| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Ludwig Institute for Cancer Research, Royal Melbourne Hospital, Victoria, Australia;
Monash University, Department of Obstetrics and Gynaecology, Monash Medical Centre, Clayton, Victoria, Australia; and
Molecular/Cancer Biology Laboratory, Ludwig Institute for Cancer Research, University of Helsinki, Helsinki, Finland
3Correspondence: Ludwig Institute for Cancer Research, Post Office Box 2008 Royal Melbourne Hospital, Victoria 3050, Australia. E-mail: marc.achen{at}ludwig.edu.au
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
17,000-fold lower affinity than mature VEGF-D, indicating the importance of processing for interaction with this receptor. Further, we show that members of the proprotein convertase (PC) family of proteases promote VEGF-D processing, which facilitates the VEGF-D/VEGFR-2 interaction. The PCs furin and PC5 promote cleavage of both propeptides, whereas PC7 promotes cleavage of the C-terminal propeptide only. The finding that PCs promote activation of VEGF-D and other proteins with roles in cancer such as matrix metalloproteinases, emphasizes the importance of these enzymes as potential regulators of tumor progression and metastasis.—McColl, B. K., Paavonen, K., Karnezis, T., Harris, N. C., Davydova, N., Rothacker, J., Nice, E. C., Harder, K. W., Roufail, S., Hibbs, M. L., Rogers, P. A. W., Alitalo, K., Stacker, S. A., Achen, M. G. Proprotein convertases promote processing of VEGF-D, a critical step for binding the angiogenic receptor VEGFR-2.
Key Words: furin • PC5 • PC7 • angiogenesis • lymphatic
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). VEGF-D (2
, 3)
is a member of this family that can induce both angiogenesis and lymphangiogenesis (4
5
6)
by activating VEGF receptor-2 (VEGFR-2) and VEGFR-3 (2)
, receptor tyrosine kinases located on the surface of endothelial cells (7
, 8)
. VEGF-D promoted tumor angiogenesis and lymphangiogenesis, leading to enhanced tumor growth and metastatic spread of tumor cells via the lymphatic vessels in animal models of cancer (9
, 10)
. Furthermore, clinicopathological studies have suggested that expression of VEGF-D in prevalent human cancers can correlate with metastatic spread and patient survival (for review, see refs. 11
, 12
).
VEGF-D consists of a central VEGF homology domain (VHD), containing the binding sites for VEGFR-2 and VEGFR-3, and N- and C-terminal propeptides flanking the VHD (2)
. It is initially secreted from the cell as homodimers of the full-length protein that can be proteolytically processed to remove the N- and C-terminal propeptides (13
, 14)
. Previous studies involving pulse-chase analysis indicated that VEGF-D is predominantly processed in the extracellular environment—no intracellular processing was observed (13)
. Removal of both propeptides was shown to increase the affinity of VEGF-D for its receptors, although these studies were carried out using full-length VEGF-D that could be processed—hence, the preparations of full-length VEGF-D could have contained traces of partially processed and mature VEGF-D (13)
. Nevertheless, these studies indicated that VEGF-D processing is important for receptor interactions and that the proteolytic enzymes which activate VEGF-D may be important regulators of tumor angiogenesis and lymphangiogenesis. Further evidence from in vivo studies using adenoviral gene delivery to rabbit hind limb skeletal muscle confirmed that VEGF-D processing is a critical determinant of its biological effects, as fully processed VEGF-D induced both angiogenesis and lymphangiogenesis, whereas expression of the full-length form induced only lymphangiogenesis (4)
. This observation is consistent with the hypothesis that processing is particularly important for the interaction of VEGF-D with VEGFR-2, the receptor that signals for angiogenesis (7)
.
Although it has been shown that the serine protease plasmin can activate VEGF-D in vitro (15)
, the proteolytic enzymes that activate VEGF-D in vivo are not well characterized. Proprotein convertases (PCs) are a family of serine proteases that activate a range of proproteins with important roles in cancer biology, including growth factors, growth factor receptors, integrins, and matrix metalloproteinases (for a review, see ref. 16
). The PCs are active within organelles of the secretory pathway or at the cell surface, and they can be shed into the extracellular milieu. The majority of PCs (furin, PC1/PC3, PC2, PC4, PACE4, PC5/PC6, and PC7/PC8/LPC) activate their substrate proteins by cleavage immediately after the consensus sequence (K/R)-(X)n-(K/R) where n = 0, 2, 4 or 6 residues (17)
. Some of the substrates for PCs, such as TGF-ß (18)
, platelet-derived growth factor (PDGF)-A (17)
, and VEGF-C (19)
, are members of the cystine-knot structural superfamily of growth factors as are all members of the VEGF family of ligands (20)
.
Here, we compare a mutant of full-length VEGF-D, that cannot be proteolytically processed, with mature fully processed VEGF-D to monitor the effects of processing on receptor binding. This revealed a dramatic enhancement of VEGFR-2 binding affinity due to processing, the degree of which had been previously underestimated (13)
, which can explain the distinct effects of different forms of VEGF-D in vivo. Further, we identify members of the PC family of proteases that can promote the proteolytic processing of VEGF-D and thereby generate bioactive forms of this growth factor.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Inhibition of PCs
PCs were inhibited with the broad range PC inhibitor Decanoyl-RVKR-chloromethylketone (dec-RVKR-CMK; Bachem) dissolved in methanol. The 293EBNA cell lines used expressed either VEGF-D-FULL-N-FLAG or VEGF-D
C-FLAG, described previously (13)
, or VEGF-DN-FLAG, a truncated derivative of human VEGF-D consisting of the VHD, tagged at the amino terminus with the FLAG epitope, and the C-terminal propeptide. These cell lines were seeded into 24-well plates at a density of 8 x 104 cells/well in fresh medium supplemented with dec-RVKR-CMK to a final concentration of 1, 10, 50, or 100 µM. Solvent controls were carried out by culturing cells in medium supplemented with equivalent concentrations of methanol alone. Cells were incubated for 24 h and VEGF-D protein in the conditioned media was immunoprecipitated and analyzed by Western blot analysis.
Western blot analysis
VEGF-D proteins were immunoprecipitated using A2 rabbit polyclonal antiserum that binds the VHD (13)
or M2 (anti-FLAG) antibody (Sigma-Aldrich) and were analyzed by Western blot analysis with biotinylated anti-VEGF-D polyclonal antibodies recognizing the VHD (R&D Systems, Minneapolis, MN, USA) followed by detection using streptavidin-horseradish peroxidase conjugate and chemiluminescence (Pierce Biotechnology, Rockford, IL, USA), essentially as described previously (13
, 21)
. For analysis of furin, conditioned media were collected or cells were lysed at 24-hour intervals after transfection. Cells were lysed in PBS, 1% Triton X-100 at 4°C and cleared by centrifugation at 14,000 rpm for 10 min. SDS-PAGE and Western blot analysis was carried out essentially as described previously (13)
using monoclonal anti-human furin antibody (MON-152 antibody, Sapphire Bioscience, Redfern, NSW, Australia) with detection using HRP-conjugated rabbit anti-mouse antibodies (Bio-Rad Laboratories, Hercules, CA, USA) and chemiluminescence (Pierce Biotechnology).
Plasmid constructs
The coding regions for human furin, PC5, and PC7 were amplified by polymerase chain reaction (PCR) with Pfx DNA polymerase (Invitrogen) from plasmids containing the cDNAs for these proteins (ATCC). The oligonucleotides used were for furin: 5'-GCG AAG CTT CCA AGG AGA CGG GCG CTC CAG GG and 5'-GCG TCT AGA TCA TCA GAG GGC GCT CTG GTC TTT; for PC5: 5'-GCG GCG GCC GCC CTT AGT GCG CGG AAC CAG CCA and 5'-GCG TCT AGA TCA TCA GCC TTG AAA TGT ACA TGT T; for PC7: GCG AAG CTT GCA CAA CAT GAG TGT GAC GTG G and 5'-GCG TCT AGA TCA TCA GCA GAT CTG CTC CTC CTT. The PCRs involved an initial incubation for 5 min at 94°C, followed by 15 cycles of 94°C for 30 s (denaturation) followed by 52°C for 30 s (annealing) and 68°C for 3 min (extension). PCR products were cloned into pcDNA3 (Invitrogen) and verified by nucleotide sequencing, and the expression constructs for furin, PC5 and PC7 were designated pcDNA3:furin, pcDNA3:PC5 and pcDNA3:PC7, respectively.
Transfection of mammalian cells
293EBNA and LoVo cells were transiently transfected with expression plasmids using Genejuice (Merck Biosciences, Whitehouse Station, NJ, USA) or Fugene 6 (Roche Diagnostics, Indianapolis, IN, USA) transfection reagents according to manufacturers’ instructions. Cells stably transfected with furin construct were allowed to reach 40–50% confluency and were then transfected using lipofectamine transfection reagent (Invitrogen) according to manufacturer’s protocol.
Analysis of binding to soluble receptors
Plasmid constructs expressing soluble fusion proteins consisting of the extracellular domains of human VEGFR-1, VEGFR-2, or VEGFR-3 joined to the Fc portions of human IgG, referred to as VEGFR-1-Ig (R&D Systems), VEGFR-2-Ig (22)
, and VEGFR-3-Ig (23)
were transiently transfected into 293EBNA cells. After 24 h, the culture medium was removed and replaced with medium lacking FBS. Cells were incubated for a further 48 h, and conditioned media were removed and incubated with protein A-sepharose beads overnight, then washed twice in binding buffer (0.5% (w/v) BSA, 0.02% (v/v) Tween 20, 10 µg/ml heparin in PBS). Receptor/Ig constructs coupled to the protein A-sepharose were incubated for 3 h at room temperature with conditioned media from LoVo cells expressing VEGF-D, collected by centrifugation and washed twice with binding buffer. The bound VEGF-D was subsequently analyzed by Western blot analysis.
Site-directed mutagenesis
The DNA encoding VEGF-D-FULL-N-FLAG (13)
was subjected to mutagenesis by PCR with the Quikchange Site-Directed Mutagenesis Kit (Stratagene) essentially as described by the manufacturer. The R88S mutation was introduced by amplification with the following primers: 5'-GCA TCC CAT CGG TCC ACT TCC TTT GCG GCA ACT TTC TAT G and 5'-CAT AGA AAG TTG CCG CAA AGG AAG TGG ACC GAT GGG ATG C; the R85S mutation was made in the resulting plasmid by amplification with 5'-CTC GCT CAG CAT CCC ATT CCT CCA CTT CCT TTG CGG and 5'-CCG CAA AGG AAG TGG AGG AAT GGG ATG CTG AGC GAG and the R204S and R205S mutations were both introduced into the resulting plasmid by amplification with 5'-CCA TAC TCA ATT ATC AGC AGC TCC ATC CAG ATC CCT GAA G and 5'-CTT CAG GGA TCT GGA TGG AGC TGC TGA TAA TTG AGT ATG G. The desired mutations, as well as the absence of any unwanted mutations, in the resulting DNA fragments were confirmed by nucleotide sequencing.
Furin knockdown
shRNA primers targeting human furin mRNA were designed using an in-house primer design algorithm (K. W. Harder and S. Lay). Two shRNA primers were generated to target the sequences 5'-GAC GAT GGC ATC GAG AAG AA-3' and 5'-CAC ACA GAT GAA TGA CAA-3'. Primers were annealed by heating a mix of 60 pM of both primers in 50 µl annealing buffer (0.1 M potassium acetate, 2 mM magnesium acetate, 30 mM HEPES pH 7.4) at 94°C for 4 min then 70°C for 10 min and allowed to cool to room temperature. Annealing was verified in 3% low melting point agarose gel. The shRNA construct was then ligated into pMSCVpac mU6, which is a derivative of pMSCV (Clontech). The shRNAs were validated by assessing reduction of furin protein levels, resulting from cotransfections of shRNA and furin expression constructs in 293EBNA cells or other cell lines stably overexpressing furin.
Immunohistochemistry
Tissue sections for immunohistochemistry were cut from archival, paraffin-embedded samples of human aorta and human endometrium obtained from the Centre for Women’s Health Research Tissue Bank under the ethics approval of Southern Health Human Research Ethics Committee C. Primary antibodies were mouse monoclonal anti-human D2–40 (clone D2–40 from Signet Laboratories), mouse monoclonal anti-human furin (MON-152 from Alexis Biochemicals), and mouse monoclonal anti-human VEGF-D (MAB286 from R&D Systems) with mouse IgG1 (Dako) as isotype-matched control in all cases. The D2–40 antibody was detected using biotinylated secondary antibody followed by streptavidin conjugated to alkaline phosphatase (Universal LSAB+ Kit; Dako, Copenhagen, Denmark) and visualization using the Vector Blue Substrate Kit (Vector Laboratories, Burlingame, CA, USA). The furin and VEGF-D antibodies were detected using biotinylated secondary antibodies (Zymed, Burlingame, CA, USA) followed by horseradish peroxidase-conjugated streptavidin (Vectastain ABC Kit; Vector Laboratories) and visualization using diaminobenzidine tetrahydrochloride (Sigma, St. Louis, MO, USA).
Biosensor analysis
Soluble fusion proteins consisting of the extracellular domains of human VEGFR-2 or VEGFR-3 joined to the Fc portions of human IgG (R&D Systems) were coupled to the carboxymethylated dextran layer of a sensor chip using NHS/EDC amine coupling chemistry for analysis of the binding kinetics using a BIAcore 3000 optical biosensor (Biacore, Uppsala, Sweden) (24)
. Following immobilization of the receptors, the residual activated ester groups were blocked by treatment with 1 M ethanolamine hydrochloride, pH 8.5, followed by washing with 10 mM diethanolamine to remove noncovalently bound material. Samples for analysis were diluted in HBS running buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, 3.4 mM EDTA, 0.005% Tween 20). The integrity of the bound VEGFR-2 and VEGFR-3 was assessed by binding of purified VEGF-DNC-FLAG (13)
. Data were analyzed using BIAevaluation 4.1 (BIACORE, Uppsala, Sweden) using regions of the curves where a 1:1 Langmurian model appeared to be operative. The VEGF-D ligands used in this analysis were purified by anti-FLAG affinity chromatography as described previously (13)
.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
) and at positions 204 and 205 (adjacent to the site at which the C-terminal propeptide is cleaved from the VHD between positions 205 and 206 (13)
) to serine residues (Fig. 1
A). The resulting mutant derivative was designated VEGF-DSSTS.IISS and was expressed in 293EBNA cells in comparison to a full-length form of VEGF-D, in which the sequences at the cleavage sites had not been mutated (this derivative was previously designated VEGF-D-FULL-N-FLAG; see Fig. 1A
, ref. 13
). As expected, a portion of VEGF-D-FULL-N-FLAG in the conditioned culture media was proteolytically processed as indicated by the presence of 31 kDa and 21 kDa bands in addition to full-length material of 50 kDa (Fig. 1B
). The 31-kDa band was shown previously to be a species consisting of the N-terminal propeptide and the VHD, whereas the 21-kDa band corresponds to mature VEGF-D (13)
. In contrast, all of the VEGF-DSSTS.IISS from the conditioned culture media consisted of the full-length 50 kDa form (Fig. 1B
). This finding demonstrated that the mutations in VEGF-DSSTS.IISS prevent cleavage of both propeptides from the VHD and that proteolytic processing is not required for VEGF-D to be secreted from the cell. In addition, these results show that mutation of the site most commonly used for cleavage of the N-terminal propeptide from the VHD (the "Major" site; see Fig. 1A
) is sufficient to completely block cleavage of the N-terminal propeptide, suggesting that in wild-type VEGF-D, cleavage at the "Minor" site may only occur after cleavage at the "Major" site.
|
To determine the relative affinities of the ligands, VEGF-DSSTS.IISS, VEGF-D-FULL-N-FLAG and mature VEGF-D (designated VEGF-DNC (13)
) for VEGFR-2 and VEGFR-3, we analyzed the relative binding kinetics for these interactions by biosensor analysis using surface plasmon resonance detection (23)
. The preparations of VEGF-DSSTS.IISS and VEGF-DNC used for this analysis each consisted of one VEGF-D species devoid of other forms of VEGF-D, but the preparation of VEGF-D-FULL-N-FLAG did contain traces of partially processed, as well as full-length material, reflecting the capacity of wild-type full-length VEGF-D to be proteolytically processed in the 293EBNA cell cultures (13)
from which these proteins were purified (see Fig. 1B
). The binding constants for the interactions were obtained by analysis of the initial dissociation phase to obtain the kd, which was then used to constrain a global analysis of the association region of the curves, using regions of the curves where a 1:1 Langmurian model appeared to be operative. Significantly, the apparent affinity of the interaction between mature VEGF-D and VEGFR-2 appeared 17,000-fold greater than that for the unprocessed mutant VEGF-DSSTS.IISS (Fig. 1C
), indicating the critical importance of processing for the binding of VEGF-D to this receptor. In contrast, the affinity of mature VEGF-D for VEGFR-3 was only 18-fold greater than that for VEGF-DSSTS.IISS. The affinity of the VEGF-D-FULL-N-FLAG preparation for VEGFR-2 was 94-fold greater than that for VEGF-DSSTS.IISS, reflecting the presence of processed derivatives in the preparation of VEGF-D-FULL-N-FLAG.
An inhibitor of PCs blocks processing of VEGF-D
The sites at which the N- and C-terminal propeptides of VEGF-D are cleaved from the VHD are immediately after RSTR88 and IIRR205, respectively (although there is also a secondary "minor" site for cleavage of the N-terminal propeptide, used less frequently than RSTR88, immediately after IETL99) (Fig. 1A
) (13)
and conform to the consensus cleavage site for PCs, at which cleavage occurs immediately after (K/R)-(X)n-(K/R), where n = 0, 2, 4, or 6 residues (17)
. Furthermore, PCs are known to process other members of the cystine-knot superfamily of growth factors (18
, 19
, 25)
, of which VEGF-D is a member. We therefore examined the possibility that PCs promote processing of VEGF-D.
To establish if any of the PCs might be involved in VEGF-D processing, the broad-range PC inhibitor Decanoyl-Arg-Val-Lys-Arg-chloromethylketone (dec-RVKR-CMK) (26)
was assessed for its capacity to block this processing in the 293EBNA cell line that was shown previously to process VEGF-D (13)
. Dec-RVKR-CMK contains a consensus PC cleavage sequence, the Arg-Val-Lys-Arg peptide moiety, and binds to the substrate recognition site of the PCs via this moiety, leading to irreversible inhibition of the enzyme due to alkylation of the active site by the chloromethylketone group (26)
. Dec-RVKR-CMK inhibits all PCs, with Ki values in the low nM range (27)
. This inhibitor was tested on 293EBNA cell lines expressing three distinct derivatives of human VEGF-D, 1) full-length human VEGF-D tagged at the N terminus with the FLAG epitope (VEGF-D-FULL-N-FLAG); 2) VEGF-DN-FLAG, a derivative of human VEGF-D lacking the N-terminal propeptide that is tagged with the FLAG epitope at the N terminus of the VHD; 3) VEGF-DC-FLAG, a derivative of human VEGF-D, in which the C-terminal propeptide has been deleted and replaced with the FLAG epitope (13)
.
The 293EBNA cell lines were cultured in media supplemented with dec-RVKR-CMK, and secreted VEGF-D proteins from cell culture supernatants were immunoprecipitated and analyzed by Western blot analysis with antibodies that recognize the VHD. Treatment of the 293EBNA VEGF-D-FULL-N-FLAG cells with 100 µM dec-RVKR-CMK (Fig. 2
A) resulted in complete inhibition of VEGF-D processing, as demonstrated by the absence of the lower MW VEGF-D species generated by proteolysis, namely the 31-kDa species consisting of the N-terminal propeptide and VHD and of the 21 kDa VHD species (13)
. At lower concentrations of dec-RVKR-CMK, the proteolysis of VEGF-D-FULL-N-FLAG was incompletely inhibited, indicating the dose-dependent effect of the inhibitor. We next examined the effect of dec-RVKR-CMK on processing of VEGF-DN-FLAG and VEGF-DC-FLAG to monitor the role of PCs in the cleavage of each propeptide individually (Fig. 2B, C
). In the absence of the inhibitor, a portion of these proteins is processed to generate the 21-kDa mature VHD. However, the inhibitor blocked both of these processing events in a dose-dependent fashion. These findings demonstrate that dec-RVKR-CMK blocks cleavage of both the N- and C-terminal propeptides from VEGF-D and thereby suggests that members of the PC family play a role in promoting these proteolytic events. We next tested the effect of dec-RVKR-CMK on VEGF-D processing by Balbc/3T3 cells that naturally produce this growth factor (i.e., in cells that had not been transfected so as to express VEGF-D). Incubation of these cells with 100 µM dec-RVKR-CMK almost completely blocked processing (Fig. 2D
), indicating the involvement of the PC family in processing of VEGF-D by these cells.
|
Furin, PC5, and PC7 promote processing of VEGF-D
The LoVo human colon carcinoma cell line lacks enzymatically active furin due to the presence of mutations in both alleles of the Furin gene (28
, 29)
. To establish whether LoVo cells might be unable to process VEGF-D, they were transiently transfected with expression constructs for VEGF-D-FULL-N-FLAG, VEGF-DN-FLAG, and VEGF-DC-FLAG, and the expressed proteins in cell culture supernatants were analyzed by immunoprecipitation and Western blot analysis to examine VEGF-D processing in a furin-deficient background (Fig. 3
A). Results were compared with processing of the same proteins in 293EBNA cells. None of the three VEGF-D derivatives was processed at all on expression in LoVo cells, whereas all three were proteolytically processed to some degree when expressed in 293EBNA cells (Fig. 3A
). Even prolonged exposure of the blots did not reveal any processed VEGF-D derivatives in the conditioned media of the transfected LoVo cells (data not shown). These findings demonstrate that LoVo cells are incapable of cleaving either the N- or C-terminal propeptides from the VHD of VEGF-D and therefore offer a processing-deficient background in which to assess the capability of PCs to promote processing of VEGF-D.
|
Cotransfection experiments with expression constructs for VEGF-D and various PCs were carried out in LoVo cells to analyze the effect of individual PCs on VEGF-D processing. Furin, PC5, and PC7 were chosen for these studies, as they are broadly expressed in the adult (30
, 31)
, as is VEGF-D (2)
, enhancing the likelihood of localization of these PCs and VEGF-D in the same regions of tissue in vivo. LoVo cells were transiently transfected with expression constructs for VEGF-D-FULL-N-FLAG, VEGF-DN-FLAG, or VEGF-DC-FLAG in combination with constructs for furin, PC5, or PC7, and the processing of VEGF-D was analyzed by immunoprecipitation/Western blot analysis of cell culture supernatants (Fig. 3B
). No processing of VEGF-D-FULL-N-FLAG was observed in the absence of PC expression constructs, as expected; however, cotransfection with furin resulted in almost complete conversion of the 50 kDa full-length VEGF-D to the 21-kDa mature VHD species (Fig. 3B
, top), indicating that furin can promote cleavage of both the N- and C-terminal propeptides from the VHD. Cotransfection with PC5 resulted in partial conversion of full-length VEGF-D to the mature form (21 kDa) and to a form consisting of the N-terminal propeptide and the VHD (31 kDa), showing that PC5 can also promote cleavage of both propeptides from the VHD. Cotransfection with PC7 led to partial conversion of VEGF-D to the 31-kDa form containing the N-terminal propeptide and the VHD but not to the mature VHD species (this form could not be detected even after prolonged exposure of the blots), suggesting that PC7 can promote cleavage of the C-terminal propeptide, but not the N-terminal propeptide, from the VHD. In none of these cases was a 44-kDa species containing the VHD and C-terminal propeptide detected, indicating that the C-terminal propeptide is cleaved much more efficiently than the N-terminal propeptide. These findings were further supported by the results of cotransfections with VEGF-DN-FLAG or VEGF-DC-FLAG (Fig. 3B
, bottom two panels). As expected, furin, PC5, and PC7 promoted cleavage of the C-terminal propeptide from VEGF-DN-FLAG, whereas furin and PC5, but not PC7, promoted cleavage of the N-terminal propeptide from VEGF-DC-FLAG.
Given that furin is considered to be a very broadly expressed PC and its up-regulation has been correlated with tumor progression and invasiveness (for a review, see (32)
), we further explored the involvement of this enzyme in VEGF-D processing by targeting its expression in HeLa cells using shRNA. The furin shRNA construct used was demonstrated to reduce furin protein levels in the cell considerably (Fig. 3C
). Furthermore, this shRNA construct almost totally blocked VEGF-D processing in HeLa cell cultures as the relative abundance of mature VEGF-D (21 kDa) and of the species consisting of the N-terminal propeptide and the VHD (31 kDa) was greatly reduced in comparison to full-length VEGF-D (50 kDa) in the presence of the furin shRNA construct (Fig. 3C
).
Furin promotes formation of mature VEGF-D that binds VEGF receptors
Coexpression of full-length VEGF-D and furin in LoVo cells revealed that this protease can promote cleavage of the propeptides from the VHD. To test whether the mature form of VEGF-D generated in the presence of furin is functional, we studied the capacity of this cleaved VEGF-D to bind soluble chimeric proteins containing the extracellular domain of VEGFR-2 or VEGFR-3 joined to the Fc portions of human IgG (designated VEGFR-2-Ig or VEGFR-3-Ig). VEGF-D proteins in the medium of LoVo cells cotransfected with plasmid expression constructs for VEGF-D-FULL-N-FLAG and furin were precipitated with VEGFR-2-Ig and VEGFR-3-Ig and analyzed by Western blot analysis with antibodies that bind the VHD (Fig. 4
A, B). Both VEGFR-2-Ig and VEGFR-3-Ig precipitated the 21-kDa VHD species from the medium, demonstrating that furin promotes formation of mature VEGF-D that can bind the same receptors as the previously characterized mature form of VEGF-D expressed in 293EBNA cells (13)
.
|
Localization of VEGF-D and furin in vivo
The studies above indicate that furin can promote proteolytic processing of VEGF-D when these molecules are expressed in cell lines in culture. However, the question arises as to whether or not VEGF-D and furin are similarly localized in human tissue such that this processing could occur in vivo. Given the role of VEGF-D as an angiogenic and lymphangiogenic growth factor, this issue was explored by studying the localization of VEGF-D and furin in the human aorta that contains a well-developed lymphatic network in its adventitia (Fig. 5
). The lymphatic network at the outer edge of the adventitia was detected using the D2–40 antibody (Fig. 5A
), which binds to the mucin-type glycoprotein podoplanin and stains the endothelium of lymphatic vessels but not of blood vessels (33
, 34)
. VEGF-D was detected near these lymphatic vessels in a region of what histopathologically appeared to be a perivascular mononuclear inflammatory cell infiltrate around and adjacent to small vessels in the adventitia (Fig. 5B
). The mononuclear inflammatory cells were also positive for furin (Fig. 5C
). These overlapping staining patterns indicate that, at least in inflammatory cells located in this region of tissue, furin and VEGF-D can be expressed in the same cells, so these proteins are in sufficiently close proximity for a direct interaction to occur between them. We explored the expression of these proteins further in human endometrium, a tissue rich in blood vessels, using thin serial sections to allow identification of cells expressing both VEGF-D and furin. Many cells were identified associated with the luminal epithelium and the stroma that were immunopositive for both VEGF-D and furin (Fig. 5E, F
), demonstrating that these two proteins can be located in the same cells in vivo.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
17,000-fold greater affinity for VEGFR-2 than unprocessed material. It was previously reported that mature VEGF-D has a 290-fold greater affinity for VEGFR-2 than full-length material; however, the analysis was conducted with a form of full-length VEGF-D that could be processed and, hence, this could have been contaminated with partially processed VEGF-D or mature VEGF-D (13)
. Thus the effect of VEGF-D processing on the affinity for VEGFR-2 was likely underestimated. Our finding that processing has such a dramatic effect on VEGFR-2 binding is consistent with previous studies based on adenoviral delivery into rabbit hind limb skeletal muscle, which showed that full-length VEGF-D could promote lymphangiogenesis but not angiogenesis, whereas mature VEGF-D, lacking both propeptides, promoted both lymphangiogenesis and angiogenesis (4)
. Given that VEGFR-2 signals predominantly for angiogenesis (7)
, whereas VEGFR-3 signals predominantly for lymphangiogenesis (6)
, these in vivo findings correlate well with the capacity of mature VEGF-D to bind both VEGFR-2 and VEGFR-3 with high affinity in contrast to full-length VEGF-D, which binds only VEGFR-3 with high affinity (Fig. 6
).
|
Given the importance of the proteolytic processing of VEGF-D for receptor interactions, we sought to investigate the enzymes responsible for this processing. Preliminary studies indicated that inhibition of all PCs with dec-RVKR-CMK blocks proteolytic processing of human VEGF-D and that the putative PC cleavage sites near each end of the VHD are required for cleavage of the N- and C-terminal propeptides. Furthermore, VEGF-D does not undergo any processing in LoVo cells lacking active furin, although processing can be established in these cells by expressing furin, PC5, or PC7; furin and PC5 being able to promote removal of both propeptides, whereas PC7 promotes removal of the C-terminal propeptide only (Fig. 6)
. The more restricted activity of PC7 is in agreement with the observation that a pair of adjacent basic residues, immediately preceding the scissile peptide bond, is essential for PC7 activity (35)
—such a pair of basic residues is not present in either the "Major" or "Minor" N-terminal propeptide cleavage sites of VEGF-D (Fig. 1A
). Importantly, the mature form of human VEGF-D generated on coexpression with furin binds VEGFR-2 and VEGFR-3, as does the previously characterized mature form expressed in 293EBNA cells (13
, 36)
. Further evidence for the involvement of furin in VEGF-D processing arose from studies with furin shRNA, which demonstrated that knockdown of furin expression led to a significant reduction in the degree of VEGF-D processing in HeLa cells.
Our previous studies of VEGF-D synthesis and secretion by pulse-chase analysis and other methods demonstrated that the proteolytic processing of this protein takes place predominantly in the extracellular environment (13)
. Although furin is localized within the trans-Golgi network (TGN), it also undergoes regulated trafficking through the endosomes to the cell surface (30
, 31
, 37)
, and a soluble form of active furin has been reported in the medium of several cell types (38
, 39)
, possibly the product of proteolytic removal of the furin transmembrane region by an undetermined protease. Hence, the localization of furin is consistent with the observation that VEGF-D is processed outside the cell or at the cell surface. Furthermore, purified full-length VEGF-D can be processed to the mature form in cell-free conditioned media in a PC-dependent manner (M. G. Achen and S. J. Loughran, unpublished observation), suggesting that a secreted or shed form of furin, or of another PC, contributes to VEGF-D processing. Similar localization of VEGF-D and furin in samples of human aorta and colocalization of these two proteins in cells in the human endometrium indicate that these proteins can be in sufficiently close proximity in vivo for a direct interaction between them to occur.
The VHD of VEGF-D can have an alternative N terminus to that generated by cleavage of the N-terminal propeptide at the cleavage site analyzed in this study (13)
. The alternative N terminus is 11 amino acids C-terminal from the cleavage site studied here. The majority of mature VEGF-D detected in the conditioned medium of 293EBNA cells secreting full-length VEGF-D is generated by proteolysis at the cleavage site studied here; hence, this site is known as the "Major" site, and the alternative cleavage site is considered the "Minor" site (13)
. The sequence of the "Minor" site does not contain the basic residues required for PC-mediated cleavage, indicating that PCs are not responsible for proteolysis at this site. Inhibition of VEGF-D processing with dec-RVKR-CMK completely abrogated proteolytic cleavage of the N-terminal propeptide from the VHD, i.e., at both cleavage sites, suggesting that PC-induced cleavage of the N-terminal propeptide at the "Major" site may be a prerequisite for subsequent cleavage at the "Minor" site by another protease.
Expression of PCs is upregulated in several human cancers, including those of the lung, breast, and head and neck (40
41
42)
, as well as in numerous tumor cell lines (32)
. Overexpression of PCs in tumors correlates with the development of a more aggressive phenotype (32
, 42)
presumably via activation of proproteins associated with tumor progression. Among the substrates of the PCs with potential roles in the development and progression of tumors are growth factors and receptors (e.g., TGF-ß (18)
, hepatocyte growth factor (43)
, PDGF-A (17)
, PDGF-B (25)
, insulin-like growth factor-1 (IGF-1) (44)
, IGF-1 receptor (45)
); molecules involved in cellular adhesion, such as integrins (46
, 47)
; and enzymes associated with degradation of the extracellular matrix (ECM) and metastasis (stromelysin-3 (48)
, membrane-type matrix metalloprotease 1 (49)
). We now know that processing of VEGF-D, a protein that promotes tumor growth and metastatic spread via the lymphatic vessels in animal models of cancer (9
, 10)
, can also be promoted by PCs as is the case for VEGF-C (19)
, a protein closely related to VEGF-D that also promotes lymphogenous metastasis (12
, 50
51
52)
. VEGF-C has the same arrangement of structural domains as VEGF-D (53)
, and it has been shown that furin, PC5, and PC7 promote cleavage of the C-terminal propeptide from the VHD of VEGF-C, although the enzyme(s) that cleave the N-terminal propeptide are unknown (19)
.
Manipulation of the PCs in animal models of cancer has demonstrated a relationship between the PCs and vascularization of tumors. Inhibition of the PCs in a xenograft tumor model resulted in a decrease in the incidence of tumor formation, as well as reduced tumor size relative to controls, in which PC activity was unperturbed (54)
. Furthermore, in the same study, inhibition of the PCs resulted in reduced tumor vascularization (54)
. In addition, Siegfried et al. showed that blocking conversion of pro-VEGF-C to mature VEGF-C, by mutation of a PC cleavage site, inhibited angiogenesis, lymphangiogenesis, and tumor growth in a mouse tumor model (19)
. The findings that PCs contribute to activation of proteins that play diverse, critical roles in tumor biology, including VEGF-D, indicate that these enzymes may be useful targets for novel anticancer therapeutics.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
2 Current address: University of British Columbia, Department of Microbiology and Immunology, Life Sciences Centre, 2350 Health Sciences Mall, Vancouver, BC, Canada.
Received for publication November 2, 2006. Accepted for publication November 28, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
-tubulin using cell lysates after transfection of the cells with plasmid encoding furin shRNA ("Furin shRNA"), expression vector lacking shRNA sequence ("Vector only"), or vector encoding an irrelevant shRNA that was not targeting furin ("Control shRNA"). The right panel shows Western blot analysis, using an antibody to the VHD of VEGF-D, of conditioned media from HeLa cells after transfection with the same plasmids as for the left panel. The domain compositions of the detected VEGF-D proteins are shown to the right with "N," "C," and "F" denoting the N- and C-terminal propeptides and the FLAG epitope, respectively. Sizes of Mr standards (kDa) are shown to the left.
