|Home | About | Journals | Submit | Contact Us | Français|
The development of tumor vasculature is thought to occur through two complementary processes: sprouting angiogenesis from preexisting blood vessels of the host, and vasculogenesis, which involves the spontaneous development of vessels through specific recruitment, differentiation, and vascular incorporation of circulating endothelial cells (EC), endothelial progenitor cells (EPC), or potentially bone marrow-derived cells. Recent reports, however, have challenged the belief that bone marrow-derived cells contribute to tumor neovascularization, claiming an exclusive role for sprouting angiogenesis in tumor blood vessel development. In the present study, we explored the recruitment behavior of bone marrow-derived lin-c-kit+Sca-1+ stem cells to subcutaneously implanted Lewis lung carcinoma in a syngeneic bone marrow transplantation model. We observed that although lin-c-kit+Sca-1+ and their derived cells demonstrate significant recruitment to carcinomas in vivo, they do not appear to functionally contribute to tumor neovascularization. Furthermore, our results support the hypothesis that new vessel formation in carcinomas occurs primarily through endothelialization from adjacent and preexisting vasculature.
Vascularization is inherent to tumor growth, development, and metastasis. Tumor vasculature develops through two complementary processes: sprouting angiogenesis from preexisting blood vessels of the host [1,2], and vasculogenesis, the spontaneous development of vessels through specific recruitment behaviors of circulating endothelial cells (EC), endothelial progenitor cells (EPC), or potentially bone marrow-derived cells, followed by differentiation and vascular incorporation [3–9]. Recent reports, however, have challenged the belief that bone marrow-derived cells contribute to tumor neovascularization, claiming an exclusive role for sprouting angiogenesis in tumor blood vessel development [10,11].
Specific subsets of the adult bone marrow appear to be involved in neoangiogenic processes in a variety of physiological and pathologic settings [3,12,1,13–18]. More specifically, murine bone marrow-derived EPCs, bearing the phenotype lin-c-kit+Sca-1+VE-cadherin+VEGFR2+, have been shown to corecruit to tumors with lin-c-kit+Sca-1+Thylo hematopoietic stem cells (HSC) and hematopoietic progenitor cells (HPC), especially of the VEGFR1+ myeloid phenotypic subset, and participate in tumor neovascularization [6,19,20]. A role for ex vivo expanded embryonic EPC in vasculogenesis has also been explored in detail [21,22]. The dynamic mobilization of these proangiogenic progenitor cells from the bone marrow compartment into the circulation involves a chemokine-mediated and cytokine-mediated activation of proteases, in combination with the upregulation of various adhesion molecules . Recent reports have indicated a role for bone marrow eNOS in the mobilization of these progenitor and stem cells [24,25]. The recruitment of these recently mobilized cells to sites of tissue injury or pathology also must require specific adhesion molecules, chemokines, and associated molecules. A recent study has indicated a role for selectins in the multistep recruitment and differentiation of embryonic EPC within tumor microvasculature , and neuronal progenitor cell recruitment has been demonstrated to be dependent on a combination of α4-integrin and SDF-1α .
In the present study, we explored the recruitment behavior of bone marrow-derived lin-c-kit+Sca-1+ stem cells to subcutaneously implanted Lewis lung carcinoma (LLC) in a syngeneic bone marrow transplantation model. Although our data conclusively demonstrate that lin-c-kit+Sca-1+ and their derived cells traffic to carcinomas in vivo, our results do not reveal a role for these recruited cells in tumor neovascularization. In fact, our results support the hypothesis that new vessel formation arises through endothelialization from adjacent and preexisting vasculature.
RPMI 1640, DMEM, l-glutamine, penicillin-streptomycin, sodium bicarbonate solution, Dulbecco's PBS (DPBS)- (no Ca2+/Mg2+), DPBS+ (with Ca2+/Mg2+), and HBSS were from Cambrex Bio Products (Walkersville, MD). Antibiotin microbeads and lineage depletion Midi MACS kit were obtained from Miltenyi Biotec (Auburn, CA). FCS was from Cellgro (Herndon, VA). Additional molecular biology-grade chemicals were from either Fisher Scientific (Suwanee, GA) or Sigma Chemical Co. (St. Louis, MO).
Mouse Lineage Panel (biotinylated antibodies; CD3e, CD11b, CD45R, Ly6G/C, and TER-119), purified rat anti-mouse CD31 (clone MEC13.3, rat IgG2aκ), CD16/32 (Fc γRIII/II, clone 2.4G2, rat IgG2bκ), phycoerythrin (PE)-conjugated rat antimouse Ly-6A/E (Sca-1, clone D7, rat IgG2aκ), and allophycocyanin (APC)-conjugated rat anti-mouse CD117 (c-kit, clone 2B8, rat IgG2bκ) were obtained from BD Pharmingen (San Jose, CA). Purified anti-mouse/human smooth muscle actin (SMA) was obtained from Lab Vision Corp. (Fremont, CA), Cy3 goat-antirat IgG (H+L) and Texas Red anti-rabbit IgG (H+L) were obtained from Caltag Laboratories (Burlingame, CA) and Vector Laboratories (Burlingame, CA), respectively. Streptavidin-PE was from Miltenyi Biotec.
C57BI/6 mice were purchased from Charles River Park (Wilmington, MA). C57BL/6-Tg (ACTbEGFP) 1Osb/J EGFP-transgenic mice were obtained from Jackson Laboratories (Bar Harbor, ME). All mice were maintained and bred in our approved institutional pathogen and viral-free housing facilities. Mice were used between 7 and 9 weeks of age (20–25 g). For imaging experiments, animals were anesthesized either through intraperitoneal injection with ketamine/xylazine (80/12 mg/kg) or isofluorane anesthesia, and sacrificed by CO2 asphyxiation as approved by the panel on euthanasia at the American Veterinary Association.
LLC cells were a generous gift from Dr. Timothy Browder (Children's Hospital, Boston, MA). Cells were cultured in DMEM supplemented with 10% FCS, l-glutamine, sodium bicarbonate, and penicillin-streptomycin, grown to confluence and split biweekly in a 1:5 split ratio. For tumor implantation studies, the cells were dissociated enzymatically, washed by centrifugation, and resuspended in normal saline for injection.
Murine bone marrow cells were recovered from the tibias and femurs of C57BI/6 mice by flushing with RPMI 1640, disaggregated, and filtered through a 30-µm nylon filter. Cells were then washed, resuspended to 1 x 108 cells/ml in DPBS- containing 1% heat-inactivated FCS, and incubated for 5 minutes with anti-mouse CD16/CD32 (1 µl/106 cells) on ice, followed by incubation with mouse lineage panel antibodies (2 µl/106 cells) for 15 minutes on ice. The cells were washed twice in DPBS- containing 0.5% BSA and 20 mM EDTA, pH 7.2, and then selected with streptavidin microbeads (2 µl/106 cells) for 15 minutes at 4°C, washed twice, and separated magnetically in a lineage depletion column to recover the unbound lineage-negative fraction.
Whole bone marrow (wbm), BMlin-, and lineage positive cells (BMlin+) were separately incubated for 30 minutes on ice with a biotinylated cocktail of antibodies directed against lineage markers (lineage panel). In some instances, the cells were also incubated in parallel with PE-Sca-1 Ab and APC-c-kit Ab. The cells were washed twice in RPMI 1640 containing 5% FCS and incubated with streptavidin-PE (dilution 1:11) for 5 to 7 minutes at 4°C for lineage marker recognition. The cells were washed 3x and fixed in 2% paraformaldehyde. Fluorescence of 10,000 cells per sample was then determined in multiple channels on a FACSCalibur flow cytometer (Becton Dickinson, Mountain View, CA) and presented as quadrant plots and single parameter histograms on a four-decade scale.
C57BI/6 mice were implanted bilaterally with subcutaneous injections of 5 x 106 LLC in 100 µl of normal saline. After 10 days, BMlin- cells were isolated, resuspended to 2.5 x 106 /ml in RPMI 1640, and incubated with 0.5 mCi of 111indium oxine (Amersham Health, Princeton, NJ) for 60 minutes at 37°C. Cells were then washed twice to remove unbound reagent and tested for viability using trypan blue exclusion. Labeling efficiency and sample radioactivity were estimated in a gamma counter. Pools of 200,000 111indium-labeled BMlin- cells were then injected into six LLC tumor-bearing C57BI/6 mice through the tail vein. Twenty-four and 72 hours postinjection, three animals each were sacrificed, their organs harvested, and radioactive counts were estimated in a gamma counter. Radioactivity in each organ is expressed as a fraction of the injected dose per organ.
Blood volume fractions for different organs were estimated by combining experimental volume and density measurements with blood and interstitial volume fraction data based on a mathematical model . Experimental radioactivity measured in a small blood sample drawn from the animal prior to sacrifice was then scaled up to account for total blood radioactivity per organ. The percent injected dose/organ could therefore be corrected for the vascular contribution in that respective organ. In control experiments, similar labeling and injection procedures of BMlin+ cells were performed.
Ten C57BL/6 WT mice were irradiated at 10.25 Gy and, 1 day later, injected intravenously with a combination of 4.5 x 105 BMlin+ carrier bone marrow cells recovered from syngeneic wild-type C57BI/6 mice, and 5 x 104 EGFP-BMlin- stem cells isolated from the wbm of EGFP-transgenic mice to generate EGFP chimeras (WT mice carrying an EGFP hematopoietic system, n = 3 for each time point). On days 3 and 17 postinjection, these mice were subcutaneously implanted with LLC as described previously. Tumors were also implanted bilaterally in WT nonirradiated C57BL/6 control mice (n = 2) and EGFP-transgenic mice (n = 3). On days 10 to 14 following tumor implantation, imaging was performed and the tumors were harvested for correlative histologic analysis.
For epifluorescence imaging studies, both the chimeric and WT control mice were anesthesized, and tumors were exposed through fine incisions in the skin. The tumors, still attached to the animal, were then inverted on a glass coverslip and imaged on an inverted Axiovert 100TV fluorescence microscope (Carl Zeiss Imaging, Inc., Thornwood, NY) fitted with a Cool Snap HQ camera (Roper Scientific, Tucson, AZ). Images were acquired at x10 and x20 magnification under both bright light and FITC channels for subsequent image analysis. The animals were then sacrificed and the tumors excised for indirect immunofluorescence staining. Based on previously published methods , excised LLC solid tumors were fixed in 2% paraformaldehyde, equilibrated in 18% sucrose, embedded in OCT, snap-frozen in liquid N2, and serially sectioned in 7- to 10-µm sections. Tissue sections were blocked with DPBS+ containing 1% mouse serum for 30 minutes at 37°C and then incubated for 45 minutes at 37°C with primary monoclonal antibody (mAb) directed against mouse CD31 and SMA (20 µg/ml), rinsed 3x in DPBS+, and detected with Cy3 or Texas Red-conjugated secondary mAbs (1/100). Coverslips were then washed and mounted using Vectashield (Vector Laboratories). Fluorescence staining was visualized using a Nikon Eclipse 80i upright microscope equipped for fluorescence and coupled to a Photometrics Cascade 512B camera (Roper Scientific). Images were acquired with x10 and x20 objectives.
We magnetically depleted wbm isolated from adult C57BI/6 mice to recover pluripotent lineage negative (lin-) stem cells. As revealed in a histogram plot for lineage marker recognition (Figure 1), the lin- subset overlaps with the negative isotype control, whereas the lineage positive population (lin+) representing the differentiated components of the wbm is highly positive for lineage marker expression. The wbm primarily contains partially differentiated submature cells, as recognized by the lineage markers CD3e, CD11b, CD45R, Ly6G/C, and TER-119, with a minor fraction of immature pluripotent cells that are lineage negative. As indicated in the lower panels, >95% of the recovered lin- stem cell population is positive for CD117 (c-kit, panel B) and Sca-1 (panel C) expression. Taken together, our characterization data reveal a highly enriched adult stem cell population of the phenotype lin-c-kit+Sca-1+, which we termed BMlin- cells.
To explore the hypothesis that adult BMlin- cells home to implanted tumors in vivo, six adult C57BI/6 mice bearing 10-day-old subcutaneous LLC tumors were injected intravenously with 111indium oxine-labeled stem cells and sacrificed 24 and 72 hours after stem cell injection. Radioactive counts in harvested organs were expressed as a percentage of the total injected dose. The radioactivity provides a direct measure of the number of cells within each organ. Our biodistribution study (Figure 2) revealed that approximately 2% of the intravenously injected stem cells trafficked to tumors within 24 hours postinjection (shaded bars) and were retained in the tumor for at least 72 hours (open bars). In addition to tumor homing, approximately 50% of the injected dose was cleared by the liver, whereas <5% of the injected stem cells was found in the kidneys, lungs, and fat. The blood contained approximately 3% of the injected radioactive dose, most likely due to labeled circulating cells; thus, the percent injected dose per organ was corrected for the predicted radioactive contribution of the blood pool in that organ. Our data suggest, therefore, that the majority of injected BMlin- cells were “noncirculating” organ or tumor resident cells at the times studied.
The biodistribution studies indicated a level of BMlin- cell recruitment to LLC tumors in the initial 72 hours following adoptive transfer of labeled cells; however, these studies did not provide specific information on the intratumoral distribution of the recruited cells. To address this question, we generated EGFP/C57BI/6 chimeric animals expressing EGFP in their hematopoietic system and used these animals as tumor hosts. At 14 days following tumor injection, the live tumors were imaged to determine the contribution of hematopoietic cells to the vasculature of the subcutaneous tumors (Figure 3). Real-time epifluorescence imaging of exteriorized subcutaneous tumors in EGFP-chimeric mice demonstrated recruitment of BMlin- cells and their progeny to LLC in vivo. Although a majority of the tumor-recruited cells appeared to reside in the extravascular tissue, we did observe a few firmly arrested EGFP cells within the tumor vasculature (Figure 3C, inset); however, none of the fluorescent cells was observed to contribute directly to vessel structure. A similar recruitment behavior was observed for tumors implanted either within 3 days of bone marrow transplantation when only EGFP-BMlin- cells were present in the circulation, or over 2 weeks following bone marrow repopulation when mature progeny cells were also present, suggesting that both BMlin- stem cells and progeny cells are recruited to LLC tumors. As expected, no EGFP signal was detected in the vasculature or in the surrounding tumor tissue in WT control mice (Figure 3A).
To further interrogate the contribution of BMlin- cells, or their progeny, to tumor composition, we performed correlative histology of the same tumors. Following epifluorescence imaging, tumors from WT control mice and EGFP- chimeras were harvested, sectioned, and stained for CD31 and SMA (Figure 4) to identify endothelial and smooth muscle cells of the tumor vasculature. Dual-channel images of tumor sections from EGFP-chimeric mice confirmed the presence of recruited EGFP cells (green) within tumor tissue (28 cells/mm2), but recruited cells did not coregister with staining for components of the tumor vasculature (red), indicated by the lack of EGFP correlation with CD31 (Figure 4, panel C) or SMA (panel D). In statistical analysis of multiple tissue sections, a maximum of 6 cells/mm2 was observed to costain with CD31 and EGFP, less than 20% of the total EGFP cells per square millimeter, and no cells costained with SMA. In contrast, when LLC were injected into transgenic ubiquitously EGFP-expressing animals, indirect immunofluorescence staining revealed that all of the blood vessels (red) in these tumors, as detected by CD31 (panel E) and SMA (panel F), were EGFP+ (green). As expected, no EGFP cells were detected in tumors from WT control mice (panels A and B). In entirety, these data do not support a role for bone marrow-derived stem cells in the vascularization of LLCs, but in fact indicate that vessel growth occurs through the participation of adjacent preexisting vasculature.
A number of recent reports have suggested a role for bone marrow-derived cells in tumor vasculogenesis. The relevant subsets of the adult wbm that appear to functionally participate in neovascularization are the angioblast-like EPC and the facilitative HSC or HPC cells [29,3,4,6,7,30,8]. However, the degree of recruitment of these cells has not been effectively quantitated, the mechanisms of recruitment are not understood, and a specific role for these recruited cells in tumor vessel growth and development has not been fully elucidated. Thus, we employed a primitive lin-c-kit+Sca-1+ stem cell population (BMlin-) that is phenotypically inclusive of these specific neoangiogenic subsets, in a syngeneic model of stem cell recruitment to tumors, to both quantify BMlin- recruitment and address the question of BMlin- function in tumor neovascularization.
From our biodistribution data (Figure 2), we have determined quantitatively that BMlin- cells are recruited to subcutaneously implanted LLCs, and that these cells are present in the tissue and not simply circulating through the tumor vasculature. However, the fraction of injected cells that are recruited to the tumor environment is far lower than that observed for recruitment of human CD34+ cells to the bone marrow in immunodeficient mice , for recruitment of antigen-specific CD8+ T cells to antigen-expressing tumors , or for the recruitment of neuronal progenitors to the same LLCs in vivo . These data suggest that BMlin- cells exhibit a limited potential for the contribution to actively and aggressively growing tumors. However, it could be argued that very few long-term repopulating BMlin- cells are actually required to initiate these events, just as very few true HSC are required to repopulate the bone marrow. It is possible that a few BMlin- cells are recruited, and then go on to expand and differentiate in the tumor environment. The biodistribution data alone are not capable of determining the degree of cell proliferation, and thus we developed the bone marrow EGFP-chimeric animals to address these questions specifically.
Fluorescence imaging of live LLCs implanted into EGFP-chimeric mice at two different time points following bone marrow transplantation, coupled with indirect immunofluorescence staining of the harvested tumor sections, revealed a number of resident EGFP cells within the tumor mass. However, other than the presence of a few arrested EGFP cells within the tumor circulation, our data do not support a role for the recruited hematopoietic cells in tumor neovascularization, evident by the lack of coregistration of CD31/SMA staining and EGFP expression in tumors carried by EGFP-chimeric animals. However, in EGFP-transgenic mice that express EGFP ubiquitously, all tumor vessels identified exhibited both EGFP expression and staining for CD31 and SMA. Together, these data support the hypothesis that tumor vessel growth in LLCs occurs through sprouting of the existing host circulation—not through spontaneous vasculogenesis of circulating progenitor cells. Our observations are supported by a recent elegant study that examined the recruitment of Tie2-expressing mononuclear EPC to tumors . The authors observed Tie2-expressing cells in the tumor mass and in close proximity to tumor vessels after long-term growth (12 weeks) of recipient tumors, but did not observe any contribution of these cells to the tumor vasculature directly.
Although the present study was supported by recent reports, there remains a significant controversy as to the functional incorporation of hematopoietic cells in sites of neovascularization, and how these conflicting reports may be reconciled. There are specific and significant differences in the tumor models exploited in these studies, which may go some way toward explaining the differing observations reported. Probably of most critical consequence is the type of tumor used. Recent reports that demonstrated vascular incorporation of EPC or other hematopoietic cells have used either colon cancer cell lines  or tumors of neuronal origin [4,21], which may represent a significant departure from the LLC model employed in the current study. Even more interesting are the reports indicating that tumor growth is significantly retarded in the absence of a normally functioning hematopoietic system, or when hematopoietic subsets have been modified to limit their recruitment or potential [4,6,9]. At initial interpretation, these data would suggest that hematopoietic cells do contribute directly to tumor vessel growth; however, recently, it has been hypothesized that HSC or HPC represent a “helper” subset that directs and stabilizes neovascularization within tumors, but does not itself participate in vasculogenesis . This hypothesis is supported not only by the retarded growth of tumors in the absence of functional HSC, but also by the location of these cells, in close proximity to new vessels in the developing tumor mass.
In conclusion, we have demonstrated that whereas long-term repopulating stem cells are recruited in low levels to subcutaneously implanted LLCs in syngeneic animals, neither them nor their progeny contributes directly to vasculo-genesis in this model. Together, these data would suggest caution in interpretation of the role of these cells in any site of spontaneous or induced neovascularization.
The authors would like to thank Herlen Alencar and Dong-Eog Kim for their assistance with intravenous injections and epifluorescence imaging, respectively. E.B.F. is a Feodor Lynen Research Fellow of the Alexander von Humboldt Foundation.
1This work was supported, in part, by grants from the National Institutes of Health [CA96978 (J.R.A.); CA85240, CA86355, and CA79443 (R.W.); and HL65584 (R.E.G.)].