Purification and Functional Characterization of CD34-Expressing Cell Subsets Following Ex Vivo Expansion of Umbilical Cord Blood-Derived Endothelial Colony-Forming Cells

Introduction

From adaptable high-volume arteriole systems to fenestrated microvascular networks, endothelial cells serve as building blocks to form the intraluminal lining of blood vessels and provide a highly regulated interface between circulating blood and vascularized tissue compartments [1,2]. The dynamic array of endothelial cell phenotypes is governed by tissue-specific microenvironments that provide bioactive cues and an architectural matrix to direct endothelial specification during organ development [3,4]. To illustrate this concept, grafts of non-neural somite tissue transplanted within a neural microenvironment formed the organization of the blood–brain barrier observed in adjacent host tissues [5]. On the other hand, transplanted brain tissues developed a fenestrated interface within coelomic tissue beds, supportive of the notion that endothelial specification occurs in association with the microenvironment [5].

Contrary to tissue-resident primary endothelial cells, endothelial cell lines often demonstrate a plastic and variable phenotype in vitro, although the functional relevance of this dynamic surface marker expression is poorly understood [6 –11]. Specifically, it appears that the removal of endothelial cells from their native environment and subcultured in vitro can induce loss of organ-specific phenotype [12,13]. It has been previously demonstrated that phenotypic expression profiles across several progenitor cells lineages, such as hematopoietic and mesenchymal progenitor cells, can rapidly change based on microenvironmental cue experience during ex vivo expansion [14 –18]. Thus, the classification of endothelial cell subsets within expanding endothelial progenitor/precursor cells (EPC) in vitro remains a challenge.

The conceptual framework for EPC arose from seminal experiments by Asahara defining the ability of CD34⁺ peripheral blood mononuclear cells (MNC) to home sites of ischemia and promote vascularization after transplantation [10]. This work inspired three distinct culture techniques to isolate and propagate EPC-like colonies from human blood or bone marrow sources. As a result, three inter-related cell populations were established: colony-forming unit–endothelial cells (CFU-EC) [10,19], circulating angiogenic cells (CACs) [20 –22], and endothelial colony-forming cells (ECFC) [23 –25]. It was later uncovered that the early outgrowth populations (CFU-EC and CAC) comprised primarily myeloid hematopoietic cells with a proangiogenic secretory phenotype [15,20,26 –28].

In contrast, late outgrowth ECFC derived from plastic-adherent MNC gave rise to highly proliferative endothelial cells able to form vascular networks in vivo [6]. Although it is agreed that only ECFC give rise to pure endothelial cells in vitro, a lack of consensus surrounding the term “EPC” to describe freshly isolated versus cells expanded in culture remains [6,9]. Several surface markers or levels of enzymatic activity have been explored to further define and enrich EPC before colony establishment [9]. For example, freshly isolated CD34⁺/CD45⁻, CD133, CD146, or ALDH^hi cells can enrich ECFC [29 –32]. However, it is poorly understood whether these same phenotypes can reselect and purify for a subpopulation of ECFC that retain progenitor-like characteristics following expansion in vitro.

The detection of circulating endothelial cell phenotypes in peripheral blood has been used as clinical diagnostic to noninvasively predict cardiovascular disease risk [19] and recovery rates [33] in patients with cardiovascular complications. Preclinical studies have demonstrated the therapeutic relevance of freshly isolated EPC; in addition, the purification of EPC has uncovered molecular mechanisms underlying tissue homeostasis or disease [34].

It remains precedent to further our understanding of the distinctions between EPC harvested from fresh and/or expanded ECFC sources. Specifically, this knowledge is crucial for accurately comparing results obtained during clinical trials using endothelial cells in applications of regenerative medicine [34]. There are currently over 350 clinical trials initiated, which indicate the use of EPC or endothelial populations (www.clinicaltrials.gov). Notably, the selection parameters for EPC ranged from cell selection using CD34 [35] and/or VEGFR2⁺ and CD133⁺ [36] to unpurified MNC cultured similar to hematopoietic CACs [37,38]. Nonetheless, these early clinical studies have consistently demonstrated safety and potential therapeutic benefit in the field of regenerative medicine.

The clinical application of ECFC will likely require extensive ex vivo expansion to obtain large numbers of ECFC required for cell therapies. Thus, it remains critical to elucidate maturation-induced phenotype changes for expanded ECFC to better direct the utility of therapeutic ECFC subsets [39,40]. We hypothesized that ECFC cultures contain a hierarchy of rare, proliferative progenitor cells that drive expansion and subsequently generate more differentiated endothelial cells that demonstrate dampened proliferative rates and blunted vascular regenerative function.

Herein, we demonstrate that umbilical cord blood (UCB)-derived ECFC expressed CD34 in a density-dependent manner and the expression of CD34 coincided with increased expression of mature endothelial cell surface antigens, but decreased clonogenicity, growth kinetics, and tube-forming capacity compared to CD34⁻ ECFC in vitro. In addition, CD34^+/− ECFC subsets were equally able to form vessel-like structures within Matrigel plugs after subcutaneous transplantation into the flank of NOD.SCID mice. Collectively, this study provides evidence that the phenotype of UCB-ECFC is highly dynamic in vitro, although the identity of a functionally potent EPC-like subset within expanded ECFC remains elusive.

Materials and Methods

Results

ECFC can be segregated based on CD34 expression and/or ALDH activity

To validate the purity of ECFC cultures, flow cytometry was used to assess both endothelial and hematopoietic cell surface marker expression. To confirm endothelial identity, cultured UCB-derived ECFC highly expressed CD31, VEGFR2, VE-Cadherin, and CD105 (>97%) with <5% hematopoietic contamination measured using CD45 (Supplementary Fig. S1B–F). Analyses of markers commonly used to assess progenitor cell status (ALDH activity, CD34) or migratory competence (CXCR4) after ex vivo expansion are shown in Fig. 1A. Expanded ECFC were 64.5% ± 17.9% ALDH^hi and 28.0% ± 14.7% ALDH^lo (Fig. 1B). In contrast, CD34 (3.3% ± 1.3%) and CXCR4 (1.4% ± 0.6%) expression on cultured ECFC were consistently low (Fig. 1B).

OPEN IN VIEWER

FIG. 1.

Putative progenitor marker expression assessment by flow cytometry on expanded UCB-ECFC. Expanded UCB-ECFC (at passage 4) were assessed for the expression of ALDH, CD34, and CXCR4 by flow cytometry to identify progenitor cell populations. Representative flow plots of ECFC expressing (A) ALDH, CD34, and CXCR4, where Aldefluor™ fluorescence was gated based on DEAB and surface markers gated based on isotype controls. (B) The majority of ECFC were CD34⁻ or CXCR4⁻ (>80%), and ALDH^hi cells represented >60% of the ECFC population. Neither (C) CD34 nor (D) CXCR4 was differentially co-expressed on cells with high ALDH activity. (E) External CXCR4 expression was increased on CD34⁺ cells. (F) To assess intracellular expression of CD34 or CXCR4, ECFC were fixed and permeabilized before staining with antibodies toward CD34 and CXCR4. Approximately 90% of ECFC expressed intracellular CXCR4, whereas CD34 expression was unchanged after permeabilization of ECFC. Data represent the mean ± SD using ECFC derived from four UCB samples (N = 4, **P < 0.01, ***P < 0.001). ALDH, aldehyde dehydrogenase; ECFC, endothelial colony-forming cells; SD, standard deviation; UCB, umbilical cord blood.

UCB-derived hematopoietic progenitor cells with high ALDH activity often co-express CD34 [18,40,48,49]; however, CD34 or CXCR4 expression was not enriched within either ALDH^lo or ALDH^hi ECFC subset (Fig. 1C, D). In contrast, CXCR4 expression was significantly enriched within the CD34⁺ ECFC subset, compared to CD34⁻ ECFC (Fig. 1E). Several reports have described dynamic internalization and externalization patterns of both CXCR4 and CD34; therefore, we determined whether ECFC possessed intracellular stores of CXCR4 or CD34 following fixation and permeabilization before flow cytometry. Notably, the vast majority of ECFC contained intracellular stores of CXCR4 (Fig. 1F). On the other hand, CD34 expression in ECFC appeared to be exclusive to the outer plasma membrane leaflet in ECFC (Fig. 1F).

Although low expression of CD34 was observed on culture-expanded ECFC, the CD34⁺ ECFC demonstrated increased CXCR4 expression, which may represent increased migratory capacity previously demonstrated for CXCR4⁺/CD34⁺ SCID-repopulating hematopoietic progenitor cells [50], and for putative circulating endothelial progenitor cells [51].

CD34⁻/CXCR4⁻ ECFC demonstrated increased colony-forming capacity

To assess progenitor-like functions in vitro, ECFC at passage 4 were sorted and plated as single cells based on either ALDH activity or CD34 or CXCR4 expression (Fig. 2), and assessed for clonal colony formation. When acquired without selection using forward and side scatter properties, 1 in 3 (34%) of ECFC established colonies (Fig. 2). After single-cell deposition of ECFC based on low versus high ALDH activity, the clonogenic capacity of ALDH^hi ECFC was equal to ALDH^lo ECFC (Fig. 2A, B). In contrast, CD34⁻ ECFC efficiently formed colonies (51.3% ± 5.85%), while CD34⁺ ECFC did not (16.3 ± 4.7) (Fig. 2C, D). Similarly, CXCR4⁻ ECFC formed significantly more colonies than CXCR4⁺ ECFC (Fig. 2E, F). Although ALDH activity did not select for increased clonogenic capacity, the common CD34⁻ and CXCR4⁻ populations retained the potential for clonal expansion.

OPEN IN VIEWER

FIG. 2.

The absence of CD34 expression marked increased colony-forming capacity in UCB-ECFC. Colony-forming capacity of single ECFC purified by expression of (A, B) ALDH, (C, D) CD34, or (E, F) CXCR4 alone or in combination (G, H). ECFC lacking CD34, (D) CD34, or (F) CXCR4 expression exhibited increased colony-forming capacity compared to CD34⁺ or CXCR4⁺ cells. ECFC expressing CD34 with high ALDH activity formed colonies with equal efficiency compared to counterparts lacking CD34 expression. (H) CD34⁻/CXCR4⁻ ECFC more efficiently formed colonies compared to ECFC with expression of either CD34 or CXCR4. Data represent the mean ± SD using ECFC derived from three UCB samples (N = 3, *P < 0.05, **P < 0.01).

Next, we performed FACS using combinations of these aforementioned progenitor cell markers. Single-cell purification of ALDH^loCD34⁺ and ALDH^hiCD34⁺ ECFC consistently established less colonies than CD34⁻ counterparts (Fig. 2G). Interestingly, CD34⁻/CXCR4⁻ cells demonstrated the highest colony formation compared to any other combination selected for CD34 and CXCR4 expression (Fig. 2H). Collectively, these data suggest that ECFC expressing CD34 or CXCR4 have a reduced ability to establish colonies from single cells in vitro. Considering the notable stores of CXCR4 within ECFC compared to ECFC CD34 expression (Fig. 1F), we focused subsequent analyses on ECFC with or without CD34 expression.

Tube-forming capacity was increased by CD34⁻ ECFC

To assess the relationship between progenitor marker expression and endothelial cell function in vitro, we assessed the tubule-forming capacity of ECFC subsets purified for ALDH activity or CD34 expression. Briefly, 120,000 purified CD34^+/− or ALDH^hi/lo ECFC were seeded in growth factor-reduced Geltrex in EBM or fully supplemented EGM-2. Similar to our previous results with clonogenic capacity, selection for low versus high ALDH activity did not alter the tubule-forming capacity of ECFC (Fig. 3A, B). In contrast, CD34⁻ ECFC demonstrated increased tubule formation under minimal EBM or fully supplemented EGM conditions, compared to CD34⁺ ECFC (Fig. 3C, D).

OPEN IN VIEWER

FIG. 3.

ECFC tube-forming capacity was decreased in CD34⁺ ECFC. After ex vivo expansion, ECFC were sorted based on (A, B) ALDH activity or (C, D) CD34 expression and plated on Geltrex™ for 20 h. Representative photomicrographs of tube formation for (A) ALDH^hi/lo and (C) CD34^+/− are shown. Whether using basal media (EBM) without growth factors or fully supplemented growth media (EGM), (B) tube-forming ability was similar for ECFC with high or low ALDH activity. (D) Alternatively, ECFC that lacked CD34 expression formed less tubules than CD34⁻ ECFC. Data represent the mean ± SD using ECFC derived from five UCB samples (N = 5; *P < 0.05, **P < 0.01). Scale bar = 100 μm. EGM, endothelial growth media.

CD34 expression on ECFC is reversible and elevated with increasing cell density

CD34 is enriched on founder ECFC populations; however, kinetics of CD34 expression during culture is less understood. Accordingly, we demonstrated that ECFC at passage 1 were highly enriched with CD34⁺ cells (Fig. 4A). However, CD34 expression rapidly decline during early passages and established a steady-state frequency during serial passaging (passage ≥3). We observed an initial trend toward accelerated growth kinetics within CD34⁻ cells compared to CD34⁺ ECFC (Fig. 4B); however, equal cell numbers were obtained after 6 days of culture with both ECFC subsets. Cell density has been previously suggested to influence CD34 expression [7]. Consistent with these findings, we observed a 6.5-fold increase in CD34 expression when cells were cultured to high density (>90% confluent) before passage compared to when ECFC were harvested and passaged at <80% confluency (normal density) (Fig. 4C, D).

OPEN IN VIEWER

FIG. 4.

ECFC cultures showed temporal and cell density-dependent CD34 expression that impacted colony and tubule formation. (A) UCB-ECFC at early passage (P1) were enriched for CD34⁺ cells, and the frequency of CD34⁺ cells rapidly declined and stabilized at later passage (after P2). (B) CD34⁻ ECFC (passage 4) demonstrated modestly accelerated growth kinetics compared to CD34⁺ cell. (C) Schematic of ECFC passaged at normal density (70%–80% confluent) versus high density (>90% confluent). (D) CD34 expression was increased under high-density culture, whereas ALDH activity or CXCR4 expression remained stable. (E) CD34⁺ and CD34⁻ cells selected from high-density cultures demonstrated equivalent colony-forming capacity, whereas CD34⁻ cells passaged at normal cell density showed increased colony-forming capacity. (F) CD34⁻ ECFC selected from high-density conditions demonstrated an increased capacity to form tubules. Dynamics of CD34 expression relative to cell density following FACS purification and replating of (G) CD34⁻ ECFC or (H) CD34⁺ ECFC. Data represent the mean ± SD using ECFC derived from four UCB samples (N = 4; *P < 0.05, **P < 0.01, ***P < 0.001). FACS, fluorescence-activated cell sorting.

In contrast, other progenitor cell markers (ALDH and CXCR4) remained stable when ECFC reached confluence >90% (Fig. 4D). Next, we sought to assess the functional implications of increased CD34 expression with increased cell density. Interestingly, we did not observe differences in colony formation observed between CD34⁺ and CD34⁻ ECFC after collection at high (>90%) cell density (Fig. 4E). Similarly, the ECFC subsets purified for CD34 after growth to high density did not demonstrate any difference in the ability to form tubes in vitro (Fig. 4F). These findings indicated that culture of ECFC to confluency increases CD34 expression and abrogated functional differences in colony and tubule formation in vitro.

It was unclear whether CD34 expression was turned on in CD34⁻ cells under high cell density conditions. Accordingly, we investigated CD34 expression dynamics over 7 days alongside the expansion of purified CD34⁻ (Fig. 4G) and CD34⁺ purified ECFC (Fig. 4H) at passage 4. Interestingly, CD34⁻ cells consistently increased CD34 expression as cell density was increased, suggesting CD34⁻ gave rise to CD34⁺ cells in vitro. The frequency of CD34⁺ ECFC gradually decreased to ∼20% over the first 96 h of culture (Fig. 4G), after which CD34 expression increased gradually as cell density increased. Collectively, these findings suggest that CD34 expression was reversible and cell density dependent; moreover, elevated CD34⁺ expression under high-density conditions does not impact endothelial functions in vitro.

Surface proteomics on CD34 selected ECFC subsets revealed enrichment of mature endothelial cell pathways on CD34⁺ cells

In an effort to identify other surface markers that may correlate with an endothelial progenitor-like state in vitro, we employed silica-bead capture followed by LC-MS/MS proteomics analyses. We identified >4,800 proteins captured on silica-beads (Fig. 5A), however ∼25% of identified proteins were associated with cell surface or plasma membrane annotations, referred to as surfaceome. Specifically, we identified >800 proteins associated with the surfaceome of either CD34⁻ or CD34⁺ ECFC subsets (Fig. 5B). The surfaceome signature in itself did not distinguish either ECFC subset (Fig. 5C), as 935 proteins were commonly expressed on both subsets. Nonetheless, 23 and 10 proteins were exclusively detected in the surfaceome of CD34⁻ versus CD34⁺ ECFC, respectively (Fig. 5D and Table 1). This included the exclusive detection of CXCR4 in CD34⁺ ECFC, which corroborated our earlier flow cytometry analysis (Fig. 1).

OPEN IN VIEWER

FIG. 5.

The surfaceome of CD34⁺ ECFC and CD34⁻ ECFC. FACS-purified CD34⁻ or CD34⁺ ECFC were temporarily seeded in EGM-2 to allow adherence. (A) Silica-bead protein capture performed on FACS-purified CD34⁻ or CD34⁺ ECFC identified >4,800 proteins were detected on both ECFC subsets. (B, C) Additional filtering for proteins associated with GOCC enrichment for the external membrane identified >900 unique proteins to generate a (C) surfaceome fingerprint for CD34⁻ and CD34⁺ ECFC. (D) Twenty-three and 13 proteins were exclusively detected in the surfaceome of either CD34⁻ or CD34⁺ ECFC, respectively (also see Table 1). Nine hundred thirty-five cell surface proteins were detected co-expressed on CD34⁺ and CD34⁻ ECFC samples. (E) Twenty and 70 proteins were significantly increased in the surfaceome of CD34⁻ versus CD34⁺ ECFC, respectively. (F) GO Biological Processes and Reactome Enrichment analysis the surfaceome of ECFC identified an enrichment of proteins implicated in (G) cell-substrate adhesion interactions and (H) regeneration on EC34⁺ ECFC. GOCC, GO Cellular Compartment.

OPEN IN VIEWER

Table 1.

Proteins Exclusively Identified on the Surfaceome of CD34⁻ Versus CD34⁺ Endothelial Colony-Forming Cells

Gene name	Protein name	Sequence coverage	% of CD34−	% of CD34+
INPPL1	Phosphatidylinositol 3,4,5-trisphosphate 5-phosphatase 2	2.1	50	0
IGHM	Ig mu chain C region	3.5	25	0
TSPAN5	Tetraspanin-5	9	50	0
ORC2	Origin recognition complex subunit 2	3.3	100	0
TUSC3	Tumor suppressor candidate 3	2.9	25	0
TJAP1	Tight junction-associated protein 1	10.2	50	0
SIRT2	NAD-dependent protein deacetylase sirtuin-2	13.7	25	0
KLHL7	Kelch-like protein 7	4.1	25	0
OR9I1	Olfactory receptor 9I1	3.8	25	0
KIRREL	Kin of IRRE-like protein 1	5.4	100	0
PLXNA4	Plexin-A4	2.3	100	0
ORC6	Origin recognition complex subunit 6	4	25	0
ENTPD6	Ectonucleoside triphosphate diphosphohydrolase 6	3.6	50	0
C6orf89	Bombesin receptor-activated protein C6orf89	5.4	50	0
SRPX	Sushi repeat-containing protein SRPX	4.7	75	0
MCOLN1	Mucolipin-1	2.6	50	0
SLC17A5	Sialin	9.4	50	0
DPP3	Dipeptidyl peptidase 3	3.5	50	0
SLC20A1	Sodium-dependent phosphate transporter 1	3.2	100	0
XPO6	Exportin-6	2.9	100	0
GRB14	Growth factor receptor-bound protein 14	9.1	75	0
SHB	SH2 domain-containing adapter protein B	2.8	75	0
DCBLD2	Discoidin, CUB and LCCL domain-containing protein 2	7.6	100	0
HLA-B	HLA class I histocompatibility antigen	3	0	100
LIMS1	LIM and senescent cell antigen-like-containing domain protein 1	7.1	0	100
ITGA1	Integrin alpha-1	3.1	0	25
B2M	Beta-2-microglobulin	16.8	0	50
HLA-C	HLA class I histocompatibility antigen	3	0	100
RPS6KB1/2	Ribosomal protein S6 kinase beta-2	4.4	0	50
RAPGEF2	Rap guanine nucleotide exchange factor 2	2.1	0	50
HLA-B	HLA class I histocompatibility antigen	3.3	0	100
VAMP2	Vesicle-associated membrane protein 2	39.7	0	100
ABCG2	ATP-binding cassette subfamily G member 2	3.9	0	75
ALPL	Alkaline phosphatase	10.7	0	75
THBD	Thrombomodulin	2.4	0	100
CD34	Hematopoietic progenitor cell antigen CD34	9.6	0	100

Furthermore, we identified 935 proteins that were detected in >50% of samples from each ECFC subset. Twenty and 70 proteins were differentially expressed twofold (P < 0.05) on either CD34⁻ or CD34⁺ ECFC subsets, respectively (Fig. 5E). Comparison of these 90 proteins against the GO Biological Process and Reactome annotation databases identified an enrichment of proteins with functions toward endothelial cell development, vascular regeneration, VEGF-VEGFR2 signaling, and cell-substrate adhesion interactions on the CD34⁺ ECFC population (Fig. 5F).

AXL and COL13A1, associated with cell-substrate adhesion and/or regeneration, were elevated in the surfaceome of CD34⁻ ECFC (Fig. 5G). In contrast, angiotensin-converting enzyme (ACE) and angiopoietin 2 (ANGPT2) were enriched within the surfaceome of CD34⁺ ECFC (Fig. 5H). Unfortunately, very few unfractioned ECFC demonstrated expression of COL13A1 or AXL on the external cell surface membrane, Cell fixation and permeabilization revealed that >90% of ECFC expressed these proteins on the inside of the cell (Supplementary Fig. S2). Further investigation is warranted to understand the relevance of exclusive or differentially expressed proteins between CD34⁻ and CD34⁺ ECFC subsets.

In vivo transplantation of CD34^+/− ECFC subsets in Matrigel enabled the formation of vessel-like structures

To assess the function of CD34⁺ versus CD34⁻ ECFC subsets to form tubes in vivo, CD34⁺ or CD34⁻ ECFC were purified and injected subcutaneously into the flank of immunodeficient NOD.SCID mice within a Matrigel plug. After harvesting the plugs at 7 days, plugs implanted with both CD34⁺ and CD34⁻ ECFC showed macroscopic areas of erythrocyte infiltration (Supplementary Fig. S4). Indeed, Matrigel plugs implanted with CD34⁺ or CD34⁻ ECFC generated vessel-like structures that were HLA positive (Fig. 6A). However, infiltration of murine CD31 cells within the Matrigel was rare and equivalent to vehicle control (Fig. 6B). On rare occasion, the only interaction between human and mouse endothelial cells was found in the Matrigel plugs containing the CD34⁻ ECFC (Fig. 6C). Collectively, these data suggested that both the CD34⁺ and CD34⁻ ECFC subsets equally generate vessel-like structures within Matrigel implanted in vivo.

OPEN IN VIEWER

FIG. 6.

Purified CD34^+/− ECFC demonstrated equal vascularization in Matrigel plugs subcutaneously injected into immunodeficient mice. CD34^+/− ECFC were purified and combined with mouse SMC within Matrigel and subcutaneously implanted into NOD.SCID mice. The Matrigel plugs extracted after 7 days were assessed for vessel formation within the implants. Vessel structures were visualized by (A) immunofluorescent staining for human (HLA⁺) or mouse (HLA⁻) endothelial (CD31⁺) cells with DAPI nuclear co-staining. (B) Quantification of human and mouse CD31⁺ cells revealed little recruitment of mouse-derived endothelial cells and a similar number of human cells within the Matrigel plugs. (C) Representative photomicrograph of CD34⁻ ECFC co-localized with mouse CD31⁺ endothelial cells (human cells—orange arrow; mouse endothelial cells—white arrows). Data represent the mean ± SEM using ECFC derived from three UCB samples (n = 5). SMC, smooth muscle cells.

Discussion

ECFC remain the gold standard for progenitor-like endothelial cells that possess potent proangiogenic properties in vitro and in vivo. Initially, ECFC can be efficiently propagated in culture; however, the proliferative capacity of expanded progeny is dampened as phenotypic and molecular transformations occur during serial passages, ultimately resulting in culture-induced senescence [6]. The ability to distinguish highly proliferative progenitor cells within heterogeneous ECFC cultures would potentially enable the isolation of more potent provascular cell subsets for vascular therapies.

In this study, we assessed a combination of known progenitor cell functions and cell surface marker expression during culture [52,53] to identify a subpopulation of ECFC with enhanced progenitor cell characteristics. Specifically, we employed in vitro assays to assess putative progenitor cell functions, including growth kinetics, single-cell colony formation, and tubule formation. Our findings suggest CD34 expression was variable and cell density dependent and may not be a strong marker for the selection of culture-expanded ECFC with progenitor cell-like characteristics. Similarly, CD34 expression did not segregate ECFC with vessel-like forming capacity in vivo.

In an attempt to identify additional markers associated with progenitor-like function, we also employed silica-bead plasma membrane capture followed by label-free proteomics analysis to compare the surfaceome of CD34⁻ ECFC versus CD34⁺ ECFC. Membrane proteins highly expressed on CD34⁺ ECFC suggested a mature endothelial cell surface phenotype; however, subcutaneous implantation of CD34⁺ ECFC into NOD.SCID mice formed comparable vessel-like structures in Matrigel plugs compared to CD34⁻ ECFC. Collectively, our data suggested that similar to cultured hematopoietic progenitor cells, CD34 expression was dynamic and reversible on expanded ECFC; moreover, selection using CD34 identified ECFC with reduced progenitor cell characteristic in vitro.

Hematopoietic stem/progenitor cells (HSPC) and EPC originate from a common hemangioblast during fetal development [26,29,54]; thus, shared phenotypic markers have been used to enrich HSPC and EPC cells from several blood-derived human sources [6,26,29,53 –55]. CD34 has been used for over a decade to identify HSPC [29] and a considerable number of studies have selected CD34⁺ cells before the establishment of ECFC cultures [7,29,31,55]. In response to injury, EPC expressing CXCR4 will home sites of ischemia by the SDF-1/CXCR4 axis and contribute to angiogenesis and vasculogenic processes [50 –52,56]. It also is well established that progenitor cells will rapidly differentiate or undergo dynamic phenotypic changes in response to the “unnatural” microenvironment of culture. As a result, it remains poorly understood whether cell surface markers on de novo isolated or uncultured EPC are retained on ECFC expanded in vitro.

In addition to surface markers, high ALDH activity was enriched in de novo isolated HSPC and EPC populations and loss of high ALDH activity coincided with HSPC differentiation in vitro [14,40]. We determined that ALDH activity did not correlate with the expression of CD34 or CXCR4, as previously observed by Ferreras et al. [31]. Although similar isolation techniques were utilized in this study, the cell source being from peripheral blood of an adult versus UCB in our study may explain the differences observed across these experiments. It is also interesting to note that each outgrowth endothelial cell (OEC) line had different CD34 expression after isolation, where some OEC expressed CD34 to a high degree and other OEC minimally expressed CD34 [31]. Furthermore, selection of ECFC using high versus low ALDH activity did not segregate ECFC with enhanced colony- or tubule-forming capacity in vitro.

Alternatively, CXCR4 and/or CD34 expression on expanded ECFC was uncommon (<5%) and served as a negative selection maker for colony- or tubule-forming capacity in vitro. Previous reports have demonstrated expression of CXCR4 and CD34 on the plasma membrane of cells is dynamic and can rapidly transition between intracellular and extracellular compartments [55 –57]. Thus, we assessed ECFC for intracellular stores of CXCR4 or CD34 after cell permeabilization. Indeed, CXCR4 expression was located within the intracellular compartment of ECFC, thus was deemed unsuitable as a selection maker due to the possible capacity for rapid externalization.

Although we were unable to demonstrate a correlation between ALDH activity and CXCR4 expression on the surface of ECFC, Tu et al. observed higher CXCR4 transcription within the ALDH^low cells [58], and induced CXCR4 expression on ALDH^high ECFC augmented endothelial cell homing to the site of in vivo injury. Notably, the only cytokine constituent in the culture media used by Tu et al. was b-FGF, and provides a possible explanation for the differential observations with other studies using complete EGM-2 culture media. Guan et al. have previously demonstrated endothelial phenotype and function were determined by the media used for endothelial cell culture [59]. This carries large implications regarding differences in surface marker expression when comparing different studies using divergent media constituents and tissue sources for ECFC isolation.

To consolidate our findings with Tu et al., we demonstrated that ALDH expression did not enrich for cells with elevated CXCR4 expression. This implies that the capacity to externalize CXCR4 in the ALDH^low cellular fraction upon exogenous stimulation is more potent. For example, stimulation of L-selectin recipient cells by cell-extrinsic ligands drives CXCR4 externalization [60]. An important point to note is many studies involving the profiling of CXCR4 expression on ECFC by flow cytometry fail to profile the CXCR4 content within the cell. Based on our results, ECFC contain internal CXCR4 stores; thus, studies demonstrating increased CXCR4 by flow cytometry may only provide a snapshot of a cell's capacity to respond to external cues leading to CXCR4 externalization.

CD34 frequency was highest at passage 1 and decreased to a steady state by passage 3 for each ECFC line tested. Purified CD34⁺ ECFC at passage 4 formed significantly fewer colonies and tubules in vitro than CD34⁻ ECFC, similar to studies by Ferreras et al. [31]. Conversely, other investigators have suggested that CD34⁺ ECFC have increased capacity to form tubule-like structures in vitro [7,8]. The divergence in these data from previous studies may be partially accounted for by the selection of tissue source (ie, placental vs. umbilical cord blood) and/or culture conditions. Specifically, it appears UCB-derived ECFC align better with peripheral blood-derived ECFC compared to placental-derived ECFC [7,31].

In agreement with Tasev et al., we demonstrate the expression of CD34 on ECFC was dependent on cell density and coincided with decreased tubule-forming capacity [7]. In contrast, we did not observe consistent changes to the expression of ALDH or CXCR4 after ECFC were grown to high-density confluency. To further demonstrate the dynamic nature of CD34 expression, we assessed ECFC density alongside the CD34⁺ cell frequency over the course of a 7-day culture using FACS-purified CD34⁻ versus CD34⁺ ECFC. CD34⁻ ECFC gradually increased CD34 expression over time as cell density increased. Similarly, FACS-purified CD34⁺ ECFC rapidly reduced CD34 expression early during culture and increased CD34 expression as cell density increased. We also observed that CD34⁻ ECFC demonstrated accelerated growth kinetics compared to CD34⁺ ECFC.

Since CD34 expression on ECFC was dynamic and increased at high cell densities, we next assessed the impact of confluency on endothelial cell functions in vitro. Although CD34⁺ cell frequency increased at high cell densities, the statistical differences in clonogenicity and tubule forming observed between CD34⁺ and CD34⁻ ECFC were abrogated when harvested at high-density conditions. Thus, increasing CD34 expression on ECFC appeared to be a response to cell-cell and/or cell-microenvironment cues that did not lead to functional changes in vitro. Collectively, these findings suggest a need to standardize culture conditions to prevent culture artifact, such as elevated CD34 expression. Moreover, we speculate CD34 expression reflects a transitional state of ECFC in vitro that warrants further investigation.

We employed silica-bead surface protein capture followed by label-free LC-MS/MS to identify prospective surface markers that correlate with progenitor functions. We identified >900 plasma membrane-associated proteins that were associated with cell-substrate adhesions, response to wounding, and regeneration. Specifically, we identified a collection of mature endothelial cell surface markers, such as CD31, ACE, VEGFR2, and ANGPT2, were significantly enriched on CD34⁺ ECFC. In contrast, proteins highly expressed on the surface of the CD34⁻ ECFC did not possess obvious categorical association.

One protein of interest on CD34⁻ ECFC was AXL, a protein tyrosine kinase that regulates cell proliferation and survival [61]. AXL is also postulated to attenuate immune responses [62], a function of progenitor cells in tissues with low turnover rates [63], a characteristic of endothelial cells in vivo [64]. Unfortunately, we were limited in this study by the lack of conjugated human-specific antibodies for flow cytometry on the outer leaflet of the membrane. For example, antibodies tested against AXL did not detect cell surface expression, instead considerable intracellular stores were detected. In addition, we validated the outer membrane expression of ACE/CD143 on ECFC was highly enriched within the CD34⁺ ECFC subpopulation. ACE is expressed on mature endothelial cells in vivo and acts as the rate-limiting enzyme in the renin-angiotensin pathway by processing angiotensinogen I to angiotensinogen II in blood plasma [65]. We speculated ACE represented a promising marker to predict why a large proportion of CD34⁺ cells was unable to make colonies or lacked tubule formation function in vitro. In addition, the expression of ACE has been previously demonstrated to be linked to cell density [66]. Unfortunately, sorting for ACE⁺ versus ACE⁻ ECFC revealed no differences in the generation of colonies after single-cell deposition.

Our surfaceome analysis identified a significant enrichment of Mucolipin 1 (MCOLN1), a member of the transient receptor potential cation channel family, on CD34⁻ ECFC. This membrane-associated protein primarily plays a role in calcium release from late endosomes and lysosome vesicles. Previous studies have implicated the mobilization of intracellular Ca2⁺ stores during CXCR4 signaling to enhance ECFC migration through PI3K/AKT signaling [67]. Moreover, endothelial cell proliferation has been linked to the mobilization of intracellular Ca2⁺ [68,69]. Indeed, our enrichment analyses indicated a significant enrichment of proteins involved with calcium-mediated signaling. Thus, it is tempting to speculate that the increased presence of MCOLN1 on CD34⁻ ECFC could, in part, drive the clonogenic capacity and tube-forming functions of ECFC.

Following in vitro analyses, we began to speculate that CD34 expression may simply reflect the microenvironment of culture and cellular state during expansion. Therefore, CD34⁻ and CD34⁺ ECFC were transplanted subcutaneously within Matrigel plugs to assess endothelial cell function in vivo. Surprisingly, CD34⁻ and CD34⁺ ECFC generated vessel-like structures and occasionally appeared to integrate with murine endothelial cells. Thus, ECFC vessel-forming function retained vessel forming in vivo independent of CD34 expression.

In summary, our data suggest that CD34 expression on ECFC after culture expansion in vitro did not mark functionally enhanced EPC. Endothelial progenitor cell phenotype and functional capacity were altered after establishment in vitro and was dependent on culture density [11,70]. Furthermore, CD34⁺ and CD34⁻ ECFC subsets were functionally equivalent in vivo, despite measurable functional differences in colony and tube formation in vitro. These data suggest that surrogate in vitro analyses need to be supported by in vivo studies to assess the function of endothelial progenitor cells relevant to applications of regenerative medicine. Collectively, standardization for the isolation, culture, and phenotype should be assessed before potential use in vascular regenerative therapies.