Published OnlineFirst August 6, 2013; DOI: 10.1158/1078-0432.CCR-12-3685
Clinical Cancer Human Cancer Biology Research
Characterization of CD45 /CD31þ/CD105þ Circulating Cells in the Peripheral Blood of Patients with Gynecologic Malignancies
Hyun-Kyung Yu1,2, Ho-Jeong Lee1,5, Ha-Na Choi1, Jin-Hyung Ahn1, Ji-Young Choi3, Haeng-Seok Song3, Ki-Heon Lee3, Yeup Yoon1, Lee S. H. Yi2, Jang-Seong Kim4, Sun Jin Kim5, and Tae Jin Kim3
Abstract Purpose: Circulating endothelial cells (CEC) have been widely used as a prognostic biomarker and regarded as a promising strategy for monitoring the response to treatment in several cancers. However, the presence and biologic roles of CECs have remained controversial for decades because technical standards for the identification and quantification of CECs have not been established. Here, we hypothesized that CECs detected by flow cytometry might be monocytes rather than endothelial cells. þ Experimental Design: The frequency of representative CEC subsets (i.e., CD45 /CD31 , CD45 / þ þ þ þ CD31 /CD146 , CD45 /CD31 /CD105 ) was analyzed in the peripheral blood of patients with gyne- þ cologic cancer (n ¼ 56) and healthy volunteers (n ¼ 44). CD45 /CD31 cells, which are components of CECs, were isolated and the expression of various markers (CD146, CD105, vWF, and CD144 for endothelial cells; CD68 and CD14 for monocytes) was examined by immunocytochemistry. þ þ Results: CD45 /CD31 /CD105 cells were significantly increased in the peripheral blood of patients þ þ with cancer, whereas evaluation of CD45 /CD31 /CD146 cells was not possible both in patients with cancer and healthy controls due to the limited resolution of the flow cytometry. Immunocytochemistry þ þ analyses showed that these CD45 /CD31 /CD105 cells did not express vWF and CD146 but rather þ þ CD144. Furthermore, CD45 /CD31 /CD105 cells uniformly expressed the monocyte-specific markers þ þ CD14 and CD68. These results suggest that CD45 /CD31 /CD105 cells carry the characteristics of monocytes rather than endothelial cells. þ þ Conclusions: Our data indicate that CD45 /CD31 /CD105 circulating cells, which are significantly increased in the peripheral blood of patients with gynecologic cancer, are monocytes rather than endothelial cells. Further investigation is required to determine the biologic significance of their presence and function in relation with angiogenesis. Clin Cancer Res; 1–11. 2013 AACR.
Authors' Affiliations: 1Mogam Biotechnology Research Institute, Yongin; Introduction 2Department of Biological Science, Sungkyunkwan University, Suwon; 3Department of Obstetrics and Gynecology, Cheil General Hospital and Overcoming resistance to therapy is the ultimate goal of Women's Healthcare Center, Kwandong University College of Medicine, Seoul; 4Biomedical Translational Research Center, Korea Research Insti- the development of novel treatment modalities in cancer tute of Bioscience and Biotechnology, Daejeon, Republic of Korea; and (1). The biologic heterogeneity and genetic instability of 5Department of Cancer Biology, The University of Texas MD Anderson cancer cells are significant barriers for the design of effective Cancer Center, Houston, Texas therapies. Therefore, relatively more homogenous and Note: Supplementary data for this article are available at Clinical Cancer genetically stable host factors have been suggested as alter- Research Online (http://clincancerres.aacrjournals.org/). native targets (2). Angiogenesis, which is one of the com- Corresponding Authors: Jang-Seong Kim, Biomedical Translational mon and crucial steps in the development and progression Research Center, Korea Research Institute of Bioscience and Biotechnol- ogy, 111 Gwahangno, Yuseong-gu, Daejeon 305-806, Republic of Korea. of solid tumors, is a host-dependent process and, conse- Phone: 82-42-860-4270; Fax: 82-42-879-8498; E-mail: quently, has been introduced as an attractive target of cancer [email protected]; Sun Jin Kim, Department of Cancer Biology, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd, treatment (3). A significant number of drugs designed to Houston, TX 77030. Phone: 1-713-563-4653; Fax: 1-713-563-5489; interrupt the establishment of tumor-associated vasculature E-mail: [email protected]; and Tae Jin Kim, Department of Obstet- by neutralizing vasculogenic factors is currently in clinical rics and Gynecology, Cheil General Hospital and Women's Healthcare Center, Kwandong University College of Medicine, 1-19 Mukjeong-dong, trials and some of them have been approved for clinical use Jung-gu, Seoul 100-380, Republic of Korea. Phone: 82-2-2000-7577; Fax: in patients with cancer (4, 5). However, understanding the 82-2-2000-7183; E-mail: [email protected] mechanisms of angiogenesis and establishing validated doi: 10.1158/1078-0432.CCR-12-3685 markers that accurately reflect the pharmacologic effects of 2013 American Association for Cancer Research. antiangiogenic therapeutics remain major challenges (6).
www.aacrjournals.org OF1
Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2013 American Association for Cancer Research. Published OnlineFirst August 6, 2013; DOI: 10.1158/1078-0432.CCR-12-3685
Yu et al.
Therefore, establishing a method to validate the identity Translational Relevance of CECs detected by flow cytometry and to assess their Circulating endothelial cells (CEC) are recognized as a biologic significance is critical before expanding their marker of tumor angiogenesis and a predictor of prog- clinical use. nosis, as well as a target for therapy. However, their Considering the angiogenic role of monocytes and the presence in the peripheral blood of patients with cancer technical hurdles of flow cytometry, we hypothesized that and role in angiogenesis remain controversial. Our þ þ a subset of circulating cells detected by flow cytometry results showed that CD45 /CD31 /CD105 cells were þ using conventional CEC markers might be monocytes rath- significantly increased, whereas CD45 /CD31 / þ er than endothelial cells. To show this, we first evaluated CD146 cells were not detected in the peripheral blood the flow-cytometric techniques and markers currently in of patients with gynecologic cancer. In addition, the use and analyzed the frequency of representative CEC þ þ þ expression of monocyte-specific markers such as CD14 subsets (i.e., CD45 /CD31 , CD45 /CD31 /CD146 , þ þ þ þ and CD68 in CD45 /CD31 /CD105 cells suggested CD45 /CD31 /CD105 ; refs. 25–28) in the peripheral that they are monocytes rather than endothelial cells. blood of patients with gynecologic cancer and healthy Collectively, our data suggest that the accuracy of con- volunteers. To identify the genuine lineage of those cells, þ ventional flow-cytometric analyses for identifying CECs we isolated CD45 /CD31 cells (a common denominator should be meticulously reevaluated in its technical and of CECs) and assessed the expression of various markers for biologic aspects. Moreover, further investigation is nec- endothelial cells or monocytes by immunocytochemistry. essary to establish the biologic significance of the pres- þ þ ence of CD45 /CD31 /CD105 monocytes and their Materials and Methods function in relation to angiogenesis. Subjects Peripheral blood samples (1–2 mL) were collected from 44 healthy donors (12 men and 32 women; age, 28–54 years) and 56 patients with gynecologic cancer including 8 Because circulating endothelial cells (CEC) are likely to patients with endometrial cancer (age, 39–59 years), 24 contribute to new vessel formation (7) and their levels in the with cervical cancer (age, 30–71 years), and 24 with ovarian blood change in response to pro- or antiangiogenic drugs cancer (age, 23–67 years; Supplementary Table S1). All (8–10), the measurement of CECs (total CECs including healthy volunteers were free of any medications and had progenitor cells) has been regarded as a promising strat- no cardiovascular disease. The Institutional Review Board at egy for monitoring tumor angiogenesis. Several studies Kwandong University College of Medicine (Seoul, Republic reported a significant increase in the number of CECs in of Korea) approved all protocols, and informed consent was patients with cancer with progressive disease (i.e., lym- obtained from all subjects. phoma, breast cancer, renal cancer, etc.; refs. 11–13). In addition, studies have shown that CEC kinetics and Antibodies for flow cytometry viability correlate well with clinical outcomes of patients The following monoclonal antibodies directly conjugat- with cancer undergoing antiangiogenic treatment (14, ed with fluorescein isothiocyanate (FITC), phycoerythrin 15). (PE), peridinin chlorophyll A protein (PerCP), or allophy- However, the presence and role of CECs have been cocyanin (APC) were used for flow-cytometric analysis: controversial for decades because trials using CECs as a anti-CD31 FITC (WM-59 clone), anti-CD61 FITC (VI-PL2 diagnostic parameter or a therapeutic target did not clone), anti-CD3 PE (SK7 clone), anti-CD19 PE (HIB19 produce consistent results. Moreover, markers that accu- clone), anti-CD31 PE (WM-59 clone), anti-CD41a PE rately identify CECs have not been established because (HIP8 clone), anti-CD56 PE (MY31 clone), anti-CD146 PE many circulating cells such as monocytes express many of (P1H12 clone), anti-CD45 PerCP (2D1 clone), and anti- the same markers as endothelial cells (16). Consequently, CD14 APC (M5E2 clone). Isotype-matched FITC-, PE-, the detection and estimation of CECs have remained PerCP-, and APC-conjugated control antibodies were pur- challenging as they comprise a small proportion of chased from BD Biosciences. Anti-CD105 PE (SN6 clone) peripheral blood cells, and there is no consensus in the and isotype-matched PE-conjugated control antibodies immunophenotype of CEC (17). The application of var- were purchased from Serotec. Anti-CD31 APC (WM-59 ious combinations of markers or different enumeration clone) and isotype-matched APC-conjugated control anti- techniques to the detection of CECs has produced dis- bodies were purchased from eBioscience. crepant results in the amount and immunophenotype of CECs (18–20). Furthermore, certain authors have ques- Preparation of peripheral blood mononuclear cells tioned whether CECs detected by flow cytometry are Peripheral blood was collected from healthy volunteers authentic endothelial cells and their function in angio- and patients with cancer using EDTA as an anticoagulant genesis, if any (17, 21). Recent studies have suggested and processed within several hours after collection as fol- that the actual angiogenic cell types incorporated into lows: whole blood was diluted 1:1 (vol/vol) with PBS newly formed vessels are myeloid cells such as monocytes containing 0.5% bovine serum albumin (BSA) and 2 (22–24) rather than CECs and/or their progenitor cells. mmol/L EDTA and overlaid onto an equal volume of Ficoll
OF2 Clin Cancer Res; 2013 Clinical Cancer Research
Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2013 American Association for Cancer Research. Published OnlineFirst August 6, 2013; DOI: 10.1158/1078-0432.CCR-12-3685
Characterization of Circulating Endothelial Cells
Paque (GE healthcare). Samples were centrifuged at 1,800 FcR blocking reagent and stained with fluorescence-labeled rpm for 25 minutes at room temperature with no brake. The monoclonal antibodies against CD45 and CD31. The cells mononuclear cell layer was carefully collected and washed were fixed with 4% PFA and then sorted with a FACS Aria twice with cold PBS containing 0.5% BSA and 2 mmol/L flow cytometer. A 70-mm nozzle (BD Biosciences), a sheath EDTA at 4 C. Red blood cells were lysed with 0.38% pressure of 20 to 25 pounds per square inch, and an ammonium chloride solution. The final mononuclear cell acquisition rate of 2,000 to 3,000 events per second were preparation was resuspended with PBS containing 0.5% used according to the guidelines for FACS Aria users (BD BSA and 2 mmol/L EDTA and then subjected to flow- Biosciences). cytometric analysis. The viability of the mononuclear cells used for the analyses was determined by the dye exclusion Immunofluorescence staining test, and cells with a viability of 99% or more were used for Immunocytochemical fluorescence labeling of cells was further experiments. conducted as previously described (24). Briefly, CD45 / þ CD31 cells isolated by flow cytometry were cytospun onto Flow cytometry and cell sorting glass slides and washed 3 times with PBS for 3 minutes. To Isolated peripheral blood mononuclear cells (PBMC; 107 stain intracellular antigens, cells were permeabilized with cells per mL of blood) were pretreated with FcR blocking 0.5% Triton X-100 (Sigma Chemical Co.). To prevent cross- reagent (Miltenyi Biotec) to block nonspecific antibody reaction with antibodies used to stain cells in the flow- binding and incubated on ice for 25 minutes with a panel cytometric analysis, Fab-fragment blocking was conducted of monoclonal antibodies (summarized in Table 1). Cells overnight at 4 C with an antibody from the same host were washed with PBS containing 0.5% BSA and 2 mmol/L species of antibody, which was a F(ab0)2 fragment from EDTA and fixed with 4% paraformaldehyde (PFA; Electron goat anti-mouse immunoglobulin G (IgG) (Jackson Immu- Microscopy Sciences). The antibody-labeled cells were ana- noResearch Laboratories) in this study. After the blocking lyzed using a FACS Aria flow cytometer (BD Biosciences) step, cells were stained with the following antibodies: equipped with 2 lasers (488 nm and 633 nm). Data were mouse anti-human CD105 monoclonal antibody (mAb; analyzed with FlowJo software (Tree Star, Inc.,) or FACS 1:100, Serotec), mouse anti-human CD146 mAb (1:100 Diva (BD Biosciences). For the analysis of CEC candidates, dilution, Chemicon), mouse anti-human CD144 mAb at least 100,000 singlet lymphocytes were isolated and (1:100, Reliatech GmbH), mouse anti-human CD68 mAb þ þ the frequencies of CD45 /CD31 , CD45 /CD31 / (1:100 dilution, DAKO), polyclonal rabbit anti-human þ þ þ CD146 , or CD45 /CD31 /CD105 cells were analyzed vWF Ab (1:100 dilution, DAKO), or mouse anti-human and expressed as a percentage of the singlet lymphocyte CD14 mAb (1:20 dilution, DAKO), followed by labeling þ population. For sorting of CD45 /CD31 cells, PBMCs with the corresponding secondary antibodies conjugated prepared from patients with cancer were pretreated with with FITC or Texas Red. For double-staining experiments,
Table 1. Antibody panels used for flow-cytometric analysis and isolation of CD45 /CD31þ cells
FITC PE PerCP APC For immunotyping Controls for compensation IgG1 IgG1 IgG1 IgG2a CD31 IgG1 IgG1 IgG2a IgG1 CD31 IgG1 IgG2a IgG1 IgG1 CD45 IgG2a IgG1 IgG1 IgG1 CD14 Samples CD31 CD3 CD45 CD14 CD31 CD19 CD45 CD14 CD31 CD56 CD45 CD14 CD31 CD146 CD45 CD14 For CEC detection Controls for compensation IgG1 IgG1 IgG1 - CD31 IgG1 IgG1 - IgG1 CD31 IgG1 - IgG1 IgG1 CD45 - FMO control CD31 IgG1 CD45 - Samples CD31 CD146 CD45 - CD31 CD105 CD45 - For isolation of CD45 /CD31þ cells Controls for compensation - - IgG1 IgG1 - - IgG1 CD31 - - CD45 IgG1 Sample - - CD45 CD31
www.aacrjournals.org Clin Cancer Res; 2013 OF3
Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2013 American Association for Cancer Research. Published OnlineFirst August 6, 2013; DOI: 10.1158/1078-0432.CCR-12-3685
Yu et al.
the protein and fragment blocking steps were repeated event frequencies. In line with the study by Cui and collea- before treating with the second primary antibody to gues (31), gating with FMO controls was shown to be a prevent cross-reaction. Cell nuclei were counterstained more efficient method to increase the accuracy and speci- with Hoechst 33342 dye (Invitrogen). The slides were ficity of the positive signals in polychromatic flow-cyto- then washed 3 times with PBS for 3 minutes each and metric detection of rare events, such as CECs, than the use of mounted in Vecta-shield (Vector Laboratories). Images isotype controls. were acquired with a LSM510 Meta DuoScan confocal On the basis of these results, we established a gating system (Zeiss). strategy to determine the frequencies of CECs in PBMCs, as described in Fig. 2B. In brief, cells were stained with Statistical analysis a panel of antibodies in parallel with FMO controls. The The Mann–Whitney U test was used to determine the singlet lymphocyte population was identified on a FSC/ statistical significance of differences in the frequencies of SSC plot and subgated onto a bivariant antigen plot to þ CEC candidates between the peripheral blood collected identify CD45 /CD31 cells. These cells were further sub- þ þ from patients with cancer and healthy volunteers. All sta- gated to identify the corresponding CD146 or CD105 tistical tests were two-sided. P values less than 0.05 were subpopulation. considered significant. Flow-cytometric analysis of CECs in PBMCs Results A number of protein markers including CD31, CD34, Establishment of flow cytometry gating strategies for CD105, CD146, CD144, and VEGF receptor-2 have been the measurement of circulating endothelial cells used to define CECs. However, there is no truly specific First, we found that monocytes showed higher levels of marker to identify CECs because those markers are also autofluorescence than lymphocytes (data not shown), indi- expressed in other type of cells (32). A generally accepted þ cating that they should be analyzed separately for fluores- definition of CECs is CD45 /CD31 cells expressing cence compensation. In flow-cytometric analysis, the detec- CD146 or CD105 (11), but this definition needs to be tion of equal levels of autofluorescence in positive and modified. According to the gating strategy described above, þ negative populations for each single stain indicates that the we examined the frequencies of CD45 /CD31 , CD45 / þ þ þ þ fluorescent compensation is appropriate (29). As an initial CD31 /CD146 , and CD45 /CD31 /CD105 cells in the step to establish the efficient gating strategies for the detec- PBMCs of patients with cancer (n ¼ 56) and healthy tion of CECs by flow cytometry, we attempted to determine volunteers (n ¼ 44) by flow cytometry. The frequency of þ þ the subset(s) of PBMCs expressing the CD45 /CD31 CD45 /CD31 cells was significantly higher in the singlet phenotype. PBMCs were plotted according to the forward FSClow/SSClow population of patients with cancer (median, scatter (FSC) versus side scatter (SSC) profiles and FSClow 1.365%; range, 0.110–26.85%) than in that of healthy /SSClow (Fig. 1A, left), FSChigh/SSCmid (Fig. 1B, left), and volunteers (median, 0.183%; range, 0.027–3.980%; P < FSCmid/SSChigh (data not shown) fractions were gated as 0.0001) as shown in Fig. 3A. In contrast with a previous þ þ lymphocytes, monocytes, or granulocyte subpopulations, report (11), the frequency of CD45 /CD31 /CD146 cells respectively. Subpopulations of cells with different CD45 in healthy volunteers (median, 0%; range, 0–0.003%) and and CD31 expression patterns were further analyzed for the patients with cancer (median, 0.001%; range, 0–0.016%) expression of CD146 or lineage-specific markers including was lower than the cutoff values of the FMO control group CD3 (T lymphocyte), CD14 (monocyte), CD19 (B lym- (median, 0.007%; range, 0–0.021%), indicating that esti- phocyte), and CD56 (NK cells). The FSClow/SSClow subset mation of the frequency of those cells is not possible both in þ was mostly composed of CD45 /CD31 cells expressing cancer patients and healthy controls due to the limited CD3, CD19, or CD56 antigens (Fig. 1A), whereas the resolution of the flow cytometry. Actually, when the isotype þ þ þ FSChigh/SSCmid subset mainly included CD45 /CD31 control was used, the frequencies of CD45 /CD31 / þ þ cells expressing the CD14 antigen (Fig. 1B). CD45 /CD31 CD146 cells were significantly higher (median, 0.041%; cells were detected only in the FSClow/SSClow fraction (Fig. range, 0.007–0.132%; Fig. 2C), underscoring the impor- 1A) but neither in the FSChigh/SSCmid (Fig. 1B) nor FSCmid/ tance of using FMO controls. SSChigh fractions (data not shown). Meanwhile, CD146 expression was detected only in the þ In the polychromatic flow-cytometric analysis used in the CD45 /CD31 subpopulation of the FSClow/SSClow popu- present study, adequate threshold was assessed by fluores- lations (Fig. 1A). These CD146-positive cells also expressed cence-minus-one (FMO) gating, which consists of analyz- CD3 (data not shown). These results, together with those of ing cells stained with all antibodies except the one being previous studies showing that CD146 is present in a subset þ tested (30). To investigate the effects of the gating controls of activated T lymphocytes (21, 33), indicate that CD146 þ on the actual event frequencies, the flow-cytometric analysis cells in the CD45 /CD31 lymphocyte subpopulation are T of the singlet lymphocyte fraction was conducted using lymphocytes. þ þ either the isotype control or the FMO control (Fig. 2A). The frequency of CD45 /CD31 /CD105 cells in Because the negative threshold of the FMO control (Fig. 2A, healthy volunteers (median, 0.003%; range, 0–0.027%) b) was higher than that of the isotype control (Fig. 2A, a), showed no significant differences statistically when com- gating with FMO controls could decrease the false-positive pared with FMO control (P > 0.05). On the other hand, the
OF4 Clin Cancer Res; 2013 Clinical Cancer Research
Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2013 American Association for Cancer Research. Published OnlineFirst August 6, 2013; DOI: 10.1158/1078-0432.CCR-12-3685
Characterization of Circulating Endothelial Cells
A – /CD31 + CD45 CD56 CD45 CD45 CD45 CD3 CD14 CD19 CD146 + /CD31 + CD45 CD56 CD45 CD45
+ –
CD45 /CD31 CD45 CD3 CD14 CD19 CD146 + CD45+/CD31+ /CD31 – CD45 CD56 CD45 CD45 CD45–/CD31+ CD45
CD3 CD14 CD19 CD146
B – /CD31 + CD45 CD56 CD45 CD45 CD45 CD3 CD14 CD19 CD146 + /CD31 + CD45 CD56 CD45 CD45 CD45+/CD31–
CD45 CD3 CD14 CD19 CD146
CD45+/CD31+ + /CD31 – CD45–/CD31+ CD45 CD56 CD45 CD45 CD45
CD3 CD14 CD19 CD146
Figure 1. CD45 and CD31 expression in PBMCs obtained from patients with cancer. PBMCs were initially gated into (A) FSClow/SSClow and (B) FSCmid/SSCmid fractions to include mainly lymphocytes and monocytes, respectively, and singlet cells were selected on the basis of FSC-Height versus FSC-Area plots. The cells were then subdivided into CD45þ/CD31 , CD45þ/CD31þ, and CD45 /CD31þ cells. The cells were further analyzed for the expression of lineage-specific markers (CD3, CD14, CD19, and CD56) and CD146. Lymphocytes gated on FSC/SSC plots were mostly CD45þ/CD31 (A, left), whereas monocytes were mostly CD45þ/CD31þ (B, left). In addition, CD45þ/CD31 cells were composed of CD3þ, CD19þ, CD56þ, and CD146þ cells (A and B, top right), whereas CD45þ/CD31þ cells were mostly CD14þ (A and B, middle right), regardless of lymphocyte and monocyte gating based on the FSC/SSC plot. In particular, FSClow/SSClow gating showed CD45 /CD31þ cells (A and B, left) that did not express lineage-specific markers or CD146 (A and B, bottom right).