The Proteolytic Activation of Vascular Endothelial Growth Factor­C

antigen (PSA) and cathepsin D. Processing by diﬀerent proteases results in distinct forms of "ma­ ture" VEGF­C, that diﬀer in their aﬃnity and their receptor activation potential. This processing is tightly regulated by the CCBE1 protein. CCBE1 regulates the acti­ vating cleavage of VEGF­C by ADAMTS3 and PSA, but not by plasmin. During em­ bryonic development of the lymphatic sys­ tem, VEGF­C is activated primarily by the ADAMTS3 protease. In contrast, it is be­ lieved that plasmin is responsible for wound healing lymphangiogenesis and PSA for tumor­associated pathological lym­ phangiogenesis. Cathepsin D has also been implicated in tumor lymphangiogenesis. In addition, cathepsin D in saliva might acti­ vate latent VEGF­C upon wound licking, there by accelerating wound healing. The molecular details of proteolytic activation of VEGF­C are only recently extensively ex­ plored, and we likely do not know yet all ac ti vating proteases. It appears that the ac­ tivity of VEGF­C is regulated for diﬀerent spe ciﬁc functions by diﬀerent proteinases. Al though VEGF­C clearly plays a pivotal role for tumor progression and metastasis in experimental animal studies, the rele­ vance of most correlative studies on the role of VEGF­C in human cancers is quite li ­ mi ted until now, also due to the lack of meth ods to diﬀerentiate between inactive and active forms. Gene therapy with AdVEGF-C. An increasingly popular therapy for breast cancer­associated lymphedema is autologous lymph node transplantation [88,89]. In pre­clinical studies, the treatment success (integration of the transplanted lymph node into the local lymphatic network) could be improved by the simultaneous administration of VEGF­C [90]. With this strategy, Lymfactin® has successfully completed Phase I clinical trials and is now in Phase II. As a further development of Lymfactin® , a simultaneous administration of VEGF­C with the VEGF­C­activating ADAMTS3 and/or CCBE1 is being discussed.


Summary
The enzymatic cleavage of the protein back bone (proteolysis) is integral to many biological processes, e.g. for the break down of proteins in the digestive system. Specific proteolytic cleavages are also used to turn on or off the activity of proteins. For example, the lymphangiogenic vascular endothelial growth factorC (VEGFC) is synthesized as a precursor molecule that must be converted to a mature form by the enzymatic removal of C and Nterminal pro peptides before it can bind and activate its receptors. The constitutive Cterminal cleavage is mediated by proprotein con vertases such as furin. The subsequent ac tivating cleavage can be mediated by at least four different proteases: by plasmin, ADAMTS3, prostatespecific antigen (PSA) and cathepsin D. Processing by different proteases results in distinct forms of "ma ture" VEGFC, that differ in their affinity and their receptor activation potential. This processing is tightly regulated by the CCBE1 protein. CCBE1 regulates the acti vating cleavage of VEGFC by ADAMTS3 and PSA, but not by plasmin. During em bryonic development of the lymphatic sys tem, VEGFC is activated primarily by the ADAMTS3 protease. In contrast, it is be lieved that plasmin is responsible for wound healing lymphangiogenesis and PSA for tumorassociated pathological lym phangiogenesis. Cathepsin D has also been implicated in tumor lymphangiogenesis. In addition, cathepsin D in saliva might acti vate latent VEGFC upon wound licking, there by accelerating wound healing. The molecular details of proteolytic activation of VEGFC are only recently extensively ex plored, and we likely do not know yet all ac ti vating proteases. It appears that the ac tivity of VEGFC is regulated for different spe cific functions by different proteinases. Al though VEGFC clearly plays a pivotal role for tumor progression and metastasis in experimental animal studies, the rele vance of most correlative studies on the role of VEGFC in human cancers is quite li mi ted until now, also due to the lack of meth ods to differentiate between inactive and active forms.

Proteinases (protein cleaving enzymes)
Proteinases (or pro te ases) are enzymes that cleave proteins by hydrolysing the peptide bonds of the protein backbone ( Figure 1). ey occur in side (intracellular) and out side (extracellular) of cells, and are es sential for a multitude of cell and body functions. For example, proteinases process antigens in the course of an im mune reaction for antigen presentation, they break down damaged or un neces sary proteins (e.g. in lysosomes) and they digest food proteins in the gastro in tes tinal tract. In the stomach, for example, pepsin is generated from the precursor pepsinogen by autoproteolysis at a low pH, and in the intestine, trypsin is generated from the prec ursor trypsino gen by auto catalysis.

Activation of proteins
Many pro teins are produced as inactive pre cur sors and are activated by proteolytic cleavage when their function is required. e proteinases them selves are also produced as inactive pro-proteinases and must be acti vated by pro teolytic removal of their pro peptides. is is of uppermost impor tance, since the uncon trolled activi ty of proteinases would other wise destroy cells and decompose the extra cellular matrix (ECM). e perhaps best known pro teo ly tical ly con trolled processes include blood coagu la tion, the limitation of blood clotting and its reversal, i.e. the dis solution of blood clots [1,2]. Many bood clotting factors are proteinases, which in turn acti vate other pro te in ases, etc. ("proteolytic cas cade") to catalyze the proteolytic con version of so luble fibrinogen into poly merizing fibrin in the final step of the blot clotting cascade.

Activation of growth factors
Many growth factors and also some cyto kines are produced as inactive precur sors, which only become active through pro teo lytic cleavage ("pro cessing"). Among the better known growth factors that are activated through proteolytic cleavage are e.g. the Transforming Growth Factor-β (TGF-β) [3], but also the lymph an gioge nic growth factors VEGF-C and VEGF-D. Many studies have analyzed the me cha nisms and the regulation of VEGF re cep tor activation by VEGFs [4], whereas rela tive ly little is known about the upstream pro cesses of mo bili sation and activation of VEGFs.

The VEGF family
e biology of the growth factors VEGF-C and VEGF-D has been described in detail in a previous review Proteolysis. The hydrolytic cleavage of a peptide bond (orange) of a protein (blue background) into two fragments (red background). The peptide bonds of the protein backbone are shown as thick lines. The amino acid side chains are symbolized as green circles. Without enzymatic catalysis by proteinases this chemical reaction is extremely slow.
The growth factors VEGF-A, VEGF-C and VEGF-D and their receptors. The growth and function of blood and lymphatic vessels is controlled by Vascular Endothelial Growth Factors (VEGFs). VEGFA is the quintessential growth factor for blood vessels, while VEGFC is the quintessential growth factor for lymphatic vessels. VEGFA is recognised by VEGF receptor1 (VEGFR 1) and VEGF receptor2 (VEGFR2). VEGFC and VEGFD are recognised by VEGF receptor3 (VEGFR3) and, under certain circumstances, also by VEGFR2. VEGFR1 is largely specific for endothelial cells of blood vessels and VEGFR3 for endothelial cells of lymphatic vessels. In contrast, VEGFR2 is found on both vessel types. If, for example, active VEGFC or VEGFD binds to VEGFR3 on the lymphatic endothelial cell surface, the signal is transduced into the cell nucleus, where it provokes a proliferative and migratory response, thus initiating vessel growth.  [5]. For this reason, only a short introduction follows, in which the relevant pro perties and cha rac teristics of VEGF-C and VEGF-D are explained. VEGF-C and VEGF-D belong to the VEGF family (see also Figure 2 for a gra phical short overview of VEGF-A, VEGF-C, VEGF-D and their re ceptors).
Charac teristic for the members of the VEGF fa mily is the VEGF Homology Domain (VHD) as the central and do minant structural element. is domain is almost 100 amino acids long and has a characteristic arrangement of eight cysteine (C) amino acid residues (CX 22 CPXCVX 3 RCXGCCX 6 CX 33-35 CXC), which form disulfide bridges among them selves and thereby give the VEGFs a very stable core. is core also forms the re cep tor binding epitope and thus determines to which of the three VEGF receptors (-1, -2 and -3) a VEGF binds. In addi tion to this core, most VEGFs have other domains that are either upstream (N-terminal) or downstream (C-ter mi nal) from the VHD ( Figure 3). ese ad ditional domains give VEGFs the abi li ty to interact with other binding part ners.
For example, different isoforms of VEGF-A have C-terminal heparin binding domains of varying strength, with which they bind heparan sulfate proteo gycans. is variation in binding strength causes a more or less pro minent immobilization on cell sur faces and the extracellular matrix (ECM), which in turn results in distinct ac ti vity pro files of the isoforms. Proteases such as plasmin can convert the longer ECM-bound VEGF-A iso forms into shor ter, more diffusible iso forms [6,7]. Cleavage by different matrix me tallopro teinases (MMPs), especially MMP-3, converts e.g. the main isoform VEGF-A 165 into a shorter, non-heparan sul fate-binding isoform [8].

The C-terminal domain of VEGF-C
Similar to VEGF-A, VEGF-C and VEGF-D are also im mo bi lized on cell sur faces and the ECM via their C-terminal domain [9]. In contrast to all other VEGF family members, the Cter minal domain of VEGF-C and VEGF-D blocks the growth factor activity [10]. Most likely, this do main steri cally hinders access to the re cep tor bin ding site. is assumption would also explain why the C-terminal domain of VEGF-C is almost twice as large as its VHD. e origin of the protein sequence of the C-terminal domain is mysterious, since no homo logous sequences seem to exist in the genomes of verte brates. Homologous pro teins are, however, found in the sali vary secretions of some silk worm mos quito larvae, e.g. Chi ro no mus tentans [11]. For this reason, this domain has also been called silk homology do main, although its amino acid se quence is unre lated to the classical silk proteins.

Hypoxia regulates angiogenesis, but how is lymphangiogenesis regulated?
VEGF-A, which is mainly re spon sible for the for mation of blood vessels, is tightly regulated at the trans crip ti onal level. If the oxygen sup ply to a

Schematic representation of the domain organisation of VEGF growth factors using VEGF-C/D and VEGF-A as examples.
The VEGF growth factors consist of the central VEGF homology domain (in grey) and optional accessory domains (in blue and magenta). The proteolytic cuts usually take place between the domains (in red). The characteristic cysteine patterns of the VEGF family and the Cterminal propeptide are represented by yellow and white lines, respectively.

Figure 3
Schematic representation of known control loops in angiogenesis and lymphangiogenesis. The production of VEGFA and VEGFC is usually selflimiting due to negative feedback. As soon as a sufficient oxygen supply has been established or the tissue pressure normalized, the signaling for blood vessel or lymphatic growth is reduced. tissue is insufficient (hy po xia), the production of VEGF-A is switched on, which in turn leads to blood vessel growth and normalisation of the oxygen pressure [12]. In contrast, VEGF-C pro duction hardly im proves tissue oxy genation, but it does im prove tissue drai nage and immune cell trafficking (see Figure 4). Presumably for this reason and in contrast to VEGF-A, the pro duction of VEGF-C is controlled by pro inflammatory signals and not or only insignificantly by hypoxia [13][14][15]. In addition, VEGF-C can contribute to the limitation of inflammatory reactions by in creased drainage [16,17] and im mu no modu lation [18]. VEGF-C also plays a vital role for the lymph ves sels of the small in tes tinal villi (lacteals). ese in fat absorption specia lised vessels require for their mainta nance the permanent stimulation by VEGF-C [19], which is produced by ma crophages in response to the mi crobial intestinal flora [20].
Increased interstitial tissue pressure am pli fies the growth of lymph vessels via the pressure-dependent signal trans duc tion of VEGF receptor-3 (mechano trans duction) mediated by β1 integrin and integrin-linked kinase (ILK), there by normalising tissue pressure [21,22]. Whether tissue pres sure also has an in fluence on VEGF-C produc tion or activa tion is not known yet.

During embryonic development VEGF-C is activated by ADAMTS3
Mutations in the Collagen-and Calci um-Binding EGF domain-containg pro tein 1 (CCBE1) gene are respon sible for the systemic lymphatic dysplasia in Hennekam Syndrome Type I [25]. CCBE1 regulates the proteinase ADAMTS3, which is the primary protein ase that activates VEGF-C du ring em bryonic growth [26,27]. e ADAMTS proteinases are cell surface or ECM-localized multidomain enzymes close ly related to the ADAM pro teinases. In contrast to the membrane-bound ADAM proteinases, the ADAMTS proteinases are secreted and con tain one or more repeats of the throm bospondin type 1 motif. Some fun ctions of this protein family, such as pro collagen processing or pro teo glycan clea vage, have been linked to the regula tion of angio gene sis [28]. Due to its struc ture, ADAMTS3 belongs to gether with ADAMTS2 and ADAMTS14 to the procollagenase group [29] and also cleaves, at least in vi tro, procollagen Npro peptide [30].
If the function of ADAMTS2 is muta tionally im paired, proteolytic collagen maturation is disturbed and a con nective tissue defect is the con sequence (Ehlers-Danlos syndrome, derma to sparaxis type) [31]. In contrast, pa tients without or with compromised ADAMTS3 genes show no deficits in col lagen synthesis, but distinct defects in the development of the lymphatic sys tem [32,33].
Although the biosynthesis of VEGF-D is very similar to that of VEGF-C (shown sche matically in Figure 5)

Activation of VEGF-C in wound healing by plasmin and cathepsin D
e restoration of oxygen supply and im mune function through blood and lymph vessels are paramount for wound healing. An acceleration of wound healing by VEGF-C was first ob served in animal experiments in 2004 [36,37]. When platelets are activa ted, VEGF-C is released from the α-Schematic representation of the proteolytic activation of VEGF-C. VEGFC is synthesized as a precursor with a size of 58 kDa. This unprocessed form (also called "preproVEGFC") is more than twice as large as the mature VEGFC and, after the signal peptide has been cleaved off during transport into the endoplasmic reticulum, is converted into proVEGFC in the transGolgi network by the proprotein convertases PC5, PC7 and especially furin. This occurs by cleaving the polypeptide chain Cterminally to the VHD (marked by a yellow triangle). If furin is blocked, unprocessed VEGFC is not converted into proVEGFC [23]. ProVEGFC can bind but not activate VEGFR3 and therefore acts as a competitive inhibitor of active VEGFC, which has been shown both in vitro and in vivo [24]. Only a further proteolytic cut Nterminally of the VHD (marked by red triangles) converts proVEGFC into the biologically active form. Mature VEGFC has by far the highest affinity for the binding and activation of VEGFR 2 and 3 [10]. Unprocessed VEGFC is hardly detectable in the cell culture supernatant and probably occurs physiologically only inside the cell [10].

Figure 5
Schematic representation of the proteolytic processing sites in the amino acid sequences of VEGF-C and VEGF-D. The activation of VEGFC and VEGFD is achieved by proteolytic cleavage of the protein between the Nterminal domain and the VEGF homology domain. The further C terminally (in the figure to the right) the cleavage of VEGFC occurs, the lower the biological activity of the resulting VEGFC form [42]. The cleavage of VEGFC by plasmin within the VEGF homology domain leads to a complete inactivation of VEGFC [26]. Alternatively, VEGFC can also be inactivated through sequesteration by soluble isoforms of its receptors [44].  gra nules [38]. Plas min, which is later in volved in the dis solution of the tempo rary fibrin ma trix [39], probably acti vates both plate let-derived VEGF-C and latent VEGF-C (pro-VEGF-C which was em bed ded in the extra cel lular matrix [40,41]).
Cathepsin D is another activator of VEGF-C [42]. Ac celerated wound healing by saliva [43] can per haps be partly attri bu ted to the acti va tion of VEGF-C by Cathepsin D, which is found in saliva. However, many other en zymes are re leased during wound heal ing which have a fairly broad substrate spe ci fi ci ty, e.g. MMP-3, and which may contribute to the acti va tion of VEGF-C as well as to the re lease/acti vation of ECM-sequestered (in active) VEGF-A.

Activation of VEGF-C by prostate specific antigen (PSA, KLK3)
Somewhat surprisingly, but not en tire ly unanticipated [45], it turned out that PSA (pros tate specific anti gen), which is con troversially used in prostate cancer screening, can acti vate VEGF-C [42]. Despite being the most fre quently used blood test for early can cer detection, it is less well known that PSA is a proteinase whose main bio logical task is to liquefy the gel-like con sistency of the male ejaculate, which allows the sperm cells to swim [46]. VEGF-A had been detected in seminal fluid more than 20 years ago [47,48], and was later shown to have a po si tive effect on the mo ti lity of sperma tozoa [49]. How ever, only recently it was recognized that also VEGF-C is pre sent in the male ejaculate and that the ac ti vation of this seminal VEGF-C oc curs concurrently with the li que faction of the ejaculate by PSA [42]. Whether seminal VEGF-C is an epiphe nomenon or has any function for re pro duction has not yet been clarified. VEGF-C is certainly required for the im plan tation of the embryo into the endo me trium, where it acts on the blood vessels [33]. However, VEGF-C could also play a role in the implan tation-associated immune modu la tion [50] or it might -as already described for VEGF-A [49] -have a direct chemo tactic or chemokinetic effect on sper matozoa.

The key position of CCBE1 as cofactor of activation
When it is important to react quickly to chang ing demands, regulation at the Schematic representation of the hypothetical mechanism of action of CCBE1. The Ctermi nal domain of proVEGFC (dark blue) blocks the access of enzymes to proteolytically sensitive se quences (shown on the left half of the dimer). CCBE1 causes a conformational change in VEGFC and thus exposes the proteolytic target site (shown on the right half of the dimer). The CCBE1 effect on VEGFC activation has been demonstrated for ADAMTS3 and KLK3/PSA , and it is suspected for Cathepsin D. After the activation of VEGFC, e.g. at the primary interface of plasmin (#1), Cathepsin D can shorten the protein with an additional cut ("secondary activation"). The shorter the Ntermi nal end of active VEGFC, the weaker it binds to and activates its receptors. With a proteolytic cut at the secondary plasmin cleavage site (#2), VEGFC loses all activity towards VEGFR2 and VEGFR3.

Figure 7
The different activation paths of VEGF-C. The proteolytic cleavage of proVEGFC by ADAMTS3 activates and mobilizes VEGFC. The activation of VEGFC can take place in four different settings: VEGFR3bound but inactive VEGFC can start signaling immediately after proteolytic activation (activation mode 1), whereas HSPGbound VEGFC must first dissociate from the HSPG and translocate to VEGFR3 (activation mode 2). The activation of VEGFC can also take place in the soluble phase (activation mode 3). Immunohistochemically, however, the vast majority of pro VEGFC, CCBE1 and ADAMTS3 are found bound to the extracellular matrix (ECM, activation mode 4) or on cell surfaces (activation modes 1 and 2). CCBE1 fulfills two independent functions for VEGFC activation: the Cterminal domain accelerates the proteolytic cleavage, while the N terminal domain recruits proVEGFC to efficiently form the trimeric activation complex.  gene expresion level introduces a delay due to the upstream pro cessess of trans cription and translation. e production and on-demand activation of in active ("latent") VEGF-C bypasses this delay. A si mi lar form of storage and activation is known e.g. from TGF-β [51]. e heparin-binding isoforms of VEGF-A are also reversibly in activated by binding to extracellular pro teins and can be re ac ti vated if required, e.g. by plasmin-mediated proteo lytic cleav age [6]. A summary of all pre viously published VEGF-C acti vating enzymes and the exact positions of the cleavage sites is shown in Figure 6. e CCBE1 protein re gulates the VEGF-C-activating func tion of the ADAMTS3 proteinase. CCBE1 con sists of two domains: the N-terminal domain, which is formed by three EGFlike repreats, and the C-terminal domain. which consists of two collagen mo tifs. Both do mains are able to ac cele rate the acti va tion of VEGF-C by ADAMTS3 independently. e N-termi nal domain of CCBE1 is res pon sible for the colocalization of VEGF-C and ADAMTS3 with CCBE1 to form the ac ti vation complex, and the C-term inal do main accelerates the catalytic clea vage of VEGF-C by ADAMTS3 [40]. Pre su ma bly, CCBE1 removes the mas king of the proteolytic target site of VEGF-C, which is normally blocked by its own C-terminal domain ( Figure  7). e different acti va tion paths of VEGF-C with regard to the localisation of the activation com plex are explained in Figure 8.
Hennekam Syndrome (HS) is a rare con genital disease with a generalized lymph edema as its main feature. At first, mutations in the CCBE1 gene were identified as the cause, but meanwhile, mutations in three different 93 LymphForsch 23 (2) 2019

REVIEW ARTICLES
Four proteinases are known in the literature as VEGF-C activators. Plasmin occupies a special position because it inactivates VEGFC during prolonged exposure by cutting at a secondary site. Gene therapy with AdVEGF-C. An increasingly popular therapy for breast cancerassociated lymphedema is autologous lymph node transplantation [88,89]. In preclinical studies, the treatment success (integration of the transplanted lymph node into the local lymphatic network) could be improved by the simultaneous administration of VEGFC [90]. With this strategy, Lymfactin® has successfully completed Phase I clinical trials and is now in Phase II. As a further development of Lymfactin® , a simultaneous administration of VEGFC with the VEGFCactivating ADAMTS3 and/or CCBE1 is being discussed.

Figure 9
REVIEW ARTICLES genes are known to trigger HS. e func tion of two of these genes (CCBE1 and ADAMTS3) within the VEGF-C sig nal transduction pathway is known. It is assumed that the third gene (FAT4) also has an important function with in the VEGF-C signal transduction path way.

Activation of VEGF-C in tumours
VEGF-C and its activation are in dispen sable for the development of the lym phatic system [33,52], and in the adult organism, at least some lymphatic networks need a constant supply of VEGF-C for their maintenance [19]. To prevent lymphatic dysfunction, the amount of active VEGF-C must be pre cise ly regulated. A degregulation with severe consequences can e.g. be trig gered by tu mors.
e relationship between VEGF-Ame di ated blood vessel formation and tu mour growth has been well studied and is also spe ci fically blocked in antian gio genic tu mour therapy, e.g. by the anti body drug be va cizumab (Avastin) [53]. It has always been assumed that the majority of tumours never be come cli ni cally relevant because they do not ac quire the ability to stimulate blood ves sel growth [54]. Without switching on VEGF-A production and without the re sul ting vascularisation ("angioge nic switch"), these tumours can never grow lar ger than a few mil li metres because they lack sufficient oxy gen and nutrients [55,56].
How ever, tumours can produce not only VEGF-A but also VEGF-C. e ef fects of VEGF-C on tumour growth oc cur at several levels: 1. VEGF-C can activate VEGFR-2 and thus replace VEGF-A as an angioge nic factor [57].

Tumor cells themselves can ex -
press VEGF re cep tors and be stimu lated in an autocrine or paracrine fashion by VEGF-C [59]. 4. VEGF-C can stimulate lymph vessel growth and thus promote metastasis [60][61][62].
Unlike for the blockade of VEGF-A, there is no approved drug therapy for the blo ck ade of VEGF-C. is lack might result from the fact that pro te oly tic activation produces many dif ferent forms of VEGF-C. Effective blocking would likely need to block all forms of VEGF-C in addition to all forms of VEGF-D, as VEGF-D can pro vide si mi lar signals for tumour growth as VEGF-C [63].
Which proteinases activate VEGF-C in tu mour diseases has not yet been ex pe ri mentally investigated, but cathep sin D and PSA are likely to play a role for at least certain tumour types. e expression of cathepsin D has long been correlated with tumour me ta stasis [64]. Although, in contrast to ca -thep sin D, the corre la tion between PSA and tumour develop ment has been studied much more in ten sively, va rious studies have come to dif ferent con clusions regarding a tu mour-promo ting function of PSA [65][66][67][68][69]. Some au thors postulate that PSA pro motes early tumour growth but in hi bits its de v elop ment in later stages [70]. In any case, with the activation of VEGF-C by ca thep sin D and PSA, pos si ble mechanis tic links have been iden ti fied, which al lows to experimentally address and ans wer these and similar questions.

Pro-VEGF-C or active VEGF-C?
e vast majority of studies on the role of VEGF-C in tumor growth describe the correlation of VEGF-C levels with di sease progression. However, none of these studies distinguishes between ac tive, mature VEGF-C and inac tive pro-VEGF-C. is can be attribu ted to the fact that pro-VEGF-C has

REVIEW ARTICLES
only been known to be inactive since 2014 and that no commercially avai lable test does distinguish between the two forms. For RNA-based expression analyses (e.g. Gene-Chip®, RNA-Seq) such differentiation is essentially impos sible, since all VEGF-C forms are trans lated from the same mRNA transcript of the VEGFC gene. A dif fe ren ti ation of the different VEGF-C forms could be achieved with an antibodybased test (ELISA, Western blot), but such a test has not been developed yet. More over, the majority of com mer cially avai lable antibodies against VEGF-C are not even capable of detecting VEGF-C with the necessary sensitivity [42]. It is there fore not surprising that the research data are confusing.
Meanwhile, the number of clinical stu dies that correlate VEGF-C ex pression of tumours with the course of the di sease has exceeded three hundred (Pub med query: https://mjlab.fi/ pubmed1). Some studies have found a link between VEGF-C levels and disease progression [71,72], while others could not demonstrate such a link [73]. In any case, con trol led animal experiments mostly con firm the instru mental role of VEGF-C for tu mour metasta sis [60,74,75], and mo le cular bio lo gical mechanisms have also been iden tified for the relationship [76].

Activation of VEGF-C for prolymphangiogenic therapies
Although lymphedema can be treated, the aim of research remains a causal the ra py, because lymph drainage and ban daging only help to control the symp toms of the underlying lymphatic insufficiency. With Bestatin (Ubenimex) and Lym factin®, the first trials for drug-based lymphedema therapies have been star ted in the recent years. How ever, the Be sta tin studies of the US com pany Eiger BioPharmaceuticals have been dis con tinued in autumn 2018 aer the second phase since neither primary nor se con dary objec -tives had been achieved [77]. In contrast, the phase 2 studies of Lym fa ctin®, sponsored by the Finnish phar ma ceuti cal start-up Herantis, have just been ex pand ed [78]. e two drugs are based on dif ferent mechanisms of action. Following the observation that the anal gesic keto profen relieves lymph edema symp toms in a mouse model [79], the keto profen-like but more spe cific Besta tin was selected for clinical trials [80]. Ketoprofen and Bestatin are non-steroidal anti-inflamma tory drugs, and not much detail is known about their influence on the lymphatic sys tem. In contrast, Lym fac -tin® is a ge ne tically engineered biophar ma ceu ti cal that is based on the body's own VEGF-C production aer ad mi ni strat ion of a re combinant adeno viral vector (AdVEGF-C, see Figure  9), whose me cha nism of action is well re searched [81][82][83]. Depending on the area of ap pli ca tion, the availability of en do ge nous pro te inases and CCBE1 for the ac ti vation of VEGF-C for Lymfactin®/AdVEGF-C could, however, be a li mi t ing factor. In animal expe riments, e.g. mu scle tissue reacted to VEGF-C-pro vid ing gene therapy only with mo de rate lymph angio genesis. Only when VEGF-C was co-ad min istered with CCBE1 the lymph angio genic re sponse be came strong [26]. Because the lymphatic system is impor t ant not only for drainage but also for im munity, it is not surprising that VEGF-C has been identified as a pharma cological target for several diseases af fecting the im mune system. ese include chro nic inflammatory bowel disease [84], psoriasis [85] and rheu mato id ar thri tis [17], but also neuro de gene rative di seases such as multiple sclero sis and Alz heimer's disease [86]. Also intruiging, albeit controversial, is a report about the successful therapy of myo cardial infarction in an animal mo del with a single dose of VEGF-C [87].
It should be noted that these prolymph angiogenic applications pursue an ob jec tive that is contrary to that of tu mour therapy. In lymphedema and im mune diseases, the typical goal is to increase expression and activation of VEGF-C, whereas in tu mour therapy, the goal is to block VEGF-C expression or activation. Balancing these op posing goals might prove a complex task, par ti cularly in the case of edema which oc curs as a result of surgical can cer treat ment.   (4) (2)