An ecotoxicological characterization of nanocrystalline cellulose (NCC)

Nanotoxicology 4 (2010), pp. 255-270

An Ecotoxicological Characterization of Nano Crystalline Cellulose (NCC)

TIBOR KOVACS1, VALERIE NAISH1, BRIAN O’CONNOR1

CHRISTIAN BLAISE2, FRANCOIS GAGNÉ2, LAUREN HALL3, VANCE TRUDEAU3

AND PIERRE MARTEL1

1FPInnovations – Paprican Division, 570 boul. St.-Jean, Pointe-Claire, QC, Canada H9R [email protected], tel: 514-630-4101, ext 2363, fax: [email protected], tel: 514-630-4101, ext 2355, fax: [email protected], tel: 514-630-4121, fax: [email protected], tel: 514-630-4101, ext 2352, fax: 514-630-4134

2Environment Canada, Fluvial Ecosystem Research, Aquatic Ecosystem Protection Division, 105 McGill Street, Montreal, QC, Canada H2Y [email protected], tel: 514-496-7094, fax: [email protected], tel: 514- 496-7105, fax: 514-496-7143

3University of Ottawa, Centre for Advanced Research in Environmental Genomics, Department of Biology, 30 Marie-Curie, Room 370, Gendron Building, Ottawa, ON, Canada K1N [email protected], tel: 613-562-5800, ext 6165, fax: [email protected], tel: 613-562-5800, ext 6015, fax: 613-562-5486

Correspondence: Tibor Kovacs, Email: [email protected]

1

Abstract

The pulp and paper industry in Canada is developing technology for the production and use of nanocrystalline cellulose (NCC). A key component of the developmental work is an assessment of potential environmental risks. Towards this goal, NCC samples as well as carboxyl methyl cellulose (CMC), a surrogate of the parent cellulosic material, were subjected to an ecotoxicological evaluation. This involved toxicity tests with rainbow trout hepatocytes and nine aquatic species. The hepatocytes were most sensitive (EC20s between 10 and 200 mg/L) to NCC, although neither NCC nor CMC caused genotoxicity. In tests with the nine species, NCC affected (IC25) the reproduction of the fathead minnow at 0.29 g/L, but no other effects on endpoints such as survival and growth occurred in the other species at concentrations below 1 g/L, which was comparable to CMC. Based on this ecotoxicological characterization, NCC was found to have low toxicity potential and environmental risk.

Keywords: Cellulose, crystallinity, fishes, nanocrystalline materials, toxicity, yield

2

1. Introduction

There are numerous economic factors challenging the viability of the pulp and paper industry in Canada. Because of this, the industry requires the development of transformative technologies to solidify its continued existence and contributions to society. Currently, work on transformative technologies is on-going in three areas: i) biorefinery (the extraction of chemicals and energy from the wood source other than pulp/paper), ii) nanotechnology and iii) novel paper products that take advantage of the unique nature of the Canadian fiber supply. Central to this initiative is that the new technologies be developed in an environmentally sustainable manner.

Of the three transformative technologies, the greatest environmental concern relates to nanotechnology. Nanomaterials have unique properties often differing significantly from their parent compound and may also have unique environmental effects and risks (Klaine et al. 2008; U.S. EPA 2007; Nowack and Bucheli 2007; Biswas and Chang-Yu 2005). In the case of the pulp/paper industry, a likely application of nanotechnology will involve cellulose-based nanomaterials (Klemm et al. 2006; Hubbe et al. 2008). Cellulose is an integral part of a tree’s composition and pulp/paper is essentially made from cellulose fibres (Smook 2002). In turn, the nano-scale building blocks of cellulose can be released through hydrolysis of the cellulose with sulfuric acid resulting in nanocrystalline cellulose (NCC). The NCC has distinctive properties and exceptional strength characteristics that can have many applications within and outside the pulp/paper industry (Klemm et al. 2006; Hubbe et al. 2008; Revol et al. 1998). Even though the NCC manufacturing technology is in the developmental stage, environmental assessments were undertaken simultaneously to help guide the technology in an environmentally compatible manner.

A thorough environmental and risk assessment of any material requires the study of its fate, exposure and toxicity and such information about NCC is entirely lacking. Fate and exposure of a product still in the developmental stage, such as NCC, are very difficult to predict and assess. That is why the decision was made to start the environmental risk assessment with toxicological testing. In this task, a key requirement was to choose the most relevant toxicity testing framework and methods. Current plans are for the production of NCC from kraft pulp at existing mills. The NCC will be produced as a (2 to 5%) suspension followed by spray drying in order to prepare a transportable solid product targeted for high value applications. Although steps would be taken to minimize any release of this high-value material, the consequences of accidental losses need to be considered. At the manufacturing plant, air dispersion of NCC would be restricted to the immediate vicinity of the work area and environmental contamination outside the mill will be unlikely. Similarly, significant soil contamination is improbable. The most probable release of NCC on a broader scale would be an accidental spill that could end up in the receiving waters via the mill’s wastewater stream. Consequently, the objective of this work was to undertake a toxicological assessment of NCC with aquatic organisms and thereby begin the risk assessment process concerning NCC manufacturing.

2. Material and Methods

2.1 Test Materials

2.1.1 Pilot plant production of nano crystalline cellulose (NCC)

The FPInnovations–Paprican pilot plant produced about 1 kg of NCC in a batch process from bleached kraft pulp (~ 4 kg). The pulp was milled to obtain a uniform size of 1 mm. This was followed by hydrolysis with the addition of sulphuric acid under specific time, concentration and temperature conditions. Once hydrolysis was complete, the reaction was quenched with deionized water. The next step was the acid removal phase where the NCC suspension was concentrated to 2–5% by weight followed by homogenization to obtain uniform particle size (70–90 nm). The colloidal suspension was passed through

3

an ultraviolet sterilization unit and then filtered through a 500 nm filter. This suspension of NCC had a pH of 2.5 to 2.7 and was used for toxicity testing.

For most toxicity tests a stock solution was prepared from the original NCC suspension by dilution with well water (pH: 7.8, hardness: 240 mg/L as CaCO3, alkalinity: 170 mg/L as CaCO3) to a concentration of 0.68% NCC. The pH of the stock solution ranged from 6.0 to 6.5. A few tests were done (see below) with the NCC suspension as is, that is no dilution with well water and these were called tests without pH adjustment. In contrast to certain nanoparticles that are difficult to suspend in water and can form aggregates on storage (Klaine et al. 2008), NCC is readily suspended in water and it will stay suspended on storage (see Figure 1).

(Place Figure 1 here)

The toxicity tests were performed on 12 batches of NCC (Batches 1 to 12). Typically, each batch of NCC produced in the pilot plant was placed in several containers. For batches 2 and 5, the toxicity tests were performed with subsamples of the original consisting of Batches 2a, 2b, and 2c and Batches 5a and 5b, respectively. For all other batches, the toxicity tests were done with one subsample. Hence, toxicity tests were done on 12 NCC batches and 15 NCC samples.

2.2 NCC Characterization

Each batch of NCC was characterized by:• pH• % dry weight of 2 to 5%. For this, a 10 mL aliquot of each sample was dried at 100°C for 3 hours

(APHA,AWWA,WPCF 1988). • Conductivity (µS/cm) with an YSI conductivity meter (APHA,AWWA,WPCF 1988).• Relative particle size as determined by photon correlation spectroscopy (PCS) (ISO 13321 1996). In

solutions containing colloidal material, the particles remain in suspension due to Brownian movement. With PCS, a laser bounces light off the particles in solution. The speed of the particle movement is determined and correlated with particle size since smaller particles will move faster than larger ones. All samples were pre-filtered through a 0.7 µm filter to remove larger particles prior to PCS.

• Polydispersity index (PI) is reported along with the particle size. The PI is basically the standard deviation for the mean particle size analysis (Malvern Instruments Ltd.). It is reported on a scale of 0 to 1 with 1 representing a sample with a variety of particle sizes.

• Total sulphur concentration was determined by ICP (APHA,AWWA,WPCF 1988) This number was used to calculate the charge density of each NCC batch (Dong et al. 1998).

2.3 Carboxyl Methyl Cellulose (CMC)

Carboxyl methyl cellulose (a non-nano form of cellulose), which has carboxy methyl groups bound to some hydroxyl groups of the glucose monomers making up the cellulose polymer, was purchased from Hercules (Wilmington, Delaware) and was also tested for toxicity in comparison with NCC. A stock solution of CMC was prepared by dissolving 20 g in 1 L of deionized water at 50ºC. CMC solutions of 0.68% were prepared in well water in order to match the NCC sample preparation procedure. The pH of the stock solution varied from 7.6 – 7.8.

2.4 Toxicity Testing

4

The toxicity testing strategy (see Table I) included i) a toxicity monitoring program, and ii) an in-depth toxicity assessment component.

(Place Table I here)

2.4.1 Toxicity monitoring tests

Acute lethal tests on 12 NCC batches were done with Daphnia magna and Ceriodaphnia dubia and on four of the 12 batches with rainbow trout (Oncorhynchus mykiss). For the daphnid tests, typically four to five concentrations of NCC were tested by serial dilution of the NCC stock with moderately hard water (U.S. EPA 1985a) prepared as a mixture of well water (pH: 8.3; hardness: 210 mg/L as CaCO3; alkalinity: 170 mg/L as CaCO3) and distilled water. Only one NCC concentration (1 or 10 g/L) was tested on rainbow trout because of the larger volume requirements and the limited quantities of NCC available. Tests on rainbow trout and D. magna followed Environment Canada protocols (Environment Canada 2000a; Environment Canada 2000b) except that trout were maintained at 13° ± 1°C instead of 15° ± 1°C. The C. dubia tests (Environment Canada 2007) were done with the same NCC solutions (concentrations) as in the D. magna tests. Ten Ceriodaphnia were placed in a volume of 40 mL in a container with similar surface area to volume ratio as the Daphnia test. The tests were done at 25°C.

The chronic sublethal toxicity of nine of the 12 NCC samples (Batches 2c, 3, 4, 5a, 5b, 6, 8, 9, 10) was determined with the 7 d C. dubia survival and reproduction test following Environment Canada protocols (Environment Canada 2007).

2.4.2 In-depth toxicity tests

Five (Batches 4, 5b, 7, 8 and 10) of the 12 batches of NCC were subjected to in-depth testing.

Fish reproduction test. One of the NCC batches (Batch 7) was used to assess effects on the reproductive capacity of the fathead minnow (Pimephales promelas) according to a protocol described by Kovacs et al. (2007). In this test, adult males and females were exposed to five concentrations (0.03 to 0.48 g/L diluted in well water) of NCC for 10 days and the egg production of the NCC-exposed fish was compared to the egg production by control fish kept in well water. The tests were done in 15 L glass aquaria containing two males and four females. Each concentration and control was done in duplicate.

Multi-trophic micro-assays. NCC Batch 4 was used in multi-trophic micro-assays with four species which included: i) bacteria, Vibrio fischeri (Microtox), ii) algae, Pseudokirchneriella subcapitata, iii) micro-crustacean, Thamnocephalus platyurus (Thamno Toxkit assay) and iv) the cnidarian, Hydra attenuata. The tests with the four species were performed with the original NCC suspension (no pH adjustment) and the pH adjusted to 6.8 with NaOH. These were diluted to multiple concentrations (that varied from test to test) with the appropriate dilution medium for each test.

Vibrio fischeri was exposed to NCC for 15 minutes and the luminescence of the bacteria was compared to controls by using the Microtox Toxicity Analyzer according to Environment Canada protocols (Environment Canada 1992a). Exponentially growing cells of the freshwater Pseudokirchneriella subcapitata were exposed to NCC using the microplate technique (Blaise and Vasseur 2005). At the end of the 72 hour exposure period, the algal cell growth was determined with an electronic particle counter. The ThamnoToxkit assay [MicroBioTests Inc.] was used to measure the acute lethal toxicity of NCC to the freshwater crustacean Thamnocephalus platyurus. The test organisms are provided as dormant cysts which are reconstituted in the hatching and dilution medium. The toxicity test is conducted on a microplate [MicroBiotests Inc] and the test duration is 24 hours. The Hydra attenuata was exposed to

5

NCC for 96 h in microplates according to the method described by Blaise and Kusui (1997). The endpoints assessed include mortality and morphological anomalies.

In addition to the test with NCC Batch 4, the Hydra test was performed a second time (Batch 5b) with the original NCC suspension (no pH adjustment) and dilution with well water to make up the 0.68% stock solution.

Zebrafish embryo development assays. Two of the NCC batches (8 and 10) were used to assess the effect on the development of zebrafish (Danio rerio) embryos in the first five days following fertilization. The embryos were collected from several breeding pairs 1 hour after fertilization. The embryos were randomized and divided into pools of 60–90, which were exposed to varying concentrations of NCC (1, 3 and 6 g/L) and CMC (0.5, 1, 3, and 6 g/L), diluted in embryo rearing medium (Westerfield 2000). Exposures were conducted in petri dishes containing 35 mL of exposure medium, which was replaced every 24 hours. Mortality, hatching rates and morphological development of the embryos were monitored and compared to control groups raised in plain embryo rearing medium.

In vitro rainbow trout hepatocyte assays. NCC Batch 4 was also used in tests with rainbow trout hepatocytes to assess various indicators of cell viability and function. Primary cultures of immature rainbow trout hepatocytes were prepared according to an adaptation of the double perfusion method using albumin and citrate to liberate hepatocytes (Gagné 2005). Briefly, livers from n = 4 rainbow trout (8–12 cm long) were used for the preparation of primary cultures of hepatocytes. Hepatocytes were plated (n = 6 wells per treatment) in 24-well microplates at 1 × 106 cells/mL in Liebovitz (L-15) medium containing 10 mM Hepes-NaOH, pH 7.4, 100 units of penicillin, 100 µg/mL streptomycin and 0.25 µg/mL amphotericin B. Cells were exposed to increasing concentrations of either NCC or CMC: 0, 16, 80, 400, and 2000 mg/L for 48 h at 15°C. At the end of the exposure period, the microplates were centrifuged at 500 × g for 5 min and the culture medium was removed by aspiration. Cells were washed twice in 0.2 mL of phosphate-buffered saline (PBS: 150 mM NaCl, 5 mM KH2PO4, and 1 mM NaHCO3, pH 7.5) to remove any traces of NCC. Cell density and viability assessments were performed immediately after cell wash. For cell density, absorbance at 600 nm was measured using a microplate reader (Powerwave Reader, USA). The remaining cell suspension was stored at -85°C until analysis.

Cell viability was evaluated by the fluorescein dye retention assay as described by Gagné (2005). The levels of labile zinc were determined in crude cell lysate according to a fluorescent probe method (Gagné and Blaise 1996). Briefly, a portion of the cell suspension was thawed at 4°C and homogenized. A 50 µL aliquot of the undiluted homogenate was mixed with 150 µL of 50 µM TSQ (N-[6-methoxy-8-quinolyl]-p-toluene sulfonamide) probe in 10% DMSO in phosphate-buffered saline. Fluorescence was measured at 360 nm excitation and 460 nm emissions in a microplate reader (Chameleon-II, Bioscan, USA). Standard solutions of zinc chloride were prepared for calibration. The data were expressed as relative zinc equivalents/cell density. Increased levels of labile zinc in hepatocytes exposed to a substance can be interpreted as a possible indicator of cell uptake and interaction with proteins/amino acids in the cells that normally bind the zinc.

The potential uptake and adsorption of NCC in cells was also estimated on the basis of cell carbohydrate levels. The total levels of carbohydrate in crude cell lysate were determined by the anthrone reaction (Jermyn 1975). Briefly, to each sample a volume of anthrone reagent (0.1g anthrone in 50 ml concentrated sulfuric acid) was added in a 1:3 proportion. After thorough mixing, all samples were heated for 15 min at 80C and immediately cooled down in a cold-water bath at the end of the incubation. Sucrose was used for calibration of the assay and optical density was read at 620 nm. The data were expressed as µg of carbohydrate/cell density.

6

The levels of heat shock proteins (HSPs), 72 kDa, were determined according to an enzyme immunoassay (EIA). The inducible molecular chaperones or HSPs are important for cell viability and are typically expressed in response to stress. Approximately 1 ug total of S15 fraction was added to each well in 50 mM carbonate coating buffer, pH 9.6 and added to high-protein-binding-capacity microplate wells (Immulon-4 microplate) overnight at 4°C. After this incubation period, the wells were washed twice with 200 µL PBS and blocked with PBS containing 1% albumin for 30 min at room temperature. The wells were washed once with PBS and 100 µL of Hsp72 polyclonal antibody (Assay Designs, MI; SPA-812) at a 1/5000 dilution in PBS containing 0.5% albumin. The wells were incubated at 37°C for 60 min. Afterwards, the wells were washed three times with PBS and 100 µL of the secondary antibody (rabbit IgG antibody linked to peroxidase) diluted 1/5000 in PBS containing albumin, as described above. The wells were washed four times in PBS and peroxidase activity was determined using the highly sensitive luminescent substrate (BM Chemiluminescence ELISA Substrate (POD), Roche Applied Science). Luminescence was measured using a microplate multireader in luminescence mode (Chameleon II, Bioscience, USA). The data were expressed in ng/ml of Hsp72 corrected against cell density.

Lipid peroxidation as an index of oxidative cell damage was determined in hepatocyte homogenates by the thiobarbituric acid method which measures the levels of malonaldehyde which is liberated during the breakdown of polyunsaturated lipids hydroperoxides (Wills 1987). Thiobarbituric acid reactants (TBARS) were determined by fluorescence at 535 nm for excitation and 600 nm for emission using a fluorescence microplate reader (Bioscan, Chameleon-II). The data were expressed as µg of thiobarbituric acid reactants (TBARS)/cell density (absorbance at 600 nm).

To check for potential genotoxicity, DNA strand breaks were determined according to the DNA precipitation assay (Oliver 1988) using the Hoescht dye for DNA quantitation (Bester et al. 1994). Genomic DNA was precipitated by K-assisted SDS precipitation (associated with proteins) in alkali, which leaves protein-free single and double-stranded DNA in solution. DNA quantification was achieved using Hoescht dye at a concentration of 100 nM in 200 mM Tris-HCl., pH 8.5, containing 400 mM NaCl and 4 mM sodium cholate. Salmon sperm DNA was used for calibration and fluorescence readings were performed at 360 nm excitation and 460 nm emissions (Bioscan, Chameleon-II). The data were expressed as µg of supernatant DNA/cell density.

2.5 Toxicity Threshold Estimations

All of the toxicity thresholds were calculated in accordance with Environment Canada guidelines (Environment Canada 2005). The LC50s for the acute lethal tests were estimated by the binomial method using a program by Stephan (1978). The chronic/sublethal toxicity thresholds were calculated using CETIS v1.7.0revW software (Tidepool Scientific Software). For the Ceriodaphnia and the fathead minnow reproduction tests, the inhibition concentration causing a 25% decrease (i.e., IC25) in the number of young and eggs produced, respectively, was estimated by monotonic smoothing and linear interpolation, with the confidence intervals calculated by a bootstrap method. All other endpoints (e.g., EC50 or concentration causing a specific effect to 50% of the organisms) were estimated by linear regression.

3. Results

3.1 NCC Characterization

The NCC characterization results are shown in Table II. There was minor variability in the characteristics of the NCC produced in the pilot plant. (The variability of the subsamples of Batches 2 and 5 was minimal.) All the batches had pH < 3.0, with a range of 2.0 to 2.9. With two exceptions, the mean NCC particle size ranged from 65.3 to 85.17 nm. The particle size of the two outliers was 52.6 (Batch 1) and

7

96.3 nm (Batch 4). Batch 1 had a different system for acid removal and this probably caused the lower particle size. For the particle size information it should be noted that the analysis assumes the particles have a spherical diameter. Since NCC is known to be rod-like in nature, the particle size data are used to monitor batch quality, rather than for determining absolute size dimensions. The typical dimensions of NCC are 200 nm × 10 nm × 5 nm.

The % NCC in most of the samples ranged from 1.6 to 2.9. For three of the samples, the % NCC weight was 4 (Batch 1), 3.4 (Batch 6) and 4.6 (Batch 7). The mean sulphur content expressed as % weight on NCC in aqueous solution was 0.83 (range 0.66 – 0.98 % S). The sulphur content of most NCC samples was between 0.7 and 0.98. Batch 11 had the lowest sulphur content (0.66% S) and this was likely due to a lower acid charge used during hydrolysis. The surface charge density estimates followed the same trend as the sulphur content. Batch 11 had the lowest surface charge density and samples 6 and 9 had surface charge densities just exceeding or equal to 3.0 × 10-3 mol e /kg. The variation in the width of the particle size distribution is evaluated by the polydispersity index (PI). The mean PI was 0.251 with a range 0.227 – 0.273 indicating that variability within the particle size analyses was fairly low.

(Place Table II here)

3.2 Toxicity Monitoring Tests

3.2.1 Acute lethal tests

A summary of the acute lethal toxicity tests is provided in Table III.

(Place Table III here)

In most of the tests, an LC50 could not be estimated because there was less than 50% mortality (in most cases 0% mortality) at the highest concentration tested. It was not considered necessary to routinely test NCC at concentrations greater than 5 g/L. All 15 Daphnia magna acute tests had an LC50 > 1 g/L. In 11 of the 15 tests the LC50 was > 5 g/L. In three cases, the LC50s could be estimated as 2.5 g/L (Batch 2a), 3.2 g/L (Batch 2b) and 2.1 g/L for Batch 3. Twelve of the 13 acute tests conducted with Ceriodaphnia dubia had an LC50 > 1 g/L and seven of these had LC50 value >5 g/L. As for D. magna, the LC50 of sample b of Batch 2 was 3.2 g/L. The LC50 of Batch 3 (1.6 g/L) and Batch 7 (4.3 g/L) was slightly lower than in tests with D. magna. Batch 2c had the lowest LC50 of 0.3 g/L.

The CMC caused no mortality at the highest concentrations in tests with both D. magna and C. dubia. The LC50 of NaCl (reference toxicant) in tests with D. magna and C. dubia was 5.7 g/L and 1.7 g/L, respectively.

In three of the rainbow trout tests, the LC50 was > 1 g/L and one test had an LC50 > 10 g/L. There were no signs of behavioral or morphological (reddening or clogging of the gills) abnormalities. Trout were more tolerant to NaCl (96 h LC50 ~ 15.9 g/L) than the daphnid species.In addition to the acute lethal testing, nine NCC batches were tested using the 7 d Ceriodaphnia reproduction test. The results are shown in Table IV along with results for CMC and sodium chloride.

(Place Table IV here)

With the one exception (Batch 2c), the IC25 for NCC was >1 g/L. The IC25s for CMC and NaCl in the Ceriodaphnia reproduction test were >1 g/L or just above 1 g/L.

8

The low LC50 and IC25 of Batch 2c in the Ceriodaphnia acute and chronic tests were probably caused by the addition of sodium azide for preservative purposes. This was done for some of the early (sub)samples produced in the pilot plant. The addition of sodium azide was discontinued for later samples.

3.3 In-depth Toxicity Testing

3.3.1 Fathead minnow reproduction test

The cumulative egg production data are shown in Figure 2. The mean egg production by the control fish was 68 eggs per female. The mean egg production (64 to 75 eggs per female) of the fish exposed to 0.03 to 0.24 g/L NCC was virtually the same as the egg production by the control fish. The mean egg production of the fish at 0.48 g/L NCC was 17 eggs per female. The IC25 for these data was estimated to be 0.29 g/L (0.19–0.31 g/L).

(Place Figure 2 here)

3.3.2 Multi-trophic micro-assays

The results of the whole organism tests with four aquatic species are shown in Table V.

In tests with NCC Batch 4 and no pH adjustment, the most sensitive response was the occurrence of morphological abnormalities in the Hydra (EC50 of 0.06 g/L). All other species and the mortality endpoint in Hydra were affected by NCC concentration > 0.1 g/L. The response pattern was the same with pH adjusted Batch 4, although at much higher threshold concentrations (0.36 g/L to 13.2 g/L). Since Hydra was the most sensitive species, another sample of NCC (Batch 5) was also tested with and without pH adjustment. In this case, the pH of the (2.72%) NCC suspension was achieved by dilution with well water to a consistency of 0.68% (as was done for the toxicity monitoring tests described above) rather than with NaOH. The pH of the 0.68% NCC suspension was 6.4. This sample was less effective in causing morphological changes in the Hydra with EC50 values of 0.36 g/L (without pH adjustment) and >6.8 g/L (with pH adjustment).

(Place Table V here)

3.3.3 Zebrafish assay results

In both batches tested (NCC Batch 8 and 10) on zebrafish embryos, the LC50 was found to exceed 6 g/L (Table VI). Toxicity of CMC to zebrafish was found to be slightly greater than that of NCC with the LC50 3-6 g/L. No morphological abnormalities were noted in zebrafish exposed to either substance.

(Place Table VI here)

In preliminary trials (data not shown) and in final testing of NCC Batches 8 and 10, there was a trend for delayed hatching of embryos over the first 54 hrs of exposure. However, by the end of the 96 h exposure period, only minor effects on hatching were noted even at the highest doses (Table VI). For both batches of NCC studies, IC50 values representing the ability of NCC to inhibit hatching, exceeded 6 g/L. Hatching rates were reduced more significantly by CMC (IC50 3-6 g/L), and it was noted that CMC accumulated around pre-hatching embryos, likely forming a physical barrier which led to reduction in hatching.

3.3.4 In vitro rainbow trout hepatocyte assay

9

Hepatocyte morphology was examined by phase contrast microscopy in the attempt to identify changes of cell shape or size upon exposure to CMC and NCC (Figure 3).

(Place Figure 3 here)

For CMC, no apparent changes of cell morphology occurred for concentrations up to 2000 mg/L. For NCC, cell swelling was observed at 16 mg/L. For concentrations of > 80 mg/L, cells were less abundant and swelled about twice the normal indicating signs of cell death. These results were corroborated by cell viability determined using the fluorescein dye retention principle (Table VII). NCC was more cytotoxic than CMC with an estimated EC20 and EC50 of 34 and 245 mg/L for NCC, respectively. In contrast, EC20 and EC50 values were >2000 mg/L for CMC.

(Place Table VII here)

The hepatocytes exposed to NCC had increased levels of labile zinc (Table VII), especially at the higher concentrations tested. The EC20 was 171 mg/L and the EC50 was >2000 mg/L. The CMC had no effect on labile zinc levels in hepatocytes.

The levels of HSP (protein chaperones of the 72 kDalton family) in hepatocytes exposed to NCC and CMC did not respond in a concentration dependent manner (see Figure 4). NCC and CMC affected the HSP in a similar manner. At the lower concentrations, there was a decrease in HSP with an EC20 of 10 mg/L and 14 mg/L for NCC and CMC, respectively (Table VII). The EC50s were >2000 mg/L for NCC and CMC. At 2000 mg/L, NCC and CMC caused a similar increase in HSP levels.

(Place Figure 4 here)

There was no increase of cell-associated carbohydrates in cells exposed to CMC (Table VII). In contrast, there were increased carbohydrates associated with NCC-exposed cells. The EC20 and EC50 was <16 mg/L and 23 mg/L, respectively.

The results of oxidative stress tests showed (Table VII) that only NCC was able to increase LPO activity in cells. The EC20 and EC50 were 100 mg/L and 434 mg/L, respectively.

The results of the genotoxicty assay are also shown in Table VII. Neither NCC nor CMC produced changes in DNA strand breaks at concentrations up to 2000 mg/L.

4. Discussion

Presented is the first comprehensive toxicological characterization of NCC with aquatic organisms. The need for an extensive toxicological characterization of nanomaterials involving multiple species and endpoints is illustrated by the work of Blaise et al. (2008) who showed a wide range of sensitivity in tests with organisms representing different trophic levels. As NCC is a new material yet to be introduced to the market place, our goal was to characterize its potential toxicity at multiple trophic levels with the intention to reduce ecotoxicological uncertainties that typically exist with new substances.

Characterization of NCC was conducted in two phases with tests involving nine species and the examination of a broad range of endpoints (acute/chronic, lethal/sublethal). As NCC production was still in the developmental stage, the first step was to monitor the toxicity of the different batches of NCC produced in a pilot plant to gain insight as to possible variability. For this phase of the work, standard tests were performed with rainbow trout, D. magna and C. dubia, which are species that are used for regulatory testing in Canada. As such, this monitoring work also allowed for the assessment of the

10

potential of NCC production to influence effluent regulatory compliance. The second step involved a more in-depth ecotoxicological characterization of selected NCC samples with six more species and rainbow trout hepatocytes in vitro to gain insight as to environmental compatibility beyond the regulatory context. Both phases of the work included comparison of the results of NCC toxicity tests to the results of toxicity tests with CMC, a surrogate of the parent cellulosic material. This broad toxicological assessment will assist in making an accurate risk assessment of NCC.

4.1 Worst-case Scenario for NCC Losses at a Production Facility

Current plans are to produce NCC from kraft pulp and, as such, the first NCC production facilities will be located at bleached kraft mills. While it is unlikely that major spillage could happen under tight management of NCC production given its high value, it is important to develop the worst case scenario for risk assessment. A first-generation NCC production facility is expected to produce about 1 ton of NCC per day as a 2 to 5% suspension in water. The pH of this suspension will be between 2 and 3 and such a low pH could deleteriously affect aquatic organisms. This was also evident in the comparison of pH adjusted and non-adjusted NCC suspensions tested in the multi-trophic micro-assays. However, the pH of effluent discharges from mills is regulated provincially and the typical discharges from mills are in the range of pH 6.5 to 7.8 (Kovacs et al. 2005; Martel et al. 2004). As such, the focus of this work was the actual toxicity of the NCC itself and not the low pH of the NCC suspension.

Assuming no loss through the treatment system, the worst-case NCC concentration that would occur in the effluent of a kraft mill, should a complete spillage of one day’s production occur, can be estimated. The water usage of pulp/paper mills in Canada can vary from about 20 m 3/t of pulp to over 167 m3/t of pulp and the pulp production ranges from 393 to 1600 t/d (Kovacs et al. 2005; Martel et al. 2004). Based on these figures, the average mill discharges roughly 55,000 m3 of effluent per day. The retention time of a typical activated sludge biotreatment system is about 18 h meaning the holding capacity of the system is about 41,250 m3 (55,000/1.33). Consequently, if 1 ton of NCC was lost in one event, the worst-case concentration of NCC in the mill’s effluent would be approximately 0.024 g/L or 24 mg/L (1000 kg/41, 250 m3). For receiving waters, the average kraft mill effluent concentration in Canadian rivers is 1% (Kovacs 1986). The predicted NCC concentration in the immediate vicinity of a mill discharge would be approximately 0.24 mg/L. In addition to the least likely loss of a whole day’s production, consideration should also be given to smaller but continuous losses during manufacturing. To assess such a scenario, assuming a continuous loss of 1% (an amount that can be considered as excessive and used here only for a worst-case scenario involving continuous losses), the NCC concentration in the final effluent would be 0.18 mg/L (10 kg/55,000 m3). In such a case, the average concentration of NCC in the river would be 1.8 μg/L, which is more than 160,000-fold lower than the IC25 of our most sensitive test (FHM egg production).

4.2 Monitoring Tests: Variability and Regulatory Implications

The batches of NCC produced in the pilot plant over a year period were monitored by means of acute lethal tests using rainbow trout, D. magna and C. dubia as well as chronic sublethal (reproduction) tests using C. dubia.

With one exception (LC50 of 0.3 g/L), all the LC50 values exceeded 1 g/L. In fact, most samples (7 of 13) had LC50 values exceeding 5 g/L and in one test with trout, the LC50 was >10 g/L. Of the three species used, C. dubia was the most sensitive. Regarding the variability of the results, the first three batches of NCC had the lowest LC50 values and these samples represent the earlier days of pilot plant production. However, there were no obvious and consistent differences in the NCC physical and chemical characteristics (see Table II) between the early and late samples that could provide specific leads concerning the differences in LC50s.

11

Irrespective of the variability, NCC production will virtually have no possibility of affecting regulatory toxicity of effluent discharges from kraft mills. In Canada, the effluent (at full-strength or 100% concentration) from pulp/paper mills must not cause more than 50% mortality in rainbow trout during a 96 h exposure and more than 50% mortality in D. magna during a 48 h exposure (Fisheries Act 1991). Even in the case of a complete loss of a day’s NCC production, the concentration of NCC in the effluent (i.e., 24 mg/L) would be orders of magnitude (50 to 250) lower than the concentrations that could cause mortality.

As was the case for acute lethal toxicity, with one exception, the IC25 values of NCC affecting C. dubia reproduction exceeded 1 g/L. There was slight variability in the effects caused by NCC, but these could not be related to specific physical/chemical characteristics and may have been more reflective of test variability than variability of NCC. The C. dubia reproduction test is used for the regulatory Environmental Effects Monitoring (EEM) program in Canada (Environment Canada 1992b) to gauge effluent quality over time. Based on the results of this work, the production of NCC and possible losses associated with this production is predicted to have no influence on the results of such tests.

The acute lethal and chronic sublethal toxicity tests with Ceriodaphnia provided an opportunity to compare acute-chronic toxicity. The acute to chronic ration (ACR) is used to identify substances that may require special consideration. For example, Kenaga (1982) found that in tests with Daphnia, about 53% of the 135 chemicals tested had ACRs in the 1 to 9 category. Chemicals with higher ACRs, considered to have specialized or unusual toxicity, tended to be pesticides and some metals. It was difficult to calculate an acute to chronic effects ratio for NCC with certainty because many of the acute lethal and chronic sublethal tests with Ceriodaphnia had no effect at the highest concentration of NCC. By selecting only the results where an LC50 and IC25 could be estimated and by averaging these LC50 and IC25 values, the ACR is ~ 2. Clearly, based on the ACR, NCC is not in a category of substances that has unusual toxicity that would require special attention.

4.3 In-depth Testing

4.3.1 Whole organism and bacterial light inhibition tests

The whole organism tests were performed on six species (two fish, a bacterium, an invertebrate and an alga) and examined endpoints including mortality, growth, reproduction and morphological abnormalities. The reproduction of the fathead minnow was the most sensitive endpoint with an IC25 of 0.29 g/L. Of the multi-trophic micro-assays, the manifestation of morphological abnormalities in Hydra had the lowest EC50 of 0.36 g/L. The other organisms were affected at NCC concentrations well in excess of 1 g/L, as was the case in the regulatory tests described above. The acute/sublethal toxicity ratio for the tests with Hydra was about 1.5, just less than the acute/chronic ratio of two estimated for tests with Ceriodaphnia. The tests with zebrafish allowed for the evaluation of the influence of NCC on early life development. This is considered to be a sensitive and critical life stage for toxicological assessments (McKim 1985). There was no indication that NCC was toxic at this stage of fish development up to a concentration of 6 g/L. In summary, the more in-depth whole-organism tests confirmed the findings of the monitoring work in that possible losses during the manufacturing of NCC are not likely to be toxic at concentrations expected to be found in bleached kraft mill effluents.

4.3.2 In vitro rainbow trout hepatocytes

The tests with hepatocytes provided another perspective regarding the potential interaction of NCC with living organisms. Interaction at the cellular level can be informative regarding the mechanism of effect and as indicators of early warning concerning whole organism toxicity (Sprague 1990). Indeed, tests with

12

hepatocytes were more sensitive than whole organism tests. The EC20 values for cell viability and lipid peroxidation were 34 and 100 mg/L, respectively. However, because some of the endpoints in the hepatocyte tests were more sensitive than the endpoints in the whole organism tests, it raises questions about NCC uptake and bioavailability in aquatic species. It is possible that whole organisms were less sensitive because NCC was not readily absorbed and thus could not affect cells (i.e., hepatocytes) within the body. This important question should be resolved with future investigations.

Concerning possible mechanism of effect, the loss of cell viability in trout hepatocytes exposed to NCC was accompanied with an increase in cell-associated carbohydrates suggesting that the carbohydrates, a possible indicator for NCC in cells, was linked to loss in cell viability. Further evidence for this comes from labile zinc measurements. NCC is known to contain sulphates on the carbohydrate backbone. Because of the presence of these anionic groups, divalent metals (such as the naturally occurring zinc) could bind to NCC and, in fact, the levels of labile zinc in cells were found to increase with NCC concentrations. Two other cellular indicators were affected by NCC: lipid peroxidation and heat shock proteins. The thresholds for lipid peroxidation in terms of EC20 and EC50 were 100 and 400 mg/L, respectively. This finding raises the possibility that at such concentrations the NCC may interact with the cell membranes, a likely possibility due to the polymeric nature of NCC. As for heat shock proteins, t he levels of protein chaperones of the 72 kDalton family measured in this study is used as a biomarker response to highlight stress at the protein conformation level which could be a result of nanomaterial-induced steric hindrance in cells. However, in this case, the results indicated that cellulosic material caused both a decrease (at lower concentrations) and an increase (at 2000 mg/L, the highest concentration tested) in heat shock protein levels making conclusions about potential mechanistic effects difficult. More importantly, the pattern of effect was similar for both NCC and CMC (Figure 4) suggesting that the effect was not related to the nanosize of NCC. Nevertheless, because there were some differences between NCC exposed cells and controls, albeit in a non consistent manner, the possibility of other interactions should be considered such as metal mobilization and oxidative stress. Finally, the lack of changes in DNA strand breaks in cells exposed to NCC suggests that NCC is not genotoxic to the fish liver cells. This is particularly noteworthy as genotoxicity has been identified as one of the most important mechanisms of toxicity caused by other nanomaterials such as silver nanoparticles, certain fullerenes, quantum dots and titanium dioxide (Klaine et al. 2008).

In summary, the NCC proved to be “bioavailable” or closely associated to hepatocytes. This association was implicated with the loss of cell viability and to metal mobilization and enhanced lipid peroxidation. These effects seemingly occurred before and during the loss of cell viability. Nanomaterials could thus act as carriers of other contaminants as shown here with zinc mobilization experiments. This “vectorisation” possibility could potentiate the toxicity of other contaminants found in pulp and paper effluents and thereby exacerbate (or lessen) their effects. More research studies are warranted to examine this more closely in addition to stability studies in effluents (i.e., the persistence and breakdown products which could have toxic outcomes and influence the toxicity of the effluent as well).

4.4 Determining the Potential Hazard of NCC Toxicity Ranking and Comparisons

The toxicity of a particular substance or material can be categorized according to the concentration required to cause an effect. One example is the categorization of pesticides in terms of acute lethal (LC50) toxicity (Kamrin 1997). For aquatic species, the categorization (in terms of mg/L LC50) is:

Toxicity categorization LC50, mg/LVery highly toxic <0.1Highly toxic 0.1 to 1.0Moderately toxic >1.0 to 10

13

Slightly toxic >10 to 100Practically nontoxic >100

Using this categorization, on the basis of the acute lethal toxicity testing, NCC can be classified as practically nontoxic. The relatively low toxicity of NCC is confirmed by the results of (chronic) sublethal tests with whole organisms where the threshold of effects also occurred at concentrations >100 mg/L. The cytotoxicity assays with hepatocytes, which were more sensitive than whole organism tests, give the indication that NCC is in the slightly toxic to practically non toxic category depending on the endpoint examined. The U.S. EPA’s Office of Pollution Prevention and Toxics (OPPT) has also provided hazard criteria for determining the thresholds of low, medium and high aquatic toxicity for new substances (Smrchek et al. 1993). The criteria for concern are:

Criteria Acute toxicity, mg/L Chronic toxicity, mg/LHigh concern <1 <0.1Medium concern >1 to ≤100 >0.1 to ≤10Low concern >100 >10

Based on these criteria, NCC can be considered a new material of low concern.

Another important aspect of toxicity characterization of nanomaterials is comparison of toxicity to the material in the so-called bulk or “non nano” form as well as to other nanomaterials which have been tested for toxicity by the same or similar methods. In our study, some tests with CMC were conducted in comparison to NCC. The whole organism tests we performed indicated that the threshold concentrations for acute and (chronic) sublethal effects were similar for NCC and CMC. In zebrafish early life stage tests, both LC50 values and IC50 values representing interference with hatching indicated that CMC was more potent than NCC. In addition, in tests with D. magna and C. dubia, the LC50 and IC25 for NCC were not lower than the LC50 or IC25 for sodium chloride. In contrast, NCC was found to affect various endpoints in hepatocyte assays and this was not evident in tests with CMC indicating the greater sensitivity of hepatocytes to the nano-form of cellulose.

Some nanomaterials assessed by tests similar to the ones used in this study have been found to be more toxic than NCC. For example, in D. magna acute lethal tests with titanium dioxide and fullerene (Lovern and Klaper 2006), the 48 h LC50s were reported to be as low as 5.5 and 0.46 mg/L, respectively. Also in tests with D. magna, the 48 h LC50 of three metal-based nanomaterials and three carbon-based nanomaterials ranged from 1.5 to 162 mg/L (Zhu et al. 2009). In this case, in contrast to NCC, the bulk counterparts were found to be less toxic to Daphnia for five of the six nanomaterials. In comparison to these nanomaterials, NCC is 1000 to 10000 fold less toxic.

Blaise et al. (2008) have also used the multi-trophic micro-assay, the rainbow trout hepatocyte acute cytoxicity assay as well as two other small-scale bioassays to test 11 nano-powders (nine metallic and two carbon-based: single wall carbon nano tube and fullerene C60). Effects on hepatocytes occurred in concentration ranges of 1 to 10 mg/L and > 100 mg/L. The tests with primary producers (P. subcapitata) and Hydra were found to be the most sensitive to seven of the nano-powders. The toxicity thresholds ranged from <0.5 to >500 mg/L, representing a very broad range of responses. Of the 11 nano-powders tested, four caused effects in the 0.1 to 1.0 mg/L range (very toxic), another four in the 1 to 10 mg/L range (toxic) and three were found to cause effects only at the 10–100 mg/L (harmful) and >100 mg/L (not toxic) concentration ranges. The rainbow trout hepatocyte assays used in this study were also used to test the toxicity of aged cadmium-telluride quantum dots (Gagné et al. 2008). The quantum dots affected the hepatocytes, including DNA damage, at a threshold concentration of 0.1 mg/L. In comparison to these

14

studies, NCC caused effects at the 10–100 mg/L concentration range in some hepatocyte tests and all other whole-organism effects occurred at concentrations well above 100 mg/L.

Results of acute lethal tests with rainbow trout and chronic reproduction tests with Ceriodaphnia exposed to nanomaterials are not readily available for comparison purposes. However, rainbow trout were used in a test with titanium dioxide and a 1 mg/L concentration after 14 d exposure caused oxidative stress in gills (Federici et al. 2007). In a test with D. magna, C60 fullerene caused inhibition of molting and reproduction at concentrations of 2.5 and 5 mg/L, respectively (Oberdörster et al. 2006). The early life stage of zebrafish was found to be a sensitive means of assessing and comparing the toxicity of several nanomaterials, such as fluorescent semiconductor nanocrytals or quantum dots. Of the five quantum dots tested, the LC50 values for larvae ranged from 7 to 42 μM Cd equivalents (King-Heiden et al. 2009). The LC50 for CdCl2 was 409 μM Cd equivalents. Since NCC did not affect zebrafish in the g/L range, this comparison again demonstrates the much lower toxic potential of NCC. The early life stage of zebrafish was also used in tests with various carbon-based nanomaterials. Buckminsterfullerene (nC60) caused effects at 1.5 mg/L whereas fullerol caused no effects at 50 mg/L (Zhu et al. 2007). Single-walled carbon nanotubes affected the fish only at concentrations ≥120 mg/L (Cheng et al. 2007). There do not appear to be reports of fish reproduction tests with nanomaterials for comparison with NCC. Nonetheless, since the reproduction of the fathead minnow was the most sensitive whole organism test in this study, and because fecundity is considered one of the most sensitive endpoints in toxicity tests (Giesey and Graney 1989), it may be advisable to consider such tests for the environmental assessments of other nano materials.

4.5 Overall Assessment

Prospective aquatic risk assessment of a new material/substance is often hampered by uncertainty due to lack of sufficient toxicological information. In such cases, assessment factors (European Commission (1996) or uncertainty factors (Smrchek et al 1993; U.S. EPA 1985b) are used to compensate for the uncertainty. The aim is to ensure that risk assessment will be protective for the most sensitive species and endpoint.

The uncertainty/assessment factor is 1000 when very limited information is available, such as the result of only an acute lethal toxicity test with one species. The uncertainty/assessment factors can be eliminated or reduced in multiples of ten by undertaking a thorough toxicological assessment as was done for NCC. This included i) multiple tests to assess sample variability and acute tests with more than three species to reduce the factors to 100 and ii) short-term chronic tests with three species to reduce the factors to 10. Field testing is required to reduce the uncertainty/ assessment factor to one and this was not done for NCC. Instead, a deliberate attempt was made not only to undertake tests with aquatic species representing the various trophic levels but also to use tests (e.g., P. subcapitata, D. magna, V. fischeri) specifically recommended for toxicity characterization of nanomaterials (Kahru et al. 2008) and to study endpoints of specific concern for nanomaterials, such as genotoxicity (Klaine et al. 2008). In addition, this study included sensitive and critical life stages in fish involving the early developmental stage and reproduction. Therefore, even though this is the first ecotoxicological characterization of NCC, the reduction/elimination of uncertainty/ assessment factors can be made with some confidence.

With an uncertainty/assessment factor of 10 applied to the most sensitive responses in this study, NCC does not appear to have the potential to harm aquatic organisms at concentrations that could occur in receiving waters even under worst-case scenarios. The most sensitive whole organism response to NCC was fish reproduction with an IC25 of 0.29 g/L. If we apply a factor of 10 to this, the threshold concentration becomes 29 mg/L, a concentration that is well above what could ever be expected to occur in receiving waters. On the basis of the hepatocyte test results, with 34 mg/L being the threshold affecting cell viability, the assessment factor of 10 would bring the threshold concentration to 3.4 mg/L. Even such concentrations would not be expected to occur in receiving waters.

15

In summary, the initial ecotoxicological characterization of NCC has not revealed serious environmental concerns. Nevertheless, further testing will be necessary along with fate, assessment of potential NCC uptake and exposure studies to complete the risk assessment for NCC.

5. Acknowledgements

We thank George Sacciadis and Guy Njamen of FPInnovations – Paprican Division for providing the pilot plant samples of NCC as well as the physical/chemical characterization of these samples. We thank Joelle Auclair from Environment Canada for undertaking the trout hepatocytes test and biochemical analyses. Michelle Ricard, FPInnovations – Paprican Division, reviewed the report and offered useful suggestions for improvements. We are also grateful for funding of this work provided to FPInnovations – Paprican Division under NRCan’s Forest Industry Long-Term Competitiveness Strategy program.

6. References

APHA, AWWA, WPCF. 1988. Standard Methods for the Examination of Water and Wastewater. 20 th

Ed., American Public Health Association, American Water Works Association, Water Pollution Control Federation, Washington, DC.

Bester MJ, Potgieter HC, Vermaak WJH. 1994. Cholate and pH reduce interference by SDS in the determination of DNA with Hoescht. Anal. Biochem. 223: 299–305.

Biswas P, Chang-Yu W. 2005. Nanoparticles and the environment. J. Air Waste Manage. Assoc. 55: 708–746.

Blaise C, Gagné F, Férard JF, Eullaffroy P. 2008. Ecotoxicity of selected nano-materials to aquatic organisms. Environ. Toxicol. 23: 591–598.

Blaise C, Vasseur P. 2005. Algal microplate toxicity test. In: Blaise C. and Férard, JF, editors. Small-scale Freshwater Toxicity Investigations. Volume 1. Springer, Dordrecht, pp 137–179.

Blaise C, Kusui T. 1997. Acute toxicity assessment of industrial effluents with a microplate-based Hydra attenuata assay. Environ. Toxicol. Water Qual. 12: 53–60.

Cheng J, Flahaut E, Cheng SH. 2007. Effect of carbon nanotubes on developing zebrafish (Danio rerio) embryos. Environ. Toxicol. Chem. 26: 708–716.

Dong XM, Revol J-F, Gray D. 1998. Effect of microcrystalline preparation conditions on the formation of colloid crystals of cellulose. Cellulose, 5: 19–32.

Environment Canada. 2007. Biological test method: Test of reproduction and survival using the Cladoceran Ceriodaphnia dubia. EPS 1/RM/21 Second Edition.

Environment Canada. 2005. Guidance document on statistics methods for environmental toxicity tests. EPS 1/RM/46.

Environment Canada. 2000a. Biological test method: Reference method for determining acute lethality of effluent to Rainbow Trout. EPS 1/RM/13 Second Edition.

16

Environment Canada. 2000b. Biological test method: Reference method for determining acute lethality of effluent to Daphnia magna. EPS 1/RM/14 Second Edition.

Environment Canada. 1992a. Biological test method: Toxicity test using luminescent bacteria (Photobacterium phosphoreum). EPS 1/RM/24.

Environment Canada. 1992b. Environmental effects monitoring studies. Sections 28–35 of the Pulp and Paper Effluent Regulations. Canada Gazette Part II 126(11), 1982–1983.

European Commission. 1996. Technical Guidance Document in Support of Commission Directive 93/67/EEC on Risk assessment for New and Notified Substances and Commission Regulation (EC) No. 1488/94 on Risk Assessment for Existing Substances. Part II: Environmental Risk Assessment. Office for the Official Publications of the European Communities, Luxembourg.

Federici G, Shaw BJ, Handy RD. 2007. Toxicity of titanium dioxide nanoparticles to rainbow trout (Oncorhynchus mykiss): Gill injury, oxidative stress, and other physiological effects. Aquat. Toxicol. 84: 415–430.

Fisheries Act. 1991. Pulp and Paper Effluent Regulations. Canada Gazette Part II 126(11): 1967–2006.

Gagné F, Maysinger D, André C, Blaise C. 2008. Cytotoxicity of aged cadmium-telluride quantum dots to rainbow trout hepatocytes. Nanotoxicol. 2(2):113–120.

Gagné F. 2005. Acute toxicity assessment of liquid samples with primary cultures of rainbow trout hepatocytes. In Blaise C and Férard JF. Small-scale Freshwater Toxicity Investigations. Volume 1. Springer, Dordrecht, pp 453–472.

Gagné F, Blaise C. 1996. Available intracellular Zn as a potential indicator of heavy metal exposure in rainbow trout hepatocytes. Environ. Toxicol. Water Qual. 11: 319–325.

Giesey JP, Graney RL. 1989. Recent developments in and intercomparisons of acute and chronic bioassays and indicators. Hydrobiologia 188/189: 21–60.

Hill AJ, Teraoka H, Heideman W, Peterson RE. 2005. Zebrafish as a model vertebrate for investigating chemical toxicity. Toxicol. Sci. 86: 6–19.

Hubbe MA, Rojas OJ, Lucia LA, Sain M. 2008. Cellulosic nanocomposites: A review. Bioresources 3(3): 929–980.

ISO 13321. 1996. Particle size analysis – photon correlation spectroscopy. International Organization for Standardization.

Jermyn MA. 1975. Increasing the sensitivity of the anthrone method for carbohydrate. Anal. Biochem. 68: 332–335.

Kahru A, Dubourguier HC, Blinova I, Ivask A, Kasemets K. 2008. Biotests and biosensors for ecotoxicology of metal oxide nanoparticles: A minireview. Sensors 8: 5153–5170 (this ref was #53 but is not mentioned in the text anywhere).

17

Kamrin MA. 1997. Pesticide profiles: Toxicity, environmental impact, and fate. Lewis Publishers, Boca Raton, FL.

Kenaga EE. 1982. Predictability of chronic toxicity from acute toxicity of chemicals in fish and aquatic invertebrates. Environ. Toxicol. Chem. 1: 347–358.

King-Heiden TC, Wiecinski PN, Mangham AN, Metz KM, Nesbit D, Pedersen JA, Hamers RJ, Heideman W, Peterson RE. 2009. Quantum dot nanotoxicity assessment using the zebrafish embryo. Environ. Sci. Technol. 43: 1605–1611.

Klaine SJ, Alvarez PJJ, Batley GE, Fernandes TF, Handy RD, Lyon DY, Mahendra S, McLaughlin MJ, Lead JR. 2008. Nanomaterials in the environment: Behavior, fate, bioavailability, and effects. Environ. Toxicol. Chem. 27: 1825–1851.

Klemm D, Schumann D, Kramer F, Heßler N, Hornung N, Schmauder H-P, Marsch S. 2006. Nanocelluloses as innovative polymers in research and application. Adv. Polym. Sci. 205: 49–96.

Kovacs T, Martel P, Ricci M. 2007. A short-term adult fathead minnow reproduction test for investigation of cause and investigation of solution work involving pulp and paper effluents. Water Qual. Res. J. Can. 42: 91–100.

Kovacs T, Martel P, Ricci M, Michaud J, Voss R. 2005. Further insights into the potential of pulp and paper mill effluents to affect fish reproduction. J. Toxicol. Environ. Health, Part A 68: 1621–1641.

Kovacs T. 1986. The effects of bleached kraft mill effluent on freshwater fish: A Canadian perspective. Water Pollut. Res. J. Can. 21: 91–118.

Lovern SB, Klaper R. 2006. Daphnia magna mortality when exposed to titanium dioxide and fullerene (C60) nanoparticles. Environ. Toxicol. Chem. 25: 1132–1137.

Malvern Instruments Ltd., Worcestershire, UK, http://www.malvern.com

Martel P, Kovacs T, Voss R. 2004. Survey of pulp and paper mill effluents for their potential to affect fish reproduction. In: Borton DL, Hall TJ, Fisher RS, Thomas J, editors. Pulp and Paper Mill Effluent Environmental Fate and Effects. Lancaster, PA: DEStech., pp 78–92.

McKim JM. 1985. Early life stage toxicity tests. In: Rand GM, Petrocelli SR, editors. Fundamentals of Aquatic Toxicology. Methods and Applications. Hemisphere, Washington, DC, pp 58-94.

Nowack B, Bucheli TD. 2007. Occurrence, behavior and effects of nanoparticles in the environment. Environ. Pollut. 150: 5–22.

Oberdörster E, Zhu S, Blickley TM, McClellan-Green P, Haasch ML. 2006. Ecotoxicology of carbon-based engineered nanoparticles: Effects of fullerene (C60) on aquatic organisms. Carbon 44: 1112–1120.

Oliver PL. 1988. DNA precipitation assay: A rapid and simple method for detecting DNA damage in mammalian cells. Environ. Mol. Mutagen 11: 487–495.

18

Revol J-F, Godbout L, Gray DG. 1998. Solid self-assembled films of cellulose with chiral nematic order and optically variable properties. J. Pulp Paper Sci. 24(5): 146-149.

Smook GA. 2002. Handbook for Pulp and Paper Technologists. Angus Wilde Publications Inc., Vancouver, BC.

Smrchek J, Clements R, Morcock R, Rabert W. 1993. Assessing ecological hazard under TSCA: Methods and evaluation of data. In: Landis WG, Hughes JS, Lewis MA, editors. Environmental Toxicology and Risk Assessment, ASTM 1179, American Society for Testing and Materials, Philadelphia, PA, pp 22–39.

Sprague JB. 1990. Aquatic toxicology. In: Schrack CB, Moyle PB, editors. Methods for Fish Biology. The American Fisheries Society, Bethesda, MD.

Stephan CE. 1978. Methods for calculating an LC50. In: Mayer FL, Hamelink JL, editors. Aquatic Toxicology and Hazard Evaluation. Amer. Soc. Testing and Materials, Philadelphia, PA, ASTM STP No. 634, pp 65–84.

U.S. EPA. 2007. Nanotechnology White Paper. U.S. Environmental Protection Agency, EPA 100/B-07/001.

U.S. EPA. 1985a. Short-term methods for estimating the chronic toxicity of effluents and receiving waters to freshwater organisms. U.S. Environmental Protection Agency, Report EPA/600/4-85/014, WB Horning and CI Weber (eds), 162 p. Cinncinnati, OH.

U.S. EPA. 1985b. Technical Support Document for Water Quality-Based Toxics Control. U.S. Environmental Protection Agency, Office of Water, Washington, DC.

Westerfield M. 2000. The zebrafish book: A guide for the laboratory use of zebrafish (Danio rerio). 4th

ed., University of Oregon Press, Eugene, OR.

Wills ED. 1987. Evaluation of lipid peroxidation in lipids and biological membranes. In: Snell K. and Mullock B, editors. Biochemical Toxicology: A Practical Approach, IRL Press, Washington, DC, pp 127–150.

Zhu X, Zhu L, Chen Y. 2009. Acute toxicities of six manufactured nanomaterial suspensions to Daphnia magna. J. Nanopart. Res. 11: 67–75.

Zhu X, Zhu L, Li Y, Duan Z, Chen W, Alvarez JJ. 2007. Developmental toxicity in zebrafish (Danio rerio) embryos after exposure to manufactured nanomaterials: Buckminsterfullerene aggregates (nC60) and fullerol. Environ. Toxicol. Chem. 26: 976–979.

19

Table I. Bioassays used to determine the toxic potential of NCC.Toxicity Test Assessment Endpoint Measurement

EndpointReference

Monitoring TestsRainbow trout Acute lethality 96 h-LC50 Environment Canada, 2000a Daphnia magna Acute lethality 48 h-LC50 Environment Canada, 2000b Ceriodaphnia dubia Acute lethality 48 h-LC50 Environment Canada, 2007 Ceriodaphnia dubia Chronic sublethal

reproduction7 d-IC25 Environment Canada, 2007

In-Depth Whole Organism TestsFathead Minnow Chronic sublethal

reproduction10 d-IC25 Kovacs et al., 2007

Bacterial test Vibrio fischeri (Microtox® toxicity test)

Acute sublethal light inhibition (after a 15-min exposure)

15 min-IC25 Environment Canada, 1992

Algal test (Pseudokirchneriella subcapitata microplate assay)

Chronic sublethal growth inhibition (after a 72-h exposure)

72 h-IC25 Blaise and Vasseur, 2005

Micro-crustacean test (ThamnoToxkit assay)

Acute lethality (after a 24-h exposure)

24h-LC50 Microbiotests Inc., http://www.microbiotests.be/

Cnidarian test (Hydra attenuata assay)

Acute sublethality indicated by morphology changes (after a 96-h exposure)

96h-EC50 Blaise and Kusui, 1997

Zebrafish (Danio rerio) Acute sublethal embryo development

96 h-EC50 Hill, 2005

Rainbow Trout Hepatocyte AssaysFish cell test (rainbow trout primary hepatocyte test)

Acute cytotoxicity (after a 48-h exposure)

48h EC20 and EC50

Gagné 2005; Gagné and Blaise 1996; Jermyn 1975; Wills 1987; Oliver 1988

20

21

Table II. Physical/chemical characteristics of NCC.

NCC Batch pH Weight,

%Conductivity,

µS/cm

S, % Weight in NCC Solution

Charge Density, mol e/Kg NCC

Mean Particle Size, nm

PolydispersityIndex

1 2.6 4 1645 - - 52.6 0.2522a 2.9 1.7 563 0.77 2.4 × 10-3 77.3 0.2272b 2.8 1.6 522 0.78 2.4 × 10-3 77.3 0.2272c 2.8 1.7 552 0.77 2.4 × 10-3 78.8 0.2293 2.7 2 865 0.98 3.1 × 10-3 83.0 0.2734 2 1.9 1048 0.93 2.9 × 10-3 96.3 0.2715a 2.6 2.7 1236 0.8 2.5 × 10-3 85.2 0.2585b 2.7 2.7 1258 0.79 2.5 × 10-3 85.2 0.2586 2.6 3.4 1229 0.7 2.2 × 10-3 74.0 0.2367 2.5 4.6 1980 0.79 2.5 × 10-3 76.0 0.2618 2.7 1.9 1111 0.9 2.8 × 10-3 70.2 0.238

9 2.6 2.7 1054 0.95 3.0 × 10-3 70.0 0.27310 2.6 2.6 982 0.92 2.9 × 10-3 69.0 0.25711 2.4 2.8 679 0.66 2.1 × 10-3 69.6 0.26612 2.2 2.9 988 0.85 2.7 × 10-3 65.3 0.235

Mean 2.6 2.6 1048 0.83 2.6 × 10-3 75.0 0.251SEM 0.06 0.2 104 0.03 0.09 × 10-3 2.6 0.004

SEM = Standard error of mean

21

Table III. Results of acute lethal toxicity tests with three aquatic species.

Substance Batch NumberLC50 g/L (95% confidence interval)

Daphnia (48 h) Rainbow Trout (96 h) Ceriodaphnia (48 h)NCC 1 > 1 >1 > 1

2a 2.5 (1 - 5) - 1.4 (1 – 2.5)2b 3.2 (1 – 10) - 3.2 (1 – 10)2c > 1 - 0.3 (0.1 – 1)3 2.1 (1 – 5) >1 1.6 (1-2.5)4 > 5 >1 > 55a > 5 - > 55b > 5 - > 56 > 5 - > 57 > 5 - 4.38 > 5 >10 > 59 > 5 - > 5

10 > 5 - -11 > 5 - -12 > 5 - > 5

CMC > 10 - 3.2 (1 – 10)NaCl   5.7 (4 – 8) 15.9 (14 – 18) 1.7 (1 – 2)

22

Table IV. Results of chronic sublethal toxicity tests with C. dubia.

Substance Batch Number IC25, g/L(95% confidence interval)

NCC 2c 0.2 (0.14 -0.22)3 > 14 > 15a > 15b > 16 1.48 > 29 1.610 > 2

Test NumberCMC 1 > 1

2 > 13 > 14 1.1 (0.9-1.1)

NaCl 1 1.1 (1.0-1.1)2 1.1 (0.8-1.2)3 1.4 (0.96-2.0)4 1.2 (1.1-1.2)

23

Table V. Results of multi-trophic micro-assays (results in g/L NCC with 95% confidence interval)

NCC Batch pH pH

AdjustedV. fischer P. subcapitata T. platyurus H. attenuata

IC 25 IC25 LC50 LC50 EC504 2.3 no 0.46 0.12 3.54 0.36 0.06

(0.1-2.1) (0.08-0.18) (2.5-5) (0.26-0.5) (0.04-0.08)

4 6.8 NaOH > 10 > 2.5* 13.2 14.22 2.6nc nc (12.2-14.24) (10-20) (2-3.22)

5b 2.7 no - - - 0.55 0.36(0.46-0.66) (0.29-0.46)

5b 6.4 well water

- - - > 6.8 > 6.8

* Tests could not be done above 2.5 g/L due to interference with Coulter Counter cell counts at higher concentrations. nc - not calculable

24

Table VI. Results of acute toxicity tests with zebrafish.

Substance Batch number

96 h LC50, g/L

96 h IC50, g/L

NCC 8 >6 >610 >6 >6

CMC - 3-6 3-6

25

Table VII. Results of the rainbow trout hepatocyte assays.

EndpointNCC, mg/L CMC, mg/L

EC20 EC50 EC20 EC50

Cell viability 34 245 >2000 >2000Available zinc 171 > 2000 >2000 >2000Heat shock proteins 10 >2000 14 >2000Total sugars 5 23 >2000 >2000Lipid peroxidation 100 434 930 > 2000DNA strand break >2000 >2000 >2000 >2000

nd – no significant effects determined.

26

Figure Captions:

Figure 1. A 5% suspension of nanocrystalline cellulose (NCC).

Figure 2. Mean (± SEM; n=2) number of eggs produced by fathead minnow exposed to various concentrations of NCC.

Figure 3. Phase-contrast microscopy of primary trout hepatocytes upon 48 h exposure to different concentrations of carboxyl methyl cellulose (CMC) and NCC. Magnification: × 200.

Figure 4. Heat shock protein (HSP) levels in rainbow trout hepatocytes exposed to NCC and CMC.

27

28

Fig 1Figure 1.A 5% suspension of nanocrystalline cellulose (NCC).

0

30

60

90

Control 0.03 0.06 0.12 0.24 0.48

Concentration of NCC, g/L

Mea

n Eg

gs p

er F

emal

e

Fig 2

Figure 2. Mean (± SEM; n=2) number of eggs produced by fathead minnow exposed to various concentrations of NCC.

29

Figure 3. Phase-contrast microscopy of primary trout hepatocytes upon 48 h exposure to different concentrations of carboxyl methyl cellulose (CMC) and NCC. Magnification: X 200

30

0 16 80 400 2000

Concentration, mg/L

6000

10000

14000

18000

22000NCCCMC

HS

P 7

2, n

g/ce

ll de

nsity

0 16 80 400 2000

Concentration, mg/L

6000

10000

14000

18000

22000NCCCMC

HS

P 7

2, n

g/ce

ll de

nsity

Fig 4

Figure 4. Heat shock protein (HSP) levels in rainbow trout hepatocytes exposed to NCC and CMC.