The Influence of a Human Embryonic Stem Cell – Derived Microenvironment on Targeting of Human Solid Tumor Xenografts

The awareness of the important role that the surrounding tissue microenvironment and stromal response play in the process of tumorigenesis has grown as a result of in vivo models of tumor xenograft growth in immunocompromised mice. In the current study, we used human embryonic stem cells in order to study the interactions of tumor cells with the surrounding microenvironment of differentiated human cell tissues and structures. Several cancer cell types stably expressing an H2A-green fluorescence protein fusion protein, which allowed tracking of tumor cells, were injected into mature teratomas and developed into tumors. The salient findings were: (a) the observation of growth of tumor cells with high proliferative capacity within the differentiated microenvironment of the teratoma, (b) the identification of invasion by tumor cells into surrounding differentiated teratoma structures, and (c) the identification of blood vessels of human teratoma origin, growing adjacent to and within the cancer cell–derived tumor. Mouse embryonic stem cell–derived teratomas also supported cancer cell growth, but provided a less suitable model for human tumorigenesis studies. Anticancer immunotherapy treatment directed against A431 epidermoid carcinoma cell–related epitopes induced the complete regression of A431-derived tumor xenografts following direct i.m. injection in immunocompromised mice, as opposed to corresponding tumors growing within a human embryonic stem cell–derived microenvironment, wherein remnant foci of viable tumor cells were detected and resulted in tumor recurrence. We propose using this novel experimental model as a preclinical platform for investigating and manipulating the stromal response in tumor cell growth as an additional tool in cancer research. (Cancer Res 2006; 66(7): 3792-801)


Introduction
There has been a growing awareness in recent years that tumorigenic properties are markedly affected by the surrounding tissue microenvironment at both primary and metastatic sites (1).Thus, the role of the cellular microenvironment in tumorigenesis has become an intense area of research (2).It has been shown that tumor progression is associated with extensive remodeling of adjacent tissues to provide a supportive microenvironment for cancer cell proliferation, migration, invasion, and tumor neovascularization required for supporting cancer growth (1,(3)(4)(5)(6)(7).As an example, human prostate carcinoma-associated fibroblasts can promote tumorigenic transformation in initiated human prostate epithelial cells (8).Recent studies have shown that the connective tissue stroma surrounding neoplastic breast tissue was induced into a state that ''remembered'' and favored subsequent neoplastic growth in naı ¨ve breast epithelium (9,10).In addition, epigenetic changes in stromal cells of breast carcinomas seem to contribute to tumor progression (11).Appreciation of the importance of the stromal response has led to the development of novel anticancer therapeutic agents targeted to frustrate stromal response factors, which support progressive tumor growth.Targets for investigation have included proteases, heparanase, and other enzymes expressed by cancer cells or by adjacent stromal cells, which degrade extracellular matrix components and facilitate the release of cytokines and growth factors which stimulate angiogenesis, or support the growth and invasion of cancer cells (12)(13)(14).A great deal of attention has been directed at the development of antiangiogenic molecules in particular, some of which have reached clinical trials (15)(16)(17)(18)(19)(20)(21).
The investigation of in vivo models using tumor xenograft growth in immunocompromised mice has also provided evidence for the importance of the stromal response in tumorigenesis and the response to antitumor therapies (15)(16)(17)(18)(19)(20)(21).However, in such models, it has been the murine, rather than the human stromal response, which has been the target of investigation or experimental therapeutic intervention.Therefore, in a recently published study (22), we have used the potential of human embryonic stem cells (ESC) to generate in vivo teratomas comprised of a wide variety of differentiated tissues and structures (23,24).This human cellular microenvironment was used for studying the interactions of human tumorigenesis properties in a human cellular microenvironment.The salient findings were: (a) the observation of growth of tumor cells with high proliferative capacity within the mixed differentiated microenvironment of the teratoma, (b) the identification of invasion by tumor cells into surrounding differentiated teratoma structures, and (c) the identification of blood vessels of human teratoma origin, growing adjacent to and within the tumor.
In order to show the potential usefulness of this experimental model as a platform for testing future anticancer treatments, we both extended it to a variety of additional solid tumor-derived cell types and in addition sought to compare the effects of known anticancer treatments, using the novel experimental model with those obtained in direct tumor xenografts in mice without teratomas.For this purpose, we have used an immunotoxin therapy that has already been tested in clinical trials on the basis of studies which showed its efficacy in the shrinkage of A431-derived tumor xenografts in immunocompromised mice (25,26).We compared the effect of this therapy in a human ESC (hESC)-derived cellular microenvironment with that reported using the standard model of direct tumor cell injection into murine tissue.The results suggest a difference in tumor behavior and response to therapy in tumors developed within the hESC-derived microenvironment.Such differences might reflect differences in the tumor interactions with the surrounding stroma.
Reporter plasmid and stable transfection.A431 cells and all the cell lines described above were stably transfected with pEGFP-N1 expression vector (Clontech, Palo Alto, CA) which contains a cDNA coding region for the green fluorescence protein (GFP) fused downstream to the histone H2A (ref.30; kindly provided by M. Brandeis, Hebrew University, Jerusalem, Israel), as described previously (22).Transfection was carried out using FuGENE6 reagent (Roche, Indianapolis, IN) and 400 Ag/mL G418 (Life Technologies Grand Island, NY) for selection of stable clones (A431-GFP clones).In addition, A431 cells were stably transfected with the pDsRed2-Nuc expression vector (BD Biosciences, Palo Alto, CA) which contains three copies of the SV40 large T-antigen nuclear localization signal that directs the expression of the red fluorescence protein to the nucleus.
Animals.Severe combined immunodeficient (SCID)/beige mice were purchased from Harlan Laboratories, Ltd., Jerusalem, Israel.The mice were housed and maintained in specific pathogen-free conditions.The facilities and experimental protocols were approved by the Committee for Oversight of Animal Experimentation at the Technion-Israel Institute of Technology, Haifa, Israel.The mice were injected when they were 4 to 8 weeks of age in accordance with institutional guidelines.
Teratomas and tumor formation.For teratoma formation, undifferentiated human ES (hES) cells were harvested using 1 mg/mL collagenase type IV (Life Technologies) and injected into the hind limb of SCID/beige mice (f5 Â 10 6 cells per injection).Teratomas were palpable after 6 weeks.At 8 weeks following initial injection of H9.1 cells, 3 Â 10 6 A431-GFP cells were injected into the teratoma and were allowed to grow for an additional 10 days.Control noninjected teratomas, and A431-GFP-injected teratomas were harvested at the same time.Tumors derived from direct injection of 3 Â 10 6 A431-GFP cells into the hind limb musculature were harvested 10 days following injection.Formation of a mouse ESC-derived teratoma was accomplished by injection of 3 Â 10 6 undifferentiated mESC into the hind limb of SCID/beige mice.Ten days following the injection of mES cells, the teratoma size was already f20 mm.A431-GFP cells (2 Â 10 6 cells per injection) were injected into the teratoma at day 14.Harvesting of the teratomas bearing tumor was obligatory at day 20 because it reached 25% of the size of the mouse.
Histologic analysis.Teratomas were harvested, fixed for 48 hours in 10% neutral buffered formalin, transferred into 70% ethanol and processed using a routine wax-embedding procedure for histologic examination.Sixmicrometer paraffin sections were mounted on Super FrostPlus microscope slides (Menzel-Glaser, Germany) and stained with H&E.
Photomicrographs.Digital presentations were generated using an Olympus BX51 microscope equipped with Olympus DP70 camera.Pictures were processed using the analySIS 5 software (Soft Imaging Systems, Munster, Germany).
RNA extraction from teratomas and teratomas bearing tumors.Tissues were homogenized in TRI reagent (Molecular Research Center, Inc., Cincinnati, OH), 10 mL/g.The homogenate was incubated for 15 minutes with gentle agitation at room temperature.1-Bromo-3-chloropropane (Sigma, St. Louis, MO) was added at a ratio of 1 mL per 10 mL TRI reagent and the homogenate was incubated for an additional 15 minutes with agitation at room temperature.Phases were separated by centrifugation at 17,200 Â g for 15 minutes at 4jC.One volume of phenol was added to the aqueous phase and additional agitation for 15 minutes at room temperature was done.Following the addition of 2 mL chloroform isoamylalcohol, phases were separated again by centrifugation at 17,200 Â g for 15 minutes at 4jC.This procedure was repeated following the addition of 0.5 volumes of phenol and 0.5 volumes of chloroform isoamylalcohol to the aqueous phase and at the final step following the addition of one volume of chloroform isoamylalchol.Precipitation of the RNA from the aqueous phase was done by adding 0.1 volumes of 3 M sodium acetate and two volumes of ethanol, overnight incubation at À20jC, and centrifugation at 17,200 Â g for 15 minutes at 4jC.The RNA pellet was washed with icecold 70% ethanol and dissolved in diethyl pyrocarbonate (DEPC)-treated water.Genomic DNA was removed from the RNA samples using 50 units of RNase-free DNaseI at 37jC for 1 hour.RNA samples were diluted to 1 Ag/AL with DEPC-H 2 O.
Semiquantitative one-step RT-PCR.Forty nanograms of total RNA extracted from each tissue was subjected to semiquantitative one-step realtime reverse transcription-PCR using the QuantiTect SYBR Green RT-PCR kit (Qiagen, Inc., Santa Clarita, CA) and the Rotor-gene 2000 (Corbett Research, Sydney, Australia).Each reaction mix contained the following: 40 ng RNA, 100 ng of each primer, 12.5 AL 2Â PCR buffer, 0.5 AL RT mix, and DEPC-H 2 O to 25 AL.The following conditions were used: for reverse transcription, 30 minutes at 50jC; for the PCR initial activation step, 15 minutes at 95jC; for amplification, 15 seconds at 94jC, 30 seconds at 50jC to 60jC, and 30 seconds at 72jC for 35 to 45 cycles, depending on the abundance of the target gene.Melting curve analysis of the RT-PCR products was done in order to verify the specificity and identity of the PCR products.The specificity of the RT-PCR products was examined by agarose gel electrophoresis.For each sample, real-time reverse transcription-PCR analysis was done with a reference gene (b-actin gene) as an internal control.Primers used for gene amplification are listed in Supplement 2.
Expression, refolding, and purification of soluble B3Fv-PE38 fusion protein.B3(Fv)-PE38 recombinant immunotoxin was expressed using a T7 promoter-based expression system in Escherichia coli BL21 cells as previously described (25,26).The recombinant protein accumulated in intracellular inclusion bodies which were isolated, purified, reduced, and refolded in a redox-shuffling renaturation system as previously described (25,26).Protein was purified by ion exchange chromatography using Q-Sepharose and MonoQ columns.

Results
hESC-derived cellular microenvironment provides a supportive setting for human tumor growth.In a recently published study (22), we injected HEY-GFP ovarian cancer cells that stably express a histone H 2 A-GFP fusion protein, into hESC-derived teratomas and showed the growth and invasion of tumor cells within a human cellular microenvironment derived from in vivo differentiation of hESCs.
To examine whether this experimental model is also applicable to a wide variety of solid tumor-derived cell lines, we have modified several tumor cell lines so as to constitutively express GFP-H 2 A, in order to facilitate the tracking of tumor cell growth and invasion using immunohistochemistry staining.The cell lines used are the PC3 and LNCap (prostate cancer), DU-145 (prostatic adenocarcinoma), H226 (non-small cell lung carcinoma), MDA-MB-468 (breast adenocarcinoma), SW620 (colorectal adenocarcinoma), U-87 (glioblastoma), and several melanoma cell lines and the A431 epidermoid carcinoma (see representative examples in Fig. 1).Paraffin sections of harvested teratomas containing tumors stained with H&E revealed a homogeneous mass of tumor cells within the characteristic mixed differentiated structures of the teratomas.Within the mass of tumor cells, a high number of cells in mitosis and numerous small blood vessels were observed.A striking and consistent observation was that in all cases, tumor cell viability was greater in the setting of growth within the hESCderived microenvironment compared with corresponding tumors growing directly following i.m. injection in the mice without teratomas-indicating that the ESC-derived cellular microenvironment might provide a supportive setting for human tumor growth.
Figure 1A and B shows representative H&E stained paraffin sections of teratoma-bearing A431-derived tumors harvested 10 days following injection.This tumor is very aggressive; it shows a high degree of invasiveness, migration, and development of apparently detached new nodules within the teratomas.In addition, this tumor is highly angiogenic as shown by the density and complexity of blood vessels developed adjacent to and within the tumor (Fig. 1B).The high degree of invasiveness and migration of A431-derived tumors is shown by using H&E staining (Fig. 1C)

Cancer Research
Cancer Res 2006; 66: (7).April 1, 2006 and immunohistochemistry staining with anti-GFP antibodies (as the cancer cells are engineered to express GFP in the cell nucleus; Fig. 2A and B), and also with anti-EGFR antibodies because A431 cells express high levels of EGFR (ref.31; Fig. 2C and D).These staining modalities reveal the development of a major tumor within the teratomas, invasion of tumor cells and creation of new nodules within the teratoma tissues reminiscent of micrometastases.In another set of experiments, A431 cells stably expressing the red fluorescence protein in the nucleus were used, and a similar pattern of growth and invasion of tumor cells into the teratoma tissue were observed using a fluorescence microscope (data not shown).To this end, our results show that the cellular microenvironment derived from hESC differentiation within a mouse can serve as a supportive surrounding stromal tissue for tumor progression and metastasis.
Examination of the capability of a hESC-derived cellular microenvironment to support the growth of injected nontransformed cells, and the ability of mESC-derived teratomas to support growth of human tumors.The teratoma represents in vivo differentiation of pluripotent hESC.Hence, it is comprised of heterogeneous differentiated tissues of hESC origin that may provide the opportunity for the tumor cells to find an optimal interaction with relevant cells of the surrounding stroma.These interactions are expressed through enzymatic pathways, growth factors, and cytokines of stromal origin which have been identified as important for the growth and survival of tumor cells, and at a later stage support the processes of migration, invasion, angiogenesis, and metastasis.Given these findings-the question arose as to whether the supportive effect of the hESC-derived microenvironment was specific to transformed and tumorigenic cells or perhaps might support the growth of nontumorigenic cells of human origin as well.In order to examine the possibility that this cellular and extracellular matrix supportive composition could also potentially support the growth of normal human cells, we injected 2 Â 10 7 normal human fibroblasts BJ cells, both i.m. and intrateratoma.The BJ cells were immortalized through ectopic expression of hTERT (the catalytic subunit of telomerase) and also express ectopic GFP, which allows the detection and tracking of the cells (29).Mice were harvested at 70 days following injection of the BJ cells and although tumors were not palpable, the hind limb musculature and the teratoma tissues were subjected to histologic examination using H&E staining and immunohistochemistry with anti-GFP antibodies.None of the paraffin sections of the mouse muscle or the teratoma revealed any foci of GFP-positive cell growth, indicating that the heterogeneous composition of the teratoma cellular microenvironment does not support the growth of nontransformed, nontumorigenic human cells, even after they have been immortalized by ectopic telomerase expression.In addition, we

Cancer Research
Cancer Res 2006; 66: (7).April 1, 2006 sought to determine whether the growth of human tumor cell linederived tumors would be supported by mESC-derived teratomas.We found that A431 cells could establish tumor foci in mESCderived teratomas.However, the much more rapid growth rate of mESC-derived teratomas, compared with hESC-derived teratomas, precluded the time necessary for substantial growth of an A431 tumor within the teratoma (data not shown).The rapid growth of the mESC-derived teratoma mandated the sacrifice of the animal at 21 days because of the large size (estimated at 25% of the weight of the mouse), not providing sufficient time for testing of anticancer therapies or for characterization of the intrateratoma tumor.It should be noted that A431 cells are particularly fastgrowing, and therefore it did not seem feasible to test the growth of other human tumor cell lines, all of which grow more slowly than the A431 cells, and therefore would certainly not be amenable to study in a mESC-derived teratoma.
Induced angiogenesis and vasculogenesis of human and murine origin in a hESC-derived microenvironment.In order for tumor cells to gain access to the circulation and subsequently begin the process of invasion and metastasis at a distant site, de novo formation of blood vessels, a process known as vasculogenesis, is required (32).The establishment of a supportive tumor neovascular blood supply can derive from sprouting of capillaries from existing blood vessels, a process that is defined as angiogenesis.Alternatively, and in addition, de novo differentiation of hESC into circulating endothelial cell precursors occurs, and serves as a source for endothelial cells (33).Pluripotent hESC have the potential to differentiate into endothelial cells in vitro (34,35) and in vivo in teratomas (22,36,37).Upon binding of angiogenic factors to specific receptors, these endothelial cells proliferate, invade the basal lamina, migrate, differentiate, and develop new capillary tubes that connect to the vascular network.Hence, for experimental models in which the tumor develops directly within the mouse hind limb musculature in the absence of surrounding hESC-derived tissue, the developing neovascular blood system is expected to derive solely from murine origin.However, the development of a tumor within a hESC-derived microenvironment might be expected to elicit a neovascular network combining blood vessels of both murine and human origin (22).Therefore, we have used immunohistochemistry to examine the origin of blood vessels in the composite structures containing A431-derived solid tumors.The specificity of the antibodies used (mouse or human) enabled us to distinguish blood vessels of human or murine origin.Because A431 epidermoid carcinoma is highly aggressive and vasculogenic, and showed massive growth within 10 days in the teratomas, the vascular network observed within the tumor is comprised mostly of small and immature blood vessels.More established and mature blood vessels can be observed within the teratomas, and blood vessels of intermediate size can be observed in the region of the teratomas adjacent to the tumor (see Fig. 1).Therefore, in this specific analysis, several human embryonic vascular markers have been examined using immunohistochemistry with human-specific antibodies that show no cross-reactivity with their murine counterparts.These include endothelial cell-specific markers: CD34, CD31/PCAM-1, the endothelial specific tyrosine kinase receptor Tie-2 (see Fig. 3A, a-d) and the vWF (Fig. 3B, a), smooth muscle cell-specific marker: desmin and a-smooth muscle actin (a-SMA; Fig. 3B, b, c, and d), that were used to identify pericytes that are most commonly viewed as smooth muscle cells adjacent to the endothelial cells of the microvasculature and immature small blood vessels (38,39).Of note, because the teratoma vascular network is comprised of blood vessels of murine and of human origin, it is possible to observe fusion of human and murine endothelial cells in the same blood vessel as shown in Fig. 3A (b).In this case, staining with antihuman-specific CD34 shows discontinuous staining in one blood vessel (small arrow) as opposed to negative staining in another blood vessel (big arrow) that is entirely of mouse origin.As controls, (a) paraffin sections of A431-derived tumors growing directly in the mouse hind limb and (b) secondary antibody alone were used (data not shown).Blood vessels of exclusively murine origin were identified using immunohistochemistry with rabbit polyclonal anti-mouse CD31 (see Supplement 1).Of note, blood vessels within the teratoma tissue itself, either in a teratoma alone or in a teratoma-bearing tumor, were positively stained with either rabbit polyclonal anti-mouse CD31 or with mouse monoclonal anti-human CD31 (data not shown), reflecting neoangiogenesis of murine origin that supports early stages of teratoma development, followed by neoangiogenesis of human origin presumably as a consequence of hES differentiation into endothelial cell precursors (38,39).Thus, these immunohistostaining results show that the A431-derived tumors elicit a neoangiogenesis response of mixed human and murine origin when injected intrateratoma, and growing within a hESC-derived teratoma, and an exclusively murine response when growing i.m.
Semiquantitative analysis for measuring the expression level of human-specific vasculogenesis-related genes.The foregoing results prompted us to examine the expression levels of human-specific vasculogenesis-related genes at the molecular level, using semiquantitative RT-PCR as a supplement to the immunohistostaining approach.For this purpose, total RNA was extracted from teratomas and teratomas bearing A431-derived tumors.For controls, RNA was extracted from A431-derived tumors growing directly in the mouse hind limb, A431 cells growing in a culture dish and from the hind limb musculature of a mouse.The difficulty in predicting the exact location of the tumor within the teratoma tissues and the fact that the teratoma tissue consists of a variety of differentiated cells and tissues derived from in vivo differentiation of hESC, compelled us to extract the RNA from the entire teratoma, teratoma-bearing tumor or tumor tissues.As shown in Fig. 4, teratoma and teratoma-bearing tumors (samples 1 and 5) express human endothelial cell-specific genes CD34, CD31, vWF, Tie-2, angiopoietin 1 (Ang-1), and low levels of angiopoietin 2 (Ang-2).Ang-1 and Ang-2 are, respectively, the Tie-2 receptor stimulating agonist and the Tie-2 receptor antagonist.In parallel, these tissue samples also express human-specific a-SMA, a specific marker for vascular smooth muscle cells, and the sprouting and remodeling gene, vascular endothelial cadherin.A431 cells growing in vitro and in vivo (samples 4 and 2, respectively) also express CD31, vWF, a-SMA, and vascular endothelial cadherin, expression profiles that have previously been reported in highly aggressive and invasive cancer cells (40,41).None of these human-specific genes were expressed in mouse muscle tissue (sample 3).The expression of mouse-specific blood vessel-related genes CD34, CD31, and a-SMA was clearly evident in teratomas, teratoma-bearing tumors, tumors growing i.m., and in mouse muscle tissues as expected.Taken together, these molecular analyses of results support our previous observation that in the process of tumorigenesis, tumor cells growing within a hESC-derived microenvironment elicit the induction of neoangiogenesis of human origin.
The effect of anticancer immunotherapy on tumorigenesis using the experimental model.In order to show the potential usefulness of the experimental model as a platform for testing future anticancer treatments, it was necessary to first examine the effect of established anticancer treatments, using the novel experimental model and compare the findings with those obtained in direct tumor xenografts in mice without teratomas.For this purpose, we have tested a well studied anticancer immunotherapy regimen using a recombinant immunotoxin antibody that reacts with the Lewis Y epitope, a well characterized antigen that is highly expressed in many carcinomas (25,26).When this anticancer immunotherapy was given to mice bearing A431 tumor xenografts following direct i.m. injection, complete eradication of A431 tumors and full recovery of the mice was observed (42).To examine the effectiveness of this immunotherapy in mice bearing A431 tumors within hES-derived teratomas, we carried out a series of experiments which included six different groups of mice (n = 5 per group).hES cells (5 Â 10 6 ) were injected into the hind limb of SCID/beige mice and teratomas were allowed to develop for 7 weeks.A431 cancer cells (2 Â 10 6 ) were injected into the teratomas and tumors were allowed to develop for 5 days within the teratomas.Groups were treated as follows: (a) mice bearing teratomas only, (b) mice bearing tumor-containing teratomas, (c) mice bearing tumor-containing teratomas treated with immunotoxin-2 Ag/100 AL was injected into the tail vein thrice, every other day (the recombinant immunotoxin synthesis procedure is described in Materials and Methods), (d) mice bearing teratomas treated as above, (e) mice bearing teratomas treated with PBS, ( f ) mice bearing tumor-containing teratomas treated with PBS.The schematic presentation of the immunotherapy and complete protocol are shown in Fig. 5.For comparison, we have also injected A431 cells directly into the hind limb of SCID/beige mice, and treated them in the same manner.
Two days following the final treatment, all teratomas and tumors were harvested and subjected to routine paraffin embedding procedures for histologic examination.Six-micrometer paraffin sections were stained with H&E or subjected to immunohistochemistry using anti-GFP and anti-EGFR to detect tumor cells within the teratomas.The following results were obtained: (a) direct injection of A431 cells into the mouse hind limb resulted in the development of highly angiogenic tumors.Tumor cells were GFP-and EGFR-positive using immunohistochemistry. Blood vessels within the tumor stained positively for mouse-specific CD31 antibody, and negatively for human-specific anti-vWF, a-SMA, and CD34 antibodies.Injection of immunotoxin into the tail vein resulted in the complete regression and full recovery of the mouse.Histologic examination revealed a small remnant of nonviable tumor cells with small and condensed nuclei that stained negatively for EGFR (Fig. 6C and D).Tumor recurrence was not evident following 10 days of recovery.(b) The histologic appearance of teratomas bearing tumors without treatment and teratomas bearing tumors treated with PBS, showed a homogeneous mass of cells with the characteristic morphology of abnormal carcinoma cells with high proliferative capacity as observed by positive staining for proliferating cell nuclear antigen and Ki67 within the teratoma tissues (data not shown) and invading the surrounded differentiated teratoma structures as indicated by GFP-and EGFR-positive cells interspersed among teratoma-derived connective tissue, adipocytes, connective tissue, and neurovascular bundles.New foci of viable tumor cells were also observed as previously shown (Fig. 2).(c) Immunotherapy treatment resulted in the regression of tumor growth within the teratomas, however, remnants of small foci of viable tumor cells varying in size from 100 to 600 Am were observed, suggesting that this immunotherapy treatment, although highly effective on direct A431 tumor xenografts, was less effective in treating the same tumors growing within a hESC-derived cellular microenvironment (Fig. 6A and B).These small foci of viable tumor cells were positively stained using immunohistochemistry with anti-EGFR, anti-GFP and with the recombinant immunotoxin antibody (Fig. 6A and B).The appearance of small tumor foci was evident even following treatment with progressively escalating doses of the immunotoxin (4 Ag/100 AL, injected five times, every other day).These tumor remnants subsequently succeeded in reestablishing large new tumor foci following cessation of this anticancer therapy.(d) No effect on differentiated teratoma structures was observed following treatment with the immunotoxin.(e) Immunohistochemistry of sections derived from tumor-bearing teratomas with and without immunotoxin treatment using human-specific vWF, a-SMA, and CD34 antibodies revealed the existence of established and immature small blood vessels of human origin in the teratomas and in the tumors.Taken together, these results suggest that tumors growing in the teratoma environment, resisted full regression, and even treatment with escalating doses of immunotoxin failed to completely eradicate remnants of tumor.

Discussion
The appreciation of the pivotal role that tumor-host cross-talk plays in cancer cell growth, invasion, and metastatic processes has motivated the establishment of mouse models of cancer progression (43).This valuable experimental system allowed the examination of several mechanisms involved in tumorigenesis such as growth control mechanisms and recruitment of blood vessels which modulate carcinogenesis via cellular interactions deriving from the relationship between the extracellular matrix and the surrounding stromal cells (44).In addition, the tumor xenograft mouse model has provided a powerful tool for testing anticancer reagents and in particular antiangiogenic molecules targeted to frustrate the stromal response (16,(18)(19)(20)(21).However, this model does not offer several advantages that could be obtained in a model in which the tumor cells have the opportunity to interact with a surrounding microenvironment of heterogeneous human cell types, conducive to the cross-talk between human tumor cells and their cellular microenvironment (10).
We have previously reported an experimental approach using human stem cell technology to generate a novel experimental platform in which human tumor cells can be grown in a preclinical setting in a tissue microenvironment consisting of differentiated cells derived from hESC for investigating tumorigenesis properties (22).Accordingly, in the present study, we used this model to examine multiple different human tumor cell types, and to determine if there is a difference in the response to anticancer therapy when tumor cells are grown in this novel experimental model, compared with the conventional murine xenograft model.We have also extrapolated this approach to freshly harvested human tumor cells and shown that the hESC-derived teratoma provided an extremely supportive environment (data not shown).A striking and consistent observation was that in all cases, tumor cell viability was greater in the setting of growth within the hESCderived microenvironment, compared with corresponding tumors growing directly following i.m. injection in the mice without teratomas-indicating that the hESC-derived cellular microenvironment provided a supportive setting for human tumor growth.This likely emanates from mutual interactions of the tumor cells with one or more human cell or tissue types in the teratoma.Importantly, the composition of heterogeneous human cellular tissues that might establish an optimal niche for tumor development, was not able to support the growth of nontransformed but immortalized human fibroblasts, indicating that the microenvironment is supportive to tumor cells, and that our model should facilitate the possible unraveling of novel interactions specific to tumor cells with their surrounding microenvi-ronment.On the other hand, human cancer cells injected into a mESC-derived teratoma, could also develop into a tumor and exhibited similar tumor growth within the mESC-derived teratoma.However, this mESC-derived experimental platform was not conducive to further investigation or testing.First, mESC-derived teratomas developed according to a time frame appropriate for mouse cell growth and differentiation i.e., teratomas appeared at 5 days following injection and reached a 2cm diameter at day 10.A431 cells were injected at day 14 following injection and harvesting was obligatory by day 21, because the teratoma-bearing tumors already reached 25% of the mouse size, necessitating sacrifice of the mouse.This tight time-frame for mESC might be even more problematic when longer time-frames are needed to develop a murine ESC-derived intrateratoma tumor as in the case of HEY ovarian cancer cells, which are much more slow-growing (22); Furthermore, the vasculogenesis in mESC-derived teratomabearing tumors is entirely of murine origin, and therefore not conducive to studies of tumor induced human cell derived vasculogenesis pathways or their interruption with possible human-specific anticancer drugs.
Clusters of small blood vessels of hESC origin which are characteristic of tumor-induced neovascularization were also prominently observed in tumor-embedded human teratomas and within the tumor nodules themselves.Sprouting of small blood vessels from existing blood vessels is very often observed as a node at the border between the tumor and the teratoma tissue indicating that during tumorigenesis, tumor cells within the teratomas induce the process of spontaneous in vivo vasculogenesis.It should also be noted that the mixed or heterogeneous microenvironment of the teratoma is of particular advantage in these studies because we have no prior knowledge of the preferred tissue type from which the neoangiogenesis process will develop.
We also went on to compare the response of A431-derived solid tumor to an anticancer therapy in a standard tumor xenograft model with the corresponding responses using the intrateratoma model.In this study, we tested a well-established experimental anticancer regimen using a recombinant immunotoxin antibody that reacts with the Lewis Y epitope.When this anticancer immunotherapy was given to mice bearing A431 tumor xenografts following direct instrumuscular injection, complete regression of A431 tumors and full recovery of the mice was observed, consistent with earlier reports (25,26).In striking contrast, tumors growing in the hESC-derived intrateratoma model resisted full regression, and even treatment with escalating doses of immunotoxin failed to completely eradicate remnants of the tumor.These tumor remnants subsequently succeeded in reestablishing large new tumor foci following cessation of the anticancer therapy.Accordingly, we suggest that the growth of human tumor cells is promoted within the heterogeneous cellular microenvironment of the hES-derived teratoma, due to a favorable set of interactions among the variety of cell types, with the establishment of a more conducive niche for relevant subsets of tumor cells to interact with corresponding subsets of cells within the teratoma.These interactions may support the growth of the tumor cells and contribute to their resistance to anticancer therapy.We postulate that these interactions will differ from tumor

Cancer Research
Cancer Res 2006; 66: (7).April 1, 2006 cell type to tumor cell type, and may more closely mimic the behavior of tumors in the clinical setting.Because such a heterogeneous cellular microenvironment is not available to human tumors growing directly as xenografts in the mouse tissue-the model we have developed should facilitate the unraveling of novel interactions between tumor and nontumor human cells that are of relevance to tumorigenesis and to anticancer therapy.

Figure 2 .
Figure 2. Epidermoid carcinoma tumor cells within the teratoma: immunostaining with anti-GFP and anti-EGFR antibodies.The high degree of invasiveness and migration of A431-derived tumors is shown using immunohistochemistry with anti-GFP antibodies (A and B ) and anti-EGFR antibodies (C and D ).Staining with anti-GFP appears in the cell nucleus as A431 cells stably express the GFP protein in the cell nucleus.Staining with anti-EGFR reveals staining in the cytoplasm and in the cell membrane.Invasion of tumor cells into the teratomas (D ).Bars, 200 Am (A), 100 Am (B), 500 Am (C ), and 50 Am (D ).

Figure 3 .
Figure 3.A and B, detection of neoangiogenesis and vasculogenesis of human origin within teratoma-bearing tumor.Human-specific vascular markers were detected using immunohistochemistry with human-specific antibodies in paraffin sections of teratomas bearing A431-derived tumors.Staining of endothelial-specific markers; CD34 (A, a and b), CD31 (c ), Tie-2 (d), and vWF (B, a).Smooth muscle cell-specific markers; desmin and aSMA in (B, b ) and (C ), respectively.Anti-aSMA staining of small blood vessels adjacent to and within the tumor reveals induction of neoangiogenesis of human origin within the teratoma.B, (d) higher-power magnification of the inset from (B, c ), shows specific staining of smooth muscle cells adjacent to endothelial cells of immature small blood vessels.A, bar, 50 Am (a, b , and c ); 100 Am (d); B, bar, 100 Am (a, b, and c ); 20 Am (d).

Figure 4 .
Figure 4. Expression of human-specific vasculogenesis-related genes.Total RNA was extracted from teratoma, A431-derived tumor, mouse muscle, A431 cells cultivated in vitro, and teratoma-bearing A431-derived tumor.Aliquots of RNA were subjected to one-step RT-PCR analysis for detection of human-and mouse-specific vasculogenesisrelated genes.b-Actin served as an internal control.NTC, no template control.

Figure 5 .
Figure 5. Immunotherapy of A431-derived tumor within a teratoma using recombinant immunotoxin.See Materials and Methods for details of experimental protocol.

Figure 6 .
Figure 6.Remnant of A431 tumor tissue within a teratoma following immunotherapy: Immunostaining with anti-EGFR antibodies.Immunotherapy treatment of A431-derived tumors growing within a hESC-derived microenvironment resulted in incomplete tumor regression.Small foci of viable tumor cells varying in size from 100 to 600 Am are evident in (A) and (B ) at higher-power magnification.The same immunotherapy treatment of A431 tumors developed directly in the hind limb musculature of a mouse resulted in complete regression and full recovery of the mouse.Small remnants of nonviable cells are evident in (C ) and (D ).Bar, 500 Am (A), 100 Am (B and D ), and 200 Am (C ).