At the time of transplantation, tumor fragments contain “passenger” cells: endothelial cells and other stromal cells from the original host. Here, we investigated the fate of genetically labeled endothelial and nonendothelial stromal cells after transplantation in syngeneic mice. We report that angiogenic stroma associated with tumor or adipose tissue persists when transplanted, remains functional, and governs the initial neovascularization of grafted tissue fragments for more than 4 weeks after implantation. Surprisingly, the passenger endothelial cells survive longer than other stromal cells, which are replaced by host-activated fibroblasts after 3 weeks. The transplantability of tumor stroma suggests that the angiogenic potential of a tumor xenograft, which determines its viability, depends on the presence of passenger endothelial cells and other stromal cells within the xenograft. These studies of tumor tissue transplantation provide a platform for exploring the role of epithelial–stromal interactions in studies of tumor heterogeneity and drug resistance.
In recent years, intense research has begun to reveal the intricate relationship between tumors and their resident host stromal cells (1, 2). To grow, spontaneously arising tumors require neovascularization: During cancer progression, new blood vessels develop from preexisting adjacent vasculature or by direct recruitment of progenitor cells (3, 4). The role and phenotype of the host stromal cells in angiogenesis is just beginning to emerge (5, 6, 7, 8, 9, 10).
Tumor xenograft/isograft models often rely on the transplantation of tumor tissue fragments that already possess endothelial cells and nonendothelial stromal cells. We have previously demonstrated that neoplastic cells from these fragments proliferate in the new host and activate host fibroblasts (5), but the fate of the transplanted endothelial cells and their role in tumor growth is only now being uncovered (11), whereas the fate of the nonendothelial stromal cells remains unknown. To evaluate the survival of non-neoplastic, tumor-associated “passenger” stromal cells, we grew tumors by injecting neoplastic cells into newly generated transgenic mice harboring green fluorescent protein under the control of the ubiquitous EF1α promoter. In subsequent experiments, we dissected the contribution of specific cell populations by using transgenic mice expressing green fluorescent protein under the Tie2 promoter (for labeling of endothelial cells), and the vascular endothelial growth factor (VEGF) promoter (for activated fibroblasts; Fig. 1 A). After transplanting fragments of these source tumors into wild-type, immunocompetent, syngeneic mice, we used intravital microscopy and histology to monitor green fluorescent protein expression in the transplanted tissue. In separate experiments, we monitored the fate of passenger stromal cells in transplanted adipose tissue, a highly angiogenic non-neoplastic tissue.
Materials and Methods
Cell Lines and Transgenic Mice.
We used the TG1-1 mammary carcinoma cell line, which was established from spontaneous tumors that arose in transgenic FVB mice carrying the c-neu oncogene under the control of a mouse mammary tumor virus promoter (12). To generate source tumors, 1 × 106 neoplastic cells were suspended in 50 μL of HBSS (Invitrogen, Grand Islands, NY) and injected into the mammary fat pads of transgenic mice that expressed green fluorescent protein driven by the ubiquitous EF1α promoter (EF1α-green fluorescent protein/FVB). The EF1α-green fluorescent protein mice were generated in the FVB background using a construct with EF1α promoter sequences driving green fluorescent protein, sandwiched by 800 bp of the core matrix attachment region from the human β globin 3′ DNase I hypersensitive site and SV40 polyadenylation sequences (5). To specifically track the tumor-associated endothelial cells and fibroblasts, neoplastic cells were injected into the mammary fat pads of transgenic mice that expressed green fluorescent protein driven by either the endothelial-specific Tie2 promoter (Tie2-green fluorescent protein/FVB; The Jackson Laboratory, Bar Harbor, ME; ref. 13) or the VEGF promoter (VEGF-green fluorescent protein/FVB; refs. 5 and 14). After 3 weeks of growth, small fragments (∼1 mm3) of tissue from the source tumors were extracted and transplanted into wild-type FVB mice. Fragments were transplanted s.c. (dorsal chamber model; ref. 15; n = 4), intracranially (cranial window model; ref. 16; n = 4), or in the mammary fat pad (for long-term orthotopic studies; ref. 17; n = 5).
In a separate experiment, small pieces of adipose (fat) tissue (2–3 mm3) from Tie2- or VEGF-green fluorescence protein mice were transplanted into dorsal skinfold chambers or cranial windows of wild-type FVB mice (n = 4 mice). During the course of the experiment, both donor and recipient mice received high-fat diets (which contained 65% fat) to promote adipogenesis (14).
The transplants were analyzed by in vivo multiphoton laser-scanning microscopy twice per week, as described previously (8). Fluorescence angiography was carried out using i.v. injections of rhodamine-conjugated dextran (Mr 2,000,000) as described previously (8). These techniques permitted concurrent molecular, anatomical, and functional imaging.
Tumor tissues were harvested at the end of the intravital observation period (4 weeks for dorsal skinfold chambers; 8 weeks for cranial windows) and at 7 weeks after transplantation into the mammary fat pad. Tumor tissues were analyzed histologically as described previously (18). Perfusion was detected using biotinylated lectin [0.1 mL of 1 mg/mL biotinylated Lycopersicon esculentum (tomato) lectin; Vector Labs, Burlingame, CA], and the functional vessels in 10-μm sections were stained using Texas Red-labeled streptavidin (Alexa-594; Molecular Probes, Eugene, OR). The nuclear dye 4′,6-diamidino-2-phenylindole (Molecular Probes) was used for tissue counterstaining, using protocols recommended by the manufacturer. This method allowed us to identify functional vascular structures in tumors while preserving the morphology of the tissue and the expression of fluorescent proteins.
Quantification of Green Fluorescent Protein Expression.
Eight-image stacks, each spanning a depth of 40 μm, were acquired using the multiphoton laser-scanning microscope with a 20× water immersion objective. Maximum-intensity projections were obtained and processed into binary images. The fractional area occupied by green fluorescent protein-positive cells was determined using the NIH Image software (14). At each time point, five regions of interest within the area of the implanted fragment were randomly chosen for each mouse and were analyzed to determine the fractional area occupied by surviving endothelial cells (in transplants from Tie2-green fluorescent protein mice) or activated fibroblasts (in transplants from VEGF-green fluorescent protein mice).
Statistical differences were calculated using Student’s t test. P < 0.05 was considered to be statistically significant.
Results and Discussion
Tumor-Associated Stromal Cells Survive after Transplantation.
We grew tumor tissue fragments containing stromal cells that expressed green fluorescent protein under the control of the ubiquitous EF1α promoter. This ensured that all stromal cells originating from the transplant would remain labeled regardless of their lineage. Implantation of neoplastic cells into EF1α-green fluorescent protein/FVB mice, followed by transplantation of the resulting tumors into wild-type FVB mice, resulted in the formation of a “capsule” of green fluorescent protein-positive cells, and functional tumor vessels with a green fluorescent protein-positive endothelial cell lining and an outer coat of green fluorescent protein-positive fibroblast-like cells (Fig. 1 B; Supplemental Movie 1). The expression of green fluorescent protein in the endothelium of functional tumor vessels was continuously detectable by intravital microscopy throughout the 4-week observation period. In contrast, the green fluorescent protein-positive stromal cells other than endothelial cells were no longer detectable in the same intravital images after 4 weeks (not shown). However, this could not be determined with utmost precision because it is difficult to distinguish green fluorescent protein-expressing endothelial cells from nonendothelial cells based on morphology alone.
Passenger Endothelial Cells in Tumor Transplants Survive and Multiply in the New Host.
To address the transplantability of stromal cells in a more specific and quantitative manner, we grew tumor tissue fragments containing endothelial cells that expressed green fluorescent protein under the control of the endothelial-specific Tie2 promoter. Immediately after transplantation into wild-type mice that did not harbor the gene for green fluorescent protein, these fragments exhibited green fluorescent emission from the passenger endothelial cells, consistent with a high level of Tie2 expression both in orthotopic and ectopic implantation sites (not shown). The expression of green fluorescent protein in the tumor endothelium remained continuously detectable by intravital microscopy throughout the observation period of 4 to 8 weeks (Fig. 2,A–C), despite the initial lack of blood flow for approximately 1 week after transplantation. The fractional area occupied by green fluorescent protein-positive endothelial cells originating from the source tumor fragment increased substantially, and these cells lined the vascular networks within the tumor during the first 4 weeks (Fig. 2; Supplemental Movie 2). Histological analysis of frozen tumor sections confirmed the green fluorescent protein expression in functional endothelium within the tumor mass, whereas green fluorescent protein was absent in the endothelial cells of the surrounding skin (Fig. 2,E and F). Fragments that were grown intracranially contained green fluorescent protein-positive vasculature that was stable for more than 2 months (Fig. 2 C). These findings were not dependent on the site of implantation and were confirmed using two other tumor cell lines (MCa-8 mammary carcinoma and LA-P0297 lung adenocarcinoma; data not shown), which were established from spontaneous tumors that arose in FVB mice in our colony.
To determine the contribution of host endothelial cells to the angiogenic vasculature within a tumor transplant, we conducted the converse experiment: Neoplastic cells were implanted into the mammary fat pads of wild-type FVB mice, and fragments of the resulting tumors were transplanted s.c. into Tie2-green fluorescent protein mice. These transplants did not exhibit detectable green fluorescent protein-positive tumor endothelium during the 4-week period of observation, despite the evident formation of many new functional vessels and the presence of green fluorescent protein-positive endothelium in the surrounding skin tissue (not shown). This indicates that, in these models, tumor angiogenesis was almost entirely dependent on the transplanted endothelial cells rather than on endothelial cells recruited from the local host tissue or from circulating endothelial precursors.
In an orthotopic model of mammary carcinoma, we transplanted mammary tumor tissue from Tie2-green fluorescent protein transgenic mice into the mammary fat pad of wild-type FVB mice. Tumors in the mammary fat pad are not accessible to intravital microscopy (IVM) without invasive surgery. However, after excision and staining, green fluorescent protein-positive endothelial cells from the transplanted tissue were identifiable by histology in the first weeks after transplantation (Fig. 2,G). In this model, some of the tumors grew relatively rapidly (reaching ∼1 cm diameter at 7 weeks), which may have prompted the recruitment of host-derived endothelial cells: In the converse experiment, in which tumor tissue from wild-type FVB mice was transplanted into Tie2-green fluorescent protein mice, functional host-derived green fluorescent protein-positive vasculature was visible in large tumors 7 weeks after transplantation (Fig. 2 H).
Stromal Cells in Tumor Transplants Survive Initially and Express Vascular Endothelial Growth Factor in the New Host.
We first implanted neoplastic cells in VEGF-green fluorescent protein/FVB mice. In the resulting tumors, nonendothelial stromal cells expressed green fluorescent protein when the VEGF promoter was activated (5). We found that VEGF-expressing tumor-activated fibroblasts survived transplantation, forming a capsule at the tumor margin (Supplemental Movie 3; refs. 5 and 8) and providing VEGF to adjacent endothelial cells in the tumor stroma (ref. 8; Fig. 3,A and B). Some of the VEGF-producing stromal cells migrated from the tumor into the adjacent tissue. Although these cells persisted for approximately 3 weeks in all implantation sites, their fractional area decreased substantially over time, both in the tumor mass and adjacent tissue (Fig. 3 C). We have previously shown that, within this time frame, host-resident fibroblasts invade growing tumor tissue (5).
Angiogenic Stroma from Nonmalignant Tissue Is Also Transplantable.
We transplanted adipose tissue (14)—a non-neoplastic tissue with high angiogenic activity—from Tie2-green fluorescent protein mice to wild-type mice. The adipose tissue vascular network survived transplantation and sustained new vessel development and remodeling (Fig. 4,A), just as the tumor vascular network did. After initial remodeling, the vasculature of the fat stabilized (Fig. 4,B) and persisted for more than 3 months (not shown). In fat tissue transplanted from VEGF-green fluorescent protein mice, VEGF expression was detected in both adipocytes and mesenchymal cells in the fat tissue (Fig. 4).
Tumor tissue exhibits uncontrolled growth, with rapid and continuous vascular remodeling. In contrast, the green fluorescent protein-positive vascular network in adipose tissue was more stable, and much of it remained intact but unperfused until day 12, when it was possible to observe anastomosis between the transplanted vasculature and the vasculature of the underlining tissue (Fig. 4 C and D). This suggests that the survival and growth of transplanted tissue may depend not only on the inclusion of stromal cells and interstitial matrix, but also on their organization.
It has been known for some time that tumor tissue grafts exhibit a highly variable growth rate after transplantation, even when they are derived from the same source tumor (19). One mechanism may be mutagenesis promoted by the transplant itself. Our results suggest another possible explanation for this heterogeneity at the cellular level: Tumor fragments contain their own endothelial cells and nonendothelial stromal cells, which survive in the new host and participate in angiogenesis. This survival occurs even in immunocompetent hosts and may depend on the presence of activated stromal cells that produce growth factors such as VEGF. The tumor environment produces profound changes in the phenotype of endothelial cells and other stromal cells, and tumor-activated stromal cells are known to promote tumor growth and progression. This was seen in studies in which tumor cells were co-implanted with transformed endothelial cells (20) or carcinoma-activated fibroblasts (6, 10). Thus, the angiogenic potential of a fragment is dependent on its particular composition: The number, phenotype, and organization of stromal cells is likely to vary from one portion of a tumor to another. Additional work is necessary to fully elucidate the molecular mechanisms by which stromal cells promote the formation and stabilization of vessels in transplanted tissue (21), as well as the extent of stromal cell migration to more distal sites.
Tumors are highly heterogeneous, and clinical treatments often fail because they affect some portions of a tumor but not others. This phenomenon is often attributed to the genetic instability and variability of neoplastic cells, and one of the attractions of antiangiogenic therapy is that its targets—endothelial cells and nonendothelial stromal cells—are genetically stable. However, a recent study of human head and neck carcinomas has found that tumor-associated stromal cells can exhibit multidrug resistance, precluding the reliable assessment of tumor chemosensitivity (22). This surprising finding indicates that stromal cells, as well as the mutations that accumulate in neoplastic cells, may contribute to drug resistance and that additional study of the heterogeneity of stromal cell phenotypes within a tumor (23) is warranted. The fact that small fragments of mature tumors can be transplanted while retaining much of their original cellular composition and organization means that large, heterogeneous tumors can be dissected into separate sections and studied piecemeal, that such tumors can be reduced in size to fit within transparent window models, and that stromal cells from one transgenic mouse can be studied in the physiological context of another mouse. Transplant models may provide a platform for the study of neoplastic-stromal cell interactions in vivo.
While our manuscript was in press we analyzed the transplantability of stromal cells in tumor fragments from VEGF-green fluorescent protein mice, which were allografted to severe combined immuno-deficient (SCID) mice. We found that transplanted stromal cells showed increased viability in SCID mice; a large number of green fluorescent protein-positive cells survived for as long as five weeks.
Grant support: National Cancer Institute NIH grants PO1-CA80124 and R24-CA85140. D. Duda was a recipient of a Cancer Research Institute fellowship.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article can be found at Cancer Research Online (http://cancerres.aacrjournals.org).
Requests for reprints: Rakesh K. Jain, E. L. Steele Laboratory, Department of Radiation Oncology, Massachusetts General Hospital, Harvard Medical School, 100 Blossom Street, Boston, MA 02114. E-mail: firstname.lastname@example.org
We thank J. Kahn, S. Roberge, and C. Smith for outstanding technical assistance and Dr. Isaiah J. Fidler for helpful suggestions.