ORIGINAL PAPER

CXCL4L1 and CXCL4 signaling in human lymphaticand microvascular endothelial cells and activated lymphocytes:involvement of mitogen-activated protein (MAP) kinases, Srcand p70S6 kinase

Katrien Van Raemdonck • Mieke Gouwy •

Stefanie Antoinette Lepers • Jo Van Damme •

Sofie Struyf

Received: 9 November 2012 / Accepted: 13 January 2014

� Springer Science+Business Media Dordrecht 2014

Abstract CXC chemokines influence a variety of bio-

logical processes, such as angiogenesis, both in a physio-

logical and pathological context. Platelet factor-4 (PF-4)/

CXCL4 and its variant PF-4var/CXCL4L1 are known to

favor angiostasis by inhibiting endothelial cell proliferation

and chemotaxis. CXCL4L1 in particular is a potent

inhibitor of angiogenesis with anti-tumoral characteristics,

both through regulation of neovascularization and through

attraction of activated lymphocytes. However, its under-

lying signaling pathways remain to be elucidated. Here, we

have identified various intracellular pathways activated by

CXCL4L1 in comparison with other CXCR3 ligands,

including CXCL4 and interferon-c-induced protein

10/CXCL10. Signaling experiments show involvement of

the mitogen-activated protein kinase (MAPK) family in

CXCR3A-transfected cells, activated lymphocytes and

human microvascular endothelial cells (HMVEC). In

CXCR3A transfectants, CXCL4 and CXCL4L1 activated

p38 MAPK, as well as Src kinase within 30 and 5 min,

respectively. Extracellular signal-regulated kinase (ERK)

phosphorylation occured in activated lymphocytes, yet was

inhibited in microvascular and lymphatic endothelial cells.

CXCL4L1 and CXCL4 counterbalanced the angiogenic

chemokine stromal cell-derived factor-1/CXCL12 in both

endothelial cell types. Notably, inhibition of ERK signaling

by CXCL4L1 and CXCL4 in lymphatic endothelial cells

implies that these chemokines might also regulate

lymphangiogenesis. Furthermore, CXCL4, CXCL4L1 and

CXCL10 slightly enhanced forskolin-stimulated cAMP

production in HMVEC. Finally, CXCL4, but not

CXCL4L1, induced activation of p70S6 kinase within

5 min in HMVEC. Our findings confirm that the angio-

static chemokines CXCL4L1 and CXCL4 activate both

CXCR3A and CXCR3B and bring new insights into the

complexity of their signaling cascades.

Keywords Chemokines � CXCL4 � CXCL4L1 � p38

MAPK � Src kinase � ERK � cAMP � p70S6 kinase �CXCR3A � CXCR3B

Introduction

Among the family of chemotactic cytokines, the CXC

chemokines are emerging as a subgroup of particular

interest due to their role in angiogenesis [1, 2]. A clear

distinction can be made between CXC chemokines that

bind to the CXC chemokine receptor 2 (CXCR2), which

mediates angiogenic signaling, and those not binding

CXCR2, a subgroup to which angiostatic properties have

been ascribed [3, 4]. The latter also includes the CXCR3

ligands. Recently, several alternative variants of the

CXCR3 receptor have been described (CXCR3A,

CXCR3B and CXCR3alt), and it has been evidenced that

CXCL4/Platelet factor-4 (PF-4), its variant CXCL4L1/PF-

4var, CXCL9/Monokine induced by interferon-c (Mig),

CXCL10/Interferon-c-induced protein 10 (IP-10) and

Katrien Van Raemdonck and Mieke Gouwy have contributed equally

as first authors.

Electronic supplementary material The online version of thisarticle (doi:10.1007/s10456-014-9417-6) contains supplementarymaterial, which is available to authorized users.

K. Van Raemdonck � M. Gouwy � S. A. Lepers �J. Van Damme � S. Struyf (&)

Laboratory of Molecular Immunology, KU Leuven - University

of Leuven, Department of Microbiology and Immunology,

Rega Institute for Medical Research, Minderbroedersstraat 10,

3000 Leuven, Belgium

e-mail: sofie.struyf@rega.kuleuven.be

123

Angiogenesis

DOI 10.1007/s10456-014-9417-6

CXCL11/Interferon-inducible T cell alpha chemoattractant

(I-TAC), all bind at least one or more of the CXCR3

variants [2, 5–7].

CXCL4 and its variant CXCL4L1 have been the sub-

ject of a continuous line of research conducted by our

group. Both mature proteins are composed of 70 amino

acids and only differ in three amino acids situated in their

carboxy-terminal region (Pro58 ? Leu, Lys66 ? Glu,

Leu67 ? His; CXCL4 ? CXCL4L1) [8–10]. Both che-

mokines are released from activated blood platelets,

together with various others mediators of vascular

remodeling [11]. Indeed, CXCL4 and CXCL4L1 are

known to shift the angiogenic/angiostatic balance in favor

of angiostasis, by inhibiting endothelial cell proliferation

and migration [12–15]. CXCL4L1 in particular is a potent

inhibitor of angiogenesis, both in vitro and in vivo, and

accordingly prevents development and metastasis of var-

ious tumors [16]. This crucial link between tumor growth

and neovascularization was first recognized in 1971 and

since then various diseases have been shown to go hand

in hand with remodeling of the vascular network [17, 18].

We therefore presume CXCL4L1 to play a role in many

other angiogenic pathologies, especially those character-

ized by inflammation, such as cancer and diabetic reti-

nopathy [19].

On the molecular level, both CXCL4 and its variant

CXCL4L1 bind CXCR3A and CXCR3B, probably

mediating chemotactic activity (on activated T lympho-

cytes and dendritic cells) and angiostatic activity,

respectively [5–7]. Indeed, the angiostatic activity of

CXCL4L1 on endothelial cells can be neutralized with an

anti-CXCR3 antibody, and the chemotactic activity for

dendritic cells is inhibited by anti-CXCR3 or pertussis

toxin treatment [7]. In addition, interaction with glycos-

aminoglycans, chondroitin sulfate for CXCL4 in particu-

lar, is important for multimerization, establishment of the

chemokine gradient and intracellular signaling [20, 21].

For CXCL4L1, the underlying intracellular signaling

mechanisms remain to be unraveled. Considering the

clinical relevance of its angiostatic and anti-tumoral

characteristics, we aspired to gain insight into the sig-

naling cascade(s) triggered by CXCL4L1. Rather than

unraveling a specific cascade from head to toe, receptor to

transcription factor, we have investigated a spectrum of

different potential signaling pathways by focusing on

some of their key components. The selection of signaling

pathways was based on previous work of others on either

CXCL4 or CXCR3 ligand signaling [21–23]. In this

publication, we elaborate on the evidence of involvement

of p38 mitogen-activated protein kinase (p38 MAPK), Src

kinase, extracellular signal-regulated kinase (ERK) 1/2,

p70S6 kinase (p70S6K) and adenylyl cyclase in CXCL4

and/or CXCL4L1 signaling.

Methods

Recombinant human CXCL4L1 production

We made use of the Bac-to-Bac� Baculovirus Expression

System (as described and supplied by Invitrogen; Carlsbad,

CA) to establish production of recombinant human CXCL4L1

by baculovirus-infected insect cells (Sf9). The protein of

interest was isolated from the cell medium by three consec-

utive chromatographical steps as described for purification of

recombinant CXCL4L1 from Escherichia coli [16].

Cell cultures

Signal transduction assays were conducted in vitro on

various cell types assumed to be responsive to CXCR3

ligands such as the chemokines of interest, CXCL4 and

CXCL4L1. Two human microvascular endothelial cell

preparations (HMVEC) were tested, from dermal (Lonza,

Verviers, Belgium) and retinal (Cell Systems, Kirkland,

WA) origin. The two HMVEC preparations responded

similarly, and results were pooled. Both dermal and retinal

HMVEC were cultured in endothelial basal medium-2

(EBM-2) enriched with endothelial growth medium-2 MV

Bulletkit (Lonza). Furthermore, human dermal lymphatic

microvascular endothelial cells were purchased from

Lonza (cat no. CC-2812) and cultured following the

manufacturer’s instructions, which are identical to those for

microvascular endothelial cells. In addition, we used Chi-

nese hamster ovary (CHO) cells, transfected to express

human CXCR3A (CHO/CXCR3A), human CXCR3B

(CHO/CXCR3B) and appropriate controls [7, 24]. These

CHO-transfected cell lines were cultured in Ham’s F-12

medium (Invitrogen) supplemented with 10 % (v/v) fetal

calf serum (FCS; Sigma-Aldrich, St. Louis, MO), 1 mM

sodium pyruvate, 400 lg/ml geneticin and 0.1 g/100 ml

sodium bicarbonate. Lymphoblasts derived from PHA-

activated (PHA-L cat no. L2769, 2 lg/ml; Sigma-Aldrich)

PBMC purified from buffy coats (blood transfusion center,

Red Cross, Leuven; gender of donors unknown) were

grown in RPMI 1640 medium (Lonza) with 10 % (v/v)

FCS, 50 U/ml human recombinant IL-2 (PeproTech,

Rocky Hill, CT), 0.1 g/100 ml sodium bicarbonate and

50 lg/ml gentamycin.

Signal transduction assays

Adherent cell cultures

Prior to signal transduction assays, CHO-transfected cells,

HMVEC or lymphatic endothelial cells were seeded in

6-well plates. Once cell cultures reached a confluency

level of 80 % or more, cells were starved from growth

Angiogenesis

123

factors overnight by replacing their growth medium with

similar medium, only without growth supplements

(HMVEC and lymphatic endothelial cells) or FCS (CHO

cells). At least 15 min prior to stimulation, the starvation

medium was replaced with starvation medium ?0.5 g/

100 ml bovine serum albumin (BSA; Sigma-Aldrich).

Tenfold dilutions of the inducers were prepared in the

appropriate medium, either EBM-2/BSA or Ham’s F-12/

BSA. In the context of this study, we stimulated cells for

varying time periods, from 5 min up till 2 h, at 37 �C/

5 % CO2 with either human natural CXCL4 (isolated

from stimulated platelets and purified to homogeneity as

previously described) [13], recombinant human CXCL4

(PeproTech), recombinant human CXCL4L1 (produced as

described above) or recombinant human CXCL10 (Pep-

roTech). Additionally, stimulations with human synthetic

CXCL12(1–68) [25] were performed, both as a single

stimulus and in combination with CXCL4 or CXCL4L1.

Afterward the cells were washed three times with ice-cold

PBS to stop the stimulation, and lysis buffer was added

[either 6 M ureum; 1 mM ethylenediaminetetraacetic acid

(EDTA); 0.5 % (v/v) Triton X-100; 5 mM NaF in PBS;

pH 7.2–7.4 or 50 mM Tris–HCl; 0.1 % SDS; 1 % NP-40;

0.5 % deoxycholic acid; 150 mM NaCl in PBS; pH 8 for

use in ELISA or Western blotting, respectively]. Cell

residue was removed from the collected lysates by cen-

trifugation (4 �C, 10 min, 1,666g). Though the protease

inhibitor cocktail for mammalian tissue as well as phos-

phatase inhibitor cocktails 1 and 2 (Sigma-Aldrich) were

used, it should be noted that samples are still relatively

unstable and especially sensitive to thawing. All sample

aliquots were therefore tested only once, immediately

after thawing on ice. The phosphorylation state of p38

MAPK (cat no. DYC869), p70S6K (cat no. DYC896), Src

kinase (cat no. DYC2685) or ERK 1/2 (cat no.

DYC1018B) was verified by DuoSet� IC ELISA’s as

supplied by R&D Systems (Minneapolis, MN). Simulta-

neously, total protein content was determined in a bi-

cinchoninic acid (BCA) protein assay (Pierce, Rockford,

IL). Alternatively, the phosphorylation state of said

kinases was examined through western blotting as

described below.

Western blotting

A previously described protocol was adapted to quantita-

tive multiplex protein detection using fluorescently labeled

secondary antibodies [26]. Reduced samples were sepa-

rated on 12 % SDS-PAGE gradient gels, which were

semidry blotted onto Hybond-LFP (low fluorescent PVDF;

GE Healthcare, Little Chalfont, UK) membranes. Mem-

branes were blocked overnight at 4 �C with 2 % ECL

blocking agent (GE Healthcare) in wash buffer [TBS

?0.1 % Tween-20 (TBST)] and then simultaneously

incubated at room temperature (RT) for 2 h with a phos-

pho-specific antibody against the studied kinase combined

with an antibody against an internal control. Antibodies

used to detect activated kinases were anti-phospho p38

MAPK (Thr180/Tyr182) mouse monoclonal (mAb) anti-

body (1/200 dilution; Cell Signaling, Danvers, MA) and

anti-phospho-ERK (E-4) mAb (1/300; Santa Cruz, Santa

Cruz, CA). Equal gel loading was verified using anti-ERK 1/2

rabbit antibody (dilution 1/1,000; Cell Signaling). Mem-

branes were washed and then incubated for 2 h at RT with

ECL Plex goat anti-mouse Cy5 (1/1,500 dilution) and ECL

Plex goat anti-rabbit Cy3-conjugated antibody (1/2,500

dilution) (GE Healthcare). After incubation with secondary

antibodies, blots were washed in TBST, followed by

washes in TBS. Dried membranes were scanned and

quantified with the Ettan DIGE Imager and ImageQuant

software, both from GE Healthcare.

cAMP assay

HMVEC were stimulated as described above for the

kinase activation assays. Briefly, we stimulated HMVEC

for 30 or 120 min at 37 �C/5 % CO2 with CXCL4,

CXCL4L1 or CXCL10. Additionally, stimulations with

forskolin (Enzo Life Sciences, Belgium) were performed,

both as a single stimulus and in combination with

CXCL4, CXCL4L1 or CXCL10. Afterward the cells were

washed three times with ice-cold PBS to stop the stimu-

lation, and lysis buffer (0.1 M HCl) was added to stop

endogenous phosphodiesterase activity. Cell residue was

removed from the collected lysates by centrifugation

(4 �C, 10 min, 1,666g). The amount of cAMP in the

lysate was determined with the cAMP complete ELISA

kit (acetylated format), which is a competitive immuno-

assay (Enzo Life Sciences).

Lymphoblasts

Before induction, lymphoblasts were centrifuged (10 min,

200g), resuspended at 159106 cells/ml in starvation med-

ium (RPMI 1640 with 0.5 g/100 ml BSA) and incubated

under these conditions for 15 min at RT. Tenfold dilutions

of the stimuli, in this case CXCL4, CXCL4L1 and

CXCL10, were prepared in RPMI/BSA medium. Sus-

pended cells (200–1,000 ll) were added to the concen-

trated chemokine mixes, distributed over conical 15-ml

centrifugation tubes. After stimulation for 30 min at 37 �C,

ice-cold PBS (10 ml/tube) was added to stop the induction.

The still intact cells were spun down in a precooled cen-

trifuge (10 min, 200g). Finally, the pellets were resus-

pended in lysis buffer, and samples were further handled as

described above for adherent cells.

Angiogenesis

123

Statistical analysis

The Mann–Whitney test was used to compare means from

two independent groups. Dependent groups were compared

using the Sign test. A p value of \0.05 was considered to

indicate a statistically significant difference. The number of

independent experiments is indicated by n.

Results

CXCL4 and CXCL4L1 upregulate p38 MAPK

phosphorylation in CHO/CXCR3A transfectants

Kasper et al. [27] found CXCL4 to activate the p38 MAPK

pathway in neutrophils. This inflammatory pathway is acti-

vated by several chemokines in leukocytes. As both CXCL4

and CXCL4L1 attract activated lymphocytes, probably via

CXCR3A, the phosphorylation status of p38 MAPK in

CXCR3A-transfected CHO cells after CXCL4 and

CXCL4L1 treatment was studied (Fig. 1a–c). Results show

p38 MAPK phosphorylation to be upregulated after 30 min,

both upon CXCL4 and upon CXCL4L1 stimulation. In

particular, natural CXCL4 (1,000 ng/ml) significantly ele-

vated the phosphorylation status to 121.6 ± 4.9 %

(p = 0.0304, n = 4). In addition, recombinant (rec) and

natural (nat) CXCL4 were equally active at 3,000 ng/ml in

inducing phosphorylation of p38 kinase (data not shown).

Hence, all further experiments were performed with natural

CXCL4. CXCL4L1 induced an increase to 154.2 ± 17.8,

136.0 ± 7.4 and 146.8 ± 18.4 % at 100 (p = 0.0210,

n = 13), 300 (p = 0.0304, n = 4) and 1,000 ng/ml

(p = 0.0809, n = 3), respectively (Fig. 1a). For compari-

son, a third more intensively studied CXCR3 ligand,

CXCL10, was also used as a stimulus. CXCL10 (500 ng/ml)

increased p38 MAPK phosphorylation on average to

126.1 ± 9.3 %, yet this change was not statistically signif-

icant (p = 0.1405, n = 10). In addition, we performed time-

course experiments. Figure 1b and c (ELISA and Western

blot, respectively) show that treatment of the cells with

100 ng/ml of CXCL4L1 resulted in enhanced phosphoryla-

tion of p38 MAPK. A maximal response was reached at

30 min. The response was transient in that baseline levels

were measured after 60 min. We did not observe upregula-

tion of p38 phosphorylation in CHO/CXCR3B or control-

b Fig. 1 CXCL4 and CXCL4L1 activate the p38 MAPK pathway in

CHO/CXCR3A transfectants. CXCR3A- (a–c) or CXCR3B-transfec-

ted (d) CHO cells were stimulated in duplicate with natural CXCL4,

recombinant CXCL4L1 or recombinant CXCL10 to evaluate p38

MAPK phosphorylation. Cells were stimulated for 30 min with

various concentrations of CXCR3 ligands (a, d) or with 100 ng/ml of

CXCL4L1 for different time intervals (b, c). The ratio of phospho-

p38 MAPK/total protein content (ELISA; a, b and d) or phospho-p38

MAPK/total ERK (Western blotting; c) was calculated for all cell

lysates. Results (mean % of control ± SEM) are expressed relative

to the phosphorylation status of p38 MAPK after control treatment.

a n = 3–13; b n = 3–9; c n = 2–3; d n = 3; *p \ 0.05 (Mann–

Whitney test)

Angiogenesis

123

transfected CHO cells after stimulation with CXCL4L1

(100–1,000 ng/ml), CXCL4 (100–1,000 ng/ml) or CXCL10

(500 ng/ml) at 30 min after treatment (Fig. 1d and data not

shown).

CXCL4 and CXCL4L1 upregulate Src kinase

phosphorylation in CHO/CXCR3A transfectants

Earlier work by Bonacchi et al. [28] has already implicated

Src kinase in CXCR3 signaling, following activation of

hepatic stellate cells by CXCL9 and CXCL10. We there-

fore also studied the effects of CXCL4 and CXCL4L1

stimulation on the activation of Src kinase more in depth,

using CHO/CXCR3A transfectants (Fig. 2a, b). Results

reveal a significant increase in the phosphorylation status

within 5 min following either CXCL4 (3,000 ng/ml;

132.8 ± 17.8 %, p = 0.0416, n = 10), CXCL4L1 (300

and 1,000 ng/ml; 120.8 ± 9.3 %, p = 0.0398, n = 5;

150.8 ± 28.7 %, p = 0.0179, n = 6, respectively) or

CXCL10 stimulation (500 ng/ml; 129.0 ± 6.6 %, p =

0.0003, n = 12) (Fig. 2a). In a separate set of experiments,

we investigated the response to 100 ng/ml of CXCL4L1

over a longer time period. Src kinase phosphorylation

peaked 15 min after stimulation and a return to baseline

levels was observed after 60 min (Fig. 2b). In contrast, Src

kinase was not phosphorylated in response to CXCL4L1

(100–1,000 ng/ml), CXCL4 (300–3,000 ng/ml) or

CXCL10 (500 ng/ml) in CHO/CXCR3B or control-trans-

fected CHO cells (Fig. 2c and data not shown).

CXCL4 and CXCL4L1 upregulate ERK

phosphorylation in lymphoblasts

Next, we decided to proceed with a physiologically rele-

vant CXCR3A-expressing cell type, which is functionally

responsive to CXCL4L1 in chemotaxis assays [7]. We tried

to confirm the phosphorylation of Src kinase in activated

lymphoblasts, but unfortunately, these cells did not express

sufficient levels of this tyrosine kinase to be detected by

ELISA (data not shown). However, as Bonacchi et al. [28]

demonstrated that in hepatic stellate cells the MAPK ERK

is downstream of Src in the CXCL10/CXCR3 signaling

pathway, we decided to investigate whether CXCL4L1

signaling in activated lymphocytes involves phosphoryla-

tion of ERK. Moreover, CXCL4-induced phosphorylation

of ERK in human T lymphocytes has been studied in detail

previously [23]. Within 30 min of stimulation with all three

CXCR3 ligands, the ERK pathway in lymphoblasts was

activated (Fig. 3). CXCL4L1 significantly increased

phosphorylation to 125.4 ± 9.2 % (p = 0.0047, n = 8) at

300 ng/ml and to 159.5 ± 22.5 % (p = 0.0031, n = 9) at

1,000 ng/ml. Similarly, CXCL4 activated ERK at

3,000 ng/ml (187.3 ± 37.3 %; p = 0.0006, n = 15) and

10,000 ng/ml (174.3 ± 46.8 %; p = 0.0508, n = 8),

whereas for CXCL10 the minimal effective concentration

was 30 ng/ml (156.8 ± 23.1 %, p = 0.0011, n = 11).

Fig. 2 CXCL4L1, CXCL4 and CXCL10 activate Src kinase in CHO/

CXCR3A transfectants. The Src kinase pathway was investigated in

CXCR3A-(a, b) or CXCR3B-transfected CHO cells (c) upon stim-

ulation with either natural CXCL4, recombinant CXCL4L1 or

recombinant CXCL10. Cells were stimulated for 5 min with various

concentrations of CXCR3 ligands (a, c) or with 100 ng/ml of

CXCL4L1 for different time intervals (b). The ratio of phospho-Src

kinase/total protein content was calculated for all cell lysates. Results

(mean ± SEM) are expressed relative to the phosphorylation status of

Src kinase after buffer treatment. a n = 5–12; b n = 3–6; c n = 4–6;

*p \ 0.05; ***p \ 0.001 (Mann–Whitney test)

Angiogenesis

123

CXCL4 and CXCL4L1 counteract CXCL12-induced

ERK phosphorylation in lymphatic and microvascular

endothelial cells

In view of the importance of the ERK pathway in signaling

of angiogenic endothelial growth factors, we analyzed the

effect of CXCL4L1 on ERK activation by CXCL12 in

HMVEC. Earlier studies have established the inhibitory

effect of CXCL4 on ERK activation induced by FGF-2 as

well as by CCL2 in endothelial cells [26, 29]. To extend

these findings to CXCL4L1 signaling, we investigated

ERK phosphorylation in HMVEC after 5-min stimulation

with CXCL4 or its variant, either as single stimulus or as

inhibitor counteracting the pro-angiogenic chemokine

CXCL12 (Fig. 4a). Upregulated phosphorylation of ERK

by CXCL12 (30 ng/ml) was significantly inhibited by

additional treatment of HMVEC with CXCL4 (300 and

1,000 ng/ml; p = 0.0455, n = 8 and p = 0.0019, n = 12,

respectively) or CXCL4L1 (100 ng/ml and 300 ng/ml;

p = 0.0098, n = 12 and p = 0.0265, n = 11, respec-

tively). CXCL4L1 (300 ng/ml) reduced CXCL12-induced

upregulation of phospho-ERK by as much as 58.9 %

(58.4 % upregulation reduced to 24.0 % upregulation).

CXCL4 (1,000 ng/ml; p = 0.0233, n = 7) also signifi-

cantly counteracted higher concentrations of CXCL12

(100 ng/ml). These data were obtained through the use of

phosphorylation-specific ELISAs, but were also corrobo-

rated by quantitative detection of phosphorylated ERK by

Western blotting (Fig. 4b; Supplementary figure S1). ERK

activation induced by CXCL12 (135.5 ± 11.3 %;

30 ng/ml; n = 4) was shown to be inhibited by CXCL4L1

at 100 ng/ml (112.1 ± 5.5 %; n = 3) or 300 ng/ml

(120.0 ± 12.5; n = 4) in HMVEC (Fig. 4b).

We repeated these experiments in a second type of

endothelial cells, i.e., lymphatic endothelial cells (Fig. 4c).

On the latter cell type, CXCL4L1 (100–300 ng/ml) also

significantly inhibited CXCL12-induced ERK activation,

when 30 ng/ml of agonist was used (p = 0.0433, n = 9

and p = 0.0269, n = 9, respectively). Lymphatic endo-

thelial cells were quite responsive to CXCL4L1. CXCL4,

however, had a more variable effect on CXCL12-induced

signaling, and therefore, statistical significance was not

reached (p = 0.4497, n = 6) (Fig. 4c).

These findings on CXCL4 or CXCL4L1 inhibition of

ERK signaling by angiogenic factors in HMVEC and

lymphatic endothelial cells fit with their function in the

angiogenic/angiostatic balance. We must, however,

Fig. 3 In lymphoblasts, CXCL4, CXCL4L1 and CXCL10 upregulate

ERK phosphorylation. Lymphoblasts were stimulated for 30 min with

recombinant CXCL4L1 (300–3,000 ng/ml), natural CXCL4

(300–10,000 ng/ml) or recombinant CXCL10 (30 and 100 ng/ml).

The ratio of phospho-ERK/total protein content was calculated for

cell lysates after 30 min incubation with various stimuli. Results

(mean ± SEM) are expressed relative to the phosphorylation status of

ERK after control treatment. n = 3–16; *p \ 0.05; **p \ 0.01

(Mann–Whitney test)

Fig. 4 CXCL4 and CXCL4L1 counteract angiogenic CXCL12-

induced ERK signaling in microvascular and lymphatic endothelial

cells. ERK phosphorylation was evaluated after stimulation of

HMVEC (a and b) or lymphatic endothelial cells (HLEC) (c) with

synthetic CXCL12 (30 and 100 ng/ml in a; 30 ng/ml in b, c),

recombinant CXCL4L1 (30–300 ng/ml), natural CXCL4 (300 and

1,000 ng/ml) or combinations of CXCL12 with a CXCR3 ligand. The

ratio of phospho-ERK/total protein content (ELISA; a, c) or phospho-

ERK/total ERK (Western blotting; b) was calculated for cell lysates

after 5 min incubation with a single chemokine stimulus or combined

chemokine treatment. Results (mean ± SEM) are expressed relative

to the phosphorylation status of ERK after buffer treatment.

a n = 3–13; b n = 3–4; c n = 6–9 ***p \ 0.001 (Mann–Whitney

test versus CO); �p \ 0.05; ��p \ 0.01 (Sign test versus CXCL12)

Angiogenesis

123

conclude that the CXCL4L1 signaling pathway in activated

lymphocytes versus endothelial cells involves activation

versus inhibition of the MAPK ERK, respectively. This can

most probably be attributed to the cell type-dependent

CXCR3 subtype expression [5, 19, 30].

CXCL4L1 and CXCL4 affect adenylyl cyclase activity

in microvascular endothelial cells

Although three variants of CXCR3 have been described,

experimental evidence suggests that endothelial cells

selectively express CXCR3B and that CXCR3B mediates

the angiostatic activity of CXCL4 [5, 19]. We have already

shown that CXCL4L1 also binds to CXCR3B [7], but not

yet investigated downstream signaling pathways. Because

CXCR3B is reported to couple to a Gas protein and to

stimulate production of cAMP after CXCL4-activation, we

explored this signaling pathway also for CXCL4L1.

Therefore, we treated HMVEC for 30 (Fig. 5) or 120 min

(data not shown) with CXCL4L1 (30 and 100 ng/ml),

CXCL4 (1,000 ng/ml) or CXCL10 (1,000 ng/ml) as single

stimulus or in combination with 1 lM of forskolin

(n = 2–5). Although the chemokines consistently potenti-

ated the forskolin-induced cAMP production in HMVEC

(both after 30 and 120 min), as single stimulus these che-

mokines did not enhance adenylyl cyclase activity in

HMVEC (Fig. 5). This is in contrast to the results reported

by Lasagni et al. [5]. It must be noted, however, that these

authors used CXCR3B-transfected cells and not primary

endothelial cells.

CXCL4, but not CXCL4L1, evokes a transient peak

in p70S6K phosphorylation in HMVEC

Besides ERK, which appears to mediate the effects of

CXCL4 and its variant on a range of different cell types

and in various biological processes, we also investigated

activation of p70S6K, which is specifically known for its

anti-apoptotic properties and has been linked to angio-

genesis as well [31]. Some studies have also provided

evidence of p70S6K activation by the chemokines CCL2

and CXCL12 in tumor cells [32, 33]. Moreover, this less-

studied pathway has been suggested to offer a new

potential target in several malignant diseases [34]. Con-

sidering the lack of previous reports on CXCR3-mediated

effects on p70S6K, we chose to study p70S6K phosphor-

ylation over time following stimulation of HMVEC with

either CXCL4 (1,000 ng/ml) or CXCL4L1 (100 ng/ml)

(Fig. 6). We postulated that these pro-apoptotic, angiostatic

ligands would reduce p70S6K activation. Controversially,

CXCL4 (1,000 ng/ml) was shown to significantly increase

p70S6K phosphorylation to 193.6 ± 51.8 % within 5 min

of stimulation (p = 0.0378; n = 10) (Fig. 6a). A less

pronounced, (not significant), upregulation was noted at

later time points. Following the brief initial peak, average

p70S6K phosphorylation by CXCL4 normalized com-

pletely within 60 min. On the other hand, CXCL4L1

(30–300 ng/ml) had no significant impact on p70S6K at

any studied time interval. The discordance between both

variants was not due to our choice of dosage, as CXCL4

was also shown to increase p70S6K phosphorylation to

204.9 ± 36.9 % (p = 0.0074; n = 3) and 150.5 ± 32.0 %

(p = 0.0402; n = 6) within 5 min at 100 and 300 ng/ml,

respectively (Fig. 6b). Finally, a third CXCR3 ligand,

CXCL10 (500 ng/ml), did also not significantly alter

p70S6K phosphorylation after 5 min of stimulation, sug-

gesting that the CXCL4-induced activation of p70S6K is

not a common CXCR3-mediated event, but might rather be

GAG-dependent (Fig. 6b).

Discussion

CXCL4 has distinguished itself from the superfamily of

chemokines as an atypical player being involved in a

remarkable variety of biological processes. Moreover, its

role in diseases, specifically angiogenesis-related patholo-

gies, has been widely recognized [19]. Anti-tumoral char-

acteristics of CXCL4 and particularly CXCL4L1 have been

a subject of interest to our group for some years. CXCL4-

activated signaling has been described to some extent

alongside its more thoroughly studied and understood fel-

low CXCR3 ligands, CXCL9, CXCL10 and CXCL11.

CXCL4L1-activated pathways on the other hand were not

Fig. 5 CXCL4L1, CXCL4 and CXCL10 potentiate forskolin-

induced production of cAMP in microvascular endothelial cells.

Production of cAMP was evaluated after stimulation of HMVEC with

buffer, recombinant CXCL4L1 (30 and 100 ng/ml), natural CXCL4

(300 and 1,000 ng/ml) or recombinant CXCL10 (1,000 ng/ml) alone

or in combination with 1 lM of forskolin. Cells were stimulated for

30 min before lysis. Results (mean ± SEM) are expressed relative to

cAMP levels after buffer treatment. n = 2–5

Angiogenesis

123

yet studied. In this study, we provide a first insight into

CXCL4L1 signaling.

We have identified the MAPKs as downstream targets of

both CXCL4 and CXCL4L1. CXCL4 and CXCL4L1

activate p38 MAPK in CXCR3A-transfected cells and

ERK in activated lymphocytes. In activated lymphocytes,

ERK is phosphorylated in response to CXCL4 and

CXCL4L1 treatment, probably establishing these ligands’

stimulatory effect on migration, whereas CXCL12-medi-

ated ERK activation in HMVEC and lymphatic endothelial

cells was inhibited. These last findings match the angio-

static effect of CXCL4 and CXCL4L1 [13, 35]. Our study

thus identifies ERK as an interesting, ambivalent player in

CXCL4 and CXCL4L1 signaling, giving insight into the

flexible, complicated cascade that contributes to the variety

of effects of these chemokines in different cell types and

hence their involvement in a broad range of biological

processes and diseases.

Rather surprising was the transient activation of p70S6K

in HMVEC after stimulation with CXCL4, but not with

CXCL4L1. The in principal anti-apoptotic, angiogenic

characteristics of p70S6K [31] do not seem to add up with

the described effects of CXCL4 on endothelial cells and

angiogenesis. We have no clear explanation for this

observation, though it is interesting to note that CXCL4L1,

which is a more potent angiostatic chemokine than CXCL4

[13], does not exhibit this brief activation of p70S6K. The

absence of counteractive p70S6K phosphorylation in

CXCL4L1-treated cells may give more weight to the an-

giostatic signals initiated by the CXCL4L1 receptor. A

possible explanation for these results may be the difference

between CXCL4 and CXCL4L1 in affinity for glycos-

aminoglycans (e.g., heparin and chondroitin sulfate). Du-

brac et al. [15] nicely showed that the replacement of the

leucine present in CXCL4 on position 67, by a histidine in

CXCL4L1 on this position causes a decrease in GAG-

binding coinciding with an increase in angiostatic poten-

tial. Kuo et al. [36] very recently identified this Leu67/

His67 replacement to be the primary cause for a different

3D structure of CXCL4L1 compared to other chemokines

and CXCL4. They demonstrated that CXCL4L1 displays a

unique C-terminal helix orientation that modifies the

overall chemokine shape and surface properties because

the C-terminal helix protrudes into the aqueous space

exposing the entire helix. It seems therefore justified to

causatively correlate these three observations, i.e., that the

replacement causes a structural change in the C-terminus,

which lowers the affinity for GAGs and is linked with

higher angiostatic potential.

Kasper and Petersen [21] recently summarized the

available information on the signaling pathways activated

by CXCL4. A distinction must be made between cells

responsive to CXCL4 via GAG moieties on proteogly-

cans, such as monocytes and neutrophils, and cell types

expressing CXCR3. In lymphoblasts, CXCR3 couples to

the pertussis toxin-sensitive ERK/MAPK, Akt/phosphati-

dylinositol 3-kinase (PI3K) and phospholipase C path-

ways [22]. Korniejewska et al. [23] confirmed ERK and

Akt activation downstream of CXCR3 after CXCL4

treatment of the cells, but observed important differences

between CXCL4 and the classical, inducible CXCR3

ligands, e.g., lack of down-regulation of the receptor from

the cell surface after ligand-receptor binding. Other

groups studied CXCR3 signaling in airway epithelial

cells, mast cells, hepatic stellate cells and vascular peri-

cytes [28, 37, 38]. Indeed, a large variety of cells express

CXCR3 isoforms and the functional outcome after

receptor activation is very diverse, ranging from

Fig. 6 CXCL4, but not CXCL4L1, increases anti-apoptotic signaling

through p70S6K in microvascular endothelial cells. HMVEC were

stimulated in duplicate for different time periods (5–120 min) with

either natural CXCL4 (1,000 ng/ml) or recombinant CXCL4L1

(100 ng/ml) (a). Alternatively, cells were stimulated with varying

concentrations of recombinant CXCL4L1, natural CXCL4 or

recombinant CXCL10 for 5 min (b). The ratio of phospho-p70S6K/

total protein content was calculated for all cell lysates. Results

(mean ± SEM) are expressed relative to the phosphorylation status of

p70S6K after control treatment. a n = 3–10; b n = 3–10; *p \ 0.05;

**p \ 0.01 (Mann–Whitney test)

Angiogenesis

123

migration (activated T lymphocytes, NK cells, airway

epithelial cells, hepatic stellate cells), inhibition of

migration (endothelial cells), proliferative (mesangial

cells) to anti-proliferative effects (endothelial cells) and

induction of apoptosis (endothelial cells) [19, 28, 39]. In

general, proliferative and positive migratory effects are

supposed to be mediated by CXCR3A, whereas inhibition

of chemotaxis, anti-proliferative and apoptotic effects are

postulated to be provoked via CXCR3B. Our results are

concordant with this hypothesis. Indeed, in CXCR3A-

transfected cells and activated lymphocytes, the

CXCL4L1-activated MAPK signal transduction pathway

is similar. In contrast, we could not observe Src, nor p38

MAPK signaling in the CHO/CXCR3B cells although

these cells are capable of binding both CXCL4 and

CXCL4L1 [7]. CXCR3B is natively expressed in endo-

thelial cells, in which CXCL4L1 induced a subtle rise in

cAMP production, suggesting that its receptor in these

cells couples to a Gas protein and results in inhibition of

migration, effects described previously for CXCR3B.

Finally, we provided evidence that CXCL4L1 also

affects lymphangiogenesis. Lymphangiogenesis, the pro-

cess of new lymphatic vessel formation, is largely inactive

in normal physiology, but is activated in developing tumors

and inflammation [40]. In cancer patients, lymphangiogenesis

has been associated with increased metastasis and poor

prognosis because it provides an extra escape route for

tumor cells. Because this process is quiescent during

homeostasis and activated under pathological conditions,

interrupting lymphangiogenesis offers therapeutic poten-

tial. Even more so since the lymphatic vasculature could be

considered to be a loophole through which disseminating

tumor cells bypass currently used anti-angiogenic therapies.

Similarly to vascular angiogenesis, lymphangiogenesis is

stimulated by CXCL8 and CXCL12 [41, 42]. Direct inhi-

bition of this process by CXCR3 ligands, however, had not

yet been studied in depth. Our study now demonstrates that

in lymphatic endothelial cells CXCL4L1 counteracted

CXCL12-induced activation as was observed in micro-

vascular endothelial cells. In both endothelial cell types, we

detected selective expression of CXCR3B (and not

CXCR3A; data not shown). Therefore, we can postulate

that the anti-tumoral activity of CXCL4L1 is based on

three different mechanisms: inhibition of the formation of

new blood vessels, blockade of lymphatic vessel develop-

ment and attraction of anti-tumoral lymphocytes. Because

CXCL4L1 blocks the angiogenic activity of chemokines

(CXCL8 and CXCL12) and growth factors (basic fibroblast

growth factor and VEGF), it is a promising therapeutic. In

this study, we revealed several signal transduction events

that could be implicated in mediating its physiological

effects. However, further research is needed to reveal the

full signaling network of CXCL4L1.

Acknowledgments The authors would like to thank Jolien Daniels,

Lien Leutenez, Noemie Portner and Isabelle Ronsse for excellent

technical assistance. M. Parmentier kindly provided the CXCR3-

transfected cells. This work was supported by the Fund for Scientific

Research of Flanders (FWO-Vlaanderen project G.0D66.13N and

G.0773.13, Belgium), the Interuniversity Attraction Poles Programme

initiated by the Belgian Science Policy Office (I.A.P. project P7/40)

and the Concerted Research Actions of the Regional Government of

Flanders (GOA12/017 and GOA13/014).

Conflict of interest None.

References

1. Rosenkilde MM, Schwartz TW (2004) The chemokine system—a

major regulator of angiogenesis in health and disease. APMIS

112:481–495

2. Keeley EC, Mehrad B, Strieter RM (2011) Chemokines as

mediators of tumor angiogenesis and neovascularization. Exp

Cell Res 317:685–690

3. Balestrieri ML, Balestrieri A, Mancini FP, Napoli C (2008)

Understanding the immunoangiostatic CXC chemokine network.

Cardiovasc Res 78:250–256

4. Rossi D, Zlotnik A (2000) The biology of chemokines and their

receptors. Annu Rev Immunol 18:217–242

5. Lasagni L, Francalanci M, Annunziato F, Lazzeri E, Giannini S,

Cosmi L, Sagrinati C, Mazzinghi B, Orlando C, Maggi E, Marra F,

Romagnani S, Serio M, Romagnani P (2003) An alternatively

spliced variant of CXCR3 mediates the inhibition of endothelial

cell growth induced by IP-10, Mig, and I-TAC, and acts as func-

tional receptor for platelet factor 4. J Exp Med 197:1537–1549

6. Mueller A, Meiser A, McDonagh EM, Fox JM, Petit SJ, Xanthou

G, Williams TJ, Pease JE (2008) CXCL4-induced migration of

activated T lymphocytes is mediated by the chemokine receptor

CXCR3. J Leukoc Biol 83:875–882

7. Struyf S, Salogni L, Burdick MD, Vandercappellen J, Gouwy M,

Noppen S, Proost P, Opdenakker G, Parmentier M, Gerard C,

Sozzani S, Strieter RM, Van Damme J (2011) Angiostatic and

chemotactic activities of the CXC chemokine CXCL4L1 (platelet

factor-4 variant) are mediated by CXCR3. Blood 117:480–488

8. Deuel TF, Keim PS, Farmer M, Heinrikson RL (1977) Amino

acid sequence of human platelet factor 4. Proc Natl Acad Sci U S

A 74:2256–2258

9. Green CJ, Charles RS, Edwards BF, Johnson PH (1989) Identi-

fication and characterization of PF4varl, a human gene variant of

platelet factor 4. Mol Cell Biol 9:1445–1451

10. Eisman R, Surrey S, Ramachandran B, Schwartz E, Poncz M

(1990) Structural and functional comparison of the genes for

human platelet factor 4 and PF4alt. Blood 76:336–344

11. Sabrkhany S, Griffioen AW, Oude Egbrink MG (2011) The role

of blood platelets in tumor angiogenesis. Biochim Biophys Acta

1815:189–196

12. Maione TE, Gray GS, Petro J, Hunt AJ, Donner AL, Bauer SI,

Carson HF, Sharpe RJ (1990) Inhibition of angiogenesis by

recombinant human platelet factor-4 and related peptides. Sci-

ence 247:77–79

13. Struyf S, Burdick MD, Proost P, Van Damme J, Strieter RM

(2004) Platelets release CXCL4L1, a nonallelic variant of the

chemokine platelet factor-4/CXCL4 and potent inhibitor of

angiogenesis. Circ Res 95:855–857

14. Sarabi A, Kramp BK, Drechsler M, Hackeng TM, Soehnlein O,

Weber C, Koenen RR, von Hundelshausen P (2011) CXCL4L1

inhibits angiogenesis and induces undirected endothelial cell

Angiogenesis

123

migration without affecting endothelial cell proliferation and

monocyte recruitment. J Thromb Haemost 9:209–219

15. Dubrac A, Quemener C, Lacazette E, Lopez F, Zanibellato C, Wu

WG, Bikfalvi A, Prats H (2010) Functional divergence between 2

chemokines is conferred by single amino acid change. Blood

116:4703–4711

16. Struyf S, Burdick MD, Peeters E, Van den BK, Dillen C, Proost

P, Van Damme J, Strieter RM (2007) Platelet factor-4 variant

chemokine CXCL4L1 inhibits melanoma and lung carcinoma

growth and metastasis by preventing angiogenesis. Cancer Res

67:5940–5948

17. Folkman J (1971) Tumor angiogenesis: therapeutic implications.

N Engl J Med 285:1182–1186

18. Thijssen VL, van Beijnum JR, Mayo KH, Griffioen AW (2007)

Identification of novel drug targets for angiostatic cancer therapy;

it takes two to tango. Curr Pharm Des 13:3576–3583

19. Vandercappellen J, Van Damme J, Struyf S (2011) The role of the

CXC chemokines platelet factor-4 (CXCL4/PF-4) and its variant

(CXCL4L1/PF-4var) in inflammation, angiogenesis and cancer.

Cytokine Growth Factor Rev 22:1–18

20. Johnson Z, Proudfoot AE, Handel TM (2005) Interaction of

chemokines and glycosaminoglycans: a new twist in the regula-

tion of chemokine function with opportunities for therapeutic

intervention. Cytokine Growth Factor Rev 16:625–636

21. Kasper B, Petersen F (2011) Molecular pathways of platelet

factor 4/CXCL4 signaling. Eur J Cell Biol 90:521–526

22. Smit MJ, Verdijk P, Raaij-Helmer EMH, Navis M, Hensbergen

PJ, Leurs R, Tensen CP (2003) CXCR3-mediated chemotaxis of

human T cells is regulated by a G(i)- and phospholipase

C-dependent pathway and not via activation of MEK/p44/p42

MAPK nor Akt/PI-3 kinase. Blood 102:1959–1965

23. Korniejewska A, McKnight AJ, Johnson Z, Watson ML, Ward

SG (2011) Expression and agonist responsiveness of CXCR3

variants in human T lymphocytes. Immunology 132:503–515

24. Proost P, Schutyser E, Menten P, Struyf S, Wuyts A, Opdenakker

G, Detheux M, Parmentier M, Durinx C, Lambeir AM, Neyts J,

Liekens S, Maudgal PC, Billiau A, Van Damme J (2001) Amino-

terminal truncation of CXCR3 agonists impairs receptor signaling

and lymphocyte chemotaxis, while preserving antiangiogenic

properties. Blood 98:3554–3561

25. Struyf S, Noppen S, Loos T, Mortier A, Gouwy M, Verbeke H,

Huskens D, Luangsay S, Parmentier M, Geboes K, Schols D, Van

Damme J, Proost P (2009) Citrullination of CXCL12 differentially

reduces CXCR4 and CXCR7 binding with loss of inflammatory

and anti-HIV-1 activity via CXCR4. J Immunol 182:666–674

26. Nawaz MI, Van Raemdonck K, Mohammad G, Kangave D, Van

Damme J, Abu El-Asrar AM, Struyf S (2013) Autocrine CCL2,

CXCL4, CXCL9 and CXCL10 signal in retinal endothelial cells

and are enhanced in diabetic retinopathy. Exp Eye Res 109:67–76

27. Kasper B, Brandt E, Bulfone-Paus S, Petersen F (2004) Platelet

factor 4 (PF-4)-induced neutrophil adhesion is controlled by src-

kinases, whereas PF-4-mediated exocytosis requires the addi-

tional activation of p38 MAP kinase and phosphatidylinositol

3-kinase. Blood 103:1602–1610

28. Bonacchi A, Romagnani P, Romanelli RG, Efsen E, Annunziato F,

Lasagni L, Francalanci M, Serio M, Laffi G, Pinzani M, Gentilini P,

Marra F (2001) Signal transduction by the chemokine receptor

CXCR3 - Activation of Ras/ERK, Src, and phosphatidylinositol

3-kinase/Akt controls cell migration and proliferation in human

vascular pericytes. J Biol Chem 276:9945–9954

29. Sulpice E, Bryckaert M, Lacour J, Contreres JO, Tobelem G

(2002) Platelet factor 4 inhibits FGF2-induced endothelial cell

proliferation via the extracellular signal-regulated kinase pathway

but not by the phosphatidylinositol 3-kinase pathway. Blood

100:3087–3094

30. Feldman ED, Weinreich DM, Carroll NM, Burness ML, Feldman

AL, Turner E, Xu H, Alexander HR Jr (2006) Interferon gamma-

inducible protein 10 selectively inhibits proliferation and induces

apoptosis in endothelial cells. Ann Surg Oncol 13:125–133

31. Fenton TR, Gout IT (2011) Functions and regulation of the

70 kDa ribosomal S6 kinases. Int J Biochem Cell B 43:47–59

32. Roca H, Varsos Z, Pienta KJ (2008) CCL2 protects prostate

cancer PC3 cells from autophagic death via phosphatidylinositol

3-Kinase/AKT-dependent survivin up-regulation. J Biol Chem

283:25057–25073

33. Hwang JH, Hwang JH, Chung HK, Kim DW, Hwang ES, Suh

JM, Kim H, You KH, Kwon OY, Ro HK, Jo DY, Shong M (2003)

CXC chemokine receptor 4 expression and function in human

anaplastic thyroid cancer cells. J Clin Endocrinol Metab

88:408–416

34. Hidalgo M, Rowinsky EK (2000) The rapamycin-sensitive signal

transduction pathway as a target for cancer therapy. Oncogene

19:6680–6686

35. Bullard LE, Qi X, Penn JS (2003) Role for extracellular signal-

responsive kinase-1 and-2 in retinal angiogenesis. Invest New

Drug 44:1722–1731

36. Kuo JH, Chen YP, Liu JS, Dubrac A, Quemener C, Prats H,

Bikfalvi A, Wu WG, Sue SC (2013) Alternative C-terminal helix

orientation alters chemokine function: structure of the anti-

angiogenic chemokine, CXCL4L1. J Biol Chem

288:13522–13533

37. Shahabuddin S, Ji R, Wang P, Brailoiu E, Dun N, Yang Y, Aksoy

MO, Kelsen SG (2006) CXCR3 chemokine receptor-induced

chemotaxis in human airway epithelial cells: role of p38 MAPK

and PI3 K signaling pathways. Am J Physiol Cell Physiol

291:C34–C39

38. Willox I, Mirkina I, Westwick J, Ward SG (2010) Evidence for

PI3K-dependent CXCR3 agonist-induced degranulation of

human cord blood-derived mast cells. Mol Immunol

47:2367–2377

39. Groom JR, Luster AD (2011) CXCR3 ligands: redundant, col-

laborative and antagonistic functions. Immunol Cell Biol

89:207–215

40. Alitalo A, Detmar M (2012) Interaction of tumor cells and

lymphatic vessels in cancer progression. Oncogene

31:4499–4508

41. Zhuo W, Jia L, Song N, Lu XA, Ding Y, Wang X, Song X, Fu Y,

Luo Y (2012) The CXCL12-CXCR4 chemokine pathway: a

novel axis regulates lymphangiogenesis. Clin Cancer Res

18:5387–5398

42. Choi I, Lee YS, Chung HK, Choi D, Ecoiffier T, Lee HN, Kim

KE, Lee S, Park EK, Maeng YS, Kim NY, Ladner RD, Petasis

NA, Koh CJ, Chen L, Lenz HJ, Hong YK (2013) Interleukin-8

reduces post-surgical lymphedema formation by promoting

lymphatic vessel regeneration. Angiogenesis 16:29–44

Angiogenesis

123