Covalent chemical functionalization enhances the biodegradation of graphene oxide

Full details on the synthesis of the organic precursors and the different functionalized GO and their characterizations are reported in Supporting Information.

GO, GOTEG, GOAZC and GODHBA powder (1 mg of each sample) were dispersed separately in 5 ml phosphate buffer saline (PBS) and sonicated for 2 min. HRP (1.2 mg) pre-solubilized in 2 ml of PBS was then added. To enable the enzymatic activity, 10 µl of a 0.1 m solution of H₂O₂ (final concentration 142 µM) were added once per day for all duration of the experiments. The control samples (without addition of HRP) of all GO conjugates were also prepared in the same way adding the same amount of H₂O₂ (final concentration 142 µM) every day for all duration of the experiments. All suspensions were stirred in the dark at r.t. for the entire duration of the experiments. Aliquots (200 µl) were withdrawn at 0, 7, 12 and 20 d and stored at  −20 °C in the dark until characterization by different techniques. In the case of GO and GOTEG, the degradation experiments were carried out for 30 d, after refreshing HRP (1.0 mg) after 20 d.

The interaction of HRP with the surface of GO sheets was evaluated by native polyacrylamide gel electrophoresis (PAGE) after incubating HRP with the different conjugates. Briefly, 50 µg of each GO conjugate was dispersed in PBS (10 µl) and sonicated for 2 min. Then HRP (60 µg in 10 µl PBS) was added to each suspension and allowed to interact during 20 min incubation. At the end of the incubation time, Laemmli buffer was added to the mixture and the samples were loaded on a Mini PROTEAN^®TGX^™ 10% precasted gel (BioRad). Gel electrophoresis was then run under non-reducing conditions to evaluate the interaction between the enzyme and GO using a Mini-PROTEAN II apparatus (BioRad) and applying 150 V during 40 min. The gels were then fixed using aqueous solution (methanol 45% and 1% acetic acid) followed by staining with Coomassie Blue (2 h at room temperature) and scanning densitometry was performed with a GS-800 Calibrated Densitometer (BioRad). The densitometry analysis was performed using the ImageJ 1.48f software. The intensities of the bands corresponding to HRP after incubation with the different GO samples were expressed relatively to the band of HRP alone (i.e. considered as 100% of intensity).

All GO models were generated using the building function provided by PyMOL (Script provided by the ERG research group and PyMOL Molecular Graphics System, Version 1.7.4 Schrödinger, LLC) [35]. Four different kinds of graphene oxide models were generated: GO, GOTEG, GOAZC and GODHBA. The functional groups (TEG, AZC, DHBA) were attached at the edge in the middle of the GO sheets. The horseradish peroxidase (HRP) structure was taken from the RCSB Protein Data Bank (PDBID: 1H5A, chain A).

The input files required for the docking (in pdbqt format) for both receptor and ligands were generated using the ADT tools package provided by AutoDock4.0. The docking was performed using AutoDock Vina [36]. We used the following parameters: a cubic box was built around the protein, with the center of the protein as center of the cube (x  =  5.766, y  =  5.952, z  =  13.735 for HRP). The grid map was calculated automatically by the Vina software. The dimensions of the cube were 126  ×  126  ×  126 points as x, y and z sizes for both receptors, with a spacing of 0.4 Å between grid points. All other parameters were kept as default in AutoDock Vina. All docking experiments were launched on 4 CPUs. The docking results were analyzed using PyMOL for visual inspection. All conformations in each predicted binding sites were examined separately.

The starting GO used in this work was obtained from Grupo Antolin (Spain) in a powder form, easily dispersible in water. First, we prepared the triethylene glycol diamine conjugated GO (GOTEG) via epoxide ring opening reaction (figure 1) [37, 38]. This GO conjugate was used as control in the biodegradation study. Indeed, TEG linker has been introduced between the GO surface and catechol in GODHBA conjugate (see below). The amount of amino functions was measured by Kaiser test resulting in 191 µmol g⁻¹ of GOTEG [39, 40]. FTIR spectroscopy analysis of GOTEG confirmed the covalent conjugation of TEG diamine to GO (figure S1) (stacks.iop.org/TDM/5/015020/mmedia). This was also supported by TGA (figure S4). Next, GO was treated with 3,4-dihydroxybenzoic acid (catechol moiety) linked to diaminotriethylene glycol (compound 5, DHBA-TEG-NH₂; see Supporting Information for the preparation and characterization), leading to the formation of GODHBA (figure 1). The coupling of DHBA-TEG-NH₂ to GO was confirmed by FTIR spectroscopy (figure S2) and TGA (figure S5). In the FTIR spectrum of GODHBA (figure S2), the absence of epoxide peak at 1230 cm⁻¹ confirmed the ring opening of the epoxides by DHBA-TEG-NH₂. In addition, peaks at 2914 and 2850 cm⁻¹ correspond to C–H vibrations of the methylene groups of TEG linker.

Zoom In Zoom Out Reset image size

To explore the possibility of other HRP ligands, we prepared a second conjugate containing a coumarin derivative. Initially, GO-Alkyne was synthesized via epoxide ring opening by treating GO with 1-amino-3-butyne, similarly to GOTEG. GO-Alkyne was characterized by FTIR spectroscopy and TGA (figures S3 and S6). Then, we conjugated 3-azido-7-hydroxycoumarin (AZC) to GO-Alkyne by click chemistry (see experimental section for details) [41]. The strong peak at ~1605 cm⁻¹ and the small peak at 1077 cm⁻¹ in the FTIR spectrum of GOAZC (figure S3) are indicative of the triazole ring [42–44].

In order to characterize the dispersibility and the stability of the different samples, we measured the zeta potential (ξ) of all GO conjugates dispersed in water at pH 7. We found that GO suspension had ξ  =  −40.0 mV (standard deviation (SD): 1.1), which was significantly increased to  −28.7 mV (SD: 3.05) in the case of GOTEG. This is due to the introduction of a certain amount of amino functions on the surface of GO. The presence of amino functions in GOTEG sample was confirmed by the positive surface potential at pH 2, ξ  =  5.23 mV (SD: 1.4). The reduction of ξ is due to a charge compensation between the positively charged amino groups and the high concentration of oxygenated species still present on the surface of GOTEG. The zeta potential of GOAZC and GODHBA was  −28.33 mV (SD: 1.36) and ξ  =  −33.13 mV (SD: 1.49), respectively. The slightly lower negative value for GODHBA could be due to the additional two hydroxy groups of DHBA. Next, the morphology of all GO conjugates was analyzed by TEM (figure S7). The flat 2D morphology was maintained after functionalization in the case of GOAZC and GODHBA (figures S7(B) and (C), respectively), whereas in the case of GOTEG moderately aggregated sheets were observed (figure S7(D)).

The main objective of this study was to assess the capacity of the functional groups on GO to enhance the biodegradability of the material. For this purpose, we carried out the degradation experiments with all GO conjugates by incubating with HRP followed by the addition of hydrogen peroxide once in a day to maintain the same catalytic activity along the entire period [24, 29]. Briefly, 1 mg of each sample was dispersed in 5 ml of PBS (figure S8) and treated with HRP and H₂O₂. The first sign of degradation was given by the change in the color of the GO suspensions. In particular, there was a significant reduction in the color intensity of GOAZC and GODHBA suspensions after 20 d (figure S8), whereas moderate color changes were instead observed for GO and GOTEG suspensions even after 20 d. We decided to continue further the degradation of GO and GOTEG up to 30 d. After this period these suspensions were still not translucent suggesting incomplete degradation. The disappearance of the dark brownish color of GOAZC and GODHBA was attributed to the degradation of GO conjugates by enzymatic catalysis [24, 29, 32]. These initial observations suggested that GO functionalized with catechol and coumarine derivatives was degraded faster than GO, while the addition of diaminotriethylene glycol likely slowed down the rate of degradation.

To confirm the visual observations, we applied a series of analytical, spectroscopic and microscopic techniques. The changes in the dimensions (i.e. lateral size) of all GO conjugates before and after treatment with HRP/H₂O₂ was followed by dynamic light scattering (DLS) [33]. The size of GO dispersed in PBS before the degradation was comprised between ~800 and 3000 nm (figure 2) with few traces of material around 10 µm. The dimensions were reduced to ~400–600 nm after 30 d treatment with HRP. These data suggest that the degradation of this type of GO by HRP was occurring only moderately. The size of GOAZC passed from ~1000 nm to ~300 nm, after 20 d. In the case of GODHBA, 20 d action of HRP resulted in a clear reduction of the sheets size mostly below 100 nm. A slight effect was instead evidenced on GOTEG sheets even after 30 d (variation from 700–1500 nm to 400–650 nm), indicating that the degradation has not taken place effectively. Overall, DLS studies supported that the degradation was effective and much faster in the case of GOAZC and GODHBA over GO and GOTEG, respectively. The degradability trend derived from DLS analysis is: GODHBA  ⩾  GOAZC  >  GO  >  GOTEG, in good agreement with the color changes of the suspension. Simultaneously, to assess the effect of H₂O₂ alone on degradation, DLS measurements were also performed on all GO conjugates treated with hydrogen peroxide. In this case the dimensions of the samples were not affected (figure 2, last column) [33, 34].

Zoom In Zoom Out Reset image size

Next, we employed Raman spectroscopy to obtain further insights about the effects of the degradation process on the structure of the GO conjugates during the enzymatic treatment. Raman spectroscopy was used for the systematic analysis of the oxidation level of graphitic materials during their degradation by various peroxidases [26, 31–33]. Raman spectra of all GO conjugates at the different time points of degradation were recorded (figure 3). The degradation state or oxidation level of GO was quantified by measuring the intensity ratios of D/G bands in each spectrum [45]. Figure 3(A) shows Raman spectra of GO. The characteristic D (~1350 cm⁻¹) and G (~1600 cm⁻¹) bands of GO are clearly visible with D/G ratio of 0.91 at day 0 [33, 46]. After 7 d treatment with HRP/H₂O₂, D/G ratio was increased to ~1.03 and reached ~1.21 after 12 d. After 20 d, the intensities of D and G bands were significantly reduced compared to the beginning, almost reaching the baseline by day 30, however maintaining a similar D/G ratio (1.02). In the same interval of time, we also observed a few distinct spectra showing high intense D and G bands (figures S9(A) and (B)), revealing that the degradation of GO was not uniformly occurring and incomplete even after 30 d. This result is not in agreement with the previous report by Star and co-workers [29], where after 20 d complete degradation of GO was observed by HRP. The dissimilar results can be ascribed to the type of GO used in the two studies [29, 33]. GO previously reported [29] was synthesized using well-known Hummers method, while GO used in this work was commercially produced from carbon fibers. Both GOs can vary in the amount of oxygen functional groups and aqueous dispersibility [33], which could explain their different degradability by HRP.

Zoom In Zoom Out Reset image size

Raman spectra of GOAZC at day 0 had D/G ratio of 0.79 (figure 3(B)). After 7 d, a significant reduction in the intensities of D and G bands was observed, leading to an increase of D/G ratio to 0.84, sign of enzymatic degradation. Twelve day treatment resulted in the disappearance of both bands which was complete by 20 d. This result strongly evidenced how the degradation of GOAZC was much faster and complete compared to GO. In the case of GODHBA, D/G ratio was 0.99 at day 0 (figure 3(C)). After 7 d treatment, the D/G ratio reached 1.14, with a significant decrease and broadening of D and G bands. After 12 d, the intensities of both bands were negligible, revealing a complete degradation of this conjugate. This was confirmed at day 20. Finally, GOTEG had D/G ratio of 0.88 at day 0 (figure 3(D)). D/G ratio reached 1.06 after 7 d and slightly increased to 1.10 at day 12, while the intensities of D and G bands gradually decreased. Between day 20 and 30, two kinds of spectra were observed. In the first type, both D and G bands were less intense but D/G ratio was almost unchanged (~1.07). The second type corresponds to spectra with high intense D and G bands (figures S10(A) and (B)). This result indicates that the degradation of GOTEG is incomplete even after 30 d. The degradation results seem to be slower than starting GO.

In view of these results, we concluded that the specific functional groups on the surface of GO are playing a crucial role in the activity of the enzyme in the presence of hydrogen peroxide. During the treatment with HRP, the increase in the D/G ratio was dependent on the time of treatment, suggesting the formation of defects on the graphitic surface of respective GO sheets due to the oxidation caused by HRP catalysis [29]. The trend for the biodegradability of GO conjugates from the Raman spectroscopy analysis follows as: GOAZC  ⩾  GODHBA  >  GO  >  GOTEG, consistent with DLS observations. We also assessed the D/G intensity ratio for all the control GO samples treated only with H₂O₂ without HRP. After 20 d, the D/G ratio was 1.05, 0.94 and 1.02 for GO, GOAZC and GODHBA, respectively, while it was 1.05 for GOTEG treated for 30 d (figure S11). In contrast to HRP treatment, the incubation of the GO conjugates with H₂O₂ alone for the same time did not affect significantly neither the D/G ratio nor the intensity of the D and G bands compared to the values at day 0.

To investigate the morphological changes during the degradation, we characterized all the GO conjugates at different time points by TEM. The starting GO sheets at day 0 with HRP are displayed in figure 4. After 12 d incubation, porous GO sheets started to appear, gradually increasing until day 20. However, the degradation of GO was not complete, since we still observed some aggregated sheets (figure S12). Additional 10 d treatment with HRP did not change significantly the morphology of the GO sheets (figure S13). In the case of GOAZC (figure 4), after 12 d HRP treatment, most of the sheets were converted into 'holey sheets' and broken down into tiny fragments. However, all these porous sheets were transformed into nanoparticles after 20 d. For GODHBA (figure 4), the degradation was more significant after 12 d, where complete rupturing of the sheets into tiny fragments was observed. By 20 d, all these tiny fragments were mostly absent revealing nearly complete degradation of GODHBA. In the case of GOTEG (figure 4), the size of the sheets was reduced only after 20 d, showing moderate porous morphology along with many aggregated sheets (figure S14), indicating that GOTEG was not degraded uniformly, similarly to GO. These results were also observed with Raman spectroscopy analysis of GOTEG. On other hand, the control experiments performed by only adding H₂O₂ to the respective GO samples did not affect sheet morphologies of any of the samples (figure S15).

Zoom In Zoom Out Reset image size

Overall, the TEM results were in good agreement with DLS and Raman analyses, showing GOAZC and GODHBA much quicker and effective degradation than GO and GOTEG. In addition, the combination of Raman, DLS and TEM indicated the key role of the specific functional groups (AZC and DHBA) in the degradation of GO. Since both catechol and coumarin derivatives are known to be good reducing substrates for HRP, they clearly favor the interaction with the enzyme leading to the effective degradation of carbon materials, as we demonstrated for MWCNTs functionalized with the same molecules [34]. We can hypothesize that the presence of free amines on the surface of GOTEG reduces the interaction between GO and HRP due to electrostatic repulsion. To support this hypothesis, we studied the interaction between GO and HRP by gel electrophoresis and molecular modeling.

A native polyacrylamide gel electrophoresis (PAGE) was carried out to acquire more information on the interaction of HRP with GO conjugates. HRP was incubated with each GO conjugate for 20 min before running the gel. Some differences in the respective bands in the gel were observed for each GO conjugates (figure 5(A)). In fact, the densitometry analysis (figure 5(B)) shows that all GO conjugates are able to slightly reduce the migration of HRP, confirming the reduced intensity of HRP gel band at  ≈44 kDa (figure 5(A)) [29]. This analysis suggests that functionalized GODHBA interacted with HRP a little better than pristine GO, likely because the additional hydroxyl groups of DHBA may increase the electrostatic interaction between the sheets and protein. Previous reports revealed that there is a specific interaction between GO [29, 47], or carboxylated CNTs [48, 49] and HRP compared to reduced GO or pristine CNTs. Our analysis suggests that HRP was adsorbed onto the GO sheets dependent on the surface functionalization, influencing, consequently, the enzymatic degradation.

Zoom In Zoom Out Reset image size

Finally, we performed molecular docking studies to better understand the interaction of HRP with the different surface modified GO conjugates. The complexes between GO sheets and HRP are illustrated in figure 6. On the basis of the simulation, the closest distance between the heme active site of HRP and each GO conjugate sheet was measured. For the GO/HRP complex, we found a distance of 14.8 Å, which is slightly higher than the value reported previously for the same interaction (12.8 Å) [29]. This is due to the difference in the size of the respective GO sheets and the amount of oxygenated groups present on the surface of GO. In the case of GOAZC, the nearest distance between the AZC molecule and the heme is 12.6 Å. The distance between DHBA molecule of GODHBA and HRP active site is 17.2 Å. Although the distance is higher compared to GO/HRP, the GO sheet is nicely projected towards HRP. In the case of GOTEG, the position of the sheet is in a totally different side of the protein, far away from the active site of the enzyme (>28.3 Å) likely due to a nonspecific interaction.

Zoom In Zoom Out Reset image size

As discussed earlier, the electrostatic repulsion between the positively charged amine groups of GOTEG and the cationic nature of HRP plays a major role to keep graphene far away from the heme site. This result was also supported by gel electrophoresis (figure 5). The molecular docking study revealed that HRP could bind more tightly to GOAZC over GO. Because of the projection of 7-hydroxy azidocoumarin group conjugated to GO, the shortest distance between the heme active site and the surface of GOAZC was attained (figure 6). In addition, we also identified the number of amino acid residues of HRP within 5 Å proximity to the surface of all GO conjugates (table S1), since these residues were found earlier to affect the degradation process [29]. Compared to GO, there are 9 additional amino acid residues for GOAZC and 5 extra residues around GODHBA. Overall, we believe that these additional residues could lead to a better binding of HRP to GODHBA and GOAZC compared to GO and GOTEG. The vicinity and the position of the different GO conjugates in respect to the enzyme, as assessed by the simulations, further confirm the degradability of the GO samples and the speed of the process.

The different characterization methodologies presented in this study confirmed that the degradation of GO functionalized with coumarin and catechol moieties was faster and almost complete in comparison to native GO and amine-functionalized GOTEG. Previous studies highlighted that GO or oxidized CNTs should be in close proximity to the heme active site of HRP for more effective degradation [24, 31]. Our molecular models revealed that GOAZC and GODHBA have additional amino acid residues of HRP within 5 Å distance compared to GO and GOTEG, which lead to a better interaction with HRP, as also supported by gel electrophoresis. Hence, the combination of close proximity to the heme-active site of HRP and additional amino acid stabilization could be responsible for a more effective degradation of GOAZC and GODHBA in comparison to GO and GOTEG. In addition, most of these amino acids participated in the catalysis of oxidation reactions [24, 50, 51]. To highlight the importance of the covalent functionalization to enhance the biodegradation of GO conjugates, we have performed a control experiment using a non-covalent mixture between GO and DHBA. The simple presence of DHBA in solution does not accelerate the degradation process of GO, proving that the ligand needs to be covalently attached to the surface of the nanomaterial (figure S16). Raman spectroscopy analysis showed the increase in D/G ratio as well as the broadening of D and G bands during HRP treatment. This is due to the gradual increase of defects on the surface of GO conjugates along with the cleavage of C–C or C–O bonds by radical intermediates of HRP [52–54]. On the other hand, oxidation of GOAZC and GODHBA could start via either oxidizing the functional molecules (coumarin or catechol) attached to GO or enhancing the enzymatic activity of HRP, since both were proven to be efficient reducing substrates for HRP [55–60]. The free OH group in the position 7 of benzopyran ring (AZC) seems to play an important role in the enhanced degrading action of HRP, likely acting as an effective donor of radical species [23, 34, 48, 61]. Catechol derivatives (DHBA) act as redox mediators by favoring electron transfer, thereby preventing enzyme inactivation [34, 57, 62]. Recently, we have also demonstrated that AZC and DHBA conjugation to MWCNTs had improved degradation over oxidized MWCNTs [34]. We can conclude that the degradation of GODHBA and GOAZC also followed a similar pathway of structural decomposition.