Physical, chemical, and toxicological characterization of fibrillated forms of cellulose using an in vitro gastrointestinal digestion and co-culture model

Abstract

Fibrillated cellulose is a next-generation material in development for a variety of applications, including use in food and food-contact materials. An alternative testing strategy including simulated digestion was developed to compare the physical, chemical, and biological characteristics of seven different types of fibrillated cellulose, following European Food Safety Authority guidance. Fibrillated forms were compared to a conventional form of cellulose which has been used in food for over 85 years and has Generally Recognized as safe regulatory status in the USA. The physical and chemical characterization of fibrillated celluloses demonstrate that these materials are similar physically and chemically, which composed of the same fundamental molecular structure and exhibit similar morphology, size, size distribution, surface charge, and low levels of impurities. Simulated gastrointestinal and lysosomal digestions demonstrate that these physical and chemical similarities remain following exposure to conditions that mimic the gastrointestinal tract or intracellular lysosomes. A toxicological investigation with an advanced intestinal co-culture model found that exposure to each of the fibrillated and conventional forms of cellulose, in either the pristine or digested form at 0.4% by weight, showed no adverse toxicological effects including cytotoxicity, barrier integrity, oxidative stress, or inflammation. The results demonstrate the physical, chemical, and biological similarities of these materials and provide substantive evidence to support their grouping and ability to read-across data as part of a food safety demonstration.

Introduction

Cellulose is the most abundant natural polymer on earth, which widely found in plants where it provides support and structure to cell walls. Other natural sources of cellulose include invertebrates, algae, bacteria, and fungi [1]. Purified cellulose fibers are usually obtained from wood pulp and serve as a base material to create a variety of different morphological forms and functional derivatives [2–4]. Each of these cellulose derivatives has numerous commercial applications. The role of cellulose in the food industry is ubiquitous, most widely used as an anti-caking agent. Several cellulosic materials have been used in food for over a century and are classified as Generally Recognized as Safe (GRAS) by the US Food and Drug Administration.

Fibrillated cellulose is produced by mechanically processing cellulose fibers to release the structural building block: individual cellulose fibrils. Fibrillated cellulose has a fibrillar structure, consisting of an entangled network of fibers and fibrils of varying lengths and widths. This morphology gives fibrillated cellulose high tensile strength and enables high capacity thermal and mechanical applications [5]. The unique properties of fibrillated cellulose enable it to be used as a food additive, food coating, and in food-contact packaging materials. As a food coating, fibrillated cellulose can protect, produce during growth, and extend the shelf life. In food packaging, fibrillated cellulose is being developed as a coating, barrier, and additive in both paper- and plastic-based products [6–9].

Although fibrillated cellulose is expected to be as safe as conventional cellulose for food uses, efforts are underway to demonstrate the safety and ensure safe and responsible commercialization. A consortium of 10 organizations, including six industrial partners manufacturing fibrillated cellulose, has partnered with academic, governmental, and commercial labs to ensure the safe use of fibrillated cellulose in food. Most commonly, safety assessments for food additives rely on rodent feeding studies. As part of their efforts, the consortium commissioned 7-, 14-, and 90-day dietary studies (following Organisation for Economic Co-operation and Development (OECD) Test Guidelines 407, 408) of a non-commercial form of fibrillated cellulose (Ref FC) with a conventional and GRAS authorized cellulose material, Solka Floc® (Ref CC) [10]. No adverse effects were found following fibrillated or conventional cellulose exposure for 90 days, where up to 4% of the diet was comprised of cellulose.

Humans lack digestive enzymes to break down insoluble fibers such as cellulose, and therefore, similar to conventional cellulose, fibrillated cellulose is not expected to be metabolized in the mammalian gut and excrete in the feces. For this reason, studies examining celluloses interactions with the gastrointestinal tract was not completed with this phase of testing; however, as part of the 90-day study in rats, additional histological examinations of the gastrointestinal epithelium were completed and no signs of irritation or inflammation were documented due to either fibrillated or conventional cellulose exposure. Furthermore, future absorption, distribution, metabolism, and excretion studies are currently being planned for fibrillated cellulose as part of its safety demonstration.

It is desirable to avoid animal testing of the industrial forms of fibrillated cellulose for regulatory and ethical reasons. As part of its safety demonstration, the consortium commissioned the design of additional physical, chemical, and biological studies in vitro to demonstrate the similarity of six industrial forms of fibrillated cellulose (C20-C25) to the forms of cellulose which underwent 90-day animal testing (Ref FC, Ref CC) to provide substantive evidence to support their grouping and read-across. A read-across method is utilized to demonstrate the similarity of substances and allow toxicological data from one substance to be used for another substance as a basis for assessing safety. The designed testing strategy follows guidance released by the European Food Safety Authority (EFSA) and represents an accepted approach for screening-level testing of food ingredients (EFSA 2018). The strategy is intended to limit the need for excessive animal testing by demonstrating the safety of chemically similar materials. There is a need to implement an in vitro testing strategy and interpret results regarding the physical, chemical, and biological characterization of fibrillated cellulose.

Most in vitro gastrointestinal toxicity models use the heterogeneous human epithelial colorectal adenocarcinoma cell line, Caco-2 [11–13]. When cultured as a confluent cell population, Caco-2 cells express the characteristics of enterocytic differentiation, i.e. absorptive cells that function as an intact, but permeable barrier. However, use of this cell line as a representative model of the intestinal epithelium is limited due to the absence of additional cell types important in contributing to intestinal epithial structure and function such as mucus-producing or immune cells [14]. To address these shortcomings, a co-culture model of Caco-2 cells together with mucus-secreting goblet cells, HT29-MTX, and immune-active lymphocytes, Raji B which act to differentiated Caco-2 cells into microfold (M) cells, were used. Extensive characterization of this co-culture demonstrates that it recapitulates the complexity of the human gastrointestinal tract compared with Caco-2 cells alone [14]. The co-culture divulges biochemical, phenotypical, and functional effects of substance exposure to an intestinal epithelium. This model was used in the in vitro testing strategy to characterize a variety of toxicological endpoints following exposure to both pristine- and digested-forms of fibrillated cellulose, including cytotoxicity, membrane integrity, oxidative stress, and pro-inflammatory markers.

There are data gaps in the literature with respect to an extensive physical, chemical, and toxicological characterization of industrial forms of fibrillated cellulose intended for food-use in both the pristine form and following simulated gastrointestinal and lysosomal digestion [15]. Recent toxicological assessments of pristine and simulated digested cellulose materials have been performed using in vitro cell-based model systems alongside a characterization of their physical and chemical properties [15–17]. These studies have demonstrated the value of characterizing the physical, chemical, and toxicological properties of a substance in its pristine (undigested) form and comparing it to its digested form as part of a safety demonstration in food [18]. Some of the more relevant physical and chemical properties in particle and fiber toxicology that may impact toxicological outcomes include primary particle/fiber size, the size and surface charge of the material when it is suspended in the aqueous phase, and the crystallinity of the material. Thoroughly characterizing these properties for fibrillated cellulose is not only important as part of its safety demonstration but can also help elucidate how these properties may relate to biological responses (i.e. structure-activity-relationships) such as an effect on cellular uptake due to a lack of dispersion or tendency to agglomerate, or how these properties potentially influence cellulose interactions with biological targets such as membrane receptors [19, 20].

Materials and Methods

Experimental Overview

Figure 1 describes the experimental design used for this study. Reference Materials, Fibrillated Cellulose (Ref FC, University of Maine Process Development Center, Orono, ME, USA) and Conventional Cellulose (Ref CC, Solka Floc®, Solvaira Specialty LP, North Tonawanda, NY, USA), were supplied along with six industrially produced fibrillated cellulose materials (C20, C21, C22, C23, C24, and C25). Once received, each material underwent standardized sample dispersion, simulated gastrointestinal digestion (at 0.25, 0.5, 1, and 4 h), and simulated lysosomal digestion (at 0.5, 2, 24, and 72 h). Samples suspended following a standardized dispersion were labeled ‘pristine cellulose materials’; samples, which were digested through simulated gastrointestinal fluid, were labeled’; and samples treated with simulated lysosomal fluid incubation were labeled ‘lysosomal digested cellulose materials’. All 40 samples of pristine and digested cellulose materials and the 32 lysosomal digested cellulose materials were physically and chemically characterized with transmission electron microscopy (TEM), dynamic light scattering (DLS), and inductively coupled plasma-mass spectrometry (ICP-MS) techniques. The pristine and simulated gastrointestinal digested cellulose samples were then biologically tested in an in vitro gastrointestinal co-culture model (Caco-2, HT29-MTX, and Raji B) over four time points (1, 6, 24, and 48 h) and measured for metabolic activity, proinflammatory effects, oxidative stress and cell viability, and barrier integrity over 7 days of culture.

Figure 1.

Experimental design as described by the flow of samples throughout the study. Two reference (Ref FC, CC) and six fibrillated cellulose materials (C20-C25) were acquired and catalogued, dispersed in water in equivalent concentrations (2%) and labeled as ‘pristine cellulose materials’. Cellulose samples were digested in one of two simulated fluids (i.e. gastrointestinal or lysosomal). Pristine, gastrointestinal digested, and lysosomal digested cellulose materials were characterized for several physical and chemical properties. Both pristine and gastrointestinal digested samples were then exposed to an in vitro intestinal co-culture model and assessed for induced biological activity.

Materials

Solka Floc (grade FCC200, CAS number 9004-34-6) was purchased from Solvaira Specialty LP. Solka Floc is GRAS in the USA and sold commercially for food use. It serves as a conventional cellulose control group (herein Ref CC) to compare to the seven fibrillated forms of cellulose. Fibrillated cellulose is produced through mechanical diminution of wood pulp. Fibrillated cellulose that has undergone acute and sub-chronic (OECD TG 407, 408) testing in rats was purchased from the University of Maine Process Development Center and served as a reference material (Ref FC; 20% wt). Six industrially produced forms of fibrillated cellulose were provided by industrial partners (C20-C25) ranging from 8 to 67% wt.

Preparation of Cellulose Suspensions

Each sample was weighed out on an analytical balance (Mettler Toledo, Columbus, OH, USA) and transferred to a centrifuge tube. Water (MilliQ 18.2 Ohms) was added to create 2% cellulose suspensions. To ensure dispersion, the solution was vortexed using a Vortex Genie 2 (Scientific Industries, Bohemia, NY, USA) for 10 min prior to analyses. Some analyses required additional dilutions, which were performed in ultrapure water unless otherwise stated.

Preparation of Simulated Gastrointestinal Digested Cellulose and Physical and Chemical Characterization

Physical, chemical, and biological characterization of the eight materials before and after simulated in vitro gastrointestinal and lysosomal digestion followed guidance released by EFSA. In vitro gastrointestinal digestion exposed cellulose to chemical conditions, enzymes, and salts that mimic physiological conditions representative of those in the mouth, stomach, and intestinal compartment, simulating digestion along the gastrointestinal tract, following an internationally agreed upon simulated digestion model for food [17].

To produce the oral phase digestate, 20 gl of cellulose (dry matter content) was suspended in 75 U/ml of salivary alpha-amylase. Calcium chloride (CaCl₂) and ultrapure water were added to achieve a final concentration 0.75 mM. The resultant mixture was incubated for 2 min at 37°C. This mixture was diluted by half with gastric fluid digestate, which included 2000 U/ml of pepsin from porcine gastric mucosa supplemented with CaCl₂ in phospholipids (0.17 mM) with pH-adjusted to 3.0 with hydrochloric acid (HCl). The resultant mixture was shaken and incubated at 37°C for 2 h. This mixture was diluted by half with intestinal phase fluid, which included pH-adjustment to 7.0 with sodium hydroxide (NaOH) immediately followed with 100 U/ml of pancreatin supplemented with CaCl₂ and bile (10 mM). Fibrillated celluloses were incubated in the simulated intestinal fluid at 37°C for one of four timepoints: 15, 30 min, 1, and 4 h. For longer term storage, all simulated digested samples were stored at 4°C for up to 2 weeks.

Preparation of Simulated Lysosomal Digested Cellulose and Physical and Chemical Characterization

In vitro lysosomal digestion exposed cellulose materials to conditions within lysosomes, simulating intracellular digestion conditions. A suspension of artificial lysosomal fluid (ALF) was created to simulate the lysosomal digestion of cellulose materials. ALF has been developed for investigating the long-term durability of samples after phagocytosis into cells. Phagolysosomal pH of alveolar and interstitial macrophages is 4.5; the acidity of this intracellular fluid may explain the higher level of solubility of metals that are phagocytized versus those that remain extracellular [21]. To produce the lysosomal digestate, 20 g/l of cellulose (dry matter content) was suspended in a complex mixture of sodium chloride (NaCl, 3.21 g/l), sodium hydroxide (NaOH, 6.0 g/l), citric acid (20.8 g/l), calcium chloride (CaCl, 0.128 g/l), disodium phosphate (NaHPO₄, 0.071 g/l), sodium sulfate (Na₂SO₄, 0.039 g/l), magnesium chloride (MgCl₂, 0.106 g/l), glycerol (0.059 g/l), trisodium citrate (C₆H₅Na₃O₇, 0.077 g/l), sodium tartrate (C₄H₄O₆Na₂, 0.09 g/l), sodium L-lactate (C₃H₅NaO₃, 0.065 ml), sodium pyruvate (C₃H₅O₃Na, 0.086 g/l), and formaldehyde (1 ml) [22–25]. ALF was pH-adjusted to 4.5 and cellulose materials were incubated at 37°C for one of four timepoints: 30 min, 2, 24, or 72 h. For longer term storage, all simulated digested samples were stored at 4°C for up to 2 weeks.

Particle Size and Surface Charge Analysis

To determine the hydrodynamic diameter (HDD), dispersity index (DI), and zeta potential (ζ potential), DLS techniques were acquired with a Zetasizer Nanoseries Nano-ZS (Malvern Pananalytical, Almelo, Netherlands). A dilution of the stock solution was required for this analysis to ensure precision. Cellulose stock suspensions were diluted to 0.01% with ultrapure water and transferred to a disposable folded capillary cell DTS 1060 (Malvern Pananalytical, Almelo, Netherlands). For HDD and DI, each sample was scanned for 10 s, 11 times, in triplicate. A 173° backscatter angle was used in general purpose mode. For zeta potential measurements, the Helmholtz–Smoluchowski model was utilized at 25 runs, in triplicate, for each sample in auto report mode. To ensure accuracy, the mean count rate was greater than 1000 counts per measurement for each sample.

TEM Preparation and Imaging

Cellulose dispersions were diluted to 0.01%. Formvar carbon-coated copper grids 200 (EMS, Hatfield, PA, USA) were loaded with each sample for 5 min. Once dried, the fibrillated cellulose samples were imaged on a TEM (JEM-1010, JEOL Inc., Akishima, Tokyo, Japan). Images were taken at an accelerating voltage of 60 kV with a spot size of 2.0.

Scanning Electron Microscopy Preparation and Imaging

Similar to TEM preparation and imaging, cellulose dispersions were diluted to 0.01% for even loading across mounting pin stubs. Once dried, the conventional cellulose sample was imaged on a scanning electron microscope Versa 3D, FEI Focused Ion Beam (Field Electron and Ion Company, FEI, Hillsboro, OR, USA). Images were taken at an accelerating voltage of 30 kV.

Metal Impurities

Prior to analysis via ICP-MS (Agilent 7900, Santa Clara, CA, USA), sample preparation was performed. Samples were prepared by acid digestion in a hot block set at 95°C in a plastic digestion vessel with 1 ml of a 2% cellulose suspension. Nitric acid (HNO₃, 4 ml) and hydrochloric acid (HCl, 1 ml) were added to the vessel and a plastic watch glass was placed over the top. After 1 h, samples were removed and allowed to cool at room temperature, upon which additional HNO₃ was added to ensure complete digestion. This process was repeated in triplicate until cellulose samples were clear and colorless. The sample digests were placed in the autosampler and run in triplicate along with an internal standard which contained lithium, scandium, germanium, rhodium, indium, terbium, lutetium, and bismuth (Agilent, Santa Clara, CA, USA). Data acquisition and analysis was completed with MassHunter Software (Agilent).

Co-Culture Model Assembly

Each cell-type used in the co-culture model was first maintained as separate individual cultures. Briefly, human colon carcinoma Caco-2 cells (ATCC, Manassas, VA, USA) utilized at passages between 38 and 45, HT29-MTX (Sigma, St. Louis, MO, USA) mucus-producing, and human Burkitt’s lymphoma Raji B (ATCC) cell lines were obtained. Dulbecco’s Modified Eagle Medium/F-12 (DMEM/F12; 1:1), fetal bovine serum, 10,000 U/ml penicillin, 10 mg/ml streptomycin (MP Biomedicals, Santa Ana, CA, USA), and trypsin–EDTA (Invitrogen, Waltham, MA, USA) were used to maintain cultures. Transwell® polycarbonate inserts (12 wells, pore diameter of 4 μm polycarbonate) were purchased from Midland Scientific, Inc., Omaha, NE, USA.

Co-culture model assembly of Raji B, Caco-2, and HT29-MTX cells were seeded at a ratio of 9:9:1, respectively, with the Caco-2 and HT29-MTX cells on the apical chamber of Transwell® inserts and Raji B cells in the basolateral side [14]. The cells were maintained under identical condition to the monocultures with medium changes every other day. To assemble the layers, Raji B cells are first plated at a known seeding density with culture medium. The Transwell® is inserted and Caco-2 and HT29-MTX cells are seeded on the insert in ratios as listed. Total volume is increased to 2 ml (1.5 ml on the well; 0.5 ml on the insert) of DMEM/F12; 1:1 culture medium. Caco-2 cells take approximately 5 days to differentiate with Raji B exposure. The characterization of this advanced intestinal co-culture characterization has been previously reported [16].

Culture Exposure to Cellulose Materials

After ~ 70–80% confluency, cells were exposed to reference or industrially produced fibrillated cellulose materials (Ref FC, C20, C21, C22, C23, C24, C25, and Ref CC). Resuspended pristine (in ultrapure water; 2%) and digested (in simulated gastrointestinal fluid; 2%) cellulose were diluted to 0.4% in cell culture media (DMEM/F12 1:1) for exposure (i.e. in 24-well plate 500 μl of media that contains 100 μl of resuspended cellulose material). Cellulose was added into the apical chamber of Transwell® plate. Exposed cells were incubated for 1, 6, 24, or 48 h in 37°C 5% CO₂ incubator. Following incubation, the co-culture was washed with phosphate buffer solution (PBS), replaced with fresh media; cellular responses were determined with the following assays: Cytotoxicity/cell viability, gastrointestinal barrier integrity impairment, oxidative stress, and proinflammatory response markers. All toxicological assays were optimized and validated; the experimental design included negative control treatments (untreated), vehicle control treatments (either deionized water or simulated gastrointestinal fluid), and positive control treatments. For cytotoxic endpoints, rotenone was the positive control, a pesticide known to induce cytotoxicity through the induction of oxidative stress [26]. For inflammation, lipopolysaccharides (LPS) served as a positive control. LPS are found in the outer membrane of gram-negative bacteria and are known to induce pro-inflammatory mediator production and release [27]. Hydrogen peroxide (H₂O₂), a known inducer of oxidative stress, was used as a positive control treatment in flow cytometry experiments examining this endpoint.

Cytotoxicity/Cell Viability

Metabolic activity was measured using the MTS [(3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium)] assay (MTS, Madison, WI, USA). Raji B, Caco-2, and HT29-MTX cells were seeded separately and also in co-culture in a ratio of 9:9:1 and incubated for 24 h at 37°C in 5% CO₂ environment. Following experimental exposures, MTS solution was added to each cell culture per manufacturer’s direction [28, 29]. After 2 h incubation, the absorbance was measured at 490 nm on a plate reader (Synergy, BioTek, Winooski, VT, USA).

Gastrointestinal Barrier Integrity Impairment

Transepithelial/transendothelial electrical resistance (TEER) is an indicator of membrane integrity and permeability [30]. Data collected may be investigated to monitor cell culture confluence, monolayer formation, and epithelial barrier function [31].

Caco-2, HT29-MTX, and Raji B cells were seeded individually at 13,000 cells/cm² density on Transwell® inserts (membrane area 1.1 cm², pore size 4 μm, Corning®, Corning, NY, USA) in triplicate. The co-culture was seeded in triplicate as follows: Caco-2 cells at 6000 cells/cm² density in the Transwell® insert, Raji B cells at 6000 cells/cm² density in the Transwell®, and HT29-MTX cells at 1000 cells/cm² density were seeded on the top of the Caco-2 cells in the Transwell® insert. Cells were allowed to settle for 1 day. On day 2, cells were exposed to cellulose suspensions (pristine or digested, 0.4% final exposure concentration) or control treatments. TEER measurements were taken immediately after exposure and denoted as Day 2 AM. Six hours later, TEER was measured again and recorded as Day 2 PM. On day 3, culture media was aspirated and replaced with fresh media. TEER measurements were taken immediately after media exchange and denoted as Day 3 AM. TEER was measured again after 6 h (recorded as Day 3 PM), 18 h (Day 4 AM), and 24 h (Day 4 PM). Media was replenished again on day 5 and TEER was recorded twice a day through day 7.

TEER was measured twice daily for 7 days using an epithelial voltohmmeter (EVOM) with chopstick electrodes (World Precision Instruments, LLC, Sarasota, FL, USA). Each measurement was an averaged triplicate; each reading was from a different point in the same well. The electrical resistance of the cellular monolayer, measured in ohms, is a quantitative measure of barrier integrity, and the TEER measurements on the co-culture model may be interpreted to simulate gastrointestinal barrier integrity [32].

Pro-Inflammatory Response Markers

The pro-inflammatory response marker interleukin 6 (IL-6) was measured using an Enzyme-linked immunosorbent Assay (ELISA; Invitrogen, Carlsbad, CA, USA). Raji B, Caco-2, and HT29-MTX cells were seeded separately and also in co-culture in a ratio of 9:9:1 and incubated for 24 h at 37°C in 5% CO₂ environment. Following experimental treatments, cells were washed with PBS and lysed with RIPA Buffer (Invitrogen). Lysate was collected and centrifuged at 12,000 rpm. Lysated samples and standards were plated per manufacturer’s direction, and absorbance was read on a plate reader (Synergy H1, BioTek) at 450 nm (620 nm as a reference wavelength) [33, 34].

Oxidative Stress

For flow cytometric staining, a 20-μl volume of cells harvested from a co-culture previously exposed to pristine or digested cellulose (15, 30 min, 1, and 4 h) was added to 180 μl of 0.5 μM CellROX Green (ex/em: 508/525 nm; ThermoFisher) or 0.5 μM SyTox Red (ex/em: 640/658 nm; ThermoFisher) in PBS in 96-well V-bottom plates and incubated for 30 min at 37°C. Samples were analyzed within 2 h on a FACSVerse flow cytometer (BD Biosciences) equipped with a blue and red laser. The flow rate was set to 3.0 μl/s with a 150-μl injection volume, 100-μl mixing volume, 250-μl/s mixing speed, five mixes, and a wash volume of 800 μl. For each sample, 50,000–200,000 events were collected.

Statistical Analyses

For each data set, a two-way analysis of variance (ANOVA; alpha = 0.05) was performed [35]. ANOVA was also implemented to confirm the parametric two-way ANOVA results [36]. In order to confirm assessment, Tukey’s Honest Significant Differences was conducted. All statistical analyses were performed in Prism 8.3.0 (GraphPad, San Diego, CA, USA). For the biochemical assays, the study included four replicates which were performed in triplicate (n = 12).

Results

Physical and Chemical Characterization before and after Simulated Digestions

Electron microscopy micrographs show the morphology of fibrillated celluloses (Fig. 2A–G), consisting of an entangled network of fibers and fibrils of varying widths. In comparison, conventional cellulose (Fig. 2H) has a lower aspect ratio than fibrillated cellulose and does not form an entangled network of fibers. Generally, conventional cellulose has a morphology that is tens to hundreds of microns in length and width. ICP-MS results indicate low levels of metal impurities for all reference and fibrillated cellulose materials (0.03–5.86 ppb), in the low parts-per-billion range (Supplemental Fig. 1), and well below what was measured in tap water (9385.5 ppb).

Figure 2.

Morphology of the pristine cellulose materials used in this study as imaged by TEM: (A) Ref FC, (B) C20, (C) C21, (D) C22, (E) C23, (F) C24, (G) C25, and (H) Ref CC. The scale bar in the images represents 600 nm, unless otherwise noted.

Figure 3 shows representative electron micrographs of the digested cellulose materials. Figure 3A shows the Ref FC sample after 4-h incubation in simulated gastrointestinal digestion fluid and Fig. 3B shows the C22 sample after 4-h incubation in simulated gastrointestinal digestion fluid. Images provide visual evidence that fibrillated cellulose does not change after subjection to oral, stomach, and intestinal phase fluids when compared with the same cellulose material suspended in ultrapure water. Similar findings were observed when the cellulose materials were subjected to ALF.

Figure 3.

Representative images of the digested cellulose materials used in this study as imaged by TEM: (A) Ref FC and (B) C22. The scale bar in the TEM images represents 200 nm. The scale bar in the scanning electron microscopy image represents 100 μm.

The HDD of each reference and fibrillated cellulose material was measured using DLS in both pristine form as well as following simulated gastrointestinal (Fig. 4) and lysosomal digestion (Supplemental Table 2). The x-axis lists each of the reference and fibrillated cellulose materials and the time incubated in the final intestinal phase of the simulated gastrointestinal digestion (15, 30 min, 1, and 4 h). The y-axis corresponds with the HDD values normalized on a logarithmic micrometer scale. This normalization was necessary to produce comparable values among all fibrillated cellulose samples in comparison to the significantly larger (P < 0.05) Ref CC. No change in HDD was observed following simulated digestion and no significant difference in HDD was observed among any of the fibrillated forms of cellulose. Among the fibrillated forms, C22 had the largest HDD, while C25 had the smallest; however, these differences were not significant. Similarly, there was no statistical significance observed when comparing the HDDs of pristine versus simulated lysosomal digested cellulose samples (Supplemental Table 2).

Figure 4.

Average HDDs (± standard deviations) for reference and fibrillated celluloses (n = 6). Tabular data sets are included in Supplemental Table 1. (^*: P < 0.05 for Ref CC against all fibrillated forms for each digestion time period. No other statistical significance was observed.)

The DI of each reference and fibrillated cellulose material was measured using DLS (Fig. 5). The x-axis lists each of the reference and fibrillated cellulose materials and the time incubated in the final intestinal phase of the simulated gastrointestinal digestion (15, 30 min, 1, and 4 h). The y-axis corresponds with the unitless DI values. Among all reference and fibrillated cellulose materials, no significant differences in dispersity existed. All eight of the cellulose materials have broad size distributions, characterized by DI > 0.4. The DI for Ref FC and Ref CC was 0.55 and 0.65, respectively, while C20-C25 ranged from 0.53 to 0.92 (Fig. 5; Supplemental Tables 1 and 2). C22 and C25 had the lowest DI among fibrillated materials although the differences were not significant. No statistical significance was observed when comparing the DIs of pristine versus simulated gastrointestinal digested cellulose samples. Similarly, there was no statistical significance observed when comparing the DIs of pristine versus simulated lysosomal digested cellulose samples (Supplemental Table 2).

Figure 5.

DIs (DI; ± standard deviations) for reference and fibrillated celluloses (n = 6). Tabular data sets are included in Supplemental Table 1. No statistically significant difference in DI was observed.

The zeta potential of each reference and fibrillated cellulose material was measured (Fig. 6), with the x-axis listing each of the reference and fibrillated cellulose materials and the time incubated in the final intestinal phase of the simulated gastrointestinal digestion (15, 30 min, 1, and 4 h). Zeta potential values (mV) are shown on the y-axis. The zeta potential for all fibrillated forms of cellulose (Ref FC, C20-C25) was negative, as expected from previous literature reports [37]. The smallest zeta potentials were measured for Ref CC, and the largest zeta potentials were measured for fibrillated material C20. There were no differences in zeta potential measurements of pristine versus simulated gastrointestinal digested cellulose samples (P < 0.05). Similarly, there was no statistically significant difference in zeta potential of pristine versus simulated lysosomal digested cellulose samples (Supplemental Table 2).

Figure 6.

Average zeta potentials (± standard deviation) for reference and fibrillated cellulose materials (n = 6). Tabular data sets are included in Supplemental Table 1. (^*: P < 0.05 Ref FC against all fibrillated forms for each digestion time period. No other statistical significance was observed).

Cytotoxicity/Cell Viability

Metabolic activity was measured using MTS absorbance after 1-, 6-, 24-, and 48-h exposures to pristine and digested cellulose (Fig. 7). Figure 8A shows metabolic activity in control cells. Negative and vehicle control groups included untreated cells in media-only and cells that were exposed to deionized water (vehicle control for all pristine cellulose materials) or simulated gastrointestinal fluid (vehicle control for all digested cellulose materials). Rotenone (50 μM) was used as a positive control for induced metabolic activity as an indicator of mitophagy and apoptosis. Figure 8B–I shows the metabolic activity of co-cultures exposed to reference and fibrillated cellulose, in pristine and 4-h digested form, for four time points. All exposures demonstrate a similar trend of increasing MTS absorbance over time (1–48 h), expected with viable growth of the co-culture. Exposure to cellulose materials had no impact on cellular viability. There was no significant change in MTS absorbance between Ref FC, C20-C25, and Ref CC compared with their corresponding vehicle control treatments. Similarly, no significant change in MTS absorbance was observed between cells exposed to pristine versus digested forms, for all cellulose materials, for all timepoints (P < 0.05).

Figure 7.

(A) Depiction of the co-culture model used in this study. Cells were seeded at a seeding density ratio 9:9:1 and were maintained in standard cell culture conditions (5% CO₂, 10% humidity) for 2 days. (A) Brightfield micrograph showing the cellular morphology of the co-culture used in the study. Brightfield micrographs of (C) Ref FC and (D) Ref CC. Darkfield micrographs of (E) Ref FC and Ref CC. Scale bar represents 100 μm.

Figure 8.

The effects of pristine and digested cellulose reference and fibrillated materials on co-cultured gastrointestinal model, measured by metabolic activity. MTS absorbance in cells exposed to (A) Controls [untreated cells, deionized water (pristine cellulose vehicle control), simulated gastrointestinal fluid (digested cellulose vehicle control), and 50 μM rotenone (positive control)]; and pristine and digested (B) Ref FC, (C) C20, (D) C21, (E) C22, (F) C23, (G) C24, (H) C25, and (I) Ref CC. Data for cellulose that have undergone 4-h intestinal digestion are shown. (^*: P < 0.05 for rotenone exposure compared to Ref FC exposure for each time period. No other statistical significance was observed). Sample included four replicates performed in triplicate (n = 12).

Gastrointestinal Barrier Integrity Impairment

Barrier integrity was measured with TEER readings over 1-week following a 24-h exposure period to reference and fibrillated cellulose (both pristine and digested forms). Figure 9A shows barrier integrity over 7 days in untreated cells, and cells exposed to vehicle controls (deionized water, the vehicle control for all pristine cellulose materials, or simulated gastrointestinal fluid, the vehicle control for all digested materials). Figure 9B–I show TEER following exposure to fibrillated and conventional cellulose materials (pristine and 4-h digested) and demonstrates consistent results among all exposures (vehicle controls, pristine, and digested cellulose materials). Fluctuations between morning and evening measurements were observed. Cells were dosed the evening of the second day (Day 2 PM). The observed dip is caused by the addition of cellulose material, which momentarily interferes with the electrical resistance measurement but is not reflective of actual changes in cell co-culture barrier integrity this is hypothesized due to the change in uniformity of density with the addition of cellulose material [32]. Furthermore, the overall trend of increased resistance over time in each individual data set is similar among pristine versus digested cellulose exposure groups in the study.

Figure 9.

The effects of pristine and digested cellulose reference and fibrillated materials on co-cultured gastrointestinal model barrier integrity as measured by TEER. TEER measurements in cells exposed to (A) controls [untreated cells, deionized water (pristine cellulose vehicle control), simulated gastrointestinal fluid (digested cellulose vehicle control), and 50-μM rotenone (positive control)]; and pristine and digested (B) Ref FC, (C) C20, (D) C21, (E) C22, (F) C23, (G) C24, (H) C25, and (I) Ref CC. Data for cellulose that have undergone 4 h intestinal digestion are shown.

Pro-Inflammatory Response

Proinflammatory response was measured with IL-6 expression following 1-, 6-, 24-, or 48-h exposures to reference and fibrillated cellulose materials (both pristine and 4-h digested forms). Figure 10A shows IL-6 responses in co-cultures exposed to media-only (negative controls), deionized water exposure (vehicle control for pristine cellulose materials), simulated gastrointestinal fluid (vehicle control for digested cellulose materials), and a positive control for IL-6, LPS. The results demonstrate minimal background expression of IL-6 in negative and vehicle control treated cells. Experiments were repeated for the Ref FC (Fig. 10B) and Ref CC (Fig. 10D) materials. IL-6 expression was higher immediately following exposure (1- and 6-h time points) compared with later time points for most materials and control treatments. One exception was noted for Ref CC in its pristine form, which had increased IL-6 concentration at the 48-h post exposure timepoint. However, the IL-6 concentration is <1.0 pg/ml which is insignificant compared with the positive control. A standard curve was completed to verify results. Fibrillated cellulose materials (Fig. 10C) after 24-h post-exposure had minimal pro-inflammatory responses, which are similar to background and vehicle control activity (Fig. 10A). Furthermore, no difference was observed in IL-6 expression between most of the pristine versus digested forms of each fibrillated cellulose material tested, except samples C21 and C22. However, the IL-6 expression in co-cultures exposed to all forms of reference (Ref FC, Ref CC) and fibrillated cellulose (C20-C25) in both pristine and digested forms after 24 h were lower than the corresponding negative and vehicle control groups.

Figure 10.

The effects of pristine and digested cellulose reference and fibrillated materials on co-cultured gastrointestinal model as measured by proinflammatory response via enzyme-linked immunosorbent assay (ELISA). IL-6 expression of cells exposed to (A) controls [untreated cells, deionized water (pristine cellulose vehicle control), simulated gastrointestinal fluid (digested cellulose vehicle control), and 50-μM LPS (positive control) plotted on a log base 10 scale; and pristine and digested (B) Ref FC; (C) C20, C21, C22, C23, C24, and C25; and (D) Ref CC. IL-6 expressions of cells after exposure to C20 through C25 for 24-h incubation period, are shown. Data for cellulose that have undergone 4-h intestinal digestion are shown. (^*: P < 0.05 for LPS exposure compared to Ref FC exposure for each time period. No other statistical significance was observed). Sample included four replicates performed in triplicate (n = 12).

Oxidative Stress

Figure 11 shows the oxidative stress status of gastrointestinal co-cultures exposure to pristine and digested cellulose materials, assessed via flow cytometry. Results from control treatments, including untreated cells, cells exposed to 100 mM H₂O₂ (positive control), deionized water (pristine cellulose vehicle control), or simulated gastrointestinal fluid (digested cellulose vehicle control) are shown in Fig. 11A. Over 90% of the cell populations stained as alive with no oxidative stress for untreated cells, cells exposed to deionized water, and cells exposed to simulated gastrointestinal fluid (15, 30 min, 1, and 4 h). In contrast, ~ 80% of cells exposed to hydrogen peroxide stained as alive but oxidatively stressed. At longer incubation times (1 and 4 h), approximately 4 and 5% of the cell population were assessed as dead, respectively. In addition to controls, cells were exposed to Ref FC, the six unique fibrillated cellulose materials, and Ref CC in both pristine and digested forms (Fig. 11B–D). In all exposure scenarios, over 90% of all cell populations were assessed as alive with no oxidative stress over all time periods used in the study. Following exposure, none of the cellulose materials resulted in the measurable amounts of oxidative stress.

Figure 11.

The effects of pristine and digested cellulose reference materials on oxidative stress and cell viability via flow cytometry. Bar graphs show the populations of cells after exposure to (A) Controls [untreated cells, 100-mM hydrogen peroxide (H₂O_2; positive control), deionized water (pristine cellulose vehicle control), and simulated gastrointestinal fluid (digested cellulose vehicle control)]; and pristine and digested (B) Ref FC; (C) C20, C21, C22, C23, C24, and C25; and (D) Ref CC. Results of cells after exposure to C20 through C25 for 1-h incubation period are shown.

Discussion

In this study, we compared the physical, chemical, and toxicological properties of six unique fibrillated cellulose materials (C20, C21, C22, C23, C24, and C25) against those of a reference fibrillated cellulose material (Ref FC) and a reference conventional cellulose (Ref CC). After each material was dispersed in water at equivalent concentrations, fibrillated celluloses (C20–C25) were found to be similar to Ref FC in terms of morphology, HDD, DI, zeta potential, and impurities. Although the same molecular composition, morphologically, fibrillated celluloses (C20–C25) are physically dissimilar to Ref CC. Conventional cellulose is a lower aspect-ratio material that is over an order of magnitude larger in diameter compared with fibrillated forms. These data demonstrate that fibrillated celluloses (C20–C25) are both physically and chemically similar to Ref FC, suggesting that these materials may induce similar biological behavior. To test this hypothesis, fibrillated celluloses and their corresponding reference materials were subjected to a tiered toxicology testing strategy.

Physical and chemical characteristics of pristine cellulose materials were compared against the characteristics of materials which underwent simulated gastrointestinal and lysosomal digestion. Generally, digested materials were similar in physical and chemical disposition to their pre-digested forms for both simulated lysosomal and gastrointestinal digestion. This suggests that the simulated digestion process did not significantly change the characteristics of these materials. The testing strategy developed here includes a dosing regimen coupled with a sample preparation methodology simulating realistic exposure conditions of gastrointestinal and lysosomal digestion of cellulose material occurring along the gastrointestinal tract. The results demonstrate the physical, chemical, and biological similarities of these materials and provide substantive evidence to support their grouping and read-across as part of a food safety demonstration.

While measuring the HDD, DI, and zeta potential of pristine and digested cellulose was accomplished with relative ease, other aspects of characterizing cellulose-based material post-digestion proved challenging. Guidance released by the EFSA to demonstrate the safety of novel materials in food and feed recommends characterization of materials post-digestion with electron microscopy; however, this proved challenging due to the cellulose being significantly diluted through the simulation protocols. Of the TEM images collected, image analyses suggest that there is no change in morphology post-digestion. Furthermore, collecting fibril length measurements was not possible for fibrillated material due to the highly entangled network of fibers and digestion additives. The difficulty of imaging and measuring fibrillated cellulose is well documented in the literature [20].

The similar physical and chemical characteristics of fibrillated forms of cellulose, both pre- and post-digestion, suggest that the materials may induce similar biological responses; the in vitro toxicity assessments reported here demonstrate this. Fibrillated and conventional celluloses, in both the pristine form as well as after simulated intestinal digestion, did not induce any adverse effects in an intestinal co-culture model. Cells were exposed to 2% cellulose (final exposure concentrations up to 0.4% by weight) in ultrapure water or simulated digestive fluid and assessed up to 48 h. Neither induction of cytotoxicity, oxidative stress, and barrier integrity impairment nor inflammatory response was observed. These toxicological data demonstrate the similarity of these materials in two ways: (i) fibrillated materials (C20, C21, C22, C23, C24, and C25) have similar biological responses as the reference materials (Ref FC and Ref CC) and (ii) pristine forms of cellulose materials have similar biological responses to their digested forms. These results are supported by other simulated digestion and in vitro studies examining fibrillated cellulose used a gastrointestinal tract simulator to digest fibrillated cellulose (0.75 and 1.5% w/w). Exposure of digested fibrillated cellulose to a triculture model of the intestinal epithelium found no significant effects on cytotoxicity, reactive oxygen species, or monolayer integrity [16].

Conclusion

Taken together, the physical, chemical, and toxicological data sets demonstrate that fibrillated celluloses (C20-C25 and Ref FC) are physically, chemically, and biologically similar, and these similarities remain following simulated gastrointestinal digestion. These results provide substantive evidence to support their grouping and read-across of available toxicity data (e.g. 90-day dietary study) to all fibrillated forms examined here. Furthermore, despite some differences in the physical and chemical properties between fibrillated cellulose (C20-C25 and Ref FC) and conventional cellulose (Ref CC), all cellulose materials behave similarly in the intestinal co-culture model. These data suggest that despite differences in morphologies, cellulose materials behave similarly biologically being composed of the same fundamental molecular structure and have no observable cytotoxic or immune effects at concentrations up to 0.4% by weight. This conclusion is supported by recent animal testing (OECD 407, 408) which found no adverse outcomes in rats fed either Ref FC or Ref CC for 90-days up to 4% of the diet [37]. The fibrillated forms of cellulose are demonstrated to have a similar safety to conventional cellulose materials such as Solka Floc, which has been used as a food additive for well over 100 years and is GRAS in the USA.

Funding

The authors would like to thank the Department of Environmental Science and Vireo Advisors, LLC (grant # 32370206) and the funding partners of the Food Safety Study Consortium, including the public private partnership between the US Endowment for Forestry and Communities and the USDA Forest Service Forest Products Laboratory for financial support of this work.

Conflict of Interest Statement

The authors have no conflicts of interest.