Clin Exp Metastasis (2013) 30:345–356

DOI 10.1007/s10585-012-9541-x

RESEARCH PAPER

Tumor location and nature of lymphatic vessels are key
determinants of cancer metastasis
Ramin Shayan • Rachael Inder • Tara Karnezis • Carol Caesar •

Karri Paavonen • Mark W. Ashton • G. Bruce Mann •
G. Ian Taylor • Marc G. Achen • Steven A. Stacker

Received: 23 July 2012 / Accepted: 28 September 2012 / Published online: 3 November 2012
Ó Springer Science+Business Media Dordrecht 2012

Abstract Tumor metastasis to lymph nodes is a key lymphangiogenic growth factors are not the sole determi-
indicator of patient survival, and is enhanced by the nants of metastasis. Here we explored the influence of tumor
neo-lymphatics induced by tumor-secreted VEGF-C or proximity to lymphatics capable of responding to growth
VEGF-D, acting via VEGFR-3 signalling. These targets factors on nodal metastasis in a murine VEGF-D over-
constitute important avenues for anti-metastatic treatment. expression tumor model. We found that primary tumor
Despite this new understanding, clinical observations link- location profoundly influenced VEGF-D-mediated lymph
ing metastasis with tumor depth or location suggest that node metastasis: 89 % of tumors associated with the flank
skin metastasised, in contrast with only 19 % of tumors
located more deeply on the body wall (p \ 0.01). Lymphatics
SAS and MGA are consultants to Vegenics Ltd and are stock holders in metastatic tumors arose from small lymphatics, and dis-
in Circadian Technologies. played distinct molecular and morphological profiles com-
pared with those found in normal lymphatics. Smaller
Electronic supplementary material The online version of this lymphatic subtypes were more abundant in skin (2.5-fold,
article (doi:10.1007/s10585-012-9541-x) contains supplementary
material, which is available to authorized users. p \ 0.01) than in body wall, providing a richer source of
lymphatics for VEGF-D? skin tumors, a phenomenon also
R. Shayan  R. Inder  T. Karnezis  C. Caesar  K. Paavonen  confirmed in human samples. This study shows that the
M. G. Achen  S. A. Stacker
proximity of a VEGF-D? primary tumor to small lymphatics
Ludwig Institute for Cancer Research, Royal Melbourne
Hospital, Post Office Box 2008, Parkville, VIC 3050, Australia is an important determinant of metastasis. These observations
may explain why tumor location relative to the lymphatic
R. Shayan  G. B. Mann network is prognostically important for some human cancers.
Department of Surgery, Royal Melbourne Hospital, University
of Melbourne, Parkville, VIC 3050, Australia
Keywords Metastasis  Lymph node  Lymphatics 
R. Shayan  M. W. Ashton  G. I. Taylor Prognosis  VEGF-D
Jack Brockhoff Reconstructive Plastic Surgery Research Unit,
Royal Melbourne Hospital and Department of Anatomy, Faculty
Abbreviations
of Medicine, Dentistry and Health Sciences, University of
Melbourne, Parkville, VIC 3050, Australia BW Body wall
DAB Diaminobenzidene
R. Shayan  T. Karnezis  C. Caesar  M. G. Achen  IL Initial lymphatics
S. A. Stacker (&)
LEC Lymphatic endothelial cell
Tumour Angiogenesis Program, Peter MacCallum Cancer
Centre, St. Andrews Place, East Melbourne, VIC 3002, Australia LN Lymph node
e-mail: Steven.stacker@petermac.org LVD Lymphatic vessel density
LYVE Lymphatic vessel endothelial hyaluronan
R. Shayan  T. Karnezis  C. Caesar  M. G. Achen 
receptor
S. A. Stacker
Sir Peter MacCallum Department of Oncology, Np Neuropilin
University of Melbourne, Parkville, VIC, Australia PC Pre-collector

123

This Introduction fluid is then transported through deep dermal ‘pre-collec- tor’ lymphatics. 32]. high-grade sar. We also define tumor lymph- remain key indicators in the pathological assessment of atics in terms of features that distinguish normal lym- many types of cancer. and examine independent markers of metastatic risk and poor prognosis how the nature of the nearby lymphatic vessels influences [14–16] (summarised in [13. it is possible that the effect of primary tumor location VEGF-D expression correlated with increased lymphatic on metastasis is based. which is ‘absorptive’ from ‘conductive’ lymphatic vessels combines promoted by the expression of the lymphangiogenic growth assessments of anatomical location. the SK Skin exact reason for this remains unclear. mitotic-rate and nuclear/cytoplasmic ratio are also specific response of collecting lymphatic vessels to VEGF- useful prognostic indicators that guide clinical manage. In contrast. but metastasise lymphogenously locations of the body may be an important determinant of when grown near the skin [24]. in tumor metastasis. 2]. 32]. For example. the the generation of ‘absorptive’ tumor lymphatics [34]. 32]. features such as the anatomical tumor location remain obese diabetic mice important determinants of LN spread [21]. The fact or VEGF-D. For example. the deepest from these smaller calibre vessels but not larger collecting tumor fronts in colorectal cancer express higher VEGF-C lymphatics. Our findings suggest that initial/capillary invading melanoma ‘cell front’ is a key prognostic deter. lymphatic subtypes have different distributions within the thological studies of human tumors in which VEGF-C or body. 10. and size and location of a tumor. This effect is verified using comas rarely spread via lymphatics. this process relies on the presence of lymph- expression of these growth factors is independently related atics that are able to sprout in response to the lymphan- to poor prognosis [22]. other SCID/NOD Severe combined immunodeficiency/non. for review see [12]. on the degree to which vessel density (LVD). Therefore. established [33]. Consequently. lymphatics and pre-collecting lymphatics (referred to minant. a smooth muscle cell (SMC) [31] and pericyte many human tumors [1. 11]. This suggests that. and more recently. the location. tumor metastasis. morphology. Clinical and experimental data coating [27. large. 123 . being able to interact with lymphatic endothelial cells variations in the lymphatic subtypes occurring in different (LECs) in vitro [23]. the depth of an phatic subtypes. or VEGF-A [6] in mouse lymphatic-marker profile [26–29. and with tumor-derived lymphangiogenic growth factors can access metastasis [7–9]. which absorb lymph fluid. D involving prostaglandin signalling leading to vessel ment of cancer patients. 17–20]). anatomical location and embryological development. Further. giogenic growth factors.346 Clin Exp Metastasis (2013) 30:345–356 PECAM-1 Platelet endothelial cell adhesion molecule-1 addition to requisite lymphangiogenic growth factors. in part. and factors VEGF-C or VEGF-D [3–5]. The latter have valves. and their signalling pathway via the VEGFR-3/ that the presence of SMCs and pericytes on certain blood Np-2 receptor complex. While VEGF-D expression does enhance LN or VEGF-D levels than more superficial tumor tissue. and a LEC-lined lumen that channels indicate that LN metastasis is associated with tumor-related lymph to LN [25–29]. to the more deeply located ‘collecting’ Lymph node (LN) involvement is an important diagnostic lymphatics [25–29]. genic growth factor expression and nodal spread in patients Here we explore the significance of tumor location for has led to suggestions that VEGF-C and VEGF-D may be VEGF-D-induced tumor lymphangiogenesis. and perform distinct functional roles [25–30]. it was observed that invasive collectively as small lymphatics) are the source vessels breast cancers involving or near overlying skin have higher that sprout to form tumor ‘neo-lymphatics’. and depth of tumor invasion thereby. dilation to favour transport of tumor cells has recently been Immunohistochemical correlation between lymphangio. despite sarcoma cells equivalent VEGF-C over-expressing tumors. and that metastasis rates and worse prognoses than size-matched VEGF-D-expression in mouse tumors induces sprouting tumors located more deeply [21]. lymphatic subtypes also behave differently in response to Apart from ‘LN-status’. However. the lymphatic vasculature commences in the superficial dermis as ‘initial’ or ‘capillary’ lymphatics. are important targets for anti. depending on the tumor type [13]. our ability to distinguish lymphangiogenesis (lymphatic vessel growth). and metastasis. a basement and prognostic parameter for the clinical management of membrane. parameters such as tumor depth/ lymphangiogenic growth factors [27. in the primary tumor. In skin. Currently. 5. SMC Smooth muscle cell It has been previously recognised that different subtypes VEGF Vascular endothelial growth factor of lymphatic vessels vary in morphology. Given that distinct cancer models. Nevertheless. vessels is thought to mediate a ‘non-sprouting’ or quiescent lymphangiogenic treatment as an avenue to inhibit tumor vessel phenotype [31] raises the possibility that different metastasis [3. These findings are supported by histopa. molecules such as VEGF-C lymphatics with responsiveness to these ligands.

Japan) or fixed in 4 % against rat (DAKO Corp. rabbit and goat Ig (1:1000) (DAKO) and wholemount staining..0 L/min oxygen administered for 1 min. Tissues were fixed in 4 % PFA prior to paraffin post-mortem for histological examination. mouse (Bio-Rad.2 L/min oxygen for maintenance) admin. ELISA. blot as described previously [5. Three other embedding and immunohistochemical vessel quantification constructs were made in which either or both of the N. Carpinteria. Louis. The left ears of 20 female SCID/NOD and 15 samples were mounted using SlowFade Antifade kit. Australia) were woun- [Invitrogen. Cell lines expressing a combination of mouse VEGF-A164 and human VEGF-C were also gen- Primary antibodies used: MAb against mouse PECAM-1 erated in 293-EBNA-1 cells and formed the basis of (1:1000) (BD Pharmingen. Samples were analysed using a MO) was used to detect F-actin. human LYVE-1 (1:1000) and VEGF-D with the body wall (BW) or with the overlying skin (SK). Inhalational anaesthesia (4 % isoflurane in 3. This protein LYVE-1-stained tumor sections were photographed and the 123 . CA) when required. podoplanin. paraformaldehyde (PFA) prior to paraffin embedding or Hercules.. MN). Carpinteria. Perth. VEGF-DDC and VEGF-DDN [39]. tems kit (Minneapolis. Tumors and regional LNs were removed consent.Clin Exp Metastasis (2013) 30:345–356 347 Materials and methods lacks the N-terminal propeptide of VEGF-D. or fluorescent high setting. and tumor samples tested for VEGF-D a volume of 2500 mm3 (flank) or a diameter of 2 mm expression by immunoprecipitation followed by Western (ear) (as per local Animal Ethics Committee limits). Billerica. R. paraffin embedded sections was performed with pH 6. Wholemount ear samples were im- rabbit anti-mouse LYVE-1 (1:1000) (Fitzgerald Industries. SMA (1:500) and LN were analyzed for metastatic tumor cells. CA). at *90–94 °C for two consecutive treatments DAB chromogen. retrospectively. manufacturer’s instructions. Carlsbad. tumors were. CA). CA) and human D2-40 antibody to normal tissue sections were immunostained for LYVE-1. and Nikon TE2000-E and C1 included anti-rat. Each of these constructs contained a FLAG sequence to facilitate Antibodies protein purification. then adjusted to 3 % isoflurane in 0. VEGF-DDNDC [38]. type IV collagen and NG2. Phalloidin (Sigma. as per ded with a standard punch [40] and harvested at day 21. Carlsbad. St. Lymphatic quantification and statistical analysis VEGF-DDN consists of residues 93–354 of human VEGF- D tagged at the N-terminus with the FLAG octapeptide. Tumor and (DAKO. peroxidase-conjugated Ig Tissue-Tek OCT (Sakura. aSMA. MA). previously described [29]. munofluorescently stained using overnight incubations Concord. (Molecular Probes). podoplanin (1:500) (Vector Laboratories. Burlingame. anti-rabbit. Minneapolis. of 5 min each (citrate buffer was topped-up between). with ABC amplification (Vector Labo. CA]. Burlingame. New South Wales. MN) as per manufacturer’s Australia) and animals sacrificed prior to tumors reaching instructions. and animals were euthanized using inhaled mammoplasty surgery that was collected at the time of carbon monoxide (as per local Animal Ethics Committee surgery with appropriate institutional approval and patient specifications). grouped as being associated Minneapolis. (1:300) and M2 (anti-FLAG) (1:500) (Sigma. MN). Polyclonal antibodies used: biotin-conjugated anti-rat Ig Mouse tissues were harvested and either frozen in (DAKO Corp. MO). in SCID/NOD mice (ARC. CA). at ratories. CA). Tokyo. St. Human tissue samples istered via a TM41 CIG anaesthetic machine (North Ryde. Carpinteria. Louis. CA). Quantification of LVD within flank tumors was performed The three amino acid residues T.0 Immunostaining of paraffin and frozen sections involved a citrate buffer.or analysis (see below). anti-mouse and anti-hamster confocal microscope. C57Bl/6 mice (WEHI. Microwave antigen retrieval of antibodies (used at 1:500) (Invitrogen. Australia) was used to perform tumor Human tissue samples were surplus tissue from reduction inoculations. Melbourne.flank LYVE-1 (1:1000) and podoplanin (1:200) (R&D Systems. collagen type IV control tumors. western blotting and immunoprecipitation Generation and analysis of tumors and wound samples studies 293EBNA-1 flank [5] and ear tumors were established as ELISA for VEGF-D was performed using an R&D Sys- described [35–37]. C-terminal domains of human VEGF-D were deleted. CA). (1:50) (R&D Systems. fluorescence Nikon Eclipse 90i upright microscope with a Alexa fluorescent-conjugated secondary antibodies Nikon DXM1200c camera. San Diego. 38]. Eighty-four VEGF-D? and twenty VEGF-D. and N are present using a pixel counting method in which multiple areas of between FLAG and the VEGF-D sequence. F-Actin. MA) and NG2 antibody (1:200) to pericytes (4 °C with gentle agitation) for LYVE-1 or podoplanin as (Milipore.

VEGF-DDC. podoplanin. 2). In general. 1a). to identify the lymphatic 293-EBNA-1 tumors secreting variant forms of the VEGF. 0/4) (data not shown). extending for variable distances We previously showed that tumor xenografts generated towards the tumor centre. 1b). Statistical analyses were performed as previously The use of multiple LEC markers showed that most tumor described [5]. staining underlying musculature of the body wall (designated ‘‘BW’’ for F-actin. VEGF-DDN. terms of different LEC marker profiles has improved parable VEGF-D-expression was detected in both tumor immunohistochemical differentiation of these subtypes in types as assessed by immunoblotting and ELISA (Fig. In order to assess the effect of primary tumor and VEGF-D. upper and lower left pan- Primary tumor location influences metastasis els). demonstrated that SK tumors had a 3. Results In addition. mouse VEGF-A164 and VEGF-C (SK 12/14. and com. tumor lymphatics in SK tumors were mor- phologically abnormal (Fig. BW 1/10) or with tumors over- Ottawa. the few LYVE-1?/ found to have no baseline level of VEGF-C or VEGF-A podoplanin? lymphatics present in indolent VEGF-D? BW secretion [5]. The sizes of SK and BW VEGF-D? tumors were not significantly different (both were 600–700 mm3 after The recent description of lymphatic vessel subtypes in 23 days and grew at similar rates throughout). 1b). LN metastasis of SK and BW tumors—and the role played skin and BW lymphatics in both animal and human spec. 1d. we established 293EBNA-1 dilated and were not seen in multiple serial sections tumors expressing full-length human VEGF-D (VEGF-D? (Fig. Similar results were observed with mural SMCs [25–29]. regional LN.tumors. This model was tions for up to 200 lm. Supplemental Fig.01) (Fig.348 Clin Exp Metastasis (2013) 30:345–356 identified vessels were quantified using Corel Draw soft. Canada).01) (Fig. metastasised than the BW (Fig. SK 13/16 metastasis to were stained for LYVE-1 and podoplanin (Fig. cent wholemount staining with LYVE-1 or podoplanin. 41] revealed a vari- When the flank skin was reflected to reveal the primary able and patchy covering of SMA? cells on the LYVE-1?/ tumors post-mortem. expressed proteins commonly used to analyse lymphatics. Tumor LVD quantitation using LYVE-1. To explore the discrepancy found between the rates of using the LVAP software plugin for ImageJ [37]. BW 0/8. upper right panel). ware (Coral Draw Graphics Suite X3. 1d. BW tumor models were quantified following immunofluores. be they SK or BW tumors. While dermal initial lymphatics In order to exclude these findings being specific to this were shown to express both LYVE-1 and podoplanin. by the differences in lymphatic subtypes in the two ana- imens were quantified using the LVAP software on tomical locations—immunohistochemistry and quantitation immunohistochemical sections. once the lymphatic vessels of tumor lymphatics were performed (Fig.control tumors. 1b). BW 2/8. they frequently do so [5]. they could be classified as either podoplanin? tumor lymphatic population (Fig. LECs in the subcutaneous collecting lymphatics express similar studies using variant forms of VEGF-D and VEGF. and are associated with C were performed. but not LYVE-1. 123 . 5]. often spanning up to half the from 293EBNA-1 transformed embryonal kidney cells do tumor tissue sections (*300 lm). 3/16 or 19 % BW tumors spread to LNs. collagen IV and the NG2 antibody revealed tumors) (Fig. Histological examination of ipsilateral an absence of anchoring filaments and pericytes. using two-tailed student’s t test. vessels in mouse skin and BW. 1c). plemental Fig. or wounded ear skin or ear factors. Corel Corporation. Similarly. when stably transfected to individual vessels were continuous in multiple serial sec- express VEGF-D.05. On many occasions. 1d. as previously described [5]. 1d. In addition.5-fold higher LVD than BW tumors (p \ 0. d) using had been identified using a combination of anti-LYVE and antibodies to podoplanin [27] and LYVE-1 [4. Lymphatic expressing a combination of other lymphangiogenic growth vessels in normal ear skin. not metastasize. lower adherent to skin (designated ‘‘SK’’ tumors) or to the right panel. as shown in Fig. In contrast. however. and axillary LNs revealed that SK tumors spread to LNs much scant basement membrane on tumor lymphatics (data not more frequently than BW tumors (40/45 or 89 % SK shown). 2a). mouse flank tissue sections D polypeptide (VEGF-DDNDC. were present on tumor lymphatics [29. were not highly branched or location on LN metastasis. expressing no VEGF-D smooth muscle actin (SMA) to ascertain whether SMCs (VEGF-D. SK6/10. Statistical lymphatics co-expressed LYVE-1 and podoplanin (Sup- significance was attributed to p values \ 0. none of the Small lymphatic subtypes are more abundant in the skin VEGF-D. LYVE-1?/podoplanin? lymphatics in to regional LNs metastatic VEGF-D? SK tumors were predominantly peripherally located. tissue samples [25–29]. Therefore. SK 14/17. 1). lymphangiogenic growth factor over-expression model. tumours spread to LNs. in the flanks of SCID/NOD mice [5]. 1c.tumors). LEC- anti-podolplanin vessel staining techniques. p value \ 0. Staining with an antibody to tumors) and control tumors. In contrast.

VEGF-D? BW and VEGF-D. error bar = SEM. 2a.01 Student’s t test.01 student’s t test. Scale bar 100 lm. or BW (‘BW’ tumors). Podoplanin immunostaining of lymphatics in SK tumors reveals lymphatics forming branched septa that permeated the tumor from the periphery. for tumor-derived human VEGF-D (lower right panel). 2b). although a similar density of collecting lymphatics were present in the BW as in skin.flank tumors. VEGF-D. upper panels) of the mouse flank. BW body wall tumor. staining profiles and anatomical location within these regions. as seen in upper left and lower left panels. **p value \ 0. error bar = SEM. d LYVE-1. 2b). VEGF-D? BW and VEGF-D. Black arrows in upper right panel and lower left indicate tumor lymphatics and white arrows indicate LEC-lined spaces. Western blotting of tumor lysate for VEGF-D (*55 kDa band indicates full-length form of VEGF-D) indicates high levels of VEGF-D protein expression in both BW and SK tumors (lower left panel). and this is confirmed by ELISA analysis of serum from mice bearing BW or SK tumors. lower panels) present in both BW and skin is consistent with the expected staining profile of collecting lymphatics [25–29]. The level of LYVE-1 staining on these vessels was con- sistent with that of negative control antibodies (data not shown). SK skin tumor. podoplanin and LVYE-1/SMA immunohis- tochemical staining of VEGF-D? SK tumors and VEGF-D? BW tumors. metastasising at higher rates than equivalent body wall tumors. quantitation of the density of LYVE-1?/Podoplanin? (‘small’ lymphatics) versus LYVE-1-/Podoplanin? (col- lecting lymphatics) within the skin and BW was performed (Fig. b The relative frequency of metastasis from VEGF-D? SK. The average density of ‘small’ lymphatics (a combination of initial and pre-collecting vessels) was *2.flank tumors. Using this differential expression profile of LEC-mark- ers to identify the respective lymphatic vessel subtypes. to axillary lymph nodes is shown graphically as the % of lymph node metastases (upper panel). 1 Skin-adherent VEGF-D? tumors generate higher densities of c abnormal lymphatics. error bar = SEM.5-fold higher in skin than that in the BW (Fig. as in upper right and lower right panels.01 student’s t test. An additional subset of LYVE-1-/Podoplanin? lymphatic vessels (Fig. Serial mouse flank sections revealed that. **p value \ 0. **p value \ 0. spanning much of the tissue section and forming LEC-lined spaces.Clin Exp Metastasis (2013) 30:345–356 349 Fig. were located in both the BW and skin (Fig. 123 . We did not differentiate between initial and pre-collector lymphatics due to similar morphology. (Color figure online) LYVE-1?/podoplanin? structures (Fig. Units shown in 106 pixels/910 objective field. 2a).tumors are pooled samples from BW and SK tumor-bearing mice in lower right panel of b. a Macroscopic and histological appearance of 293-VEGF-D1 tumors in mice. Fluorescent co-staining for SMA (red) with LYVE-1 (green) in the lower right panel showed intermittent and variable SMCs associated with tumor lymphatics. Tumors (T) were adherent to either the skin (S) undersurface (‘SK’ tumors). c Quantification analysis of LYVE-1? lymphatic vessels in VEGF- D? SK. consistent with initial and pre-collector lymphatics (so-called ‘small’ lymphatics) [25–29]. 2a. 2b). there was a significant dif- ference between the densities of the other lymphatic sub- types present in the two locations (Fig.

A artery. 2 **p value \ 0. PC. podo podoplanin. **p \ 0. and high power (920) in podo LYVE-1 LYVE-1 upper middle panel) or Mouse skin Mouse skin Podoplanin (upper right panel) in serial sections of mouse skin.e. I initial lymphatics.350 Clin Exp Metastasis (2013) 30:345–356 Fig. Scale bar 100 lm. a Immunohistochemical staining identifies abundant S B initial lymphatics in mouse skin stained with either LYVE-1 (at low power (94) in upper left panel. I. B body wall. b Quantification of the density B ** of different lymphatic vessel 4 I+PC LVD (number lymphatics/ subtypes present in different Coll anatomical locations of the 3 mouse skin. Tissue sections were Anatomical Location (Mouse) stained with an antibody to LYVE-1. c Human breast ‘skin to body wall’ tissue 0 was taken and represented an Body Wall Skin area from the skin (S) to the fascia (F). Units shown in LVD/109 field. pre-collector 1 lymphatics. Coll 12 collecting lymphatic vessel. 2 Analysis of lymphatic subtypes in mouse and human A Mouse skin Mouse skin Mouse skin tumors and tissues. collecting lymphatic vessel. initial lymphatics. error 8 bar = SEM 4 0 Skin Body Wall Anatomical Location (Human) 123 . Human skin C Human skin d Quantification of human lymphatic subtypes S immunostained with antibody against LYVE-1 and D2-40 F demonstrated the density of small and collecting lymphatic subtypes in the superficial and deep human tissues. Coll 10 power field) LVD lymphatic vessel density. S skin. error bar = SEM. comparing the skin (in the LYVE-1 dermis) with the body wall (deep to the deep fascia). Coll.01 student’s t test. Black arrows LYVE-1 podo indicate tumor lymphatics. i. Units shown in 10 power field) LVD/10 9 field. PC pre- D ** I+PC LVD (number lymphatics/ collector lymphatics.01 Student’s t test. V vein. LYVE-1 and podoplanin staining of body wall serial V V sections (lower panels) demonstrates LYVE-1-/ A A podoplanin? collecting lymphatic vessels in mouse flank tissue.

Blind-ended. i. less branched. 2c. Quantification of ‘small’ lymphatics in the skin compared with that in the the number and characteristics of the sprouts arising in an BW (Fig. A combined analysis of This pattern of staining was observed in 4 of 10 meta- anatomical location. [27] and anti-human LYVE-1 antibodies. VEGF-D? ics (Fig. same comparison of mouse tissue (Fig. left panel) and In samples of ear skin bearing VEGF-D? tumors. 3a). Unlike the istics of the mouse tumor lymphatics were attributable to deep dermal lymphatics. 3a). 4) in these human tissues ear tumors induced small lymphatic vessel sprouts and demonstrated a similar ratio between the lymphatic sub. 3b). 5) enced by VEGF-D? tumors displayed higher numbers of were consistent with those described in VEGF-D? flank lymphatic sprouts. Further. ear tumor was performed and compared with those arising in the ear skin wounding model (Fig. connections were observed between tumor lymphatics and types in the skin and those in deeper tissues.. 3a). stained for podoplanin (using anti-human D2-40) and deranged lymphatic structures were seen (Fig. 3b). (Fig. the morphological and molecular features logical models uncovered several differences. LYVE-1?/podoplanin? lymphatics. analysed. 2b. 2d. normal lymphatic subtypes. archived samples of human melanoma and surrounding tinuous with the large. which were stained using the anti-human podoplanin doplanin? lymphatics adjacent to the tumor mass (Fig. deranged. six breast reduction operations (Fig. irregular LYVE-1?/po. VEGF-D. superficial dermal lymphatics adjacent To confirm observations of the lymphatics in our animal to VEGF-D? tumors were dilated compared to lymphatics models and to exclude the possibility that the character- in the skin of normal control animals (Fig. d).e. there is a significantly greater density of lymphatics in the ear subcutis (Fig. ear skin influ- static ear tumor sections (Fig. the cervical toward the wound stimulus (Fig. that LYVE-1 (Fig. located more superficially to straigh- BW applies in human tissues. by dilated LEC-lined spaces and often extending over multiple serial sections (see Fig. disorganized. Quantitative analysis of the the VEGF-D? flank tumor model was adapted to the ear as resulting neo-lymphatics was performed (Fig. 2d. 3d) as seen in our mouse tumor models. 3b). tumors. 3c). strongly LYVE-1?/podoplanin? initial small lymphatic vessels observed between mouse skin and lymphatics were seen. Ear tumor models identify the presence and origin of pathological lymphatics An ear wounding model verifies the nature of pathological lymphatic sprouting In order to explore the differences between lymphangio- genic responses of the distinct subtypes of lymphatic ves- To analyse lymphangiogenesis in an alternative patholog- sels and to enable us to examine lymphatics in metastatic ical setting.Clin Exp Metastasis (2013) 30:345–356 351 In order to establish if the difference in the density of highly branched. (Fig. as seen in the the native small lymphatic network (Fig. nor any direct connections there is a similar density of collecting lymphatics in both from tumor lymphatics were seen to involve the collecting the skin and BW. highly however a shorter average sprout length was observed than branched. highly branched was first performed to verify our ability to differentiate lymphatics) (Fig. right panel). predominantly peripherally located. 43]. Control VEGF-D- tumors lacked lymphatic structures (data not shown). The previously described (see ‘‘Materials and Methods’’) newly-generated lymphatics originated from pre-existing (Fig. 42. 3b). tips per sprout and lymphatic loops. ter. In contrast. we investigated podoplanin? dermal lymphatics did not appear to be con. d). 2 mm circular ear wounds were created VEGF-D? tumors using wholemount immunofluorescence. 1d). 2c. left panel). None of the mice with VEGF-D. discussed above. Quantifi- LN chain. valve-bearing LYVE-1-/podoplanin? ously from skin to deep fascia (inclusive) were taken from collecting vessels (Fig. discussed below). Quantitation of the density correlated with molecular and morphological features of LYVE-1?/D2-40? small lymphatics (Fig. as described [40]. 1d) or ear tumors (Fig. i. observed in sections from metastatic VEGF-D? flank mental Fig. Supplemental Fig. 14/15 mice with VEGF-D? ear tumors cation of the lymphatic sprouts driven by the two patho- did so. Compared of the lymphatics identified on immunostaining of meta- with wound-induced lymphangiogenesis. In Lymphatic vessels in human tumors addition. whilst neither lymphatic sprouting. 123 . 3b) [43]. 2b. as has been shown in VEGF-C over-expressing tumors [35. Analysis To analyse sprouting from lymphatics in response to revealed a similar abnormal lymphatic morphology (i. morphological characteristics and static tumors but in none of 10 non-metastatic tumors LYVE-1/podoplanin staining was used.e. the most superficial LYVE-1?/ xenograft transplantation alone [26]. 3b. however. skin. 3c) [37]. punctuated in ear wounds (Fig.. wholemount immunofluorescence of ear tissue predominantly peripheral.e. 3d. 3c.ear tumors had small lymphatics (as in the ear tumor model). samples extending continu. Supple. 3) and LYVE-1-/D2-40? collecting lymphat. Supplemental Fig. 3d). and sprouted LN metastases to the regional draining LNs.

352 Clin Exp Metastasis (2013) 30:345–356 A Mouse ear Mouse ear T Tumor margin Tumor margin Tumor margin B T T Tumor margin T LYVE-1 LYVE-1 LYVE-1 LYVE-1 Skin near tumor Collecting vessels-ear LYVE-1 T LYVE-1 podoplanin C 25 ** * 160 20 120 Sprout length (pixels) Events/field 15 80 10 * * 40 5 0 0 S C S C S C S C S C S C S C S C T W T W T W T W Sprout Tips/ Loops Sprout number Sprout length D Mouse ear Melanoma LYVE-1 W Podoplanin 123 .

Open arrows indicate dilated dermal distinct lymphatic subtype functionally adapted to favour initial lymphatics and arrowheads denote pathological neo-lymphatics LN metastasis. *p \ 0. Right lower panel depicts tion arose as to why SK tumors were able to recruit neo- normal collecting lymphatic vessels stained for podoplanin (green) lymphatics more readily than BW tumors. but were deficient in anchoring may occur because a tumor in the relatively thin ear will be filaments. the flank tumor model was BW VEGF-D? tumors. ation [32]. 43]. Inset in upper panel gene expression of other growth factors in the endothelial demonstrates lymphatic endothelial structures (denoted by white cells of regional LNs in this model associated with vascular arrows). T tumor. Scale bar 100 lm in left panel and 50 lm lymphatics in these locations. type or the VEGF-D? mouse ear tumor. therefore. W wound. that the tumors were of different adapted to the ear. **p \ 0. 42]. The lymphatics in the to the enhanced LVD found in metastatic VEGF-D? VEGF-D? metastatic tumors lacked basement membranes tumors. Toge- large LEC-lined spaces. open white arrows indicate lymphatic (collecting) lymphatic vessels. in particular contributing over-expressing tumors [24. open SK and BW tumors was related to the different LVDs and/ arrows indicate lymphatic vessels sprouting towards the tumor. 36. as were lymphatic sprouts driven by an underwent regular branching. a location allowing visualisation and sizes or exhibited different VEGF-D-expression. Scale bar 50 lm. Given that sprouting can in right panel. particu. 123 . 42. To investigate the lymphatic paring lymphatic sprouting parameters between the tumor and wound vessel subtypes in the normal flank. undergoing no discernable sprouting. d Left panel depicts wholemount LYVE-1 immunofluorescent staining of lymphatics adjacent to a and pre-collector) and LYVE-1-/podoplanin? ‘large’ punch wound in a mouse ear. and associated during lymphatic development [29] or regener- human melanomas. VEGF-D (for review see [45]). C collecting lymphatics. Inset or altered lymphatic vessel morphology. 36. Inset in left panel depicts enlarged view enhanced surface area that facilitates interaction with tumor of ear tumor denoted by broken line box. or in the ability of these vessels to respond to tumors and ear punch biopsy wounds. at higher power. larly proximity to small lymphatic vessels. despite smaller sizes than their flank counterparts. a Macroscopic appearance (left panel) and LYVE. c Quantitative analysis of lymphatic sprouts arising from LYVE-1? small lymphatic vessels adjacent to VEGF-D? ear tumors. the Discussion difference in relative density of the lymphatic subtypes in the two locations could account for differences in the LVDs Our results indicate that primary tumor location. tumor. closed white arrows indicate normal lymphatics from which densities of collecting lymphatics in the skin and BW. These tumor lymphatics displayed an formed at margin of the tumor. and spanned multiple serial sec. VEGF-D-induced sprouts from nearby structures seen in VEGF-D? metastatic tumors exhibited native small skin lymphatic vessels were continuous with morphology distinct from normal lymphatics. the ques- in left panel depicts enlarged view of ear tumor denoted by white box. all features reminiscent of those in VEGF-C pathological lymphangiogensis. Lower left panel depicts sprouting from normal initial lymphatics immediately adjacent to the VEGF-D? remodelling and spread of tumor cells to regional LNs [44]. Two other possible were the active component of the skin that lead to the explanations for the different rates of spread by SK and enhanced LVD in SK tumors. were quantification of both normal and tumor lymphatics in excluded. The high metastasis rate amongst VEGF-D? ear and pericytes. and encouraged further lymphatic b Wholemount immunofluorescent staining of VEGF-D? mouse ear tumors for LYVE-1 (green) in upper panels. and the inability for VEGF-D. 35. ther. there lymphatic sprouts arose at the points denoted by arrowheads. these lymphatic sprouting models suggest that small- tions. They tumor lymphatics. mal lymphatic capillaries. 43] or VEGF-D [37]. wound-healing. S small between LYVE-1?/podoplanin? ‘small’ (initial/capillary lymphatics. Tumor lymphatics may. like nor. for LYVE-1 (green). The presence of SMC on tumor lymphatics is nearer to skin. 3 Analysis of lymphatics in mouse ear tumors and wounds.01. (Color figure online) occur from small lymphatics in response to VEGF-C [35. T tumor. as LYVE-1-expression has been metastasise confirms the pre-requisite for a lymphangio- shown to be down-regulated on LECs once SMCs become genic growth factor. we differentiated models in the ear (Student’s t test). immediately adjacent to the margin of was that there may be differences in the character. Broken line denotes tumor margin. and thus metastasis rates of SK and BW tumors. Right was a significant discrepancy between the densities of small panel depicts immunohistochemical staining of metastatic human melanoma for podoplanin. the lymphatic three-dimensions. suggesting expansive areas of LEC ‘sheet-like’ calibre lymphatics are the source of neo-lymphatics in structures. cells and growth factors. and were LYVE-1?/podoplanin?. Whilst there were similar sprouts. Clin Exp Metastasis (2013) 30:345–356 353 b Fig. sprouting to promote entry of metastatic cells to the lym- like’ lymphatic morphology and arrowheads denote the normal native phatic network. T tumor. We have also previously observed altered lymphatic network at the tumor margin. represent a 1 immunostained section (right panel) of VEGF-D? 293-EBNA tumors grown in the mouse ear. density of the pre-existing lymphatics near the primary Scale bars 50 lm. influences LVD To test our hypothesis that small lymphatic subtypes in the tumor and LN metastasis. occasional dilatation to form alternative pathological stimulus. arrows indicate ‘sheet. but has not been observed to occur from intact collecting lymphatics [33. tumors.tumors to somewhat surprising. Wholemount immunofluorescent staining of lymphatics in the It is likely that the difference between metastasis rates of skin near the VEGF-D? mouse ear tumor. Hence.05 com. On analysis of the primary tumors. The implication (demarcated by white arrows).

In unable to access the lymphatic network and spread to regional lymph nodes. Intact collecting lymphatics may instead respond to lymphangiogenic growth factors by circumfer- Lymph ential enlargement and dilatation [35. B Tumor Type Overall. of the tumor or tumor thickness and vertical growth phase in nantly have access. The BW tumor is therefore to generate tumor lymphangiogenesis and metastasis. such as breast carcinomas [21]. and generates ‘neo-lymphatic vessels’ from the small-calibre dermal lymphatics under the influence of VEGF-D. 36]. collecting recent evidence that traditional indicators (such as the site lymphatics to which more deeply located tumors predomi. IL initial lymphatic. 43]. a hypothesis summarised in in dermal and subcutaneous layers of normal mammalian skin. 41] and pericyte coating [26. Larger blood vessels with associated mural SMC and IL DERMIS pericytes are said to represent a ‘quiescent. 4 Schematic representation of lymphatic vessel subtypes found and thus promote metastasis. and the lack of sprouting from collecting lymphatics near SK Tumor VEGF-D? ear tumors. 15]. 21. A Epidermis atics in the ear did not sprout into or towards tumors. which have valves. but only circumferential enlargement of collecting Neo-lymphatics Tumor Tumor Collecting lymphatics. would have a minants of clinical staging and prognosis [1. and whilst it may lie adjacent to ‘non-sprouting’ tumor. rather than sparser. non-sprouting phenotype’ [31. a continuous basement membrane. therefore. this study demonstrates that whilst lymphangi. 14. They indi. 4. non-sprouting. 32]). and were rela- BODY WALL BW tumor ted directly to rates of metastasis [33]. b Summary contrast. The SK tumor is seen near the cate that while VEGF-D is permissive for lymphatic spread.354 Clin Exp Metastasis (2013) 30:345–356 In contrast to small lymphatics. to small longstanding clinicopathological knowledge and more lymphatics. they must have access to pre-existing Pre-Collecting LV lymphatics capable of responding to stimulatory cues. 48]. suggest that intact collecting lymphatics (which have a basement membrane. 42.‘non- sprouting’ role in LN metastasis [33. greater propensity to spread lymphogenously. Taken together. 46]. deeper in the dermis and drain into subcutaneous collecting lymph- establishing tumor lymphatics with a distinct morphological atics. in models consisting of high-LVD metastatic cells spread lymphatic VEGF-C over-expressing tumors [35. 36. 42. skin. Col- lecting lymphatics. chiefly. Fig. The low LVD in VEGF-D? BW PC tumors that have access only to collecting lymphatics. Collecting lymphatics may therefore be unable to sprout towards tumors to form absorptive tumor lymphatics [34]. LV type present SK Tumor BW Tumor ogenic growth factors are necessary for lymphatic vessel Initial LV growth in tumors. Metastatic Tumor cells may represent an analogous ‘non-sprouting lymphatic SUBCUTANEOUS phenotype’. may additionally influence lymphangiogenesis have not 123 . there are no lymphatics nearby with the of sprouting is critical for lymphangiogenic growth factors capacity to sprout and infiltrate the tumor. 28. the collecting lymph. play an important. One of the limitations of this study VEGF-C or VEGF-D expression has been suggested as a is that other cell types in the tumor microenvironment that prognostic indicator in several cancer types [7. Whilst should not be neglected. PC pre-collector lymphatic. 36. there is a paucity of small lymphatic vessels in the of lymphatic vessels (LV) occurring in the vicinity of SK and BW BW that limits the abundance of lymphatics in BW tumors. The pre-collecting vessels are grew into SK tumors from the skin in response to VEGF-D. our data indicate that lymphatics a Superficial dermal initial lymphatics absorb lymph for transporta- tion to the pre-collecting vessels. Our Collecting LV animal models have shown that primary tumor location exerts an effect on the ability to generate tumor lymphatics Fig. and an SMC [32. Hence human tumors that may originate melanoma) remain key predictors of metastasis and deter- close to skin. a thicker wall including SMC coverage and and molecular profile that promote metastasis. 8. This Node phenomenon was recently demonstrated in the case of VEGF-D mediated alterations. and that access to lymphatic vessel subtypes capable collecting lymphatics. It is also consistent with previous observations of sprouting by small lymph- LEGEND atics. tumors Our results also indicate that human tumors formed near epithelial surfaces may have access. 43]. 47. The BW tumor its effect may be limited by the anatomical location of a arises deeply. adventitia. that were driven by enhanced levels of specific prostaglandins.

Br J Cancer 94:1355–1360 pathological tumor neo-lymphatics may express a novel 12. Kitadai Y. This 59. Histopathology. RS is supported by the Raelene Boyle Sporting 19. Achen MG. Achen MG. Gorski DH (2003) Vascular endothelial growth factor C mRNA expression correlates with stage of progression in Acknowledgments The authors thank M. Mandriota SJ. Cancer Sci 95:32–39 23. and lymphatic markers have been demonstrated on 5. Mol Med 7:598–608 ment. Feng D et al (2002) Vascular permeability factor/vascular endothelial growth factor induces lymphangio- angiogenic growth factors in vitro and in vivo. Schietroma C. di Tomaso E et al (2002) Lymphatic melanoma in the 21st century. when considered in conjunction with the level of 539–549 VEGF-C or VEGF-D expression. J. however. Dis Colon Rectum 54:35–40 3. 14. and the RACS Surgeon Scientist 20.04310. Hibbs for critical reading of this lymphatic density in primary cutaneous melanomas predicts risk manuscript.2012. 4. VEGF-C expression. Philadelphia. Mel. genesis as well as angiogenesis. Vasile E. Cancer Cell 13:331–342 11. Y.1365-25 tissue samples were in accordance with NH&MRC guidelines. Padera TP. Among these. Lippincott. Jeltsch M et al (2001) Vascular endo. macrophages have been sug. Goydos JS. 15. derived from pathologi. Carcinogenesis 27:1729–1738 provide therapeutic targets for restricting metastasis [11]. doi:10. APMIS 112: tumor. Erickson LA. Stacker SA. Nat Med 7:192–198 ment. Cancer 98: 789–797 inflammatory cell populations. Tammela T. Caunt M. principles 25. treatment. Mak J. Burgess and M. Alitalo K. Rao RD et al (2007) Malignant 24. Achen MG. and may 13.1111/j. Stacker SA. Zhang for histology services. 26.x. SAS and esis and cancer metastasis. Oman L et al (2006) Is risk of axillary lymph node metastasis associated with proximity of breast cancer Ethical standard Ethics approval for research using human and to the skin? Breast Cancer Res Treat 100:319–328 animal samples was obtained from the Royal Melbourne Hospital 22. Cianfarani F. Liang WC et al (2008) Blocking neuropilin-2 cal samples. Australia. Petrova TV (2005) Lymphangiogenesis in NH&MRC. Onogawa S. Kadambi A. Stacker SA (2005) Focus on lymphan- molecular signature could provide additional avenues for giogenesis in tumor metastasis. Hellman S. respectively. the current tumor model was chosen Med 7:186–191 due to low baseline levels of production of other lymph. Jussila L et al (2002) Lymphangiogen- and Medical Research Council of Australia (NH&MRC). Kesar and staff at the patients with melanoma. Bevacqua SJ. EMBO J 20:672–682 433–449 123 . Francois for density correlates with nodal status. Ann NY Acad Sci 1131:225–234 (RACS) Foundation Scholarship. Murali R et al (2012) Variations in tumor imaging. Slagle-Webb B. DeVita VT. carcinoma and sarcoma tumor cell adhesion to lymphatic endothelium. work was funded partly by a Program Grant from the National Health 17. Rosenberg A (2001) Cancer. 6th edn. This work was supported by funds from the Operational an X-linked/AP-1 regulated putative onco-angiogen in human Infrastructure Support Program provided by the Victorian Govern. Breast Cancer Res Treat 91:125–132 figures. Achen MG et al (2001) VEGF-D is Program. Taylor for assistance in generating prognosis in breast cancer. part 2: staging. Lacal PM et al (2003) Vascular secreted into the systemic circulation are no greater than endothelial growth factor-C expression correlates with lymph baseline in the control tumor model. Kosuge D et al (2002) Vascular endo- demonstrated that in the adult. Cunningham JE. Stacker SA. Nagy JA. McColl BK. Markovic SN. The possibility that genesis to prevent tumour metastasis. glioblastoma multiforme. Hawighorst T. Weber E. Jones PF. Clin Cancer Res 9:5962–5967 Animal Facility at the Ludwig Institute for Cancer Research. Tsujimoto M et al (2005) Lymph vessel bourne. prognosis. Hewett PW. Achen MG (2004) Lymphangiogenic We propose that the location of a primary epithelial growth factors as markers of tumor metastasis. Animal experiments and treatment of archived human of lymph node metastasis. Nat Rev Cancer 2:573–583 MGA are supported by Senior Research Fellowships from the 18. 10. Debinski W. node metastasis and prognosis of patients with colorectal carci- noma. may provide early and accurate prognostic function inhibits tumor cell metastasis. Stacker SA (2006) Lymphatic vessels in cancer metastasis: bridging the gaps. recent studies have 8. J Exp Med 196:1497–1506 the levels of these factors produced at a local level or 7. Nature (Lond) 438:946–953 Australia Fellowship. Scavelli C. 1669–1675 9. 21. and provision of NG2 antibody. Baldwin ME et al (2001) VEGF-D pro- macrophage populations of the mouse embryo [49]. PA cations for T1 colorectal tumors. To motes the metastatic spread of tumor cells via the lymphatics. Shayan R. Further. Lymphology 23:4–14 1. and metastasis in the absence of functional intratumor lymphatics. Nakamura Y. Mayo Clin Proc 82:490–513 Science 296:1883–1886 2. Mann GB. Jackson DG et al (2001) Induction of gested as key pro-lymphangiogenic cells during develop. S. Stacker SA (2006) Targeting lymphangio- information regarding metastatic risk. Achen MG. and the Ludwig Institute for Cancer VEGF-C and VEGF-D at the invasive edge correlates with lymph Research Animal Ethics Committee. Stacker SA (2008) Molecular control of lymphatic Chance Foundation and Royal Australasian College of Surgeons metastasis. Smith KJ. SAS would like to acknowledge the support of the Pfizer development and human disease. Achen MG. Cancer Cell 7:121–127 identifying tumors that are more likely to spread. inflammatory cells such as thelial growth factor-D expression is an independent prognostic macrophages do not recapitulate their developmental role in marker for survival in colorectal carcinoma. White JD. Shayan R. for assistance with mouse experiments. Agliano M et al (2004) Lymphatics at the thelial growth factor-C-mediated lymphangiogenesis promotes crossroads of angiogenesis and lymphangiogenesis. Burke DA et al (2011) Lymphatic vessel and practice of oncology. Welch DR. Nat counter this. Cody for assistance with 16.Clin Exp Metastasis (2013) 30:345–356 355 been studied. Jussila L. In addition. Juri AL. 6. Yasuoka H. Tanaka S et al (2004) Expression of Human Ethics Research Council. Williams RA. Skobe M. Caesar C. tumor lymphangiogenesis by VEGF-C promotes breast cancer metastasis. Diez de Pinos SM et al (1990) Quan- References titation of human melanoma. Karnezis T. as was once hypothesised [50]. despite the presence of node localization of human melanoma metastases. and A. Cancer Res 62: lymphangiogenesis. M. Williams and distribution in the mucosa and submucosa and potential impli- Wilkins. J Anat 204: tumour metastasis.

Adams RH. Egeblad M. Cancer Res 65:4739–4746 48. Hagendoorn J et al (2006) Imaging steps of studies on the invasion of melanomas in initial lymphatics of lymphatic metastasis reveals that vascular endothelial growth human skin. Deutsch A. Alitalo K (2007) Molecular regulation of angiogenesis enlargement and sprouting. Jeltsch M.356 Clin Exp Metastasis (2013) 30:345–356 27. Pajusola K et al (2005) Vascular endothelial cell 122:1765–1772 growth factor receptor 3-mediated activation of lymphatic 47. Baldwin ME. Boswell N et al (2003) Molecular logical. Stadelmann WK. Saaristo A. Skobe M (2003) Lymphatic endothelium: morpho. Cancer Res 66: dermal lymphatic vessel calibre during development by regulat- 8065–8075 ing lymphatic endothelial cell proliferation. Cancer Cell 21:181–195 Growth Factors 30:283–296 34. Hematol Oncol Clin North Am 12:767–796. J Biol Chem 274:32127–32136 123 . collecting lymphatic endothelium. Stenvers K. Stacker SA (2012) Vascular endothelial growth tumor metastasis by regulating prostaglandins produced by the factor-D: signalling mechanisms. Tammela T. metastasis and vessel remodelling. Uutela M et al (2007) Distinct vas- of lymphangiogenesis. Rozen W. Farnsworth RH. Wirzenius M. Karnezis T. Cao Y. J Invest Dermatol 98:64–67 factor-C increases metastasis by increasing delivery of cancer 49. Stacker SA. Rajantie I. molecular and functional properties. biology and clinical relevance. Shayan R et al (2011) A role for 32. Shayan R. Van der Auwera I. Stacker SA. node dissection and transplantation. cation in solid human tumours. Pepper MS. Pollard JW et al (2010) Macrophages define cells to lymph nodes: therapeutic implications. Karnezis T. thelial growth factor-C accelerates diabetic wound healing. endothelial growth factor C promotes tumor lymphangiogenesis lature. Gordon EJ (2012) Deciphering the roles of macro- Factors 25:417–425 phages in developmental and inflammation stimulated lymphan- 38. Proc Natl Acad Sci USA 95:548–553 39. Achen MG. Farkkila A et al (2006) Vascular endo- consensus on the methodology of lymphangiogenesis quantifi. BioEssays 24:1030–1040 cular endothelial growth factor signals for lymphatic vessel 31. Plast Reconstr Surg 35. Nissen S et al (1992) Ultrastructural 36. J Cell Biol 163: and functional analyses of the contractile apparatus in lymphatic 209–213 muscle. Nat Rev Mol Cell Biol 8:464–478 44. Karnezis T. Karkkainen MJ et al (2001) Vascular in ephrinB2 is required for the remodeling of lymphatic vascu. Lubach D. Br J Cancer 95:1611–1625 Am J Pathol 169:1080–1087 28. FASEB J 17:920–922 29. Reintgen DS (1998) Prognosis in malignant endothelium is crucial for tumor cell entry and spread via melanoma. Am J Pathol 170:334–346 weight loss: implications for perforator flaps. Genes Dev 19:397–410 and intralymphatic tumor growth. Achen MG. Harvey NL. Kukk E et al (1998) Vascular endothelial giogenesis. Nat Med 13:1458–1466 Cancer Res 71:6547–6557 33. Tammela T. Muthuchamy M. Vasc Cell 4:15 growth factor D (VEGF-D) is a ligand for the tyrosine kinases VEGF receptor 2 (Flk1) and VEGF receptor 3 (Flt4). Azzali G (2007) Tumor cell transendothelial passage in the 46. Bailey J et al (2005) PDZ interaction site 42. Growth 50. Shayan R. Isaka N. Tsantikos E et al (2007) A system for 3899–3910 quantifying the patterning of the lymphatic vasculature. He Y. Karpanen T. vi lymphatic vessels. Shayan R. Rao S. Cancer Res 61:1786–1790 30. 41. Holopainen T et al (2007) Therapeutic bone morphogenic protein-4 in vascular endothelial growth fac- differentiation and maturation of lymphatic vessels after lymph tor-D mediated tumor growth. Saaristo A. Tammela T. Bernard S et al (2008) Perforator dilatation absorbing lymphatic vessel of transgenic adenocarcinoma mouse induced by body weight gain is not reversed by subsequent prostate. Hoshida T. Gashev A. Makinen T. Gordon EJ. Caesar C et al (2012) VEGF-D promotes 45. Tille JC et al (2006) First international 40. Caesar C et al (1999) Biosynthesis of vascular endothelial growth factor-D involves proteolytic processing which generates non-covalent homodimers. J Exp Med 204:1431–1440 and lymphangiogenesis. Achen MG (2002) Molecular control 43. Development 137: 37. Adams RH.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. .