Yunbing
Tan
a,
Ruyi
Li
b,
Hualu
Zhou
a,
Jinning
Liu
a,
Jorge
Muriel Mundo
a,
Ruojie
Zhang
a and
David Julian
McClements
*a
aDepartment of Food Science, University of Massachusetts Amherst, Amherst, MA 01003, USA. E-mail: mcclements@foodsci.umass.edu
bState Key Laboratory of Food Science and Technology, Nanchang University, 8 Nanchang, Jiangxi 330047, PR China
First published on 9th December 2019
Recently, the standardized in vitro digestion model (“INFOGEST method”) used to evaluate the gastrointestinal fate of foods has been revised and updated (Brodkorb et al., 2019, Nat. Protoc., 2019, 14, 991–1014). Under fed state conditions, the calcium level used in this model is fixed and relatively low: 0.525 mM. In practice, the calcium concentration in the human gut depends on the nature of the food consumed and may vary from person-to-person. For this reason, we examined the impact of calcium concentration on the gastrointestinal fate of a model nutraceutical delivery system. The effect of calcium level (0.525–10 mM) on lipid digestion and β-carotene bioaccessibility in corn oil-in-water nanoemulsion was investigated using the INFOGEST method. At all calcium levels, the lipids were fully digested, but this could only be established by carrying out a back titration (to pH 9) at the end of the small intestine phase. Conversely, the bioaccessibility of β-carotene decreased with increasing calcium levels: from 65.5% at 0.525 mM Ca2+ to 23.7% at 10 mM Ca2+. This effect was attributed to the ability of the calcium ions to precipitate the β-carotene-loaded mixed micelles by forming insoluble calcium soaps. The ability of calcium ions to reduce carotenoid bioaccessibility may have important nutritional implications. Our results show that the bioaccessibility of hydrophobic carotenoids measured using the INFOGEST method is highly dependent on the calcium levels employed, which may have important consequences for certain calcium-rich foods. Moreover, we have shown the importance of carrying out a back titration to accurately measure free fatty acid levels in the presence of low calcium levels.
In vivo human feeding studies are the most accurate methods of determining the impact of specific colloidal delivery systems on the bioaccessibility and bioavailability of nutraceuticals.8 In these studies, the levels of nutraceuticals and their metabolites are typically measured in the bloodstream, urine, and/or feces over time. Despite their accuracy, however, human feeding studies are rarely used to test nutraceutical bioavailability because of their high cost, long time, potential safety issues, and ethical concerns.7,12 Moreover, they do not provide any insights into the physicochemical mechanisms involved. For this reason, in vitro digestion models have been developed that are relatively simple and inexpensive to perform, and allow researchers to rapidly screen many different formulations during the early stages of delivery system development.13 A recent review article has highlighted that there is often a good qualitative correlation between the results obtained using in vivo and in vitro methods, which supports the utilization of the latter for testing nutraceutical bioaccessibility.8
A number of in vitro digestion protocols have been developed over the past decade to test the gastrointestinal fate of foods, which vary in their attempts to mimic the complex physiochemical and physiological processes occurring inside the human gut.13,14 These models typically come to some compromise between accurately simulating the complexity of the human gut and having a protocol that is inexpensive, simple, rapid, and repeatable. Recently, there have been attempts to standardize the in vitro digestion models used in both the food and pharmaceutical areas.15,16 The advantage of having a standardized model is that results on different samples or from different laboratories can be compared. In the food industry, the most well-established and widely-used in vitro digestion model is that developed by the INFOGEST consortium.16 Despite its widespread use, this method often leads to widely differing results when studying lipid digestion in foods, ranging from almost complete digestion17 to only modest digestion.18 In reality, humans are known to digest and absorb the vast majority of the lipids (triacylglycerols) they consume. It is, therefore, important to establish the origin of the observed variations in lipid digestion determined using the INFOGEST method. Moreover, this static in vitro digestion procedure has recently been refined to improve its accuracy and reliability,13 and so it is useful to examine the efficacy of this refined method for studying lipid digestion.
One factor that is known to play an important role in lipid digestion in foods is the level of calcium present.19,20 Calcium may act as both a promoter or inhibitor of lipid digestion depending on the system.21 On one hand, cationic calcium ions can promote flocculation of oil droplets, especially in emulsions stabilized by anionic emulsifiers, resulting in less surface area for lipolysis reaction. On the other hand, calcium ions can facilitate lipid digestion due to their ability to act as co-factors for pancreatic lipase and the ability to precipitate free fatty acids generated at the lipid droplet surfaces. There have, therefore, been several attempts at understanding the impact of calcium ions on lipid digestion.19,20
The digestive conditions within the INFOGEST method are based on the fed state of human digestion. Nevertheless, the calcium level used in the simulated intestinal fluids (0.6 mM) is considerably below those reported in human intestinal fluids.22,23 Besides, some foods contain high levels of calcium, such as cheeses and milk, which can lead to much higher calcium levels (up to 10 mM) in the small intestine during digestion.23–25 These relatively high calcium levels may interfere with the lipid digestion process.
The objective of the current study was to examine the impact of calcium levels on the digestion of lipids in nanoemulsion-based delivery systems using the INFOGEST method. Moreover, we also examined the impact of calcium levels on the bioaccessibility of a model hydrophobic nutraceutical (β-carotene) encapsulated within the nanoemulsions. This carotenoid has been associated with several diet-related health benefits due to its pro-vitamin A and antioxidant activities.26 Structurally, β-carotene consists of two beta-rings held together by a long polyene chain, and hence it is a highly hydrophobic polyunsaturated molecule that has low water-solubility and is highly prone to oxidation. A considerable research effort has, therefore, been carried out to develop colloidal delivery systems to encapsulate and protect this carotenoid, as well as to boost its bioavailability.27,28 Nevertheless, the fate of these systems within the INFOGEST method has rarely been examined.
In summary, the main objective of this research is to investigate the effects of different calcium levels on lipid digestion and β-carotene bioaccessibility of a nanoemulsion-based delivery system using the INFOGEST method. The physical and structural properties of the nanoemulsion were measured during in vitro digestion to identify the possible underlying mechanisms involved. The information generated by this research should lead to a better understanding of the lipid digestion process, as well as to the creation of more efficacious nutraceutical delivery systems. Moreover, it provides valuable insights into the critical role that calcium ions play in the standardized INFOGEST model.
The particle size of the samples from the micelle phases were measured by dynamic light scattering (Zetasizer Nano ZS, Malvern Instruments Ltd, Malvern, Worcestershire, UK). Phosphate buffer solution (5 mM, pH 7.0) was used to dilute samples so as to avoid multiple scattering.
The release of the FFA was measured using a pH stat method, and calculated from the amount of NaOH needed for titration (857 Titrando, Metrohm USA Inc., Hillsborough, FL, USA). During the small intestine phase, the volume of NaOH solution required to maintain the system at pH 7.0 was measured. Afterwards, a back titration to pH 9.0 was applied to neutralize any non-ionized FFAs. A blank test was carried out using a blank sample with the same composition as the test sample, but no oil, to subtract the contribution of any non-lipid components to the digestion profiles. After digestion, the separation of the micelle and sediment phases was achieved by centrifugation (Sorvall Lynx 4000 centrifuge, Thermo Scientific, Waltham, MA, USA) of the intestine samples at 30970g (18000 rpm) for 50 min at 4 °C.
The bioaccessibility, recovery, and stability (%) of β-carotene were then calculated using the following equations:
Here, Cmicelle, Csediment, Cdigesta, and Cinitial are the concentrations of β-carotene in samples collected from the mixed micelle, sediment, total intestine digesta, and initial nanoemulsion, respectively.
The nanoemulsion was then subjected to the simulated oral phase. The light scattering experiments suggested that the particle size of the nanoemulsions remained fairly similar after exposure to the simulated oral conditions (Fig. 1a), but the confocal microscopy images showed that some of the oil droplets had flocculated (Fig. 1c). The magnitude of the negative surface potential on the protein-coated oil droplets decreased significantly (p < 0.05) (Fig. 2). These effects might have occurred because the change in ionic strength altered the electrostatic interactions, or the presence of mucin promoted depletion or bridging flocculation.33
The nanoemulsion was then exposed to simulated gastric conditions by adjusting the pH to 3.0 and adding pepsin to reach a level of 2000 U mL−1 in the final mixture. The physical properties of the nanoemulsion changed appreciably within this stage. The mean particle diameter increased to over 30 μm, and the position of the peak in the particle size distribution shifted upwards, with most of the particles being in the range from 1 to 300 μm (Fig. 1a and b). The confocal microscopy images indicated that excessive droplet flocculation occurred under simulated gastric conditions (Fig. 1c). Additionally, the negative surface potential of the particles in the nanoemulsion decreased to around −1.9 mV. Moreover, visual observations showed that the nanoemulsions became unstable to phase separation, with a thin cream layer being seen on the top and a clear serum layer at the bottom (Fig. 3).
Fig. 3 Photos of corn oil nanoemulsion during simulated gastrointestinal tract (GIT) digestion. Abbreviation: small intestine (SI). |
Previous studies suggest that the flocculation state of oil droplets when they first enter the small intestine phase impacts their digestion: when the droplets are highly flocculated, the lipase is unable to easily access the surfaces of all the oil droplets, thereby inhibiting digestion.34 For this reason, we adjusted the gastric chyme to pH 7.0 before adding the other intestinal components (calcium, bile salt, and pancreatic enzymes) so as to focus on the impact of pH on the flocculation state of the oil droplets. The sample obtained by this process is referred to as the “SI-Initial phase”. The light scattering and microscopy measurements indicated that there was a pronounced decrease in droplet flocculation when the system was adjusted to pH 7 (Fig. 1). This effect can mainly be attributed to an increase in the negative charge on the protein-coated oil droplets (Fig. 2), which recovered to a value fairly similar to that of the oral phase. Thus, the increased electrostatic repulsion between the oil droplets in the emulsions reduced the degree of flocculation. Nevertheless, there was evidence of some large individual oil droplets present in the SI-initial phase, which suggests that droplet coalescence as well as flocculation had occurred during incubation in the gastric phase. Visual observations of the nanoemulsions in the SI-initial phase showed that they consisted of homogenous nanoemulsions without any obvious phase separation (Fig. 3). The good stability of the nanoemulsions to gravitational separation can be attributed to the fact that most of the individual oil droplets were relatively small and not flocculated.
The nanoemulsions were then exposed to the full intestinal phase by adding the appropriate levels of digestive enzymes, bile salts, and mineral ions. The confocal microscopy images (Fig. 1c) and pH stat measurements (see later) indicated that the majority of the lipids had been digested by the end of the small intestine phase. Visually, the nanoemulsions had a watery-turbid appearance, with no cream layer at the top, again suggesting that most of the oil droplets had been fully digested (Fig. 3). The particle size distribution measurements indicated that the digested samples contained a broad range of different sized particles, which were probably micelles, liposomes, insoluble calcium soaps, and protein aggregates (Fig. 1b). The ζ-potential of the digested samples was highly negative, which can be attributed to the presence of various kinds of anionic species, such as proteins, peptides, bile salts, phospholipids, and free fatty acids (Fig. 2).35
The properties of the particles in the mixed micelle sample (0.525 mM calcium, pH 7) were also measured (Fig. 2 and 4). Micelles are too small to be detected by static light scattering and so dynamic light scattering was used instead. The mean particle diameter of the mixed micelle phase was around 208 nm (Fig. 4a), which is considerably larger than the dimensions of individual micelles (<10 nm).36,37 This phenomenon is probably because the samples also contained liposomes and calcium soaps.38–40 The magnitude of the negative ζ-potential on the particles in the mixed micelle phase was relatively high (−59.4 mV), which can again be attributed to the presence of various anionic species (Fig. 2). The mixed micelle solution was slightly turbid suggesting that it contained particles large enough to scatter light strongly (Fig. 3).
Overall, the physical and structural properties of the nanoemulsions during passage through the INFOGEST GIT model were fairly similar to those found using other in vitro digestion models.31,41 The main exception was that the micelle phase formed after lipid digestion was more turbid in the INFOGEST method.
The release of FFAs over time was measured during the small intestine phase using the pH-stat method by adding enough alkaline solution to maintain a neutral pH (Fig. 5a). Similar to previous studies, the amount of FFA increased greatly during the first 1000 s, then gradually increased afterwards.42 However, under the fed conditions used in the standardized INFOGEST method, which employed only 0.525 mM calcium, the fraction of FFAs released by the end of the small intestine phase was only 39.4% (Fig. 5b). This value is much lower than that reported (>90%) for similar nanoemulsions tested using an alternative in vitro digestion method.31 This phenomenon may have occurred because most of the FFAs were non-ionized at neutral pH when low calcium levels were used.22 For this reason, we used an additional back-titration step to pH 9 to detect the total number of FFAs released. In this case, a much higher final FFA value was reached, i.e., around 128% (Fig. 5b). Indeed, the final FFA value was considerably higher than the expected value of 100%, assuming that two FFAs were released per triacylglycerol molecule. Several other studies have reported FFA values >100% at the end of lipid digestion.22,43 This effect might have been attributed to the production of more than two FFAs per triacylglycerol due to alkaline hydrolysis of the triacylglycerol molecules at high pH values.44
Overall, this study shows that a high proportion of the FFAs generated during the small intestine phase under neutral conditions are non-ionized, and so could not be titrated by sodium hydroxide. Interestingly, this effect is not seen in alternative in vitro digestion models that use much higher calcium levels, which may have been because the presence of the calcium ions altered the equilibrium between the ionized and non-ionized states of the FFAs. In summary, for the INFOGEST method, a back titration to pH 9 is required to measure the faction of non-ionized FFAs present in the samples after digestion.
The β-carotene concentration in the intestinal samples, as well as its distribution in the mixed micelle and sediment phases, was also measured at the calcium level (0.525 mM) normally used in the INFOGEST method (Fig. 6a). Most of the β-carotene was solubilized in the mixed micelle phase after digestion, leading to a relatively high bioaccessibility (65.5%) (Fig. 6b), which is in agreement with previous studies.41 The recovery rate of the carotenoid was around 97.9% (Fig. 6b), which indicates that most of the β-carotene in the intestinal digesta was located in either the mixed micelle or sediment phases. The stability of the β-carotene encapsulated within the nanoemulsions was relatively high (84.4%), which indicated that the carotenoid was relatively stable to chemical degradation within the simulated GIT.
At low calcium levels (0.525 and 1.5 mM), the D3,2 values were 0.424 and 0.499 μm respectively, and the particle size distributions were bimodal. The confocal microscopy images indicated that there were only a few relatively large oil-rich particles present in these samples after digestion. These particles may have been oil droplets formed by the non-ionized free fatty acids. When the calcium level was increased to 5 mM, the mean particle diameters increased significantly (p < 0.05) to 0.692 μm and the particle size distribution broadened. The confocal microscopy images indicated that there were some large spherical lipid-rich particles, as well as many smaller irregular lipid-rich particles. When the calcium level was increased further to 7.5 and 10 mM, the D3,2 values remained fairly similar, but there was a change in the shape of the particle size distribution. Besides, there were more particles of a ring shape or an irregular shape in the confocal images at these high calcium levels. We hypothesize that the cationic calcium ions promoted the aggregation of some of the small anionic mixed micelles, leading to the production of calcium soaps. Previous studies have also reported that high levels of calcium can promote the formation of insoluble calcium soaps.24,43
The surface potential of the particles in the digesta became increasingly less negative as the calcium ion level was increased (Fig. 7a). This effect is consistent with the binding of cationic calcium ions to the anionic mixed micelles. At sufficiently high calcium levels, the formation of calcium soaps may have been promoted because of the decrease in surface potential on the mixed micelles and the tendency for bridging to occur.
The release of FFAs during digestion in the small intestine phase (pH 7.0) was monitored by titration with sodium hydroxide using the pH stat method (Fig. 5a). The FFA release profiles of the nanoemulsion all followed a fairly similar pattern irrespective of the calcium levels used: there was a sharp increase during the first 1000 s followed by a gradual increase later. The fraction of titrated FFAs by the end of the digestion period increased from around 39% to 95% as the calcium concentration was increased from 0.525 to 10 mM. Interestingly, the final FFA value measured at pH 7 was linearly related to the calcium concentration (Fig. 5b). As discussed earlier, most FFAs generated during lipid digestion are not titratable at low calcium levels because they are in a non-ionized form.22 Therefore, a back titration to pH 9 was carried out after the small intestine phase to measure the total amount of FFAs produced. Under these conditions, all the samples were seen to be fully digested after the intestinal phase with a final FFA value ranging from 116 to 128%. This result highlights the critical importance of carrying out the back-titration step under the low calcium conditions used in the standardized INFOGEST method. The actual FFA release kinetics were determined by multiplying the measured values with a correction factor (C), which was defined as the ratio of the final FFA values measured at pH 9 and pH 7 (Fig. 5c). The corrected FFA release profiles indicate that the calcium level used had little impact on the lipid digestion curve. The calcium concentration used in the in vitro digestion model therefore appears to impact the ionization state of the FFAs, rather than the total amount of FFAs generated. We hypothesize that the calcium ions altered the ionization equilibrium of the carboxylic acid groups (2[–COOH] + Ca2+ ↔ Ca(–COO)2 + 2H+), thereby releasing the titratable protons. Thus, most FFAs could be titrated when higher calcium levels are employed in an in vitro GIT model. Alternatively, this result suggests that the in vitro digestion model could be simplified (no back titration) if relatively high calcium levels (10 mM) are used in the small intestine phase.
The β-carotene concentrations of the whole small intestine samples, as well as those of the micelle and sediment phases, were measured at different calcium levels (Fig. 6a). The β-carotene concentrations in the whole small intestinal samples did not change appreciably (4.3 to 4.8 μg ml−1) when the calcium level was changed. Moreover, the β-carotene remained relatively stable under different calcium level during the GIT (75.3 to 84.4%) (Fig. 6b), which suggests that calcium do not have a major impact on the chemical stability of the carotenoid during the in vitro GIT digestion. The recovery of the β-carotene from the micelle and sediment phases combined was between about 89.2 to 98.2% of that of intestinal phase (Fig. 6b), which indicated that most of the β-carotene was released from the oil phase after digestion.
Nevertheless, the bioaccessibility of β-carotene decreased from 65.5 to 23.7% as the calcium concentration was increased from 0.525 to 10 mM (Fig. 6b). In particular, the β-carotene concentration in the micelle phase decreased from around 3.02 to 0.78 μg ml−1, what in the sediment phase increased from around 1.32 to 3.20 μg ml−1 (Fig. 6a). This result indicates that there was a change in the location of the carotenoids within the digesta at the end of the small intestine phase. During lipid digestion, most of the encapsulated β-carotene should be released from the oil droplets and solubilized within the mixed micelles formed by bile salts and FFAs.12 In the presence of calcium ions, however, the micelles tend to aggregate and eventually form particles that are large enough to sediment to the bottom of the samples.31 Indeed, an increasing amount of sediment was observed in the samples as the calcium concentration was increased (Fig. 8a). As a result, some of the β-carotene-loaded mixed micelles are incorporated into insoluble calcium soaps, which decreases the measured bioaccessibility. These results show that the bioaccessibility of carotenoids is highly sensitive to the level of calcium present in the small intestine, which may have important practical applications. For instance, consumption of carotenoid-rich foods with calcium-rich foods could lead to a decrease in carotenoid bioaccessibility. Even so, it will be important to establish whether a similar effect is observed in vivo using animal or human feeding studies.
Fig. 8 Photos of the samples from intestinal phase with different calcium levels: (a) sedimentation after centrifuge; (b) micelle phase. |
In addition, we also observed that the mixed micelle phase was highly turbid at lower calcium levels but became increasingly clear at higher calcium levels (Fig. 8b). This phenomenon may have occurred because there were a lot of mixed micelles (micelles and vesicles) present at low calcium levels, which scattered light strongly and made the mixed micelle phase looks turbid. Conversely, at high calcium levels, many mixed micelles precipitate and sediment to the bottom of the samples so the mixed micelle phase looks clearer.
Fig. 9 Photos of samples from control sample (no lipid) and micelle phase of high and low calcium levels at different pH values. |
The surface potential of the mixed micelle samples remained negative across the entire pH range (from 6 to 9), which is because all of the colloidal particles present are anionic. At pH 6, the ζ-potential of the low calcium sample (−32.3 mV) were similar to that of high calcium sample (−29.9 mV) (Fig. 4b). This phenomenon may have occurred because there was a high fraction of non-ionized FFAs at pH 6, since the pKa value of the carboxylic acid groups is higher than this value.46 As a result, the magnitude of the negative surface potential was lower. On the other hand, calcium ions to bind to the surfaces of the mixed micelles in high calcium sample and thus reduced the negative potential. The impact of pH value on the surface potential depended on the calcium levels present. For low calcium levels, the magnitude of the ζ-potential increased significantly (p < 0.05) with increasing pH: from −32.3 mV at pH 6 to −63.6 mV at pH 9 (Fig. 4b). This effect can be attributed to the increasing ionization of the FFAs as the pH is increased over the pKa values of the carboxylic acid groups. For high calcium levels, increasing the pH had little effect on the surface potential (−25 to −34 mV), which may have been because the ionized FFAs were bound and precipitated by the calcium ions.
The impact of pH and calcium levels on the turbidity of the mixed micelle samples was also investigated (Fig. 9). In this case, control samples were also used, which were obtained by collecting the mixed micelle phase from samples containing no oil. For the control samples, the mixed micelle phase was cloudy at pH 6 but transparent from pH 7 to 9, irrespective of calcium level. The turbidity of the control samples at pH 6 was presumably due to the fact that they contained proteins (whey protein, pancreatin, and pancreatic lipase) with isoelectric points close to this value. At pH 6, the turbidity of the control samples containing 10 mM calcium was slightly less than those containing 0.525 mM calcium, which might have occurred because the cationic calcium ions promoted some precipitation of the anionic proteins.
In the test (nanoemulsion) samples, the mixed micelle solutions were cloudy at pH 6 and became increasingly clear as the system was raised to pH 9. This effect was observed at both calcium levels used, but the turbidity of the mixed micelle phase was less in the samples containing the higher calcium levels. This effect may have been because the cationic calcium ions promoted aggregation of the anionic proteins and mixed micelles. For the test samples, the pH-dependent change in turbidity of the mixed micelle phase might be due to both protein and mixed micelle aggregation. As the pH is increased, both the proteins and fatty acids become more negatively charged, which should increase the electrostatic repulsion between the colloidal particles, thereby opposing aggregation.
In summary, these results showed that the change of pH value in the micelles samples could modify the size and charge properties as well as the solubility of the components insides, and these effects are dependent on the calcium level applied. Consequently, it might place possible impacts on the bioaccessibility of the solubilized nutraceuticals.
Interestingly, increasing the level of calcium in the small intestine phase decreased the bioaccessibility of the carotenoids, which was attributed to the ability of the cationic calcium ions to precipitate the anionic β-carotene-loaded mixed micelles. As a result, the carotenoids were no longer present in the mixed micelle phase generated during lipid digestion. The ability of calcium ions to reduce carotenoid bioaccessibility may have important implications for the nutritional benefits of these nutraceuticals. Consuming carotenoid-rich foods with calcium-rich foods could lead to a reduction in carotenoid bioavailability and efficacy.
This journal is © The Royal Society of Chemistry 2020 |