Impact of heterogeneously crosslinked calcium alginate networks on the encapsulation of β-carotene-loaded beads

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


INTRODUCTION
Alginate is a natural ionic polysaccharide that finds countless applications in a range of life-science related environments, like thickener and stabilizer in foods and beverages or encapsulating material for living cells (Chan, Lee, & Heng, 2006). Indeed, the attractiveness and versatility of this material lies on its low toxicity (Gombotz & Wee, 1998), film-forming ability and biodegradability (Benavides, Villalobos-Carvajal, & Reyes, 2012). Furthermore, the widely-recognised capacity of its sodium salt to form crosslinked networks in the presence of polyvalent cations, e.g. Ca 2+ , has received greatest attention from the scientific community (Russo, Malinconico, & Santagata, 2007).
Within food applications, microencapsulation offers the possibility of customising the functionality of food products by incorporating active compounds that are protected from the environment and released within the matrix under controlled conditions (Soukoulis et al., 2017). Particularly, the use of alginate networks as encapsulating material for a range of phytochemicals, volatile additives, enzymes and probiotic bacteria has been intensively researched. For instance, Lupo et al. (Lupo, Maestro, Porras, Gutiérrez, & González, 2014) explored the use of alginate microbeads as carriers for cocoa polyphenols to offer alternatives to commercial synthetic antioxidants in foods, while Tan et al (Tan et al., 2018) investigated the stability of tocotrienols encapsulated in alginate-chitosan microcapsules, compared to non-encapsulated tocotrienols in bulk oil, during storage and in a food model system. Among the different phytochemicals, β-carotene is characterised by an extended hydrocarbon backbone and high degree of unsaturation, which leads to low water solubility, low bioaccessibility and poor chemical stability (Zhang, Zhang, & McClements, 2016). Its encapsulation has been the focus of numerous studies, given the potential health benefits as antioxidant and anti-inflammatory properties of this compound and the challenges linked to its incorporation in food products, many of which could be addressed by encapsulation. For instance, the entrapment of β-carotene into a delivery system could prevent its fast degradation while providing effective protection against oxidation (Soukoulis et al., 2017;Zhang et al., 2016).

J o u r n a l P r e -p r o o f Journal Pre-proof
The performance of calcium alginate networks as coatings and delivery systems tends to be defined by their thickness, porosity, permeability, mechanical strength and swelling behaviour. Some of these properties are significantly affected by the molecular structure of alginate as well as the kinetic governing the process of crosslinking. Several studies in literature have focused on exploring the performance of calcium alginate networks as coatings, where different crosslinking densities have been proposed by varying the concentration of cations (used as gelling agents) and the gelation time, or by comparing alginates with different amounts of guluronic fraction. For example, Li et al. (Li, He, Huang, Li, & Chen, 2015) investigated how concentration of CaCl 2 affected the gelation process and physical properties of calcium alginate films, while Russo et al. (Russo et al., 2007) compared the mechanical properties and permeability of alginate matrices with different levels of crosslinking obtained by immersing alginate films with different ratios of guluronic acid on a 2% CaCl 2 aqueous solution. Furthermore, Palatel et al. (Patel, AbouGhaly, Schryer-Praga, & Chadwick, 2017) studied the effect of ionotropic gelation residence time on the crosslinking of alginate with CaCl 2 and their impact on the properties of alginate particles.
However, little consideration has been made on the role that heterogeneity of the crosslinks may play on the capability of the matrix to limit the diffusion of molecules from the environment and, therefore, on the ability to minimise the degradation of the sensitive core materials. Externally gelled alginate systems tend to be heterogeneous with a gradient of zones that are crosslinked to different extents. While at the surface gel formation is instantaneous, there is a constant diffusion of Ca 2+ towards the core, and of alginate from the core towards the gelled zone (Quong, Neufeld, Skjåk-Braek, & Poncelet, 1998). Despite the crucial role that such heterogeneity could play on the barrier properties of the network, this is often overseen in the literature, while it is also implicitly assumed that similar alginate networks will be developed by using comparable amount of guluronic fraction and concentration of gelling agent. For instance, in one of the studies conducted by Zhang et al. (Zhang et al., 2016), addressing the physicochemical stability and bioaccessibility of βcarotene entrapped in alginate beads, the authors assumed preliminary that higher concentration of J o u r n a l P r e -p r o o f Journal Pre-proof alginate should lead to enhanced chemical stability of β-carotene within the microparticles.
However, their results did not confirm this.
The present study highlights the relevant role that heterogeneity of crosslinking plays on determining the barrier properties of alginate networks and aims at demonstrating that homogeneously crosslinked calcium alginate matrixes show an improved performance as carriers for β-carotene, since they allow for enhanced chemical stability of the core material. To this end, different concentrations of sodium alginate (0.25-2%) and calcium chloride (2.5-5%) were employed to develop a range of calcium alginate networks, in the form of films and beads, with different crosslinking densities. Finally, thickness, weight, solubility, swelling index, mechanical properties, calcium content, microstructure (SEM) and FTIR spectra of films were analysed, while the chemical stability of β-carotene over storage at 25, 35 and 45 °C when encapsulated in the calcium alginate beads was also evaluated.

Preparation of films
Films were prepared by solvent casting. Alginate (0.25, 0.5, 1, 1.5 and 2% w/v) was dissolved in deionised water by stirring on a hot plate at 60 °C and, then, sonicated in a heated bath (Bransonic® CPXH, Fisher Scientific, UK) (Chan et al., 2006;Zhang et al., 2016). When the solutions became completely clear, they were left to cool down for 24 h at room temperature and approximately 30 J o u r n a l P r e -p r o o f Journal Pre-proof mg of each solution was poured onto a 60 mm diameter petri dish. Films were left to dry on a levelled surface at 37 °C for 60 h, and peeled off.

Preparation of crosslinked alginate films
Calcium alginate films were prepared by a dry-cast external gelation. The crosslinking solution was prepared by dissolving 2.5 and 5 % w/v of CaCl 2 powder in deionised water (Li et al., 2016). Each dried sodium alginate film was immersed in 300 ml of crosslinking solution for 24 h, and then washed once with a solution of 4 % w/v of NaCl and three times with deionised water to remove any surface unbound cations. The excess water was removed with a clean tissue paper, and, finally, the films were stored at room temperature and 65% relative humidity for 18 h before further use.

Determination of the thickness
Film thickness (t) was measured using a digital micrometre (IP65, Oxford Precision, UK). Six measurements were taken at the centre and around the perimeter of the film, for both alginate and crosslinked alginate films.

Apparent solubility and swelling index
Crosslinked samples were cut into square pieces of approximately 20 x 20 mm, weighted (w o ), immersed in 50 mL of distilled water and placed in a shaking water bath (Precision TSSWB15, ThermoScientific, UK) at 37 °C for 8 hours. After this, the films were weighted (w 2 ). The apparent solubility (S a ) was calculated by the difference between the initial weight (w o ) and the weight (w 1 ) of the films dried at room temperature until constant weight was achieved, and expressed as a fraction of the initial weight. The swelling index (S i ) was calculated as: where w 2 was the weight of the swelled filtered film and w 0 the initial weight. Five samples were analysed per condition.

Determination of the Calcium content
The calcium content of the crosslinked films was determined by using a flame photometer (500 701, Jenway, UK), where the samples (round shape, 60 mm diameter) were placed into 20 ml of J o u r n a l P r e -p r o o f Journal Pre-proof concentrated nitric acid (70% w/w) and heated for 10 minutes on a hot plate with stirring until reaching the boiling point. Once the films were completely dissolved, deionised water was added to the solutions up to 100 ml of final volume and then analysed (n= 4 per condition). A calibration curve with concentrations of calcium ranging from 10 to 100 ppm was calculated priorly. Calcium content was expressed as mg Ca per g of alginate.

Mechanical properties
The crosslinked and non crosslinked alginate films were cut in rectangular shape of 60mm length ( ) and 20 mm width (W). Model 3400 Instron (Instron Engineering Corporation, UK) was used to analyse the mechanical properties by setting the following operation conditions: initial grip separation of 50 mm and crosshead speed of 10 mm/min. The tensile strength (TS) was calculated by means of equation 2. (Eq.2) where F max was the maximum load at breaking point, and A was the cross-sectional area calculated by multiplying the thickness (T) by the width (W). Elongation at break percentage [ɛ(%)] of the films was calculated using the equation 3.
Where was the final length of the film at breakpoint and was the initial length of the film.
Young´s modulus (E) was calculated as the slope in the linear elastic region (0-5% of strain). At least five samples were analysed per condition.

Preparation of oil-in-water (o/w) emulsion
Coarse β-carotene-loaded (0.01% w/w, Sigma-Aldrich, UK) o/w emulsions were prepared following the methodology described by Soukoulis et al. (Soukoulis et al., 2017). The droplet size in these coarse emulsions was further reduced using a homogeniser (T25 Ultra Turrax, IKA, UK) at J o u r n a l P r e -p r o o f Journal Pre-proof 14,000 rpm for 5 min. These were used immediately after to prepare the alginate beads as described below.

Formation of alginate beads
Aqueous solutions containing different amounts of alginate (1 or 2% w/w) were prepared as described in section 2.2.1. Alginate solutions and β-carotene-loaded emulsions were then mixed together (1:1 weight ratio) for 1 h with continuous stirring to form a mixture with 3% oil (w/w) and either 0.5 or 1% alginate (w/w). Then, externally crosslinked alginate beads were produced by extruding dropwise the β-carotene-loaded-alginate solution into 300 mL of 5% CaCl 2 under constant stirring. The beads were held in the CaCl 2 solution for 1 h at ambient temperature and under agitation to promote crosslinking. Finally, the alginate beads were collected by filtration and washed with deionised water and 4% NaCl to remove any excess ions from their surfaces.
Subsequently, the beads were stored at 4 • C for 24 h to remove any residual external water, and final beads weight was determined. The encapsulation efficiency [EE(%)] was calculated using equation 4. As control, alginate beads that did not contain any β-carotene were prepared following a similar procedure. (Eq.4)

Storage study
The β-carotene loaded alginate beads (0.5 and 1%) and β-carotene loaded bulk oil were stored in a controlled-temperature incubator at 25, 35 and 45 • C in darkness for a maximum of 32 days.
Samples were withdrawn periodically every 4 days to analyse the content of β-carotene as described in the section below.

Extraction and quantification of β-carotene
β-carotene was isolated from the freshly made and stored alginate beads (0.5% and 1%) and bulk oil samples using a solvent extraction by adapting the method described by Biehler   Since the samples only contained single carotenoids (β-carotene), quantification was done spectrophotometrically (Cary 50 UV-Vis, Agilent Technologies, UK). β-carotene content in the samples (μg٠100 g -1 ) was calculated using the Beer-Lambert law (equation 5). (Eq.5) where A is the measured absorbance, ε λ the molar absorption coefficient of β-carotene in hexane at 453 nm (139200 L٠mol -1 ٠cm -1 ) (Craft & Soares, 1992), c the molar concentration of βcarotene(mol/L), and d the width of the cuvette. As blank, Canola oil and alginate beads that did not contain any β-carotene were subjected to similar extraction procedures. All measurements were repeated three times.

Kinetic modelling of β-carotene degradation
To obtain the kinetic parameters explaining loss of β-carotene content in the bulk oil and alginate bead samples during storage, the amount of β-carotene detected in the samples was plotted vs. time at all temperatures studied. Zero, first and second order kinetics were hypothesized by applying the corresponding reaction rate expression. Then, the order that best fitted the experimental data (data not shown) was selected. Following this criterion, first-order kinetics (equation 6) were used to describe degradation of β-carotene over time. The time for the concentration of a compound to fall to half its initial value (half-life, t 1/2 ) was also determined by equation 7.
(Eq.6) (Eq.7) where C represents the concentration (mg٠100g -1 ) of the compound at the specific time t; C 0 the concentration (mg٠100g -1 ) of each compound at time zero; and k the first-order constant (days -1 ).

Journal Pre-proof
For all the measurements, best fit between the experimental and predicted data was assessed by means of the adjusted regression coefficient (R 2 -ad.) (equation 8), considering that the higher the R 2 -ad. value the better the fit.
where m is the number of observations; j the number of model parameters and SSQ represents the sum of squares.

Statistical analysis
All the quantitative data were presented as mean ± standard deviation, unless noted otherwise.
Furthermore, all variables were checked for normality. Analyses of variance by using ANOVA were used to test for significant differences among the used alginate concentrations and conditions for a given variable. Furthermore, Pearson correlation coefficients were used to quantify the association between pairs of variables.

. Thickness, weight and Calcium content
In this work thickness, weight and Calcium content were used as indicators of the microstructural characteristics of the alginate-based networks. Thickness, weight and Calcium content of the alginate films before and after crosslinking are shown in Table 1. Scanning Electronic Microscopy (SEM) with EDS analysis was also used to further explore microstructural changes in the samples linked to gelation (Supplementary Material, Figure S1, Figure S2 and Table S1). Alginate films appeared very flexible and visually homogeneous with no brittle areas or bubbles. After crosslinking with 2.5 and 5% CaCl 2 , the films increased significantly (p<0.05) in thickness (50-119%) for all the studied concentrations of sodium alginate; nevertheless, no statistically significant changes (p>0.05) on their weights were observed with no relevant loss of polymer during gelation.
Similar range of values and observations on similar films before and after crosslinking have been widely reported in literature (Benavides et al., 2012;Chan et al., 2006). Gelation takes place when the alginate matrix is in a swollen (hydrated) state, since cross-links increase the rigidity of the alginate chains, after drying they will keep occupying more space (Crossingham, Kerr, & Kennedy, 2014). However, they also experienced some shrinkage associated to the process of gelation due to chain aggregation, which tends to be more noticeable in densely and homogeneously crosslinked networks (Blandino et al., 1999;Chan et al., 2006).  Figure S3). As expected, the spectra showed a shift on the asymmetric -COO-and symmetric -COO-peaks after gelation, which was associated to the replacement of sodium ions by calcium ions. However, this shift was of similar magnitude for all the samples. As reported by Blandino et al. (Blandino, Macias, & Cantero, 1999), after instant crosslinking at the junction zones, the process of gel formation is controlled by the diffusion of Ca 2+ across the chains of alginate, which then binds to the unoccupied binding sites on the polymer. Thus, at high concentrations of alginate, the diffusion of Ca 2+ is considerably restricted due to the high levels of entanglement, and the majority of the crosslinks remains on the surface. Therefore, the obtained films can be considered highly heterogeneous crosslinked matrices. This hypothesis seems to be supported by SEM micrographs (Supplementary Material, Figure S1), as higher degree of heterogeneity was found in crosslinked samples with 1-2% alginate. Higher atomic concentration of calcium was also found in the surface of films with 1-2% of alginate (Supplementary Material, Figure S2 and Table S1).

Apparent solubility and swelling index
Apparent solubility and swelling index of calcium alginate films have been traditionally used as indicators of the extent of crosslinking (Li et al., 2015;Li et al., 2016;Rhim, 2004) and in the literature, lower solubility and swelling index have been assumed to be associated with higher J o u r n a l P r e -p r o o f crosslinking density. Figure 1 reports solubility and swelling index values for the crosslinked samples. As control, uncrosslinked alginate films completely dissolved after 10 min of immersion in water, while the apparent solubility of crosslinked films ranged from 4.3±0.9% to 15.4±1.6% for films containing 2% and 0.25% of alginate respectively (Figure 1(A)). These values were in good agreement with those reported in literature for samples with similar concentrations of alginate and CaCl 2 (Chan et al., 2006;Li et al., 2016). Furthermore, at 5% CaCl 2 , films with higher concentration of alginate showed significantly lower values of apparent solubility and swelling index (p<0.05), while a lower concentration of CaCl 2 (2.5%) led to a significant increase (p<0.05) of the swelling index (Figure 1(B)), without affecting the samples apparent solubility (p>0.05). To be noticed, despite the lower swelling index at higher CaCl 2 concentration, the samples containing 1 and 2% alginate showed no differences in calcium content. This could indicate that, rather than the extent of crosslinking, the apparent solubility and the swelling index of the films were strongly affected by the distribution of the crosslinks. Particularly, the densely crosslinked surface of films containing 1-2% alginate may have simply restricted the absorption of water within the network and, therefore, limited the swelling and the loss of polymer chains through dissolution, as confirmed by the weight variation values ( Table 1) These results indicated that the concentration of alginate and crosslinking agent strongly affected the degree of swelling of the network and the density and heterogeneity of crosslinks. Therefore, it is also likely to impact on the performance of the network as encapsulating material. In addition, the degree of swelling experienced by the alginate network during immersion in the gelling solution may also play a role in the behaviour of the matrix upon hydration, as indicated by the significant correlation found between the apparent solubility (r = -0.701) and swelling index (r = -0.755) and the thickness of the dry calcium alginate films (p<0.05). A similar observation was reported by Li et al. (Li et al., 2015). Consequently, all the studied properties of the calcium alginate networks need to be carefully considered when used in encapsulating applications, since their overall performance may be dramatically impacted by the heterogeneity of crosslinking and swelling degree. Figure 2 shows the representative stress-strain curves obtained for the 2% alginate film before and after crosslinking. Figure 3 reports the values of the elongation (ɛ) and tensile strength (TS) at break and Elastic modulus (E) of the prepared films. Regarding the elongation at break, the crosslinked samples were characterised by higher values compared to the uncrosslinked films. As example, at 1% alginate content, crosslinked samples showed values of 1.73±0.33% with 2.5% CaCl 2 and 3.17±0.27% with 5% CaCl 2 while we evaluated an elongation at break of 1.01±0.17% for the uncrosslinked samples.

Mechanical properties
Furthermore, it can be noticed that ɛ increased proportionally with the increase of alginate content for the uncrosslinked films, while after crosslinking the films presented a different trend. The highest ɛ values were found for the 0.5-1% alginate samples, with a significant decrease (p<0.05) in the samples with higher alginate content (Figure 3(A)).  For the tensile strength at break (Figure3(B)), the increase of alginate content led to higher values for both uncrosslinked and crosslinked conditions. Furthermore, we observed that films with 1.5% alginate crosslinked with 5% CaCl 2 showed the highest TS value (146.8±10.9 MPa), indicating an increase of the resistance due to the reaction between alginate and Ca 2+ , and consequently the formation of the "egg-box" structure. This behaviour is in agreement with published results (Costa et al., 2018).
Similarly, uncrosslinked samples showed higher values also for the elastic modulus. As example, we calculated 12702.5±1385.1 MPa and 2135.9±748.0 MPa at 1% alginate for the uncrosslinked and 5% CaCl 2 crosslinked films, respectively. Furthermore, in all conditions, samples showed a alginate content for the uncrosslinked and crosslinked samples respectively, followed by a significant decrease (p<0.05) at 2% alginate content. This was also confirmed in the E values calculated for the crosslinked samples at 2.5% CaCl 2 . This trend was in accordance with similar works reported in literature (Cuadros, Skurtys, & Aguilera, 2012).

Multivariate analysis
PCA analysis of the characterisation variables was performed to determine whether the structural characteristics of the crosslinked films allowed to discriminate between the different processing conditions. Figure 4 shows the biplot used to assess the data structure and the loading of the first two components (PC1 and PC2), that explained 75.91% of the total variance. Particularly, calcium content (Ca) and apparent solubility (S a ) exhibited large positive loadings on PC1, that may be related with the crosslinked density, while the variables associated with the mechanical properties had equally a negative loading. For PC2 the elasticity [ɛ(%)] had the largest positive loading, while elastic modulus (E) and the hydration index (S i ) presented large negative loadings.
To be noticed, those samples were also located in the positive region of PC2 (18.85 %). According to these results, those films, presenting similar and comparable characteristics, were selected for the further formation of the β-carotene loaded beads.
J o u r n a l P r e -p r o o f

PC2
According to Gama and de Sylos (Gama & de Sylos, 2007), the principal cause of carotenoid losses is oxidative degradation, which depends on the availability of oxygen, accelerated by heat, light, enzymes, metals, and co-oxidation with lipid hydroperoxides. To further investigate the protective role of the calcium alginate coatings, the kinetics of degradation of β-carotene in all the samples was studied ( Table 2). Degradation of β-carotene was appropriately described by first-order kinetics and t 1/2 and k values were well within range of those reported in the literature (Benlloch-Tinoco et al., 2015).

°C 25 °C
J o u r n a l P r e -p r o o f Encapsulation significantly (p<0.05) improved the chemical stability of β-carotene ( Figure 5). The degradation rate of β-carotene dispersed in bulk oil was at least 1.5 times higher than in any other sample. Larger differences were observed between oil and calcium alginate (0.5 % and 1%) samples at higher storage temperatures ( Table 2). Beads with 0.5% alginate were 1.5-2 times more efficient at preventing degradation of β-carotene than 1% calcium alginate beads (p<0.05). This means that if β-carotene-loaded beads were to be stored at room temperature, for example, half of the β-carotene content would be degraded after 22 days for 0.5% alginate beads, compared to 13 days for 1% alginate beads. Similar observations have been reported by Zhang et al. (Zhang et al., 2016).
These results highlight that calcium alginate networks that are more extensively crosslinked ( Table   1) and homogeneous in nature (par 3.1.1), e.g. 0.5% alginate, are notably more efficient at protecting β-carotene from oxidation over storage. Firstly, diffusion of oxygen, which typically takes place directly through the polymeric matrix (capillary diffusion) and/or as a result of solubilisation in the film (activated diffusion) (Donhowe, 1994), may be restricted by the reduced macromolecular mobility associated with crosslinking. Alternatively, homogeneous networks, for being more flexible (lower elastic modulus, Figure 3(C)), may experience a more uniform contraction during drying, which prevents excessive shrinkage, cracking and/or collapse, and J o u r n a l P r e -p r o o f therefore are able to restrict the diffusion of oxygen more efficiently than those that are stiffer (higher elastic modulus, Figure 3(C)) and more heterogeneous (e.g. 1% alginate).

CONCLUSIONS
The high levels of entanglement typically found in uncrosslinked films with high concentrations of sodium alginate (≥1%) markedly effected the microstructural characteristics of the calcium alginate network formed upon crosslinking via external gelation with Ca 2+ . When using CaCl 2 (≥2.5%) as gelling solution the extent of crosslinking in alginate films (≥1%) was highly limited by the heterogeneous nature of the network formed. Calcium alginate networks with a more homogeneous distribution of crosslinks (<1% alginate, 5% CaCl 2 ) showed more desirable physical and mechanical properties for encapsulation purposes, as indicated by their higher calcium content and elasticity. Improved barrier properties were also observed in more homogeneously and extensively crosslinked calcium alginate beads, as indicated by the enhanced chemical stability of β-carotene in 0.5% alginate samples. These findings highlight the need to adequately understand how crosslinking affects the microstructure of calcium alginate networks if their use as encapsulating material for food applications is to be exploited in an efficient and controlled manner.

CRediT author statement
Joel Giron Hernandez: Investigation, Formal Analysis, Writing-Reviewing and Editing.

Acknowledgments
This study was supported financially by the Global Challenges Research Fund programme from Northumbria University.

Data availability
The data that support the findings of this study are available upon reasonable request to the corresponding author.
J o u r n a l P r e -p r o o f