Carbon nanotubes and graphene, the nanoscale sp² allotropes of carbon, have garnered widespread attention as a result of their remarkable electrical, mechanical, and optical properties and the promise of new technologies that harness these properties. Today, these breakthroughs no longer remain just a vision. The Project on Emerging Nanotechnologies has documented over 30 products containing these carbon nanomaterials (CNMs), in the form of composites, dispersions, and solid-state devices, from a broad selection of industries: aerospace, consumer electronics, cosmetics, and sporting goods.¹ The manufacture and eventual disposal of these products may result in the release of CNMs into the environment and subsequent exposure of humans, animals, and vegetation to these CNMs. Given possible pro-inflammatory and toxic effects of CNMs, much attention has been focused on their health effects and interactions with living systems,^2–6 with other review articles addressing the distribution, toxicity, and persistence of CNMs in the environment.^7–9

Recently, several groups have reported the degradation of CNMs catalyzed by naturally occurring enzymes from plants and humans. These findings suggest that the potentially detrimental effects of CNMs associated with their biopersistence may be mitigated thereby enabling these materials to be used more widely in a safe manner. Here, we review these findings and discuss possible mechanisms of enzymatic degradation. We relate the oxidative potencies of these enzymes to those of inorganic oxidants employed to oxidize/degrade CNMs. We further elucidate the role of chemical functionalization in CNM degradation. Finally, we discuss data relating to the oxidative biodegradation of CNMs both in vitro¹⁰ and in vivo¹¹ as potential means to mitigate the distribution and toxicological effects of CNMs after exposure.

The scientific literature is rich with reports regarding the covalent modification of CNMs^12–15 with those relating to oxidizing processes summarized in Scheme 1. Strong oxidants, such as H₂SO₄ and HNO₃, disrupt the sp² carbon lattice of unfunctionalized CNMs and introduce oxygen-containing functional groups. Typical morphological changes accompanying such oxidation are a decrease in length for nanotubes¹⁶ and the appearance of holes in graphene sheets.¹⁷ Sustained oxidation leads to a loss of material from the carbon framework in the form of by-products such as CO, CO₂, as well as aliphatic and polyaromatic hydrocarbons.^10,18 While it was clearly demonstrated in the literature that strong oxidants can oxidize CNMs, these nanomaterials were largely considered to be biopersistent under physiological conditions and in the environment. We, however, hypothesized that like biodegradable polymers CNMs may undergo enzymatic oxidative modifications and biodegradation both in vivo and in the environment.

To date, primarily two peroxidase enzymes, horseradish peroxidase (HRP, Figure 1A) and myeloperoxidase (MPO, Figure 1B) have been studied for the degradation/biodegradation of CNMs. For over a century, HRP (molecular weight ~44 kDa), an enzyme secreted by the root of the horseradish plant (Armoracia rusticana), has been utilized in chemical synthesis, biotechnology, and bioremediation.¹⁹ Another enzyme, MPO, consisting of two identical dimers with a total weight ~144 kDa, is expressed in inflammatory cells (predominantly neutrophils) of animals and humans. MPO generates reactive intermediates and oxidants that serve to destroy infecting bacteria.^20–22

Both HRP and MPO contain a heme group as their active sites, which is a Fe(III) protoporphyrin IX (Figure 1C). When the enzymes are inactive, the heme group is in the ferric resting state (Figure 1D). In the presence of hydrogen peroxide (H₂O₂), the heme group undergoes a protein-assisted conversion to a ferryl oxo iron (Fe⁴⁺=O) porphyrin p cation radical known as Compound I, which is subsequently returned to the ferric resting state in two sequential, one-electron transfer steps.^19,22 In the first step, the transient intermediate, Compound II, forms when the porphyrin p cation radical is reduced as a substrate (AH₂) is oxidized.^19,22 AH₂ is further oxidized when the ferryl oxo iron is reduced to the ferric resting state.^19,22 Under the acidic conditions of the neutrophil and in the presence of chloride (Cl^–), Compound I of MPO abstracts two electrons from Cl^– to yield hypochlorous acid (HOCl) as it returns to the ferric resting state.²²

Table 1 lists the redox potentials of HRP and MPO. While HRP is a weaker oxidant than MPO, the redox potentials of HRP and MPO are comparable to those of nitric acid (0.957 V) and ozone (1.24 V), respectively.²³ Moreover, HOCl, which has a redox potential of 1.63 V,²³ functions as a stronger oxidant than MPO at each of its steps of the peroxidase cycle.

We demonstrated that carboxylated single-walled carbon nanotubes (SWCNTs), underwent morphological changes (shortening and deformation) when statically incubated at 4°C in the presence of HRP and H₂O₂ (~40 μM).25 Utilizing transmission electron microscopy (TEM), SWCNTs with a mean length ~500 nm (Figure 2A) were observed to shorten by ~45% over 8 weeks, and mostly residual globular material was detected at 12 weeks (Figure 2B). In contrast, pristine SWCNTs failed to degrade when incubated with HRP and H₂O₂ (~80 μM), from which we theorized that defects or functionalized sites were important facilitators of enzymatic action.²⁶ Bianco and coworkers corroborated these findings in a study of HRP/H₂O₂ treatment over 30 days, using Raman spectroscopy to monitor changes in the carbon lattice (Figure 2C).²⁷ The initial Raman spectrum exhibited a strong D band, which was attributed to SWCNT carboxylation. After 7 days of incubation, the D band weakened, suggesting that highly carboxylated nanotubes were oxidized first; on day 30, no Raman signal was observed because the nanotubes were completely oxidized. During the degradation process, the SWCNTs were observed to open via an unzipping mechanism yielding graphene sheets. This result was consistent with recent reports by Tour and coworkers²⁸ and Dai and coworkers,²⁹ which revealed that chemical treatments could longitudinally unzip both SWCNTs and MWCNTs to produce graphene nanoribbons (GNRs, Scheme 2).

More recently, the investigation of CNM degradation was expanded to multiwalled carbon nanotubes (MWCNTs). In one study, Bianco and coworkers subjected chemically carboxylated MWCNTs to HRP/H₂O₂ treatment, and over 60 days, MWCNTs were observed to shorten between 25 to 63% (Figure 3A). The observed shortening contrasts the unzipping observed for SWCNTs treated under identical conditions.²⁷ Independently, during an 80 day study, we investigated the process of MWCNT degradation catalyzed by HRP/H₂O₂.³⁰ TEM, including high-resolution imaging, revealed that carboxylated MWCNTs decreased in both length and diameter (Figure 3B), but unlike SWCNTs treated with HRP/H₂O₂, MWCNTs failed to disappear.³⁰ The decrease in length, in agreement with the report by Bianco and coworkers,²⁷ suggested that degradation proceeded from the ends of the nanotubes. The concurrent decrease in diameter could be explained by exfoliation of defect-containing outer walls, which were unzipped via the enzymatic oxidation, yielding nanotubes with smaller diameters and possessing mostly pristine outer walls (Figure 3C). The exposure of these newly revealed, more pristine layers coincided with a slowdown in the rate of degradation. Again, we could infer that defects or functionalized sites were important facilitators of enzymatic action. This notion was further supported by our experiments with nitrogen-doped MWCNTs, which were effectively degraded by the HRP/H₂O₂ system over 90 days.³⁰

The importance of functionalization for CNM oxidation/degradation by HRP/H₂O₂ was further supported by experiments with graphene oxide (GO), a two-dimensional sheet of oxidized graphitic carbon comprising one layer of carbon atoms decorated with oxygen functionalities like epoxides and carboxylates (Figure 4A). GO incubated with HRP/H₂O₂ was observed via TEM and Raman spectroscopy to undergo enzymatic oxidation.³¹ After 8 days of incubation, with daily additions of H₂O₂, holes with a diameter of 2.1 ± 0.6 nm formed on the basal plane of the GO sheet.³¹ By day 10, the diameter of the holes increased more than 12-fold to 26.7 ± 12.8 nm (Figure 4B), and by day 20, a majority of the GO was oxidized completely. Raman spectroscopy corroborated the TEM observations. In the first 4 days, the D/G intensity ratio increased from 1.1 to 1.6, and on day 20, both the D and G bands disappeared from the signal, indicating progressive oxidation leading to the elimination of the graphitic lattice (Figure 4C). For comparison, the effects of HRP/H₂O₂ treatment on a chemically reduced graphene oxide (RGO) were studied. RGO has the same single-sheet geometry of GO but is mostly devoid of GO’s oxygen-containing functional groups. In contrast to GO, the HRP/H₂O₂ treatment failed to oxidize RGO (Figure 4D). Thus, the defect sites introduced by chemical oxidation appeared to have a significant role in the subsequent enzymatic oxidation of the CNM, irrespective of its geometry.

Having examined CNM degradation catalyzed by HRP/H₂O₂, we turn to the report of MPO-catalyzed degradation. With the knowledge that human myeloperoxidase (hMPO) degrades implantable polymers such as poly(ester-urea-urethane), we hypothesized that hypochlorite and reactive radical intermediates generated by hMPO would catalyze the oxidation of carboxylated SWCNTs. Nanotube suspensions incubated with the oxidants hMPO/H₂O₂ or hypochlorite appeared lighter in color after 24 hours, indicating a decrease in graphitic material (Figure 5A). Moreover, the SWCNT suspension in which NaCl was added to hMPO/H₂O₂ exhibited significant oxidation as evidenced by its nearly clear appearance after 24 hours. This result suggested an important role of the strong oxidant, hypochlorous acid, in MPO-catalyzed degradation of CNMs.³² Raman spectroscopy conducted at intervals up to 24 h showed a progressive decrease in the G band until it was quenched, indicating complete degradation of graphitic lattice (Figure 5B). Vis–NIR absorbance spectroscopy revealed a suppression of the characteristic semiconducting (S₂) and metallic (M₁) transition bands of SWCNTs in the same interval (Figure 5C). Finally, scanning electron microscopy (SEM) provided direct observation of the morphological changes to the SWCNTs resulting from the different incubation conditions (Figure 5D). For example, after the hMPO/H₂O₂/NaCl treatment, the cylindrical structure that characterizes SWCNTs disappeared (Figure 5D, center left); small quantities of carbonaceous residue remained after incubation with hMPO/H₂O₂ (Figure 5D, center right). Together these measurements showed that hMPO with H₂O₂ catalyzed the degradation of carboxylated SWCNTs, a process further accelerated by the formation of hypochlorite.

Based on the literature, Table 2 summarizes the reported degradation of CNMs catalyzed by different enzymes. The different functional groups on the CNMs are noted. In a later section, in vitro and in vivo biodegradation of CNMs will be explored.

A few trends can be gleaned from the ex vitro data presented in Table 2. First, minimal enzyme-catalyzed oxidation of pristine (i.e. nonfunctionalized) CNTs and graphene (i.e. RGO) was demonstrated by the available characterization techniques. Alternatively, carboxylated CNTs incubated with either HRP or MPO were further oxidized. Kane and coworkers proposed a mechanism that correlated CNT oxidation to the disruption of the graphitic lattice incurred by functionalization.³⁴ Strong oxidizing acids employed to functionalize SWCNTs typically yield epoxide, peroxide, hydroxyl, carbonyl, and carboxylate functional groups.³⁴ During carboxylation, three bonds belonging to the participating carbon atoms are broken, rendering this functionality attached to the graphitic backbone by only one bond (Scheme 3). Thus, carboxylated CNTs appeared most susceptible to enzymatic oxidation, because only one bond needed to be cleaved. This assertion was further supported by the results shown in Figure 3 for MWCNTs and their proposed oxidation mechanism. After the outer layers containing functional groups (i.e. carboxylates) were enzymatically oxidized and exfoliated, the more pristine inner layers failed to undergo oxidation.

We hypothesized that the carboxylates are important for one additional reason: their negative charge in aqueous media facilitates the proper orientation/proximity of the peroxidase enzyme’s heme active site to the SWCNT substrate. We employed molecular modeling studies to characterize the role of functional groups in enzyme-substrate interactions. A calculation of the docking conformations for carboxylated SWCNTs on HRP predicted that the carboxylated ends, which are negatively charged in an aqueous suspension, would orient themselves toward a positively charged arginine residue, Arg178, near HRP’s heme site (Figure 6A).²⁶ In contrast, pristine SWCNTs, which failed undergo degradation, were predicted to orient themselves with the distal end of the enzyme, away from the active heme site. Similarly, for MPO, molecular simulations suggested two possible sites for interaction between the enzyme and SWCNTs.¹⁰ The first site, located at the proximal end of the heme group, contained catalytically active tyrosine residues, Tyr 293 and 313, and arginine residues, Arg 294, 307, and 507;¹⁰ again, the positively charged arginine residues interact strongly with the negatively charged carboxylated nanotubes (Figure 6B). The second site, at the distal end of the heme group and distant from the tyrosine residues, appeared to bind the pristine SWCNTs.

The molecular modeling of HRP was expanded from the docking of one-dimensional SWCNTs to that of two-dimensional graphene sheets.³¹ For this geometry, favorable binding poses were identified only on graphene’s basal plane (Figure 6C) because the edges of the sheet failed to provide sufficient surface area for graphene-HRP stabilization through hydrophobic interactions. Along the basal plane, the heme active site was situated approximately 8.7 Å from the surface of GO. In contrast, the separation distance for RGO was 11.5 Å. The presence of epoxide and alcohol groups on the basal plane of GO may be responsible for orienting the heme active site with enough proximity to the GO thereby enabling its oxidation catalyzed by HRP/H₂O₂.

Scheme 4 depicts the work function under vacuum (V, ~5 eV) and standard potential (E⁰, +0.5 V) of the valence band for SWCNTs.^35,36 Since HRP, MPO, and MPO’s byproduct, HOCl, are stronger oxidants (i.e. demonstrate more positive redox potentials), these species should be spontaneously reduced under standard conditions by accepting electrons from the valence band of SWCNTs and thereby oxidizing them. While thermodynamically HRP and MPO have the potential to spontaneously oxidize SWCNTs, in reality CNT oxidation/degradation may be influenced by enzyme/substrate interactions and is facilitated by defects.

It has been demonstrated that hypochlorite results in the introduction of acid “defects” sites on the surface of CNTs.³⁷ Therefore, for MPO-catalyzed biodegradation, it is possible that the initial oxidation of CNMs is predominantly due to the effects of hypochlorite (as a result of its very high oxidizing potential); the accumulation of oxidative “defects” may pave the way for subsequent more effective oxidation via both mechanisms (i.e. hypochlorite and peroxidase reactive intermediates with lower oxidizing potentials).

What is evident from these proposed mechanisms is the need for further investigations to elucidate the enzyme-catalyzed degradation process at the molecular level. By identifying the by-products of degradation, it will be possible to construct more specific reaction schemes describing the mechanism of the enzyme-catalyzed processes. While this Account has focused on the heterolytic bond cleavage of H₂O₂ catalyzed by peroxidase enzymes, an alternative mechanism involving the homolytic cleavage of H₂O₂ catalyzed by transition metals such as iron can also impact CNMs. For example, Smalley and coworkers demonstrated that a metal-catalyzed process could be employed for the purification of SWCNTs,³⁸ and we established that the radicals generated through the homolytic cleavage of H₂O₂ in the presence of iron and hemin (i.e. via the Fenton reaction) can degrade SWCNTs.²⁶

While ex vitro experiments with HRP and MPO demonstrated the ability of these enzymes to catalyze degradation of CNMs, they did not address the significance of these reactions in vitro and in vivo. Thus, the enzymatic oxidation of CNMs has been studied in living systems. For carbon nanotubes (CNTs), we employed Raman spectroscopy to monitor carboxylated SWCNTs located within neutrophils, inflammatory cells containing high concentrations of MPO. The intensity of the D band relative to the G band increased noticeably going from 2 h to 8 h of incubation within the neutrophils (Figure 7A, B). This change suggests the introduction of defects to the nanotubes via enzymatic oxidation. Neves et al. tracked the process further by measuring the Raman signal from oxidized, RNA-wrapped double-walled CNTs incubated with human prostate adenocarcinoma cells for up to 18 h.³⁹ They reported a steady increase in the D/G intensity ratio with incubation time (Figure 7C) and inferred that this progression was due to defect-inducing alterations in the outer wall of the nanotubes. Unfortunately, they did not pinpoint the actor or mechanism within the cells that altered the structure of the nanotubes’ wall. These are interesting experiments in that RNA is also negatively charged; therefore, the positively charged domains in the putative catalyst would act similarly to the peroxidases described above.

Our in vivo murine experiments afforded the most important evidence for the role of MPO in oxidative biodegradation of CNTs. In this study, the pulmonary inflammation response of mice exposed to carboxylated SWCNTs via pharyngeal aspiration was compared between wild-type (w/t) mice and MPO-knockout (k/o) mice (i.e. mice with and without MPO, respectively).¹¹ One of the primary challenges in assessing CNT biodegradation entails quantitatively evaluating the content of the CNM in vivo. An interesting opportunity is offered by the Hyperspectral Imaging methodology (CytoViva) whereby spectral signatures characteristic of CNTs are collected into a spectral library and subsequently utilized for the detection of CNT in tissue sections (Figure 8A). Furthermore, photoacoustic (PA) imaging technology⁴⁰ was utilized to assess the content of SWCNTs in sections of the lungs from w/t and MPO-k/o mice (Figure 8B). The decreased PA intensity 28 days after pharyngeal aspiration of SWCNTs was statistically significant in w/t mice but not in MPO-k/o mice (Figure 8C). Combined with the data obtained by several additional quantitative protocols (Raman microscopy, NIR spectroscopy, TEM analysis), we concluded the inflammatory response in MPO k/o mice was stronger than in w/t C57B1/6 mice, and that SWCNT oxidation and clearance from the lungs of MPO k/o animals was markedly less effective (Figure 8D).¹¹ Interestingly, a recent report documented in vivo degradation of CNT in the brain of animals.⁴¹

Biopersistence of CNMs has been long viewed as the major factor contributing to the toxic effects of these nanomaterials in the body. The discovery of the enzymatic CNT degradation processes,^{10,11,25–27,30–32} opened new opportunities for the regulation of CNT distribution and fate in vivo by controlling inflammatory response and/or employing SWCNT-metabolizing enzymes. Indeed, the number of studies relating to the field of CNMs and biodegradation has grown vigorously. For example, PubMed revealed 16 entries using the search terms “biodegrade/biodegradation” and “carbon nanotubes” before 2008; an additional 48 publications have been added since then. Although research in the field has greatly expanded, significant amount of work is still warranted. To date, most research on enzyme-catalyzed biodegradation has focused on HRP ex vitro and MPO in vitro and in vivo; it is possible that other enzymes such as eosinophil peroxidases (EPO) and lactoperoxidase (LPO) could be involved in the biodegradation process. Furthermore, other hemo-proteins with peroxidase activity merit investigation, including different forms of cytochrome P450, cytochrome c, etc. The specific roles and significance of these enzymes in biodegradation and regulation of inflammatory responses should be explored. Additionally, the assessment of the intermediates of biodegradation represents another area of research that merits investigation because the likely products—aliphatic and oxidized polyaromatic hydrocarbons—will provide details about the mechanism of CNM biodegradation and may exert their own specific, possibly toxic, effects on the body, which may also have to be taken into account.

As a result of the revolutionary research on enzyme-catalyzed biodegradation, one can envision futuristic applications in areas like drug delivery or imaging applications. For example, enzymes that result in biodegradation could be encapsulated inside nitrogen-doped nanocapsules,⁴² a CNM that could be employed to deliver drug or imaging cargo along with all the components for its self-elimination from the body. More practically, relatively non-toxic and inexpensive enzymes may be instrumental in environmental/occupational biodegradation of CNMs contamination introduced via spills, waste products in landfills, etc. To this end, white-rot basidiomycete fungi that secrete lignin peroxidase (LiP) were exposed to a dispersion of C₆₀(OH)_n and proceeded to oxidize this CNM over 32 weeks.⁴³ This result suggests the existence of a bioremediation pathway for oxidized CNMs that might otherwise accumulate in the environment. Finally, an interesting avenue of research will be applying enzymatic biodegradation to other non-carbonaceous nanomaterials to mitigate toxicity.