Physicochemical aspects of mucosa surface

Abstract

Interactions between the surface of the intestinal mucus and the molecular components of a food, toxin or drug control the latter's bioabsorption. Understanding the colloidal basis of such interactions requires a thorough thermodynamic characterization of the mucus layer. Towards this aim, surface thermodynamics of porcine intestinal mucin are studied in this work by the aid of inverse gas chromatography (IGC), at infinite dilution and a temperature range (33–60 °C) well below the glass transition. The affinity of several molecular probes, both apolar and polar of different functionalities, onto the mucin surface is evaluated in terms of dispersive and specific interactions calculated by processing the chromatographic retention profiles. Well defined Gaussian peaks, typical of Henry's adsorption, have been obtained for apolar probes. The absence of any thermal transitions is confirmed by the linear drop of the surface free energy with temperature rise, within a typical energy range for carbohydrate polymers (γ^d_s = 33–40 mN m⁻¹). For specific polar probes, however, non-Gaussian tailing peaks have been recorded, indicative of desorption retardation phenomena. The potential of polar probes to deploy specific interactions with mucin surface is interpreted on the basis of the homomorph concept. It is inferred that a kind of molecular sieving mechanism assisted by the high polarity of probes seems to favor retention. Evaluation of the surface acid–base interaction potential demonstrates that the mucosal barrier is Lewis amphoteric with predominant basic character. Complementary data obtained via TMDSC, TGA, XRD, and FTIR are discussed in conjunction.

---

1. Introduction

Mucus is a highly complex viscoelastic medium that provides a defensive barrier across many epithelial surfaces including the respiratory, reproductive and gastrointestinal (GI) tracts. In the gut it serves many functions, including lubricating the passage of particles of food; acting as a barrier to pathogens, destructive enzymes and toxic substances and as a permeable gel layer for the exchange of nutrients with the underlying epithelium.¹ Secreted mucins typically have MW above 2 MDa and possess a highly swollen structure in solution. They comprise of long and stiff polypeptide rods, densely covered with glycans, which bind a lot of water that is responsible for the typical gel-like properties of mucins.^2–4

Under normal circumstances, the intestinal epithelial surface is covered by this mucosal viscous liquid/gel. It is intuitive that prior to digestion of the individual components of a food, a toxin, or a drug, these have to be adsorbed first onto the mucosal barrier. After this adsorption, they should diffuse through this semi-liquid colloidal layer prior to reaching the epithelial wall. This suggests that the successful bioabsorption of a food, toxin, or drug molecular component could be reduced into a physicochemical problem of adsorption onto the mucus, and then of thermodynamic compatibility with the mucus macromolecular components. Under this rationale, understanding the underlying thermodynamics of the mucosal layer can allow for a thorough description of digestion at the molecular level, contributing to a wide number of fields, such as drug design, food digestion, and hazard assessment due to exposure in toxins.

Inverse gas chromatography (IGC) is one of the most powerful and sensitive techniques for characterizing the thermodynamic properties of materials.⁵ In the present study this method was used for the determination of the thermodynamic functions for adsorption of test compounds onto the surface of porcine mucin in the Henry's law region (i.e. infinite dilution or equivalently zero surface coverage), providing the first description of the basic thermodynamic parameters of intestinal mucus.

2. Materials & methods

2.1. Mucin structure

The examined mucin fraction was isolated from the epithelial surface of porcine jejunum, as described in detail elsewhere.¹ In brief, fresh porcine small intestine was rinsed with 67 mM phosphate buffer (pH 6.7) containing 0.02% w/v sodium azide and protease inhibitor cocktail (Roche Diagnostics GmbH, Mannheim, Germany) in order to remove debris. Mucus was removed by scraping the epithelial surface of the intestine; debris was further removed by extracting the ex vivo mucus overnight at 18–22 °C, the extraction buffer being 10 mM sodium phosphate at pH 6.5 containing 4 M guanidinium hydrochloride, 5 mM EDTA, 5 mM N-ethylmaleimide and 0.02% (w/v) sodium azide. The mucus solution was adjusted to a density of 1.4 g mL⁻¹ with CsCl and was then centrifuged (55 000 rpm at 10 °C for a duration of 62 h). 1 mL aliquots were sampled, absorption at 280 nm was measured, and 2 μL of each fraction was spotted and stained with alcian blue. UV and alcian blue-positive aliquots were pooled and diluted in extraction buffer lacking guanidinium hydrochloride (final guanidinium concentration 0.5 M), adjusted in density to 1.4 g mL⁻¹ with CsCl, and centrifuged again (50 K rpm at 10 °C for 96 h). One-mL aliquots were sampled, measured at 280 nm and stained with alcian blue. The fraction at 1.4–1.55 g mL⁻¹ was strongly alcian blue-positive and had very weak absorption at 280 nm, identifying it as the mucin fraction. This fraction was freeze dried and used.

2.2. Chromatographic column preparation

The chromatographic column was made specifically for the purpose of the present IGC experiment, using a short stainless steel 1/4′′ diameter tube that was washed in an ultrasonic acetone bath and then dried prior to use. Approximately 2 g of the above mucus was packed in the column by means of vertical tapping. Column loading was assisted with a mechanical vibrator and a vacuum pump. The end of the column was plugged with a small piece of glass wool and was connected to the vacuum pump. In order to precondition the biopolymer matrix and to remove any contaminants and humidity, the column was first operated at the maximum working temperature (60 °C) and the applied carrier gas flow rate (about 20 mL min⁻¹). The same column conditioning was repeated at the beginning of every working day until the detector signal was stabilized. The column specifications are listed in Table 1.

Table 1 Column specifications and chromatographic conditions

Column length (mm)	150
Column O.D. (mm)	6.4
Column I.D. (mm)	4.7
Mass loading (g)	2.08
Injector temp. (°C)	150
Detector temp. (°C)	200
Oven temp. (°C)
• Conditioning	60
• 1^st exper. series	40, 45, 50, 60
• 2^nd exper. series	33, 35, 37, 38, 39, 40, 41, 42, 43, 44
Room temp. (°C)
• 1^st exper. series	20 ± 2
• 2^nd exper. series	25 ± 1
Flow rate (mL min^−1)
• 1^st exper. series	20.1 ± 0.5
• 2^nd exper. series	18.5 ± 0.3

---

2.3. Physico-chemical characterization

The examined mucin sample was taken from the packing material of the chromatographic column upon completion of the IGC experiments, so it has been conditioned up to 60 °C for a long period of time. Temperature modulated differential scanning calorimeter scans (TMDSC) were performed using a TA Instrument (TA Q2000). A heating rate of 5 °C min⁻¹ with a temperature modulation amplitude of 1 °C and a period of 60 s were applied. Nitrogen gas flow of 50 mL min⁻¹ was purged into the DSC cell. The sample mass was about 5 mg inside a hermetically sealed aluminum pan. Conditioning thermal cycles were carried out prior to TMDSC analysis in order to remove any remained humidity as well as to erase thermal history. During these cycles the sample was heated just before the degradation onset above 140 °C and quenched down to −20 °C for a couple of times. Equilibration was performed at the limits of the pretreatment cycles. TMDSC analysis was carried out in the temperature range from −20 up to +230 °C. Thermogravimetric analysis (TGA) was performed using a TA Instrument (TGA-Q50) at a heating rate of 5 °C min⁻¹ and with a nitrogen gas flow of 40 mL min⁻¹. The sample mass was about 6 mg and heated from room temperature up to 250 °C. Fourier transform IR (FTIR) spectra were obtained using a Thermo Nicolet 380 spectrometer fitted with a SmartOrbit reflection accessory (Thermo Electron Corporation, Madison, WI). X-ray diffraction (XRD) measurements were accomplished for the as received mucin material with a PANalytical X'Pert Pro diffractometer (PANalytical, Almelo, The Netherlands), using monochromatic CuKα radiation. The XRD patterns were recorded in steps of 0.02° intervals and 1 s counting time at each step.

2.4. IGC experimental setup

Inverse gas chromatography measurements were carried out on a Hewlett Packard HP 5890 gas chromatograph, equipped with a flame ionization detector (FID). High purity helium (99.999%) was used as the carrier gas. The flow rate of helium was measured with a soap bubble flowmeter at room temperature and was set at about 20 mL min⁻¹. Two series of IGC experiments were performed at oven temperature ranges of 40–60 °C and 33–44 °C, and at varied intervals. Room temperature was 20 ± 2 °C and 25 ± 1 °C in the first and the second experimental series, respectively. The applied chromatographic conditions are tabulated in Table 1.

A volume of 0.1 μL of probes were manually injected with a 1 μL Hamilton syringe. It has been verified that this injected volume is at least three orders of magnitude lower than the carrier gas volume that passed the column until the complete elution of the probe vapor. Under these conditions of practically infinite dilution, the probe–probe interactions can be regarded negligible so that the system can be considered to be dominated by stationary phase–probe interactions. A series of probes of known properties and of the highest commercially available purity (analytical or HPLC grade) were injected into the packed column of mucin. n-Alkanes comprising of 6 up to 10 carbon atoms were used as non-polar probes in order to evaluate the London dispersive interactions with the column matrix. Several polar probes of basic, acidic and amphoteric character, namely methyl acetate, ethyl acetate, tetrahydrofuran (THF), chloroform, acetonitrile, pyridine, ethanol and n-butanol, were utilized in studying the specific, non-dispersive interactions (Lewis acid–base) with the stationary phase of mucus. Physical and chemical properties of the tested probes are provided in ESI (Table S.1†). Methane was used as the non-retained marker for the calculation of the packed column dead volume. For every tested temperature, repetitive injections were performed for each probe and the average net retention volume, V_n, was calculated by the retention time, t_R of at least two injections with a standard deviation of less than 4% in t_R.

2.5. IGC theoretical background for data reduction

In IGC study, all physicochemical parameters are derived from measurements on the chromatogram for the determination of the retention times of the probe, t_R, and the unretained marker, t_M. The retention time, t_R, is the time the average molecule of solute takes to travel the length of the column and coincides to the midpoint of the symmetrical breakthrough curve. A part of this time (t_M) is required by the solute simply to pass through the mobile phase space from inlet to outlet. This time is found by the elution of an unretained marker such as methane. In practice, the observed peak profiles of the probes are not perfectly symmetrical even at infinite dilution. Peak distortion is usually due to non-linear (concentration-dependent partitioning effects) or non-ideal (mass transfer limitations between phases) chromatography, and it is more likely to be observed in the elution profile of polar probes.^6–8 A precise and meaningful way to identify t_R of peaks of either Gaussian or non-Gaussian shape is provided by the theory of statistical moments. The first statistical moment corresponds to the elution time of the centre of gravity of the peak, since it is obtained by weighting the elution time of each point in the peak by its concentration, and as that, it constitutes a good approximation of the true retention time at infinite dilution:^6–8

The fundamental parameter in IGC study is the net retention volume, V_n, that relates directly the thermodynamic interaction of the volatile probe with the surface. It expresses the volume of the carrier gas required to sweep out the injected amount of a given probe from the column, and it is calculated from the collected data of the retention times t_R and t_M according to the relation:⁶

where V_n is expressed in cubic centimeters, F_M is the carrier gas flow rate measured in cm³ s⁻¹, at ambient pressure P′ (bar) and temperature T_F (K), by using a (soap bubble) flowmeter. T is the column temperature in Kelvin, P_w is the vapor pressure of water in units of bar at ambient temperature (T_F), and j is the James–Martin compressibility correction factor. P_w is calculated by the parametric formula:^9,10For water, the parameters of eqn (3) are equal to A = 73.64, B = −7.2582 × 10³, C = −7.3037, D = 4.1653 × 10⁻⁴ and E = 2, and they are valid in the temperature range of 273.16 K < T_F < 647.13 K. The correction factor j is dimensionless and is given by:⁶where P_i, P_o is the carrier gas pressure at the column inlet and outlet, respectively, with the last taken as barometric (P_o = 1.01325 bar).

The net retention volume at 0 °C per unit weight of the stationary phase is expressed as specific retention volume, V⁰_g:⁶

where m_s is the mass of the stationary phase of the column. The specific retention volume is characteristic of any particular combination of solute, stationary phase and carrier gas. It is related to many equilibrium properties and serves to the graphical determination of the glass transition temperature (T_g) of the polymeric stationary phase by plotting ln V⁰_g versus the reciprocal of the absolute temperature. In the absence of any transition, ln V⁰_g varies linearly with 1/T, however, when the polymer undergoes glass transition, a typical Z-shaped retention diagram can be observed.^6,11 Such a retention diagram is most likely to be observed for the retention of poor solvents, such as hydrocarbons that are only adsorbed on the surface of a polymer. Good solvents like polar probes usually present linear retention diagrams in an extended temperature range that includes T_g, probably because of their ability to interact specifically with the polymer and to penetrate into the bulk even at temperatures far below the glass transition.¹²

Based on basic thermodynamic considerations at infinite dilution, where the probe–probe interactions are negligible, the free energy of adsorption can be related to the net retention volume according to the relation:¹³

ΔG^ads = −RT ln V_n + C (6)

where R is the universal gas constant (8.314 J K⁻¹ mol⁻¹). The constant C depends on the choice of the reference state.^13,14 It can be written as a function of the probe vapor pressure in the standard gaseous state, p_s,g, the surface pressure, π, and the specific surface area of the stationary phase, A:¹⁵

According to the basic thermodynamic relation, the free energy of adsorption is related to the enthalpy of adsorption and the change of the entropy as following:

ΔG^ads = ΔH^ads − TΔS^ads (8)

In respect to eqn (6) and (8), the calculated first-order temperature derivatives of ΔG^ads result in the equality:Combination of eqn (6), (8) and (9) gives:

ΔH^ads = C (10)

2.5.1 Dispersive interactions. Near zero coverage, where the interactions between the probe molecules themselves can be assumed negligible, the injected probe interacts solely with the solid stationary phase by either dispersive or specific forces. The dispersive interactions are developed due to London forces between instantaneous dipoles, and occur between the probe and the solid surface no matter how dissimilar their chemical natures may be.^13,16 Surface ability to undergo dispersive interactions is represented quantitatively by the dispersive component of the surface energy, γ^d_s.^17,18 In IGC technique, the determination of γ^d_s is normally based on the measurement of the retention volume of a series of n-alkanes, since their interactions with the surface can be regarded of purely dispersive nature.¹⁵ The most frequently applied method for the determination of γ^d_s is the one proposed by Dorris & Gray that considers the increment of the free energy of adsorption per methylene group:¹⁴where N is the Avogadro's number, and ΔG^CH₂ is the free energy of adsorption of a single methylene group of cross-sectional area α^CH₂ = 6 Ų. ΔG^CH₂ is obtained experimentally from the retention volumes V_n+1 and V_n that are measured for two subsequent n-alkanes of n + 1 and n carbon atoms, respectively:¹³

Also in eqn (11), γ^CH₂ is the surface energy of a solid body constituted solely of methylene groups (polyethylene-type macromolecule). It is given by the expression:

γ^CH₂ = 35.6 + 0.058(293.15 − T) = 36.8 − 0.058t (13)

where t is the temperature in celsius degrees and T is the absolute temperature in kelvin.^13,19

2.5.2 Specific interactions. When a polar probe is injected in the chromatographic column, both dispersive and specific interactions between the test solute and the solid surface are exhibited. It is generally assumed that these two types of interactions are independent of each other and are additive.²⁰ Thus, the specific component of the free energy of adsorption, ΔG^sp, is determined from the difference between the free energy of adsorption of the polar probe and its dispersive contribution (ΔG^d), and leads to an increased net retention volume, V_n, for the polar probe compared to that of a reference apolar solute, V_n,ref:^13,18

In that sense, it is hypothesized that the adsorption of the reference solute is exclusively due to London forces and gives the same rise in the free energy of adsorption with the dispersive interactions of the polar probe. Consequently, on a plot of RT ln V_n versus a physicochemical property of the reference state of solutes, the difference of ordinates between the point corresponding to the specific polar probe and the reference line of n-alkane series gives the specific component, ΔG^sp.^13,21

For the representation of the reference state of the test solutes, three methods prevail in the literature; (I) the product of the surface area and the square root of the dispersion component of the surface energy (αγ^d_l), (II) the logarithm of the saturation vapor pressure, and (III) the boiling point. Panzer and Schreiber²² compared the three methods for the graphical representation of RT ln V_n data of various probes on polycarbonate and found almost identical values for ΔG^sp as well as for the specific component of the enthalpy of adsorption, ΔH^sp. The authors proposed the third representation method as the most convenient, since only boiling point data are readily available in the literature for numerous solutes. The assumption that the similarity in boiling points of the polar test solutes and the reference n-alkanes reflects a similarity in degree of nonspecific adsorption, implies that the degree of nonspecific interaction is related to the ease of condensation of the adsorbates.²³

However, scaling the degree of nonspecific interaction with condensability may not always lead to successful correlations. Donnet and coworkers,²⁰ and van Asten and coworkers¹⁵ showed that the aforementioned three correlation methods failed in the analysis of the adsorption data on solids of relatively high γ^d_s, since they all result in the calculation of negative ΔG^sp for most polar probes. However, they found reasonable ΔG^sp values by plotting RT ln V_n data against the molar deformation polarization of the adsorbates:

where (n² − 1)/(n² + 1) is the Lorenz–Lorentz refraction ratio and n is the refractive index measured with optical frequency radiation. Lorenz formula assumes that a material is made up of spherical molecules through which light travels slower than in the vacuum.²⁴ Since the ratio M/ρ of the molar mass with respect to density expresses the molar volume, V_m, eqn (15) can be rewritten as:

From eqn (15) and (16), it is obvious that P_D is expressed in volume units and is a function of data that are readily available in literature²⁵ or they can be easily calculated by parametric formulas.^9,10 Thus, the representation of the reference state with respect to molar deformation polarization offers a convenient calculation method for ΔG^sp, whilst the relation between the dispersive interactions and P_D can be justified theoretically: the London dispersion forces arise from the interaction of fluctuating electronic dipoles with induced dipoles in neighboring atom or molecules, and they depend on the electrical properties of the interacting elements and the distance between them.¹⁷ Polarizability has been introduced by London²⁶ in order to express the ease with which the molecule's electrons can be displaced by an electric field, and is directly proportional to the induced dipole moment. Straightforward, therefore, is the direct relation between the dispersion interactions and polarizability.²⁷

2.5.3 The homomorph concept. The dispersive contribution in the free energy of adsorption is attempted in the present study to be determined by representing the polar probes with reference apolar solutes in the context of homomorph concept. This concept is not novel in thermodynamic literature and proposes that homomorphs are structurally similar molecules having the same or closely similar molecular dimensions.²⁸ For the specification of the corresponding homomorph of a particular adsorbate, various approaches are presented in the literature and all assume similarity in physical or London interactions. Among these approaches, differences can be detected in the applied criterion for the specification of homomorphs, which could be the similarity of either polarizabilities^29,30 or molecular mass and dimensions.^31,32 Here, the homomorph of a non-hydrocarbon is considered to be a hydrocarbon molecule that retains the structure and types of bonds of the original molecule and simulates as close as possible its volume or solvation cavity. For the specification of the hydrocarbon homomorphs, the heteroatoms of the non-hydrocarbon (i.e. polar probes) such as O and N are replaced by a carbon atom or a proper CH_x group (CH, CH₂ or CH₃) in respect to the required valences and bonds. Bulkier halogen atoms are replaced by larger alkyl groups. In particular, it is considered that the ideal substituent of Cl atoms falls between methyl and ethyl group.³³ For the determination of ΔG^sp, the RT ln V_n data are plotted against the molar volume (V_m) of the tested adsorbates, with the polar probes to be represented by the molar volume of their corresponding homomorphs. Again, the difference of ordinates between the specific polar probe and the point of n-alkanes reference line at the same abscissa gives the specific component of the free energy of adsorption.

2.5.4 Assessment of the acid–base character of solid surface. The surface ability to interact specifically with a given solvent by forces other than London dispersive is essentially ascribed to Lewis acid–base interactions or otherwise electron donor–acceptor interactions. According to this concept, strong specific interactions can be only developed between an acid and a base, whilst an adsorber–adsorbate pair of the same character (both acids or bases) exchanges nearly zero specific interactions no matter how high their polarity is.¹³ In order to assess the acid–base properties of the surface, the enthalpy of the specific interactions, ΔH^sp, is determined by the slope of the linear plot of ΔG^sp/T as a function of reciprocal temperature, T⁻¹. The plot is depicted for a set of polar probes of acid or base character or of both characters (amphoteric):

The polar character of the probes is classified on a relative scale by empirical parameters of Gutmann approach.³⁴ The evolution stages of this approach were discussed in detail by Mukhopadhyay and Schreiber in their comprehensive review on the considerations about acid–base interactions.³⁵ As the authors stated, Gutmann scale is essentially an arbitrary and relative scale of the electron donor or electron acceptor ability of organic solvents rather than a universal and absolute scale of their acid–base characteristics. However, a significant practical importance has been gained in materials characterization by IGC technique, since it uses only two parameters and permits the evaluation of the acid or base and amphoteric functionalities of the tested solids.^22,35,36 Quantitative determination of acid–base interaction parameters for the tested solid surface is based on an empirical function that relates the specific interactions between solid surface and a polar probe:^16,37

−ΔH^sp = K_ADN + K_BAN* (18)

where DN and AN* are Gutmann's donor and modified acceptor numbers, respectively, and K_A and K_B are indices that reflect the acidity (electron acceptor) and basicity (electron donor), respectively, of the surface. Modifications of the above function with the insertion of additional parameters were received limited attention due to quite high empiricism and the marginal improvement of the results.^35,38 Rearranging eqn (18), ΔH^sp/AN* is plotted versus DN/AN* for a series of polar probes, producing a straight line of slope K_A and intercept K_B. With respect to the ratio of the two indices, the solid surface is classified as basic if K_B/K_A > 1, while it is considered as acidic if K_B/K_A < 1.^39–41

3. Results & discussion

3.1. Physicochemical characterization

Thermal analysis of the mucin of the column material via TMDSC and TGA showed a humidity content of only 4% that should be absorbed during the sample's handling. No crystallization or melting transitions were observed, while the decomposition onset was detected above 140 °C. A second-order phase transition, characteristic of glass transition, was identified at 82.5 °C. Meanwhile, XRD analysis demonstrated that the mucin material is completely amorphous. These findings confirm that the solid stationary phase is certainly an amorphous glassy up to the maximum working temperature of 60 °C in IGC study, and provide the necessary conditions for the thermodynamics applied in the determination of surface properties (discussed in §2.5.).

The significant compositional features of secreted mucins were identified by FTIR analysis. In particular, the obtained spectra displays the characteristic fingerprint regions of amides I and II, of glycans and of hydrogen-bonding between hydroxyl groups, which can be related to the polypeptide rigid backbone, the glycosylation coverage of the stiff robs, and the water bound on glycans, respectively.

The obtained results of the physicochemical characterization via TMDSC, TGA, XRD and FTIR are discussed in detail in ESI.†

3.2. Surface characterization by IGC

Two series of IGC experiments were performed in the temperature ranges of 40–60 °C and 33–44 °C at varied intervals (Table 1). Selection of these low temperature ranges and their respective intervals was based on the need to obtain data below the glass transition (T_g = 82.5 °C) of the amorphous mucin matrix, as well as close to the physiological temperature of a pig's functioning intestine, as described below. More specifically, the investigation of the range between 30 and 60 °C aimed at the surface characterization through the assessment of its ability to exhibit dispersive and specific interactions. An increment of 10 °C was considered adequate from 50 to 60 °C, while the closer interval of 5 °C between 40 and 50 °C was intended to the detection of any weak transitions of the polymer that could not be clearly depicted by DSC. Focus on the low temperature transitions was further given by the investigation of the lower range between 33 and 44 °C of the second experimental series that deals with the retention of only n-alkanes. The lower limit of 33 °C was set as to comply with the limitations of the GC oven, while the upper limit of 44 °C approximated a maximum tolerable temperature of a living porcine body. The body temperature of pigs normally varies between 39 and 40 °C, and can rise by up to 4 °C in serious pathological situations.⁴² Any changes in IGC response within this narrow range that lies between health and serious illness of pigs may signal essential changes in the mucin function.

3.2.1 Eluted peaks profile. The IGC measurements were performed under infinite dilution conditions, assuring negligible interactions between the probe molecules. Fig. 1 presents the chromatographic peaks obtained for the elution of the tested probes at 40 °C, using logarithmic scale in X axis for reasons of better representation at the largest elution times. The depicted peak profiles are representative of the probes elution at every tested temperature. As shown in Fig. 1a, Gaussian peaks were obtained for all tested n-alkanes, indicating that the measurements were performed at the linear part (known as the Henry's region) of the adsorption isotherms. It was verified that both the peak shape and the retention times (t_R) remained unchanged for repetitive injections of varied microvolumes (0.05–3 μL). Small deviations from symmetry can be detected in Fig. 1b for the faster-eluting polar probes (i.e. methyl acetate, ethyl acetate, THF and chloroform), while a substantial broadening and tailing at the last edge is obvious in Fig. 1c for the peak profiles of the slower-eluting polar probes (i.e. acetonitrile, pyridine, n-butanol and ethanol).

Fig. 1 Eluted peak profile of (a) n-alkanes, and (b) faster, (c) slower eluting polar probes with the packing material of mucin at a column temperature of 40 °C.

At infinite dilution, asymmetric spreading of gradually broadening peaks can be attributed to slow kinetic processes such as slow desorption from sites of high adsorption energy and very slow diffusion-controlled mass transfer between phases.⁶ It is regarded, in general, that adsorption kinetics of mucins are very slow.⁴³ Moreover, it is well-known that below T_g, the rate of diffusion of the probe through the polymer is too slow to permit appreciable bulk interaction, rendering surface adsorption as the only possible mechanism for probe–polymer interactions. Above T_g, the increasing permeability of the polymer due to extended thermal segmental motions permits the solvent diffusion, thus, changing the retention mechanism to bulk absorption.⁴⁴ Nastasovic and coworkers⁷ observed for benzene elution at infinite dilution and a wide temperature range that includes T_g, peak skewing and broadening at the last edge, by testing a modified copolymer of poly(glycidyl methacrylate-co-ethylene glycol dimethacrylate) with ethylene diamine (PGME-en). Since they obtained well-defined Gaussian peaks for alkanes (C5–C8) and other polar probes (chloroform, ethyl acetate, diethyl ether, tetrahydrofuran and cyclohexane) using the same polymer and identical conditions, they ascribed the distortion of benzene peak to diffusion resistance by the polymer, possibly due to partitioning of bulk absorption even at temperatures below T_g. Polars absorption below T_g was also reported by Guillet and coworkers¹² who investigated the interaction of acetic acid, butyl alcohol, a-chloronaphthalene, naphthalene, and hexadecane with poly(N-isopropylacrylamide). They found that at temperatures below T_g, only the first two solutes are able to penetrate the polymer and diffuse into the bulk due to a mechanism associated to their ability to form hydrogen bonds and to interact more strongly with the polymer. The rest solutes were considered as weakly interacting with the polymer, because they exhibit smaller excess free energy of mixing below T_g probably due to their larger size. In other research works, the tailing of the last peak edge at very low loadings of polar probes was ascribed to concentration dependent adsorption caused by the energetic heterogeneity of the active sorption sites.^15,45,46

Based on the above literature remarks, it is assumed that the observed considerable peak distortion for acetonitrile, pyridine, n-butanol and ethanol should be ascribed to desorption retardation due to the ability of the specific polar probes to access either adsorption sites of higher energy or extra sites into the polymer bulk. Further investigation of the concentration dependence is required in order to clarify whether the tailing elution is provoked due to the energetic heterogeneity of the surface, i.e. concentrated adsorption capacity in special sites, or due to mass transfer limitations through the polymer, i.e. supplementary absorption mechanism from the bulk. The present data suffice, however, to conclude that the stronger sorption of the specific probes is not due to special acid–base interactions, since they exhibit different polar functionalities, but it should be related to the combination of their smaller size (V_m) and/or higher condensability (T_b).

3.2.2 Dispersive interactions. The evaluation of surface ability to exhibit dispersive interactions is based on the retention data of n-alkanes that were obtained in the two experimental series of temperature ranges 40–60 °C and 33–44 °C. In Fig. 2, the perfect linear form of the retention diagram of n-alkanes indicates the absence of any transition over the whole tested temperature range (33–60 °C). Considering that the polymer is in the glassy state in this range, it is inferred that retention occurs mainly by surface adsorption.¹¹ The observed rise of V⁰_g with decreasing temperature and increasing carbon chain length, manifests that adsorption is governed by physical interactions favored by probes condensability. Analogous conclusions are drawn by the reduction of the RT ln V_n component of the free energy of adsorption with carbon number decrease and temperature rise, which is depicted in Fig. 3 (and ESI in Fig. S.5†).

Fig. 2 Retention diagram of n-alkanes at (a) 40–60 °C (1^st experimental series), and (b) 33–44 °C (2^nd experimental series).

Fig. 3 Variation of RT ln V_n data of n-alkanes with carbon number at (a) 40–60 °C, and (b) 33–44 °C.

The slope of the linear regression of RT ln V_n data versus the carbon number corresponds to the increment of the free energy of adsorption per methylene group, ΔG^CH₂, and determines the dispersive component of the surface energy, γ^d_s, according to Dorris & Gray method. The coefficients of the linear regression are provided in ESI (Table S.2†). The temperature dependence of γ^d_s is shown in Fig. 4 and is characterized by the linear decrease of γ^d_s with T, which is indicative of the absence of any surface modifications by a temperature dependent process (e.g. phase transitions, chemical or physical changes in non-conditioned polymers). The decrease is attributed to the weakening of dispersive interactions with temperature rise due to entropic contribution to the surface free energy as well as to limited condensation of the probes.^15,47,48 The values of γ^d_s vary from about 40 mN m⁻¹ at 33 °C to about 32 mN m⁻¹ at 60 °C, and are typical for natural carbohydrate polymers such as lignocellulosic polymers like cellulose, pine wood and kenaf,⁴⁷ plant fibers⁴⁹ and okra hydrocolloid.^39,40 The depicted confidence intervals of γ^d_s values were calculated by usual calculation procedures of error propagation. The width of intervals together with the small differences in the conditions of the two experimental series (carrier flow rate and room temperature), could explain the slight deviation between the two lines of γ^d_s vs. T.

Fig. 4 Temperature dependence of the dispersive component of the surface energy, γ^d_s, determined by the two experimental series at 40–60 °C and 33–44 °C.

3.2.3 Specific interactions. Assessment of the ability of the mucin surface to exhibit specific interactions was based on retention data of acidic (chloroform, n-butanol), basic (THF, pyridine), and amphoteric (methyl acetate, ethyl acetate, acetonitrile, ethanol) probes in the temperature range of 40–60 °C (1^st experimental series). The retention diagram and the temperature variation of RT ln V_n data are presented in Fig. 5. The near-perfect linear correlation of ln V⁰_g versus 1/T with no discontinuity suggests the absence of transitions. In addition, the decrease of temperature or the increase of the probes condensability (T_b) seems to favor the rise of V⁰_g and RT ln V_n, however, without clear scaling. Particularly, the highest and lowest values of these data are presented by ethanol and ethyl acetate, respectively, despite the similar boiling points of the two solutes. Furthermore, the temperature effect on RT ln V_n data is negative for ethanol, n-butanol and acetonitrile, negligible for pyridine, THF, chloroform and ethyl acetate, and positive for methyl acetate. As a matter of fact, two groups of polar probes can be discriminated of stronger and weaker retention with respect to V⁰_g and RT ln V_n values, or of slower and faster desorption with respect to the elution profiles (Fig. 1b and c), however, without relevance to their specific polar character. This fact, on the one hand, implies the presence of both acidic and basic active sites in the mucin surface, and, on the other hand, suggests the significant effect of the exhibited specific interactions and their temperature dependence on the retention of the polar probes that diverges from the typical behavior of physical sorption. It is also pointed out that regardless the tailing of the eluted peaks, the confidence intervals of RT ln V_n data are narrow (Fig. 5b), confirming that the use of the first statistical moment for the elution time ensures the accuracy and reproducibility of retention calculations.

Fig. 5 Adsorption of the polar probes at 40–60 °C: (a) the retention diagram, and (b) the temperature variation of RT ln V_n component of the free energy of adsorption.

The specific component of the free energy of adsorption of polar probes, ΔG^sp, was estimated by subtracting the respective contribution of dispersive interactions, as depicted schematically for 40 °C in Fig. 6. Analogous graphs were made for every tested temperature (not presented here). The reference state of the probes was represented in terms of boiling point (Fig. 6a), molar deformation polarization (Fig. 6b) and molar volume (Fig. 6c). It was assumed reasonable, the molar volume of adsorbates not to be taken constant but to be calculated for every tested temperature by appropriate equations.^9,10 Molar deformation polarization values were also calculated for every tested temperature from the molar volume and refractive index data (Table A.1 of ESI†) by applying eqn (16). In the plot of RT ln V_n versus V_m, the polar probes were represented by the molar volume of their corresponding hydrocarbon homomorphs. The applied selection considerations for the specification of homomorphs concern, at first, the selection of hydrocarbon molecules that retain the structure and types of bonds of the corresponding polar solutes and bear an appropriate CH_x group in the position of O and N heteroatoms, or an ethyl group in the position of Cl. These considerations are refined below in respect to the estimated values of the enthalpy of specific interactions (ΔH^sp), and corrected homomorph structures are identified for specific probes. Election of ΔH^sp as evaluation criterion of the correlation methods is considered appropriate, since it expresses in a quantitative manner the potential of a particular adsorbent–adsorbate pair to exhibit specific interactions. Details for the correspondence of the tested polar solutes with homomorph hydrocarbons and the physical properties of the specified homomorphs are presented in Tables S.3 and S.4 of ESI.†

Fig. 6 Estimation of the specific component of the free energy of adsorption at 40 °C by the representing reference state of probes in terms of: (a) boiling point, (b) molar deformation polarization, and (c) molar volume. In the last representation, the polar probes are matched to the molar volume of their corresponding hydrocarbon homomorphs.

Fig. 6 shows that in all correlation methods, the free energy of polar probes lie above the defining reference line of London dispersive interactions, indicating the presence of specific interactions. However, the contribution of the specific interactions estimated by the correlation of boiling point is much lower compared to those calculated by the other two correlations, where the data points for the polar probes are situated more to the left-hand side of the n-alkanes reference lines. As shown in Fig. 7, the representations of reference state by P_D and V_m attribute higher ΔG^sp values, which, for most polar probes, compare well between each other within the specified confidence intervals. Such an agreement is reasonable since the similarity in the molecular size and shape between the polar solute and its corresponding homomorph implies an equality of mean electronic polarizabilities.²⁹ However, for methyl acetate, ethyl acetate and chloroform, the correlation of P_D results in ΔG^sp values much higher than those obtained by the correlation of V_m, while the later show an increasing trend with temperature. In contrast, the correlation of P_D displays the expected decline of the specific interactions with temperature rise due to molecular dilatation and the resulted increasing distances between the interacting molecules. The depicted confidence intervals can accommodate any fluctuation from this trend. Deviations from this trend can be also noticed for most polar probes in the correlation of T_b.

Fig. 7 Temperature influence on the specific component of the free energy of adsorption of the polar probes at 40–60 °C determined by the correlation of: (a) boiling point, (b) molar deformation polarization, and (c) molar volume of corresponding hydrocarbon homomorphs.

A clearer picture of the temperature effect on the specific interactions can be drawn by the tabulated ΔH^sp values in Table 2, which were estimated for each correlation by applying eqn (17). The correlation of boiling point results in positive ΔH^sp values for most polar probes, and obviously suggests the intensification of specific interactions with increasing temperature. A similar temperature effect is suggested by the V_m correlation for chloroform and the two ethers. However, this effect could only be true if the polymeric matrix exhibited changes with temperature rise that could unfold additional active sites for specific interactions. In contrast, the linearity of both retention diagrams and temperature dependence of γ^d_s indicates the absence of such modifications. It is established, therefore, that boiling point fails to represent the contribution of the dispersive interactions of the polar probes. The failure of V_m correlation for the specific polars could be attributed to their representation by quite large hydrocarbon homomorphs. Corrections of these particular homomorphs by using smaller substituents by a CH_x group, increase −ΔH^sp above the respective values obtained by the correlation of P_D. Thus, it is considered that the ideal hydrocarbon homomorph falls between isobutene and 2-methyl-1-butene for methyl acetate, 2-methyl-1-butene and 2-methyl-1-pentene for ethyl acetate, and isobutene and 3-ethyl-pentane for chloroform. In contrast, corresponding pyridine to toluene, that is the next higher aromatic hydrocarbon of benzene, gives a better agreement with the correlation of P_D. The necessity for a higher substitution for nitrogen could probably be attributed to its lone electron pair that is not involved into the aromatic system of π-electrons but projects outward the ring. For the rest polar probes, the initial selection considerations of replacing the functional group by an appropriate CH_x group are considered convenient; hence, THF, acetonitrile, n-butanol and ethanol correspond to cyclopentane, propyne, n-pentane and propane, respectively. Of course, these are empirical considerations since the very concept of the homomorphy is an empirical one.

Table 2 Comparison of −ΔH^sp values obtained for the polar probes by the correlations of boiling point, molar deformation polarization, and molar volume of corresponding homomorphs

Polar probe	−ΔH^sp (kJ mol^−1)
	Boiling point	Molar deformation polarization	Molar volume of homomorphs	Correction^a of homomorphs
a Consult Tables S.3 and S.4 of ESI.
Methyl acetate	−10.94 ± 2.14	−0.98 ± 2.77	−5.65 ± 2.86	1.35 ± 2.94
Ethyl acetate	−4.39 ± 1.68	4.55 ± 2.13	−1.13 ± 2.14	5.50 ± 2.33
THF	−0.77 ± 2.51	7.90 ± 3.07	8.42 ± 3.40
Chloroform	−0.15 ± 1.23	6.16 ± 1.83	−7.64 ± 1.59	11.49 ± 1.93
Acetonitrile	12.71 ± 3.45	35.43 ± 4.77	39.94 ± 5.01
Pyridine	−7.41 ± 3.21	7.04 ± 3.50	14.43 ± 4.08	8.32 ± 3.81
n-Butanol	5.70 ± 3.17	23.03 ± 3.87	22.37 ± 3.84
Ethanol	10.53 ± 3.64	30.07 ± 4.58	37.69 ± 4.54

---

Based on the above remarks and considering the molecular properties of the polar probes and their corresponding homomorphs (Tables S.1 and S.4 of ESI†), it is inferred that the potential of most polar probes to exhibit dispersive interactions resembles that of hydrocarbons with similar molecular structure and size but of much lower condensability, especially for linear solutes. This fact reveals, on the one hand, that the similarity in boiling points between polar solutes and hydrocarbons does not reflect a similarity in the degree of dispersive interactions, and, on the other hand, that the accessibility of polar probes in the active sorption sites and the deployed specific interactions are assisted by a kind of molecular sieving mechanism. Such a mechanism can explain the high free energy of adsorption of ethanol, n-butanol and acetonitrile, that is greatly contributed by the specific component ΔG^sp. Undoubtedly, the high free energy of pyridine adsorption could not be favored by a molecular sieving mechanism, but it should be attributed to the ability of the aromatic ring to exhibit π-electron interactions and, as a result, high polarity.²³ It follows, then, that implementation of the homomorph concept for the interpretation of the enthalpy of specific interactions, although empirical, it provides with meaningful information for the dominant mechanism of retention of polar probes.

3.2.4 Acid–base properties. The specific interactions established between a polar probe and the solid polymer surface, are ascribed to Lewis acid–base interactions. The dominant polar character of the polymer surface is evaluated with respect to the Lewis acid, K_A, and Lewis base, K_B, indices that are obtained by the slope and the intercept, respectively, of the linear regression of ΔH^sp/AN* data against DN/AN*. For mucin surface, these calculations were performed using the set of ΔH^sp values of all tested polar probes that were obtained by the correlation of P_D. Regression results give K_A = 0.048 ± 0.003 and K_B = 0.839 ± 0.307 with R² = 0.9708, thus, the ratio K_B/K_A is equal to 17.54 ± 6.54. These values suggest that the mucin surface accommodates both electron acceptor and donor moieties of low acidity and moderate basicity, respectively, as compared with the K_A and K_B values of twenty kinds of polymers summarized by Baoli and coworkers.³⁶ Consequently, mucin surface is characterized, like most polymers, as amphoteric with predominant basic functionality.

4. Conclusions

The present work gives an insight into the surface adsorption onto the mucin layer of the porcine intestinal epithelial surface, as a first clue of the underlying thermodynamics that govern the processes of bioabsorption and digestion. The study is focused on IGC adsorption experiments at temperatures well below the glass transition temperature (T_g = 82.5 °C) of the amorphous mucin matrix. Very low probe loadings at infinite dilution were applied of several molecules both apolar (C6–C10 n-alkanes) and polar of basic, acidic and amphoteric functionality.

Determination of the surface characteristics showed that the retention of n-alkanes onto the mucus layer is favored by the temperature decrease and the carbon number increase, while it regresses linearly with these two factors. These variations suggest that the adsorption of apolar molecules onto mucin is governed by physical interactions, as well as that mucus layer does not face any transitions or surface modifications within the tested temperature range. The absence of any transition within the applied temperature range was also supported by TMDSC analysis. The ability of mucus layer to undergo dispersive interactions due to London forces was evaluated in terms of the dispersive component of the surface energy, γ^d_s. It was found that γ^d_s decreases linearly with temperature within moderate values (32–40 mN m⁻¹), which are typical for natural carbohydrate polymers. Assessment of the Lewis acid–base surface properties suggested that the mucus layer accommodates both electron acceptor and donor moieties of low acidity and moderate basicity, and thus it should be classified as an amphoteric polymer with predominant basic functionality.

Divergences from the typical behavior of physical sorption in the Henry's region were noticed for polar probes, with a greater extent for acetonitrile, pyridine, n-butanol and ethanol. This fact was attributed to the significant effect of the exhibited specific interactions that have as a result the stronger retention of polar probes and the retardation of their desorption from mucin surface. Reasonable free energy values of the specific interactions were obtained by representing the reference state of probes in terms of molar deformation polarization. In contrast, the most widely used representation of boiling point fails to express the expected decline of the specific interactions with temperature rise due to molecular dilatation. Implementation of the homomorph concept for the interpretation of the free energy and enthalpy of specific interactions, provided with meaningful information for the dominant mechanism of retention of polar probes. In particular, it was considered that the accessibility of polar probes in the active sorption sites and the deployed specific interactions are favored by their small molecular dimensions and high polarity.

Acknowledgements

The authors would like to thank Dr E. S. Papastergiadis, Dr M. Papageorgiou, and Dr I. Tsivintzelis for the kind assistance in XRD, thermal analysis, and IGC set up, respectively. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

References

1. A. N. Round, N. M. Rigby, A. Garcia de la Torre, A. Macierzanka, E. N. Clare Mills and A. R. Mackie, Biomacromolecules, 2012, 13, 3253–3261.
2. A. P. Corfield, Biochim. Biophys. Acta, 2015, 1850, 236–252.
3. K. Jumel, I. Fiebrig and S. E. Harding, Int. J. Biol. Macromol., 1996, 18, 133–139.
4. M. E. V. Johansson, H. Sjövall and G. C. Hansson, Nat. Rev. Gastroenterol. Hepatol., 2013, 10, 352–361.
5. S. Mohammadi-Jam and K. E. Waters, Adv. Colloid Interface Sci., 2014, 212, 21–44.
6. J. R. Conder and C. L. Young, Physicochemical Measurements by Gas Chromatography, Wiley, Chichester, 1979.
7. A. B. Nastasovic, A. E. Onjia, S. K. Milonjic and S. M. Jovanovic, J. Polym. Sci., Part B: Polym. Phys., 2005, 43, 524–2533.
8. J. R. Conder, G. J. Rees and S. McHalle, J. Chromatogr., 1983, 258, 1–10.
9. T. E. Daubert and R. P. Danner, Data Compilation Tables of Properties of Pure Compounds, AIChE Symp Ser. No. 203, American Institute of Chemical Engineers, New York, 1985.
10. T. E. Daubert and R. P. Danner, Physical and Thermodynamic Properties of Pure Compounds: Data Compilation, Hemisphere, New York, 2001.
11. A. B. Nastasovic and A. E. Onjia, J. Chromatogr., A, 2008, 1195, 1–15.
12. O. Smidsrød and J. E. Guillet, Macromolecules, 1969, 2, 272–277.
13. J. Schultz, L. Lavielle and C. Martin, J. Adhes., 1987, 23, 45–60.
14. G. M. Dorris and D. G. Gray, J. Colloid Interface Sci., 1980, 77, 353–362.
15. A. van Asten, N. van Veenendaal and S. Koster, J. Chromatogr., A, 2000, 888, 175–196.
16. E. Papirer, J.-M. Perrin, B. Siffert and G. Philipponneau, J. Colloid Interface Sci., 1991, 144, 263–270.
17. F. M. Fowkes, Ind. Eng. Chem., 1964, 56, 40–52.
18. A. Voelkel, Chemom. Intell. Lab. Syst., 2004, 72, 205–207.
19. A. Voelkel, B. Strzemiecka, K. Adamska and K. Milczewska, J. Chromatogr. A, 2009, 1216, 1551–1566.
20. S. Dong, M. Brendlé and J. B. Donnet, Chromatographia, 1989, 28, 469–472.
21. A. Voelkel, Stud. Surf. Sci. Catal., 1996, 99, 465–477.
22. U. Panzer and H. P. Schreiber, Macromolecules, 1992, 25, 3633–3637.
23. D. J. Brookman and D. T. Sawyer, Anal. Chem., 1968, 40, 106–110.
24. R. J. W. Le Fèvre, Adv. Phys. Org. Chem., 1965, 3, 1–90.
25. Physical Constants of Organic Compounds, Section 3 in CRC Handbook of Chemistry and Physics, ed. D. R. Lide, Internet version 2005, http://www.hbcpnetbase.com, accessed, October 2015.
26. F. London, Trans. Faraday Soc., 1937, 33, 8–26.
27. J. M. Prausnitz, R. N. Lichtenthaler and E. G. de Azevedo, in Molecular Thermodynamics of Fluid-phase Equilibria, Prentice Hall, NJ, 3rd edn, 1999, ch. 4, pp. 57–121.
28. H. C. Brown, G. K. Barbaras, H. L. Berneis, W. H. Bonner, R. B. Johannesen, M. Grayson and K. L. Nelson, J. Am. Chem. Soc., 1953, 75, 1–6.
29. H. P. Diogo, M. E. Minas da Piedade, J. J. Moura Ramos, J. A. Simoni and J. A. Martinho Simões, J. Chem. Educ., 1993, 70, A227–A233.
30. Y. Miranda, Á. Piñeiro and P. Brocos, J. Chem. Eng. Data, 2007, 52, 429–437.
31. M. Gepert and S. Ernst, J. Solution Chem., 2003, 32, 831–852.
32. M. P. Breil, I. Tsivintzelis and G. M. Kontogeorgis, Fluid Phase Equilib., 2011, 301, 1–12.
33. C. Panayiotou, Phys. Chem. Chem. Phys., 2012, 14, 3882–3908.
34. V. Gutmann, The donor–acceptor approach to molecular interactions, Plenum Press, New York, 1978.
35. P. Mukhopadhyay and H. P. Schreiber, Colloids Surf., A, 1995, 100, 47–71.
36. S. Baoli, Z. Qianru, J. Lina, L. Yang and L. Bin, J. Chromatogr. A, 2007, 1149, 390–393.
37. C. S. Flour and E. Papirer, Ind. Eng. Chem. Prod. Res. Dev., 1982, 21, 666–669.
38. T. Hamieh, M. Rezzaki, Y. Grohens and J. Schultz, J. Chim. Phys. Phys.–Chim. Biol., 1998, 95, 1964–1990.
39. M. Dimopoulou, C. Ritzoulis and C. Panayiotou, Colloids Surf., A, 2014, 475, 37–43.
40. M. Dimopoulou, I. Tsivintzelis, C. Ritzoulis and C. Panayiotou, RSC Adv., 2016, 6, 20916–20925.
41. S. K. Papadopoulou, G. Dritsas, I. Karapanagiotis, I. Zuburtikudis and C. Panayiotou, Eur. Polym. J., 2010, 46, 202–208.
42. P. G. G. Jackson and P. D. Cockcroft, in Handbook of pig medicine, Elsevier, UK, 2007, ch. 1, pp. 1–15, https://books.google.co.uk/books/about/Handbook_of_Pig_Medicine.html?id=BFQYKo6q3rUC, accessed, March 2015.
43. O. Svensson and T. Arnebrant, Curr. Opin. Colloid Interface Sci., 2010, 15, 395–405.
44. D. G. Gray and J. E. Guillet, Macromolecules, 1974, 7, 244–247.
45. E. Papirer, A. Vidal, W. M. Jiao and J. B. Donnet, Chromatographia, 1987, 23, 279–285.
46. T. I. Bakaeva, C. G. Pantano, C. E. Loope and V. A. Bakaev, J. Phys. Chem. B, 2000, 104, 8518–8526.
47. M. A. Tshabalala, J. Appl. Polym. Sci., 1997, 65, 1013–1020.
48. T. Hamieh, M. Rezzaki and J. Schultz, J. Colloid Interface Sci., 2001, 233, 343–347.
49. N. Cordeiro, C. Gouveia, A. G. O. Moraes and S. C. Amico, Carbohydr. Polym., 2011, 84, 110–117.

---

Footnote

† Electronic supplementary information (ESI) available: Physicochemical characterization of mucin (TMDSC, TGA, XRD, FTIR analyses), and supplementary IGC data. See DOI: 10.1039/c6ra23051b

---