscope (Olympus 1X81) and Cell^P imaging software (Olympus). Digital photomicrographs of each single well were taken at a 4× magnification and the total number of tubes, branching points, and the length of the tubes as well as the sum of the lengths of all tubes were calculated for each well.

Whole-genome gene expression analysis of mCRP and pCRP-treated EPCs

Total RNA was extracted from 2 × 106 EPCs after 72 h of culture in EGM-2 medium in the presence or absence of mCRP (1 μg/ml), pCRP (5 μg/ml) or PBS. The RNA was obtained using Qiagen® RNeasy protect mini kit™ following the manufacturer’s instructions. RNA concentration and integrity was analysed by NanoDrop (Thermo Fischer Scientific, Waltham, USA) and MultiNA microchip electrophoresis (Shimadzu Biotech, Kyoto, Japan) according to the manufacturer’s recommendations. Total RNA was amplified with the TotalPrep™ RNA Amplification Kit (Ambion, UK) and applied to Illumina® Human WG-6 v3.0 Expression BeadChip kits according to the manufacturer’s instructions. Fluorescent bead intensity was transformed into gene expression level via Illumina Genome Studio, including quantile normalisation [27] and background subtraction. The exported reports were analysed on GeneSpring GX10, Partek GS and arraytools (http://linus.nci.nih.gov/pub/rsimon/ ArrayTools). Genes with a raw signal <250 and a detection confidence >0.8 in at least three arrays were filtered out, leaving 6,064 genes for further analysis. Whisker box plot was used to confirm the quintile normalisation eliminated systematic cross array variations in the overall dynamic range. Quality control of the samples and controlling for systematic bias were performed using principal component analysis and unsupervised hierarchical clustering to show that samples segregated according to treatment groups. Differentially expressed genes were selected based on a Bayesian “volcano plot” of expression fold change greater than twofold and significance p value of <0.05, including Benjamini Hochberg false discovery correction. Differential expression of the chosen genes across the treatment groups was assessed using supervised hierarchical clustering [14] that measures the proximity of distribution of samples and genes. The interpretation of the resulting gene lists was performed using gene ontology Web interface (http://david.abcc.ncifcrf.gov/), Gene Set Analysis [6], protein–protein interaction KEGG database [22] and Ingenuity Pathways Analysis [49].

Quantitative real-time PCR analysis

First-strand cDNA was synthesised using 350 ng of total RNA, obtained as described in the microarray section, and random primers in a 10 μl reverse transcriptase reaction mixture using TaqMan® Reverse Transcription Reagents (Applied Biosystems) following the manufacturer’s recommendations. Quantitative real-time PCR assays were carried out with Applied Biosystems 7500 real-time PCR system using SYBR® Green Mastermix (Applied Biosystems, Carlsbad, USA). Primers were designed in-house and synthesised by GeneWorks (Hindmarsh, Australia). PCR amplification was performed in a 96-well plate with a final volume of 20 μl reaction mixture in each well. For each sample, 21 ng of cDNA was loaded in duplicates with 1× SYBR® Green Mastermix and 10 μM of the following primers for: Mx1 sense 5′-CACTGCGCAGGGACCGGAATT-3′ and anti-sense 5′-TCCTGTAGCCTCCGACCCAGAA-3′; OAS2 sense 5′-GCTCCCGGCCCACCAAACTA-3′ and anti-sense 5′-TGGGGGCAAAGACCCCTTTGG-3′; OAS3 sense 5′-GGACCCTGCAGTTGGGCAGT-3′ and anti-sense 5′-CCCATGTGGGGTCAGCTGGG-3′; IFIT3 sense 5′-ACCGGGACCCCAGCTTTTCAG-3′ and anti-sense 5′-AGCTGTGGAAGGATTTTCTCCAGGG-3′; IFI6 sense 5′-TCCGGGCTGAAGATTGCTTCTCTT-3′ and anti-sense 5′- ACTTTTTCTTACCTGCCTCCACCCC-3′; IFI44 sense 5′-GAGATGTGAGCCTGTGAGGTCCAA-3′ and anti-sense 5′TTTACAGGGTCCAGCTCCCACTCA-3′; IFI27 sense 5′- CCGTAGTTTTGCCCCTGGCC-3′ and anti-sense 5′-CATGGGCACAGCCGCCATG-3′; IFI44L sense 5′- ATGTGACTGGCCAAGCCGTAGT-3′ and anti-sense 5′TGCCCCATCTAGCCCCATAGTGT-3′; PROK2 sense 5′-TGGGAGACAGCTGCCATCCAC-3′ and anti-sense 5′-AGCCTGGCAGACATGGGCAA3′; DDIT3 sense 5′-TCAGAGCTGGAACCTGAGGAGAGA-3′ and anti-sense 5′-ATGGGGAGTGGCTGGAACAAGC-3′. Relative expression of the genes was obtained using the differences in cycle threshold (Ct) between the sample and 18S ribosomal RNA (ΔC). The difference in gene expression for the treated samples compared to the PBS control samples was calculated (ΔΔCt) and the fold difference was calculated as 2ΔΔCt.

Treatment with human IFNα2A

The expanded HPCs were collected and transferred to fibronectin (10 μg/ml)-coated plates, cultured and differentiated into EPCs for 3 days in fresh EGM-2 media in the presence or absence of IFNα2A (0.1–10 ng/ml) (Sigma), mCRP (1 μg/ml) or pCRP (5 μg/ml), respectively. A control group of HPCs treated with EGM-2 media containing PBS (diluted 1:40) was also included. Changes in gene expression were determined using real-time PCR. In addition, uptake of AcLDL and binding of ulex lectin, CFU assay and endothelial tube formation assay were investigated as described above.

Statistical analysis

All experiments were performed with EPCs from at least three different donors, respectively. Mean and SD were used for descriptive statistics. Statistical analysis was performed using Sigma Stat. Differences between means were assessed by analysis of variance (ANOVA) and post hoc test was carried out according to Bonferroni. p values <0.05 were considered to be statistically significant.

Results

EPCs were treated with mCRP, pCRP or PBS for a period of 72 h in fibronectin-coated six-well plates (10 μg/ml). Thereafter EPC viability, phenotype and function were assessed. In addition, differences in gene expression profiles were assessed in EPCs derived from three different cord blood donors before and after mCRP, pCRP and PBS treatment, respectively.

Assessment of mCRP and pCRP toxicity

The effect of a 72-h treatment period of mCRP and pCRP on EPC viability was investigated using a cytotoxicity luminescence assay. First, the number of dead cells was measured and secondly the total number of cells was measured. The relative number of viable cells (relative live cell luminescence, RLU) obtained in the assay is demonstrated in Fig. 1a. There was no toxicity associated with mCRP treatment (1, 5, or 25 μg/ml for 72 h). In fact, mCRP significantly (p < 0.05–0.001) increased the viability/cell numbers in a dose-dependent manner (Fig. 1a). In accordance to mCRP, pCRP at the concentrations 1 and 5 μg/ml for 72 h was not toxic to the EPCs. However, at the concentration of 25 μg/ml, the viability of the cells was strongly reduced (p < 0.001, Fig. 1a). In addition to the luminescence-based cytotoxicity assay, which is based on protease activity released from cells through the loss of membrane integrity, the effect of mCRP and pCRP treatment on EPCs was also determined by annexin V binding. In accordance with the cytotoxicity assay, only pCRP at the concentration 25 μg/ml induced an increased binding of annexin V, thereby indicating a pro-apoptotic/necrotic effect on EPCs (Fig. 1b).
https://static-content.springer.com/image/art%3A10.1007%2Fs00395-011-0191-y/MediaObjects/395_2011_191_Fig1_HTML.gif
Fig. 1

Cytotoxicity assay after 72 h of EPC treatment with mCRP and pCRP. a Viable cells were measured by the CytoTox-Glo™ cytotoxicity assay. The bars represent the relative luminesce unit (RLU) calculated for the viable cells, i.e. dead cells deducted from total cells, mean and SD of n = 3. b. Histograms of one representative experiment showing the binding of the apoptosis marker Annexin V to EPCs using a FITC-labelled Annexin V antibody detected by flow cytometry

In the following descriptive and functional investigations, we focused on the comparison of 1 and 5 μg/ml mCRP and pCRP, respectively, to adjust for molarity of the pentameric structure of pCRP.

pCRP but not mCRP treatment decreased the number of AcLDL and ulex lectin double-positive cells

One phenotypic characteristic for EPCs is their ability to take up AcLDL and to bind ulex lectin. The uptake of AcLDL and the binding of ulex lectin were investigated by immunofluorescence staining using Dil-labelled AcLDL and FITC-labelled ulex lectin. The percentage of cells double positive for uptake of Dil-labelled AcLDL and binding of FITC-labelled ulex lectin after treatment with 1 μg/ml mCRP, 5 μg/ml pCRP and PBS was 36 ± 7, 42 ± 11 and 12 ± 3%, respectively (Fig. 2a). The percentage of double-positive cells indicating EPC phenotype was significantly decreased after pCRP treatment (12 ± 3%, p < 0.05, Fig. 2a). In contrast, EPCs treated with mCRP (1 μg/ml) were not affected and showed a similar percentage of double-positive cells compared to the PBS control.
https://static-content.springer.com/image/art%3A10.1007%2Fs00395-011-0191-y/MediaObjects/395_2011_191_Fig2_HTML.gif
Fig. 2

The effect of mCRP (1 μg/ml) or pCRP (5 μg/ml) on phenotype and function of EPCs. a Binding of ulex lectin and uptake of AcLDL after 72 h of culture of EPCs in the presence or absence of mCRP or pCRP visualised with FITC-labelled ulex lectin and Dil-labelled AcLDL, respectively. The photomicrographs show a typical optical field from one representative experiment. The bar graph shows the percentage of double-positive cells, mean and SD of n = 3. b The number of colony forming units (CFU-Hill) visualised by Giemsa staining. The bars represent the total number of colonies per well in a 24-well plate, mean and SD of n = 3. c Endothelial tube formation assay after 72 h of culture of EPCs in the presence or absence of mCRP or pCRP. The treated EPCs were co-cultured with HUVECs for additional 4, 16 and 24 h, respectively. The photomicrographs show the results of a representative experiment. The length of the tubuli was assessed after 16 h. The box plot represents the mean (dotted line), median (solid line) and SD of three independent experiments

Functional characteristics of EPCs are differentially affected by mCRP and pCRP

EPCs were treated with mCRP (1 μg/ml), pCRP (5 μg/ml) or PBS and then assessed for their ability to form colony forming units (CFU-Hill) and to support endothelial tube formation. Treatment of EPCs with pCRP significantly increased the number of CFUs (40 ± 5, p < 0.001), whereas mCRP (19 ± 2) did not affect the number of CFUs compared to PBS (25 ± 3) treated cells (Fig. 2b). The effect of mCRP (1 μg/ml) and pCRP (5 μg/ml) treatment on the ability of EPCs to support formation of endothelial tubuli in a co-culture with human umbilical vein endothelial cells was determined using a Matrigel™ assay. Tube formation occurred at 4 h of incubation on Matrigel™ and was maximal at 16 h of incubation (PBS-panel, Fig. 2c). After 24 h, the tube network started to break up, and this was facilitated when pCRP-treated EPCs were used (Fig. 2c). Neither mCRP nor pCRP affected the number of tubuli or branching points. However, the length of the tubuli was slightly increased by EPCs treated with mCRP, whereas the ability of these cells to support tube formation was significantly reduced after treatment with pCRP, respectively (p < 0.001, Fig. 2c).

Differential effect of mCRP and pCRP on the gene expression profile of EPCs

RNA for whole-genome gene expression analysis was extracted after a 72-h culture period of in vitro expanded cord blood-derived CD34+ cells on fibronectin-coated dishes. Cells were treated with mCRP (1 μg/ml), pCRP (5 μg/ml) or PBS in EGM-2. In addition, RNA was extracted before the 72-h culture period to determine the background of differentially expressed genes induced by the EPC differentiation process during the 72-h culture period on fibronectin-coated dishes. The microarray analysis was undertaken in three biological repeats (three different cord blood donors) on beadchips containing >48,000 different probes. Box and whisker plots confirmed that the normalisation of the arrays achieved comparable dynamic range of the different samples (supplementary Fig. 1). In contrast, principal component analysis (PCA) demonstrated that the gene expression, induced by either mCRP or pCRP, was systematically distinct across the three donors (Fig. 3a). Hierarchical clustering of 320 differentially expressed genes further demonstrated that mCRP and pCRP treatment differentially induced coordinate programmatic gene expression and differentiation (Fig. 3b).
https://static-content.springer.com/image/art%3A10.1007%2Fs00395-011-0191-y/MediaObjects/395_2011_191_Fig3_HTML.gif
Fig. 3

a Principal component analysis (PCA) of the source of variation in the sample gene expression. This was an exploratory multivariate statistical technique that was used to simplify the complex microarray changes that occur in three individuals (patient 10, 18 and 20) and in response to different valency CRP. This is done by reducing the dimensionality of the data matrix by finding r new variables (that sum the expression of multiple genes into single axes); shown here are the average signal of each sample along the three-dimensional virtual space of the first three principal components. The plot clearly shows that the variation in gene expression is dominated by the CRP treatment and not the patient source. b Hierarchical clustering of the gene expression response to CRP valency; 320 differentially expressed genes are plotted according to their degree of respective co-expression. Columns represent samples, while rows represent genes. The treatment of each sample is listed at the bottom. The degree of correlation between genes (left) or samples (top) is plotted in a tree view fashion

The gene expression analysis comparing EPCs treated with mCRP to EPCs treated with PBS detected 42 differentially expressed genes, of which 28 were significantly up-regulated and 14 were down-regulated (at least twofold change, p < 0.05) (Table 1). Comparing expression patterns of EPCs treated with pCRP to PBS-treated EPCs detected 86 differentially expressed genes, of which 42 were up-regulated and 44 were down-regulated (at least twofold change, p < 0.05) (Table 2). As demonstrated in Table 1, analysis of the differentially expressed genes revealed that mCRP and pCRP showed opposing effects, meaning that the genes up-regulated by mCRP were down-regulated by pCRP, and vice versa.
Table 1

Genes up- and down-regulated in EPCs treated with mCRP (1 μg/ml)

Gene symbol

Gene name

mCRP treatment

pCRP treatment

Fold change

p Value

Fold change

p Value

Up-regulated genes

 IFI44L

Interferon-induced protein 44-like

10.04

0.001

−2.36

0.61

 MX1

Myxovirus (influenza virus) resistance 1

7.58

0.008

−2.68

0.46

 OAS2

2′-5′-Oligoadenylate synthetase 2, 69/71kDA

6.84

0.001

−2.56

0.31

 LY6E

Lymphocyte antigen 6 complex, locus E

6.18

0.003

−2.64

0.33

 IFI27

Interferon, alpha-inducible protein 27

6.11

0.040

−2.52

0.63

 IFI44

Interferon-induced protein 44

6.05

0.0005

−2.43

0.35

 IFI6

Interferon, alpha-inducible protein 6

5.09

0.0004

−2.51

0.20

 EPSTI1

Epithelial stromal interaction 1 (breast)

4.94

0.026

−2.06

0.94

 OAS3

2′-5′-Oligoadenylate synthetase 3, 100 kDa

4.28

0.005

−2.90

0.10

 IFIT3

Interferon-induced protein with tetratricopeptide repeats 3

3.87

0.048

−2.92

0.22

 SAMD9L

Sterile alpha motif domain containing 9-like

3.73

0.002

−3.04

0.02

 GLTPD2

Glycolipid transfer protein domain containing 2

3.65

0.009

−2.02

0.96

 EIF2AK2

Eukaryotic translation initiation factor 2-alpha kinase 2

3.42

0.007

−2.14

0.65

 DDX60

DEAD (Asp–Glu–Ala–Asp) box polypeptide 60

3.28

0.031

−2.24

0.57

 SPINT4

Serine peptidase inhibitor, Kunitz type 4

3.23

0.026

−2.18

0.64

 IRF7

Interferon-regulatory factor 7

3.14

0.032

−2.22

0.57

 PARP9

Poly (ADP-ribose) polymerase family, member 9

3.10

0.031

−2.69

0.12

 GPR87

G protein-coupled receptor 87

3.08

0.017

−2.09

0.77

 RETSAT

Retinol saturase (all-trans-retinol 13, 14-reductase)

3.08

0.022

−2.30

0.38

 CLDN4

Claudin 4

3.04

0.032

−2.27

0.46

 IRF9

Interferon-regulatory factor 9

3.01

0.032

−2.53

0.18

 OASL

2′-5′-Oligoadenylate synthetase-like

2.97

0.038

−2.46

0.23

 PARP12

Poly (ADP-ribose) polymerase family, member 12

2.88

0.037

−2.50

0.17

 STAT1

Signal transducer and activator of transcription 1

2.87

0.009

−2.29

0.24

 HIST1H2AK

Histone cluster 1, H2ak

2.85

0.039

−2.25

0.43

 PEX11A

Peroxisomal biogenesis factor 11 alpha

2.75

0.014

−2.50

0.06

 POMT2

Protein-O-mannosyltransferase 2

2.61

0.003

−2.43

0.09

 EPB41

Erythrocyte membrane protein band 4.1

2.56

0.014

−2.37

0.07

Down-regulated genes

 MLLT3

Myeloid/lymphoid or mixed-lineage leukaemia; translocated to, 3

−3.48

0.007

2.08

0.80

 SOX1

SRY (sex determining region Y)-box 1

−3.25

0.002

2.04

0.86

 SRD5A2

Steroid-5-alpha-reductase, alpha polypeptide 2

−3.16

0.008

2.55

0.10

 ACSM1

Acyl-CoA synthetase medium-chain family member 1

−3.16

0.003

2.28

0.27

 RNF133

Ring finger protein 133

−3.02

0.035

2.16

0.64

 ROR1

Receptor tyrosine kinase-like orphan receptor 1

−2.99

0.032

2.03

0.92

 ZEB1

Zinc finger E-box-binding homoeobox 1

−2.99

0.001

2.02

0.91

 ZNF300

Zinc finger protein 300

−2.96

0.011

2.12

0.64

 PRAMEF21

PRAME family member 21

−2.93

0.038

2.37

0.30

 SRD5A3

Steroid 5 alpha-reductase-3

−2.87

0.024

2.26

0.37

 RPL21

Ribosomal protein L21

−2.64

0.008

2.41

0.05

 CARD18

Caspase recruitment domain family, member 18

−2.60

0.025

2.33

0.15

 PCDHGB6

Protocadherin gamma subfamily B, 6

−2.50

0.021

2.57

0.01

 GRIN1

Glutamate receptor, ionotropic

−2.42

0.008

2.49

0.004

Up- and down-regulated genes by mCRP (n = 3) compared to PBS-treated control EPCs were selected based on a Bayesian “volcano plot” of expression with a fold change in expression level greater than twofold, and a p value <0.05 as determined by student’s t test

The fold change effect of pCRP treatment, as compared to PBS-treated control EPCs, for the differentially expressed genes was also determined

Table 2

Genes up- and down-regulated in EPCs treated with pCRP (5 μg/ml)

Gene symbol

Gene name

mCRP treatment

pCRP treatment

Fold change

p Value

Fold change

p Value

Up-regulated genes

 MFAP4

Microfibrillar-associated protein 4

3.60

0.0010

−2.03

0.855

 RABGGTB

Rab geranylgeranyltransferase, beta subunit

3.51

0.0013

−2.05

0.793

 FAM116B

Family with sequence similarity 116, member B

3.16

0.0040

−2.00

0.994

 DDIT3

DNA damage-inducible transcript 3

3.08

0.0004

−2.07

0.578

 PROK2

Prokinectin 2

3.08

0.0486

−2.13

0.690

 RPL7A

Ribosomal protein L7a

3.05

0.0311

−2.43

0.215

 PABPC4

Poly(A)-binding protein, cytoplasmic 4 (inducible form)

3.04

0.0042

−2.07

0.704

 CELSR1

Cadherin, EGF LAG seven-pass G-type receptor 1

3.01

0.0104

−2.12

0.570

 FRMD7

FERM domain containing 7

3.01

0.0499

−2.10

0.739

 ZNF695

Zinc finger protein 695

3.00

0.0199

−2.04

0.859

 HERC2

Hect domain and RLD2

2.99

0.0286

−2.00

0.998

 RPL13L

Ribosomal protein L13 pseudogene 5

2.98

0.0094

−2.12

0.566

 ZNF277

Zinc finger protein 277

2.96

0.0002

−2.11

0.325

 LRAT

Lectin retinol acyltransferase

2.90

0.0157

−2.12

0.572

 STAG3L4

Stromal antigen 3-like 4

2.87

0.0329

−2.25

0.368

 ABHD14B

Abhydrolase domain containing 14B

2.86

0.0488

−2.17

0.554

 AVPI1

Arginine vasopressin-induced 1

2.85

0.0062

−2.21

0.258

 DYM

Dymeclin

2.82

0.0005

−2.05

0.652

 ROM1

Retinal outer segment membrane protein 1

2.81

0.0099

−2.30

0.163

 RAC3

ras-related C3 botulinum toxin substrate 3

2.81

0.0076

−2.10

0.558

 CCDC104

Coiled-coil domain containing 104

2.81

0.0002

−2.15

0.164

 SLC16A10

Solute carrier family 16, member 10

2.80

0.0082

−2.15

0.398

 PHACTR1

Phosphatase and actin regulator 1

2.80

0.0079

−2.04

0.813

 PCOLCE2

Procollagen C-endopeptidase enhancer 2

2.76

0.0379

−2.10

0.657

 ZNF581

Zinc finger protein 581

2.75

0.0207

−2.11

0.602

 CRP

C-reactive protein, pentraxin related

2.72

0.0400

−2.15

0.528

 SIRPG

Signal-regulatory protein gamma

2.70

0.0129

−2.24

0.210

 NNMT

Nicotinamide N-methyltransferase

2.70

0.0115

−2.07

0.685

 EIF2A

Eukaryotic translation initiation factor 2A

2.69

0.0134

−2.22

0.249

 FABP2

Fatty acid-binding protein 2, intestinal

2.69

0.0067

−2.13

0.411

 RGMA

RGM domain family, member A

2.69

0.0146

−2.08

0.636

 POLR1E

Polymerase (RNA) polypeptide E, 53 kDa

2.61

0.0358

−2.18

0.385

 LETMD1

LETM1 domain containing 1

2.60

0.0168

−2.20

0.258

 FTHL11

Ferritin, heavy polypeptide-like 11

2.59

0.0422

−2.25

0.261

 CSNK2A2

Casein kinase 2, alpha prime polypeptide

2.57

0.0036

−2.27

0.061

 CCDC140

Coiled-coil domain containing 140

2.53

0.0332

−2.17

0.346

 PAX7

Paired box 7

2.49

0.0343

−2.25

0.190

 ZFAND1

Zinc finger, AN 1-type domain 1

2.49

0.0175

−2.21

0.175

 RAB39

RAB39, member RAS oncogene family

2.44

0.0350

−2.25

0.156

 PCDHGB6

Protocadherin gamma subfamily B, 6

2.44

0.0115

−2.38

0.021

 GRIN1

Glutamate receptor, inotropic, N-methyl d-aspartate 1

2.37

0.0041

−2.31

0.008

 RPL21

Ribosomal protein L21

2.30

0.0474

−2.49

0.008

Down-regulated genes

 ABCA13

ATP-binding cassette, sub-family A (ABC1), member 13

−5.39

0.0009

2.02

0.931

 LTF

Lactotransferrin

−5.11

0.0427

2.27

0.652

 PDAP1

PDGFA associated protein 1

−5.04

0.0005

2.23

0.353

 IFIT1

Interferon-induced protein with tetratricopeptide repeats 1

−4.39

0.0264

2.86

0.175

 TXNIP

Thioredoxin-interacting protein

−3.75

0.0207

2.03

0.917

 UGT1A3

UDP glucoronsyltransferase 1 family, polypeptide A3

−3.60

0.0453

2.05

0.891

 RIMS3

Regulating synaptic membrane exocytosis 3

−3.47

0.0004

2.04

0.906

 PKD2L1

Polycystic kidney disease 2-like 1

−3.47

0.0191

2.34

0.336

 MT1L

Metallothionein 1L (gene/pseudogene)

−3.37

0.0006

2.07

0.666

 OXTR

Oxytocin receptor

−3.30

0.0464

2.04

0.906

 SIGLEC5

Sialic acid-binding Ig-like lectin 5

−3.29

0.0350

2.03

0.912

 NCF1

Neutrophil cytosolic factor 1

−3.25

0.0308

2.00

0.987

 LRRC39

Leucin rich repeat containing 39

−3.20

0.0082

2.14

0.549

Up- and down-regulated genes by pCRP (n = 3) compared to PBS-treated control EPCs were selected based on a Bayesian “volcano plot” of expression with a fold change in expression level greater than twofold, and a p value <0.05 as determined by student’s t test

The fold change effect of mCRP treatment, as compared to PBS-treated control EPCs, for the differentially expressed genes was also determined

mCRP, but not pCRP, stimulates the expression of interferon-α responsive genes

To benchmark differential gene expression in EPCs in response to either mCRP or pCRP treatment to existing microarray literature and to gain more biological insight from the gene expression pattern specific for mCRP and pCRP treatment, we have employed Ingenuity Pathway Analysis® (IPA) and Gene Set Enrichment Analysis [41]. Those genes that were up-regulated in response to mCRP exhibited extensive cross interactions, centred around interferon-α (Table 1; Fig. 4). Treatment with pCRP led to a highly different response in the gene expression profile centred around the MAPK pathway, DDIT3/GADD153 and consequent cell growth (ribosomal subunits), and cell cycle regulators such as p21/CDKN1A (Table 2, Fig. 5). Genes with significant down-regulation in response to mCRP and pCRP treatments are described in Tables 1 and 2. An IPA network analysis of selected down-regulated genes in response to mCRP and pCRP, respectively, is depicted in Supplementary Figures 2 and 3.
https://static-content.springer.com/image/art%3A10.1007%2Fs00395-011-0191-y/MediaObjects/395_2011_191_Fig4_HTML.gif
Fig. 4

Ingenuity pathways analysis. Network of up-regulated genes in response to 1 μg/ml mCRP. Networks of gene/gene product interaction were generated using IPA (Ingenuity® Systems, http://www.ingenuity.com). Genes or gene products are represented as nodes, and the biological relationship between two nodes is represented as an edge (line). All edges are supported by at least one published reference. Solid edges represent a direct relationship and dashed edges represent an indirect relationship. The red node colour represents up-regulation in response to mCRP and the colour intensity indicates the degree of up-regulation. The shape of each node represents the functional class of the gene product, as shown in the legend

https://static-content.springer.com/image/art%3A10.1007%2Fs00395-011-0191-y/MediaObjects/395_2011_191_Fig5_HTML.gif
Fig. 5

Ingenuity pathways analysis. Network of up-regulated genes in response to 5 μg/ml pCRP. Networks of gene/gene product interaction were generated using IPA (Ingenuity® Systems, http://www.ingenuity.com). Genes or gene products are represented as nodes, and the biological relationship between two nodes is represented as an edge (line). All edges are supported by at least one published reference. Solid edges represent a direct relationship and dashed edges represent an indirect relationship. The red node colour represents up-regulation in response to pCRP. The shape of each node represents the functional class of the gene product, as shown in the legend of Fig. 4

The gene expression was verified by quantitative real-time PCR with RNA isolated from mCRP (1 μg/ml), pCRP (5 μg/ml) or PBS-treated EPCs and expressed as fold change in gene expression compared to the PBS-treated control (Fig. 6). Confirming the results of the gene array, IFI44L was the most up-regulated gene in response to mCRP treatment (29.3-fold ± 17) followed by MX1 (8.3-fold ± 2.2) and OAS2 (8.6 ± 2.4) (Fig. 6). Treatment with pCRP but not mCRP led to increased expression of DDIT3 (4.4 ± 0.97) and PROK2 (2.2 ± 0.86) (Fig. 6).
https://static-content.springer.com/image/art%3A10.1007%2Fs00395-011-0191-y/MediaObjects/395_2011_191_Fig6_HTML.gif
Fig. 6

Validation of the expression of selected highly upregulated genes in mCRP (1 μg/ml) and pCRP (5 μg/ml)-treated EPCs by quantitative real-time PCR. Validation of gene expression by quantitative real-time PCR. The mean in gene expression from EPCs derived from three individual cord blood donors was obtained using differences in cycle threshold between the gene and 18 s (ΔCt). The fold change in difference (ΔΔCt) in gene expression in the treated samples compared to the PBS control samples was determined (2ΔΔCt) and expressed in the diagram as mean and SEM of n = 3. Grey bars represent pCRP and black bars represent mCRP-treated EPCs

The effects of IFNα2A on phenotypic and functional characteristics of EPCs

The percentage of EPCs double positive for the uptake of AcLDL and binding of ulex lectin was significantly higher (p < 0.001) in the presence of IFNα2A or mCRP compared to pCRP (Fig. 7a). Also, compared to pCRP-treated EPCs, IFNα2A or mCRP-treated EPCs gave rise to significantly fewer colonies (p < 0.01) (Fig. 7b). Compared to PBS, mCRP or IFNα2A-treated EPCs, pCRP-treated EPCs showed a reduced ability to stimulate endothelial tube formation assessed by the length of endothelial tubes in a Matrigel™ assay with HUVECs (Fig. 7c).
https://static-content.springer.com/image/art%3A10.1007%2Fs00395-011-0191-y/MediaObjects/395_2011_191_Fig7_HTML.gif
Fig. 7

The effect of mCRP (1 mg/ml), pCRP (5 mg/ml) or IFNa2A (1 ng/ml) on phenotype and function of EPCs. a Binding of ulex lectin and uptake of acetylated LDL after 72 h of culture of EPCs in the presence or absence of pCRP, mCRP and IFNa2A. The bar graph shows the percentage of cells double positive for binding of ulex lectin and uptake of acetylated LDL as compared to the total number of cells in three random optical fields (20× magnification). Mean and SD of n = 3. The statistical significance was determined by one-way ANOVA with Tukey’s post hoc test, ***p<0.001. b The bar graph shows the number of colony forming units (CFU-Hills) visualised by Giemsa staining, formed by EPCs that had been cultured in the presence of pCRP, mCRP or IFNalpha2 for 72 h (mean and SD of three individual donors, n = 3). The statistical significance was determined by one-way ANOVA with Tukey’s post hoc test, **p < 0.01. c The boxplot shows the lengths of the tubuli formed in a co-culture of HUVECs and EPCs (n = 3) in an endothelial tube formation assay. The EPCs had been cultured in the presence of pCRP, mCRP or IFNa2A for 72 h prior to the assay. The photomicrographs were taken when a clear tubuli network was observed after 16 h of incubation. The statistical significance was determined by ANOVA on ranks with Dunn’s post hoc test, *p < 0.05

Discussion

This study aimed to assess the effects of pentameric (p)CRP and monomeric (m)CRP on function and differentiation of EPCs. The generation of mCRP by dissociation of pCRP on the cell membrane of activated platelets has been recently described as a pathophysiological mechanism to localise inflammatory reactions, e.g. at developing atherosclerotic plaques [15]. The regenerative capability and the numbers of circulating EPCs have been associated with atherosclerotic plaque development and thus we hypothesised that CRP, particularly mCRP, is a potential regulator of EPC function.

The two major findings of this study are: (1) mCRP and pCRP at concentrations similar to the reported serum levels of CRP in patients with atherosclerotic disease [2] induce an opposing gene expression profile in cord blood-derived EPCs as well as differential effects on the functional capacity of EPCs. (2) The highly up-regulated genes in response to mCRP but not to pCRP treatment are centred around the pro-inflammatory actions of interferon-α and are highly related to up-regulated genes in patients with systemic lupus erythematodes (SLE). Interferon-α2A treatment induced a functional response in EPCs similar to mCRP treatment.

Differential effects of mCRP and pCRP on monocyte and platelet function have been observed in previous studies of our group and others [15, 16, 40]. EPC function following pCRP treatment has been extensively examined in previous studies [7, 19, 38, 64], but the differential effects of CRP isoforms on EPCs have not been investigated yet. EPCs are thought to play a key role in the regeneration of injured/inflamed endothelium, and reduced functional capacity of EPCs has been correlated to serum levels of CRP [9]. Therefore, CRP (either monomeric or pentameric) may directly account for the impaired EPC function in patients with chronically elevated serum CRP levels in atherosclerotic disease. Our in vitro study confirms the results of previous studies showing decreased viability and induction of apoptosis with pCRP concentrations >10 μg/ml [5, 19, 52, 64]. Furthermore, we are able to confirm that pCRP, but not mCRP, decreases the number of cells double positive for the uptake of AcLDL and the binding of ulex lectin, which has been described as a common EPC phenotype [24, 29, 48, 69]. The ability of EPCs to stimulate endothelial tube formation in co-cultures with HUVECs has been used to describe the pro-angiogenic functions of EPCs [8, 24, 64]. In line with other reports [64], our results indicate that pCRP directly impairs endothelial tube formation and this occurred at the concentration of 5 μg/ml, which was not cytotoxic to the EPCs, thereby indicating that EPC differentiation rather than viability in response to pCRP treatment impaired the ability of EPCs to support endothelial tube formation. This interpretation is supported by the decreased number of AcLDL and ulex lectin double-positive cells after pCRP treatment and the significant increase in the number of CFU in response to pCRP treatment.

The increase in CFUs may be explained by the up-regulation of two genes (DDIT3 and PROK2) in response to pCRP treatment. PROK2 has been shown to increase the number of CFUs and to enhance progenitor cell mobilisation [36] and DDIT3 (also known as GADD153 or CHOP), although also associated with the response to oxidative stress [12, 42], increased the number of CFUs in erythroid cells [10]. Interestingly, pCRP treatment of human coronary vascular smooth muscle cells with 5 μg/ml of pCRP also led to a >2-fold increase in expression of DDIT3 (GADD153) [5], which has been reproduced by others with higher concentrations of pCRP (25 μg/ml) [52].

In view of our observations showing increased CFUs with low concentrations of pCRP and decreased viability and increased apoptosis with higher concentrations, we hypothesise that moderately elevated levels of pCRP under non-inflammatory conditions may have a different impact on EPC differentiation and function as compared to inflammatory conditions. This is supported by the observation that the endothelial colony forming capacity of EPCs is positively correlated to pCRP levels in healthy volunteers [9]. Whereas under inflammatory conditions, such as those present in atherosclerotic disease, localised dissociation of pCRP to mCRP may occur [15]. This in turn could then have a negative biological effect on EPC numbers and function as observed in patients with atherosclerosis [55, 67, 68].

In our in vitro study, mCRP treatment of EPCs did not decrease the numbers of AcLDL and ulex lectin positive cells and there was also no decrease in their capacity to support endothelial tube formation with mCRP treatment, which furher highlights the differential effects of the two CRP isoforms that has been shown in vitro in various cell types [28, 3235, 56, 72]. We compared the concentrations of 1 μM mCRP with 5 μM pCRP to account for the differences in molarity between pCRP (115 kDa) and mCRP (23 kDa).

The substantial differences in the effects of mCRP and pCRP were confirmed in whole-genome gene expression analysis. The genes up-regulated by mCRP are known to be responsive to the pro-inflammatory interferon-α. Among the 28 highly up-regulated genes in response to mCRP treatment were nine interferon-inducible genes (IFI44L, MX1, IFI27, OAS3, IFI44, Ly6E, EPISTI1, STAT1). Interestingly, these genes matched highly up-regulated genes described in the microarray analysis of tissue from synovial biopsies of patients with systemic lupus erythematosus [44]. In addition, the three genes that showed the strongest up-regulation in response to mCRP (IFI44L, MX1 and OAS2) were also found to be the most up-regulated genes in peripheral blood mononuclear cells (PBMCs) of paediatric patients with SLE [4]. Importantly, patients with SLE are more prone to the development of atherosclerosis [57] and especially women with SLE and coronary artery disease (CAD) have a poor outcome after percutaneous coronary intervention (PCI) [39]. Interferon-α (measured via expression of the interferon-α responsive genes IFI44 and MX1) has been linked to abnormal vascular repair caused by EPC dysfunction in patients with SLE thereby generating the hypothesis that interferon-α triggered EPC dysfunction is involved in the accelerated atherosclerosis in patients with SLE [11, 30]. This is supported by the observation that increased expression of MX1 in PBMCs of SLE patients and elevated levels of pCRP are independently associated with EPC dysfunction [37]. Furthermore, impaired function of EPCs was accompanied by increased expression of MX1 in an animal model of SLE [60]. Finally, auto-antibodies against mCRP have been described in serum from patients with active SLE [59], which generates the hypothesis that elevated mCRP levels lead to interferon-α-mediated impairment of EPCs in patients with SLE, thereby impairing vascular repair and accelerating the development of atherosclerosis. Of note, our data do not directly demonstrate a pro-atherosclerotic effect of mCRP. However, for interferon-α and β, such a direct pro-atherosclerotic effect has very recently been demonstrated in mouse models of atherosclerosis [21, 43]. Most interestingly, up-regulation of MX1, OAS1, OAS2 and IFIT3 mRNA in ruptured atherosclerotic plaques in human carotid arteries has been found [21].

By confirming the fundamentally different pathological functions of mCRP and pCRP, our findings may help to explain the clinical situation in vivo. The positive impact of pCRP on endothelial colony forming capacity of EPCs observed by us has also been described in healthy volunteers [9]. Under inflammatory conditions, such as atherosclerosis, pCRP is dissociated to mCRP [15] and thus exerts substantially different functions, potentially fuelling the inflammatory response, as suggested by the results of our whole-genome gene expression analysis that suggests pro-inflammatory properties for mCRP. The localised dissociation of pCRP to mCRP in inflammation could potentially mediate the negative biological effect on EPC numbers and function observed in patients with atherosclerosis [55, 67, 68].

Our study focussed on the in vitro effects of two isoforms of CRP on cord blood-derived EPCs to explore the functional, phenotypical and gene expression profiles of monomeric and pentameric CRP towards a rare cell population with potent pro-angiogenic capabilities involved in the natural repair of vascular damage. A limitation of our study is that the results obtained in our in vitro system cannot directly be translated into in vivo situation where a more complex system of different vascular cells acts in concert to maintain the vascular integrity, and each cell type may have a different response to the two isoforms of CRP. Furthermore, the use of cord blood-derived cells may bear the risk of a differential response in gene expression in comparison to adult EPCs. However, most of the functional properties of cord blood-derived EPCs resemble those of EPCs derived from adult donors, and EPCs derived from both young and old donors have successfully been used in the past to treat cardiovascular disease in animal models and human subjects [25, 46, 54, 70]. Interestingly, a recent report describes a weaker specific interferon-α-mediated immune response in cord blood-derived mononuclear cells as compared to mononuclear cells from adult donors [65]. Therefore, our gene expression analysis could also underestimate the pro-inflammatory effects of mCRP.

In conclusion, we provide evidence for a differential, highly opposing effect of monomeric and pentameric C-reactive protein on human umbilical cord blood-derived EPCs. Monomeric CRP induced the up-regulation of pro-inflammatory, interferon-responsive genes in EPCs, whereas pentameric CRP exhibited a primarily non-inflammatory gene response. These data support the concept that localised mCRP generation thus provides a means to locally regulate EPC function in vascular homoeostasis.

Acknowledgments

This work has been supported by the National Health and Medical Research Council (NHMRC) of Australia (Peter K), the Australian Research Council (future fellowship, Peter K), the German Research Foundation (AH 185/1-1, Ahrens I), the Henning and Johan Throne-Holst’s Foundation (Domeij H) and the German Academic Exchange Service DAAD (D/10/47670, Albrecht M). Dr. Martha Lappas is the recipient of an NHMRC RD Wright Fellowship (Grant No. 454777). The authors gratefully acknowledge the skilful technical assistance of Ruusu-Maria Merivirta, in the stem cell culture; the assistance of the Clinical Research Midwives Gabrielle Fleming, Astrid Tiefholz and Anne Beeston; and the Obstetrics and Midwifery staff of the Mercy Hospital for Women for their co-operation.

Supplementary material

395_2011_191_MOESM1_ESM.ppt (623 kb)
Supplementary figures (PPT 623 kb)
395_2011_191_MOESM2_ESM.doc (57 kb)
Supplementary methods (DOC 57 kb)

Copyright information

© Springer-Verlag 2011

Authors and Affiliations

  • I. Ahrens
    • 1
    • 2
  • H. Domeij
    • 2
  • S. U. Eisenhardt
    • 3
  • D. Topcic
    • 2
  • M. Albrecht
    • 1
    • 2
  • E. Leitner
    • 2
  • K. Viitaniemi
    • 2
  • J. B. Jowett
    • 4
  • M. Lappas
    • 5
  • C. Bode
    • 1
  • I. Haviv
    • 4
  • K. Peter
    • 2
  1. 1.Department of Cardiology and AngiologyUniversity Hospital FreiburgFreiburgGermany
  2. 2.Atherothrombosis and Vascular BiologyBaker IDI Heart and Diabetes InstituteMelbourneAustralia
  3. 3.Department of Plastic and Hand SurgeryUniversity Hospital FreiburgFreiburgGermany
  4. 4.The Blood and DNA Profiling FacilityBaker IDI Heart and Diabetes InstituteMelbourneAustralia
  5. 5.Department of Obstetrics and Gynaecology, Mercy Hospital for WomenUniversity of MelbourneMelbourneAustralia