**1. Introduction**

Currently, aliphatic polyesters are the widely used class of biodegradable polymers for the preparation of various biomedical materials [1–3]. The application of poly(lactic acid) (PLA), poly(glycolic acid-co-lactic acid) (PLGA), poly(ε-caprolactone) (PCL) and polyhydroxybutyrate (PHB) in medicine is due to their biocompatibility with human tissues and their ability to biodegrade to nontoxic metabolites [4]. In addition to their applications as drug delivery systems [3] and surgical suture threads [5], they are widely studied as materials for regenerative medicine including scaffolds for bone tissue engineering [6,7]. However, the hydrophobic properties of aliphatic polyesters impair the adhesion of cells while the lack of a sufficient number of the reactive groups limits the possibility of modifying those polymers with bioactive molecules [8,9]. Moreover, the mechanical properties

of neat materials based on aliphatic polyesters should be improved so that they can be considered as scaffolds for bone regeneration [10,11].

Among the strategies allowing the regulation of properties of a polymer material, the preparation of biocomposites via the incorporation of biofunctional and/or reinforcing nano- or microparticles in the polymer matrix can be a matter of choice [10,12]. Such particles could improve the mechanical properties [11,13,14] or create the time–spatial distribution of biological ligands [15,16], while the polymer matrix would be degraded in vivo. Moreover, the ability to influence mineralization is one of the key properties of the resulting materials for bone tissue regeneration [16,17]. In this regard, hydroxyapatite [18] or ceramic nanoparticles [19] are often used as fillers both to improve mechanical properties and induce the mineralization of implanted materials.

Recently, it was shown that the application of graphene as a filler can improve the mechanical properties of polymer materials [20,21]. The use of graphene derivatives such as graphene oxide (GO) or reduced graphene oxide (rGO), which contain hydroxylic and carboxylic groups, can also improve the interaction of composites with cells and biomolecules and enhance cell growth, cell differentiation and cell proliferation [22,23]. At the same time, GO and rGO similar to graphene enhances the mechanical properties of polymer materials [24,25]. Recently, Krystyjan et al. and Mohamad et al. prepared the starch/chitosan [26] and chitosan/PLA-based [27] composites containing GO as filler, respectively. In both cases, the addition of GO improved the mechanical properties of polymer materials. Furthermore, the developed polymer composites containing GO showed bacteriostatic activity but were not toxic to human cells. Luo et al. have reported that GO incorporated on electrospun PLGA demonstrated osteoinductive properties when compared to the scaffolds without GO [28].

One of the main drawbacks in preparing of many composites is the aggregation of the fillers in a polymer matrix [29,30]. To overcome this obstacle, many studies have focused on modifying fillers to increase their phase compatibility and homogeneous dispersion state. Surface grafting with oligomers or polymers is a well-known approach to improve the filler distribution in the polymer matrix [31–33]. Recently, we have successfully applied this approach to improve the compatibility of the hydrophobic PLA and PCL with hydrophilic filler (nanocrystalline cellulose) by its grafting with amphiphilic polypeptide [32]. Such modification allowed the enhancement of the composite mechanical characteristics and also favored the matrix mineralization [12,34].

In the case of graphene-based materials, the strong π–π interaction contributes to a pronounced aggregation of graphene and, as a result, its poor distribution in the polymer matrix. Partly, this effect can be overcome by the use of GO or rGO. Furthermore, GO and rGO can give more stable dispersions in water and polar solvents than graphene. In particular, Gu et al. reported the modification of GO with polyamide in ethanol followed by melt spinning [35]. The developed composites provided excellent mechanical properties. Wang et al. proposed a modification of GO with poly(lactic acid) to adjust GO surface properties and enhance phase compatibility with polymer matrix (PLA) [36].

The goal of this study was the development of a technique for the modification of aminated rGO (rGO-Am) with oligomers of glutamic acid (oligo(Glu)) as a moiety with osteoinductive properties [16]. The grafting of oligo(Glu) from the rGO-Am surface was performed by the ring-opening polymerization of N-carboxyanhydride of γ-benzyl ester of glutamic acid initiated by primary amino groups of rGO-Am. The modification of rGO-Am was carefully tested by a number of physicochemical methods. Neat and modified rGO-Am was applied as a filler to prepare PCL-based biocompatible composite films whose mechanical and biological properties were investigated and compared. The biocompatibility of dispersions of the fillers as well as composite films with human MG-63 osteoblast-like cells was studied in order to assess the applicability of the developed composites as potential materials for bone tissue regeneration.

#### **2. Materials and Methods**

#### *2.1. Materials*

An aqueous dispersion of GO was purchased from Graphene Technologies (Moscow, Russia, www.graphtechrus.com accessed on 30 June 2019). γ-benzyl-L-glutamate (Glu(OBzl)) (>99%), triphosgene (98%) and α-pinene (98%) used for the synthesis of N-carboxyanhydride as well as *n*-hexylamine (98%) used as an initiator were purchased from Sigma-Aldrich (Darmstadt, Germany). All organic solvents used in this work were purchased from Vecton Ltd. (St. Petersburg, Russia) and distilled before use.

Low-molecular-weight poly(glutamic acid) (PGlu) used for comparison was synthesized in IMC RAS as described earlier [32] and had the following characteristics according to size-exclusion chromatography (SEC): *M*<sup>w</sup> = 10,800, *M*<sup>n</sup> = 7700 and Ð = 1.40. According to <sup>1</sup>H NMR, the used PGlu sample contained 15% of residual benzyl groups (PGlu(OBzl)).

Membranes for dialysis with molecular weight cutoff (MWCO) 1000, 3500, 6000−8000 and 12,000–14,000 were purchased from Orange Scientific (OrDialDClean regenerated cellulose dialysis tubing, Anaheim, CA, USA). Vivaspin concentrators used for ultrafiltration were products of Sartorius (Göttingen, Germany).

Human MG-63 osteoblast-like cell line was obtained from the Vertebrate Cell Culture Collection (Institute of Cytology RAS, St. Petersburg, Russia). They were cultured at 37 ◦C in humidified 5% CO<sup>2</sup> and cultured in EMEM (Lonza, St. Louis, MO, USA) containing 10% fetal bovine serum (FBS; HyClone, Logan, UT, USA), 1% NEAA (FBS; HyClone, Logan, UT, USA) and 1% penicillin/streptomycin (Sigma-Aldrich, Darmstadt, Germany). In all, 4~6 passage cells were used for this study. On attaining 80–90% confluency, the cells were trypsinized with trypsin-EDTA (Sigma-Aldrich, Darmstadt, Germany).

#### *2.2. Synthesis of Aminated Graphene*

The synthesis of the rGO-Am was carried out according to the previously described method [37]. In brief, GO suspension of 0.05 wt.% concentration was centrifuged at 18,100 g for 20 min (Sigma S-16 centrifuge, Sigma-Aldrich, Darmstadt, Germany), the supernatant was decanted away, and the sediment was transferred into a fluoroplastic cup and rinsed with 46% HBr (Sigma-Aldrich, Darmstadt, Germany). The obtained suspension was further heated at 80 ◦C for 48 h in air while stirring. Afterward, the synthesized intermediate (brominated graphene) was copiously washed by isopropyl alcohol using a glass filter (of 40 µm of pore size), the sediment was placed into a fluoroplastic cup, and rinsed with a saturated solution of ammonia in isopropyl alcohol. The obtained suspension was stirred for 24 h, resulting in the formation of rGO-Am. To purify the synthesized rGO-Am from the residuals of the reaction mixture, it was washed several times with isopropyl alcohol using a glass filter (of 40 µm of pore size).

#### *2.3. Modification of Aminated Graphene with Oligomers of Glutamic Acid*

Two milliliters of distilled and dried ethyl acetate were added to a glass Schlenk tube containing 110 mg of rGO-Am and the mixture was ultrasonicated with an ultrasonic probe UP 50H Hielscher Ultrasonics (Hielscher, Teltow, Germany) at 15–20% power for 1 min. The obtained dispersion was purged with argon for 3 min.

N-carboxyanhydride of γ-benzyl ester of glutamic acid (Glu(OBzl) NCA; monomer) was prepared using a standard protocol as described elsewhere [38,39]. Then, 230 mg of monomer was dissolved in 2 mL of dry ethyl acetate and the solution was purged with argon for 10 min. The resulting solution was added to the rGO-Am dispersion, ultrasonicated with an ultrasonic probe for 1–2 min, sealed, again purged with argon for 10 min, and incubated at 35 ◦C for 2 days. After this, 5 mL of DMF was added to the reaction mixture. The dispersion was ultrasonicated in the ultrasound bath for 2 min and finally centrifuged at 15,300 g for 15 min. After the removal of the supernatant, a fresh portion of DMF (5 mL) was added to the sediment, and the procedure of redispersion and centrifugation was repeated again. The washing procedure was repeated five times. Finally, the modified rGO-Am was redispersed in 7 mL of methanol, centrifuged as described

above, and the sediment was dried in a vacuum at 22 ◦C for 12 h. The product weight was 103 mg (~30%).

To remove the benzyl protective groups, 2 mL of the DMSO/TFA (1/1, *v/v*) mixture was added to 50 mg of rGO-Am-oligo(Glu(OBzl)). The reaction was carried out in an ice bath (0–5 ◦C) for 1 h. After this time, the ice bath was removed, 50 mL of DMSO/TFMSA solution (4/1, *v/v*) was added, and the dispersion was stirred for additional 1 h at 22 ◦C. After that, 6 mL of DMF was added to the resulting dispersion and dialysis against water using a membrane bag with MWCO = 1000 was carried out for 5 days. The purified product was freeze-dried and stored at 4 ◦C before use. The product weight was 33 mg (~66%).

#### *2.4. Characterization of rGO-Am and Its Derivatives*

The survey, C 1s, N 1s and O 1s X-ray photoelectron spectra (XPS) were measured using a Thermo Fisher ESCALAB 250Xi XPS system (Thermo Fisher Scientific, Waltham, MS, USA) with a monochromatic Al Kα X-ray source (*hv* = 1486.68 eV). For all the spectroscopic measurements, the studied materials were deposited onto the silicon wafers by the drop-casting of 20–30 <sup>µ</sup>L of the corresponding isopropyl suspension, of <sup>5</sup> <sup>×</sup> <sup>10</sup>−<sup>1</sup> wt.% of concentration with subsequent drying overnight at room temperature. Prior to the measurements, samples were evacuated down to a pressure *P* = 10−<sup>9</sup> Torr at least for 20 h to remove all adsorbates. The spectra were calibrated with respect to the Au 4f7/2 line (84.0 eV).

CasaXPS© software (Version 2.3.16Dev52, Casa Software Ltd.; Teignmouth, United Kingdom) was used for the deconvolution and quantification of the acquired C 1s, N 1s and O 1s X-ray photoelectron spectra. All the spectra were fitted with a Shirley background. In the case of C 1s spectra, a set of one asymmetric Doniach-Sunjic function ((DS0.10,100)(GL90)) and six symmetric Gaussian−Lorentzian convoluted functions of 70–30% ratio (GL(30)) was applied. At the same time, three symmetric Gaussian−Lorentzian converged functions of 70–30% ratio (GL(30)) were used for the deconvolution of the N 1s and O 1s spectra. The χ 2 minimization was ensured using the nonlinear least squares routine. Afterward, the C/O ratios and the relative concentration of the carbon atoms in different states were calculated.

Fourier-transform infrared spectroscopy (FTIR) was performed using IRAffinity-1 S Shimadzu (Shimadzu, Kyoto, Japan). The spectra were recorded for 0.2 mg of sample evenly distributed in 20 mg in a KBr tablet.

The hydrodynamic diameter and electrokinetic potential of neat and modified rGO-Am particles were determined by dynamic and electrophoretic light scattering (DLS and ELS, respectively). All measurements were performed in water at concentration of particles equal to 0.5 mg/mL using a Zetasizer Nano-ZS (Malvern Instruments, Malvern, United Kingdom) equipped with a He–Ne laser at 633 nm at scattering angles of 173◦ and 25◦ .

The morphology of the studied materials was evaluated with scanning electron microscopy (SEM) using JSM-7001F Jeol microscope (Jeol Ltd., Tokyo, Japan). Optical microscopy in transmitted and reflected light was carried out with the use of the Nikon Eclipse E200 microscope (Nikon Corp., Tokyo, Japan).

Thermogravimetric analysis (TGA) was performed with the use of a DTG-60 Shimadzu (Shimadzu, Kyoto, Japan) in an air atmosphere at a constant heating rate of 5 ◦C/min. The analysis was performed in the temperature range from 40 to 600 ◦C using the prehomogenized samples.

#### *2.5. Synthesis of PCL*

Polymerization of ε-caprolactone was carried out in bulk at 130 ◦C for 20 h. Before polymerization, the weighted portion of the monomer was placed into a Schlenk flask and purged with argon for 10 min. The ratio of monomer to stannous octoate was 3600. After synthesis, the polymer was dissolved in a minimum amount of chloroform and precipitated into cold methanol. The precipitated polymer was dried in a vacuum (ca. 150 Pa). PCL yield was 91%.

The molecular weight characteristics (*M*<sup>w</sup> and *M*n) and dispersity (*Ð*) of the polymer obtained were determined by SEC with the use of a Shimadzu HPLC system (Shimadzu, Kyoto, Japan) consisting of a pump LC-10AD VP, system controller SCL-10A VP, and refractometric detector RID-10A (Shimadzu, Canby, OR, USA) supplied with a Rheodyne 725i injection valve (Rohnert Park, CA, USA) and two columns of Agilent PLgel MIXED-D (7.5 × 300 mm, 5 µm) (Agilent, Santa-Clara, CA, USA). The analysis was carried out in THF at a temperature of 40 ◦C, a flow rate of the mobile phase of 1.0 mL/min. Calculations of molecular weight were fulfilled regarding polystyrene standards with molecular weights in the range of 2000–450,000. Data processing was performed using LC Solution Shimadzu software (version 1.25, Shimadzu, Kyoto, Japan). In addition, the intrinsic viscosity (η) of the synthesized PCL was calculated after measurements of viscosity for a set of solutions with different concentrations in CHCl<sup>3</sup> using Ostwald's capillary viscosimeter.

#### *2.6. Manufacturing of Composite Films*

The pure and composite films based on PCL were manufactured using the previously developed procedure for the PCL composite films with nanocrystalline cellulose [12]. Briefly, a 5% PCL solution in CHCl<sup>3</sup> (6.5 mL) was poured inside the glass cylinder (i.d. = 75 mm) with fixed cellophane bottom. The solution was left for 12 h in the air for chloroform evaporation. After cellophane removal, the obtained PCL-based films were dried in the air thermostat at 45 ◦C for 3 days. In the case of composite films, unmodified or modified rGO-Am was dispersed in PCL solution in chloroform and shortly ultrasonicated with the ultrasonic probe for 15–20 s at 15–20% power. The amount of a filler was equal to 0.5 and 1.0 wt.%. For neat rGO-Am, 3 wt.% of the filler was used also for comparison. Other manipulations were done as described above for the manufacturing of unfilled PCL films.

## *2.7. Mechanical Testing*

Mechanical properties of the films were studied under uniaxial extension using bandlike specimens of 2 mm × 20 mm using the AG-100kNX Plus Shimadzu universal mechanical system (Shimadzu, Kyoto, Japan). The thickness of the tested specimens was 90 ± 10 µm. The extension speed was 10 mm/min. The characteristics of the tested samples were calculated basing on the results of measurements for 7–11 fragments of the material and the results are given as average value ± SD.

#### *2.8. Biological Evaluation*

The MTT assay was employed to assess the cell viability and proliferation of the MG-63 on interaction with the material. This is a colorimetric assay measuring the reduction of yellow 3-(4,5-dimethythiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) substrate to an insoluble purple formazan product by mitochondrial succinate dehydrogenase enzyme. The MTT enters the cells where it gets reduced to an insoluble, dark purple colored formazan product. Neat and modified rGO-Am dispersions were evaluated in the concentration range from 4 to 1000 µg/mL using an adhesive 96-well plate (n = 3). In the case of films, the round-shaped specimens with a dimeter of 5 mm were glued by BF-6 medical glue (Tula Pharmaceutical Company, Tula, Russia) to the bottom of the nonadhesion 96-well plate (n = 3). Sterilization of glued films was performed by their exposure under a UV light of wide spectrum for 10 min. Next, 100 <sup>µ</sup>L of medium containing <sup>1</sup> <sup>×</sup> <sup>10</sup><sup>4</sup> cells was added to each well. The plates were incubated at 37 ◦C with 5% CO2. As a control for dispersions, MG-63 cells were cultured in a culture medium in an adhesive 96-well plate, while pure PCL material was considered as a control for composite films. After 3 days of incubation, MTT (Sigma, St. Louis, MO, USA) solution was added to each well and incubated for 2 h at 37 ◦C with 5% CO2. After the incubation time, the above solution was discarded, and the colored formazan crystals formed were solubilized by adding 50 µL of dimethyl sulfoxide. The absorbance was read at 570 nm in a multiwell plate reader (Thermo Fisher Multiscan Labsystems, Waltham, MA, USA). The absorbance values were plotted using MS Excel software.

#### **3. Results and Discussion** *3.1. Synthesis of rGO-Am* 260

*Polymers* **2021**, *13*, x FOR PEER REVIEW 6 of 18

#### *3.1. Synthesis of rGO-Am* Figure 1a displays the survey spectra of the GO presented by dominant O 1s and C 261

Figure 1a displays the survey spectra of the GO presented by dominant O 1s and C 1s lines at *hv* = 532.5 eV and *hv* = 284.7 eV with the absence of other spectral features, verifying the purity of the initial material. Upon the amination, the O 1s line substantially diminished, while the N 1s peak at ca. *hv* = 400.1 eV appeared, pointing out the successful introduction of amines by the substitution of the oxygenic moieties in the treated GO. The concentration of nitrogen was estimated to be ca. 3.02 at.%. To further analyze the composition of nitrogen-containing groups embedded upon the applied treatment, high-resolution N 1s spectra were acquired and processed (Figure 1b). Three distinct peaks positioned at *hv* = 398.9 eV, *hv* = 399.8 eV and *hv* = 401.9 eV were discerned, which are related to the presence of pyridines, amines (both primary and secondary), and implemented graphitic nitrogen, respectively [40,41]. As seen, amines are the dominant form of the introduced nitrogen functionalities with their relative content estimated to be up to 88.4%. Accordingly, the concentration of the amines in the synthesized rGO-Am is ca. 2.63 at.%. 1s lines at *hv* = 532.5 eV and *hv* = 284.7 eV with the absence of other spectral features, ver- 262 ifying the purity of the initial material. Upon the amination, the O 1s line substantially 263 diminished, while the N 1s peak at ca. *hv* = 400.1 eV appeared, pointing out the success- 264 ful introduction of amines by the substitution of the oxygenic moieties in the treated GO. 265 The concentration of nitrogen was estimated to be ca. 3.02 at.%. To further analyze the 266 composition of nitrogen-containing groups embedded upon the applied treatment, high- 267 resolution N 1s spectra were acquired and processed (Figure 1b). Three distinct peaks 268 positioned at *hv* = 398.9 eV, *hv* = 399.8 eV and *hv* = 401.9 eV were discerned, which are re- 269 lated to the presence of pyridines, amines (both primary and secondary), and imple- 270 mented graphitic nitrogen, respectively [40,41]. As seen, amines are the dominant form 271 of the introduced nitrogen functionalities with their relative content estimated to be up 272 to 88.4%. Accordingly, the concentration of the amines in the synthesized rGO-Am is ca. 273 2.63 at.%. 274

5% CO2. As a control for dispersions, MG-63 cells were cultured in a culture medium in 251 an adhesive 96-well plate, while pure PCL material was considered as a control for com- 252 posite films. After 3 days of incubation, MTT (Sigma, St. Louis, MO, USA) solution was 253 added to each well and incubated for 2 h at 37 °C with 5% CO2. After the incubation 254 time, the above solution was discarded, and the colored formazan crystals formed were 255 solubilized by adding 50 μL of dimethyl sulfoxide. The absorbance was read at 570 nm 256 in a multiwell plate reader (Thermo Fisher Multiscan Labsystems, Waltham, MA, USA). 257 The absorbance values were plotted using MS Excel software. 258

**3. Results and Discussion** 259

**Figure 1.** The XPS characterization of initial GO and rGO-Am layers: (**a**) survey spectra, (**b**) high-resolution N 1s spec- 276 tra and (**c**) high-resolution C 1s spectra. 277 **Figure 1.** The XPS characterization of initial GO and rGO-Am layers: (**a**) survey spectra, (**b**) high-resolution N 1s spectra and (**c**) high-resolution C 1s spectra.

The elimination of the oxygenic groups upon the consecutive bromination and 278 amination is pointed out by the C 1s spectra of the GO and rGO-Am (Figure 1c). The ini- 279 tial GO is functionalized by the hydroxyls and epoxides at the basal plane of the gra- 280 phene layer along with the carbonyls and carboxyls at its edges. The presence of these 281 functional groups is signified by the C–O–C&C–OH, C=O and COOH peaks centered at 282 The elimination of the oxygenic groups upon the consecutive bromination and amination is pointed out by the C 1s spectra of the GO and rGO-Am (Figure 1c). The initial GO is functionalized by the hydroxyls and epoxides at the basal plane of the graphene layer along with the carbonyls and carboxyls at its edges. The presence of these functional groups is signified by the C–O–C&C–OH, C=O and COOH peaks centered at *hv* = 286.8 eV, *hv* = 288.2 eV and *hv* = 289.1 eV, respectively [42,43]. The C–V peak is related to the nonterminated carbon atoms of the vacancy defects and Stone–Wales defects [44]. The relative concentration of the oxygenic groups and carbon atoms in various states is presented in Table 1, whereas the C/O ratio was calculated to be 2.51. After the amination, all the spectral features related to the oxygenic groups were absent with the retention of only a dominant C=C peak at ~284.6 eV, corresponding to the pristine π-conjugated graphene network, and a C–C peak at ~285.1 eV related to nonconjugated C–C bonds [43]. The concentration of the preserved oxygenic groups is less than 3 at.% with the corresponding C/O ratio of rGO-Am of 32.51 (Table 1). In addition, a C–N peak at ~285.9 eV appears due to the formation of carbon–nitrogen bonds upon the introduction of amines and other nitrogen species [37,45]. Given XPS data on the composition of functional groups and the

275

size distribution of the rGO-Am derived from both the measurements using LD technique and the acquired SEM images of arrays of flakes (Supplementary Materials, Section S1, Figure S1), the molar ratio of the presented amines was estimated. The calculated values are ca. 7.1 mmol/g, which corresponds to 5.4 <sup>×</sup> <sup>10</sup><sup>9</sup> amine groups per 1 g of rGO-Am. The details on the calculation of the molar ratio are presented in Supplementary Materials, Section S2.

**Table 1.** Composition of functional groups (at.%), carbon in different states and C/O ratio estimated from the analysis of the deconvoluted C 1s X-ray photoelectron spectra of GO and rGO-Am.

