**4. Discussion**

VEGF165 was cleaved at position 124/125 in the heparin/neuropilin-binding region that resembles a classic, multi-basic, FSAP cleavage site. Because the disulphide bonds in this region were still intact, there was no difference in mobility of the protein under non-reducing conditions, only under reducing conditions. Only the larger forms of VEGF, 165 but not 121, bound to FSAP, thus conferring specificity to the process. FSAP also bound to neuropilin but not to VEGF-R2, and partially inhibited VEGF165 binding to neuropilin. The binding and cleavage by FSAP was more pronounced in the presence of heparin, which is cofactor for FSAP. As previously reported, plasmin cleaved VEGF after position 110, thereby completely separating the receptor-binding domain from the heparin/neuropilin-binding domain [33,34]. Cleavage in the C-terminal region of VEGF165 occurs naturally in vivo, and such cleaved forms have been observed in tumor ascites [35] as well as wound fluids [34]. Such cleavage is likely to reduce the association of VEGF long forms with the matrix and increase their mobility, thus raising the possibility of activating VEGF receptor as well as modulating their binding to co-receptors [27].

In proliferation, migration, and signal transduction assays on HUVEC, no consequence of the VEGF165 cleavage was observed. This cleavage did not a ffect the induction of proliferation of VEGFR-transfected BAF3 cells either. This can be explained by the fact that the position of the cleavage in VEGF165, in relation to the disulphide bond, was such that the molecule was still held together and preserved some of its characteristics. Other proteases such as urokinase, matrix metalloprotease (MMP), elastase, and tissue kallikrein also cleave the long forms of VEGF in this region and modulate its activity in di fferent ways depending on the site of cleavage.

Matrigel-embedded growth factors provide a model to study neovascularization in vivo [29]. Growth factor-reduced matrigel is the matrix of mouse sarcoma, which contains 1851 unique proteins including classical matrix proteins such as laminin, collagen IV, entactin/nidogen, fibronectin, and heparan sulphate proteoglycans [36]. Heparin increases angiogenesis in this model by multiple mechanisms that probably include modifying growth factor's presentation to the cellular receptors [37]. In this model, FSAP inhibited the activity of VEGF165, bFGF, and a combination of both, and microvascular density was decreased. We used BS1 and vWF as markers for endothelial cells and α-SMC actin as a marker for pericytes/VSMC. Individually, the staining of each marker was reduced by FSAP, indicating that FSAP directly inhibited both cell types. FSAP may also influence paracrine interactions between them, potentially, leading to the same end result. In this model, bFGF and VEGF165 were known to upregulate PDGF-BB, and this in turn may have been responsible for the recruitment of the smooth muscle cells [38]. Smooth muscle cells and pericytes can indirectly contribute to the regulation of neo-vascularization through various paracrine/juxtacrine mechanisms [39], and we have previously shown that FSAP can inhibit PDGF-BB [18]. The inhibitory e ffect of FSAP on bFGF may be related to a complex interaction between FSAP and bFGF, as described before [19]. This result also illustrates the fact that the inhibitory e ffect of FSAP in this model was not limited to VEGF165, and a completely di fferent mechanism of action, independent of the growth factor, cannot be excluded.

In view of the lack of the e ffect of FSAP on VEGF-mediated proliferation and migration of endothelial cells in vitro*,* the strong reduction of angiogenesis in the matrigel model was remarkable. Because FSAP cleavage of VEGF165 led to a VEGF121-like molecule, the di fference in the in vitro and in vivo results must be related to the di fference between the two isoforms. It was possible that in the matrigel model, the VEGF was presented to the cells in a form that was bound to the matrix, and that the haptotaxis-e ffect was altered by FSAP. Evidence for proteolysis-mediated modulation of haptotaxis of matrix-anchored VEGF has been demonstrated before [40]. The cleavage of VEGF by FSAP may also alter the sequestration of the growth factor by the matrix or its spatial complexity, both of which are known to be important for VEGF activity [41]. VEGF can also activate leukocytes and other VEGFR-bearing cells, which may indirectly mediate the inhibition of angiogenesis by FSAP. An e ffect of VEGF in the matrigel model through changes in vascular permeability was also a possibility. The time course of the in vitro experiments, ranging from minutes to 24 h, were di fferent from the in vivo matrigel experiments that lasted a few days, and this could also account for the di fferences in the results.

The induction of ischemia led to angiogenesis in the gastrocnemius muscle in the hind limb ischemia model. Mice without endogenous FSAP showed no changes in their angiogenesis response in this model, indicating that endogenous FSAP was not involved in this process [16]. However, collateral vessel growth was increased in the adductor muscle. Application of exogenous FSAP, directly into the adductor muscle, decreased collateral growth there, but increased angiogenesis in the gastrocnemius muscle. This was most likely a response to the decreased collateral growth in the vessels feeding the gastrocnemius muscle [16]. In the neointimal growth model, exogenous FSAP decreased neointima

formation [15] and a lack of endogenous FSAP increased it [18]. Thus, in two independent models of vascular growth and repair, FSAP seemed to be involved in regulating vascular remodeling but not angiogenesis. This also fits with the lack of difference in neovascularization in *Habp2-*/*-* mice in the matrigel model (see Table 1 for an overview).


**Table 1.** Summary of the effects of FSAP on vascular growth and repair processes.

We chose to study the hind limb ischemia model to investigate processing of long forms of VEGF because we have performed studies using this model in FSAP-deficient mice [16]. However, we did not detect any VEGF165 in this tissue and it was not possible to find any direct evidence for changes in cleavage of this VEGF isoform. However, we did observe an increase in VEGF121 in *Habp2-*/*-* in the absence of any increase in mRNA. Further experiments would be required to study VEGF isoform-specific mRNA and protein to provide compelling evidence for this hypothesis. In this model system there was an upregulation of uPA and MMP-9 at the mRNA and protein level in the gastrocnemius muscle, which could also account for the fact that only the short form of VEGF was detected [16]. Such experiments would be more conclusive in performed in relation to angiogenesis in the eye or the brain because they display the sharpest concentration gradients of VEGF165 [27].

Although endogenous FSAP did not regulate angiogenesis and neovascularization, it did influence remodeling of vessels in general. VEGF165 cleavage, binding properties towards neuropilin, and angiogenesis in matrigel were inhibited by exogenous FSAP. Thus, FSAP has the potential to modulate VEGF165-mediated angiogenesis that may be relevant in some pathophysiological conditions.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/8/11/1396/s1, Figure S1: Cleavage of VEGF165 and VEGF121 by FSAP, Figure S2: Microvascular density in matrigel plugs, Figure S3: *Vegf-A* mRNA in the hind limb ischemia model.

**Author Contributions:** Ö.U. and J.H. performed experiments, analyzed the data, and edited the manuscript. S.M.K. designed the research, obtained the funding, analyzed data, and drafted the manuscript.

**Funding:** Funding was from the Deutscheforschungsgemeinschaft, Germany.

**Acknowledgments:** We would like to thank Lars Muhl, Susanne Tannert-Otto, Thomas Schmidt Wöll, and Baerbel Fuehler for their excellent technical assistance. The gift of VEGFR cells from Steven Stacker and Marc Achen (Ludwig Cancer Research Institute, Melbourne Branch, Australia) and Kari Alitalo (Helsinki Branch, Finland) is greatly appreciated. This study was done as part of the MD/PhD thesis of Özgür Uslu at the Justus Liebig University, Giessen, Germany.

**Conflicts of Interest:** The authors have no financial or other conflicts of interest.
