β-Lactoglobulin nanofibrils: The long and the short of it

Accepted Manuscript β-Lactoglobulin nanofibrils: the long and the short of it Simon M. Loveday, Skelte G. Anema, Harjinder Singh PII:

S0958-6946(16)30310-7

DOI:

10.1016/j.idairyj.2016.09.011

Reference:

INDA 4088

To appear in:

International Dairy Journal

Received Date: 28 July 2016 Revised Date:

14 September 2016

Accepted Date: 14 September 2016

Please cite this article as: Loveday, S.M., Anema, S.G., Singh, H., β-Lactoglobulin nanofibrils: the long and the short of it, International Dairy Journal (2016), doi: 10.1016/j.idairyj.2016.09.011. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1

1

ACCEPTED MANUSCRIPT β-Lactoglobulin nanofibrils: the long and the short of it

2 3 4

6

Simon M. Lovedaya,b*, Skelte G. Anemac, Harjinder Singha,b

7 8

SC

9

RI PT

5

10

M AN U

11 12

a

Riddet Institute, Massey University, Private Bag 11 222, Palmerston North, New Zealand.

13

b

Massey Institute of Food Science and Technology, Private Bag 11 222, Palmerston North,

14

New Zealand.

15

c

18 19 20

TE D

17

EP

16

Fonterra Research and Development Centre, Palmerston North, New Zealand.

*Corresponding author. Tel.: +64 6 951 7259

22

E-mail address: [email protected] (S. M. Loveday)

AC C

21

2

ACCEPTED MANUSCRIPT

23

____________________________________________________________________________

24

ABSTRACT

25 The bovine whey protein β-lactoglobulin will self-assemble into amyloid-like nanofibrils when

27

heated at pH ≤ 3 and temperatures ≥ 75 °C. These fibrils have diameters < 10 nm and length

28

that can exceed 10 µm. The length and stiffness of fibrils depends on ionic strength; heating at

29

low ionic strength produces long semi-flexible fibrils, whereas heating with added salts produces

30

highly flexible ‘worm-like’ fibrils that are an order of magnitude shorter. Over the last two

31

decades there has been a substantial research effort focused on imaging fibrils, manipulating

32

conditions to accelerate fibril formation, and characterising microstructural and rheological

33

properties of fibril dispersions. Here we review the major mechanistic findings, explore potential

34

applications for fibrils as food ingredients, and highlight the major gaps in knowledge

35

surrounding these intriguing self-assembled structures.

36

____________________________________________________________________________

AC C

EP

TE D

M AN U

SC

RI PT

26

37

3

ACCEPTED MANUSCRIPT

Table of contents

38 1.

Introduction

40

2.

Nanofibril assembly processes

41

3.

Process conditions modulate assembly kinetics and fibril morphology 3.1. pH, ionic strength and ion-specific effects

43

3.2. Heating temperature

44

3.3. Shear effects

45

4.

Fibrils as food ingredients 4.1. Shifting pH and ionic strength

47

4.2. Drying and rehydration

48

4.3. Crosslinking fibrils

49

4.4. Gelling

50

4.5. Interfacial and colloidal properties of fibrils

51

4.6. Safety

53

References

AC C

54

Conclusions

TE D

5.

EP

52

M AN U

46

SC

42

RI PT

39

55

1.

4

ACCEPTED MANUSCRIPT

Introduction

56 57

The term amyloid arises from abnormal tissue deposits in the brain that stain positively with iodine; these deposits were discovered in 1854 and they were found to consist of

59

nanofibrils in 1959 (Sipe & Cohen, 2000). The characteristic stacked β-sheet internal structure

60

of the amyloid nanofibrils is now understood to be a generic folding pattern that almost all

61

proteins are capable of assuming (Goldschmidt, Teng, Riek, & Eisenberg, 2010; Otzen, 2010).

62

The term ‘amyloid’ usually refers protein structures assembled in vivo, and includes a wide

63

variety of non-pathological proteins such as the curli proteins involved in bacterial biofilm

64

formation, chorion proteins reinforcing the egg shells of silk moths, fish and other organisms,

65

and Pmel17 protein involved in human melanosome biogenesis (Berson et al., 2003; Dueholm,

66

Albertsen, Otzen, & Nielsen, 2012; Iconomidou & Hamodrakas, 2008).

SC

M AN U

67

RI PT

58

‘Amyloid-like’ is often used in connection with fibrils assembled in vitro that share the same ‘stacked β-sheet’ structure, and the amount of fibrils can be quantified using fluorescent

69

dyes that bind to this structure, particularly thioflavin T (ThT) and Congo red (Nilsson, 2004).

70

The prefix ‘nano’ refers to the fact that individual fibrils have diameters well below 20 nm.

71

Subsequent reference to fibrils in this paper refers to amyloid-like nanofibrils.

72

TE D

68

For the milk proteins, amyloid-like fibrils have been shown to form from α-lactalbumin (Goers, Permyakov, Permyakov, Uversky, & Fink, 2002; Ipsen & Otte, 2007), αS2-casein and κ-

74

casein (Thorn, Ecroyd, Carver, & Holt, 2015), and bovine serum albumin (Usov, Adamcik, &

75

Mezzenga, 2013; Veerman, Sagis, Heck, & van der Linden, 2003b). However, by far the most

76

common source material for milk protein-derived fibrils is β-lactoglobulin (β-lg), which readily

77

assembles into fibrils when heated at low pH and low ionic strength. The β-lg can be isolated

78

using a simple salt precipitation method (Dave et al., 2013) from whey protein isolate (WPI) - a

79

commercial milk protein fraction that is derived by ultrafiltration of whey, and typically contains >

80

90% protein, most of which is β-lg (Foegeding, Luck, & Vardhanabhuti, 2011).

81

AC C

EP

73

Fibrils that form in mixed whey protein solutions appear to comprise only β-lg (Bolder,

82

Hendrickx, Sagis, & van der Linden, 2006), and β-lg appears to be the only dairy protein

83

capable of rapidly forming very long fibrils during heating under suitable conditions. These heat-

5

ACCEPTED MANUSCRIPT

induced β-lg fibrils will be the focus of this review. Amyloid-like aggregation is compared with

85

other protein aggregation patterns in a recent review by Mezzenga and Fischer (2013), and

86

polymer physics aspects of fibril aggregation were discussed by Adamcik and Mezzenga

87

(2012). Fig. 1 shows a typical example of amyloid-like fibrils formed by heating bovine β-lg at

88

low pH and low ionic strength (top). Individual fibrils have a diameter of 5–10 nm and a length

89

that can exceed 10 µm, i.e., an aspect ratio in the order of 1000 or more.

RI PT

84

The β-lg amyloid-like fibrils that assemble at low ionic strength and low pH are

90

considered semi-flexible according to biopolymer physics definitions. Fibril flexibility is quantified

92

as persistence length, lp, which typically varies from 1.5 to 4 µm when measured with TEM or

93

AFM images (Loveday et al., 2010; Schleeger et al., 2013). It is the extreme length and

94

flexibility of β-lg fibrils that allows fibril dispersions to form self-supporting physical entanglement

95

networks that gel at much lower protein concentration than other types of β-lg aggregates

96

(Kavanagh, Clark, & Ross-Murphy, 2000b; Loveday, Rao, Creamer, & Singh, 2009), though

97

with a narrower linear viscoelastic shear rate range (Wu et al., 2016).

M AN U

SC

91

In addition to forming water-holding gel networks, β-lg fibrils can also adsorb at

99

interfaces, form fibril-fibril and fibril-polymer superstructures, and they can be surface-

TE D

98

functionalised via simple chemical crosslinking reactions. The morphology and functionality of β-

101

lg fibrils can be tuned by manipulating heating conditions. Here we will explore the unique

102

properties of β-lg fibrils and the potential they have as high-value components of food

103

ingredients.

105 106 107

2.

AC C

104

EP

100

Nanofibril assembly processes

For β-lg and other proteins with a significant degree of tertiary structure, partial

108

denaturation must occur before fibril assembly is possible. Unfolding of the tertiary structure

109

allows β-sheets to come into contact and form hydrogen-bonded stacks. Denaturation with

110

concentrated alcohols or urea will induce β-lg to assemble into fibrils over several days (Gosal,

111

Clark, & Ross-Murphy, 2004a; Hamada & Dobson, 2002), whereas denaturation and assembly

112

occur within hours at pH <3 and temperatures >75 °C. When β-lg is heated to 80 °C at pH 2 and

6

ACCEPTED MANUSCRIPT

113

at low ionic strength, 80–90% of the tertiary structure is disrupted within 10 min (Dave et al.,

114

2013).

115

Low ionic strength and low pH ensure that electrostatic repulsion among positivelycharged β-lg monomers and peptides inhibits random aggregation. The combination of low pH

117

and high temperature also drives hydrolysis of the protein. Partial hydrolysis or ‘nicking’

118

facilitates fibril assembly (Mishra et al., 2007), but excessive hydrolysis is likely to dramatically

119

reduce fibril yield because non-core regions would be cleaved from assembly-competent

120

regions, eliminating non-core regions from the fibril fraction. Akkermans, Venema, van der Goot,

121

Boom, and van der Linden (2008a) showed that hydrolysis of β-lg by an AspN proteinase at pH

122

8 and 37 °C followed by incubation of the hydrolysate at pH 2 and room temperature led to the

123

assembly of fibrils several hundred nanometres long. The ThT fluorescence increased during

124

the pH 2 incubation, but much of the hydrolysate was present as random aggregates.

SC

M AN U

125

RI PT

116

Fibril assembly is depicted schematically in Fig. 2. Assembly begins with nucleation, a process in which proteins/peptides assemble reversibly at first, then irreversibly once a nucleus

127

reaches a critical size (Akkermans et al., 2006; Arnaudov, deVries, Ippel, & vanMierlo, 2003;

128

Aymard, Nicolai, Durand, & Clark, 1999). Fibril growth occurs by linear addition of building

129

blocks to a nucleus, and the extraneous non-core regions of building blocks form a ‘coat’ of

130

unstructured peptide chains that can be further hydrolysed (Usov & Mezzenga, 2014).

131

Exhaustion of building blocks ultimately limits the growth of fibrils. This nucleation-limited

132

assembly process leads to sigmoidal growth curves consisting of lag, growth and stationary

133

phases. The lag phase can be significantly reduced or eliminated by adding pre-formed fibrils or

134

‘seeds’ consisting of fractured fibrils (Bolder, Sagis, Venema, & van der Linden, 2007; Dunstan,

135

Hamilton-Brown, Asimakis, Ducker, & Bertolini, 2009; Loveday, Su, Rao, Anema, & Singh,

136

2012b) or peptide strands corresponding to β-sheet-rich regions of β-lg (Hamada et al., 2009).

137

AC C

EP

TE D

126

Single-stranded fibrils are sometimes termed filaments or protofibrils, and they can

138

assemble spontaneously into superstructures consisting of two or more aligned protofibrils.

139

Three or more fibrils will align laterally to form twisted ribbons or tapes rather than close-packed

140

arrangements (Adamcik et al., 2010). Ribbons are typically observed only after >5 h of heating,

7

ACCEPTED MANUSCRIPT

141

and β-lg ribbons of up to 17 protofibrils have been observed (Lara, Adamcik, Jordens, &

142

Mezzenga, 2011).

143

Kroes-Nijboer, Venema, Bouman, and van der Linden (2011) sought to understand the relation between β-lg hydrolysis and fibril assembly using a mathematical model prefaced on

145

the assumption that hydrolysis (modelled with a single exponential equation) was a mandatory

146

precursor for assembly. Fitting this model to data obtained for 1% β-lg at pH 2 that was heated

147

and stirred at 80–110 °C produced the surprising finding that the model rate constant for the

148

attachment of a building block (assumed to be a peptide) to a fibril was independent of

149

temperature. The authors believed that the rate-limiting step of fibril assembly was the diffusion

150

of building blocks to the point of attachment, rather than the attachment reaction.

SC

There are several problems with this model. Firstly, it assumes that all β-lg peptides are

M AN U

151

RI PT

144

152

equally capable of assembling into fibrils. It is clear from mass spectrometry analysis of the

153

peptides in fibrils that the β-sheet rich regions of the β-lg sequence have a greater propensity to

154

form fibrils (Akkermans et al., 2008b; Dave et al., 2013). The mass yield of fibrils from heated β-

155

lg varies from <5% to 68% (Hettiarachchi, Melton, Gerrard, & Loveday, 2012). Secondly, if diffusion were the rate-limiting step in the fibril assembly pathway then there

157

should be a direct relationship between viscosity and assembly rate. Using sorbitol and glycerol

158

as viscosity-enhancing cosolvents it was shown that the rate of fibril assembly is not related to

159

viscosity for 1% (w/w) β-lg heated at pH 2 (Dave, Loveday, Anema, Jameson, & Singh, 2014b).

160

These findings are consistent with other work using for short β-hairpin peptides (Sukenik &

161

Harries, 2012; Sukenik et al., 2011) and insulin (Murray et al., 2015).

EP

AC C

162

TE D

156

Thirdly, large peptides and unhydrolysed monomers have been observed in amyloid-like

163

nanofibrils formed from β-lg and other proteins. It was shown that the large peptides and intact

164

monomers are detectable in β-lg fibrils that assemble with microwave heating at 80 °C

165

(Hettiarachchi et al., 2012). Amyloid-like fibrils that form when hen egg-white lysozyme is

166

heated at pH 1.6 and 65 °C similarly contain intact monomers (Mishra et al., 2007). Under

167

chemically-denaturing conditions that are not conducive to peptide bond hydrolysis, β-lg will

168

nonetheless assemble into fibrils, e.g., with 3–5 M urea at 37 °C and neutral pH (Hamada et al.,

8

ACCEPTED MANUSCRIPT

169

2009; Rasmussen et al., 2007) or in concentrated alcohols at ambient temperature (Gosal et al.,

170

2004a).

171

An alternative explanation for the observations of Kroes-Nijboer et al. (2011) is that hydrolysis occurs in parallel with assembly during heating at low pH. In this hypothesis,

173

hydrolysis is not necessarily a precursor for assembly, but partial hydrolysis accelerates self-

174

assembly by reducing steric constraints, i.e., small peptides assemble more rapidly than large

175

peptides or intact denatured monomers. The inhibitory effect of steric hindrance on assembly is

176

evident from the way in which glucosylation and lactosylation extend the lag phase and slow the

177

rate of fibril growth (Dave, Loveday, Anema, Jameson, & Singh, 2014a).

SC

178

RI PT

172

Under conditions where hydrolysis is rapid and assembly is relatively slow, e.g., 1–2% protein, 80–90 °C and pH 2, hydrolysis begins prior to assembly (Kroes-Nijboer et al., 2011;

180

Oboroceanu, Wang, Brodkorb, Magner, & Auty, 2010), which is unsurprising. Factors that shift

181

the relative rates of hydrolysis and assembly to favour the latter are likely to lead to the

182

incorporation of larger peptides and monomers into fibrils, which would potentially improve the

183

yield of fibrils and their ability to be gelled and surface-functionalised. The assembly of

184

unhydrolysed β-lg is observed with microwave heating (Hettiarachchi et al., 2012), but the

185

peptide sequences of β-lg fibrils formed under other ‘fast assembly’ conditions, e.g. heating at

186

lower temperature, higher protein concentration and/or increased ionic strength, are not yet

187

available in the literature.

EP

TE D

M AN U

179

The self-assembly process can also be perturbed by the addition of polyhydroxy

189

alcohols (polyols), such as glycerol and sorbitol. These compounds are preferentially excluded

190

from β-lg and also influence the thermal denaturation temperature of β-lg. Dave et al. (2014b)

191

showed that both polyols decreased the rate of β-lg self-assembly, but had no effect on the

192

morphology of fibrils. Sorbitol inhibited self-assembly by stabilising β-lg against unfolding and

193

hydrolysis, resulting in fewer fibrillogenic species, whereas glycerol inhibited nucleation without

194

inhibiting hydrolysis. Slowing down self-assembly with polyols suggested that self-assembly is

195

not diffusion-limited under these conditions and it is possible to decouple hydrolysis from self-

196

assembly.

197

AC C

188

198

3.

ACCEPTED MANUSCRIPT Process conditions modulate assembly kinetics and fibril morphology

3.1.

pH, ionic strength and ion-specific effects

9

199 200 201 202

The solution pH has a strong influence on protein reactions because it determines the ionisation states of the terminal carboxyl and amino groups as well as the ionisable amino acids

204

– primarily Arg, His, Lys, Asp and Glu. The isoelectric point of β-lg is ~5.2 (Swaisgood, 1982),

205

and a pH below 2.5 favours dissociation of dimers into monomers (Mercadante et al., 2012). At

206

low ionic strength, long semi-flexible fibrils form in heated β-lg solutions between pH 1.6 and pH

207

3. Aggregates are much shorter and more flexible or ‘wormlike’ at higher pH and have a

208

persistence length an order of magnitude lower than semi-flexible fibrils (Kavanagh, Clark, &

209

Ross-Murphy, 2000a; Loveday et al., 2010; Mudgal, Daubert, & Foegeding, 2009). Fig. 3

210

illustrates these changes in morphology with changing pH, and associated pH effects on the

211

kinetics of fibril assembly.

M AN U

SC

RI PT

203

The ionic strength modulates the strength of electrostatic interactions between charged

213

species in solution. When β-lg is heated at low pH and low ionic strength (which usually means

214

no added electrolytes other than acid), long semi-flexible fibrils form, whereas wormlike fibrils

215

predominate at higher ionic strength. This corresponds with faster assembly at higher ionic

216

strength for a given concentration, as judged by ThT fluorescence (Loveday et al., 2010) and in

217

situ dynamic light scattering (Arnaudov & de Vries, 2006). The relatively strong electrostatic

218

repulsive forces at low ionic strength appear to favour the attachment of building blocks to

219

growing fibrils in a well-ordered systematic arrangement, whereas faster growth at higher ionic

220

strength is more haphazard and chaotic.

EP

AC C

221

TE D

212

The somewhat analogous effects on fibril morphology of raising ionic strength or raising

222

pH reflect the fact that both lead to a weakening of electrostatic repulsion. However, lowering

223

pH leads to faster fibril assembly (Fig. 3H) despite favouring higher positive charge, which

224

indicates that other factors are also involved. The peptide bonds adjacent to Asp are vulnerable

225

to acid hydrolysis (Schultz, 1967), and faster acid hydrolysis at low pH would favour faster

226

assembly because smaller peptides are less sterically constrained than larger peptides. The

10

ACCEPTED MANUSCRIPT

227

converse situation of heating at high pH and high ionic strength (e.g., pH 3, 50 mM NaCl and 40

228

mM CaCl2, heating at 100 °C) favours slow hydrolysis and rapid assembly, which produces very

229

short fibrils with irregular shapes (Loveday, Wang, Rao, Anema, & Singh, 2011b).

230

Relatively little is known about the relationship between β-lg fibril morphology and the physical and functional properties of fibril dispersions. The yield of fibrils varies with heating

232

conditions (Hettiarachchi et al., 2012; Loveday et al., 2011b) from <5% to 68% (w/w), which

233

makes it difficult to obtain dispersions of equivalent fibril mass fraction from two different heat

234

treatments.

Kavanagh et al. (2000b) heated β-lg solutions in a rheometer while making small-

SC

235

RI PT

231

amplitude oscillatory measurements of ‫ ܩ‬ᇱ , the elastic modulus, and they extrapolated the gel

237

ᇱ cure curve to estimate the long-time elastic modulus, ‫ܩ‬ஶ . Fibril gels formed at pH 3 had higher

238

ᇱ ‫ܩ‬ஶ than those formed at pH 2.5 or pH 2.0, but the difference was not large.

239

M AN U

236

A similar approach was used to determine how the type and amount of salts affect the rheological properties of fibril dispersions. When fibrils were produced by heating 5% (w/w) WPI

241

ᇱ at 80 °C, adding MgCl2 or CaCl2 produced faster gelation and higher ‫ܩ‬ஶ than adding NaCl or

242

KCl at the same molarity, and there were no apparent differences in fibril morphology (Loveday,

243

Su, Rao, Anema, & Singh, 2012a).

244

TE D

240

ᇱ Raising the CaCl2 concentration from 0 to 80 mM tripled the ‫ܩ‬ஶ of 10 % w/w WPI ᇱ (Loveday, Su, solutions heated at 80 °C, but a further increase to 120 mM CaCl2 halved ‫ܩ‬ஶ

246

Rao, Anema, & Singh, 2011a). Fibril dispersions formed by heating 2% WPI at 80 °C with

247

different levels of CaCl2 had the lowest viscosity with 40 mM CaCl2, and higher viscosity with

248

60–120 mM CaCl2 (Loveday et al., 2011b).

AC C

249

EP

245

The nonlinear effects at both 2% and 10% WPI correspond with the observation that

250

mixtures of long semi-flexible fibrils and short wormlike fibrils are formed at intermediate ionic

251

strength (Arnaudov & de Vries, 2006; Loveday et al., 2010). It was hypothesised that the

252

nonlinear effects are attributable to different fibril entanglement regimes that result from different

253

fibril growth habits and fibril nucleation densities (Loveday et al., 2011a).

254 255

11

ACCEPTED MANUSCRIPT 3.2.

Heating temperature

256 257

Proteins undergo thermal denaturation because the temperature affects the balance of forces stabilising protein structure, e.g., internal hydrogen bonds, van der Waals interactions,

259

and hydrophobic interactions. For β-lg the rate of denaturation depends on the temperature and

260

ionic environment (Loveday, 2016).

261

RI PT

258

Acid hydrolysis of peptide bonds is also temperature-dependent, i.e., the hydrolysis rate increases with increasing temperature (Lawrence & Moore, 1951). At pH 2 β-lg monomer

263

hydrolysis can be modelled with a single exponential equation, and rate constants follow the

264

Arrhenius relation quite well over the 80–110 °C temperature range (Kroes-Nijboer et al., 2011).

265

To form fibrils, the heating temperature must be high enough to denature β-lg. Further raising

266

the temperature shortens the lag phase and increases the maximal growth rate (Loveday,

267

Wang, Rao, Anema, & Singh, 2012c). Fibril yield is slightly decreased above 80 °C, but this

268

does not affect the viscosity of fibril dispersions. Fibril morphology also changes with the

269

heating temperature, and various multi-stranded structures are observed above 100 °C (Fig. 5).

TE D

M AN U

SC

262

The assembly of unfolded and/or hydrolysed β-lg into fibrils will be affected by

270

temperature via thermal effects on diffusion rates and molecular mobility; however, the diffusion

272

of building blocks to sites of attachment on growing fibrils is not the rate-limiting process for β-lg

273

assembly at pH 2, 80 °C and 1% (w/w) protein (Dave et al., 2014b). Thermal effects on fibril

274

assembly are difficult to separate from thermal effects on protein hydrolysis, because the two

275

processes occur in parallel.

AC C

EP

271

276 277

3.3.

Shear effects

278 279

Processes that create shear, such as stirring, agitation, or sonication drastically shorten

280

the lag phase of fibril formation for β-lg and other proteins. Dunstan et al. (2009) reported that

281

shearing at 50–200 s-1 and 20 °C in a laminar flow cell stimulated the assembly of β-lg fibril

12

ACCEPTED MANUSCRIPT

282

‘seeds’ and formed highly flexible fibrils with a ‘string of beads’ morphology. Stirring with a stir

283

bar creates a non-uniform shear field in which the shear rate varies with location. This type of

284

shear stimulated fibril assembly more dramatically than uniform shear at 150 s-1, and it

285

produced stiffer, more highly structured fibrils than uniform shearing (Dunstan et al., 2009).

286

Akkermans et al. (2006) showed that shearing pre-heated β-lg solutions at 20 °C and 200 s-1 in a rheometer accelerated fibril assembly (measured with flow-induced birefringence).

288

Solutions containing 0.5% (w/w) β-lg at pH 2 were held at 80 °C for up to 10 h then cooled and

289

sheared at 20 °C for up to 5 h. Short pulses of shearing produced fibrils with the similar mean

290

length but wider length distribution than with continuous shearing. In subsequent experiments

291

Akkermans et al. (2008c) reported that shearing at up to 337 s-1 during heating stimulated fibril

292

formation at 4–5% (w/w) β-lg (though not at lower concentrations), and excess shear was

293

detrimental to fibril integrity, a phenomenon also observed by Hill, Krebs, Goodall, Howlett, and

294

Dunstan (2006) at 300 s-1. Similarly, rotor-stator shearing and high pressure homogenisation

295

breaks fibrils into shorter fragments (Jung, Gunes, & Mezzenga, 2010; Serfert et al., 2014).

296

Several mechanisms have been proposed to explain the effects of shear on fibril assembly.

297

Dunstan et al. (2009) showed that shearing can partially disrupt the β-barrel structure of β-lg,

298

converting it into an α-helix-rich precursor capable of seeding fibril growth. They believed that

299

laminar flow under controlled shear conditions inhibited the transition of pre-fibril aggregates

300

into mature fibrils, whereas variable shear created by a stir bar in a cuvette gave rise to low-

301

shear regions in which fibrils could anneal and attain a mature structure.

SC

M AN U

TE D

EP

AC C

302

RI PT

287

Akkermans et al. (2006) proposed that the onset of shear flow converted denatured

303

proteins into ‘pre-aggregates’ capable of nucleating fibril growth, and inferred that orthokinetic or

304

flow-assisted aggregation was not responsible for shear-enhanced fibril formation at 20 °C

305

following heating. However they believed that orthokinetic flux, i.e., an enhanced supply of

306

building blocks to fibril tips as a result of shear, did occur when heating and shearing were

307

simultaneous (Akkermans et al., 2008c).

308 309

Bolder et al. (2007) used a stirring bar to shear 2% WPI solutions during heating at 80 °C, and reported dramatically increased conversion of β-lg into fibrils as a result of shear. They

13

ACCEPTED MANUSCRIPT

310

proposed two scenarios to explain increased conversion – either enhanced orthokinetic flux or

311

the fragmentation of growing fibrils to create new active growth sites, i.e., secondary nucleation.

312

The secondary nucleation hypothesis was tested by Arosio, Beeg, Nicoud, and Morbidelli (2012) using a population balance model of fibril length distributions, which were

314

measured by atomic force microscopy. They concluded that secondary nucleation due to fibril

315

breakage occurred in both stagnant and continuously shaken samples, and to a larger extent

316

when samples were shaken. However, they believed that secondary nucleation could not

317

entirely account for the elimination of the lag phase by shaking, and that a primary nucleation

318

effect also occurred.

SC

RI PT

313

Shearing effects on amyloid-like fibril formation by β-lg and other proteins were reviewed

319

by Bekard, Asimakis, Bertolini, and Dunstan (2011). They highlighted the denaturing effect of

321

air-water and water-solid interfaces, and noted that shear-related enhancement of aggregation

322

may be related to increased turnover of the air-water interface in sheared systems. This

323

hypothesis was invoked by Arosio et al. (2012) to explain how shaking enhanced primary

324

nucleation.

326

4.

TE D

325

M AN U

320

Fibrils as food ingredients

327

Relatively extreme process conditions are needed to induce β-lg to form fibrils, and the

EP

328

utility of fibrils as food ingredients depends on their behaviour under conditions more relevant to

330

food systems. This section examines how β-lg fibrils are affected by changing conditions, and

331

how the functionality of fibrils may be enhanced by post-assembly processing such as

332

crosslinking or surface modification. Whey protein fibrils may have applications in biophysical or

333

biotechnological fields (Bolisetty & Mezzenga, 2016; Sasso et al., 2014), but the focus here is

334

on food systems.

AC C

329

335 336 337

4.1.

Shifting pH and ionic strength

338

14

ACCEPTED MANUSCRIPT

The isoelectric pH of β-lg is ~5.2, and β-lg fibrils tend to aggregate around this pH. Akkermans et al. (2008d) published micrographs of WPI fibrils that were aggregated at pH 5

340

and 7 and loosely clustered at pH 4 or 8. Kinetic effects may be important for pHs above the pI,

341

i.e., a slow transition through the pI may allow for aggregation that is only slowly reversible as

342

the pH is further increased, whereas a rapid transition through the pI region may avoid

343

aggregation altogether. In the experiments of Akkermans et al. (2008d) the ThT fluorescence

344

approximately halved when the pH was raised to 8, although they noted that pH and clustering

345

may have affected ThT binding.

Veerman, Baptist, Sagis, and van der Linden (2003a) reported no change in fibril

SC

346

RI PT

339

morphology when the pH was adjusted from 2 to 7 or 8 while the sample was kept cold, but

348

others have found that at pH 6-8 fibrils are shorter and more irregular, and often surrounded by

349

small amorphous particles (Akkermans et al., 2008d; Jones et al., 2011; Loveday et al., 2011a;

350

Wu et al., 2016). This suggests that the integrity of fibrils may be partially compromised near

351

neutral pH.

352

M AN U

347

We compared the pH sensitivity of WPI fibrils produced with 0 or 80 mM CaCl2 (Loveday et al., 2011a). Fibril dispersions made in the presence of 80 mM CaCl2 precipitated at a wider

354

pH range than those without CaCl2, but this effect was mitigated by dialysing. This showed that

355

wormlike fibrils formed at high ionic strength have similar intrinsic stability to long semi-flexible

356

fibrils. Dialysing WPI fibrils made with 40 or 120 mM CaCl2 produced relatively small changes in

357

viscosity, but adjusting the pH of dialysed fibril dispersions from 2 to 6.7 resulted in a decrease

358

in viscosity by approximately a decade (Loveday et al., 2011a).

EP

AC C

359

TE D

353

Bolisetty, Harnau, Jung, and Mezzenga (2012) plotted a phase diagram for β-lg fibril

360

dispersions of 0.1–2% with 0–500 mM NaCl that was added after heating at low ionic strength.

361

At constant concentration they noted a transition from gel to translucent solution to phase-

362

separated solution as the ionic strength was increased. Increasing concentration produced an

363

isotropic-nematic phase transition and samples were birefringent at higher concentrations.

364

Interestingly, the G’ increased by an order of magnitude when NaCl concentration was raised

365

from 0 to 100 mM, and decreased by almost the same amount with a further increase to 400 mM

366

(Bolisetty et al., 2012). This may be related to the observation that NaCl concentration affects

15

ACCEPTED MANUSCRIPT

367

the pitch of twisted β-lg fibril ribbons and tapes through a balance between the mutual repulsion

368

of surface charges, which promotes twisting, and elastic torsional forces, which resist twisting

369

(Adamcik & Mezzenga, 2011). It may be speculated that the increase in viscosity with 100mM

370

added salt could be related to better entanglement that results from longer pitch (less twisting).

371 4.2.

Drying and rehydration

RI PT

372 373 374

There is limited literature on the changes in fibril structure or properties that result from drying and rehydration. Mudgal, Daubert, and Foegeding (2011) reported that the concentration

376

of fibrils in solution had a strong impact on the viscosity of lyophilised and rehydrated fibrils, i.e.,

377

viscosity was higher and more sensitive to protein concentration for fibril dispersions made with

378

>6.9% protein. Fibrils formed entangled ‘microgels’ at these concentrations; this was thought to

379

enhance viscosity both before and after drying. A similar observation was reported in the patent

380

literature (Veerman, 2006), in that lyophilised fibrils formed at 6% WPI produced more stable

381

foams than fibrils formed at 3% WPI, even when diluted to the same protein content. It was also

382

observed that lyophilising had a minimal effect on the foam-stabilising properties of WPI fibrils.

TE D

M AN U

SC

375

The effect of drying and rehydration Wormlike and semi-flexible fibrils β-lg fibrils were

384

formed at low or high ionic strength from 2% WPI and lyophilised. The viscosity of rehydrated

385

fibril dispersions at various concentrations was monitored (Loveday et al., 2012b). The wormlike

386

fibrils formed with 80 mM CaCl2 produced more viscous solutions prior to lyophilising, by more

387

than an order of magnitude (Fig. 6). Lyophilising and rehydrating at the same concentration

388

diminished viscosity for both semi-flexible and wormlike fibrils. Viscosity was highly

389

concentration-dependent for rehydrated wormlike fibrils, whereas semi-flexible fibrils produced

390

only slight increases in viscosity when concentration was increased from 2% (w/w) to 4% and

391

10% (w/w).

AC C

EP

383

392 393 394

4.3.

Crosslinking fibrils

16

ACCEPTED MANUSCRIPT

395

Fibrils naturally form entanglement networks that are responsible for their viscosity-

396

enhancing effects, and several groups have investigated ways to crosslink fibrils to enhance this

397

effect.

398

Veerman et al. (2003a) developed a protocol for Ca2+-induced cold gelation of β-lg fibrils at pH 7 or 8. The critical percolation concentration was lowest with 10 mM CaCl2 at both pH 7

400

(0.12 ± 0.07%, w/w) and 8 (0.44 ± 0.08%, w/w), and it increased (weaker gelling) with higher

401

CaCl2 concentrations. The authors believed that 10 mM CaCl2 produced an optimal amount of

402

crosslinking, and more crosslinking produced by higher CaCl2 concentrations gave rise to

403

bundle formation and less efficient gelation.

SC

404

RI PT

399

Wu et al. (2016) treated β-lg fibrils with a reducing agent (dithiothreitol, DTT) and the enzyme transglutaminase, which catalyses the formation of a covalent bond between a primary

406

amine such as in lysine, and the carboxyamide group of a glutamine residue. Compared with

407

ᇱ by an order of DTT alone, a combination of DTT and transglutaminase increased the ‫ܩ‬ஶ

408

magnitude and reduced the critical gelation concentration by three quarters. The difference was

409

attributed to thicker strands within the crosslinked gel.

Munialo, de Jongh, Broersen, van der Linden, and Martin (2013) introduced fibril-fibril

TE D

410

M AN U

405

crosslinks by derivitising WPI fibrils with a reagent that grafts free thiol groups onto lysines and

412

terminal amines. Thiolation of pre-formed fibrils caused clustering and tightening but otherwise

413

did not affect fibril morphology; however, thiolated WPI had limited ability to form fibrils. When

414

acidified slowly with glucono-δ-lactone (GDL), thiolated fibrils formed a gel at approximately the

415

same time as unthiolated fibrils, but thiolation increased the final G’ from 265 ± 2 Pa to 664 ± 3

416

Pa with 0.9% (w/w) protein.

AC C

417

EP

411

Fibrils can be crosslinked noncovalently using specific highly-methylesterified pectins

418

(Hettiarachchi et al., 2016a). At pH 3, a citrus pectin with a degree of methylesterification (DM)

419

of 86 ± 2% was capable of binding laterally-aligned β-lg fibrils into nanotapes with mean width of

420

180 nm. Nanotape formation was less pronounced with DM of 67.4 ± 0.1% and absent with

421

47.8% (Hettiarachchi et al., 2016b). At pH 2, 4 and 6 nanotape assembly was minimal;

422

assembly was absent or random at pH 7 or with other poly- and oligosaccharides (Hettiarachchi

423

et al., 2016b). Fig. 7 illustrates the microscopic appearance of fibril-pectin nanotapes.

17

ACCEPTED MANUSCRIPT

Nanotapes disassembled when 100 mM NaCl was added (Hettiarachchi et al., 2016a) and when

425

the pectin was hydrolysed with a pectinase mixture or demethylesterified with pectin

426

methylesterase (Hettiarachchi et al., 2016b). The ability of any given pectin to assemble β-lg

427

fibrils into nanotapes apparently depends on a high degree of methylesterification, blockwise

428

distribution of non-methylesterified residues, and sufficient length to bind to two fibrils

429

simultaneously (Hettiarachchi et al., 2016a). Binding appears to involve a combination of

430

electrostatic and hydrophobic interactions.

431 4.4.

Gelling

SC

432 433

Mention has already been made of how pH, mono- and divalent cations, and

M AN U

434

RI PT

424

temperature affect fibril gel properties in simple systems, but reports on fibril-enriched foods are

436

extremely scarce. In the patent literature there is mention that whey protein fibrils can improve

437

the mouthfeel of custard, and increase the viscosity of drinking yoghurt more effectively than

438

nonfibril whey protein (Veerman, 2006), but we have not found corresponding reports in the

439

scientific literature.

TE D

435

In our unpublished work, WPI fibrils were lyophilised, and they were used to increase the

441

protein content of skim milk in acid gel systems by 1%. When the fibrils were added to unheated

442

skim milk, and the mixture subsequently heated before being slowly acidified to form the gels,

443

some increase in acid gel firmness was observed, but this was substantially less than could be

444

achieved by a similar addition level of WPI. When the fibrils were added to a heated skim milk

445

system and this mixture acidified to form a gel, the gel firmness decreased (Table 1). Similar

446

results were observed with both long semi-flexible fibrils (made without added salts) and short

447

worm-like fibrils (made with 120 mM CaCl2 and subsequently dialysed).

448

AC C

EP

440

In processed cheese systems, fibrils replaced 1% of the casein protein. The cheese

449

samples with both long semi-flexible fibrils and short worm-like fibrils maintained a similar

450

firmness to the original cheese sample made entirely with casein. In contrast, the cheese

451

samples containing unheated WPI (dissolved and lyophilised in the same way as fibrils) were

452

substantially softer, being similar to a cheese with 2% less protein (Table 2). Thus the fibrils

18

ACCEPTED MANUSCRIPT

453

have some functionality in cheese that is at a pH of ~5.7, but this is not substantially more than

454

typical of casein. Fibril dispersions have high viscosities at low concentration when in pure solution and at

456

the pH of formation. However, when the pH is increased, the viscosity is diminished (Loveday et

457

al., 2011a). Many food systems are at pH close to the isoelectric point of β-lg (~pH 5.1) and

458

fibrils prone to aggregation at this pH (Akkermans et al., 2008d). Thus functionality of fibrils in a

459

food system is expected to be diminished simply due to the higher pH of typical food matrices.

460

In addition, fibrils will be diluted with other components when used in a complex food matrix.

461

This will further diminish their effectiveness at increasing viscosity and water holding through

462

entanglements.

SC

RI PT

455

464

4.5.

M AN U

463 Interfacial and colloidal properties of fibrils

465 466

Denatured β-lg monomers stabilise oil-water interfaces very effectively, and β-lg nanofibrils can either stabilise or flocculate emulsions. Blijdenstein, Veerman, and van der

468

Linden (2004) reported that semi-flexible fibrils induced depletion flocculation of an oil-in-water

469

emulsion at an estimated minimum concentration of 0.003% (w/w) fibrils; whereas the

470

corresponding concentration for wormlike fibrils was 0.4% (w/w). Peng, Kroes-Nijboer, Venema,

471

and van der Linden (2016) further explored fibril effects on colloidal stability, using latex

472

microbeads with different size and charge. They reported that as fibril concentration increased,

473

the effect of fibrils varied from bridging flocculation at low concentration to steric and/or

474

electrostatic stabilisation then depletion flocculation at intermediate concentration, and finally

475

depletion stabilisation at the highest concentrations.

EP

AC C

476

TE D

467

Jung et al. (2010) showed that a β-lg fibril solution created by heating at 90 °C and pH 2

477

for 5 h produced a rapid drop in air-water surface tension followed by a prolonged slow

478

adsorption phase. The drop in surface tension at air-water interfaces was notably slower if non-

479

fibril material was removed by dialysis prior to tensiometer measurements, which led the

480

authors to conclude that the rapid initial drop was due to non-fibril protein. Dialysing decreased

19

ACCEPTED MANUSCRIPT

481

the interfacial yield stress at oil-water interfaces and increased the yield strain in large-

482

amplitude rotational shear experiments.

483

Jordens et al. (2014) showed that fibrils adsorbed at an oil-water interface aligned into 2dimensional liquid crystalline domains at pH 2, and they were more randomly aligned at pH 3

485

due to lower charge. Interfacial tension was similar at both pHs, but at pH 2 the fibril-stabilised

486

interface had high shear moduli, whereas at pH 3 the dilatational moduli were higher.

RI PT

484

487

In other work Serfert et al. (2014) reported that fibril-coated fish oil microcapsules had higher oxidative stability than control microcapsules coated with native β-lg. The fish oil in fibril-

489

coated capsules oxidised slightly more slowly, despite smaller capsule size.

SC

488

It was shown by Oboroceanu, Wang, Magner, and Auty (2014) that the foam-stabilising

491

abilities of WPI fibrils improved with pH adjustment from 2 to 7. WPI fibrils gave higher overrun

492

and longer stability than WPI heated at pH 7, and they were superior to egg white in some

493

cases. High pressure homogenisation fractured fibrils, but was not detrimental to foam-

494

stabilising abilities. The foam-stabilising ability of WPI fibrils was noted in patent literature

495

(Veerman, 2006), in which WPI fibrils were said to increase the hardness of meringues.

496 497 498

4.6.

Safety

TE D

M AN U

490

Given the structural resemblance between β-lg amyloid-like fibrils and pathological

500

amyloids, the safety of whey-protein derived nanofibrils is not a trivial matter. The ability of

501

amyloid-like nanofibrils to nucleate the growth of other fibrils appears to be closely linked to

502

sequence homology; i.e., fibrils of a given protein cannot nucleate unrelated proteins, and

503

sequence variations of a given protein in different species present a barrier to cross-species

504

nucleation (Krebs, Morozova-Roche, Daniel, Robinson, & Dobson, 2004).

505

AC C

EP

499

Human breast milk lacks β-lg, but tear lipocalin and glycodelin share structural

506

similarities with β-lg (Edwards & Jameson, 2014). It is conceivable that β-lg fibrils might

507

nucleate amyloid-like fibril assembly of these proteins, if they ever encountered one another in

508

the body. However an encounter would require β-lg nanofibrils to pass through the stomach

20

ACCEPTED MANUSCRIPT

509

intact, cross the intestinal epithelium and circulate to the nasolacrimal ducts (tear lipocalin) or

510

the endometrium (glycodelin), which is highly unlikely. Bateman, Ye, and Singh (2010) showed that β-lg fibrils are rapidly destroyed by pepsin

512

under simulated gastric digestion conditions. However, they observed the appearance of short

513

fibril-like structures after prolonged hydrolysis, and proposed that small peptides produced by

514

pepsinolysis of fibrils may be able to reassemble (Bateman, Ye, & Singh, 2011). We note that

515

nonfibrillar denatured β-lg is rapidly hydrolysed by pepsin (Peram, Loveday, Ye, & Singh, 2013),

516

and β-lg is often deliberately denatured in food manufacture, so hydrolysis will be rapid in both

517

fibrillar and denatured β-lg. However peptide sequences and their assembly behaviours may be

518

different, and more research is needed in this area.

The conjugation of metal nanoparticles onto fibrils has been shown to enhance cellular

M AN U

519

SC

RI PT

511

uptake of the metals in dendritic and MCF7 breast cancer cells (Bolisetty et al., 2014), an effect

521

described as a ‘shuttle’ mechanism. Silver nanoparticles conjugated to β-lg nanofibrils were

522

cytotoxic in vitro, but conjugating gold or palladium nanoparticles onto fibrils produced no effect

523

on cell viability and no measurable immune response. Bolisetty et al. (2014) highlighted the

524

potential for amyloid-like fibrils to facilitate targeted drug delivery. We note that fibrils could

525

potentially form complexes in vivo with inorganic nanoparticles taken up from the environment,

526

thereby potentiating cytotoxic effects.

TE D

520

Lassé et al. (2015) examined the protease resistance and cell viability effects of amyloid-

EP

527

like fibrils produced from whey, soy, kidney bean or egg proteins. All food protein fibrils were

529

degraded by pepsin, pancreatin and proteinase K, though at varying rates. The viability of Hec-

530

1a cells was slightly diminished with kidney bean fibrils, but in all other cases the cell viability

531

was unchanged or even enhanced by fibrils. The authors suggested that fibrils may have been

532

a source of nutrition to the cells.

AC C

528

533 534

5.

Conclusions

535 536 537

Over the last two decades there has been a substantial research effort focused on manipulating conditions to accelerate β-lg fibril formation, imaging fibrils and characterising

21

ACCEPTED MANUSCRIPT

538

microstructural and rheological properties of fibril dispersions. The vast majority of research has

539

focused on the long semi-flexible fibrils formed at low ionic strength, perhaps because they are

540

amenable to modelling with polymer physics concepts. The shorter wormlike fibrils have

541

received little attention, despite evidence that they are more robust than semi-flexible fibrils and

542

gel very efficiently (Loveday et al., 2012b). Direct comparisons of semi-flexible and wormlike fibrils would require the preparation of

544

corresponding dispersions at equal volume fraction of fibrils under identical ionic conditions, as

545

well as the removal of non-fibril species from both dispersions. This is experimentally involved

546

because of confounding effects of ionic strength on fibril yield, but it is well within the capabilities

547

of researchers who have been active in the field.

SC

RI PT

543

The aforementioned controversy about the role of hydrolysis in fibril formation has arisen

549

partly because of a disconnect between studies using protein concentrations below 5% w/w (the

550

majority of food laboratories) and those using 8–17% (w/w) (Gosal, Clark, & Ross-Murphy,

551

2004b; Kavanagh et al., 2000b), and partly because wormlike fibrils have been neglected.

552

Knowledge of the composition of fibrils formed under ‘rapid assembly’ conditions, such as high

553

protein concentration or moderate ionic strength, would help to resolve this controversy.

TE D

M AN U

548

The conditions required for β-lg to self-assemble into fibrils are extreme by food

555

manufacturing standards. The 304 grade stainless steel that is often used in food processing

556

plants would be corroded by contact with protein solutions at pH <3 heated to temperatures

557

above 80 °C for several hours, so equipment fabricated with corrosion-resistant materials (e.g.,

558

316 stainless steel) would be required. The acidity is the major problem; it may be possible to

559

raise the pH at which β-lg will form fibrils by chemically modifying ionisable amino acid

560

sidechains (Alting, de Jongh, Visschers, & Simons, 2002; Hoare & Koshland Jr., 1966), though

561

it is doubtful whether this could be done with food-grade reagents.

562

AC C

EP

554

The yield of β-lg or WPI fibrils produced at laboratory scale with protein concentrations ≤

563

5% (w/w) has been reported to vary from <5% to 68% (w/w) (Hettiarachchi et al., 2012), which

564

indicates a relatively inefficient assembly process under these conditions. Having said that,

565

there is little research on the relationship between fibril formation conditions and fibril yield, so

566

higher yields may be achievable. To our knowledge, fibril yields at protein concentrations > 5%

22

ACCEPTED MANUSCRIPT

567

w/w have not been reported, perhaps because fibrils form gels at such conditions and it is

568

difficult to remove non-fibril material from within gelled fibril networks.

569

WPI is relatively expensive for a protein-based food ingredient, and combined with the extreme conditions required to produce β-lg fibrils, the relatively low fibril yields and

571

disappointing preliminary results with model cheese and yoghurt systems (Tables 1 and 2), this

572

suggests that specialised high-value applications such as encapsulation are the most likely

573

avenue for commercialisation of β-lg fibrils.

574

RI PT

570

Very little is known about the behaviour of β-lg fibrils in food systems. The presence salts, sugars, and biopolymers are likely to influence fibril properties and functionality, as are

576

structures such as oil-water and air-water interfaces and granular or crystalline particles.

577

Microbial metabolism in fermented foods may modify fibrils in unknown ways, and phase or

578

state changes are likely to affect fibril interactions with other food components. Many

579

researchers (including ourselves) have noted the potential for fibrils to provide thickening,

580

viscosifying, emulsifying or encapsulating functionalities, and it is time to test this potential

581

under realistic conditions.

584 585 586

References

Adamcik, J., & Mezzenga, R. (2012). Proteins fibrils from a polymer physics perspective.

EP

583

TE D

582

M AN U

SC

575

Macromolecules, 45, 1137–1150. Adamcik, J., Jung, J. M., Flakowski, J., De Los Rios, P., Dietler, G., & Mezzenga, R. (2010).

588

Understanding amyloid aggregation by statistical analysis of atomic force microscopy

589 590 591 592 593

AC C

587

images. Nature Nanotechnology, 5, 423–428.

Adamcik, J., & Mezzenga, R. (2011). Adjustable twisting periodic pitch of amyloid fibrils. Soft Matter, 7, 5437–5443. Akkermans, C., Venema, P., Rogers, S., van der Goot, A., Boom, R., & van der Linden, E. (2006). Shear pulses nucleate fibril aggregation. Food Biophysics, 1, 144–150.

594

23

ACCEPTED MANUSCRIPT

Akkermans, C., Venema, P., van der Goot, A. J., Boom, R. M., & van der Linden, E. (2008a).

595

Enzyme-induced formation of β-lactoglobulin fibrils by AspN endoproteinase. Food

596

Biophysics, 3, 390–394.

597

Akkermans, C., Venema, P., van der Goot, A. J., Gruppen, H., Bakx, E. J., Boom, R. M., et al. (2008b). Peptides are building blocks of heat-induced fibrillar protein aggregates of β-

599

lactoglobulin formed at pH 2. Biomacromolecules, 9, 1474–1479.

RI PT

598

Akkermans, C., van der Goot, A. J., Venema, P., van der Linden, E., & Boom, R. M. (2008c).

601

Formation of fibrillar whey protein aggregates: Influence of heat and shear treatment,

602

and resulting rheology. Food Hydrocolloids, 22, 1315–1325.

603

SC

600

Akkermans, C., van der Goot, A. J., Venema, P., van der Linden, E., & Boom, R. M. (2008d). Properties of protein fibrils in whey protein isolate solutions: Microstructure, flow

605

behaviour and gelation. International Dairy Journal, 18, 1034–1042.

M AN U

604

606

Alting, A. C., de Jongh, H. H. J., Visschers, R. W., & Simons, J. W. F. A. (2002). Physical and

607

chemical interactions in cold gelation of food proteins. Journal of Agricultural and Food

608

Chemistry, 50, 4682–4689.

612 613 614 615 616 617 618 619 620

TE D

611

formation of β-lactoglobulin fibrils. Biomacromolecules, 4, 1614–1622. Arnaudov, L. N., & de Vries, R. (2006). Strong impact of ionic strength on the kinetics of fibrilar aggregation of bovine β-lactoglobulin. Biomacromolecules, 7, 3490–3498.

EP

610

Arnaudov, L. N., deVries, R., Ippel, H., & vanMierlo, C. P. M. (2003). Multiple steps during the

Arosio, P., Beeg, M., Nicoud, L., & Morbidelli, M. (2012). Time evolution of amyloid fibril length distribution described by a population balance model. Chemical Engineering Science,

AC C

609

78, 21–32.

Aymard, P., Nicolai, T., Durand, D., & Clark, A. (1999). Static and dynamic scattering of βlactoglobulin aggregates formed after heat-induced denaturation at pH 2.

Macromolecules, 32, 2542–2552. Bateman, L., Ye, A., & Singh, H. (2010). In vitro digestion of β-lactoglobulin fibrils formed by heat treatment at low pH. Journal of Agricultural and Food Chemistry, 58, 9800–9808.

621

24

ACCEPTED MANUSCRIPT

Bateman, L., Ye, A., & Singh, H. (2011). Re-formation of fibrils from hydrolysates of β-

622

Lactoglobulin fibrils during in vitro gastric digestion. Journal of Agricultural and Food

623

Chemistry, 59, 9605–9611.

625 626

Bekard, I. B., Asimakis, P., Bertolini, J., & Dunstan, D. E. (2011). The effects of shear flow on protein structure and function. Biopolymers, 95, 733–745. Berson, J. F., Theos, A. C., Harper, D. C., Tenza, D., Raposo, G., & Marks, M. S. (2003).

RI PT

624

Proprotein convertase cleavage liberates a fibrillogenic fragment of a resident

628

glycoprotein to initiate melanosome biogenesis. Journal of Cell Biology, 161, 521–533.

629

Blijdenstein, T. B. J., Veerman, C., & van der Linden, E. (2004). Depletion-flocculation in oil-in-

631

water emulsions using fibrillar protein assemblies. Langmuir, 20, 4881–4884. Bolder, S. G., Hendrickx, H., Sagis, L. M. C., & van der Linden, E. (2006). Fibril assemblies in

M AN U

630

SC

627

632

aqueous whey protein mixtures. Journal of Agricultural and Food Chemistry, 54, 4229–

633

4234.

634

Bolder, S. G., Sagis, L. M. C., Venema, P., & van der Linden, E. (2007). Effect of stirring and seeding on whey protein fibril formation. Journal of Agricultural and Food Chemistry, 55,

636

5661–5669.

637

TE D

635

Bolisetty, S., Harnau, L., Jung, J. M., & Mezzenga, R. (2012). Gelation, phase behavior, and dynamics of β-lactoglobulin amyloid fibrils at varying concentrations and ionic strengths.

639

Biomacromolecules, 13, 3241–3252.

EP

638

Bolisetty, S., Boddupalli, C. S., Handschin, S., Chaitanya, K., Adamcik, J., Saito, Y., et al.

641

(2014). Amyloid fibrils enhance transport of metal nanoparticles in living cells and

642 643 644 645

AC C

640

induced cytotoxicity. Biomacromolecules, 15, 2793–2799.

Bolisetty, S., & Mezzenga, R. (2016). Amyloid-carbon hybrid membranes for universal water purification. Nature Nanotechnology, 11, 365–371.

Dave, A. C., Loveday, S. M., Anema, S. G., Loo, T. S., Norris, G. E., Jameson, G. B., et al.

646

(2013). β-Lactoglobulin self-assembly: Structural changes in early stages and disulfide

647

bonding in fibrils. Journal of Agricultural and Food Chemistry, 61, 7817–7828.

648

25

ACCEPTED MANUSCRIPT

Dave, A. C., Loveday, S. M., Anema, S. G., Jameson, G. B., & Singh, H. (2014a). Glycation as

649

a tool to probe the mechanism of β-lactoglobulin nanofibril self-assembly. Journal of

650

Agricultural and Food Chemistry, 62, 3269–3278.

651

Dave, A. C., Loveday, S. M., Anema, S. G., Jameson, G. B., & Singh, H. (2014b). Modulating βlactoglobulin nanofibril self-assembly at pH 2 using glycerol and sorbitol.

653

Biomacromolecules, 15, 95–103.

RI PT

652

Dueholm, M. S., Albertsen, M., Otzen, D., & Nielsen, P. H. (2012). Curli functional amyloid

655

systems are phylogenetically widespread and display large diversity in operon and

656

protein structure. PLoS One, 7, e51274.

657

SC

654

Dunstan, D. E., Hamilton-Brown, P., Asimakis, P., Ducker, W., & Bertolini, J. (2009). Shearinduced structure and mechanics of β-lactoglobulin amyloid fibrils. Soft Matter, 5, 5020-

659

5028.

660

M AN U

658

Edwards, P. J. B., & Jameson, G. B. (2014). Structure and stability of whey proteins. In A.

661

Thompson, M. Boland & H. Singh (Eds.), Milk proteins: From expression to food (2nd

662

ed., pp. 202–242). San Diego, CA, USA: Academic Press. Foegeding, E. A., Luck, P., & Vardhanabhuti, B. (2011). Milk protein products | whey protein

TE D

663 664

products. In J. W. Fuquay (Ed.), Encyclopedia of dairy sciences (2nd edn., pp. 873–878).

665

San Diego, CA, USA: Academic Press.

Goers, J., Permyakov, S. E., Permyakov, E. A., Uversky, V. N., & Fink, A. L. (2002).

EP

666

Conformational prerequisites for α-lactalbumin fibrillation. Biochemistry, 41, 12546–

668

12551.

669 670 671 672 673 674

AC C

667

Goldschmidt, L., Teng, P. K., Riek, R., & Eisenberg, D. (2010). Identifying the amylome, proteins capable of forming amyloid-like fibrils. Proceedings of the National Academy of Sciences of the United States of America, 107, 3487–3492.

Gosal, W. S., Clark, A. H., & Ross-Murphy, S. B. (2004a). Fibrillar β-lactoglobulin gels: Part 1. Fibril formation and structure. Biomacromolecules, 5, 2408–2419. Gosal, W. S., Clark, A. H., & Ross-Murphy, S. B. (2004b). Fibrillar β-lactoglobulin gels: Part 2.

675

Dynamic mechanical characterization of heat-set systems. Biomacromolecules, 5, 2420–

676

2429.

677 678 679

26

ACCEPTED MANUSCRIPT

Hamada, D., & Dobson, C. M. (2002). A kinetic study of β-lactoglobulin amyloid fibril formation promoted by urea. Protein Science, 11, 2417–2426. Hamada, D., Tanaka, T., Tartaglia, G. G., Pawar, A., Vendruscolo, M., Kawamura, M., et al.

680

(2009). Competition between folding, native-state dimerisation and amyloid aggregation

681

in β-lactoglobulin. Journal of Molecular Biology, 386, 878–890. Hettiarachchi, C. A., Melton, L. D., Gerrard, J. A., & Loveday, S. M. (2012). Formation of β-

RI PT

682 683

lactoglobulin nanofibrils by microwave heating gives a structural assembly different to

684

conventional heating. Biomacromolecules, 13, 2868–2880.

Hettiarachchi, C. A. (2013). β-Lactoglobulin nanofibrils and their interactions with pectins. PhD

SC

685

thesis, University of Auckland, Auckland, New Zealand. Retrieved 25 July 2016 from

687

http://hdl.handle.net/2292/21095

688

M AN U

686

Hettiarachchi, C. A., Melton, L. D., McGillivray, D. J., Loveday, S. M., Gerrard, J. A., & Williams,

689

M. A. K. (2016a). β-Lactoglobulin nanofibrils can be assembled into nanotapes via site-

690

specific interactions with pectin. Soft Matter, 12, 756–768.

691

Hettiarachchi, C. A., Melton, L. D., Williams, M. A. K., McGillivray, D. J., Gerrard, J. A., & Loveday, S. M. (2016b). Morphology of complexes formed between β-lactoglobulin

693

nanofibrils and pectins is influenced by the pH and structural characteristics of the

694

pectins. Biopolymers, 105, 819–831.

697 698 699 700 701 702 703

EP

696

Hill, E. K., Krebs, B., Goodall, D. G., Howlett, G. J., & Dunstan, D. E. (2006). Shear flow induces amyloid fibril formation. Biomacromolecules, 7, 10–13. Hoare, D. G., & Koshland Jr., D. E. (1966). A procedure for the selective modification of

AC C

695

TE D

692

carboxyl groups in proteins. Journal of the American Chemical Society, 88, 2057–2058.

Iconomidou, V. A., & Hamodrakas, S. J. (2008). Natural protective amyloids. Current Protein and Peptide Science, 9, 291–309.

Ipsen, R., & Otte, J. (2007). Self-assembly of partially hydrolysed α-lactalbumin. Biotechnology Advances, 25, 602–605. Jones, O. G., Handschin, S., Adamcik, J., Harnau, L., Bolisetty, S., & Mezzenga, R. (2011).

704

Complexation of β-lactoglobulin fibrils and sulfated polysaccharides.

705

Biomacromolecules, 12, 3056–3065.

706

27

ACCEPTED MANUSCRIPT

Jordens, S., Ruhs, P. A., Sieber, C., Isa, L., Fischer, P., & Mezzenga, R. (2014). Bridging the

707

gap between the nanostructural organization and macroscopic interfacial rheology of

708

amyloid fibrils at liquid interfaces. Langmuir, 30, 10090–10097.

709

Jung, J. M., Gunes, D. Z., & Mezzenga, R. (2010). Interfacial activity and interfacial shear rheology of native β-lactoglobulin monomers and their heat-induced fibers. Langmuir,

711

26, 15366–15375.

712

RI PT

710

Kavanagh, G. M., Clark, A. H., & Ross-Murphy, S. B. (2000a). Heat-induced gelation of globular proteins: Part 3. Molecular studies on low pH β-lactoglobulin gels. International Journal

714

of Biological Macromolecules, 28, 41–50.

715

SC

713

Kavanagh, G. M., Clark, A. H., & Ross-Murphy, S. B. (2000b). Heat-induced gelation of globular proteins: 4. Gelation kinetics of low pH β-lactoglobulin gels. Langmuir, 16, 9584–9594.

717

Krebs, M. R. H., Morozova-Roche, L. A., Daniel, K., Robinson, C. V., & Dobson, C. M. (2004).

M AN U

716

718

Observation of sequence specificity in the seeding of protein amyloid fibrils. Protein

719

Science, 13, 1933–1938.

721

Kroes-Nijboer, A., Venema, P., Bouman, J., & van der Linden, E. (2011). Influence of protein hydrolysis on the growth kinetics of β-lg fibrils. Langmuir, 27, 5753–5761.

TE D

720

Lara, C., Adamcik, J., Jordens, S., & Mezzenga, R. (2011). General self-assembly mechanism

723

converting hydrolyzed globular proteins into giant multistranded amyloid ribbons.

724

Biomacromolecules, 12, 1868–1875.

726 727 728 729

Lassé, M., Ulluwishewa, D., Healy, J., Thompson, D., Miller, A., Roy, N., et al. (2015). Evaluation of protease resistance and toxicity of amyloid-like food fibrils from whey, soy,

AC C

725

EP

722

kidney bean, and egg white. Food Chemistry, 192, 491–498.

Lawrence, L., & Moore, W. J. (1951). Kinetics of the hydrolysis of simple glycine peptides. Journal of the American Chemical Society, 73, 3973–3977.

730

Loveday, S. M., Rao, M. A., Creamer, L. K., & Singh, H. (2009). Factors affecting rheological

731

characteristics of fibril gels: The case of β-lactoglobulin and α-lactalbumin. Journal of

732

Food Science, 74, R47–R55.

28

ACCEPTED MANUSCRIPT

733

Loveday, S. M., Wang, X. L., Rao, M. A., Anema, S. G., Creamer, L. K., & Singh, H. (2010).

734

Tuning the properties of β-lactoglobulin nanofibrils with pH, NaCl and CaCl2.

735

International Dairy Journal, 20, 571–579. Loveday, S. M., Su, J., Rao, M. A., Anema, S. G., & Singh, H. (2011a). Effect of calcium on the

737

morphology and functionality of whey protein nanofibrils. Biomacromolecules, 12, 3780–

738

3788.

739

RI PT

736

Loveday, S. M., Wang, X. L., Rao, M. A., Anema, S. G., & Singh, H. (2011b). Effect of pH, NaCl, CaCl2 and temperature on self-assembly of b-lactoglobulin into nanofibrils: A central

741

composite design study. Journal of Agricultural and Food Chemistry, 59, 8467–8474.

742

SC

740

Loveday, S. M., Su, J., Rao, M. A., Anema, S. G., & Singh, H. (2012a). Whey protein nanofibrils: Kinetic, rheological and morphological effects of group IA and IIA cations.

744

International Dairy Journal, 26, 133–140.

745

M AN U

743

Loveday, S. M., Su, J., Rao, M. A., Anema, S. G., & Singh, H. (2012b). Whey protein

746

nanofibrils: The environment-morphology-functionality relationship in lyophilization,

747

rehydration, and seeding. Journal of Agricultural and Food Chemistry, 60, 5229–5236. Loveday, S. M., Wang, X. L., Rao, M. A., Anema, S. G., & Singh, H. (2012c). β-Lactoglobulin

749

nanofibrils: Effect of temperature on fibril formation kinetics, fibril morphology and the

750

rheological properties of fibril dispersions. Food Hydrocolloids, 27, 242–249.

753 754 755 756 757

EP

752

Loveday, S. M. (2016). β-Lactoglobulin heat denaturation: A critical assessment of kinetic modelling. International Dairy Journal, 52, 92–100. Mercadante, D., Melton, L. D., Norris, G. E., Loo, T. S., Williams, M. A. K., Dobson, R. C. J., et

AC C

751

TE D

748

al. (2012). Bovine β-lactoglobulin is dimeric under imitative physiological conditions: Dissociation equilibrium and rate constants over the pH range of 2.5–7.5. Biophysical

Journal, 103, 303–312.

Mezzenga, R., & Fischer, P. (2013). The self-assembly, aggregation and phase transitions of

758

food protein systems in one, two and three dimensions. Reports on Progress in Physics,

759

76, 046601.

760 761

Mishra, R., Sörgjerd, K., Nyström, S., Nordigården, A., Yu, Y.-C., & Hammarström, P. (2007). Lysozyme amyloidogenesis is accelerated by specific nicking and fragmentation but

29

ACCEPTED MANUSCRIPT

762

decelerated by intact protein binding and conversion. Journal of Molecular Biology, 366,

763

1029–1044.

764

Mudgal, P., Daubert, C. R., & Foegeding, E. A. (2009). Cold-set thickening mechanism of β-

765

lactoglobulin at low pH: Concentration effects. Food Hydrocolloids, 23, 1762–1770.

766

Mudgal, P., Daubert, C. R., & Foegeding, E. A. (2011). Effects of protein concentration and CaCl2 on cold-set thickening mechanism of β-lactoglobulin at low pH. International Dairy

768

Journal, 21, 319–326.

769

RI PT

767

Munialo, C. D., de Jongh, H. H. J., Broersen, K., van der Linden, E., & Martin, A. H. (2013). Modulation of the gelation efficiency of fibrillar and spherical aggregates by means of

771

thiolation. Journal of Agricultural and Food Chemistry, 61, 11628–11635.

SC

770

Murray, B., Rosenthal, J., Zheng, Z., Isaacson, D., Zhu, Y., & Belfort, G. (2015). Cosolute

773

effects on amyloid aggregation in a nondiffusion limited regime: Intrinsic osmolyte

774

properties and the volume exclusion principle. Langmuir, 31, 4246–4254.

776 777

Nilsson, M. R. (2004). Techniques to study amyloid fibril formation in vitro. Methods, 34, 151– 160.

Oboroceanu, D., Wang, L., Brodkorb, A., Magner, E., & Auty, M. A. E. (2010). Characterization

TE D

775

M AN U

772

of β-lactoglobulin fibrillar assembly using atomic force microscopy, polyacrylamide gel

779

electrophoresis, and in situ Fourier transform infrared spectroscopy. Journal of

780

Agricultural and Food Chemistry, 58, 3667–3673.

EP

778

Oboroceanu, D., Wang, L., Magner, E., & Auty, M. A. E. (2014). Fibrillization of whey proteins

782

improves foaming capacity and foam stability at low protein concentrations. Journal of

783

AC C

781

Food Engineering, 121, 102–111.

784

Otzen, D. (2010). Functional amyloid: Turning swords into plowshares. Prion, 4, 256–264.

785

Peng, J., Kroes-Nijboer, A., Venema, P., & van der Linden, E. (2016). Stability of colloidal

786

dispersions in the presence of protein fibrils. Soft Matter, 12, 3514–3526.

787

Peram, M. R., Loveday, S. M., Ye, A., & Singh, H. (2013). In vitro gastric digestion of heat-

788

induced aggregates of b-lactoglobulin. Journal of Dairy Science, 96, 63–74.

789

30

ACCEPTED MANUSCRIPT

Rasmussen, P., Barbiroli, A., Bonomi, F., Faoro, F., Ferranti, P., Iriti, M., et al. (2007).

790

Formation of structured polymers upon controlled denaturation of β-lactoglobulin with

791

different chaotropes. Biopolymers, 86, 57–72.

792

Sasso, L., Suei, S., Domigan, L., Healy, J., Nock, V., Williams, M. A. K., et al. (2014). Versatile multi-functionalization of protein nanofibrils for biosensor applications. Nanoscale, 6,

794

1629–1634.

RI PT

793

Schleeger, M., vandenAkker, C. C., Deckert-Gaudig, T., Deckert, V., Velikov, K. P., Koenderink,

796

G., et al. (2013). Amyloids: From molecular structure to mechanical properties. Polymer,

797

54, 2473–2488.

SC

795

Schultz, J. (1967). [28] Cleavage at aspartic acid. Methods in Enzymology, 11, 255–263.

799

Serfert, Y., Lamprecht, C., Tan, C. P., Keppler, J. K., Appel, E., Rossier-Miranda, F. J., et al.

M AN U

798

800

(2014). Characterisation and use of β-lactoglobulin fibrils for microencapsulation of

801

lipophilic ingredients and oxidative stability thereof. Journal of Food Engineering, 143,

802

53–61.

805 806 807

Biology, 130, 88–98.

TE D

804

Sipe, J. D., & Cohen, A. S. (2000). Review: History of the amyloid fibril. Journal of Structural

Sukenik, S., & Harries, D. (2012). Insights into the disparate action of osmolytes and macromolecular crowders on amyloid formation. Prion, 6, 26–31. Sukenik, S., Politi, R., Ziserman, L., Danino, D., Friedler, A., & Harries, D. (2011). Crowding

EP

803

alone cannot account for cosolute effect on amyloid aggregation. PLoS One, 6, e15608.

809

Swaisgood, H. E. (1982). Chemistry of milk protein. In Developments in dairy chemistry (Vol. 1,

810 811 812 813

AC C

808

pp. 1–59). London, UK: Applied Science Publishers.

Thorn, D. C., Ecroyd, H., Carver, J. A., & Holt, C. (2015). Casein structures in the context of unfolded proteins. International Dairy Journal, 46, 2–11.

Usov, I., Adamcik, J., & Mezzenga, R. (2013). Polymorphism in bovine serum albumin fibrils:

814

Morphology and statistical analysis. Faraday Discussions, 166, 151–162.

815

Usov, I., & Mezzenga, R. (2014). Correlation between nanomechanics and polymorphic

816

conformations in amyloid fibrils. ACS Nano, 8, 11035–11041.

817

31

ACCEPTED MANUSCRIPT

Veerman, C., Baptist, H., Sagis, L. M. C., & van der Linden, E. (2003a). A new multistep Ca2+-

818

induced cold gelation process for β-lactoglobulin. Journal of Agricultural and Food

819

Chemistry, 51, 3880–3885.

820

Veerman, C., Sagis, L. M. C., Heck, J., & van der Linden, E. (2003b). Mesostructure of fibrillar bovine serum albumin gels. International Journal of Biological Macromolecules, 31, 139–

822

146.

RI PT

821

Veerman, C. (2006). Method for improving the functional properties of a globular protein, protein

824

thus prepared, use thereof and products containing the protein. Patent PCT/EP03/13678

825

(US2006/0204454A1)

826

SC

823

Wu, X., Nishinari, K., Gao, Z., Zhao, M., Zhang, K., Fang, Y., et al. (2016). Gelation of βlactoglobulin and its fibrils in the presence of transglutaminase. Food Hydrocolloids, 52,

828

942–951.

M AN U

827

AC C

EP

TE D

829

1

ACCEPTED MANUSCRIPT

Figure legends

2 Fig. 1. Transmission electron micrograph of bovine β-lg fibrils formed by heating at pH 2 and 80

4

°C (top, Loveday et al. unpublished); schematic illustration of the stacked β-sheet internal

5

structure of fibrils and the coiling behaviour of a pair of entwined fibrils (bottom). Schematic

6

reprinted with permission from Usov and Mezzenga (2014), ACS Nano, 8, 11035–11041.

7

Copyright 2014 American Chemical Society.

8

RI PT

3

Fig. 2. Schematic representation of amyloid-like fibril formation at denaturing temperatures, low

10

pH and low ionic strength. Adapted from Dave et al. (2014b), Biomacromolecules, 15, 95–103.

11

Copyright 2014 American Chemical Society. Fibril structure adapted from Protein Data Bank

12

entry 2FKG.

M AN U

SC

9

13

Fig. 3. Atomic force microscopy images from Serfert et al. (2014) showing the influence of pH

15

on the morphology of β-lg aggregates formed during heating for 5 h at 90 °C without added

16

electrolytes (top); transmission electron microscopy image of fibrils at pH 1.6 and kinetics of

17

fibril assembly at 80 °C and pH 1.6 to 2.4, as measured via ThT fluorescence (Loveday et al.,

18

2010). Data at pH 2 were almost identical to those at pH 2.2, and are not shown for that reason.

19

Top image reprinted from Journal of Food Engineering, 143, 53–61, Copyright (2014), with

20

permission from Elsevier. Bottom images reprinted from International Dairy Journal, 20, 571–

21

579, Copyright (2010) with permission from Elsevier.

EP

AC C

22

TE D

14

23

Fig. 4. Gel cure data at 80 °C for 5% (w/w) WPI solutions at pH 2 with no added salt (control) or

24

with 120 mM NaCl, KCl, MgCl2 or CaCl2 (top) (Loveday et al., 2012a); transmission electron

25

micrographs of fibril mixtures formed at intermediate ionic strength after heating 1 % β-lg at pH

26

2 for 6 h at 80 °C (bottom) (Loveday et al., 2010). Top figure reprinted from International Dairy

27

Journal, 26, 133–140, Copyright (2012) with permission from Elsevier. Bottom figure reprinted

28

from International Dairy Journal, 20, 571–579, Copyright (2010) with permission from Elsevier.

29

ACCEPTED MANUSCRIPT

30

Fig. 5. Effect of heating temperature on whey protein fibril assembly kinetics and morphology.

31

Reprinted from Food Hydrocolloids, 27, 242–249, Copyright (2012) with permission from

32

Elsevier.

33 Fig. 6. Viscosity of fibril dispersions before lyophilizing or after lyophilizing and rehydrating in

35

water at 2%, 4%, or 10% (w/w). Measurements were made after stirring overnight at 4 °C to

36

ensure complete hydration. Fibrils made with 80 mM CaCl2 could not be rehydrated at 10% w/w

37

because a thick gel was formed, preventing even dispersion and hydration. Different symbols

38

show apparent viscosity at different shear rates, and vertical bars are min and max of two

39

replicates. Reprinted with permission from Loveday et al. (2012b), Journal of Agricultural and

40

Food Chemistry, 60, 5229-5236. Copyright (2012) American Chemical Society.

M AN U

SC

RI PT

34

41

Fig. 7. Overview of fibril-pectin nanotape morphology. Images a to c, transmission electron

43

microscopy; d, cryo-electron microscopy; e and f, scanning electron microscopy. Images d and

44

e from Hettiarachchi et al. (2016b) used with permission of the authors. Images a and c from

45

Hettiarachchi (2013) used with permission of the author. Images b and f from Hettiarachchi et

46

al. (2016a) with permission from the Royal Society of Chemistry.

EP

48

AC C

47

TE D

42

ACCEPTED MANUSCRIPT Table 1 Effect of WPI or fibrils on the final oscillatory elastic modulus (G’) of acid milk gels.

Final G' (Pa)

Fibrils or WPI added before heating WPI Salt-free semiflexible fibrils Salt-free control Dialysed wormlike fibrils Dialysed control

538 343 439 324 497

Fibrils or WPI added after heating Skim milk Salt-free semiflexible fibrils Salt-free control Dialysed wormlike fibrils Dialysed control

202 176 60 117 83

AC C

EP

TE D

M AN U

SC

RI PT

Treatment

ACCEPTED MANUSCRIPT Table 2

Treatment Casein control (18% protein) Salt-free semiflexible fibrils Salt-free control Dialysed wormlike fibrils Dialysed control Casein control (16% protein)

Firmness (g) 222.0 226.0 130.8 263.9 91.9 92.7

a

RI PT

Firmness of model processed cheese samples. a

AC C

EP

TE D

M AN U

SC

One control sample contained 18% protein and the other contained 16% protein. All other samples contained 18% protein, but 1% of the protein was from either fibrils (semiflexible or worm-like) or their precursor solutions containing non-fibril protein material that was dialysed and lyophilised in the same way as fibril dispersions.

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

partially denatured

peptides

non-fibril peptides

EP

AC C

fibrils with extra loops

TE D

M AN U

SC

native

RI PT

I NCREA SIN G H EAT IN G T IME

non-fibril peptides

‘shaved’ fibrils

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT