Application of Polychromatic Flow Cytometry to Identify Novel Subsets of Circulating Cells with Angiogenic Potential

Abstract

Defining whether human circulating pro-angiogenic cells represent a subset of the hematopoietic system and express CD45 or are hematopoietic derivatives that do not express CD45 (and are called endothelial progenitor cells) remains controversial. We have previously developed a polychromatic flow cytometry (PFC) protocol to isolate subsets of hematopoietic cells and we now identify the circulating pool of CD34⁺CD45^dim cells representing functional circulating hematopoietic stem and progenitor cells (CHSPCs) that can be separated on the basis of AC133 expression and report that the AC133⁺ subset of the CHSPCs enhances the growth of tumor blood vessels in vivo in immunodeficient mice. In addition, the ratio of AC133⁺ pro-angiogenic CHSPCs to AC133⁻ non-angiogenic CHSPCs unambiguously correlates with the severity of the clinical state of patients with peripheral arterial disease. In sum, a PFC protocol validated via in vitro and in vivo analyses, can be used to interrogate the roles of human hematopoietic elements in the growth and maintenance of the vasculature.

INTRODUCTION

Though a powerful technique, conventional flow cytometric protocols are highly susceptible to and in fact require the operator to make subjective decisions in the process of data acquisition and interpretation [1, 2]. As additional fluorescence values are increasingly added to define more cell parameters, the advent of polychromatic flow cytometry (PFC) emerged to implement the rigorous controls and standards necessary to discriminate multi-parametric data (for more detailed information on the technical advances, see the supporting information and Table S1) [1]. Nowhere is this level of scrutiny of greater concern than in rare and dim event analysis where great care must be taken to ensure the reliability of collected data [3]. Besides occurring in frequencies on the cusp of reproducible detection (0.001–0.1%), cells of endothelial and hematopoietic origin proposed to be engaged in angiogenesis are often discriminated using a combination of antigens with low, dull, or a continuum of cell surface expression [4, 5]. Limitations of conventional flow cytometry, compounded by antigen promiscuity between the endothelial and hematopoietic lineages has necessitated consideration of the newer methods and approaches of PFC in defining the various circulating cell subsets involved in neoangiogenesis [6–10]. For example, current debate often centers on CD45 expression in putative endothelial progenitor cell (EPC) subsets [11, 12]. Previously, the distinction between a CD34⁺ progenitor cell that expressed CD45 versus one that was CD34⁺ and CD45⁻ was deemed critical for the discrimination of myeloid cells that mimicked endothelial morphology in culture (colony forming unit-Hill; CFU-Hill) versus bona fide endothelial cells found in blood that form vessels upon implantation following in vitro expansion (endothelial colony forming cells; ECFCs). While the original putative EPCs were first described as CD45⁻CD34⁺AC133⁺KDR⁺ cells, recent use of flow cytometers with more channels of resolution have since revealed that the EPCs described as CD45⁻ may in fact be CD45^dim [11]. However, these debates exist because the conventional flow cytometry methods employed to analyze these cell populations have lacked the controls necessary to accurately depict the rare event and low antigen expression profiles desired [13].

A PFC protocol was applied to discriminate previously undetected phenotypic and functional heterogeneity within the commonly referenced circulating progenitor cells (CPCs)[14]. Our PFC profile utilized a panel of antigens including rodent monoclonal antibodies to human CD45, CD34, CD31, and AC133, reagents to exclude false positive events, and methods to improve data analysis. Using PFC, the CPC subset identified as CD45^dimCD34⁺CD31⁺ and heterogeneous in AC133 expression is now reported to be comprised of circulating hematopoietic stem and progenitor cells (CHSPCs) that engraft in NOD/SCID mice of which a subset display pro-angiogenic tumor growth promoting activity in vivo. Thus, the application of PFC techniques and approaches to the EPC field have permitted clarification of the cells involved in neoangiogenesis and may now permit broader application of these cells to therapeutic and diagnostic applications.

MATERIALS AND METHODS

Blood Samples

Peripheral blood (PB) samples (16–32mls) were collected from 20 healthy adult donors (10 male and 10 female, age range 20–40 years) and umbilical cord blood (CB) samples (20–100mls) were collected from 15 full-term newborns. The Institutional Review Board at the Indiana University School of Medicine approved all protocols, informed consent was obtained from adult donors, and cord blood collection was deemed exempt. Granulocyte colony stimulating factor (G-CSF) mobilized peripheral blood (mPB) CD34⁺ cells were kindly provided through a Program of Excellence in Gene Therapy grant from Shelly Heimfeld at the Fred Hutchinson Cancer Research Centre, Seattle, WA, USA.

Clinical Samples and Subject Characteristics

PB samples (16mls) were collected from 9 patients (pts) with peripheral artery disease (PAD) (6 male and 3 female, age range 50–81) along with age and gender matched controls. PAD pts ranged in Rutherford class 1–6, with co-morbidity related to coronary artery disease (CAD) in 4 patients and chronic obstructive pulmonary disease (COPD) in 3 pts.

Isolation of Mononuclear Cells

Blood was diluted 2:1 with phosphate buffered saline without calcium or magnesium (PBS, Invitrogen, Grand Island, NY, USA) and underlaid with Ficoll (GE Healthcare, USA). Mononuclear cells (MNCs) were isolated by centrifuging blood at 740g for 30 minutes at room temperature. The MNCs were removed and washed two times in PBS without calcium or magnesium (Invitrogen, Grand Island, NY, USA) with 2% fetal bovine serum (FBS, Hyclone, Logan, UT, USA). Cells were counted on a hemacytometer.

Antibodies and Staining Reagents

In order to resolve the rare and/or dim populations of interest, specific antigen and fluorochrome conjugate coupling was optimized for the six-antibody plus viability marker staining panel described below as previously described [1, 14–16]. The following primary conjugated monoclonal antibodies were used: anti-human CD31 fluoroscein isothyocyanate (FITC, BD Pharmingen, San Diego, CA, USA, cat. no. 555445), anti-human CD34 phycoerythrin (PE, BD Pharmingen, cat. no. 550761), anti-human AC133 allophycocyanin (APC, Miltenyi Biotec, Auburn, CA, USA, cat. no. 130-090-826), anti-human CD14 PECy5.5 (Abcam, Cambridge, MA, USA, cat. no. ab25395), anti-human CD45 APC-AlexaFluor (AF) 750 (Invitrogen, cat. no. MHCD4527), anti-human CD235a (glyA, R&D Systems, Minneapolis, MN, USA, cat. no. MAB1228) conjugated to Pacific Blue (PacB, Invitrogen), anti-human CD3 FITC (BD Pharmingen, cat. no. 555339), anti-human CD4 FITC (BD Pharmingen, cat. no. 555346), anti-human CD7 FITC (BD Pharmingen, cat. no. 555360), anti-human CD8 FITC (BD Pharmingen, cat. no. 555634), anti-human CD10 FITC (BD Biosciences, cat. no. 340925), anti-human CD11b PECy7 (BD Pharmingen, cat. no. 557743), anti-human CD13 (BD Pharmingen, cat. no. 558744), anti-human CD19 PE (BD Pharmingen, cat. no. 555413), anti-human CD33 APC (BD Pharmingen, cat. no. 551378), anti-human CD34 PECy7 (BD Biosciences, cat. no. 348791) anti-human CD41a APC (BD Pharmingen, cat. no. 559777), CD45RA FITC (BD Pharmingen, cat. no. 555488), CD56 CD71 HLA-DR FITC (BD Pharmingen, cat. no. 555811) IgG FITC (BD Pharmingen, cat. no. 555748), IgG PE (BD Pharmingen, cat. no. 555749), IgG APC (BD Pharmingen, cat. no. 555751), IgG PECy5.5 (Invitrogen, cat. no. MG118), IgG APC-AF750 (Invitrogen, cat. no. MG127), IgG PacB (Invitrogen, cat. no. S-11222), the amine reactive viability dye, ViViD (Invitrogen), and DAPI (Invitrogen).

PFC Immunostaining

A total of 10⁷ PB MNCs were suspended in 720μl PBS with 2% FBS and incubated for 10 minutes at 4°C with 180μl human Fc blocking reagent (Miltenyi Biotec). Cells were aliquoted into 9 tubes and stained with the following antibodies. Subsequently, 100μl of the cell suspension was distributed into nine sample tubes with the following pre-titered antibodies: (1) unstained; (2) Isotypes: 4μl IgG FITC, 15μl IgG PE, 10μl IgG APC, 10μl IgG PECy5.5, 5μl IgG APC-AF750 and 4μl IgG PacB; (3) FITC FMO: 15μl CD34 PE, 10μl AC133 APC, 10μl CD14 PECy5.5, 5μl CD45 APC-AF750, 4μl glyA PacB and 1μl ViViD; (4) PE FMO: 4μl CD31 FITC, 10μl AC133 APC, 10μl CD14 PECy5.5, 5μl CD45 APC-AF750, 4μl glyA PacB and 1μl ViViD; (5) APC FMO: 4μl CD31 FITC, 15μl CD34 PE, 10μl CD14 PECy5.5, 5μl CD45 APC-AF750, 4μl glyA PacB and 1μl ViViD; (6) PECy5.5 FMO: 4μl CD31 FITC, 15μl CD34 PE, 10μl AC133 APC, 5μl CD45 APC-AF750, 4μl glyA PacB and 1μl ViViD; (7) APC-AF750 FMO: 4μl CD31 FITC, 15μl CD34 PE, 10μl AC133 APC, 10μl CD14 PECy5.5, 4μl glyA PacB and 1μl ViViD; (8) V450 Channel FMO: 4μl CD31 FITC, 15μl CD34 PE, 10μl AC133 APC, 10μl CD14 PECy5.5, and 5μl CD45 APC-AF750; and (9) full panel: 4μl CD31 FITC, 15μl CD34 PE, 10μl AC133 APC, 10μl CD14 PECy5.5, 5μl CD45 APC-AF750, 4μl glyA PacB and 1μl ViViD (six-antibody/viability marker panel). Table S2 is included in the supporting information to better illustrate the staining panel. Cells were incubated with antibodies for 30 minutes at 4°C, washed twice in PBS with 2% FBS, and fixed in 300μl 1% paraformaldehyde (Sigma Aldrich, St. Louis, MO, USA). Additionally, anti-mouse Ig BD CompBeads (BD Biosciences, Bedford, MA, USA) were stained with each of the individual test antibodies to serve as single-color compensation controls. Prior to use, each lot of antibody was individually titered as previously described [17] to determine the optimal staining concentration. In some experiments, cells were incubated with either glyA PacB or ViViD to determine the individual contribution of RBCs or dead/apoptotic cells, respectively.

Flow Cytometry Acquisition and Sorting

Stained fixed MNC samples were acquired on a BD LSRII flow cytometer (BD, Franklin Lakes, NJ, USA) equipped with a 405nm violet laser, 488nm blue laser and 633nm red laser (for filter specifications see Table S3) [14]. Prior to acquiring any data, photomultiplier tube (PMT) voltages were calibrated to the highest signal to background ratio as previously described [18]. At least 300,000 events were acquired for each sample. Data was acquired uncompensated, exported as FCS 3.0 files and analyzed using FlowJo software, version 8.7.3 (Tree Star, Inc., Ashland, OR, USA).

For sorting, unfixed MNC samples were run on a BD FACSAria flow cytometer (BD) equipped with a 407nm violet laser, 488nm blue laser and 630nm red laser. PMT voltages were also calibrated and using single stained compensation beads a compensation matrix was created. Cells were collected into sterile PBS supplemented with 2% FBS. Data was exported as FCS 3.0 files and also analyzed using FlowJo software, version 8.7.3 (Tree Star).

Colony Assays

To assess the hematopoietic progenitor colony forming potential of CD31⁺CD34^brightCD45^dimCD133⁺ cells from PB, 500 freshly sorted cells or 10,000 MNCs were suspended in 0.66% to 1.0% agar (Becton Dickinson) in the presence of 1000U/ml human interleukin (IL)-1α, 200U/ml human IL-3, 100ng/ml human macrophage colony stimulating factor (M-CSF), and 100ng/ml human stem cell factor (SCF) (all from Peprotech, Rocky Hill, NJ, USA) as previously described [19]. Cells were plated in 35mm Petri dishes in triplicate and scored for low proliferative potential – colony forming cells (LPP) and high proliferative potential – colony forming cells (HPP-CFCs) on day 14. Sorted sub-populations were also assayed for the presence of multi-potential granulocyte, erythroid, macrophage, megakaryocyte progenitors (i.e. CFU-GEMMs) using MethoCult® GF H4434, Complete Methylcellulose Kit (StemCell Technologies, Vancouver, BC, Canada) according to the manufacturer’s protocol.

Matrigel Tube Forming Assays

To assess the presence of functional endothelial cells within sub-populations of PB, freshly sorted CD31⁺CD34^brightCD45^dimCD133⁺ cells (3,500–5,000 per well) were seeded onto Matrigel-coated (BD Biosciences) 96-well plates as previously described [6, 20]. Wells were examined by visual microscopy every two hours for capillary-like tube formation. Early passage cultured CB ECFCs were used as a positive control.

qPCR

For pre-amplification, 1,000 to 3,000 sorted cells were first lysed and reverse transcribed exactly according to the manufacturer’s protocol using the TaqMan PreAmp Cells-to-CT Kit (Applied Biosystems, Foster City, CA, USA). All reverse transcription (RT) reactions were performed for 60 minutes at 37°C and 5 minutes at 95°C. Pooled cDNA was pre-amplified with TaqMan Gene Expression Assays (Applied Biosystems) with primers for CD34, CD45, CD31, AC133 and ACTB (β-actin, Applied Biosystems). Pre-amplification was performed for 10 minutes at 95°C, 10 cycles of 15 seconds at 95°C and 4 minutes at 60°C. A quantitative real-time PCR (qPCR) analysis was performed with the ABI PRISM 7500 sequence detection system (Applied Biosystems). Cycling conditions consisted of a 2 minute hold at 50°C for uracilo-N-glycosylase degradation, 10 minute hold at 95°C for enzyme activation, 40 cycles of 15 second denaturation at 95°C, 1 minute annealing and elongation at 60°C. Relative quantification of triplicate samples was performed using the delta-C_T method and expressed as fold increase relative to ACTB.

Cytospin Analysis

Cellular content of fluorescence activated cell sorting (FACS) derived CPC populations was evaluated by counting, cytospin preparation and Wright-Giemsa staining as previously described [21]. Identification of cell types was done by visual inspection under 100× magnification and photomicrographs of cytospins were taken with an Olympus DP20 on an Olympus BX50 microscope.

Mice

NOD/SCID mice, 6–8 weeks old, were housed according to protocols approved by the Laboratory Animal Research Facility and adhered strictly to National Institutes of Health guidelines and protocols were approved by Indiana University Animal Care and Use Board.

Transplantation of NOD/SCID Mice

All animals were given a sub-lethal dose of 300cGy total body irradiation 4 hours before transplantation. The mPB CD34⁺ cells (10⁵ per mouse), sorted CD31⁺CD34^brightCD45^dimAC133⁺ sub-population (i.e. CPCs) from mPB CD34⁺ cells (10⁵ per mouse), or cells not contained in the CD31⁺CD34^brightCD45^dimAC133⁺ sort gate (i.e. non-CPCs) (2.5×10⁴ per mouse) were re-suspended in PBS and transplanted by tail vein injection. To assess engraftment, mice were sacrificed 8–12 weeks after transplantation and both femurs were flushed with a total of 2×10⁶ cells collected and stained with anti-human CD45 and CD34 antibodies. Approximately 500,000 events per sample were collected on a BD LSRII flow cytometer. Analysis was performed with FlowJo software version 8.7.3.

Determination of Human CPC Function in a Melanoma Xenograft Model

NOD.CB17-Prkdc^scid/J (NOD/SCID) mice were subcutaneously injected with 2×10⁶ C32 human melanoma cells (ATCC) and tumor growth monitored. Once tumors reached ~50mm³, mice were injected with 5×10⁴ AC133⁺ CPCs, AC133⁻ CPCs (non-CPCs), bulk CD34⁺ cells, or vehicle control (PBS). Tumor growth was monitored by caliper and the volume determined by the following formula: mm³ = (width)² × length × 0.5. The fold increase in tumor growth was determined by comparing tumor volume over time to the base line tumor volume. At the end of the experiment, mice were euthanized, tumors harvested, and the weight of each tumor determined. Data are presented as the mean±sem. Statistical significance was determined using a 2-sided student’s t-test to calculate p values.

Statistical Analysis

Statistical analysis was performed using GraphPad Prism software, version 5.01 for Windows (GraphPad Software, San Diego, CA, USA). Data was tested for normality using the D’Agostino-Pearson normality test (alpha=0.05), and normal data sets were compared using two-tailed Student’s t test or one-way ANOVA.

RESULTS

Frequency Analysis and Characterization of CD31⁺CD34^brightCD45^dimAC133⁺ cells (CPCs)

We performed a similar methodological comparison of the flow cytometry methods to enumerate the CD31⁺CD34^brightCD45^dimAC133⁺ putative CPCs [4, 5, 12, 14, 22]. We initially isolated PB MNC samples from 10 healthy, young adult volunteers and stained the cells with the six monoclonal antibodies (CD34, CD45, CD31, AC133, glyA, and CD14) and the viability marker (ViViD) with ‘fluorescence minus one’ (FMO) controls or isotype controls as described [4, 14]. Stained samples were acquired on a digital BD LSRII flow cytometer and assessed for CD31⁺CD34^brightCD45^dimAC133⁺ (CPCs) events using two different analysis schemas as shown in Figure 1. First, stained MNCs were analyzed in conventional logarithmic dot plots (Fig. 1a–d) [4] and CPC identification was determined by placement of population gates, which were based on isotype controls exactly as previously described for conventional flow cytometry (Fig. 1a–d) [4]. Data were manually compensated using singly stained cell controls as reported [4]. In contrast, MNCs from the same donor were analyzed in bi-exponential contour plots to identify CD31⁺CD34^brightCD45^dimAC133⁺ cells after exclusion of contaminating monocytes, red blood cells (RBCs) and dead cells (Fig. 1e–i)[14]. Prior to analysis, automated compensation was applied based on single-color bead controls. To objectively identify CPCs, regional gates were applied based on the use of proper FMO gating controls as previously described [14].

Figure 1. Frequency analysis of CD31⁺CD34^brightCD45^dimAC133⁺ cells.

Two strategies for frequency analysis of CD31⁺CD34^brightCD45^dimAC133⁺ cells from a ficoll MNC preparation of PB stained with the six-antibody/viability panel are shown. In the first strategy (a–d), manually compensated data collected on a digital flow cytometer were visualized in plots with logarithmic scaling. MNCs (red gate in a) were identified on a FSC/SSC plot and sub-gated onto a bi-variant antigen plot to identify CD34^brightAC133⁺ cells (dark blue gate in b). CD34^brightAC133⁺ MNCs were further sub-gated to identify the CD45^dim sub-population (light blue gate in c). CD31 expression on the resulting CD34^brightCD45^dimAC133⁺ MNCs was confirmed on a CD31 histogram (d). In the first strategy (a–d), gate boundaries were set using Boolean gating and negative isotype controls. In the second strategy (e–i), uncompensated data was collected on a digital flow cytometer, compensated after acquisition by using software, and visualized in plots with bi-exponential scaling. MNCs (red gate in a) were identified on FSC/SSC plot and then CD14⁻ cells (orange gate in e) were identified. All CD14⁻ cells were then assessed for viability and glyA expression (f). CD14⁻glyA⁻ViViD⁻ cells (pink gate in f) were sub-gated onto a bi-variant antigen plot to identify CD14⁻glyA⁻ViViD⁻CD34^brightAC133⁺ cells (dark blue gate in g). Viable CD14⁻glyA⁻CD34^brightAC133⁺ cells are further sub-gated to identify the CD45^dim sub-population (light blue gate in h). CD31 expression on the resulting viable CD14⁻glyA⁻CD34^brightCD45^dimAC133⁺ cells was confirmed on a CD31 histogram (i). In the second strategy (e–i), FMO gating controls were used to set gate boundaries.

To compare the reproducibility and margin of error between the two methods, PB MNCs were harvested, and CPCs were identified utilizing the methods outlined in Figure 1. In samples analyzed using the four-antibody panel and logarithmic dot plots (Fig. 1a–d), 0.290±0.218% (mean±s.d., n=10, range 0.170–0.900) of gated MNCs were CD31⁺CD34^brightCD45^dimAC133⁺. In comparison, when cell preparations from the same donors were analyzed using bi-exponential contour plots and FMO gating controls with the exclusion of monocytes, RBCs, and dead/apoptotic cells (Fig. 1e–i), CD31⁺CD34^brightCD45^dimAC133⁺ cells constituted only 0.134±0.0347% (mean±s.d., n=10, range 0.0800–0.200; logarithmic method vs. bi-exponential method, p=0.0380 by two-tailed, unpaired Student’s t test) of viable glyA⁻CD14⁻ cells. Examination of Figure 1g reveals a distinct CD31⁺CD34^brightCD45^dimAC133⁻ population that is nearly indistinguishable from the neighboring CD31⁺CD34^brightCD45^dimAC133⁺ population in Figure 1b. Exclusion of the unwanted myelo-erythroid events, the use of contour plots and bi-exponential display for data depiction, and selection of an antigen panel for maximal resolution of dull populations has culminated in the enhanced resolution depicted in Figure 1g, facilitating the detection of previously unreported heterogeneity within the putative CPC population [14]. Moreover, Figure 1c depicts a CD45^bright population (dark blue) due to monocyte and dead cell contamination that has been eliminated in the equivalent PFC plot (Fig. 1h). FMO gating control analysis provides the only objective means of distinguishing between a broad negative population and an adjacent dull positive subset and in this case, confirms the visually apparent demarcation of AC133 expression [14]. The combined frequencies of the CD31⁺CD34^brightCD45^dimAC133⁻ and CD31⁺CD34^brightCD45^dimAC133⁺ populations illustrated in Figure 1g, confirm the observed frequency gated by conventional methodology in Figure 1b, 0.310±0.024% (mean±s.d., n=10, range 0.270–0.400; logarithmic method vs. bi-exponential method, p=0.7973 by two-tailed, unpaired Student’s t test). To confirm the antigen profile, we performed quantitative RT-qPCR and demonstrated that FACS purified and isolated CPCs as gated in Figure 1g transcribe mRNA for the cell surface antigens CD34, CD31 and AC133 (Fig. 2a). Importantly, CPCs also transcribe mRNA for CD45 (Fig. 2a). Thus, the putative CPC population enumerated by conventional flow cytometry methods is comprised of two phenotypically distinct cellular subsets. Application of PFC analysis for CPC enumeration reveals heterogeneity [14] and produces a smaller range of values with a lower standard deviation, which is critical for clinical comparative studies.

Figure 2. Characterization of CD31⁺CD34^brightCD45^dimAC133⁺ cells.

(a) Semi-quantitative RT-PCR analysis of CD34, CD45, CD31 and AC133 expression in PB MNCs (■) and sorted CD31⁺CD34^brightCD45^dimAC133⁺ cells (□). (b) Representative photomicrographs of cytospin from FACS CD31⁺CD34^brightCD45^dimAC133⁺ cells stained with Wright-Giemsa. Original magnification, 100×. (c) Representative photomicrographs of CFU-GEMM colonies derived from culture of sorted PB CD31⁺CD34^brightCD45^dimAC133⁺ cells plated in CFU-GEMM colony assays. Scale bar represents 200μm. For all characterization assays, similar results were seen in 3 independent experiments using cells from different donors.

Though previous reports of heterogeneous CPCs found correlations with extent of tumor progression and CVD risk, the functional identity of these progenitors is unknown [23]. Therefore, we performed experiments to determine the functional phenotype of CPCs in human PB identified with PFC. To better ascertain the identity of the CPCs, we isolated, pelleted, re-suspended, and deposited the CPCs onto slides and performed Wright-Giemsa staining on the CD31⁺CD34^brightCD45^dimAC133⁺ cells. Remarkably, morphological analysis revealed hematopoietic blast cells or progenitor cells, which potentially represent hematopoietic stem and progenitor cells (HSPCs) (Fig. 2b). Based on expression of multiple hematopoietic cell surface antigens (CD34, CD45 and AC133) and cellular morphology via immunocytochemistry, we next tested whether CPCs displayed functional properties of primitive HSPCs or EPCs in colony forming assays. Both CD31⁺CD34^brightCD45^dimAC133 ⁻ and CD31⁺CD34^brightCD45^dimAC133⁺ CPC populations formed primitive hematopoietic progenitor cell colonies including LPP-CFCs, HPP-CFCs, and CFU-GEMMs (Fig. 2c) at a frequency range of 1:7–1:20 in each population. Neither population of CPCs yielded ECFCs in established in vitro clonal assays or formed capillary-like tubes with lumens in Matrigel^™ and neither population formed vessels in vivo.

Since both CPC populations formed primitive multi-lineage hematopoietic cell colonies and expressed HSPC antigens, we tested whether these populations in fact contained NOD/SCID engrafting hematopoietic stem cells (HSCs). NOD/SCID mice were sub-lethally irradiated and subsequently transplanted intravenously with purified mPB CD34⁺ cells (positive control for HSPCs), CD31⁺CD34^brightCD45^dimAC133⁺ or CD31⁺CD34^brightCD45^dimAC133⁻ CPCs. Mice were sacrificed after 8–12 weeks, and human cell engraftment was measured in the mouse bone marrow (BM) by the presence of human CD45⁺ cells using species-specific monoclonal antibodies. Transplanted mouse BM was also analyzed for the presence of human CD19, CD33 and CD34 expressing cells; markers used to determine multi-lineage potential of engrafted human cells in NOD/SCID mice. Strikingly, mice transplanted with either CD31⁺CD34^brightCD45^dimAC133⁺ or CD31⁺CD34^brightCD45^dimAC133⁻ CPCs demonstrated multi-lineage engraftment, which is the hallmark of transplantable NOD/SCID repopulating HSPCs (Table 1). Thus, the previously identified human CPCs are comprised of a heterogenous mixture of HSPCs and would be best described as circulating HSPCs (CHSPCs) rather than the less descript term CPCs.

Table 1.

Engraftment analysis of NOD/SCID mice transplanted with mPB CD34⁺ cells, sorted CPCs or sorted non-CPCs.

	% CD45+	% of CD45+ cells that are CD34+	% of CD45+ cells that are CD33+	% of CD45+ cells that are CD19+
CPC Mouse 1	6.00	10.7	12.9	14.4
CPC Mouse 2	4.12	18.6	12.6	17.4
CPC Mouse 3	4.78	9.22	13.2	14.2
non-CPC Mouse 1	6.03	10.9	15.7	17.3
non-CPC Mouse 2	6.85	12.0	15.3	16.7
mPB CD34+ Mouse 1	4.66	15.5	15.6	13.8
mPB CD34+ Mouse 2	4.35	15.2	13.2	14.1
mPB CD34+ Mouse 3	3.74	22.3	19.7	27.2
mPB CD34+ Mouse 4	3.83	11.3	16.1	18.5

CHSPC Heterogeneity Corresponds with Disparate Angiogenic Potential

While the CD34⁺ cells, CD31⁺CD34^brightCD45^dimAC133⁺ CHSPCs or CD31⁺CD34^brightCD45^dimAC133⁻ CHSPCs all showed similar rates of NOD/SCID engraftment, we wanted to compare their capacity for promoting angiogenesis in an in vivo model. Thus, NOD/SCID mice bearing human melanoma xenografts were intravenously injected with equal numbers of CB CD34⁺ cells, CD31⁺CD34^brightCD45^dimAC133⁺ CHSPCs or CD31⁺CD34^brightCD45^dimAC133⁻ CHSPCs and tumor growth was monitored over time in each cohort (Fig. 3a). Surprisingly, mice injected with CD31⁺CD34^brightCD45^dimAC133⁺ CHSPCs demonstrated a 23.12±0.15% (mean±sem n=8, range 18.35–29.00) fold increase in tumor growth as compared to tumor bearing animals treated with CD31⁺CD34^brightCD45^dimAC133⁻ CHSPCs (7.20±0.15% mean±sem n=8, range 5.31–9.87), the parental population of CD34⁺ cells (5.98±0.23% mean±sem n=8, range 4.99–7.59) and PBS control (9.17±0.14% mean±sem n=8, range 5.30–11.96; AC133⁺ CHSPCs vs. PBS, p=0.001 by two-tailed, unpaired Student’s t test). When the explanted tumors were removed from the host mice in each cohort and weighed, animals injected with the CD31⁺CD34^brightCD45^dimAC133⁺ CHSPCs tumors were significantly heavier (0.89±0.04g mean±sem n=8) than the tumors removed from animals that were injected with PBS (0.51±0.06g mean±sem n=8) or CD34⁺ cells (0.53±0.06g mean±sem n=8; AC133⁺ CHSPCs vs. PBS or CD34⁺ cells, p<0.001) (Fig. 3b). Furthermore, it was apparent that the animals injected with the CD31⁺CD34^brightCD45^dimAC133⁺ CHSPCs displayed a greater tumor vasculature that may be the most plausible explanation for the enhanced tumor volume measured in this group of animals. Though the transplanted CD31⁺CD34^brightCD45^dimAC133⁺ CHSPCs promoted tumor angiogenesis and tumor growth, we found no evidence of human CD31, CD33 or CD34 expressing cells/endothelium in the murine tumor vessels. Thus, CD31⁺CD34^brightCD45^dimAC133⁺ CHSPCs are pro-angiogenic and accelerate tumor growth in a statistically significant manner and are functionally distinct from the non-angiogenic CD31⁺CD34^brightCD45^dimAC133⁻ CHSPCs. Moreover, lineage phenotyping of pro-angiogenic and non-angiogenic CHSPCs revealed further heterogeneity (Fig. 3c). Pro-angiogenic CHSPCs expressed a preponderance of myeloid cell surface markers (CD11b, CD13, CD33) while the non-angiogenic CHSPCs displayed more lymphoid cell surface markers (CD3, CD4, CD7, CD10, CD56) (Fig. 3c). This data demonstrates that both pro-angiogenic and non-angiogenic CHSPCs are not ECFCs or CPCs with endothelial potential but are comprised of hematopoietic progenitor cells, myeloblasts, and HSCs with NOD/SCID repopulating ability.

Figure 3. CPC heterogeneity corresponds with disparate angiogenic potential and lineage markers.

(a) Intravenous injection of purified CPCs results in significant increases in melanoma xenograft growth. NOD/SCID mice bearing human melanoma xenografts were intravenously injected with CD31⁺CD34^brightCD45^dimAC133⁺ CPCs, CD31⁺CD34^brightCD45^dimAC133⁻ CPCs, bulk CD34⁺ cells or PBS control (n=8 per cohort) and tumor growth monitored over time in each cohort. The fold increase in tumor growth was determined by comparing tumor volume over time to base line tumor volume. *p<0.001, CD31⁺CD34^brightCD45^dimAC133⁺ CPC versus CD31⁺CD34^brightCD45^dimAC133⁻CPC, CD34 or PBS control. (b) At the end of the study, tumors from each cohort of mice were harvested and weighed. *p<0.001, CD31⁺CD34^brightCD45^dimAC133⁺ CPC versus bulk CD34⁺ cells or PBS control. (c) CD31⁺CD34^brightCD45^dimAC133⁺ CPCs (■) and CD31⁺CD34^brightCD45^dimAC133⁻ CPCs (□) were assessed for percent expression of lymphoid (CD3, CD4, CD7, CD8, CD10, CD19 and CD45RA) and myeloid lineage markers (CD11b, CD13, CD33 and CD71).

Ratio of Circulating Progenitor Subsets Denotes Disease State in Peripheral Arterial Disease Patients

The disparate angiogenic potential of the two CHSPC subsets (as functionally determined in the tumor xenograft model) observed within the putative CPC population led us to hypothesize that these two progenitor fractions may regulate different aspects of vascular homeostasis. To test this hypothesis, we measured the frequency of the pro-angiogenic CHSPCs versus the non-angiogenic CHSPCs in a patient population with vascular dysfunction leading to PAD. In patients with diagnosed PAD we have identified a significant decrease in the ratio of pro-angiogenic CHSPCs to non-angiogenic CHSPCs (0.79±0.16050% mean±sem, n=9, range 0.140–1.52) as compared to age and gender matched control subjects (1.81±0.09433% mean±sem, n=9, range 1.231–2.130; healthy vs. PAD, p=0.0001 by two-tailed, unpaired Student’s t test) (Fig. 4a–c). Interestingly, patients with PAD and healthy controls were indistinguishable when the total CHSPC population (CD31⁺CD34^brightCD45^dimAC133^+/−) was enumerated and compared. These data suggest that future studies to define biologic changes in the pro-angiogenic fraction of CHSPCs may be informative as to deficiencies they may exhibit in vascular repair in subjects with PAD.

Figure 4. Ratio of circulating progenitor subsets denotes disease state in peripheral arterial disease patients.

Representative PFC analysis of PB MNCs from healthy controls (a) and PAD patients (b) stained with the six-antibody/viability marker panel and assessed for (CD31⁺CD34^brightCD45^dimAC133⁺) pro-angiogenic CHSPCs (dark blue gate) and (CD31⁺CD34^brightCD45^dimAC133⁻) non-angiogenic CHSPCs (dark red gate). FMO gating controls determine gate boundaries. (c) The ratio of pro-angiogenic CHSPCs to non-angiogenic CHSPCs of PAD patients shows a significant decrease when compared to age and gender matched controls (n=9, ***p=0.0001 by two-tailed, unpaired Student’s t test).

DISCUSSION

A PFC protocol [14] was utilized to isolate certain putative hematopoietic subsets now found to be comprised of CHSPCs which are defined by cell surface antigen expression, colony assay, morphologic analysis, and in vivo function. The circulating hematopoietic cells identified are validated herein as cells that function in neoangiogenesis and serve as potential biomarkers of CVD or tumor progression. Another reported protocol for CPC enumeration that correlates with tumor progression risk is now clarified to identify hematopoietic progenitor cells, myeloblasts and engrafting HSCs [4]. Thus, we present data that provides an analytical method for enumerating circulating blood cells that participate in new blood vessel formation at homeostasis and in subjects with abnormal cardiovascular health.

Confusion around the function of, EPCs, circulating endothelial progenitors (CEPs), and CPCs in vascular repair and regeneration at homeostasis or in response to injury or disease is linked to lack of consensus regarding quantitative measures to isolate each cell type using in vitro colony assays, immunomagnetic separation (IMS), or conventional flow cytometry approaches [11, 24, 25]. More recently, the heterogeneous nature of EPC populations (to include non-endothelial precursors) has been recognized (reviewed in [24]) and an even broader term, CPC, is now used to encompass circulating cells with pro-angiogenic activity [4, 26]. Use of the term CPC, without functional validation of the cell types comprising this fraction has not been helpful in understanding the mechanisms of cellular action purported to emerge from these flow cytometry “events”. It is widely recognized that a new approach in defining the parameters and properties of cells involved in neoangiogenesis is required for the field to make advancements in clinical treatments [12].

Many different proposals for using conventional flow cytometry approaches to identify EPCs, and CPCs are published (reviewed in [24]). The CPC population identified using the conventional flow cytometry approach [4] and the novel CHSPC population isolated using the PFC protocol [14] are now demonstrated to be comprised of hematopoietic cells at different stages of differentiation. All observed cells belong to the HSPC pool, a significant proportion of which display in vitro hematopoietic colony forming cell activity and others engraft in immunodeficient mice. Hematopoietic cells are known to participate in angiogenesis [27]. Consistent with their biological function, it is not surprising then that increased concentrations of CPCs correlate with risk for tumor recurrence and patient responsiveness to anti-angiogenic therapies [23].

Use of the PFC protocol has permitted a clear distinction between CHSPCs with pro-angiogenic function and those lacking in angiogenic supportive activity, based upon AC133 expression[14]. AC133 is proposed as a marker for circulating EPCs and has been used in combination with CD34 and/or CD31 and/or KDR as a biomarker in patients with CVD, cancer, sepsis, or renal failure [28, 29]. The PFC approach [14] permits isolation of CHSPC subsets based upon AC133 expression and only the CD31⁺CD34^brightCD45^dimAC133⁺ CHSPC subset possesses pro-angiogenic activity in promoting angiogenesis and human melanoma tumor growth in an immunodeficient mouse explant model system. This particular CHSPC subset was enriched in cells displaying a variety of myeloid cell surface antigens in addition to displaying in vitro and in vivo HSPC functions. This subset did not display any vasculogenic ability in vivo when examined for the presence of human endothelium within the explanted human tumors within the immunodeficient mice. Thus, the pro-angiogenic CHSPC appears to be enriched in pro-angiogenic functions but lacks postnatal vasculogenic activity. Future examination of these cells to identify the specific molecular pathways for promoting tumor angiogenesis may be insightful and permit a better understanding of mechanisms for blocking tumor angiogenesis. It will also be important to understand whether the CHSPCs are specifically recruited to the tumor site and differentiate into myeloid cells such as Tie2⁺ monocytes that are known to promote tumor angiogenesis [10].

An interesting insight into the variance of circulating concentrations of pro-angiogenic and non-angiogenic CHSPCs was observed in patients with PAD. A significant decrease in the ratio of pro-angiogenic to non-angiogenic CHSPCs was measured in the bloodstream of patients with PAD as compared to healthy control subjects. Future studies to evaluate potential differences in the gene expression and function of the CHSPCs in normal subjects and those with PAD at various stages of their disease may permit a new insight into whether or not this cell population contributes to PAD disease progression.

In summary our PFC method has now been shown to identify subsets of circulating cells in human PB which promote angiogenesis. Use of this method for prospective identification of these cells should facilitate human clinical studies and functional biological characteristics of each defined cellular subset.

Abbreviations

PFC

polychromatic flow cytometry

EPC

endothelial progenitor cell

CPC

circulating progenitor cell

ECFC

endothelial colony forming cell

CFU-Hill

colony forming unit – Hill

CHSPCs

circulating hematopoietic stem and progenitor cell

BM

bone marrow

CFU-GEMM

colony forming unit – granulocyte erythroid macrophage megakaryocyte

PB

peripheral blood

CB

cord blood

mPB

mobilized peripheral blood

PAD

peripheral artery disease

CAD

coronary artery disease

COPD

chronic obstructive pulmonary disease

PBS

phosphate buffered saline

MNC

mononuclear cells

FBS

fetal bovine serum

PMT

photomultiplier tube

IL-1α

interleukin 1 alpha

M-CSF

macrophage-colony stimulating factor

SCF

stem cell factor

LPP-CFC

low proliferative potential-colony forming cells

HPP-CFC

high proliferative potential-colony forming cells

RT-qPCR

real time quantitative polymerase chain reaction

FACS

fluorescence activated cell sorting

FMO

fluorescence minus one

RBC

red blood cell

HSPC

hematopoietic stem and progenitor cell

CEP

circulating endothelial progenitor

IMS

immunomagnetic separation