Developmental Biology 381 (2013) 365–376

Contents lists available at ScienceDirect

Developmental Biology

0012-16http://d

n CorrBiology,German

E-m

journal homepage: www.elsevier.com/locate/developmentalbiology

Maldevelopment of dermal lymphatics in Wnt5a-knockout-mice

Kerstin Buttler a,n, Jürgen Becker a, Tobias Pukrop b, Jörg Wilting a

a Centre of Anatomy, Department of Anatomy and Cell Biology, University Medicine Goettingen, Goettingen, Germanyb Department of Hematology and Oncology, University Medicine Goettingen, Goettingen, Germany

a r t i c l e i n f o

Article history:Received 22 January 2013Received in revised form12 June 2013Accepted 28 June 2013Available online 11 July 2013

Keywords:LymphangiogenesisLymphatic endothelial cellWNT5AROR1RYKFDZ3DKK2FABP4AChELGALS3BPENTPD1

06/$ - see front matter & 2013 Elsevier Inc. Ax.doi.org/10.1016/j.ydbio.2013.06.028

espondence to: Centre of Anatomy, DepartUniversity Medicine Goettingen, Kreuzberg

y. Fax: +49 551 397067.ail address: buttler.kerstin@med.uni-goettinge

a b s t r a c t

Maintenance of tissue homeostasis and immune surveillance are important functions of the lymphaticvascular system. Lymphatic vessels are lined by lymphatic endothelial cells (LECs). By gene micro-arrayexpression studies we recently compared human lymphangioma-derived LECs with umbilical veinendothelial cells (HUVECs). Here, we followed up on these studies. Besides well-known LEC markers,we observed regulation of molecules involved in immune regulation, acetylcholine degradation andplatelet regulation. Moreover we identified differentially expressed WNT pathway components, whichplay important roles in the morphogenesis of various organs, including the blood vascular system. WNTsignaling has not yet been addressed in lymphangiogenesis. We found high expression of FZD3, FZD5 andDKK2 mRNA in HUVECs, and WNT5A in LECs. The latter was verified in normal skin-derived LECs. Withimmunohistological methods we detected WNT5A in LECs, as well as ROR1, ROR2 and RYK in both LECsand HUVECs. In the human, mutations of WNT5A or its receptor ROR2 cause the Robinow syndrome.These patients show multiple developmental defects including the cardio-vascular system. We studiedWnt5a-knockout (ko) mouse embryos at day 18.5. We show that the number of dermal lymphaticcapillaries is significantly lower in Wnt5a-null-mice. However, the mean size of individual lymphaticsand the LEC number per vessel are greater. In sum, the total area covered by lymphatics and the totalnumber of LECs are not significantly altered. The reduced number of lymphatic capillaries indicates asprouting defect rather than a proliferation defect in the dermis of Wnt5a-ko-mice, and identifies Wnt5aas a regulator of lymphangiogenesis.

& 2013 Elsevier Inc. All rights reserved.

Introduction

The maintenance of tissue homeostasis is one of the importantfunctions of the lymphatic vascular system. Developmental defectslike primary lymphedema and lymphangioma lead to functionaldisorders with decreased quality of life for patients. Lymphan-gioma is usually considered as a vascular malformation with foci ofproliferating lymphatic endothelial cells (LECs). Its etiology isunknown (Filston, 1994; Okazaki et al., 2007; Witte et al., 2001).Studies of the last years have identified markers for LECs, e.g., thehyaluronan receptor Lyve-1, the glycoprotein podoplanin and thetranscription factor Prox1 (Banerji et al., 1999; Breiteneder-Geleffet al., 1999; Wigle and Oliver, 1999). Therefore, LECs are now welldistinguishable from blood vascular endothelial cells (BECs). Thedevelopment of specific antibodies provided the opportunity toseparate these two types of endothelial cells by fluorescenceactivating cell sorting and magnetic cell separation (Hirakawaet al., 2003; Kriehuber et al., 2001).

ll rights reserved.

ment of Anatomy and Cellring 36, 37075 Goettingen,

n.de (K. Buttler).

In previous studies we characterized LECs from dermal lym-phangiomas of two children (Norgall et al., 2007). We performedgene microarray analyses in comparison with human umbilicalvein endothelial cells (HUVECs). Approximately 1200 genes wereregulated in LECs. As expected, the above noted lymphendothelialmarkers were highly expressed. We now followed up on thesestudies and focused on molecules, which have not yet beenstudied in the context of lymphangiogenesis. Among these wefound WNT5A (wingless-type MMTV integration site family,member 5A), which was regarded as a regulator of angiogenesis.WNTs belong to a family of secreted glycoproteins that operate viathree pathways with various receptors and co-receptors. Theycontrol fundamental mechanisms during embryonic developmentlike proliferation, migration and cell fate specification (Dale, 1998;Logan and Nusse, 2004). Dysregulation of WNT signaling pathwayshas been observed under pathological conditions, e.g., in cancer,rheumatoid arthritis and Alzheimer′s disease (De Ferrari andInestrosa, 2000; Polakis, 2000; Sen et al., 2000).

In the canonical signaling pathway, WNT proteins bind toFrizzled (FZD) receptors and form a complex with low densitylipoprotein (LDL) receptor-related protein 5 or 6 (LRP5/6) at thecell surface. Hence, the intracellular proteins Dishevelled (Dsh)and Axin become activated with subsequent accumulation of

K. Buttler et al. / Developmental Biology 381 (2013) 365–376366

β-catenin in the cytoplasm and the nucleus. The nuclear responsesinvolve the transcription factor lymphoid enhancer-binding factor1/T cell-specific transcription factor (LEF1/TCF). Lack of WNTligands entails the degradation of β-catenin, controlled by theformation of a complex comprising axin, adenomatous polyposiscoli (APC) and glycogen synthase kinase-3β (GSK3β) (MacDonaldet al., 2009). Additionally, non-canonical Wnt pathways are theWnt/calcium (Ca2+) pathway, which activates phospholipase C,and the planar cell polarity (PCP) pathway, acting through Rhoand Rock (Rho-associated kinase) as well as Rac1 and JNK in aβ-catenin independent manner.

Much of the activity of WNT5A is transduced via non-canonicalpathways and the transmembrane receptors ROR2 (receptor tyr-osine kinase-like orphan receptor 2), ROR1 and RYK (receptor-liketyrosine kinase) (Fukuda et al., 2008; Keeble et al., 2006; Mikelsand Nusse, 2006). Thereby, the non-canonical pathways possessthe ability to inhibit the canonical one. This seems to depend onthe availability and activation status of the ligands and receptors,as well as the involved cell types (Pandur et al., 2002; Seifert andMlodzik, 2007; van Amerongen et al., 2008). Recent studies haveproposed essential roles of Wnts in angiogenic processes, e.g.,vascular morphogenesis, vessel sprouting and elongation (Cironeet al., 2008). However, Wnts have not been investigated inlymphangiogenesis so far. Here, we studied lymphatic vessels ofWnt5a-knockout (ko) mouse embryos in comparison to hetero-zygous and wild type mice. Immunohistological staining wasperformed to characterize the morphology and frequency oflymphatics in the dermis. Number and shape of lymphatic vesselsof the dermis were affected in Wnt5a-ko-mice. Our studies show afunction for Wnt5a in the development of dermal lymphaticcapillaries.

Materials and methods

Animals

Wnt5atm1Amc B6 mice were bought from The Jackson Labora-tory (Bar Harbor, Maine, USA). Heterozygous animals were bred toobtain embryos at embryonic day (ED) 12.5 and 18.5. Wild type,heterozygous and homozygous ko-mice were obtained at nearMendelian rates. Embryos were genotyped by the application ofmultiplex polymerase chain reactions (PCR) with the followingprimers: Wild type forward, 5′GAG GAG AAG CGC AGT CAA TC3′;mutant forward, 5′GCC AGA GGC CAC TTG TGT AG3′ and commonreverse, 5′CAT CTC AAC GGC CTC AT3′. The products possessedfragment lengths of 400 base pairs (bp) for the mutant allele, and480 bp for the wild type allele. PCR was performed under standardconditions and analyzed by ethidium bromide staining of agarosegels (Supplementary Fig. 1).

Isolation of lymphatic endothelial cells and cell culture

LECs were isolated from lymphangiomas of two male patients,10 and 4 months of age. Lymphangiomas were located in theaxillary region of patient A, and in the upper arm of patient B. Thestudies were approved by the hospital's ethics committee andwere performed with the informed consent of the patient'sparents. Explants of endothelial cells were cultured in Petri dishesover 4–6 weeks and, using the cell sweeping procedure two timesper week, contaminating stromal cells were removed. This proce-dure has been described in detail before (Folkman et al., 1979).Purity of the cultures was controlled by CD31 and Prox1 doublestaining, and reached nearly 100% in the studied cultures. Usingstandard procedures, human umbilical vein endothelial cells(HUVECs) were prepared from umbilical cords, with the informed

consent of the parents. Only pure CD31+ cultures were used formicro-array analyses. LECs and HUVECs were maintained in ahumidified incubator at 37 1C and 5% CO2 using EBM-2 mediumand supplements (Lonza, Basel, Switzerland). We added 250 ng/mlhuman VEGF-C to the LEC culture (ReliaTech, Braunschweig,Germany). To compare our lymphangioma-derived LECs with LECsfrom healthy persons, we used foreskin-derived LECs (PromoCell,Heidelberg, Germany). The cells were isolated from juvenile fore-skin and studied in passages 4–5. Cells were cultured in EBM-2medium with VEGF-C as described above. MDA-MB231 breastcancer cells were used as positive controls for real-time RT-PCRexperiments. Cells were cultured in RPMI 1640 medium and 10%FCS (Lonza, Belgium).

RNA micro-array analyses

Here we followed up on micro-array analyses, which werepartially published before (Norgall et al., 2007). Briefly, analyseswere performed at the transcriptome core facility of the UMG(http://www.microarrays.med.uni-goettingen.de) with 44k wholegenome oligo micro-arrays (Agilent Technologies, Santa Clara, CA,USA). For hybridization of the probes, a mixture of Cy5 and Cy3labeled probes was prepared. Per Array 0.7–1.5 mg of Cy-labeledDNA was used. Hybridization was performed according to theAgilent ′60-mer oligo micro-array processing protocol. Differen-tially expressed genes were identified by ANOVA-procedure. Theresulting raw p-values were adjusted with the Benjamini–Hoch-berg method to control the false discovery rate.

RNA isolation and real-time RT-PCR

RNA was prepared from freshly harvested cultured cells usingTri-Reagent solution (Life Technologies, Darmstadt, Germany) andOmniscript reverse transcriptase kit (Qiagen, Hilden, Germany)according to the manufacturers recommendation as describedbefore (Becker et al., 2010). Real time analyses were performedat least in triplicate on an ABI StepOne-plus device (Life Technol-ogies) using the Fast SYBR-Green Master Mix (Life Technologies) assuggested by the manufacturer. The following oligo-nucleotideprimers were used (5′–3′): RYK fwd: agttcgttggatggctcttg, RYK rev:gagttcccacagcgtcactc; ROR1 fwd: caacaaacggcaaggaggtg, ROR1 rev:atcctggacttgcagtggga; WNT5A fwd: ctcgccatgaagaagtccat, WNT5Arev: acattgcacttccagccatc; β-Actin fwd: gcatcccccaagttcacaa,β-Actin rev: aggactgggccattctcctt. Oligo-nucleotides were synthe-sized by IBA (Goettingen, Germany). Relative quantification wascalculated using the δδCt-method and Microsoft Excel Software.β-Actin was used as reference gene and lymphangioma LECs(LEC2) were used as calibrator and set as 1.

Western blot

Cell pellets of HUVECs and LECs were prepared with Laemmlilysis-buffer (Roth, Karlsruhe, Germany) containing 15% β-mercaptoethanol and subjected to SDS–PAGE. They were blottedsemi-dry on a nitrocellulose membranes (Amersham Biosciences,UK). Blocking solution consisting of 5% milk powder in TBS and0.1% Tween 20 (TBS-T) was used to incubate the blots for one hour.Primary antibodies against monoclonal rat anti-human Wnt5a(R&D Systems, #Mab-645, dilution 1:5000), polyclonal rabbitanti-human ROR1 (Cell Signaling Technology, #4102; dilution1:3000), monoclonal mouse anti-human ROR2 (provided by A.Schambony, University Erlangen-Nürnberg, Germany; dilution1:50) and mouse anti-human α-tubulin (Merck Millipore, #05-829; dilution 1:5000) dissolved in TBS-T containing bovineserum albumin (5%) were incubated overnight at 4 1C. Afterrinsing in TBS-T, secondary peroxidase-conjugated anti-rabbit

K. Buttler et al. / Developmental Biology 381 (2013) 365–376 367

IgG, anti-mouse IgG and anti-rat IgG (Santa Cruz Biotechnology,#2004, #2005, #2006; each diluted 1:6000) were applied for onehour followed by rinsing in TBS-T. Antibody detection was per-formed employing ECL prime western blotting detection reagent(GE Healthcare Europe, Freiburg, Germany). Images were capturedwith a luminescence image analyzer (LAS-4000mini, FujifilmEurope, Düsseldorf, Germany).

Histology and immunohistology

For paraffin histology, ED 18.5 mouse embryos were dissectedand placed overnight in Bouin's fixative. Specimens wereembedded in paraffin, sectioned into 8 mm slices and mountedonto slides. Hematoxylin and eosin (HE) or trichrome staining wasperformed. For immunofluorescence staining, ED 12.5 embryosand tissues from ED 18.5 embryos were fixed in 4% paraformalde-hyde, rinsed in phosphate-buffered saline (PBS), transferred to 10%and 30% sucrose in PBS, and embedded in Tissue Freeze Medium(Sakura Finetek Europe, NL). Cryosections were incubated withbovine serum albumin (BSA), and stained with rat anti-mouseCD31/PECAM-1 (1:50; BD Pharmingen, San Diego, USA), rabbitanti-mouse LYVE-1 (1:100; Regeneron, Tarrytown, NY, USA), rabbitanti-human Prox1 (1:500; ReliaTech, Braunschweig, Germany),

Fig. 1. (A and B) Expression of endothelial/lymphendothelial markers in human LECsendothelial marker CD31/PECAM-1 (green) in the cell membrane and the transcription fabut are negative for Prox1. Counterstaining with Dapi (blue). Scale bar in (A) 30 mm. (C) Retypes are: LEC2 (lymphangioma-derived LECs), HUVEC, LEC9 (foreskin-derived LECs),calibrator and set as 1. (D and E) Immunocytology with anti-WNT5A antibodies (green) sScale bar in (E) 30 mm.

rabbit anti-human ROR1 (Cell Signaling Technology, #4102) andmouse anti-human ROR2 (provided by A. Schambony). In negativecontrols primary antibodies were omitted. After rinsing, thesecondary Alexa 488-conjugated goat anti-rat IgG, Alexa 594-conjugated goat anti-mouse IgG and Alexa 594-conjugated goatanti-rabbit IgG (Molecular Probes, Eugene, USA) were applied(1:200). Specimens were counterstained with Dapi (1:40,000; LifeTechnologies, NY, USA) and subsequently mounted under coverslips with Fluoromount-G (Southern Biotechnology Associates,Birmingham, GB). Double staining of human LECs and HUVECswith anti-CD31 and anti-Prox1 antibodies was performed inchamber slides as previously (Norgall et al., 2007). A short(1 min) permeabilization with Triton was implemented after BSAincubation. Histological analyses were done with Axio Imager.Z1(Zeiss, Göttingen, Germany). The macroscopic images of the mouseembryos were made with a Nikon D70 camera.

Quantitative analyses and statistics

Images for quantification were coded and evaluated blindly inmicroscopic fields at 200� magnification (three different animalsfor each group; 14 sections per animal; 3–4 fields per section).Both wild type and heterozygous ko-embryos were used as

and HUVECs. (A) Isolated LECs from lymphangioma patients co-express the pan-ctor Prox1 (red) in the nucleus. (B) HUVECs show higher CD31/PECAM-1 expressional-time RT-PCR of human cell lines with probes against WNT5A, ROR1 and RYK. Celland breast cancer cell line MDA-MB231 as a positive control. LEC2 were used ashows higher expression in LECs than in HUVECs. Counterstaining with Dapi (blue).

K. Buttler et al. / Developmental Biology 381 (2013) 365–376368

controls, since our preliminary studies did not show any differ-ences between these mice. Number of lymphatics, their area andmean size, and the number of Prox1-positive dermal LECs (totalcount app. 2800 cells) were evaluated. For comparisons of themean, the unpaired two-tailed t-test, with Welch's correctionwhere necessary, and the Mann–Whitney-test were used. Statis-tical analyses were carried out with GraphPad Prism 5 software(Vers. 5.03; GraphPad Software, Inc., La Jolla, CA). Data arepresented as mean7SEM. Statistical significance is indicated inthe figures by asterisks.

Fluorescence micro-lymphangiography

To study the functionality of the lymphatics we used ED 18.5mouse embryos and injected 200–300 nl of 2000 kDa FITC-dextran (Sigma; 8 mg/ml PBS) into the paws of control embryosand the most distal part of the legs of the homozygous ko-embryos (n¼12). Embryos were studied with a fluorescencestereomicroscope (Leica). Images were taken at regular intervalsup to 15 min after injection.

Results

RNA-expression of human LECs compared to HUVECs

We have studied the transcriptome of LECs isolated fromlymphangioma tissues in comparison with HUVECs by micro-array analyses (Agilent 44k whole human genome oligo micro-arrays). Our investigations reveal regulation of approximately1200 genes in the LECs. The lymphendothelial character of thecells was confirmed by the prominent expression of molecules,which are characteristic for lymphatic endothelium, such as thetranscription factor Prox1 (Wilting et al., 2002). Its expression wasapproximately 38-fold higher in LECs than in HUVECs. Doublestaining of LECs with anti-CD31/PECAM-1, a pan-endothelialmarker, and anti-Prox1 antibodies exhibited co-expression of bothmarkers in LECs; CD31 in the cell membrane and Prox1 in thenucleus (Fig. 1A), whereas HUVECs only showed CD31-positivity

Fig. 2. (A) Western blot analyses of lysates from lymphangioma-derived LECs and HUVEand tubulin (loading control; ∼50 kDa). WNT5A is found in LECs, the receptors in both oskin from ED 18.5 mouse embryos with antibodies against WNT5A (green) and Prox1 (Scale bar 30 mm.

(Fig. 1B). Other molecules also known to be characteristic of LECswere highly expressed in our two LEC lines, too, like podoplanin(107-fold), Lyve-1 (5-fold), reelin (41-fold) and macrophage man-nose receptor (19-fold). We also found molecules that have notbeen studied intensively in lymphatics, such as fatty acid bindingprotein 4 (FABP4), which was 52-fold higher in LECs. Immunohis-tochemical studies of multiple human tissues preformed with‘The Human Protein Atlas’ (http://www.proteinatlas.org/; seeUhlen et al. (2005)) showed expression of FABP4 in humanlymphatics, whereas FABP5 was found in both lymphatics andblood vessels (Supplementary Fig. 2A and B). Also, acetylcholines-terase (AChE), which has been found in the cell membrane ofHUVECs (Carvalho et al., 2005) was 5.6-fold higher in LECs, andwas expressed in both BECs and LECs (Supplementary Fig. 2C).In one of the two LEC lines we found 17-fold increased expressionof LGALS3BP (lectin, galactoside-binding, soluble, 3 binding pro-tein¼MAC2BP), which binds the macrophage associated lectinLGALS3 (Koths et al., 1993). We observed protein expression ofLGALS3BP in human lymphatics (Supplementary Fig. 2D). Incontrast, in HUVECs we found the highest (240-fold) expressionfor ENTPD1 (Ectonucleoside triphosphate diphosphorylase 1,CD39), an inhibitor of platelet activation (Enjyoji et al., 1999),and observed immunopositivity in human blood vessels, but not inlymphatics (Supplementary Fig. 2E). These data show that ourmicro-array analyses produced reliable results.

We then focused on the members of the WNT signaling path-way. WNT5A showed a significant 16-fold higher expression in thetwo lymphangioma-derived LEC lines than in HUVECs, whereasother family members (WNT1, WNT6, WNT7a, WNT7b, WNT8b,WNT11) were almost equally expressed. To study if the highWNT5A levels in our LECs might be pathological, we comparedthem to normal foreskin-derived LECs with qRT-PCR. The normalLECs showed an even higher value (22% more) than the lymphan-gioma LECs (Fig. 1C). The different WNT5A expression levels ofLECs and HUVECs were also evident in immunocytological ana-lyses (Fig. 1D and E).

In the micro-array analyses, the receptors of the WNT signalingpathway ROR1, ROR2, RYK and LRP5/6 displayed no significantdifferences between LECs and HUVECs, whereas FZD3 and FZD5

Cs with antibodies against WNT5A (∼42 kDa), ROR1 (∼135 kDa), ROR2 (∼135 kDa),f the cell lines. Positions of size markers are indicated. (B) Immunohistology of thered) shows expression of WNT5a in lymphatics. Counterstaining with Dapi (blue).

Fig. 3. Immunostaining of adult mouse dermis with: (A) anti-CD31 and (B) anti-Ror2 antibodies. (C) Merged picture. Ror2 is expressed in blood vessels (arrows) and invessels that appear to be lymphatics (arrowhead). Scale bar in (C) 25 mm. Staining with (D) lymphendothelial marker anti-podoplanin and (E) anti-Ror2 antibodies. (F)Merged picture. Ror2 is expressed in lymphatic vessels (arrowhead), and in adjacent hair follicles. Scale bar in (F) 20 mm.

Fig. 4. Immunohistochemical staining of human tissues. Taken from 〈www.proteinatlas.org〉. (A and B) Antibodies against ROR2 show staining of blood vascular andlymphatic (L) endothelium in (A) small intestine (female 56 years, pat id 3343) and (B) skin (male 83 years, pat id 2156). (C) Antibodies against RYK show staining of bloodvascular and lymphatic (L) endothelium in colon (female 62 years, pat id 3753). Scale bar in (C) 25 mm.

K. Buttler et al. / Developmental Biology 381 (2013) 365–376 369

were 2.5-fold higher in HUVECs. The qRT-PCR studies confirmedequal expression of ROR2 (not shown) and RYK, while ROR1 wasconsiderably higher in HUVECs (Fig. 1C). Correspondingly, inwestern blot analyses we found WNT5A in LECs but hardly anyin HUVECs, while both cell lines expressed ROR1 and ROR2(Fig. 2A). We used these antibodies in human foreskin and in thedermis of mice. In the dermis of ED 18.5 mouse embryos we foundexpression of Wnt5a in lymphatics (Fig. 2B). We could not detectany signals for Ror1 in the vasculature, but found Ror2 in CD31-positive vessels of the mouse dermis (Fig. 3A–C). Since CD31 isexpressed in both blood vessels and lymphatics, we applied theLEC marker podoplanin and observed its colocalization with Ror2in lymphatics (Fig. 3D–F).

Since our antibodies did not reveal staining of human foreskin,we used ‘The Human Protein Atlas’ (http://www.proteinatlas.org/;see Uhlen et al. (2005)). We observed ROR2 signals in theendothelium of numerous fetal and adult tissues. Thereby bothblood vessels and lymphatics were positive in various organs(Fig. 4A and B). The WNT5A co-receptor RYK was also found inboth types of vessels (Fig. 4C). Additionally, in human dermis FZD3was expressed in blood vessel endothelium, but hardly in lym-phatics (Supplementary Fig. 3A). In our gene micro-array analysesthe WNT modulator Dickkopf homologue 2 (DKK2) was signifi-cantly higher (6-fold) in HUVECs, and we observed immunostain-ing with anti-DKK2 in blood vessels of the human placenta and invarious organs (Supplementary Fig. 3B and C) [Fig. 4A–C and

Supplementary Fig. 3A–C were photographed from http://www.proteinatlas.org/].

The inactivation of Wnt2, Wnt5a, Fzd4, Fzd5, Wnt7a+Wnt7beach induces blood vascular defects (Dejana, 2010). By now,functions of Wnt signaling have not been studied in lymphangio-genesis. Since we observed expression of WNT5A in human andmurine LECs and found its receptors ROR1, ROR2 and RYK incultured LECs and in the lymphatics of various organs, we studiedthe development of lymphatics in Wnt5a-ko-mice.

Morphology of Wnt5a-knockout-mice

Homozygous Wnt5a-ko-mice die postnatally. Therefore, weinvestigated the lymphatic network at embryonic day (ED) 18.5.(Preliminary studies at ED 12.5 did not show obvious differences ofthe jugular lymph sacs and we therefore concentrated on the laterstages). Macroscopically, heterozygous ko-embryos appeared nor-mal, whereas homozygous ko-embryos exhibited several defectssuch as a reduced body length, shortening of the mandible,head, limbs and tail (Fig. 5A), as shown before (Yamaguchi et al.,1999). Both small and large intestine of the ko-mice showed asevere reduction in size (Supplementary Fig. 4A), which hasrecently been attributed to a massive decrease of enterocyteproliferation (Cervantes et al., 2009). Despite this massive dyspla-sia there were no signs of intestinal edema, and anti-Lyve1staining of the intestinal lymphatics did not show any obvious

Fig. 5. Wnt5a heterozygous and homozygous ko-mice at embryonic day (ED) 18.5. (A) Lateral view shows massive defects of the homozygous ko-animal (right) such asreduced body length, shortening of the mandible, head, limbs and tail. Heterozygous embryo exhibits normal phenotype (left). (B and C) The skin of the ko-mouse seems tobe modified and indicates a deep transversal crease in the abdominal skin (arrow). Scale bar 500 mm.

K. Buttler et al. / Developmental Biology 381 (2013) 365–376370

differences between homo- and heterozygous ko-mice (Supple-mentary Fig. 4B and C). During the macroscopic evaluation of theED 18.5 embryos, the skin attracted our attention. Whereas theoverall size of the homozygous ko-embryos and the anlagen ofseveral organs were severely reduced, the skin appeared to be toolarge for the embryos, and there existed a deep transversal creasein the abdominal skin (Fig. 5B and C).

Morphology and lymphatics of the skin

We isolated skin from the dorsal thoracic region and specimensfrom the abdominal body wall (consisting of skin, abdominal

muscle and peritoneum). The latter were stained with trichrometo investigate the microscopic structure. The wild type andheterozygous animals showed a regular arrangement of epidermisand dermis, abdominal muscle with an outer and an inner fascia,and peritoneum. Typically, in the dermis a dermal muscle wasfound (Fig. 6A and B). This morphology was severely altered in thehomozygous ko-mice. The connective tissue compartment wasless well organized, showed no signs of fascia development, andappeared to be increased in size mainly on the expense of theabdominal muscles, which showed signs of degeneration (Fig. 6Cand D). In control embryos, a regular arrangement of abdominalskeletal muscles was present, as well as a layer of dermal muscle

Fig. 6. Trichrome staining of the abdominal body wall of ED 18.5 mice. (A) Wild-type embryos possess a regular arrangement of epidermis and dermis with dermal muscle(dm), abdominal muscles (am), outer fascia (of) and inner fascia (if). (B) Higher magnification of the boxed area in (A) showing a dense plexus of vessels immediatelyunderneath the epidermis (arrows in (A) and (B)). (C) In Wnt5a-null-mice the structure of the body wall is altered. The dermal muscle (dm) appears to be normal, but thereare massive signs of degeneration of the abdominal muscles (am). (D) Higher magnification of the boxed area in (C). Dermal vessels are sparse, but vessels with a large lumen(arrowhead in (C) and (D)) are seen in the dermis underneath the dermal muscle (dm). Scale bar in (C) 20 mm, (D) 10 mm.

K. Buttler et al. / Developmental Biology 381 (2013) 365–376 371

(Fig. 6A and B). The dermal muscle was almost normally devel-oped in the homozygous ko-mice, whereas the abdominal musclesshowed signs of degeneration (Fig. 6C and D). Immediately under-neath the epidermis, above the dermal muscle, a well-developedvascular plexus was seen in control embryos. This was largelyabsent in the Wnt5a-ko-mice, and we observed vessels with alarge lumen in the deeper parts of the dermis, below the dermalmuscle (Fig. 6C and D).

In order to identify lymphatic networks we performed immu-nohistological staining of dermis, taken from the dorsal thoracicregion and the abdominal region of ED 18.5 mice, with LEC-specific antibodies. Using antibodies against CD31 and Lyve-1,we detected a regular pattern of small lymphatic capillaries in thethoracic skin of control mice (Fig. 7A–C). In contrast, homozygousko-mice possessed lymphatics with a large lumen, which becameevident by staining with CD31 and Lyve-1 (Fig. 7D–F), as well asCD31 and Prox1 (Fig. 7G). In the abdominal dermis, identicalmorphological alterations were evident. Small initial Lyve1+ lym-phatics were found at regular intervals in the dermis of wild typeand heterozygous ko-mice (Fig. 8A). In contrast, homozygous ko-mice possessed larger lymphatic vessels located more deeply inthe dermis (Fig. 8B). Blood vessels were Lyve-1-negative, butexpressed CD31/PECAM-1 more strongly than the lymphatics.We obtained the same results with antibodies against Prox1 andCD31. Lymphatics co-expressed the two markers. Again, numeroussmall lymphatics were found in the dermis of control mice, whilefewer but larger lymphatics were seen in Wnt5a-null-mice (Fig. 8Cand D). There obviously is a generalized defect in the developmentof dermal lymphatic capillary networks in Wnt5a-ko-mice.

Quantitative analyses of dermal lymphatics

The immunostaining of the dermal lymphatics allowed us toquantify their number and size, as well as the number of Prox1/CD31-positive LECs. Since heterozygous-ko mice showed no

quantitative differences to wild type mice, we combined bothgenotypes into one control group. We found that the number ofdermal lymphatics was significantly (po0.05) reduced in Wnt5a-ko-mice compared to control mice (Fig. 9A). However, the area oflymphatics in the dermis did not differ between the two groups(Fig. 9B), suggesting major alterations of the size of individuallymphatics in Wnt5a-ko-embryos. In fact, our data reveal asignificant increase (po0.0001) in the mean size of the dermallymphatics in these mice (Fig. 9C). We also counted the Prox1/CD31+ LECs to study whether the reduced number of lymphaticsin the Wnt5a-null-mice might be due to a proliferation defect.Interestingly, the number of LECs did not discriminate betweenthe two groups, whereas the mean number of endothelial cells perlymphatic vessel was significantly (po0.0001) higher in Wnt5a-ko-embryos (Fig. 9D and E). The data show that despite a nearlynormal number of dermal LECs, the number of lymphaticswas significantly lower in Wnt5a-ko-mice, suggesting a defect insprouting.

Function of dermal lymphatics

To study the function of dermal lymphatics we injected2000 kDa FITC-dextran into the paws of the ED 18.5 embryos. ThisFITC-dextran is taken up by the lymphatics and transportedcentripetally (Hirashima et al., 2008). Since the homozygousko-embryos do not develop paws (see Fig. 5), we injected themarker into the most distal part of the legs. In both wild-type andheterozygous ko-embryos we observed an immediate uptake ofthe marker into the local lymphatic network and the transportwithin lymphatic collectors, which are located adjacent to themain blood vessels of the limbs (Fig. 10A and B). In Wnt5a-nullembryos the uptake of FITC-dextran into the lymphatic capillarieswas greatly delayed and transport in lymphatic collectors was notobserved (Fig. 10C). This shows that both the uptake of interstitial

Fig. 7. Immunostaining of dermal lymphatics in the dorsal thoracic region of ED 18.5 wild-type vs. Wnt5a-ko-mice. Counterstaining of nuclei with Dapi (blue). (A–C)Expression of CD31 (green) and Lyve-1 (red). (A) Merged picture. In wild-type embryos numerous small lymphatics of the dermis (arrows) co-express CD31 and Lyve-1.(B) CD31, (C) Lyve-1. Scale bar in (A) 30 mm. (D–F) Huge but irregularly distributed lymphatics in homozygous ko-mice also express CD31 and Lyve-1 (arrows). Scale bar in(D) 40 mm. (G) Staining with antibodies against CD31 (green) and Prox1 (red). Note lymphatics (arrows) with large lumen. Scale bar 45 mm.

Fig. 8. Immunostaining of dermal lymphatics in the abdominal region of ED 18.5 Wnt5a-ko-mice and controls. (A and B) Expression of CD31 (green) and Lyve-1 (red). (A) Inheterozygous embryos numerous small lymphatics of the dermis (arrows) co-express CD31 and Lyve-1. (B) Lymphatics (arrow) of homozygous ko-mice are considerablylarger, but fewer, and located more deeply in the dermis. Nuclear staining with Dapi (blue). Scale bar in (A) 50 mm. (C and D) Staining with antibodies against CD31 (green)and Prox1 (red) (C) Heterozygous embryos possess a regular pattern of initial lymphatics (arrows) in the dermis, with co-expression of CD31 in the cell membrane and Prox1in the nucleus. (D) The huge but irregularly distributed lymphatics in homozygous ko-mice also express both markers (arrows). Counterstaining of nuclei with Dapi (blue). Ep—epidermis; Scale bar in (C) 50 mm.

K. Buttler et al. / Developmental Biology 381 (2013) 365–376372

Fig. 9. Quantitative analyses of number, area and mean size of dermal lymphatics, and the number of Prox1-positive dermal LECs in ED 18.5 embryos. (A) The number oflymphatics per microscopic field (200� magnification) is significantly lower in Wnt5a-null-embryos compared to controls. (B) The areas covered by lymphatics do not differsignificantly. (C) The mean size of individual dermal lymphatics is significantly larger in Wnt5a-ko-mice. (D) There is no statistically significant difference in the total numberof Prox1-positive dermal LECs per microscopic field (200� magnification). (E) The mean number of Prox1-positive LECs per vessel is significantly higher in Wnt5a-ko-mice.Results represent means7SEM; *po0.05; **po0.01; ***po0.001.

K. Buttler et al. / Developmental Biology 381 (2013) 365–376 373

fluid and the transport of lymph are impaired in Wnt5a-ko-embryos.

Discussion

Characteristics of the transcriptome of human BECs and LECs

We have isolated and purified LECs from the dermis of twolymphangioma patients (Norgall et al., 2007) and compared theirtranscriptome with that of freshly isolated HUVECs. One has to beaware that molecular differences have been observed betweenin vitro and ex vivo endothelial cells (Amatschek et al., 2007; Wicket al., 2007). Nevertheless, in LECs we found high expression ofgenes that are well-known LEC markers, such as vascular endothe-lial growth factor receptor (VEGFR)-3, Prox1, podoplanin, reelinand Lyve-1 (Hirakawa et al., 2003; Kriehuber et al., 2001; Norgallet al., 2007; Becker et al., 2012; Podgrabinska et al., 2002; Wiltinget al., 2000), suggesting that our micro-array data are reliable. Thisis further supported by the fact that we found many of thetranscriptionally regulated genes expressed as proteins (www.proteinatlas.org) in BECs or LECs; e.g., fatty acid binding protein4 (FABP4), acetylcholinesterase (AChE), LGALS3BP (lectin, galacto-side-binding, soluble, 3 binding protein¼MAC2BP), Ectonucleo-side triphosphate diphosphorylase 1 (ENTPD1, CD39).

FABP4 is found with high specificity in LECs. It is a smallcytoplasmic protein that controls metabolic and inflammatorypathways (Furuhashi et al., 2007). Under specific diet, the totalweight gain in FABP4 �/� mice is higher than that in wild typecontrols (Hotamisligil et al., 1996). AChE is a very rapid enzymethat hydrolyzes the neurotransmitter acetylcholine. Outside thenervous system, AChE has been found in the cell membrane ofHUVECs (Carvalho et al., 2005). The expression in HUVECs isconfirmed by our studies, however, we observed even higherlevels in LECs. The functions of AChE in LECs remain to be studied,however, the close spatial relation between peripheral nerves and

the lymphatics, as well as the fact that high systemic AChE levelsare toxic, may provide an explanation for this finding.

We found high expression of LGALS3BP in LECs. LGALS3BP is a90 kDa protein most highly expressed in the gastrointestinal tractand up-regulated in the serum of patients with AIDS (Ullrich et al.,1994). The stimulatory effect of LGALS3BP on natural killer cellsand lymphokine-activated killer cells (Ullrich et al., 1994) impliesan active role for LECs in the immune defense. The reliability of ourtranscriptome analyses is also supported by the finding ofENTPD1/CD39 in HUVECs but not in LECs. CD39 is an inhibitor ofplatelet activation (Enjyoji et al., 1999). It hydrolyzes extracellularATP and ADP to AMP, which is further converted to adenosine bythe enzyme 5′-nucleotidase. ADP is a powerful agonist for plateletrecruitment and adhesion, and becomes immediately degraded byCD39 in BECs. On the other hand, adenosine is a potent antagonistof platelet activation (Kaarbo et al., 2003) and is produced by5′-nucleotidase, which, notably, was the first specific markeridentified for LECs (Ji and Kato, 2001). Together, these data showthat both BECs and LECs can inhibit platelet activation, althoughwith different enzymes. This may explain why lympho-venousanastomoses, which are produced surgically for the treatment oflymphedema patients, can be functionally active for a long time(Ingianni, 2003).

During our further analyses we then focused on the membersof the WNT-signaling pathway, because several of the familymembers play crucial roles in hemangiogenesis (Dejana, 2010),but have not been studied during lymphangiogenesis.

Wnt signaling in hemangiogenesis and lymphangiogenesis

Our micro-array analyses provided evidence for a significantlyhigher expression of FZD3, FZD5 and DKK2 in HUVECs, while wedetected a 16-fold higher value for WNT5A in lymphangioma-derived LECs. We detected even higher levels of WNT5A in normalforeskin-derived LECs. As regards the expression of the receptorsROR1 and RYK we found equal values in normal and lymphangioma

Fig. 10. Fluorescence micro-lymphangiography of ED 18.5 embryos. Injection ofFITC-dextran into the paw of the hind limb. (A) Transport of the marker inlymphatic collectors after 2 min in wild-type mice. (B) Transport of the marker inlymphatic collectors after 2 min in heterozygous ko-mice. (C) No transport inlymphatic collectors is visible in Wnt5a-null mice after 10 min. There is someuptake of the marker in proximal lymphatic networks.

K. Buttler et al. / Developmental Biology 381 (2013) 365–376374

LECs. Using immunohistology, we found WNT5A in murine dermallymphatics and in cultured human LECs, but not in HUVECs. TheWNT5A receptors ROR2 and RYK were found on both blood vesselsand lymphatics.

WNT5A is a secreted glycoprotein and a member of a family of19 ligands binding at least 10 Frizzled receptors and several co-receptors. WNTs are highly conserved and essential in develop-mental processes like proliferation, migration, differentiation andapoptosis (Logan and Nusse, 2004). These are also important stepsin the formation and maintenance of the vascular system (Dejana,2010). In vitro, the expression of WNTs has been demonstrated inprimary BECs of human and mouse origin (Goodwin et al., 2006;Wright et al., 1999). These investigation identified mRNA of nearlyall WNTs (except for WNT1, 3A, 7B, 8B, 16) as well as theirreceptors and co-receptors in HUVECs, human microvascularendothelial cells (which are a mixture of BECs and LECs), andhuman/mouse aortic smooth muscle cells. In vivo, expression hasbeen shown in the embryonic yolk sac (Wnt5a, Wnt10b) and infetal vessels of the mouse placenta (Wnt2) (Ishikawa et al., 2001;Monkley et al., 1996). Wnt2-null-mice have an abnormal placentalvasculature with a reduced number of blood capillaries (Monkleyet al., 1996).

WNT5A and its receptors in angiogenesis and lymphangiogenesis

Our studies have shown a significant effect of Wnt5a on dermallymphatics. Macroscopically the dermis of ED 18.5 mice appearsoversized, obviously due to increased connective tissue spaces andaltered vascularization. Other organs did not reveal this kind ofdysplasia, showing heterogeneity of Wnt5a-functions in differentorgan systems. Such heterogeneity has also been found in terms ofhemangiogenesis, where data on Wnt5a are highly divers andobviously context dependant. Transfection of WNT5A into HUVECshas not revealed any proliferative effects or increased stimulationof β-catenin/TCF activity in these cells, whereas Wnt1 raised cellproliferation, stabilization of β-catenin and transcription of repor-ter genes (Wright et al., 1999). However, other studies havereported that WNT5A induces proliferation of HUVECs andHDMECs (Cheng et al., 2008; Masckauchan et al., 2006), butreduces the proliferation of aortic ECs (Goodwin et al., 2007). Stillothers have not observed any effects of WNT5A on angiogenesis e.g., in squamous cell carcinoma (Huang et al., 2005). Our studiesshow significantly higher expression of WNT5A in dermallymphangioma-derived and foreskin LECs as compared to HUVECs.Accordingly, we found maldevelopment of dermal lymphatics inWnt5a-ko-mice. We studied these mice at ED 18.5 because theydie postnatally due to hypoplasia of multiple organ systems(Yamaguchi et al., 1999).

In contrast to the hypoplasia of multiple organ systems, theskin of Wnt5a-null mouse embryos appeared oversized. In theabdominal body wall we observed maldevelopment of varioustissues, e.g., disaggregation of the abdominal skeletal muscle andthe absence of the inner and outer muscular fascia. These spaceswere filled with loose connective tissue. Hyperplasia of adiposeand connective tissue is a typical sequela of lymphedema patients(Witte et al., 2001), but we must consider that multiple tissuealterations take place simultaneously in Wnt5a-null mice. Thereasons for the muscular degeneration are not clear. Cell autono-mous effects are possible, since Wnt5a has been shown to increasesatellite cell proliferation in adult myogenesis (Otto et al., 2008).

The vascular pattern in the dermis of Wnt5a-ko-mice isseverely altered. Instead of the dense plexus of subepidermalvessels, there are dilated vessels in deeper positions (below thedermal muscle) of the skin. The number of initial lymphatics in thedermis is significantly reduced in Wnt5a-null-mice. In contrast,the area covered by lymphatics is the same as in control embryos,due to an increased mean diameter of individual vessels. The totalnumber of LECs, as measured by Prox1/CD31 double staining, isnearly the same as in the control embryos. This shows that Wnt5aregulates the morphogenesis of dermal lymphatics, and does not

K. Buttler et al. / Developmental Biology 381 (2013) 365–376 375

seem to have significant effects on the proliferation of dermalLECs. There obviously is defective sprouting of dermal lymphaticsin Wnt5a-null-mice resulting in fewer but larger vessels withmore LECs per vessel. The functionality of the initial lymphatics isgreatly reduced, as shown by the delay of FITC-dextran uptake andthe absence of centrifugal transport. We cannot yet comment onthe existence of lymphatic collectors in the Wnt5a-ko-mice. Ofnote, canonical Wnt-signaling via β-catenin enhances the expres-sion of the transcription factor Foxc2 in a model of skeletal muscledevelopment (Savage et al., 2010). FOXC2 regulates the differentia-tion of lymphatic collectors and valves, and its loss-of-functionmutations cause lymphedema–distichiasis (Fang et al., 2000). Theexistence of quadruple FOX-binding sites in the highly conservedintron 1 of the WNT5A gene (Katoh, 2009) may indicate afunctional loop.

Robinow syndrome

The Wnt5a-ko-mice recapitulate many features characteristicfor the Robinow syndrome. This is caused by autosomal-recessiveor compound-heterozygous mutations of the ROR2 gene, or byautosomal-dominant heterozygous mutations of the WNT5A gene(Person et al., 2010; van Bokhoven et al., 2000). Robinow patientspossess abnormal morphogenesis of the face and external genita-lia, along with short-limbed dwarfism and vertebral segmentationanomalies. Additionally, pulmonary atresia with ventricular septaldefects have been observed (Webber et al., 1990). Yet, there are noreports on the structure and function of the lymphatic vascularsystem in the affected patients, and they do obviously not sufferfrom generalized edema. However, considering the massivelythickened dermis of the forearm of numerous affected infants(see 〈http://www.robinow.org/〉) it appears possible that thereis a localized failure of lymph flow in this region. Therefore, ourdata obtained in mice may hold true for the human.

Acknowledgments

We thank Mrs. Ch. Zelent, Mrs. S. Schwoch, Mrs. M. Schaffrinski,Mr. B. Manshausen and Mr. M. Schulz for their excellent technicalassistance. Many thanks to Dr. A. Schambony (Department ofBiology, University Erlangen-Nürnberg, Germany) for providingthe mouse anti-human ROR2 antibody and to Mr. R. Dungan fortaking makro-photos of the mouse embryos. We are grateful toDr. K. Lange, Department of Medical Statistics, University MedicineGöttingen, for support of the statistical analyses. We thank Drs. G.Salinas-Riester and L. Opitz from the transcriptome-analysis corefacility of the UMG, Göttingen, for their help. The studies weresupported by the German Research Counsil, DFG (Project 2 of theForschergruppe 942; BI 703/3-1).

Appendix A. Supporting information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.ydbio.2013.06.028.

References

Amatschek, S., Kriehuber, E., Bauer, W., Reininger, B., Meraner, P., Wolpl, A.,Schweifer, N., Haslinger, C., Stingl, G., Maurer, D., 2007. Blood and lymphaticendothelial cell-specific differentiation programs are stringently controlled bythe tissue environment. Blood 109, 4777–4785.

Banerji, S., Ni, J., Wang, S.X., Clasper, S., Su, J., Tammi, R., Jones, M., Jackson, D.G.,1999. LYVE-1, a new homologue of the CD44 glycoprotein, is a lymph-specificreceptor for hyaluronan. J. Cell Biol. 144, 789–801.

Becker, J., Pavlakovic, H., Ludewig, F., Wilting, F., Weich, H.A., Albuquerque, R.,Ambati, J., Wilting, J., 2010. Neuroblastoma progression correlates with down-regulation of the lymphangiogenesis inhibitor sVEGFR-2. Clin. Cancer Res. 16,1431–1441.

Becker, J., Frohlich, J., Perske, C., Pavlakovic, H., Wilting, J., 2012. Reelin signalling inneuroblastoma: migratory switch in metastatic stages. Int. J. Oncol 41, 681–689.

Breiteneder-Geleff, S., Soleiman, A., Kowalski, H., Horvat, R., Amann, G., Kriehuber,E., Diem, K., Weninger, W., Tschachler, E., Alitalo, K., Kerjaschki, D., 1999.Angiosarcomas express mixed endothelial phenotypes of blood and lymphaticcapillaries: podoplanin as a specific marker for lymphatic endothelium. Am. J.Pathol. 154, 385–394.

Carvalho, F.A., Graca, L.M., Martins-Silva, J., Saldanha, C., 2005. Biochemicalcharacterization of human umbilical vein endothelial cell membrane boundacetylcholinesterase. FEBS J. 272, 5584–5594.

Cervantes, S., Yamaguchi, T.P., Hebrok, M., 2009. Wnt5a is essential for intestinalelongation in mice. Dev. Biol. 326, 285–294.

Cheng, C.W., Yeh, J.C., Fan, T.P., Smith, S.K., Charnock-Jones, D.S., 2008. Wnt5a-mediated non-canonical Wnt signalling regulates human endothelial cellproliferation and migration. Biochem. Biophys. Res. Commun. 365, 285–290.

Cirone, P., Lin, S., Griesbach, H.L., Zhang, Y., Slusarski, D.C., Crews, C.M., 2008. A rolefor planar cell polarity signaling in angiogenesis. Angiogenesis 11, 347–360.

Dale, T.C., 1998. Signal transduction by the Wnt family of ligands. Biochem. J 329 (Pt2), 209–223.

De Ferrari, G.V., Inestrosa, N.C., 2000. Wnt signaling function in Alzheimer's disease.Brain Res. Brain Res. Rev. 33, 1–12.

Dejana, E., 2010. The role of wnt signaling in physiological and pathologicalangiogenesis. Circ. Res. 107, 943–952.

Enjyoji, K., Sevigny, J., Lin, Y., Frenette, P.S., Christie, P.D., Esch 2nd, J.S., Imai, M.,Edelberg, J.M., Rayburn, H., Lech, M., Beeler, D.L., Csizmadia, E., Wagner, D.D.,Robson, S.C., Rosenberg, R.D., 1999. Targeted disruption of cd39/ATP dipho-sphohydrolase results in disordered hemostasis and thromboregulation. Nat.Med. 5, 1010–1017.

Fang, J., Dagenais, S.L., Erickson, R.P., Arlt, M.F., Glynn, M.W., Gorski, J.L., Seaver, L.H.,Glover, T.W., 2000. Mutations in FOXC2 (MFH-1), a forkhead family transcrip-tion factor, are responsible for the hereditary lymphedema–distichiasis syn-drome. Am. J. Hum. Genet. 67, 1382–1388.

Filston, H.C., 1994. Hemangiomas, cystic hygromas, and teratomas of the head andneck. Semin. Pediatr. Surg. 3, 147–159.

Folkman, J., Haudenschild, C.C., Zetter, B.R., 1979. Long-term culture of capillaryendothelial cells. Proc. Natl. Acad. Sci. USA 76, 5217–5221.

Fukuda, T., Chen, L., Endo, T., Tang, L., Lu, D., Castro, J.E., Widhopf II, G.F., Rassenti, L.Z., Cantwell, M.J., Prussak, C.E., Carson, D.A., Kipps, T.J., 2008. Antisera inducedby infusions of autologous Ad-CD154-leukemia B cells identify ROR1 as anoncofetal antigen and receptor for Wnt5a. Proc. Natl. Acad. Sci. USA 105,3047–3052.

Furuhashi, M., Tuncman, G., Gorgun, C.Z., Makowski, L., Atsumi, G., Vaillancourt, E.,Kono, K., Babaev, V.R., Fazio, S., Linton, M.F., Sulsky, R., Robl, J.A., Parker, R.A.,Hotamisligil, G.S., 2007. Treatment of diabetes and atherosclerosis by inhibitingfatty-acid-binding protein aP2. Nature 447, 959–965.

Goodwin, A.M., Sullivan, K.M., D'Amore, P.A., 2006. Cultured endothelial cellsdisplay endogenous activation of the canonical Wnt signaling pathway andexpress multiple ligands, receptors, and secreted modulators of Wnt signaling.Dev. Dyn. 235, 3110–3120.

Goodwin, A.M., Kitajewski, J., D'Amore, P.A., 2007. Wnt1 and Wnt5a affectendothelial proliferation and capillary length; Wnt2 does not. Growth Factors25, 25–32.

Hirakawa, S., Hong, Y.K., Harvey, N., Schacht, V., Matsuda, K., Libermann, T., Detmar,M., 2003. Identification of vascular lineage-specific genes by transcriptionalprofiling of isolated blood vascular and lymphatic endothelial cells. Am. J.Pathol. 162, 575–586.

Hirashima, M., Sano, K., Morisada, T., Murakami, K., Rossant, J., Suda, T., 2008.Lymphatic vessel assembly is impaired in Aspp1-deficient mouse embryos. Dev.Biol. 316, 149–159.

Hotamisligil, G.S., Johnson, R.S., Distel, R.J., Ellis, R., Papaioannou, V.E., Spiegelman,B.M., 1996. Uncoupling of obesity from insulin resistance through a targetedmutation in aP2, the adipocyte fatty acid binding protein. Science 274,1377–1379.

Huang, C.L., Liu, D., Nakano, J., Ishikawa, S., Kontani, K., Yokomise, H., Ueno, M.,2005. Wnt5a expression is associated with the tumor proliferation and thestromal vascular endothelial growth factor—an expression in non-small-celllung cancer. J. Clin. Oncol. 23, 8765–8773.

Ingianni, G., 2003. Microsurgical lympho-venous anastomosis in the treatment ofsecondary lymphoedema of the upper extremity. Handchir. Mikrochir. Plast.Chir. 35, 216–220.

Ishikawa, T., Tamai, Y., Zorn, A.M., Yoshida, H., Seldin, M.F., Nishikawa, S., Taketo, M.M.,2001. Mouse Wnt receptor gene Fzd5 is essential for yolk sac and placentalangiogenesis. Development 128, 25–33.

Ji, R.C., Kato, S., 2001. Histochemical analysis of lymphatic endothelial cells inlymphostasis. Microsc. Res. Tech. 55, 70–80.

Kaarbo, M., Crane, D.I., Murrell, W.G., 2003. RhoA is highly up-regulated in theprocess of early heart development of the chick and important for normalembryogenesis. Dev. Dyn. 227, 35–47.

Katoh, M., 2009. Transcriptional mechanisms of WNT5A based on NF-kappaB,Hedgehog, TGFbeta, and Notch signaling cascades. Int. J. Mol. Med. 23, 763–769.

Keeble, T.R., Halford, M.M., Seaman, C., Kee, N., Macheda, M., Anderson, R.B.,Stacker, S.A., Cooper, H.M., 2006. The Wnt receptor Ryk is required for

K. Buttler et al. / Developmental Biology 381 (2013) 365–376376

Wnt5a-mediated axon guidance on the contralateral side of the corpuscallosum. J. Neurosci. 26, 5840–5848.

Koths, K., Taylor, E., Halenbeck, R., Casipit, C., Wang, A., 1993. Cloning andcharacterization of a human Mac-2-binding protein, a new member of thesuperfamily defined by the macrophage scavenger receptor cysteine-richdomain. J. Biol. Chem. 268, 14245–14249.

Kriehuber, E., Breiteneder-Geleff, S., Groeger, M., Soleiman, A., Schoppmann, S.F.,Stingl, G., Kerjaschki, D., Maurer, D., 2001. Isolation and characterization ofdermal lymphatic and blood endothelial cells reveal stable and functionallyspecialized cell lineages. J. Exp. Med. 194, 797–808.

Logan, C.Y., Nusse, R., 2004. The Wnt signaling pathway in development anddisease. Annu. Rev. Cell Dev. Biol. 20, 781–810.

MacDonald, B.T., Tamai, K., He, X., 2009. Wnt/beta-catenin signaling: components,mechanisms, and diseases. Dev. Cell 17, 9–26.

Masckauchan, T.N., Agalliu, D., Vorontchikhina, M., Ahn, A., Parmalee, N.L., Li, C.M.,Khoo, A., Tycko, B., Brown, A.M., Kitajewski, J., 2006. Wnt5a signaling inducesproliferation and survival of endothelial cells in vitro and expression of MMP-1and Tie-2. Mol. Biol. Cell 17, 5163–5172.

Mikels, A.J., Nusse, R., 2006. Purified Wnt5a protein activates or inhibits beta-catenin-TCF signaling depending on receptor context. PLoS Biol. 4, e115.

Monkley, S.J., Delaney, S.J., Pennisi, D.J., Christiansen, J.H., Wainwright, B.J., 1996.Targeted disruption of the Wnt2 gene results in placentation defects. Devel-opment 122, 3343–3353.

Norgall, S., Papoutsi, M., Rossler, J., Schweigerer, L., Wilting, J., Weich, H.A., 2007.Elevated expression of VEGFR-3 in lymphatic endothelial cells from lymphan-giomas. BMC Cancer 7, 105.

Okazaki, T., Iwatani, S., Yanai, T., Kobayashi, H., Kato, Y., Marusasa, T., Lane, G.J.,Yamataka, A., 2007. Treatment of lymphangioma in children: our experience of128 cases. J. Pediatr. Surg. 42, 386–389.

Otto, A., Schmidt, C., Luke, G., Allen, S., Valasek, P., Muntoni, F., Lawrence-Watt, D.,Patel, K., 2008. Canonical Wnt signalling induces satellite-cell proliferationduring adult skeletal muscle regeneration. J. Cell Sci. 121, 2939–2950.

Pandur, P., Maurus, D., Kuhl, M., 2002. Increasingly complex: new players enter theWnt signaling network. BioEssays 24, 881–884.

Person, A.D., Beiraghi, S., Sieben, C.M., Hermanson, S., Neumann, A.N., Robu, M.E.,Schleiffarth, J.R., Billington Jr., C.J., van Bokhoven, H., Hoogeboom, J.M., Mazzeu,J.F., Petryk, A., Schimmenti, L.A., Brunner, H.G., Ekker, S.C., Lohr, J.L., 2010.WNT5A mutations in patients with autosomal dominant Robinow syndrome.Dev. Dyn. 239, 327–337.

Podgrabinska, S., Braun, P., Velasco, P., Kloos, B., Pepper, M.S., Skobe, M., 2002.Molecular characterization of lymphatic endothelial cells. Proc. Natl. Acad. Sci.USA 99, 16069–16074.

Polakis, P., 2000. Wnt signaling and cancer. Genes Dev. 14, 1837–1851.

Savage, J., Voronova, A., Mehta, V., Sendi-Mukasa, F., Skerjanc, I.S., 2010. CanonicalWnt signaling regulates Foxc1/2 expression in P19 cells. Differentiation 79,31–40.

Seifert, J.R., Mlodzik, M., 2007. Frizzled/PCP signalling: a conserved mechanismregulating cell polarity and directed motility. Nat. Rev. Genet. 8, 126–138.

Sen, M., Lauterbach, K., El-Gabalawy, H., Firestein, G.S., Corr, M., Carson, D.A., 2000.Expression and function of wingless and frizzled homologs in rheumatoidarthritis. Proc. Natl. Acad. Sci. USA 97, 2791–2796.

Uhlen, M., Bjorling, E., Agaton, C., et al., 2005. A human protein atlas for normal andcancer tissues based on antibody proteomics. Mol. Cell Proteomics 4,1920–1932.

Ullrich, A., Sures, I., D'Egidio, M., Jallal, B., Powell, T.J., Herbst, R., Dreps, A., Azam, M.,Rubinstein, M., Natoli, C., et al., 1994. The secreted tumor-associated antigen90K is a potent immune stimulator. J. Biol. Chem. 269, 18401–18407.

van Amerongen, R., Mikels, A., Nusse, R., 2008. Alternative wnt signaling is initiatedby distinct receptors. Sci. Signalling 1, re9.

van Bokhoven, H., Celli, J., Kayserili, H., van Beusekom, E., Balci, S., Brussel, W.,Skovby, F., Kerr, B., Percin, E.F., Akarsu, N., Brunner, H.G., 2000. Mutation of thegene encoding the ROR2 tyrosine kinase causes autosomal recessive Robinowsyndrome. Nat. Genet. 25, 423–426.

Webber, S.A., Wargowski, D.S., Chitayat, D., Sandor, G.G., 1990. Congenital heartdisease and Robinow syndrome: coincidence or an additional component of thesyndrome? Am. J. Med. Genet. 37, 519–521.

Wick, N., Saharinen, P., Saharinen, J., Gurnhofer, E., Steiner, C.W., Raab, I., Stokic, D.,Giovanoli, P., Buchsbaum, S., Burchard, A., Thurner, S., Alitalo, K., Kerjaschki, D.,2007. Transcriptomal comparison of human dermal lymphatic endothelial cellsex vivo and in vitro. Physiol. Genomics 28, 179–192.

Wigle, J.T., Oliver, G., 1999. Prox1 function is required for the development of themurine lymphatic system. Cell 98, 769–778.

Wilting, J., Schneider, M., Papoutski, M., Alitalo, K., Christ, B., 2000. An avian modelfor studies of embryonic lymphangiogenesis. Lymphology 33, 81–94.

Wilting, J., Papoutsi, M., Christ, B., Nicolaides, K.H., von Kaisenberg, C.S., Borges, J.,Stark, G.B., Alitalo, K., Tomarev, S.I., Niemeyer, C., Rossler, J., 2002. Thetranscription factor Prox1 is a marker for lymphatic endothelial cells in normaland diseased human tissues. FASEB J. 16, 1271–1273.

Witte, M.H., Bernas, M.J., Martin, C.P., Witte, C.L., 2001. Lymphangiogenesis andlymphangiodysplasia: from molecular to clinical lymphology. Microsc. Res.Tech. 55, 122–145.

Wright, M., Aikawa, M., Szeto, W., Papkoff, J., 1999. Identification of a Wnt-responsive signal transduction pathway in primary endothelial cells. Biochem.Biophys. Res. Commun. 263, 384–388.

Yamaguchi, T.P., Bradley, A., McMahon, A.P., Jones, S., 1999. A Wnt5a pathwayunderlies outgrowth of multiple structures in the vertebrate embryo. Devel-opment 126, 1211–1223.