Injury-Driven Stiffening of the Dermis Expedites Skin Carcinoma Progression

Cutaneous squamous cell carcinoma (cSCC) is among the most commonly diagnosed cancers with invasive and metastatic potential. Although mostly moderately progressive and easily resectable, approximately 4% of cSCCs metastasize (1). Besides UV irradiation as the major cause of cSCC in the general population (2), also chronic inflammation, scarring, papilloma virus infection, or genetic factors have been postulated to contribute (1, 3).

In certain genetic skin disorders, the incidence of cSCC is highly elevated (1). One example is recessive dystrophic epidermolysis bullosa (RDEB), a skin fragility disorder (4). The cumulative risk of invasive cSCC in the most severe form of RDEB is > 90% by the age of 55, and 80% of the patients die of metastatic tumors within 5 years after their first cSCC (4).

RDEB is caused by mutations in the COL7A1 gene encoding collagen VII (C7), an epidermal–dermal adhesion protein and a component of the anchoring fibrils (5). Loss of C7 functions leads to friction-induced separation of the skin layers, clinically visible as blisters, and healing with scarring. Typically, aggressive cSCCs arise at strongly scarred sites (5).

Very little is known about the molecular mechanisms underlying the belligerent behavior of RDEB-cSCC (6). Although C7 supports adhesion and migration of keratinocytes in vitro (7), it is unlikely that its loss drives carcinogenesis, because keratinocytes do not usually come into direct contact with C7 in situ. Further, no consistent genetic changes distinguish RDEB-cSCC from less-aggressive common cSCCs, except for COL7A1 mutations (8).

Alternatively, the dermal tissue may play an enabling role for carcinoma progression in RDEB. Many cancers generate an adjacent stroma with altered composition and mechanical properties of the extracellular matrix (ECM) that facilitates invasion (9, 10). However, important tissue-specific differences exist (11), and the influence of the ECM on cancer development cannot be generalized for all tumor types.

Here, we addressed the mechanisms of the dermal contribution to cSCC progression using in vivo and in vitro genetic models, RDEB mice, and three-dimensional (3D) organotypic skin cultures populated with RDEB and cSCC cells. Complete absence of C7 in mice causes early postnatal lethality (12). Therefore, we used a C7 hypomorphic mouse model, here referred to as the RDEB mouse, which shows all signs of human RDEB (13, 14). The results of our study indicate that in RDEB, cSCCs do not interact with a self-generated tumor stroma, but rather with a pre-existing, genetically altered and injury-modified stiff microenvironment, which enables rapid carcinoma progression. In therapeutic terms, inhibition of advancing tissue stiffness reduced invasion and is likely to have an impact on carcinoma prevention in RDEB.

Studies using patient material were approved by the ethics committee of the University of Freiburg (approval no. 127/04) and conducted according to the declaration of Helsinki. The patients provided informed consent for use of their biomaterials for research. All animal experiments were approved by the regional review board (Regierungspräsidium Freiburg, Freiburg, Germany; approval no. G-08/50).

The RDEB mouse, which has been described in detail elsewhere (13, 14), was kept on mixed C57BL/6, 129sv background. The mice were housed in the pathogen-free facility at the University of Freiburg and given food and water ad libitum; in addition, the RDEB mice were provided with a soft food diet for extra support.

Cutaneous tumors were induced by two-stage 7,12-Dimethylbenz[a]anthracene (DMBA)/12-O-tetradecanoylphorbol-13-acetate (TPA) chemical carcinogenesis (15) following the treatment regimen displayed in Fig. 2A. One week after the final TPA application, the mice were sacrificed, tumors counted, and photographed with Canon power shot S3IS camera (Canon). Tumor-primed skin was isolated from mice treated with the initial dose of DMBA and twice-weekly applications with TPA for a period of 7 weeks. Untreated skin specimens were obtained from back skin of 10-week-old mice. Skin and tumor specimens were embedded either in optimal cutting temperature compound (Sakura) or paraffin for immunohistochemical or immunofluorescence analyses.

All patients had a genetic diagnosis of RDEB, and complete C7 deficiency was molecularly confirmed. Skin biopsy specimens were taken from perilesional skin adjacent to blisters on limbs; age- and site-matched biopsies from healthy human donors served as controls. RDEB-cSCC tumor specimens were derived from surgical excision material. RDEB-cSCC keratinocytes (RDEB-SCCK) were isolated from primary invasive RDEB-cSCC and spontaneously immortalized in culture. Normal human and RDEB patient keratinocyte cell lines were generated by transduction with retroviruses containing HPV-E6/E7 oncogenes. SCC25 cell lines were originally isolated from oral squamous cell carcinomas, generously provided by Dr. James Rheinwald (Department of Dermatology, Brigham and Women's Hospital and Harvard Skin Disease Research Center, Boston, MA). Control keratinocytes and fibroblasts were cultured under standard conditions in Keratinocyte Growth Medium (Gibco) and DMEM (Gibco) with 10% FCS, respectively.

3D cocultures were set up as follows: rat tail collagen I (BD Biosciences) and matrigel (BD Biosciences) were mixed to a final concentration of 4.0 mg/mL and 1.35 mg/mL, respectively. After neutralizing with 1 mol/L NaOH, 1 × 10⁶ fibroblasts/mL were added, and 1.0 mL of this mixture was pipetted into each 12-well insert (BD sciences). Next day, 5 × 10⁵ RDEB-SCCK were added onto each gel, and the cultures were raised to the air-liquid interface. Ascorbic acid (50 μg/mL) and the following concentrations of inhibitors were added freshly to the culture medium, every other day: SB505124 (10 μmol/L); PF573228 (5 μmol/L); β-aminopropionitrile (BAPN; 600 μmol/L); and LY294002 (20 μmol/L). Each experiment was repeated at least three times in duplicates. The cocultures were harvested at 21 days, fixed in 4% paraformaldehyde and embedded in paraffin, and analyzed for invasion by hematoxylin and eosin (H&E) staining. The invasion index, presented as the average values of (1 − [non-invading area/total area]), was calculated by quantifying the total area over which RDEB-SCCK cells had dispersed (including invading and non-invading cells) and the area of non-invading cells.

Additional materials and methods are described in Supplementary Methods.

Due to the loss of physiologic adhesion provided by C7, RDEB skin undergoes cycles of mechanically induced blistering, healing with scarring and, ultimately, develops cSCC with poor outcome. Morphologic hallmarks of human RDEB skin are dermal–epidermal microblistering, dense matrix, and inflammatory infiltrates in the dermis (Fig. 1A). H&E staining of early lesions suspected of RDEB-cSCC revealed invading epithelial cell columns and tumor cell islands in the upper dermis, accompanied by inflammatory infiltrates (arrows, Fig. 1A). Immunofluorescence staining confirmed complete lack of C7 in RDEB skin and cSCC used for this study (Fig. 1B). Earlier data from our laboratory have demonstrated that loss of C7 increases expression of TGFβ (13, 14) that has been widely implicated in inflammation, fibrosis, and cancer progression. In order to identify the TGFβ-expressing cell types, we performed double immunofluorescence staining for TGFβ and vimentin, a marker specific to fibroblasts. High levels of TGFβ expression were observed in RDEB skin and RDEB-cSCC fibroblasts (Fig. 1C), which was in accordance with previous findings (6, 16, 17). The staining pattern also suggested that TGFβ was highly expressed by inflammatory cells. The TGFβ signaling activity in both RDEB skin and RDEB-cSCC was confirmed by the elevated expression of pSMAD2/3 as compared with normal human skin (Fig. 1D). The staining also suggested slightly elevated levels in RDEB skin epidermis. Therefore, next we determined the levels of TGFβ expression in keratinocytes isolated from RDEB and non-RDEB skin and SCC. In accordance with previous findings (14), qPCR analysis revealed that RDEB keratinocytes expressed slightly but not significantly more TGFB1 than normal human keratinocytes. Similarly, TGFB1 transcripts were slightly but not significantly upregulated in RDEB-cSCC as compared with the non–RDEB-SCCK, indicating limited contribution of keratinocyte-derived TGFB1 expression to differences in tumor progression (Supplementary Fig. S1). In order to determine the extent of fibrotic remodeling in the dermal microenvironment, collagen I and picrosirius red stainings were carried out. Both RDEB skin and RDEB-cSCC not only contained excessive collagen I and dense collagen fibrils (Fig. 1E and F), but also activated (myo) fibroblasts (Fig. 1G). These results demonstrate that RDEB skin not yet bearing tumors exhibits fibrotic changes of the dermal microenvironment similar to a well-developed solid tumor stroma that supports tumor progression. Importantly, the results implicate injury-driven tumor-unrelated dermal changes as a determiner of the aggressive nature of RDEB-cSCC.

To corroborate our hypothesis that the unusual aggressiveness of cSCC in RDEB depends on injury-induced changes in RDEB skin, we subjected RDEB mice to two-stage chemical carcinogenesis. This procedure generates homogeneous mutations primarily in the Hras1 gene followed by repeated cycles of injury (15). Five-week-old wild-type and RDEB mouse littermates were topically treated with DMBA followed by twice-weekly promotion with TPA for 14 weeks (Fig. 2A). In the wild-type mice, the onset of tumors commenced at week 6 of TPA promotion and reached 89% by week 9, whereas the onset of tumors in the RDEB mice was delayed until week 8 and reached 93% by week 11 of TPA promotion (Fig. 2B). At the end of the treatment, the majority of wild-type mice showed multiple benign papillomas (15) with an average of 2.27 ± 1.74 tumors per mouse (Fig. 2C and D). In contrast, RDEB mice displayed erythematous flat papules with an overlying scale/crust (average 1.07 ± 0.47 tumors per mouse; Fig. 2C and E) that were strongly reminiscent of RDEB-cSCC in patients (Fig. 2F). Careful histologic analysis showed that 80% of the lesions in RDEB mice were invasive (Fig. 2H and I). To assess the tumor cells undergoing epithelial-to-mesenchymal transition (EMT), one of the hallmarks of cancer progression, tumor sections were stained for E-cadherin. The expression of E-cadherin, an epithelial cell adhesion marker, was progressively lost in RDEB tumors, as compared with wild-type tumors, indicating the loss of epithelial identity (Fig. 2K). Further, RDEB tumors displayed increased tumor cell proliferation and reduced rate of apoptotic cell death compared with wild-type tumors (data not shown). In addition to invasive cSCC, one RDEB mouse had lung metastasis. Although not significant, this finding supports the observation of increased aggressiveness of the induced RDEB mouse tumors (Fig. 2M; Supplementary Fig. S2). Taken together, chemical carcinogenesis on RDEB mouse skin leads to tumors that phenocopy naturally occurring RDEB-cSCC in patients and can serve as models to assess the mechanisms underlying tumor progression.

The finding that chemically induced tumors in RDEB mice differed from wild-type papillomas clearly supported key involvement of the dermis in RDEB-cSCC. Studies on human tissues (Fig. 1) and our previous work (13, 14) implicated altered TGFβ activity as a major determinant of the disparate tumor-influencing properties of wild-type and RDEB mouse dermis. Indeed, immunostaining revealed that both the level of TGFβ and its downstream activity, as measured by pSMAD2/3, were highly increased in 7-week DMBA/TPA-treated murine RDEB skin, hereafter referred to as tumor-primed skin (Fig. 3). Similar to human RDEB skin (Fig. 1C), double immunofluorescence staining showed colocalization of vimentin and TGFβ in the fibroblasts in both tumor-primed RDEB skin and tumors (Fig. 3A). A slight decrease in TGFβ levels and pSMAD2/3 activity was observed in immediate RDEB-cSCC tumor microenvironment as compared with tumor-primed RDEB skin not yet carrying tumors (Fig. 3A and B). These findings were validated by qPCR and Western blot analyses (Fig. 3A and C). The presence of heightened TGFβ signaling and the lower number of tumors in RDEB mice reflect the paradoxical effect TGFβ has on tumor initiation and growth (18).

Next, we assessed the effects of TGFβ activity on the biomechanical properties of the ECM in tumor-primed murine RDEB skin and showed that they were similar to those observed in mechanically injured human RDEB skin. Significantly increased collagen I deposition was observed in tumor-primed murine skin and cSCC (Fig. 4A). To validate the increase as TGFβ-related, murine RDEB fibroblasts (RDEBF) were cultured in the presence or absence of the selective TGFβ type 1 receptor inhibitor SB505124 (Supplementary Fig. S3A). The substantial reduction of collagen I synthesis observed in the presence of SB505124 demonstrated that TGFβ regulated collagen I expression in RDEBF.

In line with this, picrosirius red staining revealed significantly increased collagen fibril thickness in murine RDEB dermis as compared with controls (Fig. 4B). High fibril thickness/rigidity can be associated with collagen cross-linking (10, 19) and, indeed, strongly elevated levels of the collagen cross-linking enzyme, lysyl oxidase (LOX), were found in the ECM of tumor-primed murine and injured human RDEB skin and tumors (Fig. 4C; Supplementary Fig. S4). LOX is expressed by both keratinocytes and fibroblasts, and is essential for the normal development of the skin (20, 21). In order to identify LOX-expressing cell types, costaining with antibodies to cytokeratins (keratinocyte marker) and vimentin (fibroblast marker) was carried out. LOX was present in both keratinocytes and fibroblasts. However, its expression was strongly and specifically increased in fibroblasts in tumor-primed murine and injured human RDEB skin and tumors (Fig. 4C; Supplementary Fig. S4) and also in the metastatic region of RDEB lung (Supplementary Fig. S2). Further, to assess the influence of TGFβ on LOX expression, murine wild-type and RDEB keratinocytes and fibroblasts were cultured in presence of recombinant TGFβ or TGFβ type I receptor inhibitor SB505124. Western blot analysis showed that whereas LOX expression in keratinocytes was largely unresponsive to TGFβ stimuli, TGFβ greatly governed LOX expression in fibroblasts (Supplementary Fig. S3B). Further, the analysis demonstrated that presence or absence of C7 did not affect the ability to respond to TGFβ (Supplementary Fig. S3B). Stimulation with TGFβ increased both the proenzyme and mature LOX, whereas the inhibitor attenuated both (Supplementary Fig. S3B). In accordance with the fact that both the TGFβ expression and stiff matrix in RDEB skin facilitate a phenotypic switch of fibroblasts to αSMA-positive myofibroblasts (22), αSMA-positive cells were more prominent in tumor-primed murine RDEB skin than in controls (Supplementary Fig. S5A). Thus, the unrestrained TGFβ activity in C7-deficient RDEB skin concomitantly drives expression of collagen I and LOX, resulting in a stiffer proinvasive dermal matrix.

The effects of biomechanically altered ECM on RDEB-cSCC cell signaling were first assessed with immunofluorescence staining using antibodies to β1 integrin and focal adhesion kinase (FAK). Enhanced levels of β1 integrin and activated pFAK^tyr397 were observed in RDEB tumors (Fig. 4D; Supplementary Fig. S5B and S5C), specifically colocalized at the surface of invading RDEB-cSCC cells. This, in combination with increased active pAKT^ser473 in the tumors (Supplementary Fig. S5B and S5C), indicated focal adhesion/stiff ECM-mediated mechanosignaling. The fact that treatment of human RDEB-SCCK with PF573228, an inhibitor of pFAK^tyr397 autophosphorylation, reduced pAKT^ser473 validated the focal adhesion-mediated pAKT signaling axis (Fig. 4E).

Taken together, these data conclusively demonstrate an invasion-permissive chain of events in RDEB skin: loss of C7 and repeated mechanical or chemical injury lead to unrestrained TGFβ activity; this results in gradual stiffening of the dermal matrix; the stiff microenvironment triggers integrin β1–mediated mechanosignaling in cSCC cells, which promotes malignant progression.

3D organotypic cocultures (23) were used to evaluate cell–matrix interactions relevant for RDEB-cSCC progression and to assess possibilities to modify the process. Cocultures constructed with human RDEB-SCCK as epithelial cells and normal human fibroblasts (NHF) formed a skin-like bilayered structure, whereas RDEBF promoted invasion of RDEB-SCCK (Fig. 5A). Significant increase of collagen fibril density, as indicated by picrosirius red staining, was evident in RDEBF-containing cocultures (Fig. 5B). Atomic force microscopy (AFM) demonstrated that 3D collagen matrigels populated with RDEBF were significantly stiffer and had about 3-fold higher Young's modulus than 3D collagen matrigels constructed with NHF (Fig. 5C). Treatment of these matrices with SB505124 reduced stiffness substantially (Fig. 5C).

The picrosirius red staining intensity was correlated with stiffness values obtained by AFM. A clear positive correlation was observed between the median values of Young's modulus of 3D matrices and median values of area intensity of picrosirius red stainings obtained from their respective 3D organotypic cocultures (Fig. 5D). Hence, in this system, picrosirius red staining intensity directly corresponded with stiffness and could be used as a read-out for it.

With the perspective of identifying therapeutic targets to prevent or at least delay RDEB-cSCC, different strategies were used to inhibit matrix stiffness, mechanosignaling, and tumor cell invasion in 3D organotypic cultures. The first strategy targeted early events in RDEB fibrosis by inhibiting TGFβ signaling with SB505124, which efficiently reduced matrix stiffening and invasion (Fig. 5A, B, E, and F). The second strategy targeted mid events using BAPN, an irreversible LOX inhibitor. This treatment reduced collagen fibril thickness and invasion (Fig. 5A, B, E, and F). To inhibit invasion at later stages, i.e., in extensively fibrotic stiff dermis, we targeted stress sensing in tumor cells and used the ATP-competitive FAK inhibitor PF573228 and the phosphoinositide 3-kinase inhibitor LY294002. These compounds blocked invasion without altering dermal stiffness (Fig. 5A, B, E, and F). Taken together, depending on the extent of injury-driven dermal fibrosis, RDEB-cSCC progression can be combated by inhibiting specific targets along the tissue stiffness–promoting and responding pathway.

Almost every adult patient with severe generalized RDEB will develop aggressive cSCC, a leading cause of premature death in this disorder (4). Except for surgical removal, there is no therapy for RDEB-cSCC; development of treatments has been hampered by lack of understanding of the underlying cellular and molecular events. Here, we delineated the contributing mechanisms (Fig. 6) and generated new knowledge that facilitates design of prophylactic and therapeutic measures for delaying tumor progression and extending cancer-free periods in RDEB.

The triggers of malignant transformation in the epidermis were not in our focus, because non-RDEB and RDEB-cSCC display the same nature of genetic changes, apart from the individual COL7A1 mutations (8). For this reason, tumors were induced by DMBA treatment, which generates a rather homogeneous oncogenic change, i.e., the Hras1 mutation c.182A>T, instead of other methods like exposure to UV light, which results in heterogeneous genetic changes (15).

The observation that RDEB mice developed fewer but more aggressive tumors implicated TGFβ activity as a major regulatory factor (18). TGFβ displays a paradoxical role on tumor growth. During initial tumorigenesis, it inhibits proliferation of keratinocytes, but later a switch occurs from a tumor suppressor to an enhancer (24); evidence of this in our study can be seen by the later appearance of tumors in RDEB mice. Similarly to RDEB mice, mice overexpressing TGFβ in the skin are resistant to formation of papillomas upon chemical carcinogenesis, but display enhanced carcinoma progression (24). Epidermal TGFβ signaling certainly contributes to RDEB-cSCC and accelerates initial invasion, along with a weakened basement membrane in RDEB skin, as indicated by our previous work (14). However, the present analyses indicate primarily dermal contribution of TGFβ in cSCC progression. Further, we recently showed that, in addition to TGFβ, RDEB mice display elevated levels of inflammatory cytokines, such as TNFα and IL6, in injured tissue and in circulation. This unrestrained inflammatory response contributes to progression of fibrosis in RDEB (17). Thus, inflammation is a crucial driver of the microenvironmental remodeling promoting RDEB-cSCC progression. Here, we focused on extracellular dermal mechanisms independent of tumor cells and showed that abundant TGF-β, increased ECM production, thicker dermal fibrils, and higher matrix stiffness were key features in tumor-primed murine and mechanically injured human RDEB skin.

Individual extracellular molecules can influence carcinoma progression (25). However, the present study suggests that not individual molecules, but the dermal matrix as a 3D scaffold, with its specific biophysical properties, offers an enabling environment for cSCC progression in RDEB. This is supported by two observations: (i) the gradual stiffening of chemically injured murine RDEB dermis promoted tumor progression; and (ii) treatments that limited mechanosensing or softened RDEBF containing organotypic cultures without affecting protein expression hindered invasion by RDEB-cSCC cells (Fig. 5).

The DMBA/TPA-induced tumors on RDEB mice, in contrast with those in wild-type mice, displayed hallmarks of cancer, i.e., increased proliferation, lower cell death, signs of EMT, invasion, and ability to metastasize. In vitro and in vivo analyses showed that these events were mediated through the integrin-FAK-AKT mechanosensing pathway. Our observations are in line with the growing awareness that biophysical characteristics of tumor-generated stroma influence different cancers. The tissue/tumor constellations are variable and context-dependent, but a stiff environment has been reported to lower expression of tumor suppressors, increase telomerase activity, promote cell-cycle progression, lead to apoptotic resistance, promote malignant transformation by enhancing TGFβ-induced EMT, and increase invasion and metastasis (11). However, it is very clear that the influence of the ECM on cancer development must be delineated specifically for each tissue/carcinoma constellation (11). Here, we show that RDEB-cSCC fate is profoundly affected by the ECM, but not via tumor-conditioned stroma. Instead, injury-driven cancer-independent biophysical alterations of the dermis are pivotal in this context.

The emerging concept of repeated injury, unrestrained fibroproliferative healing, and subsequent tissue stiffening as enablers of aggressive carcinoma behavior have high relevance for other pathologies, too, including burn scar neoplasms and SCC complicating common leg ulcers (26). However, tissue stiffening alone is not sufficient to drive carcinoma. For example, progressive systemic sclerosis, a multiorgan autoimmune disease, exhibits endogenously stiffened skin but not cSCC proneness. In contrast, systemic sclerosis patients who have pulmonary fibrosis, which is injury-related, have an increased risk of lung cancer (27, 28). Hence, our observations and published data from other investigators collectively argue that injury-driven changes in the biophysical performance of the mesenchyme are a general determinant of carcinoma progression (Fig. 6).

Causative treatment for RDEB will offer ultimate protection against aggressive cSCC, but in spite of progress, clinical implementation of such therapies is still in distant future (5). Our study offers promise for a disease-modulating, cSCC-limiting therapy. Three major mechanisms were identified as potential druggable targets, namely excessive TGFβ activity, tissue stiffening, and dysregulated mechanosensing. It is important to state that the drugs BAPN and LY294002 used to target LOX and PI3K, respectively, cannot be clinically used due to dermal toxicity and adverse effects (29, 30). However, several more clinically relevant FDA/European Medicines Agency–approved drugs already exist for the identified targets, or candidate drugs are in advanced stages of clinical trials. Examples are losartan or neutralizing TGFβ antibodies to target initial fibrotic processes, LOX antibodies to reduce crosslinking, and small molecules inhibiting FAK activity to correct mechanosensing in tumor cells and keratinocytes (11, 31). Such drugs could be repurposed for rapid clinical implementation as prophylactic cSCC therapy in RDEB and other related, more common conditions.

No potential conflicts of interest were disclosed.

Conception and design: V.R. Mittapalli, A. Nyström, L. Bruckner-Tuderman

Development of methodology: V.R. Mittapalli, J. Madl, S. Löffek, W. Römer

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): V.R. Mittapalli, J. Madl, D. Kiritsi, J.S. Kern

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): V.R. Mittapalli, J. Madl, D. Kiritsi, J.S. Kern, W. Römer, A. Nyström, L. Bruckner-Tuderman

Writing, review, and/or revision of the manuscript: V.R. Mittapalli, S. Löffek, D. Kiritsi, A. Nyström, L. Bruckner-Tuderman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W. Römer

Study supervision: L. Bruckner-Tuderman

The authors are grateful to Drs. Anja Fritsch, Claus-Werner Franzke, and Yinghong He for helpful discussions, and Sylke Lange and Margit Schubert for excellent technical assistance.

This work was supported in part by the German Excellence Initiative (EXC 294), by grants from the German Research Foundation, DFG, to L. Bruckner-Tuderman and D. Kiritsi (numbers BR1475/12-1, KI1795/1-1), Debra International and the German Federal Ministry for Education and Research, BMBF, under the frame of Erare-2, the ERA-Net for Research on Rare Diseases (SpliceEB) to A. Nyström. W. Römer acknowledges the support from the Ministry of Science, Research and the Arts of Baden-Württemberg (Az: 33-7532.20) and the starting grant of the European Research Council (Programme “Ideas” – call identifier: ERC-2011-StG 282105).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.