Alan
Mackie
*a,
Simon
Gourcy
b,
Neil
Rigby
ac,
Jonathan
Moffat
d,
Isabel
Capron
e and
Balazs
Bajka
cf
aSchool of Food Science and Nutrition, University of Leeds, Leeds, LS2 9JT, UK. E-mail: a.r.mackie@leeds.ac.uk
bUniv Angers, Inst Univ Technol, F-49016 Angers, France
cInstitute of Food Research, Norwich Research Park, Norwich, NR47UA, UK
dAsylum Research, an Oxford Instruments Company, High Wycombe, HP12 3SE, UK
eINRA, Biopolymeres Interact Assemblages UR1268, F-44316 Nantes, France
fDepartment of Nutritional Sciences, King's College London, London, SE1 9NH, UK
First published on 30th January 2019
It is well recognised that the average UK diet does not contain sufficient fibre. However, the introduction of fibre is often at the detriment of the organoleptic properties of a food. In this study on the gastrointestinal fate of nanoparticles, we have used cellulose nano-crystals (CNCs) as Pickering stabilising agents in oil in water emulsions. These emulsions were found to be highly stable against coalescence. The CNC and control emulsions were then exposed to simulated upper gastrointestinal tract digestion and the results compared to those obtained from a conventional protein stabilised emulsion. Finally the digested emulsions were exposed to murine intestinal mucosa and lipid and bile absorption was monitored. Importantly, the results show that the CNCs were entrapped in the intestinal mucus layer and failed to reach the underlying epithelium. This entrapment may also have led to the reduced absorption of saturated lipids from the CNC stabilised emulsion versus the control emulsion. The results show the potential of CNCs as a safe and effective emulsifier.
The use of particles to stabilise emulsion droplets has been known for a long time and was initially described by Pickering.9 As a result, such colloidal systems have become known as Pickering emulsions. They have a number of advantages over conventional emulsions associated with high stability against coalescence. This is in part because of the high energy associated with displacing the particles from the interface.10 For large particles, the energy may be of the order of 1000 kT depending on the conditions. This is in contrast small molecule surfactants where the energy required may only be a few kT. In addition to the high energy required to remove them from the interface, the surface properties of the particles can dictate the type of emulsion formed. Thus, for largely hydrophilic particles where the contact angle (θ) at the interface is small (θ < 90°) the particles drive the formation of oil in water emulsions. The large thickness of the particle coating then provides robust steric stabilisation against coalescence. In the food arena there are also a number of Pickering stabilised emulsion and foam examples where the particles may be fat crystals or protein aggregates.2 In each case, the properties of the particles determine the properties of the colloid.
Cellulose nanocrystals (CNCs) have been studied since the 1950s but have recently regained interest as a Pickering stabilising agent.11 Cellulose is very widely available from many different sources and can be milled or otherwise processed into nanoparticles with relative ease. The use of rods as Pickering agents has been studied theoretically,12 showing the importance of the aspect ratio of the rods and the size of the emulsion droplets in the precise form of the 2-D structures formed by the rods. This has implications for tuneable delivery systems. Indeed, there are many examples in the literature of CNCs used to stabilise emulsions.13,14
Although colloidal systems are widely used in the food industry because of their extremely acceptable organoleptic properties, they have also been associated with highly processed food. This represents a challenge due to the effectiveness of such multi-phase systems ability deliver nutrients to the body. As a result, there is now interest in assessing ways of making such systems healthier, such as the addition of dietary fibre. Soluble fibres are normally considered the most effective as they can delay nutrient absorption through a number of different mechanisms. However, because of the complex nature of the digestive process, there are several other systems that have potential. For example, the use of dietary fibre is one tool that can be used to lower risk factors for cardiovascular disease and type 2 diabetes mellitus.15 Several studies have shown efficacy for a range of different dietary fibres. For example, the cereal dietary fibre β-glucan has been shown to lower cholesterol. Although the precise mechanism is not known, it is thought to involve the sequestering of bile acids, probably through entrapment of mixed micelles.
The aim of this study was to determine the fate of cellulose nanoparticles in the upper GI tract. For this purpose we formed cellulose nano-crystal (CNC) stabilised Pickering emulsions and exposed them to simulated upper GI tract digestion. This was followed by exposure to murine intestinal tissue where the absorption of both free fatty acids and bile acids was determined. Using this approach, we aimed to assess the fate of the CNCs following ingestion and measure the effect of the cellulose on the absorption of bile and lipid.
![]() | ||
Fig. 1 An AFM measurement, topography (left) and phase (right), of typical CNC's deposited from aqueous solution onto mica and imaged under air. |
All solid phase extraction steps were performed with flow under gravity except activation and brief drying steps. The chloroform extract was applied to a Bond Elut solid phase extraction cartridge (amino propyl, 500 mg, Agilent UK) immediately previously activated with 5 ml of hexane and very briefly dried under vacuum before sample application. The cartridge was eluted with 2 × 1 mL of chloroform:
isopropanol (2
:
1 v/v) under gravity using the mixture to rinse out the sample tube and to wash the walls of the SPE cartridge to ensure complete loading of the sample. The column was eluted with a further 2 mL of chloroform
:
isopropanol (2
:
1 v/v), these fractions were combined and contained the neutral lipids. The cartridge was then eluted with 4 mL of chloroform
:
methanol
:
acetic acid (100
:
2
:
2 v/v), this fraction contained the free fatty acids. The solution of neutral lipids was evaporated to dryness at 40 °C (Genevac) and dissolved in 0.5 ml of hexane while warming at 40–55 °C in a water bath to ensure complete dissolution of any lipid crystals formed during the evaporation. The mixture was applied to a second amino propyl cartridge activated as before. A further 0.5 mL of hexane was used to dissolve any remaining lipid and wash the walls of the SPE cartridge to ensure complete loading before it was developed with a further 3.0 mL of hexane followed by 4.0 mL of hexane
:
dichloromethane
:
chloroform (88
:
10
:
2 v/v), these extracts were combined and contained the triglyceride components. The cartridge was washed with 4.0 mL of hexane
:
ethyl acetate (95
:
5 v/v) followed by 4.0 mL of hexane
:
ethyl acetate (85
:
15 v/v) and the washings discarded. The cartridge was eluted with 4.0 mL of hexane
:
ethyl acetate (20
:
80 v/v), this fraction contained the diglyceride components. The monoglyceride components were then eluted with 4.0 mL of chloroform
:
methanol (2
:
1 v/v). After evaporation to dryness at 40 °C (Genevac), the lipid fractions were dissolved in 0.5 ml of toluene (containing 0.1 mg ml−1 butylated hydroxytoluene, Sigma) in screw-capped glass tubes equipped with Teflon lined caps. Methanol (1.0 ml containing 2% sulphuric acid (Sigma)) was added and the tubes were flushed with nitrogen gas before being incubated overnight at 50 °C in a shaking incubator at 150 rpm (Incu-Shake Max, SciQuip, UK). Following cooling the solution was neutralised by the addition of 1.0 ml of 0.25 M potassium bicarbonate, 0.5 M potassium carbonate solution and mixed for 30 seconds. The mixture was extracted with 1.0 ml of hexane and following centrifugation (100g, one minute) the supernatant was collected and evaporated to dryness before being redissolved in 100 μl hexane and transferred to a vial for subsequent analysis.
FAMES were analysed on a Varian CP 3800 GC equipped with a split-splitless injector and a Saturn 2000 single quad MS detector (Agilent, UK) fitted with a 100 m SP2560 column, 0.25 mm ID, 0.2 μm film thickness (Sigma, UK). 1.0 μL of sample was injected, the injector was held at 240 °C, splitless for 1.5 minutes then split at 50:
1 for 4.5 minutes before the split was reduced to 10
:
1 for the remainder of the separation. The column oven was held at 50 °C for 2 minutes before ramping to 150 °C at 100 °C per minute, held for one minute, ramped to 240 °C at 4 °C per minute before being held at this temperature for the remainder of the separation. Response factors were determined by analysing a gravimetric FAME standard (Supelco 37 Food FAME Mix (Sigma)).
Having produced a stable emulsion, it was then exposed to simulated gastrointestinal digestion based on the standard Infogest conditions.17 The initial stage of digestion was 2 hours of gastric simulation at pH 3. Under these conditions, the CNC emulsion was less flocculated than the initial emulsion as shown in Fig. 3A and droplets size was unchanged. The confocal micrograph in Fig. 3B shows the emulsion droplets surrounded by a heterogeneous layer of CNCs demonstrating the propensity of the CNCs to aggregate, even in the surface layer of the droplets. In contrast, there was a significant, and rapid decrease in droplet size once the emulsion was passed into the intestinal phase of digestion. Prolonged exposure to the intestinal conditions digested essentially all the triglyceride emulsion leaving only particles of a few hundred nanometres. This is confirmed from the image in Fig. 3C, showing that the lipid droplets were no longer surrounded by a shell of CNCs but were instead surrounded by a variety of self-assembled structures formed from lipid hydrolysis products, primarily fatty acids and monoglycerides. The CNCs that are visible in the image are clearly in aggregates that are no longer associated with the lipid droplets. Indeed, some of the CNCs associated with the 10 μm droplet in the bottom right of the image can be seen in the process of being removed from the droplet interface as a coherent sheet. The image also highlights the propensity of the lipid hydrolysis products to form lamella structures.
![]() | ||
Fig. 3 The surface area weighted size distribution of the CNC emulsion through digestion. The confocal images are of samples taken after the gastric (B) and intestinal (C) phases of digestion. |
In order to determine the fate of the CNCs once exposed to the intestinal mucosa, Ussing chambers containing murine intestinal mucosa had the apical chamber filled with a four-fold diluted sample of control or CNC emulsion following in vitro digestion. The images in Fig. 4 show the intestinal mucosa after 90 minutes exposure to CNCs. It is clear that the CNCs were trapped in the mucus layer, as there is no evidence of calcofluor staining in the mucosal tissue (Fig. 4A). Indeed, at higher magnification, there is little evidence that the CNCs were able to penetrate the mucus layer to any extent (Fig. 4B).
In addition to determining the final location of the CNCs, the impact of the CNCs on the transport and absorption of bile acids (BA) and free fatty acids (FFAs) was also measured. This was achieved by measuring concentrations in both the apical and basolateral chambers as a function of time following in vitro digestion of the control and CNC stabilised emulsions. Epithelial barrier integrity was maintained throughout the experimental period as demonstrated trans-mucosal resistance (Fig. 5A) and no significant differences were observed between control and CNC groups. Fig. 5B shows bile absorption by ileal segments as the decrease in concentration over the length of the experiment. By 30 min, significantly less bile acid had been absorbed in samples containing the digested CNC stabilised emulsion compared to the control. No difference was observed in the basolateral buffer samples. This is most likely because values were at or below the limit of detection (data not shown) and any bile remaining in the tissue samples was not measured.
The amount of individual fatty acids present as a proportion of the total fatty acid pool is outlined in Table 1. Overall, there was no significant difference in the total FFAs between the control and CNC stabilised emulsions even though a trend toward lower absorption was observed in Fig. 6B. Individually, Linoleic acid (C18:2) showed a significant proportion remaining on the apical side of the tissue after 90 minutes with the control being more than 10% lower. While this fatty acid made up the highest proportion of the total FFA profile, little was absorbed across the intestinal barrier and although there were differences in the basolateral proportions observed, these were not significant due to the low concentrations involved. However, significantly less stearic acid (C18:0) and eicosanoic acid (C20:0) were absorbed from the CNC emulsion following the 90 min incubation. This result suggests that CNC stabilisation of the emulsions may differentially affect absorption of saturated (Fig. 6A) and unsaturated (Fig. 6B) FFAs with inhibition of the absorption of saturated fatty acids.
Apical | Basolateral | |||||||||
---|---|---|---|---|---|---|---|---|---|---|
Control | CNC | p-Value | Control | CNC | p-Value | |||||
Mean | SEM | Mean | SEM | Mean | SEM | Mean | SEM | |||
C16:0 | 6.856 | 0.268 | 6.694 | 0.229 | 0.328 | 1.475 | 0.275 | 0.884 | 0.186 | 0.11 |
C16:1 | 0.504 | 0.072 | 0.591 | 0.176 | 0.346 | 0.182 | 0.116 | 0.281 | 0.134 | 0.608 |
C18:0 | 4.152 | 0.185 | 4.048 | 0.145 | 0.332 | 0.879 | 0.158 | 0.411 | 0.069 | 0.022 |
C18:1 | 30.89 | 1.351 | 29.21 | 0.983 | 0.169 | 7.924 | 0.729 | 6.224 | 1.616 | 0.41 |
C18:2 | 44.69 | 1.062 | 50.02 | 1.753 | 0.023 | 0.931 | 0.285 | 0.769 | 0.361 | 0.667 |
C20:0 | 0.334 | 0.088 | 0.258 | 0.067 | 0.253 | 0.966 | 0.061 | 0.435 | 0.128 | 0.011 |
Total | 87.426 | 1.219 | 90.821 | 1.917 | 0.105 | 12.357 | 1.219 | 9.004 | 1.917 | 0.211 |
The surface area weighted size distribution and confocal imaging demonstrates that the CNC stabilised emulsions resist digestion through the gastric phase. In contrast to more conventional protein stabilised emulsions,22 CNC stabilised emulsions showed reduced flocculation in gastric conditions although this is unlikely due to changes in the electrostatic interactions as the charge is unlikely to have changed.23 However, the shift in particle size following the addition of pancreatin (which includes lipase) in the intestinal phase suggests the lipid within the emulsion was readily available for hydrolysis. Thus, there were some areas of the lipid droplets that were not protected by a CNC coating, allowing bile and more importantly, the lipase/co-lipase complex to access the oil/water interface. This is consistent with the 3D confocal reconstructions demonstrating areas of no CNC coverage and previous studies that suggest 40% coverage with un-sulphated CNCs.16 However, it is at odds with previous studies using CNCs and whey protein that were able to retard lipolysis.24 Indeed, in this case, significant digestion has occurred by 15 min and at 120 min, virtually all triglycerides had been hydrolysed with the mean particle size dropping to 100 nm, likely comprising CNC aggregates and larger self-assembled structures (the primary peak in Fig. 3A), and mixed micelles. Both the size distribution and the micrographs have shown that the CNCs are likely to form discrete aggregates in the intestinal lumen that not associated with lipid droplets. These data suggest that CNCs are an effective material for stabilising lipid emulsions, allowing survival in a gastric environment but disassembly under small intestinal conditions. Thus allowing effective delivery of lipid and lipid soluble compounds. Subsequently, the fate of the CNCs in the gut lumen is as aggregated clusters of crystals with a size of circa 100 nm.
In order to interact with the gut wall, particles must pass through the intestinal mucus layer. Murine tissue explants were used to ascertain the fate of the CNCs after in vitro digestion and their influence on lipid and bile absorption in the small intestine. Ussing-type perfusion chambers were used to determine permeability across the intact intestinal mucosal barrier, including the mucus layer. This study provides the first clear evidence that CNCs become trapped in the intestinal mucus layer post-digestion and are unable to reach the underlying enterocytes. Thus the proposed routes of absorption of the CNCs envisaged by others,25 such as through M-cells, through enterocytes by passive diffusion, through enterocytes by transcytosis or through the paracellular route, seem not to be realised.
Recent evidence has demonstrated that dietary polysaccharides can alter cholesterol and lipid homeostasis, potentially through the sequestration of bile acids, inhibiting their recycling and decreasing permeability of the mucus layer, slowing diffusion of digestion products self-assembled into mixed micelles. The cholesterol-lowering potential of dietary fibre is currently thought to involve the sequestering of bile acids, either through a direct interaction or as a result of increased luminal viscosity and entrapment of mixed micelles. Recent evidence demonstrates inhibition of fat digestion in vitro,26 potentially through interactions between water-soluble dietary fibre and pancreatic lipase. While soluble cellulose esters have demonstrated significant bile acid binding in vitro,27 less is known about the functional properties of the insoluble fibres, such as cellulose, within the gut environment. While this study demonstrates a decrease in bile absorption in tissue explants from the murine ileum, the mechanism and potential interaction with intestinal mucus remain unclear.
The hydrolysis of dietary fat is extremely efficient in the mammalian small intestine, but the range of digestion products is extremely diverse.28 However, all of the products readily self-assemble into structures of different size and only the smallest can diffuse through the mucus layer and be absorbed. While the mean micelle radius of sunflower oil following in vitro digestion is 3.05 nm ± 0.05 nm (ref. 29) they can also vary in size depending on the concentrations of saturated and unsaturated free fatty acids, bile salts, cholesterol, etc. Indeed, decreasing bile concentrations significantly increases micelle size30 and this will affect diffusion through the mucus layer.
Although saturated fatty acids only account for approximately 11% of the total fatty acid pool in this study, the data provides the first evidence that an insoluble dietary fibre may differentially regulate absorption of saturated and unsaturated free fatty acids in the small intestine, although further studies will be required to validate this finding. Early studies investigating lipid digestion demonstrated slower absorption of saturated compared to unsaturated fatty acids.31 The selective absorption may relate to the differences observed in bile acid uptake and/or the size of the mixed micelles formed during digestion through altered micelle packaging. Alternatively, several studies have demonstrated interactions between several types of dietary fibre and the mucus layer.32–36 It is proposed that the fibre could block the pores in the mucus network altering mucus permeability and reducing lipid micelle diffusion. The micrographs (Fig. 4) demonstrate that CNCs become trapped by the mucus layer and this may reduce its permeability to mixed micelles thereby reducing diffusion in a similar way to soluble fibres.37,38 If so, the fibre decreases the effective pore size of the mucus and limits diffusion of particulate species. These include mixed micelles, which are self-assembled structures of ≥10 nm and so are most likely to be affected. The mass transfer of other macronutrients such as peptides and reducing sugars are less likely to be influenced.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8nr05860a |
This journal is © The Royal Society of Chemistry 2019 |