Enhanced Mitogenic Activity of Recombinant Human Vascular Endothelial Growth Factor VEGF121 Expressed in E. coli Origami B (DE3) with Molecular Chaperones

We describe the production of a highly-active mutant VEGF variant, α2-PI1-8-VEGF121, which contains a substrate sequence for factor XIIIa at the aminoterminus designed for incorporation into a fibrin gel. The α2-PI1-8-VEGF121 gene was synthesized, cloned into a pET-32a(+) vector and expressed in Escherichia coli Origami B (DE3) host cells. To increase the protein folding and the solubility, the resulting thioredoxin-α2-PI1-8-VEGF121 fusion protein was co-expressed with recombinant molecular chaperones GroES/EL encoded by independent plasmid pGro7. The fusion protein was purified from the soluble fraction of cytoplasmic proteins using affinity chromatography. After cleavage of the thioredoxin fusion part with thrombin, the target protein was purified by a second round of affinity chromatography. The yield of purified α2-PI1-8-VEGF121 was 1.4 mg per liter of the cell culture. The α2-PI1-8-VEGF121 expressed in this work increased the proliferation of endothelial cells 3.9–8.7 times in comparison with commercially-available recombinant VEGF121. This very high mitogenic activity may be caused by co-expression of the growth factor with molecular chaperones not previously used in VEGF production. At the same time, α2-PI1-8-VEGF121 did not elicit considerable inflammatory activation of human endothelial HUVEC cells and human monocyte-like THP-1 cells.


Introduction
Therapeutic angiogenesis is a promising approach for treating patients with cardiovascular diseases, and is also critical in engineering vascularized tissue replacements. Vascular endothelial by immobilizing the growth factor into a polymer matrix, where the release can be controlled by the degradation rate of the polymer.
In this study, we prepared a mutant variant α 2 -PI 1-8 -VEGF 121 , first designed by Zisch et al. [30], containing a substrate sequence (NQEQVSPL) for factor XIIIa that would enable covalent incorporation of VEGF 121 into the fibrin network. VEGF-loaded fibrin matrices have been shown to increase the growth activity of vascular endothelial cells [30]. They can therefore be used for coating vascular prostheses or other cardiovascular implants (stents, heart valve replacements) in order to accelerate their endothelialization. To improve its folding and solubility, the protein was expressed in fusion with thioredoxin. Linkage of the gene of interest to a second 'carrier' or 'partner' gene avoids the problems associated with heterologous protein expression in E. coli [31,32]. Glutathione-S-transferase, maltose-binding protein, thioredoxin and NusA have been successful in producing correctly folded and soluble recombinant proteins in bacteria [32,33].
Thioredoxin (Trx) has been shown to facilitate the soluble expression of a number of mammalian growth factors and cytokines [34]. It is a small, ubiquitous protein that is involved in many physiological functions, and acts both intra-and extracellularly [35]. It works as an important antioxidant that plays a key role in maintaining the reducing environment in the cells. Outside the cell, however, thioredoxin acts as a growth factor or cytokine and stimulates angiogenesis [35][36][37]. In mammalian cells, thioredoxin is encoded by two Trx genes. The Trx1 isoform occurs in the cytosol and nucleus, while the Trx2 isoform is expressed in the mitochondria. Homozygous knock-out of either isoform in mouse was found to be lethal [37]. Thioredoxin of E. coli encoded by the TrxA gene is a single polypeptide chain composed of 109 amino acid residues with a molar weight of 11.7 kDa, and is structurally related to human thioredoxin.
Besides fusion with thioredoxin, a combination of pET-32a(+) vector and E. coli Origami B (DE3) host cells was chosen to increase the soluble protein fraction. This combination was used for several proteins that were difficult to express with a structure related to VEGF [38][39][40].
In the present study, protein folding was encouraged by co-expression with recombinant molecular chaperones GroES/EL encoded by independent plasmid pGro7. The α 2 -PI 1-8 -VEGF 121 prepared by the strategy presented here had a 3.9-8.7 times greater effect on endothelial cell proliferation than commercially-available VEGF 121 . Since the final α 2 -PI 1-8 -VEGF 121 preparation probably contained a small amount of thrombin, which was used for cleavage of the fusion partner and which is able to stimulate the proliferation of endothelial cells [41][42][43], the effect of appropriate concentrations of thrombin on endothelial cell proliferation were also observed. In addition, VEGF can cause immune activation of cells, which can even lead to implant rejection [44,45]. We therefore also tested the potential of VEGF to induce the production of pro-inflammatory cytokines and chemokines in human vascular endothelial cells and human monocyte-like THP cells.

Materials and Methods Materials
Medium and EGM-2 SingleQuots were supplied by Lonza, Czech Republic. Concentration cell Amicon Ultra (cut-off 10 kDa) and membrane filters (pore size 0.22 μm) were from Millipore, USA. Dialysis membrane Spectra/Por 6 (MWCO: 1000) was obtained from Spectrum Labs, USA. Coomassie Brilliant Blue R-250 was purchased from Serva, USA. Bradford reagent was from Bio-Rad, Germany. The SDS-PAGE molecular weight protein marker was supplied by GE Healthcare, UK. L-Arabinose, the Resazurin-based In Vitro Toxicology Assay Kit, human monocyte-like THP-1 cells, RPMI-1640 medium, lipopolysaccharides from E. coli 026:B6 and thrombin from human plasma were purchased from Sigma-Aldrich, Germany.

Construction of α 2 -PI 1-8 -VEGF 121 expression vector
The gene sequence encoding the modified variant of human VEGF 121 (α 2 -PI 1-8 -VEGF 121 ), which contains the additional factor XIIIa substrate sequence NQEQVSPL [30] at the aminoterminus of mature VEGF 121 , was synthesized by Generay Biotech, China, according to the published sequences (GenBank). The codon bias was optimized with respect to the E. coli preferred codon usage. The gene was subcloned between restriction sites Msc I and Xho I of the expression vector pET-32a(+) after the thioredoxin gene. The nucleotide sequence of the final construct (pET32-VEGF 121 ) was confirmed by DNA sequencing.

Expression of Trx-α 2 -PI 1-8 -VEGF 121
The construct was used to transform E. coli Origami B (DE3) competent cells. The strain was co-transformed with the expression vector pGro7 encoding GroEL/ES chaperone genes. E. coli Origami B (DE3) cells were grown in 2YT medium (10 g of yeast extract, 16 g of tryptone, and 5 g of NaCl per liter; 200 mL in 500-mL Erlenmeyer flasks) with ampicillin (100 mg/L) and chloramphenicol (35 mg/L) in an orbital shaker for ca. 3 h at 37°C and at 220 rpm. When the OD at 600 nm reached a value of ca 0.5, isopropyl-β-D-thiogalactopyranoside (IPTG) was added to a final concentration of 0.02 mM, and the temperature was lowered to 25°C. The expression of GroEL/ES was induced by arabinose (1.7 g/L), which was added at the same time point as IPTG. After 20 h of incubation, the cells were washed with the buffer (50 mM Tris, pH 7.4, 100 mM NaCl, 2.5 mM CaCl 2 ), pelleted by centrifugation, and the pellets were stored frozen at -80°C until required.

Recombinant α 2 -PI 1-8 -VEGF 121 purification
The harvested cells were then disrupted by sonication, and the cell debris was removed by centrifugation (13 000 rpm, 25 min, 4°C). The supernatant was loaded onto a Talon Metal Affinity Resin column equilibrated with the same buffer. After washing with washing buffer (50 mM Tris, pH 7.4, 100 mM NaCl, 2.5 mM CaCl 2 , 1 mM imidazole), the recombinant protein was eluted with the same buffer containing 200 mM imidazole. The purified fusion protein Trx-α 2 -PI 1-8 -VEGF 121 was dialyzed against Tris buffer to remove imidazole, and then it was cleaved by human thrombin (9.5 NIH units/ mg of fusion protein, 4 h, RT). The cleaved thioredoxin fusion part was removed by a Talon Metal Affinity Resin column equilibrated with Tris buffer. The purified α 2 -PI 1-8 -VEGF 121 was analyzed by 12% SDS-PAGE followed by Coomassie Brilliant Blue staining.

Endothelial cell proliferation followed by the xCELLigence system
The effect of purified VEGF 121 on proliferation of the endothelial cells was evaluated by the xCELLigence System (Roche Applied Science). This device consists of microtiter plates containing interdigitated gold microelectrodes. This enables label-free, real-time monitoring of cell growth and viability based on electrical impedance measurements.
First, the E-plate background signal corresponding to the culture medium was set up. The E-plate was seeded with HUVEC cells (3 500 per well) in 150 μL EBM-2 basal medium supplemented with ascorbic acid, hydrocortisone, heparin, gentamicin, amphotericin-B and 2% fetal bovine serum from EGM-2 SingleQuots and different concentrations of recombinant α 2 -PI 1-8 -VEGF 121 (20,50, and 100 ng/mL) or commercial VEGF 121 I and VEGF 121 II as standards (20,50, and 100 ng/mL). The hEGF, VEGF, R3-IGF-1 and h-FGF-beta that are also present in the SingleQuots supplement were not added into the EBM-2 basal medium. Cell growth was monitored every 15 minutes for up to 7 days.
To test the effect of thrombin on proliferation of the endothelial cells, the E-plate was seeded with HUVEC cells (3 500 per well) in 150 μL of EBM-2 basal medium with ascorbic acid, hydrocortisone, heparin, gentamicin, amphotericin-B and 2% fetal bovine serum from EGM-2 SingleQuots. Subsequently, the effect of thrombin (0, 0.01, 0.05, 0.1 and 1.0 NIH U/mL) on the proliferation of HUVEC cells in the presence of VEGF 121 standards (0, 20, 50 and 100 ng/mL) was monitored every 15 minutes for up to 7 days.
In vitro toxicology assay kit, resazurin-based A resazurin 4 mM stock solution was filter-sterilized and stored at -20°C. Cell proliferation was observed for up to 7 days. On day 2, 4 and 7, the cells were twice washed with PBS and incubated in 4 μM resazurin diluted in EGM-2 culture medium without added growth factors for 4 hours at 37°C and 5% CO 2 . The relative cell count was quantified by fluorescence measurement (excitation 530 nm, emission 590 nm) on a Synergy HT Multi-Mode Microplate Reader (BioTek).

Protein concentration
The protein concentration was determined by the Bradford method [46] with bovine serum albumin (BSA) as the standard, or by the Human VEGF ELISA Kit (Invitrogen, #KHG0112), in accordance with the manufacturer's instructions.

Potential immune activation of cells by α 2 -PI 1-8 -VEGF 121
The pro-inflammatory potential of α 2 -PI 1-8 -VEGF 121 was estimated by production of selected cytokines and chemokines (Table 1)  Recombinant α 2 -PI 1-8 -VEGF 121 was added into both types of media at a concentration of 0 ng/ mL, 20 ng/mL, and 50 ng/mL. The cells were cultured for 3 and 6 days, and then the media were collected for cytokine and chemokine analysis. The media without α 2 -PI 1-8 -VEGF 121 , taken from THP-1 or HUVEC cells, were used as a negative control.
As a positive control, lipopolysaccharides from Escherichia coli 026:B6 (Sigma, Cat. No. L2654) were added into the cell culture media (i.e. RPMI or EGM-2) at concentrations of 10 ng/mL, 100 ng/mL, and 1000 ng/mL on day 2 and 5. The media were collected 24 hours later for cytokine and chemokine analysis [47]. The positive controls contained no α 2 -PI 1-8 -VEGF 121 .
A Human Luminex Performance Assay Base Kit, Panel A (R&D Systems, Cat. No. LUH000) was used to analyze IL-1α, IL-1β, GM-CSF, TNF-α, MCP-1 and IL-8. The array uses colorcoded microparticles, which are pre-coated with specific antibodies against cytokines or chemokines. The microparticles, incubated with samples, bind the analytes of interest. After washing, a biotinylated antibody cocktail specific to the analytes of interest is added into each well. After washing and removing the unbound antibody, streptavidin-phycoerythrin conjugate is added into each well and binds the biotinylated antibody. Finally, one laser of the Luminex analyzer determines the magnitude of the phycoerythrin signal, and the other laser determines a microparticle-specific signal of the analyte bound.
The array was processed according to the manufacturer's protocol with some modifications (using Uniplate-Microplate Devices, 96-well, U Bottom, Whatman TM, instead of the original plate). Briefly, the samples were centrifuged at 200 g for 7 min. The microplate was pre-wetted with 100 μL of Wash Buffer, centrifuged at 900 g for 10 min. The microparticle concentrate was centrifuged at 1000 g for 30 sec, resuspended in the same solution and diluted according to the manufacturer's protocol. The diluted microparticles (50 μL) were added into each well, followed by 50 μL of standards and media samples. Covered by an aluminium foil, the plate was incubated for 3 hours at room temperature at 500 rpm, subsequently centrifuged at 900 g for 10 min and washed 3 times with 100 μL of Wash Buffer. Then 50 μL of diluted Biotin Antibody Cocktail was added to each well. Covered by an aluminium foil, the plate was incubated for 1 hour at room temperature at 500 rpm and then washed 3 times with 100 μL of Wash Buffer. Then 50 μL of diluted Streptavidin-Phycoerythrin solution was added to each well, incubated for l hour, and subsequently the wells were washed 3 times with 100 μL of Wash Buffer. The microparticles were then resuspended in 100 μL of Wash buffer and the cytokine concentrations were analyzed on Luminex LABScan 3D (Luminex, Netherlands) from 3 parallel samples. The cytokine concentrations were normalized per 100 000 cells, similarly as in our earlier study [47].

Results
Expression and purification of α 2 -PI 1-8 -VEGF 121 The gene sequence encoding the modified variant of human VEGF 121 (α 2 -PI 1-8 -VEGF 121 ) was synthesized and subcloned between restriction sites Msc I and Xho I of the expression vector pET-32a(+) after the thioredoxin (Trx) gene and the hexahistidine tag. The resulting expression vector pET32-VEGF 121 was used to transform the E. coli Origami B (DE3) strain along with plasmid pGro7 encoding bacterial chaperones GroEL/GroES. The recombinant protein Trx-α 2 -PI 1-8 -VEGF 121 was expressed as a fusion protein composed of thioredoxin, histidine tag, thrombin cleavage site, factor XIIIa substrate sequence NQEQVSPL derived from α 2 -plasmin inhibitor and VEGF 121 (Fig 1, S1 Text). A high level of recombinant protein expression was achieved after the induction of recombinant bacteria with 0.02 mM IPTG and subsequent culture growth at 25°C for 20-22 h. The ratio of the soluble protein fraction and the insoluble protein fraction was determined by SDS-PAGE. An analysis with Gel Analyzer software (http://www.gelanalyzer.com) showed that 43% of the Trx-α 2 -PI 1-8 -VEGF 121 was expressed in soluble form (Fig 2, lanes 3 and 4). Trx-α 2 -PI 1-8 -VEGF 121 was purified from the soluble fraction of cytoplasmic proteins using Talon Metal Affinity Resin (Fig 2, lane 6). After purification, the fusion partner was cleaved out with thrombin. The molecular mass of the Trx-α 2 -PI

Mitogenic activity of α 2 -PI 1-8 -VEGF 121
To test the mitogenic activity of α 2 -PI 1-8 -VEGF 121 , the proliferative effect on HUVEC cells was investigated using the xCELLigence system and a resazurin-based assay. The activity of recombinant α 2 -PI 1-8 -VEGF 121 was compared to the activity of VEGF 121 from commercial sources. For this purpose, we used a VEGF 121 variant expressed in E. coli and a variant expressed in mammalian HEK cells. Three concentrations of VEGF 121 (20, 50 and 100 ng/mL) were used for the evaluation. The concentrations of each VEGF 121 variant were determined by ELISA analysis. The EGM-2 cultivation medium without growth factors served as a negative control. The evaluation showed that the recombinant α 2 -PI 1-8 -VEGF 121 manufactured in this work had a greater impact on HUVEC proliferation than the two tested commercially-available VEGF 121 variants. This effect was most apparent at a concentration of 50 ng/mL of VEGF 121 (see Fig 3, S3 Fig), but it was also observed at VEGF 121 concentrations of 20 and 100 ng/mL (see S1 and S7 Figs).
The plateau value of the cell index (evaluated at 165 hours, i.e. on day 7) for α 2 -PI 1-8 -VEGF 121 was 4.30 ± 0.20; for VEGF 121 I it was 1.11 ± 0.11; for VEGF 121 II it was 0.72 ± 0.10, and for the culture medium without VEGF 121 the value was 0.43 ± 0.06 (Fig 3). In other words, PI 1-8 -VEGF 121 expressed and purified under the described conditions showed an approximately 3.9 times greater effect on endothelial growth in comparison with the mitogenic activity of the VEGF 121 I standard, and an approximately 6.0 times greater effect when compared to the mitogenic activity of VEGF 121 II (Fig 3).
The data obtained in the xCELLigence system were further supported by an independent proliferation assay based on the fluorescent indicator resazurin, performed on the 2 nd , 4 th and 7 th day of cultivation. Major differences were observed after 165 hours (day 7) of incubation in a cultivation medium containing 50 ng/mL of VEGF 121 preparations, where the relative cell counts were 11.5 ± 0.6% for VEGF 121 I and 13.3 ± 2.7% for VEGF121 II, in comparison with recombinant α 2 -PI 1-8 -VEGF 121 (100%, Fig 4). This effect was also observed at VEGF 121 concentrations of 20 and 100 ng/mL (see S2 and S4 Figs). Thus, α 2 -PI 1-8 -VEGF 121 increased endothelial proliferation 8.7 times in comparison with VEGF 121 I and 7.5 times in comparison with VEGF 121 II (Fig 4).

Effect of thrombin on HUVEC proliferation
As revealed by the real-time monitoring of cell growth using the xCELLigence system, the addition of thrombin in concentrations of 0.01, 0.05, 0.1 and 1.0 NIH U/mL to the media with    VEGF 121 standards (concentrations of 20, 50 and 100 ng/mL) did not significantly change the proliferation activity of HUVEC cells. The growth dynamics of HUVEC in the presence of the thrombin + VEGF 121 standards were similar as in the presence of the VEGF 121 standards only (Fig 5, S5 Fig).
The mitogenic activity of the preparations (controls) was evaluated using real-time monitoring of HUVEC cell proliferation. The cells were monitored every 15 minutes for 163 hours. The results shown here are mean ± SEM (n = 4). VEGF 121 I expressed in E. coli (50 ng/mL) was used as a standard.

Cell immune activation by VEGF 121
The Luminex Performance Assay revealed that THP-1 monocyte-like cells, cultured with α 2 -PI 1-8 -VEGF 121 (in a concentration of 0, 20 or 50 ng/mL), did not produce Il-1α, Il-1ß and GM-CSF cytokines in a concentration detectable in the cell culture media. However, when the THP-1 cells were stimulated with bacterial lipopolysaccharide (LPS, concentrations from 10 to 1000 ng/mL), i.e. an endotoxin often used as a positive control in studies of cell immune activation, these cells produced Il-1α, Il-1ß, and GM-CSF in high concentrations (Fig 6A-6C, S6  Fig). However, α 2 -PI 1-8 -VEGF 121 stimulated the THP-1 cells to produce TNF-α, MCP-1 and IL-8. The production of these molecules was proportional to the α 2 -PI 1-8 -VEGF 121 concentration, but it was rather transient and was apparent in considerable amounts only for 3 days. After 6 days of culture, the concentration of TNF-α, MCP-1 and IL-8 in the cell culture media dropped to very low values (Fig 6D-6F, S6 Fig). However, LPS massively stimulated the production of TNF-α, MCP-1 and IL-8 by THP-1 cells at both time intervals, so their concentrations in the media exceeded the standard calibration curve of the array, even at the lowest LPS concentration of 10 ng/mL. As for the HUVEC cells, α 2 -PI 1-8 -VEGF 121 stimulated the production of GM-CSF, but this production was markedly lower that the values obtained after stimulation with LPS, even in the lowest concentration (Fig 6G, S6 Fig). Thus, the potential of our recombinant VEGF 121 to induce cell immune activation and an inflammatory reaction can be qualified as relatively low.

Discussion
Most studies on VEGF expression in bacteria describe its purification from insoluble inclusion bodies by the denaturing and refolding method, where the recovery of protein biological activity may be a problem [30,[48][49][50][51][52][53]. However, only a small number of studies have shown the heterologous expression of VEGF in E. coli in the soluble protein fraction [38,54]. Our work is one of the rare cases in which VEGF is expressed in E. coli in a soluble form.
VEGF belongs to the cystine knot superfamily of growth factors. The cystine knot consists of nine cysteine residues that are present within the VEGF 121 structure and play a role in protein dimerization. Due to the absence of a hydrophobic core region, the cystine knot is believed to be the major determinant of protein stability [55]. This complex structure may be responsible for the higher requirements on protein folding quality when expressed heterologously. In accordance with studies that have also dealt with the expression of proteins with a similar structure [38][39][40], pET-32a(+) vector and E. coli Origami B (DE3) host cells were chosen for α 2 -PI 1-8 -VEGF 121 production. Moreover, to ameliorate protein folding, α 2 -PI 1-8 -VEGF 121 was co-expressed with GroEL/ES chaperones, which have been reported to increase the solubility, the yield, and in some cases even the biological activity of several recombinant proteins [56][57][58][59][60][61][62]. To the best of our knowledge, co-expression of VEGF protein with molecular chaperones has not been published before.
About 43% of recombinant thioredoxin-α 2 -PI 1-8 -VEGF 121 fusion protein was expressed in the soluble protein fraction. The corresponding molecular mass was found to be 32.5 kDa (Fig 2), which was in a good agreement with the expected theoretical size of the fusion protein (29.2 kDa). Two other dominant protein bands present in the soluble protein fraction on an SDS-PAGE gel correspond to a molecular mass of ca 58 and 16 kDa (Fig 2, column l and 3). These protein bands were interpreted as molecular chaperones GroEL and GroES. The putative GroEL protein was present in both the soluble protein fraction and the insoluble protein fraction (Fig 2, lanes 3, 4). The apparent presence in a fraction of unbound proteins suggests a weak interaction with Talon Metal Affinity Resin, which was used for affinity chromatography (Fig 2, lane 5).
During dialysis following affinity chromatography, an unspecified amount of the protein precipitated. Thus, although the expression of the soluble protein fraction was high (Fig 2), the overall yield of 1.4 mg per L of culture was lower than expected. The yield of recombinantly prepared VEGF mentioned in related studies varies on a case-by-case basis. Some studies do not present any yield [49,54] or the data are unclear [38]. In other works, the yield of purified active VEGF 121 or VEGF 165 is between 1 mg and 5 mg per L of bacterial culture [50,52,53], which is similar to the yield of mutant VEGF 121 purified in this work. The yield of recombinantly expressed α 2 -PI 1-8 -VEGF 121 by Zisch et al. was 11 mg/L [30].
The SDS-PAGE record of purified α 2 -PI 1-8 -VEGF 121 did not show any contamination, but it is possible that small traces of thrombin used for cleavage of the thioredoxin fusion part remained in the solution. It has been reported that a low concentration of thrombin is able to stimulate the proliferation of endothelial cells. In earlier studies, the addition of thrombin to  HUVEC cells cultivated in serum-free conditions resulted in significant dose-and time-dependent stimulation of endothelial cell proliferation, with a maximal effect at concentrations of thrombin of ca 0.04-0.08 NIH U/mL [41][42][43]. However, in our study, similar or even higher concentrations of thrombin (0.01, 0.05, 0.1 and 1.0 NIH U/mL) had no significant influence on the HUVEC proliferation (Fig 5). A positive effect of thrombin on the very high mitogenic activity of α 2 -PI 1-8 -VEGF 121 can therefore be excluded.
After cleavage of the thioredoxin fusion part by thrombin, the mitogenic activity of α 2 -PI 1-8 -VEGF 121 was evaluated by two independent methods, namely by real-time monitoring of endothelial cell proliferation by the xCELLigence system, and by the fluorescent indicator resazurin. Both assays revealed that α 2 -PI 1-8 -VEGF 121 expressed and purified under the conditions described here showed a significantly greater effect on endothelial cell growth than the mitogenic activity of the VEGF 121 I and II standards. This effect was the most apparent after 7 days, when, at a concentration of 50 ng/mL, α 2 -PI 1-8 -VEGF 121 increased the endothelial cell proliferation approximately 3.9 to 8.7 times more than the VEGF 121 I and II (Figs 3 and 4). The concentrations of all VEGF 121 preparations used in this work were determined by VEGF-ELISA, under the same experimental conditions (see S1 Table). This suggests that the relative VEGF 121 activity was not influenced by incorrect determination of the VEGF 121 concentration. The highly mitogenic α 2 -PI 1-8 -VEGF 121 created in this study (even with a relatively low yield) could be used effectively for improving of endothelialization of various biomaterials, where it could be incorporated in lower concentrations than those usually applied. This option would be helpful for reducing undesirable side-effects of VEGF, such as inflammation, induced by overexpression of VEGF [45]. Recombinant α 2 -PI 1-8 -VEGF 121 contains an additional N-terminal substrate sequence (NQEQVSPL) for factor XIIIa, but there is no evidence that such a short peptide should influence the biological activity of VEGF 121 . Studies comparing the biological activities of various VEGF preparations have not found any significant differences [53,63]. The biological activity was also comparable to that of the commercial standard in the case of VEGF proteins joined to uncleaved fusion partners, such as thioredoxin or GST [49,50].
Work evaluating the biological activity of recombinant VEGF with an uncleaved C-terminal GST fusion partner shows that the presence of GST did not affect the correct assembly of dimers and the display of residues critical for receptor recognition [50]. A similar finding was reported in a study describing the functionality of VEGF 165 N-terminally fused to thioredoxin, where the authors found that N-terminal extension decreased the affinity of VEGF fusion proteins to VEGFR-2, but at saturated concentrations these proteins were as efficient as VEGF 165 of the correct size [49]. Unlike the activity of our α 2 -PI 1-8 -VEGF 121 , which was 8.7 times higher than the activity of the VEGF 121 I standard, the activity of α 2 -PI 1-8 -VEGF 121 prepared without chaperones [30] was only comparable with the activity of unmodified, E. coli-derived VEGF 121 . However, in many papers there is no comparison of biological activity with a commercial VEGF standard [38,48,51,52,54].
It has been shown that administering the thioredoxin gene can stimulate the proliferation of mesenchymal stem cells [64]. When thioredoxin is added into the media in the form of a protein, concentrations as high as 5-70 μg/mL were observed to have an impact on the cell growth [65]. Regarding the level of efficient thioredoxin concentration, the high mitogenic activity of the manufactured VEGF observed in our experiments should not be related to the traces of hypothetical thioredoxin contamination, which were far below 50 ng/mL. The his-tagged E. coli thioredoxin, which was used in this work as a fusion partner, was cleaved out and removed by affinity chromatography. Its elimination was confirmed by SDS-PAGE analysis of the final VEGF 121 preparation. In addition, as mentioned above, the biological activity of recombinant VEGF was not increased even if its fusion partner, thioredoxin, was not cleaved out [49,50].
Thus, a positive role of thrombin, NQEQVSPL or thioredoxin in the increased mitogenic activity of our recombinant α 2 -PI 1-8 -VEGF 121 can be ruled out. However, it should be noted that the commercial VEGF 121 standards that were available to us (i.e., purchased from Prospecbio, Cat. No. CYT-343 and CYT-116) were obtained in the form of a lyophilized powder recommended to be reconstituted in PBS. Proteins in general may lose their biological activity during lyophilization or reconstitution. Presumably, all this could reduce the biological activity of the VEGF 121 standards.
The high mitogenic activity α 2 -PI 1-8 -VEGF 121 created in this study may be caused by coexpression of the growth factor with molecular chaperones that have not been used so far in VEGF production. A similar phenomenon was observed in the case of several recombinant proteins expressed in E. coli. For example, the relative binding activity of anti-B-type natriuretic peptide scFv [57], and the specific enzyme activities of nitrilase [56], and also cold-active lipase Lip-948 [66], were increased by co-expression with molecular chaperones. In the case of VEGF, chaperones could either stabilize an optimal folding structure during the expression, or could selectively solubilize a highly active VEGF fraction. However, this interesting but complicated issue needs further investigation.
Our analysis of cell immune activation suggested that the potential of α 2 -PI 1-8 -VEGF 121 to induce the production of inflammatory cytokines and chemokines in cells and their release into the cell culture media is relatively low. This conclusion is based on our findings that upon stimulation by α 2 -PI 1-8 -VEGF 121 , human monocyte-like THP-1 cells did not release measurable quantities of IL-1α, IL-1β and GM-CSF, and released only small amounts of IL-8, TNF-α and MCP-1 in comparison to cells stimulated by LPS. HUVEC cells were able to release only GM-CSF, but its concentration was again very low compared with the values obtained after LPS stimulation. These results can be considered as favorable for our intended future use of α 2 -PI 1-8 -VEGF 121 for incorporation into fibrin matrices for potential modification of cardiovascular implants and scaffolds for tissue engineering in order to improve their endothelialization or vascularization. However, in other studies, both recombinant and natural VEFG molecules have been reported to act as pro-inflammatory factors, which activated cells of the immune system (leucocytes, lymphocytes, monocytes and macrophages), and also vascular endothelial cells to produce pro-inflammatory cytokines, chemokines, and adhesion molecules of immunoglobulin and selectin families [44,45]. In addition, recombinant VEGF molecules are capable of inducing the production of antibodies when administered into organisms in vivo. This has been used for producing vaccines against tumors [67,68]. A vaccine based on human recombinant VEGF combined with a bacterial adjuvant has even been tested in a phase I clinical trial on patients with advanced solid tumors [69].

Conclusion
In this work we have described a new procedure by which a highly active mutant variant α 2 -PI 1-8 -VEGF 121 can be produced. The mutant protein structure was first designed by Zisch et al. [30], and was used for preparing fibrin gels with incorporated VEGF 121 . The new procedure, based on co-expressing thioredoxin-α 2 -PI 1-8 -VEGF 121 with recombinant molecular chaperones GroES/EL, resulted in mitogenic activity of α 2 -PI 1-8 -VEGF 121 that was 3.9-8.7 times higher than the mitogenic activity of commercial VEGF 121 standards. Very high mitogenic activity and a low effect on inducing the inflammatory activation of human endothelial HUVEC cells and human monocyte-like THP-1 cells make this α 2 -PI 1-8 -VEGF 121 variant attractive for fibrin-based biomaterials releasing VEGF 121 , e.g. for coating cardiovascular implants in order to improve their endothelialization.