Molecular Mechanisms of Melanoma Metastasis

(1)

503

Molecular Mechanisms of Melanoma Metastasis

Mohammed Kashani-Sabet

C

ONTENTS

INTRODUCTION

BIOLOGY OF TUMOR METASTASIS

SYSTEMIC GENE DELIVERY IN THE IDENTIFICATION OF MELANOMA PROGRESSION GENES

SYSTEMIC DELIVERY OF RIBOZYMES TO TUMOR-BEARING MICE

VASCULAR FACTORS IN THE PROGRESSION OF CUTANEOUS MELANOMA

MOLECULAR ANALYSIS OF THE TUMOR VASCULATURE IN PRIMARY MELANOMA

CONCLUSIONS AND PERSPECTIVES

REFERENCES

28

Summary

The recent development of powerful and efficient systemic gene delivery techniques has allowed the identification of a number of genes and pathways critical to the progression of melanoma in murine models. Using this approach, the nuclear factor-NB (NF-NB)-signaling pathway has been identified as playing an important role in the metastatic progression of melanoma. Moreover, recent prognostic analyses using a large cohort of patients from the University of California at San Francisco Melanoma Center database have identified interactions between tumor cells and the tumor vasculature that play a dominant role in the survival associated with melanoma. Translational studies using tissue microarrays have identified expression of the p65 subunit of NF-NB to be tightly correlated with the development of tumor vascularity and vascular invasion. Finally, recent studies on tumor lymphangiogenesis have clarified the nature of the relevant vasculature in the progression of cutaneous melanoma. This chapter will discuss the recent understanding of the molecular mechanisms and pathways underlying the development of traditional and novel histological prognostic markers in melanoma.

Key Words: Melanoma; metastasis; ribozymes; vascularity.

INTRODUCTION

Malignant melanoma is a disease with an unpredictable clinical behavior. Although patients with thin tumors have an excellent 5-yr survival based on the most recent analysis performed by the American Joint Committee on Cancer (AJCC) staging system for melanoma (1), a small but clinically important subset of these patients with thin

From: From Melanocytes to Melanoma: The Progression to Malignancy Edited by: V. J. Hearing and S. P. L. Leong © Humana Press Inc., Totowa, NJ

(2)

tumors relapse and die as a result of their melanoma. Conversely, a substantial pro- portion of patients with thick tumors are free of metastasis. Thus, the development of prognostic markers that provide important prognostic information regarding the risk of relapse and death independent of the tumor thickness represents an important area of melanoma research. In addition, understanding the molecular basis of these pathological prognostic factors is crucial to the identification of molecular markers of melanoma outcome that may represent important targets for novel therapies for advanced melanoma.

BIOLOGY OF TUMOR METASTASIS

Based on recent prognostic analyses of melanoma, lymph node metastasis is well- recognized as one of the most important prognostic factors for predicting survival in patients with cutaneous melanoma (1,2). The identification of the pathological events and underlying molecular factors that are crucial for the development of both lymph node and distant metastasis would improve our risk assessment of melanoma patients, resulting in potentially different algorithms for identifying patients who may benefit from sentinel lymph node dissection or systemic adjuvant therapies.

Recent studies in murine models have shed light on the biology of tumor metastasis.

Metastasis is a highly complicated biological process that has been broken down into a number of different steps: detachment from the primary tumor, migration, invasion into lymphatic or blood vessels (intravasation), survival in the lymphatic or blood circula- tion, egress from the lymphatic or blood vessels (extravasation), angiogenesis, and cell proliferation (3). Many of these steps require the interaction of the tumor cell with the extracellular matrix or with vascular endothelial cells. As a result, the identification of molecular factors that play a role in these interactions would greatly improve our understanding of steps crucial to the metastatic cascade. Intriguingly, recent studies have revealed that distal steps in the cascade (i.e., those following extravasation) are crucial to the development of clinically apparent metastases (4). Thus, the important factors that result in the success or failure of tumor metastasis appear to occur at the level of metastatic progression. Although much of cancer research has been devoted to the identification of genes involved in tumor initiation (such as activating mutations in oncogenes and losses in tumor suppressor genes), until recently, little attention has been paid to the identification of genes that drive metastatic progression and the onset of lethal tumor metastasis. Understanding genes and pathways that promote the distal steps of the metastatic cascade may be important in the identification of novel targeted therapies for lethal melanoma metastasis.

SYSTEMIC GENE DELIVERY IN THE IDENTIFICATION OF MELANOMA PROGRESSION GENES

Recent advances in gene delivery have demonstrated the utility of cationic liposome–

DNA complexes (CLDCs) in the identification of genes and pathways crucial to the metastatic progression of melanoma. An extensive discussion of gene delivery is beyond the scope of this chapter; however, a brief review of systemic gene transfer techniques will be undertaken. For systemic gene therapy to be successful, the genetic construct of interest (comprising the complementary [c]DNA of a gene with tumor suppressor activity, or an anti-gene, such as antisense, ribozyme, or small interfering RNA [siRNA], targeting a tumor-promoting gene) must be delivered systemically and expressed effi-

(3)

ciently and for prolonged periods in the tissues to which the melanoma metastasizes.

First-generation studies used viral methods, such as retroviruses and adenoviruses, to effect gene transfer. However, these approaches have significant limitations (5,6), including the development of neutralizing immune responses that restrict the repeated administrations of the antitumor genetic agent required to produce biologically and clinically relevant antitumor effects. As a result, nonviral gene transfer approaches are of interest. Given the net negative charge of DNA, cationic liposomes or polymers have been extensively evaluated for their ability to deliver different cDNAs in several model systems (7). However, first-generation CLDCs were limited by a short duration of expression of the delivered gene. These approaches resulted in peak expression at 24 h, which quickly plateaued and then reached background levels by 1 wk after systemic injection of CLDCs (8).

The recent development of a novel expression plasmid using elements from the Epstein-Barr Virus (EBV) genome has resulted in significant improvements in the du- ration of transferred gene expression (8). This expression plasmid includes the family of repeats and the Epstein-Barr nuclear antigen-1 gene, which encodes a protein that binds site specifically to sequences within the family of repeats. Incorporation of these elements into the expression plasmid results in enhanced retention of the plasmid in the nucleus, without the ability to replicate, thereby circumventing potential safety con- cerns. In initial studies, a single injection of CLDC driven by the human cytomegalovirus promoter and encoding the human granulocyte colony-stimulating factor cDNA resulted in therapeutic levels of granulocyte colony-stimulating factor expression that translated into granulocytosis and an elevated total white blood cell count. Moreover, this EBV- based vector could be repeatedly injected into immunocompetent mice, each time pro- ducing high levels of delivered gene expression (8).

Given the development of an efficient and durable systemic nonviral gene delivery system, the utility of this system was examined in mouse models of tumor metastasis. In early studies, this CLDC-based approach was used to identify several genes whose expression produced significant antimetastatic activity in the B16 model of melanoma metastatic progression (9). Thus, a single injection of CLDC containing the p53, GM- CSF, angiostatin, and CC3 genes each produced significant antimetastatic activity in the C57Bl/6 mice bearing metastatic B16-F10 lung tumors (9). Each of these genes was shown to mediate its antimetastatic effects through potent inhibition of angiogenesis of lung metastases. Intriguingly, CLDC-based intravenous delivery of the human wild- type p53 gene transfected nearly 80% of metastatic tumor cells in the lung, thereby demonstrating the efficiency of this systemic gene delivery approach.

SYSTEMIC DELIVERY OF RIBOZYMES TO TUMOR-BEARING MICE

More recently, this CLDC approach has been used to deliver various antigene approaches, including antisense nucleotides, siRNA, and ribozymes. A detailed review of these approaches is beyond the scope of this chapter, and the reader is directed to several recent reviews of these modalities (10,11). To date, the majority of studies of this CLDC-based approach have used hammerhead ribozymes. Ribozymes are catalytic RNAs that can cleave a specific RNA sequence (10). Ribozymes have demonstrated efficacy in the suppression of gene expression and in the identification of gene function (12). The development of ribozymes as gene therapy agents has been slowed by the development of an efficient gene delivery approach.

(4)

Recent studies have examined, in detail, the utility of CLDC-based delivery of hammerhead ribozymes in the identification of melanoma progression genes and as a rational strategy for systemic gene therapy of melanoma metastasis. Our studies have defined the time course of ribozyme expression and suppression of target gene expression when using this approach. Thus, CLDC encoding a ribozyme targeting murine telomerase RNA was injected on days 3 and 10 after intravenous inoculation of B16 melanoma cells into C57Bl/6 mice. Ribozyme levels in the lung were serially assayed by quantitative real-time polymerase chain reaction. Peak ribozyme expression was observed on day 17, and detectable levels persisted until day 21 (13). This level of ribozyme expression resulted in significant suppression of murine telomerase RNA levels (as determined by the TaqMan assay) and in the enzymatic activity of telomerase (as determined by the telomeric repeat amplification protocol). These results showed that the novel EBV- based expression plasmid could also produce durable levels of ribozyme expression in tumor-bearing mice, resulting in profound suppression of target gene expression levels and functional activity.

Further studies examined the ability of this systemic CLDC-based ribozyme delivery approach to suppress the metastatic progression of melanoma in tumor-bearing mice.

The earliest study used ribozymes to target the nuclear factor-NB (NF-NB)-signaling pathway, given its importance to tumor metastasis (14). A single injection of CLDC, encoding a hammerhead ribozyme targeting either the p65 or p50 subunit of NF-NB, 7 d after tumor cell inoculation produced significant effects on the metastatic progression of B16-F10 cells to the lung (15). Intriguingly, both the anti-p65 and anti-p50 ribozymes were shown to be superior in their effects on the development of B16 lung metastases than a disabled ribozyme containing a single-base mutation in the catalytic core and, thus, devoid of catalytic activity. These studies strongly suggested the cleavage of target transcripts as the dominant mechanism by which ribozymes mediate antitumor activity when delivered by this CLDC-based approach.

Additional studies performed using the anti-NF-NB/p65 ribozyme shed light on the cell types and mechanisms targeted by downregulation of NF-NB in tumor-bearing animals. Prior studies showed metastatic tumor cells in the lung and vascular endothelial cells to be the cell types most efficiently transfected by this CLDC-based approach (16).

Thus, several cell types in the lung were examined for levels of NF-NB/p65 protein expression by immunohistochemistry (15). These studies revealed that p65 expression was significantly reduced in four cell types in the lung, including metastatic tumor cells, vascular endothelial cells, as well as alveolar and bronchial lining cells. However, the greatest level of suppression was seen in tumor cells and vascular endothelial cells, suggesting that the cell types with the most efficient levels of transfection are those that are most susceptible to ribozyme action. Moreover, injection of stable B16 transformants expressing the anti-p65 ribozyme into syngeneic mice resulted in significantly reduced metastatic capacity, suggesting a correlation between level of melanoma cell NF-NB/

p65 protein expression and metastatic potential (15). In addition, several different func- tional pathways critical to the metastatic phenotype were examined, including apoptosis, angiogenesis, cell cycle, and tumor cell invasion, each of which has been shown to be regulated by NF-NB. Anti-NF-NB/p65 ribozyme action showed no significant effects on the apoptotic and mitotic indices of metastatic tumor cells, and had no effect on tumor angiogenesis in the lung. However, the ability of B16 cells to invade into matrigel and in the Boyden chamber assay was dramatically reduced by ribozyme expression. These

(5)

results suggested that NF-NB mediates its prometastatic effects in melanoma primarily through effects on tumor invasion, rather than through apoptosis, mitosis, or angiogenesis.

Additional studies have used CLDC-based delivery to identify the genes regulated by NF-NB that could mediate its effects on tumor invasion. Initial studies focused on integrin E3 and platelet endothelial cell-adhesion molecule-1, which form a ligand–receptor pair linked to cell adhesion and invasion (17), and are known to be regulated by NF-NB.

Systemic administration of ribozymes targeting integrin E3 and platelet endothelial cell- adhesion molecule-1 also significantly suppressed the metastatic progression of B16 cells (15), suggesting that this ligand–receptor pair may help mediate the effects of NF-NB on tumor invasion and metastasis in tumor cells and/or vascular endothelial cells.

Subsequently, gene expression profiling was used to identify additional genes involved in NF-NB-mediated effects on tumor invasion. cDNA microarray analysis of the B16 clone stably expressing the anti-p65 ribozyme compared with the control transformant clone demonstrated the upregulation of FKBPr38, a previously uncharacter- ized member of the FK-506 binding protein (FKBP) gene family, in ribozyme-expressing cells. We then used systemic gene transfer to assign novel anti-invasive and antimetastatic functions to FKBPr38, and subsequently to FKBP12 (18). cDNA microarray analysis was then combined with FKBP gene transfer to show that FKBP gene expression coordinately induced the anti-invasive syndecan-1 gene and suppressed the proinvasive MMP-9 gene. Conversely, suppression of FKBP12 expression by anti- FKBP12 siRNA treatment significantly reduced cellular levels of syndecan-1 protein, whereas it increased matrix metalloproteinase (MMP)-9 levels and tumor invasiveness.

Taken together, these studies have identified a network of adhesion- and matrix-remod- eling genes that appear to be responsible for the ability of the NF-NB pathway to control tumor invasion and metastasis.

VASCULAR FACTORS IN THE PROGRESSION OF CUTANEOUS MELANOMA

Recent prognostic analyses from the University of California at San Francisco Mela- noma Center database have identified novel histological prognostic factors that appear to play an important role in the progression of cutaneous melanoma. Our initial study (19) examined the role of vascular invasion as a prognostic marker. Vascular invasion is a known marker of poor outcome in a number of different malignancies, but has not been traditionally recognized as an independent marker of survival in melanoma. As a result, presence or absence of vascular invasion is not routinely recorded in most melanoma pathology reports. We examined the role of vascular invasion in a database of 526 patients with primary cutaneous melanoma who had 2 yr of follow-up or documented first relapse. In this data set, 78 patients were identified with two patterns of vascular involvement: definite vascular invasion, with melanoma cells identified within vascular spaces, and incipient vascular invasion, with tumor cells abutting the tumor vasculature, without evidence of definite invasion. The presence of either type of vascular involvement significantly increased the risk of relapse and death, and reduced the survival associated with melanoma. Intriguingly, vascular involvement increased the risk of both regional and distant metastasis. The impact of vascular involvement on these outcomes was similar to that of ulceration. Furthermore, the risk of metastasis and death from tumors exhibiting incipient invasion was as high as those exhibiting definite vascular invasion, suggesting the poor prognosis associated with this finding. By multivariate

(6)

analysis, vascular involvement was an independent predictor of relapse-free and overall survival. There was no difference between the two patterns of vascular involvement with respect to overall relapse or death rate, and rate of regional nodal or distant metastasis.

However, the relapse-free survival of cases with uncertain vascular invasion was significantly lower than those with definite invasion. This observation suggested that the group with incipient vascular invasion occupies an important earlier step in tumor progression.

In addition, we evaluated the impact of morphological patterns of tumor vascularity in the outcome associated with melanoma (20). Four patterns of tumor vascularity were recorded: absent, sparse, moderate, and prominent. By univariate analysis, increasing tumor vascularity was associated with a significantly increased risk of relapse (consist- ing of both regional and distant metastasis) and death associated with melanoma. In addition, increased vascularity correlated with significantly reduced relapse-free and overall survival. By multivariate analysis, tumor vascularity was an independent predictor of relapse-free and overall survival. Intriguingly, the prevalence of ulceration increased with increased vascularity, such that ulceration was present in almost 50% of cases with the prominent pattern of tumor vascularity.

Finally, we analyzed the interactions between these vascular factors in the progres- sion of cutaneous melanoma (21). To begin with, Cox regression analysis of factors evaluated by the AJCC melanoma staging committee reproduced the powerful impact of tumor thickness and ulceration in the University of California at San Francisco data set, thereby establishing comparability between the two databases and analyses. Thus, when the six factors used by the AJCC were incorporated into the model, tumor thickness and ulceration emerged as the most powerful prognostic factors, with risk ratios strik- ingly similar to those demonstrated by the AJCC analysis. Intriguingly, when vascular involvement and tumor vascularity were added to the model, they emerged as the most powerful predictors of melanoma survival by multivariate analysis (21). With the inclu- sion of these two factors in the model, ulceration was no longer an independent predictor of overall survival. Finally, increasing tumor vascularity was shown to predispose to the development of vascular invasion, and the presence of both of these factors increased the prevalence of ulceration. These results suggested that melanomas invade more commonly into newly formed vessels than pre-existing ones, and that ulceration may reflect the interactions between the tumor and the tumor vasculature, as evidenced by tumor vascularity and vascular involvement.

Based on these observations, we have developed a model of melanoma progression that takes into account the dominant role played by these vascular factors in the progression of cutaneous melanoma and in the development of ulceration, which has now been incorporated into the AJCC staging classification for melanoma (1). In this model, tumor vascularity represents the initial vascular factor that is necessary for the development of melanoma metastases. The presence of increased vascularity increases the likelihood (and potentially enables the development) of both vascular invasion and ulceration.

Although tumors can metastasize with the presence of increased tumor vascularity alone, the presence of either vascular invasion or ulceration (or both) greatly increases the likelihood of metastatic spread.

(7)

MOLECULAR ANALYSIS OF THE TUMOR VASCULATURE IN PRIMARY MELANOMA

Given the powerful role played by these vascular pathways in melanoma progression, determining the molecular events that underlie the development of these factors would potentially result in the identification of novel molecular prognostic factors for melanoma. It is intriguing to note that melanoma cell invasiveness was the dominant mechanism identified by our gene targeting studies in mice, as well as the dominant prognostic factor in patients with cutaneous melanoma. Thus, we hypothesized that the NF-NB pathway, which drove tumor metastasis in murine models by virtue of effects on tumor invasion, could also be involved in mediating the development of tumor invasiveness in human melanoma specimens. To examine this possible association further, we performed a matched-pair analysis of human primary melanomas using tissue microarray technology. We identified 24 cases with documented vascular involvement and increased tumor vascularity and 24 controls with no vascular involvement and limited tumor vascularity but matched for a number of other known melanoma prognostic factors, including tumor thickness, gender, age, anatomical location, histogenetic subtype, and Clark level. The pairs were not matched for ulceration, given that the presence of vascular factors would automatically create an imbalance in the presence of ulceration.

Immunohistochemical assay of the tissue array with an antibody targeting NF-NB/p65 showed higher levels of p65 protein in 17 cases with vascular factors present. In addition, the cases more commonly expressed higher levels of p65 expression than the controls (21). These results suggested a correlation between elevated expression of NF-NB/p65 and the presence of increased tumor vascularity and vascular involvement in human melanoma tissues. Given the powerful impact of these histological prognostic factors on melanoma survival, these studies also suggest that NF-NB may serve as a molecular marker of melanoma outcome.

With the important role of NF-NB in melanoma progression, the identification of downstream genes that may mediate the development of tumor vascularity would be of great interest. In this regard, recent studies examining the role of lymphangiogenesis in melanoma progression have yielded significant insights into the nature of the relevant tumor vasculature in melanoma. These studies have identified vascular endothelial growth factor (VEGF)-C and VEGF-D as potent lymphangiogenic agents, and lym- phatic endothelial hyaluronan receptor-1 as a specific marker for lymphatic vessels (22).

Dadras et al. (23) used lymphatic endothelial hyaluronan receptor-1 immunohistochem- istry on primary melanoma specimens to examine the correlation of lymphatic vessel density and melanoma metastasis to the regional lymph nodes. In a matched-pair study of 18 melanoma patients with lymph node metastasis and 19 patients without metastasis, a significantly higher number of intratumoral lymphatic vessels and a significant increase in peritumoral lymphatic vascular area was observed in metastatic vs nonmetastatic primary melanomas. In contrast, there was no difference in the blood vessel density (as determined by CD31 immunohistochemistry) between the matched pairs. By multivariate analysis, only peritumoral lymphatic vascular area was found to be an independent predictor of lymph node metastasis. Increasing lymphatic density was also associated with significantly reduced disease-free and overall survival. These results suggest lymphangiogenesis as a novel molecular prognostic marker for melanoma. The further identification of relevant lymphangiogenic factors in melanoma would be of great inter-

(8)

est. In this regard, it is intriguing to note that one candidate lymphangiogenic factor, VEGF-C, has been shown to be regulated by NF-NB (24), thereby further implicating the NF-NB pathway in the development of increased tumor vascularity by virtue of promoting lymphangiogenesis in primary cutaneous melanoma.

CONCLUSIONS AND PERSPECTIVES

In conclusion, recent studies have demonstrated the dominant impact of interactions between melanoma cells and the tumor vasculature in the progression of melanoma. As a result, tumor vascularity, vascular involvement, and tumor lymphangiogenesis have emerged as powerful prognostic factors for melanoma. These studies have clarified the contribution of lymphatic vs blood vessels in the metastatic cascade of primary melanoma. These studies have also shown that coordinate regulation of the NF-NB-signaling pathway plays an important role in mediating these effects, resulting in the lethal metastasis of melanoma.

REFERENCES

1. Balch CM, Buzaid AC, Soong SJ, et al. Final version of the American Joint Committee on Cancer staging system for cutaneous melanoma. J Clin Oncol 2001;19:3635–3648.

2. Gershenwald JE, Thompson W, Mansfield PF, et al. Multi-institutional melanoma lymphatic mapping experience: the prognostic value of sentinel lymph node status in 612 stage I or II melanoma patients.

J Clin Oncol 1999;17:976–983.

3. Pantel K, Brakenhoff RH. Dissecting the metastatic cascade. Nat Rev Cancer 2004;4:448–456.

4. Chambers AF, Groom AC, MacDonald IC. Dissemination and growth of cancer cells in metastatic sites. Nature Rev Cancer 2002;2:563–572.

5. Sandrin V, Russell SJ, Cossett FL. Targeting retroviral and lentiviral vectors. Curr Top Microbiol Immunol 2003;281:137–178.

6. Imperiale MJ, Kochaneck S. Adenovirus vectors: biology, design, and production. Curr Top Microbiol Immunol 2004;273:335–357.

7. Liu Y, Fong S, Debs RJ. Cationic liposome-mediated gene delivery in vivo. Methods Enzymol 2003;373:536–550.

8. Tu G, Kirchmaier AL, Liggitt D, et al. Non-replicating Epstein-Barr virus-based plasmids extend gene expression and can improve gene therapy in vivo. J Biol Chem 2000;275:30,408–30,416.

9. Liu Y, Thor A, Shtivelman E, et al. Systemic gene delivery expands the repertoire of effective anti- angiogenic agents. J Biol Chem 1999;274:13,338–13,344.

10. Bagheri S, Kashani-Sabet M. Ribozymes in the age of molecular therapeutics. Current Molecular Med 2004;4:489–506.

11. Scherer LJ, Rossi JJ. Approaches for the sequence-specific knockdown of mRNA. Nat Biotech 2003;21:1457–1465.

12. Kashani-Sabet M. Ribozyme therapeutics. J Invest Dermatol Symp Proc 2002;7:76–78.

13. Nosrati M, Li S, Bageri S, et al. Anti-tumor activity of systemically delivered ribozymes targeting murine telomerase RNA. Clin Cancer Res 2004;10:4983–4990.

14. Aggarwal BB. Nuclear factor-kappaB: the enemy within. Cancer Cell 2004;6:203–208.

15. Kashani-Sabet M, Liu Y, Fong S, et al. Identification of gene function and functional pathways by systemic plasmid-based ribozyme targeting in adult mice. Proc Natl Acad Sci USA 2002;99:

3878–3883.

16. Liu Y, Mounkes LC, Liggitt HD, et al. Factors influencing the efficiency of cationic liposome- mediated intravenous gene delivery. Nature Biotechnol 1997;15:167–173.

17. Piali L, Hammel P, Uherek C, et al. CD31/PECAM-1 is a ligand for alpha v beta 3 integrin involved in adhesion of leukocytes to endothelium. J Cell Biol 1995;130:451–460.

18. Fong S, Mounkes L, Liu Y, et al. Functional identification of distinct sets of anti-tumor activities mediated by the FKBP gene family. Proc Natl Acad Sci USA 2003;100:14,253–14,258.

(9)

19. Kashani-Sabet M, Sagebiel RW, Ferreira CM, Nosrati M, Miller JR 3rd. Vascular involvement in the prognosis of primary cutaneous melanoma. Arch Dermatol 2001;137:1169–1173.20.

Kashani-Sabet M, Sagebiel RW, Ferreira CM, et al. Tumor vascularity in the prognostic assessment of primary cutaneous melanoma. J Clin Oncol 2002;20:1826–1831.

21. Kashani-Sabet M, Shaikh L, Miller JR 3rd, et al. NF-kappa B in the vascular progression of melanoma.

J Clin Oncol 2004;22:617–623.

22. Banerji S, Ni J, Wang SX, et al. LYVE-1, a new homologue of the CD44 glycoprotein, is a lymph- specific receptor for hyaluronan. J Cell Biol 1999;144:789–801.

23. Dadras SS, Paul T, Bertoncini J, et al. Tumor lymphangiogenesis: a novel prognostic indicator for cutaneous melanoma metastasis and survival. Am J Pathol 2003;162:1951–1960.

24. Tsai PW, Shiah SG, Lin MT, et al. Up-regulation of vascular endothelial growth factor C in breast cancer cells by heregulin-beta 1. A critical role of p38/nuclear factor-kappa B signaling pathway.

J Biol Chem 2003;278:5750–5759.