Biocompatibility of graphene oxide intravenously administrated in mice—effects of dose, size and exposure protocols

Abstract

Graphene oxide (GO) shows great promise in in vivo drug delivery and therapy applications. However, several reports have reported an in vivo toxicity of GO. In this study, we found that the toxicity of GO intravenously injected into mice could be tuned by dose, size and exposure protocols of GO. The exposure to a single dose of 2.1 mg kg⁻¹ (single-high-dose exposure) small size GO or large size GO caused macrophage nodule formation in the lungs of the mice, and the exposure to seven repeated doses of 0.3 mg kg⁻¹ (multiple-low-dose exposure) large size GO also induced small macrophage nodule formation, serious lymphocyte infiltration around the bronchioles in the lungs of the mice, and even death of the mice. Nephritic inflammatory reactions were also observed after the multiple-low-dose exposure to large size GO. However, no obvious lung toxicity but hepatic inflammatory infiltration was observed after the exposure to multiple-low-dose small size GO. GO accumulation in the macrophage nodules was verified by Raman mapping. These findings will benefit the applications of GO in the future, especially in biomedical fields.

---

1. Introduction

The rapid development in nanotechnology has spawned numerous novel applications and products. The subsequent issues on biosafety of the new nanomaterials are attracting more and more attention.^1,2 A number of research studies have been conducted on the biological effects of nanomaterials, such as carbon nanomaterials, polymer nanoparticles and quantum dots, at both in vitro and in vivo levels.^3–5 After over a decade of exploration, researchers realized that the biological effect of nanomaterials is complex and largely affected by the exposure dose and methods as well as the nature of the materials, e.g. chemical composition, size, structure and surface properties.^5–7

Very recently, the newly developed material, graphene, and its derivatives have become a research focus due to their unique properties in electronics, chemistry, and mechanics, and due to their two-dimensional carbon structure.^8,9 Among the graphene derivatives, graphene oxide (GO) has been widely explored for in vitro and in vivo drug delivery and imaging, taking advantage of its high solubility and stability in physiological solutions, low cost and scalable production, and facile biological/chemical functionalization.^10–14

However, the biocompatibility of GO is inconclusive. Several studies supported its good biocompatibility.^15–18 Nevertheless, granuloma formation, inflammation, and thrombus formation in mice have also been observed after GO exposure.^19–22 The different experimental conditions and GO characteristics, such as exposure methods, dose, GO size and surface properties, may be the reasons for the inconsistency. Small size GO, especially after modification, was reported to be more biocompatible at a low concentration.^15–18,23 Nevertheless, the large delocalized-election system, which is always essential for their biological applications,²⁴ might be damaged by chemical modification. A high dose exposure to GO would be needed for its application, which could therefore induce toxicity. The multiple-low-dose injection is always an alternative method for drugs that are not appropriate as a single-high-dose injection. Actually, many nanotoxicity research studies have been done by the multiple-injection method but inconsistent results were obtained.^25–28 Therefore, a systematic safety evaluation of nanomaterials using both single and multiple injection methods is essential for assessing the bioapplications of GO.

Herein, we report the effects of size, dose and dosing frequency on the toxicity of GO in mice. The mice were exposed intravenously (i.v.) to GO and a saline control following Scheme 1, which shows the exposure schedule, i.e. the mice were given a single dose of 0.3 mg kg⁻¹ or 2.1 mg kg⁻¹ of both size GO, or multiple doses of 0.3 mg kg⁻¹ small or large size GO every other day for 15 days. After i.v. exposure, the toxicity of GO in the main organs, including liver, lungs, kidneys and spleen, was evaluated to provide a general toxicological profile of GO in the mice. It was found that the multiple-low-dose exposure to small size GO was safer for mice.

Scheme 1 Animal exposure schedule. ICR mice were i.v. injected with GO or 0.9% saline. For the single-dose exposure, mice were given either 0.3 mg kg⁻¹ or 2.1 mg kg⁻¹ s-GO or l-GO; for the multiple-dose exposure, mice were given seven times of 0.3 mg kg⁻¹ s-GO or l-GO every other day. Examinations were performed on day 15.

2. Materials and methods

2.1 Preparation and characterization of GOs

Both s-GO (small size GO) and l-GO (large size GO) were prepared and characterized carefully following our previous report.⁶ Briefly, a GO suspension, prepared following the modified Hummers method (see ESI†), was further heated at 120 °C for 20 min to generate small GO sheets. The obtained suspension was centrifuged at 36 000g for 50 min to separate s-GO (supernatant) and l-GO (residue). GO samples were dispersed in ultrapure water to prepare the stock suspension (1.0 mg mL⁻¹). The concentration of the GO samples was measured by drying and weighing GO in an aliquot of the suspension.

The shape and thickness of the GO sheets were characterized by atomic force microscopy (AFM; SPM-9600, Shimadzu, Japan). The particle size distribution and ζ-potential of GO in water were measured by a nanosizer (DLS, NanoZS90, Malvern, UK).

2.2 Animal exposure and sampling

All animal experiments were performed in compliance with the institutional ethics committee regulations and guidelines on animal welfare (Animal Care and Use Program Guidelines of Peking University), and approved by Peking University.

Male CD-1 (ICR) mice (∼25 g) were obtained from the Peking University Animal Center, Beijing, China. They were housed in plastic cages (five mice per cage) and kept on a 12 h light/dark cycle. Food and water were provided ad libitum. After acclimation, the mice were randomized into groups.

Mice were i.v. injected with the GO suspensions and the saline control through the tail vein following Scheme 1. The body weight and behaviour were recorded every day after the first exposure.

At day 15, the mice were sacrificed and blood/organ samples were collected for toxicological assays. Blood plasma samples were collected from blood (1.0 mL) by anti-coagulation with sodium citrate (0.1 μL 3.2% (w/v)) and centrifugation (3000g for 10 min). Liver, lungs, spleen and kidneys were collected and weighed for the calculation of organ indices (organ weight/body weight). Two pieces of each organ were cut off and fixed in a 4% formaldehyde solution. The rest was stored at −80 °C.

2.3 Determination of plasma coagulation and biochemical parameters

Following the standard procedures, activated partial thromboplastin time (APTT) and fibrinogen (Fib) were measured with an automated coagulometer (ACL 9000, Instrumentation Laboratory, Lexington, USA). Biochemical assays were performed using a Hitachi 7170A clinical automatic chemistry analyzer (Hitachi Ltd., Tokyo, Japan). Lactate total bilirubin (TBIL), alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), uric acid (UA), urea (UREA), creatinine (CRE) and dehydrogenase (LDH) were measured using commercial kits (Bühlmann Laboratories, Switzerland).

2.4 Histological observations

For histological observations, the formalin-fixed tissue samples were embedded in paraffin, thin-sectioned and mounted on glass microscope slides for hematoxylin-eosin (H&E) staining, and followed by light microscopy examination.

2.5 Apoptosis assay

Cell apoptosis of organs was evaluated by using the terminal deoxynucleotidyl transferase-mediated dUTP nick-end-labeling (TUNEL) technique on the organ sections. All the reagents used were purchased from Dingguo Biotechnology Co., Beijing, China and the instruction was followed exactly as described in our previous work.²⁹

2.6 Oxidative stress assay

For the assays to measure the reduced glutathione (GSH) level and malondialdehyde (MDA) level, each organ sample was three times minced and homogenized at 4 °C (10 s each time, intermittent for 30 s) to yield a 10% (w/v) homogenate. The homogenates were centrifuged at 3000g for 10 min to obtain the supernatants. The protein concentration of the supernatants was determined with the method of Bradford, using bovine serum albumin as the standard. The reduced GSH level of the supernatants was examined by using spectrophotometric diagnostic kits (Nanjing Jiancheng Biotechnology Institute, China). The results of GSH are expressed as mg GSH (g protein)⁻¹. The lipid peroxidation indicator, MDA, was determined by the method of thiobarbituric acid reactive species (Nanjing Jiancheng Biotechnology Institute, China). The levels of MDA are expressed as nmol MDA (mg protein)⁻¹, using 1,1,3,3-tetraethoxypropane (TEP) as the standard. The measurements of GSH and MDA were performed following the manufacturer’s instructions.

2.7 Micro-Raman mapping of GOs in lung tissues

The lung tissue slides were focused in a Raman microscope (Renishaw, UK) at ×20 magnification and excited with a 785 nm laser (100 mW). Images were obtained by scanning an area in 7 μm × 7 μm steps, collecting the Raman spectrum at each spot (2 s integration time) and plotting the integral of the area under the G-peak (around 1600 cm⁻¹, characteristic peak of GO) in the corresponding spot to form the area image. Both the H&E staining slide and the apoptosis assay slide of the lung tissues were measured.

2.8 Statistical analysis

All data are presented as the mean of more than three individual observations with the standard deviation. The significance has been calculated using the Student's t-test. A difference is considered to be significant if p < 0.05.

3. Results

3.1 GO samples

Both s-GO and l-GO used in this work are the same as those used in our previous paper.⁶ The GO sheets showed a typical G-band (1600 cm⁻¹) in the Raman spectrum and similar contents of oxygen-containing groups. Most GO sheets were single layers (the thickness is around 0.9 nm) and had a size in the range of several micrometers for l-GO (2.2 ± 1.4 μm) and 0.54 ± 0.26 μm for s-GO (Fig. S1 in ESI†). In the water suspension, the average hydrodynamic diameters were 914 nm for l-GO and 243 nm for s-GO. In a word, the two GO samples have very similar properties except for the size.

3.2 Effects of GO on mouse life expectancy, body weight and organ indices

We found that the mice can hardly live 2 weeks after one single intravenous injection of 10 mg kg⁻¹ b.w. GO. Zhang et al. also reported that mice treated with 10 mg kg⁻¹ b.w. GO for 14 days showed significant pathological changes, including granulomatous lesions, pulmonary edema, inflammatory cell infiltration and fibrosis throughout the lungs, due to the high accumulation and low clearance of GO in the mice.²² In addition, we found that the biodistribution of GO could be dramatically tuned by the concentration of GO and 2 mg kg⁻¹ b.w. is a transitional dose, where the biodistribution is significantly different from 1 mg kg⁻¹ b.w. GO.⁶ Therefore, 2.1 mg kg⁻¹ b.w. (the single dose of 2.1 mg kg⁻¹ b.w. and the seven-repeated dose of 0.3 mg kg⁻¹ b.w.) was chosen as the high dose in this study.

In the multiple-dose l-GO group, two mice were found dead on day 8 and one more was dead on day 13. The mice in this group exhibited the following clinical abnormalities: a thin appearance, their fur was upright and less movement. No abnormal clinical signs or death was seen in the other groups and all the mice were in good condition at the time of sacrifice.

The effects of GO on the body weight and the organ indices of the liver, lungs, spleen and kidneys were monitored. The body weights of GO treated mice were not significantly different from those of the control mice (Fig. 1). Although, the body weights of the multiple-dose l-GO group mice were slightly (but not significantly) lower than those of the control and multiple-dose s-GO group mice.

Fig. 1 Body weights of the mice i.v. injected with s-GO, l-GO or saline for 15 days from the first injection (n = 8, except specified differently). (A) The single dose of s-GO at 2.1 mg kg⁻¹ (s-GO 2.1) was compared to the single dose of saline (control); (B) the single dose of l-GO at 2.1 mg kg⁻¹ (l-GO 2.1) was compared to the single dose of saline (control); (C) the single dose of s-GO at 0.3 mg kg⁻¹ (s-GO 0.3) was compared to the single dose of saline (control); (D) the single dose of l-GO at 0.3 mg kg⁻¹ (l-GO 0.3) was compared to the single dose of saline (control); (E) the multiple doses of s-GO at 0.3 mg kg⁻¹ (s-GO) were compared to the multiple doses of saline (control); (F) the multiple doses of l-GO at 0.3 mg kg⁻¹ (l-GO) were compared to the multiple doses of saline (control) (^#n = 6, ^##n = 5).

The organ indices are commonly used in the toxicological evaluation to provide a general impression of toxicity. The data are summarized in Table 1. No organ index of the single-dose groups is significantly changed compared to the control group. Neither the size nor the dose alters the organ indices after the single-dose exposure. However, a size-related change in the organ index is observed in mice after the multiple-dose exposure to GO. The liver index of the group exposed to multiple doses of s-GO is significantly lower than that of the control group, which indicates organ atrophy or degenerative changes, etc., but its lung index is not markedly changed. Conversely, for the group exposed to multiple doses of l-GO, the lung index is significantly higher than that of the control group, which indicates organ congestion, edema or hypertrophy, etc., but the liver index keeps being unchanged. It has been reported that the distribution of GO in the liver and lungs was size-dependent, s-GO mainly accumulated in the liver and l-GO mainly accumulated in the lungs.⁶ Clearly, the GO size-related change in the liver and lung indices closely relates to the GO accumulation in these organs. No obvious difference is observed in the spleen and kidney indices among all the groups.

Table 1 Organ indices of the GO-exposed and control mice. Data represent mean ± S.D. (n = 5)

Animal groups		Organ indices^a (mg g^−1)
		Liver	Lungs	Spleen	Kidneys
a (Organ weight/body weight) × 1000. *Significantly different to the control group with p < 0.05.
Single-dose	Control	57 ± 2	5.2 ± 0.6	4.3 ± 1.0	16 ± 1.0
	0.3 mg kg^−1 s-GO	54 ± 3	6.5 ± 0.6	5.0 ± 1.1	16 ± 1.0
	0.3 mg kg^−1 l-GO	52 ± 6	6.6 ± 0.4	4.5 ± 0.4	16 ± 2.0
	2.1 mg kg^−1 s-GO	50 ± 4	6.5 ± 0.6	4.9 ± 0.6	16 ± 0.9
	2.1 mg kg^−1 l-GO	58 ± 4	6.9 ± 1.1	4.8 ± 0.7	16 ± 0.5
Multiple-dose	Control	59 ± 2	6.9 ± 0.5	5.6 ± 0.7	16 ± 0.8
	0.3 mg kg^−1 s-GO	54 ± 3*	6.8 ± 0.7	6.1 ± 0.9	15 ± 1.0
	0.3 mg kg^−1 l-GO	60 ± 5	8.0 ± 0.3*	7.2 ± 2.0	15 ± 1.0

---

3.3 Effects of GO on the plasma coagulation and biochemical parameters of the mice

It has been reported that GO induced thrombi after i.v. exposure.^20,21 Therefore, the plasma coagulation parameters were assayed to evaluate the toxicity of GO in blood. Table 2 shows the typical coagulation parameters, Fib (fibrinogen) and APTT (activated partial thromboplastin time). However, the levels of Fib and APTT keep being normal, regardless of the GO samples and the dosing frequency. The finding is in accordance with the conclusion of Sasidharan et al. that graphene was non-thrombogenic by testing the possibility of graphene interference with the prothrombin time (PT) and the activated partial thromboplastin time ratio (APTTr).¹⁷

Table 2 Plasma coagulation parameters of the GO-exposed and control mice. Data represent mean ± S.D. (n = 5)^a

Animal groups		Fib (g L^−1)	APTT (s)
a Fib: fibrinogen; APTT: activated partial thromboplastin time.
Single-dose	Control	3.09 ± 0.30	21.8 ± 2.1
	0.3 mg kg^−1 s-GO	2.82 ± 0.43	24.8 ± 3.5
	0.3 mg kg^−1 l-GO	3.12 ± 0.74	23.3 ± 2.6
	2.1 mg kg^−1 s-GO	2.89 ± 0.18	22.0 ± 2.1
	2.1 mg kg^−1 l-GO	3.10 ± 0.32	21.3 ± 0.8
Multiple-dose	Control	3.26 ± 0.30	23.7 ± 1.5
	0.3 mg kg^−1 s-GO	2.88 ± 0.19	20.7 ± 1.9
	0.3 mg kg^−1 l-GO	2.94 ± 0.22	21.7 ± 2.5

---

Next, the plasma biochemical parameters were measured and the results are summarized in Table 3. After the GO injection, the levels of the biochemical parameters, including lactate total bilirubin (TBIL), alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), uric acid (UA), urea (UREA), creatinine (CRE) and dehydrogenase (LDH), are similar among the control group and the GO-treated groups, except that a significant increase in the UREA, CRE and LDH levels is observed in the multiple-dose l-GO group. The increases in UREA and CRE, which are important indicators of nephritic injury, infer a possible kidney injury induced by the multiple-dose exposure to l-GO. As for the cytoplasmic enzyme LDH, it is an indicator of alveolar macrophage injury in the pulmonary toxicity study, and also a general indicator of hepatic and nephritic injuries. The high level of the LDH activity manifests that the organ injury is caused by the l-GO exposure.

Table 3 Plasma biochemical parameters of the GO-exposed and control mice. Data represent mean ± S.D. (n = 5)^a

Animal groups		TBIL (μmol L^−1)	ALT (IU L^−1)	AST (IU L^−1)	ALP (IU L^−1)	UA (μmol L^−1)	UREA (μmol L^−1)	CRE (μmol L^−1)	LDH (IU L^−1)
a TBIL: total bilruin; ALT: alanine aminotransferase; AST: aspartate aminotransferase; ALP: alkaline phosphatase; UA: uric acid; CRE: creatinine; LDH: lactate dehydrogenase. *Significantly different to the control group with p < 0.05.
Single-dose	Control	6.4 ± 0.3	40.2 ± 12.1	64.5 ± 10.6	75.2 ± 11.1	34.4 ± 11.7	7.5 ± 1.2	15.2 ± 3.5	250.6 ± 76.3
	0.3 mg kg^−1 s-GO	6.2 ± 0.2	32.4 ± 9.2	58.7 ± 13.4	96.5 ± 20.4	26.8 ± 9.7	7.8 ± 1.5	15.8 ± 2.6	229.0 ± 57.2
	0.3 mg kg^−1 l-GO	6.2 ± 0.2	33.1 ± 3.8	60.9 ± 3.0	74.6 ± 14.7	29.4 ± 9.4	7.4 ± 0.6	15.6 ± 4.0	284.2 ± 71.1
	2.1 mg kg^−1 s-GO	6.3 ± 0.2	31.7 ± 5.7	53.0 ± 4.6	77.6 ± 10.4	38.0 ± 14.0	8.7 ± 1.0	20.2 ± 5.0	237.0 ± 25.2
	2.1 mg kg^−1 l-GO	6.3 ± 0.3	35.4 ± 7.4	56.5 ± 5.9	86.8 ± 13.1	52.2 ± 38.3	8.1 ± 1.1	16.8 ± 4.2	263.0 ± 72.9
Multiple-dose	Control	6.3 ± 0.3	37.4 ± 12.5	65.0 ± 17.9	73.2 ± 8.6	47.6 ± 28.5	6.6 ± 0.8	15.6 ± 1.1	201.6 ± 51.2
	0.3 mg kg^−1 s-GO	6.3 ± 0.2	31.2 ± 8.7	63.0 ± 5.9	74.5 ± 8.2	42.8 ± 11.2	7.1 ± 2.0	15.5 ± 3.4	237.3 ± 28.5
	0.3 mg kg^−1 l-GO	6.1 ± 0.2	40.8 ± 5.2	65.1 ± 3.8	86.3 ± 11.6	34.3 ± 10.3	9.1 ± 0.7*	18.3 ± 1.7*	274.0 ± 20.2*

---

3.4 Histopathological observations

The histology photographs of the lung, liver and kidney tissues are shown in Fig. 2–5. In the lungs, many GO aggregates are observed in macrophage nodules after the mice were treated with a single dose of 2.1 mg kg⁻¹ s-GO or l-GO, but no other lung damage is observed. No macrophage nodules or other lung damages are observed in the mice after exposed to the multiple-dose s-GO. However, s-GO enriched macrophage nodules as well as serious inflammation infiltrations are observed in the mice exposed to the multiple doses of l-GO. The alveolar walls thicken and swell, and the alveolar cavity shrinks (Fig. 2). In addition, the lymphocytes seriously infiltrate the smooth muscle layer of the bronchioles and vessels after the multiple-dose exposure to l-GO.

Fig. 2 Representative histopathological changes of the lungs of the GO-exposed and control mice in the H&E section. The doses of s-GO and l-GO are 2.1 mg kg⁻¹ for the single-dose exposure and 0.3 mg kg⁻¹ (every other day, in total seven injections) for the multiple-dose exposure. The arrows indicate the lung macrophage nodules full of GO; the circles indicate the lymphocyte infiltration around the bronchioles. The scale bar represents 100 μm.

Fig. 3 Apoptosis analysis of the lungs of the GO-exposed and control mice by the TUNEL method. The doses of s-GO and l-GO are 2.1 mg kg⁻¹ for the single-dose exposure and 0.3 mg kg⁻¹ (every other day, in total seven injections) for the multiple-dose exposure. The black solid arrows indicate the lung macrophage nodules full of GO; the circles indicate the lymphocyte infiltration around the bronchioles. The scale bar represents 100 μm.

Fig. 4 Representative histopathological changes of the liver of the control and GO-exposed mice in the H&E section. The doses of s-GO and l-GO are 2.1 mg kg⁻¹ for the single-dose exposure and 0.3 mg kg⁻¹ (every other day, in total seven injections) for the multiple-dose exposure. The arrow indicates small focal-like inflammatory cells accumulating around the central vein. The scale bar represents 100 μm.

Fig. 5 Representative histological changes of the kidneys of the control and GO-treated mice in the H&E section. The doses of s-GO and l-GO are 2.1 mg kg⁻¹ for the single-dose exposure and 0.3 mg kg⁻¹ (every other day, in total seven injections) for the multiple-dose exposure. The arrows indicate the renal glomerulus and renal glomerular capsule interspace. In the multiple-dose l-GO group, the glomerular capsule interspaces disappeared (hollow arrows), which might result from swollen glomerulus cells. The scale bar represents 100 μm.

However, no thrombus is observed, which is in accordance with the results of the plasma coagulation parameters (Table 2). The apoptosis of the cells in the lung sections was tested by the TUNEL method. Similar to the histopathological observations, black and brown lung macrophage nodules are full of GO, as well as lymphocyte infiltrations around the bronchioles are observed in the GO-exposed groups (Fig. 3). However, the apoptosis levels are similar among all the exposure and control groups.

No obvious hepatic damage is found after the single-dose exposure to GO (Fig. 4). But after the multiple-dose exposure to s-GO, there are some small focal-like inflammatory cells that accumulate around the central veins of the liver. This may be ascribed to the overload of particles in the liver after the low-dose exposure to s-GO.

The histopathological changes of the kidneys in the mice are shown in Fig. 5. A serious swelling in the renal glomerulus and a close capsular space are found only after the multiple-dose exposure to l-GO. The remarkable renal tubule injury is in accordance with the plasma biochemical assay (Table 3). No significant change is observed in the other groups compared to the control mice. As for the spleen, no obvious damage was induced in the GO-exposed mice (Fig. S2†).

3.5 Oxidative stress

The oxidative stress caused by the GO samples in the main organs was measured to reveal the possible toxicological pathway. However, as shown in Fig. 6, the GSH level and MDA level in the liver, spleen and lungs remain unchanged in all the groups. Therefore, there was no observed oxidative damage to these organs.

Fig. 6 The oxidative stress in the control mice and the GO-exposed mice. (A) The GSH level in the main organs for the multiple-dose groups; (B) the GSH level in the main organs for the single-dose groups; (C) the MDA level in the main organs for the multiple-dose groups; (D) the MDA level in the main organs for the single-dose groups.

3.6 GO in the lung tissues

The Raman G-peak signal of GO is at around 1600 cm⁻¹, which is characteristic of graphite carbon. Under the Raman microscope, paraffin-embedded mouse lung sections show focal increases in the G-peak signal in the macrophage nodules both in the H&E staining section and the TUNEL section, indicating enriched GO at the grey ranges (macrophage nodules) (Fig. 7), while no G-band signal was observed in any of the control tissue samples (data not shown). This evidences that GO exists in the lung macrophage nodules after the GO exposure. The accumulation of GO in the lungs was confirmed by the spectrometric method (Fig. S3†).

Fig. 7 Micro-Raman mapping of the lung section of the mice after exposure to a single dose (2.1 mg kg⁻¹) of l-GO. The images were obtained in 7 μm × 7 μm steps. (A) The H&E section of the lung. (B) The Raman mapping of the same area as shown in A. (C) The lung section stained by the TUNEL method. (D) The Raman mapping of the same area as shown in C.

4. Discussion

GO exhibits high solubility and stability in a physiological solution and has been used in drug delivery, bioimaging, etc.^10,30 However, the in vivo behavior of graphene-based nanomaterials still remains largely unknown. Herein, the size and dose effects on the toxicity of GO were evaluated in an animal model. The biological consequences of the GO exposure under different conditions are summarized in Table 4.

Table 4 Biological consequences of the GO exposure under different conditions^a

Animal groups		Organ indices		Lymphocyte infiltration		Lung microphage nodule		Renal glomerulus swelling	Plasma biochemical parameters
		Liver	Lungs	Liver	Lungs	Number	Size
a + Significant toxicity observed. ++ Severe toxicity observed.
Single-dose	0.3 mg kg^−1 s-GO
	0.3 mg kg^−1 l-GO
	2.1 mg kg^−1 s-GO					+	++
Multiple-dose	2.1 mg kg^−1 l-GO					+	++
	0.3 mg kg^−1 s-GO	+		+
	0.3 mg kg^−1 l-GO		+		+	++	+	+	+

---

The pulmonary toxicity is a focal point of the toxicity of GO. In fact, GO has shown obvious pulmonary toxicity with different exposure methods.^19,31 GO caused an acute and sustained inflammatory response in the lungs and pleural space by pharyngeal aspiration or direct intrapleural injection.³¹ After exposure by intratracheal injection, lung macrophages with a homogeneous black cytoplasm throughout the lungs were observed, and the GO aggregates induced peribronchial inflammation, alveolar exudates and mild fibrosis in mice.¹⁹ In addition, after exposure by a single i.v. injection, inflammatory cell infiltration, fibrosis and lung nodule formation in lungs could also be easily observed,^22,32 which is in accordance with our observation. In fact, the formation of macrophage nodules in lungs was generally observed in animals exposed to other carbon nanomaterials, such as carbon nanotubes (CNTs). The intratracheal instillation of CNTs induced the formation of granuloma in lungs.^33,34 Similar macrophage nodules could also be seen after a single i.v. injection of CNTs.³²

We found that the grey lung macrophage nodules were full of GO by the Raman mapping technique (Fig. 7), which has been widely used to observe single-walled CNTs (SWCNTs) in vivo, taking advantage of the evident SWCNT G-band Raman signal at ∼1580 cm⁻¹.²⁹ Although the G-band intensity shown in GO is orders of magnitudes lower compared to that of SWCNTs,^35–37 the Raman spectroscopic method has been successfully used to qualitatively track GO in mouse lungs and liver by measuring their homogenates in our group.⁶ Here, for the first time we show the GO distribution in tissue directly and associate the accumulation of GO with pathological changes. The Raman imaging revealed GO clearly without any interference. For example, the low contrast of GO and the grey color of the stained inflammatory cells under an optical microscope may hide the true distribution of GO in tissues. We found that the macrophage nodules were full of GO, but GO did not accumulate in lymphocytes but around the bronchioles (data not shown). It suggested that the lymphocytes recruited by the GO-induced injury were not able to trap GO, though these cells have a vigorous phagotrophic ability.

According to our results, the oxidative stress is not the dominant toxicological mechanism of the i.v. exposed GO. Although oxidative stress is a broadly existent phenomenon when cells are exposed to GO,²³ the protective effect of existing proteins should be noticed. When GO is incubated with serum, due to the high protein adsorption ability of GO, the interaction between GO sheets and proteins and thus the cytotoxicity of GO are largely attenuated.³⁸ In such a case, the protein adsorption on GO might protect the organs against the oxidative damages.

No significant hepatic index, pathological change and oxidative stress were observed in mice after exposure to GO, except a slight inflammatory response after the multiple-low-dose exposure to s-GO. This was consistent with the previous reports that GO could induce a slight hepatic toxicity.^12,32 In fact, neither l-GO nor high concentration s-GO distributed considerably in liver.⁶

The low hepatic toxicity of GO was different from our previous report on pristine SWCNTs.²⁹ SWCNTs increased the levels of serum biochemical parameters indicating a hepatic injury, including ALT and AST, after a single-dose i.v. exposure. But the low hepatic toxicity of GO observed was similar to that of the functionalized SWCNTs with a higher hydrophilicity. For example, no sign of liver injury was shown in mice at 28 days after a single-dose i.v. exposure to taurine functionalized multi-walled CNTs (MWCNTs), even though 78% of the MWCNTs were found to be accumulated in the liver.³⁹ In a word, the good hydrophilicity is an important factor for the biocompatibility of GO.

The excretion of l-GO through the kidneys might be one reason for the renal injury. There are also some papers reporting the excretion of GO from the body. Zhang et al. found that the clearance of GO from the kidneys was size-dependent. Large GO particles were intercepted and then highly accumulated in the lungs, while the small size GO was quickly eliminated through the renal route.²² In this study, for the multiple-dose exposure to GO, the renal damage of the mice was GO size-dependent too. s-GO neither accumulated in the kidneys nor induced renal damage.

Similar results were reported by Yang et al., who observed the clearance of small size PEG functionalized GO (10–30 nm) without an obvious kidney damage in mice after i.v. exposure.³⁶ For l-GO in this paper, the renal damage might be attributed to the failed clearance of the large GO sheets.

“The dose makes the poison” is an everlasting truth. Previous studies confirm that the exposure dose is a key factor that affects the toxicity of GO too. Wang et al. found that low and middle-dose GO (0.1 mg and 0.25 mg per mouse, respectively) did not cause obvious toxicity in mice, while high-dose GO (0.4 mg per mouse) induced chronic toxicity, such as lung nodule formation and even death.³² Zhang et al. compared the toxicity of GO in mice at 1.0 mg kg⁻¹ and 10.0 mg kg⁻¹ at 14 days after a single-dose i.v. injection and found that GO was biocompatible in most tissues, including liver, spleen and kidneys, but induced lung pathological changes at the higher dose.²²

In this work, we also observed a dose effect, by comparing the low and high single-dose exposed groups. At 15 days after the single-low-dose exposure, neither l-GO nor s-GO induced any change in the organ indices or plasma biochemical parameters. But after the single-high-dose exposure to s-GO or l-GO, toxicities, such as organ index changes and lung nodule formation, were observed.

Given the same total exposure dose (2.1 mg kg⁻¹), the toxicity can also be modulated by changing the exposure frequency. We observed macrophage nodules induced by GO in the mouse lungs after the single-high-dose exposure to s-GO (2.1 mg kg⁻¹), however, no such nodules were found after the multiple-low-dose exposure to s-GO (0.3 mg kg⁻¹, 7 times), though the total exposure doses were identical. Our previous research has found that the dose could regulate the distribution of GO in mice.⁶ After the single-dose exposure to GO, the accumulation of s-GO in the lungs increased with an increasing GO dose, because GO at a higher concentration readily interacted with proteins, forming larger GO–protein complexes which might be retained in the lungs. Whereas, the majority of s-GO could pass through the lung capillary after each low-dose exposure, the final accumulation of s-GO in the lungs after the multiple-low-dose exposure was lower than that after the single-high-dose exposure.

The toxicity of l-GO was modulated by changing the exposure frequency as well. Unlike s-GO, the multiple-low-dose exposure to l-GO led to more accumulation of l-GO and hence a more severe toxicity, even death, compared to the single-high-dose exposure to l-GO. One possible mechanism is that l-GO formed larger numbers of smaller GO–protein complexes at a lower concentration.⁶ These smaller complexes can enter the capillary, create more injury points and hence more inflammatory cells.

The size of the GO sheets affected their toxic property. The multiple-dose exposure to l-GO induced a serious lymphocyte infiltration around the bronchioles in the lungs, as well as obvious renal damage. The multiple-dose exposure to s-GO didn't significantly induce lung or kidney damages, but induced a liver index increase and an inflammatory cell infiltration. The size related different distribution behaviours of s-GO and l-GO were clearly demonstrated: s-GO mainly distributed in the liver, whereas l-GO mainly in the lungs after the low concentration exposures. However, the size effect, on both the biodistribution and toxicity effects of GO, would be hidden by the formation of the large protein-complex at the high concentration.

5. Conclusions

In this work, the toxicity of s-GO and l-GO after different i.v. exposure protocols was examined. The toxicity of GO i.v. exposed to mice was tuned by the dose, size and exposure of GO. The single-high-dose exposure (2.1 mg kg⁻¹) to s-GO or l-GO caused macrophage nodule formation in the lungs and the multiple-low-dose exposure (seven repeated doses of 0.3 mg kg⁻¹) to l-GO also induced small macrophage nodule formation as well as a serious lymphocyte infiltration around the bronchioles in the lungs and the death of mice. The size is another key factor influencing the toxicity of GO. The lower toxicity of the multiple-low-dose exposure to s-GO than that of the multiple-low-dose exposure to l-GO implicates that GO with a smaller size could be more benign to mice. For biomedical applications in the future, the size and exposure protocols (including dose and dosing frequency) of GO should be optimized and strictly controlled.

Notes and references

1. L. H. Reddy, J. L. Arias, J. Nicolas and P. Couvreur, Chem. Rev., 2012, 112, 5818.
2. National Research Council. Committee to Develop a Research Strategy for Environmental, Health, and Safety Aspects of Engineered Nanomaterials, The National Academies Press, Washington, D.C., 2012, pp. 214–219.
3. K. Yang, L. Feng, X. Shi and Z. Liu, Chem. Soc. Rev., 2013, 42, 530.
4. K. T. Yong, W. C. Law, R. Hu, L. Ye, L. W. Liu, M. T. Swihart and P. N. Prasad, Chem. Soc. Rev., 2013, 42, 1236.
5. L. M. Kaminskas, B. J. Boyd and C. J. H. Porter, Nanomedicine, 2011, 6, 1063.
6. J.-H. Liu, S.-T. Yang, H. Wang, Y. Chang, A. Cao and Y. Liu, Nanomedicine, 2012, 7, 1801.
7. Arnida, M. M. Janat-Amsbury, A. Ray, C. M. Peterson and H. Ghandehari, Eur. J. Pharm. Biopharm., 2011, 77, 417.
8. D. Chen, H. Feng and J. Li, Chem. Rev., 2012, 112, 6027.
9. K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva and A. A. Firsov, Science, 2004, 306, 666.
10. L. M. Zhang, J. G. Xia, Q. H. Zhao, L. W. Liu and Z. J. Zhang, Small, 2010, 6, 537.
11. W. Zhang, Z. Y. Guo, D. Q. Huang, Z. M. Liu, X. Guo and H. Q. Zhong, Biomaterials, 2011, 32, 8555.
12. Y. Wang, Z. Liu, J. Wang, J. Liu and Y. Lin, Trends Biotechnol., 2011, 29, 205.
13. Y. Z. Pan, N. G. Sahoo and L. Li, Expert Opin. Drug Deliv., 2012, 9, 1365.
14. H. Hong, K. Yang, Y. Zhang, J. W. Engle, L. Feng, Y. Yang, T. R. Nayak, S. Goel, J. Bean, C. P. Theur, T. E. Barhart, Z. Liu and W. Cai, ACS Nano, 2012, 6, 2361.
15. K.-H. Liao, Y.-S. Lin, C. W. Macosko and C. L. Haynes, ACS Appl. Mater. Interfaces, 2011, 3, 2607.
16. W. Paul and C. P. Sharma, Trends Biomater. Artif. Organs, 2011, 25, 91.
17. A. A. Sasidharan, L. S. Panchakarla, A. R. Sadanandan, A. Ashokan, P. Chandran, C. M. Girish, D. Menon, S. V. Nair, C. N. R. Rao and M. Koyakutty, Small, 2012, 8, 1251.
18. K. Yang, J. M. Wan, S. Zhang, B. Tian and Y. J. Zhang, Biomaterials, 2012, 33, 2206.
19. M. C. Duch, G. R. S. Budinger, Y. T. Liang, S. Soberanes, D. Urich, S. E. Chiarella, L. A. Campochiaro, A. Gonzalez, N. S. Chandel, M. C. Hersam and G. M. Mutlu, Nano Lett., 2011, 11, 5201.
20. S. K. Singh, M. K. Singh, M. K. Nayak, S. Kumari, S. Shrivastava, J. J. A. Gracio and D. Dash, ACS Nano, 2011, 5, 4987.
21. S. K. Singh, M. K. Singh, P. P. Kulkarni, V. K. Sonkar, J. J. A. Gracio and D. Dash, ACS Nano, 2012, 6, 2731.
22. X. Y. Zhang, J. L. Yin, C. Peng, W. Q. Hu, Z. Y. Zhu, W. X. Li, C. H. Fan and Q. Huang, Carbon, 2011, 49, 986.
23. Y. L. Chang, S.-T. Yang, J.-H. Liu, E. Dong, Y. W. Wang, A. Cao and Y. Liu, Toxicol. Lett., 2011, 200, 201.
24. Y. He, Y. Lin, H. W. Tang and D. W. Pang, Nanoscale, 2012, 4, 2054.
25. C. Lasagna-Reeves, D. Gonzalez-Romero, M. A. Barria, I. Olmedo, A. Clos, V. M. S. Ramanujam, A. Urayama, L. Vergara, M. J. Kogan and C. Soto, Biochem. Biophys. Res. Commun., 2010, 393, 649.
26. K. Inoue, R. Yanagisawa, E. Koike, M. Nishikawa and H. Takano, Free Radical Biol. Med., 2010, 48, 924.
27. A. M. Pinto, I. C. Goncalves and F. D. Magalhaes, Colloids Surf., B, 2013, 111, 188.
28. A. M. Jastrzebska, P. Kurtycz and A. R. Olszyna, J. Nanopart. Res., 2012, 14, 1320.
29. S.-T. Yang, X. Wang, G. Jia, Y. Gu, T. Wang, H. Nie, C. Ge, H. Wang and Y. Liu, Toxicol. Lett., 2008, 181, 182.
30. L. Zhou, W. Wang, J. Tang, J. H. Zhou, H. J. Jiang and J. Shen, Chem. – Eur. J., 2011, 17, 12084.
31. A. Schinwald, F. A. Murphy, A. Jones, W. MacNee and K. Donaldson, ACS Nano, 2012, 6, 736.
32. K. Wang, J. Ruan, H. Song, J. Zhang, Y. Wo, S. Guo and D. Cui, Nanoscale Res. Lett., 2011, 6, 1.
33. D. B. Warheit, B. R. Laurence, K. L. Reed, D. H. Roach, G. A. M. Reynolds and T. R. Webb, Toxicol. Sci., 2004, 77, 117.
34. C. W. Lam, J. T. James, R. McCluskey and R. L. Hunter, Toxicol. Sci., 2004, 88, 126.
35. Z. Liu, C. DAvis, W. Cai, L. He, X. Chen and H. Dai, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 1410.
36. K. Yang, J. Wan, S. Zhang, Y. Zhang, S.-T. Lee and Z. Liu, ACS Nano, 2011, 5, 516.
37. K. N. Kudin, B. Ozbas, H. C. Schniepp, R. K. Prud'homme, I. A. Aksay and R. Car, Nano Lett., 2008, 8, 36.
38. W. Hu, C. Peng, M. Lv, X. Li, Y. Zhang, N. Chen, C. Fan and Q. Huang, ACS Nano, 2011, 5, 3693.
39. X. Y. Deng, S.-T. Yang, H. Y. Nie, H. Wang and Y. Liu, Nanotechnology, 2008, 19, 075101.

---

Footnote

† Electronic supplementary information (ESI) available: Preparation of graphene oxide and the histological changes of the spleen after GO exposure. See DOI: 10.1039/c4tx00044g

---