Effect of Non-Covalent Interactions on Arabinoxylan–Protein Cross-Linking and Gluten-Free Batter Stability

Abstract

Maize arabinoxylans (AX) and proteins (maize gluten meal, MGM) can partially replace gluten in gluten-free (GF) breads by forming polymer networks. This study investigated how non-covalent interactions (hydrophobic, electrostatic, or hydrogen (H) forces) influenced viscoelasticity, gas retention and enzymatic AX–protein cross-linking in simplified GF model batters using two maize AX extracts (commercial MAX; xylanase-extracted M-XEAX). Batter stability strongly depended on AX structure and formulation type. MGM-only controls were mainly governed by hydrophobic and electrostatic forces, while AX-based batters relied primarily on H-bonds and electrostatic interactions. Combining MGM and AX increased batter stiffness, dominated by electrostatic and H-interactions. Enzymatic coupling reinforced the AX–protein network when both H and electrostatic forces were present, whereas hydrophobic interactions partly hindered these associations. Changes in viscoelasticity (G′) did not fully align with gas retention behaviour. In MGM-containing batters, gas retention was predominantly governed by H and electrostatic interactions. AX-based batters showed extract-dependent responses: electrostatic or H-interactions hindered gas stabilisation in M-XEAX, while their suppression supported gas-holding in enzyme-treated MAX batters. AX-MGM systems generally showed reduced gas expansion, indicating the contribution of multiple non-covalent interactions. Overall, batter stability strongly depended on AX structure, MGM addition, the balance of non-covalent interactions and the resulting network strength.

1. Introduction

A major challenge in the production of gluten-free (GF) baked goods is the absence of gluten. This protein forms a strong, covalent network responsible for the viscoelastic dough properties (i.e., gas-holding, stability), baking performance and overall product quality, including textural and sensory characteristics [1]. As a consequence, GF breads typically exhibit reduced structural integrity, limited gas retention capacity, lower loaf volume and reduced crumb cohesiveness [2,3].

To overcome these techno-functional limitations of GF bread, various formulation strategies have been implemented, including the application of fibre-based hydrocolloids and proteins extracted from GF cereals. Among these, maize arabinoxylans (AX) and zein have shown particular potential as natural gluten substitutes due to their network-forming ability, improving texture and overall product quality [4,5,6]. Both AX and proteins can form covalent cross-links, which makes them particularly interesting for structuring GF batters. However, the extent of these interactions strongly depends on the polymer structure (molecular size, branching, functional side group content/accessibility), and on the presence of catalysing enzymes [7].

In addition to covalent bonding, non-covalent interactions, including hydrogen, hydrophobic or electrostatic interactions, can either weaken or reinforce polymer associations, thereby affecting network formation as well as mechanical and functional batter properties [8]. Recent studies have demonstrated that non-covalent interactions contribute to the rheological behaviour and structural performance of GF batters [9,10,11]; however, their specific role and relative importance in batter stabilisation and protein–AX cross-linking remains insufficiently understood. These effects are highly system-dependent and are influenced by the processing conditions; the ingredients, including different polysaccharide- and protein-based extracts; as well as their composition and structure.

To address these knowledge gaps, this study investigated how non-covalent interactions affect arabinoxylan–protein cross-linking and thereby influence key gluten-free batter properties, including viscoelasticity and gas retention. For this purpose, two structurally distinct maize AX extracts were used in combination with maize protein in a model GF batter recipe to investigate substrate-specific interaction effects. Non-covalent interactions between polymers were supressed using sodium chloride, urea and sodium dodecyl sulphate (SDS), which primarily, yet not exclusively, affect electrostatic, hydrogen and hydrophobic interactions respectively, as reported in previous studies [11,12]. These agents may additionally influence ionic strength and/or structural polymer properties [13]; however, such secondary effects were not addressed in detail in the present study. The role of non-covalent interactions on viscoelastic behaviour and gas retention capacity were assessed through rheological time sweeps and fermentation trials. In addition, enzymatic cross-linking of AX and protein was applied to examine the contribution of non-covalent interactions to network formation and strength in GF batters. It was hypothesised that the AX extract and the presence of maize gluten meal (MGM) would significantly affect the extent of different non-covalent interactions, resulting in substantial changes in viscoelasticity and gas retention capacity. Furthermore, it was assumed that supressing non-covalent interactions would alter the efficiency of AX–protein cross-linking. Overall, evaluating these interactions is essential to understand how non-covalent interactions influence polymer cross-linking mechanisms and contribute to batter stability.

2. Materials and Methods

2.1. Materials

Maize gluten meal (MGM), maize bran for AX extraction and native maize starch were provided by AGRANA Beteiligungs-AG (Vienna, Austria). Commercial maize arabinoxylan (MAX) was donated by ClonBio Group Ltd. (Dublin, Ireland). Pre-gelatinised maize starch was obtained from Kröner-Stärke GmbH (Ibbenbüren, Germany). Sumizyme X was supplied by Takabio LCC (Beaucouzé, France), while glucose oxidase (GOX) from Aspergillus niger and horseradish peroxidase (HRP) were purchased from Sigma-Aldrich (Steinheim, Germany). Yeast was obtained from Lesaffre Austria AG (Wiener Neudorf, Austria). All reagents were purchased from Carl Roth GmbH + Co. KG (Karlsruhe, Germany) and Sigma-Aldrich (Steinheim, Germany).

2.2. Arabinoxylan Extraction

Xylanase-extracted maize AX (M-XEAX) was extracted from maize bran. The pilot scale extraction process followed a modified protocol based on Bender et al. (2018) [4,9]. Modifications included the xylanase treatment (Sumizyme X, 20,000 U/g; addition of 0.1 g/1000 g bran) of a maize bran suspension (1:10, w/v distilled water; pH 5) for 2 h at 55 °C under stirring. The resulting supernatant was subsequently neutralised (pH 7), decanted (Type P 600, No. 2076, Sharples, West Chester, PA, USA), and purified by ultrafiltration using a 100 kDa cut-off membrane (MANN + HUMMEL Water & Fluid Solutions GmbH, Wiesbaden, Germany) on an MMS SW18 Pilot System (MMS AG, Urdorf, Switzerland) at 50 °C, an average flow rate of 2.3 L/h and 1.3 bar transmembrane pressure. After ultrafiltration, two sequential diafiltration steps were carried out until a concentration factor of 4 was reached. Nutritional composition (including total phenolic/ferulic acid content [14]) of the freeze-dried AX extracts, including MAX and M-XEAX, and the maize protein can be found in

Table A1. Nutritional composition (in %, based on dry matter) of maize arabinoxylan (AX) extracts and maize gluten meal (MGM).

Extracts	Proteins	Dietary Fibre	AX 1	A/X Ratio	Starch	Free Glucose	Ash	Total Bound Phenolic Content 2 (FA 3 Content)	Total Conjugated Phenolics	Total Free Phenolics
MAX	<8 4	>80 4	Arabinose: 31 4 Xylose: 51 4	0.61	<0.5 4	- 5	-	0.04 ± <0.01 (FA: 0.011 ± <0.001)	-	<0.01
M-XEAX 6	5.1 ± 0.1	1.4 ± 0.9 (insoluble) 32.0 ± 0.8 (soluble)	28.4 ± 1.5	0.77	14.5 ± 0.6	1.1 ± 0.1	0.7 ± <0.1	0.48 ± 0.03 (FA: 0.16 ± 0.01)	0.43 ± <0.01	0.65 ± 0.03
MGM 6	66.7 ± 0.4	4.4 ± 1.5	-	-	12.2 ± 0.4	-	2.0 ± <0.1	-	-	-

¹ Calculated as the sum of arabinose and xylose, ² expressed as g ferulic acid/100 g sample, ³ ferulic acid, ⁴ according to producer, ⁵ not determined or detectable, ⁶ data provided by Sukop et al. (2025) [9]. Values represent mean ± standard deviation, results were performed in triplicates.M-XEAX: xylanase-extracted maize arabinoxylan, MAX: commercial maize AX.

. The molecular size distribution of the AX extracts was assessed by size-exclusion chromatography (SEC-HPLC) following an established protocol described previously [9] (see Figure A1).

2.3. Batter Preparation

An overview of the GF model batter formulations, applied components and concentrations is provided in

Table 1. Composition of gluten-free (GF) model batters containing xylanase-extracted (M-XEAX) or commercial maize arabinoxylan (MAX), combined with maize gluten meal (MGM). Addition of chemical reagents (sodium chloride, NaCl; urea; sodium dodecyl sulphate, SDS) and enzymes (glucose oxidase, GOX; horseradish peroxidase, HRP) was carried out depending on the formulation.

Components	AX-Based Batters	AX+MGM-Based Batters	MGM-Based Batters
Buffer (50 mM, pH 5.5) (mL)	3.0	3.15 1	3.15 1
Native maize starch (g)	2.82	2.82	2.82
Pre-gelatinised starch (g)	0.18	0.18	0.18
M-XEAX or MAX (%) 2	2.6	2.6	– 3
MGM (%) 2	– 3	3	3
NaCl (mol/L)	Optional 0.3 mol/L	Optional 0.3 mol/L	Optional 0.3 mol/L
Urea (mol/L)	Optional 0.3 mol/L	Optional 0.3 mol/L	Optional 0.3 mol/L
SDS (%) 2	Optional 3% (w/w, starch basis)	Optional 3% (w/w, starch basis)	Optional 3% (w/w, starch basis)
GOX + HRP	Optional 0.02 U/mg	Optional 0.02 U/mg	Optional 0.02 U/mg

¹ Adjusted according to MGM water-holding capacity (1.11 mL buffer/g MGM); ² w/w additions of pure AX/protein/SDS on a starch basis; ³ Respective component was not added.

.

Control batters containing AX were prepared by dissolving 2.6% (w/w) of either M-XEAX (0.274 g/batter) or MAX (0.095 g/batter) in 3.0 mL 50 mM phosphate–citrate buffer (providing an initial pH of 5.5), and incubating for 1 h at 20 °C. The AX solution was mixed with dry ingredients (starch, pre-gelatinised starch, and MGM in AX + MGM formulations; see Table 1) using a spatula for 2 min. MGM-only batters (3%) were prepared following the same procedure, except that AX was omitted. To disrupt non-covalent interactions, buffers containing urea (0.3 mol/L; 0.054 g/batter), NaCl (0.3 mol/L; 0.053 g/batter) or SDS (3% w/w, based on total starch; 0.089 g/batter) were individually added, as described in previous studies. The applied concentrations were selected to enable controlled modulation of non-covalent interaction pathways [11,12]. Cross-linking was induced by applying GOX and HRP (each at 0.02 U/mg AX) immediately before analysis.

2.4. Rheological Analysis

Time-dependent oscillatory tests were performed using an MCR 302e rheometer (Anton Paar, Graz, Austria) equipped with a smooth parallel plate geometry (PP25 and INSET I-PP50/SS/P2, 25 mm, stainless steel). Batters were loaded onto the lower plate, compressed into a gap of 1 mm, trimmed, and the exposed edges were covered with low-viscosity silicone oil. Samples were allowed to rest for 6 min, followed by measurements at a constant shear strain of 0.02% and a frequency of 1 Hz for 30 min at 25 °C. The applied strain was selected within the linear viscoelastic region based on preliminary amplitude sweeps [9].The viscoelastic batter properties were evaluated by monitoring the storage (G′) and loss modulus (G″). The loss tangent (tan δ) was calculated as the ratio of G″ to G′.

2.5. Batter Expansion Analysis

Batters were prepared according to Section 2.3 with, additionally, 2% yeast and 1.5% sucrose (w/w; based on total starch). For fermentation trials, formulations were adjusted to a starch basis of 5 g (~10 g total batter mass). To preserve yeast functionality, sodium dodecyl sulphate (SDS) addition was omitted and a combination of sodium chloride (NaCl) and urea (0.3 mol/L each) was used instead to identify non-additive effects from which we could infer hydrophobic stabilisation. The batter was transferred to a 50 mL glass beaker and incubated for 30 min at 30 °C and 85% relative humidity (Easy Plus Pro fermentation chamber, Manz Backtechnik, Creglingen, Germany). Batter height was measured using a ruler along the beaker wall before and after fermentation. Batter expansion rate was calculated according to Equation (1):

$$\mathrm{Batter} \mathrm{expansion} \mathrm{rate} \left(\%\right) ={\displaystyle \frac{\mathrm{Batter} \mathrm{height} \mathrm{after} \mathrm{fermentation} \left(\mathrm{cm}\right)}{\mathrm{Batter} \mathrm{height} \mathrm{before} \mathrm{fermentation} \left(\mathrm{cm}\right)}}\times 100$$

(1)

2.6. Statistical Analysis

All measurements were conducted at least in triplicate (n ≥ 3) and statistically analysed using Statgraphics centurion 19–X64, (Statpoint Technologies, VA, USA). Results are presented as mean ± standard deviation. A one-way analysis of variance (ANOVA) and Fisher’s least significance difference tests were used to estimate significant differences between samples (p-value < 0.05), particularly between each batter system.

3. Results and Discussion

3.1. Time-Dependent Rheological Behaviour of Gluten-Free Batters

Figure 1 presents the time-dependent viscoelastic response (G′) of batters containing MGM, MAX or M-XEAX when H-bonds, hydrophobic or electrostatic interactions are mainly modulated by urea, SDS, or NaCl, displaying their roles in batter stability and in AX–protein cross-linking (raw data: Tables S1 and S2).

Across all model batters, the elastic modulus G′ exceeded the viscous modulus G″ throughout the time sweeps (tan δ < 1; see Supplementary Tables S1–S3), which indicated predominantly elastic behaviour of gel-like matrices, as commonly reported for GF formulations [9,11]. To facilitate comparison between formulations, the G′, G″ and tan δ of each model batter (after 30 min) are summarised in

Table 2. Storage modulus (G′), loss modulus (G″) and loss tangent (tan δ) of model batters (after 30 min) with commercial maize (MAX) or xylanase-extracted maize arabinoxylans (M-XEAX), with or without maize gluten meal (MGM) and enzymatic treatment. Values are given as mean ± SD (n ≥ 3). Different letters within one model system indicate significant differences (p < 0.05).

Model Batter Formulation		G′ [Pa]	G″ [Pa]	tan δ
MGM	Control	748 ± 224 b	81.1 ± 22.7 a	0.108 ± 0.009 a
	Control + Enzymes	509 ± 244 a,b	60.9 ± 28.1 a	0.120 ± 0.005 a,b
	Control + NaCl	460 ± 78 a,b	52.6 ± 8.6 a	0.114 ± 0.012 a,b
	Control + NaCl + Enzymes	653 ± 167 b	76.1 ± 15.6 a	0.117 ± 0.009 a,b
	Control + Urea	769 ± 479 b	87.8 ± 50.8 a	0.114 ± 0.008 a,b
	Control + Urea + Enzymes	380 ± 101 a, b	50.2 ± 11.8 a	0.132 ± 0.008 b
	Control + SDS	456 ± 216 a,b	87.0 ± 43.6 a	0.191 ± 0.017 c
	Control + SDS + Enzymes	265 ± 15 a	53.4 ± 5.5 a	0.201 ± 0.025 c
MAX	Control	1058 ± 159 a,b	230.8 ± 33.1 a,b	0.218 ± 0.004 a
	Control + Enzymes	993 ± 331 a,b	239.9 ± 57.3 a,b	0.242 ± 0.030 c
	Control + NaCl	1276 ± 52 b	267.7 ± 17.1 b,c	0.210 ± 0.005 a
	Control + NaCl + Enzymes	1135 ± 179 a,b	249.2 ± 33.0 a,b,c	0.220 ± 0.006 a
	Control + Urea	968 ± 114 a,b	216.1 ± 22.2 a,b	0.223 ± 0.006 a,b
	Control + Urea + Enzymes	843 ± 225 a	204.5 ± 47.5 a	0.242 ± 0.015 b,c
	Control + SDS	1043 ± 95 a,b	307.9 ± 26.0 c	0.295 ± 0.003 d
	Control + SDS + Enzymes	839 ± 35 a	256.2 ± 11.1 a,b,c	0.305 ± 0.010 d
M-XEAX	Control	724 ± 38 a,b	173.0 ± 10.8 a,b	0.239 ± 0.010 a,b,c
	Control + Enzymes	619 ± 111 a	160.4 ± 24.3 a,b	0.259 ± 0.004 c
	Control + NaCl	646 ± 295 a,b	158.4 ± 51.3 a,b	0.245 ± 0.030 b,c
	Control + NaCl + Enzymes	541 ± 65 a	146.4 ± 10.3 a,b	0.270 ± 0.015 b,c
	Control + Urea	508 ± 84 a	133.5 ± 18.4 a	0.263 ± 0.014 b,c
	Control + Urea + Enzymes	631 ± 157 a	157.0 ± 22.2 a,b	0.249 ± 0.036 b,c
	Control + SDS	963 ± 349 b,c	197.3 ± 62.7 b	0.205 ± 0.013 a
	Control + SDS + Enzymes	1155 ± 115 c	260.2 ± 23.8 c	0.225 ± 0.003 a,b
MAX-MGM	Control	2664 ± 1483 b	431.3 ± 246.1 b,c	0.162 ± 0.012 a,b
	Control + Enzymes	1526 ± 439 a	271.9 ± 66.0 a,b	0.178 ± 0.014 c
	Control + NaCl	1767 ± 215 a,b	276.4 ± 31.2 a,b	0.156 ± 0.004 a
	Control + NaCl + Enzymes	1231 ± 84 a	220.0 ± 10.5 a	0.179 ± 0.004 b,c
	Control + Urea	1361 ± 123 a	223.1 ± 16.7 a	0.164 ± 0.003 a,b,c
	Control + Urea + Enzymes	1096 ± 111 a	200.0 ± 18.7 a	0.182 ± 0.004 c
	Control + SDS	2320 ± 18 a,b	567.5 ± 29.3 c,d	0.245 ± 0.013 d
	Control + SDS + Enzymes	2354 ± 342 a,b	677.2 ± 186.0 d	0.288 ± 0.038 e
M-XEAX-MGM	Control	1356 ± 320 b,c,d	213.1 ± 52.0 a	0.157 ± 0.008 a
	Control + Enzymes	976 ± 7 a	162.4 ± 10.7 a	0.166 ± 0.010 a
	Control + NaCl	1241 ± 296 a,b,c,d	194.9 ± 37.4 a	0.157 ± 0.009 a
	Control + NaCl + Enzymes	1044 ± 123 a,b,c	175.3 ± 11.8 a	0.168 ± 0.010 a
	Control + Urea	1031 ± 94 a,b	170.7 ± 14.7 a	0.166 ± 0.004 a
	Control + Urea + Enzymes	984 ± 520 a	171.0 ± 90.6 a	0.174 ± 0.095 a
	Control + SDS	1471 ± 325 d	345.5 ± 63.4 b	0.235 ± 0.018 b
	Control + SDS + Enzymes	1420 ± 84 c,d	351.6 ± 13.6 b	0.248 ± 0.012 b

Abbreviations: SDS refers to sodium dodecyl sulphate; NaCl to sodium chloride.

.

3.1.1. Effect of Model Batters Added with Protein

The MGM control batter (see Figure 1) exhibited a small rise in G′ that rapidly plateaued at 748 Pa, indicating weak protein–protein interactions. While SDS and NaCl reduced G′ to 450 Pa and 456 Pa, urea did not affect it. Moreover, only SDS addition significantly increased tan δ, revealing less solid-like behaviour and a weaker batter structure. Overall, this suggested that hydrophobic and electrostatic forces primarily governed the batter’s stability.

Enzyme addition did not lead to effective cross-linking in the control batter as G′_t=30min decreased to 509 Pa, indicating a radical-induced disruption of weak non-covalent protein associations [5]. A further decline was observed with urea or SDS (to 380 Pa and 265 Pa, respectively), suggesting that H- and hydrophobic interactions could have played a role in protein cross-linking. NaCl addition increased G′ values (+42%), highlighting that electrostatic shielding favoured protein cross-linking [15]. Overall, hydrophobic and electrostatic interactions governed MGM batter stability, while suppressing the electrostatic repulsions enhanced batter strength and facilitated protein cross-linking.

3.1.2. Effect of Model Batters Added with Arabinoxylans

Regarding the AX-added batters, two maize AX extracts differing in ferulic acid content and substitution patterns (see Table A1) were used to generate networks of differing strengths. The control batter with MAX reached a G′_t=30min of 1058 Pa (tan δ = 0.218). The addition of urea and SDS caused a small reduction in G′ (968 and 1043 Pa, respectively), and there was a pronounced increase with NaCl (1276 Pa). While G′ remained largely unaffected, tan δ significantly increased with SDS addition, indicating a less structured batter matrix. This showed only little or no contribution of H- and hydrophobic interactions in the batter, as is also seen in the past literature [16]. NaCl likely reduced intermolecular repulsion by electrostatically shielding the negatively charged ferulic acid carboxyl groups, promoting tighter AX associations [15].

Control batters with M-XEAX (see Figure 1) showed the lowest G′_t=30min of 724 Pa (tan δ = 0.239). Unlike MAX, M-XEAX batters exhibited opposite responses to the chemicals, with a pronounced increase in G′ upon SDS addition (963 Pa) and a decrease with NaCl (619 Pa). In contrast to MAX batters, SDS treatment in M-XEAX systems was accompanied by lower tan δ values, indicating a more structured network. This behaviour was attributed to the structural differences of the AX (i.e., molecular weight, A/X ratio and phenolic acid side chains) [7,17] with M-XEAX likely containing more exposed hydrophobic moieties (e.g., phenolic residues, unsubstituted xylose residues). Urea addition caused a small G′ reduction (G′_t=30min: 508 Pa) similar to MAX, highlighting that H-bonds were important in maintaining the elasticity of the batter.

In M-XEAX batters, enzymatic treatment increased G′ by 19% with SDS and 24% with urea; however, it led to a 16% reduction when adding NaCl. The higher ferulic acid content provided sufficient reactive sites for enzymatic coupling, enabling additional AX cross-linking when either hydrophobic or H-interactions were predominantly suppressed. In contrast, the low ferulic acid content in MAX limited enzymatic cross-linking in all batters, remaining generally low. Accordingly, only the enzyme-treated MAX control batter showed a significant increase in tan δ, while G′ remained largely unaffected. Overall, these findings emphasised that the structure and phenolic content of the AX strongly determined the type and strength of non-covalent interactions dominating batter stability.

3.1.3. Effect of Model Batters Added with Arabinoxylans and Protein

Batter stiffness (G′) increased in all mixed-polymer formulations (see Figure 1), accompanied by overall lower tan δ values (compared to control batters), with MAX–MGM showing the highest stiffness (G′_t=30min: 2664 Pa, tan δ: 0.162). In MAX–MGM batters, urea caused the largest G′ decrease (to 1361 Pa), followed by NaCl (1767 Pa) and SDS (2320 Pa). As observed for single MAX batters, only SDS significantly increased tan δ. This indicated a major contribution of H-bonds and electrostatic forces to network strength, while SDS promoted a more viscous-dominated behaviour. In contrast, H-bonds appeared less pronounced in batters containing only MGM or MAX. For M-XEAX–MGM batters (Figure 1), batter stiffness increased with SDS (+8.5%) and declined with NaCl (−8.5%) and urea (−23%). Compared with MAX–MGM batters, hydrophobic interactions destabilised M-XEAX–MGM batters and significantly affected viscoelastic behaviour, while H-bonds and electrostatic forces contributed similarly to structure. Enzyme addition reduced G′ in all AX–MGM batters except in MAX with SDS. Moreover, all MAX–MGM formulations showed increased tan δ values upon enzyme treatment. In these mixed-polymer systems, stable AX–protein coupling required both H-bonding and electrostatic interactions, while hydrophobic forces could partly hinder these associations.

3.2. Batter Expansion of Gluten-Free Batters

Figure 2 displays the effect of modulated non-covalent interactions on gas retention in batters containing AX and/or MGM (for raw data see Table S4).

3.2.1. Gas Retention in Protein-Based Model Batters

As SDS impairs yeast activity and thus gas formation, NaCl and urea were combined in batters to identify non-additive effects from which we could infer hydrophobic stabilisation. For MGM batters, the control expanded by 194%, decreasing slightly with the addition of NaCl (185%), urea (180%) and NaCl + urea (166%), which suggested only a small contribution from electrostatic and H-interactions. The predominant suppression of H-bonds likely weakened the hydrophilic protein regions extending into the aqueous, continuous batter phase, resulting in less efficient stabilisation of gas bubbles [18]. Enzyme treatment significantly increased gas retention in MGM batters after mainly suppressing H-bonds (urea), likely induced by protein unfolding, and enhanced the exposure of reactive sites for enzymatic cross-linking [19]. The predominant suppression of both electrostatic and H-interactions led to a drop in gas retention (−10.2%), attributed to partial protein unfolding and enhanced cross-linking, as well as the self-association of exposed hydrophobic domains.

In comparison to time sweep measurements, NaCl significantly reduced bulk elasticity over time but did not affect gas-holding capacity, indicating that electrostatic forces contributed to MGM batter stiffness rather than gas stabilisation. Moreover, urea addition reduced batter expansion without affecting G′, whereas its combination with enzymes resulted in the highest gas retention, despite a pronounced drop in elasticity.

3.2.2. Gas Retention in AX-Based Model Batters

AX-only batters behaved similarly, except for urea addition. Gas-holding ability was generally higher when MAX was added (MAX: 233%; M-XEAX: 194%) and significantly increased with NaCl (246%) but declined with urea (211%) and both reagents (218%), indicating that electrostatic shielding improved gas retention, likely by reducing repulsion among charged groups. In the case of M-XEAX, the expansion was higher with NaCl (206%) and urea (247%) but decreased when both reagents were combined (170%). Gas-holding differences among AX were attributed to the substitution pattern of M-XEAX, creating more interaction-prone sites. By partially hindering either H-bonds or electrostatic repulsions, AX may align and stabilise the gas bubbles more easily. Under these conditions, hydrophobic effects appear secondary to the roles of H- and electrostatic interactions.

Enzyme treatment of MAX batters significantly enhanced the batter expansion with NaCl (+19.3%) and urea (+15.9%), suggesting that electrostatic or H-interactions slightly hinder gas-cell stabilisation. The suppression of electrostatic interactions facilitated the formation of a more rigid network in batters containing M-XEAX, restricting gas expansion (−10.2%).

Consistent with rheological results, AX-only batters showed a positive relationship between elasticity and gas stabilisation, with MAX-based systems displaying both higher G′ values and greater expansion. Conversely, urea addition to M-XEAX batters reduced G′ while markedly improved expansion rates, demonstrating that gas stabilisation was not solely governed by bulk elasticity. Although enzymatic treatment increased G′ in M-XEAX systems containing SDS or urea, this trend was not observed in gas-holding capacity. In MAX batters, enzyme addition reduced elasticity, while NaCl- or urea-containing batters exhibited substantially improved gas stabilisation.

3.2.3. Gas Retention in AX and Protein-Based Model Batters

For mixed-polymer batters, MGM-M-XEAX batters showed the lowest expansion across all batters, with only minor changes upon reagent addition, in contrast to MGM-MAX, where higher effects were seen. This behaviour suggested that non-covalent interactions only contributed to gas stabilisation in the weaker MGM-MAX network, whereas in MGM-M-XEAX the higher cross-linked polymers likely compensated for these interactions during fermentation. Enzymatic treatment significantly increased gas-holding in both AX-MGM systems, irrespective of AX type, and behaved similarly to their non-treated controls. Interestingly, this enzyme-driven stabilisation was only observed in mixed control batters, but remained absent in the single-polymer formulations.

Although all AX-MGM batters exhibited consistently higher G′ values than AX-based systems, the incorporation of MGM led to reduced gas stabilisation in both mixed formulations. In MAX-MGM systems, all reagents decreased both elasticity and gas retention, with the extent dependent on the applied chemical. Conversely, in M-XEAX-MGM batters, the suppression of only hydrophobic interactions increased batter stiffness; however, gas retention appeared to be balanced by all non-covalent interactions. Across all mixed formulations, no consistent relationship between rheological properties and gas-holding capacity were observed upon enzyme treatment.

Overall, non-covalent contributions to batter elasticity, gas retention and covalent cross-linking efficiency were highly dependent on the AX extract and MGM addition. Furthermore, enhanced bulk elasticity did not necessarily correspond to improved gas retention, which may reflect interfacial stabilisation mechanisms that are not fully captured by viscoelasticity, also discussed in previous studies [20,21]. A summary of the dominant interaction patterns observed across the model batters, and their effects on rheological and gas-holding behaviour, is provided in

Table 3. Overview of non-covalent interactions and their influence on batter stability (G′, tan δ) as well as gas retention in model systems with commercial maize (MAX) or xylanase-extracted maize arabinoxylans (M-XEAX), with and without maize gluten meal (MGM) and enzymatic treatment.

Model Batters	Effects of Non-Covalent Interactions on Batter Stability (G′)	Gas Retention Behaviour	Relationship Between G′ and Gas Retention
MGM	Generally low elasticity: hydrophobic forces, H-bond suppression additionally reduced G′; increased tan δ with SDS	Minor changes; balanced contribution of non-covalent interactions	No clear relationship
MAX	Highest G′ among both AX-based systems; electrostatic shielding increased G′; minor stabilising effects of hydrophobic forces, H-bond; SDS increased tan δ (without major G′ changes)	Higher gas retention overall; electrostatic shielding improved expansion	Positive relationship
M-XEAX	Suppression of hydrophobic interactions increased G′; SDS reduced tan δ	Gas retention generally improved by suppression of electrostatic, hydrophilic interactions	No clear relationship
MAX-MGM	Highest G′ overall; mainly stabilised by H-bonds, hydrophobic forces; SDS significantly increased tan δ	Balanced contribution of non-covalent interactions; gas retention decreased compared to MAX batters	Inverse relationship
M-XEAX-MGM	Moderate G′ changes depending on interaction suppression; negative effect of hydrophobic interactions on batter stability; SDS increased tan δ despite higher stiffness	Lowest expansion; decreased compared to M-XEAX batters	No clear relationship
Enzyme-treated	Increased G′ in M-XEAX systems with suppressed hydrophobic, hydrophilic forces; suppression of hydrophobic interactions in mixed batters potentially increased enzyme efficiency; MAX(-MGM) exhibited higher tan δ without G′ enhancement.	Did not consistently improve gas retention; positive effects in MGM batters (+urea), MAX batters (+NaCl/urea), mixed systems	No clear relationship

Abbreviations: SDS refers to sodium dodecyl sulphate; NaCl to sodium chloride.

.

4. Conclusions

This study analysed the effects of non-covalent interactions on rheological behaviour and batter expansion in GF model systems formulated with different maize AX extracts and proteins. Overall, formulation type and polymer structure significantly affected interaction mechanisms, and thereby batter and gas-cell stabilisation.

MGM-based systems showed lower batter elasticity and were mainly controlled by hydrophobic and electrostatic forces; enzyme treatment did not strengthen these matrices unless electrostatic repulsion was reduced. In AX-only systems, viscoelastic behaviour depended strongly on the applied AX extract and structure: MAX showed little sensitivity to predominant H-bond or hydrophobic suppression and benefited from electrostatic shielding. M-XEAX, being higher in ferulic acid content and more branched, behaved oppositely, indicating a greater role for hydrophobic and H-interactions due to the more abundant cross-linking sites. When AX and MGM were combined, batter stiffness increased over single-polymer systems. However, enzyme addition revealed that effective AX–protein coupling was favoured by both H-bonding and moderate electrostatic shielding for batter stabilisation. Hydrophobic interactions were secondary or destabilising in polymer blends, especially for M-XEAX-MGM.

Batter expansion analyses did not consistently reflect changes in rheological behaviour. AX-only batters showed a positive, AX-dependent relationship between elastic behaviour and gas retention, governed by distinct non-covalent interactions. In contrast, mixed-polymer systems showed reduced gas stabilisation despite increased batter stiffness. Enzyme-treated batters further indicated that batter expansion was strongly dependent on the AX extract, MGM addition and the prevailing non-covalent interactions, and differed from trends observed in bulk elasticity.

These results provided the first insights into how substrate-dependent non-covalent interactions influenced batter and gas-holding stability in GF model systems. Nevertheless, these findings are restricted to the applied maize-derived AX and protein extracts and to simplified model batters rather than commercial GF formulations containing additional components. Future research should therefore investigate a broader spectrum of AX and protein substrates and evaluate their behaviour in more complex batters to better understand how the structural properties of different polymers from various sources affect intermolecular interactions, batter strength and overall stability.