PEDS Advance Access originally published online on October 25, 2006
Protein Engineering Design and Selection 2006 19(12):537-545; doi:10.1093/protein/gzl041
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Emulsifying performance of modular ß-sandwich proteins: the hydrophobic moment and conformational stability
Department of Pharmaceutical Sciences, University of Strathclyde 27 Taylor Street, Glasgow, UK
1To whom correspondence should be addressed. E-mail: chris.walle{at}strath.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Our understanding of protein emulsifying properties is largely based on analysis of emulsifiers found in milk and seed. The 9th–10th type III fibronectin domain pair retains full biological activity following emulsification–encapsulation into polyester microspheres, for controlled delivery, but the conformational criteria determining emulsification efficiency (EE) are unknown. Here, we have generated a series of mutants of this ß-sandwich protein, changing the hydrophobic moment and conformational stability, to investigate the structure–emulsification relationship. Predictive modelling of the hydrophobic moment of ß-strands and mutations known to increase conformational stability were used to generate the series. The proteins were tested for their emulsion stability and EE for oil-in-water mixtures. We show that the stabilization of emulsions by ß-sandwich proteins is best predicted by conformational stability during equilibrium denaturation in ionic surfactant. In contrast, the EE of these proteins is inversely related to an increase in their surface hydrophobicity following unfolding in surfactant. We also describe a novel ß-sandwich emulsifier with strong EE. The requirement for interdomain flexibility to achieve maximum emulsion stability and EE is also shown. This work increases our understanding of the mechanisms involved in protein emulsification and will be of use to the microencapsulation of proteins into polyester microspheres via emulsion-extraction protocols.
Keywords: Beta-sandwich/protein/structure/emulsification/unfolding
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Protein-stabilized emulsions have been studied extensively with respect to the properties of milk-proteins and seed-proteins (Creamer et al., 1998
; van der Ven et al., 2001; Burnett et al., 2002). Emulsions stabilized by proteins are of great importance to the food industry and are becoming increasingly important to the pharmaceutical industry; specifically, protein controlled release formulations via emulsion–encapsulation into polymeric microspheres (Carino et al., 2000; Egilmez et al., 2000; Robertson et al., 2001). During emulsification, proteins are generally accepted to adsorb at the oil/water interface and undergo conformational changes, generally unfolding (Sah, 1999; van de Weert et al., 2000). These changes promote hydrophobic and electrostatic interactions (Dickinson and Matsumura, 1991; Damodaran et al., 1998), and, where possible, disulphide-mediated polymerization (Dickinson and Matsumura, 1991), which are thought to stabilize the viscoelastic protein film around oil droplets.
Emulsifying proteins studied to date are often composed of 
-helical and/or random coil structure in aqueous solution and may lose or gain -helical structure during interfacial adsorption (Caessens et al., 1999; Burnett et al., 2002). The interfacial activity of the globular protein myoglobin has been shown to be due to alignment of amphipathic -helices at the interface (Poon et al., 1999), correlating with the calculated hydrophobic moment (Poon et al., 2001b). Thus, secondary structure clearly plays a role in emulsion stability. ß-lactoglobulin provides an example of a ß-sheet protein with emulsifying activity, comprising mainly ß-sheet and random coil but also -helical structure (Qin et al., 1998). The structural criteria governing the emulsifying properties of modular ß-sandwich proteins are less clear. Recently, the modular protein fibronectin (FN), was shown to strongly adsorb to the oil/water interface, displacing BSA even when at lower concentrations (Vaidya and Ofoli, 2005). This was suggested to be a consequence of the flexibility of FN in solution, afforded by its modular nature and cryptic self-association sites (Johnson et al., 1999), such that energetically favourable conformations at the interface were rapidly adopted. Thus, the emulsifying properties of modular ß-sandwich proteins depend upon domain–domain interactions in addition to secondary structure criteria. An understanding of their mechanism will be of importance to pharmaceutical formulation via emulsification and polymer encapsulation (Bouissou et al., 2004; Mordenti et al., 1999).
Here we have used the human fibronectin 9th–10th type III domain pair (9-10FNIII, Figure 1) as a scaffold on which to superpose residues bringing about change to the hydrophobic moment and conformational stability. Starting from a scaffold whose emulsification properties are uncharacterized may appear counter-intuitive, but the approach has several attractions: (i) it will employ and so test current hypotheses of protein structural criteria for emulsion stability/EE, (ii) the FN-type III fold is structurally well characterized down to specific contributions from individual residues (Leahy et al., 1996; Plaxco et al., 1997; Spitzfaden et al., 1997), and this is of great potential to the design of recombinant proteins with a range of specific characteristics aimed at dissecting emulsification properties at the structural level and (iii) de-novo design of emulsifying properties has previously been reported for a peptidic series (Saito et al., 1995), suggesting that a similar approach with a protein series is not unfeasible.
|
The work described uses site-directed mutagenesis to increase the hydrophobic moment (a measure of amphipathicity) of individual ß-strands within each domain. We also explore the role of domain stability using previously described 9-10FNIII mutants, which are progressively more stable to chemical denaturation (van der Walle et al., 2002
; Altroff et al., 2004). We have sought to avoid residues specifically aimed at causing hydrophobic surface patches on natively folded protein, because the emulsifying role of such regions is well understood (Toren et al., 2002). Similarly, our use of mutants of distinct conformational stability negates the use of chemical denaturant to evaluate the emulsification properties of partially to fully unfolded protein (Poon et al., 2001a). Structural changes in sodium dodecyl sulphate were used to acquire data regarding putative conformational changes and emulsifying properties (or ‘performance’ in the case of empirical data regarding emulsion stability) in relation to protein adsorption at the oil/water interface. |
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Calculation of the mean hydrophobic moment and mean hydrophobicityn
The X-ray crystallographic structure of 9-10FNIII shows each domain having seven ß-strands folded in a 3+4 arrangement into two opposing sheets exposed to solvent to varying extents (Leahy et al., 1996). For each ß-strand the mean hydrophobic moment, 
µH
, and mean hydrophobicity, H, was calculated. Table I shows those ß-strands falling outside the ‘surface seeking’ classification described by Eisenberg et al., i.e. a high µH above 0.25 and H between 0.3 and –0.3 (Eisenberg et al., 1982). In order to increase the hydrophobic moment of these ß-strands, amino acid substitutions were designed to increase µH, which should accordingly produce proteins with increasing surface-active potential. As shown in Table I, only localized amino acids substitutions were required to bring about 3- to 14-fold increments in µH, bringing the ß-strands into the surface-seeking classification. Table II shows the notations used for the resultant mutants in the following text.
|
|
Calculation of solvation energies and surface areas
Models of 9-10FNIII and the 9-10FNIII mutants expressed and tested were built using the 3D jigsaw comparative modelling server (http://www.bmm.icnet.uk/servers/3djigsaw/) (Bates et al., 2001). The models were refined by minimization and molecular dynamics using the GROMACS program (Lindahl et al., 2001). The quality of the models during refinement was continually assessed using the biotech validation suite for protein structures (http://biotech.ebi.ac.uk:8400/) (Laskowski et al., 1996; Pontius et al., 1996). For the refined models, the solvation energy and conditional hydrophobic accessible surface area were measured using the program CHASA (Cooper et al., 2005). Solvation energies were also compared with calculation by other methods (see Table III), performed using the solvation module of the Insight II program (Accelrys Software Inc., Cambridge, UK).
|
Construction of cDNA clones
The 9-10FNIII wild-type cDNA cloned into pRSET-a was received as a generous gift from Prof. H. Mardon, University of Oxford. Mutation of the 9-10FNIII cDNA template as directed by the amino acid substitutions described in Table I was made following the QuickchangeTM protocol (Stratagene, Amsterdam, The Netherlands). The constructs 9-10FNIII-P and 9-10FNIII-CC were created as previously described (Altroff et al., 2004; van der Walle et al., 2002). Table I also shows those constructs with increasing hydrophobic moment, created in similar fashion using appropriately designed primer pairs (Invitrogen, Paisley, UK). Showing only the sense strand, 9-10FNIII-IEI was achieved using 5'-CACTTCAGTGGGAGACCTATAGAAATTCGGGTGCCCCACTCTCGG, and 9-10FNIII-IDLEVRV was achieved in two stages using 5'-CAGTTTCTGATGTTCCGATTGACCTGGAAGTTGTTGC and then 5'-CCGATTGACCTGGAAGTTCGTGTTGCGACCCCCACCAGCC (mutations underlined).
Expression and purification of recombinant proteins
Polyhistidine-tagged 9-10FNIII and mutants were expressed in Escherichia coli BLR (DE3) pLysS and purified from the cell soluble fraction using metal chelation affinity chromatography as previously described (Altroff et al., 2001). Eluted protein fractions were desalted into 10 mM NaH2PO4, 50 mM NaCl pH 7 ('phosphate buffer') and concentrated to between 1 and 10 mg/ml. Purity was assessed by SDS–PAGE and size exclusion chromatography using a calibrated Superdex 75 HR 10/30 column (Amersham Biosciences); the column was equilibrated in phosphate buffer and proteins eluted under a flow of 0.5 ml/min, monitoring the eluate at 280 nm. Protein concentrations were calculated using the A280 value and a calculated extinction coefficient (
) of 21 620 M–1cm–1.
Measurement of emulsion performance
The turbidimetric method of Pearce and Kinsella (1978) was used to determine the emulsion activity index (EAI) (Pearce and Kinsella, 1978). Proteins were diluted into phosphate buffer to give final concentrations of 1.0 mg/ml and 1 ml added to 0.2 ml of peanut oil and homogenized (30 s, 14 000 rpm, IKA T18 homogenizer with a S18N 10G dispersing tool). The emulsion was immediately diluted 100-fold into 0.1% w/v SDS solution in phosphate buffer and the absorbance measured at 500 nm. The EAI = (2 x 2.303 x A500)/(
.C), where = the oil volume fraction and C = % w/v protein. Measurements were repeated in triplicate for independently prepared samples.
The method of van der Ven et al. (2001) was used to make quantitative measurements of EE. Emulsions were prepared by homogenization as above and diluted into 50 ml phosphate buffer within a stirred (2000 rpm) small volume sample dispersion unit (Malvern, UK). Oil droplet size was then measured by laser diffraction using Mie scattering theory with a laser obscurity 
12% until little further change was observed (Mastersizer 2000, Malvern Instruments, UK). The curves shown represent the change in the median diameter, d(0.5), of the oil droplets with time. Reduction of 9-10FNIII-CC was achieved by addition of 20 mM dithiothreitol (DTT) prior to homogenization. The EE (%) was calculated as the percentage difference in droplet size at 75 min with respect to the droplet size at 5 min (immediately after equilibration following dilution); negative values, therefore, indicating stronger EE than positive values. Measurements were repeated at least in duplicate for independently prepared samples.
The method of Willumsen and Karlson was used to calculate the emulsion stability index (ESI) (Willumsen and Karlson, 1997). Proteins were diluted appropriately into 6 ml phosphate buffer and 0.6 ml peanut oil added to give final protein concentrations of 300, 150, 37.5 and 18.75 µg/ml, and emulsions prepared as above. Sedimentation and oil droplet coalescence (creaming) were assessed by visual inspection at regular intervals up to 2 weeks, measuring the height of the separated, coalesced oil phase relative to the height of the remaining emulsion. The ESI was calculated as the percentage ratio of the emulsion height to the sum of the emulsion height plus the height of separated oil phase. The ESI value, therefore, represents the strength of the emulsion surface. Measurements were repeated in quadruplicate for independently prepared samples.
Chemical denaturation and fluorescence measurement
Equilibrium unfolding experiments in guanidine.HCl (GuHCl) were performed for the proteins as previously described (Altroff et al., 2001), following the method of Pace and Scholtz (1997) for the calculation of the Gibbs free energy (
G) between the folded and unfolded states, extrapolated to 0 M GuHCl to obtain the conformational stability of the protein, G(H2O). Briefly, proteins samples were diluted in GuHCl and fluorescence measured on a Perkin Elmer 50B spectrofluorimeter, at 25°C (
ex 290 nm, em 350 ± 3 nm). All fluorescence measurements were repeated at least in duplicate for independently prepared samples.
Equilibrium unfolding experiments in SDS were performed for the proteins incubated in 0 to 3 mM SDS in phosphate buffer. The accepted critical micelle concentration (CMC) of SDS in 50 mM NaCl is 3.8 mM. The phosphate buffer used here contained 50 mM NaCl and 10 mM phosphate; using microcalorimetry, measurement of the CMC was made by repeated 10 µl injections of 50 mM SDS into 1.4 ml phosphate buffer (MicroCalTM VP-ITC, CA, USA) and gave a value of 2.5 mM. Unfolding of the recombinant proteins in increasing concentrations of SDS was followed for step-wise additions of 15 µl of 50 mM SDS into 0.2 mg/ml protein solution in phosphate buffer. Samples were allowed to equilibrate and run as described for unfolding in GuHCl.
Unfolding of the recombinant proteins in SDS over time was monitored for addition of 0.6 mM SDS (final concentration) to 0.25 mg/ml protein solution in phosphate buffer, immediately monitoring the change in fluorescence until no further change was observed (around 60 min; ex, 290 nm; em 350 nm). The surface hydrophobicity of natively folded protein and SDS-unfolded protein, following incubation in 0.6 mM SDS as above, was measured by step-wise addition of 8-anilino-1-naphthalenesulfonic acid ammonium salt (ANS) (Sigma, Dorset, UK). To 2 ml of 0.1mg/ml (5 µM) protein solution, 15 x 5 µl additions, then 1 x 25 µl addition of 5 mM ANS, were made (0–240 µM), monitoring the change in fluorescence using ex 350 nm and em 490 nm. Analysis of data was made as previously described with modification (Moro et al., 2001). Briefly, fluorescence readings were adjusted such that 240 µM ANS gave a reading of 1. The maximum fluorescence value for the protein sample (Fmax), found at saturating ANS concentrations, is a function of the number of ANS binding sites per protein molecule. The ANS concentration at half Fmax is equal to the apparent dissociation constant of the protein–ANS complex (Kd), with the Fmax/Kd ratio being the protein surface hydrophobicity index (PSH), the higher the index the more hydrophobic the protein surface.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Domain surface hydrophobicity is not perturbed by IDLEVRV/IEI mutations
In creating the 9-10FNIII-IEI and 9-10FNIII-IDLEVRV mutants the aim was to increase the hydrophobic moment of 9FNIII and 10FNIII, respectively, rather than patching a hydrophobic surface to 9-10FNIII. Of the five ß-strands in 9-10FNIII falling outside the surface seeking classification (Eisenberg et al., 1982) (Table I), only the IDLEVRV and IEI mutations proved compatible with the expression system. The calculated expression levels of the purified recombinant proteins were as follows (units, mg/l culture): 9-10FNIII, 10; 9-10FNIII-IDLEVRV, 3; 9-10FNIII-IEI, 19; 9-10FNIII-CC, 12; 9-10FNIII-P, 20. The purification of all mutants and 9-10FNIII proceeded to homogeneity and ran as a single peak on size-exclusion chromatography with elution times matching the internal 9-10FNIII standard (Figure 2).
|
Change to the surface hydrophobicity of natively folded proteins was measured experimentally by ANS titration and theoretically by molecular modelling. For 9-10FNIII and all mutants tested, titration of ANS into protein solution was indistinguishable from titration of ANS into phosphate buffer. Since the proteins in aqueous solution did not interact with ANS this suggested that the mutations made negligible change to surface hydrophobicity; 9-10FNIII-IDLEVRV and 9-10FNIII-IEI mutations, therefore, selectively conferred increased hydrophobic moment as intended. These fluorescence data were supported by calculation of the solvation energies of the proteins. Comparison across the data generated by all the modelling protocols used showed that the solvation energies for the proteins did not differ substantially from one another (Table III). A slight increase to the conditional hydrophobic accessible surface area for 9-10FNIII-IEI and 9-10FNIII-CC was observed. This is probably due to the two isoleucine substitutions in 9-10FNIII-IEI and formation of the cystine upon disulphide oxidation in 9-10FNIII-CC. (The conditional hydrophobic accessible surface area takes into account solvation of neighbouring polar atoms and is more sensitive to hydrophobicity than conditional accessible surface area calculations (Cooper et al., 2005
). Emulsion stability is dominated by conformational stability
The mutants 9-10FNIII-CC and 9-10FNIII-P have previously been demonstrated to remain natively folded in solution, the disulphide bridge altering interdomain tilt angle (Altroff et al., 2004). The two-step unfolding curves for 9-10FNIII-IDLEVRV and 9-10FNIII-IEI in GuHCl are in good agreement with these previous studies (Figure 3) (van der Walle et al., 2002); a large difference in 9FNIII and 10FNIII thermodynamic stability being observed (Table IV). It is reasonable then to believe that the mutants, like 9-10FNIII, maintained their modular nature.
|
|
Unfolding of the wild-type and mutant domain pairs in SDS occurred over one transition (Figure 4) and was, therefore, modelled as a two-state process (Table IV). Unfolding in SDS was clearly more dependent on concentration compared with unfolding in GuHCl: values for m being an order of three greater (the denaturant concentration at 50% protein unfolding, [SDS]1/2, being in the mM range). The free energy of 9-10FNIII-CC unfolding was greater than for the other mutants and 9-10FNIII. Comparing the equilibrium parameters for the proteins in SDS to GuHCl showed a somewhat altered trend for conformational stability of the series. The
G(H2O) values for 9-10FNIII and 9-10FNIII-P are remarkably similar but m for the same pair are different (4140 and 2990 kcal/mol/M, respectively).
|
The emulsifying stability index (ESI) for the proteins at concentrations of 18.75 µg/ml clearly showed that emulsions stabilized with 9-10FNIII-P or 9-10FNIII-CC were most stable (Figure 5). The trend for decreasing emulsion stability placed the more conformationally stable mutants ahead of those conferring increased hydrophobic moment and 9-10FNIII (wild-type). By linear regression analysis, equilibrium parameters for protein unfolding in GuHCl were poor indicators of emulsion stability. However, regression analysis of the proteins' ESI values versus respective [SDS]1/2 values gave an R2 value of 0.900 (Table V). Thus, for ß-sandwich proteins, prior knowledge of protein unfolding in SDS would appear to be a reasonable predictor of emulsion stability.
|
|
Observed differences in EE for the protein series
Figure 6 shows the change to the median emulsion oil droplet size over time. A monomodal size-distribution for each data point shown was observed (data not shown). The emulsions, therefore, did not consist of two individual populations of droplet size as is observed for heterogeneous preparations of protein or peptide emulsifiers (van der Ven et al., 2001). This is likely to reflect the relatively high purity that can be achieved with recombinant proteins.
|
For 9-10FIII-stabilized emulsions droplet size was seen to transiently decrease. Although this decrease may have been due to removal of larger drops from the system by surface adsorption, this was unlikely given the absence of a fall for BSA-stabilised emulsions and the monomodal droplet size-distribution. The decrease in droplet size was more prolonged for 9-10FNIII-CC and 9-10FNIII-P stabilized emulsions (15–20 min) before levelling off. In contrast, for 9-10FNIII and 9-10FNIII-IEI stabilized emulsions, the transient fall was followed by a prolonged increase in droplet size, presumably due to coalescence, with 9-10FNIII-IDLEVRV lying in-between. The distinct changes in droplet size over time seen between the 9-10FNIII proteins indicated clear differences in their EE.
Table V shows the overall percentage change in droplet size (EE, %); the larger the decrease in size the stronger the EE. 9-10FNIII-CC and particularly 9-10FNIII-P showed good EE, 9-10FNIII-IDLEVRV was comparable to BSA [a known emulsifier, (Rampon et al., 2004)], with 9-10FNIII and 9-10FNIII-IEI showing poor EE. Therefore, the ß-sandwich structure does not prohibit robust EE despite this conformation being poorly represented in typical protein emulsifiers. Interestingly, reduction of the disulphide bridge of 9-10FNIII-CC improved EE such that droplet size closely matched the 9-10FNIII-P stabilized emulsion. The coalescence observed after 60 min is likely to have been due to re-oxidation, which occurs rapidly for 9-10FNIII-CC. This clearly demonstrates the dependence on interdomain flexibility for EE.
Determination of the EAI provided a complementary measure of the proteins' EE. Table V shows that EAI was greatest (EE was higher) for 9-10FNIII-P and 9-10FNIII-CC and lowest for wild type 9-10FNIII, with 9-10FNIII-IDLEVRV and 9-10FNIII-IEI lying in-between; the trend matching the EE data. EAI was greater for 9-10FNIII-P and 9-10FNIII-CC than for BSA, providing further evidence of their good emulsifying efficiency. However, linear regression analysis of EE (%) and EAI against equilibrium parameters did not suggest strong dependence between conformational stability and EE (data not shown). In order to enable further interpretation of the EE data we investigated the rate of protein unfolding and subsequent surface hydrophobicity.
Emulsification efficiency is dependent on protein surface hydrophobicity (PSH)
The rate of unfolding of 9-10FNIII in phosphate buffer containing 0.6 mM SDS (2.06 x 10–3 s–1) was around 300 times slower than unfolding of 10FNIII in 2.5 M guanidine isothiocyanate (0.742 s–1) (Cota and Clarke, 2000). Broadly similar rates of unfolding were observed for the mutant 9-10FNIII proteins (Figure 7 and Table V). The slow rates of protein unfolding in SDS versus guanidinium is typical (Ternstrom et al., 2005), there being orders of magnitude difference in denaturant concentrations. The rate constants of protein unfolding did not correlate with respective ESI values or EE data. However, with respect to the extent of unfolding (seen as the relative increase in fluorescence intensity over 1 h incubation in SDS, Figure 7), a general trend was noted to correspond to EE—the smaller the extent of unfolding the greater the EE. The extent of unfolding in SDS was, therefore, studied in more detail using the aromatic hydrophobic probe, ANS.
|
Following equilibration in SDS, ANS bound to the unfolded proteins to varying degrees, implying that the proteins unfolded exposing buried hydrophobic residues (Figure 8). After reaching saturated levels of bound ANS, it was noted that fluorescence subsequently tended to fall. This has been previously noted and is suggested to represent ANS-induced oligomerization of the protein species, reducing the number of available binding sites (Peri et al., 1990
). Given that PSH reflects the cumulative index of unfolding (Moro et al., 2001), this was most extensive for 9-10FNIII and least for 9-10FNIII-P and 9-10FNIII-CC. Regression analysis of the PSH values versus EE (%) and EAI gave R2 values of 0.932 and 0.809, respectively (Table V and Figure 9). Therefore, the interpretation of the EE data is consistent with the EAI data. Thus, the EE of the proteins was dependent on limited, rather than extensive, unfolding and exposure of hydrophobic surface.
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The value in increasing the hydrophobic moment of ß-strands was found to be limited with respect to improving emulsion stability (ESI). Our results showed that 9-10FNIII-IEI stabilized the emulsion somewhat better than 9-10FNIII (
2-fold), but this was not the case for 9-10FNIII-IDLEVRV. It is possible that the IEI mutation on the 9FNIII C' strand contributed to stabilization of the protein interfacial film, whereas the IDLEVRV mutation on the 10FNIII A strand did not (see Leahy et al., 1996, for strand notations). Although in situ structural analysis would be required to prove this, the data suggest that 10FNIII unfolding at the interface was not complete. This would be in agreement with the much greater conformational stability of 10FNIII over 9FNIII; the IDLEVRV and IEI mutations did not appreciably alter the conformational stability of 10FNIII and similar G(H2O) values for 9FNIII were found with respect to the wild-type. Although the dependence of unfolding on denaturant concentration (m) for 9-10FNIII-IDLEVRV was lower compared with the wild-type, m may vary noticeably between single amino acid mutants (Shirley et al., 1989).
In contrast, a large increase in the ESI was found for 9-10FNIII-P (P1408 mutation increasing the conformational stability but not the hydrophobic moment), with a similar but smaller increase for 9-10FNIII-CC. Although 9-10FNIII-CC includes P1408 it also includes the disulphide, which strongly restricts interdomain mobility (Altroff et al., 2004). This suggests that interdomain flexibility is required for maximal emulsion stability. It is interesting to note that FN was shown to displace BSA from the oil/water interface and the flexibility of its modular nature may play a role in this observation (Vaidya and Ofoli, 2005). The dependence of the proteins' ESI on its respective [SDS]1/2 value implies that a certain degree of conformational stability in ß-sandwich proteins tends to favour emulsion stability. This may initially appear contrary to protein unfolding upon interfacial adsorption but does not necessarily involve a complete loss of secondary structure. For example, ß-casein is predominantly random coil in solution but adopts -helical structure on adsorption (Caessens et al., 1999), adsorption of 2S albumins involves loss of -helical and disordered structure in favour of ß-sheet (Burnett et al., 2002) and AlnA requires maintenance of its ß-barrel structure in order to stabilize oil-in-water emulsions (Toren et al., 2002). In contrast, Poon et al. (2001a) demonstrated that chemical denaturation of globular and milk proteins in most cases increased EE. However, the denaturation used was mild, disrupting quaternary structure, and the authors concluded that the effect was protein-specific and depended on the relative importance of disulfide bonds, which is in agreement with the observed differences in ESI for 9-10FNIII-P and 9-10FNIII-CC.
Emulsification efficiency was quantified by the change in droplet size during shear. The initial transient fall in droplet size seen in the 9-10FNIII-stabilized emulsions possibly represented break-up of bridged flocs following the initial dilution of the emulsion on addition to the dispersion unit. The subsequent changes in droplet size, followed by a leveling-off (except for wild type 9-10FNIII) would have been dependent on the shear of the mixer in the dispersion unit limiting droplet coalescence till equilibration, determined by the relative emulsification efficiencies of the proteins. Comparing the trends in droplet size for 9-10FNIII-IDLEVRV and 9-10FNIII-IEI versus 9-10FNIII suggested that the hydrophobic moment of the ß-strands in either domain somewhat improved EE, particularly the IDLEVRV mutation. However, the relative improvement in EE was greatest for 9-10FNIII-P and, to a lesser extent, 9-10FNIII-CC. This trend initially appeared to reflect the ESI data, though linear regression did not show dependence on [SDS]1/2 values, suggesting a somewhat different mechanism underpins the ESI and EE data. This may not be surprising since emulsion stability relies on a measure of the strength of the emulsion surface, which may not necessarily have been directly related to the interfacial activity of the proteins. In contrast the measurement of droplet size with shear may be more sensitive to protein adsorption or desorption from the oil/water interface, stabilizing or destabilizing the viscoelastic film surrounding the oil droplets.
Emulsification efficiency was clearly dependent on PSH (Table V and Figure 9). This is not to say that complete attenuation of protein unfolding would yield a strong emulsifying (ß-sandwich) protein. It is clear from earlier work that surface hydrophobicity is correlated to EE and interfacial tension (Kato and Nakai, 1980). Similarly, where notable hydrophobic stretches in primary structure are evident these may be associated with EE (Toren et al., 2002). However, recent reports demonstrate that the relationship is not straightforward, as also seen in our data. Losso and Nakai, (2002) described protein bridging as destabilizing with respect to the dispersion of oil droplets, conjugation of (hydrophilic) poly ethyleneglycol to the protein reduced bridging-destabilization, implicating protein association via exposed hydrophobic domains. Creamer et al. (1998) showed that stabilization of the 
-casein interfacial film involved oligomerization, which may have been as much dependent on disulphide interchange as hydrophobic interaction. Moreover, hydrophobic domains drive the oligomerization of ß-casein (Caessens et al., 1999) but in doing so this limits EE (Poon et al., 2001a). Thus, it becomes apparent that there exists an optimal surface hydrophobicity for each protein emulsifier, which may exist either for the native or unfolded conformer. In the case of the ß-sandwich domain pair studied here, the optimal hydrophobicity for EE is best represented by SDS-unfolded 9-10FNIII-P.
The dependence of the EE data on PSH was corroborated by the same trend being observed for the EAI data, which provided a complementary measure of the EE of the protein (Pearce and Kinsella, 1978). In both EE and EAI data sets the EE of 9-10FNIII-CC was slightly less than 9-10FNIII-P, suggesting that restricted domain–domain mobility reduced EE (as with emulsion stability). Reduction of the disulphide bond elegantly confirmed this by increasing the EE to that of 9-10FNIII-P; upon reoxidation the EE returned to that seen for 9-10FNIII-CC (Figure 6). In aiming to achieve maximum emulsion stability, it is noteworthy to relate the emulsion performance to the interdomain flexibility between the 9th and 10th FNIII domains, which is relatively high on the basis of an unusually low buried surface area of 333 Å2 (Leahy et al., 1996), and the tilt and twist of the 9-10FNIII-P and 9-10FNIII-CC domain pairs, which is 28° and 349° and 5° and 355°, respectively (Altroff et al., 2004). The conformational changes to FN, a modular ß-sandwich protein, observed during interfacial adsorption to hydrophilic or hydrophobic surfaces (Renner et al., 2005) also points to the importance of interdomain flexibility in emulsion stability. It is currently unclear how the kinetics of ß-sandwich domain unfolding during emulsification relate to interfacial adsorption or desorption. This will form the basis of further work, focusing on conformational changes to ß-sandwich proteins upon adsorption to an interface.
It should be noted that the emulsifying performance of the protein series has been related to fundamental properties of conformational stability and unfolding but not directly to an understanding of protein surface area and interfacial tension. Similarly, it is not yet possible to assign a ‘hydrophile–lipophile balance’ (HLB) value, as is the case for typical sufactants (e.g. Tweens/Spans), on the basis of a protein's physico-chemical properties; this has been previously noted by (Poon et al., 2001a). Ideally, the oil/water interfacial tensions for the proteins would have supported the emulsion data. However, application of the Langmuir–Blodgett trough proved problematic because of dilution of the stock protein into the bulk such that the formation of a film was non-evident. To address this issue, we are currently investigating the behaviour of the proteins at surfaces via neutron scattering and Dual Polarisation Interferometry. Similarly, measurements of force distance profiles for ß-sandwich proteins acquired by Atomic Force Microscopy (Ng et al., 2005) may provide a mechanistic link between the protein's fundamental properties and its emulsifying performance.
|
Footnotes |
|---|
Edited by Hagan Bayley
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
This work was supported by the Biotechnology and Biological Sciences Research Council, grant no. 86/E19380 to C.V.D.W. We thank M. Nutley and A. Cooper (BBSRC Microcalorimetry facility, Glasgow, UK), and E. Schmidt for support in protein analysis.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Altroff H., van der Walle C.F., Asselin J., Fairless R., Campbell I.D., Mardon H.J. (2001) J. Biol. Chem. 276:38885–38892.
Altroff H., Schlinkert R., van der Walle C.F., Bernini A., Campbell I.D., Werner J.M., Mardon H.J. (2004) J. Biol. Chem. 279:55995–56003.
Bates P.A., Kelley L.A., MacCallum R.M., Sternberg M.J.E. (2001) Proteins 39–46.
Bouissou C., Potter U., Altroff H., Mardon H., van der Walle C. (2004) J. Control. Release 95:557–566.[CrossRef][Web of Science][Medline]
Burnett G.R., Rigby N.M., Mills E.N.C., Belton P.S., Fido R.J., Tatham A.S., Shewry P.R. (2002) J. Colloid Interf. Sci. 247:177–185.[CrossRef][Web of Science][Medline]
Caessens P.W., De J, ongh H.H., Norde W., Gruppen H. (1999) Biochim. Biophys. Acta 1430:73–83.[CrossRef][Medline]
Carino G.P., Jacob J.S., Mathiowitz E. (2000) J. Control. Release 65:261–269.[CrossRef][Web of Science][Medline]
Cooper A., Kennedy M.W., Fleming R.I., Wilson E.H., Videler H., Wokosin D.L., Su T.J., Green R.J., Lu J.R. (2005) Biophys. J. 88:2114–2125.
Cota E. and Clarke J. (2000) Protein Sci. 9:112–120.[Web of Science][Medline]
Creamer L.K., Plowman J.E., Liddell M.J., Smith M.H., Hill J.P. (1998) J. Dairy Sci. 81:3004–3012.[Abstract]
Damodaran S., Anand K., Razumovsky L. (1998) J. Agric. Food Chem. 46:872–876.[CrossRef]
Dickinson E. and Matsumura Y. (1991) Int. J. Biol. Macromol. 13:26–30.[CrossRef][Web of Science][Medline]
Egilmez N.K., Jong Y.S., Sabel M.S., Jacob J.S., Mathiowitz E., Bankert R.B. (2000) Cancer Res. 60:3832–3837.
Eisenberg D., Weiss R.M., Terwilliger T.C. (1982) Nature 299:371–374.[CrossRef][Medline]
Janin J. (1979) Nature 277:491–492.[CrossRef][Medline]
Johnson K.J., Sage H., Briscoe G., Erickson H.P. (1999) J. Biol. Chem. 274:15473–15479.
Kato A. and Nakai S. (1980) Biochim. Biophys. Acta 624:13–20.[Medline]
Laskowski R.A., Rullmannn J.A., MacArthur M.W., Kaptein R., Thornton J.M. (1996) J. Biomol. NMR 8:477–486.[Web of Science][Medline]
Leahy D.J., Aukhil I., Erickson H.P. (1996) Cell 84:155–164.[CrossRef][Web of Science][Medline]
Lindahl E., Hess B., van der Spoel D. (2001) J. Mol. Model. 7:306–317.
Losso J.N. and Nakai S. (2002) J. Agric. Food Chem. 50:1207–1212.[CrossRef][Web of Science][Medline]
Mordenti J., et al. (1999) Toxicol. Sci. 52:101–106.
Moro A., Gatti C., Delorenzi N. (2001) J. Agric. Food Chem. 49:4784–4789.[CrossRef][Web of Science][Medline]
Ng S.P., Rounsevell R.W., Steward A., Geierhaas C.D., Williams P.M., Paci E., Clarke J. (2005) J. Mol. Biol. 350:776–789.[CrossRef][Web of Science][Medline]
Ooi T., Oobatake M., Nemethy G., Scheraga H.A. (1987) Proc. Natl Acad. Sci. USA 84:3086–3090.
Pace C.N. and Scholtz J.M. (1997) Protein Structure, A Practical Approach. (IRL PressIn Creighton T.E. (Ed.). , Oxford)299–321.
Pearce K.N. and Kinsella J.E. (1978) J. Agric. Food Chem. 26:716–723.[CrossRef]
Peri C., Pagliarini E., Iametti S., Bonomi F. (1990) J. Dairy Res. 57:101–108.[Medline]
Plaxco K.W., Spitzfaden C., Campbell I.D., Dobson C.M. (1997) J. Mol. Biol. 270:763–770.[CrossRef][Web of Science][Medline]
Pontius J., Richelle J., Wodak S.J. (1996) J. Mol. Biol. 264:121–136.[CrossRef][Web of Science][Medline]
Poon S., Clarke A.E., Schultz C.J. (1999) J. Colloid Interf. Sci. 213:193–203.[CrossRef][Web of Science][Medline]
Poon S., Clarke A.E., Schultz C.J. (2001a) J. Agric. Food Chem. 49:281–286.[CrossRef][Web of Science][Medline]
Poon S., Clarke A., Currie G., Schultz C. (2001b) Biosci. Biotechnol. Biochem. 65:1713–1723.[CrossRef][Medline]
Qin B.Y., Bewley M.C., Creamer L.K., Baker H.M., Baker E.N., Jameson G.B. (1998) Biochemistry 37:14014–14023.[CrossRef][Medline]
Rampon V., Brossard C., Mouhous-Riou N., Bousseau B., Llamas G., Genot C. (2004) Adv. Colloid Interf. Sci. pp. 87–94.
Renner L., Pompe T., Salchert K., Werner C. (2005) Langmuir 21:4571–4577.[CrossRef][Web of Science][Medline]
Robertson A., MacColl G.S., Nash J.A., Boehm M.K., Perkins S.J., Bouloux P.M. (2001) Biochem. J. 357:647–659.[CrossRef][Web of Science][Medline]
Sah H. (1999) J. Pharm. Sci. 88:1320–1325.[CrossRef][Web of Science][Medline]
Saito M., Ogasawara M., Chikuni K., Shimizu M. (1995) Biosci. Biotechnol. Biochem. 59:388–392.[Medline]
Shirley B.A., Stanssens P., Steyaert J., Pace C.N. (1989) J. Biol. Chem. 264:11621–11625.
Spitzfaden C., Grant R.P., Mardon H.J., Campbell I.D. (1997) J. Mol. Biol. 265:565–579.[CrossRef][Web of Science][Medline]
Ternstrom T., Svendsen A., Akke M., Adlercreutz P. (2005) Biochim. Biophys. Acta 1748:74–83.[Medline]
Toren A., Segal G., Ron E.Z., Rosenberg E. (2002) Environ. Microbiol. 4:257–261.[CrossRef][Medline]
Vaidya S.S. and Ofoli R.Y. (2005) Langmuir 21:5852–5858.[CrossRef][Web of Science][Medline]
van de Weert M., Hoechstetter J., Hennink W.E., Crommelin D.J. (2000) J. Control. Release 68:351–359.[CrossRef][Web of Science][Medline]
van der Ven C., Gruppen H., de Bont D.B., Voragen A.G. (2001) J. Agric. Food Chem. 49:5005–5012.[CrossRef][Web of Science][Medline]
van der Walle C.F., Altroff H., Mardon H.J. (2002) Protein Eng. 15:1021–1024.
Vila J., Williams R.L., Vasquez M., Scheraga H.A. (1991) Proteins 10:199–218.[CrossRef][Web of Science][Medline]
Wesson L. and Eisenberg D. (1992) Protein Sci. 1:227–235.[Web of Science][Medline]
Willumsen P.A. and Karlson U. (1997) Biodegradation 7:415–423.[CrossRef]
Received March 13, 2006; revised June 28, 2006; accepted September 9, 2006.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||