Digging deeper: structural background of PEGylated fibrin gels in cell migration and lumenogenesis

Fibrin is a well-known tool in tissue engineering, but the structure of its modifications created to improve its properties remains undiscussed despite its importance, e.g. in designing biomaterials that ensure cell migration and lumenogenesis. We sought to uncover the structural aspects of PEGylated fibrin hydrogels shown to contribute to angiogenesis. The analysis of the small-angle X-ray scattering (SAXS) data and ab initio modeling revealed that the PEGylation of fibrinogen led to the formation of oligomeric species, which are larger at a higher PEG : fibrinogen molar ratio. The improvement of optical properties was provided by the decrease in aggregates' sizes and also by retaining the bound water. Compared to the native fibrin, the structure of the 5 : 1 PEGylated fibrin gel consisted of homogenously distributed flexible fibrils with a smaller space between them. Moreover, as arginylglycylaspartic acid (RGD) sites may be partly bound to PEG-NHS or masked because of the oligomerization, the number of adhesion sites may be slightly reduced that may provide the better cell migration and formation of continuous capillary-like structures.


Introduction
One of the critical issues in new organ and tissue fabrication is vascularization that is usually caused by cell migration and lumenogenesis and ensures the nutrient and oxygen supply and metabolite removal. Despite the progress achieved, most studies are based on spontaneous vessel formation within tissue-engineered constructs. [1][2][3] In such cases, the formed microvasculature is highly random and cannot provide a sufficient supply of nutrients and oxygen because of the lack of anastomoses and large spacing between vessels. Therefore, one should guide the angiogenesis within newly fabricated tissues that can be achieved via biochemical and mechanical cues provided by the microenvironment. 4 In tissue engineering, the microenvironment is mostly formed by biomaterials (scaffolds, hydrogels, etc.). Among them, brin has been shown to be an effective tool to produce capillary-like networks and can be used as a biomaterial design platform to fabricate pre-vascularized tissues. [5][6][7][8][9] Fibrin is forme via the thrombin-associated cleavage of brinogen, a blood plasma protein, followed by its polymerization. 10 Natural brin rapidly degrades and is not transparent, so several modications have been suggested to overcome these limitations and, moreover, those have been shown to increase its angiogenic potential. [11][12][13] Among them, PEGylation (modication with functionalized polyethylene glycol (PEG)) is of particular interest.
Previously, it was shown that PEGylation allowed not only achieving the gel transparency and better stability, but also improving its angiogenic properties (e.g. they supported the formation of lumens). 11,14,15 However, the structural aspects remain undiscussed despite their evident importance in designing angiogenesis-guiding biomaterials. Earlier, we determined the shape of brinogen in solution using small angle X-ray scattering (SAXS). 16 Frisman et al. 17 investigated the structural properties of PEG-brinogen conjugates formed via a Michael-type addition of thiols to acrylate-functionalized PEG. However, the latter procedure is time-and labor-consuming and requires protein refolding that is hard to be controlled and causes signicant changes in the protein native structure. To avoid the aforementioned problems, another common type of brin PEGylation using NHS-functionalized PEG was studied by several groups. 12,15,[18][19][20][21] This type of modication proceeding through the covalent binding of amino groups does not require protein refolding and is easy to carry out. However, despite the good results in biological experiments, no study was dedicated to the structural properties of this modication, though this knowledge could help in understanding cell-matrix interactions and possible ways to tune them. In this study, we therefore set the goal to reveal the structural aspects of the PEGylated brin hydrogel that provide a favorable environment to cells and may facilitate the angiogenesis.

FT-IR spectral analysis
PEG binding to the brinogen backbone was proven by FT-IR spectra. Fig. 1A shows the corresponding spectra of NHSfunctionalized PEG and native and PEGylated (5 : 1, 10 : 1) brinogen. Succinimidyl ester triplet band (1739, 1782, and 1813 cm À1 ) is specic for PEG-NHS; it was not observed in the modied brinogen spectra because of the reaction with primary amino groups followed by imide ring opening. Samples of the brinogen modications had an increased intensity band at 1100 cm À1 that proves the PEG-derived (C-O) units insertion. This band varies depending on the molar ratio. The same peaks as those in the PEG-NHS spectrum for amide I and amide II bands (1650 and 1530 cm À1 ) were noticed in the spectra of the PEGylated brinogen samples. The increased C-O units and amide peaks proved the successful PEG insertion into the brinogen backbone.
Thermal gravimetry analysis Fig. 1B and C show the temperature dependences of the mass loss (thermal gravimetry, TG) acquired during the thermal destruction (TD) of native and PEGylated brinogen (B) and native and PEGylated brin (C). Fibrin samples were more thermally stable than brinogen: there were small shis (app. 20 C) to the high-temperature region of the initial mass loss temperature on the TG curves. This does not appear surprising because brin is a polymer representing monomer chains within brinogen which crosslink aer thrombin-associated cleavage. On the TG curves, there is a region with a small mass loss that may be associated with the carbon residue formation. The carbon residue content was different: for the brin samples, it was app. 10% higher than that for the brinogen samples that was the evidence for the successful brin cross-linking. The PEGylated brin samples had a slightly higher carbon residue content than native brin samples had that presumably conrmed a higher cross-linking degree of the PEGylated brin. However the small amount of PEG in those samples might have resulted in such difference of the TG curves, as well.

Turbidity analysis
The samples prepared from 10 : 1 and 5 : 1 PEGylated brinogen were transparent whereas the gels from native brinogen became turbid upon thrombin crosslinking ( Fig. 2A). The visible light transmission through the modied brin gels was signicantly increased compared to that for the native gel: 66% at a wavelength of 380 nm and 98%at 780 nm (vs. 1% and 63%, respectively, for the native brin gel) ( Fig. 2A). Both modications had almost the same absorbance spectra ( Fig. 2A). The observed peak in a range of 900-1000 nm ( Fig. 2A) corresponded to the well-known absorbance peak of water.

Confocal laser scanning microscopy
The microscale structure and porosity of samples were assessed using confocal laser scanning microscopy ( Fig. 2B and 3B). The morphology of native brin (Fig. 3B) was presented by bundles consisting of numerous thin bers. The structure of the PEGylated brin samples ( Fig. 3B) was occulent with hardly distinguishable short bers. The samples' porosity varied (Fig. 2B): the lowest porosity was revealed for gels prepared from 5 : 1 PEGylated brinogen (23.8% AE 7.8); native and 10 : 1 PEGylated brin gels had a similar porosity level (47.9% AE 4.4 and 44.2% AE 6.9, respectively) that was signicantly higher than that of 5 : 1 PEGylated brin gel.

Atomic force microscopy
The AFM measurements of Young's moduli of the prepared gels ( Fig. 2C-E) showed that, in general, their values were slightly inhomogeneous for the PEGylated brin gels and more heterogeneous for the native brin gel, in which we observed localized regions with signicantly higher Young's moduli (Fig. 2E). The average Young's modulus values are presented in Fig. 2C. All the gels were relatively so (<1 kPa); PEGylation led to a decrease in Young's modulus by 18% for the 5 : 1 PEGylated brin gel and by 50% for the 10 : 1 PEGylated brin gel. However, the apparent viscosity (Fig. 2D) of the 5 : 1 PEGylated gel was 77% higher, while viscosity of the 10 : 1 PEGylated gel did not change signicantly relative to the native gel. Fig. 3C demonstrates topography maps showing different scales of inhomogeneity in the gels.

Scanning electron microscopy
PEGylation resulted in alteration of the bers' morphology and their packing within a gel (Fig. 3A). Compared to the gel from native brinogen, the samples from both protein modications showed ber thickening and a densely packed ber architecture. The gel from 5 : 1 PEGylated brinogen had a occulent structure with a non-uniform pore distribution. The gel formed from 10 : 1 PEGylated brinogen had an almost poreless dense ber network.

SAXS analysis
To estimate the sizes and shapes of native type brinogen and PEGylated brinogen constructs in a solution and in a hydrogel, we performed SAXS measurements at different protein concentrations and at the two molar ratios of brinogen to PEG (1 : 5 and 1 : 10, respectively). The processed SAXS data and the computed distance distribution functions are summarized in Table S1 † and Fig. 4 and 5. The experimental radius of gyration (R g ) and the maximal distance (D max ) of native brinogen in the PBS buffer (14.1 AE 0.1 nm and 50.0 AE 1.0 nm, respectively) suggest a rather elongated structure. The p(r) function displayed an asymmetric tail (Fig. 5) typical for elongated particles. The PEG addition led to signicant increases in R g and D max up to 19.5 AE 0.2 nm and 75.0 AE 1.5 nm (in the case of 1 : 5 brinogen-PEG molar ratio) and to 23.9 AE 0.3 nm and 90.0 AE 2.0 nm (in the case of 1 : 10 brinogen-PEG molar ratio), respectively. The experimental molecular mass (MM) of brinogen in the PBS buffer measured as 415 AE 50 kDa suggested that the protein formed associates consisting of two dimeric species in solution (with the theoretical MM of a monomer being 106 kDa). This was corroborated by the excluded volume (V p ) of the particle, 670 AE 50 nm 3 , in agreement with the empirical ndings that the hydrated volume of a compact protein in nm 3 is generally larger than the MM in kDa approximately by a factor of 1.7. In the applied concentration range, the predominant brinogen species were associates consisting of two dimeric brinogen molecules. The addition of PEG promoted further oligomerization of PEG-conjugated brinogen molecules shiing the average MM to 620 AE 60 kDa for 5 : 1 PEG-brinogen constructs and to 920 AE 150 kDa for 10 : 1 PEG-brinogen constructs. Such structural alterations of brinogen can be explained by the covalent PEG attachment to the brinogen polypeptide chain. Thus, the PEG modication of brinogen initiated the appearance of species with a higher MM (presumably associates of three and four dimeric molecules) in corroboration with the PEG-dependent oligomerization (Fig. 4).
The macromolecular shapes of individual molecules were reconstructed by ab initio modeling using the experimental Xray scattering data. Low resolution models of the native and PEGylated brinogen in a solution and in a hydrogel (Fig. 6) were built using ab initio shape determination programs as , and its distribution maps (E); apparent viscosity (Pa s), mean AE SD (D). Young's moduli were slightly inhomogeneous for the PEGylated fibrin gels and more heterogeneous for the non-modified gel, in which regions with significantly higher Young's moduli were observed. PEGylation led to a decrease of the Young's modulus. * Fibrin gel vs. PEGylated 5 : 1 fibrin gel, p < 0.05; ** PEGylated 5 : 1 fibrin gel vs. PEGylated 10 : 1 fibrin gel, p < 0.05; *** PEGylated 10 : 1 fibrin gel vs. fibrin gel, p < 0.05.
This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 4190-4200 | 4193 described in the Experimental section. They provided good ts to the experimental data with discrepancies of c 2 ¼ 0.90 O 1.26 (Fig. 4, solid lines and Table S1 †). Individual molecules were found to resemble elongated hairpin-like particles. One has to note that SAXS patterns only contain the contribution of the brinogen backbones and not that of the attached PEG because the electron density of PEG is almost the same as that of the solvent. The PEG modication resulted in the expansion of the length and cross-section of the tted model shapes, but the overall shape topology remained the same. The gelation led to brinogen oligomerization in the case of PEGylated constructs and non-specic polymerization in the case of wt brinogen. The X-ray scattering patterns and the restored ab initio models of 10 : 1 PEGylated brinogen and brin were very similar and characteristic of self-assembled elongated objects that points to the stable organization of brinogen oligomeric species in a solution and low inuence of gelation on the local structure.

Discussion
The small-angle scattering data and SEM and confocal laser scanning micrographs suggest that conjugated brinogen self- assembles into elongated objects and their dimensions are dictated by the crosslinking mechanism (Fig. 7). The number of potential PEG-NHS binding sites on brinogen polypeptide chains is quite high (a-chain -87, b-chain -77, g-chain -60). If all PEG-NHS molecules contributed to crosslinking of brinogen molecules, it would result in the formation of large aggregates. However, as follows from the SAXS data, one can observe only a moderate degree of oligomerization (from tetramers for wt brinogen to hexamers/octamers for PEG-brinogen conjugates), in which only a few of PEG molecules contribute to the crosslinking. One of the explanations is that the used PEG-NHS has two active terminal groups and a sufficient spacer length so that it can link to different sites of the same brinogen polypeptide chain and does not promote the oligomerization. At the same time, the larger molar ratio of PEG to brinogen molecules expectedly led to a higher oligomerization degree.
However, compared to the native brin, the PEG-brin conjugates were rather small. This may be caused by PEG coupling with the protein backbone leading to retaining the bound water and hence decreasing the chaotic protein assembly and formation of long bers. These smaller water-shelled aggregates scattered light only insignicantly and ensured the gel transparency in the visible light region (Fig. 2A). The same This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 4190-4200 | 4195 effects were reported for elastin-like polypeptide, 22 soy protein isolate, pea albumin, and casein. 23 The gel transparency can signicantly facilitate studying cell properties in a 3D culture when methods such as immunochemical staining or colorimetric assays (e.g. AlamarBlue) are required. Moreover, inside a transparent gel, the cells' behavior and functioning can be reliably modulated via light irradiation. [24][25][26][27] It is well known that the mechanical properties of the microenvironment can inuence the cells' behavior, [28][29][30] and the matrix stiffness is claimed to be a crucial cue that regulates angiogenesis. 31 The structural difference between native and modied brin ensures changes in their elastic properties. The 5 : 1 and 10 : 1 PEGylated brin gels were soer than the native one; but the 5 : 1 PEGylated gel was also more viscous. The apparent viscosity data could be interpreted in the framework of the poroelasticity theory, which relates viscous relaxation to the diffusion of a solvent through the porous polymer network. 32 Accordingly, the increase in the apparent viscosity of the 5 : 1 PEGylated brin gel can be explained by smaller pore sizes than those in other gels. This correlates with the porosity measurements and SAXS data. Fig. 2B and 3A and B show that the native brin gel was more porous than the 5 : 1 PEGylated brin gel. Despite the poor porosity observed in the SEM images, the 10 : 1 PEGylated brin gel has almost the same porosity as the native one ( Fig. 2B and 3B). This can be explained by the need to dry samples before conducting the SEM study, so that the particles revealed by SAXS (Table S1 † and Fig. 6) may stick to each other and adhere to a surface. Thus, the inner structure of the 5 : 1 PEGylated brin gel appears to be formed by exible macroparticles (brils) with a smaller space between them (Fig. 7).  The cell adhesion to brin is mostly provided by two arginylglycylaspartic acid (RGD) sites located on the a-chain through integrins aVb3, aIIbb3, a5b1, etc. 33 RGD sites containing arginine with a primary amino group may be masked because of the described above oligomerization or coupled with PEG-NHS. Therefore, cells may have fewer adhesion sites in the modied brin than they have in the native one. This is supported by the results described: immunocytochemical staining did not reveal the integrin aVb3 expression by broblasts encapsulated within the PEGylated brin gel; 12 transmission electron microscopy showed almost no junction formed by mesenchymal stromal cells from the umbilical cord with the hydrogel when compared to those within the non-modied gel. 15 As shown by Korff and Augustin,34 despite the support of forming outgrowths in the beginning, the RGD peptides obstructed the sprouting of tubules. Thus, as cells may adhere less to brin bers, they may easier migrate within the modied brin gel and form capillary-like structures without the disruption of their integrity, that is advantageous in comparison with the unmodied gel. A new study to test this assumption is warranted, based on a detailed analysis of cellmatrix interactions and integrins expression.

Preparation of thrombin and brinogen solutions
Lyophilized brinogen and thrombin from bovine plasma (Sigma-Aldrich, Germany) were dissolved in sterile phosphate buffered saline (PBS) to the concentrations of 25 and 50 mg mL À1 and 100 U mL À1 , respectively. Before their use, the protein and enzyme solutions were stored at À20 C.

FTIR-spectroscopy and thermal gravimetric analysis
Lyophilized samples of gels and their components were studied using a Spotlight 400N FT-NIR Imaging System (PerkinElmer, USA). Differential scanning calorimetry (DSC) was carried out using a STA 6000 simultaneous thermal analyzer (PerkinElmer, This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 4190-4200 | 4197 USA). Samples, 10 mg, were destructed in a nitrogen medium at a gas ow rate of 40 mL min À1 and linear heating rate of 20 C min À1 . Mass losses were registered to 3-10 mg; the relative errors of measuring the temperature and thermal effect were AE1.5 C and AE2%, respectively. The destruction process was described as a temperature dependence of the mass loss (thermal gravimetric analysis).

Turbidity assay
To assess the turbidity of the prepared samples, we measured absorbance spectra using a spectrophotometer (Varian, 50 Scan, Cary).

Confocal laser scanning microscopy and porosity measurement
The procedures were performed as described elsewhere. [35][36][37] Briey, before polymerization, brinogen solutions were mixed with brinogen conjugated with AlexaFluor-488 (Invitrogen, USA) at a ratio 50 : 1. Samples were prepared on slides and analyzed using an LSM 880 confocal laser scanning microscope equipped with an AiryScan module and GaAsP detector (Carl Zeiss, Germany; 40Â water immersion objective). Porosity was measured in ten images from three samples using the ImageJ soware (NIH, USA).

Atomic force microscopy
The measurements were performed using a Bioscope Resolve atomic force microscope (Bruker, USA). The force-distance curves were acquired in the force volume mode using CP-PNP-BSG colloidal probes (NanoandMore GmbH, Germany) with a 5 mm borosilicate glass microsphere attached to the 200 mmlong cantilever. The spring constant of the cantilever measured by the thermal tune method was 0.056 N m À1 . The deection sensitivities of cantilevers were measured in respect to the fused silica standard (Bruker). All the measurements were conducted at the temperature of 25 C, in the PBS medium. The force-distance curves were processed using the MATLAB soware (MathWorks, USA). The elastic modulus E (Pa) was extracted from force-distance curves by tting according to the Hertzian model of contact mechanics using the extend curves. The apparent viscosity was extracted from the hold region between the extend and retract phases (stress-relaxation experiments) using the standard linear solid model and numerical algorithm proposed in ref. 38

Scanning electron microscopy
The gel structure was visualized by scanning electron microscopy using a FEI Scios microscope at 2 kV in the secondary electron mode using an Everhart-Thornley detector (ETD) and a FEI Quanta 200 3D microscope at 20 kV in the environmental mode at a pressure of 50 Pa using a large eld detector (LFD).
Synchrotron radiation X-ray scattering data were collected on the EMBL P12 beamline on the storage ring PETRA III (DESY, Hamburg, Germany) using an automated sample changer and a vacuum setup with a 1.5 mm capillary at 20 C. 39 The data were recorded using a 2M PILATUS detector (DECTRIS, Switzerland) at a sample-detector distance of 3.0 m and a wavelength of l ¼ 0.124 nm, covering the range of momentum transfer 0.02 < s < 5.0 nm À1 (s ¼ 4p sin q/l, where 2q is the scattering angle). No measurable radiation damage was detected by comparison of twenty successive time frames with 50 millisecond exposures. The data were averaged aer normalization to the intensity of the transmitted beam using a Becquerel pipeline, 40 the scattering of the buffer was subtracted and the difference data were extrapolated to the zero solute concentration using PRIMUS. 41 Independently, SAXS experiments were performed for hydrogel samples obtained by mixing PEG-brinogen solutions at 25 mg mL À1 with the thrombin solution (at a concentration of 0.2 U per 1 mg of brinogen). The hydrogel samples were put into the cells composed of two carton windows (cell thickness 1 mm) centered on the beam path and measured at room temperature 20 C.
The radius of gyration R g of solute brinogen molecule and the forward scattering I(0) were evaluated using the Guinier approximation at small angles (s < 1.3/R g ) assuming the intensity was represented as I(s) ¼ I(0)exp(À(sR g ) 2 /3) and from the entire scattering pattern by the program GNOM. 42 In the latter case, the distance distribution function p(r) and the maximum particle dimension D max were also computed. The molecular masses (MM) of the molecules were evaluated by a calibration against a reference solution of bovine serum albumin. The excluded volume of the hydrated molecule (V p ) was calculated using the Porod approximation: in which the intensity I(s) was modied by subtraction of an appropriate constant from each data point, forcing the s À4 decay of the intensity at higher angles for homogeneous particles demanded by the Porod's law. The programs DAMMIN 43 and its fast version DAMMIF 44 were employed to construct low resolution ab initio bead models of wildtype (wt) and PEGylated brinogen that best tted the scattering data. DAMMIN employs a simulated annealing (SA) procedure to build a compact bead conguration inside a sphere with the diameter D max that ts the experimental data I exp (s) to minimize the discrepancy: where N is the number of experimental points, c is a scaling factor and I calc (s j ) and s(s j ) are the calculated intensity and the experimental error at the momentum transfer, respectively. Fieen independent DAMMIN runs were performed for each scattering prole in the "slow" mode, using default parameters and no symmetry assumptions (P1 symmetry). The models resulting from independent runs were superimposed using the program SUPCOMB 45 and aligned models were averaged using DAMAVER 46 to generate a consensus three-dimensional shape.

Statistical analysis
Experiments were conducted at least thrice to ensure validity of the results, and the data shown are from single experiments yielding similar results to the triplicate experiments. For any given experiment, each data point represents the mean AE standard deviation (SD). The analysis was performed using the one-way analysis of variance (ANOVA). Differences were assumed to be statistically signicant if the probability of chance occurrence (P value) was less than 0.05.

Conclusion
The architecture of vessels within tissue-engineered constructs is crucial for their successful engrament. Therefore, to promote angiogenesis and hence fabricate viable tissues, the deep understanding of many factors involved is required. Previously, we have shown that the 5 : 1 PEGylated brin gel possesses a high angiogenic potential compared to the native brin gel. 11,14,15 However, the structural rearrangement caused by brin modication remained unclear. The results presented in this study allowed us to suggest that the inner structure of the PEGylated 5 : 1 brin gel can be presented by exible macroparticles with a smaller space between them. Since PEG-NHS may partly bind to or mask the RGD sites, the number of adhesion sites for encapsulated cells may be slightly reduced that may provide better cell migration and formation of nondisrupted capillary-like structures. However, this latter suggestion requires the detailed analysis of cell-matrix interactions and integrins expression that warrants the future studies of PEG-modied brin gels.