Skip Navigation


Carcinogenesis Advance Access originally published online on October 19, 2006
Carcinogenesis 2007 28(3):595-610; doi:10.1093/carcin/bgl188
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
28/3/595    most recent
bgl188v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Willhauck, M. J.
Right arrow Articles by Stark, H.-J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Willhauck, M. J.
Right arrow Articles by Stark, H.-J.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2006. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

Reversion of tumor phenotype in surface transplants of skin SCC cells by scaffold-induced stroma modulation

Michael J. Willhauck1, Nicolae Mirancea1,3, Silvia Vosseler2, Alessandra Pavesio4, Petra Boukamp3, Margareta M. Mueller2, Norbert E. Fusenig1 and Hans-Jürgen Stark1,3,*

1 Division of Carcinogenesis and Differentiation, German Cancer Research Center (DKFZ) Im Neuenheimer Feld 280, D-69120 Heidelberg, Germany
2 Group of Tumor Microenvironment, German Cancer Research Center (DKFZ) Im Neuenheimer Feld 280, D-69120 Heidelberg, Germany
3 Division of Genetics of Skin Carcinogenesis, German Cancer Research Center (DKFZ) Im Neuenheimer Feld 280, D-69120 Heidelberg, Germany
4 Fidia Advanced Biopolymers (FAB), Abano Terme Italy

*To whom correspondence should be addressed at: Division of Genetics of Skin Carcinogenesis (A110), German Cancer Research Center (DKFZ), Im Neuenheimer Feld 280, D-69120 Heidelberg, Germany. Tel: +49 6221 42 3446; Fax: +49 6221 42 3457; Email: hj.stark{at}dkfz.de


   Abstract
Top
 Abstract
Introduction
 Materials and methods
 Results
 Discussion
 Supplementary material
 References
 
Interactions between cancer cells and the tissue microenvironment play an essential role in controlling tumor development and progression. Here, we report that stromal modulation induced by a biodegradable meshwork (Hyalograft 3D) inhibited tumor vascularization and invasion of the locally invasive low-grade malignant human HaCaT-ras II-4 keratinocytes in a surface xenotransplantation assay. The scaffold caused formation of an active granulation tissue that shifted to a fibrotic-type connective tissue with accumulation of myofibroblasts and collagen bundles. Most importantly, in transplants with scaffolds, the epithelial-stromal border was normalized developing an ultrastructurally complete basement membrane (BM) including hemidesmosomes. The observed reversion of the tumor phenotype was not due to decreased tumor cell proliferation but correlated with (i) normalization of epidermal differentiation, (ii) condensation of extracellular matrix (ECM) and (iii) reduction of peritumoral protease activity Furthermore, inhibited invasion was paralleled by eliminated tumor vascularization. This was substantiated by a diminished endothelial VEGF-receptor (VEGFR) expression and, in turn, by a concomitant increase in the ECM components thrombospondin-1 (TSP-1) and endostatin, known to impair angiogenesis. Even in transplants of the metastatic high-grade malignant HaCaT-ras A-5RT3 keratinocytes the anti-invasive effect of the scaffold-modulated stroma prevailed. Tumor vascularization and invasion was reduced and the epithelial tissue partially normalized including formation of stretches of BM. This clearly demonstrates that the scaffold-modulated connective tissue not only blocks tumor invasion but reverts the tumor phenotype. These novel findings underline the controlling function of tumor stroma and open new strategies of cancer therapy by targeting tumor stroma elements.

Abbreviations: {alpha}-SMA, alpha smooth muscle actin; BM, basement membrane; col XVIII, collagen type XVIII; ECM, extracellular matrix; HA, hyaluronic acid; Hyalograft 3D, a fleece-like non-woven fibrous material (~0.4 mm thick) consisting of HA; MMP, matrix metalloproteinase; SCC, squamous cell carcinomas; VEGFR2, vascular epithelial growth factor receptor 2


   Introduction
Top
 Abstract
 Introduction
Materials and methods
 Results
 Discussion
 Supplementary material
 References
 
Malignant tumor cells activate the host stromal microenvironment through production and secretion of stimulatory growth factors and cytokines that induce a desmoplastic reaction in the stroma representing a critical step for carcinoma growth and progression (1–3). This locally activated host microenvironment, the tumor stroma, in turn modifies the proliferative and invasive behavior of the tumor cells (3,4). Although the importance of the stromal microenvironment for tumor development and control is increasingly appreciated, the molecular mechanisms underlying the interactions between stromal and tumor compartments are still mostly obscure (5).

The activated tumor–stroma consisting primarily of fibroblasts, endothelial cells and inflammatory cells, as well as, different extracellular matrix (ECM) components, is very similar to a generic wound repair response exhibiting elevated production of ECM components, growth factors and matrix remodeling enzymes to create a growth-promoting microenvironment (3). Stromal activation appears to be initiated during early stages of carcinoma development and evolve with cancer progression (6), where it enhances tumor progression by stimulating angiogenesis and by promoting cancer cell proliferation, survival and invasion (7–9). In contrast to the transient activation of the stroma during tissue regeneration in wound healing, malignant cells perpetually stimulate host stromal and vascular cells and thus create a continuous permissive field for malignant cell invasion (10). This cancer invasion is a hallmark of malignant tumors and can be viewed as a derangement in the proper sorting of cell populations, causing a violation of normal tissue boundaries (1).

In normal tissue the regular architecture is maintained by basement membrane (BM) delineation of tissue boundaries and cell–cell communication. This ensures tissue integrity and suppresses inappropriate intermixing of cells from different tissue types. Malignant cells, however, acquire increasing resistance to the regulatory signals, which control appropriate sorting of parenchymal tissue cells during morphogenesis and wound healing (11).

In experimental hetero-transplant systems in nude mice it has been demonstrated that paracrine regulatory mechanisms of the organ-microenvironment play a decisive role for tumor take, invasion and metastatic rate of many human tumors (4).

  1. Rapidly proliferating organs specifically enhance growth of local and metastatic tumors suggesting paracrine growth regulation of tumor cells (12).
  2. Moreover, preinduction of granulation tissue at the tumor cell transplantation side, drastically enhanced the invasion of carcinomas (13).
  3. Finally, there is increasing evidence that fibroblasts isolated from different tissues can have stimulatory influence on tumor cells concerning infiltrative growth and protease production (14–16). Thus, activated stroma as found in wound granulation tissue seems to exert stimulating effects on tumor cell growth and facilitates their invasion (17).
On the other hand, the crosstalk between mesenchyme and epithelium is also essential in controlling normal tissue development and differentiation and is crucial for normal tissue regeneration (18,19). As a consequence, normalization of the altered epithelial stromal interactions in tumors by blocking the stromal activation should lead to reversion of the tumor phenotype. Indeed, there have been several reports on reversion of the tumor phenotype by a normal tissue context in different experimental systems (11,20,21). In vitro studies have shown that normal dermal fibroblasts suppress the growth of neighboring transformed keratinocytes and fibroblasts via cell–cell contact and paracrine interactions (22–24). Such tumor–stroma interrelationship has been well documented in a surface transplantation model of tumor invasion where early tumor–stroma interactions can be studied in detail (10,25,26). In the study presented here we use this model to explore the scaffold-induced stromal alterations and demonstrate that modulation of the stromal context by a biodegradable scaffold to a fibrotic-type connective tissue blocks or considerably delays tumor cell invasion depending on tumor stage and reverts the malignant epithelial tumor phenotype to a premalignant tissue.


   Materials and methods
Top
 Abstract
 Introduction
 Materials and methods
Results
 Discussion
 Supplementary material
 References
 
Cells and culture conditions
The low-grade malignant cell line HaCaT II-4 is derived from the immortalized human keratinocyte cell line HaCaT (27) after transfection with the c-Ha-ras oncogene (27). Upon subcutaneous (s.c.) injections and in surface transplants II-4 cells grow to well keratinized squamous cell carcinomas (SCCs). The high-grade malignant and metastatic clone A-5RT3 was established by recultivation of heterotransplants of benign-tumorigenic HaCaT-ras A5 cells in nude mice as described previously (28). All cells were grown in enriched minimal essential medium (4x MEM) supplemented with 5% FCS and 200 µg/ml geneticin as described previously (25).

Scaffold material
Hyalograft 3D, a fleece-like non-woven fibrous material (~0.4 mm thick) consisting of hyaluronic acid (HA) esterified to 95% with benzyl alcohol was provided by Fidia Advanced Biopolymers (Abano Terme, Italy). This material, formerly named Hyaff 11, has been reviewed thoroughly by Campoccia (29).

Surface transplantation assay
Cells were transplanted onto the dorsal muscle fascia of 7–9-week-old nude mice (Swiss/c nu/nu back crosses) as monolayer cultures growing on collagen type 1 gels using a silicone chamber device as described previously (30,31). For analyzing the influence of scaffold-induced granulation tissue on tumor growth and invasion, Hyalograft 3D materials (2 x 2 cm2 dimension and 0.4 mm thickness) were placed onto the back muscle fascia immediately before the cell-carrying collagen gels were inserted on top. After 1, 2, 4 and 6 weeks mice were sacrificed (n = 3 per timepoint) and transplants were dissected en bloc, embedded in Tissue-Tek (Miles Laboratories, Elkhart, IN), and frozen in liquid nitrogen vapor for preparation of cryostat sections. For histopathology transplants were fixed in 3.7% formaldehyde, dehydrated and embedded into paraffin. Cryostat as well as paraffin sections were evaluated after standard hematoxylin and eosin (H&E) staining. For labeling of proliferating cells, mice were injected into tail veins with 100 µl of 5-bromodeoxyuridine (BrdU) and 5-bromodeoxycytidine (65 mM each) in 0.9% NaCl 1.5 h before being sacrificed (30). Transplantation experiments were repeated three times.

Antibodies
Rat monoclonal antibody (mAb) against mouse CD31 and mouse mAB against ß4-integrin subunit was obtained from BD PharMingen (Heidelberg, Germany), guinea pig polyclonal anti-serum against cytokeratins (pan), mouse mAb against Keratin 1/10 and mouse mAb against Keratin 8 from Progen (Heidelberg, Germany), rat mAb against {alpha}6 integrin subunit from Chemicon (Temecula, CA, USA), sheep polyclonal antibody against BrdU from NatuTec (Frankfurt am Main, Germany), rabbit polyclonal antibody against tenascin-C from Telios Pharmaceuticals (San Diego, USA), biotinylated mouse mAb against a-smooth muscle actin from Progen (Heidelberg, Germany), rabbit polyclonal antibody against mouse collagen type IV from Novotec (Lyon, France), and rabbit polyclonal antibody against nidogen from Merck Bioscience GmbH (Schwalbach, Germany). Goat antisera against mouse vascular epithelial growth factor receptor-2 (VEGFR2) and mouse MMP-2 were purchased from R&D Systems (Wiesbaden-Nordenstadt, Germany), a rat mAb against thrombspondin-1 from Immunotech (Marseille, France) and a rabbit antiserum against MMP-1 from Sigma-Aldrich (Taufkirchen, Germany). The following reagents were kindly donated by colleagues: sheep antiserum against MMP-9 (G. Murphy, Cambridge, UK), goat antiserum against human uPA (M. Kramer, Institute for Immunology, University of Heidelberg, Germany), rabbit anisera against Loricrin (D. Hohl, CHUV, Lausanne, Switzerland), and Laminin-5 (P. Marinkovich, Shriners Hospital, Portland, OR, USA). Secondary antibodies were obtained from Dianova (Hamburg, Germany) and Hoechst 33258 bisbenzimide for nuclear staining from Sigma-Aldrich (Taufkirchen, Germany).

Indirect immunofluorescence assay
For immunofluorescence staining, frozen sections were fixed for 5 min in 80% methanol at 4°C and 2 min in acetone at –20°C, and rehydrated in phosphate-buffered saline (PBS). For BrdU localization in DNA, sections were additionally denatured in 2 M HCl for 10 min at room temperature and washed (3 x 10 min). Primary antibodies were incubated in 12% BSA/PBS at RT for 2 h or 4°C over night. After washing in PBS (3 x 10 min) sections were incubated with appropriate secondary antibodies along with 5 µg/ml Hoechst bisbenzimide for staining of cell nuclei. Prior to embedding in Permafluor (Immunotech, Marseille, France) sections were washed again (3 x 10 min) in PBS. Stained sections were examined at an Olympus AX-70 microscope equipped with epifluorescence optics and recorded with a CCD-camera (F-View 12, Soft Imaging Systems (SIS), Münster, Germany) applying AnalySIS Pro 6.0-software. Immunostainings that had to be compared quantitatively were captured at identical illumination conditions, with identical exposure time and system settings for digital image processing.

cDNAs
For in situ hybridization a 1.8 kb fragment of human MMP-1 cDNA (32) (a kind gift of Professor Peter Angel, DKFZ, Heidelberg, Germany) and a 2048 bp fragment of human MMP-2 cDNA corresponding to bases 433–2481 (a gift of Professor Karl Tryggvason, Oulu, Finland) were used.

In situ hybridization
In situ hybridization was performed as described previously(26) In brief, DIG-labeled RNA probes for human matrix metalloproteinase (MMP)-1 and -2 were prepared using T7-, SP6- or T3- RNA-polymerase (for antisense and sense, respectively) according to the manufacturer's instructions (Roche, Mannheim, Germany). Cryostat sections were fixed in 4% paraformaldehyde, pre-treated, hybridized and washed at high stringency as described elsewhere. DIG was labeled by anti-DIG-AP (Roche, Mannheim, Germany) and alkaline phosphatase reaction was detected by NBT/BCIP (Gibco-Life Technologies/Invitrogen, Eggenstein-Leopoldshafen, Germany). After DIG in situ hybridization of MMP-1 or MMP-2, counterstaining was performed by indirect immunofluorescence with antisera against pankeratin and collagen type IV. Sections were photographed in the fluorescence mode for the counterstaining and the bright field mode for the hybridization signals to which the false color yellow was assigned by means of AnalySIS Pro 6.0-Imaging Software.

In situ zymography
Gelatinolytic activity was demonstrated in unfixed cryostat sections using as a substrate DQ gelatin (EnzChek; Molecular Probes, Leiden, The Netherlands) according to Ref. 33. Cryostat sections (6 µm) were incubated with 40 µg/ml DQ gelatin for 30 min at room temperature. After washing (3 x 10 min) sections were stained for MMP-9 and Hoechst using immunofluorescence techniques as described above.

Transmission electron microscopy
Fresh samples of transplants were prefixed in ice-cold 4% glutaraldehyde in 0.05 M sodium cacodylate buffer, pH 7.3, for 3 h and post-fixed in 2% osmium tetroxide in 0.1 M sodium cacodylate buffer for 2.5 h at 4°C. Tissue blocs were then washed with distilled H2O, stained en bloc with 0.5% aqueous uranyl acetate overnight at 4°C and again washed with distilled H2O. Following dehydration through two graded series of ethanol and infiltration with propylene oxide specimens were embedded in Epon 812-equivalent (glycidether 100, Serva, Heidelberg, Germany) and finally polymerized at 60°C for 48 h. Semi-thin sections of 1 µm were stained with 0.1% toluidine blue for light microscopy. Ultrathin sections (50–90 nm) were cut at a Reichert-Jung ultramicrotome, counterstained with uranyl acetate and subsequently lead citrate and examined with a Zeiss EM10B electron microscope.

Morphometric analysis
Quantification of BrdU incorporation and vessel density was performed morphometrically using analySIS software (Soft Imaging Systems, Münster, Germany). In brief, cellular proliferation was visualized by BrdU incorporation, evaluated in 2–4 vision fields (2.38 mm2) per animal (2–3 animals per timepoint) and brought into relation with the total number of cell nuclei in the vital tumor parts. As tissue architecture is well discernible by the pattern of DNA stained nuclei, BrdU-labeled cells clearly could be attributed to the tumor and the stromal compartment. Mean vessel density was determined using whole mount pictures (3.44 mm2) from CD31 immunostained transplants (4–6 animals per timepoint from two independent experiments) and evaluating the ratio of CD31 stained areas, 300 µm below and 300 µm within the tumor, in percent. Results are presented as means ± standard deviation, statistical significance was calculated using students t-test (Microsoft Excel 2002) (P < 0.01).


   Results
Top
 Abstract
 Introduction
 Materials and methods
 Results
Discussion
 Supplementary material
 References
 
Rapid induction of granulation tissue and angiogenesis by Hyalograft 3D-scaffolds
Transplantation of the biodegradable non-woven scaffold Hyalograft 3D onto the back muscle fascia of nude mice induced a highly activated granulation tissue within 2 weeks (Figure 1). The dry scaffold, when attached to the fascia, was rapidly soaked with fibrin-rich wound fluid and blood components. After 1 week, post-transplantation, fibroblasts and hematopoetic cells had populated the fiber-meshwork and after 2 weeks the whole scaffold was highly vascularized and densely populated with stromal cells depositing increasing amounts of ECM (Figure 1A). In addition, a large number of inflammatory cells, predominantly monocytes, had infiltrated the meshwork (not shown here). The increasing number of stromal fibroblasts had synthesized a dense provisional matrix composed of fibronectin, tenascin-C and thrombospondin, components typically found in wound granulation tissue. This material filled the spaces between the scaffold fibers as examplified for tenascin-C (Figure 1B). Additionally, a large number of newly formed CD31-stained vessels were seen throughout the whole scaffold whereas mature vessels, co-stained with an antibody to alpha smooth muscle actin ({alpha}-SMA) that marks smooth muscle cells (Figure 1C) were only found beneath. While vessel density remained high in the granulation tissue surrounding the scaffold, the number of vessels inside the scaffold decreased after 4 weeks (see also Figure 4B and D).


Figure 1
View larger version (59K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 1 Rapid formation of granulation tissue in Hyalograft 3D-scaffolds. (A) Histological cross section of a 2-week-old transplant of Hyalograft 3D-scaffold (Hy) placed on the back muscle (M) fascia of a nude mouse. H & E staining. (B) Immunostaining of tenascin-C (green) in the newly formed provisional matrix in 2-week-old transplants of Hyalograft 3D-scaffold, counterstained with DAPI (in blue). (C) Double immunostaining of a 2-week-old transplant of Hyalograft 3D-scaffold visualizing newly formed capillaries stained with an antibody to CD 31 (in red) and mature vessels below the scaffold co-stained with an antibody to {alpha} smooth muscle actin ({alpha}SMA, green) to label smooth muscle cells. (D) Histologic cross section of a 4-week-old transplant with Hyalograft 3D-scaffold showing the dense granulation tissue formed between the fibers and foreign body giant cells (FBGC) surrounding the fibers of the scaffold. H&E-staining. Bars 200 µm (A), 100 µm (B and C), 50 µm (D).

 


Figure 2
View larger version (153K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 2 Blocked tumor growth and invasion of HaCaT II-4 transplants on Hyalograft 3D-scaffold-modulated stroma. Histology of H&E-stained sections (A–F) and analysis of proliferation by means of BrdU-incorporation and –immunostaining (G and H). (A) In 4-week-old control transplants (Con) of HaCaT II-4 cells on muscle fascia, a well-differentiated invasive epithelial tumor tissue (T) has formed on a well vascularized stroma (S) with stromal strands penetrating into the epithelium. (B) The granulation tissue (stroma, S) formed on top of a Hyalograft scaffold (Hy) in 4-week-old transplants has not yet completely replaced the collagen gel (C) and is separated from the compact non-invasive tumor epithelium (T). (C) At 6 weeks, large invasive tumor masses penetrate deep into the granulation tissue (S), while keratinized stratum-corneum-like (sc) tissue is situated at the upper surface. (D) On the Hyalograft 3D-scaffold the heavily keratinized (sc) but not-invading epithelium is still separated in some areas by remnant collagen gel (C) from the granulation tissue (S) on top of the scaffold (Hy). (E) At higher magnification in control transplants invasive tips of the invasive tumor epithelium (T) exhibit many mitoses (arrows) and infiltrate into the stroma (S) rich in blood vessels (V). (F) On Hyalograft-scaffold modulated stroma (S) the compact and well delineated tumor epithelium (T) is well organized under thick stratum corneum-like (SC) keratinized layers and exhibits no invasive features, even 6 weeks after transplantation. (G) Proliferating epithelial cells were quantified by counting BrdU-labeled epithelial cells and calculating the ratio to all viable cells visualized by DAPI DNA stain. Bars represent means with standard deviation of three vision fields per animal and two to three animals analyzed per timepoint in two different experiments. (H) BrdU-labeled nuclei were randomly distributed throughout the living HaCaT II-4 cell layers in 4-week-old transplants on Hyalograft 3D-scaffold (Hy) modulated stroma (S). The demarcatation of the epithelial border clearly showed cycling cells also in the stromal part. Bars: 200 µm in A–D; 50 µm in E; 100 µm in F and H.

 


Figure 3
View larger version (139K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 3 Normalized differentiation in HaCaT II-4 transplants on Hyalograft 3D-scaffolds. Frozen sections of 4-week-old (A–F) transplants on muscle fascia (Con) and Hyalograft 3D-scaffold (Hy). Tumor cells (T) were stained with antibodies against keratin K1/10- (A and B) and K8 (E and F) in red and against ß4 Integrin (C and D) in green. The stroma (S) was counterstained in green with antibodies to nidogen (A and B) and perlecan (E and F) and in red to laminin 5 (C and D). Nuclei were marked (in blue) by the DAPI DNA stain. Bars: 50 µm (A and B); 100 µm (C–F).

 


Figure 4
View larger version (119K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 4 Blocked HaCaT II-4 tumor vascularization by scaffold-modified granulation tissue. Comparison of time-matched transplants without (Con; A, C and E) and with scaffold (Hy; B, D and F). (A–D) Frozen sections of 4- and 6-week-old transplants stained with pan-specific keratin antiserum to label tumor cells (T) in green and with an antibody to CD31 (red) to visualize endothelial cells of vessels both in the stroma (S) and in the tumor (T) parenchyma (arrows) (A–D). While invasive tumor masses had formed in control transplants (A and C), transplants on Hyalograft 3D-scaffolds (Hy) developed a compact epithelium with thick parakeratotic stratum corneum (sc) (B and D). The faint staining of scaffold fibers with the fluorescein-labelled second antibody (green) was unspecific here. (E and F) Immunohistochemical detection of VEGFR2 (in green) in combination with staining of CD31 (in red). VEGFR2 was present and colocalized with CD31 exclusively in those vessels that were located within the stromal strands penetrating the invading tumor masses in control transplants (E, arrows). In Hyalograft 3D-transplants VEGFR2 was considerably reduced and limited to few vessels in close proximity to the tumor (f, arows). In both cases, the mature, larger vessels distant from the tumors are negative for VEGFR2. Nuclei are stained in blue with DAPI. Bars: 200 µm (A–D); 100 µm (E and F) (G) Evaluation of vessel density in tumor stroma: Vessels stained by a CD31 antibody were quantified by computer-assisted morphometry on frozen sections within areas ranging from <300 µm up to 500 µm within the tumor epithelium and given as percentage of labeled area. Bars represent means with standard deviation measurements in three vision fields per animal using transplants of 4–6 animals per timepoint obtained from two independent experiments. Statistical significance was proven by students t-test indicated by asterisks (P < 0.01).

 
Starting from Week 2, increased numbers of fiber-associated multinucleated cells were observed for representing foreign body giant cells (Figure 1D). These cells, typical for a foreign body induced granulation tissue, persisted over the entire observation period also when the scaffolds started to degrade as indicated by swelling of fibers from 6 weeks on.

Altered growth behavior of low-grade malignant HaCaT II-4 cells on scaffold-modulated granulation tissue
To analyze the impact of such a scaffold-induced granulation tissue on growth and invasion of tumor cells, low-grade malignant skin SCC cells (HaCaT II-4) were transplanted together with the scaffold onto nude mice. The transplant consisted of a monolayer of HaCaT II-4 cells grown on a collagen type I gel and placed on top of the scaffold material that had been inserted onto the back muscle fascia immediately before. As a control, tumor cells grown on collagen gels were transplanted directly onto the muscle fascia without scaffold in between. After 1 week, post-transplantation, the epithelial take rate (~95%) and stromal response were comparable in both scaffold and control groups, exhibiting similar thickness and organization. After 2 weeks, however, differences became apparent. In controls, HaCaT II-4 cells had induced a well-formed granulation tissue with stromal cells and blood vessels completely populating the collagen gel leading to an immediate contact of tumor and stromal cells. At this timepoint, vessels started to infiltrate the epithelial tissue and the tumor cells began to invade the granulation tissue. In contrast, transplants with Hyalograft 3D exhibited no invasion and stromal cells had colonized only the lower half of the collagen gel, so that the granulation tissue and the vessels did not reach the tumor epithelium (not shown here).

These differences in tumor and stroma phenotype became even more striking after 4 and 6 weeks (Figure 2). In control transplants large tumor masses had formed that grew invasively into the underlying stroma with stromal strands deeply infiltrating the tumor parenchyma (Figure 2A, C and E). In transplants on scaffold material, the epithelial cells had formed a thick multilayered avascular tissue that did not invade the underlying stroma (Figure 2B and D). The granulation tissue that had formed within and around the scaffold had not yet fully replaced the collagen matrix. Even after 6 weeks, a thin rim of collagen gel remained visible in some areas of the transplants clearly separating tumor parenchyma and stroma (Figure 2D). When the granulation tissue had completely replaced the collagen gel, the compact, hyperplastic and differentiated tumor epithelium exhibited a well delineated border to the adjacent stroma without any features of invasion as seen in control transplants (Figure 2F).

This reversion from an invasive to a premalignant non-invasive tumor was not caused by reduced or blocked tumor cell proliferation, as was substantiated by quantifying BrdU-positive (DNA synthesizing) epithelial cells. Their ratio was very similar in the HaCaT II-4 transplants with or without scaffold at all timepoints analyzed (Figure 2G). In addition, these proliferating cells were similarly localized in control as well as scaffold-transplants being randomly distributed in the lower viable cell layers (Figure 2H). Furthermore, a significant role of apoptotis in establishing different phenotypes could be excluded. TUNEL-assay data not only showed equivalent ratios of apoptotic cells but also their identical localization confined to the parakeratotic compartment of HaCaT II-4 epithelia in both conditions (data not shown).

Stroma modulation restores regularly differentiated tissue architecture of malignant HaCaT II-4 cells
The normalized tissue architecture of the malignant epithelia on scaffold-modulated stroma as seen in histology (see Figure 2F) prompted us to analyze keratinocyte differentiation in more detail. HaCaT II-4 cells are capable of expressing the typical keratinocyte differentiation markers, such as keratins K1 and 10 that label all suprabasal keratinocytes in the epidermis (34). However, in control transplants—as in established tumors—these keratins were irregularly distributed and predominantly seen in central keratinized tumor areas (Figure 3A). On Hyalograft 3D-scaffold modulated granulation tissue an almost normal suprabasal labeling for K1/10 with a rather homogeneous intensity in all vital layers was obtained (Figure 3B). These epithelia were further characterized by a flattened, though parakeratotic stratum corneum and a normalized distribution of filaggrin and loricrin as well as the integrin subunits {alpha}6 and ß1 (data not shown). Also the pattern of the ß4-integrin subunit, normally restricted to hemidesmosomes, is a sensitive marker for epithelial organization. While it was highly expressed in basal and suprabasal tumor cells of the control transplants at 6 weeks (Figure 3C), it became largely restricted to the basal layers of the tumor epithelium on scaffold-modulated stroma (Figure 3D), substantiating the trend towards normal epidermal tissue architecture.

On the other hand, atypical keratins, such as the simple epithelia-type keratin K8, not present in normal epidermis, but often detected in SCCs (34,35), were continuously expressed in the basal part of the tumor epithelia in control transplants (Figure 3E). In contrast, differentiating HaCaT II-4 epithelia on scaffold-modulated granulation tissue still showed some K8-staining at 2 weeks (data not shown). However, in 4 and 6-week-old transplants only a few positive cells persisted, mostly in the uppermost layers. Thus, the progressive downregulation of this atypical keratin also argues for the normalization of the epithelium (Figure 3F).

Suppressed tumor vascularization in transplants on Hyalograft 3D-scaffolds
To determine the role of the stroma in reverting the epithelial phenotype, we first analyzed the developing granulation tissue with particular emphasis on the vasculature.

In control transplants, vessels came close to the tumor tissue as early as in the second week and after 4 weeks stromal strands with vessels penetrated deeply into the tumor parenchyma while the tumor cells had infiltrated into the granulation tissue (Figure 4A). At 6 weeks, large and well-vascularized tumor masses had developed (Figure 4C). In marked contrast, in transplants on Hyalograft 3D-scaffolds the vascularization of the zone adjacent to the tumor cells was delayed from the beginning. Only after 4 weeks, a well-vascularized granulation tissue had formed on top of the scaffold, but was still separated from the non-invasive epithelium. No vessels were observed inside the tumor parenchyma (Figure 4B). Even after 6 weeks, only the parakeratotic stratum corneum had increased (Figure 4D), but the compact tumor epithelium remained completely avascular, non-invasive and clearly delineated from the vascularized granulation tissue underneath. As shown in Figure 4B and D, the scaffold fibers (Hy) can be easily distinguished in the lower stroma due to the unspecific staining of the scaffold filaments with the secondary antibody. To analyze the mechanisms that underlie this difference in the angiogenic phenotype, the activation state of the vascular endothelial cells was proven by immunostaining for the VEGFR2, an accepted marker for actively sprouting vessels (25) In the control transplants, the endothelial cells of the small vessels in the stromal strands within the tumors contained significant amounts of VEGFR2. In contrast, the endothelial cells in the mature large vessels distant from the tumor almost completely lacked VEGFR2 expression (Figure 4E). On Hyalograft 3D-scaffold, only very faint VEGFR2 staining was detected in few vessels and only directly adjacent to the invasion inhibited tumor cells (Figure 4F). When vessel density was quantified morphometrically, the area covered by vessels was very similar in 2-week-old transplants but was significantly reduced in 4 and 6-week-old transplants on Hyalograft 3D-scaffolds (Figure 4G). As the underlying granulation tissue was well vascularized in both experimental conditions, this discrepancy was mainly due to the lack of intraepithelial vessels. Thus, the scaffold-modulated stromal tissue obviously exerted an inhibitory effect on penetration of vessels into the tumor parenchyma.

Stromal differentiation in scaffold-modulated granulation tissue
To further elucidate the mechanism causing this difference in tumor phenotype we analyzed cell proliferation, cellular composition and ECM in the stromal compartment. The proliferative activity of stromal cells, evaluated by BrdU labeling, showed no major difference between the control and the scaffold-modulated granulation tissue at all timepoints analyzed (not shown). Similarly, colonization by stromal cells, i.e. influx of neutrophil granulocytes, mast cells and macrophages, as well as their distribution in the granulation tissue below the tumor parenchyma was undistinguishable in both experimental conditions (see Supplementary Figure S1). However, when staining for {alpha}-SMA, a characteristic component of smooth muscle cells and myofibroblasts, remarkable differences became obvious. In control transplants, smooth muscle cells were predominantly detected in the walls of larger vessels while free myofibroblasts were rare at all time points (Figure 5A). In contrast, the number of myofibroblasts steadily increased in the granulation tissue on scaffolds, reaching a dense accumulation after 4- and 6 weeks (Figures 5B and 7H).


Figure 5
View larger version (133K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 5 Reversion of tumor phenotype paralleled by altered stromal composition. (A and B) Immunofluorescence analysis on 4-week-old HaCaT II-4 transplants on either muscle fascia (Con A and C) or Hyalograft 3D-scaffold (Hy B and D) with antibodies to {alpha}-SMA (in red) and collagen typ I (in green). In control transplants {alpha}-smooth muscle actin-containing cells were restricted to blood vessels (arrows in A), while in Hyalograft 3D-transplants the stroma harbored dense aggregates of myofibroblasts (myoF in B) at the periphery of the scaffold-material. This was paralleled by a more compact texture of collagen type I. (C and D) The antiangiogenic ECM-component TSP-1 (stained in red, green counterstaining of keratin) was completely lacking in control transplants with invasive tumor growth (c) whereas it was contiguously deposited underneath the reverted tumor on Hyalograft 3D (D). Nuclei were stained with DAPI DNA stain. Scale bars 100 µm.

 


Figure 6
View larger version (185K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 6 Ultrastructure of the tumor–stroma border in HaCaT II-4 transplants. (A) In 3-week-old control transplants of HaCaT II-4 cells many tumor cell (T) projections protrude in the adjacent stroma (S) filled with vesicles, ECM material and altered vessels (V). (B) Higher magnification presents tumor projections (arrows), vesicles and a swollen capillary endothelial cell (EC) with altered interendothelial junctions (arrowheads) in more detail (Ery = erythrocyte in capillary lumen). (C) In 6-week-old transplants on Hyalograft 3D-scaffolds the tumor cells (T) are clearly delineated by a continuous BM (arrows) from the stroma with fibroblasts (F) and remnant vesicles (V). (D) At higher magnification the tumor–stroma border shows a continuous lamina lucida (LL) and lamina densa (LD) separating tumor cells from the stroma rich in collagen bundles (C). Hemidesmosomal junctions (HDs) at the tumor cell membrane are connected with the enriched network of intermediate filament bundles (IF). (E) The granulation tissue formed on top of Hyalograft 3D-scaffolds is composed of parallel arranged activated fibroblasts (F) and myofibroblasts (myoF) embedded in a dense ECM of collagen bundles (C). (F) At higher magnification the periphery of a myofibroblast reveals rough endoplasmic reticulum (RER) and bundles of microfilaments (mf), aligned parallel to the cell membrane (arrowheads) forming fibronexus-type connections with extracellular ECM fibers in a collagen fiber (C)-rich environment, also detailed in the insert (arrows). Bars: 1.6 µm (A), 0.6 µm (B), 3.2 µm (C), 0.2 µm (D), 3.2 µm (E), 0.8 µm (F), 0.6 µm (insert).

 


Figure 7
View larger version (139K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 7 Decreased protease expression and activity in malignant HaCaT II-4 transplants upon Hyalograft 3D-scaffolds. Analysis of 6- (A, G and I) and 4-week-old transplants (c,e) of HaCaT II-4 cells on muscle fascia (Con) and on Hyalograft 3D-scaffold (Hy, B, H and J and D and G, respectively). (A and B) Fluorescent in situ zymography with DQ-gelatine detecting protease activity by green fluorescence. Sections were counterstained in red with an antibody to MMP-9. (C and D) Analysis of MMP-1 mRNA expression by in situ hybridization (ISH, presented in yellow false color), counterstained with collagen type IV antiserum (in red). Tumor cells were labelled with panreactive keratin antiserum (in green); (E and F) MMP-2 mRNA expression in HaCaT II-4 transplants on muscle fascia (E) and on Hyalograft 3D-scaffold (F). In situ hybridization with RNA probes specific for MMP-2 (in yellow) combined with staining of collagen type IV (red) and keratin (green). In general, MMP mRNA expression was dramatically reduced in transplants on Hyalograft 3D-scaffolds (D and F) in comparison to controls (C and E). Additionally, in the ISH-specimens the collagen type IV staining of epithelial BM was abolished due to the harsh pretreatment involved in the ISH-procedure (compare with J). G and H) Immunofluorescence staining of control (G) and scaffold- transplants (H) for MMP-2 (in green) and {alpha}-SMA (in red). According to its m-RNA distribution, MMP-2 protein appears beneath invasive tumor masses in control transplants, while only faint staining is detected close to the myofibroblasts in scaffold-transplants. (I and J) Demonstration of uPA [stained in red) in control (I) and scaffold-transplants (J)]. In contrast to invasive tumors, the non-invasive tumor epithelia on Hyalograft 3D-stroma are devoid of uPA, but show a linear deposition of collagen type IV (in green, arrows in J), whereas in controls only vascular BMs are demarcated (I). All images counterstained with DAPI DNA stain (blue). Scale bars: 50 µm.

 
A variety of ECM-components belonging to several functional groups of ECM was analyzed. Typical constituents of provisional matrix indicative for regenerative processes such as tenascin and fibronectin did not reveal major differences on both types of stroma (not shown). Also the BM components nidogen (Figure 3A and B), laminin-5 (Figure 3C and D) and perlecan (Figure 3E and F) had similar expression patterns, while laminin-10 (not shown) and collagen type IV were significantly increased in the epithelial BM on scaffold modulated stroma (see Figure 7I and J). The linear deposition of these major BM constituents indicates regular BM formation and fits to the ultrastructural findings (see below and Figure 6D). The main structural elements of interstitial dermal ECM are collagen fibrils composed of collagens type I and type III. Both of them were enhanced and more densely packed in the stromal tissue formed on Hyalograft 3D-scaffold (shown for collagen type I in Figure 5A and B).

The suppressed tumor vascularization prompted us to also analyze for ECM-components known to be inhibitors of angiogenesis. We have previously shown that thrombospondin-1 (TSP-1) is very effective in blocking angiogenesis in experimental SCCs (36). TSP-1 was absent in control transplants but contiguously deposited beneath the non-invading tumor epithelia of transplants on Hyalograft 3D (Figure 5C and D). As presented in Supplementary material (Figure S2 a and b), a similar distribution was found for endostatin, a potent antiangiogenic factor and proteolytic fragment of the BM constituent collagen type XVIII (col XVIII) (37,38). It is one aspect of endostatin's mode of anti-angiogenic action that it mimics the presence of col XVIII typical for a mature BM thereby preventing endothelial cells from further proliferating and sprouting (39,40). Therefore, the staining pattern obtained reflected functional significance even though the antiserum used could not discriminate between endostatin and col XVIII. Thus, the finding that the vessels in the tumor parenchyma were mostly devoid of endostatin/col XVIII indicated their status of unrestricted proliferative activity. On the other hand, the vessels in the transplants on scaffold-induced stroma showed an intense staining consistent with their quiescent state.

Ultrastructural normalization of the tumor–stroma border zone
The impact of stroma modulation on tissue organization was further characterized by ultrastructural analyses. In 3-week-old control transplants of HaCaT II-4 cells the invading epithelia had formed many filopodia and membrane vesicles filling a broad, loosely organized border zone to the underlying granulation tissue (Figure 6A and B). Large vesicles accumulated with further tumor growth so that the border line between tumor and stroma was difficult to identify. In contrast, 6-week-old transplants on Hyalograft 3D-modulated stroma showed a clear delineation between the compact tumor cells and the stromal area and also number and size of vesicles were significantly reduced (Figure 6C). At higher magnification, the sharp delineation of the tumor–stroma border was even more evident and included typical features of an epithelial BM (lamina densa) as well as hemidesmosomes at the epithelial cell membrane (Figure 6D). In addition, as compared to control transplants the keratinocytes contained an increased density of intermediate filament bundles, partly inserting in hemidesmosomes. All these observations fully correspond to the enhanced cellular organization and polarization seen in histology and immunofluorescence staining (see Figures 2 and 3).

The normalized tumor epithelium in scaffold transplants was situated on top of a cell-rich stromal tissue with densely packed and mostly horizontally oriented fibroblasts surrounded by accumulations of thick collagen filament bundles (Figure 6D and E). This confirmed the intensified immunohistochemical reactions found for collagen types I (see Figure 5A and B) and III and decorin (not shown here). Also in good agreement with the immunohistochemical findings, many of the fibroblasts in the scaffold-modulated stroma shared characteristics of myofibroblasts. They displayed abundant rough endoplasmic reticulum and microfilament bundles parallel to the cell membrane (Figure 6E and F) forming the typical fibronexus with ECM bundles of fibronectin (insert to Figure 6F). Since none of these features were observed in the stroma of control transplants, they most likely were causally related to the state of inhibited invasion.

Modulated stroma affects epithelial protease expression and activity
Since proteases are essential for tumor cell invasion we further analyzed protease expression patterns and proteloytic activity in both types of HaCaT II-4 tumor transplants. Fluorescent in situ zymography revealed clear differences in gelatinolytic activity. While control transplants exhibited distinct levels of gelatinolysis, which were mostly confined to the tumor–stroma border where tumor cells actively invaded the host tissue (Figure 7A), gelatinolytic activity was not detectable in tumor cells on Hyalograft 3D-scaffolds (Figure 7B). To distinguish which proteases were involved and to determine whether the altered protease activities were based on inhibition of protease activities or reduced protease expression, we performed in situ hybridization with RNA-probes for MMPs 1, 2 and 9, as well as, immunostainings with antibodies against the respective MMPs. In control transplants, we found expression of MMP-1 mRNA primarily in the basal tumor cells facing the stroma (Figure 7C). In contrast, in the corresponding epithelia on scaffolds only minor MMP-1 mRNA expression was detected (Figure 7D). In concordance with the RNA levels, an intense peritumoral protein staining for MMP-1 in controls was opposed by a faint subepithelial reaction in scaffold transplants (data not shown). Even more striking differences were seen for MMP-2. In situ hybridization demonstrated high MMP-2 mRNA levels in control transplants once the tumor cells had started to invade the adjacent stroma. The expression reached its maximum after 4 weeks (Figure 7E). On the other hand, in the tumor cells of scaffold transplants no MMP-2 signal appeared at any time point analyzed (Figure 7F), though few weak MMP-2 signals were visible in the myofibroblasts located in the fibrotic stroma above the scaffold (Figure 7F). Accordingly, high levels of MMP-2 protein were found in the stroma adjacent to the invasive HaCaT II-4 epithelium in control transplants (Figure 7G), whereas only little MMP-2 protein could be identified in the myofibroblasts-rich areas of scaffold transplants (Figure 7H).

Whereas MMPs 1 and 2 are attributed to the HaCaT II-4-keratinocytes, MMP-9 is a stroma derived protease mainly produced by granulocytes. In this study, MMP-9 mRNA expression was undetectable in both types of granulation tissue (not shown). Furthermore, MMP-9 containing granulocytes showed a similar and sparse distribution arguing against a significant contribution of this enzyme in the observed gelatinolysis (Figure 7A and B).

As an additional protease, known to be positively correlated with tumor cell invasion in SCCs (41), urokinase-type plasminogen activator (uPA) was included in our study. By immunostaining, uPA was localized in the periphery of invading tumor masses in 6-week-old control tumors (Figure 7I). In contrast, uPA was not detectable in tumor epithelia grown on Hyalograft 3D-scaffolds at any timepoint analyzed; instead a well developed epithelial BM containing increased amounts of collagen type IV became apparent (Figure 7J).

In summary, these results establish that reduced activities of matrix degrading enzymes in the tumor cells accompany the phenotypic tumor reversion on scaffold-modulated stroma.

Partial reversion of tumor phenotype in high-grade malignant HaCaT A-5RT3 transplants
To analyze whether this normalizing effect on the tumor phenotype through Hyalograft 3D-scaffold-induced stromal modulation was restricted to low grade tumor cells, the same set of transplantation experiments was performed using the high-grade malignant HaCaT A-5RT3 cell line (27,28). These cells form rapidly growing poorly differentiated and metastasizing tumors following s.c. injection of as few as 2 x 105 cells so that animals have to be sacrificed after 4 weeks. In surface transplants on collagen gels, A-5RT3 cells rapidly induce the formation of a strongly vascularized granulation tissue that replaces the collagen gel within 1 week so that epithelial invasion and stromal infiltration into the epithelium is already obvious after 2 weeks (26). Thus, in 3-week-old control transplants, large invasive and well-vascularized tumor tissues had formed (Figure 8A). In contrast, stromal tissue modulated by Hyalograft 3D-scaffolds significantly inhibited tumor expansion and invasion. In 3-week-old transplants a multilayered epithelial tissue with only minimal invasion was observed (Figure 8B). After 5 and 6 weeks, post-transplantation, the cells in controls had formed extended invasive lobular tumor masses, that were well vascularized (Figure 8C). On the other hand, the tissue formed on the scaffold-modulated stroma strongly resembled a compact epithelium. These epithelia were poorly vascularized and clearly delineated to the underlying stroma (Figure 8D). This was also well reflected at the ultrastructural level. Control transplants of A-5RT3 cells exhibited infiltrating tumor cells into a loose stromal border zone, a typical feature of a rapidly invading carcinoma (Figure 8E). In contrast, the epithelial tissue on scaffold-modulated stroma was clearly demarcated and showed stretches of condensed material with typical features of a lamina densa and hemidesmosomes at the basal epithelial cell membrane (Figure 8F). Moreover, the neighboring stroma was densely packed with activated fibroblasts in-between collagen bundles (data not shown).


Figure 8
View larger version (128K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 8 Impaired growth, invasion, and vascularization of high-grade malignant HaCaT A-5RT3 cells on Hyalograft 3D-scaffolds. (A and B) Histological sections of 3-week-old control transplants on muscle fascia (A) and Hyalograft 3D-scaffold (Hy, B) with invading tumor masses (T) in control stroma (S) and poorly invading tumor epithelium (T) confronting stroma (S) on top of Hyalograft 3D-scaffold (Hy), H&E-staining. (C and D) Immunostaining of frozen sections of 5-week-old transplants on control (C) and scaffold modulated (D) stroma (S), labeling tumor cells with a pan keratin antibody (green) and vessels by a CD31 antibody (red) exhibiting many intratumoral vessels (v) in control but few in transplants on Hyalograft 3D (Hy) scaffolds (arrows, D). (E and F) Ultrastructure of tumor–stroma border of 5-week-old HaCaT A-5RT3 transplants on control (E) and scaffold modulated (F) stroma (S). (E) Infiltrative tumor cells (T) and protrusions in the ECM and fibroblast (F) rich stroma. (F) Tumor cell (T) border at higher magnification demonstrating patches of lamina densa (arrows) separating the thickened tumor-cell membrane with type II-hemidesmosomes (arrowheads) from the ECM rich stroma (s). Bars: 200 µm (A and B), 400 µm (C and D), 4 µm (E), 0.4 µm (F).

 
Taken together, our data clearly demonstrate an invasion-inhibiting and tumor-phenotype-reverting effect of the scaffold-modulated stroma that is not restricted to low-grade malignant tumors but still affects, though to a lesser extent, high-grade malignant SCCs.


   Discussion
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
Supplementary material
 References
 
The data presented here add to the growing body of evidence that alterations of the microenvironment substantially affect malignant tumor growth and invasion (4–10,13–17,20). The biodegradable non-woven filamentous scaffold of esterified hyaluronan (Hyalograft 3D) rapidly induced formation of a granulation tissue. This scaffold-modulated stroma resulted in long-term inhibition of tumor cell invasion and reversion of the tumor phenotype when co-transplanted with low-grade malignant human skin SCC cells. The phenotypic reversion included (i) architectural normalization of epithelial tissue organization and polarization as evident by a normalized expression of typical markers for regular epidermal differentiation and dermo–epidermal junction, (ii) suppression of atypical differentiation markers and (iii) the appearance of a structured BM at the tumor–stroma border zone. Although tumor cell-induced angiogenesis was not inhibited initially, infiltration of vessels into the tumor parenchyma, in this assay intimately associated with tumor cell invasion (25,26), was completely blocked. These alterations were paralleled by distinct changes in the neighboring stroma with augmentation of ECM including anti-angiogenic ECM components, decrease of peritumoral protease activity as well as accumulation of myofibroblasts and collagen bundles, indicative of a fibrotic tissue. The changes in tumor phenotype exerted by an altered stromal environment were most striking in transplants of low-grade malignant cells, but a similar trend became obvious also in transplants of the high-grade metastatic cell line HaCaT A-5RT3. This underscores that tumor surveillance mediated by the stromal microenvironment is highly efficient at all stages of tumor development and progression.

Stroma modulates tumor growth and invasion
The microenvironment of a solid tumor is the result of a complex interaction between neoplastic cells, stromal cells (fibroblasts and endothelial cells), and inflammatory cells (1,42,43). Thus, tumors have to be regarded as complex tissues in which mutant cancer cells have subverted normal stromal cells to serve as active collaborators in their neoplastic agenda. Successful tumor cells are those that have acquired the ability to co-opt their normal neighbors by inducing them to release growth stimulating signals and matrix-degrading proteases (42,43). The supportive role of an activated tumor stroma for tumor invasion was exemplified by the increased invasion of carcinoma cells when transplanted onto a preinduced granulation tissue (13,44). Similarly, rat adenocarcinoma cells only exhibited an invasive phenotype when implanted in an experimentally induced subcutaneous granulation tissue, but not when surrounded by an undisturbed subcutaneous tissue (17). Moreover, regenerating and rapidly proliferating organs specifically enhanced growth of local and metastatic tumors (14–16). On the other hand, studies have demonstrated that normal tissue fibroblasts could also have ‘normalizing’ effects on early-stage or premalignant tumor epithelia (14,45), while aging fibroblasts are thought to rather exert tumor promoting activity (46).

Modulation of tumor phenotype by Hyalograft 3D in surface transplants
Detailed in vivo studies on the characterization of tumor stroma elements and their interaction with tumor cells are problematic in conventional injection assays for tumor induction (47). Therefore, we have elaborated a matrix-inserted surface transplantation model, which provides several crucial advantages. First, matrix-attached cells are transplanted and provide high take rates. Second, the inserted matrix allows the development of an extensive granulation tissue induced by tumor-cell derived factors (25,31,48). Using this heterotransplantation model we have shown that tumor cell invasion and malignant tumor phenotype can be modulated by alteration of the tumor stroma, e.g. by (i) a disturbed balance between serine proteases and plasminogen activator inhibitor-1 (PAI-1) (48,49); (ii) epithelial overexpression of platelet-derived growth factor-B (PDGF-B) (50); (iii) forced overexpression of TSP-1 (36) and iv) inhibition of angiogenesis by blockade of VEGFR2 signaling (25,26).

Here we utilize this surface transplantation assay for analyzing the effect of the stromal tissue induced by the Hyalograft 3D-scaffold on tumor growth. Comparing control and scaffold transplants, differences in the tumor-cell induced granulation tissue became obvious already after 2 weeks with a delayed penetration of stromal tissue through the collagen matrix. These differences were still maintained at 4 weeks. Most importantly, the infiltration of blood vessels into the tumor epithelium was completely prevented in scaffold-modulated HaCaT II-4 cell transplants. In agreement with earlier reports (25,26), this resulted in suppressed tumor invasion. As we present here, these effects are accompanied by a dramatic drop in VEGFR expression in the endothelial cells along with an augmentation of ECM molecules that are potent inhibitors of angiogenesis (36–38). Definitely, angiogenic stimuli by the SCC cells in our transplantation assay are only active over short distances (51) thereby accounting for the restriction of induced angiogenesis to the tumor area. However, the scaffold-modulated stroma provides not only ECM elements that act directly antiangiogenic but seems also to instigate the tumor cells to convert their invasive and proangiogenic behavior. We found that cellular stromal components such as inflammatory cells are not the critical effectors since neither macrophages nor neutrophils showed major differences between control and Hyalograft 3D transplants (see Supplementary Figure S1). Instead, stromal ECM composition might gain in importance as functional stromal feature in this context.

Interestingly, these stromal modifications had no effects on tumor cell proliferation, as measured by BrdU-labeling of the transplants, thus eliminating the possibility of reduced supply of tumor cells with nutrients and growth factors. Moreover, normal keratinocytes on collagen gel cotransplanted with Hyalograft 3D-scaffolds developed into a normally structured and differentiated epidermis with equal speed and quality as seen in transplants without scaffolds (H.-J. Stark, M. J. Willhauck and N. E. Fusenig, unpublished data). All this makes it highly unlikely that the scaffold itself influenced tumor cell growth.

The influence of hyaluronic acid (HA) that is introduced in form of Hyalograft 3D on the surrounding stroma has to be considered carefully because involvement of HA in stromal control on tumor phenotype was previously reported (52). HA is enriched in tissues undergoing rapid turn over, such as wound granulation tissue and tumor stroma. In vitro, HA promoted growth and spread of tumor cells and in vivo, a HA-rich environment correlated with tumor progression and enhanced tumor invasion (53). Accordingly, treatment with hyaluronidase caused regression of human breast cancer xenografts in SCID mice (54). However, it also was shown that, depending on the degree of polymerization, HA did either stimulate or inhibit angiogenesis (55–58). The Hyalograft 3D used in this study consists of highly esterified HA that is initially protected against hyaluronidase-mediated degradation (59,60). In fact, only after 6 weeks transplants showed some degradation as indicated by swelling and erosion of the fibers. Since the scaffold-mediated effect on tumor invasion was already visible during the first 4 weeks, i.e. when a fully intact fiber meshwork is still present, free HA is unlikely to participate in tumor phenotype modulation. In line with this, we have preliminary evidence that also less esterified material (Hyalofil) that disintegrates within a week after s.c. implantation is not increasing the effect but rather leads to a somewhat reduced modulation, while another non-degradable meshwork of polypropylene had the same effect on the granulation tissue as the Hyalograft 3D and inhibited HaCaT II-4 cell invasion (M. J. Willhauck, H.-J. Stark and N. E. Fusenig, unpublished data). From all this, we conclude that it is not the chemistry of HA that is crucial for the observed phenotypic reversion, but the rigid scaffold structure in the stroma.

Upon implantation, rigid structures generally elicit foreign body reactions. These are characterized by an initial inflammatory response with a granulation tissue containing macrophages that fuse to foreign body giant cells and degrade the implant (61,62). Different from the granulation tissue beneath control transplants without scaffolds, transplants containing Hyalograft 3D scaffolds induced those foreign body reactions with accumulating multinucleated giant cells that appeared already after 2 weeks and persisted for months (H.-J. Stark, M. J. Willhauck and N. E. Fusenig, unpublished data). These giant cells are supposed to stimulate fibroblast activation and ECM production inducing a fibrotic metaplasia (63). In fact, such a foreign body reaction was shown to prevent methylcholanthrene-induced fibrosarcoma formation in mice (64). Accordingly, the ‘desmoplastic response’ often found in the stroma of human carcinomas has been interpreted as a host response by ‘walling off’ the tumor and preventing invasion (65). Another prominent example for tumor inhibition by fibrosis are cirrhotic livers, which are less susceptible to metastatic spread than normal livers (66–69). Here as well, the resistance to metastasis is supposedly due to altered ECM composition and increased content of MMP-inhibitors (70).

On the other hand, as reported for man and rodents (71,72), implanted foreign body structures can also elicit tumor formation, suggesting that scaffolds could be tumorigenic per se. Importantly, however, sarcomas developed solely in contact with smooth surfaces and only after long latency periods (73,74). This is quite different from the Hyalograft 3D meshwork that (i) represents a porous, ‘rough surface’ foreign body, (ii) induces a histologically different non-sarcoma-type foreign body tissue and (iii) is finally degraded in the transplants. Therefore, a tumorigenic contribution of this material is largely excluded.

Taken together, there are several lines of strong evidences identifying fibrosis as a potent anti-tumor effect that is apparently independent of how the fibrotic state is developed. Obviously, by implanting Hyalograft 3D-scaffold, such a specifically active fibrotic state was achieved in the induced stromal tissue ultimately leading to the phenotypic reversion of the malignant tumor.

Functional aspects of altered stromal composition in Hyalograft 3D-transplants
Myofibroblasts, which are regarded as a differentiated phenotype of fibroblasts, are important players in the modulated fibrotic granulation tissue. Their characteristic feature is the expression of {alpha}-smooth muscle actin and specific ultrastructural criteria (75). Myofibroblasts are also found in cancer-induced stromal tissues and may have positive or negative influence on tumor progression (76,77). Around in situ tumors, myofibroblasts strongly express lysyloxidase, an enzyme involved in collagen and elastin crosslinking. The resulting increase in connective tissue tenacity has been suggested as anti-tumor defense mechanism (78).

Thus, the increase of myofibroblasts in the Hyalograft 3D-scaffold-modulated granulation tissue may indicate a role of these activated fibroblasts in the blockade of tumor invasion, e.g. by enhanced production and secretion of ECM components. Accordingly, we observed not only the ultrastructural accumulation of regular collagen fibril bundles but also elevated levels and more compact textures of collagen type I and type III demonstrating a considerable condensation of the ECM. However, the effect of the scaffold-induced stromal tissue clearly exceeds that of a passive physical barrier for tumor invasion by also triggering a remarkable downregulation of protease activity in the tumor cells. The peritumoral proteolysis detectable in this tumor model by in situ zymography is not associated with MMP-9 containing neutrophils, but is largely attributed to the expression of MMPs 1, 2 and uPA by the tumor cells. Those proteolytic enzymes are all well characterized factors fostering the process of keratinocyte activation, migration and tumor invasion (79–83) and we have preliminary evidence that the expression of MMP-1 by the tumor cells represents a critical step in establishing an invasive tumor phenotype in our keratinocyte tumor model (S. Vosseler and M. M. Mueller, unpublished data). Responsible for their downregulation on Hyalograft 3D-scaffold are obviously specific functional properties of the modulated stroma comprising (i) instructive cues from the ECM, as well as, (ii) paracrine signals sent out by the myofibroblasts in concert with the foreign body giant cells. Several stroma-derived paracrine signals that are supportive for the invasive growth of SCCs have been well defined e.g. TNF-{alpha} and HGF (84,85). However, to the best of our knowledge, no conclusive data are available about distinct tumor inhibiting effects of stroma-derived cytokines. Instead, a growing body of evidence has been emerging that underlines the tumor-suppressive potential of the ECM component decorin in different types of tumors (86–88). Indeed, our preliminary data indicate that decorin is increased in Hyalograft 3D-induced stroma (data not shown). As ‘matrikine’ it was shown to represent a ligand of the EGF-receptor; yet in part it elicites downstream responses antagonistic to EGF (89–91). Therefore, decorin in combination with the EGFR signaling cascade is a promising candidate likely to contribute to the described tumor reverting effect of scaffold-modulated stroma. In addition to the anti-angiogenic properties of certain ECM components, these mechanistic aspects are relevant issues in tumor–stroma-interaction and require further investigative efforts.

In summary, the data reported here further corroborate the important regulatory role of the tissue microenvironment in tumor development and progression. Moreover, they suggest that the different components of the tumor stroma are possible targets for tumor therapy.


   Supplementary material
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Supplementary material
References
 
Supplementary material is available at carcinogenesis online.


   Acknowledgments
 
We thank Gaby Blaser, Angelika Krischke, Iris Nord and Heinrich Steinbauer for expert technical assistance, Martina Kegel for help in preparing the manuscript, Dirk Breitkreutz for numerous valuable discussions and Wiltrud Lederle for many helpful informations and the DIG-labeled MMP-1-probe. This work was supported by EU-grant QLK3-CT-2002-02136 to N.E.F. and LSHG-CT-2005-503447 to P.B., a grant from the Deutsche Krebshilfe e.V. (Verbund Beltinger 106415) to P.B. and by Fidia Advanced Biopolymers, Abano Terme, Italy.

Conflict of Interest Statement: None declared.


   References
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Supplementary material
 References
 

  1. Liotta L.A. and Kohn E.C. (2001) The microenvironment of the tumour-host interface. Nature 411:375–379.[CrossRef][Medline]
  2. Bhowmick N.A. and Moses H.L. (2005) Tumor–stroma interactions. Curr. Opin. Genet. Dev. 15:97–101.[CrossRef][ISI][Medline]
  3. Dvorak H.F. (2003) Rous-Whipple award lecture. How tumors make bad blood vessels and stroma. Am. J. Pathol. 162:1747–1757.[Free Full Text]
  4. Fidler I.J. (2003) The pathogenesis of cancer metastasis: the ‘seed and soil’ hypothesis revisited. Nat. Rev. Cancer 3:453–458.[CrossRef][ISI][Medline]
  5. Kunz-Schughart L.A. and Knuechel R. (2002) Tumor-associated fibroblasts (part I): Active stromal participants in tumor development and progression? Histol. Histopathol. 17:599–621.[ISI][Medline]
  6. Tuxhorn J.A., Ayala G.E., Smith M.J., Smith V.C., Dang T.D., Rowley D.R. (2002) Reactive stroma in human prostate cancer: induction of myofibroblast phenotype and extracellular matrix remodeling. Clin. Cancer Res. 8:2912–2923.[Abstract/Free Full Text]
  7. Noel A. and Foidart J.M. (1998) The role of stroma in breast carcinoma growth in vivo. J. Mammary Gland. Biol. Neoplasia. 3:215–225.[CrossRef][ISI][Medline]
  8. Cunha G.R., Hayward S.W., Wang Y.Z., Ricke W.A. (2003) Role of the stromal microenvironment in carcinogenesis of the prostate. Int. J. Cancer 107:1–10.[CrossRef][ISI][Medline]
  9. Tlsty T.D. (2001) Stromal cells can contribute oncogenic signals. Semin. Cancer Biol. 11:97–104.[CrossRef][ISI][Medline]
  10. Mueller M.M. and Fusenig N.E. (2004) Friends or foes - bipolar effects of the tumour stroma in cancer. Nat. Rev. Cancer 4:839–849.[CrossRef][ISI][Medline]
  11. Javaherian A., Vaccariello M., Fusenig N.E., Garlick J.A. (1998) Normal keratinocytes suppress early stages of neoplastic progression in stratified epithelium. Cancer Res. 58:2200–2208.[Abstract/Free Full Text]
  12. Morikawa K., Walker S.M., Nakajima M., Pathak S., Jessup J.M., Fidler I.J. (1988) Influence of organ environment on the growth, selection, and metastasis of human colon carcinoma cells in nude mice. Cancer Res. 48:6863–6871.[Abstract/Free Full Text]
  13. Gabbert H. (1985) Mechanisms of tumor invasion: evidence from in vivo observations. Cancer Metastasis Rev. 4:293–309.[CrossRef][ISI][Medline]
  14. Fabra A., Nakajima M., Bucana C.D., Fidler I.J. (1992) Modulation of the invasive phenotype of human colon carcinoma cells by organ specific fibroblasts of nude mice. Differentiation 52:101–110.[CrossRef][ISI][Medline]
  15. Camps J.L., Chang S.M., Hsu T.C., Freeman M.R., Hong S.J., Zhau H.E., von Eschenbach A.C., Chung L.W. (1990) Fibroblast-mediated acceleration of human epithelial tumor growth in vivo. Proc. Natl Acad. Sci. USA 87:75–79.[Abstract/Free Full Text]
  16. Cornil I., Theodorescu D., Man S., Herlyn M., Jambrosic J., Kerbel R.S. (1991) Fibroblast cell interactions with human melanoma cells affect tumor cell growth as a function of tumor progression. Proc. Natl Acad. Sci. USA 88:6028–6032.[Abstract/Free Full Text]
  17. Gregoire M. and Lieubeau B. (1995) The role of fibroblasts in tumor behavior. Cancer Metastasis Rev. 14:339–350.[CrossRef]