Impaired Clearance and Enhanced Pulmonary Inflammatory/Fibrotic Response to Carbon Nanotubes in Myeloperoxidase-Deficient Mice

Advancement of biomedical applications of carbonaceous nanomaterials is hampered by their biopersistence and pro-inflammatory action in vivo. Here, we used myeloperoxidase knockout B6.129X1-MPO (MPO k/o) mice and showed that oxidation and clearance of single walled carbon nanotubes (SWCNT) from the lungs of these animals after pharyngeal aspiration was markedly less effective whereas the inflammatory response was more robust than in wild-type C57Bl/6 mice. Our results provide direct evidence for the participation of MPO – one of the key-orchestrators of inflammatory response – in the in vivo pulmonary oxidative biodegradation of SWCNT and suggest new ways to control the biopersistence of nanomaterials through genetic or pharmacological manipulations.


Introduction
Biopersistence of carbon nanotubes (CNT) -resulting from their inherent durability [1,2] is one the major stumbling blocks on the way of their broad biomedical applications. This is because engineered CNT represent a possible health risk due to their ability to cause pulmonary inflammation, severe oxidative stress and early onset fibrosis in animals [3,4,5,6]. They also exert genotoxic effects [7] possibly associated with carcinogenesis (e.g., induction of mesotheliomas) [8,9]. These health concerns have been associated -to a large extent -with the reported long ''lifespan'' of SWCNT in the lung thus necessitating exploration of possible metabolic pathways leading to their biodegradation.
Although different types of chemical oxidative cutting of CNT -lengthwise and shortening -have been described, they require harsh oxidants (e.g., sulphuric acid plus H 2 O 2 or KMnO 4 ) [10]. Recently, we reported that reactive intermediates of horseradish peroxidase are effective in enzymatic oxidative biodegradation of CNT and graphene oxide [11,12,13]. Moreover, myeloperoxidase (MPO), an abundant enzyme of inflammatory cells (neutrophils), -involved in the principal defense mechanisms of innate immunity -was also effective in oxidative biodegradation of CNT in biochemical models and in cells yielding the products that did not cause pulmonary inflammation in mice [14]. However, the relevance of this mechanism for CNT biodegradation in vivo was lacking. Here, we employed MPO knockout B6.129X1-MPO (MPO k/o) vs wild-type C57Bl/6 mice (w/t) and demonstrated that clearance and oxidation of single wall carbon nanotubes (SWCNT) in the lungs after their pharyngeal aspiration was markedly less effective whereas inflammatory response was more robust in the former than in the latter. Our results provide direct evidence for the participation of MPO in pulmonary biodegradation of SWCNT in vivo. Based on these data, new genetic and pharmacological approaches can be developed to regulate the biopersistence of nanomaterials in tissues.

Results and Discussion
Pulmonary inflammatory responses in w/t and MPO k/o mice exposed to SWCNT Upon pharyngeal aspiration, SWCNT elicited typical inflammatory responses in w/t and MPO k/o mice documented by increased production of pro-inflammatory cytokines (TNF-a, IL-6 and MCP-1) 1 day after the exposure ( Fig. 1a, b, c). BAL cytology indicated a robust and early (day 1) accumulation of neutrophilsslightly weaker in MPO k/o mice vs w/t animals (Fig. 1d), followed by a sequential appearance of macrophages in both groups of animals (with a peak at day 7) (data not shown) [4]. At day 28 post exposure, the amounts of PMN in BAL fluid from exposed MPO k/o mice and w/t mice are not different from those in the respective control groups of animals (Fig. 1d). In BAL, phagocytized SWCNT were detected inside PMNs and macrophages (Fig. 1e). Because of the enrichment of PMNs with MPO, we compared the SWCNT content in these cells. An ,10 times greater number of PMNs from MPO k/o mice had SWCNT inclusions than those from w/t animals (Fig. 1f), in agreement with the lack of MPO-driven biodegradation in MPO k/o mice.
Assessments of fibrosis by measurements of collagen deposition on day 28 post-exposure revealed its significantly higher amounts in the lungs of MPO k/o mice compared to w/t animals (Fig. 2a). A stronger fibrogenic response in k/o vs w/t mice was also evident from quantitative morphometry of the thickness of fibrous collagen in the alveolar connective tissue (Fig. 2b). An increase of the thickness was observed already at day 1 post-exposure and progressed further by day 28 compared with air control groups. A markedly higher collagen content of the alveolar wall was detected Figure 1. Characterization of pulmonary inflammatory responses to SWCNT in w/t and MPO k/o mice at day 1 after pharyngeal aspiration exposure. a-c. Levels of pro-inflammatory cytokines (a -TNF-a; b -IL-6; c -MCP-1) in BAL fluid of w/t and MPO k/o mice. Mean 6 SEM (n = 6 mice/group). *p,0.05, vs control PBS-exposed mice. d. Content of PMNs in BAL fluid of w/t and MPO k/o mice. Mean 6 SEM (n = 6 mice/group). *p,0.05, vs control PBS-exposed mice. e. Typical microscopic images of inflammatory cells in BAL fluid with SWCNT inclusions (red arrows). f. Content of PMNs with engulfed SWCNT in BAL fluid of w/t and MPO k/o mice. Mean 6 SEM (n = 6 mice/group). *p,0.05, vs w/t mice. doi:10.1371/journal.pone.0030923.g001 in the lung of MPO-deficient mice compared to w/t animals (Fig. 2b).

Assessment of SWCNT aggregates in the lung of w/t and MPO k/o mice
To evaluate the content of SWCNT and their aggregates we employed quantitative imaging by illuminating lung sections with the light in the spectral range of 750-840 nm selectively absorbed by SWCNT. In the images presented in Fig. 3a, the lung tissue is shown in red, the SWCNT image was inverted such that the nanotubes look bright and pseudocolored in green. The volume occupied by SWCNT at day 1 post-exposure (,07-1.0% of the total lung volume) markedly decreased by day 28 (to 0.02-0.03%). Elimination of SWCNT from the lung of w/t mice was almost 2 times more effective than in MPO k/o animals (Fig. 3b). Additionally, we employed another approach for collecting SWCNT images: we scanned unstained paraffin embedded lung sections using INCell Analyser 1000 high content screening; optically detectable SWCNT aggregates were counted in each field throughout the lung sections. The content of SWCNT aggregates in each field on day 1 post-exposure was approximately the same in both groups of mice but markedly decreased by day 28 after exposure (Fig. 3c). Assuming the content of aggregates on day 1 in both groups as 100%, the relative content of SWCNT aggregates dropped to ,20% in w/t samples but remained at the level of .60% in the MPO k/o lung sections -a more than 3-fold difference.
To obtain further evidence for ''cutting'' of SWCNT by MPO, we performed TEM analysis of the lung samples after solubilization of the tissue and evaluated the size distribution of SWCNT and their aggregates (Fig. 4). On day 1 after exposure, ,35-40% of SWCNT were represented by large aggregates of .0.75 mm whereas 20-25% of aggregates were .0.5 mm and 35-40% had the size range from 0.5-0.75 mm in w/t and k/o animals. This distribution had changed markedly at day 28 after exposure in favour of small aggregates. The ''cutting'' pattern was different in w/t and k/o animals: a ,2-fold increase for aggregates ,0.5 mm, mostly at the expense of the decreased amount of large aggregates, was observed in the lungs of w/t mice whereas these changes in the SWCNT size distribution were much less pronounced in MPO k/o mice (Fig. 4b, c). This suggests that MPO was involved in the cutting of SWCNT into smaller fragments -in accord with our previous in vitro experiments [14].

Raman spectroscopic evaluation of defects in SWCNT in the lungs of w/t and MPO k/o mice
MPO-driven biodegradation occurs via oxidative modification of SWCNT with the appearance of characteristic defects detectable by Raman spectroscopy [14]. Therefore, we utilized two different Raman spectroscopic techniques to determine whether these oxidation-associated defects could be detected in the samples of the lung from exposed mice. First, we performed Raman spectroscopy after solubilization of the lung tissue. To assess oxidative biodegradation of SWCNT, the D-band/G-band ratios were calculated, whereby an increase in the ratio indicates oxidative degradation [15,16]. Typical Raman spectra documenting the signals from non-oxidized SWCNT with a pronounced G band and a very weak D-band were recorded on day 1 postexposure resulting in low D-band/G-band ratios for both w/t and MPO k/o animals (Fig. 5a). On day 28 post-exposure, the Dband/G-band ratio remained almost unchanged for the samples from k/o animals whereas an ,2-fold increase was found in the lungs of w/t mice. Next, we obtained Raman maps of ten different areas that contained SWCNT within each lung tissue section. Fig. 5b shows examples of bright-field images of samples from w/t and MPO k/o mice, respectively, at day 1 post-exposure, with a red box indicating the area where the Raman maps were acquired; (5c, 1-4) shows typical Raman maps obtained based on the clusters (5c, 5-8) per pixel. These bright-field images and Raman maps were generated for every sample on days 1 and 28 after the SWCNT exposure. SWCNT degradation increased over time. Importantly, 28 days post exposure there was a significant difference (p = 0.0277) between the D-band/G-band ratios of the SWCNT indicative of a markedly higher SWCNT degradation in w/t mice than in MPO k/o animals (Fig. 5d).
While the difference in clearance of SWCNT from the lungs of w/t and MPO k/o mice was significant, it is also obvious that additional, yet to be identified, factors contribute to biodegradation and elimination of SWCNT. Indeed, SWCNT clearance occurred, although at a lower rate, in MPO k/o animals. Contribution of other heme-peroxidases to the clearance of SWCNT is quite possible. Degradation of SWCNT by MPO during the peroxidase cycle whereby the reaction with H 2 O 2 yields a reactive intermediate, compound I, which oxidizes chloride by a single 2-electron transfer to produce the hypochlorous acid (HOCl). Both oxidants -reactive MPO intermediates and hypochlorous acid -are involved in the oxidative degradation of nanotubes [14] due to the breakage of C-C and C-H bonds [17]. Reactive intermediates can be also generated by other hemeperoxidases such as eosinophil peroxidase, lactoperoxidase, thyroid peroxidase as well as by hemoproteins with peroxidase activity (hemoglobin, myoglobin, cytochromes P450). In fact, the ability of horseradish peroxidase (HRP) and lactoperoxidase to oxidatively biodegrade SWCNT in vitro has been documented [11,12,18]. While it is possible that alternative peroxidase pathways may participate in pulmonary biodegradation of SWCNT the results of the current study -demonstrating a significantly lower clearance of SWCNT from the lungs of MPO k/o mice vs w/t animals -are supportive of the essential role of MPO in this process. While this study utilized SWCNT, it is highly likely that multi-walled carbon nanotubes, fullerenes, graphene and other carbonaceous particles may also undergo MPOcatalyzed modifications and degradation [19].
In contrast to in vitro incubation systems, non-covalent coating of SWCNT with proteins and lipids in vivo [20,21] occurring during their interactions with components of biofluids will inevitably affect recognition patterns and metabolic pathways of the particles. Our recent studies indicate that in the lung, SWCNT are effectively coated with the major surfactant lipids and proteins. Hence, the nature of the nanomaterial ''coronation'' by proteins and/or lipids may impact on the degree of recognition and biodegradation by immune-competent cells.
The major bactericidal function of MPO via oxidative attack on pathogens has been well characterized [22,23]. The ability of MPO to oxidatively modify organic xenobiotics and drugs has been also documented [24,25]. Here, we describe a hitherto unknown role of MPO in the body to oxidatively biodegrade carbonaceous particles, specifically SWCNT. One can presume that this is one of the essential and very ancient catalytic roles of the enzyme aimed at oxidative degradation and clearance of different types of carbonaceous materials to which animals and humanoids have been exposed from primordial times -such as particles generated by intentional (e.g., during cooking of food or technological high temperature processing of metals and other materials) or unintentional (forest fires, volcano eruptions) exposures. With the advent of nanotechnologies, health concerns prompted exploration of endogenous metabolic pathways competent in clearance of these notoriously biopersistent, inflammogenic and genotoxic, hence potentially high risk materials, including SWCNT. The current work demonstrates that MPO is involved in SWCNT degradation in vivo thus offering new opportunities for controlled regulation of SWNT's ''life-span'' in tissues and circulation.

Preparation and characterization of SWCNT
SWCNT (CNI Inc., Houston, TX) produced by the high pressure CO disproportionation process (HiPco) technique [26] employing CO in a continuous-flow gas phase as the carbon feedstock and Fe(CO) 5 as the iron-containing catalyst precursor, and purified by acid treatment to remove metal contaminates [27] were used in the study. Chemical analysis trace metal (iron) in SWCNT was performed at the Chemical Exposure and Monitoring Branch (DART/NIOSH, Cincinnati, OH) using nitric acid dissolution and inductively coupled plasma-atomic emission spectrometry (ICP-AES). Analysis revealed that SWCNT comprised of 0.23 weight % iron. SWCNT were routinely tested for bacterial endotoxin (LPS) contamination using the endpoint chromogenic LAL method, as previously described [28]. The mean diameter and surface area of SWCNT was 1-4 nm and 1040 m 2 /g. Surface area was determined by Brunauer, Emmett, and Teller (BET) analysis, and diameter and length was measured by TEM.
The chemical cutting of SWCNT was performed as reported previously [29]. Purified SWCNT were dispersed in 3:1 mixture of concentrated H 2 SO 4 and 30% aqueous H 2 O 2 and sonicated in ultrasonic bath (Branson 1510 SonifierH, output power of 70 W at 40 KHz) for 24 hrs at 0uC. The dispersion was then heated to 70uC for 10 min for ''polishing'' the nanotubes. This solution was then diluted 10-fold by deionized water and filtered through PTFE membrane (100 mm pore size). The collected sample was thoroughly washed with deionized water and vacuum dried at 110uC for 30 min. Obtained short SWCNT were dispersed in 25 mM HEPES buffer (pH 7.4; containing 150 mM NaCl) by sonication. For purity assessment and characterization of SWCNT we used several standard analytical techniques. TEM was employed to determine the length distribution (Fig. S1a). Raman spectroscopy was implemented to visualize the D and G bands (Fig. S1b). Diffuse reflectance infrared Fourier Transform spectroscopy (DRIFTS) was also performed (methods S1, Fig.  S1c).

Animals
Specific pathogen-free adult (8-10 week) female C57BL/6 mice (w/t) and B6.129X1-MPO mice (MPO k/o) were supplied by Jackson Lab (Bar Harbor, ME) and weighed 20.061.9 g when used. C57BL/6 mice are widely used as background for the development of transgenic/knockout models as well as for safety and efficacy testing. B6.129X1 MPO-deficient mice were created by targeted disruption of the MPO gene in RW4 embryonic stem cells. The mutant allele was transferred to the C57BL/6J background using a marker assisted screening protocol. Polymorphisms between parental strain 129/SvJ and recipient strain C57BL/6J were screened at approximately 20 cM intervals to select for the mice containing the most C57BL/6J composition. Heterozygous mice (approximately 91% C57BL/6J) were intercrossed and all experiments used 2-6-month-old female knockout, heterozygote, and wild-type littermate animals. The absence of MPO in MPO k/o mice was proved by Northern blot or Western blot analysis of bone marrow. Neutrophils and monocytes from both peripheral blood and bone marrow of MPO k/o animals failed to exhibit endogenous peroxidase activity. In contrast, eosinophils, which possess eosinophil peroxidase, demonstrated peroxidase staining [30].
Animals were housed one mouse per cage receiving HEPA filtered air in AAALAC-approved NIOSH animal facilities. All animals were acclimated in the animal facility under controlled temperature and humidity for one week prior to use.

Experimental Design
Experimental protocols for the present study included pharyngeal aspiration exposure of w/t and MPO k/o mice to 40 mg/ mouse SWCNT, while the corresponding control mice were administered sterile Ca 2+ +Mg 2+ -free phosphate-buffered saline (PBS) vehicle. Particulate instillation was performed as described in Methods S1. Mice were sacrificed on days 1 and 28 following the exposure. Inflammation was evaluated by total cell counts, cell differentials, and accumulation of cytokines in the bronchoalveolar lavage (BAL) fluid. Fibrogenic responses to exposed materials were assessed by morphometric measurements and collagen deposition.

Obtaining bronchoalveolar lavage (BAL) from mice
Mice were sacrificed with intraperitoneal injection of sodium pentobarbital (.100 mg/kg) and exsanguinated. The trachea was cannulated with a blunted 22 gauge needle, and BAL was performed using cold sterile PBS at a volume of 0.9 ml for first lavage (kept separate) and 1.0 ml for subsequent lavages. Approximately 5 ml of BAL fluid per mouse was collected in sterile centrifuge tubes. Pooled BAL cells for each individual mouse were washed in PBS by alternate centrifugation (8006g, 10 min, 4uC) and resuspened. BAL aliquots were frozen at 280uC until processed. Procedures used for BAL cell counting and cytokine analysis is described in Methods S1.

Sirius red staining
The distributions of type I and type III collagen in the lung tissue were determined by morphometric evaluation of the Sirius red-stained sections [3]. To identify collagen fibers under the microscope, deparaffinized and dehydrated lung sections were stained with F3BA/picric acid for 1-2 h, washed with 0.01N HCL for 1 min, and counterstained with Mayer's hematoxylin for 2 min [31]. Type I and III collagen stained by Sirus red was visualized, and 6 randomly selected areas were scored under polarized microscopy and average thickness of Sirius red positive connective tissues in the alveolar wall was quantitatively measured. Volume and surface density was measured using standard morphometric analyses of points and intercept counting [32]. Average thickness of the Sirius red positive connective tissues of the alveolar wall was computed from two times the ratio of volume density of points to the surface density of the alveolar wall.

Lung collagen measurements
Whole lungs from each mouse were homogenized in 0.7 ml of 0.5 M acetic acid containing pepsin (Accurate Chemical and Scientific Corporation, Westbury, NY) with 1:10 ratio of pepsin: tissue wet weight. Total lung collagen content was determined by quantifying total soluble collagen using the Sircol Collagen Assay kit (Accurate Chemical and Scientific Corporation, Westbury, NY).

Solubilization of lungs
Lungs were homogenized using Biospec Products Inc, OK, USA tissue homogenizer in 500 mL of deionized water. After homogenization, 1% SDS was added and samples were heated to 100uC for 10 min. This was followed by sonication using the ultrasonic benchtop cleaner (Branson 2510, output power of 70 W at 40 KHz) for 10 min. After cooling the contents to room temperature, Tris-HCl buffer (final concentration 30 mM) (pH 8.0) was added. Finally, an 18 hrs incubation of the material with 100 mg/mL of proteinase K at 50uC was undertaken to completely solubilize the lungs.

Raman maps
Raman maps of lung tissue sections were acquired using an NTEGRA Spectra microscopic system integrated with Raman spectrometer (NT-MDT, Russia) 473 nm cobalt laser operated at 10%, 1006 oil objective, 2 s exposure of static spectra centred at 2070 cm 21 . Ten maps (32632 points, 20 mm620 mm or 10 mm610 mm) were recorded for each tissue section. The Raman maps were then imported into Matlab 7 (The Mathworks, USA) for analysis, where all of the spectra were normalized, smoothed, baseline corrected and K-means clustering was carried out. Raman spectroscopy of solubilized lung tissue as well as transmission electron microscopy was performed as described in Methods S1.

Image analysis
Lung preparation for microscopic evaluation was performed as described in Methods S1. Two color images were collected from mounted sections of lung using a Hamamatsu Orca ER cooled CCD camera (Hamamatsu, Bridgewater, NJ) on a Nikon 90i upright microscope (Melville, NY), equipped with a high speed linear encoded xy stage (Prior Scientific, Rockland, MA), and fluorescence imaging and bright field imaging capabilities. The motorized stage coupled with Elements software (Nikon NIS, Melville, NY) allows collection of large area images which consist of multiple stitched and aligned frames. Each image was a 10610 image field at 106 magnification. To collect the image of the lung tissue the autofluorescent signal from the paraffin embedment was detected using 530 nm excitation and 560-600 nm emission. To visualize nanotubes sections were illuminated with white light and blocking filter (750-840 nm transmission) was placed between the specimen and the camera. In this conditions nanotubes selectively absorbing light in the range 750-800 nm were seen as black spots. For image processing, the area of the lung in the section profile was used to outline the lung, and the autofluorescent signal was used to threshold and generate tissue area measurements in each section. The same section profile region was overlaid on the nanotube image and a dark objects thresholded to generate an area measurement of the nanotubes within the lung. In the final images shown in Fig. 3a the lung tissue is shown in red, the tube image was inverted (such that the tubes are now bright) and pseudocolored green. The scale bar is 1 mm.

Detection of SWCNT aggregates
Unstained paraffin embedded lung tissue sections from MPO k/ o and w/t mice were scanned using the IN Cell Analyser 1000 high content screening and analysis system to distinguish between the organic tissue and the inorganic SWCNT (under bright field imaging settings). Optically detectable SWCNT aggregates were counted and their distribution and sizes were calculated using IN Cell Investigator software.

Statistics
The results are presented as means 6 S.D. values from at least three experiments, and statistical analyses were performed by oneway ANOVA. The statistical significance of differences was set at p,0.05.

Disclaimer
The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the National Institute for Occupational Safety and Health. Figure S1 Characterization of SWCNT employed in the study. a. Histogram detailing the length distribution of SWCNT. The mean length was determined to be 6766329 nm employing a sample size of 100. The insert depicts a TEM micrograph (500 nm scale bar) for the SWCNT sample. b. Raman spectrum for SWCNT; the D-and G-bands are marked on the spectrum. c. The spectrum obtained utilizing diffuse reflectance infrared Fourier Transform spectroscopy (DRIFTS). The unit for the ordinate axis is Kubelka-Munk (KM). (TIF)

Supporting Information
Methods S1 Supplemental methods.