CEC a heterogenous and rare cell type of the peripheral blood

CEC have been shown to contribute significantly to angiogenesis in ischemia, inflammation, wound healing and tumor progression [1]. In cancer patients CEC measurement in peripheral blood has been proposed a non-invasive tool to assess tumor angiogenesis [2] and monitor antiangiogenic therapies [3] CEC comprise a heterogenous cell population consisting of endothelial cells shed from the vessel wall [4], bone marrow-derived CEP [5] and endothelial precursors originating from monocytic cells [6] (figure 1). CEC are only a tiny subset of the mononuclear cell fraction of peripheral blood rendering their quantification a challenging task. Endothelial colony formation assays are labor-intensive, time-consuming, poorly standardised and may give rise preferentially to monocytic cells [7]. Currently, the most common technique applied for CEC quantification is multicolor flow cytometry. Based on this method absolute numbers of CEC reported in the literature vary greatly ranging from 0 to 7.900 CEC/mL in peripheral blood of healthy controls and from 5 to 39.000 CEC/mL under pathological conditions [2, 8]. Thus, there is an urgent need for more reliable and standardized methods for CEC quantification.

In this study both methods, flow cytometry and qPCR, were evaluated to compare the ability to detect mature peripheral blood-derived endothelial colony forming cells (ECFC; CEC phenotype) and cord blood-derived progenitors (CEP phenotype) spiked into PBMNC of healthy volunteers.

Phenotype analysis of cord blood progenitor cells and ECFC by flow cytometry

At first, the phenotypes of cord blood derived progenitor cells and ECFC was determined by flow cytometry (figure 2). As expected, ECFC were CD45^- CD31⁺ (figure 2A). Further subtyping revealed that cells were CD34^-KDR⁺. Moreover, cells were CD133^- and CD144⁺ (data not shown). Due to the two-step isolation procedure cord blood progenitor cells displayed a CD45^-/low CD34⁺ phenotype (figure 2B). The majority of CD34⁺ cells displayed a CD31⁺ CD133^- phenotype corresponding to hematopoietic precursor cells. In accordance with previous reports only a small subset of cells displayed a CD31⁺ CD133^low phenotype (5–10% of the CD34+ cells)[9]. Therefore, only few cord blood-derived progenitor cells were detectable in the CEP window during the subsequent spiking experiments. Moreover, progenitor cells were KDR^- and CD144^- (data not shown).

Detection of progenitors spiked into peripheral blood samples

In five independent experiments cord blood derived progenitors from different donors were spiked in triplicates into PBMNC from healthy volunteers at concentrations ranging from 0.001 to 1%. As depicted in figure 3A, flow cytometry allowed detection of progenitors with a CEP phenotype at concentrations ≥ 0.01% (mean: 4 cells/10⁶ PBMNC vs 1 cell in unspiked controls, p = 0.017).

Quantification of both CD34 and CD133 gene transcripts proved to be a reliable approach for detecting spiked progenitor cells in PBMNC samples with 10-fold greater sensitivity than flow cytometry. At a frequency of only 0.001% progenitor cells were detectable by a 3.9-fold increased gene expression for CD133 (SD 2.2) and a 3.4-fold increase for CD34 (SD 2.3) in comparison to unspiked controls (p = 0.03; figure 3B). Although linear, the increase of CD34 and CD133 gene transcripts was not proportional to the number of cells spiked, possibly due to technical reasons (e.g., cell clumping) or cell death as progenitor cells are more fragile compared to mature cells. Due to their low expression on progenitors as described above, CD144 and KDR gene expression was not analyzed.

Detection of ECFC spiked into peripheral blood samples

In five independent experiments freshly detached autologous ECFC were spiked into PBMNC samples of the respective donor at concentrations ranging from 0.001 to 1%. Using flow cytometry ECFC spiked into PBMNC samples were detectable at frequencies ≥ 0.01% (mean: 68 cells/10⁶ PBMNC vs 20 cells in unspiked controls, p = 0.004; figure 4A).

QPCR was at least 10-fold more sensitive than flow cytometry for ECFC detection. Indeed, at a frequency of 0.001% relative gene expression of KDR was increased 48.3-fold (SD 52.1; p = 0.027) and that of CD144 40.8-fold (SD 58.5; p = 0.04) compared to unspiked controls (figure 4B). Due to their low expression level on ECFC as described above, CD34 and CD133 gene transcripts were not determined in this setting.

Detection of CEC/CEP in peripheral blood of cancer patients and healthy volunteers

Applying 4-color flow cytometry no significant differences concerning CEC/CEP numbers were found between cancer patients and age-matched healthy controls (figure 5A). We then applied qPCR for quantifying gene transcripts of endothelial progenitor cell markers CD34, CD133 and endothelial cell markers KDR and CD144. Overall, gene transcripts for CD34 and CD144 were found abundantly expressed (i.e., >100 copies/μg RNA) in the total study cohort whereas only very few KDR or CD133 gene transcripts were detectable (figure 5B). Subgroup analysis showed no influence of gender, age or tumor stage on any of the markers studied. No significant elevation of any of the endothelial marker gene transcripts determined was found in locally advanced rectal cancer patients. In contrast, in the cohort of patients with metastatic NSCLC a significantly increased CD34 and KDR gene expression was found (p = 0.028 and p = 0.002, respectively).

Acquisition of blood samples

Isolation of progenitor cells from cord blood

Human umbilical cord blood samples (n = 10) were obtained at birth from full-term newborns. Blood samples were collected in heparinized tubes and stored at 8°C no longer than 12 h before flow cytometric analysis and mRNA extraction, respectively. MNC were isolated by Ficoll density gradient centrifugation (Lymphoprep^®, Nycomed, Norway). Progenitor cells were enriched by a two step immunomagnetic bead separation protocol by negative selection for CD45 and subsequent positive selection for CD34⁺ (CD34 isolation kit, CD45 microbeads, Miltenyi Biotec). Progenitor cells were spiked into PBMNC of healthy volunteers at frequencies ranging from 0.001 to 1%.

Generation of endothelial colony forming cells (ECFC) from peripheral blood

Autologous ECFC cultures were generated as described previously [7, 13]. Briefly, PBMNC from five healthy donors were isolated by Ficoll density gradient centrifugation, resuspended in EGM-2 medium (Cambrex) and placed into a six-well-plate coated with type I collagen (from kangaroo, Sigma-Aldrich). After 24 h, non-adherent cells were removed by changing the medium. Autologous ECFC were spiked into PBMNC of the corresponding donor at frequencies ranging from 0.001 to 1%.

Flow cytometry

Flow cytometric detection and enumeration of of CEC/CEP was carried out according to a recently published protocol [14]. Except for the anti-CD34 antibody monoclonal antibodies were chosen exactly as suggested by the authors. In brief, after Fc-blocking (Fc-receptor blocking antibody, Miltenyi Biotec) PBMNC were incubated in triplicates with antibodies specific for CD31-FITC (BD Pharmingen), CD34-PC7 (Beckman Coulter), CD45 PerCP (BD Pharmingen), CD133-PE (Miltenyi Biotec) or VEGF-R2 (KDR)-PE (R&D Systems). Appropriate fluorochrome-conjugated isotype-matched murine IgG antibodies (BD Pharmingen) were used as controls for each staining procedure. After incubation for 30 min at 4°C, cells were washed, resuspended in 300 mL PBS and analyzed in a Cytomics-FC-500 cytometer using the Cytomics RXP-Software (Beckman Coulter; figure 2). CEC were defined as CD31⁺/CD34⁺/CD45^-/CD133^- and CEP were defined as CD31⁺/CD34⁺/CD133⁺/CD45^-/low cells. All experiments were carried out in triplicates with analysis of at least 5 × 10⁵ cells per run.

Quantitative PCR

External standards

For absolute quantification CD34, CD133, KDR and CD144 cDNAs were subcloned by the use of the PCR-Script cloning kit (Stratagene). Plasmid copies were calculated as follows: amount (copies/μL) = 6 × 10²³ (copies/mol) × concentration (g/μL)/MW (g/mol). Standard curves were generated by logarithmic dilutions of triplicates of the amplicon-containing plasmids in a complex matrix of COS-7 cDNA. These external standard curves were used to calculate copy numbers per μg total RNA in all PBMNC samples using the iCycler software (BioRad).

Statistical analysis

Statistical analysis was performed with the GraphPad Prism 5 software for Windows. All tests of statistical significance were two-sided. Student's t-test was used for the analysis of spiking experiments. Kruskal Wallis H test was applied to study differences between healthy controls, rectal cancer patients and NSCLC patients.